article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
over the past three years , two of the largest - ever hst stellar spectroscopic guest observer programs have been undertaken . the `` advanced spectral library ( astral ) '' project ( pi = t. ayres ) consists of two treasury programs , the first in cycle 18 on `` cool stars '' ( go-12278 ) and the second in the current cycle 21 on `` hot stars '' ( go-13346 ) .
these programs are designed to collect a definitive set of representative , high - resolution ( r@xmath030,000 - 100,000 ) , high signal / noise ( s / n@xmath1100 ) spectra , with full uv coverage ( @xmath01150 - 3100 ) of prototypical stars across the hr diagram .
the space telescope imaging spectrograph ( stis ) is ideally suited for this purpose and is the sole instrument used for the acquisition of the program spectra .
the 8 f - m evolved late - type stars observed in astral / cool stars include iconic objects like betelgeuse , procyon , and pollux .
in addition , this sample has been supplemented with hst observations from a separate chandra program , so that two main sequence stars ( @xmath2 cen a & b ) are included in this definitive dataset .
astral / hot stars is in the process of observing 21 early - type stars , which span a broad range of spectral types between early - o and early - a , including main sequence and evolved stars , fast and slow rotators , as well as chemically peculiar and magnetic objects .
the targets include equally iconic sirius , vega , and the classical wind source zeta puppis .
all of these extremely high - quality stis uv echelle spectra will be available from the hst archive .
high - level science products , produced by the astral science team using the best - available techniques to calibrate and merge the individual spectra into a definitive , single spectrum , can be found at http://casa.colorado.edu/@xmath0ayres / astral/. in section 2 we describe the details of the observing programs and spectra now available for community use . in section 3
we present some illustrative examples of the on - going scientific analyses by the astral science team , including a study of the outer atmospheres and winds of the two evolved m stars in the sample , the m3.4 iii giant gamma cru and the m2 iab supergiant alpha ori and a first look at a `` high definition '' uv spectrum of the magnetic chemically peculiar `` ap '' star hr 465 .
these serve as examples of the kind of research that these data can enable , as the data will support investigations of a broad range of problems stellar , interstellar , and beyond by many investigators , for many years into the future .
in this section , we describe the content of these programs and the observing strategies used to optimize the final , high - level science products being delivered .
the cool stars program was carried out in hst cycle 18 from october 2010 through november 2011 .
it was awarded 146 orbits to record eight bright late - type stars , including well - known stars like betelgeuse and procyon .
cool stars are faint in uv , requiring @xmath020 orbits per target to achieve good s / n at full wavelength coverage with the stis echelles .
the resolution of the spectra vary from r = 30,000 to 110,000 depending on wavelength and target brightness , as the goal was to obtain the highest resolution possible with reasonable exposure times in a given wavelength regime .
this sample has been supplemented with observations of equivalent quality from a separate chandra program , which had hst observations as a component , so that two main sequence stars ( @xmath2 cen a & b ) are included in this dataset , along with the 8 evolved stars .
the science team includes many co - investigators , covering a variety of topical areas and providing complementary ground - based observations and relevant atomic physics information for the analysis of the spectra .
the astral - i program was driven by a desire to enable the study of : 1 ) magnetic activity , which is most conspicuous at the highest energies ( fuv and x - rays ) , 2 ) stellar mass loss , which is most prominently seen in the uv resonance lines of abundant elements such as h i , o i , c ii , fe ii , and mg ii , 3 ) the interstellar medium , also seen in the same uv resonance lines , and 4 ) the complex atmospheric dynamics exhibited by these stars , due to the mysterious chromospheric and coronal heating mechanism(s ) , seen in the fuv hot lines " , such as si iii , si iv , c iv , etc . the resulting spectra , in compressed / downsampled form to permit display in a single figure ,
are illustrated in fig .
we note that these data are the spectral equivalent of the hubble imaging `` deep fields '' , in that one can zoom in on the final , co - added spectrum repeatedly and reveal more and more features and information , due to their very high s / n and resolution*. centauri a & b ) for comparison to the astral sample , which is mostly giants .
typical late - type stars show photospheric continuum longward of 2000 ( and mg ii @xmath32800 emission ) , but shortward display numerous chromospheric emission lines .
strong 1300 - 1600 absorptions in @xmath2 ori are cold circumstellar co bands . ] the observing strategy used to maximize the quality of these spectra , in terms of s / n and resolution , within the time available , is shown in fig .
2 . the primary guidelines used to formulate this strategy were to : 1 ) use highest - resolution mode / setting available and narrowest slit possible , within a reasonable total exposure time , 2 ) exposures were divided into subexposures to mitigate fixed pattern noise , and 3 ) at least one observation of each type was scheduled near a peak - up to ensure the best possible flux and wavelength calibrations .
post - processing of the spectra corrects wavelength distortions identified in previous studies of stis wavecal lamps , updated with new calstis dispersion solutions from t. ayres `` deep lamp project '' and
special techniques were used to optimally splice and merge the orders of the echellegrams .
the hot stars program focuses on the warm half of the h - r diagram .
it was awarded 230 orbits in the current cycle 21 , starting in august 2013 and continuing through most of 2014 .
the program observes 21 targets , including star - gazer favorites vega , sirius , and regulus , and classic windy - star zeta puppis ; normal and chemically peculiar stars , fast and slow rotators ; magnetic and other exotica .
eleven targets have been fully observed to date .
the program spectra are illustrated schematically in fig .
3 , which shows compressed astral spectra in color and `` place - holder '' iue spectra in black for stars not yet observed in the astral - ii program . we summarize in fig .
4 the observing strategy used to optimize the signal - to - noise and resolution of each spectrum within the allocated observing time . we note in particular that this program was only made feasible due to the execution of a special go calibration program , between the times of the astral - i and astral - ii programs , to commission several previously `` available but unsupported '' neutral density slits on stis .
the previously supported slits had a huge gap in attenuation .
the standard , clear slits transmit too much light and would have saturated the detectors , while the supported neutral density slits ( nd=2 and nd=3 ) attenuate the light by 2 and 3 orders of magnitude ( factors of 100x and 1000x ) , respectively , and would have required too much observing time to be practical .
t. ayres in go program 12567 ( bridging stis s neutral density desert " ) therefore calibrated a set of `` available but unsupported '' nd ( neutral density)-filtered long slits ( 31x0.05nda / ndb / ndc ) that can be used with the stis echelles , and which provide intermediate attenuations between the standard ( nd=0 ) spectroscopic slits ( 0.2x0.06 , 0.2x0.09 , or 0.1x0.03 ) and the ( only two ) supported nd slits : 0.2x0.05nd ( nd=2 ) and 0.3x0.05nd ( nd=3 ) .
these intermediate nd slits provide attenuations of nd=0.6 , 1.0 , and 1.4 ( factors of 4x , 10x , and 25x ) .
the use of these slits can be seen in fig .
4 . in this figure
, tall rectangles represent 96-minute hst orbits ; 3 - 4 orbits constitute a `` visit '' ( circled numerals ) .
colored rectangles indicate echelle settings ; vertical size is wavelength coverage .
the shorter rectangles are high - resolution settings , taller are medium - resolution settings , and horizontal size is exposure duration .
the lower hatched zone highlights fuv ( 1150 - 1700 ) ; upper clear band is nuv ( 1700 - 3100 ) .
the guidelines for the observing strategies were similar to those used for astral - i : 1 ) use highest - resolution mode / setting and narrowest spectral slit possible within reasonable exposure times , 2 ) break up exposures to mitigate fixed pattern noise , and 3 ) schedule at least one observation of each type near peak - up .
in addition , special neutral density slits , as noted above were used for brighter objects .
here we present a couple of examples of some of the early science results from the astral - i and astral - ii programs .
one of the studies being performed as part of the astral - i program is an investigation of the dynamics and wind of the coolest stars . as part of this
, we have used observations of fe ii lines of different strength to map the wind acceleration in the two m - stars in the astral - i sample , @xmath2 ori ( m2 iab ) and @xmath4 cru ( m3.4 iii ) .
the fe ii profiles vary with the intrinsic strength of the lines , as the photons detected at earth originate higher in the atmosphere of the star for lines of higher opacity .
therefore , as we view lines of increasing line strength , two things happen : 1 ) the chromospheric emission lines increase in width due to opacity - broadening , and 2 ) the overlying wind absorptions become stronger and shift to bluer wavelengths as the wind velocity increases with height .
this behavior is shown in fig .
5 , which shows that the @xmath2 ori wind acceleration appears small or non - existent compared to that seen by carpenter et al .
( 1997 , apj , 479 , 970 ) in ghrs observations taken in 1992 , while the @xmath4 cru wind seems to be very similar to that seen previously in ghrs spectra .
the astral - ii program includes observations of the ap star hr 465 , which was chosen as a prototypical example of an a - type magnetic cp star .
this spectrum is of extremely high quality in terms of both signal / noise and resolution and reveals a bright continuum that is absolutely blanketed by a sea of absorption features - so much so , that it is doubtful that a true continuum point exists anywhere in the 1200 - 3200 range .
hr 465 has a global magnetic field @xmath02200 gauss , which varies with a period of @xmath023.3 years @xcite .
it has a spectral type of b9pe ( simbad ) , and a t@xmath5 @xmath010,723 k @xcite and log(g ) = 4.01 @xcite .
the strength of the cr and eu features are also known to vary with a @xmath023 year period @xcite . @xcite analyzed iue spectra and found strong iron - peak element lines but an underabundance on some ions of low atomic number , such as c. pt ii was seen in 1970 s iue data , but au ii , hg ii , and bi ii were uncertain and rare earth elements appeared weaker than in optical spectra .
we have begun determinations of element abundances , concentrating on the heavy ( exotic ) elements of most interest , in the astral - ii spectrum of hr 465 , which was taken during rare earth element ( ree ) minimum state .
our abundance measurements are obtained by comparison of the observed spectra with synthetic spectra , as shown in figures 6 and 7 . | the `` advanced spectral library ( astral ) '' project ( pi = t. ayres ) consists of two treasury programs : the cycle 18 `` cool stars '' ( go-12278 ) program and the cycle 21 `` hot stars '' ( go-13346 ) program .
the primary goal of these programs is to collect , for the use of the astronomical community over the coming decades , a definitive set of representative , high - resolution ( r@xmath030,000 - 100,000 ) , high signal / noise ( s / n@xmath1100 ) spectra , with full uv coverage ( @xmath01150 - 3100 ) of prototypical stars across the hr diagram , utilizing the high - performance space telescope imaging spectrograph ( stis ) .
the cycle 18 program obtained spectra of 8 f - m evolved late - type stars , while the cycle 21 program is in the process of observing 21 early - type stars , which span a broad range of spectral types between early - o and early - a .
all of these data will be available from the hst archive and , in post - processed and merged form , at http://casa.colorado.edu/@xmath0ayres / astral/. these data will enable investigations of a broad range of problems stellar , interstellar , and beyond for many years into the future .
we describe here the details of the observing programs , including the program targets and the observing strategies utilized to optimize the quality of the spectra , and present some illustrative examples of the on - going scientific analyses , including a study of the outer atmospheres and winds of the two evolved m stars in the sample and a first look at a `` high definition '' uv spectrum of a magnetic chemically peculiar `` ap '' star . |
we are concerned with the evolution of compact hypersurfaces @xmath0 moving according to the general non - local law of propagation @xmath1 where @xmath2 is the normal velocity of @xmath3 which depends , through the evolution law @xmath4 , on time , on the position of @xmath5 , on the set @xmath6 enclosed by @xmath3 , on the unit normal @xmath7 to @xmath3 at @xmath8 pointing outward to @xmath6 and on its gradient @xmath9 which carries the curvature dependence of the velocity . when such motion is local , i.e. , when @xmath4 does not depend on @xmath6 , and satisfies _ the inclusion principle _ or _
geometrical monotonicity _ ,
i.e. , when , at least formally , the inclusion @xmath10 at time @xmath11 implies @xmath12 for any @xmath13 , it is proved by souganidis and the first author @xcite that the motion can be defined and studied by the _ level set approach _ , which was introduced by osher and sethian @xcite for numerical calculations and then developed , from a theoretical point of view , by evans and spruck @xcite for the mean curvature motion and by chen , giga and goto @xcite for general velocities .
this approach replaces the geometrical problem with a degenerate parabolic partial differential equation called the _ geometric _ or _ level set equation .
_ this equation is designed to describe the desired evolution via the 0-level set of its solution .
more precisely , the existence and uniqueness of the level set solution @xmath14 allows to define @xmath3 as being the set @xmath15 . in recent years , there has been much interest on the study of front propagations problems in cases when the normal velocity of the front depends on a non - local way of the enclosed region like .
this interest was motivated by several types of applications like dislocations theory or fitzhugh - nagumo type systems or volume dependent velocities that we describe below .
it is worth pointing out that the level set approach still applies for motions with nonlocal velocities provided that the inclusion principle holds , following the ideas of slepcev @xcite .
but , in many of the above mentioned applications , one faces non - monotone surface evolution equations . for such class of problems
, the level set approach can not be used directly since the classical comparison arguments of viscosity solutions theory fail and therefore , the existence and uniqueness of viscosity solutions to these equations become an issue .
though the existence properties for such motions seem now to be well understood ( see @xcite ) , this is not the case for uniqueness . in particular , there are not many uniqueness results for curvature dependent velocities . as far as the authors know , there are only two works by forcadel @xcite and forcadel and monteillet @xcite which investigate the motion arising in a model for dislocation dynamics which is included by our general equations .
the aim of this article is to consider cases where we have , at the same time , a non - local velocity which induces a non - monotone evolution together with a curvature dependence ( we explain later on the state of the art for such problems and why the curvature dependence creates a specific difficulty ) . more specifically , we describe a method to show short time uniqueness results for the general motion .
we now describe some typical applications we have in mind .
we first consider a model for dislocation dynamics @xmath16 where @xmath17 denotes the indicator function of a subset @xmath18 of @xmath19 and @xmath20 , @xmath21}\to{\mathbb{r}}$ ] are given functions and we write @xmath22 here , @xmath23 denotes the @xmath24-dimensional unit sphere . the term @xmath25 is called the _ anisotropic _ ( or _ weighted _ ) _ mean curvature _ of @xmath3 at @xmath8 ( in the direction of @xmath7 ) .
see for instance giga @xcite .
typically , the reasonable assumptions in this context are the following : @xmath26 is a positive and bounded function , @xmath27 are bounded , continuous functions which are lipschitz continuous in @xmath8 variable ( uniformly with respect to @xmath28 variable ) , @xmath29;l^{1}({\mathbb{r}}^n))$ ] and @xmath30 is a positively homogeneous function with degree @xmath31 .
the surface evolution equation without the last term in the right hand side is well - known as typical models of the dislocation dynamics ( see @xcite for a derivation and the physical background ) .
we next consider asymptotic equations of a fitzhugh - nagumo type system as an example of interface dynamics coupled with a diffusion equations , @xmath32 where @xmath33 are bounded and lipschitz continuous with @xmath34 .
this system has been investigated by giga , goto and ishii @xcite and soravia and souganidis @xcite . finally , we consider equations depending on the measure of the fronts like @xmath35 where the function @xmath36 is lipschitz continuous .
a typical example is @xmath37 for some @xmath38 which has been investigated by chen , hilhorst and logak in @xcite ( see also @xcite ) . as we already mentioned it above , these examples are not only nonlocal but also non - monotone surface evolution equations .
indeed , in , the kernel @xmath39 may change sign and , in and in , the functions @xmath40 , @xmath41 may be non - monotone .
we also refer to @xcite for some monotone non - local geometric equations . by using the framework which we present in this paper , we give short time uniqueness results for , and .
there are many results of existence and uniqueness for the simplest case of motions of , i.e. , @xmath42 and @xmath43 without a curvature term .
a short time existence and uniqueness result was first obtained in @xcite .
but then most of the results were obtained for curvature - independent velocities ( @xmath44 ) : long time existence and uniqueness results were obtained when the velocity is positive , i.e. , @xmath45 by alvarez , cardaliaguet and monneau in @xcite and by the first two authors in @xcite by different methods .
the first two authors with cardaliaguet and monneau in @xcite presented a new notion of weak solutions ( see definition [ def weak sol ] ) of the level set equation for without a curvature term , gave the global existence of these weak solutions and analysed the uniqueness of them when holds .
a similar concept of solutions already appeared in @xcite . in the companion paper @xcite ,
the first two authors with cardaliaguet and monteillet proposed a new perimeter estimate for the evolving fronts with uniform interior cone property and by using this , they extended the uniqueness result for dislocation dynamics equations and provided the uniqueness result for asymptotic equations of a fitzhugh - nagumo type system , still under the positiveness assumption . in this paper , we do not use the perimeter estimate in an essential way but either elementary measure estimates or , in the most sophisticated cases , the interior cone property ( see lemma [ lem - for - unique ] ) . since the studies by @xcite , it is now well - known that estimates on _ lower gradient bound _ and perimeter of @xmath31-level sets of viscosity solutions of associated local equations are key properties to obtain existence and uniqueness results for nonlocal equations derived from , and .
let us describe the main difficulty of our problem and , to do so , we consider the level set equations of the simplest case of or here . considering the non - local part as a given function ,
we are led to the study the ( local ) initial value problem @xmath46 where @xmath47 and @xmath48})$ ] are bounded and lipschitz continuous with respect to the @xmath8 variable .
one of our main results is a short time lower gradient bound estimate for the viscosity solution of , i.e. , @xmath49 for first - order eikonal equations , lower gradient bound comes naturally from the barron - jensen s approach ( see @xcite ) . for second - order equations like
, it is affected by the _ diffusion _ " term and the _ non - empty interior _ difficulty and therefore we can not expect that the property holds generally and for long - time
. indeed , in @xcite , see also @xcite , they consider the simple example of with @xmath50 and smooth @xmath51 such that @xmath52 on the initial front @xmath53 they prove that , up to choose suitable @xmath51 , fattening may occur for arbitrary @xmath54 i.e. , the front may develop an interior .
it is precisely this reason which implies that there are not many results on the nonlocal second - order equations like the level sets equations of , and and a short - time result is optimal .
existence results were obtained by @xcite but they concern merely existence of weak solutions defined by definition [ def weak sol ] . as stated above , there are two works @xcite which give uniqueness results for the motion . the difference between our results and
theirs is that , in @xcite , only the evolution of hypersurfaces which can be expressed by graphs of functions is considered while , in @xcite , the arguments are based on minimizing movement for and they are completely different from our arguments which are based on the theory of viscosity solutions .
moreover , for the existence of minimizing movement for the simplest case of , the assumptions that @xmath55 is symmetry and @xmath27 are smooth enough are essentially used .
therefore , uniqueness results for the examples and are not covered by @xcite . another difference with existing results in the literature is that @xmath4 is allowed to change sign in , contrary to @xcite where is one of the main assumption to get uniqueness
. it may give rise of fattening , see ( * ? ? ?
* proposition 4.4 ) , and it is another explanation of the short time result . finally , we explain the key idea to obtain for viscosity solutions of . in order to get it , we make the following assumption on @xmath56 .
there exist constants @xmath57 , @xmath58 and @xmath59 such that @xmath60 for all @xmath61 $ ] . then we prove that such a property is preserved for the solution of , at least for short time , i.e. , @xmath62 for all @xmath63 $ ] , @xmath64 $ ] and some @xmath65 , @xmath66 $ ] , where @xmath67\to[0,\infty)$ ] is a non - increasing continuous function such that @xmath68 the assumption on @xmath51 is inspired by ( * ? ?
* theorem 4.3 ) , where it is formulated only for the sign - distance function . a similar result to
may be found in @xcite where it is used to prove uniqueness results for the mean curvature motion for entire graphs . if @xmath51 is a smooth function with @xmath69 on the compact hypersurface @xmath70 , then the assumption is satisfied with @xmath71 and if there exists a smooth solution of the level set equation , then holds for short time .
but , on one hand , the general degenerate parabolic and nonlinear equations we consider do not have classical solutions in general , and on the other hand , the above assumption on @xmath51 is valid in cases when @xmath72 is not a smooth hypersurface , which is also an important point here .
the proof of uses in a crucial way the geometric property of and a _ continuous dependence result _ for parabolic problems ( which is , by the way , of independent interest ) .
we refer to @xcite and references therein for the detail of the continuous dependence result for elliptic and parabolic problems .
we derive lower gradient estimate from formally here .
we have @xmath73 for all @xmath64 $ ] with @xmath74 as @xmath75 . dividing @xmath76 in the above and taking a sufficiently small @xmath77 $ ] ,
we get the lower estimate .
we also obtain the interior cone property of fronts by .
the paper is organized as follows : in section 2 , we state a continuous dependence result for a class of equations which encompasses level set equations associated to . in section 3
, we obtain the key estimate and derive the lower - bound gradient and perimeter estimates of @xmath31-level sets of viscosity solutions of local equations . in section 4 , we consider the _ level set _ equation of and give the proof for the short time uniqueness result ( theorem [ uniqueness ] ) .
section 5 is devoted to existence and uniqueness results for the level set equations of , and as applications of theorem [ uniqueness ] .
* notations . * for some @xmath78
, we denote by @xmath79 the @xmath80-dimensional euclidean space equipped with the usual euclidean inner product @xmath81 , and by @xmath82 the space of @xmath83 symmetric matrices .
we write @xmath84 for @xmath85 @xmath86 and @xmath87 for @xmath88 .
the symbols @xmath89 and @xmath90 denote the @xmath80-dimensional lebesgue and hausdorff measures , respectively .
we write @xmath91 for the transpose of the matrix @xmath92 and @xmath93 . finally ,
for @xmath38 , we write @xmath94 and @xmath95
. * acknowledgements .
* we are grateful to a. chambolle , e. jakobsen and l. rifford for their comments and advice .
this work was done while h. mitake was visiting the laboratoire de mathmatiques et physique thorique , universit de tours .
his grateful thanks go to the faculty and staffs .
in this section , we are concerned with the equation @xmath96 @xmath97 , @xmath98 is the unknown function , @xmath99 , @xmath100 and @xmath101 stand respectively for its time and space derivatives , and hessian matrix with respect to @xmath8 variable .
we use the following assumptions .
* @xmath102\times({\mathbb{r}}^{n}\setminus\{0\ } ) \times{\mathcal{s}}^{n})$ ] . *
the equation is _ degenerate parabolic _ ,
i.e. , @xmath103 for any @xmath104\times({\mathbb{r}}^{n}\setminus\{0\})$ ] and @xmath105 with @xmath106 , where @xmath107 stands for the usual partial ordering for symmetric matrices . * for any @xmath108}$ ] , @xmath109 , where @xmath110 ( resp . , @xmath111 )
is the upper - semicontinuous envelope ( resp . , lower semicontinuous envelope ) of @xmath112 .
* there exist @xmath113 , @xmath114 such that @xmath115 for any @xmath116 , @xmath117 , @xmath118 , @xmath119 , @xmath105 and some @xmath120 satisfying @xmath121 where @xmath122 with @xmath123 .
we note that , in this section , we do not assume that @xmath112 is geometric .
[ continuous dependence ] let @xmath124 be functions on @xmath125\times({\mathbb{r}}^{n}\setminus\{0\ } ) \times{\mathcal{s}}^{n}$ ] satisfying assumptions ( a1)(a4 ) .
let @xmath126})$ ] be , respectively , a bounded viscosity subsolution and viscosity supersolution of with @xmath127 for @xmath128 .
assume that there exists @xmath129 such that @xmath130\ ] ] for either @xmath131 or @xmath132 , and that there exists @xmath117 such that @xmath133\ ] ] for both @xmath131 and @xmath134 then there exists @xmath135 which depends only on @xmath136 , @xmath137 , @xmath138 and @xmath139 such that @xmath140 for all @xmath141 $ ] .
an assumption like ( a4 ) is natural in viscosity theory to obtain continuous dependence results of the type and the regularity of the solution ( cf . )
is a key ingredient too , see @xcite .
in example [ ex-1 ] below , we show that ( a4 ) holds in the cases we are interested in .
note that are not restrictive assumptions when dealing with front propagation problems , see @xcite .
let @xmath142 and @xmath143 .
we shall later fix @xmath144 .
consider @xmath145 } \{u_1(x , t)-u_2(y , t)- \frac{|x - y|^4}{{\varepsilon}^4}-kt\}.\ ] ] noting , it is clear that the supremum is attained at @xmath146 $ ] for small @xmath147 .
we consider the case where @xmath148 $ ] . in view of ishii s lemma , for any @xmath149
, there exist @xmath150 and @xmath151 ( see @xcite for the notation ) such that @xmath152 where @xmath18 is the matrix defined by .
the definition of viscosity solutions immediately implies the following inequalities : @xmath153 hence we have @xmath154 using that @xmath155 or @xmath156 is lipschitz continuous with respect to @xmath8 variable , we get , by standard estimates , @xmath157})}$ ] , @xmath158})}$ ] and @xmath137 .
we now distinguish two cases : ( i ) for any @xmath142 , @xmath159 ; ( ii ) there exist @xmath160 such that @xmath161 as @xmath162 , @xmath163 for any @xmath164 .
we first consider case ( i ) . in view of ( a4 ) , we have @xmath165 sending @xmath166 , we get @xmath167 in case ( ii ) , we have @xmath168 . due to
, we have @xmath169 , @xmath170 and @xmath171 . by ( a2 ) , we have @xmath172 therefore ,
we get @xmath173 set @xmath174 and then the two above cases can not hold ; this means that necessarily we have @xmath175 .
therefore , for any @xmath176 $ ] , @xmath177 an optimization with respect to @xmath147 yields @xmath178 for some @xmath179 .
[ ex-1 ] we consider the functions @xmath180}\times({\mathbb{r}}^n\setminus\{0\ } ) \times{\mathcal{s}}^{n}$ ] defined by @xmath181 for @xmath128 , where @xmath182 are compact metric space and @xmath183 , @xmath184 are , respectively , real - valued functions and @xmath185 matrix valued functions for some @xmath186 on @xmath125\times({\mathbb{r}}^{n}\setminus\{0\})$ ] with a possible singularity at @xmath187 we assume that the functions @xmath188 , @xmath184 satisfy the following conditions by replacing @xmath4 by @xmath188 , @xmath184 for any @xmath189 , @xmath190 , @xmath128 , respectively : @xmath4 are continuous on @xmath191}$ ] and for some @xmath192 ( independent of @xmath193 ) , @xmath194 for all @xmath195 , @xmath141 $ ] , @xmath196 .
let @xmath116 , @xmath118 , @xmath197 , @xmath105 satisfy for some @xmath117 and let @xmath18 be the matrix given by .
we omit the dependence of @xmath193 for simplicity of notation .
we calculate that @xmath198 where @xmath199 and @xmath200 for some @xmath201 , where @xmath202 is the canonical basis of @xmath19 , @xmath203 and @xmath204 . due to
, we have @xmath205 from the above computations , it follows that the inequality ( a4 ) holds by replacing @xmath206 and @xmath207 by @xmath208 and @xmath209 , respectively .
therefore , if the @xmath210 s are solutions of with @xmath127 given by , @xmath211 then the conclusion of theorem [ continuous dependence ] holds and reads @xmath212 for all @xmath141.$ ] finally , note that , applying theorem [ continuous dependence ] with @xmath213 gives comparison and uniqueness for .
[ appr - arg ] for the applications we have in mind , a continuous dependence result for equations with a measurable dependence in time will be needed .
we do not state a precise result here but we mention that it can be obtained by an easy approximation argument .
we consider the initial value problem in this section @xmath214 we make the following assumption on @xmath56 throughout this section
. * @xmath215 and @xmath216 for all @xmath217 and there exists @xmath218 such that @xmath219 for all @xmath220 .
* there exist constants @xmath221 , @xmath58 and @xmath222 such that @xmath223 , \end{aligned}\ ] ] where @xmath224 . [ rmk - c1-lisse - geom ] without loss of generality , we may assume that @xmath225 is a smooth bounded lipschitz continuous function and henceforth we will assume it from now on . indeed , let @xmath226 for @xmath142 be an approximate function of @xmath225 , then we have @xmath227 if @xmath228 is enough small , then we have @xmath229.\ ] ] let @xmath230 such that @xmath231 then , for @xmath232 and @xmath233 enough small , @xmath52 in @xmath234 and , setting @xmath235 we have @xmath236 where @xmath237 is a modulus of continuity of @xmath238 in @xmath239 and @xmath240 therefore ( i2 ) holds for @xmath241 and @xmath242 $ ] for @xmath243 enough small .
moreover , the implicit function theorem implies that @xmath72 is a @xmath244 hypersurface .
conversely , assume that @xmath72 is a @xmath244 hypersurface with the _ unique nearest point property _ ( that is , there exists a neighborhood @xmath245 of @xmath72 such that , for all @xmath246 there exists a unique @xmath247 such that @xmath248 ) .
then the signed distance function @xmath249 to @xmath72 is @xmath244 ( see @xcite ) .
it follows that ( i1 ) , ( i2 ) hold with @xmath51 such that @xmath250 in a neighborhood of @xmath72 and @xmath51 is a suitable regularization of @xmath249 elsewhere . more generally , when considering front propagation problems , it may be convenient to have a characterization of ( i2 ) in geometrical terms .
such a result does not seem obvious .
however , we have partial results in the following lemma , the proof of which is given in the appendix with additional comments .
a subset @xmath251 is _ star - shaped _ with respect to @xmath252 if , for every @xmath253 the segment @xmath254 belongs to @xmath255 it is _ star - shaped with respect to a ball _
@xmath256 if @xmath18 is star - shaped with respect to every @xmath257 [ i2geom ] let @xmath258 be an open bounded set with boundary @xmath259 * ( star - shaped with respect to a ball domains ) the set @xmath260 is star - shaped with respect to a ball , i.e. , there exists a compact subset @xmath261 and @xmath262 such that @xmath263 } { \overline}{b}(\alpha x , ( 1-\alpha)r_0),\ ] ] if and only if there exists @xmath264 such that @xmath265 and ( i1 ) , ( i2 ) hold in @xmath245 with @xmath266 in this case , @xmath72 is locally the graph of a lipschitz continuous function . *
if there exists @xmath143 such that @xmath72 is locally the graph of a lipschitz continuous function with constant @xmath267 then there exists @xmath51 such that , ( i1 ) and ( i2 ) hold .
hereinafter , we set @xmath268 from remark [ rmk - c1-lisse - geom ] , we may assume that @xmath225 is a smooth bounded lipschitz continuous function and , replacing @xmath243 by a smaller constant in order that @xmath269 we obtain that @xmath270 is a @xmath244-diffeomorphism in @xmath19 with @xmath271 we assume ( a1)(a3 ) and make the following additional assumptions on @xmath112 throughout this section . *
the function @xmath112 is _ geometric _ , i.e. , @xmath272 for all @xmath273 , @xmath274 , @xmath275}\times({\mathbb{r}}^{n}\setminus\{0\})\times{\mathcal{s}}^{n}$ ]
. * there exists @xmath276 such that @xmath277 for any @xmath278}\times({\mathbb{r}}^{n}\setminus\{0\})$ ] and @xmath105 . * for any @xmath117 , there exists @xmath279 such that @xmath280 for any @xmath281 $ ] and for any @xmath282 , @xmath118 , @xmath141 $ ] , @xmath119 , @xmath105 satisfying and @xmath283 given by .
* there exists at least one viscosity solutions of which satisfies @xmath284\ ] ] for some @xmath285 .
let us make some comments about these new assumptions : ( a5 ) is needed to use the level set approach to describe front propagation ( see @xcite for instance ) . assumption ( a6 ) is satisfied for a wide class of quasilinear equations under interest in this paper , see example [ ex-2 ] .
a consequence of ( a5 ) and ( a6 ) is : for any @xmath117 , there exists @xmath286 such that @xmath287\times ( { b}(0,r)\setminus \{0\ } ) \times{\mathcal{s}}^{n},\ ] ] which is a crucial property to obtain hlder continuity in time for the solutions of , see proposition [ regularity ] .
assumption ( a7 ) is a natural condition to obtain a preservation of the initial property ( i2 ) during the evolution .
this condition is related to ( a4 ) ; it is worthwhile to notice , as it was done at the end of example [ ex-1 ] , that such a condition gives uniqueness for the solutions of .
existence of solutions to is assumed in ( a8 ) because it is not the point in this paper , see @xcite for some conditions which guarantee existence .
more precisely , we have the following result about solutions of and the proof is given in appendix : [ regularity ] there exists a unique viscosity solution @xmath288})$ ] of and we have @xmath289 , @xmath290 $ ] , where @xmath291 are positive constants which depend only on @xmath292 and @xmath293 respectively .
now , we state the main result of this section .
[ key estimate ] there exist @xmath294 @xmath295 ( @xmath243 is given by ( i2 ) and satisfies and a non - increasing continuous function @xmath67\to[0,\infty)$ ] which depend only on @xmath296 @xmath297 @xmath298 @xmath299 @xmath300 @xmath301 @xmath302 @xmath303 such that @xmath304 and @xmath305 satisfies @xmath306,\ ] ] where @xmath307 in the appendix , we can extend in @xmath308 @xmath309,\end{aligned}\ ] ] where @xmath310 and @xmath311 are introduced in lemma [ psi ] .
we first prove @xmath312 for all @xmath108}$ ] , @xmath313 $ ] , some constant @xmath314 , which is depends only on @xmath315 @xmath316 and @xmath317 ( note that @xmath318 does not depend on @xmath303 contrary to @xmath310 which depends on @xmath303 through @xmath243 because of ) .
fix @xmath313 $ ] . set @xmath319 and @xmath320 for all @xmath108}$ ] . since @xmath112 is geometric and @xmath311 is a nondecreasing function , the functions @xmath321 satisfy v_t+ h(_(x),t , d_(_(x))^tdv(x , t ) , + d_(_(x))^td^2v(x , t)d_(_(x ) ) + d^2_(_(x))dv(x , t ) ) = 0 & in @xmath322 , + v(x,0)=u_0(_(x ) ) & in @xmath19 , w_t+h(x , t , dw , d^2w)=0 & in @xmath322 , + w(x,0)=(u_0(x)+_0 ) & in @xmath19 in the viscosity sense ( see ( * ? ? ?
* theorem 4.2.1 ) for instance ) .
let @xmath323 be the constant in ( a8 ) and recall that @xmath324 in lemma [ psi ] .
for any @xmath325 $ ] , @xmath326 , which implies that @xmath327 .
therefore , we only need to show that for any @xmath328 $ ] inequality holds .
note that @xmath329 by assumptions ( a6 ) , ( a7 ) and theorem [ continuous dependence ] with @xmath330 and @xmath331 , we get , for any @xmath332 $ ] , @xmath333 for all @xmath141 $ ] , which implies . setting @xmath334 and @xmath335 for all @xmath64 $ ] , we obtain the conclusion .
the first important consequence is a lower - gradient bound estimate on the front .
[ lower bound ] we have @xmath336 where @xmath337 and @xmath338 are given in theorem [ key estimate ] .
theorem 3.1 in @xcite implies that we can not expect global in time lower gradient estimates for solutions of with general initial data like ( i1 ) , ( i2 ) , even if we assume some positiveness assumptions on the velocity like in @xcite . before giving the proof of this result ,
we continue by stating another consequences of theorem [ key estimate ] .
we need to introduce some notations . for any @xmath141 $ ] ,
@xmath339 $ ] , we set @xmath340 and define the cone with vertex @xmath341 , axis @xmath342 and parameters @xmath343 by @xmath344 } { \overline}{b}(z+ae , a\,\frac{\rho}{\theta } ) \\ { } & = \ , \{z+ae+a\,\frac{\rho}{\theta}\xi \mid a\in[0,\theta],\xi\in{\overline}{b}(0,1)\}. \end{aligned}\ ] ] the following result means that the evoluting fronts have the interior cone property . [ cone ] for any @xmath345 $ ] and @xmath64 $ ] , @xmath346 where @xmath347 when @xmath18 is a subset of @xmath348 we will write , by abuse of notation , @xmath349 for the perimeter of @xmath18 .
notice that it does not always correspond to the usual definition of perimeter .
the two definitions coincide for instance when the boundary is locally the graph of a lipschitz function , which is often the case in our applications . for further details , see ( * ? ? ?
* section 5 and remark p.183 ) or @xcite .
[ estimate perimeter ] there exists a constant @xmath350 which depends only on the constants appearing in theorem [ key estimate ] such that @xmath351 for all @xmath345 $ ] and @xmath352.$ ] we turn to the proofs .
take any function @xmath353 satisfying @xmath354 and @xmath355 for all @xmath356 for some @xmath357 , where @xmath337 is given by theorem [ key estimate ] . by theorem [ key estimate ] and mean - value theorem ,
we have @xmath358 dividing by @xmath359 in the above and letting @xmath77 $ ] go to @xmath31 , we obtain the conclusion . fix @xmath345 $ ] , @xmath64 $ ] and @xmath360 . by theorem [ key estimate ]
, we have @xmath361.\ ] ] set @xmath362 . for any @xmath363 , we have @xmath364 which implies that @xmath365 for any @xmath313 $ ] .
therefore , we have @xmath366 } { \overline}{b}(z+{\lambda}\nu(z ) , { \lambda}\frac{\rho(t)}{{\overline}{{\lambda } } } ) = \bigcup_{a\in[0,\theta(z ) ] } { \overline}{b}(z+a\frac{\nu(z)}{|\nu(z)| } , a\frac{\rho(t)}{\theta(z ) } ) \subset { \overline}{{\omega}}_{t}^{r}.\ ] ] before doing the proof of corollary [ estimate perimeter ] , we recall the following lemma .
[ perimeter ] let @xmath367 be a compact subset of @xmath19 having the interior cone property of parameters @xmath368 and @xmath369 .
then there exists a positive constant @xmath370 such that for all @xmath117 , @xmath371 set @xmath372 .
let @xmath373 be the functions in corollary [ cone ] and set @xmath374 and @xmath375 } \theta(z)$ ] . by theorem [ key estimate ] and corollary [ cone ] , we see that @xmath376 and we have , @xmath377.\ ] ] due to lemma [ perimeter ]
, there exists @xmath378 such that @xmath379 for all @xmath380 $ ] , @xmath345 $ ] .
we end this section with an application .
[ ex-2 ] we consider the function @xmath381 where the functions @xmath382 and @xmath383 satisfy for all @xmath189 , @xmath190 , respectively .
we add the following assumptions on @xmath384 : @xmath385 for all @xmath386 , @xmath108}$ ] , @xmath387 and some @xmath388 .
these assumptions are related to ( a5 ) .
a typical example is @xmath389 and then the second - order term is the so - called mean curvature term .
we claim that the function @xmath112 satisfies ( a1)(a8 ) .
it is easy to check that the function @xmath112 satisfies ( a1)(a5 ) .
we check that the function @xmath112 satisfies ( a7 ) .
note that , by , , we have @xmath390 for @xmath391 small enough and any @xmath392 . by abuse of notations
, we write @xmath393 instead of @xmath384 for any @xmath394 .
we compute @xmath395 and @xmath396 for some @xmath397 and any @xmath398 and @xmath281 $ ] . by using the same computations as example [ ex-1 ] , we have @xmath399 for some @xmath400 and any @xmath282 , @xmath118 , @xmath141 $ ] , @xmath197 with @xmath401 , @xmath105 satisfying and @xmath283 given by , which implies that @xmath112 satisfies ( a7 ) .
we finally check that @xmath112 satisfies ( a8 ) . at first
, the constant function @xmath402 is obviously a subsolution of .
we set @xmath403 with @xmath404 and define the function @xmath405}\to{\mathbb{r}}$ ] by @xmath406 where @xmath407 .
we prove that @xmath408 is a viscosity supersolution of .
it is easily seen that @xmath409 on @xmath19 .
indeed , for all @xmath410 , we have @xmath411 and , for all @xmath412 , @xmath413 ( see ( i1 ) ) .
we have @xmath414 , @xmath415 and @xmath416 for any @xmath417 and @xmath418 .
note that @xmath419 for all @xmath141 $ ] .
set @xmath420 .
we calculate that @xmath421 set @xmath422 and take @xmath423 for @xmath424 so that @xmath425 is an orthonormal basis .
then we have @xmath426 and @xmath427 for @xmath424 .
therefore , @xmath428 since @xmath429 and @xmath430 by .
moreover , @xmath431 and @xmath402 is obviously a supersolution on @xmath432 .
setting @xmath433 , we see that ( a8 ) is satisfied in view of the comparison theorem for viscosity solutions of .
in this section , we consider the initial value problem of the nonlocal and non - monotone geometric equations which is derived from , through the _ level set approach _ ( see @xcite ) , @xmath434(x , t , du , d^{2}u)=0 & & \textrm{in } \ { \mathbb{r}^n\times(0,t ) } , \\ & u(\cdot,0)=u_0 & & \textrm{in } \
{ \mathbb{r}}^{n}. \end{aligned } \right.\ ] ] for any function @xmath435},[0,1])$ ] , @xmath436 $ ] denotes a real - valued function of @xmath437\times({\mathbb{r}}^n\setminus\{0\})\times{\mathcal{s}}^{n}$ ] . for almost any @xmath438 $ ] , @xmath439(x , t , p , x)$ ]
are continuous functions on @xmath440 with a possible singularity at @xmath163 . for all @xmath441 , @xmath442(x , t , p , x)$ ]
are measurable functions . for any @xmath435},[0,1])$ ] , @xmath436=h$ ] satisfies ( a2 ) , ( a3 ) , ( a5 ) .
furthermore , we make the following assumptions ( h1)(h5-(i ) ) or ( h5-(ii ) ) and ( i1 ) , ( i2 ) on @xmath56 throughout this section . * for any @xmath435},[0,1])$ ] , equation has a bounded uniformly continuous @xmath443-viscosity solution @xmath444 $ ] .
moreover , there exist constants @xmath445 independent of @xmath435},[0,1])$ ] such that @xmath446 for all @xmath108}$ ] and @xmath447 for all @xmath448 $ ] . * for any @xmath449 $ ] and @xmath450;l^{1}({\mathbb{r}}^n))$ ] such that @xmath451 is compact for any @xmath452 $ ] , @xmath436\in c({\mathbb{r}}^n\times[0,\tau ] \times({\mathbb{r}}^{n}\setminus\{0\})\times{\mathcal{s}}^{n})$ ] . * the functions @xmath436 $ ] satisfy ( a6 ) with @xmath453 $ ] uniformly for any @xmath435},[0,1])$ ] .
* the functions @xmath436 $ ] satisfy ( a7 ) with @xmath453 $ ] uniformly for any @xmath435},[0,1])$ ] . * for any @xmath117 , there exists @xmath279 such that @xmath454(y , t , p , y)| \nonumber\\ \le \ , & c_{h}\bigl(\frac{|x - y|^{4}}{{\varepsilon}^{4}}+ { \kappa}_{\chi_1 , \chi_2}(x , t ) + \frac{{\kappa}_{\chi_1 , \chi_2}^{2}(x , t)|x - y|^{2}}{{\varepsilon}^{4 } } + \rho\|a^{2}\| \bigr ) \label{assump main ineq}\end{aligned}\ ] ] for any @xmath455},[0,1])$ ] and for any @xmath282 , @xmath118 , @xmath141 $ ] , @xmath197 and @xmath105 satisfying and @xmath283 given by , where @xmath456 * one has inequality by replacing @xmath457 by @xmath458 where @xmath459 is the green function defined by @xmath460 * for any @xmath435},[0,1])$ ] , if @xmath461 , then the nonlinearity @xmath462(x , t , p , x)\ ] ] satisfies ( a1)-(a4 ) and @xmath463 \to u[\chi]$ ] uniformly in @xmath464 $ ] as @xmath465 .
assumptions ( h1)(h4 ) are modifications of ( a1)(a7 ) in order to be able to deal with the nonlocal equation . while ( h5-(i ) ) and ( h5-(ii ) ) are specially designed to encompass dislocation type equations or fitzhugh - nagumo type systems .
finally ( h6 ) is the assumption which allows to use theorem [ continuous dependence ] through an approximation argument ( cf .
remark [ appr - arg ] ) .
further detailed examples are given in section [ sec : applic ] .
we use the following definition of weak solutions introduced in @xcite which is inspired by @xcite .
[ def weak sol ] let @xmath466}\to{\mathbb{r}}$ ] be a continuous function .
we say that @xmath305 is a weak solution of if there exists @xmath435},[0,1])$ ] such that * @xmath305 is an @xmath443-viscosity solution of @xmath467(x , t , du , d^2u)=0 & \textrm{in } \
{ \mathbb{r}^n\times(0,t ) } , \\ & u(\cdot,0)=u_0 & \textrm{in } \ { \mathbb{r}}^n , \end{aligned } \right.\ ] ] * for almost every @xmath417 , @xmath468 moreover , we say that @xmath305 is a classical solution of if in addition , for almost all @xmath141 $ ] , @xmath469 [ weak is classical ] if there exists a weak solution @xmath288})$ ] of , then @xmath305 is classical in @xmath470 for some @xmath65 which depends on @xmath296 @xmath297 @xmath298 @xmath299 @xmath300 @xmath301 @xmath302 @xmath303 .
let @xmath471},[0,1])\times c({\mathbb{r}^n\times[0,t]})$ ] be an @xmath443-viscosity solution of .
we prove that @xmath472 for some @xmath65 .
we use ( h6 ) and set @xmath473 $ ] .
we recall that @xmath474 is the viscosity solutions of with @xmath475 for all @xmath476 . by the comparison theorem for local equations , proposition [ regularity ] we have @xmath477 , @xmath290 $ ] and some @xmath478 which is independent of @xmath479 . in view of ascoli - arzel theorem , the stability ( see @xcite ) and
the uniqueness ( see @xcite ) of @xmath443-viscosity solutions of , we have @xmath480 locally uniformly on @xmath191}$ ] for @xmath288})$ ] which is the @xmath443-viscosity solution of .
moreover , in view of corollary [ lower bound ] , we have @xmath481 for some @xmath65 . by the usual stability result of viscosity solution
, we get @xmath336 which implies that @xmath482 for a.e .
@xmath483 in view of ( * ? ? ?
* corollary 1 in p. 84 ) .
therefore , we get . by proposition
[ weak is classical ] , we have @xmath484;l^{1}({\mathbb{r}}^n))$ ] for any weak solution @xmath305 of . in view of ( h2 ) , we see that @xmath485(x , t , p , x)$ ] is continuous on @xmath486 $ ] for any @xmath487 .
we state our main result .
[ uniqueness ] if there exist weak solutions of the initial - value problem , they are classical and unique in @xmath488,$ ] where @xmath337 is given by theorem [ key estimate ] . we formulate the main ingredient of the proof of the above theorem as a lemma .
[ lem - for - unique ] let @xmath489 and @xmath338 be a continuous function on @xmath490 $ ] such that @xmath491 for any @xmath492 $ ] and @xmath493\to{\mathbb{r}}$ ] be a bounded lipschitz continuous function with respect to @xmath8 variable which satisfies , and on @xmath490.$ ] then we have @xmath494 for any @xmath495 , \end{gathered}\ ] ] where @xmath496 is a constant depending on @xmath497 and @xmath498 is a constant depending on @xmath499 , @xmath233 , @xmath500 , @xmath137 , @xmath501 , @xmath502 .
we first prove the estimate .
we have @xmath503 for @xmath504 , since @xmath505 we claim that @xmath506 for @xmath507 $ ] and @xmath508 small enough .
we recall that @xmath509 is a @xmath244-diffeomorphism when @xmath76 satisfies .
to prove the claim , let @xmath510 $ ] such that @xmath511 and set @xmath512 we distinguish two cases .
if @xmath513 then , by , @xmath514 for @xmath515 if @xmath516 then , by , @xmath517 for @xmath518 @xmath519 and @xmath492.$ ] finally , holds if @xmath508 is such that @xmath520 by a change of variable , we have @xmath521 for small @xmath522 and therefore small @xmath76 , since @xmath523 from and , it follows @xmath524 by .
we next prove the estimate .
note that implies the lower gradient estimate @xmath525 from the increase principle of ( * ? ? ? * lemma 2.3 ) , we get @xmath526 where @xmath527 . therefore , noting that implies that @xmath6 has a interior cone property as we can see in the proof of corollary [ cone ] , by ( * ? ? ?
* lemma 4.4 ) for some @xmath528 depending on @xmath499 , @xmath233 , @xmath500 , @xmath529 , @xmath137 , @xmath501 , @xmath502 , we have @xmath530 where @xmath531
suppose that there exist viscosity solutions @xmath155 and @xmath156 of .
let @xmath532 $ ] which will be fixed later and set @xmath533}|(u_{1}-u_{2})(x , t)|.\ ] ] in view of theorem [ continuous dependence ] and ( h5-(i ) ) or ( h5-(ii ) ) , we have @xmath534 where @xmath535 } \int_{{\mathbb{r}}^n } \mathbf{1}_{\{u_{2}(\cdot , t)\ge0\}}(y)|\,dy \label{dislocation}\\ \textrm{and } \nonumber\\ { \overline}{{\kappa}}_\tau:= \sup_{x\in{\mathbb{r}}^{n } , t\in[0,\tau ] } \int_{0}^{t}\int_{{\mathbb{r}}^n } g(x - y , t - s)|\mathbf{1}_{\{u_{1}(\cdot , s)\ge0\}}(y ) -\mathbf{1}_{\{u_{2}(\cdot , s)\ge0\}}(y)|\,dyds . \label{fn}\end{gathered}\ ] ] note that @xmath536 we fix @xmath537 where @xmath337 is given by theorem [ key estimate ] .
take @xmath538 small enough in order that @xmath539 the lower - bound gradient estimate ( corollary [ lower bound ] ) holds on @xmath540 $ ] and , for all @xmath541,$ ] @xmath542 moreover , take @xmath543 such that @xmath544,\ ] ] where @xmath367 is the constant give by proposition [ regularity ] . by continuity of @xmath545 , @xmath546 which achieve the same initial condition @xmath51 ,
it is always possible to find @xmath547 small enough in order that the above condition holds . by lemma [ lem - for - unique ]
, we have @xmath548 for some @xmath549 depending on @xmath550 and @xmath551 for some @xmath552 depending on @xmath553 .
therefore , we get @xmath554 for some constant @xmath555 which is independent of @xmath556 . for @xmath556 small enough , we have @xmath557 .
it follows @xmath558 on @xmath559 $ ] .
we consider @xmath560\}$ ] .
if @xmath561 , then we can repeat the above proof from time @xmath562 instead of @xmath31 . finally , we have @xmath558 on @xmath563 $ ] for all @xmath564 which gives the conclusion .
in the companion paper @xcite , the framework to show existence of weak solutions of is given and as applications , existence results for weak solutions of level set equations appearing in dislocations theory and in the study of fitzhugh - nagumo systems are presented ( see ( * ? ? ?
* sections 3.2 , 4.2 ) ) . in this section , we give uniqueness results for viscosity solutions of such equations .
we consider the level set equation of the evolution of hypersurfaces : @xmath565 where we use the notations in introduction . here
@xmath566 , @xmath567}\to{\mathbb{r}}$ ] , @xmath568 are given functions which satisfy the following assumption ( a ) : * @xmath569 , @xmath570 for all @xmath571 , where @xmath572 and @xmath573 is lipschitz continuous ; * @xmath574 ; l^{1}({\mathbb{r}}^n))$ ] , @xmath575})$ ] , @xmath576;l^{1}({\mathbb{r}}^n))$ ] ; * there exist constants @xmath577 such that , for any @xmath578 and @xmath141 $ ] @xmath579 * @xmath580 for some positively homogeneous function @xmath581 of degree @xmath582 , i.e. , @xmath583 for @xmath273 , @xmath196 , which satisfies @xmath584 @xmath585 the function @xmath30 is called the _ cahn - hoffman _ vector and the last term in the right hand side of is the anisotropic curvature of @xmath3 at @xmath8 given by @xmath586 .
we refer the reader to the monograph by giga @xcite and the references therein for more details . for reader
s convenience , we derive the level set equation of , see @xcite .
we have @xmath587 since @xmath588 is positively homogeneous of degree @xmath582 , @xmath589 is positively homogeneous of degree @xmath31 for all @xmath590 , i.e. , @xmath591 for all @xmath273 , @xmath571 .
differentiating in @xmath40 , setting @xmath592 and noting that @xmath593 yields @xmath594 where @xmath595 .
equality yields @xmath596 where @xmath597 and @xmath598 where @xmath599 .
introducing an auxiliary function @xmath466}\to{\mathbb{r}}$ ] such that @xmath600 on @xmath3 and @xmath601 in @xmath6 for all @xmath141 $ ] , we note @xmath602 and it follows @xmath603 in the above equalities , we used the homogeneity of degree @xmath31 of @xmath30 and of degree @xmath402 of @xmath604 .
set @xmath605(x , t , p):= m(-\hat{p})\bigl(c_0(\cdot , t)\ast\chi(\cdot , t)(x ) + c_1(x , t)\bigr ) , \\ { \sigma}(p):= \sqrt{m(-\hat{p } ) } \bigl(d^{2}{\gamma}(-\hat{p})\bigr)^{1/2 } r_{p } , \\
h_{d}[\chi](x , t , p , x):= -c[\chi](x , t ,
p)|p| -{{\rm tr}\,}\bigl({\sigma}(p){\sigma}^{t}(p)x\bigr ) , \end{gathered}\ ] ] for any @xmath435},[0,1])$ ] , @xmath275}\times({\mathbb{r}}^n\setminus\{0\})\times{\mathcal{s}}^{n}$ ] .
the level set equation of is the equation @xmath606 ( x , t , du , d^{2}u)=0 \quad\textrm{in } \ { \mathbb{r}^n\times(0,t)},\ ] ] which is a particular case of .
[ dislocation thm ] under assumptions ( a ) , ( i1 ) and ( i2 ) , the initial value problem with @xmath607 has at least a weak solution in @xmath191}$ ] .
moreover , weak solutions are classical and unique in @xmath563 $ ] for some @xmath608 $ ] which depends only on @xmath609 @xmath316 and @xmath317 .
we refer to @xcite for the short time existence and uniqueness of the solution of a dislocation dynamics equation with a mean curvature term under different assumptions .
the existence of weak solutions is proved in ( * ? ? ?
* theorem 3.3 ) . by using theorem [ uniqueness ] , we prove a short time uniqueness .
it is easy to check that ( h2 ) , ( h3 ) are satisfied . due to the arguments in example [ ex-2 ] and assumptions ( a ) ( i ) , ( iv )
, we see that ( h4 ) is satisfied .
we prove that @xmath610 satisfies ( h1 ) and ( h5-(i ) ) .
we first check ( h5-(i ) ) .
set @xmath611(x , t , p)$ ] for @xmath278}\times({\mathbb{r}}^n\setminus\{0\})$ ] , @xmath612},[0,1])$ ] and @xmath128 .
note that @xmath613 we finally check ( h1 ) .
let @xmath614})\times{l^{\infty}}({\mathbb{r}^n\times[0,t]},[0,1])$ ] be a @xmath443-viscosity solution of .
we extend the functions @xmath615(x,\cdot , p)$ ] to be equal @xmath31 when @xmath616 and @xmath617 .
we set @xmath618(x,\cdot , p)\ast\zeta_{n}(t ) , \end{gathered}\ ] ] for all @xmath278}\times({\mathbb{r}}^n\setminus\{0\})$ ] , where @xmath619 for @xmath620 and @xmath621 is a standard mollification kernel .
then we have @xmath622}\times({\mathbb{r}}^n\setminus\{0\}))$ ] , @xmath623 locally uniformly in @xmath624 as @xmath625 .
moreover , the @xmath626 s are lipschitz continuous with respect to @xmath8 variable with the constant @xmath627 and bounded ( independently of @xmath479 ) .
let @xmath628 be the viscosity solutions of @xmath629 for all @xmath476 .
by example [ ex-2 ] and proposition [ regularity ] , we have @xmath630 } , \quad u_{n}(x , t)=-1 \ \textrm{on } \ ( { \mathbb{r}}^n\setminus b(0,r(t)))\times[0,t ] , \\ & |u_{n}(x , t)-u_{n}(y , t)|\le c|x - y| , \quad \end{aligned}\ ] ] for all @xmath578 ,
@xmath290 $ ] and some @xmath201 , where @xmath631 is the function introduced in example [ ex-2 ] . by ascoli - arzel theorem , the uniqueness of @xmath443-viscosity solutions of with @xmath436=h_{d}[\chi]$ ] and ( * ? ? ?
* theorem 1.1 ) , we have @xmath480 locally uniformly on @xmath191}$ ] and @xmath305 still satisfies properties .
the above mollification argument can be an alternative way of getting the approximation property we need ( cf .
( h6 ) ) by a regularization of @xmath436 $ ] by @xmath436(x,\cdot , p , x)\ast\zeta_{n}(t)$ ] instead of @xmath632(x , t , p , x)$ ] .
we consider the following system : u_t=((v)+div())|du| & in @xmath322 , + v_t - v= g^+(v)_\{u0}+ g^-(v)(1-_\{u0 } ) & in @xmath322,[fn asymp ] + u(,0)=u_0 , v(,0)=v_0 & in @xmath19 , which is obtained as the asymptotic as @xmath633 of the following fitzhugh - nagumo system arising in neural wave propagation or chemical kinetics ( see ( * ? ? ?
* theorem 4.1 ) ) : @xmath634 where @xmath635 the functions @xmath636 and @xmath637 appearing in are associated with @xmath408 and @xmath638 . we make the following assumptions ( b ) : * the function @xmath639 is bounded and of class @xmath640 with @xmath641 ; * the functions @xmath642 are lipschitz continuous on @xmath643 with a lipschitz constant @xmath644 and there exist @xmath645 such that @xmath646 * @xmath647 , @xmath648 for all @xmath649 and some @xmath650 . for @xmath435},[0,1])$ ] , we write @xmath651 for the solution of @xmath652 and set @xmath615(x , t):={\alpha}(v(x , t))$ ] .
then problem reduces to @xmath653(x , t ) + { { \rm div}\,}\bigl(\frac{du}{|du|}\bigr)\bigl)|du(x , t)| = 0 & & \textrm{in } \ { \mathbb{r}^n\times(0,t ) } , \\ &
u(\cdot,0)=u_0 & & \textrm{in } \
{ \mathbb{r}}^{n } , \end{aligned } \right.\ ] ] which is a particular case of .
[ fn thm ] under assumptions ( b ) , ( i1 ) and
( i2 ) , the initial value problem has at least a weak solution in @xmath191}$ ] .
moreover , it is classical and unique in @xmath563 $ ] for some @xmath654 which depends only on @xmath655 @xmath316 and @xmath317 .
the existence of weak solutions is proved in ( * ? ? ?
* theorem 3.4 ) .
see also @xcite .
it is easy to check that ( h2 ) and ( h3 ) are satisfied . due to similar arguments as those in the proof of theorem [ dislocation thm ]
, we see that ( h1 ) and ( h4 ) are satisfied .
we prove that ( h5-(ii ) ) is satisfied . for @xmath656},[0,1])$ ] , the solutions of are given by @xmath657 for @xmath128 . by the proof of (
* ? ? ? * theorem 4.1 ) , we have @xmath658(x , t)-c[\chi_{2}](x , t)|\le l_{{\alpha}}({\overline}{g}-{\underline}{g})e^{3l_{g}t } \int_{0}^{t}\int_{{\mathbb{r}}^n } g(x - y , t - s ) @xmath108}$ ] .
this completes the proof .
we consider the following evolution of hypersurfaces : @xmath659 where the function @xmath36 is lipschitz continuous .
a typical example is @xmath37 for some @xmath38 which has been studied by chen , hilhorst and logak in @xcite ( see @xcite also ) .
the authors prove that the limiting behaviour of the following reaction - diffusion equation u_t = u+f(u,_u ) & in @xmath660 , + = 0 & on @xmath660 , + u(,0)=g^ & in @xmath661 is characterized by the motion of hypersurface with @xmath37 .
the level set equation of is the nonlocal equation : @xmath662 for some positive constant @xmath663 it is worthwhile to mention that we can not expect the global existence of weak solutions without any restriction of the growth of @xmath41 , because the front can blow up at a finite time . a growth condition to ensure a global existence result
is given below in ( c-(ii ) ) , see also ( * ? ? ?
* section 4.2 ) .
we can easily check that the equation satisfies assumptions ( h1)-(h4 ) and ( h5-(i ) ) .
so , the following result holds . under assumptions ( i1 ) and
( i2 ) , for any @xmath664 , there exists a constant @xmath665 $ ] and at least a weak solution @xmath614})\times{l^{\infty}}({\mathbb{r}^n\times[0,t]},[0,1])$ ] of the initial value problem for such that @xmath666 for any @xmath667 $ ] .
moreover , the weak solution is classical and unique in @xmath563 $ ] for some @xmath668 $ ] which depends only on the lipschitz constant of @xmath41 , @xmath669 , @xmath670 , @xmath671 @xmath316 and @xmath317 .
now , adding the assumption ( c ) : * the assumption ( i2 ) holds with @xmath672 in @xmath245 , * there exist @xmath673 such that @xmath674 we can show a global existence and uniqueness result of solutions of for small @xmath588 .
more precisely , we get [ measure global thm ] under assumptions ( i1 ) , ( i2 ) and ( c ) , there exists a positive constant @xmath675 such that , for any @xmath676 , there exists a unique viscosity solution of in @xmath191}$ ] .
[ measure key estimate ] let @xmath305 be the viscosity solution of u_t = c(t)|du| + tr((i-)d^2u ) & in @xmath322 , + u(,0)=u_0 & in @xmath19 , where @xmath588 is a positive constant and @xmath677)$ ] is a nonnegative given function .
then we have , for any @xmath678 $ ] , @xmath679,\ ] ] where @xmath136 is a positive constant which depends only on @xmath670 and @xmath112 and @xmath310 is the constant given by lemma [ psi ] ( we may assume that @xmath680 ) .
the proof of lemma [ measure key estimate ] is very similar to that of theorem [ key estimate ] , but since we would like to explain how to use the positiveness of @xmath681 and note the dependence of @xmath136 in lemma [ measure key estimate ] , we give it here .
let @xmath311 be the function given by lemma [ psi ] and fix @xmath313 $ ] .
setting @xmath682 and @xmath683 for all @xmath108}$ ] , we consider @xmath145 } \{w(x , t)-v(y , t)- \frac{|x - y|^4}{{\varepsilon}^4}-\tilde{k}t\}.\ ] ] for @xmath684 . by similar arguments as those used in theorem [ continuous dependence ] , there exist @xmath685 $ ] for some @xmath285 and small enough @xmath147 such that the supremum attains at @xmath686 and @xmath687 and @xmath688 satisfy .
note that the lipschitz constant of @xmath305 is @xmath316 in this case .
we have @xmath689 .
the definition of viscosity solutions immediately implies that we have @xmath690 it follows @xmath691 we note that @xmath692 is a positive definite bounded matrix and @xmath693 by . in view of the positiveness of @xmath694 we get , for some constant @xmath695 which may change line to line , @xmath696 sending @xmath166 and taking @xmath697 , we necessarily have @xmath175 .
thus we have for any @xmath108}$ ] @xmath698 setting @xmath699 we get @xmath700 where @xmath701 depends only on @xmath702 this implies a conclusion . in ( * ? ? ?
* section 4 ) , the global existence result for weak solutions of is given .
due to lemma [ measure key estimate ] , we see that if @xmath703 then @xmath704.\ ] ] therefore , we get @xmath705 , \ { \lambda}\in[0,{\overline}{{\lambda}}].\ ] ] a careful review of the proof of theorems [ uniqueness ] gives the conclusion .
[ star perimeter ] inequality implies that the @xmath706-level set of @xmath707 for @xmath706 close to 0 , are star - shaped domains with respect to a ball with center @xmath708 see lemma [ i2geom ] .
in particular , they are locally lipschitz continuous graphs
. then we can get perimeter estimates without using lemma [ perimeter ] . indeed , noting that , from the lower gradient estimate ( corollary [ lower bound ] ) and the increase principle ( * ? ? ?
* lemma 2.3 ) , for small @xmath147 , @xmath709 we have , for any @xmath64 $ ] , @xmath710 in view of the co - area formula and proposition [ regularity ] .
since the @xmath711 s are locally lipschitz continuous , the @xmath24-hausdorff measure and the perimeter in geometric measure theory coincide ( see @xcite ) .
since this latter perimeter is lower - semicontinuous with respect to the hausdorff convergence , sending @xmath633 , we get @xmath712 for all @xmath141 $ ] .
we give the proof of lemma [ i2geom ] , proposition [ regularity ] and lemma [ psi ] .
let @xmath260 be defined by and consider the function @xmath713 it is not difficult to see that , ( i1 ) and ( i2 ) hold with @xmath266 conversely , suppose that ( i1 ) and ( i2 ) with @xmath672 hold for some @xmath714 we claim that @xmath715 is star - shaped with respect to the ball @xmath716 with @xmath717 let @xmath718 and @xmath719 and define @xmath720 it suffices to show that @xmath721 on @xmath722.$ ] from ( i1 ) , ( i2 ) , we have @xmath723 for @xmath724.$ ] let @xmath725 : g>0 \ { \rm on } \ [ 0,{\lambda}]\}.\ ] ] if @xmath726 then the proof is complete .
otherwise , let @xmath147 small enough such that @xmath727 from ( i1 ) , ( i2 ) , we have @xmath728 which is a contradiction .
it completes the proof of the claim .
the proof of the fact that a star - shaped with respect to a ball domain has a locally lipschitz continuous boundary may be found in ( * ? ? ?
2.4.4 and theorem 2.4.7 ) or ( * ? ? ?
* lemma p.20 ) .
we turn to the proof of ( ii ) .
we need to recall some notations and definitions and we refer the reader to @xcite for further details .
the clarke generalized directional derivative at @xmath8 in the direction @xmath4 of @xmath729 is @xmath730 the clarke generalized derivative at @xmath8 is the closed convex set @xmath731 which is nonempty when @xmath408 is locally lipschitz continuous at @xmath732 the clarke tangent cone to @xmath733 at @xmath734 is the convex cone @xmath735 and the clarke normal cone is the polar of the latter , i.e. , @xmath736 the proof is divided in several steps . _
we claim that , for any @xmath734 and @xmath737 there exists @xmath738 such that _ @xmath739 we fix @xmath740 @xmath741 and consider neighborhoods @xmath742 of @xmath743 and @xmath744 of @xmath745 such that there exists a @xmath367-lipschitz continuous function @xmath746 with @xmath747 and @xmath748 let @xmath749 by definition , for all @xmath750 @xmath751 by ( * ? ? ?
* theorem 2.5.7 ) , since @xmath408 is lipschitz continuous , @xmath752 it follows that , for any @xmath753 @xmath754 still belongs to @xmath755 from , we get @xmath756 a first consequence is that @xmath757 necessarily .
moreover @xmath758 indeed , if @xmath759 then holds for all @xmath620 and therefore @xmath760 for all @xmath761 therefore , using that @xmath408 is @xmath367-lipschitz continuous , we get @xmath762 which leads to a contradiction for @xmath763 using again ( * ? ? ?
* theorem 2.5.7 ) , we have @xmath764 since @xmath408 is @xmath367-lipschitz continuous , it follows @xmath765 using that @xmath766 we obtain @xmath767 taking @xmath768 we obtain easily .
we denote by @xmath769 the distance to @xmath733 and by @xmath770 the signed distance to @xmath771 which is negative in @xmath772 for any set @xmath773 @xmath774 is the closure of the convex hull of @xmath255 _ 2 .
we claim @xmath775 $ ] for all @xmath734 . _
( notice it means that the generalized derivative of the signed distance does not contain 0 . )
let @xmath776 and @xmath777 a sequence of points which converges to @xmath771 such that @xmath770 is differentiable at @xmath778 assume that @xmath779 converges to @xmath780 suppose first that , up to extract a subsequence , @xmath781 then , since @xmath782 in @xmath783 and @xmath769 is differentiable at @xmath784 it means that @xmath777 has a unique closest point @xmath785 and @xmath786 but @xmath787 is the convex hull of the cone generated by such limits ( ( * ? ? ?
* exercise 8.5 , p.96 ) ) .
thus @xmath788 from ( * ? ? ?
* theorem 2.8.1 ) , @xmath789 is the convex hull of such @xmath790 which completes the proof of the claim .
there exists an open bounded neighborhood @xmath791 of @xmath771 and @xmath792 such that , for all @xmath793 @xmath794 is a nonempty compact convex subset of @xmath795 _ the subsets @xmath796 are clearly convex , closed and bounded for any @xmath797 and open subset @xmath798 it remains to prove that there are nonempty for some @xmath799 it is true for @xmath800 with @xmath801 by claims 1 and 2 . to extend this property in a neighborhood @xmath802 of @xmath803
we notice the following facts : since @xmath804 is upper - semicontinuous ( ( * ? ? ?
* proposition 2.1.5 ) ) and @xmath805 there exists @xmath806 such that @xmath807 for all @xmath808 by clarke s implicit function theorem ( * ? ? ?
* proposition 3.3.6 ) , the @xmath809-level sets of @xmath770 are lipschitz continuous in @xmath810 the lipschitz constant is controlled by the distance from 0 to @xmath811 up to take @xmath706 small , it depends only on @xmath812 then , we can repeat the previous arguments and obtain the result in @xmath802 ( up to take @xmath813 smaller than @xmath814 ) .
we then find @xmath791 by compactness of @xmath815 _ 4 .
the multi - valued map @xmath816 is lower - semicontinuous .
_ let @xmath817 let @xmath818 and @xmath819 since @xmath804 is upper - semicontinuous , by definition , there exists @xmath820 such that , if @xmath821 then @xmath822 ( where @xmath823 ) .
any @xmath824 can be written @xmath825 where @xmath826 and @xmath827 using that @xmath828 it follows @xmath829 this proves that @xmath830 which is the definition of the lower - semicontinuity of a multi - valued function .
michael s continuous selection theorem and end of the proof . _ from steps 3 and 4 , we can apply michael s continuous selection theorem ( @xcite ) : there exists a continuous map @xmath831 such that @xmath832 i.e. , for all @xmath833 and @xmath834 we have @xmath835 according to remark [ rmk - c1-lisse - geom ] , up to decrease @xmath836 we may choose @xmath837 which is smooth , bounded and lipschitz continuous . using the lipschitz continuity of @xmath770 and @xmath838 we may earn some uniformity in up to reduce @xmath839 more precisely , for every @xmath840 there exists @xmath841 and @xmath842 such that @xmath843 for all @xmath844 and @xmath845.$ ] we conclude by compactness of @xmath771 ( note that we can modify the signed distance function far from @xmath771 in order to have a bounded function ) .
star - shaped property is not sufficient to ensure ( i2 ) , see the counterexample of figure [ dessins - ens ] .
there exist some sets which satisfy ( i2 ) but they do not have a locally lipschitz boundary , see figure [ dessins - ens ] .
existence of a solution to is given by assumption ( a8 ) .
we prove the uniqueness and the lipschitz continuity regularity in @xmath8 of the solutions .
let @xmath288})$ ] be a solution of which satisfy .
let @xmath846 and set @xmath847 } \{u(x , t)-u(y , t)-e^{kt}\frac{|x - y|^{4}}{{\varepsilon}^{4}}-\eta t\}.\ ] ] let @xmath26 be attained at @xmath848 $ ] . by ( a8 ) , we may assume that @xmath849 $ ] .
we first consider the case where @xmath148 $ ] . in view of ishii s lemma , for any @xmath149
, there exist @xmath850 and @xmath851 such that @xmath852 where @xmath853 the definition of viscosity solutions immediately implies the following inequalities : @xmath854 we have @xmath855 therefore , sending @xmath166 and setting @xmath859 , necessarily we have @xmath175 .
we get for all @xmath578 , @xmath141 $ ] @xmath860 we have used the young inequality in the third inequality .
setting @xmath861 we get @xmath862 sending @xmath863 , we have @xmath864 on the one hand , by taking @xmath865 in the above inequality , we get @xmath866 and obtain the uniqueness of the solution . on the other hand , by choosing @xmath867 we obtain . we now prove .
set @xmath868 , where @xmath367 is the constant given by proposition [ regularity ] .
recalling , in view of lemma 9.1 in @xcite , there exists a constant @xmath869 such that @xmath870 for all @xmath871 , @xmath872 $ ] .
noting that @xmath873 on @xmath874 $ ] , we get @xmath870 for all @xmath875 , @xmath290 $ ] .
it is easy to see inequality if @xmath881 in view of ( i2 ) . in the case
where @xmath882 , we have @xmath883 which implies inequality if @xmath313 $ ] with @xmath884 finally , we consider the case where @xmath885 . by replacing @xmath310 by a smaller constant if necessary
, we may assume that @xmath886 for all @xmath313 $ ] .
then we have @xmath887 , which yields a conclusion .
m. bourgoing , _ viscosity solutions of fully nonlinear second order parabolic equations with @xmath888 dependence in time and neumann boundary conditions_. discrete contin .
21 ( 2008 ) , no .
3 , 763800 .
m. bourgoing , _ viscosity solutions of fully nonlinear second order parabolic equations with @xmath443 dependence in time and neumann boundary conditions . existence and applications to the level - set approach_. discrete contin .
21 ( 2008 ) , no .
4 , 10471069 . e. r. jakobsen and k. h. karlsen , _ continuous dependence estimates for viscosity solutions of fully nonlinear degenerate elliptic equations _ , electron .
j. differential equations ( 2002 ) , no .
39 , 10 pp .
( electronic ) . | we describe a method to show short time uniqueness results for viscosity solutions of general nonlocal and non - monotone second - order geometric equations arising in front propagation problems .
our method is based on some lower gradient bounds for the solution .
these estimates are crucial to obtain regularity properties of the front , which allow to deal with nonlocal terms in the equations .
applications to short time uniqueness results for the initial value problems for dislocation type equations , asymptotic equations of a fitzhugh - nagumo type system and equations depending on the lebesgue measure of the fronts are presented . |
extra neutral gauge bosons , known in the literature as @xmath0 , appear in many proposals for beyond - the - standard model ( bsm ) physics ; for a review , see for instance @xcite . here
we focus on _ minimal _ @xmath0 , previously studied in @xcite , which stand out both for their simplicity , and because they could arise in several of the above mentioned bsm scenarios , such as , _
e.g. _ , grand unified theories ( guts ) and string compactifications .
following @xcite , we consider a minimal extension of the sm gauge group that includes an additional abelian factor , labeled @xmath4 , commuting with @xmath5 .
the fermion content of the sm is augmented by one right - handed neutrino per family .
we require anomaly cancellation , as this allows us to write a renormalizable lagrangian . if family - universality is imposed , then the anomaly cancellation conditions yield a unique solution : @xmath6 , where @xmath3 and @xmath7 are baryon and lepton number respectively . , with @xmath8 arbitrary coefficients .
however , the @xmath9 component can be absorbed in the kinetic mixing in the class of models we consider .
] however , if the requirement of family - universality is relaxed , it can be shown that the following set of family - dependent charges satisfy the anomaly cancellation conditions : @xmath10 , where @xmath11 are the lepton flavours , and @xmath12 are arbitrary coefficients .
we will consider a specific example of such _ non - universal _
@xmath0 in the following . in the basis of mass eigenstates for vectors , and with canonical kinetic terms
, the neutral current lagrangian reads @xmath13 where @xmath14 is the photon field coupled to the electromagnetic current , while @xmath15 are the massive states , which couple to the currents @xmath16 respectively , obtained from @xmath17 \ , \overline{f } \gamma^\mu f \,$ ] and @xmath18 \ ,
\overline{f } \gamma^\mu f \,$ ] via a rotation of the @xmath19 mixing angle @xmath20 . the explicit expression of the latter reads @xmath21 .
thus , under our minimal assumptions , only three parameters beyond the sm ones are sufficient to describe the @xmath0 phenomenology : the physical mass of the extra vector , @xmath22 , and the two coupling constants @xmath23 . in the following discussion , we normalize these couplings to the sm @xmath24 coupling , namely @xmath25 . .
the yellow band represents the region of couplings compatible with a gut , whereas dots and lines correspond to specific benchmark models or full supersymmetric gut models , see @xcite for details.,scaledwidth=70.0% ]
because guts are one of the motivations for considering minimal @xmath0 , it is interesting to give an estimate of the constraints that a gut would imply on the weak - scale couplings @xmath26 . for choosing the boundary conditions at unification scale @xmath27 , we normalize all charges as in @xmath28 , and take @xmath29 gev .
we allow the @xmath0 coupling at unification scale to vary within the interval @xmath30 , and using the rge of the model we obtain the gut - favoured region of weak - scale couplings , shown in fig.[gut ] for the universal case @xmath31 .
since the boundary conditions at scale @xmath27 are symmetric under the reflection @xmath32 , it is evident from fig .
[ gut ] that mixing effects in the rge ( due to the non - orthogonality of the generators @xmath9 and @xmath2 ) are important .
the gut - favoured regions for non - universal models , computed along similar lines , can be found in @xcite .
the measurements providing constraints on minimal @xmath0 can be divided into two classes : electroweak precision tests and direct searches at the tevatron .
measurements performed at lep1 and at low energy mainly constrain @xmath19 mixing , whereas data collected at lep2 ( above the @xmath33 pole ) constrain effective four - fermion operators . to compute the bounds from ewpt on minimal @xmath0
, we integrate out the heavy vector and use the effective lagrangian thus obtained to perform a global fit to the data .
the results are shown in fig .
[ chi model ] , for the universal ` @xmath34 model ' , corresponding to a particular direction in the @xmath26 plane often considered in the literature .
the cdf and d0 collaborations have derived , from the non - observation of discrepancies with the sm expectations , upper limits on @xmath35 ( @xmath36 ) , @xcite . to extract bounds on minimal @xmath0 ,
we compute the same quantity at nlo in qcd , and compare it with the limits published by the experimental collaborations .
the comparison between bounds from ewpt and from the tevatron is most clear if we plot them in ( coupling _ vs. _ mass ) , for a chosen direction in the @xmath37 plane , as it is done in fig .
[ chi model ] for the @xmath34 model .
we see that bounds from ewpt have a linear behaviour , because all the effects due to the @xmath0 in the low - energy effective lagrangian depend on the ratio @xmath38 , whereas bounds from the tevatron become negligible above a kinematic limit , which is of the order of 1 tev .
thus for low masses the tevatron data give the strongest limits , while above a certain value of @xmath22 ( which is of the order of 500 gev for the @xmath34 model ) , bounds from ewpt are stronger . in particular , for models compatible with guts the strongest bounds are those given by ewpt . at 7 tev , and 400 pb@xmath39 at 10 tev ) for the @xmath34 model
( right ) present bounds and discovery prospects of the lhc at 7 tev and 50 pb@xmath39 for the _ muonphilic _ model with @xmath40 .
for @xmath41 , both the bounds from the tevatron and the lhc reach are indeed weaker , because of finite - width effects not included in the figure , but the general message is unaffected .
the yellow bands correspond to the gut - favored region , see section [ sect : gut].,title="fig:",scaledwidth=50.0% ] at 7 tev , and 400 pb@xmath39 at 10 tev ) for the @xmath34 model .
( right ) present bounds and discovery prospects of the lhc at 7 tev and 50 pb@xmath39 for the _ muonphilic _ model with @xmath40 . for @xmath41
, both the bounds from the tevatron and the lhc reach are indeed weaker , because of finite - width effects not included in the figure , but the general message is unaffected .
the yellow bands correspond to the gut - favored region , see section [ sect : gut].,title="fig:",scaledwidth=50.0% ]
the present schedule foresees that in 2010/2011 the lhc will run at 7 tev in the center of mass , collecting up to 1 fb@xmath39 of integrated luminosity @xmath7 .
therefore , it is interesting to ask whether there are any minimal @xmath0 which are both allowed by present constraints and accessible for discovery in such early phase . to answer this question ,
we have performed a nlo analysis similar to the one used in extracting bounds from tevatron data , requiring the @xmath0 signal to be at least a @xmath42 fluctuation over the sm - drell yan background .
the results are displayed for the @xmath34 model in fig .
[ chi model ] , where a comparison with present bounds is made .
we see that for @xmath43 pb@xmath39 ( the luminosity approximately foreseen at the end of 2010 ) , no discovery is possible . on the other hand , for @xmath44fb@xmath39
some unexplored regions become accessible ; however , @xmath0 compatible with guts are still out of reach , and more energy and luminosity will be needed to test them .
we have seen that universal models are strongly constrained by present data . on the other hand ,
when we consider non - universal couplings to leptons , the bounds can be significantly altered .
in particular , let us consider the case where @xmath45 , which we called ` muonphilic @xmath46 '
. let us further assume that kinetic mixing is negligible , _
i.e. _ @xmath47 . in this case , the @xmath0 has no coupling to the first and third leptonic families , in particular it has no coupling to the electron . as a consequence ,
bounds from ewpt are strongly relaxed , the only surviving constraints coming from @xmath48 and @xmath49-@xmath50 scattering ( nutev ) . on the other hand ,
the tevatron reach is limited , as already noted in section [ bounds ] , to @xmath51 tev : therefore the lhc has access to a wide region of unexplored parameter space already with a very low integrated luminosity at 7 tev , as shown in fig .
[ chi model ] .
we have discussed the present experimental bounds and the early lhc reach on minimal @xmath0 models , showing that present constraints can not be neglected when assessing the discovery potential of the early lhc . in particular , we have found that exploration of universal models , coupled to @xmath2 , may need more energy and luminosity than those foreseen for 2010/2011 , in particular for values of the couplings compatible with guts . on the other hand ,
some non - universal models which are weakly constrained by present data , such as the _ muonphilic _ @xmath0 , could be discovered at the lhc with very low integrated luminosity .
i am indebted to my collaborators f. zwirner , g. villadoro , and a. strumia .
i would also like to thank the organizers of the 2@xmath52 young researchers workshop _
physics challenges in the lhc era _ for giving me the possibility to present this work .
99 p. langacker , rev .
* 81 * ( 2008 ) 1199 , arxiv:0801.1345 [ hep - ph ] .
t. appelquist , b. a. dobrescu and a. r. hopper , phys .
d * 68 * ( 2003 ) 035012 , arxiv : hep - ph/0212073 .
e. salvioni , g. villadoro and f. zwirner , jhep * 0911 * ( 2009 ) 068 , arxiv:0909.1320 [ hep - ph ] ; e. salvioni , a. strumia , g. villadoro and f. zwirner , jhep * 1003 * ( 2010 ) 010 , arxiv:0911.1450 [ hep - ph ] . t. aaltonen _ et al .
_ [ cdf collaboration ] , phys .
* 102 * ( 2009 ) 031801 , arxiv:0810.2059 [ hep - ex ] ; t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
* 102 * ( 2009 ) 091805 , arxiv:0811.0053 [ hep - ex ] ; [ d0 collaboration ] , d0 note 5923-conf ( july 2009 ) , http://www-d0.fnal.gov/run2physics/www/results/np.htm . | we consider a class of minimal extensions of the standard model with an extra massive neutral gauge boson @xmath0 .
they include both family - universal models , where the extra @xmath1 is associated with @xmath2 , and non - universal models where the @xmath0 is coupled to a non - trivial linear combination of @xmath3 and the lepton flavours . after giving an estimate of the range of parameters compatible with a grand unified theory , we present the current experimental bounds , discussing the interplay between electroweak precision tests and direct searches at the tevatron . finally , we assess the discovery potential of the early lhc .
= 11.6pt |
execution of foreign exchange transactions at a fixed market benchmark rate is common amongst institutional investors .
appeal lies in the fact that , particularly when tracking benchmark indices in other asset classes , currency conversion can be performed by the same underlying fx rates as employed by their target index .
furthermore , aggregation of fixing trades by market makers prior to execution increases liquidity , which , in theory , enables clients to trade large fx amounts at a trusted rate , without having to worry about best execution or the risk of moving the price .
moreover , as a market fix is uniform across providers , it is also intended to build trust in client / provider relationships , as a clearly defined and measurable service arises .
the currently most important benchmark rate is the world market reuters 4 pm london fix ( wm / r 4 pm ) , which accounts for approximately @xmath0-@xmath1 of the total @xmath2 daily volume in the fx spot market @xcite .
the wm / r 4 pm fixing rate @xcite is determined using the methodology described in [ sec_wmrfix ] .
it is based on the last trade prices as well as the last best bid and offer quotes at the end of each one second interval between 15:59:30 to 16:00:30 gmt . therefore , to achieve the wm / r rate ,
service providers execute their fixing orders within this 60 second interval .
compression of large order flow into a narrow time window can be expected to give rise to a special market structure around the fixing time , which is part of what we will look at in this paper .
the described structure has recently come to the attention of the wider public , as concerns have been raised regarding market participants who may have used the construction mechanism of the wm / r 4 pm fix to influence the benchmark @xcite .
subsequently , multiple suggestions regarding how to improve the construction of the wm / r 4 pm fix have been brought forward @xcite .
these investigations mainly focus on the mitigation of eventual manipulation possibilities .
generally , little research has been done on the observable market structure around the wm / r 4 pm fix @xcite , despite the importance of the wm / r 4 pm fix in the services received by many institutional investors . in this
regard , spot rate volatility and extreme spot rate movements are of particular interest , since they are important parameters in the quest for best execution practice .
some analysis has been done in the context of general interactions between clients and dealers @xcite , with the recent focus shifting onto the impact of algorithmic order execution @xcite . in this paper , we investigate the influence of the wm / r 4 pm fix on the dynamics of both , spot rate volatility and extreme spot rate movements .
our study shows that the order compression of the fix indeed changes the market behaviour , resulting in spiking volatility within the wm / r fixing window , and an increased probability for spot rate extrema in this period .
before discussing the market dynamics around the fix , a short description of its construction shall be provided @xcite .
the fixing rate is determined for currency spot , forward , and non - deliverable forward rates .
the calculation differs between forward and spot rates .
while , for the former , a single rate snapshot at the fixing time is used as the benchmark , the spot rate calculation utilises several price quotes within an interval around the fixing time . in the following ,
we will focus on spot rates only .
spot fixings are determined for 160 currency pairs , which are split into trade currencies for liquid pairs and quote currencies for illiquid pairs . for our analysis
, we consider trade currency pairs .
market data is sourced from thomson reuters , ebs , and currenex .
currently , all trade currency data is obtained from a primary source . for currency pairs involving chf , eur , and jpy , data from ebs
currenex data supplements it as a secondary source , if the number of datapoints is too low .
rub data is solely obtained from ebs .
benchmark rates for other pairs are based on thomson reuters data only .
the fix rate is obtained by accumulating quote and trade data within a set time interval around the fixing time . at present , this interval is 1-min . for trade currencies and 2-min . for quote currencies ,
i.e. , for the 4 pm fix ( _ closing rate _ ) the accumulation times are 15:59:30 - 16:00:30 and 15:59:00 - 16:01:00 , respectively .
notably , following recent recommendations @xcite , wm / r is aiming to implement changes to its calculation methodology @xcite , which will become effective on the 15/02/2015 . thereafter
, the data sourcing window is widened to @xmath3 , i.e. , for the 4 pm fix the accumulation time changes to 15:57:30 - 16:02:30 .
additionally , trade currency data will be gathered from different sources . to this end
, thomson reuters data is pooled together with ebs and currenex data for pairs containing chf , eur , jpy , and rub .
quote currency data is sampled at the end of 15-sec .
intervals , leading to 9 datapoints in total . at each sampling point , the last valid best bid and ask quotes are recorded .
subsequently , the medians of these bid and ask quotes are calculated , whose mid - price yields the wm / r benchmark rate . for trade currencies ,
a more elaborate method is used , which is illustrated in figure [ fig_wmrmethod ] .
data is sourced at the end of each 1-sec .
interval @xmath4 , which amounts to 61 datapoints in total .
the last valid best bid and ask quotes , as well as the last rate at which a trade happened in the interval , are captured .
to calculate the fixing rate for trade currencies , firstly , the spread @xmath5 of the best bid and offer quotes is determined for each interval .
subsequently , the trade rate is categorised into a bid trade or an offer trade , depending on whether the trade hit a quote on the bid or the offer side of the order book .
the spread @xmath5 is applied to the trade to infer the opposite side of the order book , resulting in an inferred trade rate at that side .
trades that fall outside the best bid and offer quotes , i.e. , trades which were executed hitting quotes inside the order book , are excluded from the calculation . combining the inferred and actual trade rates
ideally yields subsets for the bid and the offer side containing a total of 61 datapoints each ( bid trades and offer trades in figure [ fig_wmrmethod ] ) .
notably , this assumes that a valid trade occurs within each of the 1-sec .
intervals , which is not necessarily the case . if there is insufficient trade rate data , bid and offer quotes are used instead . the exact limit on
the number of required trade datapoints is discretionary to wm / r and unpublished .
in such a case , quote data is not pooled together from different data sources , but rather data from the source with the largest number of valid quotes is used . for each side ,
i.e. , bid trades and offer trades , the median of the rates is computed . the mid - price between the median bid and median offer trades
yields the wm / r fix .
its spread is given by a predefined standard spread @xmath6 for each currency pair , which is supposed to reflect the liquidity at different times of the day .
the determination of this standard spread is discretionary .
however , if the market spread @xmath7 , defined as the mean over all spreads @xmath5 obtained from the quote data , is larger than the standard spread , @xmath7 will be used instead . applying this spread to the wm / r fix
yields its bid and ask rates . for both , trade and quote currencies ,
the initially sourced data is subject to quality checks , which compare it to the general market level .
this level is established from continuous market surveillance by capturing trade rates at every 15-sec . throughout the entire day .
quality checks happen automatically , testing for data consistency @xcite , as well as on a discretionary basis , for which no detailed methodology has been published .
the mechanism for determining the wm / r closing spot rate suffers from two main weaknesses . on the one hand
, it does not take into account volume information . ignoring spot rate modifications arising from market impact ,
trade rates obtained from the execution of large orders contribute with the same weight to the benchmark as orders at the minimum size . on the other hand , capturing solely the last quote and trade rates at the end of a macroscopic time interval ignores market microdynamics .
market trends happening towards the end of each interval have the potential to affect the data used in benchmark calculation .
for instance , the splitting of a large order into many transactions , distributed over the fixing interval , with each transaction selectively executed towards the end of the 1-sec .
wide data aggregation intervals , can result in a modification of the benchmark in the direction of the order .
for our analysis , we use spot rate data sourced from bloomberg , for a total of 12 currency pairs .
the data has a minute - by - minute resolution , and contains four different price streams .
these are the opening , the highest , the lowest , and the last spot rate obtained within each 1-min . interval .
since the opening rate corresponds to the closing rate of the previous point , our analysis only focuses on the latter three time series .
each interval is centred at the minute , i.e. , datapoints are , for example , obtained at 15:59 , 16:00 and 16:01 .
the time indices are expressed in local london time .
as occasionally there are minute entries missing in the datasets , the data is filtered for completeness , and days containing incomplete datasets have been excluded from the evaluation .
this is done because parts of our analysis rely on the comparison between spot rates at different minutes .
due to data availability , the look - back periods vary between the currency pairs as follows : [ cols= " < , < " , ]
in conclusion , we report on the observation of a change in market dynamics around 4 pm london time , introduced by the wm / r closing rate .
the compression of large order flow into the short time interval , considered for determining the wm / r fix , gives rise to local extrema in volatility .
moreover , it increases the probability for extreme movements in the spot rate .
both effects happen predominantly in anticipation of the event , i.e. , before 4 pm .
this sets the 4 pm market dynamics apart from the one observed for other market relevant events , such as information arrival .
unlike such events , whose effects lead to longer term distortions in the market dynamics , the observed spot rate extrema are confined to a short interval on the order of the fixing window size . while fx execution to the fix has its advantages in terms of convenience , our analysis shows that this comes at the expense of an increased probability for the investor to be subject to the most extreme daily market movements . in recent months
, the occurrence probability for such spot rate extrema around the fixing window has decreased .
the reduction coincides with an increase media attention paid to the wm / r closing rate benchmark .
this could be indicative for a change in the behaviour of market participants , who selectively walk away from execution at the wm / r fix to avoid these drawbacks .
world markets reuters spot and forward rates methodology guide http://www.wmcompany.com/pdfs/wmreutersmethodology.pdf http://www.wmcompany.com/pdfs/wmreuters_spot_rate_service_-_methodology_changes.pdf financial conduct authority , uk press release : _ fca fines five banks 1.1 billion for fx failings and announces industry - wide remediation program _ 12/11/2014 http://www.fca.org.uk/news/fca-fines-five-banks-for-fx-failings liam vaughan , gavin finch , ambereen choudhury _ traders said to rig currency rates to profit off clients _ bloomberg news 12/06/2013 http://www.bloomberg.com/news/2013-06-11/traders-said-to-rig-currency-rates-to-profit-off-clients.html bank of england - fair and effective markets review _ how fair and effective are the fixed income , foreign exchange and commodities markets ? _
27/11/2014 http://www.bankofengland.co.uk/markets/documents/femr/consultation271014.pdf martin d. d. evans _ forex trading and the wmr fix _ social science research network 2014 geir hidal bjnnesa , dagfinn rime _ dealer behavior and trading systems in foreign exchange markets _ vol .
571 2005 alain chaboud , ben chiquoine , erik hjalmarsson , clara vega _ rise of the machines : algorithmic trading in the foreign exchange market _ the journal of finance vol . | a market fix serves as a benchmark for foreign exchange ( fx ) execution , and is employed by many institutional investors to establish an exact reference at which execution takes place .
the currently most popular fx fix is the world market reuters ( wm / r ) 4 pm fix .
execution at the wm / r 4 pm fix is a service offered by fx brokers ( normally banks ) , who deliver execution at the fix provided they obtain the trade order until a certain time prior to 4 pm . in this paper , we study the market microstructure around 4 pm . we demonstrate that market dynamics can be distinguished from other times during the day through increased volatility and size of movements .
our findings question the aggregate benefit to the client base of using the 4 pm fix in its current form .
* keywords : * wm / reuters 4 pm fix , fx market microstructure , fx execution , market manipulation |
m dwarfs are the smallest and least luminous stars on the main sequence , yet they are the most numerous of all stellar constituents and have lifetimes longer than the current age of the universe .
determining the age of an individual main sequence , field m dwarf is quite challenging .
stellar clusters would seem to provide the ideal environments for calibrating the ages of m dwarfs ( because the age of the cluster is determined ) .
however , the large distances to clusters older than a few gyr ( and resulting faintness of cluster m dwarfs ) , preclude any detailed observations of the intermediate or oldest cluster resident m dwarfs
. age estimates for field m dwarfs must therefore rely on a number of techniques that piggyback on other age dating methods .
although several methods exist for estimating the ages of m dwarfs ( e.g. contribution by s. catalan ) , we focus this contribution on what the magnetic activity of m dwarfs can tell us about their age .
m dwarfs are host to intense magnetics dynamos that give rise to chromospheric and coronal heating , producing emission from the x - ray to the radio .
this magnetic activity has long been thought to be tied to the age of the host star .
almost 40 years ago , wilson & woolley ( 1970 ) found a link between magnetic activity ( as traced by the ca ii h & k emission lines ) and the orbital elements ( namely the inclination and eccentricity ) of more than 300 late - type dwarfs ; stars in near - circular orbits with small inclinations had the strongest activity .
they concluded that as stars age , their orbits get more inclined and more eccentric ( due to dynamical interactions ) , and thus , magnetic activity must also decline with age .
subsequent studies over the following decades have found similar connections between age and activity in low - mass stars ( wielen 1977 ; giampapa & liebert 1986 ; soderblom et al .
1991 ; hawley et al .
1996 , hawley et al .
1999 , hawley et al .
2000 ) .
although activity has been observed for decades , the exact mechanism that gives rise to the chromospheric heating is still not well - understood . in the sun ,
magnetic field generation and subsequent heating is closely tied to the sun s rotation . from helioseismology , we know that there is a rotational boundary between the inner solid - body rotating radiative zone and the outer differentially rotating convective zone ( parker 1993 ; ossendrijver 2003 ; thompson et al .
this boundary , dubbed the tachocline , creates a rotational shear that likely allows magnetic fields to be generated , preserved and eventually rise to the surface where they emerge as magnetic loops .
these loops bring heat to the sun s chromosphere and corona , driving both large stellar flares , as well as lower - level quiescent ( or persistent ) magnetic activity .
the faster a star ( with a tachocline ) rotates , the stronger its magnetic heating and surface activity .
therefore , the angular momentum evolution of solar - type stars should play an important role in determining the magnitude of the observed activity .
angular momentum loss from magnetized winds has been shown to slow rotation in solar - type stars ; as a result magnetic activity decreases .
skumanich ( 1972 ) found that both activity and rotation decrease over time as a power law ( @xmath1 ) .
subsequent studies confirmed the skumanich results and empirically demonstrated a strong link between age , rotation and activity in solar - type stars ( barry 1988 ; soderblom et al .
1991 ; pizzolato et al .
2003 ; mamajek & hillenbrand 2008 ; see contribution by e. mamajek ) .
there is strong evidence that the rotation - activity ( and presumably age ) relation extends from stars more massive than the sun to smaller dwarfs ( pizzolato et al .
2003 ; mohanty & basri 2003 ; kiraga & stepien 2007 ) .
therefore , rotation derived ages ( using gyrochronology ) may provide a useful independent estimate of age for m dwarfs when large enough calibration samples can be acquired ( see contributions by s. meibom , j. irwin , and s. barnes ) . however , at a spectral type of @xmath2m3 ( 0.35 m@xmath3 ; reid & hawley 2005 ; chabrier & baraffe 1997 ) , stars become fully convective and the tachocline presumably disappears . despite this changes ,
magnetic activity persists in late - type m dwarfs ; the fraction of active m dwarfs peaks around a spectral type of m7 before decreasing into the brown dwarf regime ( hawley et al .
1996 ; gizis et al .
2000 ; west et al . 2004 ) .
it is unclear if the rotation - age - activity relation extends to the fully convective m dwarfs .
a few empirical studies have uncovered evidence that activity and rotation might be linked in late - type m dwarfs ( delfosse et al .
1998 ; mohanty & basri 2003 ; reiners & basri 2007 ) .
in addition , recent simulations of fully convective stars find that rotation may play a role in magnetic dynamo generation ( dobler et al . 2006 ; browning 2008 ) .
however , a recent study found detectable rotation in a few inactive m7 dwarfs , indicating that the situation may in fact be more complicated ( west & basri 2009 ) .
irrespective of rotation , many studies have found evidence that the age - activity relation extends into the m dwarf regime .
eggen ( 199 ) observed a skumanich - type decrease in activity as a function of age .
larger samples of m dwarfs have added further evidence that magnetic activity decreases with age ( fleming et al . 1995 ; gizis et al .
there is also evidence that m dwarfs may have finite active lifetimes .
stauffer et al .
( 1994 ) suggested that activity may not be present in the most massive m dwarfs in the pleiades .
hawley et al . ( 2000 ) confirmed the stauffer et al .
( 1994 ) claim by observing a a sample of clusters that spanned several gyr in age .
they were able to calibrate the lifetimes for early - type m dwarfs by observing the color at which activity ( traced by h@xmath0 emission ) was no longer present . because of the small sample size , the derived hawley et al .
( 2000 ) activity ages are lower limits at a given color or spectral type .
as mentioned above , the hawley et al .
( 2000 ) study was limited to younger , nearby clusters due to the intrinsic faintness of m dwarfs ; thus , it could only probe ages of a few gyr and spectral types as late as @xmath2m3 . larger samples of m dwarfs are required to statistically derive age - activity relations that extend both in the ages they probe and the range of spectral types they cover . in this contribution , we review recent studies that have used large spectroscopic samples to investigate the relationship between age and activity in m dwarfs . in addition , we derive a statistical h@xmath0 age - activity relation for m2-m7 dwarfs .
large surveys such as the sloan digital sky survey ( sdss ; york et al .
2000 ) have created optical and infrared catalogs of tens of millions of m dwarfs ( bochanski et al .
2009 in prep ) .
in addition to the photometric data , the sdss has obtained spectra for tens of thousands of m dwarfs . recently , these surveys have been utilized to examine the statistical properties of m dwarfs , the dynamics of the milky way and detailed studies of magnetic activity ( hawley et al .
2002 ; west et al . 2004 ; west et al . 2006 ; covey et al
. 2007 ; bochanski et al .
2007a , 2007b ; west et al .
2008 ; hereafter w08 ) .
recently , the sdss data release 5 ( dr5 ) low - mass star spectroscopic sample was released ( w08 ) , containing over 44000 m and early l - type dwarfs .
radial velocities ( accurate to within @xmath25 km s@xmath4 ) and photometric distances ( from the @xmath5 colors ; see bochanski et al .
2009 in prep ) were measured for all of the stars .
proper motions ( matches to usno - b ; munn et al .
2004 ) were obtained for over 27000 of the stars , allowing for full @xmath6 , @xmath7 , @xmath8 3-d space motions for over 27000 low - mass dwarfs , the largest such sample ever assembled .
in addition , all of the spectra were run through the hammer spectral typing facility ( covey et al .
2007 ) to assign spectral types and measure the strength of the h@xmath0 emission line .
one of the hallmarks of the w08 sample is that it probes the entirety of the galactic thin disk and extends well into the thick disk ( @xmath250 - 2500 pc above the plane ) .
this is due to the fact that the majority of sdss fields are in the north galactic cap and that sdss was designed to study the distant universe ; intrinsically faint galactic m dwarfs are bright in a deep extragalactic survey .
activity in m dwarfs varies as a function of spectral type .
figure [ frac ] shows the fraction of active m dwarfs as a function of spectral type from west et al .
the active fraction is very small for early - type m dwarfs , peaks at m7-m8 and declines into the l dwarf regime .
while some of the morphology of this relation may be due to the ability ( or inability ) to host a strong dynamo ( the activity fraction decrease in the late - type m and l dwarfs is likely caused by the atmospheres becoming neutral ) , a large part of the shape may be due to age effects .
if activity has a finite lifetime that changes as a function of spectral type , and we observe a range of stellar ages , then stars with longer lifetimes will appear to have higher activity fractions . to test this hypothesis , west et al .
( 2004 , 2006 ) showed that the fraction of active m7 stars decreases as a function of vertical distance from the galactic plane ( see figure [ falloff ] ) .
stars are born in dynamically cold molecular gas near the midplane of the galaxy . over time
they undergo dynamical interactions with molecular clouds , which add energy to the stellar orbits in all dimensions ( see contribution by b. nordstrm ) .
it is this process that gives thickness to the milky way disk and allows us to use vertical distance as a proxy for age ; stars further from the plane are statistically older ( they have to be dynamically heated to reach those heights ) , while stars near the plane are statistically younger .
figure [ falloff ] shows that the younger stars , near the plane of the milky way are almost all active , whereas the fraction of active stars at larger distances above ( and below ) the galaxy is significantly reduced ; older stars have ceased being active .
the decrease of active fractions as a function of vertical height above the plane is seen for all m dwarf spectral types ( w08 ) .
w08 also showed that the amount of activity , quantified by @xmath9/@xmath10 ( the ratio of the luminosity in the h@xmath0 emission line to the bolometric flux of the star ) decreases over time .
figure [ lbol ] shows the median log(@xmath9/@xmath10 ) as a function of vertical distance from the plane .
the narrow error bars represent the spread of the values and the wide error bars indicate the uncertainty of the median relation in each bin .
the decrease as a function of height is a statistically significant over most of the spectral types .
some of the spectral types were omitted from figure [ lbol ] because they lacked a sufficient number of active stars for a robust study .
/@xmath10 ) as a function of vertical distance from the plane ( reproduced from w08 ) .
the narrow error bars represent the spread of the values and the wide error bars indicate the uncertainty of the median relation in each bin .
the decrease as a function of height is a statistically significant over most of the spectral types.,width=336 ] west et al .
( 2006 , 2008 ) devolved a 1d dynamical model that traces the vertical dynamics of stars as a function of time .
the model assumes a constant star formation rate and adds a new population of 50 stars every 200 myr for a total simulation time of 10 gyr .
each new population of stars begins with a randomly drawn velocity dispersion 8 km s@xmath4 ( binney et al . 2000 ) .
orbits are integrated using a `` leap frog '' integration technique ( press et al .
1992 ) and the vertical galactic potential from siebert et al .
( 2003 ) .
dynamical heating was simulated by altering the velocities of stars such that their velocity dispersions as a function of age match a @xmath11 relation ( wielen 1977 ; fuchs et al .
2001 ; hnninen & flynn 2002 ; nordstrm et al .
energy was added to stars that were within a certain distance from the plane .
this `` region of influence '' is a way to parameterize the cross section of interaction with molecular clouds during a plane crossing ( see west et al .
2006 for more information ) .
the majority of the molecular gas is constrained to a small range of galactic heights and the dynamical interaction depends on the proximity to the cloud , the density of the gas and the velocity of the star , all of which are absorbed in the `` region of influence '' .
the size of the `` region of influence '' was varied from @xmath120.5 to @xmath125 pc in intervals of 0.5 pc .
seventy simulations were run for each spectral type .
each simulation recorded the velocity , position , and age of each star .
w08 introduced a binary activity state to the dynamical models ; stars started their lives as active and after a finite `` lifetime '' , the activity turned off .
the active lifetimes were varied from 0.0 to 9.0 gyr in 0.5 gyr intervals for each spectral type .
the resulting model active fractions ( as a function of vertical distance from the plane ) were compared to the empirical relations using a chi - squared minimization technique and a best - fit model for each spectral type was determined .
the resulting h@xmath0 activity lifetimes are shown in figure [ lifetimes ] .
the results from the hawley et al .
( 2000 ) cluster study are overplotted for comparison ( dotted ) .
w08 found that there is a significant increase in activity lifetimes between spectral types m3 and m5 , possibly indicating a physical change in the production of magnetic fields as the stellar interiors become fully convective .
activity lifetimes of m dwarfs as determined by comparing the sdss dr5 spectroscopic data to 1d dynamical simulations ( stars ; reproduced from w08 ) .
the hawley et al . ( 2000 ) activity lifetime relation is overplotted for comparison .
as predicted , the hawley relation provides a lower limit to the ages .
w08 found that there is a significant increase in activity lifetimes between spectral types m3 and m5 , possibly indicating a physical change in the production of magnetic fields as full convection sets in .
, width=297 ] figure [ lbol ] suggests that activity is not simply a binary process , but rather that the amount of magnetic activity may decrease over time . to quantify this decrease
, we compared the median log(@xmath9/@xmath10 ) from figure [ lbol ] to the median ages in the same vertical distance bins from the w08 1d dynamical models .
the result , shown in figure [ actage ] , gives a statistical , quantitative relationship between the h@xmath0 activity of an m dwarf and it s age .
horizontal error bars represent the spread in age at a given bin .
the age - activity relations are consistent with a smooth skumanich - like activity decrease until the star reaches its activity lifetime , after which it rapidly falls to its eternal inactive state .
/@xmath10 ) as a function of age ( as derived from the median age of the same bins in the w08 dynamical models ) . the error bars represent the spread in the distributions .
the best - fit function is overplotted for each spectral type ( dashed).,width=345 ] we fit the activity - age relations in figure [ actage ] with a function of the form : @xmath13 where @xmath14 , @xmath15 and @xmath16 are coefficients , @xmath17 is the active lifetime ( gyr ) , and @xmath18 is the age measured in gyr .
the exponent @xmath16 was forced to be the same for all spectral types .
the lifetime @xmath17 coefficient was allowed to vary within the uncertainties of the derived active lifetimes from w08 ( figure [ lifetimes ] ) .
the resulting best - fit coefficients can be found in table [ coef ] .
we caution that these relations are purely statistical and may not be appropriate for use with individual stars .
additional discretion should be used with the functional fits , which may not be justifiably extrapolated beyond where data exist .
.coefficients for best - fit activity - age relation [ cols="^,^,^,^,^",options="header " , ]
large astronomical surveys have produced m dwarf samples of unprecedented size .
we have shown that using the statistical foothold of the largest low - mass spectroscopic sample ever assembled , a strong tie between magnetic activity ( as traced by h@xmath0 ) and age has been established and that m dwarfs have finite active lifetimes .
these lifetimes are a strong function of spectral type , dramatically increasing as the m dwarf interiors become fully convective .
we derived age - activity relations for m2-m7 dwarfs based on the average activity as a function of height above ( or below ) the galactic plane .
future studies will extend these relations to other spectral types .
in addition , these relations can be confirmed and calibrated using both white dwarf cooling ages in wide binary systems that have both a white dwarf and an m dwarf ( see contributions by s. catalan , m. salaris , and j. kalirai ; silvestri et al .
2006 ) , and deep cluster observations of m dwarfs ( when such observations become possible ) . , s. l. , reid , i. n. , & tourtellot , j. g. 2000 , in very low - mass stars and brown dwarfs , edited by r. rebolo and m. r. zapatero - osorio
. published by the cambridge university press , uk , 2000 .
, p.109 , ed .
r. rebolo & m. r. zapatero - osorio , 109+ | we present a study of the dynamics and magnetic activity of m dwarfs using the largest spectroscopic sample of low - mass stars ever assembled .
the age at which strong surface magnetic activity ( as traced by h@xmath0 ) ceases in m dwarfs has been inferred to have a strong dependence on mass ( spectral type , surface temperature ) and explains previous results showing a large increase in the fraction of active stars at later spectral types .
using spectral observations of more than 40000 m dwarfs from the sloan digital sky survey , we show that the fraction of active stars decreases as a function of vertical distance from the galactic plane ( a statistical proxy for age ) , and that the magnitude of this decrease changes significantly for different m spectral types . adopting a simple dynamical model for thin disk vertical heating , we assign an age for the activity decline at each spectral type , and
thus determine the activity lifetimes for m dwarfs .
in addition , we derive a statistical age - activity relation for each spectral type using the dynamical model , the vertical distance from the plane and the h@xmath0 emission line luminosity of each star ( the latter of which also decreases with vertical height above the galactic plane ) . |
in particle physics , the exact hadronization process is unknown . to determine the structure of events at high and low energies a model for hadronization from first principles
is needed but it is still missing .
there have been many attempts at modeling creation within the quark model ( qm ) formalism arising from micu s suggestion @xcite that hadron decays proceed through @xmath6 pair production with vacuum quantum numbers , i.e. @xmath7 . since the @xmath6 pair corresponds to a @xmath8 quark - antiquark state , this model is known as the @xmath8 pair - creation model @xcite .
new studies have been conducted by the clas collaboration @xcite .
these have tried to extract the flavor - dependence of the creation in two - body exclusive reactions .
the study of creation can help us to understand how quarks become observable hadrons , which now is an open problem . in this letter
, we will focus on the role of nucleon sea quarks in hadron production of baryon - meson final state for the exclusive reactions , when there are no decay chains . since the evidence for the flavor asymmetry of the proton sea was found by nmc at cern @xcite
, many studies have been carried out to explain the importance of the quark - antiquark content in the nucleon and its role in the observables .
we computed the ratios of this baryon - meson production by inferring them from the continiuum components of the nucleon using the unquenched quark model ( uqm ) for baryons@xcite .
we also extracted the probabilities of the sea quarks and their relation with creation in electroproduction .
the uqm is based on a quark model ( qm ) with continuum components , which quark - antiquark pairs with vacuum quantum numbers are added as a perturbation employing a @xmath9 model for @xmath10 pair creation .
this approach , which is a generalization of the unitarized quark model by trnqvist and zenczykowski @xcite , was motivated by the work by isgur and coworkers on the flux - tube breaking model .
they showed that the qm emerges as the adiabatic limit of the flux - tube model to which the effects of pair creation can be added as a perturbation @xcite .
the pair - creation mechanism is inserted at the quark level and the one - loop diagrams are calculated by summing over a complete set of intermediate baryon - meson states . under these assumptions ,
the baryon wave function consists of a zeroth order three - quark configuration @xmath11 plus a sum over all the possible higher fock components due to the creation of @xmath9 quark - antiquark pairs .
\ \ \ \
\label{ucqmwf } \end{gathered}\ ] ] here , @xmath13 denotes the initial baryon , @xmath14 and @xmath15 represent the asymptotic baryon and meson system , @xmath16 , @xmath17 and @xmath18 are their respective energies calculated in the rest frame of the initial baryon a , @xmath19 and @xmath20 are the relative momentum and orbital angular momentum between @xmath14 and @xmath15 respectively , and @xmath21 is the total angular momentum @xmath22 .
the operator @xmath23 creates a quark - antiquark pair in the @xmath9 state with the quantum numbers of the vacuum : @xmath24 and @xmath25 @xmath26^{(0)}_0 \ , b_4^{\dagger}(\vec{p}_4 ) \ , d_5^{\dagger}(\vec{p}_5 ) ~. \label{3p0}\end{aligned}\ ] ] here , @xmath27 and @xmath28 are the creation operators for a quark and an antiquark with momenta @xmath29 and @xmath30 , respectively .
the quark - antiquark pair is characterized by a color singlet wave function @xmath31 , a flavor singlet wave function @xmath32 , a spin triplet wave function @xmath33 with spin @xmath34 and a solid spherical harmonic @xmath35 that indicates that the quark and antiquark are in a relative @xmath36 wave .
the operator @xmath23 creates a pair of constituent quarks with an effective size .
thus , the pair creation point is smeared out by a gaussian factor whose width is given by @xmath37 . [ 0.5 ]
in the uqm , the asymptotic states are included in the wave function of the proton .
these components are related to the creation of @xmath38 pairs , with the quantum numbers of the vacuum as it is described in the @xmath8 mechanism .
here we describe the production ratios from nucleon @xmath39 in exclusive reactions .
the production ratios in exclusive reaction from proton target were just measured by the clas collaboration @xcite .
we computed the ratios , inferring from the ratio of probability to find the virtual - baryon - meson - components ( bc ) in the nucleon wave function .
@xmath40 in the electroproduction process we could assume that the proportional constant could be the same for baryon - meson final states , the production ratio from proton target , for example , could be written as follow : @xmath41 [ we have noticed that this assumption is not always true , since in the process a phase space factor should be added . however , in the recent study by mestayer _
@xcite the ratios between @xmath42 and @xmath43 are approximately independent of @xmath44 ( total hadronic energy in the center of mass frame ) .
the complete theoretical dependence of the results on the energy is the subject of subsequent article @xcite . ]
[ relvr ] comes from diagram ( a ) in the fig .
[ fig1 ] , where the virtual photon @xmath45 is absorbed by the virtual baryon @xmath14 , by the quark created in the @xmath8 vertex . under this picture , after the quark from the @xmath8 vertex absorbs the photon , it can not close the @xmath8 loop and the hadronization occurs .
since the final baryon - meson state @xmath46 is observed experimentally in exclusive reactions , only the virtual component bc in the nucleon wave function will contribute if @xmath46 virtual state is the same as the @xmath46 state that can be observed in exclusive reactions .
in the ratios , the coupling constants and other factors due to the process cancel out , thus the ratios in exclusive reactions are inferred from the sea quarks of the nucleon . here
we study the exclusive reactions where octet and decuplet baryons in combination with pseudoscalar meson could be observed , thus the other components do not contribute . to extend our study to the description of baryon - vector meson production , another process in diagram ( a )
( see fig [ fig1 ] ) should be considered , where the photon is absorbed by the antiquark in the meson c. the interference between this two diagrams would be subject of subsequent article @xcite .
the virtual photon can not couple to any valence quark , since if the virtual photon is absorbed by a valence quark the @xmath8 loop would close , diagram ( b ) in the fig .
[ fig1 ] , as a result a @xmath47 resonance is produced , that is not considered here because we are studying the exclusive reactions .
there exists another process in which the virtual photon creates a @xmath48 pair , that may contribute to the electroproduction of baryon - meson states , see fig [ fig1 ] diagram ( c ) .
the contribution of this creation process is expected to be small and is not considered here . in the uqm
the probability of finding a @xmath46 component is the product of a spin - flavor - isospin factor and a radial integral as follows : @xmath49 where the factor @xmath50 is the ratio of the color - flavor - spin - isospin ( cfsi ) factors and the integral @xmath51 is simply defined in general for the production of a baryon b and a meson c , from a baryon a , as @xmath52 the value of @xmath53 depends on the size of the harmonic oscillator wave functions for baryons and mesons , @xmath54 , and @xmath55 respectively , and the gaussian smearing of the pair - creation vertex , @xmath56 , it is taken from ref .
@xcite to be @xmath57 gev@xmath58 . in ref .
@xcite , @xmath54= 0.32 gev , @xmath55= 0.40 gev and @xmath56= 0.30 fm .
the energy denominator represents the energy difference between initial and final hadrons calculated in the rest frame of the initial baryon .
the color - flavor - spin - isospin factor plays an important role in the production rates .
in fact , for example if we consider the limit case in which , the two final bc states in the ratio are in the same isospin channels , the radial contribution is the same and the production rate will only depend on this factor .
thus the predicted ratio for @xmath59 is well understood on the basis of isospin symmetry ( it is conserved in the uqm ) , for which the value is @xmath60 .
however , the general result is determined by a combination of ratios of cfsi factors and integrals over the relative momentum , see eq .
( [ ratio ] ) . in the case of flavor symmetry ,
the ratios are determined completely by the cfsi factors .
for example , in the limit of flavor symmetry the ratio @xmath61 would be @xmath62 .
the effect of @xmath63 symmetry breaking in the uqm can be observed by a comparison with uqm result @xmath64 from table [ tab : res ] .
inspection of table [ tab : res ] shows that the observed rates are reproduced very well by our calculations .
in addition , we present the results for some other channels . moreover , using the isospin symmetry we computed the production rates from the neutron .
we are aware that the neutron target does not exist but eventually those results can be extracted from new experiments on deuterium or tritium .
cclc + ratio & ratio & uqm & exp .
@xcite + from neutron&from proton & & from proton + + + @xmath65&@xmath61 & 0.227 & @xmath66 + @xmath67&@xmath68 & 0.454 & @xmath69 + @xmath70 & @xmath71 & 0.5 & @xmath72 + @xmath73 & @xmath74 & 0.007 & @xmath75 + @xmath76 & @xmath77 & 0.014 & @xmath75 + @xmath78 & @xmath79 & 0.014 & @xmath75 + @xmath80 & @xmath81 & 0.028 & @xmath75 + @xmath82 & @xmath83 & 0.045 & @xmath75 + @xmath84 & @xmath85 & 0.090 & @xmath75 + @xmath86 & @xmath87 & 0.090 & @xmath75 + @xmath88 & @xmath89 & 0.18 & @xmath75 + [ tab : res ]
as we have already shown , the asymptotic states play an important role in exclusive two - body production , since the ratios of this process can be determined in a straightforward way by means of the uqm inferred from the virtual components produced by the quantum vacuum fluctuations
. likewise we can infer from the flavor content of the proton the creation probabilities on the @xmath46 final states . within this approach the production of the in the asymptotic states of the proton
is linked with the final states produced by electroproduction . here
, we extract the probabilities from sea quarks and we only point out the fact that the production of is related to the sea quarks in the nucleon . as an example we study the strangeness suppression factor ( ssf ) @xmath5 is defined in the literature @xcite as : @xmath90 where @xmath91 , @xmath92 and @xmath93 are the mean numbers of @xmath94 quarks and the corresponding antiquarks in the final state . the probabilities to produce a @xmath95 pair
are computed @xmath96 where @xmath97 is the operator that counts the @xmath95 pairs created in the electroproduction from nucleon , @xmath98 and @xmath99 are the all possible baryon and meson states that can be observed in exclusive reactions .
our results for the production ratios are present in table [ strange ] , the column uqm@xmath100 contains the production ratios with pseudoscalar mesons in combination with octet and decuplet baryons ( @xmath101 , @xmath102 , @xmath2 , @xmath103 , @xmath104 , @xmath105 , @xmath106 channels ) , while the column uqm@xmath107 only takes into account @xmath108 and @xmath109 mesons in combination with octet and decuplet baryons ( @xmath101 , @xmath102 , @xmath2 , @xmath103 , @xmath104 channels ) .
since the isospin symmetry is valid for the nucleon wave function in the uqm , we present our results for the production ratios of proton and neutron in table[strange ] .
the value for the strangeness suppression factor @xmath5 is in good agreement with the observed values from exclusive reactions at jlab @xcite and in high - energy production experiments at cern @xcite .
c ccccc & proton & & & + ratio & uqm@xmath100 & uqm@xmath107 & exp . & ref .
+ @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xcite + @xmath114 & @xmath115 & @xmath115 & @xmath116 & @xcite + @xmath117 & @xmath118 & @xmath119 & @xmath120 & @xcite + & & & @xmath121 & @xcite + & neutron & & & + + @xmath122 & @xmath111 & @xmath112 & + @xmath123 & @xmath115 & @xmath115 & & + @xmath117 & @xmath118 & @xmath119 & & +
in summary we computed the production ratios of baryon - meson final states in exclusive reactions and we found that our predictions of those ratios are in good agreement with the new experimental data at jlab @xcite as a first theoretical calculation that explains the new data reported in the letter @xcite .
we also computed other baryon - meson production ratios still unobserved but that in principle could be observed in exclusive reactions from proton or neutron initial states in future new experiments . in particular , for the neutron case we suggested new experiments on deuterium or tritium .
we showed that sea quarks play an important role in the electroproduction since our predictions are in agreement with the experimental results , thus we found that the meson - baryon production in exclusive reactions could provide information on the sea components of the nucleon .
therefore future experiments of exclusive reactions can provide another test of different predictions concerning the nucleon structure .
an interesting result was the production ratio @xmath124 from the proton , which is compatible with the experimental value obtained by mestayer _
et al _ at jlab @xcite , unlike the unity value used in high energy hadronization models .
those two finding reinforce each other ( the experimental one by @xcite , and our theoretical prediction ) and in our view are well explained from the fact that there is a excess on proton as seen by nmc and hermes @xcite and if one believes that the production is governed by the sea quark on proton , where the is bigger than the content , this result on the ratio different from one is a direct consequence . in the case of neutron
, we predicted @xmath125 thus the production would be higher than the production , since isospin symmetry yields a excess over in neutron , but this prediction can be tested by future experiments at jlab and lhc . on the contrary
it is worthwhile noticing that @xmath5 is the same from proton and neutron .
the extraction of quark - antiquark production in exclusive reactions from nucleon gave us a suppression of strange quark - antiquarks compared to nonstrange pairs .
this is in a good agreement with the available experimental data both from jlab @xcite and cern @xcite . finally , in ref .
@xcite the experimentalists observed that the measured rates are independent from the type and energy of any resonance formed in the intermediate state .
as such , the results should reflect some fundamental aspect of flavor production like the breaking of the underling su(3 ) symmetry induced by the sea quark components .
moreover , our picture can be extended to study the photoproduction and hadron production with meson beams
. it will be the subject of a future article @xcite , that eventually can be tested at jpark .
it is worthwhile noticing that the results for the uqm are independent of the overall parameter or strength of the @xmath9 quark - antiquark pair creation vertex , usually called @xmath126 .
the values of the remaining parameters were taken from previous work and no attempt was made to optimize their values .
the authors acknowledge interesting discussion with dr .
daniel carman ( jlab ) .
this work is supported in part by infn sezione di genova and papiit - dgapa , mexico ( grant in107314 ) .
60 n. a. trnqvist and p. zenczykowski , phys . rev .
d * 29 * , 2139 ( 1984 ) ; z. phys .
c * 30 * , 83 ( 1986 ) ; p. zenczykowski , annals phys . *
169 * , 453 ( 1986 ) .
p. geiger and n. isgur , phys .
d * 55 * , 299 ( 1997 ) . | we describe the electroproduction ratios of baryon - meson states from nucleon , inferring from the sea quarks in the nucleon using an extension of the quark model that takes into account the sea . as a result we provide , with no adjustable parameters , the predictions of ratios of exclusive meson - baryon final states:@xmath0 , @xmath1 , @xmath2 , @xmath3 , and @xmath4 .
these predictions are in agreement with the new jlab experimental data showing that sea quarks play an important role in the electroproduction .
we also predicted further ratios of exclusive reactions that can be measured and tested in future experiments .
in particular , we suggested new experiments on deuterium and tritium .
such measurements can provide crucial test of different predictions concerning the structure of nucleon and its sea quarks helping to solve an outstanding problem . finally , we computed the so called strangeness suppression factor , @xmath5 , that is the suppression of strange quark - antiquarks compared to nonstrange pairs , and we found that our finding with this simple extension of the quark model is in good agreement with the results of jlab and cern experiments .
electroproduction , strangeness , nucleon , arxiv:1601.06987 00 - 01,99 - 00 |
fast and accurate predictions of binding free energies between proteins and small organic ligands would have significant impact on designing drugs @xcite and other modulators of biological processes
. the clear relevance of protein - ligand binding affinity prediction in chemical biology and drug discovery has inspired a vast array of physics - based methods ( for a broad review , see @xcite ) , each with a different trade - off between computational accuracy and speed .
on one extreme , molecular docking focuses on speed .
docking algorithms are designed to quickly obtain plausible configurations of a protein - ligand complex .
scoring functions are then used to rank one configuration versus another .
docking programs are commonly assessed by their ability to redock ligands into crystallographic structures from which they have been removed .
comparative studies @xcite and blinded exercises @xcite consistently show that docking methods are adept at generating the native pose but are less competent at giving it the highest rank . in this context
, it is not surprising that docking scores are poorly correlated with binding free energies @xcite .
in contrast , alchemical pathway methods are based on rigorous statistical mechanics .
the methods involve sampling from a series of possibly nonphysical thermodynamic states in between end - states where the receptor and ligand are bound and unbound . in the unbound state , the receptor - ligand nonbonded interaction terms
may be switched off or the species may be physically separated . in accordance to the established statistical mechanics of noncovalent binding @xcite ,
the receptor is usually allowed full flexibility . unfortunately
, sampling a fully flexible complex from multiple statistical distributions along an alchemical pathway generally requires substantial computing resources ; it has been suggested that most published studies are not fully converged @xcite ! even sampling the ligand binding pose is challenging @xcite . to bypass this difficulty ,
most pathway calculations pursue relative binding free energies between similar molecules and assume that the binding mode does not change . in absolute binding free energy calculations , ligand
sampling issues are usually alleviated by confining the molecule to a specific pose @xcite .
within this restricted range of problems , alchemical pathway calculations are amassing a growing track record of accurate prediction ( e.g. @xcite ) . _
the success of these methods suggests that docking may be substantially improved by incorporating rigorous statistical mechanics . _ implicit ligand theory ( ilt ) @xcite , a recently derived statistical mechanics framework for noncovalent association , has the potential to inspire new methods that combine the speed of docking and the rigor of alchemical pathway methods .
ilt formally separates receptor and ligand sampling into two distinct stages .
because the first stage of receptor sampling does not require a ligand , receptor conformations can be sampled once and used with many different ligands .
in contrast , conventional alchemical pathway methods require thorough receptor sampling for every receptor - ligand pair .
the second stage is to calculate the binding potential of mean force ( bpmf ) - the binding free energy between a ligand and a _ rigid _ receptor configuration ( eq . [ eq : bpmf ] ) - for each receptor configuration .
finally , the standard binding free energy is an exponential average of bpmfs ( eq . [ eq : fe_ilt ] ) .
while multiple bpmfs are required to estimate a standard binding free energy , the overall calculation should require less computer time than conventional methods because bpmfs are easier to estimate than binding free energies with a fully flexible receptor .
the key reason for this speedup is that a bpmf calculation only requires sampling of the ligand , which usually has many fewer degrees of freedom than the complex .
furthermore , nonbonded interactions between a rigid receptor and ligand can be treated by interpolating precomputed three - dimensional grids , a strategy first developed for docking @xcite .
once the grid is stored , calculation time no longer depends on the size of the receptor .
in contrast , conventional alchemical pathway methods require frequent force evaluation between flexible receptor atoms . as the number of pairwise interactions scales as @xmath1 with n receptor atoms ( neglecting cutoffs ) , the relative efficiency of ilt - based methods will be more pronounced as receptor size increases . in the first paper on ilt
, bpmfs were estimated for simple host - guest systems @xcite . here
, the focus is on precise bpmf estimation for protein - ligand systems . @xcite
previously computed binding free energies between simple ligands and a rigid protein , t4 lysozyme .
in contrast , the present work involves more diverse systems .
@xcite were equally rigorous , but did not implement specialized methods to significantly speed their calculations compared to flexible - receptor calculations .
@xcite developed a fast method based on a number of approximations .
as in other bpmf @xcite and standard binding free energy @xcite calculations , hamiltonian replica exchange is applied .
the main methodological differences between the current and previous work are the adaptive initialization of thermodynamic states and the use of linearly - scaled interaction grids @xcite with transformation - based smoothing @xcite ; these will be discussed in detail in the section on _ theory and methods_. the method is demonstrated on the astex diverse set @xcite , a curated database of 85 high - quality protein - ligand complexes of pharmaceutical or agrochemical interest .
the algorithm is implemented in a new software package , alchemical grid dock ( algdock ) , a python module based on the molecular modeling toolkit ( mmtk ) 2.7.8 @xcite .
algdock is available under the open - source mit license at https://github.com / ccbatiit / algdock/.
this section reviews implicit ligand theory , details the algorithms in algdock , and describes the setup of the bpmf calculations on the astex diverse set . for the noncovalent association between a receptor @xmath2 and ligand @xmath3 to form a complex @xmath4 , @xmath5 , the standard binding free energy is , @xmath6 where @xmath7 is the inverse of boltzmann s constant times the temperature , @xmath8 is the standard concentration ( 1 m = 1/1660 @xmath9 ) , and @xmath10 is the equilibrium concentration of species @xmath11 .
activities have been assumed to be unity , a reasonable approximation in the limit of low concentrations .
coordinates of the complex , @xmath12 , are partitioned into receptor ( @xmath13 ) and ligand internal ( @xmath14 ) and external ( @xmath15 ) coordinates .
based on this partitioning , the interaction energy is defined as @xmath16 , where @xmath17 is an effective potential energy that includes the gas - phase potential energy @xmath18 and solvation free energy @xmath19 @xcite .
a bpmf is an exponential average of interaction energies over ligand coordinates in the binding site @xcite , @xmath20 where @xmath21 is an indicator function that takes values between 0 and 1 and specifies whether the receptor and ligand are bound or not .
according to ilt , the standard binding free energy @xmath22 is related to an exponential average of bpmfs over boltzmann - distributed receptor configurations @xmath13 , @xmath23 } dr_r } { \int e^{-\beta { \mathcal u}(r_{r } ) } dr_r } \right ) - \beta^{-1 } \ln \left ( \frac{\omega c^\circ}{8 \pi^2 } \right ) , \label{eq : fe_ilt}\end{aligned}\ ] ] where @xmath24 is the volume of the binding site .
a bpmf can also be expressed as a ratio of partition functions @xcite
, @xmath25 } ~ dr_l d\xi_l } \right ) .
\label{eq : bpmf_ratio}\end{aligned}\ ] ] in algdock , bpmfs are estimated by completing the thermodynamic cycle shown in figure [ fig : thermo_cycle ] . ,
width=322 ] mathematical expressions for individual free energy differences within the cycle are given in table [ tab : b_parts ] .
the sum of all the free energy components in table [ tab : b_parts ] is @xmath26 . .
* thermodynamic cycle for bpmfs : expressions * [ cols="^,^,^,^",options="header " , ] the least precise free energy estimates between adjacent milestones .
counts are based on all simulations or the imprecise subset , when the standard deviation of the total bpmf is greater than 5.0 .
[ tab : least_precise ] for systems with poor convergence of bpmf estimates , longer sampling is not necessarily an effective way to improve the precision of free energy estimates . while @xmath27 $ ] usually decreases monotonically with sample size , this trend is not universally true ( figure [ fig : std_fe_cycle ] ) . in some systems
there are temporary jumps in @xmath27 $ ] due to instability in determining the equilibrated cycle .
more importantly , in other systems , @xmath27 $ ] appears flat , with little or no change with increased sampling .
$ ] based on 15 independent simulations of ( a ) the ligand or @xmath28 $ ] based on the ligand simulations and 15 independent simulations of the complex starting from the ( b ) crystallographic , ( c ) docked , and ( d ) lowest - energy observed poses .
[ fig : std_fe_cycle],width=641 ] + in some systems , the variance is low but free energy estimates are dependent on the starting poses .
when results from all simulations between milestones c&d are grouped together , the standard deviation in many systems no longer converges to within 5 @xmath29 . because converged thermodynamic quantities
should not depend on initial conditions , this inconsistency implies that some simulations are not fully converged . when free energy estimates contradict , it is not completely clear which result is most consistent with the force field .
false convergence is quantified based on the minimum value , using @xmath30 $ ] .
a system is considered to be falsely converged when @xmath31 $ ] is less than 5 @xmath29 but @xmath32 $ ] is greater than 15 @xmath29 .
based on this definition , false convergence occurs in a nontrivial fraction of systems ( table [ tab : nconverged ] ) .
ultimately , poor convergence and false convergence are caused by incomplete sampling .
the most straightforward explanation for incomplete sampling is that a simulation starts and remains trapped in a local minimum distinct from the global minimum energy .
indeed , between milestones c&d , most falsely converged calculations starting from the crystal structure ( 4/5 ) and docked poses ( 7/11 ) have no starting pose within 1.0 of the lowest - energy observed pose . however , sampling the lowest - energy structure is not a sufficient condition for complete sampling , as other poses may have a lower _
free energy_. for simulations starting from the lowest - energy observed pose , four systems appear falsely converged between milestones c&d ( formatted by pdb identifier ( @xmath31 , \sigma[f , f_{min}]$ ] ) with @xmath33 in units of @xmath29 ) : 1hww ( 0.353 , 16.1 ) , 1r1h ( 3.95 , 15.3 ) , 1hp0 ( 0.496 , 16.7 ) , 1gm8 ( 0.422 , 15.6 ) . for 1hww ( figure [ fig : dock_last](c ) and
( d ) ) and 1gm8 ( figure [ fig : dock_last](g ) and ( h ) ) , starting from the crystallographic pose leads to a lower estimate of @xmath34 than the lowest - energy pose . for 1hp0 ( figure [ fig : dock_last](e ) and
( f ) ) , a docked pose has the lowest estimated @xmath34 .
1r1h does not appear to sample different poses ; the large @xmath32 $ ] is the result of an outlier with a significantly lower @xmath34 estimate than average .
the selection of intermediate states between two thermodynamic milestones of interest is a ubiquitous problem in molecular simulation .
the usual approach is an iterative trial - and - error process starting from a naive protocol , checking for issues such as replica exchange bottlenecks , and inserting and removing states as necessary .
i have developed a simple and robust approach to initialize a series of thermodynamic states based on only a single adjustable parameter , the thermodynamic speed .
due to this automated procedure , i was able to run a large number of simulations on a diverse array of protein - ligand complexes . in the vast majority of cases ,
the protocols yielded consistent replica exchange rates across neighboring thermodynamic states without further fine - tuning .
the described approach to trailblazing thermodynamic state space may find use in other classes of simulations .
replica exchange calculations in this present study include more thermodynamic states than most published molecular simulations .
conventional wisdom about replica exchange is that an optimal number of replicas will maximize efficiency . with too few replicas ,
there is limited phase space overlap between neighbors and exchange rates are vanishingly small . with too many replicas , metrics of replica exchange efficiency , such as a mean round - trip time ,
however , in a recent study involving extensive simulation of several distinct processes , it was found that if there are no bottlenecks , the number of states has little impact on the convergence of free energy estimates ( the manuscript is under preparation ) .
i chose to include a large number of states to allow for the possibility that later sampling will explore different regions of phase space and reduce the exchange rate between neighboring states .
while useful , the described thermodynamic state initialization process remains imperfect .
replica exchange is particularly beneficial when the important phase space of a thermodynamic states is a subset of the important phase space of another .
the present procedure is limited to the variation of a single thermodynamic parameter between milestones , and can do little to promote the subset relationship .
future improvements could accomodate varying multiple parameters ( e.g. separate parameters for the temperture , van der waals grids , and electrostatic grids ) in between thermodynamic states of interest , while maintaining a phase space subset relationship .
while a direct comparison is difficult , the described procedure appears less successful than the best docking programs at identifying the native pose .
@xcite found that with a larger site ( 10 ) , the top - ranked gold solution is within 2 of the experimental binding mode in 80.4% of calculations .
in contrast , only 56.5% of the present ucsf dock 6 calculations were successful , suggesting that setup and parameterization , and perhaps the docking algorithm , could be fine - tuned .
native pose identification based on algdock performs similarly to ucsf dock 6 , suggesting that replica exchange sampling from the boltzmann distribution does not significantly increase the diversity of poses . as evidenced by the variable performance as a function of starting pose(s ) ,
improving boltzmann sampling schemes should improve native pose identification . nonetheless , the sampling and estimation approach is able to attain precise bpmf estimates in the majority of systems . as for the others ,
it is unsurprising that the same approach does not work equally well in all cases . with a diverse set
, ligands and receptors will exhibit distinct levels of complexity and may be driven to bind by different factors . indeed , the inadequate sampling that led to convergence issues for several systems is probably not unique to bpmf calculations @xcite . because of the relative simplicity of the calculations , i have been able to conduct more replicates from more distinct starting points than other alchemical standard binding free energy calculations .
the sheer number of simulations has allowed us to expose problems that may not be clear from fewer calculations .
our results suggest that hamiltonian replica exchange with molecular dynamics is not always an adequate approach for sampling distinct ligand binding poses .
specialized sampling methods will likely be required in a more broadly effective protein - ligand bpmf estimation procedure .
it is also clear that for some systems , there may be limited phase space overlap between different implicit solvent models .
more precise bpmf estimation can be attained by introducing intermediate states or implementing more similar force fields for milestones a&b and d&e .
i have developed a method to initialize thermodynamic states and estimate bpmfs for protein - ligand systems .
the largest sources of imprecision are found to be ligand pose sampling and phase space overlap between implicit solvent models .
intriguingly , i found that starting from the lowest - energy pose is not a sufficient condition for converged bpmf estimates .
future studies may build on the current work to compare bpmfs with molecular docking scores as classifiers of binding .
another useful direction will be to assess the convergence of standard binding free energy estimates upon increased receptor conformational sampling .
i thank michael shirts ( university of virginia ) , john chodera ( mskcc ) , and trung hai nguyen ( iit ) for helpful discussions .
peter eastman ( simbios ) and john chodera assisted with implementing an early version of the code in openmm .
michael sherman ( simbios ) made an invaluable suggestion of using precomputed grids .
rob gardner , lincoln bryant , and balamurugan desinghu ( open science grid ) and tom milledge ( dscr ) provided assistance with computing resources .
david beratan ( duke ) was a supportive postdoctoral advisor .
openeye scientific software , inc .
and ucsf provided academic licenses to their software . at the beginning of this project ,
i was a postdoctoral scholar at duke , supported by nih 2p50 gm-067082 - 06 - 10 .
i also spent a month as an openmm visiting scholar at simbios .
53ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1007/s10822 - 010 - 9363 - 3 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0959440x11000261 [ * * , ( ) ] link:\doibase 10.1063/1.4769292 [ * * , ( ) ] link:\doibase 10.1146/annurev.biophys.36.040306.132550 [ * * , ( ) ] link:\doibase 10.1021/jm050362n [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/ci400025f [ * * , ( ) ] link:\doibase 10.1021/ci049733j [ * * , ( ) ] link:\doibase 10.1021/ci200269q [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1007/s10822 - 011 - 9497-y [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.2221683 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0022283607007760 [ * * , ( ) ] link:\doibase 10.1021/jm800524s [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0022283609011760 [ * * , ( ) ] link:\doibase 10.1021/jp807701h [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/ci200126v [ * * , ( ) ] link:\doibase 10.1073/pnas.1114017109 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/ja512751q [ * * , ( ) ] link:\doibase 10.1063/1.4751284 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/ct400995c [ * * , ( ) ] link:\doibase 10.1007/s10822 - 012 - 9552 - 3 [ * * , ( ) ] link:\doibase 10.1007/s10822 - 013 - 9689 - 8 [ * * , ( ) ] http://www.ncbi.nlm.nih.gov/pubmed/12479928 [ * * , ( ) ] link:\doibase 10.1021/jm061277y [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1073/pnas.181342398 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop ( ) , @noop * * , ( ) link:\doibase 10.1063/1.431689 [ * * , ( ) ] http://pre.aps.org/abstract/pre/v80/i4/e046705 [ * * , ( ) ] http://scitation.aip.org/getabs/servlet/getabsservlet?prog=normal&id=prltao000099000010100602000001&idtype=cvips&gifs=yes [ * * , ( ) ] http://pubs.acs.org/doi/abs/10.1021/ct1001768 [ * * , ( ) ] link:\doibase 10.1063/1.3660669 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) link:\doibase 10.1002/jcc.20035 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1002/jcc.10128 [ * * , ( ) ] link:\doibase 10.1002/jcc.20289 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1088/1742 - 6596/78/1/012057 [ * * , ( ) ] link:\doibase 10.1063/1.1359181 [ * * , ( ) ] @noop * * , ( ) | a binding potential of mean force ( bpmf ) is a free energy of noncovalent association in which one binding partner is flexible and the other is rigid . expanding on previous work with host - guest systems ,
i have developed a method to calculate bpmfs for protein - ligand systems .
the method is based on replica exchange sampling from multiple thermodynamic states at different temperatures and protein - ligand interaction strengths .
protein - ligand interactions are represented by interpolating precomputed electrostatic and van der waals grids . using a simple estimator for thermodynamic length
, thermodynamic states are initialized at approximately equal intervals .
the method is demonstrated on the astex diverse set , a database of 85 protein - ligand complexes relevant to pharmacy or agriculture .
fifteen independent simulations of each complex were started using poses from crystallography , docking , or the lowest - energy pose observed in the other simulations .
benchmark simulations completed within three days on a single processor .
overall , protocols initialized using the thermodynamic length estimator were system - specific , robust , and led to approximately even replica exchange acceptance probabilities between neighboring states . in most systems , the standard deviation of the bpmf converges to within 5 k@xmath0 t .
even with low variance , however , the mean bpmf was sometimes dependent on starting conditions , implying inadequate sampling . within the thermodynamic cycle , free energies estimated based on multiple intermediate states were more precise , and those estimated by single - step perturbation were less precise .
the results demonstrate that the method is promising , but that ligand pose sampling and phase space overlap can sometimes prevent precise bpmf estimation .
the software used to perform these calculations , alchemical grid dock ( algdock ) , is available under the open - source mit license at https://github.com / ccbatiit / algdock/. |
today , very large amounts of information are available in on - line documents . as part of the effort to better organize this information for users , researchers have been actively investigating the problem of automatic text categorization .
the bulk of such work has focused on _ topical _ categorization , attempting to sort documents according to their subject matter ( e.g. , sports vs. politics ) .
however , recent years have seen rapid growth in on - line discussion groups and review sites ( e.g. , the new york times books web page ) where a crucial characteristic of the posted articles is their _ sentiment _ , or overall opinion towards the subject matter for example , whether a product review is positive or negative .
labeling these articles with their sentimentwould provide succinct summaries to readers ; indeed , these labels are part of the appeal and value - add of such sites as www.rottentomatoes.com , which both labels movie reviews that do not contain explicit rating indicators and normalizes the different rating schemes that individual reviewers use .
sentimentclassification would also be helpful in business intelligence applications ( e.g. mindfuleye s lexant system ) and recommender systems ( e.g. , , ) , where user input and feedback could be quickly summarized ; indeed , in general , free - form survey responses given in natural language format could be processed using sentimentcategorization .
moreover , there are also potential applications to message filtering ; for example , one might be able to use sentimentinformation to recognize and discard `` flames''@xcite . in this paper ,
we examine the effectiveness of applying machine learning techniques to the sentimentclassification problem . a challenging aspect of this problem that seems to distinguish it from traditional topic - based classification is that while topics are often identifiable by keywords alone , sentimentcan be expressed in a more subtle manner .
for example , the sentence `` how could anyone sit through this movie ? ''
contains no single word that is obviously negative .
( see section [ sec : discuss ] for more examples ) .
thus , sentimentseems to require more _ understanding _ than the usual topic - based classification .
so , apart from presenting our results obtained via machine learning techniques , we also analyze the problem to gain a better understanding of how difficult it is .
this section briefly surveys previous work on non - topic - based text categorization .
one area of research concentrates on classifying documents according to their _ source _ or _ source style _
, with statistically - detected stylistic variation @xcite serving as an important cue .
examples include author , publisher ( e.g. , the _ new york times _ vs. _ the daily news _ ) , native - language background , and `` brow '' ( e.g. , high - brow vs. `` popular '' , or low - brow ) @xcite .
another , more related area of research is that of determining the _ genre _ of texts ; subjective genres , such as `` editorial '' , are often one of the possible categories @xcite
. other work explicitly attempts to find features indicating that subjective language is being used @xcite .
but , while techniques for genre categorization and subjectivity detection can help us _
recognize _ documents that express an opinion , they do not address our specific _
task of determining what that opinion actually is
. most previous research on sentiment - based classification has been at least partially knowledge - based .
some of this work focuses on classifying the semantic orientation of individual words or phrases , using linguistic heuristics or a pre - selected set of seed words @xcite .
past work on sentiment - based categorization of entire documents has often involved either the use of models inspired by cognitive linguistics @xcite or the manual or semi - manual construction of discriminant - word lexicons @xcite .
interestingly , our baseline experiments , described in section [ sec : closer ] , show that humans may not always have the best intuition for choosing discriminating words .
turney s work on classification of reviews is perhaps the closest to ours .
he applied a specific unsupervised learning technique based on the mutual information between document phrases and the words `` excellent '' and `` poor '' , where the mutual information is computed using statistics gathered by a search engine . in contrast , we utilize several completely prior - knowledge - free supervised machine learning methods , with the goal of understanding the inherent difficulty of the task .
for our experiments , we chose to work with movie reviews .
this domain is experimentally convenient because there are large on - line collections of such reviews , and because reviewers often summarize their overall sentimentwith a machine - extractable _ rating _ indicator , such as a number of stars ; hence , we did not need to hand - label the data for supervised learning or evaluation purposes .
we also note that found movie reviews to be the most difficult of several domains for sentiment classification , reporting an accuracy of 65.83% on a 120-document set ( random - choice performance : 50% ) .
but we stress that the machine learning methods and features we use are not specific to movie reviews , and should be easily applicable to other domains as long as sufficient training data exists .
our data source was the internet movie database ( imdb ) archive of the rec.arts.movies.reviews newsgroup .
we selected only reviews where the author rating was expressed either with stars or some numerical value ( other conventions varied too widely to allow for automatic processing ) .
ratings were automatically extracted and converted into one of three categories : positive , negative , or neutral . for the work described in this paper , we concentrated only on discriminating between positive and negative sentiment . to avoid domination of the corpus by a small number of prolific reviewers
, we imposed a limit of fewer than 20 reviews per author per sentimentcategory , yielding a corpus of 752 negative and 1301 positive reviews , with a total of 144 reviewers represented .
this dataset will be available on - line at . [
cols="^,<,<,<",options="header " , ]
we used documents from the movie - review corpus described in section [ sec : domain ] . to create a data set with uniform class distribution ( studying the effect of skewed class distributions was out of the scope of this study ) , we randomly selected 700 positive - sentiment and 700 negative - sentiment documents .
we then divided this data into three equal - sized folds , maintaining balanced class distributions in each fold .
( we did not use a larger number of folds due to the slowness of the maxenttraining procedure . ) all results reported below , as well as the baseline results from section [ sec : closer ] , are the average three - fold cross - validation results on this data ( of course , the baseline algorithms had no parameters to tune ) . to prepare the documents , we automatically removed the rating indicators and extracted the textual information from the original html document format , treating punctuation as separate lexical items .
no stemming or stoplists were used .
one unconventional step we took was to attempt to model the potentially important contextual effect of negation : clearly `` good '' and `` not very good '' indicate opposite sentimentorientations . adapting a technique of , we added the tag not _ to every word between a negation word ( `` not '' , `` is nt '' , `` did nt '' , etc . ) and the first punctuation mark following the negation word .
( preliminary experiments indicate that removing the negation tag had a negligible , but on average slightly harmful , effect on performance . ) for this study , we focused on features based on unigrams ( with negation tagging ) and bigrams . because training maxentis expensive in the number of features , we limited consideration to ( 1 ) the 16165 unigrams appearing at least four times in our 1400-document corpus ( lower count cutoffs did not yield significantly different results ) , and ( 2 ) the 16165 bigrams occurring most often in the same data ( the selected bigrams all occurred at least seven times ) .
note that we did not add negation tags to the bigrams , since we consider bigrams ( and @xmath0-grams in general ) to be an orthogonal way to incorporate context .
[ [ initial - unigram - results ] ] initial unigram results + + + + + + + + + + + + + + + + + + + + + + + the classification accuracies resulting from using only unigrams as features are shown in line ( 1 ) of figure [ fig : results - plain ] . as a whole ,
the machine learning algorithms clearly surpass the random - choice baseline of 50% .
they also handily beat our two human - selected - unigram baselines of 58% and 64% , and , furthermore , perform well in comparison to the 69% baseline achieved via limited access to the test - data statistics , although the improvement in the case of svms is not so large . on the other hand , in topic - based classification , all three classifiers have been reported to use bag - of - unigram features to achieve accuracies of 90% and above for particular categories @xcite and such results are for settings with more than two classes .
this provides suggestive evidence that sentimentcategorization is more difficult than topic classification , which corresponds to the intuitions of the text categorization expert mentioned above .
nonetheless , we still wanted to investigate ways to improve our sentimentcategorization results ; these experiments are reported below .
[ [ feature - frequency - vs .- presence ] ] feature frequency vs. presence + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + recall that we represent each document @xmath1 by a feature - count vector ( @xmath2 .
however , the definition of the maxentfeature / class functions @xmath3 only reflects the presence or absence of a feature , rather than directly incorporating feature frequency . in order to investigate whether reliance on frequency information could account for the higher accuracies of naive bayes and svms , we binarized the document vectors , setting @xmath4 to 1 if and only feature @xmath5 appears in @xmath1 , and reran naive bayes and @xmath6on these new vectors . as can be seen from line ( 2 ) of figure [ fig : results - plain ]
, better performance ( _ much _ better performance for svms ) is achieved by accounting only for feature presence , not feature frequency .
interestingly , this is in direct opposition to the observations of with respect to naive bayes topic classification .
we speculate that this indicates a difference between sentimentand topic categorization perhaps due to topic being conveyed mostly by particular content words that tend to be repeated but this remains to be verified . in any event , as a result of this finding , we did not incorporate frequency information into naive bayes and svms in any of the following experiments .
[ [ bigrams ] ] bigrams + + + + + + + in addition to looking specifically for negation words in the context of a word , we also studied the use of bigrams to capture more context in general .
note that bigrams and unigrams are surely not conditionally independent , meaning that the feature set they comprise violates naive bayes conditional - independence assumptions ; on the other hand , recall that this does not imply that naive bayes will necessarily do poorly @xcite .
line ( 3 ) of the results table shows that bigram information does not improve performance beyond that of unigram presence , although adding in the bigrams does not seriously impact the results , even for naive bayes .
this would not rule out the possibility that bigram presence is as equally useful a feature as unigram presence ; in fact , found that bigrams alone can be effective features for word sense disambiguation .
however , comparing line ( 4 ) to line ( 2 ) shows that relying just on bigrams causes accuracy to decline by as much as 5.8 percentage points .
hence , if context is in fact important , as our intuitions suggest , bigrams are not effective at capturing it in our setting .
[ [ parts - of - speech ] ] parts of speech + + + + + + + + + + + + + + + we also experimented with appending pos tags to every word via oliver mason s qtag program .
this serves as a crude form of word sense disambiguation @xcite : for example , it would distinguish the different usages of `` love '' in `` i love this movie '' ( indicating sentimentorientation ) versus `` this is a love story '' ( neutral with respect to sentiment ) .
however , the effect of this information seems to be a wash : as depicted in line ( 5 ) of figure [ fig : results - plain ] , the accuracy improves slightly for naive bayes but declines for svms , and the performance of maxentis unchanged . since
adjectives have been a focus of previous work in sentimentdetection @xcite , we looked at the performance of using adjectives alone .
intuitively , we might expect that adjectives carry a great deal of information regarding a document s sentiment ; indeed , the human - produced lists from section [ sec : closer ] contain almost no other parts of speech . yet , the results , shown in line ( 6 ) of figure [ fig : results - plain ] , are relatively poor : the 2633 adjectives provide less useful information than unigram presence . indeed ,
line ( 7 ) shows that simply using the 2633 most frequent unigrams is a better choice , yielding performance comparable to that of using ( the presence of ) all 16165 ( line ( 2 ) ) .
this may imply that applying explicit feature - selection algorithms on unigrams could improve performance .
[ [ position ] ] position + + + + + + + + an additional intuition we had was that the position of a word in the text might make a difference : movie reviews , in particular , might begin with an overall sentimentstatement , proceed with a plot discussion , and conclude by summarizing the author s views .
as a rough approximation to determining this kind of structure , we tagged each word according to whether it appeared in the first quarter , last quarter , or middle half of the document .
the results ( line ( 8) ) did nt differ greatly from using unigrams alone , but more refined notions of position might be more successful .
the results produced via machine learning techniques are quite good in comparison to the human - generated baselines discussed in section [ sec : closer ] . in terms of relative performance , naive bayes tends to do the worst and svms tend to do the best , although the differences are nt very large . on the other hand , we were not able to achieve accuracies on the sentimentclassification problem comparable to those reported for standard topic - based categorization , despite the several different types of features we tried .
unigram presence information turned out to be the most effective ; in fact , none of the alternative features we employed provided consistently better performance once unigram presence was incorporated . interestingly , though , the superiority of presence information in comparison to frequency information in our setting contradicts previous observations made in topic - classification work @xcite . what accounts for these two differences
difficulty and types of information proving useful between topic and sentimentclassification , and how might we improve the latter ? to answer these questions , we examined the data further .
( all examples below are drawn from the full 2053-document corpus . ) as it turns out , a common phenomenon in the documents was a kind of `` thwarted expectations '' narrative , where the author sets up a deliberate contrast to earlier discussion : for example , `` this film should be brilliant .
it sounds like a great plot , the actors are first grade , and the supporting cast is good as well , and stallone is attempting to deliver a good performance .
however , it ca nt hold up '' or `` i hate the spice girls . ... [ 3 things the author hates about them ] ... why i saw this movie is a really , really , really long story , but i did , and one would think i d despise every minute of it .
but ... okay , i m really ashamed of it , but i enjoyed it .
i mean , i admit it s a really awful movie ... the ninth floor of hell ... the plot is such a mess that it s terrible .
but i loved it . '' in these examples , a human
would easily detect the true sentiment of the review , but bag - of - features classifiers would presumably find these instances difficult , since there are many words indicative of the opposite sentiment to that of the entire review .
fundamentally , it seems that some form of discourse analysis is necessary ( using more sophisticated techniques than our positional feature mentioned above ) , or at least some way of determining the focus of each sentence , so that one can decide when the author is talking about the film itself .
( makes a similar point , noting that for reviews , `` the whole is not necessarily the sum of the parts '' . )
furthermore , it seems likely that this thwarted - expectations rhetorical device will appear in many types of texts ( e.g. , editorials ) devoted to expressing an overall opinion about some topic .
hence , we believe that an important next step is the identification of features indicating whether sentences are on - topic ( which is a kind of co - reference problem ) ; we look forward to addressing this challenge in future work . we thank joshua goodman , thorsten joachims , jon kleinberg , vikas krishna , john lafferty , jussi myllymaki , phoebe sengers , richard tong , peter turney , and the anonymous reviewers for many valuable comments and helpful suggestions , and hubie chen and tony faradjian for participating in our baseline experiments .
portions of this work were done while the first author was visiting ibm almaden .
this paper is based upon work supported in part by the national science foundation under itr / im grant iis-0081334 .
any opinions , findings , and conclusions or recommendations expressed above are those of the authors and do not necessarily reflect the views of the national science foundation .
aidan finn , nicholas kushmerick , and barry smyth .
genre classification and domain transfer for information filtering . in _ proc . of theeuropean colloquium on information retrieval research _ ,
pages 353362 , glasgow . | we consider the problem of classifying documents not by topic , but by overall sentiment , e.g. , determining whether a review is positive or negative . using movie reviews as data , we find that standard machine learning techniques definitively outperform human - produced baselines
. however , the three machine learning methods we employed ( naive bayes , maximum entropy classification , and support vector machines ) do not perform as well on sentiment classification as on traditional topic - based categorization .
we conclude by examining factors that make the sentiment classification problem more challenging . |
this paper is concerned with a detailed local analysis of the solution of the cauchy initial - value problem for the sine - gordon equation @xmath0 we will consider the number @xmath1 to be a small parameter . this type of scaling can be physically motivated in the situation that the sine - gordon equation is used to model the propagation of magnetic flux along superconducting josephson junctions @xcite .
the sine - gordon equation can also be derived in the continuum limit as a model for an array of coaxial pendula with nearest - neighbor torsion coupling @xcite .
this latter application is particularly useful for the purposes of visualization of solutions .
a dramatic separation of scales occurs in the _ semiclassical limit _
@xmath2 if the initial data @xmath3 and @xmath4 are held fixed .
as can be seen in figures [ fig : exactplots4][fig : exactplots16 ] , the semiclassical dynamics apparently consists of well - defined ( asymptotically independent of @xmath5 ) spacetime regions containing oscillations on space and time scales proportional to @xmath5 but modulated over longer scales originating with the @xmath5-independent initial conditions . an important role is played by the @xmath6-parametrized curve @xmath7 in the phase portrait of the simple pendulum , and its relation to the separatrix curve @xmath8 .
indeed , in the specific context of suitable initial data of _ pure impulse _ type , that is , for which @xmath9 , the following dichotomy has recently been established @xcite regarding the asymptotic behavior of the solution @xmath10 of for small time @xmath11 independent of @xmath5 .
if @xmath12 lies inside the separatrix , then @xmath10 is accurately modeled by a modulated train of superluminal librational waves ; but if @xmath12 lies outside the separatrix , then @xmath10 is instead accurately modeled by a modulated train of superluminal rotational waves . if @xmath13 is a point lying exactly on the separatrix curve , then the approximation theorems proved in @xcite fail to provide a uniform description of the asymptotic behavior of the solution @xmath10 near such @xmath6 .
it is clear from plots of exact solutions shown in figures [ fig : exactplots4][fig : exactplots16 ] that some essentially different asymptotic behavior is generated by separatrix crossings in the pure - impulse initial data .
in particular , a different and more complicated kind of waveform than modulated traveling waves appears to spread in time away from specific points where separatrix crossings occur in the initial data .
the coupled pendulum interpretation is useful here : if the pendula are all initially at rest in the gravitationally stable configuration and are given a spatially - localized initial impulse of sufficient strength , then some pendula have sufficient energy to rotate completely about the axis a number of times , while the pendula in the `` wings '' experience very little initial impulse and only have energy for small oscillatory motions near equilibrium ( so - called librational motion ) .
clearly this situation leads to kink generation near the transition points and a more complicated type of dynamics as the kinks struggle to separate from one another .
( left ) and @xmath14 ( right ) for the exact solution of the cauchy problem with @xmath15 for @xmath9 and @xmath16 . for this data ,
separatrix crossings occur for @xmath17 . _
_ ] but with @xmath18 .
_ _ ] but with @xmath19 .
_ _ ] to better understand the reason for the locally complicated behavior , it is useful to view the sine - gordon equation as a perturbation of the simple pendulum ode : @xmath20 we might expect that the right - hand side should be negligible for quite a long rescaled time @xmath21 as long as in the unperturbed system ( @xmath22 ) pendula located at nearby values of @xmath6 follow nearby orbits of the pendulum phase portrait
. this will be the case unless @xmath23 , the condition allowing for nearby values of @xmath6 to correspond to topologically dissimilar orbits , leading to large relative displacements in @xmath24 over finite @xmath21 and causing @xmath25 to become very large very quickly .
see figure [ fig : sepplot ] . and
@xmath26 with @xmath27 plotted parametrically in the @xmath28-plane ( blue ) .
orbits of the simple pendulum are shown with gray curves .
orbits corresponding to nearby values of @xmath6 ( red curves ) can diverge after a descent to a neighborhood of a saddle point in finite @xmath21 , if @xmath29 . _ _ ] to leading order , and for times @xmath11 of order @xmath5 we expect that orbits near @xmath30 should follow the pendulum separatrix : @xmath31 , @xmath32 , and @xmath33 .
of course as @xmath34 , the perturbation term @xmath35 will become important and will prevent the rapid divergence of trajectories shown in figure [ fig : sepplot ] . the subject of this paper is the asymptotic analysis , in the semiclassical limit @xmath2 , of the solution of the cauchy problem for the sine - gordon equation in the case that @xmath11 is small and @xmath6 is near a value where the curve @xmath12 crosses the separatrix ( at its midpoint , as in figure [ fig : sepplot ] ) .
this region of the @xmath36-plane is blown up with what turn out to be the correct scalings for easier viewing in figure [ fig : criticalzoom ] .
-plane in which the fluxon condensate is analyzed . _
_ ] therefore we are near the boundary of the regions where the solution is described in terms of modulated rotational and librational solutions of the pendulum equation .
we will show that , under suitable further assumptions , the asymptotic behavior of @xmath10 in this situation is universal , and is described by an essentially multiscale formula that is expressible in terms of modern special functions , specifically solutions of certain nonlinear ordinary differential equations of painlev type . to formulate our results precisely requires that we first set up some background material ; we hope that the reader will bear with us until [ sec : results ] where the full details will be presented . in the meantime , we can describe the semiclassical asymptotics of @xmath10 in the region illustrated in figure [ fig : criticalzoom ] by saying that this region contains a curvilinear network of isolated kink - type solutions of the sine - gordon equation ( with the approximation @xmath37 holding in between the kinks ) lying along the graphs of certain rational solutions of inhomogeneous painlev - ii equations . at spacetime points associated with poles and zeros of these solutions ,
the kinks collide in a grazing fashion as can be modeled by a more complicated exact solution of the sine - gordon equation corresponding to a double soliton .
the phenomenon of an asymptotically universal wave pattern being described by simple `` solitonic '' solutions located in space and time according to a rather more transcendental solution of a painlev - type equation was only recently observed for the first time in another problem by bertola and tovbis @xcite .
all power functions @xmath38 will be assumed to be defined for nonintegral real @xmath39 as the principal branch with branch cut @xmath40 and with @xmath41 .
if @xmath4 is a function of @xmath6 for which the equation @xmath42 has a unique solution for @xmath43 , then we denote this solution by @xmath44 .
let us use the term _ criticality _ to describe the point @xmath44 and @xmath45 , and _ near criticality _ to mean @xmath46 and @xmath47 .
for any rational function @xmath48 , let @xmath49 denote the finite set of real poles of @xmath50 and let @xmath51 denote the finite set of real zeros of @xmath50 . finally , for a finite set @xmath52 and a real number @xmath53 denote by @xmath54 the distance between @xmath55 and @xmath56 .
if @xmath57 is an oriented contour in the complex plane and @xmath58 is a function analytic in the complement of @xmath57 , we will use subscripts @xmath59 and @xmath60 to refer to the boundary values taken by @xmath61 as @xmath62 nontangentially from the left and right respectively .
we will make frequent use of the pauli matrices : @xmath63 with the sole exception of these three , we write all matrices with boldface capital letters .
we use the landau notation for most estimates , with `` big - oh '' written @xmath64 and `` little - oh '' written @xmath65 .
also , if @xmath66 are some quantities , we will use the shorthand notation @xmath67 to represent a quantity that is bounded by a linear combination of @xmath68 , that is , @xmath69 .
we will be dealing with several matrix functions in which the matrix symbol carries subscripts and superscripts , so we will use a special notation for taylor / laurent expansion coefficients of such matrix functions : if @xmath70 is such a matrix function and @xmath71 is a point about which this matrix is to be expanded , we write the taylor expansion in the form @xmath72({\tau}-{\tau}_0)^k.\ ] ] we also use this notation in the case that @xmath73 with obvious modifications .
we study the cauchy initial - value problem under exactly the same assumptions used in our earlier work @xcite : the initial conditions for satisfy @xmath9 .
[ ass : impulse ] assumption [ ass : impulse ] asserts that the initial data is of pure - impulse type .
this is important because it implies that the direct scattering problem for the solution of the cauchy problem by the inverse - scattering transform reduces from the faddeev - takhtajan eigenvalue problem to the better - understood nonselfadjoint zakharov - shabat eigenvalue problem .
the function @xmath74 is a nonpositive function of klaus - shaw type , that is , @xmath75 and @xmath74 has a unique local ( and global ) minimum .
[ ass : klausshaw ] as was shown by klaus and shaw in @xcite , assumption [ ass : klausshaw ] provides a useful and important confinement property of the discrete spectrum of the nonselfadjoint zakharov - shabat eigenvalue problem associated to the potential @xmath74 .
this allows us to formulate riemann - hilbert problem [ rhp : basic ] below on a system of contours very close to the unit circle and a transecting negative real interval , a set whose image under the function @xmath76 defined in below is a segment of the imaginary axis .
the function @xmath74 is even in @xmath6 : @xmath77 , placing the unique minimum of @xmath74 at @xmath78 .
[ ass : even ] assumption [ ass : even ] is admittedly less important ( we believe that with appropriate modifications our results all hold true without it ) but it allows for a substantial simplification of our analysis .
we point out that under assumptions [ ass : klausshaw ] and [ ass : even ] the function @xmath74 restricted to @xmath79 has a well - defined inverse function @xmath80 .
the function @xmath74 is strictly increasing and real - analytic at each @xmath43 , and the positive and real - analytic function @xmath81 can be analytically continued to neighborhoods of @xmath82 and @xmath83 with @xmath84 and @xmath85 .
[ ass : scriptgnice ] the analyticity of @xmath74 for @xmath43 and that of @xmath86 up to the endpoints of the interval @xmath87 $ ] as guaranteed by assumption [ ass : scriptgnice ] are both absolutely crucial to our method of analysis .
it is the analyticity of @xmath74 that implies that of @xmath88 defined in below , and hence of @xmath89 defined by and , and these are used to convert a `` primordial '' riemann - hilbert problem of inverse scattering that involves a large number ( inversely proportional to @xmath1 ) of pole singularities into the simpler riemann - hilbert problem [ rhp : basic ] to be formulated below .
the latter problem has no poles , but only jump discontinuities along contours , and hence is amenable to the deift - zhou steepest descent technique @xcite of rigorous asymptotic analysis .
the small number @xmath5 lies in the infinite sequence @xmath90 [ ass : epsilon ] assumption [ ass : epsilon ] is important because it minimizes the effect of _ spectral singularities _ , events occurring infinitely often as @xmath2 at which discrete eigenvalues are born from the continuous spectrum . when spectral singularities occur , the reflection coefficient has poles in the continuous spectrum , and without assumption [ ass : epsilon ] ( or some suitable approximation thereof ) the reflection coefficient can not be neglected uniformly on the continuous spectrum .
assumption [ ass : epsilon ] is important because our approach is based on neglecting the reflection coefficient entirely .
( it can be shown to be small in the semiclassical limit without assumption [ ass : epsilon ] except near points where spectral singularities can occur . )
the function @xmath91 satisfies @xmath27 .
[ ass : kinks ] it is this last assumption that ensures that there exist exactly two points @xmath92 at which the initial data curve @xmath12 crosses the separatrix of the simple pendulum phase portrait .
therefore , assumption [ ass : kinks ] guarantees that the phenomenon we wish to study in this paper actually occurs for the initial data in question .
given initial data for the cauchy problem satisfying these assumptions , we construct a sequence of exact solutions @xmath93 of the sine - gordon equation @xmath94 for @xmath95 , @xmath96 called the _ fluxon condensate _ associated with the given initial data .
while @xmath97 is an exact solution of the partial differential equation for each @xmath98 , it generally does not satisfy exactly the given initial conditions .
however , it has been proved ( * ? ? ?
* corollary 1.1 ) that @xmath99 holds modulo @xmath100 and @xmath101 where the error estimates are valid pointwise for @xmath102 and @xmath103 , and uniformly on compact subsets of the set of pointwise accuracy .
we strengthen this convergence to include the points @xmath6 near where @xmath104 in theorem [ thm : tzero ] below .
the fluxon condensate is constructed as follows .
first one defines a function @xmath105 by setting @xmath106 note that @xmath88 is a strictly decreasing function of @xmath107 where defined .
this fact allows us to define a sequence of numbers @xmath108 by solving the equation @xmath109 ( these positive imaginary numbers are approximate eigenvalues for the zakharov - shabat eigenvalue problem associated with the klaus - shaw potential @xmath74 , and is a kind of _ bohr - sommerfeld quantization rule _ for that nonselfadjoint problem . )
setting @xmath110\quad\text{and}\quad d(w):=\frac{i}{4}\left[(-w)^{1/2}-(-w)^{-1/2}\right],\quad |\arg(-w)|<\pi , \label{eq : ed}\ ] ] we define @xmath111 and @xmath112 this function is meromorphic and has no zeros where defined .
it has @xmath113 poles ( counted with multiplicity ) , the set of which we denote as @xmath114 . generically
( with respect to deformations of @xmath74 ) the poles are all simple , and there are @xmath115 of them in complex - conjugate pairs on the unit circle in the @xmath116-plane , and @xmath117 of them in pairs on the negative real axis in involution with respect to the map @xmath118 . as @xmath119 ( or @xmath120 ) the poles accumulate on the whole unit circle and the interval @xmath121 $ ] , where @xmath122 clearly , both @xmath123 and @xmath124 are independent of @xmath98 , and @xmath125 . were assumption [ ass : kinks ] not satisfied
, we would have @xmath126 , and the poles of @xmath127 would accumulate only in a complex - conjugate symmetric incomplete arc of the unit circle containing the point @xmath128 . in nongeneric cases it can happen that there is a double pole at @xmath129 .
we consider the accumulation locus of poles , @xmath130\cup s^1 $ ] to be an oriented contour , with orientation of the two intervals @xmath131 and @xmath132 both toward @xmath129 , and with orientation of the upper and lower semicircles from @xmath129 toward @xmath128 .
see figure [ fig : critncontour ] .
we also use the notation @xmath133 then , define a function @xmath134 by a cauchy integral : @xmath135 and then set @xmath136 finally , writing @xmath137 a function that turns out to have a well - defined analytic continuation to a full neighborhood of the self - intersection point of @xmath129 of the contour @xmath138 ( upon continuation from any of the four intersecting arcs ) , we set @xmath139 it can be shown ( * ? ? ?
* proposition 3.1 ) that @xmath140 holds uniformly on compact subsets of the open domain of definition , and that @xmath141 holds if @xmath142 or @xmath143 nontangentially to the real axis .
similarly , @xmath144 extends from @xmath138 to an analytic function @xmath145 on a neighborhood of @xmath129 , and @xmath146 holds uniformly on this neighborhood as well as on @xmath138 , as long as @xmath116 is bounded away from @xmath147 and @xmath148 ; however the estimate @xmath149 holds uniformly for @xmath150 .
consider the contour @xmath151 illustrated in figure [ fig : critncontour ] . of discontinuity for the sectionally analytic function @xmath152 . with the exception of the two components of @xmath153 which are oriented left - to - right
, the contour arcs are oriented with the regions labeled `` @xmath154 '' on the left , and with the regions labeled `` @xmath155 '' on the right .
the image of @xmath156 under @xmath157 with @xmath158 defined by consists of the four straight line segments : ( i ) @xmath159 , @xmath160 , ( ii ) @xmath161 , @xmath160 , ( iii ) @xmath162 , @xmath160 , and ( iv ) @xmath163 , @xmath164 . here
@xmath165 is a sufficiently small number .
the self - intersection point @xmath129 is the critical point of @xmath158 . _ _ ] let @xmath166 denote a function analytic in the domain @xmath167 , satisfying the symmetry @xmath168 and the conditions that @xmath169 and @xmath170 as @xmath171 .
we assume also that @xmath172 takes well - defined boundary values on @xmath173 in the classical sense ( hlder continuity up to the boundary ) , and that the boundary values satisfy @xmath174 for @xmath175 .
set @xmath176 then , to determine the fluxon condensate , one solves the following riemann - hilbert problem .
given @xmath172 as above , find a @xmath177 matrix function @xmath178 of the complex variable @xmath116 with the following properties : * * analyticity : * @xmath152 is analytic for @xmath179 . *
* jump condition : * @xmath152 is hlder continuous up to @xmath151 .
the boundary values @xmath180 taken by @xmath152 on the various arcs of @xmath151 are required to satisfy the following jump conditions @xmath181 @xmath182 where @xmath183 , @xmath184 and , @xmath185 * * normalization : * the following normalization condition holds : @xmath186 where the limit is uniform with respect to angle for @xmath187 .
[ rhp : basic ] it turns out that for each choice of the function @xmath172 , each @xmath96 , and each @xmath188 there exists a unique solution of this riemann - hilbert problem .
moreover , the product @xmath189 does not depend on the choice of the function @xmath172 .
the solution @xmath190 has convergent expansions near @xmath191 and @xmath192 of the form @xmath193(x , t)(-w)^{-k/2},\quad \text{for $ |w|$ sufficiently large , } \label{eq : nexpandwlarge}\ ] ] and @xmath194(x , t)(-w)^{k/2},\quad \text{for $ |w|$ sufficiently small . }
\label{eq : nexpandwsmall}\ ] ] given a function @xmath74 satisfying assumptions [ ass : impulse ] through [ ass : kinks ] , the fluxon condensate associated with @xmath74 is the family of functions @xmath195 defined modulo @xmath100 by the equations @xmath196(x , t)\right]_{11 } \quad\text{and}\quad \sin(\tfrac{1}{2}u_n(x , t)):=\left[[\fouridx{0}{0}{g}{n}{\mathbf{n}}](x , t)\right]_{21}. \label{eq : fluxoncondensate}\ ] ] note that these are independent of @xmath172 because @xmath169 .
[ def : fluxoncondensate ] while it is possible to extract formulae for derivatives of @xmath97 with respect to @xmath6 and @xmath11 by the chain rule , it is preferable to have formulae that do not require differentiation of @xmath190 , as this will require control of derivatives of error terms .
however , it can be shown also that the following formula holds : @xmath197(x , t)\right]_{12 } + \left[[\fouridx{0}{0}{g}{n}{\mathbf{n}}](x , t)^{-1}[\fouridx{0}{1}{g}{n}{\mathbf{n}}](x , t)\right]_{12 } , \label{eq : epsunt}\ ] ] and this does not require differentiation of @xmath190 with respect to @xmath11 ( it also turns out to be independent of choice of @xmath172 ) .
each function @xmath97 of the fluxon condensate is an exact solution of the sine - gordon equation with @xmath95 , and @xmath198 .
while riemann - hilbert problem [ rhp : basic ] is the most convenient starting point for our local analysis near criticality , it is not the most fundamental representation of @xmath97 .
indeed , @xmath97 is a _ reflectionless potential _ for the direct scattering problem of the lax pair for the sine - gordon equation , and this means that it can be obtained from a purely `` discrete '' riemann - hilbert problem whose solution is a rational function on the riemann surface of @xmath199 , with poles on both sheets over the points of @xmath114 .
since the number of poles is increasing with @xmath98 , some preparations are required to recast the problem in a suitable form for addressing the limit @xmath200 .
these preparations are detailed in @xcite , and they take the form of a sequence of explicit transformations resulting in the equivalent riemann - hilbert problem [ rhp : basic ] . in general , a number of choices are made along the way because the transformations that are required to enable the subsequent asymptotic analysis depend on @xmath36 ; however for @xmath201 and @xmath11 small only the simplest of the choices detailed in @xcite are required . for readers familiar with the terminology of that paper
, we are assuming that @xmath202 , and correspondingly , @xmath203 is the function called @xmath204 in @xcite while @xmath145 is the function called @xmath205 in @xcite . also ,
for our purposes we need make no distinction between the contour called @xmath206 ( which would just be @xmath207 in the case that @xmath202 ) and the contour @xmath138 , and this implies that the set @xmath208 that lies between these two contours is empty . finally , to derive riemann - hilbert problem [ rhp : basic ] from the results of @xcite we used the facts that for @xmath209 , @xmath210 .
the solution of the sine - gordon equation by inverse - scattering methods is a subject with a long history .
when the sine - gordon equation is written in terms of characteristic coordinates @xmath211 it fits naturally into the hierarchy of the ablowitz - kaup - newell - segur or zakharov - shabat scattering problem .
characteristic _ cauchy problem for the sine - gordon equation was integrated in this framework by ablowitz , kaup , newell , and segur @xcite and by zakharov , takhtajan , and faddeev @xcite .
the more physically - relevant problem of the cauchy problem in laboratory coordinates as posed in required new methodology , and the solution of this cauchy problem by the inverse scattering method was first outlined by kaup @xcite .
an account of the solution of the cauchy problem in laboratory coordinates is given in the textbook of faddeev and takhtajan @xcite .
some further analytical details needed to make the theory completely rigorous were supplied by zhou @xcite and later by cheng , venakides , and zhou @xcite . a self - contained account of the riemann - hilbert formulation of the inverse - scattering solution of the cauchy problem assuming only that at each instant of time the solution has @xmath212-sobolev regularity can be found in our paper ( * ? ? ?
* appendix a ) , and a direct proof that the sine - gordon equation preserves this degree of regularity if it is initially present is given in ( * ? ? ?
* appendix b ) .
in our recent paper @xcite we found that to describe the semiclassical limit for the cauchy problem it is useful to reformulate the riemann - hilbert problem in the complex plane of a square root of the spectral variable used in ( * ? ? ?
* appendix a ) ; this makes it easier to express the asymptotic solutions in terms of riemann theta functions of the lowest possible genus . as we view the current paper as a continuation of our work in @xcite we use the same formulation here .
let @xmath213 be defined by @xmath214 and define a positive constant @xmath215 by @xmath216 also , set @xmath217 all of our results concern the asymptotic behavior of the fluxon condensate @xmath97 in the small region near criticality where @xmath218 and @xmath219 as shown in figure [ fig : criticalzoom ] .
our first result is concerned with the relevance of the fluxon condensate associated with @xmath91 to the cauchy initial - value problem .
suppose that @xmath220 .
then uniformly for such @xmath6 , @xmath221 [ thm : tzero ] this result extends that of ( * ? ? ?
* corollary 1.1 ) to suitable neighborhoods of the points @xmath92 .
there exist multiscale approximating functions @xmath222 and @xmath223 ( defined in detail in [ sec : proofmain ] , and depending on initial data only through the constant @xmath224 ) such that @xmath225 with the error terms being uniform for @xmath226 and @xmath227 .
[ thm : main ] in fact , we will really show that the error term is significantly smaller , namely @xmath228 , over most of the small region of the @xmath36-plane where the above theorem provides an asymptotic description of the dispersive breakup of the pendulum separatrix .
now , the multiscale model provided by theorem [ thm : main ] serves to establish universality of the behavior near the critical point , but the formulae for @xmath222 and @xmath223 are rather complicated , so it is useful to give some more detailed information by focusing on smaller parts of the @xmath36-plane near criticality . to render our results in a more elementary fashion , we need to first define a certain hierarchy of rational functions
first define @xmath229 then define @xmath230 by the forward recursion @xmath231 and the backward recursion @xmath232 up to constant factors , @xmath233 and @xmath234 are ratios of consecutive _ yablonskii - vorobev _ polynomials ( see @xcite ) and logarithmic derivatives of @xmath233 and @xmath234 are the unique @xcite rational solutions of the inhomogeneous painlev - ii equations .
these observations are not necessary for us to state our results , and we will make further comments later at an appropriate point in the paper .
the following proposition characterizes the behavior of these rational functions for large @xmath235 . for each @xmath236
, @xmath237 as @xmath238 . in particular , @xmath239 for sufficiently large negative @xmath235 while @xmath240 for sufficiently large positive @xmath235 .
[ prop : unvnasympy ] it is obvious that the recursions and preserve rationality of the input functions @xmath241 , so for each @xmath236 , @xmath242 is a rational function of @xmath235 .
therefore , @xmath242 has a laurent expansion for large @xmath243 of the form @xmath244 where @xmath245 is a complex constant and @xmath246 is an exponent to be determined , and moreover this expansion is differentiable any number of times with respect to @xmath235 . in particular , it follows that @xmath247 as @xmath238 .
this shows that the final two terms on the right - hand side of the formula for @xmath248 given in are subdominant compared to the term @xmath249 , and therefore , @xmath250 this formula gives the recurrence relation for the large-@xmath235 asymptotics of @xmath242 . the asymptotic formula for @xmath233 given in follows by solving the recurrence with the base case @xmath251 , and that for @xmath234 follows from the identity @xmath252 . fix an integer @xmath253 , and suppose that @xmath219 while @xmath254 .
then @xmath255 where the phase is @xmath256 and a rescaled spatial coordinate is given by @xmath257 the error terms satisfy @xmath258 unless ( i ) both @xmath259 and @xmath260 or ( ii ) both @xmath261 and @xmath262 .
moreover , given any interval @xmath263 $ ] on which @xmath264 is a bounded function , we have @xmath265 holding uniformly for @xmath266 $ ] and @xmath254 .
[ thm : kinkapprox ] wherever the error terms vanish in the limit , it follows that @xmath267 and that @xmath268 , where @xmath269 .
if we define a function @xmath270 modulo @xmath271 by @xmath272 then it is easy to check that @xmath270 is an @xmath273-independent solution of the unscaled sine - gordon equation @xmath274 defined by with @xmath275 . for a superluminal kink ( @xmath276 ) , the plots are simply reflected through the horizontal @xmath273-axis . _ ] this exact solution represents a superluminal ( infinite velocity ) kink with unit magnitude _
topological charge _ @xmath277 ; if @xmath275 ( respectively @xmath276 ) then @xmath24 decreases ( respectively increases ) by @xmath271 as @xmath21 varies from @xmath278 to @xmath279 .
sometimes one refers to @xmath270 as a kink in the case @xmath276 and as an _ antikink _ in the case @xmath275 .
another way to describe @xmath270 is to say that it is a solution of the simple pendulum equation @xmath280 that is homoclinic to the unstable equilibrium of a stationary inverted pendulum .
see figure [ fig : kinkpicture ] .
the dominant part of the phase variable @xmath281 defined by is indeed a recentering and rescaling by @xmath282 of @xmath11 ; however @xmath281 also contains weak @xmath6-dependence through the function @xmath283 where @xmath235 is proportional to @xmath284 .
therefore , the approximation of @xmath97 described in theorem [ thm : kinkapprox ] is not an exact kink , but rather is one that is slowly modulated in the direction parallel to the wavefront .
indeed , the center of the approximating kink ( where the pendulum angle is zero ) corresponds to @xmath285 , which is a curve in the @xmath36-plane that is a scaled and translated version of the graph of the function @xmath286 .
see figure [ fig : kinkcurves ] .
an additional important observation is that the period of each approximating kink is proportional to @xmath282 , but the strip in which it lives ( indexed by the integer @xmath253 ) corresponds to a time interval of length proportional to @xmath287 ; hence the kinks are in reality widely separated from each other , and therefore in `` most '' of the domain @xmath288 and @xmath227 covered by theorem [ thm : kinkapprox ] it is fair to say that the approximate formula @xmath289 holds , which in the context of the coupled pendulum interpretation of sine - gordon means that `` most '' of the pendula are approximately in the unstable inverted configuration .
-plane in each of which theorem [ thm : kinkapprox ] provides asymptotics for @xmath97 , shown with alternating gray and white shading , with the center @xmath285 of each corresponding superluminal kink shown with a superimposed curve .
left : @xmath290 .
right : @xmath291 . in both cases @xmath292 . for different values of @xmath224
, the curves can `` march out '' of the confining strips as @xmath293 increases , unless @xmath282 is small enough.__,title="fig:",height=1]-plane in each of which theorem [ thm : kinkapprox ] provides asymptotics for @xmath97 , shown with alternating gray and white shading , with the center @xmath285 of each corresponding superluminal kink shown with a superimposed curve .
left : @xmath290 .
right : @xmath291 . in both cases @xmath292 . for different values of @xmath224 , the curves can `` march out '' of the confining strips as @xmath293 increases , unless @xmath282 is small enough.__,title="fig:",height=1 ] we may now formulate some observations about the kink centered at @xmath285 that follow from proposition [ prop : unvnasympy ] . indeed , for each @xmath236 the following is true : for @xmath235 sufficiently negative , @xmath294 , while for @xmath235 sufficiently positive , @xmath295 .
this then implies that as @xmath296 we have a series of timelike antikinks ( all of the same topological charge ) consistent with nearly synchronous rotational motion of pendula .
similarly , as @xmath297 we have alternation between kink and antikink with no net topological charge consistent with nearly synchronous librational motion of pendula .
these facts demonstrate that the asymptotic behavior near criticality matches appropriately with the established asymptotic formulae @xcite valid away from criticality that model @xmath97 as a modulated train of superluminal rotational traveling waves to the left of @xmath213 and as a modulated train of superluminal librational traveling waves to the right of @xmath213 .
it also follows from proposition [ prop : unvnasympy ] that if @xmath298 then @xmath233 blows up as @xmath299 while if @xmath300 then @xmath301 as @xmath302 .
this implies that for @xmath298 the corresponding kinks follow a logarithmic `` frown '' for large @xmath303 , while for @xmath300 they instead follow a logarithmic `` smile '' for large @xmath303 .
these features can be seen in figure [ fig : kinkcurves ] .
when the horizontal strips of the @xmath36-plane corresponding to different integral values of @xmath304 are put together , one sees that the kink - like asymptotics given by theorem [ thm : kinkapprox ] are valid throughout the region where @xmath219 and @xmath218 with the sole exception of small sub - regions near the top and bottom edges of each strip where the curve @xmath285 tries to exit the strip .
when the strips are put together there are obvious mismatches of the curves from neighboring strips in these small sub - regions as can be seen from the left - hand plot in figure [ fig : kinkcurves ] .
our final result corrects the kink asymptotics in these regions and therefore removes the mismatches .
fix an integer @xmath253 and let @xmath305 denote any of the ( necessarily simple ) real zeros of @xmath306 ( equivalently a simple zero of @xmath307 ) . given a sufficiently small positive number @xmath308 ,
suppose that @xmath309 , and that @xmath310 , where @xmath235 is defined by .
then @xmath311 where @xmath312 the error terms satisfy @xmath313 whenever @xmath314 as @xmath120 ( which excludes only the four extreme corners of the hourglass - shaped region under consideration ) .
moreover , we have @xmath315 holding uniformly for @xmath316 and @xmath317 .
[ thm : grazingapprox ] wherever the error terms vanish in the limit , it follows from that @xmath318 and @xmath319 , where with @xmath320 , a function @xmath321 is defined modulo @xmath271 by @xmath322 then it is again easy to confirm that @xmath321 is an exact solution of the unscaled sine - gordon equation .
plots of @xmath323 and @xmath324 are shown in figure [ fig : grazingcollision ] . defined by with @xmath325 . for @xmath326 the plots
are simply reflected through the vertical @xmath21-axis . _ ] this is a particular solution of the sine - gordon equation that corresponds to boundary conditions of the form @xmath327 as @xmath328 . in the proper version of scattering theory corresponding to these boundary conditions , the solution at hand is a reflectionless potential associated to a _
double eigenvalue_. in the context of the riemann - hilbert problem of inverse scattering such an object is encoded as a double pole of the matrix unknown .
such solutions were noted in the earliest days of inverse scattering by zakharov and shabat @xcite in their study of the focusing nonlinear schrdinger ( nls ) equation .
they pointed out that such solutions describe _ grazing collisions _ of solitons .
the trajectories of the solitons emerging from the interaction region are not asymptotically straight lines , but rather are logarithmic , with relative velocities of the solitons tending to zero . in the context of the sine - gordon equation
we have instead a grazing collision of superluminal kinks .
these structures serve to smooth out the mismatches of curves shown in figure [ fig : kinkcurves ] at horizontal strip boundaries . in the language of
matched asymptotic expansions , they function as internal transition layers of `` corner '' type .
the regions of validity of the asymptotic formulae given in theorems [ thm : kinkapprox ] and [ thm : grazingapprox ] are compared in figure [ fig : regionsschematic ] , which illustrates the complementary nature of these two results .
provides asymptotics in horizontal strips in the @xmath329-plane with the exception of the small rectangles shown in green , and with optimal accuracy of @xmath228 in the subregions shown in blue .
theorem [ thm : grazingapprox ] provides asymptotics in hourglass - shaped regions with the exception of where they overlap the blue regions ( where the error estimate of theorem [ thm : kinkapprox ] is optimally small ) and with optimal accuracy of @xmath228 in the included green regions ( where the error estimate of theorem [ thm : kinkapprox ] fails to be small ) and brown regions . _ _ ] it is clear from this figure that theorems [ thm : kinkapprox ] and [ thm : grazingapprox ] both provide simple asymptotic formulae for the same fields in overlapping regions . in the overlap domains , formulae of kink type provided by theorem [ thm : kinkapprox ] , and of grazing kink collision type provided by theorem [ thm : grazingapprox ]
are _ simultaneously valid_. this is actually a consequence of theorem [ thm : main ] , but it may also be checked directly , by writing both types of formulae in terms of common spatiotemporal independent variables . one way to characterize the asymptotic formulae giving the universal form of the dispersive breakup of the simple pendulum separatrix under the sine - gordon equation is to say that the wave pattern consists of waves of elementary `` solitonic '' forms that are spatiotemporally arranged according to solutions of certain nonlinear differential equations of painlev type . to our knowledge , this type of phenomenon was first observed quite recently in a paper by bertola and tovbis @xcite in which all of the local maxima of the modulus of the solution to the focusing nls equation in the semiclassical limit near the onset of oscillatory behavior ( the point of so - called elliptic umbilic catastrophe of the approximating elliptic quasilinear whitham modulational system ) are individually modeled by the same exact solution of the focusing nls equation ( the rational or peregrine breather solution ) but the locations of the maxima are far more transcendental , being determined by the location of poles in the complex plane of the so - called _ tritronque _ solution of the painlev - i equation . in the current context , the role of the tritronque solution is played by the rational functions @xmath330 solving the painlev - ii system , and interestingly , _ we require an infinite number of different painlev functions to describe the full wave pattern_.
the use of a so - called `` @xmath172-function '' to precondition a matrix riemann - hilbert problem for subsequent asymptotic analysis by the deift - zhou steepest - descent method is by now a standard tool , having first been introduced by deift , venakides , and zhou @xcite in their analysis of the korteweg - de vries ( kdv ) equation in the small - dispersion limit . recall that the scalar function @xmath172 appearing in the formulation of riemann - hilbert problem [ rhp : basic ] is subject to the basic requirements of * analyticity : @xmath172 is analytic in the domain @xmath331 , * boundary behavior : @xmath172 takes well - defined boundary values on @xmath173 in the classical sense , and the sum of the boundary values vanishes on @xmath79 , and , * normalization : @xmath332 .
we can use any such function to formulate the basic riemann - hilbert problem for @xmath333 , but it is well - known that to study the limit @xmath200 it is important that @xmath172 be chosen appropriately . in @xcite
it is explained how @xmath172 should be chosen given @xmath188 to calculate the limit @xmath200 with @xmath36 fixed .
the chosen function @xmath334 has the property that its boundary values satisfy a kind of equilibrium condition ( an analogue of below ) on a contour ( called @xmath335 in @xcite ) whose topology changes at criticality : in the simplest ( genus one ) case , near @xmath336 the equilibrium contour is either a nearly circular arc through @xmath128 with complex - conjugate endpoints near @xmath129 ( this corresponds to modulated superluminal librational waves ) or the union of a nearly circular closed contour with a transecting small interval of the real axis with both endpoints near @xmath129 ( this corresponds to modulated superluminal rotational waves ) . exactly at criticality
the two endpoints coalesce at @xmath129 and the equilibrium contour becomes the unit circle .
it turns out that proximity of endpoints of this contour is an obstruction to proving uniform asymptotics for @xmath36 near , but not exactly at , criticality . here ,
since we want to allow @xmath36 to approach the critical point @xmath337 at some rate depending on @xmath98 , we need to use a different function than @xmath334 defined in @xcite .
we will replace @xmath166 with @xmath338 whose construction and properties are described below .
it is important that while for general @xmath6 and @xmath11 , the functions @xmath166 and @xmath339 are different , at criticality they coincide ( they satisfy the same conditions , which can be shown to uniquely determine the solution ) .
the function @xmath339 is required to satisfy the following conditions : * @xmath339 is analytic if @xmath340 and @xmath341 , and it takes continuous and bounded boundary values on the unit circle @xmath342 and on the positive half - line @xmath79 .
* @xmath339 obeys the following jump conditions : @xmath343 @xmath344 * @xmath339 has the following values : @xmath345 @xmath346 to obtain a formula for @xmath339 , we first write @xmath347 in the form @xmath348 where the function @xmath349 is defined by . therefore the above conditions on @xmath347 are translated into equivalent conditions on @xmath350 : * @xmath351 is analytic if @xmath340 , @xmath341 , and @xmath352 $ ] , and it takes continuous boundary values on the unit circle @xmath342 , on the positive half - line @xmath79 , and on the interval @xmath121 $ ] .
* @xmath351 obeys the following jump conditions : @xmath353 @xmath354 @xmath355.\ ] ] in the latter condition , the two complementary segments @xmath131 and @xmath132 are taken to be oppositely oriented toward @xmath356 . * @xmath351 has the following values : @xmath357 @xmath358 finally , we write @xmath351 in terms of a new equivalent unknown @xmath359 by the piecewise substitution : @xmath360 the conditions satisfied by @xmath359 are then the following : * @xmath359 is analytic if @xmath340 and @xmath352 $ ] , and it takes continuous boundary values on the unit circle @xmath342 and the interval @xmath121 $ ] . *
@xmath359 obeys the following jump conditions : @xmath361 where the orientation of the circle is in the positive ( counterclockwise ) sense , and @xmath362,\ ] ] where now the interval @xmath121 $ ] is taken to be oriented left - to - right .
note that @xmath363 for @xmath364 , and that the right - hand side of is a single - valued function on the circle : @xmath365 * @xmath359 is uniformly bounded , and has the following value : @xmath366 we may now express @xmath359 in terms of a cauchy integral using the plemelj formula : @xmath367 where the constant @xmath368 is the most general entire function we can add to the cauchy integrals consistent with the uniform boundedness of @xmath359 .
the integral over the positively - oriented unit circle may be evaluated in closed form : @xmath369 we choose the value of the constant @xmath368 to enforce the condition : @xmath370 it follows that @xmath371 by writing @xmath347 in terms of @xmath350 ( by ) and then writing @xmath350 in terms of @xmath372 ( by ) , this completes the construction of @xmath339 .
according to , with @xmath373 and with @xmath338 defined as in [ sec : frakg ] , a condition of `` equilibrium type '' holds on the unit circle @xmath342 , regardless of the values of @xmath6 and @xmath11 .
the next step is to `` open a lens '' about the whole circle .
the lens consists of four disjoint open sets separated by the unit circle and the interval @xmath374 as shown with shading in figure [ fig : critocontour ] .
the union of the two components lying on the left ( respectively right ) of the unit circle with its orientation will be denoted @xmath375 ( respectively @xmath376 ) .
we now define a new unknown @xmath377 equivalent to @xmath152 by the substitution @xmath378/\epsilon_n}\\ 0 & 1\end{bmatrix},&\quad w\in\lambda^\pm \\\\
\mathbf{n}(w),&\quad w\in\mathbb{c}\setminus(\sigma_{\mathbf{n}}\cup\lambda^+\cup\lambda^- ) .
\end{cases } \label{eq : ointermsofn}\ ] ] here , the square root @xmath379 is well - defined as the principal branch , since the uniform approximation @xmath380 holds when @xmath98 is large .
( the definition coincides with that of @xmath377 in terms of @xmath152 given in @xcite in the case that @xmath202 and with @xmath172 replaced by @xmath347 . )
the jump contour for @xmath377 is the same as that for @xmath152 but augmented by four arcs emanating from @xmath129 representing the `` outer '' boundaries of the lens halves @xmath381 as shown in figure [ fig : critocontour ] .
, with the lens halves @xmath381 indicated with shading . arcs of the jump contour also present in the jump contour for @xmath152 ( see figure [ fig : critncontour ] ) retain that orientation , and the lens boundaries are oriented as indicated . _ _ ] it is a consequence of the equilibrium condition satisfied by the boundary values taken by @xmath339 on the unit circle that @xmath377 satisfies the piecewise - constant jump conditions @xmath382 since the equilibrium contour is a closed curve , we may remove these discontinuities from the problem by making another explicit transformation : @xmath383 unlike @xmath377 , the matrix function @xmath384 extends continuously to the unit circle , but the jump conditions it satisfies within the unit disk where it differs from @xmath377 are altered somewhat from those of @xmath377 , including a new jump of the simple form @xmath385 on the segment @xmath386 .
the contour of discontinuity for @xmath384 is illustrated in figure [ fig : critpcontour ] . .
contour arcs also contained in the jump contour for @xmath377 ( see figures [ fig : critncontour ] and [ fig : critocontour ] ) retain that orientation , and otherwise ( that is , in the interval @xmath387 ) the orientation is as indicated . _ _ ] the jump discontinuities of @xmath384 will turn out to be negligible except in a small neighborhood of @xmath129 and along the ray @xmath388 , and @xmath384 is a matrix tending to the identity as @xmath171 .
these considerations lead us to propose a model riemann - hilbert problem whose solution we expect to approximate @xmath384 away from @xmath129 : let @xmath389 be a real parameter . find a @xmath177 matrix @xmath390 with the following properties . *
* analyticity : * @xmath390 is an analytic function of @xmath116 for @xmath391 , hlder continuous up to the jump interval @xmath392 with the exception of an arbitrarily small neighborhood of the point @xmath393 . in a neighborhood of @xmath393 ,
the elements of @xmath390 are bounded by an unspecified power of @xmath394 . * * jump condition : * the boundary values taken by @xmath390 along @xmath395 and @xmath79 satisfy the following jump conditions : @xmath396 and @xmath397 as these jump conditions are both involutive , orientation of the jump contours is irrelevant . * * normalization : * the following normalization conditions hold : @xmath398 [ rhp : wodotcrit ] there are many solutions of this riemann - hilbert problem , a fact that can be traced partly to the unspecified power - law rate of growth admitted as @xmath399 .
the complete variety of solutions is not directly relevant for us ; we will simply select a family of particular solutions in the class of diagonal matrices .
indeed , for each @xmath236 , we have the solution @xmath400^{(1 - 2{m})\sigma_3}(w_*-w)^{({m}-1/2)\sigma_3}. \label{eq : poutds}\ ] ] note that if @xmath253 is held fixed , then @xmath401 and its inverse are bounded when @xmath116 is bounded away from @xmath402 . later
( see ) , we will let @xmath402 depend weakly on @xmath6 and @xmath11 in such a way that @xmath403 as @xmath404 and @xmath405 . clearly ,
any bound for @xmath406 or its inverse valid for @xmath116 bounded away from @xmath407 will hold uniformly with respect to @xmath6 and @xmath11 near criticality , and of course @xmath406 is independent of @xmath282 .
let @xmath408 be a fixed neighborhood of @xmath129 . by straightforward substitutions
, we may assume without loss of generality that within @xmath408 , the jump contour for @xmath384 consists of the real axis together with four arcs ( the lens boundaries ) meeting at some real point @xmath409 tending to @xmath407 as criticality is approached ; the way @xmath402 is determined as a function of @xmath6 and @xmath11 will be explained later ( the resulting formula being ) .
we suppose further that the lens boundaries lie along certain curves ( also to be specified later ) tangent at @xmath393 to the straight lines @xmath410 and @xmath411 .
some calculations show that the exact jump conditions for @xmath384 within @xmath408 can be expressed only in terms of the analytic function @xmath380 and another analytic function @xmath412 , which is the analytic continuation from @xmath413 of the function @xmath414 the jump matrix @xmath415 for which along each of six contour arcs meeting at @xmath393 we have @xmath416 is illustrated in figure [ fig : ptildelocaljumps ] . near @xmath417 .
the four non - real arcs of the jump contour are in general only approximately linear near the intersection point @xmath393 . _
_ ] the function @xmath145 can easily be removed from the jump conditions by making the following near - identity transformation ( locally , in @xmath408 ) : @xmath418 where @xmath419 is the piecewise - analytic function given by @xmath420 note that @xmath421 holds uniformly for @xmath422 , so indeed @xmath423 .
the jump conditions satisfied by @xmath424 near @xmath417 are as shown in figure [ fig : ptildelocaljumps ] but with @xmath145 replaced by @xmath425 in all cases .
the function @xmath426 is analytic at @xmath129 , having the taylor expansion @xmath427(w+1)^2 \\ & \quad\quad\quad\quad{}+\left[\frac{3}{16}\delta
x- \frac{1}{8}t + \nu\right](w+1)^3 \\ & \quad\quad\quad\quad{}+\left[\frac{5}{32}\delta x-\frac{15}{128}t + \frac{3}{2}\nu\right](w+1)^4 + { \mathcal{o}}((w+1)^5),\quad w\to -1 , \end{split } \label{eq : bexpansion}\ ] ] where @xmath224 is independent of @xmath116 , @xmath6 , and @xmath11 , being defined by .
all taylor coefficients are linear in @xmath428 and @xmath11 , and hence in particular the error term is uniform for bounded @xmath428 and @xmath11 .
[ prop : kexpand ] for @xmath413 and @xmath422 , we have the formula @xmath429,\ ] ] where @xmath359 is given by @xmath430 to express the boundary values @xmath431 for @xmath413 in a form admitting analytic continuation to a full neighborhood of @xmath129 ( a point that lies on the discontinuity contour @xmath121 $ ] of @xmath359 ) , let @xmath432 denote two contours from @xmath433 to @xmath434 , with @xmath435 in the upper half @xmath436-plane and @xmath437 in the lower half @xmath436-plane , and such that the two contours are mapped onto each other with reversal of orientation under the mapping @xmath438 ( in particular , this mapping permutes the points @xmath123 and @xmath124 ) .
then , also recalling the definition of @xmath439 in terms of @xmath440 and @xmath441 , @xmath442 this formula represents the analytic continuation to a full neighborhood of @xmath129 of the function @xmath443 originally defined by only for @xmath413 .
in fact , the domain of analyticity for @xmath443 has now been extended to the domain enclosed by the contour @xmath444 .
it is also obvious that @xmath443 depends linearly on @xmath6 and @xmath11 , and this property will clearly be inherited by all of its taylor coefficients . to analyze @xmath412 near the point @xmath129 to which the self - intersection point @xmath402 will converge at criticality , we begin with the following elementary taylor expansions ( convergent for @xmath445 ) : @xmath446 we also have the expansion @xmath447(w+1)\\ & \quad\quad\quad\quad{}+\left[\frac{1}{(s+1)^3}-\frac{1}{2}\frac{1}{(s+1)^2 } -\frac{1}{8}\frac{1}{s+1}\right](w+1)^2\\ & \quad\quad\quad\quad{}+\left[\frac{1}{(s+1)^4}-\frac{1}{2}\frac{1}{(s+1)^3 } -\frac{1}{8}\frac{1}{(s+1)^2}-\frac{1}{16}\frac{1}{s+1}\right](w+1)^3\\ & \quad\quad\quad\quad{}+\left [ \frac{1}{(s+1)^5}-\frac{1}{2}\frac{1}{(s+1)^4}-\frac{1}{8 } \frac{1}{(s+1)^3}-\frac{1}{16}\frac{1}{(s+1)^2}-\frac{5}{128 } \frac{1}{s+1}\right](w+1)^4\\ & \quad\quad\quad\quad{}+{\mathcal{o}}((w+1)^5 ) , \end{split}\ ] ] which is a convergent power series for @xmath448 . therefore , given contours @xmath432 as above ,
if @xmath449 is sufficiently small we may integrate term - by - term to obtain @xmath450(w+1 ) + \left[i_3-\frac{1}{2}i_2- \frac{1}{8}i_1\right](w+1)^2 \\ & \quad\quad\quad\quad{}+\left[i_4-\frac{1}{2}i_3-\frac{1}{8}i_2- \frac{1}{16}i_1\right](w+1)^3 \\ & \quad\quad\quad\quad{}+\left[i_5-\frac{1}{2}i_4-\frac{1}{8}i_3-\frac{1}{16 } i_2-\frac{5}{128}i_1\right](w+1)^4 + { \mathcal{o}}((w+1)^5 ) , \end{split } \label{eq : termbytermcrit}\ ] ] where @xmath451 by making the substitution @xmath452 , we may rewrite these integrals in the form @xmath453 therefore , we see immediately that @xmath454 while by averaging and we obtain for @xmath455 , @xmath456 , and @xmath457 the following : @xmath458 and @xmath459 also , averaging @xmath460 as given by and and comparing with and we obtain the identity @xmath461 now we calculate @xmath462 exactly .
recalling the definitions of the functions @xmath76 and @xmath463 , as well as the definition of @xmath464 , we see that @xmath465 since @xmath466 can be eliminated in favor of @xmath467 by the identity @xmath468 , we would like to introduce @xmath469 as a new integration variable .
the two contours @xmath432 are mapped under @xmath470 to two oppositely - oriented copies of the same contour . in particular , @xmath471 is a teardrop - shaped contour beginning and ending at the point @xmath472 lying on the real axis to the right of @xmath473 and encircling the point @xmath474 once in the positive sense . along the contours
@xmath432 we have the identities @xmath475 ( principal branch ) .
these considerations show that both contributions from @xmath435 and from @xmath437 are equal , and so @xmath476 next , we integrate by parts , using the fact that @xmath477 : @xmath478 collapsing the contour to the top and bottom of the interval @xmath479 and noting that @xmath480 we obtain @xmath481 now substituting from the definition and exchanging the order of integration ( see ( * ? ? ? * proposition 1.1 ) ) leads to the identity @xmath482 next we will show that @xmath483 where @xmath224 is defined by . using together with and
the definitions of @xmath76 and @xmath463 gives @xmath484 introducing @xmath469 as a new integration variable , as above , yields @xmath485 and then integrating by parts , @xmath486 note that by the subsitution @xmath487 , @xmath488 may be written in the form @xmath489 where the function @xmath86 is defined in terms of the initial condition @xmath4 by .
we need to write this formula in a way that admits analytic continuation to complex @xmath107 such as those @xmath490 . to this end , let @xmath491 where the principal branch is meant , and then note that @xmath492 where @xmath493 is a loop contour surrounding , in the negative ( clockwise ) sense , the straight - line branch cut of @xmath494 ( viewed as a function of @xmath304 ) connecting @xmath495 with @xmath496 . in writing the formula we therefore are using assumption [ ass : scriptgnice ] to guarantee analyticity of @xmath86 in a neighborhood of @xmath83 .
with the loop contour @xmath493 fixed , it is clear that @xmath488 as given by is analytic if @xmath497 varies in the region enclosed by @xmath493 .
in particular , we will assume that @xmath437 has been chosen so that @xmath471 is completely contained in this region .
since @xmath471 and @xmath349 are both fixed contours , we may exchange the order of integration upon substituting into : @xmath498 where the inner integral is now : @xmath499 ^ 2}\frac{t(m , z)\,dz}{(4-z)^{3/2}}.\ ] ] note that the contour @xmath500 ^ 2 $ ] is a teardrop - shaped contour beginning and ending at @xmath501 and encircling the part of the ( principal ) branch cut of @xmath502 between @xmath503 and @xmath504 once in the positive sense , and each @xmath505 lies outside of this closed contour . to evaluate @xmath506
we note that the branch cut of @xmath507 viewed as a function of @xmath235 is the ray from @xmath508 to @xmath509 in the direction away from @xmath501 . by simple contour deformations of @xmath500 ^ 2
$ ] taking into account that the integrand @xmath510 is integrable at @xmath509 and changes sign across all branch cuts , we obtain , for @xmath511 , @xmath512 where the path of integration is a straight line in the region of analyticity of the integrand .
since @xmath513 and @xmath514 coincide , we may absorb the sign by changing the branch of @xmath502 in the lower - half @xmath235-plane : @xmath515 with the substitution @xmath516 the formula for @xmath506 becomes @xmath517\,d\alpha\\ & = \frac{4}{(4-m)\sqrt{g(0)^2 - 4}}. \end{split}\ ] ] inserting this formula into and taking into account that the only singularity of the integrand enclosed by the contour @xmath493 is the simple pole at @xmath518 yields @xmath519 evaluating @xmath520 directly using the definition then proves that indeed @xmath521 where @xmath224 is given by . using , , , and in , and combining the resulting expansion with the expansions and the analytic formula for @xmath522 completes the proof of the proposition . from it
is clear that exactly at criticality , @xmath523 . for small @xmath524 and @xmath293 ,
the cubic degeneration unfolds , with one real and two complex conjugate roots , or three real roots , near @xmath129 .
the double critical point ( double root of @xmath525 ) unfolds generically to a pair @xmath526 of simple critical points near @xmath129 , and either these are both real or they form a complex - conjugate pair ( in both cases since @xmath527 , @xmath528 ) .
the main idea here is that to unfold the cubic degeneracy it is really only necessary to take into account the lower - order terms in the taylor expansion , and therefore it seems desirable to somehow replace @xmath522 by an appropriate cubic polynomial with coefficients depending on @xmath36 .
this issue has arisen frequently in the construction of local parametrices for matrix riemann - hilbert problems corresponding to certain _
double - scaling limits_. perhaps the first time such a replacement was made rigorous was in the paper of baik , deift , and johansson @xcite , in which a certain double - scaling limit of orthogonal polynomials on the unit circle is analyzed , and the authors construct a local parametrix by essentially truncating an analogue of the taylor expansion after the cubic term .
of course such a truncation is not exact , so there are errors incurred in modeling the jump matrices by others having cubic exponents , and the effect of these errors must be carefully controlled .
also , any artificial truncation of a taylor series can only be accurate if the local parametrix is constructed in a disk centered at the expansion point whose radius tends to zero at some rate tied to the large parameter in the problem .
this fact further implies that estimates must be supplied for norms of singular integral operators that are independent of the moving contour and it also typically means that divergence of the outer parametrix near the expansion point can lead to a sub - optimal mismatch with the local parametrix , leading in turn to sub - optimal estimates of errors .
a significant advance was made by claeys and kuijlaars in @xcite ( see in particular section 5.6 of that paper ) ; here the authors consider a similar double - scaling limit and take the approach of constructing a certain nontrivial conformal mapping @xmath529 of a fixed neighborhood of the expansion point to a neighborhood of the origin .
the conformal mapping is more - or - less explicit , and it does not depend on the parameters ( analogues of @xmath6 and @xmath11 ) driving the system to criticality . however , the point is that in a full fixed neighborhood of the expansion point , the analogue of the function @xmath522 is represented exactly ( no truncation required ) as a cubic polynomial in @xmath530 .
the difficulty that remains with this approach is that the coefficients of the cubic in @xmath530 actually depend on @xmath116 ( so in fact it is not really a cubic after all ) .
the remarkable approach of @xcite is to simply substitute the @xmath116-dependent coefficients into a known solution formula for exactly cubic exponents ( involving solutions of the linear differential equation whose isomonodromy deformations are governed by solutions of the painlev - ii equation ) , resulting in a local parametrix in terms of painlev transcendents depending ( through the coefficients of the cubic ) on @xmath116 .
confirming that such an approach actually provides a usable local parametrix requires exploiting _ a priori _ knowledge of the behavior of solutions of the nonlinear painlev - ii equation and its auxiliary linear differential equation .
if this information is available in a convenient form , the approach of claeys and kuijlaars delivers a vast improvement over earlier methods because it really works in a neighborhood of fixed size centered at the expansion point where critical points coalesce at criticality .
the technique advanced in @xcite has more recently been applied to problems of nonlinear wave theory as well ( see , for example , @xcite ) .
we choose instead to replace @xmath522 by a cubic in an exact way , an approach that provides all of the accuracy of the claeys - kuijlaars method but seems simpler and requires no _ a priori _ knowledge of the behavior of solutions of the local parametrix riemann - hilbert problem .
the approach we are going to explain now has also been used recently to study a different kind of double - scaling limit for a matrix riemann - hilbert problem in @xcite . as part of a careful study of the asymptotic behavior of exponential integrals with exponent functions having coalescing critical points ( to generalize the steepest descent or saddle point method ) , chester , friedman , and ursell @xcite showed how to construct a substitution that rendered the exponent function in the integrand in the exact form of a cubic polynomial .
their method applies in the current context to establish that , _ because the coefficient @xmath215 in the expansion of @xmath522 is strictly positive _ , there is a suitable choice of new spacetime coordinates @xmath531 and @xmath532 for which the relation @xmath533 defines an invertible conformal mapping @xmath534 between @xmath422 and @xmath535 that preserves the real axis , for @xmath36 near enough to criticality .
moreover , the new coordinates @xmath536 and @xmath436 depend continuously on @xmath36 near criticality . unlike in the claeys - kuijlaars method @xcite ,
neither @xmath536 nor @xmath436 depends on @xmath116 with the cost that the conformal mapping will now depend on @xmath36 .
the new coordinates @xmath531 and @xmath532 are to be determined so that under the critical points of the cubic on the right - hand side , namely @xmath537 , correspond to the two critical points of @xmath412 near @xmath129 when @xmath428 and @xmath11 are sufficiently small .
evaluating for @xmath526 and @xmath538 , one obtains the formulae @xmath539 ^ 2\right)^{1/3}\\ s(x , t)&=-\frac{1}{2}\left[k(w_+(x , t);x , t)+k(w_-(x , t);x , t)\right ] , \end{split } \label{eq : rsdefine}\ ] ] where in the formula for @xmath536 the real cube root is meant , and hence both @xmath536 and @xmath436 are real .
moreover , it is possible to show that @xmath540 and @xmath541 are analytic functions of @xmath6 and @xmath11 near criticality , and have two - variable taylor expansions of the form : @xmath542 to see this , first we find a unique root @xmath543 of @xmath544 such that @xmath545 at criticality and such that @xmath546 is an analytic function of @xmath36 near criticality . indeed , from the taylor expansion and the fact that @xmath547 we see that the analytic implicit function theorem applies and we obtain @xmath548 now write @xmath549 , and re - expand @xmath522 about @xmath550 .
from and the definition of @xmath551 we obtain the convergent power series expansion ( note that there is no quadratic term ) @xmath552 where the coefficients are all analytic functions of @xmath36 near criticality , and in particular , @xmath553 @xmath554 @xmath555 the radius of convergence of this series is bounded away from zero near criticality .
the equation satisfied by the critical points of @xmath443 is @xmath556 let @xmath557 be any number satisfying the equation @xmath558 note that @xmath559 is an analytic function of @xmath36 near criticality because @xmath560 . if @xmath561 then @xmath562 and @xmath550 is a double root of .
so we suppose that @xmath563 , and we rescale @xmath235 by writing @xmath564 for some new unknown @xmath565 . dividing through by @xmath566 and canceling a factor of @xmath559 then converts into the equation @xmath567 the coefficients of @xmath568 in the sum are analytic functions of the three variables @xmath6 , @xmath11 , and @xmath557 . when @xmath561 there are two solutions , @xmath569 , and by the implicit function theorem there are two corresponding solutions for @xmath570 and @xmath36 near criticality .
the two solutions are related by the symmetry @xmath571 .
we may write them in the form @xmath572 the coefficients @xmath573 are all analytic functions of @xmath36 near criticality , and the radius of convergence of the series with respect to @xmath557 is bounded away from zero near criticality .
the corresponding critical points are written in terms of @xmath235 as @xmath574 to calculate @xmath540 and @xmath541 we need to evaluate @xmath522 at the critical points .
this is accomplished by substitution of the series into : @xmath575 where we have used , and @xmath576 . here
again , the coefficients @xmath577 are certain analytic functions of @xmath36 near criticality , and the radius of convergence of the power series in @xmath578 is bounded below near criticality .
then , by definition , we have @xmath579 ^ 2\right)^{1/3}\\ & = \frac{k_1(x , t)}{k_3(x , t)^{1/3}}\left(1-\frac{1}{2}\sum_{j=1}^\infty\frac{g_{2j}(x , t)}{k_3(x , t)}\left[-\frac{k_1(x , t)}{3k_3(x , t)}\right]^j\right)^{2/3 } , \end{split}\ ] ] where we have used the definition of @xmath580 in terms of @xmath581 and .
therefore , @xmath540 is clearly an analytic function of @xmath36 near criticality .
similarly , by definition we have @xmath582\\ & = -k_0(x , t ) -\sum_{j=0}^\infty g_{2j+1}(x , t)\left(-\frac{k_1(x , t)}{3k_3(x , t)}\right)^{j+2 } , \end{split}\ ] ] which again is obviously an analytic function of @xmath36 near criticality .
the leading terms of @xmath540 and @xmath541 near criticality are easy to calculate from these formulae , with the result being .
as @xmath583 is a conformal map , its inverse is an analytic function mapping a neighborhood of @xmath584 to a neighborhood of @xmath129 .
we now define @xmath585 which is a real analytic function of @xmath6 and @xmath11 near criticality .
note also that @xmath586 ( prime denotes differentiation with respect to @xmath116 ) near criticality .
it is not difficult to obtain the taylor expansion of the conformal mapping @xmath587 about @xmath393 exactly at criticality . indeed , from and with @xmath588
, we obtain the equation @xmath589 , \quad\text{at criticality},\ ] ] and analytically blowing up the cubic degeneracy we obtain @xmath590 , \quad\text{at criticality}.\ ] ] in particular , this implies that @xmath591 also , it is clear that @xmath592 , and therefore @xmath593 .
since @xmath594 is an analytic function of @xmath6 and @xmath11 , it then follows that @xmath595 and furthermore , @xmath596 near criticality .
now let @xmath597 and @xmath55 be scaled versions of @xmath530 and @xmath536 respectively : @xmath598 we will regard @xmath55 as being bounded .
it then follows that _ without approximation _ the exponent appearing in the jump matrix @xmath599 takes the form of a simple cubic with a formally large constant term : @xmath600 to begin with , we wish to find a simpler representation for @xmath406 valid when @xmath116 is close to @xmath417 . using , we write @xmath406 in the form @xmath601 where @xmath602 is independent of @xmath282 and is analytic and nonvanishing in a neighborhood @xmath408 of @xmath393 : @xmath603 ( analyticity follows since from we have @xmath604 and @xmath530 is analytic at @xmath402 with @xmath605 . ) given a value of @xmath236 ( to be determined below ) , we wish to construct an inner model , valid for @xmath422 , of the matrix @xmath424 ( related to @xmath384 for @xmath422 via ) .
we temporarily denote this model as @xmath606 , and we want it to have the following properties : * supposing that in @xmath408 the lens boundaries are identified with arcs of the curves @xmath607 and @xmath608 , which makes them segments ( of length proportional to @xmath609 ) of straight rays in the @xmath597-plane , the inner model @xmath606 should be analytic exactly where @xmath424 is within @xmath408 and should satisfy _ exactly the same jump conditions _ that @xmath424 does within @xmath408 .
* the inner model @xmath610 should match onto the latter factors in along the disc boundary @xmath611 in the sense that @xmath610 may be analytically continued from each sector of the domain @xmath612 to the corresponding infinite sector in the @xmath597-plane , and that @xmath613 with the limit being uniform with respect to direction in each of the six sectors of analyticity .
since @xmath611 is fixed and bounded away from @xmath389 , upon scaling its image under @xmath583 by @xmath609 to work in terms of the variable @xmath597 , we see that @xmath614 corresponds to @xmath615 at a uniform rate of @xmath609 .
since @xmath606 depends on @xmath116 through @xmath616 , it is convenient to use to write the jump matrices for @xmath424 ( and hence also for @xmath606 ) in terms of @xmath597 , which shows that the jump matrices involve the product of exponentials @xmath617 .
the constant ( @xmath597-independent ) factors @xmath618 present in the jump matrices can be removed , and simultaneously the presence of @xmath282 in the normalization condition can be eliminated , by defining the equivalent unknown @xmath619 the conditions previously discussed as being desirable for @xmath610 are then easily seen to be equivalent to those of the following problem for @xmath620 : let a real number @xmath55 and an integer @xmath253 be fixed . seek a matrix @xmath620 with the following properties : * * analyticity : * @xmath620 is analytic in @xmath597 except along the rays @xmath621 , @xmath622 , from each sector of analyticity it may be continued to a slightly larger sector , and in each sector is hlder continuous up to the boundary in a neighborhood of @xmath623 . * * jump condition : * the jump conditions satisfied by the matrix function @xmath620 are of the form @xmath624 , where the jump matrix @xmath625 is as shown in figure [ fig : vlocr ] + . _ _ ] + and all rays are oriented toward infinity . *
* normalization : * the matrix @xmath620 satisifies the condition @xmath626 with the limit being uniform with respect to direction in each of the six sectors of analyticity .
[ rhp : dslocalii ] suppose that riemann - hilbert problem [ rhp : dslocalii ] has a unique solution for @xmath627 and for some integer @xmath253 .
we will now describe how to use it to create a model for @xmath384 valid in the neighborhood @xmath408 assuming that @xmath628 ; we set @xmath629 the effect of post - multiplication by @xmath630 is simply to restore the factors involving @xmath145 to the jump conditions .
we emphasize that within the fixed neighborhood @xmath408 of @xmath129 , the inner model matrix @xmath631 satisfies exactly the same jump conditions along the six contour arcs meeting at @xmath393 as does @xmath384 .
it appears reasonable to propose a global model for @xmath384 in the following form : @xmath632 the integer @xmath253 is evidently at our disposal .
we will later describe how it should be chosen .
while it may not yet be clear whether there exists a solution of riemann - hilbert problem [ rhp : dslocalii ] for any @xmath53 and @xmath236 at all , it is a standard argument that there exists at most one solution for each @xmath53 and @xmath236 , and that every solution must satisfy @xmath633 . in this short section
we suppose that @xmath53 and @xmath236 are values for which riemann - hilbert problem [ rhp : dslocalii ] has a ( unique and unimodular ) solution , and we examine some of the consequences of the existence .
we will later show that @xmath634 has a complete asymptotic expansion in descending powers of @xmath597 as @xmath615 : @xmath635 and where the matrix coefficients are the same regardless of which of the six sectors of analyticity of @xmath620 is used to compute the expansion .
given a solution @xmath620 of riemann - hilbert problem [ rhp : dslocalii ] , consider the related matrix @xmath636 since the jump contour for @xmath620 is invariant under complex conjugation , @xmath637 will be analytic in the same domain that @xmath620 is .
it is a direct calculation that @xmath637 satisfies exactly the same jump conditions as does @xmath620 .
therefore , the matrix @xmath638 extends continuously to the jump contour from both sides of each arc .
it follows from the classical sense in which the boundary values are attained ( even at the self - intersection point @xmath623 ) , and from the fact that @xmath620 is unimodular , that in fact @xmath638 is an entire function of @xmath597 .
moreover , @xmath639^{-1}\\ & { } = \lim_{\zeta\to\infty}\mathbf{z}_{m}(\zeta^*;y)^ * ( -\zeta)^{(1 - 2{m})\sigma_3/2 } = \lim_{\zeta\to\infty}\left[\mathbf{z}_{m}(\zeta^*;y ) ( -\zeta^*)^{(1 - 2{m})\sigma_3/2}\right]^ * = \mathbb{i } , \end{split}\ ] ] so by liouville s theorem , @xmath640 .
applying this symmetry to the expansion shows that the elements of the matrices @xmath641 and @xmath642 ( and in fact all of the matrix expansion coefficients ) are real numbers .
we can obtain more detailed information by exploiting a further symmetry in the special case of @xmath643 . indeed , let us compare @xmath644 with the matrix @xmath645
clearly , @xmath646 , and it is a direct matter to check that @xmath637 is analytic precisely where @xmath644 is , and @xmath637 satisfies exactly the same jump conditions on the six rays of riemann - hilbert problem [ rhp : dslocalii ] as does @xmath644 .
therefore , @xmath647 is an entire unimodular matrix function of @xmath597 . to identify this entire function as a polynomial ,
it is enough to extract the non - decaying terms in its asymptotic expansion as @xmath615 ; assuming without loss of generality that @xmath648 so that @xmath649 , we have @xmath650^{-1}\\ & { } = \mathbf{z}_0(-\zeta;y)\sigma_1\sigma_3\zeta^{\sigma_3/2 } \cdot\left(\mathbb{i}-\mathbf{a}_0(y)\zeta^{-1 } + { \mathcal{o}}(\zeta^{-2})\right)\\ & { } = \mathbf{z}_0(-\zeta;y)\zeta^{\sigma_3/2 } \begin{bmatrix}0 & -\zeta^{-1}\\\zeta & 0\end{bmatrix } \left(\mathbb{i}-\mathbf{a}_0(y)\zeta^{-1 } + { \mathcal{o}}(\zeta^{-2})\right)\\ & { } = \left(\mathbb{i}-\mathbf{a}_0(y)\zeta^{-1}+{\mathcal{o}}(\zeta^{-2})\right ) \begin{bmatrix}0 & -\zeta^{-1}\\\zeta & 0\end{bmatrix } \left(\mathbb{i}-\mathbf{a}_0(y)\zeta^{-1}+{\mathcal{o}}(\zeta^{-2})\right)\\ & = \sigma_-\zeta -\mathbf{a}_0(y)\sigma_-- \sigma_-\mathbf{a}_0(y ) + { \mathcal{o}}(\zeta^{-1}),\quad\zeta\to\infty , \end{split } \label{eq : fqminusexpansion}\ ] ] where @xmath651 and so @xmath647 may be identified with the polynomial terms in this expansion ; therefore we arrive at the identity @xmath652 since both sides represent unimodular matrices , we learn that @xmath653 we will later obtain a refinement of , namely that @xmath654 holds for all real @xmath55 .
the sign information will follow from the explicit solution of riemann - hilbert problem [ rhp : dslocalii ] for @xmath643 in terms of special functions , which will be possible precisely as a consequence of the identity .
the only parameters in the riemann - hilbert problem characterizing @xmath620 are @xmath236 and @xmath53 . assuming existence for some fixed @xmath236 and for @xmath55 in some open set
, we will now investigate some consequences of the dependence of the solution on @xmath55 .
we suppose that the expansion holds in the stronger sense that the following are also true : @xmath655={\mathcal{o}}(\zeta^{-1})\quad\text{and}\quad \frac{\partial}{\partial\zeta}\left[\mathbf{z}_{m}(\zeta;y ) ( -\zeta)^{(1 - 2{m})\sigma_3/2 } \right]={\mathcal{o}}(\zeta^{-2}),\quad\zeta\to\infty.\ ] ] the matrix @xmath656 defined by @xmath657 has jump discontinuities along the six rays mediated by jump matrices that are independent of both @xmath55 and @xmath597 .
given the compatibility of the jump matrices at @xmath623 , this implies that the matrices @xmath658 are both entire functions of @xmath597 . using the above expansions of @xmath620 and its derivatives , we easily obtain the following expansions involving @xmath659 : @xmath660\sigma_3 + { \mathcal{o}}(\zeta^{-1})\right ) e^{-\theta\sigma_3}\\ & = \left(-\frac{1}{2}\sigma_3\zeta-\frac{1}{2}\mathbf{a}_{m}(y)\sigma_3 + { \mathcal{o}}(\zeta^{-1})\right)e^{-\theta\sigma_3}\\ \frac{\partial\mathbf{l}_{m}}{\partial\zeta}&= \left(-\frac{\partial\theta}{\partial\zeta } \left[\mathbb{i}+\mathbf{a}_{m}(y)\zeta^{-1}+\mathbf{b}_{m}(y)\zeta^{-2 } + { \mathcal{o}}(\zeta^{-3})\right]\sigma_3 + { \mathcal{o}}(\zeta^{-2})\right)e^{-\theta\sigma_3}\\ & = \left(-\frac{3}{2}\sigma_3\zeta^2-\frac{3}{2}\mathbf{a}_{m}(y)\sigma_3\zeta -\frac{3}{2}\mathbf{b}_{m}(y)\sigma_3-\frac{1}{2}y\sigma_3 + { \mathcal{o}}(\zeta^{-1 } ) \right)e^{-\theta\sigma_3}\\ \mathbf{l}_{m}^{-1}&=e^{\theta\sigma_3}\left(\mathbb{i}-\mathbf{a}_{m}(y)\zeta^{-1 } + \left[\mathbf{a}_{m}(y)^2-\mathbf{b}_{m}(y)\right]\zeta^{-2 } + { \mathcal{o}}(\zeta^{-3})\right ) , \end{split}\ ] ] where @xmath661.\ ] ] we
can then easily see that both @xmath662 and @xmath663 grow algebraically in @xmath597 as @xmath615 and hence are necessarily polynomials : @xmath664\\ \mathbf{v}_{m}(y,\zeta)&=-\frac{3}{2}\sigma_3\zeta^2+\frac{3}{2 } [ \sigma_3,\mathbf{a}_{m}(y)]\zeta -\frac{3}{2}[\sigma_3,\mathbf{a}_{m}(y)]\mathbf{a}_{m}(y)+\frac{3}{2 } [ \sigma_3,\mathbf{b}_{m}(y)]-\frac{1}{2}y\sigma_3 .
\end{split}\ ] ] we may rewrite these in the form @xmath665 where @xmath666 the matrix @xmath659 is therefore a simultaneous ( and fundamental , since @xmath667 ) solution of the _ lax equations _
@xmath668 and therefore the compatibility condition @xmath669=0\ ] ] holds identically in @xmath597 .
separating the powers of @xmath597 leads to the following system of differential equations ( the _ painlev - ii system _ ) governing the quantities defined by : @xmath670 although satisfied by quantities evidently depending on @xmath236 , the painlev - ii system does not involve @xmath253 in any explicit way . eliminating @xmath671 and @xmath672 yields the coupled system of second - order painlev - ii - type equations @xmath673 in fact , functions related in an elementary way to @xmath242 and @xmath674 turn out to satisfy uncoupled second - order equations of painlev - ii type ,
but we must use the _ inhomogeneous _ form of the painlev - ii equation , and the coefficient of inhomogeneity will depend on @xmath236 . to see that this is so , we follow @xcite by multiplying the first equation of by @xmath674 , the second by @xmath242 , and subtracting : @xmath675.\ ] ] hence , the quantity @xmath676 \label{eq : nuconst}\ ] ] is a constant , independent of @xmath55 . to compute the value of @xmath677 for the particular solution of corresponding to the specific riemann - hilbert problem for @xmath620
, we first consider the direct problem for the lax equation @xmath678 , formally expanding @xmath656 for large @xmath597 by assuming the form @xmath679 where each of the coefficients @xmath680 is an off - diagonal matrix , and where @xmath681 is a diagonal matrix with the asymptotic expansion @xmath682 where all of the coefficients @xmath683 are diagonal matrices .
the value of @xmath677 will be determined from the coefficient @xmath684 . by differentiation of the formal series , we obtain @xmath685 e^{\mathbf{d}_{m}(y,\zeta ) } , \quad\zeta\to\infty,\ ] ] where the matrix coefficients are systematically determined as follows : @xmath686 and so on . by setting each @xmath687 to zero in sequence , and by separating the diagonal and off - diagonal parts of each of these equations , we obtain the following : the matrix equation @xmath688 implies @xmath689 the matrix equation @xmath690 implies @xmath691 the matrix equation @xmath692 implies @xmath693 the matrix equation @xmath694 implies @xmath695 where we have recalled the definition of @xmath677 , and
finally , the diagonal terms of the matrix equation @xmath696 imply that @xmath697 where the painlev - ii _ hamiltonian _ is @xmath698 it follows that the formal asymptotic expansion of @xmath656 for large @xmath597 takes the form @xmath699\zeta^{-1 } \\
& + \left[\mathbf{y}_{{m},2}(y)+\mathbf{d}_{{m},2}(y)+\frac{1}{2}\mathbf{d}_{{m},1}(y)^2 + \mathbf{y}_{{m},1}(y)\mathbf{d}_{{m},1}(y)\right]\zeta^{-2}\\ & + { \mathcal{o}}(\zeta^{-3 } ) , \quad\zeta\to\infty .
\end{split } \label{eq : gexpansion}\ ] ] comparing with and , we see that @xmath700 and also that @xmath701 so that in addition to the relations we have @xmath702 and @xmath703 now that the value of @xmath677 has been determined according to , a direct calculation shows that the logarithmic derivatives @xmath704 satisfy the uncoupled painlev - ii equations @xmath705 & \qquad\qquad \mathcal{q}_{m}''(y)&=2\mathcal{q}_{m}(y)^3+\frac{2}{3}y\mathcal{q}_{m}(y)-\left[\frac{1}{3}+2\lambda_{m}\right ] \\ & = 2\mathcal{p}_{m}(y)^3+\frac{2}{3}y\mathcal{p}_{m}(y)-\frac{2}{3}{m}&\qquad\qquad & = 2\mathcal{q}_{m}(y)^3+\frac{2}{3}y\mathcal{q}_{m}(y)+\frac{2}{3}({m}-1 ) .
\end{alignedat}\ ] ] note that for general @xmath236 , both equations are of inhomogeneous type . recalling the relation and the definitions of the potentials @xmath242 , @xmath674 , @xmath671 , and @xmath672 , we learn that when @xmath643 , @xmath706 , where ( assuming for the moment continuity of @xmath707 with respect to @xmath55 ) @xmath708 is a fixed sign to be determined .
according to the system of differential equations necessarily satisfied by the potentials , we can determine the values of the remaining potentials in the special case of @xmath643 : @xmath709 therefore , from and we see that in the special case of @xmath643 , the linear differential equations simultaneously satisfied by the matrix @xmath710 take the form @xmath711 let @xmath712 denote either of the columns of @xmath713 . from the first of the above two lax equations , we observe that @xmath714 and then that @xmath715 where @xmath716 is any solution of airy s equation @xmath717 .
therefore @xmath718 necessarily has the form @xmath719 and by substitution into the second equation of the lax pair we learn that the airy function @xmath716 may only depend on @xmath597 through the variable @xmath720 , in other words , we must select for @xmath716 a linear combination of @xmath721 and @xmath722 with coefficients independent of @xmath597 .
now we try to select the correct solutions of airy s equation to construct @xmath723 and hence @xmath644 in each of the six sectors of analyticity in such a way that we obtain the desired asymptotics for large @xmath597 and satisfy the required jump conditions .
this procedure will also determine the correct value of @xmath708 .
let us consider three specific choices of solution to airy s equation leading to the three solution vectors @xmath724 each of these is an entire function of @xmath597 . from the standard asymptotic formulae @xmath725 valid as @xmath238 with @xmath726 for any @xmath165 , we see that ( all of the following asymptotic statements assume @xmath55 is held fixed ) @xmath727 latexmath:[\[\mathbf{l}^0 = -\sigma\frac{(-\zeta)^{1/2}e^{(\zeta^3+y\zeta)/2}}{(48)^{1/6}\sqrt{\pi } } \begin{bmatrix}{\mathcal{o}}(\zeta^{-1})\\ 1+{\mathcal{o}}(\zeta^{-1 } ) \end{bmatrix},\quad \zeta\to\infty,\quad latexmath:[\[\mathbf{l}^\pm = -\sigma e^{\pm 2\pi i/3 } \frac{(e^{\pm i\pi/3}\zeta)^{1/2}e^{(\zeta^3+y\zeta)/2}}{(48)^{1/6}\sqrt{\pi } } \begin{bmatrix}{\mathcal{o}}(\zeta^{-1})\\1+{\mathcal{o}}(\zeta^{-1})\end{bmatrix},\quad \zeta\to\infty,\quad latexmath:[\[\mathbf{l}^\pm = \frac{(e^{\mp 2\pi i/3}\zeta)^{-1/2}e^{-(\zeta^3+y\zeta)/2 } } { ( 48)^{1/6}\sqrt{\pi}}\begin{bmatrix}1+{\mathcal{o}}(\zeta^{-1})\\ { \mathcal{o}}(\zeta^{-1})\end{bmatrix},\quad\zeta\to\infty,\quad
define a candidate for @xmath644 that satisfies the required analyticity and normalization properties among the conditions of riemann - hilbert problem [ rhp : dslocalii ] : @xmath729 it only remains to verify the jump conditions of riemann - hilbert problem [ rhp : dslocalii ] , with the help of the identity @xmath730 which implies the vector identity @xmath731 .
using this fact , it is a direct matter to confirm that the required jump conditions are indeed satisfied , _ provided we make the choice of sign @xmath732_. we have therefore proved the following .
let @xmath643 .
for every @xmath53 , riemann - hilbert problem [ rhp : dslocalii ] has a unique solution given explicitly by the formulae and with @xmath732 .
moreover , @xmath644 is an entire function of @xmath55 , @xmath733 has a complete asymptotic expansion as @xmath615 in descending integer powers of @xmath597 ( as in ) differentiable term - by - term with respect to both @xmath55 and @xmath597 , and the corresponding potentials @xmath707 , @xmath734 , @xmath735 and @xmath736 are given by with @xmath732 . [ prop : r0locsolution ] now we describe an inductive procedure for obtaining @xmath620 uniquely from the conditions of riemann - hilbert problem [ rhp : dslocalii ] for @xmath737 .
suppose @xmath53 is a value for which riemann - hilbert problem [ rhp : dslocalii ] has a ( unique ) solution @xmath620 for some @xmath236 . following ( * ? ? ?
* chapter 6 ) , we claim that @xmath738 if the matrices @xmath739 and @xmath740 can be properly chosen .
it is immediately clear that since the the prefactor is entire in @xmath597 , the right - hand side of has the same domain of analyticity , achieves its boundary values in the same classical sense , and satisfies exactly the same jump conditions , as does @xmath620 .
of course the analyticity and jump conditions in riemann - hilbert problem [ rhp : dslocalii ] are independent of @xmath253 , so it only remains to impose the normalization condition on @xmath741 given by , and this condition takes the form @xmath742 assuming the expansion ( which at the moment is justified only for @xmath643 ) , we are requiring that @xmath743 equating the polynomial part of the left - hand side to the identity yields the conditions @xmath744 which can be solved uniquely for the matrix elements of @xmath745 and @xmath746 : @xmath747 _ provided that @xmath748 _ , where we have used , , and .
similarly , from we obtain the conditions @xmath749 which can be solved uniquely for the matrix elements of @xmath750 and @xmath751 : @xmath752 _ provided that @xmath753_. when it makes sense , the formula subject to or constitutes a so - called _ discrete isomonodromic schlesinger transformation_. the recurrence relation also implies a corresponding recurrence for the potentials @xmath242 , @xmath674 , @xmath671 , and @xmath672 . indeed , expanding the right - hand side of as @xmath754 the terms proportional to @xmath755
yield the conditions @xmath756 and @xmath757 recalling the definition of the potentials @xmath241 in terms of the matrix entries of @xmath641 , we examine the @xmath758-entry of the matrix equation and easily obtain the relation @xmath759 .
then , from the differential equations satisfied by the potentials for all @xmath760 , we obtain @xmath761 which together with the implied relations for @xmath762 and @xmath763 constitutes a _ bcklund transformation _ for the painlev - ii system .
similarly , the @xmath764-element of together with yields the bcklund transformation @xmath765 which is easily seen to be inverse to .
the iterative scheme for generating the solution to riemann - hilbert problem [ rhp : dslocalii ] for general @xmath236 is then simply to start with the solution valid for all @xmath53 for @xmath643 as described by proposition [ prop : r0locsolution ] and attempt to apply the discrete isomonodromic schlesinger transformation in order to repeatedly increase or decrease the value of @xmath253 in integer steps . at the level of the implied bcklund transformations and , this procedure generates a family @xmath766 of _ rational solutions _ of the painlev - ii system . at the level of the logarithmic derivatives given by we have a family of rational solutions of inhomogeneous painlev - ii equations with certain quantized values for the inhomogeneity parameter @xmath767 .
in fact , it is known @xcite that the equation @xmath768 has a unique rational solution @xmath769 if and only if @xmath770 .
these may be represented as logarithmic derivatives of other rational functions @xmath771 that are ratios of consecutive _ yablonskii - vorobev polynomials _
@xcite , which are known to have a number of remarkable properties .
it is important for us to recognize that the poles of these solutions can occur for finite real values of @xmath55 .
for example , applying twice with @xmath772 we obtain @xmath773 and @xmath774 , and the latter has a pole at @xmath775 .
these poles correspond to values of @xmath55 for which the discrete schlesinger maps from the neighboring values of @xmath253 fail to exist , and consequently for these values of @xmath55 and @xmath253 _ there exists no solution to riemann - hilbert problem [ rhp : dslocalii]_. while there exist values of @xmath55 for which the bcklund transformations or may not make sense , we may always interpret these transformations as maps on the ring of rational functions , and as such they can be shown to have a kind of _ singularity confinement property _ : if @xmath776 is a pole of the solution for some @xmath253 , then @xmath777 is a regular point of the solution for @xmath778 .
indeed , a local analysis of the system in the spirit of painlev s original method @xcite shows that all poles of @xmath779 are necessarily simple and simultaneously occur in both functions ; if @xmath777 is a pole , then for some constants @xmath780 and @xmath781 we necessarily have @xmath782\\ { \mathcal{v}}_{m}(y)&=-k^{-1}\left[\frac{1}{y - y_0 } + \frac{y_0}{18}(y - y_0 ) + \left(\frac{1}{12}-\omega\right)(y - y_0)^2+v_3(y)\right ] , \end{split } \label{eq : uvpainleveexpansions}\ ] ] where @xmath783 and @xmath784 are analytic functions each vanishing to third order at @xmath785 .
then , using these formulae in the bcklund transformations shows that the singularities at @xmath785 in the pairs @xmath786 and @xmath787 are all removable .
moreover , both @xmath788 and @xmath789 have simple zeros at @xmath785 .
it is easy to see that the singularity confinenment property extends to the schlesinger transformations that generated . that is , if @xmath790 does not exist ( because @xmath777 is a pole of @xmath791 and @xmath792 ) , then both @xmath793 and @xmath794 do exist and they may be calculated from @xmath620 by applying schlesinger transformations and then taking the limit @xmath795 .
we summarize these results in the following proposition .
let @xmath236 .
then by iterated schlesinger transformations given by subject to or applied to the base case of @xmath643 characterized by proposition [ prop : r0locsolution ] , a matrix function @xmath620 is well - defined as a rational function of @xmath55 with simple poles .
the poles of @xmath620 are exactly the poles of the corresponding potentials @xmath779 obtained from the base case of @xmath796 by iterated bcklund transformations
. * if @xmath777 is not a pole , then there is a neighborhood @xmath797 with @xmath798 such that for @xmath799 , the matrix @xmath620 * * is analytic in @xmath55 , * * is the solution of riemann - hilbert problem [ rhp : dslocalii ] , * * has an expansion for large @xmath597 of the form differentiable with respect to both @xmath55 and @xmath597 and uniform for @xmath799 , and * * is uniformly bounded for @xmath799 and @xmath597 in compact subsets of @xmath800 . * if @xmath777 is a pole , then riemann - hilbert problem [ rhp : dslocalii ] has no solution for @xmath785 ; however both @xmath801 and @xmath802 are regular at @xmath777 ( in the sense of taking limits from @xmath803 ) , and both @xmath788 and @xmath789 have simple zeros at @xmath785 .
since the poles are simple , the bound on the elements of @xmath620 ( and its inverse , since @xmath804 where defined ) can be strengthened to @xmath805 ( or equivalently @xmath806 ) uniformly on compact sets in the @xmath597-plane .
this improved bound also applies to all of the coefficients in the expansion and to the error term as well .
in particular , this implies that at @xmath807 , the functions @xmath808 , @xmath809 , and @xmath810 all have simple poles .
[ prop : rnlocgeneral ]
on one hand , it seems quite reasonable to expect that the global model @xmath811 defined by should provide a good approximation to @xmath384 , at least if the integer @xmath253 is properly chosen .
indeed , we have shown in an earlier paper @xcite that far from criticality a global model obtained by steps completely analogous to those we have followed here ( choice of an appropriate @xmath172-function , opening of lenses , pointwise approximation of jump matrices away from points of nonuniformity and the solution of a corresponding outer model problem , and exact solution of the jump conditions near points of nonuniformity yielding inner models that match well onto the outer model at disk boundaries ) leads directly to a riemann - hilbert problem for the error of small - norm type . on the other hand , in the present case
the formally large exponents @xmath812 were removed from the jump matrices in our construction of the inner model , and we may expect them to reappear when the mismatch between the inner and outer models is calculated on @xmath611 , contaminating the error estimates when @xmath436 is small but large compared with @xmath282 . in order to properly gauge our prospects for success , we introduce a new matrix unknown @xmath813 constructed from the old unknown @xmath384 and the explicit global model @xmath811 defined by as follows : @xmath814 we might expect to be able to prove that @xmath813 is a small perturbation of the identity matrix with the use of small - norm theory applied to the riemann - hilbert problem satisfied by this matrix function . to begin to determine the nature of the riemann - hilbert problem satisfied by @xmath813 ,
firstly note that @xmath813 is analytic for @xmath422 , since for such @xmath116 , @xmath815 has determinant one and satisfies exactly the same jump conditions as does @xmath384 . for @xmath816 ,
@xmath813 is analytic except on the arcs of the jump contour for @xmath384 illustrated in figure [ fig : critpcontour ] ; it can further be checked that since both @xmath384 and @xmath817 change sign across the contour segment @xmath387 , @xmath813 also extends analytically to this segment .
the jump contour for @xmath813 also contains @xmath611 , where @xmath384 is analytic but @xmath811 has a jump discontinuity stemming from the mismatch ( because @xmath597 is large but finite on @xmath611 ) between @xmath631 and @xmath406 .
the jump contour for @xmath813 is illustrated in figure [ fig : crittcontour ] .
of discontinuity of the sectionally analytic function @xmath813 . with the exception of the circle @xmath611 which is clockwise - oriented as shown , all other arcs are oriented exactly as components of the jump contour of @xmath384 . _ _ ]
let @xmath57 denote the jump contour for @xmath813 , that is , @xmath813 is analytic for @xmath818 , and it takes its boundary values in the classical sense on this contour from each component of the domain of analyticity .
also , since both @xmath384 and @xmath811 tend to the identity matrix as @xmath171 ( the former by hypothesis and the latter by the explicit formula for @xmath406 ) , we have @xmath819 as @xmath171 . for fixed @xmath236 , it turns out that for @xmath36 sufficiently close to criticality , @xmath820 as @xmath821 for @xmath822 , with the estimate being uniform with respect to @xmath720 on the specified contour arcs and @xmath36 near criticality .
indeed , exactly at criticality we have @xmath823 where @xmath334 is as defined in ( * ? ? ?
* section 4 ) and so by continuity with respect to @xmath6 and @xmath11 we will have near criticality that the jump matrix for @xmath377 is subject to the same error estimates as recorded in ( * ? ? ?
* proposition 5.1 ) generally valid away from criticality , where the neighborhoods @xmath824 and @xmath825 in the statement of the proposition are taken to coincide with @xmath408 . making the transformation from @xmath377 to @xmath384
we learn that the jump condition for @xmath384 is of the form @xmath826 uniformly for all @xmath720 in the jump contour pictured in figure [ fig : critpcontour ] with the exception of @xmath827 and on the ray @xmath828 .
but , @xmath811 satisfies exactly the same jump conditions as does @xmath384 within @xmath408 and in the interval @xmath829 .
the estimate on the jump conditions for @xmath813 asserted at the beginning of this paragraph then follow by exact computation . for @xmath830 ( recall that @xmath831 is a small open interval containing @xmath832 as shown in figure [ fig : critncontour ] ) , both @xmath384 and @xmath833 satisfy the same jump condition ,
from which it follows that for @xmath834 but @xmath835 , @xmath836^{-1 } = \sigma_2\mathbf{p}_-(\xi)\dot{\mathbf{p}}_{{m}-}^\mathrm{out}(\xi)^{-1}\sigma_2 = \sigma_2\mathbf{q}_{{m}-}(\xi)\sigma_2 .
\label{eq : econjjump}\ ] ] for @xmath209 , it follows from the fact that @xmath837 for @xmath834 that ( * ? ? ?
* proposition 3.3 ) holds , implying that @xmath384 satisfies @xmath838 . on the other hand , by construction @xmath811 satisfies the corresponding jump condition without error term . because @xmath811 is uniformly bounded with bounded inverse , by a calculation similar to we see that @xmath839 holds uniformly for @xmath840 . to compute the jump condition for @xmath813 across the negatively - oriented boundary @xmath611 of the disk @xmath408 , first use the fact that @xmath384 is continuous across @xmath611 to obtain @xmath841= \mathbf{q}_{{m}-}(\xi)\left[\dot{\mathbf{p}}_{m}^\mathrm{in}(\xi ) \dot{\mathbf{p}}_{m}^\mathrm{out}(\xi)^{-1}\right],\quad \xi\in\partial u.\ ] ] by substitution from and
, we see that @xmath842 e^{-\frac{1}{2}s\sigma_3/\epsilon_n}\epsilon_n^{(1 - 2{m})\sigma_3/6}d_n(\xi)^{-\sigma_3 } , \label{eq : qjumpdiscboundary}\end{gathered}\ ] ] for @xmath843 , where @xmath844 .
we therefore see that for @xmath36 close enough to criticality , @xmath813 satisfies the conditions of the following type of riemann - hilbert problem .
seek a matrix @xmath813 with the following properties : * * analyticity : * @xmath813 is analytic for @xmath818 , where @xmath57 is the contour ( independent of @xmath6 , @xmath11 , @xmath253 , and @xmath282 ) pictured in figure [ fig : crittcontour ] , and in each component of the domain of analyticity is uniformly hlder continuous up to the boundary . * * jump conditions : * the boundary values taken by @xmath813 on @xmath57 are related as follows . for @xmath845 with clockwise orientation , @xmath846 with jump matrix @xmath847 defined by the right - hand side of . for @xmath175 , @xmath848 where @xmath849 holds uniformly for @xmath209 and @xmath850 for @xmath830 .
for all remaining @xmath851 we have the uniform estimate @xmath852 as @xmath120 . * * normalization : * the matrix @xmath813 satisfies the condition @xmath853 all of the above jump matrix estimates are uniform for bounded @xmath253 .
[ rhp : critq ] note that since the jump matrix for @xmath813 is not specified here precisely , the conditions of riemann - hilbert problem [ rhp : critq ] are not sufficient to determine @xmath813 , but we shall see that they are enough to approximate @xmath813 with suitable accuracy as @xmath821 .
now @xmath854 is bounded away from zero for @xmath843 , so from the normalization condition on @xmath620 given as part of riemann - hilbert problem [ rhp : dslocalii ] and understood in the stronger sense of , and from the fact that @xmath855 and its reciprocal are uniformly bounded as @xmath821 for @xmath843 , we obtain @xmath856e^{-\frac{1}{2}s\sigma_3/\epsilon_n } \epsilon_n^{(1 - 2{m})\sigma_3/6}d_n(\xi)^{-\sigma_3},\quad \xi\in\partial u.\ ] ] recall that @xmath421 uniformly for @xmath857 .
it follows that the jump matrix for @xmath813 will be a small perturbation ( of order @xmath858 for some @xmath859 ) of the identity for @xmath843 if @xmath860 subject to this condition , we may formulate a riemann - hilbert problem for @xmath813 of small - norm type , that is , a problem that can be solved by iteration .
this calculation leads to the suggestion that different choices of the integer @xmath253 will be required to control the error in different regions of the @xmath861 parameter plane , an idea that has proven useful in at least two very recent works on related problems of semiclassical analysis for nonlinear waves . in the work of claeys and grava @xcite on the small - dispersion limit of the kdv equation near the `` trailing edge '' of the oscillatory wavepacket generated by a gradient catastrophe in the solution of the limiting burgers equation
, the authors find that different choices of an integer parametrizing an outer model problem are required as the rescaled distance from the trailing edge varies .
similarly , in the work of bertola and tovbis @xcite the semiclassical solution of the focusing nls equation is examined for points near the `` primary caustic '' or `` breaking curve '' generated from an elliptic umbilic catastrophe in the solution of the formal limiting whitham system , and it is again found that different choices of parametrices must be made as a certain rescaled distance from the caustic curve varies . in both of these problems , as will be the case here , the rescaling of distance involves logarithms of the small parameter to get the balance right , and the resulting approximate solutions have a `` solitonic '' nature ( in the kdv problem they are simple solitons while in the nls problem they are soliton superpositions forming breathers ) . to implement this suggestion , we will choose a value for @xmath253 and then define corresponding regions of the @xmath861 plane in which we can make use of this value of @xmath253 .
also , we will find it necessary to relax the small - norm condition in order to cover a large enough part of the @xmath861 plane with overlapping regions , and this means that the riemann - hilbert problem solved by the matrix @xmath813 will no longer be treatable by iteration , and instead we shall have to improve our inadequate model @xmath811 by building parametrices for @xmath813 , another technique used in the papers @xcite and @xcite ( also in the more recent work @xcite of bertola and tovbis where the solution is analyzed in a neighborhood of the elliptic umbilic catastrophe point itself ) . for each integer @xmath236 we define a shifted and scaled coordinate @xmath862 equivalent to @xmath436 by the relation @xmath863 let @xmath236 . for suitably small @xmath165 and arbitrary @xmath864 ,
set @xmath865 and @xmath866 [ def : sets ] the set @xmath867 is basically a horizontal strip with excluded _ notches _ located near the poles of @xmath791 and upward - extending _ teeth _ located near the zeros of @xmath791 .
similarly , @xmath868 is a horizontal strip with excluded notches located near the poles of @xmath792 and downward - extending teeth located near the zeros of @xmath792 . since the poles of @xmath242 and @xmath674 coincide ( although their zeros do not in general ) , that is , @xmath869 , the notches of @xmath867 align with those of @xmath868
this is very important , since we will make use of the solution @xmath620 of riemann - hilbert problem [ rhp : dslocalii ] for @xmath870 , and therefore we must exclude from this union points with @xmath55-values too close to points of @xmath869 at which @xmath620 fails to exist .
see figure [ fig : twomorestrips ] .
( pink ) and @xmath868 ( blue ) for @xmath871 . for this figure , @xmath872 , @xmath873 and @xmath874 .
note the alignment of the notches .
it is for @xmath861 in the union of these regions that we select the value @xmath875 in both the solution @xmath406 of the outer model problem and the solution @xmath620 of the inner model problem . _
_ ] since @xmath876 according to , it is clear that @xmath877 and @xmath878 .
it follows that while the sets @xmath867 and @xmath879 have disjoint interiors , their union is a horizontal strip : @xmath880 the two regions interlock near the singularities of @xmath881 ( equivalently the singularities of @xmath882 ) ; the teeth of @xmath867 fit in a one - to - one fashion into the notches of @xmath879 , and the teeth of @xmath879 fit in a one - to - one fashion into the notches of @xmath867 .
see figure [ fig : twostrips ] . in figure
[ fig : criticalregionsannotated ] we illustrate the tiling of a region of the @xmath861-plane with many such strips when @xmath872 ( the strips overlap if @xmath883 ) . ( pink ) and @xmath879 ( blue ) for @xmath871 .
their union is a closed horizontal strip in the @xmath861-plane . for this figure , @xmath872 , @xmath873 and @xmath874 . _ ] ( pink ) and @xmath884 ( blue ) , for the `` tiling '' case @xmath872 . _ _ ] this latter figure illustrates another feature of the @xmath872 tiling : since by the bcklund transformations , @xmath777 is a simple zero of @xmath789 if and only if @xmath777 is a simple pole of both @xmath791 and @xmath792 if and only if @xmath777 is a simple zero of @xmath788 , the tips of the teeth of @xmath885 are matched in a one - to - one fashion onto the tips of the teeth of @xmath879 .
we will use definition [ def : sets ] in the following context : we fix a bound @xmath886 and consider only indices @xmath236 with @xmath887 ; we then choose @xmath165 sufficiently small that for all @xmath236 with @xmath887 , no two real singularities of @xmath881 or @xmath888 are closer as @xmath889 , which is one indication of failure of uniformity of the analysis in this paper when @xmath890 .
indeed , for @xmath891 fixed and @xmath46 fixed one expects a different description of the sine - gordon dynamics for sufficiently small @xmath282 in terms of higher - genus riemann theta functions . ]
than @xmath892 .
this prevents neighboring teeth / notches from overlapping at their widest point ( corresponding to @xmath893 ) .
even with the choice of the integer @xmath253 corresponding to @xmath894 it turns out that riemann - hilbert problem [ rhp : critq ] can not generally be solved with the use of small - norm theory as the jump matrix @xmath847 is not close to the identity for @xmath843 , at least not uniformly for @xmath894 .
therefore , even though we may think of @xmath813 as an `` error '' matrix gauging the mismatch between the matrix function @xmath811 and the matrix function @xmath384 for which the former was introduced as a model , we must in general carry out further modeling of @xmath813 by creating an explicit model ( parametrix ) for it , and comparing the model with @xmath813 in hopes of finally arriving at a small - norm problem for the matrix ratio .
we now proceed to build such a _
parametrix for the error _ by a procedure referred to in some papers @xcite as _ improvement of the parametrices_. let @xmath895 with @xmath887 be fixed , and suppose that @xmath896 with @xmath53 bounded . in particular
, this implies that the coordinate @xmath862 defined by is subjected to the inequalities @xmath897 . using the coordinate @xmath862 , and recalling proposition [ prop : rnlocgeneral ] along with the expansion , we write the matrix @xmath847 defined by the right - hand side of in the form ( assuming only that @xmath55 is bounded ) @xmath898 \eta_{m}(\xi)^{-\sigma_3}\\ & \quad\quad{}\cdot\exp\left(-\frac{1}{2}\left(p_{m}+\frac{1}{3}\log(\epsilon_n^{-1})\right ) \sigma_3\right)d_n(\xi)^{-\sigma_3},\quad\xi\in\partial u \end{split } \label{eq : vqomegaplus}\ ] ] because @xmath854 is bounded away from zero on @xmath611 . since @xmath855 , @xmath899 , and
their reciprocals are all bounded on @xmath611 and @xmath900 for @xmath901 , @xmath902 \eta_{m}(\xi)^{-\sigma_3}\\ & \quad\quad{}\cdot\exp\left(-\frac{1}{2}\left(p_{m}+\frac{1}{3}\log(\epsilon_n^{-1})\right ) \sigma_3\right)d_n(\xi)^{-\sigma_3 } + { \mathcal{o}}\left(\frac{\epsilon_n^{2/3}e^{p_{m}}}{|y-\mathscr{p}({\mathcal{u}}_{m})|}\right ) \end{split } \label{eq : vqomegaplus1}\ ] ] holds for @xmath901 with @xmath55 bounded and @xmath843 .
now , we want to neglect all but the @xmath764-entry of the matrix @xmath903 .
since the matrix elements of @xmath641 and @xmath642 are @xmath805 , we may do this at the cost of adding a new error term of the form @xmath904 ( dominated by the diagonal entries ) .
this error term dominates the one already present in for @xmath901 as a consequence of the inequality @xmath905 .
therefore , @xmath906 where , defining @xmath907 the explicit function appearing in is @xmath908w(w)^{-2}\right ) .
\end{split } \label{eq : ynplusdef}\ ] ] note that @xmath909 extends from @xmath843 as a meromorphic function of @xmath910 whose only singularity is a double pole at @xmath393 , and that @xmath911 also depends parametrically on @xmath862 and @xmath55 .
since @xmath912 holds uniformly for @xmath843 , another way to write is @xmath913 the following three estimates hold uniformly for @xmath843 and @xmath55 bounded .
+ in the upward - pointing `` teeth '' of @xmath867 : @xmath914 in the part of @xmath867 resembling a `` strip with notches '' : @xmath915 elsewhere in @xmath867 : @xmath916 [ prop : vqerrorplus ] first suppose that @xmath917 ( so we are considering @xmath861 to lie in the upward - pointing `` teeth '' of @xmath867 ) and that @xmath55 is bounded . in this sub - region , @xmath918 is bounded away from zero because the teeth are localized near the zeros @xmath919 of @xmath791 .
therefore @xmath920 is bounded and @xmath921 , so @xmath922 holds uniformly for @xmath843 .
putting this estimate into gives @xmath923 this proves the estimate , and it also proves the estimate in the case that @xmath924 .
now suppose instead that @xmath925 ( so we are considering @xmath861 to lie in the complementary part of @xmath867 that resembles a `` strip with notches '' ) , and that @xmath55 is bounded .
since according to proposition [ prop : rnlocgeneral ] , both @xmath791 and @xmath926 have simple poles at the points of @xmath927 , it follows that in this sub - region , @xmath928 .
but in this subregion we also have the inequality @xmath929 , so with @xmath930 in fact we have @xmath931 .
putting this information into gives @xmath932 this proves , and it also proves in the case that @xmath930 .
the following estimate holds uniformly for @xmath843 and @xmath901 with @xmath55 bounded : @xmath933 while uniform , this estimate fails to be sharp except possibly at the vertical extremes of @xmath867 where @xmath934 .
[ cor : vqerrorplus ] as @xmath120 , the function @xmath935 is not uniformly small on @xmath611 as @xmath861 ranges over @xmath867 with @xmath55 bounded .
this implies that to approximate @xmath813 for @xmath896 , it will be necessary to build a parametrix for it that takes into account the explicit terms in .
let real numbers @xmath55 and @xmath862 be given , and assume that @xmath936 .
seek a matrix @xmath937 with the following properties : * * analyticity : * @xmath937 is analytic for @xmath938 , and is uniformly hlder continuous up to the boundary of its domain of analyticity . * * jump conditions : * the boundary values taken by @xmath937 on the contour @xmath939 are related as follows : @xmath940 and @xmath941 where @xmath611 is taken to be negatively ( clockwise ) oriented and @xmath911 is defined by . * * normalization : * the matrix @xmath937 satisfies the condition @xmath942 [ rhp : eparametrix ] we solve this problem uniquely for @xmath937 as follows .
firstly , define a related unknown @xmath943 as follows : @xmath944 clearly , @xmath943 is continuous across the contour @xmath611 .
now , @xmath602 is analytic in @xmath408 , but @xmath945 has a simple zero at @xmath417 , so @xmath943 generally has a double pole at @xmath393 . to handle the only remaining jump condition for @xmath946 , namely @xmath947 for @xmath834 , we introduce @xmath948 and set @xmath949 it is now evident that @xmath950 is a rational function of @xmath951 with simple poles at @xmath952 where @xmath953 , normalized to the identity at @xmath954 , and satisfying the symmetry relation @xmath955 . to fully characterize the singularities of @xmath950
, we will relate its principal part at @xmath956 to its regular part .
note that in a neighborhood of @xmath956 , @xmath957 has the laurent expansion @xmath958 where @xmath959 represents a function analytic at @xmath956 and where @xmath960 .
\end{split } \label{eq : jkplus}\ ] ] note also that @xmath961 and @xmath962 are both real when @xmath53 and @xmath963 .
let us write the laurent expansion of @xmath950 about @xmath956 in the form @xmath964 where @xmath965 represents a matrix function analytic and vanishing to second order at @xmath956 .
since @xmath937 is meant to be analytic at @xmath393 , from we see that the principal part of @xmath950 at @xmath956 is determined from its holomorphic part as follows .
the first column vanishes : @xmath966 and the second column satisfies the linear relations @xmath967 taking into account the symmetry condition @xmath955 and the notation of the principal part of the laurent expansion , the matrix @xmath950 necessarily can be expressed in partial fractions as follows : @xmath968 where we have also used .
evaluating the first column of this expression and its derivative at the regular point ( for the first column ) @xmath956 we find that , in the notation of the regular part of the laurent expansion , @xmath969 substituting these expressions into the relations yields a closed inhomogeneous linear system for the vectors @xmath970 and @xmath971 : @xmath972 this system has a unique solution : @xmath973^{2})\\ [ j_{m}^+]^{3}-16w_*^2j^+_{m}-32w_*^2sk^+_{m}\end{bmatrix}\\ \mathbf{d}^+_{m}&=\frac{4is}{d^+_{m}}\begin{bmatrix } 4w_*(2[j_{m}^+]^{3}+s[j_{m}^+]^{2}k^+_{m}-16w_*^2sk^+_{m})\\ [ j_{m}^+]^{4}+16w_*^2[j_{m}^+]^{2}+32w_*^2sj^+_{m}k^+_{m}-32w_*^3[k_{m}^+]^{2 } \end{bmatrix } , \end{split } \label{eq : bdplus}\ ] ] where the denominator @xmath974^{2})^2 + 64w_*^2(j^+_{m}+sk^+_{m})^2\ge 256w_*^4 \label{eq : denomplus}\ ] ] is strictly positive , because @xmath389 . substituting into the partial fraction representation completes the construction of @xmath950 . by writing @xmath975 and restricting @xmath950 to the upper half - plane we then recover @xmath943 , and then from we recover the solution @xmath937 of riemann - hilbert problem [ rhp : eparametrix ]
. for each fixed integer @xmath253 with @xmath887 , riemann - hilbert problem [ rhp : eparametrix ] has a unique and explicit solution @xmath937 satisfying @xmath976 where defined . also , @xmath937 is uniformly bounded for @xmath977 and @xmath901 with @xmath55 bounded as @xmath821 .
[ prop : parametrixplus ] the above construction clearly yields a solution of riemann - hilbert problem [ rhp : eparametrix ] , and uniqueness follows easily from the conditions of that problem via a liouville argument , which also shows that @xmath937 is unimodular . since @xmath978 for @xmath979 , the uniform bound on @xmath937 will follow from a corresponding bound on @xmath943 valid when @xmath116 is bounded away from the singularity at @xmath402 , or what is the same , a bound on @xmath950 valid when @xmath980 is bounded away from both @xmath981 and @xmath982 where @xmath953 . according to the partial - fraction representation of @xmath950 , it is therefore enough to establish that the elements of the vectors @xmath970 and @xmath971 remain bounded as @xmath120 uniformly with respect to @xmath901 with @xmath55 bounded . but from we see that the components of @xmath970 and @xmath971 really only depend on @xmath983 and @xmath984 ( since @xmath389 ) and while the latter can become unbounded , it is still obvious that @xmath970 and @xmath971 can not .
fix an integer @xmath253 with @xmath887 , and suppose that @xmath985 with @xmath53 bounded . by introducing the scaled and shifted coordinate @xmath986 defined by and analyzing the jump matrix @xmath847 for @xmath843 as in the beginning of [ sec : parametrixplus ]
, we see that in the current situation a lower - triangular model for the jump matrix is more appropriate .
in particular , defining @xmath987 and then @xmath988w(w)^{-2}\right ) , \label{eq : ynminusdef}\ ] ] ( note that @xmath989 extends from @xmath843 to a meromorphic function for @xmath990 whose only singularity is a double pole at @xmath393 ) we arrive at the following .
the following three estimates hold uniformly for @xmath843 and @xmath55 bounded .
+ in the downward - pointing `` teeth '' of @xmath868 : @xmath991 in the part of @xmath868 resembling a `` strip with notches '' : @xmath992 elsewhere in @xmath868 : @xmath993 [ prop : vqerrorminus ] the proof of this statement is virtually the same as that of proposition [ prop : vqerrorplus ] .
the analogue for @xmath868 of corollary [ cor : vqerrorplus ] is then this result .
the following estimate holds uniformly for @xmath843 and @xmath985 with @xmath55 bounded : @xmath994 while uniform , this estimate fails to be sharp except possibly at the vertical extremes of @xmath868 where @xmath995 .
[ cor : vqerrorminus ] based on this approximation result for @xmath847 , we build a parametrix for @xmath813 to satisfy the following problem .
let real numbers @xmath55 and @xmath986 be given , and assume that @xmath996 .
seek a matrix @xmath997 with the following properties : * * analyticity : * @xmath997 is analytic for @xmath938 , and is uniformly hlder continuous up to the boundary of its domain of analyticity .
* * jump conditions : * the boundary values taken by @xmath997 on the contour @xmath939 are related as follows : @xmath998 and @xmath999 where @xmath611 is taken to be negatively ( clockwise ) oriented and @xmath989 is defined by . * * normalization : * the matrix @xmath997 satisfies the condition @xmath1000 [ rhp : eparametrixii ] this problem is solved completely analogously to riemann - hilbert problem [ rhp : eparametrix ] .
namely , we introduce a matrix @xmath1001 by @xmath1002 where we define @xmath1003 by @xmath1004 then , @xmath1001 is a rational function with @xmath1005 and double poles at @xmath952 where @xmath953 only , satisfying the symmetry relation @xmath1006 .
the singularities of @xmath1001 are characterized by first noting the laurent expansion of @xmath1007 about @xmath956 : @xmath1008 where @xmath959 denotes a function analytic at @xmath956 and where @xmath1009 \end{split } \label{eq : jkminus}\ ] ] are real - valued coefficients when both @xmath53 and @xmath1010 .
the laurent series of @xmath1001 about @xmath956 has the form @xmath1011 where @xmath965 represents an analytic function vanishing to second order at @xmath956 .
since @xmath997 is meant to be analytic at @xmath393 , from we learn that the principal part of the second column of @xmath1001 vanishes : @xmath1012 and the principal part of the first column satisfies the relations @xmath1013 using the symmetry condition @xmath1006 and taking into account , we see that the partial fraction expansion of @xmath1001 necessarily has the form : @xmath1014 by evaluating the second column of this expression and its derivative at @xmath956 we can express @xmath1015 and @xmath1016 in terms of @xmath1017 and @xmath1018 : @xmath1019 combining this with yields a closed inhomogeneous linear system of equations for the components of the vectors @xmath1017 and @xmath1018 : @xmath1020 as before , this system has a unique solution : @xmath1021 ^ 3 - 16w_*^2j_{m}^--32w_*^2sk_{m}^-\\ 4w_*(16w_*^2 + 2sj_{m}^-k_{m}^-+3[j_{m}^-]^2 ) \end{bmatrix}\\ \mathbf{c}_{m}^-&=\frac{4is}{d_{m}^- } \begin{bmatrix } [ j_{m}^-]^4 + 16w_*^2[j_{m}^-]^2 + 32w_*^2sj_{m}^-k_{m}^--32w_*^3[k_{m}^-]^2\\ 4w_*(2[j_{m}^-]^3+s[j_{m}^-]^2k_{m}^--16w_*^2sk_{m}^- ) \end{bmatrix } \end{split } \label{eq : acminus}\ ] ] where the denominator @xmath1022 ^ 2)^2 + 64w_*^2(j_{m}^-+sk_{m}^-)^2\ge 256w_*^4 \label{eq : denomminus}\ ] ] is again strictly positive because @xmath389 .
this essentially completes the construction of @xmath997 solving riemann - hilbert problem [ rhp : eparametrixii ] .
the analogue of proposition [ prop : parametrixplus ] in this case is the following .
( its proof is also nearly the same . ) for each fixed integer @xmath253 with @xmath887 , riemann - hilbert problem [ rhp : eparametrixii ] has a unique and explicit solution @xmath997 satisfying @xmath1023 where defined . also , @xmath997 is uniformly bounded for @xmath977 and @xmath985 with @xmath55 bounded as @xmath821 .
[ prop : parametrixminus ] the _ error _ in approximating @xmath813 by the parametrix @xmath1024 is the matrix @xmath1025 . since the jump contour @xmath611 for @xmath1024 is a subset of the jump contour @xmath57 of @xmath813 , and since @xmath1026 for @xmath1027 , it is clear that @xmath1028 is , like @xmath813 , analytic for @xmath818 , and takes its boundary values on @xmath57 in the same classical sense as we require of @xmath813 .
also , @xmath1029 as @xmath171 since this is true of both factors .
we now consider the jump conditions satisfied by @xmath1028 as a consequence of those known to be satisfied by @xmath813 from the statement of riemann - hilbert problem [ rhp : critq ] and those satisfied by @xmath1024 from either riemann - hilbert problem [ rhp : eparametrix ] or [ rhp : eparametrixii ] . if @xmath175 , then @xmath1030 .
\end{split}\ ] ] if we write the matrix in brackets above as @xmath1031 , we see that @xmath1032 wherever @xmath1033 ( that is , everywhere except in the interval @xmath831 near @xmath832 ) .
using proposition [ prop : parametrixplus ] and the fact that @xmath849 for @xmath209 , we see that under the condition that @xmath901 with @xmath55 bounded we also have @xmath1034 for @xmath209 .
similarly , proposition [ prop : parametrixminus ] shows that under the condition that @xmath985 with @xmath55 bounded the same estimate holds for @xmath1035 .
if @xmath1036 , then the parametrix has no discontinuity , so @xmath1037 therefore according to proposition [ prop : parametrixplus ] , if @xmath901 with @xmath55 bounded , we will have @xmath1038 for such @xmath720 .
analogously , proposition [ prop : parametrixminus ] shows that @xmath985 with @xmath55 bounded implies that @xmath1039 for such @xmath720 . finally , if @xmath843 , then @xmath1040 if @xmath901 with @xmath55 bounded , then according to propositions [ prop : vqerrorplus ] and [ prop : parametrixplus ] we find @xmath1041 where the size of the error term is @xmath1042 at the worst ( according to corollary [ cor : vqerrorplus ] ) but is more typically smaller according to the more precise estimates enumerated in proposition [ prop : vqerrorplus ] .
similarly , for @xmath843 , @xmath1043 so if @xmath985 with @xmath55 bounded , then from propositions [ prop : vqerrorminus ] and [ prop : parametrixminus ] we find @xmath1044 with a similar characterization of the error term . the dominant terms in the deviation of the jump matrix from the identity come from @xmath843 .
therefore , we observe that the matrix @xmath1045 satisfies the conditions of the following type of riemann - hilbert problem .
let @xmath253 be an integer with @xmath887 , and suppose that @xmath894 with @xmath55 bounded .
seek a matrix @xmath1028 with the following properties : * * analyticity : * @xmath1028 is analytic for @xmath818 , and is uniformly hlder continuous up to the boundary of its domain of analyticity . * * jump conditions : * the boundary values taken by @xmath1028 on the contour @xmath57 are related as follows : @xmath1046 and @xmath1047 where the following estimates hold for the jump matrix .
firstly , we have @xmath1048 for @xmath834 with @xmath1049 sufficiently large , so @xmath1050 is compactly supported and vanishes identically in a neighborhood of @xmath1051 .
next , we have @xmath1052 , or , more precisely , @xmath1053 , where for @xmath901 , @xmath1054 and for @xmath985 , @xmath1055 * * normalization : * the matrix @xmath1028 satisfies the condition @xmath1056 [ rhp : error ] suppose that @xmath253 is an integer with @xmath887 , and that @xmath894 with @xmath55 bounded
. then for sufficiently small @xmath282 , there exists a unique solution @xmath1028 of riemann - hilbert problem [ rhp : error ] .
the solution has expansions for small and large @xmath116 of the form @xmath1057 + [ \fouridx{0}{1}{\pm}{{m}}{\mathbf{e}}](-w)^{1/2 } + { \mathcal{o}}(w),&\quad w\to 0\\ \mathbb{i}+[\fouridx{\infty}{1}{\pm}{{m}}{\mathbf{e}}](-w)^{-1/2 } + { \mathcal{o}}(w^{-1}),&\quad w\to\infty , \end{cases } \label{eq : eexpand}\ ] ] and we have the estimates @xmath1058=\mathbb{i}+ { \mathcal{o}}(e_{m}^\pm(y , s;\epsilon_n)),\quad [ \fouridx{0}{1}{\pm}{{m}}{\mathbf{e}}]={\mathcal{o}}(e_{m}^\pm(y , s;\epsilon_n)),\quad\text{and}\quad [ \fouridx{\infty}{1}{\pm}{{m}}{\mathbf{e}}]={\mathcal{o}}(e_{m}^\pm(y , s;\epsilon_n))\ ] ] holding uniformly for @xmath894 with @xmath55 bounded , where @xmath1059 and @xmath1060 are defined by and respectively .
[ prop : error ] by the substitution @xmath1061 for @xmath1062 and @xmath1063 for @xmath1064 , we translate the conditions of riemann - hilbert problem [ rhp : error ] into a list of conditions satisfied by @xmath1065 .
this `` unfolding '' of @xmath116 to the @xmath980-plane implies that @xmath1065 is analytic on the complement of a compact contour @xmath1066 , along which it satisfies jump conditions of the form @xmath1067 where @xmath1068 satisfies the uniform estimate @xmath1069 and @xmath1070 is defined by for @xmath901 and by for @xmath985 .
since these estimates are all dominated by @xmath1042 , for sufficiently small @xmath282 , the small - norm theory of matrix riemann - hilbert problems in , say , the @xmath1071-sense implies that @xmath1065 may be constructed by iteration applied to a suitable singular integral equation , and it follows from that theory that an estimate of the form @xmath1072 actually holds whenever @xmath1073 is a compact set disjoint from the jump contour @xmath1066 . in particular
, @xmath1065 is analytic on some compact neighborhood @xmath1073 of the origin @xmath1074 , so it has a taylor expansion that we write in the form @xmath1075 + [ \fouridx{0}{1}{\pm}{{m}}{\mathbf{f}}]{\tau}+ { \mathcal{o}}({\tau}^2),\quad { \tau}\to 0,\ ] ] and by the uniform estimate on @xmath1076 we have @xmath1077=[\fouridx{0}{0}{\pm}{{m}}{\mathbf{f}}]=\mathbb{i } + { \mathcal{o}}(\|\mathbf{f}_{m}^\pm-\mathbb{i}\|_{l^\infty(k)})$ ] .
applying the cauchy integral formula to @xmath1076 on a small loop in @xmath1073 surrounding the origin then shows that @xmath1078=i[\fouridx{0}{1}{\pm}{{m}}{\mathbf{f}}]= { \mathcal{o}}(\|\mathbf{f}_{m}^\pm-\mathbb{i}\|_{l^\infty(k)})$ ] .
also , @xmath1065 tends to the identity as @xmath1079 and it is analytic for sufficiently large @xmath1080 so it has a laurent expansion of the form @xmath1081{\tau}^{-1}+{\mathcal{o}}({\tau}^{-2}),\quad { \tau}\to\infty,\ ] ] now we can take @xmath1073 as an annulus enclosing a large circle . applying the cauchy integral formula on such a circle to @xmath1076
we find that @xmath1082=-i[\fouridx{\infty}{1}{\pm}{{m}}{\mathbf{f}}]={\mathcal{o}}(\|\mathbf{f}_{m}^\pm-\mathbb{i}\|_{l^\infty(k)})$ ] .
this completes the proof . the magnitude of the expansion coefficients of @xmath1083 and of @xmath1084 in the interlocking regions @xmath867 and @xmath879 respectively is illustrated in figure [ fig : eminusiestimates ] . in particular
it is clear that the order of accuracy is @xmath228 except very close to the common boundaries between teeth and notches .
moreover , the order of accuracy is worst right along these common boundaries where it is @xmath1085 which in turn is largest just at the tips of the teeth ( @xmath1086 ) , where the estimate reduces to @xmath1042 . for expansion coefficients of @xmath1087 with @xmath901 and @xmath1088 for expansion coefficients of @xmath1089 with @xmath1090 .
the boundaries of the two interlocking regions are drawn with solid black curves . _ _ ]
beginning with the matrix @xmath373 we may now express @xmath152 in terms of explicitly known quantities and the error matrix @xmath1028 characterized in proposition [ prop : error ] .
for example , if @xmath1091 is sufficiently small then for @xmath1092 we have @xmath1093 while for @xmath1094 we have the same formula just replacing @xmath384 and @xmath401 by @xmath1095 and @xmath1096 respectively .
but from we see that @xmath1097 where @xmath953 . therefore , in terms of the expansion of @xmath1028 for small @xmath1091 , and the corresponding expansion of @xmath1024 : @xmath1098 + [ \fouridx{0}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}](-w)^{1/2 } + { \mathcal{o}}(w),\quad w\to 0 , \label{eq : qwsmall}\ ] ] we find the following formulae for certain expansion coefficients of @xmath152 ( see ): @xmath1099(x , t)=[\fouridx{0}{0}{\pm}{{m}}{\mathbf{e}}][\fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}](-1)^{{m}+1}i\sigma_2 , \label{eq : n00}\ ] ] and @xmath1100(x , t)^{-1}[\fouridx{0}{1}{\mathfrak{g}}{n}{\mathbf{n}}](x , t)= \sigma_2[\fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]^{-1 } [ \fouridx{0}{0}{\pm}{{m}}{\mathbf{e}}]^{-1 } [ \fouridx{0}{1}{\pm}{{m}}{\mathbf{e } } ] [ \fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]\sigma_2 + \sigma_2[\fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]^{-1 } [ \fouridx{0}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]\sigma_2 - \frac{1 - 2m}{s}\sigma_3 .
\label{eq : n01}\ ] ] similarly , if @xmath1091 is sufficiently large , then we have @xmath1101 and from we have @xmath1102 so in terms of the expansion of @xmath1028 for large @xmath1091 , and the corresponding expansion of @xmath1024 : @xmath1103(-w)^{-1/2 } + { \mathcal{o}}(w^{-1}),\quad w\to\infty , \label{eq : qwlarge}\ ] ] we obtain an exact formula for the coefficient @xmath1104(x , t)$ ] ( see ): @xmath1105(x , t)=[\fouridx{\infty}{1}{\pm}{{m}}{\mathbf{e } } ] + [ \fouridx{\infty}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q } } } } ] + ( 1 - 2{m})s\sigma_3 .
\label{eq : ninfty1}\ ] ]
now suppose that @xmath36 are such that @xmath1108 with @xmath55 bounded
. then we can apply proposition [ prop : error ] and the boundedness of @xmath1109=\dot{\mathbf{q}}_{m}^\pm(0)$ ] and its inverse as guaranteed by propositions [ prop : parametrixplus ] and [ prop : parametrixminus ] to obtain : @xmath1110 all holding for @xmath894 with @xmath55 bounded , where @xmath1111_{12}\\ \dot{s}_{n,{m}}^\pm(x , t)&:=(-1)^{m}[\fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{22}\\ \dot{g}_{n,{m}}^\pm(x , t)&:=[\fouridx{\infty}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{12 } - \left[[\fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]^{-1 } [ \fouridx{0}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]\right]_{21 } = [ \fouridx{\infty}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{12}+ [ \fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{21 } [ \fouridx{0}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{11 } - [ \fouridx{0}{0}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{11 } [ \fouridx{0}{1}{\pm}{{m}}{\smash{\dot{\mathbf{q}}}}]_{21}. \end{split}\ ] ] ( in the last equality we used the fact that @xmath1109= \dot{\mathbf{q}}_{m}^\pm(0)$ ] has determinant one . ) since @xmath1112 , the necessary expansion coefficients of @xmath1024 ( see and ) can be obtained directly from the partial - fractions expansions and . in this way
, we get @xmath1113&=\mathbb{i}+\frac{1}{w_*}\left(i\sigma_2[\mathbf{b}_{m}^+-is\mathbf{d}_{m}^+ ] , \mathbf{b}_{m}^+-is\mathbf{d}_{m}^+\right)\\ [ \fouridx{0}{1}{+}{{m}}{\smash{\dot{\mathbf{q}}}}]&=\frac{1}{w_*s } \left(-i\sigma_2[2\mathbf{b}_{m}^+-is\mathbf{d}_{m}^+ ] , 2\mathbf{b}_{m}^+-is\mathbf{d}_{m}^+\right)\\ [ \fouridx{\infty}{1}{+}{{m}}{\smash{\dot{\mathbf{q}}}}]&=-i\left(-i\sigma_2\mathbf{d}_{m}^+,\mathbf{d}_{m}^+\right ) , \end{split}\ ] ] implying that @xmath1114\\ \dot{s}_{n,{m}}^+(x , t)&=(-1)^{m}\left[1+\frac{b_{{m},2}^+}{w_*}+\frac{id_{{m},2}^+}{s}\right]\\ \dot{g}_{n,{m}}^+(x , t)&=\frac{1}{w_*^2}\left[iw_*(1-w_*)d_{{m},1}^++2sb_{{m},1}^++ib_{{m},1}^+d_{{m},2}^+-ib_{{m},2}^+d_{{m},1}^+\right ] , \end{split}\ ] ] where the components of the vectors @xmath1115 and @xmath1116 are given by with and we recall that @xmath953 . in the same way one sees that @xmath1117&=\mathbb{i}+\frac{1}{w _ * } \left(\mathbf{a}_{m}^--is\mathbf{c}_{m}^-,-i\sigma_2[\mathbf{a}_{m}^--is\mathbf{c}_{m}^-]\right)\\ [ \fouridx{0}{1}{-}{{m}}{\smash{\dot{\mathbf{q}}}}]&=\frac{1}{w_*s } \left(2\mathbf{a}_{m}^--is\mathbf{c}_{m}^-,i\sigma_2[2\mathbf{a}_{m}^--is\mathbf{c}_{m}^-]\right)\\ [ \fouridx{\infty}{1}{-}{{m}}{\smash{\dot{\mathbf{q}}}}]&=-i\left(\mathbf{c}_{m}^-,i\sigma_2\mathbf{c}_{m}^-\right ) , \end{split}\ ] ] implying that @xmath1118\\ \dot{s}_{n,{m}}^-(x , t)&=(-1)^{m}\left[1+\frac{a_{{m},1}^-}{w_*}+\frac{ic_{{m},1}^-}{s}\right]\\ \dot{g}_{n,{m}}^-(x , t)&=\frac{1}{w_*^2}\left[iw_*(1-w_*)c_{{m},2}^-+2sa_{{m},2}^- + ia_{{m},2}^-c_{{m},1}^--ia_{{m},1}^-c_{{m},2}^-\right ] , \end{split}\ ] ] where @xmath1017 and @xmath1018 are given by with and we have @xmath953 . finally , substituting from and
gives @xmath1119 ^ 2 + 1}\\ \dot{s}_{n,{m}}^\pm(x , t)&=(-1)^{m+1}\frac{[r_{m}^\pm]^2 - 1}{[r_{m}^\pm]^2 + 1}\\ \dot{g}_{n,{m}}^\pm(x , t)&=16w_*\frac{[j_{m}^\pm]^2k_{m}^\pm+2s[j_{m}^\pm]^3-w_*[j_{m}^\pm]^2k_{m}^\pm+32w_*sj_{m}^\pm-16w_*^2k_{m}^\pm+16w_*^3k_{m}^\pm}{([j_{m}^\pm]^2 + 16w_*^2)^2 + 64w_*^2(j_{m}^\pm+sk_{m}^\pm)^2 } , \end{split } \label{eq : dotcsgplusminus}\ ] ] where @xmath1120 ^ 2 + 16w_*^2 } , \label{eq : rplusminus}\ ] ] and where @xmath983 and @xmath984 are defined by while @xmath1121 and @xmath1122 are defined by .
[ [ proof - of - theoremthmtzero - initial - accuracy - of - fluxon - condensates . ] ] proof of theorem [ thm : tzero ] ( initial accuracy of fluxon condensates ) .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the fluxon condensate @xmath97 is an even function of @xmath6 , so it suffices to consider @xmath201 .
the line segment @xmath45 with @xmath226 is mapped under @xmath1123 into the union @xmath1124 , assuming that @xmath282 sufficiently small .
more specifically , it is mapped into the part of @xmath1124 with @xmath1125 and @xmath1126 . using the facts that @xmath1127 and
@xmath1128 , and hence also @xmath1129 , it is easy to see from the definitions and ( using also @xmath1130 ) that for @xmath45 and @xmath226 , @xmath1131 while @xmath1132 where @xmath953 . according to we then have @xmath1133 it follows from these facts and that for such @xmath6 and @xmath45 , @xmath1134 @xmath1135 and @xmath1136 furthermore , since @xmath402 is an analytic function of @xmath36 with @xmath593 , we have @xmath1137 .
therefore , we may equivalently write the above formulae in the form @xmath1138 also , since the function @xmath4 in the initial data is differentiable at @xmath213 and @xmath1139 , we can also write @xmath1140 finally , recalling , the fact that @xmath1141 for @xmath1142 with @xmath55 bounded and @xmath1126 , and the fact that @xmath1143 for @xmath1144 with @xmath55 bounded and @xmath1126 , we arrive at @xmath1145 which completes the proof of the theorem .
firstly , we wish to simplify the expressions for @xmath1146 and @xmath1147 given in by localizing near the critical point @xmath1148 , which allows us to write the coordinates @xmath540 and @xmath541 explicitly in terms of @xmath6 and @xmath11 at the cost of some small error terms according to . by analogy with the definition of @xmath862 for general @xmath236 ,
we define a shifted and scaled coordinate @xmath1149 equivalent to @xmath11 by the relation @xmath1150 recall the spatial coordinate @xmath235 exactly proportional to @xmath428 as defined by .
we will now show that there is essentially no additional cost ( beyond that already introduced in approximating @xmath1028 ) in explicitly introducing @xmath428 and @xmath11 as coordinates .
suppose that @xmath901 with @xmath55 bounded and @xmath253 fixed .
then @xmath1151 where @xmath1152 ^ 2 + 1}\quad\text{and}\quad \ddot{s}_{n,{m}}^+(x , t):=(-1)^{m+1}\frac{[\dot{r}_{m}^+]^2 - 1}{[\dot{r}_{m}^+]^2 + 1 } \label{eq : csddotplus}\ ] ] and where @xmath1153 similarly , suppose that @xmath985 with @xmath55 bounded and @xmath253 fixed .
then @xmath1154 where @xmath1155 ^ 2 + 1}\quad\text{and}\quad \ddot{s}_{n,{m}}^-(x , t):=(-1)^{m+1}\frac{[\dot{r}_{m}^-]^2 - 1}{[\dot{r}_{m}^-]^2 + 1 } \label{eq : csddotminus}\ ] ] and where @xmath1156 [ prop : csapproximate ] first , we localize near criticality by isolating the terms in @xmath1157 depending `` slowly '' on @xmath6 and @xmath11 and replacing @xmath6 with @xmath213 and @xmath11 with zero in these terms .
the `` slowly varying '' quantities are @xmath1158 and the following : @xmath1159 recalling the definitions of @xmath1160 and @xmath1161 given by and , one obtains from the exact expressions @xmath1162 ^ 2 + 16w_*^2},\quad\text{where}\quad b_{{m},12}(y)=2h_{m}(y){\mathcal{u}}_{m}(y)-{\mathcal{u}}_{m}'(y)\\ r_{m}^-&=\frac{\beta_{m}^-e^{-p_{{m}-1}}{\mathcal{v}}_{m}(y)+\gamma_{m}^-\epsilon_n^{1/3}e^{-p_{{m}-1}}b_{{m},21}(y)}{[\alpha_{m}^-\epsilon_n^{1/3}e^{-p_{{m}-1}}b_{{m},21}(y)]^2 + 16w_*^2},\quad\text{where}\quad b_{{m},21}(y)={\mathcal{v}}_{m}'(y)-2h_{m}(y){\mathcal{v}}_{m}(y ) .
\end{split}\ ] ] now , for each @xmath236 , @xmath1163 , @xmath1164 , @xmath1165 , and @xmath402 are all independent of @xmath282 , depending only on @xmath36 .
moreover , they are all analytic functions of @xmath36 near criticality . computing their limits at criticality using and together with and
we obtain : @xmath1166 and for @xmath402 we have the corresponding expression .
since by assumption @xmath253 and @xmath55 are bounded , it follows immediately that @xmath1167 and that @xmath1168 .
then , since the jacobian matrix @xmath1169 is diagonal and invertible at criticality according to , it also follows that @xmath226 and @xmath227 , and consequently all error terms of the form @xmath1170 may be replaced with @xmath1171 .
therefore , @xmath1172 ^ 2 + 16(1+{\mathcal{o}}(\epsilon_n^{2/3}))}\\ r_{m}^-&=\frac{-2^{4{m}}\nu^{2(m-1)/3}e^{-p_{{m}-1}}{\mathcal{v}}_{m}(y)(1+{\mathcal{o}}(\epsilon_n^{2/3 } ) ) + \epsilon_n^{1/3}e^{-p_{{m}-1 } } b_{{m},21}(y){\mathcal{o}}(\epsilon_n^{2/3 } ) } { [ -2^{4({m}-1)}\nu^{-1 + 2{m}/3}\epsilon_n^{1/3}e^{-p_{{m}-1}}b_{{m},21}(y)(1+{\mathcal{o}}(\epsilon_n^{2/3}))]^2 + 16(1+{\mathcal{o}}(\epsilon_n^{2/3}))}. \end{split } \label{eq : rpmind_intermediate}\ ] ] the denominators are bounded away from zero and , according to proposition [ prop : rnlocgeneral ] , @xmath1173 and @xmath1174 have simple poles at the points of @xmath869 implying that @xmath1175 therefore , the formulae can be written in the form @xmath1176 the explicit terms we have retained in @xmath1157 have the same form as @xmath1177 defined by and but written in terms of the `` fast '' variables @xmath862 , @xmath986 , and @xmath55 instead of @xmath1149 , @xmath1178 , and @xmath235 .
but , recalling the definition we deduce from that @xmath1179 and from this it follows ( using and the fact that the denominators are bounded away from zero ) that @xmath1180 also , from the definition , the definition @xmath1181 and , we see that @xmath1182 since the poles are simple for @xmath242 and @xmath674 , @xmath1183 now , @xmath1184 for @xmath870 , so @xmath1185 so that @xmath1186 and similarly @xmath1187 . for @xmath1188 and @xmath1189
the poles are double , and it follows by the same reasoning that @xmath1190 and @xmath1191 . from these considerations we have that @xmath1192 where @xmath1193 is defined by and @xmath1194 is defined by . by comparing with the definition of @xmath1059 for @xmath901 and the definition of @xmath1060 for @xmath985
we see that @xmath1195 finally , we insert the estimate into @xmath1146 and @xmath1147 given by noting that derivatives of the rational expressions in @xmath1157 that appear are uniformly bounded to complete the proof of the proposition .
we are now in a position to define the multiscale asymptotic formulae for @xmath1197 and @xmath1198 mentioned in the statement of the theorem : @xmath1199 using we find that @xmath1200 hold whenever @xmath36 are such that @xmath1201 .
the exponent of @xmath1202 is the worst - case exponent as explained in the paragraph following the proof of proposition [ prop : error ] . on the other hand ,
it is clear that this error estimate is uniformly valid whenever @xmath304 is bounded and @xmath1203 is bounded , or as is the same , whenever @xmath227 and @xmath226 .
this completes the proof of theorem [ thm : main ] .
choose any integer @xmath304 with @xmath1204 , and let @xmath1205 be fixed .
suppose that @xmath36 lies in the rectangle @xmath1206 defined by the inequalities @xmath1207 and @xmath254 . because @xmath1208 and @xmath1209 , for any _ strictly positive _ choice of the parameter @xmath864 entering into definition [ def : sets ] , the image @xmath1210 of @xmath1206 in the @xmath861-plane under the mapping @xmath1181 and @xmath532 will be contained in the horizontal strip @xmath1211 ( see ) with @xmath55 bounded , as long as @xmath282 is sufficiently small . also , for @xmath36 mapping into @xmath1212 ( respectively into @xmath1213 ) we have the uniform estimate @xmath1214 ( respectively we have the uniform estimate @xmath1215 ) .
moreover , given an interval @xmath263 $ ] with @xmath1216 on which @xmath283 is bounded , if the parameter @xmath165 entering into definition [ def : sets ] is chosen sufficiently small , the sub - rectangle @xmath1217 defined by the inequalities @xmath1218 ( recall the definition of @xmath235 in terms of @xmath428 ) and @xmath254 will be mapped into a subregion @xmath1219 of @xmath1210 in which both inequalities @xmath1220 and @xmath1221 hold .
it then follows that for @xmath36 mapping into @xmath1222 , the estimate @xmath1223 holds uniformly while for @xmath36 mapping into @xmath1224 , the estimate @xmath1225 holds uniformly .
see also figure [ fig : eminusiestimates ] .
suppose that @xmath36 corresponds to a point @xmath901 .
we wish to neglect the first term in the denominator of @xmath1193 as defined by ; we write @xmath1193 in the form @xmath1226 where @xmath281 is defined in terms of @xmath11 , @xmath282 , @xmath215 , and @xmath1227 by .
note that since @xmath1228 and @xmath1229 , and since @xmath1230 for @xmath901 , the upper bound for @xmath58 may be replaced with @xmath1231 .
therefore , as pointed out in the proof of proposition [ prop : csapproximate ] , @xmath1232 uniformly for @xmath901 , although it is typically smaller pointwise . by using this result in and recalling , it follows that whenever @xmath1233 maps into @xmath867 , @xmath1234 as shown at the beginning of the proof , the error terms involving @xmath1235 are either of magnitude @xmath1042 and in particular @xmath1236 , or , if @xmath1237 , of smaller magnitude @xmath228 .
the remaining error terms are bounded , and they are negligible unless _ both _ @xmath1238 and @xmath58 fails to be small .
but since @xmath1239 and @xmath901 , we see from definition [ def : sets ] that @xmath58 fails to be small only if @xmath1240 , or equivalently , @xmath1241 .
now , under this condition we have from that @xmath1242 , so to have @xmath1238 requires that @xmath260 .
moreover , if @xmath1237 and its image @xmath1243 lies in @xmath867 , @xmath1244 and @xmath1245 .
the inequality @xmath1246 then implies that if @xmath1237 with image in @xmath867 , then @xmath1247 and @xmath1248 are both dominated by the @xmath228 estimate of the error terms involving @xmath1235 .
now suppose instead that @xmath36 corresponds to a point @xmath1090 . as before ,
we wish to neglect the first term in the denominator of @xmath1249 as defined by , so we write @xmath1249 in the form ( using the identity @xmath1250 ) @xmath1251 again , the error bound for @xmath1252 may be replaced with @xmath1253 and so @xmath1254 holds uniformly for @xmath1090 ; it then follows by using this result in and recalling , that whenever @xmath1233 maps into @xmath879 , @xmath1255 in particular , we note that the leading terms are _ identical _ with those arising from the analysis in the case that the image of @xmath36 lies in @xmath867 . by completely analogous reasoning as in that case
, one learns that the sum of error terms is @xmath1236 unless both @xmath1238 and also @xmath1252 fails to be small , which occurs in this case only if @xmath1256 , or equivalently , if @xmath1257 . under this condition
, the definition implies that @xmath1258 , so to have @xmath1238 requires that @xmath262 .
moreover , if @xmath1237 with image in @xmath879 , then both error terms are @xmath228 .
this completes the proof of theorem [ thm : kinkapprox ] .
[ [ proof - of - theoremthmgrazingapprox - grazing - kink - collisional - asymptotics . ] ] proof of theorem [ thm : grazingapprox ] ( grazing kink collisional asymptotics ) .
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ recall that @xmath1259 is a simple zero of @xmath1260 , and hence also a simple zero of @xmath1261 .
note that from the recurrence relation it follows that @xmath1262 .
since @xmath1228 and @xmath1263 for the range of @xmath6 and @xmath11 under consideration , the region in the @xmath36-plane defined by the inequalities @xmath1264 is , for both @xmath308 and @xmath282 sufficiently small , mapped into the union @xmath1265 in the @xmath861-plane , where @xmath1243 . in particular ,
each point of the image lies either in the upward - pointing `` tooth '' of @xmath885 centered at @xmath1266 or in the downward - pointing `` tooth '' of @xmath879 centered at @xmath1266 .
it follows that for the points whose images lie in @xmath885 , the error term @xmath1267 is @xmath1042 while for the points whose images lie in @xmath879 , the error term @xmath1268 is also @xmath1042 so that in particular these error terms arising in will always tend to zero with @xmath282 .
moreover , the inequality @xmath316 implies that @xmath1269 , from which it follows that under this condition @xmath36 will be mapped into the parts of @xmath885 and @xmath879 ( the centers of two matching teeth ) in which the better error estimate of @xmath228 holds for @xmath1270 and @xmath1271 respectively .
suppose that the image of @xmath36 in the @xmath861-plane lies in the upward - pointing tooth of @xmath885 centered at @xmath1266 . in this part of the image , the uniform estimate @xmath1272 holds . by taylor expansion , @xmath1273 where the spatial coordinate @xmath1274 is defined by .
it follows from using @xmath1272 that @xmath1275 and then that @xmath1276 combining this formula with the above estimates of @xmath1277 and the formulae proves the theorem for those @xmath36 whose image in the @xmath861-plane lies within the upward - pointing tooth of @xmath885 centered at @xmath1266 .
now suppose instead that the image of @xmath36 in the @xmath861-plane lies in the downward - pointing tooth of @xmath879 centered at @xmath1266 . in this part of the image , the uniform estimate @xmath1278 holds . by taylor expansion and
the fact that @xmath1262 , @xmath1279 it then follows from using @xmath1278 that @xmath1280 and then that @xmath1281 combining this formula with the above estimates of @xmath1282 and the formulae proves the theorem for those @xmath36 whose image in the @xmath861-plane lies within the downward - pointing tooth of @xmath879 centered at @xmath1266 .
99 m. ablowitz , d. kaup , a. newell , and h. segur , `` method for solving the sine - gordon equation , '' _ phys .
_ , * 30 * , 12621264 , 1973 . j. baik , p. deift , and k. johansson , `` on the distribution of the length of the longest increasing subsequence of random permutations , '' _
j. ams _ , * 12 * , 11191178 , 1999
. a. barone , f. esposito , c. j. magee , and a. c. scott , `` theory and applications of the sine - gordon equation , '' _ rev .
nuovo cim .
_ , * 1 * , 227267 , 1971 . m. bertola and a. tovbis , `` universality in the profile of the semiclassical limit solutions to the focusing nonlinear schrdinger equation at the first breaking curve , '' _ int .
res . not .
_ , * 2010 * , 21192167 , 2010 . m. bertola and a. tovbis , `` universality for the focusing nonlinear schrdinger equation at the gradient catastrophe point : rational breathers and poles of the tritronque solution to painlev i , '' ` arxiv:1004.1828 ` , 2010 .
r. j. buckingham and p. d. miller , `` exact solutions of semiclassical non - characteristic cauchy problems for the sine - gordon equation , '' _ physica d _ , *
237 * , 22962341 , 2008 .
r. j. buckingham and p. d. miller , the sine - gordon equation in the semiclassical limit : dynamics of fluxon condensates , " ` arxiv:1103.0061 ` , 2011 .
p. cheng , s. venakides , and x. zhou , `` long - time asymptotics for the pure radiation solution of the sine - gordon equation , '' _ comm .
_ , * 24 * , 11951262 , 1999 . c. chester , b. friedman , and f. ursell , `` an extension of the method of steepest descents , '' _ proc
. cambridge philos .
_ , * 53 * , 599611 , 1957 . t. claeys and t. grava , `` painlev ii asymptotics near the leading edge of the oscillatory zone for the korteweg - de vries equation in the small - dispersion limit , '' _ comm .
pure appl .
_ , * 63 * , 203232 , 2010 . t. claeys and t. grava , `` solitonic asymptotics for the korteweg - de vries equation in the small dispersion limit , '' _ siam j. math .
_ , * 42 * , 21322154 , 2010 .
t. claeys and a. kuijlaars , `` universality of the double scaling limit in random matrix models , '' _ comm .
pure appl .
_ , * 59 * , 15731603 , 2006 .
p. a. clarkson , `` remarks on the yablonskii - vorobev polynomials , '' _ phys .
* 319 * , 137144 , 2003 .
p. a. clarkson and e. l. mansfield , `` the second painlev equation , its hierarchy and associated special polynomials , '' _ nonlinearity _ , * 16 * , r1r26 , 2003 .
p. deift , s. venakides , and x. zhou ,
new results in small dispersion kdv by an extension of the steepest descent method for riemann - hilbert problems , " _ internat .
res . notices _ , * 1997 * , 285299 , 1997 .
p. deift and x. zhou , `` a steepest - descent method for oscillatory riemann - hilbert problems : asymptotics for the mkdv equation , '' _ ann . of math .
_ , * 137 * , 295368 , 1993 . l. faddeev and l. takhtajan , _ hamiltonian methods in the theory of solitons _
, springer - verlag , new york , 1987 .
a. s. fokas , a. r. its , a. a. kapaev , and v. yu .
novokshenov , _ painlev transcendents : the riemann - hilbert approach _ ,
volume 128 , mathematical surveys and monographs , american mathematical society , providence , 2006 .
e. l. ince , _ ordinary differential equations _ , dover publications , new york , 1956 .
d. kaup , `` method for solving the sine - gordon equation in laboratory coordinates , '' _ stud .
_ , * 54 * , 165179 , 1975 . m. klaus and j. k. shaw , `` purely imaginary eigenvalues of zakharov - shabat systems , '' _ phys .
e _ , * 65 * , 3660736611 , 2002 . y. murata , `` rational solutions of the second and fourth painlev equations , '' _ funkcialaj ekvacioj _
, * 28 * , 132 , 1985 .
a. c. scott , f. chu , s. reible , `` magnetic flux propagation on a josephson transmission line , '' _ j. appl
_ , * 47 * , 32723286 , 1976 . v. e. zakharov and a. b. shabat , `` exact theory of two - dimensional self - focusing and one - dimensional self - modulation of waves in nonlinear media , '' _ sov
jetp _ , * 34 * , 6269 , 1972 . v. e. zakharov , l. takhtajan , and l. faddeev , `` complete description of solutions of the sine - gordon equation , '' _ dokl .
nauk . sssr _ ,
* 219 * , 13341337 , 1974 .
x. zhou , `` inverse scattering transform for systems with rational spectral dependence , '' _
_ , * 115 * , 277303 , 1995 . | we study the cauchy problem for the sine - gordon equation in the semiclassical limit with pure - impulse initial data of sufficient strength to generate both high - frequency rotational motion near the peak of the impulse profile and also high - frequency librational motion in the tails .
subject to suitable conditions of a general nature , we analyze the fluxon condensate solution approximating the given initial data for small time near points where the initial data crosses the separatrix of the phase portrait of the simple pendulum .
we show that the solution is locally constructed as a universal curvilinear grid of superluminal kinks and grazing collisions thereof , with the grid curves being determined from rational solutions of the painlev - ii system . |
on march 29 , 2003 one of the brightest gamma - ray bursts ( grbs ) in history was discovered and localized by the hete-2 satellite .
astronomers worldwide watched and within days were rewarded by the discovery of a type ic supernova , sn 2003dh , in precisely the same location @xcite . from the light curve and spectrum , a temporal coincidence with the burst
was also estimated to be @xmath3 days .
it is now generally agreed that the supernova and the grb came from the same explosion .
the v - band light curve and spectrum of sn 2003dh closely resembled that of another famous supernova - grb pair , sn 1998bw @xcite and grb 980425 , but with several important distinctions : * 1 ) * grb 980425 had an observed equivalent isotropic energy in gamma - rays roughly four orders of magnitude less than grb 030329 ; * 2 ) * sn 2003dh rose to maximum in less than 10 days , sn 1998bw took 16 days @xcite ; * 3 ) * sn 2003dh exhibited higher expansion speeds , up to @xmath4 ; and * 4 ) * the total energy of ejected relativistic matter in sn 1998bw was @xmath5 with @xmath6 @xcite .
this is less even than the energy in gamma - rays from grb 030329 , unless the beaming angle is very small .
it is generally agreed that sn 1998bw was a very asymmetric explosion whose high velocities may not have characterized ejecta at all angles ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , and the same seems likely to be true of sn 2003dh . despite the fact that its radiation is not beamed like a grb , an asymmetric supernova can have a quite different light curve from a spherical one of the same total energy .
the higher velocities in one direction lead to the earlier escape of radiation and , in the case of a radioactive power source , a rapid decline in the deposition efficiency of gamma - rays . if the radioactivity itself is mixed out preferentially along one axis , the efficiency of gamma - deposition is also affected , leading to a more rapid rise and decline of the luminosity . on the other hand ,
the less rapidly moving matter ejected in other directions , can continue to contribute an extended tail on the light curve , powered by the remaining radioactivity . in a pioneering study of light curves from asymmetric explosions
, @xcite modeled the light curve of sn 1998bw , but attempted to fit it into the general family of type ib / c supernovae .
their parameters were thus typical of these common events : @xmath0 mass ( @xmath7 to @xmath8 ) , ejected mass ( @xmath9 ) , and explosion energy ( @xmath10 ) .
they , and @xcite , championed the idea that asymmetrically exploded supernovae of the same energy may have different light curves .
however , in light of sn 2003dh , neither went far enough . here
we model the supernova as essentially two components - a slowly moving , high - mass equatorial ejection and a lower mass polar ejection ( see also * ? ? ?
* ; * ? ? ?
the juncture between these is almost discontinuous .
a smooth transition would give too broad a light curve .
the total mass of the progenitor , @xmath11 of helium and heavy elements , the @xmath0 mass synthesized , @xmath12 , and the total explosion energy , @xmath13 , are all quite atypical for ordinary type ib / c supernovae .
a series of explosions was calculated for two wolf - rayet ( wr ) stars of final mass @xmath14 and @xmath15 .
both models had an initial mass of @xmath15 and an initial composition appropriate to the helium core of a star with 0.1 solar metallicity .
both models were started with a surface rotation rate corresponding to @xmath16 keplerian at the equator .
it was assumed , however , that the lower mass star was in a binary system and lost its hydrogen envelope to the companion star during the expansion phase after core hydrogen depletion ( case b mass transfer ) .
this model continued to lose mass as a wr - star at a rate given by @xcite , reduced by @xmath17}$ ] ( i.e. , about a factor 3 ) , and another factor 3 to account for clumping @xcite .
the higher mass model , while lacking a hydrogen envelope , assumed no further mass loss .
this was an artificial way to create two rapidly rotating wr - stars with a range of masses . for the given starting mass , the one evolved with mass loss is probably the more realistic . in both calculations
the effects of rotationally induced mixing were included @xcite throughout the evolution and both reached iron core collapse with sufficient angular momentum to form a kerr black hole and an accretion disk .
the adopted reaction rate for the c(@xmath18,@xmath19)o reaction was 1.2 times that of @xcite .
both models had very fine surface zoning , down to less than @xmath20 .
the radii of the two stars at the time their cores collapsed were @xmath21 ( with mass loss ) and @xmath22 ( without mass loss ) .
some other properties of the presupernova stars are given in tables [ tab : proj1 ] and [ tab : proj2 ] .
explosions were simulated by placing a piston at the location of a large entropy jump , around @xmath23 , in the presupernova star when its peak infall velocity had reached about @xmath24 .
such a large change in entropy typically corresponds to a sudden decrease in density at the base of the oxygen - burning shell where mass bifurcations often develop in supernova models .
the star outside this piston was first allowed to collapse to @xmath25 at one - fourth the free fall acceleration ( inward movement of the piston ) . for the lower mass model , which only formed a small iron core ( table [ tab : proj2 ] ) , the entropy jump and piston were at @xmath26 . for the higher mass model ,
a large collapsing low - entropy core formed ( table [ tab : proj2 ] ) and two piston locations were explored .
one was located at 1.93 @xmath27 , at the edge of the iron core , the other at @xmath28 where the large entropy discontinuity occurred .
after the minimum radius was reached , the piston was moved outward supersonically , decelerating at constant fraction of gravitational acceleration until it coasted to a halt at @xmath29 .
the initial velocity and deceleration of the piston were adjusted to give the desired explosion energy .
( the right piston acceleration has been determined using a modified regular falsi algorithm . )
this explosion energy here is defined as the final kinetic energy of the ejecta for an explosion into vacuum .
a wide range of kinetic energies was explored for these isotropic explosion ( table [ tab : mods ] ) , including energies all the way up to @xmath30 .
this energy , half the binding energy of a typical neutron star , is far more than expected from any realistic neutrino - powered model , but , as we shall see , might be relevant for a small amount of mass ejected in a very asymmetric explosion in which the energy source is not neutrinos , but gravitational energy from accretion into a black hole .
nucleosynthesis was calculated as described in @xcite .
an interesting result was the observation of a maximum mass of @xmath0 , regardless of explosion energy and depending only on the mass of the progenitor and depth of the piston .
a near constant mass , @xmath31 was synthesized in the @xmath14 models ( table [ tab : mods ] ) .
the near constancy of this upper bound is a consequence of the fast expansion and increasing entropy in the center of the ejecta that leads to freeze - out of 4he rather then @xmath0 ( fig . [ fig : comp8 ] _ _ ) . eventually , turning up the energy only increases the ejection of @xmath18-particles , not of @xmath0 . because the location of the piston can not be much deeper in the @xmath14 model and because any other way of exploding the star , say by energy deposition rather than a piston ,
would give more photodisintegration , @xmath32 is the maximum @xmath0 that can be synthesized in a spherically symmetric explosion of this star .
a larger amount can be made in the @xmath15 star without mass loss because more mass sits closer to the piston . for a piston mass of @xmath28 ,
the limiting mass of @xmath0 was @xmath33 . for a piston mass of @xmath34 ,
the edge of the iron core , the limiting @xmath0 mass was @xmath35 ( table [ tab : mods ] ) .
one might get still larger @xmath0 masses by going to helium stars above 15 @xmath27 .
however , this is the largest helium core one expects for stars near solar metallicity and larger mass cores will require even greater energies to expand rapidly enough to explain sn 2003dh . that these limits are within a factor of two of the mass of @xmath0 inferred ( see below ) for both sn 1998bw and sn 2003dh is interesting , but probably coincidental .
the existence of an upper bound for shock powered models has interesting implications though ( [ sec : disc ] ) .
light curves were calculated for all the explosion models using the kepler code as described , e.g. , in @xcite .
gamma - rays from @xmath0 and @xmath36 decay were assumed to deposit locally .
the gamma - ray opacity was @xmath37 . for the diffusing radiation , opacity
was assumed to be predominantly electron scattering .
the electron density was calculated by solving the saha equation for the ionization structure at each point .
though calculated assuming a single - temperature , flux - limited diffusion , and a simple model for gamma - ray deposition , the curves should be qualitatively correct and suffice for present purposes .
[ fig : lite15 ] _ _ and the first frame of fig .
[ fig : lite8 ] _ _ show the light curves expected when the supernova experiences `` moderate '' mixing . mixing was simulated by a running average of the composition within a region of mass , @xmath38 , set here to 10% of the final mass of the star .
that is , the actual composition gradients were smoothed by averaging within a defined band of masses , and this band was itself moved out , zone by zone , from the piston to the surface .
this procedure was repeated four times in each model . for the 8.38@xmath39model with 1.25@xmath40explosion energy , for example ,
this resulted in @xmath0 being mixed far enough out that it had 50% of its central value ( i.e. , just above the piston ) at about @xmath41 and 10% at @xmath42 ( as measured from the center of the collapsed remnant ) .
the second panel of fig .
[ fig : lite8 ] _ _ shows similar light curves for the @xmath43 model when the composition is completely homogenized , i.e. , made to be the same from center top surface .
this might be the case if a vigorous asymmetric flow is responsible for exploding the star .
these are bolometric light curves , which properly should be compared only with luminosities integrated across the uvoir bands .
however , based upon our experience with sn 1998bw , we shall compare them with the observed v - band light curve of sn 2003dh @xcite .
this shows the brightness rising to maximum at 10 13 days ( rest frame ) and declining by 1 magnitude when the supernova was about 30 days old .
the authors further estimate that the supernova was `` slightly brighter '' than sn 1998bw for which a peak luminosity @xmath44 @xcite and @xmath0 mass @xmath45 to @xmath46 has been estimated @xcite .
though the @xmath0 mass ejected varies with kinetic energy and progenitor mass ( table [ tab : mods ] ) , the supernova must have made a single value .
knowing the approximate luminosity we needed at peak , another series of light curves was calculated for which the yield of @xmath0 was normalized to @xmath47 in all the explosions , even for the low mass progenitor .
this is a typical value for the very energetic explosions ( table 2 ) , is close to the @xmath48 @xcite to @xmath46 @xcite of @xmath0 inferred for sn 1998bw , and is consistent with the observation that sn 2003dh was a little brighter than sn 1998bw @xcite .
the fact that it is allowed to exceed the hydrodynamical limit derived in the previous section is justified because of the alternative way @xmath0 is made in the collapsar model ( [ sec:2d ] ) .
such energetic models with so much @xmath0 are likely to be thoroughly mixed , more so than usual supernovae . mixing proved necessary to obtain a good fit to the light curve of sn 1998bw , which accompanied grb 980425 @xcite , and
is favored by the rapid rise time in sn 2003dh .
the third panel in fig .
[ fig : lite8 ] _ _ shows the light curves expected for the @xmath14 models when the ejecta are thoroughly mixed and forced to contain a fiducial @xmath47 of @xmath0 .
this was accomplished by replacing as much of the material directly above the piston by pure @xmath0 as was needed to obtain the desired total @xmath0 mass .
the replacement was done 100@xmath49after the explosion when essentially all thermonuclear reactions have ceased , but before any mixing was applied .
[ fig : vfinal ] _ _ shows the terminal velocity for both models as a function of mass and fig . [ fig : vphot ] _ _ gives the velocity at the photosphere as a function of time and for the @xmath50 model . in fig .
[ fig : vfinal ] _ _ only the intrinsically produced @xmath0 was considered , however , the energy release from @xmath0 decay has little effect on the expansion velocities , especially for the cases with high explosion energy .
the fact that speeds as high as @xmath51 were seen on sn 2003dh on day 10 @xcite illustrates the need for at least some portion of the ejecta to have equivalent isotropic energies above @xmath52
. however , most of the models can give the milder requirement of @xmath53 on day @xmath54 .
[ sec:2d ] only the most energetic symmetric explosions , @xmath55 , rise rapidly enough , make sufficient @xmath0 , and decay rapidly enough to resemble sn 2003dh .
the velocities in these models are also consistent with what was seen ( fig .
[ fig : vphot ] _
_ ) , but the energies strain credibility and even the most energetic model would not make a powerful gamma - ray burst
. the correct model must be asymmetric .
ideally , one would like to explore the coupled grb - supernova explosion in two - dimensional code that couples radiation transport , gamma - ray deposition , and special relativistic hydrodynamics .
such calculations will surely be done , but as an expedient and for clarity in exposition , we consider here a simple , composite toy model rendered out of combinations of our one - dimensional models .
we rely here on experience with the collapsar model , especially @xcite and @xcite . a black hole forms in the middle of the star and , after a few seconds while the polar accretion rate declines , launches relativistic jets along both axes .
these jets penetrate the star and ultimately make the grb , but they do not , by themselves , make a supernova .
the peak brightness of a type i supernova of any subclass is measured by the @xmath0 that it ejects .
both the jets and the lateral shocks they launch make almost none .
_ lacking additional sources of power , the supernova accompanying a collapsar - produced grb would be weak and nearly invisible_. that additional source of power is the wind off the accretion disk @xcite .
the mass ejected is of the same order as the mass accreted by the black hole , i.e. , @xmath56 .
it s energy is uncertain , but could easily be @xmath13 , that is the binding energy of a solar mass of material at the last stable orbit around a kerr black hole times a few percent .
a similar amount of energy is released by the reassembly of @xmath47 of nucleons from the disk into bound nuclei .
_ indeed the energy in this `` wind '' may considerably exceed the energy in the grb - producing jets themselves_. its composition is likely to be chiefly @xmath0 @xcite mixed with the helium , oxygen , and other heavy elements that make up the star .
this wind drives a highly asymmetric explosion @xcite .
in fact , to first order , it blows two conical - shaped `` wedges '' out of the star , each along the rotational axis . in the case of sn 2003dh ,
because we saw the grb , one of these inverted cones was directed straight at us .
the equatorial regions are partly ejected , but partly fall into the hole , continuing to power the jet for some time after the initial explosion .
the opening angle of the conical wedges is unknown , but certainly greater than the @xmath57 opening of the grb jet itself , yet probably small enough to contain only a fraction of the stellar mass . here
we will use @xmath58 as an example .
this implies that @xmath59 of the star is ejected at very high velocity . within this @xmath58
we assume a gaussian distribution of kinetic energies between @xmath60 and @xmath61 ( fig .
[ fig : eiso ] _
_ ) as a function of angle .
we further assume a total production of @xmath1 of @xmath0 , @xmath62 of which comes out in the high velocity ( well - mixed ) wedges , and @xmath63 of which stays behind in the low velocity ejecta .
the composite light curves were computed by assuming an explosion energy as a function of colatitude , @xmath64 : @xmath65 where @xmath66 is chosen such that @xmath67 , i.e. , @xmath68 this was used to determine the flux as function of @xmath64 by interpolation in the grid of one - dimensional light curves .
the one - dimensional models were calculated with an effective mass of @xmath0 equal to @xmath69 ( last panel of fig .
[ fig : lite8 ] _ _ ) , i.e. , a mass such that @xmath70 of it gave the actual @xmath0 mass in the high - velocity ejecta , @xmath71 .
these contributions were then integrated over the projection of the sphere in the direction of the observer , along the pole ( fig . [ fig : slice ] _ _ ) .
such a procedure does not take into account the `` shadowing '' effect of the fast moving polar ejecta on the slower equatorial ejecta .
owing to its fast expansion , the polar ejecta quickly become optically thin and the late - time light curve from the equatorial ejecta should not be significantly affected .
we also do not take into account that the cone of high explosion energy may trap @xmath19-rays less efficiently than a full sphere and thus cool down faster , or similarly , that the polar `` hole '' punched into the explosion by the jet could allow photons to escape faster in this direction , or that a larger fraction of the low - energy ejecta than corresponds to the projected surface area would become visible once the polar ejecta are optically thin .
these truly two - dimensional effects may make the low - e part of the ejecta a little brighter early on and decline a bit faster at late times before the spherical model becomes optically thin .
the composite light curve corresponding to these assumptions is given in fig .
[ fig : composite]__. during the first month , the luminosity is given entirely by the high - velocity , nickel - rich ejecta . because of the high velocity and complete mixing the rise to maximum is very rapid .
because the material becomes thin and gamma - rays from radioactive decay escape , the decline from maximum is also fast .
the decline of the light curve might be even faster than suggested by the composite model , because @xmath19-rays can escape easier from a polar wedge than from a full sphere that has been used for modeling the light curves . at late times
though , the slower moving , nickel - poor ejecta dominate the light curve because gamma - ray deposition there remains highly efficient . for situations where the energy is due chiefly to radioactivity , this sort of light curve is distinctive signature of asymmetry .
models that expand slowly would not be so bright at peak and models that expand rapidly would not be so bright at late times .
the amount of @xmath0 inferred from the peak brightness would be quite different from that inferred on the tail . such seems to have been the case for sn 1998bw where a @xmath0 mass of @xmath72 was inferred for the peak @xcite and @xmath73 at late times @xcite .
[ sec : disc ] the optical light curve of the supernova that accompanies a typical grb is unique in two ways .
first , it must be disentangled from the `` optical afterglow '' , which is dominated at early times by the relativistic grb - producing jet interacting with the circum - source medium .
there will also be optical emission from slower moving , but still highly energetic supernova ejecta running into this medium .
continued , possibly time - variable output from the central engine also contributes at a declining rate to the afterglow at all wavelengths @xcite . whether one counts these afterglows as `` supernova '' or something else is largely a matter of taste .
supernova shock interactions have often been considered part of the light curve @xcite .
they are probably the origin of variations in the early optical emission seen during the first few days of sn 2003dh @xcite .
even with complete mixing , enormous explosion energy , and a large mass of @xmath0 , a supernova powered by radioactivity would not reach maximum light in two days .
second , the supernova is grossly asymmetric .
we have attempted to account for this by merging the results of several spherically symmetric models of various energies .
even so , large energies are required in the composite model , far above the grb jet energy estimated by @xcite .
this energy must be provided by another source which , in the collapsar model , is the disk wind .
any attempt to model sn 2003dh without the formation of a black hole and disk must find a different way to produce the necessary @xmath0 and energy .
the only other potential candidate at this time is a millisecond magnetar @xcite .
provided the putative pulsar can survive accretion in a massive progenitor star long enough , it will need to deposit @xmath74 in @xmath75 in order to make @xmath12 of @xmath0 in a high mass progenitor ( making this much in a low mass model is impossible ) .
the @xmath0 is required by the light curve and the high energy is required to make the star move fast enough to explain the rapid post - maximum decline of the supernova .
the one - second time scale is set by the requirement that the @xmath0 be made by a shock that raises the temperature to greater than @xmath76 as is necessary to produce @xmath0 out of lighter elements .
concentrating the pulsar s energy in a smaller solid angle , as has been done here , will give a smaller @xmath0 mass no greater than @xmath35 times that solid angle ( table [ tab : mods ] ) .
in addition , the pulsar would need to focus some part of this energy into a very relativistic narrow jet ( in order to make the grb ) at a time when the accretion rate is very high .
these potential difficulties lead us to favor the collapsar model . based upon this model , in which the @xmath0 is provided by the disk wind in an explosion that is highly asymmetric
we have developed a composite working model in which the total explosion energy is @xmath2 and the mass of @xmath0 is @xmath1 .
further observations to test the predicted light curve ( fig . [ fig : composite ] _
_ ) are encouraged .
this research has been supported by nasa ( nag5 - 8128 , nag5 - 12036 , and mit-292701 ) and the doe program for scientific discovery through advanced computing ( scidac ; de - fc02 - 01er41176 ) .
ah has been supported in part by the department of energy under grant b341495 to the center for astrophysical thermonuclear flashes at the university of chicago .
cccc iron core & @xmath77 & @xmath78 & @xmath79 + si core & @xmath80 & @xmath81 & @xmath82 + ne / mg / o core & @xmath83 & @xmath84 & @xmath85 + c / o core & @xmath86 & @xmath87 & @xmath88 + star / he core & @xmath89 & @xmath90 & @xmath91 + cccc iron core & @xmath92 & @xmath93 & @xmath94 + si core & @xmath95 & @xmath96 & @xmath97 + ne / mg / o core & @xmath98 & @xmath99 & @xmath100 + c / o core & @xmath101 & @xmath102 & @xmath103 + star / he core & @xmath104 & @xmath105 & @xmath106 + cccc ke ( @xmath107 ) & & @xmath0 ( @xmath27 ) & + 1.25 & 0.158 & 0.040 & 0.024 + 2.5 & 0.193 & 0.614 & 0.225 + 5 & 0.229 & 0.707 & 0.254 + 10 & 0.244 & 0.815 & 0.303 + 20 & 0.230 & 0.938 & 0.377 + 40 & 0.189 & 1.045 & 0.487 + 80 & 0.168 & 1.083 & 0.583 + 160 & 0.170 & 0.955 & 0.625 + star with @xmath108 explosion energy @xmath109 after core collapse , but prior to any mixing . _ right : _ same plot , but for the @xmath110 explosion .
note that @xmath111 has been removed from the surface of this calculation because it exceeded @xmath112 and posed problems for the non - relativistic hydro - code . also note the large mass of photodisintegrated matter ( 4he ) in the inner regions near the mass cut .
this limits the production of @xmath0 in the model .
[ fig : comp8],title="fig : " ] star with @xmath108 explosion energy @xmath109 after core collapse , but prior to any mixing .
_ right : _ same plot , but for the @xmath110 explosion .
note that @xmath111 has been removed from the surface of this calculation because it exceeded @xmath112 and posed problems for the non - relativistic hydro - code . also note the large mass of photodisintegrated matter ( 4he ) in the inner regions near the mass cut .
this limits the production of @xmath0 in the model .
[ fig : comp8],title="fig : " ] ) and the large piston mass ( @xmath28 ) .
the explosions were calculated in spherical symmetry with kinetic energies at infinity of 160 , 80 , 40 , 20 , 10 , 5 , 2.5 , and 1.25@xmath113 and the light curves include just the mass of @xmath0 produced explosively in the model ( table [ tab : mods ] ) .
note the pre - maximum non - monotonic evolution of the most energetic models resulting from helium and heavy element recombination .
for the most energetic explosions the initial rise for the first few hours is a numerical artifact of removing zones moving faster than @xmath114 .
[ fig : lite15 ] ] ) .
calculated in spherical symmetry with kinetic energies at infinity of 160 , 80 , 40 , 20 , 10 , 5 , 2.5 , and 1.25@xmath113 .
see also fig .
[ fig : lite15]__. _ top left : _ moderately mixed with the @xmath0 yields in table [ tab : mods ] .
_ top right : _ same models as top left , but completely mixed . _ bottom left : _ completely mixed models with a normalized abundance of @xmath0 of @xmath47 in all cases .
the peaks of the light curves are brighter owing to the increased mass of @xmath0 and the pre - maximum variability is less apparent . _ bottom right : _
same mixed calculations normalized to a @xmath0 mass of @xmath69 , as is appropriate for the high - velocity ejecta .
[ fig : lite8],title="fig:",scaledwidth=52.5% ] ) . calculated in spherical symmetry with kinetic energies at infinity of 160 , 80 , 40 , 20 , 10 , 5 , 2.5 , and 1.25@xmath113 .
see also fig .
[ fig : lite15]__. _ top left : _ moderately mixed with the @xmath0 yields in table [ tab : mods ] .
_ top right : _ same models as top left , but completely mixed . _ bottom left : _ completely mixed models with a normalized abundance of @xmath0 of @xmath47 in all cases .
the peaks of the light curves are brighter owing to the increased mass of @xmath0 and the pre - maximum variability is less apparent . _ bottom right : _
same mixed calculations normalized to a @xmath0 mass of @xmath69 , as is appropriate for the high - velocity ejecta .
[ fig : lite8],title="fig:",scaledwidth=45.5% ] ) . calculated in spherical symmetry with kinetic energies at infinity of 160 , 80 , 40 , 20 , 10 , 5 , 2.5 , and 1.25@xmath113 .
see also fig .
[ fig : lite15]__. _ top left : _ moderately mixed with the @xmath0 yields in table [ tab : mods ] .
_ top right : _ same models as top left , but completely mixed . _ bottom left : _ completely mixed models with a normalized abundance of @xmath0 of @xmath47 in all cases .
the peaks of the light curves are brighter owing to the increased mass of @xmath0 and the pre - maximum variability is less apparent . _ bottom right : _
same mixed calculations normalized to a @xmath0 mass of @xmath69 , as is appropriate for the high - velocity ejecta .
[ fig : lite8],title="fig:",scaledwidth=52.5% ] ) . calculated in spherical symmetry with kinetic energies at infinity of 160 , 80 , 40 , 20 , 10 , 5 , 2.5 , and 1.25@xmath113 .
see also fig .
[ fig : lite15]__. _ top left : _ moderately mixed with the @xmath0 yields in table [ tab : mods ] .
_ top right : _ same models as top left , but completely mixed . _ bottom left : _ completely mixed models with a normalized abundance of @xmath0 of @xmath47 in all cases .
the peaks of the light curves are brighter owing to the increased mass of @xmath0 and the pre - maximum variability is less apparent . _
bottom right : _
same mixed calculations normalized to a @xmath0 mass of @xmath69 , as is appropriate for the high - velocity ejecta .
[ fig : lite8],title="fig:",scaledwidth=45.5% ] in the @xmath15 model .
moderate mixing and only the @xmath0 produced intrinsically by the explosion were used .
note fallback of @xmath115 for the lowest explosion energy .
_ right : _ same plot , but for the @xmath50 model[fig : vfinal],title="fig : " ] in the @xmath15 model .
moderate mixing and only the @xmath0 produced intrinsically by the explosion were used .
note fallback of @xmath115 for the lowest explosion energy .
_ right : _ same plot , but for the @xmath50 model[fig : vfinal],title="fig : " ] explosions of the @xmath50 star with moderate mixing and only its intrinsic @xmath0 production .
note that the velocity drops to zero and becomes undefined thereafter when the photosphere reaches the center of the ejecta .
( a line of sight through the center would have an optical thickness of @xmath116 at this point . )
also note that zones moving faster than @xmath112 have been cut off from the calculations ; this affects the velocity at the photosphere for about half a day for the most energetic explosion , and increasingly earlier times for lower energies . _
right : _ same plot , but for ejecta whose @xmath0 mass has been increased to @xmath69 and the ejecta were `` thoroughly mixed '' .
extra heating by the increased @xmath0 abundance at large radii maintains an extended photosphere longer .
[ fig : vphot],title="fig : " ] explosions of the @xmath50 star with moderate mixing and only its intrinsic @xmath0 production .
note that the velocity drops to zero and becomes undefined thereafter when the photosphere reaches the center of the ejecta .
( a line of sight through the center would have an optical thickness of @xmath116 at this point . )
also note that zones moving faster than @xmath112 have been cut off from the calculations ; this affects the velocity at the photosphere for about half a day for the most energetic explosion , and increasingly earlier times for lower energies .
_ right : _ same plot , but for ejecta whose @xmath0 mass has been increased to @xmath69 and the ejecta were `` thoroughly mixed '' .
extra heating by the increased @xmath0 abundance at large radii maintains an extended photosphere longer .
[ fig : vphot],title="fig : " ] the energy varies from @xmath60 to @xmath61 and models normalized to @xmath69 of @xmath0 were used , being `` thorough '' mixed . for angles greater than @xmath58 the explosion energy is @xmath108 weak ; only the intrinsic @xmath0 production ( table [ tab : proj2 ] ) and mild mixing was employed.[fig : eiso ] ] ( e.g. , @xmath117 as depicted above ) , each polar cap comprises a solid angle of @xmath118 , i.e. , the bipolar jet flows out at a fraction @xmath119 of the total solid angle of @xmath120 .
this means @xmath119 of the isotropic explosion contributes to such an explosion and its energy
. however , an observer looking along the axis of the jet sees the projected surface area @xmath121 out of @xmath122 for the full circle , the full projected sphere , i.e. , a fraction @xmath123 of the isotropic light curves contributes to apparent light curve for a such positioned observer . note that for @xmath124 : @xmath125 , i.e. , the observer along the axis is favored to see an apparently brighter light curve than would correspond to the total energy of a composed light curve ( assuming the explosion is stronger in polar direction than in equatorial direction )
this is because material at high colatitude , @xmath64 , appears under a high inclination angle .
[ fig : slice ] ] _ _ ) having a distribution of kinetic energies as given in fig .
[ fig : eiso]__. for polar angles less than @xmath58 the three highest energy models in fig .
[ fig : lite8 ] _ _ were employed which each have an effective @xmath0 mass of @xmath69 . together , with appropriate angular weighting ( see text ) , these give the curve labeled `` high - e '' .
for angles greater than @xmath58 the model with energy @xmath108 in fig .
[ fig : lite8 ] _ _ was used .
the @xmath0 mass in this component was left at @xmath126 .
this curve , with appropriate weighting , is labeled `` low - e '' .
the dark solid line is the total . the total energy in the composite is @xmath2 and the total mass of @xmath0 is @xmath1 . [ fig : composite ] ] | sn 2003dh , one of the most luminous supernovae ever recorded , and the one with the highest measured velocities , accompanied gamma - ray burst 030329 .
its rapid rise to maximum and equally rapid decline pose problems for any spherically symmetric model .
we model the supernova here as a very energetic , polar explosion that left the equatorial portions of the star almost intact .
the total progenitor mass was much greater than the mass of high - velocity ejecta , and the total mass of @xmath0 synthesized was about 0.5 solar masses .
such asymmetries and nickel masses are expected in the collapsar model .
a `` composite two - dimensional '' model is calculated that agrees well with the characteristics of the observed light curve .
the mass of @xmath0 required for this light curve is @xmath1 and the total explosion energy , @xmath2 . |
investigations focusing on global changes in star - formation activity and stellar - mass build - up in field galaxies have provided significant insight into the physical evolution of galaxies and their constituent stellar populations .
it has now been well - established that the global star - formation rate density has declined by roughly an order of magnitude since @xmath7 11.5 ( e.g. , lilly 1996 ; madau 1996 ; steidel 1999 ; hopkins 2004 ; prez - gonzlez 2005 ; schiminovich 2005 ; colbert 2006 ) .
recent investigations into the details of this evolution have shown that the star - formation history of a given galaxy depends strongly on its stellar mass ( e.g. , cowie 1996 ; juneau 2005 ; bundy 2006 ; noeske 2007a , b ; zheng 2007 ) ; the peak star - formation epoch for the most massive galaxies occured earlier in cosmic history than it did for galaxies with lower masses .
x - ray studies of normal late - type galaxies ( i.e. , those that are not dominated by luminous active galactic nuclei [ agns ] ) have shown that emission provides a useful , relatively - unobscured measure of star - formation activity ( e.g. , bauer 2002a ; cohen 2003 ; ranalli 2003 ; colbert 2004 ; grimm 2003 ; gilfanov 2004a ; persic 2004 ; persic & rephaeli 2007 ; however , see barger 2007 ) . in normal galaxies ,
emission originates from binaries , supernovae , supernova remnants , hot ( @xmath0 ) interstellar gas , and o - stars ( see , e.g. , fabbiano 1989 , 2006 for reviews ) .
sensitive and studies of local late - type galaxies have shown that high - mass binaries ( hmxbs ) and low - mass binaries ( lmxbs ) typically dominate the total non - nuclear power output ( e.g. , zezas 2002 ; bauer 2003 ; soria & wu 2003 ; swartz 2003 ; jenkins 2005 ; kilgard 2005 ; however , see , e.g. , doane 2004 ) .
observations indicate that the integrated emission from hmxb and lmxb populations trace galaxy star - formation rate ( sfr ) and stellar mass ( @xmath3 ) , respectively .
for example , using observations of 32 local galaxies , colbert ( 2004 ) found that the summed non - nuclear point - source emission from a given galaxy ( @xmath8 ) can be approximated as , where @xmath9 and @xmath10 are constants .
therefore , galaxies having relatively high star - formation rates per unit mass ( specific star - formation rates , ssfrs ) generally have dominant point - source contributions from hmxbs ( e.g. , late - type star - forming galaxies ) , while those with relatively low ssfrs have point - source emission primarily from lmxbs ( e.g. , massive early - type galaxies ) .
if the binary populations are similarly dominating the normal - galaxy power output over a significant fraction of cosmic time , then there should be a rapid increase in the globally - averaged luminosity of normal star - forming galaxies with cosmic look - back time in response to the increasing global star - formation rate density ( e.g. , ghosh & white 2001 ) . in this scenario ,
hmxbs trace the immediate star - formation rate of a galaxy and lmxbs trace its star - formation history with a lag of a few gyr . with the advent of deep and surveys ( see , e.g. , brandt & hasinger 2005 for a review ) , it has become possible to study the properties of normal galaxies out to @xmath11 and @xmath12 , respectively ( see , e.g. , hornschemeier 2000 , 2003 ; alexander 2002 ; georgakakis 2007 ; georgantopoulos 2005 ; lehmer 2006 , 2007 ; kim 2006 ; tzanavaris 2006 ; rosa - gonzalez 2007 ) .
evidence for a global increase in the activity with redshift for normal late - type galaxies has since been mounting .
studies of the luminosity functions of detected normal galaxies have found that @xmath13 evolves as over the redshift range @xmath7 01.4 ( e.g. , norman 2004 ; georgakakis 2006 ; ptak 2007 ; tzanavaris & georgantopolous 2008 ) .
additionally , stacking analyses have enabled investigations of more representative optically - selected galaxy populations over the majority of cosmic history ( @xmath7 0.14 ; e.g. , brandt 2001 ; hornschemeier 2002 ; nandra 2002 ; georgakakis 2003 ; reddy & steidel 2004 ; laird 2005 , 2006 ; lehmer 2005a ) .
these studies have found that the average luminosities of normal late - type galaxies increases with redshift out to @xmath7 1.43 .
for example , using a @xmath01 ms exposure of a subregion within the deep field - north , hornschemeier ( 2002 , hereafter h02 ) tentatively observed a factor of @xmath023 increase in @xmath14 from @xmath15 to 1.4 for @xmath16 galaxies . despite these promising initial constraints , the details of the evolution of normal late - type galaxy populations , including dependences on the physical properties of galaxies ( e.g. ,
optical luminosity , stellar mass , environment , and star - formation rate ) , have remained unexplored . in this paper
, we aim to improve significantly upon constraints for the evolution of normal late - type galaxies ( e.g. , h02 ) .
we study for the first time how the properties of late - type field galaxies evolve as a function of optical luminosity , stellar mass , and star - formation rate over the redshift range of @xmath17 .
we construct late - type galaxy samples located in two of the most well - studied extragalactic survey fields , the @xmath02 ms _ chandra _ deep field - north (; alexander 2003 ) and the extended deep field - south ( ) , which is composed of the central @xmath01 ms deep field - south ( cdf - s ; giacconi 2002 ) and four flanking @xmath0250 ks observations ( lehmer 2005b ) .
these deep fields ( hereafter cdfs ) reach detection limits of @xmath0@xmath18 in the most sensitive regions and @xmath19@xmath20 over the majority of the cdfs ; these levels are sufficient to detect moderately - powerful sources ( @xmath21 ) at @xmath1 and @xmath22 , respectively .
therefore , the cdfs comprise an unprecedented data set for effectively studying the emission and evolution of cosmologically distant normal galaxies with minimal contamination from powerful agns .
the galactic column densities are @xmath23 @xmath24 for the ( lockman 2004 ) and @xmath25 @xmath24 for the ( stark 1992 ) .
all of the fluxes and luminosities quoted throughout this paper have been corrected for galactic absorption . unless stated otherwise , we quote optical magnitudes based upon the ab magnitude system for photometry and the vega magnitude system everywhere else . in the band ,
we make use of three standard bandpasses : ( soft band [ sb ] ) , ( hard band [ hb ] ) , ( full band [ fb ] ) . throughout this paper ,
we make estimates of stellar mass and star - formation rates using a kroupa ( 2001 ) initial mass function ( imf ) ; when making comparisons between these estimates and those quoted in other studies , we have adjusted all values to correspond to our adopted imf .
@xmath26 = 70 , @xmath27 = 0.3 , and @xmath28 = 0.7 are adopted throughout this paper ( e.g. , spergel 2003 ) , which imply a look - back time of 7.7 gyr at @xmath29 .
we constructed an optically selected sample of late - type galaxies within the cdfs to use for our subsequent analyses .
we restricted our galaxy selection to regions of the cdfs where _ hubble space telescope _
( ) observations were available to allow the best possible morphological classifications .
the observations in the cdfs have been carried out via the great observatories origins deep survey ( goods ; giavalisco 2004a ) and galaxy evolution from morphology and seds ( gems ; rix 2004 ; caldwell 2005 ) programs ; these surveys cover @xmath090% of the -observed regions of the cdfs with the advanced camera for surveys ( acs ) . in the goods and gems regions , photometry was available for four ( @xmath30 , @xmath31 , @xmath32 , and @xmath33 ) and two ( @xmath31 and @xmath33 ) acs passbands , respectively .
we began building our sample by selecting all galaxies having @xmath34 .
this initial selection criterion was motivated by ( 1 ) the availability of deep @xmath33 band imaging over all of the cdfs , ( 2 ) the fact that the @xmath33 emission probes mass - tracing rest - frame optical light redward of the 4000 break for galaxies at @xmath35 , which constitutes the large majority of our sample , ( 3 ) the availability of reliable spectroscopic and photometric redshifts ( see details below ) , and ( 4 ) the high multiwavelength detection fractions for these sources , which allow us to determine informative rest - frame spectral energy distributions ( seds ) . in order to isolate most effectively distant detected agns , we further restricted our sample to include only galaxies that were located in the most sensitive areas of the cdfs where the point - spread function ( psf ) was small .
we therefore chose to include sources having optical positions that were within 60 of at least one of the six aimpoints in the cdfs2 ms and @xmath01 ms , respectively , as well as table 1 of lehmer ( 2005b ) for the @xmath0250 ks . ] ; the corresponding total areal footprint is @xmath00.18
futhermore , we removed obvious galactic stars that were identified via optical spectral properties or the presence of obvious diffraction spikes in the @xmath33 band images . under these restrictions , we found 6905 galaxies .
we cross - correlated our initial sample of 6905 galaxies with available spectroscopic and photometric redshift catalogs ( e.g. , barger 2003 ; le fevre 2004 ; szokoly 2004 ; wirth 2004 ; wolf 2004 ; mobasher 2004 ; mignoli 2005 ; vanzella 2005 , 2006 ; grazian 2006 ; ravikumar 2007 ; silverman 2008 ) .
all galaxies that did not have spectroscopic redshifts were located in the where highly accurate ( median @xmath37 for galaxies with @xmath38 ) photometric redshifts were available via combo-17 ( classifying objects by medium - band observations in 17 filters ; wolf 2004 ) . in total , 6683 ( @xmath097% ) of our sources had either spectroscopic or photometric redshifts .
visual inspection of the 222 galaxies in the without redshifts indicated that these sources were mainly faint galaxies near bright stars , as well as a handful of sources that were subgalactic features within relatively nearby galaxies . whenever possible , we adopted spectroscopic redshifts as the most accurate redshifts for our galaxies . using the redshift information ,
we filtered our sample to include only sources with @xmath39 in the @xmath02 ms , @xmath40 in the @xmath01 ms , and @xmath41 in the @xmath0250 ks regions of the ; these redshift limits represent the largest distances at which we would expect to identify moderately luminous ( @xmath42 ) agns effectively in each respective field . in total , 3259 galaxies remained after filtering our sample based on redshift properties .
we used spectroscopic redshifts for 1351 ( @xmath041% ) galaxies and photometric redshifts for the remaining 1908 ( @xmath059% ) galaxies .
the optical - color distribution for field galaxies has been shown to be bimodal , separating `` red '' and `` blue '' galaxy populations ( e.g. , strateva 2001 ; hogg 2002a ; blanton 2003 ; baldry 2004 ) .
studies of the color - magnitude relation for distant galaxy populations have shown that this color bimodality is observed to persist out to at least ( see , e.g. , bell 2004a ; faber 2007 ; labbe 2007 ) , thus providing an excellent quantifiable means for separating late - type and early - type galaxy populations .
we therefore filtered our galaxy sample to include only sources that had blue rest - frame optical colors , as expected for late - type galaxies that contain young stellar populations . to estimate rest - frame @xmath43 colors and absolute @xmath44 , @xmath45 , and @xmath46 band magnitudes ( @xmath47 , @xmath48 , and @xmath49 , respectively ) for each of the 3259 galaxies in our sample
, we utilized the available optical / near - ir data .
for the , we used the @xmath44 , @xmath45 , @xmath46 , @xmath50 , @xmath51 , @xmath52 , and @xmath53 photometric catalogs from capak ( 2004 ) , as well as irac imaging ( 3.6 , 4.5 , 5.8 , and 8.0@xmath54 m ; fazio 2004 ) from the goods ( dickinson , in preparation ) .
for the , we used ( 1 ) the 17-bandpass photometry available through combo-17 , ( 2 ) @xmath55 and @xmath56 imaging from the eso imaging survey ( olsen 2006 ) , ( 3 ) @xmath55 , @xmath57 , and @xmath58 from the musyc collaboration ( taylor in preparation ; see also , moy 2003 ; gawiser 2006 ) , and ( 4 ) irac data from the goods and simple teams ( dickinson , in preparation ; van dokkum , in preparation ) . using these data , we constructed a rest - frame near - uv to near - ir sed for each galaxy .
for these seds , the rest - frame @xmath44 , @xmath45 , and @xmath46 filters are well - sampled at all relevant redshifts , with the exception of sources at @xmath59 where the wavelength range of the available data is sparse at rest - frame @xmath46 band . for sources at @xmath60 , we linearly interpolated our sed to cover the @xmath46 filter .
we convolved these seds with johnson @xmath44 , @xmath45 , and @xmath46 filter curves and computed rest - frame absolute magnitudes for each respective filter following equation 5 of hogg ( 2002b ) . for sources in the
, these computed absolute magnitudes are consistent with those presented by wolf ( 2004 ) . in figure 1
( _ small filled circles _ ) , we present rest - frame @xmath43 colors versus @xmath49 for our galaxies in six redshift ranges . for clarity
, we also show inset histograms giving the distribution of rest - frame @xmath43 colors for each redshift interval .
we utilized the rest - frame @xmath43 color to divide roughly populations of red and blue galaxies .
use of the rest - frame @xmath43 color was motivated by bell ( 2004a ) , who note that the @xmath44 and @xmath46 bandpass pair straddle the 4000 break , which is particularly sensitive to age and metallicity variations of galactic stellar populations .
the dashed lines in figure 1 show the empirically - determined redshift - dependent color divisions that separate blue and red galaxy populations ; we calculated these divisions following @xmath61 5 of bell ( 2004a ) : @xmath62 galaxies having rest - frame @xmath43 less than the values provided by equation 1 are often referred to as `` blue - cloud '' galaxies , while those with rest - frame @xmath43 greater than the division are called `` red - sequence '' galaxies .
the redshift - dependence of the blue - cloud / red - sequence galaxy division is thought to be largely due to the evolution of the mean age and dust content of the blue - cloud population , with a smaller contribution from changes in metallicity . in total
, we found 2502 blue - cloud galaxies and 757 red - sequence galaxies .
generally , the blue - cloud and red - sequence populations are composed of late - type and early - type galaxies , respectively ( see , e.g. , bell 2004b , 2005 ; mcintosh 2005 ) . in support of this point
, we have created figure 2 , which shows the fraction of galaxies in our sample having blue - cloud ( _ dashed histogram _ ) and red - sequence ( _ dotted histogram _ ) colors as a function of galaxy srsic index @xmath63 .
we utilized the hu@xmath64ssler ( 2007 ) srsic indices , which were computed using the gems @xmath33 images and the galfit ( peng 2002 ) two - dimensional light - profile fitting program .
light - profile studies of large galaxy samples have found empirically that a srsic cutoff of @xmath65 can roughly discriminate between late - type and early - type galaxies ( e.g. , blanton 2003 ; shen 2003 ; hogg 2004 ) .
galaxies with @xmath66 are generally late - type galaxies , while the majority of galaxies with @xmath67 are early - types ( _ vertical line _ in fig . 2 ) .
figure 2 shows that there is reasonable agreement between late - type and early - type galaxy populations selected using rest - frame optical colors and srsic indices . ) for our initial sample of @xmath17 01 galaxies in the having blue - cloud ( _ dashed histogram _ ) and red - sequence ( _ dotted histogram _ ) colors .
the solid vertical line indicates @xmath65 , the empirical cutoff between late - type ( @xmath66 ) and early - type galaxies ( @xmath67 ) . ,
width=359 ] we refined further our division between late - type and early - type morphologies by visually inspecting the @xmath33-band images of our entire sample of 3259 galaxies to see if there were obvious cases where the rest - frame optical colors provided an inaccurate morphological classification .
for example , late - type galaxies that are highly inclined to our line of sight may experience significant reddening of the young ( and blue ) disk population and will have red - sequence colors . also , due to variations of the stellar populations in galaxies of a given morphological class , there will be some scatter in rest - frame @xmath43 color near the division of the red - sequence and blue - cloud regimes .
this will lead to a number of `` green '' galaxies that are misclassified morphologically when the classification is based solely on rest - frame color .
we stress that such a visual inspection is only to correct source classifications that were _ obviously misclassified _ by simply using color - magnitude diagrams , and our resulting sample should not be considered a morphologically - selected sample .
based on our visual inspection , we found 76 obvious early - type galaxies with blue - cloud colors , and 142 obvious late - type galaxies with red - sequence colors ; we reclassified these sources as late - type and early - type galaxies , respectively .
we inspected histograms of the fraction of blue - cloud and red - sequence galaxies that were reclassified as a function of redshift and find no obvious redshift - related trend . in figure 3 , we show @xmath33-band postage - stamp images of ten obvious early - type galaxies with blue - cloud colors ( _ top panels _ ) , as well as ten obvious late - type galaxies with red - sequence colors ( _ bottom panels _ ) . after reclassifying these objects ,
we were left with 2568 late - type galaxies and 691 early - type galaxies with . since we were interested in studying the properties of the late - type galaxy populations
, we hereafter refer to our sample of 2568 late - type galaxies as our _ main sample _ , which we use in subsequent analyses .
the primary goal of this study is to investigate the evolution of normal ( i.e. , non - agn ) late - type galaxies and to determine how this evolution depends upon three intrinsic physical properties : optical luminosity , stellar mass , and star - formation rate .
below , we describe how we estimated each of these physical properties for the galaxies in our main sample .
we note that the populations of late - type galaxies that we are investigating here have been selected via their intrinsic physical properties , which may have changed significantly from the observed epoch to the present day .
for example , it is expected that a significant fraction of the late - type galaxies in our main sample will evolve into early - type galaxies via mergers or passive stellar evolution ( e.g. , bell 2007 ) .
ideally , we would like to study the evolution of the properties of late - type galaxies while controlling for such changes in the physical nature of each galaxy .
however , since the details of this evolution are highly complex and not well understood for a given galaxy , such a task is beyond the scope of this paper .
we therefore investigate the evolution of normal late - type galaxy populations in relation to their observed intrinsic physical properties . in order to study late - type galaxy samples
selected using an observable quantity , we made use of the @xmath45-band luminosity ( @xmath2 ) . in @xmath61 2.3 , we computed absolute @xmath45-band magnitudes @xmath48 for galaxies in our main sample using photometrically - derived seds . the @xmath45-band emission from a given late - type galaxy
will be significantly influenced by large populations of old ( @xmath68100 myr ) stars ( measured by the stellar mass ) as well as the younger and less - numerous massive stars that reside in low - obscuration star - forming regions ( measured by the star - formation rate ) . in figure 4@xmath69 , we show @xmath2 ( expressed in solar units ; where ) versus redshift for galaxies in our main sample .
the redshift - dependent selection limit of @xmath2 for our sample is set by our @xmath34 criterion , and at @xmath70 ( @xmath1 ) this limit corresponds to roughly ( ) . for comparison , in figure 4@xmath69 , we have plotted the values of @xmath2 for the milky way ( mw ) and local galaxies m101 , m82 , and the luminous infrared galaxy ( lirg ) ngc 3256 . for the mw
, we adopted @xmath71 as the approximate total disk - plus - bulge @xmath45-band luminosity ( see table 15.2 of gilmore 1990 ) .
for the local galaxies , we utilized optical photometry from the third reference catalog of bright galaxies ( rc3 ; de vaucouleurs 1991 ) and adopted distances from the iras revised galaxy sample ( sanders 2003 ) , which gives distances of 6.7 , 3.6 , and 35.4 mpc and implies corresponding @xmath45-band luminosities of @xmath72 , @xmath73 , and @xmath74 for m101 , m82 , and ngc 3256 , respectively .
as discussed in @xmath61 1 , the emission from lmxb populations scales with galaxy stellar mass .
it is therefore useful to select late - type galaxies via their stellar masses as a means for estimating the lmxb contribution to their emission . to estimate the stellar mass ( @xmath75 ) of each of our galaxies
, we exploited the tight correlation between rest - frame optical color and stellar mass - to - light ratio ( e.g. , bell & de jong 2001 ; bell 2003 ; kauffmann 2003a ; see also , borch 2006 ) . for this calculation
, we used a combination of rest - frame @xmath76 colors and rest - frame @xmath58-band luminosities .
@xmath58-band luminosities were computed by fitting all available optical / near - ir photometric data to a grid of 276 synthetic spectra generated by the pgase stellar population synthesis code ( fioc & rocca - volmerange 1997 ) .
these templates assume a single formation epoch with an exponentially decaying star - formation history ( time constants @xmath77 1 , 2 , 4 , and 7 gyr ) and a kroupa ( 1993 ) imf ; the template grid spans ages of . for each galaxy in our main sample , we convolved the best - fit template spectrum with the @xmath58-band filter function to approximate the rest - frame @xmath58-band luminosity , @xmath78 .
this method allowed us to account for significant curvature in the rest - frame near - ir seds of our sources where photometric data points did not always overlap with the rest - frame @xmath58-band .
we note that 2546 ( @xmath099.1% ) of our late - type galaxies have at least one photometric data point lying at rest - frame wavelengths between 1 and 6 @xmath54 m , which significantly constrains the sed in the rest - frame @xmath58-band .
we adopted the prescription outlined in appendix 2 of bell ( 2003 ) to estimate @xmath75 using rest - frame @xmath76 color and @xmath58-band luminosity : @xmath79 the numerical constants in equation 2 were supplied by table 7 of bell ( 2003 ) and are appropriate for our choice of @xmath76 color and @xmath78 ; the normalization has been adjusted by @xmath80 dex to account for our adopted kroupa ( 2001 ) imf ( see @xmath61 1 ) .
we compared our stellar - mass estimates with those computed by borch ( 2006 ) for 1758 ( @xmath068% of our main sample ) galaxies in the combo-17 survey and find excellent agreement between methods , with an overall scatter of @xmath00.2 dex .
-band luminosity @xmath2 ( _ a _ ) , stellar mass @xmath3 ( _ b _ ) , and star - formation rate sfr ( _ c _ ) versus redshift for sources in our main sample . for comparison , in each panel we have indicated the properties of the mw and local galaxies m101 , m82 , and ngc 3256 . in figure 4@xmath81
, we have included only sfrs for the 888 late - type galaxies in our main sample that have 24@xmath54 m counterparts ( see @xmath61 3.3 for details ) ; the dashed curve indicates roughly the sfr detection limit for sources without 24@xmath54 m counterparts . detected normal galaxies and agns
have been indicated with open circles and diamonds , respectively .
the thick gray rectangles indicate regions where galaxy populations were selected for stacking ( see @xmath824.2 and 5.1 ) . for each galaxy . @xmath3 and
sfr were computed following equations 2 and 3 , respectively . ] in figure 4@xmath83 , we show @xmath3 versus redshift for our sample of late - type galaxies .
we note that our stellar - mass estimates are broadly limited by the @xmath34 criterion used in our sample selection . at @xmath70 ( @xmath84 ) our samples
are representative for late - type galaxies with @xmath85 ( @xmath86 ) . for comparison , in figure 4@xmath83
we have plotted stellar masses for the mw and local galaxies m101 , m82 , and ngc 3256 .
we adopted a stellar mass of @xmath87 @xmath88 for the mw ( e.g. , hammer 2007 ) .
for the local galaxies , we estimated @xmath3 using a similar relation to equation 2 , except using @xmath76 color and @xmath45-band luminosity ( see appendix 2 of bell 2003 for details ) ; for this computation , we utilized @xmath76 colors from rc3 and values of @xmath2 determined in @xmath61 3.1 .
this computation gives stellar masses of @xmath89 , @xmath90 , and @xmath91 @xmath88 for m101 , m82 , and ngc 3256 , respectively .
since the emission from normal late - type galaxies is known to be strongly correlated with sfr , it is of particular interest to understand how changes in sfr have contributed to the evolution of the normal late - type galaxy population . to calculate sfrs for the galaxies in our main sample , we utilized estimates of both the dust - uncorrected ultraviolet luminosities ( @xmath92 ) originating from young stars and the infrared luminosities (;
@xmath93 ) from the dust that obscures uv light in star - forming regions ( see , e.g. , kennicutt 1998 for a review ) .
the former quantity was computed following @xmath94(2800 ) , where @xmath95(2800 ) is the rest - frame 2800 monochromatic luminosity ( see @xmath61 3.2 of bell 2005 ) .
@xmath96(2800 ) was approximated using the template seds that were constructed in @xmath61 3.2 .
the latter quantity ( @xmath93 ) was computed using observed - frame 24@xmath54 m flux densities ( i.e. , rest - frame 24@xmath54m/@xmath97 ) from observations with the mips ( rieke 2004 ) camera onboard .
we note that generally the dust - obscured star - formation activity , probed here using @xmath93 , can be measured using either the uv spectral slope or optical nebular recombination lines ( e.g. , balmer emission lines and ) ; however , the data available for our @xmath98 01.4 late - type galaxies lack the rest - frame uv and spectroscopic coverage needed to provide such useful measurements . over the cdfs ,
deep 24@xmath54 m observations were available for the goods fields ( @xmath99 30 @xmath54jy , 6@xmath100 ) .
for the remaining area ( covering the outer regions of the ) , shallower 24@xmath54 m observations ( @xmath99 120@xmath101jy ) were available through swire ( lonsdale 2003 ) .
we matched the positions of the 2568 late - type galaxies in our main sample with those from the available 24@xmath54 m source catalogs , requiring that the and centroids be offset by no more than 15 .
we found successful matches for 888 ( @xmath035% ) of our late - type galaxies .
for these sources , we converted 24@xmath54 m flux densities to @xmath93 following the methods outlined in papovich & bell ( 2002 ) .
briefly , we utilized the entire grid of 64 infrared seds provided by dale ( 2001 ) to estimate the mean conversion factor @xmath102 , which transforms observed - frame 24@xmath54 m luminosity @xmath103(24@xmath104 ) to @xmath93 as a function of redshift [ i.e. , @xmath105 .
the mean conversion factor spans the tight range of over the entire redshift range .
we note that different choices of infrared seds yield similar results at @xmath35 , but become significantly discrepant at @xmath11 ( see , e.g. , fig . 2 of papovich & bell 2002 ) .
using our estimates of @xmath92 and @xmath93 , we calculated star - formation rates for galaxies in our sample using the following equation : @xmath106 where @xmath93 and @xmath92 are expressed in units of the solar bolometric luminosity ( @xmath107 ) .
equation 3 was adopted from @xmath61 3.2 of bell ( 2005 ) and was derived using pgase stellar - population models , which assumed a 100 myr old population with constant sfr and a kroupa ( 2001 ) imf ( see bell 2003 for further details ) .
we note that for the majority of our galaxies @xmath108 , and for sources with @xmath109 and @xmath110 , the median ir - to - uv ratio @xmath111 = 2.8 and 7.1 , respectively .
our @xmath112 galaxies have median @xmath113 , which is characteristic of objects that are found in the nearby @xmath114 mpc universe .
by contrast , our @xmath110 galaxies have median @xmath115 , similar to the local lirg population found at . in figure 4@xmath81
, we show the distribution of sfrs for the 888 galaxies in our main sample that had 24@xmath54 m counterparts .
we note that for sources in our main sample that were within the goods regions , where the 24@xmath54 m observations are most sensitive , the infrared detection fraction drops from @xmath0100% for galaxies with @xmath116 to @xmath020% for galaxies with @xmath117 .
this demonstrates that our sfr completeness is limited primarily by the 24@xmath54 m sensitivity limit and that our sample of @xmath34 late - type galaxies is highly representative of galaxy populations above the apparent redshift - dependent sfr limit shown in figure 4@xmath81 ( _ dashed curve _ ) . at @xmath70 ( @xmath1 )
, this limit corresponds to sfr @xmath118 ( sfr @xmath119 ) . for comparison , in figure 4@xmath81
we have plotted the sfrs for the mw and local galaxies m101 , m82 , and ngc 3256 .
the sfr of the mw was taken to be @xmath02 ( mckee & williams 1997 ) , and the sfrs of the local galaxies were computed using equation 3 .
values of @xmath96(2800 ) were approximated from seds available through the nasa / ipac extragalactic database ( ned ) and @xmath93 was taken from the iras revised bright galaxy sample ( sanders 2003 ) .
we find sfr = 2.5 , 6.2 , and 32.2 for m101 , m82 , and ngc 3256 , respectively . as a consistency check on our uv - plus - infrared sfr estimates
sfr(uv+ir ) , we calculated radio - derived sfrs using 1.4 ghz observations , sfr(1.4 ghz ) , following equation 7 of schmitt ( 2006 ) , which we adjusted to be consistent with our adopted kroupa ( 2001 ) imf .
we matched sources in our main sample to 1.4 ghz catalogs , which were derived from observations using the vla in the ( @xmath030 @xmath54jy ; richards 1998 ) and the atca in the ( @xmath060 @xmath54jy ; afonso 2006 ; rovilos 2007 ) . using a matching radius of 15
, we found a total of 54 radio sources coincident with our late - type galaxies .
we found that 48 ( @xmath089% ) of the radio - detected sources had 24@xmath54 m counterparts , allowing for reasonable comparison between derived sfrs .
for these sources we found reasonable agreement between sfrs derived from uv - plus - infrared and radio measurements , with a mean ratio of sfr(1.4 ghz)/sfr(uv+ir ) @xmath120
. a large number of late - type galaxies ( 840 sources ) in our main sample were detected in the 24@xmath54 m observations that were not detected at 1.4 ghz .
all but three of these sources had sfr(1.4 ghz ) upper limits that were consistent with that expected from estimates of sfr(uv+ir ) ( see @xmath61 4.1 for further details ) .
we utilized the multiwavelength observations in the cdfs to obtain a census of the active galaxies in our main sample .
we began by matching the optical positions of our galaxies to the positions of point sources in the cdf catalogs of alexander ( 2003)2 ms and @xmath01 ms . ] for the @xmath02 ms and @xmath01 ms and lehmer ( 2005b)250 ks . ] for the @xmath0250 ks .
for a successful match , we required that the optical and centroids be displaced by no more than 1.5 times the radius of the positional error circles ( confidence ) , which are provided in each respective catalog .
we note that for a small number of the galaxies in our sample at @xmath121 , moderately luminous off - nuclear sources ( e.g. , ultraluminous sources [ ulxs ] ) that are intrinsically related to the galaxies may lie outside of our adopted matching radius ; we utilized the off - nuclear source catalog of lehmer ( 2006 ) to identify such galaxies and assign properties ( see details below ) .
the source catalogs were generated using wavdetect ( freeman 2002 ) with false - positive probability thresholds of and @xmath122 for the alexander ( 2003 ) and lehmer ( 2005b ) point - source catalogs , respectively .
however , as demonstrated in @xmath61 3.4.2 of alexander ( 2003 ) and @xmath61 3.3.2 of lehmer ( 2005b ) , legitimate lower significance sources , detected by running wavdetect at a false - positive probability threshold of , can be isolated by matching with relatively bright optical sources ; therefore , when matching our late - type galaxies to -detected sources , we utilized this technique .
the sky surface density for late - type galaxies in our main sample ranges from @xmath023000 deg@xmath123 in the to @xmath08800 deg@xmath123 in the @xmath0250 ks .
the large difference between these source densities is primarily due to differences in applied redshift cuts ( see @xmath61 2.2 ) .
given the fact that the positional uncertainties are generally small ( @xmath1915 ) for sources within 60 of the aim points , as is the case for sources in our main sample , the corresponding estimated number of spurious matches is small .
we estimate that when using wavdetect with a false - positive probability threshold of to search for sources in three bandpasses ( , , and ) , we expect @xmath01.8 spurious matches . when including the off - nuclear sources from lehmer ( 2006 ) , we expect an additional @xmath00.5 false sources ; this brings our total spurious matching estimate to @xmath02.3 sources for our main sample . using the matching criteria above
, we find that 225 late - type galaxies are detected in at least one of the , , or bandpasses . out of these 225 galaxies , 12 are known off - nuclear sources from lehmer ( 2006 ) . since only one off - nuclear source is detected for each host galaxy , we assume that the off - nuclear point - source dominates the total emission from each host galaxy .
we therefore adopted the properties presented in table 1 of lehmer ( 2006 ) for each off - nuclear host galaxy . ) versus the logarithm of the flux ( @xmath124 ) for detected sources in our main sample .
off - nuclear sources catalogued by lehmer ( 2006 ) have been highlighted with open squares .
the shaded region represents sources with effective photon indices @xmath125 ; we classified these sources as agn candidates ( see @xmath61 4 , criterion 1 ) . for reference ,
we have plotted lines corresponding to @xmath126 0.5 , 1 , and 2 ( _ dashed lines _ ) .
, width=351 ] -band magnitude versus @xmath124 for detected sources in our main sample .
open circles represent sources that were classified as agn candidates by criterion 1 ( see @xmath61 4.1 and fig . 5 ) , and filled circles represent all other sources ; open squares have the same meaning as in figure 5 .
the shaded area represents the region where @xmath127 ; we classified sources in this region as agn candidates ( see @xmath61 4.1 , criterion 2 ) . for reference ,
the dashed lines represent @xmath128 , @xmath129 , and 1 .
, width=351 ] the unprecedented depths of the cdfs allow for the individual detection of luminous normal galaxies over the entire redshift range of our main sample ( @xmath17 01.4 ) ; however , the majority of the detected sources in even the most - sensitive regions of the cdfs will be distant ( @xmath130 ) agns ( e.g. , bauer 2004 ) , which we want to separate from our normal late - type galaxy sample .
we identified agn candidates using four primary criteria , which utilize ( 1 ) hardness to identify luminous obscured sources , ( 2 ) to optical flux ratios to identify additional relatively - unobscured agns , ( 3 ) the sfr correlation to identify additional lower - luminosity agns that are significantly influencing the total emission , and ( 4 ) the combination of radio and infrared properties to identify additional compton - thick agns and radio - loud agns . as a final check on our agn identifications we utilized optical spectroscopic information to identify sources with obvious agn signatures
( see criterion 5 below ) . in the sections below we provide details of each criteria . 1 .
_ x - ray hardness_.one unique signature of moderately obscured ( @xmath131 @xmath24 ) agn activity is a hard spectrum . for normal galaxies ,
the collective emission from binaries dominates the total power output . on average ,
these sources have observed power - law seds with spectral slopes of for lmxbs ( e.g. , church & baluciska - church 2001 ; irwin 2003 ) and for hmxbs and ulxs ( e.g. , sasaki 2003 ; liu & mirabel 2005 ; liu 2006 ) ; the presence of a significant hot interstellar gas component will steepen the resulting spectral slope ( i.e. , produce larger @xmath132 ) . to identify obscured agns in our sample effectively , we flagged sources having effective photon indices of @xmath125 as agn candidates .
we determined @xmath133 using the to hardness ratio @xmath134 , where @xmath135 is the count rate for each bandpass . a few sources have only detections .
since these sources were not detected in the bandpass , our most sensitive bandpass , there must be a significant contribution from the bandpass such that @xmath136 ( @xmath137 ) .
we therefore classified these sources as agn candidates . in figure 5
, we show the @xmath138 hardness ratio versus the logarithm of the flux @xmath139 for the detected sources in our main sample .
the shaded area highlights the region corresponding to @xmath125 .
sources that had only upper limits on @xmath138 that lie in the shaded region were not classified as agn candidates , while those with lower limits on @xmath134 that were in the shaded region were flagged as likely agns .
we note that occasionally an luminous ulx that is too close to the nucleus of its host galaxy to be identified as off - nuclear may have @xmath140 and therefore be classified as an agn candidate via this criterion ; however , since only one identified off - nuclear source has @xmath140 ( _ open squares _ in fig . 5 ; n.b .
, upper limits are not included ) , we do not expect such rare sources to significantly affect our results .
using criterion 1 , we identified a total of 71 obscured agn candidates .
x - ray to optical flux ratio_.detailed analyses of the spectra of luminous agns in the @xmath01 ms show that the intrinsic agn power - law photon index is relatively steep , ( e.g. , tozzi 2006 ) .
therefore , luminous agns having column densities of @xmath141 @xmath24 will often have effective photon indices of @xmath142 and would not have been classified as potential agns by criterion 1 . in order to identify luminous agns with @xmath141 @xmath24
, we utilized the to optical flux ratio ( @xmath143 ) as a discriminator of agn activity ( e.g. , maccacaro 1988 ; hornschemeier 2000 ; bauer 2004 ) .
we identified sources with @xmath144 ( see criterion 3 below for justification ) as unobscured agn candidates . in figure 6 , we show the @xmath50-band magnitude versus @xmath124 for sources in our main sample ; the shaded area shows the region where @xmath145 .
sources that were classified as agn candidates via criterion 1 are denoted with open symbols . in total
, 43 detected sources satisfied criterion 2 , and 16 of these sources were uniquely identified using this criterion ( i.e. , not identified by criterion 1 ) . versus sfr for detected sources in our sample that have 24@xmath54 m counterparts .
open circles represent sources that were classified as agn candidates via criteria 1 and 2 ( see @xmath61 4.1 and figs . 5 and 6 ) , and filled circles represent all other sources ; open squares have the same meaning as in figure 5 .
we have shown the -sfr relations calibrated by bauer ( 2002b ) , ranalli ( 2003 ) , gilfanov ( 2004a ) , and persic & rephaeli ( 2007 ) ; each respective curve has been annotated in the figure .
the shaded area above the thick dashed line represents the region where @xmath146 is three times larger than its value predicted by persic & rephaeli ( 2007 ) ; we classified detections that lie in this region as agn candidates ( see @xmath61 4.1 , criterion 3 ) .
, width=359 ] 3 .
_ x - ray to sfr correlation_.taken together , criteria 1 and 2 provide an effective means for identifying agns that are affected by large absorption column densities ( criterion 1 ) and those that are notably overluminous for a given optical luminosity ( criterion 2 ) .
however , these criteria will still miss moderately luminous unobscured agns that have @xmath147 ( see , e.g. , peterson 2006 ) .
although an accurate classification for _ all _ such sources is currently not possible , the situation can be mitigated using the available multiwavelength data .
we therefore exploited the correlation between luminosity @xmath4 and sfr ( see @xmath61 1 ) to identify additional agn candidates in our main sample that have significant excesses over what is expected based on the -sfr correlation . in order to calculate the rest - frame luminosity ( @xmath148 ; where @xmath149 and @xmath150 are the photon - energy lower and upper bounds , respectively ) of a source having a power - law sed , we used the following equation : @xmath151 where @xmath152 is the observed - frame emission in the bandpass and @xmath153 is the luminosity distance . using equation 4 and an adopted photon index of @xmath154 , we calculated luminosities for detected sources in our sample . in figure 7
, we show the logarithm of the luminosity @xmath155 versus sfr ( computed following equation 3 ) for galaxies in our main sample that had 24@xmath54 m counterparts ( see @xmath61 3.3 ) .
several estimates of the correlation have been shown for reference ( bauer 2002a ; ranalli 2003 ; gilfanov 2004a ; persic & rephaeli 2007 , hereafter pr07 ) ; these correlations have been corrected for differences in bandpass and sed as well as adopted imfs .
hereafter , we adopt the correlation from pr07 for comparisons ; however , the use of other correlations would yield similar results and conclusions .
open squares show the locations of the galaxies hosting off - nuclear sources , which appear to be preferentially located near the correlation .
open circles indicate sources that were identified as agn candidates via criteria 1 and 2 . generally , these agns have @xmath146/sfr @xmath156 times that predicted by the pr07 correlation ( _ thick dashed line _ ) , which corresponds to a factor of @xmath682.5 times the rms scatter of the pr07 correlation .
we therefore classified all sources in this regime ( _ shaded region _ ) as agn candidates .
we note that sources having only upper limits were not classified as agns .
however , sources that were detected in the bandpass that had only upper limits on the sfr were classified as agn candidates if they were within the shaded region of figure 7 . using this criterion , we identified 101 potential agn candidates , of which 33 were unique to criterion 3 .
for the 124 detected sources that were not classified as agns , we found that 19 sources ( @xmath015% ) had spectral properties indicative of agns ( criterion 1 ) and only one source ( @xmath00.8% ) had an to optical flux ratio elevated in comparison to normal galaxies ( criterion 2 ) . versus redshift for the 225 detected sources in our main sample .
open circles indicate all of the agn candidates that were isolated using the four criteria outlined in @xmath61 4.1 , and filled circles represent potential normal galaxies ; open squares have the same meaning as in figure 5 .
the estimated detection limits for the @xmath02 ms , the @xmath01 ms , and the @xmath0250 ks have been indicated with dotted , dashed , and dot - dashed curves , respectively . ,
width=359 ] 4 . _ infrared and radio properties_.in addition to the detected sources , we also expect there to be additional agns that lie below the detection threshold that are not identified here . in @xmath61 3.3 , we noted using 48 sources that the sfrs derived from the uv - plus - infrared emission are generally consistent with those derived from radio ( 1.4 ghz ) observations ; however , there are a few sources that are significantly scattered ( i.e. , by more than a factor of five times the intrinsic scatter ) outside of the correlation .
these sources have either ( 1 ) excess infrared emission due to reprocessed dust emission from a highly - obscured ( @xmath157 @xmath24 ) moderately luminous ( intrinsic @xmath158@xmath159 ) compton - thick agn ( e.g. , daddi 2007a , b ) or ( 2 ) an excess of radio emission due to prominent radio - emitting jets ( e.g. , snellen & best 2001 ) . the reasonable agreement between sfr(uv+ir ) and sfr(1.4 ghz ) suggests that such agns are not prevalent in our late - type galaxy sample . ideally we would also like to explore how uv - corrected sfrs compare to the uv - plus - infrared sfr to search for potential compton - thick agns following the approach used by daddi ( 2007a , b ) for @xmath160 galaxies at @xmath161 ; however , since our galaxies generally lie at much lower redshifts that the daddi ( 2007a , b ) sources , we are unable to constrain well uv spectral slopes with our available photometry , which would be required to make dust - corrections to the observed uv sfrs ( see @xmath61 3.3 for discussion ) . for our sample of 24@xmath54m - detected galaxies
, we identified three radio - excess galaxies with sfr(1.4 ghz)/sfr(uv+ir ) @xmath162 and three ir - luminous galaxies with sfr(1.4 ghz)/sfr(uv+ir ) @xmath163 . using these criteria we identified three new agn candidates , two radio luminous ( and ) and one ir luminous ( ) ; one of these sources , the radio luminous source , was detected in the and bandpasses .
as an additional test , we utilized irac photometry ( see @xmath61 2.3 ) to search for infrared power - law sources having near - ir spectral properties characteristic of luminous agns ( e.g. , alonso - herrero 2006 ; donley 2007 ) .
when searching for power - law sources , we adopted the criteria discussed in donley ( 2007 ) .
we found that no sources in our sample satisfied these criteria , which is consistent with the finding that most irac power - law sources reside at @xmath11 .
_ optical spectroscopy_.as a final check on our agn classifications , we searched the optical spectroscopic catalogs available for cdf sources ( e.g. , barger 2003 ; le fevre 2004 ; szokoly 2004 ; wirth 2004 ; vanzella 2005 , 2006 ) to isolate additional luminous agns in our sample . in total , we found 15 galaxies in our main sample that were classified as agns via optical spectroscopy and all of these sources had been identified as agn candidates by the previous criteria .
we note that the majority of the detected agns have moderate luminosities ( intrinsic @xmath164@xmath159 ) and therefore often have high - excitation agn emission lines that are too faint with respect to stellar emission to be identified via optical spectroscopy ( e.g. , moran 2002 ) .
therefore , it is not surprising that we do not find any additional agns using this criterion . to summarize , in the band
we have detected a total of 225 ( @xmath09% ) late - type galaxies out of the 2568 sources in our main sample . using the criteria presented above
, we classified 121 detected sources as agn candidates ( with an additional two undetected agn candidates via criterion 4 ) .
the remaining 104 -detected sources that we do not classify as agn candidates are considered to be normal late - type galaxies , and we include these galaxies in subsequent stacking analyses ( see details in @xmath61 5 below ) . thus we use 2447 late - type galaxies in our stacking analyses . in figure 8 , we show @xmath155 versus redshift for detected sources in our main sample , and in table 1 ( _ available electronically _ ) we summarize their properties .
agn candidates are denoted with open circles and normal galaxies are plotted with filled circles .
we note that the above criteria are not completely sufficient to classify all detected sources that are truly agns as agn candidates ( see , e.g. , peterson 2006 ) .
such a misclassification is possible for low - luminosity agns that are only detected in the more sensitive bandpass , which have emission too weak for an accurate classification . in @xmath61 5.2 ,
we use the agn fraction as a function of luminosity to argue quantitatively that we do not expect misclassified agns ( detected only in the bandpass ) and low - luminosity agns below the detection limit to have a serious impact on our results .
ccccccccc [email protected]
& 17.19 & 0.15@xmath166 & 10.39 & 10.49 & @xmath1675.89 & e - cdf - s 03 & 217 & @xmath16515.06 + [email protected] & 22.17 & 0.15@xmath166 & 8.85 & 8.44 & 0.23 & e - cdf - s 03 & 217 & @xmath16515.45 + [email protected] & 17.06 & 0.25@xmath166 & 10.94 & 11.07 & 1.40 & e - cdf - s 03 & 206 &
@xmath167 @xmath16515.15 + [email protected] & 20.14 & 0.47@xmath168 & 10.14 & 10.43 & @xmath16719.93 & e - cdf - s 02 & 226 & @xmath16513.46 + [email protected] & 19.58 & 0.51@xmath166 & 10.84 & 10.54 & @xmath16713.17 & e - cdf - s 03 & 239 & @xmath16515.02 + cccccccccc @xmath169 9.510.0 & @xmath170 & 120 & 9 & 213.3 & 158.7 & 54.9 & 8.6 & 9.6 & 3.0 + @xmath169 9.510.0 & @xmath171 & 363 & 1 & 176.7 & 152.6 & 25.3 & 4.8 & 6.8 & 0.9 + + @xmath169
10.010.5 & @xmath172 & 28 & 2 & 66.9 & 46.5 & 20.2 & 5.2 & 5.3 & 2.1 + @xmath169 10.010.5 &
@xmath173 & 33 & 3 & 82.9 & 68.5 & 14.9 & 5.9 & 6.7 & 1.5 + @xmath169 10.010.5 & @xmath174 & 67 & 7 & 204.2 & 144.7 & 59.8 & 9.3 & 9.7 & 3.8 + @xmath169 10.010.5 & @xmath175 & 158 & 6 & 347.5 & 240.0 & 109.1 & 10.7 & 11.3 & 4.4 + @xmath169 10.010.5 & @xmath176 & 100 & 3 & 128.0 & 109.0 & 18.7 & 6.0 & 7.5 & 1.2 + @xmath169 10.010.5 & @xmath177 & 71 & 2 & 154.1 & 93.5 & 60.6 & 6.3 & 6.1 & 3.2 + @xmath169 10.010.5 & @xmath178 & 62 & 2 & 156.3 & 107.4 & 48.8 & 6.4 & 6.9 & 2.6 + @xmath169 10.010.5 & @xmath179 & 57 & 3 & 105.9 & 64.8 & 41.2 & 4.4 & 4.5 & 2.2 + the majority of the normal late - type galaxies that make up our main sample were not detected individually in the bandpass . in order to study the mean properties of these sources
, we implemented stacking analyses of galaxy populations selected by their physical properties ( see @xmath61 3 ) .
we divided our main sample into subsamples ( to be used for stacking ) of normal late - type galaxies selected by both physical properties ( i.e. , @xmath45-band luminosity , stellar mass , and star - formation rate ) and redshifts . in figure 4
, we have highlighted the divisions of our sample with thick gray rectangles , and for normal late - type galaxies in each region , we used stacking analyses to constrain average properties . for each of the subsamples
defined above , we performed stacking in each of the three standard bandpasses ( i.e. , sb , hb , and fb ; see @xmath61 1 ) .
we expect these bandpasses to sample effectively power - law emission originating from binaries ( i.e. , hmxbs and lmxbs ) with a minor contribution from hot interstellar gas in the sb for late - type galaxies at @xmath180 . in our analyses , we used data products presented in alexander ( 2003 ) for the @xmath02 ms and @xmath01 ms and lehmer ( 2005b ) for the @xmath0250 ks ( see footnotes 15 and 16 ) .
our stacking procedure itself was similar to that discussed in @xmath61 3 of steffen ( 2007 ) .
this procedure differs from past stacking analyses ( e.g. , lehmer 2007 ) in how the local background of each stacked sample is determined , and produces results that are in good agreement with the method discussed in @xmath61 2.2 of lehmer ( 2007 ) . for completeness
, we have outlined this procedure below . using a circular aperture with radius @xmath181 , we extracted source - plus - background counts @xmath182 and exposure times @xmath183 ( in units of cm@xmath36 s ) for each galaxy using images and exposure maps , respectively . for a given source
, we used only pointings with aimpoints ( see footnote 12 ) that were offset from the source position by less than 60 ; hereafter , we refer to this maximum offset as the inclusion radius , @xmath185 .
has the same meaning as it did in lehmer ( 2007 ) . ]
since the psf increases in size with off - axis angle and degrades the sensitivity for sources that are far off - axis , our choices of @xmath186 and @xmath185 have been chosen to give the maximal stacked signal with the majority of the psfs being sampled by our stacking aperture ( see @xmath61 2.2 and fig . 3 of lehmer 2007 for justification ) .
we estimate , based on stacked images of @xmath98 optically luminous ( @xmath187 ) late - type galaxies , that @xmath1910% of the normal galaxy emission originates outside of our @xmath188 radius aperture ( at ) for all stacked samples . for galaxies that were within 60 of more than one of the cdf aimpoints ,
we added source counts and exposure times from all appropriate images and exposure maps , respectively ; however , there were very few sources in our main sample that met this criterion . using background maps ( see @xmath61 4.2 of alexander 2003 for the @xmath02 ms and @xmath01 ms and @xmath61 4 of lehmer 2005b for the @xmath0250 ks ) and exposure maps , we measured local backgrounds @xmath189 and exposure times @xmath190 within a 30 pixel @xmath191 30 pixel ( @xmath0@xmath192 ) square , centered on each source with the 15 radius circle masked out . here again , if a source was within 60 of more than one of the cdf aimpoints , we summed the local backgrounds and exposure times .
we estimated the expected number of background counts in each circular aperture @xmath193 by scaling the background counts within the square by the relative exposure times of the circular aperture and the square ( i.e. , @xmath194 ) .
this approach is similar to scaling the background counts in the square by the relative areas of the circular aperture and the square ; however , by using exposure times , we are able to account more accurately for spatial variations in pixel sensitivity due to chip gaps , bad pixels , and vignetting .
furthermore , comparisons between this method and the monte carlo method used in lehmer ( 2007 ) to compute @xmath193 give excellent agreement and are most convergent for large numbers of monte carlo trials . when stacking galaxy populations , we excluded sources that were ( 1 ) classified as agn candidates ( via the criteria outlined in @xmath61 4.1 ) , ( 2 ) within 10of an unrelated source detected in the catalogs , ( 3 ) within the extent of extended sources ( see bauer 2002b , @xmath61 3.4 of giacconi 2002 , and @xmath61 6 of lehmer 2005b ) , and ( 4 ) located within 3of another late - type galaxy in our main sample .
we note that we include detected normal galaxies when stacking our samples , since we are interested in the average properties of the normal late - type galaxy population .
we tested the effects of including such sources by stacking samples both with and without detected sources included , and find similar results for both cases .
typically , the detected sources contribute of the counts in the total stacked signal .
for each of the subsamples of normal late - type galaxies outlined in figure 4 ( _ gray rectangles _ ) , we determined stacked source - plus - background ( @xmath195 ) and background counts ( @xmath196 ) to determine net counts ( @xmath197 ) . for each stacked sample , we required that the signal - to - noise ratio [ s / n @xmath198 ; where @xmath199 and @xmath200 be greater than or equal to 3 ( i.e. , @xmath6899.9% confidence ) for a detection . for stacked samples without significant detections ,
3@xmath100 upper limits were placed on the source counts .
we converted the net counts obtained from each stacked sample to absorption - corrected fluxes and rest - frame luminosities using a power - law sed with @xmath201 .
due to the fact that our 15 radius stacking aperture encircles only a fraction of the psf , our 15 radius circular aperture contains an encircled - energy fraction of @xmath0100% , 80% , and 100% for the sb , hb , and fb , respectively ; however at @xmath202 , this fraction decreases to @xmath030% , 25% , and 25% , respecively . ] for sources at relatively large off - axis angle , we calculated aperture corrections @xmath203 for each stacked source @xmath204 .
since we are calculating average counts from the summed emission of many sources of differing backgrounds and exposure times , we used a single , representative exposure - weighted aperture correction , @xmath205 .
this factor , which was determined for each stacked sample , was calculated as follows : @xmath206 where @xmath207 .
the average aperture corrections ( @xmath205 ) for sources in our main sample were @xmath01.6 , 1.8 , and 1.7 for the sb , hb , and fb , respectively . using our adopted sed ,
we estimated observed mean fluxes using the following equation : @xmath208 where @xmath209 is a bandpass - dependent count - rate to flux conversion factor that incorporates both the sed information as well as galactic extinction using the column densities listed in @xmath61 1 .
these mean fluxes were then converted to rest - frame luminosities following equation 4 , assuming a photon index of @xmath154 . ) versus the logarithm of the to optical flux ratio ( @xmath210 ) for 44 stacked samples selected via observed properties : @xmath2 ( _ filled circles _ ) , @xmath3 ( _ filled squares _ ) , and sfr ( _ filled triangles _ ) .
the characteristic mean error bar for each quantity is given in the lower right - hand corner .
the median logarithm of the to optical flux ratio is indicated with a vertical dotted line ( @xmath211 ) .
the median effective photon index for the samples that were detected in both the sb and hb is indicated with a horizontal dotted line ( @xmath212 ) .
we note that all of the stacked samples with both sb and hb detections have @xmath142 , consistent with normal galaxies . for the several stacked samples that have only sb detections , we have indicated lower - limits on @xmath132 .
the shaded regions and corresponding boundaries ( _ dashed lines _ ) represent areas where detected sources were classified as agn candidates ( for details , see discussion of criteria 1 and 2 in @xmath61 4.1 ) ; note that additional criteria were used to identify potential agns when generating our samples of normal late - type galaxies ( see criteria 35 in @xmath61 4.1 ) .
, width=359 ]
using the stacking analysis methods discussed in @xmath61 4.2 , we stacked the late - type galaxy samples presented in figure 4 ( stacked samples are denoted with _ thick gray rectangles _ ) .
these samples were selected using @xmath2 , @xmath75 , and sfr , which include 14 , 17 , and 13 stacked samples ( 44 total ) , respectively . in table 2
( _ available electronically _ ) , we tabulate our stacking results .
we found significant ( i.e. , s / n @xmath68 3 ) detections in the and bandpasses for all stacked samples . in the bandpass ,
15 out of the 44 stacked samples were detected , and these samples generally constitute the most optically luminous and massive galaxies , as well as those galaxies that are most actively forming stars . in figure 9 , we show the effective photon index ( @xmath132 ) versus the logarithm of the to optical mean flux ratio ( @xmath213 ) for our 44 stacked samples ( _ filled symbols _ ) that were selected via their observed properties .
effective photon indices were estimated using hb - to - sb count - rate ratios ( i.e. , @xmath214 ) .
all 44 stacked samples have to optical flux ratios and spectral slopes consistent with normal galaxies ( _ unshaded region _ in fig .
9 ) , suggesting that these samples are not heavily contaminated by an underlying population of agns . ) .
these samples have and and lie at @xmath215 , a redshift where large - scale sheets of galaxies and agns have been isolated previously in the ( e.g. , gilli 2003 ) .. these galaxies have spectral slopes that are either statistically scattered towards @xmath216 ( see characteristic error bar in fig . 9 ) or may have some non - negligible contribution from heavily - obscured agns . ]
we find a median logarithm of the to optical flux ratio of @xmath217 ( _ vertical dotted line _ ) , and for samples that were detected in both the hb and sb , the median effective photon index is @xmath218 ( _ horizontal dotted line _ ) .
these values are characteristic of galaxies dominated by binary populations .
-band mean luminosity ratio ( @xmath219 ) , ( _ b _ ) the to stellar - mass mean ratio ( @xmath220 ) , and ( _ c _ ) the to - star - formation - rate mean ratio ( @xmath221sfr ) versus redshift ( _ filled symbols and curves _ ) for stacked normal late - type galaxy samples selected by @xmath2 ( fig .
4@xmath69 ) , @xmath3 ( fig .
4@xmath83 ) , and sfr ( fig .
4@xmath81 ) , respectively . for comparison
, in each panel we have indicated the properties of the mw and local galaxies m101 , m82 , and ngc 3256 . quoted luminosities correspond to the bandpass and were calculated following the methods described in @xmath61 4.2 , assuming a power - law sed with photon index of @xmath201 .
symbols and curves correspond to unique ranges of @xmath2 , @xmath3 , and sfr , which are annotated in each respective figure . for reference , in figures 10@xmath69 and 10@xmath83 we have plotted the corresponding values of @xmath222 and @xmath223 , respectively , for normal late - type galaxies in the local universe ( _ open symbols _ ) using the s01 sample . in figure 10@xmath83
we show the expected lmxb contribution based on gilfanov ( 2004b ; _ dashed line _ ) .
finally , in figure 10@xmath81 we show the local -sfr relation and its dispersion ( _ dashed line with shading _ ) derived by pr07 and corrected for our choice of imf and @xmath201 .
we note that roughly all of our data points in figure 10@xmath81 lie @xmath0 dex below the pr07 relation , which is likely due to systematic differences in how @xmath4 and sfr were determined between studies .
, width=343 ] in figure 10 , we show the logarithm of the ratio of the luminosity ( hereafter , @xmath4 ) to each physical property ( i.e. , @xmath219 , @xmath223 , and @xmath221sfr ) versus redshift for our samples .
each relevant quantity for the mw and local galaxies m101 , m82 , and ngc 3256 have been shown for comparison . we adopt @xmath224 as the approximate luminosity of the mw ( grimm 2002 ) . for the local galaxies
, we utilized the luminosities from shapley ( 2001 ; hereafter s01 ) for m101 ( @xmath225 ) and m82 ( @xmath226 ) and lira ( 2002 ) for ngc 3256 ( @xmath227 ) .
we have corrected the luminosities to be consistent with our use of the bandpass and our choice of @xmath201 . for the purpose of comparing our results to those for nearby late - type galaxy populations
, we made use of the s01 sample of 183 normal local ( @xmath228 100 mpc ) spiral and irregular galaxies .
these galaxies were observed in the band using the ipc and hri ( fabbiano 1992 ) and agns , with luminous x - ray emission or spectral signatures indicative of agn activity , have been excised from the sample . to avoid the inclusion of early - type s0 galaxies , we chose to utilize 139 normal late - type galaxies from the s01 sample with morphological types @xmath229 ( hubble types ) .
the s01 sample covers ranges of @xmath2 and @xmath3 that are well - matched to our main sample and are representative of late - type galaxies in the local universe . in order to compute mean
luminosities for samples that were directly comparable with our results , we divided the s01 sample into the same intervals of @xmath2 and @xmath3 that were used for our main sample ( see figs .
4@xmath69 and 4@xmath83 ) .
since several of the s01 galaxies had only upper limits available , we computed mean luminosities and errors using the kaplan - meier estimator available through the astronomy survival analysis software package ( asurv rev .
1.2 ; isobe & feigelson 1990 ; lavalley 1992 ) ; the kaplan - meier estimator handles censored data sets appropriately . when computing these mean luminosities , we filtered the s01 samples appropriately into distance intervals to avoid the malmquist bias . in figures 10@xmath69 and 10@xmath83 ,
we show the corresponding values of @xmath230 and @xmath223 , respectively , for the s01 sample with open symbols .
by contrast , the sfrs of the local sample are generally too low ( @xmath19110 ) to provide a meaningful comparison with our distant 24@xmath54m - detected sources .
this is due to the strong positive evolution of the star - formation rate density with redshift ( see @xmath61 1 ) , which makes sfrs that are common for galaxies in our sample ( @xmath68110 ) comparatively rare at @xmath231 . from figures 10@xmath69 and 10@xmath83 , it is apparent that there is significant positive redshift evolution in @xmath222 and @xmath223 over the redshift range of . for each of the six total selection ranges of @xmath2 and @xmath3 ,
the redshift progression of luminosities is inconsistent with a constant at the @xmath23299.9% confidence level .
for the most optically luminous ( @xmath233320 @xmath234 ) and massive ( @xmath235 120 @xmath234 ) late - type galaxies at @xmath236 , @xmath237 and @xmath238 are measured to be larger than the local values of s01 by factors of @xmath239 and @xmath240 , respectively .
such values are consistent with the @xmath0@xmath241 evolution of @xmath13 found for the normal late - type galaxy population , which has been constrained using largely the most optically luminous and massive galaxies ( georgakakis 2006 ; ptak 2007 ; tzanavaris & georgantopoulos 2008 ) .
the above results confirm the increase in @xmath237 with redshift found by h02 . in past studies ,
@xmath237 has been used as a proxy for star - formation activity ( e.g. , ptak 2001 ; h02 ; lehmer 2005a ) , despite the fact that @xmath2 is likely to be somewhat sensitive to sfr ( see discussion in @xmath61 3.1 ) . as discussed in @xmath61 1 ,
@xmath4 has been shown to be strongly correlated with galaxy sfr .
correlation studies of spiral galaxies in the local universe have also found strong correlations between @xmath4 and @xmath242 such that @xmath243 ( s01 ; fabbiano & shapley 2002 ) . the nonlinear relationship between and @xmath45-band emission is thought to be due to the increase in dust obscuration with star - formation activity , which attenuates light from the @xmath45-band more effectively than it does in the band .
using our sample of 47 normal late - type galaxies with both and 24@xmath54 m detections , we found that @xmath237 was positively correlated with the uv dust - extinction measure ( @xmath93+@xmath244)/@xmath92 ( kendall s @xmath245 ; @xmath6899.99% confidence level ) , thus providing support for this hypothesis .
furthermore , it has been shown that @xmath237 is correlated with @xmath246 , suggesting that more intense emission is associated with hotter ir colors , which is indicative of intense obscured star - formation activity .
therefore , the increase in @xmath237 with redshift ( see fig .
10@xmath69 ) observed for our galaxies is likely due to an increase in their star - formation activity . upon comparing @xmath2 with @xmath3 and sfr for our galaxies
, we find that @xmath2 is well correlated with both @xmath3 ( kendall s @xmath247 ) and sfr ( @xmath248 ) ; however , the scatter in the @xmath2@xmath3 relation is significantly smaller than it is for the @xmath2sfr relation ( @xmath00.2 dex versus @xmath00.4 dex , respectively ) implying that @xmath2 traces more effectively @xmath3 rather than sfr .
it is therefore not surprising that we see similar redshift evolution of @xmath238 and @xmath237 . in figure 10@xmath83 ,
we show the estimated lmxb contribution to @xmath249 ( _ dashed line _ ) based on table 5 of gilfanov ( 2004b ) .
this value is @xmath0510 times lower than all mean values of @xmath250 , suggesting that on average lmxbs play a fairly small role in the emission from our stacked samples .
furthermore , late - type galaxies in the local universe with similar stellar masses are often found to have hmxb emission that is @xmath0210 times more luminous than that expected from lmxbs ( see _ open symbols _ in fig .
10@xmath83 and fig . 3 of gilfanov 2004c ) . for galaxy samples selected via sfr
, we find no evidence for significant evolution in @xmath221sfr ( fig .
10@xmath81 ) .
we note that _ if _ the majority of the flux from our stacked samples originated from detected sources , then the agn selection criteria 3 and 4 from @xmath61 4.1 may potentially bias these results .
however , we find that the more populous and unbiased undetected source populations dominate the stacked signals at the @xmath6880% level ( see column 14 of table 2 )
. for each of the three ranges of sfr , @xmath221sfr is consistent with a constant value , and has a best - fit ratio of @xmath221sfr @xmath251 ( @xmath252 for 12 degrees of freedom ) .
these results suggest that the contribution from lmxbs is small and that the integrated emission from our late - type galaxies is dominated by hmxbs .
this implies that the evolution of our late - type galaxy samples is likely due to changes in star - formation activity .
cccccc @xmath2 & 11 & 25.37 & @xmath253 & @xmath254 & @xmath255 + @xmath3 & 14 & 25.73 & @xmath256 & @xmath257 & @xmath258 + sfr & 10 & 9.72 & @xmath259 & @xmath260 & @xmath261 sfr versus the logarithm of the luminosity @xmath262 for normal late - type galaxies . detected sources from our main sample are indicated as small gray dots .
sources that were not detected in the bandpass but were detected in either the or bandpasses are shown as upper limits .
results from our stacking analyses of late - type galaxies selected via observed sfr are shown as large filled circles , squares , and triangles , which have the same meaning as in figure 10@xmath81 ; @xmath263 lbgs that were both uncorrected and corrected for agn contamination have been shown plotted as a filled star and open star , respectively .
for comparison , we show the local galaxy sample from pr07 , which includes normal late - type galaxies ( _ open circles _ ) and ultraluminous infrared galaxies ( ulirgs ; _ open squares _ ) ; the best - fit pr07 relation is shown as a solid curve .
the sfrs for the mw and local galaxies m101 , m82 , and ngc 3256 are indicated ( _ horizontal dashed lines _ ) . ,
width=377 ] since the emission from our late - type galaxies is dominated by star - formation processes , we note that our @xmath3-selected stacking results provide a relatively unobscured measure of the star - formation activity per unit stellar mass ( i.e. , the ssfr ; see fig .
10@xmath83 ) .
we find that at @xmath264 the emission per unit stellar mass is a factor of @xmath0 larger for galaxies with versus that observed for galaxies with . at @xmath7 1 , we find that @xmath265 is larger than its local value ( s01 ) by factors of @xmath266 and @xmath267 for late - type galaxies with and , respectively .
these results are broadly consistent with observed differences in the mean ssfrs found by zheng ( 2007 ) for @xmath264 galaxies of comparable stellar masses , and imply that the lower - mass galaxies are undergoing more significant stellar mass growth over than more massive galaxies . in order to quantify the dependences of the luminosity on redshift and physical properties
, we performed multivariate parametric fitting to our stacked data . for each galaxy sample selected via @xmath2 , @xmath3 , and sfr , we investigated the redshift evolution of the luminosity . for this analysis
, we fit our data to a power - law parametric form : @xmath268 where @xmath269 is a place holder for each of the three physical properties ( @xmath2 , @xmath3 , and sfr ) used for our sample selections , and @xmath270 , @xmath45 , and @xmath271 are fitting constants . for each sample
, we utilized our stacking results , equation 7 , and @xmath272 fitting to estimate the constants @xmath270 , @xmath45 , and @xmath271 . for our adopted three - component model
, we constrained @xmath270 , @xmath45 , and @xmath271 using 90% confidence errors ( @xmath273 ) .
the s01 local data points were not used for these fits due to differences in galaxy selection , instrument calibration , and agn identification . versus @xmath274 for late - type galaxies with
@xmath275 at @xmath276 ( _ filled circles _ ) and @xmath98 0.61 ( _ open triangles _ ) . for reference
, we have indicated the corresponding agn fraction for lyman break galaxies at @xmath263 ( _ filled stars _ ; see @xmath61 5 for further details ) .
we have indicated the median detection limit for galaxies in each redshift range ( _ downward - pointing arrows along the x - axis _ ) .
the solid curve represents the best - fit relation for @xmath277 , which was fit using all late - type galaxies with @xmath98 01 and @xmath278 in our main sample . for reference ,
we have shown the estimated agn fraction for galaxies with @xmath279 , @xmath280 , and @xmath281 ( _ dotted curves _ ; see @xmath61 5.2 for further details ) .
, width=351 ] in table 3 , we tabulate our constraints on @xmath272 , @xmath270 , @xmath45 , and @xmath271 for these fits .
we find that this particular choice ( i.e. , eqn .
7 ) of parameterization does not provide acceptable fits for galaxy samples selected via @xmath2 and @xmath3 .
however , for galaxy samples selected via sfr , we find a good fit for this parameterization ( @xmath282 for ten degrees of freedom ) .
we constrain the evolution of @xmath262/sfr to be independent of or at most weakly dependent on redshift [ @xmath283 .
based on radio observations of distant star - forming galaxies with sfr @xmath0 in the and , barger ( 2007 ) reported that the upper limits for undetected sources were below the level expected from the -sfr correlaiton , thus suggesting that the correlation may not hold in the high - redshift universe .
however , our stacking results suggest that the -sfr correlation _ does _ in fact hold for average galaxies with sfr = ( sfr = ) out to @xmath284 ( @xmath5 ) . for illustrative purposes we have created figure 11 , which shows @xmath285 sfr versus @xmath285 for normal star - forming galaxies selected from several different sources including detected galaxies from our main sample ( _ filled gray circles and limits _ ) , stacked galaxies from this study ( _ filled black symbols _ ) , local galaxies from pr07 ( _ open circles _ ) , local ultraluminous infrared galaxies from pr07 ( ulirgs ; _ open squares _ ) , and stacked @xmath263 lyman break galaxies ( lbgs ; _ stars _ ; see @xmath61 6 for details ) . for all data used in this plot
, we have normalized sfrs appropriately to be consistent with our adopted kroupa ( 2001 ) imf and have adjusted luminosities to correspond to the band using a @xmath154 power - law sed .
the best - fit -sfr correlations for local galaxies from pr07 ( _ solid curve _ ) and @xmath17 01.4 late - type galaxies from this study ( @xmath262/sfr = 39.87 ; _ dotted curve _ ) have been shown for reference .
for comparison , we have shown the sfrs for the mw and local galaxies m101 , m82 , and ngc 3256 . in this section
we estimate the contribution to our stacked signals from contaminating agns that have luminosities below our detection limit .
this analysis is similar in nature to that in @xmath82 3.1 and 3.2.2 of lehmer ( 2007 ) , which was performed for early - type galaxies .
we implement the observed cumulative agn fraction @xmath286 : the fraction of galaxies harboring an agn with luminosity of @xmath274 or greater .
hereafter , we compute @xmath286 by taking the number of candidate agns in a particular galaxy sample with a luminosity of @xmath287 or greater and dividing it by the number of galaxies in which we could have detected an agn with luminosity @xmath274 .
the latter number is computed by considering the redshift of each galaxy and its corresponding sensitivity limit , as obtained from spatially varying sensitivity maps ( see @xmath61 4.2 of alexander 2003 and @xmath61 4 of lehmer 2005b ) ; these sensitivity maps were calibrated empirically using sources detected by wavdetect at a false - positive probability threshold of @xmath288 .
the quantity @xmath286 is not only a function of @xmath274 , but is also dependent on the selection of the galaxy sample considered : in our case redshift and the physical property @xmath269 ( i.e. , @xmath2 , @xmath3 , and sfr ) of the galaxy population may plausibly play a role in @xmath286 . in our analyses , we assume that each of the respective dependencies ( i.e. , @xmath287 , @xmath289 , and @xmath269 ) are independent of each other , such that @xmath290 , where @xmath291 , @xmath292 , and @xmath293 represents the functional dependence of the cumulative agn fraction for each indicated variable @xmath274 , @xmath289 , and @xmath269 , respectively .
we made use of the bandpass because of its ability to probe relatively unattenuated emission in a regime of the spectrum where we expect there to be minimal emission from normal galaxies ( see also criterion 1 of @xmath61 4.1 for further details ) .
in total 62 ( @xmath051% ) of our 121 detected agn candidates had detections ; we use these agns in our agn fraction analyses . versus redshift for late - type galaxy samples with @xmath294 and @xmath295 .
we find no significant evolution of @xmath296 over the redshift range .
, width=351 ] we began constructing @xmath286 by estimating the shape of @xmath277 using late - type galaxies with and @xmath275 , an optical luminosity regime where we have a relatively large number of sources and are sufficiently complete out to @xmath297 ( see fig 4@xmath69 ) .
we split this sample into two subsets about @xmath22 , to test whether there is substantial evolution in the shape and normalization of @xmath277 over this redshift range . in figure 12 , we show @xmath277 for galaxies in the redshift ranges @xmath276 ( @xmath298 ; _ filled circles _ ) and ( @xmath299 ; _ open triangles _ ) . from figure 12 , we see that the overall shape and normalization of @xmath277 for late - type galaxies with @xmath300 is similar for galaxies at @xmath298 and @xmath301 .
we fit the shape of @xmath277 using least - squares fitting of the luminosity dependent cumulative agn fraction using all galaxies from with @xmath300 . for these fits
, we found that the data were well - fit by an exponential function , which we parameterized as @xmath302 , where @xmath69 and @xmath83 are fitting constants . by construction , this function is only valid for @xmath303 , which is @xmath012 orders of magnitude less luminous than a typical stacked luminosity of our late - type galaxy samples ( see table 2 and fig .
we find best - fit values of @xmath304 and @xmath305 ; in figure 12 ( _ thick black curve _ ) , we show our best - fit relation for @xmath277 .
we constrained further the redshift evolution of @xmath286 [ i.e. , the shape of @xmath296 ] by dividing our main late - type galaxy sample into five nearly independent redshift bins ( with ) and calculating @xmath286 for fixed ranges of @xmath274 and @xmath269 . in figure 13 , we show @xmath296 as a function of redshift for late - type galaxies with @xmath295 and @xmath294 , the approximate completeness limit at @xmath307 ( see fig .
4@xmath69 ) . using these data and @xmath272 fitting [ assuming a @xmath308 dependence ] ,
we constrained the redshift evolution of @xmath296 to be proportional to @xmath309 ; similar results were found for different ranges of @xmath274 and @xmath2 .
this result differs from that observed for early - type galaxies , where the agn fraction and mean agn emission has been found to evolve as @xmath0@xmath310 ( e.g. , brand 2005 ; lehmer 2007 ) . )
late - type galaxy agn number density is constant over .
for example , @xmath16 has been shown to fade by @xmath01 mag since @xmath264 ( e.g. , lilly 1995 ; wolf 2003 ; faber 2007 ) , suggesting that there are fewer optically luminous late - type galaxies in the local universe than at @xmath311 .
] we therefore conclude that there is little redshift evolution in the late - type galaxy agn fraction over the redshift range , and hereafter we assume that @xmath296 remains roughly constant out to @xmath84 . to constrain the overall dependence of @xmath286 on @xmath269 [ i.e. , @xmath312 , we calculated the cumulative agn fractions for late - type galaxy samples with @xmath269 by holding the ranges of @xmath274 and @xmath289 fixed and varying @xmath269 . in figure 14 , we show @xmath313 versus @xmath314 ( fig .
14@xmath69 ) , @xmath315 ( fig .
14@xmath83 ) , and @xmath316 ( fig .
14@xmath81 ) for @xmath317 at @xmath318 ( _ dashed curves _ ) and ( _ dotted curves _ ) .
we calculated @xmath319 for intervals of @xmath269 where we are approximately complete at @xmath22 ( for the @xmath276 interval ) and @xmath297 ( for the interval ; see fig . 4 ) .
for each sample , we again utilized least - squares fitting to approximate the @xmath320 dependence of @xmath319 .
these fits were performed using all data over the redshift range assuming a functional dependence of @xmath321 . in each panel of figure 14 , we show the best - fit solutions for @xmath319 with the gray lines .
we find that the agn fraction is strongly dependent on @xmath2 ( fig .
14@xmath69 ) and @xmath3 ( fig .
14@xmath83 ) , such that more optically - luminous and massive galaxies have larger agn fractions ; this result is consistent with other studies of the agn host galaxies ( e.g. , kauffmann 2003b ; nandra 2007 ; silverman 2007 ) .
also , the agn fraction seems to be mildly dependent on the galaxy sfr ; however , this is likely due to the fact that the sfr is larger on average for more massive galaxies .
based on the above estimates of the shapes of @xmath277 , @xmath296 , and @xmath319 , we approximated empirically @xmath286 for each choice of @xmath269 following : @xmath322 + c\log f_{\rm phys } , % \ ] ] where @xmath323 represents the normalization of each relation based upon the value of @xmath286 at @xmath324 and , , and sfr @xmath325 , for galaxy samples selected via @xmath2 , @xmath3 , and sfr , respectively .
we note that only @xmath323 and @xmath81 are dependent upon our choice of @xmath269 ; we find for the set of physical parameters @xmath326 [ @xmath2 , @xmath3 , sfr ] that @xmath327 [ @xmath328 , @xmath329 , 0.20 ] and @xmath330 [ 1.2 , 0.67 , 0.66 ] . for reference , in figure 12
we have shown curves of @xmath286 for @xmath331 , @xmath280 , and @xmath281 ( _ dotted curves _ ) .
to estimate the agn contamination expected for each of our stacked samples presented in @xmath61 5.1 , we followed closely the procedure in @xmath61 3.2.2 of lehmer ( 2007 ) . for completeness
, we outline this procedure below . for each stacked sample ,
we used equation 8 to compute cumulative agn fractions @xmath286 . in figure 15@xmath69
, we show @xmath286 for two of our @xmath2-selected samples : @xmath332 galaxies with @xmath33310@xmath334 ( _ dashed curve _ ) and @xmath335 galaxies with @xmath33610@xmath337 ( _ solid curve _ ) .
we then converted @xmath286 for each sample to a differential agn fraction @xmath338 ( i.e. , the luminosity dependent fraction of galaxies harboring agns within discrete luminosity bins of width @xmath339 ; see fig .
15@xmath83 ) .
for each sample , we calculated the luminosity dependent fraction of galaxies that were below our @xmath274 detection limit @xmath340 ( fig .
15@xmath81 ) .
we then multiplied @xmath338 and @xmath340 to estimate the luminosity dependent fraction of galaxies harboring an agn that was undetected in the exposures @xmath341 ( i.e. , @xmath342 ; see fig .
15@xmath343 ) ; these agns would _ not _ have been removed for our stacking analyses .
finally , we approximated the total agn contamination @xmath344 of each stacked sample using the following summation : @xmath345 where the summation is over all bins of @xmath339 in the range @xmath346 3945 .
we converted each value of @xmath347 to estimated values of @xmath348 and @xmath344 by assuming the contaminating agn emission roughly follows an sed described by a power - law . in order to constrain the average photon index of the power - law ,
we stacked all detected agns in our stacked samples with luminosities below @xmath349 . for these agns
, we find a stacked effective photon index of @xmath350 .
if the intrinsic value of the photon index is @xmath154 , at the median redshift of our main sample ( @xmath351 ) , @xmath352 corresponds to an intrinsic column density of @xmath353 @xmath354 @xmath24 .
if we assume @xmath355 describes well the effective sed of the undetected agns in our stacked sample , we find that agn contamination can account for ( median of @xmath05% ) of the emission from our stacked samples , suggesting that agns are not providing a significant contribution to our stacked results .
we note that the sed used in this calculation has an important effect on the overall estimate of the agn contamination .
since our estimates for contamination in the and bandpasses decrease as @xmath132 decreases , the amount of contamination in our samples may be underestimated if our choice of @xmath132 is too flat ; however , we find that for conservative choices of @xmath132 ( i.e. , @xmath356 ) that are representative of even unobscured agns ( for reference , see fig . 5 ) , the agn contamination remains low ( @xmath1940% ) and has no material effect on our results .
it is also worth noting that if there exists a large population of intrinsically luminous yet heavily obscured agns in our high - redshift galaxy samples that are undetected in the cdfs , then our conclusions above could be somewhat different .
if such a population were present and had significant influence on our stacked results , then we would expect to find stacked spectra that were relatively flat ( @xmath125 ) ; however , as discussed in @xmath61 5.1 , we find that very few of our stacked samples could have such flat spectra in our stacked samples , suggesting that such a population , if present , does not have a strong effect on our results . or greater ) ,
@xmath286 , versus @xmath357 for two of our stacked samples : @xmath332 galaxies with @xmath33310@xmath334 ( _ dashed curve _ ) and @xmath335 galaxies with @xmath35810@xmath337 ( _ solid curve _ ) .
each curve was computed following equation 8 .
( _ b _ ) differential agn fractions ( i.e. , the fraction of galaxies harboring an agn in discrete bins of width @xmath359 ) , @xmath338 , versus @xmath357 .
( _ c _ ) fraction of late - type galaxies for which we could _ not _ have detected an agn with a luminosity of @xmath274 , @xmath340 , versus @xmath357 .
( _ d _ ) fraction of galaxies harboring agns in our optically luminous faded samples that would remain undetected due to sensitivity limitations , @xmath342 , versus @xmath357 ; these galaxies would not have been removed from our stacking analyses . ] in table 2 ( cols .
2527 ) , we have provided the estimated fractional agn contribution to each stacked sample [ i.e. , @xmath360 for the fb , sb , and hb using the technique described above and an assumed @xmath361 .
we find that the estimated agn contamination is most significant for galaxy samples with large values of @xmath2 , @xmath3 , and sfr .
as noted in @xmath61 1 , the global star - formation rate density has been observed to increase with redshift out to .
investigations of the most - distant lbgs at show that the star - formation density peaks around and gradually declines toward higher redshifts ( e.g. , steidel 1999 ; giavalisco 2002 , 2004b ; bouwens 2004 , 2005 ; dickinson 2004 ) . to investigate
whether the mean activity from normal late - type galaxies follows a similar trend , we study the properties of a sample of @xmath362 lbgs , which were identified as @xmath44-band `` dropouts '' through the goods project ( see lehmer 2005a for details ) . we filtered the goods @xmath263 lbg sample ( lee 2006 ) to include only lbgs that ( 1 ) were within the central @xmath040 of the @xmath02 ms and @xmath01 ms , and ( 2 ) had rest - frame @xmath45-band luminosities that were similar to the most optically luminous ( @xmath233 320 @xmath363 ) late - type galaxies used in this study .
@xmath45-band luminosities were calculated by applying @xmath364-corrections to the @xmath33-band flux ( from goods ) , where @xmath364-corrections were derived using an sed appropriate for lbgs ( see @xmath61 2.2 of lehmer 2005a for details ) .
we found that 85 @xmath263 lbgs from the lehmer ( 2005a ) sample satisfied these two selection criteria .
we identified three detected lbgs , which all had @xmath365 . due to their high luminosities
, we classified these sources as obvious agns .
after removing these three detected agns from our @xmath263 lbg sample , we performed stacking analyses as described in @xmath61 4.2 .
we found a significant ( @xmath366 ) detection in the bandpass , which corresponds roughly to rest - frame emission . assuming an intrinsic power - law spectrum with a photon index of @xmath201
, we found a mean luminosity of for our @xmath263 lbgs , a value that agrees well with previous studies ( e.g. , lehmer 2005@xmath69 ; laird 2006 ) .
-band mean luminosity ratio @xmath222 versus the age of the universe ( for our adopted cosmology , the current age of the universe is 13.47 gyr ) for star - forming galaxies with @xmath233 @xmath363 .
for reference , redshift has been plotted along the top axis .
we have included mean values of @xmath237 for the s01 local sample of late - type galaxies ( _ open circles _ ) , our stacked samples at @xmath98 ( _ large filled circles _ ) , and @xmath263 lbgs that were both uncorrected ( _ filled star _ ) and corrected ( _ open star _ ) for agn contamination . at @xmath7 ,
we have plotted @xmath237 for individual late - type galaxies from our main sample ( _ small filled circles _ and _ upper limits _ ) ; the mean values and errors for these galaxies , computed using asurv , has been indicated . ] since our lbgs reside in the high - redshift universe , agn contamination is expected to have a more significant effect on these results than it did for our late - type galaxies .
the median luminosity detection limit is and @xmath0@xmath159 for the and bandpasses , respectively . to estimate the agn contamination
, we followed the approach outlined in @xmath61 5.2 , which made use of the luminosity dependent agn fraction ( @xmath286 ) . in figure 12
( _ filled stars _ ) , we show the luminosity dependent agn fraction @xmath367 for @xmath6 lbgs . we note that at @xmath368 , the cumulative agn fraction for @xmath6 lbgs is a factor of @xmath034 times larger than that computed for our late - type galaxy sample with similar optical luminosities . using the functional form for @xmath286 presented in equation 8 , but with a times larger normalization factor for @xmath263 lbgs ( i.e. , @xmath369 34 @xmath323 ) , we find that agn emission may plausibly account for of the stacked counts . in figure 16 , we show the to@xmath45-band mean luminosity ratio for @xmath370 320 @xmath234 star - forming galaxies ( i.e. , late - type galaxies and lbgs ) as a function of the age of the universe .
together , these data span @xmath085% of cosmic history ( i.e. , out to @xmath371 ) .
as presented in @xmath61 5.1 , /@xmath2 shows significant evolution over the redshift range @xmath17 01.4 , and after correcting for agn contamination , we find that /@xmath2 is similar for @xmath263 lbgs and @xmath1 late - type galaxies .
this result suggests that the non - agn emission for the most luminous star - forming galaxies may flatten near , which has been predicted roughly from simulations of how the normal - galaxy emission is expected to respond due to global changes in the star - formation rate density ( e.g. , ghosh & white 2001 ) . to test whether the -sfr correlation is similar for lbgs as we found for late - type galaxies at , we approximated absorption - corrected sfrs for the @xmath6 lbgs using uv band emission .
these sfrs were approximated following sfr = @xmath372 , where is the absorption - correction factor ( see giavalisco 2004b ) .
as described in @xmath61 3.3 , we approximated the uv luminosity following @xmath94(2800 ) ; however , here @xmath96(2800 ) was derived using our adopted lbg sed .
we find that the mean sfr for optically - luminous @xmath6 lbgs is . after correcting the mean stacked luminosity for agn contamination
, we find an to sfr ratio of @xmath262/sfr @xmath0 , which is suggestively lower than its value for late - type galaxies ( i.e. , @xmath262/sfr ) .
we note , however , that the estimation of the mean sfr for these galaxies is based solely on the 2800 emission and is therefore highly uncertain ; further constraints on the mean infrared luminosity of these sources would help considerably .
for reference , in figure 11 we have plotted @xmath285 sfr versus @xmath262 for the lbgs both uncorrected ( _ filled star _ ) and corrected ( _ open star _ ) for agn contamination .
it is interesting to note that the agn - corrected lbg sample appears to have @xmath4 and sfr values similar to those of local ulirgs , which are relatively underluminous for their derived sfrs .
pr07 suggest that galaxies with sfrs that begin to exceed @xmath6850100 , similar to the ulirgs , may have properties that are completely dominated by hmxbs . in such systems , other emitting populations that may normally be significant in more quiescent galaxies ( e.g. , lmxbs , hot gas , supernovae and their remnants , etc . )
would collectively contribute only a negligible fraction of the total emission , thus causing the overall galaxies to appear underluminous compared with the sfr correlation .
we note , however , that our results presented in @xmath61 5.1 suggest that hmxbs likely dominate the x - ray emission from galaxies with sfrs much lower than those of ulirgs .
another possibility is that as the absorption within the most actively star - forming galaxies increases with sfr ( see discussion in @xmath61 5.1 ) , the emission from point - source populations may become significantly obscured .
deep observations of local lirgs and ulirgs with and have found significant absorption in their point - source populations ( see , e.g. , lira 2002 ; zezas 2002 , 2003 ) thus providing some support for this possibility .
future investigations into the nature of the populations of ulirgs could help resolve this issue .
we have investigated the emission from 2568 normal late - type galaxies over the redshift range @xmath17 01.4 that lie within the deep fields ( cdfs ) .
our late - type galaxy sample was constructed primarily using color - magnitude diagrams , which incorporated rest - frame @xmath43 color and absolute @xmath46-band magnitudes , to isolate blue late - type galaxies ( see @xmath61 2 for details ) . in total ,
225 ( @xmath09% ) of our late - type galaxies were detected individually in the band .
based on and optical spectral properties , to optical flux ratios , the correlation between luminosity and star - formation rate , and comparisons between infrared and radio properties , we infer that 121 ( @xmath053% ) of the detected late - type galaxies are dominated by agn emission .
the remaining 104 detected galaxies had and multiwavelength properties consistent with those of normal late - type galaxies with emission dominated by binaries ( hmxbs and lmxbs ) . to study the emission and evolution from large representative populations of late - type galaxies ( i.e. , including the undetected sources ) , we utilized stacking analyses of galaxy populations with agn candidates removed .
we stacked normal - galaxy samples that were selected via their rest - frame @xmath45-band luminosity ( @xmath2 ) , stellar mass ( @xmath3 ) , and star - formation rate ( sfr ) in redshift bins ( see @xmath61 5.1 ) .
furthermore , we compared these results with those found for a sample of @xmath6 lbgs from lehmer ( 2005a ) . in the points
below , we summarize our key findings : \1 .
we obtained significant detections in the and bandpasses for all of our stacked samples .
we estimated that lmxbs and low - level agns provide only low - level contributions to the stacked emission from our samples and that hmxbs constitute the dominant emitting component .
therefore , for these galaxies the emission is tracing primarily star - formation activity . normal late - type galaxy samples selected via @xmath2 and @xmath3
show significant ( at the @xmath23299.9% confidence level ) evolution in their average properties from @xmath15 to 1.4 . for the most optically luminous ( @xmath373 ) and massive ( ) late - type galaxies at @xmath236 , @xmath237 and @xmath238
are measured to be larger than their local values by factors of and , respectively .
we find that late - type galaxies of lower stellar mass generally have larger to stellar - mass mean ratios ( @xmath238 ) than their higher - mass analogs . over @xmath7
0.21 , galaxies with @xmath374 are a factor of times more luminous per unit stellar mass than galaxies with @xmath375 . \3 .
we characterized the properties of 888 24@xmath54m - detected late - type galaxies in our sample .
the 24@xmath54 m data allowed us to select galaxy samples selected via their sfrs .
for these samples , we found that the luminosity is well - predicted by a constant to
sfr ratio , similar to the -sfr correlation reported by previous authors ( e.g. , pr07 ) .
this implies that the -sfr correlation holds out to at least @xmath70 , 1 , and 1.4 for galaxies with sfr @xmath0 2 , 10 , and 50 , respectively , and supports the idea that the strong evolution observed for normal late - type galaxies selected via @xmath2 and @xmath3 is likely due to strong changes in sfr .
the properties of our most optically - luminous ( @xmath233 320 @xmath363 ) late - type galaxies at @xmath236 are comparable to those for @xmath376 lbgs with similar optical luminosities , once undetected agn contamination in the lbg population has been accounted for .
this suggests that there may plausibly be a flattening in the @xmath237@xmath289 relation for optically - luminous star - forming galaxies between @xmath377 1.43 .
we estimate a mean sfr of @xmath060 for these lbgs .
we find that the observed mean luminosity is suggestively underluminous based on the -sfr correlation prediction ; this result is similar to that found for local ulirgs with comparable sfrs .
+ the above results can be improved greatly through ( 1 ) the study of late - type galaxy populations in other existing multiwavelength extragalactic survey fields that contain observations , ( 2 ) additional observations of the cdfs to even better sensitivity levels than are currently available , and ( 3 ) observations with future missions with imaging capabilities that are complementary to those of .
the areal footprint of the cdf regions used in this study totals @xmath00.18 deg@xmath36 ( see @xmath61 2.1 for details ) , which severely limits the number of galaxies being studied at @xmath378 .
the redshift range spans @xmath02.5 gyr of cosmic look - back time , compared with the @xmath06.6 gyr over , that we are most effectively studying above .
several complementary deep and wide extragalactic survey fields that have recently been conducted or are in progress can improve the present situation .
three such ideal survey fields are the @xmath0200 ks all - wavelength extended groth strip international survey ( aegis ; e.g. , nandra 2005 ; davis 2006 ) , the @xmath050 ks cosmic evolution survey ( c - cosmos ; pi : m. elvis ; see also , scoville 2007),50 ks acis - i exposures ; in these regions , the total exposure will reach @xmath0200 ks due to the overlaping exposures .
however , the analyses presented in this paper utilize only high - quality imaging at off - axis angles of @xmath1960 ( see @xmath61 2.1 for additional justification ) , where there is very little overlap in the c - cosmos pointings .
we therefore do not consider the overlapping regions of the c - cosmos exposures in this discussion .
for additional information regarding c - cosmos , see http://cfa-www.harvard.edu/hea/cos/c-cosmos.html . ] and the @xmath05 ks noao deep wide - field survey ( ndwfs ; e.g. , murray 2005 ) .
our study shows that identifying and removing contaminating agns is a crucial ingredient to studying the stacked properties of normal - galaxy samples ( see @xmath61 5.2 ) . in order to avoid significant levels of agn contamination in shallower wide - area surveys such as aegis ,
c - cosmos , and ndwfs , studies of normal galaxy populations must be limited to lower redshift intervals .
if we require a detection limit of @xmath379 over regions of each survey where the sensitivity is optimal ( i.e. , @xmath1960 from the aim - points ) , then we estimate that aegis , c - cosmos , and ndwfs could effectively be used to study normal late - type galaxy populations at redshifts less than @xmath00.5 , @xmath00.25 , and @xmath00.1 , respectively .
when factoring in the areal coverage of these fields , we estimate that aegis and c - cosmos will contain @xmath0250350 late - type galaxies with at ( i.e. , a factor of @xmath03 larger than the cdfs ) and that aegis , c - cosmos , and ndwfs taken together will have @xmath0100250 similar galaxies at @xmath380 ( i.e. , a factor of @xmath022 times larger than the cdfs ) . studying these galaxy populations
would significantly improve constraints on the properties of @xmath381 late - type galaxies .
deeper observations of the cdfs would provide additional insight into the properties of the normal late - type galaxy populations presented in this study and enable us to extend our analyses to higher redshifts .
typical galaxies in our sample have mean luminosities of . in the @xmath02 ms , where our sensitivity is greatest
, we expect that galaxies with @xmath382 should be detectable out to @xmath332 . out of the 11 late - type galaxies with @xmath383 at @xmath378 , we find detections for 8 of them ( @xmath072% ) .
deeper observations over the cdfs will allow for the detection of such galaxies out to progressively higher redshifts . for exposures of @xmath05
ms and @xmath010 ms , a source with @xmath384 should be detectable out to @xmath12 and @xmath385 , respectively . if we pessimistically assume an detection fraction of @xmath070% , we estimate that @xmath030 and @xmath050 normal galaxies would be detected individually for each respective deep exposure .
the individual detection of these galaxies would allow for improved constraints on both the luminosity functions of normal late - type galaxies at higher redshifts and the low - luminosity agn contributions to our stacked signals . at higher redshifts , an @xmath05
ms and @xmath010 ms exposure would enable us to effectively perform studies of normal late - type galaxies , similar to those presented in this paper , out to @xmath386 and @xmath387 , respectively .
such a data set would provide , for the first time , a reliable constraint on the emission from normal star - forming galaxies near the peak of the global star - formation rate density at @xmath7 1.53 .
in addition to the improvement that additional observations could provide , future missions such as and should enable new scientific investigations of distant normal galaxies .
will be able to place significant spectral constraints for sources with fluxes down to @xmath0@xmath388 , a level fainter than many of the average fluxes derived from our samples .
is planned to provide imaging with @xmath001 per resolution element and will easily probe to flux levels of @xmath0@xmath389 ( @xmath390 at @xmath297 ) . at these levels
, the first detailed investigations of the evolution of the normal late - type galaxy luminosity function can be performed effectively out to @xmath11 without confusion problems due to the crowding of large numbers of sources .
furthermore , these observations will allow for new constraints to be placed on the populations of all normal galaxies in the observable universe that are offset by more than @xmath00.8 kpc from their host - galaxy nuclei .
we thank robin ciardullo , doug cowen , mike eracleous , caryl gronwall , bin luo , brendan miller , tim roberts , and ohad shemmer for useful suggestions , which have improved the quality of this paper .
we also thank the referee for carefully reviewing the manuscript and providing detailed comments .
we gratefully acknowledge financial support from x - ray center grant g04 - 5157a ( b.d.l .
, a.t.s . ) , the science and technology facilities council fellowship program ( b.d.l ) , the royal society ( d.m.a . ) , the emmy noether program of the deutsche forschungsgemeinscaft ( e.f.b . ) , nasa ltsa grant nag5 - 13102 ( d.h.m . ) , the fellowship program ( f.e.b . ) , and nsf grant ast 06 - 07634 ( d.p.s . ) . | we report on the evolution over the last @xmath09 gyr of cosmic history ( i.e. , since @xmath1 ) of late - type galaxy populations in the deep field - north and extended deep field - south ( and , respectively ; jointly cdfs ) survey fields .
our late - type galaxy sample consists of 2568 galaxies , which were identified using rest - frame optical colors and morphologies .
we utilized stacking analyses to investigate the emission from these galaxies , emphasizing the contributions from normal galaxies that are not dominated by active galactic nuclei ( agns ) . over this redshift range
, we find significant increases ( factors of @xmath0510 ) in the mean luminosity ratio ( /@xmath2 ) and the mean ratio ( /@xmath3 ) for galaxy populations selected by @xmath2 and @xmath3 , respectively . when analyzing galaxy samples selected via sfr , we find that the mean ratio ( @xmath4/sfr ) is consistent with being constant over the entire redshift range for galaxies with , thus demonstrating that emission can be used as a robust indicator of star - formation activity out to @xmath5 .
we find that the star - formation activity ( as traced by luminosity ) per unit stellar mass in a given redshift bin increases with decreasing stellar mass over the redshift range , which is consistent with previous studies of how star - formation activity depends on stellar mass .
finally , we extend our analyses to lyman break galaxies at @xmath6 and estimate that /@xmath2 at @xmath6 is similar to its value at @xmath1 . |
cosmology and black holes as seen through the eyes of general relativity come together in the investigation of a dynamical black hole embedded in a cosmological background .
the interplay between the cosmic dynamics and the black hole gives rise to interesting phenomena and can reveal some unexpected features of the underlying theory of gravity . in this work
we restrict our attention , for simplicity , to spherically symmetric systems .
the prototypical solution of the einstein equations representing a black hole embedded in a cosmological spacetime is the schwarzschild - de sitter - kottler solution
. this metric is special since it admits a timelike killing vector and is , therefore , static in the region between the black hole horizon and the de sitter ( cosmological ) horizon .
a less well known solution is the 1933 mcvittie solution @xcite , which is a generalization of the schwarzschild - de sitter - kottler solution . in this case
the black hole is embedded in a general friedmann - lematre - robertson - walker ( flrw ) background , so that the region between the black hole horizon and cosmological horizon need not be static .
although it has been studied and celebrated by many authors @xcite , it has proved surprisingly difficult to understand ( see the recent work @xcite ) . a simplifying assumption in the study of this solution , explicitly stated in mcvittie s original paper , is the no - accretion condition @xmath0 ( in spherical coordinates , where @xmath1 is the einstein tensor ) .
this explicitly forbids any radial flow of material , which should otherwise occur whenever a spherically symmetric local inhomogeneity ( such as a central black hole ) is introduced in the background . when this is modelled however , more general solutions of einstein s theory become possible .
these include some generalized mcvittie solutions @xcite ; solutions such as those derived by husain - martinez - nuez @xcite , fonarev @xcite , sultana - dyer @xcite and mcclure - dyer @xcite ; the class of solutions found by szekeres @xcite ; lematre - tolman - bondi black hole solutions @xcite ; and other solutions @xcite . in extended theories of gravity , such as scalar - tensor and @xmath2 gravity ,
several other solutions of the relevant field equations ( which involve an extra gravitational scalar field or higher derivative terms , respectively ) have been found and sometimes discussed @xcite . the original motivation for mcvittie s work @xcite was the investigation of the effects of the cosmological expansion on local systems .
another approach to this problem later led to the construction of swiss - cheese models by einstein and straus @xcite .
however , although this problem has stimulated much discussion over the years @xcite , the scientific community as a whole is yet to arrive at an agreement about the best approach to it ( see the recent review @xcite ) . when solutions representing local inhomogeneities in cosmic backgrounds are considered , the scope of the investigation broadens .
for example , a problem of current interest is the possible spatial and temporal variation of the gravitational `` constant '' ( which becomes a scalar field in brans - dicke and scalar - tensor gravity ) @xcite .
we now know several solutions of this kind , but before enlarging the catalog further it is important to fully understand the presently known solutions ( for some of them , it is not even known whether the local inhomogeneity is associated with a black hole , a naked singularity , or another kind of object ) .
for this reason , we revisit here the no - accretion mcvittie solution , proposing a quick way of locating the associated black hole and cosmological ( apparent ) horizons and studying their evolution .
we extend the type of cosmological background to include phantom universes , which have not been considered before in relation to the mcvittie solution . with the exception of the schwarzschild - de sitter - kottler solution , which incorporates only a static background universe ,
spherically symmetric black holes in more general cosmological backgrounds are dynamical .
this significantly complicates their analyses .
since the solutions of einstein s equations corresponding to the mcvittie metric are highly dynamical , it is not convenient for us to study the event horizons ( both black hole and cosmological ) , which may not even exist .
it is more instructive to study the dynamical apparent horizons , the importance of which is being increasingly recognized in the literature @xcite .
it is known that , for dynamical cosmological black holes , apparent horizons can appear or disappear @xcite , and we would like to shed some light on this bizarre phenomenology .
the plan of this paper is as follows . in sec .
ii we briefly review the schwarzschild - de sitter - kottler solution ; this ( over-)simplified situation will serve us well when attempting to understand the more complicated phenomenology of dynamical apparent horizons . in sec .
iii we locate the apparent horizons of the mcvittie metric for non - phantom cosmological backgrounds and recover the previous results in certain limits .
we then continue with the analysis of phantom background universes .
finally , sec .
iv contains a discussion of our results and our conclusions . throughout this work we use units in which the speed of light @xmath3 and newton s constant @xmath4 are unity , and we mostly follow the notations of ref .
in particular , the metric signature is @xmath5 .
the schwarzschild - de sitter - kottler solution is the prototypical solution representing a black hole embedded in a cosmological background ( for a certain range of parameter values ) .
we will discuss the mcvittie metric by using an analogy with the schwarzschild - de sitter - kottler metric wherever possible , even though the latter corresponds to a very special situation by admitting only a _
static _ black hole in the de sitter background , and its apparent horizons are also event horizons . nonetheless , analogies are made possible by the fact that the schwarzschild - de sitter - kottler solution is contained as a special case in the mcvittie class of solutions .
the spherically symmetric schwarzschild - de sitter - kottler solution of the einstein equations has line element [ 1 ] ds^2=- ( 1 - -h^2r^2 ) dt^2 + ( 1 - -h^2r^2 ) ^-1 dr^2 + r^2 d_(2)^2 , [ eq : sdsk ] where @xmath6 is the areal radius ( of a sphere with surface area @xmath7 ) , @xmath8 is the metric on the unit 2-sphere , the constant @xmath9 is the hubble parameter of the de sitter background , @xmath10 is the cosmological constant and @xmath11 is a second parameter describing the mass of the central inhomogeneity ( _ e.g. _ , @xcite ) . in general , the locations of the apparent horizons for a spherically - symmetric system can be calculated from the radial element of the inverse metric @xmath12 @xcite .
thus the apparent horizons for the schwarzschild - de sitter - kottler solution are defined by the positive roots of the cubic equation [ 3 ] 1- -h^2r^2 = 0.[eq : sdskhor ] following the method outlined by nickalls in @xcite , these roots may be written as @xmath13 where @xmath14 . since @xmath15 and @xmath16 are both necessarily positive ( we only consider expanding universes ) , @xmath17 is negative and therefore unphysical .
we thus refer to this spacetime as having only two apparent horizons .
we refer to @xmath18 as the black hole apparent horizon , since it reduces simply to the schwarzschild horizon at @xmath19 if there is no background expansion @xmath20 , and we refer to @xmath21 as the cosmological apparent horizon , since it reduces to the static de sitter horizon at @xmath22 if there is no mass present .
the metric ( [ 1 ] ) is static in the region covered by the coordinates @xmath23 , which is comprised between these two horizons .
a number of interesting observations can be made .
first , both apparent horizons only actually exist if @xmath24 . in this case , since the metric is static between these two horizons , the apparent black hole and cosmological horizons are also event horizons and , therefore , null surfaces .
second , if @xmath25 it is easy to show that these horizons then coincide .
this case corresponds to the nariai black hole .
finally , for @xmath26 both horizons become complex - valued and therefore unphysical , and one is left with a naked singularity .
these results can be summarized as follows : @xmath27 the hubble parameter for an idealized de sitter background is a constant , whereas more realistic models incorporate a time - dependent hubble parameter . with a clear understanding of the static horizons in the schwarzschild - de sitter - kottler spacetime , we may now study the dynamical horizons which emerge by considering a more realistic time - dependent metric .
we now consider the mcvittie metric for a black hole embedded in an flrw background which is expanding with the hubble flow @xcite . for simplicity , we restrict ourselves to the case in which the background is spatially flat ( curvature index @xmath28 ) .
the line element can thus be cast in the form @xcite ds^2=-dt^2 - dtdr + r^2d_(2)^2 . [ 4 ] here @xmath29 , where @xmath30 is the scale factor of the flrw background and an overdot denotes differentiation with respect to the comoving time @xmath31 .
note that for the case of a static background in which @xmath32 and @xmath33 , the mcvittie metric actually corresponds to the schwarzschild - de sitter - kottler metric given by ( [ eq : sdsk ] ) via a simple transformation of the time coordinate @xcite . assuming a perfect fluid stress energy tensor , we may use einstein s equations to calculate forms for the density @xmath34 and pressure @xmath35 of the background fluid in mcvittie s metric .
the density turns out to correspond to the known flrw density ( t)= h^2(t ) , one may consider arbitrary flrw backgrounds generated by cosmic fluids satisfying any equation of state ( in fact , in the next section , we study a flrw universe dominated by a phantom fluid ) . for illustrative purposes however , in this section we restrict our attention to a cosmic fluid which reduces to dust at spatial infinity .
this corresponds to an equation of state parameter @xmath36 , so the pressure can be shown to be @xcite p(t , r)=(t ) ( -1 ) [ 5 ] .
other quantities may be calculated from the inverse metric , given by ( g^ ) =( cccc - & - & 0 & 0 + & & & + - & ( 1- -h^2r^2 ) & 0 & 0 + & & & + 0 & 0 & & 0 + & & & + 0 & 0 & 0 & ) . [ 6 ] the misner - sharp - hernandez mass @xmath37 @xcite contained in a sphere of areal radius @xmath6 is defined , in the case of spherical symmetry , by 1- = g^rr .
thus , we obtain m_msh= r^3 + m . which is interpreted as the contribution of the energy of the cosmic fluid contained in the ball plus the contribution of the local inhomogeneity .
this mass coincides with the hawking - hayward quasi - local mass @xcite . since for the mcvittie metric @xmath6 is an areal radius and the system is spherically symmetric , the apparent horizons can once again be calculated from @xmath12 , corresponding to [ 8 ] 1 - -h^2(t ) r^2=0
this is clearly equivalent to the schwarzschild - de sitter - kottler horizon condition given by ( [ eq : sdskhor ] ) but with a time - dependent hubble parameter .
we denote the resulting time - dependent apparent horizons @xmath38 and @xmath39 , and these correspond to the solutions @xmath18 and @xmath21 given in equation ( [ eq : horizons ] ) but with the replacement @xmath40 . since the apparent horizons for the mcvittie metric are dynamical , rather than static , their relative locations now depend on the cosmic time .
analogous to the schwarzschild - de sitter kottler case , the condition for both horizons to exist is @xmath41 , which corresponds to @xmath42 ( and @xmath43 which is always satisfied ) .
however , unlike the former case where the hubble parameter is a constant , this inequality will only be satisfied at certain times during the cosmological expansion , and not at others .
the time at which @xmath44 is unique for a dust - dominated background with @xmath45 , and we denote it @xmath46 .
the three cases may then be characterized as : * @xmath47 : at early times @xmath48 , so both @xmath38 and @xmath39 are complex and therefore unphysical .
_ there are no apparent horizons . _
* @xmath49 : at this time @xmath50 and the horizons @xmath38 and @xmath39 coincide at a real , physical location .
_ there is a single apparent horizon at @xmath51_. * @xmath52 : at late times @xmath53 , so both @xmath38 and @xmath39 are real and therefore physical .
_ there are two apparent horizons .
_ the qualitative dynamical picture which emerges from this analysis is the following and is illustrated in fig .
[ figure1 ] . , hence time @xmath31 and radius @xmath6 are measured in units of @xmath15 ( see text for details).[figure1 ] ] the lack of apparent horizons for @xmath54 leaves a naked singularity at @xmath55 , where the ricci scalar and pressure also diverge ( see below ) .
this is explained by the divergence of the hubble parameter @xmath56 in the early universe , causing the mass @xmath15 to remain supercritical , _
i.e. _ causing @xmath57 to be satisfied .
analogous to the schwarzschild - de sitter - kottler solution , a black hole horizon can not be accommodated in such a small universe . at the critical time
@xmath58 a black hole apparent horizon appears and coincides with the cosmological apparent horizon at @xmath59 . for a dust - dominated cosmological background
this may be given as @xmath60 .
this is the analog of the nariai black hole in the schwarzschild - de sitter - kottler solution , but it is instantaneous .
as time progresses , @xmath61 , the single horizon splits into a dynamical black hole apparent horizon surrounded by a time - dependent cosmological horizon .
this solution can progressively constitute a better and better toy model for a spherical , non - accreting astrophysical black hole in the late universe with @xmath62 . the black hole apparent horizon shrinks , asymptoting to the spacetime singularity at @xmath19 from above as @xmath63 , while the cosmological apparent horizon expands monotonically , tending to @xmath64 in the same limit .
the actual universe is of course not dust - dominated , and is better described by the scale factor for expansion @xmath65^{1/3},\ ] ] consistent with the spatially flat concordance model @xcite . here
@xmath66 km s@xmath67 mpc@xmath67 is the current value of the hubble parameter and @xmath68 is the current dark energy density . using this we may calculate actual values for @xmath58 and apparent horizon locations for black holes in our universe .
considering , for example , the @xmath69 black hole at the centre of the milky way , we find that a single horizon would have first appeared as early as @xmath70 secs and at a radius very close to the centre @xmath71pc .
thereafter , this would have split into two apparent horizons , which would have become increasingly separated .
note that a problem with this calculation is that it neglects mass accretion .
the results are therefore purely theoretical and would only truly be valid if this black hole had always existed at its current mass .
although in reality there were no bound structures in the universe at such an early time , this calculation does at least provide some insight into the scales involved .
let us discuss now the well known singularity @xcite .
the surface of equation @xmath72 has normal @xmath73 with norm squared n_n^= g^ n_n _ |_r=2 m .=-4m^2h^2(t ) < 0 .
@xmath74 is timelike and the surface @xmath75 is spacelike .
the ricci scalar r=-8 t^_=8 ( -3p ) = 8 ( t ) ( 4- ) diverges as @xmath76 .
this singularity separates spacetime into two disconnected regions @xmath77 and @xmath78 @xcite ; the latter region is described by the metric ( [ 4 ] ) . at the critical time @xmath79 , when @xmath80 , the normal to the surface of equation @xmath81 is @xmath82 and m^m _ = g^11(r=)= ( -mh(t ) ) = 0 .
thus the ( cosmological and black hole ) apparent horizon is instantaneously null . by differentiating the cubic equation ( [ 8 ] ) , one may solve for the rate of change in location of the apparent horizons with respect to the comoving time .
dropping the @xmath31-dependencies for simplicity , one obtains _
rearranging this , one can compare the expansion rates of the apparent horizons with that of the cosmic substratum , -h= -h ( 1 + ) .[eq : rateofexp ] this equation shows that the apparent horizons are not comoving except for trivial cases .
this explains why the black hole can not remain static but is instead forced to expand . in the case of a spatially flat flrw universe ( without the central inhomogeneity ) , it turns out that even the single cosmological horizon at @xmath83 is not comoving , since = - h .
it is widely believed that , in the absence of event horizons , an entropy can be meaningfully ascribed to apparent horizons .
the thermodynamics of these horizons has been discussed extensively @xcite .
therefore , it is interesting to ask whether the total entropy associated with both the black hole and cosmological apparent horizons is a non - decreasing function of time .
the area @xmath84 of the black hole apparent horizon is decreasing , but it is bounded from below while this behaviour is more than compensated for by the increase of the area @xmath85 of the cosmological apparent horizon .
the total horizon entropy s = s_1+s_2 = ( r_1 ^ 2 + r_2 ^ 2 ) = , where @xmath86 , is plotted in fig .
[ figure2 ] . the total horizon entropy @xmath87 ( in units @xmath88 , where @xmath89 is the boltzmann constant ) associated with the apparent horizons as a function of time . ]
since the apparent horizons emerge as a pair at @xmath49 , the horizon entropy @xmath87 exhibits a discontinuous jump from zero value at this time .
we now discuss the situation of a cosmological background dominated by a phantom fluid with equation of state satisfying @xmath90 ( @xmath91 ) and violating the weak energy condition .
the recent renewed interest in such a field has been motivated by the analysis of data from supernovae ia @xcite and the study of the effects of the accelerating universe @xcite .
the consideration of a phantom background has also led to the prediction of a big rip singularity at a finite time in the future @xmath92 @xcite .
we now consider a phantom background in the context of the mcvittie solution .
surprisingly , this is a situation which has not received much attention in previous studies .
one may consider the late time behaviour of the friedmann equation governing a phantom fluid and solve it to obtain a form for the scale factor in terms of @xmath93 and @xmath94 .
indeed the solution has been shown to be @xcite : a(t)= , where @xmath95 is a constant .
the hubble parameter may therefore be written concisely as h(t)= .
note the reverse behaviour of this function compared with the hubble parameter for a dust - dominated universe @xmath45 .
the latter diverges at the big bang singularity and gradually decreases over time , tending to zero .
the hubble parameter for a phantom fluid , however , takes on a finite value at @xmath96 and slowly increases until the big rip time , at which point it too diverges .
this suggests that the horizons around black holes embedded in a phantom fluid might behave in the opposite way to those in a dust - dominated background with @xmath97 .
indeed this does turn out to be the case , and the discussion in the previous subsection can be repeated .
the result is plotted in fig .
[ figure3 ] .
the behaviour of the mcvittie apparent horizons versus time in a phantom - dominated background universe for the parameter values @xmath98 and @xmath99 . ] we may summarize our results in an expanding universe dominated by a phantom fluid as follows . in the early universe ,
both black hole and cosmological apparent horizons exist , and are approximately located at @xmath19 and @xmath64 , respectively . as time progresses the cosmological horizon shrinks and the black hole horizon expands , until they meet and merge at the critical time @xmath79 .
thereafter they disappear , leaving behind a naked singularity . during this evolution the total apparent horizon entropy decreases and has a discontinuous jump to zero value at @xmath79 .
this behaviour is yet another manifestation of the `` weirdness '' of the phantom fluid , which seems to violate the second law of thermodynamics in many ways @xcite .
the behaviour of the apparent horizons for a phantom cosmic background was derived in ref .
@xcite for _ generalized _ mcvittie solutions , which are obtained by relaxing the mcvittie no - accretion condition and allowing for a radial energy flux onto the black hole @xcite . for simplicity of modelling , this radial flux density @xmath100 is necessarily spacelike and violates the energy conditions .
the lesson to be learnt by the present discussion of the corresponding mcvittie solution with @xmath101 is that the disappearance of the apparent horizons is not due to the fact that the accreted phantom fluid violates the weak energy condition and the total accreted mass becomes zero : it is due to the phantom character of the fluid which dictates the unusual cosmic expansion leading to the big rip , but not to accretion .
in order to understand the bizarre phenomenology of apparent horizons in the mcvittie spacetime , it is useful to first understand the schwarzschild - de sitter - kottler solution of the einstein equations
. this is a special case of the mcvittie solution .
our study of the simple , static , schwarzschild - de sitter - kottler metric has essentially revealed that a black hole can only fit in a de sitter universe if its horizon size ( determined by its mass ) does not exceed the size of the cosmological horizon . equipped with this clarity
, we have then moved on to consider the more complicated mcvittie solution , which accounts for a dynamical background and thus better represents reality .
not surprisingly , the condition for the existence of the apparent horizons in this case is analogous to the corresponding one in the static case , with the static hubble constant replaced by a dynamical hubble parameter .
this follows from the dynamical nature of the apparent horizons themselves in this case , which we are able to locate throughout their period of existence .
the absence of any ( black hole or cosmological ) apparent horizons at early times is now easily understood . at early times
the mathematical solutions suggest that the cosmological horizon would be smaller than the black hole horizon , but this is not possible since the universe at this time would be too small to accommodate a black hole apparent horizon at all .
one can not then meaningfully distinguish between the `` black hole '' and the `` universe '' in which it is embedded ; rather , the mathematical solutions represent neither and do not possess the properties of a black hole or a universe .
thus at early times , not only is there a naked singularity , but the cosmological apparent horizon is also absent .
the presence of this naked singularity prevents one from being able to derive the mcvittie solution as the development of regular cauchy data . at some finite time , given by @xmath102 for a dust - dominated background , the cosmological solution is able to catch up with the black hole solution and a single black hole / cosmological apparent horizon appears .
these then split and continue to diverge thereafter .
the mcvittie metric does not account for accretion onto the central mass .
hence the mass parameter @xmath15 is fixed and the horizon dynamics are wholly determined by the expansion of the universe . if the no - accretion assumption is relaxed however , the black hole mass itself is then also determined by the universe s expansion ( possibly with some residual freedom ) and can not be fixed _ a priori_. indeed some generalized mcvittie solutions , for which @xmath15 becomes a function of time , have already been derived @xcite . at late times
this class of solutions converges to an attractor with a well - defined mass function @xmath103 @xcite .
other solutions for cosmological lumps ( including black holes ) have also been derived and investigated without imposing the no - accretion condition in general relativity and in scalar - tensor and higher derivative gravity @xcite . in some of these studies
, the phenomenology of the apparent horizons appears to be even more bizarre than in the mcvittie case and involves the creation or disappearance also of inner black hole apparent horizons @xcite . locating the apparent horizons and understanding , at least in principle , their behaviour is not the whole story .
the recent work @xcite studying the global structure of the mcvittie solution has unveiled a new feature which is believed to be generic : radial ingoing null geodesics do not penetrate the black hole apparent horizon to reach the @xmath55 singularity , but are asymptotic to this horizon . in our opinion , this feature is not too surprising for a solution in which radial flow onto the central black hole is excluded by construction .
the property of radial ingoing null geodesics merely reflects the mcvittie no - accretion condition .
in fact , the ingoing radial null geodesics can be seen as the test - particle limit of a gravitating null dust ( which however , would be forbidden by the no - accretion condition and could not fit in the mcvittie spacetime ) . future work to fully understand this feature , as well as more general solutions representing black holes embedded in cosmological backgrounds , will be presented elsewhere .
thakurta , _ indian j. phys . _ * 55b * , 304 ( 1981 ) ; d. kastor and j.h .
traschen , _ phys .
d _ * 47 * , 5370 ( 1993 ) ; m.l .
mcclure , k. anderson , and k. bardahl , arxiv:0709.3288 ; _ phys .
d _ * 77 * , 104008 ( 2008 ) ; m. nozawa and h. maeda , _ class .
quantum grav . _
* 25 * , 055009 ( 2008 ) ; k. maeda , n. ohta and k. uzawa , _ j. high energy phys . _ * 0906 * , 051 ( 2009 ) ; c .- y . sun , arxiv:0906.3783 ; _ commun . theor . phys .
_ * 55 * , 597 ( 2011 ) ; j.t .
firouzjaee and r. mansouri , _ gen .
gravit . _ * 42 * , 2431 ( 2010 ) ; b.j .
carr , t. harada , and h. maeda , _ class .
quantum grav . _
* 27 * , 183101 ( 2010 ) ; h. culetu , arxiv:1201.3796 .
a. krasinski and c. hellaby , _ phys .
d _ * 69 * , 043502 ( 2004 ) ; p. jacewicz and a. krasinski , _ gen .
gravit . _ * 44 * , 81 ( 2012 ) ; c. gao , x. chen , y .- g .
shen , and v. faraoni , _ phys .
d _ * 84 * , 104047 ( 2011 ) .
j. pachner , _ phys .
rev . _ * 132 * , 1837 ( 1963 ) ; _ phys .
b _ * 137 * , 1379 ( 1965 ) ; w.m .
irvine , _ ann .
( ny ) _ * 32 * , 322 ( 1965 ) ; r.h . dicke and p.j.e .
peebles , _ phys .
* 12 * , 435 ( 1964 ) ; c. callan , r.h .
dicke , and p.j.e .
peebles , _ am .
j. phys . _ * 33 * , 105 ( 1965 ) ; p. death , _ phys .
d _ * 11 * , 1387 ( 1975 ) ; r.p.a .
newman and g.c .
mcvittie , _ gen .
gravit . _ * 14 * , 591 ( 1982 ) ; r. gautreau , _ phys . rev
. d _ * 29 * , 198 ( 1984 ) ; p.a .
hogan , _ astrophys .
j. _ * 360 * , 315 ( 1990 ) ; j.l .
anderson , _ phys .
_ * 75 * , 3602 ( 1995 ) ; w.b .
bonnor , _ mon . not .
soc . _ * 282 * , 1467 ( 1996 ) ; a. feinstein , j. ibanez , and r. lazkoz , _ astrophys .
j. _ * 495 * , 131 ( 1998 ) ; k.r . nayak , m.a.h .
maccallum , and c.v .
vishveshwara , _ phys .
* 63 * , 024020 ( 2000 ) ; v. guruprasad , gr - qc/0005090 ; gr - qc/0005014 ; g.a .
baker jr . , astro - ph/0003152 ; a. dominguez and j. gaite , _ europhys . lett . _ * 55 * , 458 ( 2001 ) ; t.m . davis and c.h .
lineweaver , aipc * 555 * , 348 ( 2001 ) ; g.f.r .
ellis , _ int .
a _ * 17 * , 2667 ( 2002 ) ; c. stornaiolo , _ gen .
* 34 * , 2089 ( 2002 ) ; t.m .
davis , c.h .
lineweaver , and j.k .
webb , _ am .
j. phys . _
* 71 * , 358 ( 2003 ) ; l. lindegren and d. dravins , _ astron .
astrophys . _ * 401 * , 1185 ( 2003 ) ; c.j .
gao , _ class .
quantum grav . _
* 21 * , 4805 ( 2004 ) ; t.m .
davis and c.h .
lineweaver , _ publ .
* 21 * , 97 ( 2004 ) ; t.m .
davis , phd thesis , univ . of new south wales ( astro - ph/0402278 ) ; d.p .
sheehan and v.g .
kriss , astro - ph/0411299 ; w.j .
clavering , _ am .
j. phys . _
* 74 * , 745 ( 2006 ) ; z .- h .
li and a. wang , _ mod .
lett . a _ * 22 * , 1663 ( 2007 ) ; o. gron and o. elgaroy , astro - ph/0603162 ; l.a .
barnes , m.j .
francis , j.b .
james , and g.f .
lewis , _ mon . not .
_ * 373 * , 382 ( 2006 ) ; r. lieu and d.a .
gregory , astro - ph/0605611 ; p.k.f .
kuhfittig , _ int . j. pure appl . math . _ * 49 * , 577 ( 2008 ) [ gr - qc/0608120 ] ; g.s .
adkins , j. mcdonnell , and r.n . fell , _ phys .
d _ * 75 * , 064011 ( 2007 ) ; d.l .
wiltshire , _ new j. phys . _ * 9 * , 377 ( 2007 ) ; m. sereno and p. jetzer , _ phys . rev .
d _ * 75 * , 064031 ( 2007 ) ; p.d .
noerdlinger and v. petrosian , _ astrophys . j. _ * 168 * , 1 ( 1971 ) ; h. sato and k. maeda , _ prog .
phys . _ * 70 * , 119 ( 1983 ) .
death , _ phys .
d _ * 11 * , 1387 ( 1975 ) ; f.i .
cooperstock , v. faraoni , and d.n .
vollick , _ astrophys .
j. _ * 503 * , 61 ( 1998 ) ; b. bolen , l. bombelli , and r. puzio , _ class .
quantum grav . _ * 18 * , 1173 ( 2001 ) ; m.t .
busha , f.c .
adams , r.h .
wechsler , and a.e .
evrard , _ astrophys .
j. _ * 596 * , 713 ( 2003 ) ; c. gao and s.n .
zhang , _ phys .
b _ * 595 * , 28 ( 2004 ) .
m. mizony and m. lachize - rey , _ astron .
astrophys . _
* 434 * , 45 ( 2005 ) ; d. izzo and a. rathke , astro - ph/0504634 ; f.j .
oliveira , gr - qc/0610029 ; m. lachize - rey , _ class .
quantum grav . _
* 24 * , 2735 ( 2007 ) ; j. rosales and j. sanchez - gomez , gr - qc/99810085 ; m. carrera and d. giulini , _ class .
quantum grav .
_ * 23 * , 7483 ( 2006 ) ; c. lammerzhal and o. preuss , gr - qc/0604052 ; s.g .
turyshev and j.g .
williams , _ int
* 16 * , 2165 ( 2007 ) ; h .- j .
fahr and m. siewert , gr - qc/0610034 ; j.g .
williams , s.g .
turyshev , and d.h .
boggs , _ phys .
* 98 * , 059002 ( 2007 ) ; y.v .
dumin , _ phys .
lett . _ * 98 * , 059001 ( 2007 ) ; s. nesseris and l. perivolaropoulos , _ phys .
d _ * 70 * , 123529 ( 2004 ) ; r.h .
price , arxiv : gr - qc/0508052 ; a. balaguera - antolinez and m. nowakowski , _ class .
quantum grav .
_ * 24 * , 2677 ( 2007 ) ; b. mashhoon , n. mobed , and d. singh , _ class . quantum grav . _
* 24 * , 5031 ( 2007 ) .
a. ashtekar and b. krishnan , _ phys .
lett . _ * 89 * , 261101 ( 2002 ) ; _ phys .
d _ * 68 * , 104030 ( 2003 ) ; _ living rev .
rel . _ * 7 * , 10 ( 2004 ) ; i. booth , _ can .
j. phys . _
* 83 * , 1073 ( 2005 ) ; a.b .
nielsen , _ gen .
gravit . _ * 41 * , 1539 ( 2009 ) ; m. visser , arxiv:0901.4365 .
sushkov and s .- w .
kim , _ gen .
gravit . _ * 36 * , 1671 ( 2004 ) ; v. faraoni and w. israel , _ phys .
* 71 * , 064017 ( 2005 ) ; h. maeda , t. harada , and b. j. carr , _ phys .
d _ * 79 * , 044034 ( 2009 ) ; i. bochicchio and v. faraoni , _ phys .
d _ * 82 * , 044040 ( 2010 ) .
brill , g.t .
horowitz , d. kastor , and j. traschen , _ phys .
d _ * 49 * , 840 ( 1994 ) ; h. saida , t. harada , and h. maeda , _ class .
quantum grav . _
* 24 * , 4711 ( 2007 ) ; d.n .
vollick , _ phys .
d _ * 76 * , 124001 ( 2007 ) ; y. gong and a. wang , _ phys .
lett . _ * 99 * , 211301 ( 2007 ) ; f. briscese and e. elizalde , _ phys .
d _ * 77 * , 044009 ( 2008 ) ; m. akbar and r .-
g . cai , _ phys .
b _ * 635 * , 7 ( 2006 ) ; p. wang , _ phys . rev
. d _ * 72 * , 024030 ( 2005 ) ; h. mohseni - sadjadi , _ phys .
d _ * 76 * , 104024 ( 2007 ) ; r. di criscienzo , m. nadalini , l. vanzo , and g. zoccatelli , _ phys . lett .
b _ * 657 * , 107 ( 2007 ) ; v. faraoni , _ phys . rev
* 76 * , 104042 ( 2007 ) ; m. nadalini , l. vanzo , and s. zerbini , _ phys . rev .
d _ * 77 * , 024047 ( 2008 ) ; s.a .
hayward , r. di criscienzo , l. vanzo , m. nadalini , and s. zerbini , _ class .
quantum grav . _
* 26 * , 062001 ( 2009 ) ; s.a .
hayward , r. di criscienzo , m. nadalini , l. vanzo , and s. zerbini , _ aip conf .
* 1122 * , 145 ( 2009 ) ; r. di criscienzo , m. nadalini , l. vanzo , s. zerbini , g. zoccatelli , _ phys . lett .
b _ * 657 * , 107 ( 2007 ) ; r. brustein , d. gorbonos , and m. hadad , _ phys .
d _ * 79 * , 044025 ( 2009 ) ; v. faraoni , _ entropy _ * 12 * , 1246 ( 2010 ) .
_ , _ astron
. j. _ * 116 * , 1009 ( 1998 ) ; _ astron .
j. _ * 118 * , 2668 ( 1999 ) ; _ astrophys . j. _ * 560 * , 49 ( 2001 ) ; _ astrophys . j. _ * 607 * , 665 ( 2004 ) ; s. perlmutter _
_ , _ nature _ * 391 * , 51 ( 1998 ) ; _ astrophys . j. _ * 517 * , 565 ( 1999 ) ; j.l .
_ , _ astrophys
. j. _ * 594 * , 1 ( 2003 ) ; r. knop _ et al .
_ , _ astrophys
. j. _ * 598 * , 102 ( 2003 ) ; b. barris _
_ , _ astrophys
. j. _ * 602 * , 571 ( 2004 ) .
s. nojiri and s.d .
odintsov , _ phys .
d _ * 70 * , 103522 ( 2004 ) ; i. brevik , s. nojiri , s.d .
odintsov , and l. vanzo , _ ibid . _ * 70 * , 043520 ( 2004 ) ; s. nojiri and s.d .
odintsov , _ phys .
b _ * 595 * , 1 ( 2004 ) ; e. elizalde , s. nojiri , and s.d .
odintsov , _ phys .
d _ * 70 * , 043539 ( 2004 ) ; s. nojiri and s.d .
odintsov , _ phys .
b _ * 599 * , 137 ( 2004 ) ; k. lake , _ class . quantum grav . _ * 21 * , l129 ( 2004 ) ; p.f .
gonzalez - diaz and c.l .
siguenza , _ nucl .
b _ * 697 * , 363 ( 2004 ) ; r.m .
buny and d.h .
hsu , _ phys .
b _ * 632 * , 543 ( 2006 ) ; d.h . hsu , a. jenskins , and m.b .
wise , _ ibid .
_ * 597 * , 270 ( 2004 ) ; j.a.s . lima and j.s .
alcaniz , _ ibid . _
* 600 * , 191 ( 2004 ) ; y. gong , b. wang , and a. wang , _ phys .
d _ * 75 * , 123516 ( 2007 ) ; j.a.s .
lima and s.h .
pereira , _ ibid . _ * 78 * , 083504 ( 2008 ) ; g. izquierdo and d. pavon , arxiv : gr - qc/0612092 ; h.m .
sadjadi , _ phys .
d _ * 73 * , 063525 ( 2006 ) . | the bizarre behaviour of the apparent ( black hole and cosmological ) horizons of the mcvittie spacetime is discussed using , as an analogy , the schwarzschild - de sitter - kottler spacetime ( which is a special case of mcvittie anyway ) . for a dust - dominated `` background '' universe ,
a black hole can not exist at early times because its ( apparent ) horizon would be larger than the cosmological ( apparent ) horizon .
a phantom - dominated `` background '' universe causes this situation , and the horizon behaviour , to be time - reversed . |
the dynamics of adsorbed molecules is a fundamental issue in interface science @xcite and is also crucial in a large number of emerging technologies @xcite .
adsorption at the solid - liquid interfaces arises , for instance , in many biological contexts involving protein deposition @xcite , in solutions or melts of synthetic macromolecules @xcite , in colloidal dispersions @xcite , and in the manufacture of self - assembly mono- and multi - layers @xcite .
experimental studies of surfactant molecules @xcite , proteins @xcite and synthetic polymers @xcite confined to interfaces have identified different types of surface translational motions .
one is in - surface self - diffusion of individual molecules , which has been investigated with fluorescence recovery after photobleaching ( frap ) methods @xcite .
measured self - diffusivity at liquid - solid interfaces are much smaller than bulk values @xcite and similar to bulk values in liquid - gas cases @xcite .
a second source of motion , exclusive to the liquid - fluid interface , is surface visco - elasticity .
adsorbed surface phases possess compressibility and viscosities which govern the dynamics of the surface density waves .
their relaxation kinetics have been intensively studied experimentally @xcite-@xcite and theoretically @xcite , with direct viscoelastic measurement @xcite and surface light scattering studies @xcite providing considerable support for current theories . in this paper
we explore another mechanism called _ bulk - mediated surface diffusion_. this mechanism arises at interfaces separating a liquid bulk phase and a second phase which may be either solid , liquid , or gaseous .
whenever the adsorbed species are soluble in the liquid bulk , adsorption - desorption processes occur continuously .
these processes generate surface displacement because desorbed molecules undergo fickian diffusion in the liquid s bulk , and are later re - adsorbed elsewhere .
when this process is repeated many times , an effective diffusion results for the molecules on the surface .
the importance of bulk - surface exchange in relaxing homogeneous surface density perturbations is experimentally well established @xcite . here , unlike our previous work @xcite on the adsorption - desorption mechanism
, we assume that the bulk is finite .
the main goal of this work is the observation of an effective diffusion process at the interface @xmath0 .
we calculate analytically its variance , @xmath6 , as a function of time and @xmath7 , the conditional probability of being on the surface at time @xmath8 since the particle arrived at @xmath9 , that we indicate as @xmath10 .
the organization of the paper is as follows . in the next section
we describe the model by means of a set of master equations and solve this system in fourier - laplace space .
after that , we present a special case ( the bilayer case ) where we can obtain explicit analytical results for @xmath10 and @xmath6 .
next we compare these theoretical results with numeric monte carlo simulations .
moreover , we present results obtained by numerical inverse transformation of the equations for @xmath11 and @xmath12 for the cases with more layers . in the last section we discuss the results and draw some conclusions .
let us start with the problem of a particle making a random walk in a finite cubic lattice .
the bulk is bounded in the @xmath13 direction where the particles can move from @xmath5 to @xmath14 .
the @xmath3 and @xmath4 directions remain unbounded .
the position of the walker is defined by a vector @xmath15 whose components are denoted by the integer numbers @xmath16 corresponding to the directions @xmath3 , @xmath4 and @xmath13 respectively . in order to find the solution to the eq .
( [ c6mat1 ] ) , we decompose the @xmath37 matrix according to @xmath41 where @xmath42 @xmath43 @xmath44 and @xmath45 , \nonumber \\ \delta_2 & = & \delta - \gamma .\end{aligned}\ ] ] we also define @xmath46^{-1 } , \nonumber \\ \tilde{g}^1 & = & [ s \tilde{i } - ( a(k_x , k_y ) \tilde{i } + \tilde{h}_0 + \tilde{h}_1)]^{-1}.\end{aligned}\ ] ] a formal solution of the eq .
( [ c6mat1 ] ) is @xmath47^{-1}.\ ] ] we can show , by applying the dyson formula @xcite , that @xmath48 @xmath49 the solution for @xmath50 can be obtained analytically @xcite .
the result is @xmath51 where @xmath52 and @xmath53 the constant @xmath54 is obtained by exploiting the orthonormality relations for the @xmath55 functions given by @xmath56 while the completeness relation for the @xmath55 functions is written as @xmath57 the corresponding expression for @xmath54 is @xmath58 we are now able to find the statistical quantities which describe the diffusion problem over the surface .
we are interested in the probability of finding a particle in the site @xmath59 at time @xmath8 given it was in @xmath60 at @xmath9 .
this quantity is obtained by applying the inverse laplace transform on the @xmath61 matrix element .
another direct measurable experimental magnitude @xcite is the variance ( @xmath6 ) of the probability distribution at time @xmath8 over the plane @xmath62 this quantity measures the particles dispersion over the surface and has a direct relation to the diffusion coefficient .
when @xmath63 is known , the variance is calculated in the following manner @xmath64 in order to obtain this expression , we have assumed symmetric properties for the diffusion along the @xmath3 and @xmath4 directions , that is @xmath65 .
finally the variance in the laplace space can be obtained as @xmath66 [ \tilde{g}_{11}]\mid_{k_x = k_y=0}.\ ] ]
so far , we have developed a general theory describing the important problem of the diffusion of a particle system over a surface which is surrounded by a bounded bulk .
the expressions we found , are in the laplace space and the obtention of the inverse transform is usually a non trivial task . in this section
we restrict the problem to finding a particular solution , considering the case of a system of particles moving inside a bilayer , that is the space formed by the surface where we investigate the movement of the particles and a second layer . to obtain the @xmath63 ( @xmath10 )
, we take the expression in eq .
( [ c6sol ] ) with @xmath67 and evaluate for @xmath68
. then we obtain the inverse laplace transform .
the result is @xmath69}.\ ] ] this expression establishes that this probability is only a function of the temporal adsorption and desorption rates and does not depend on the temporal rates on the plane . from this equation ,
we can also obtain the initial condition of the problem by evaluating eq .
( [ c6pz1 ] ) at @xmath70 , obtaining @xmath71 .
if we consider the long time limit , the expression has the following behavior @xmath72 this expression shows that , at this limit , the probability of finding the particle in the @xmath0 plane does not decay to zero ( as happens in unbounded systems @xcite ) , but it approaches an asymptotic value which is a function of the ratio of the temporal rates . if the adsorption rate @xmath73 is large ( compared with the desorption rate @xmath23 ) , then the probability is large , the particles go back to the plane frequently . on the other hand , if the desorption rate is large , the particles go away from the surface and so the probability decays .
these behaviors are shown and quantified by eq .
( [ c6pz1 ] ) .
in the case of the variance , the result we have obtained can be decomposed as follows @xmath74 where @xmath75 } - 1 \right ) \right .
\nonumber \\ \left .
+ \frac{2 \gamma \delta + \delta^2}{(\gamma + \delta)^2 } t\ , \ , \exp{[-(\gamma + \delta ) t ] } + \frac{\gamma^2}{(\gamma + \delta)^2 } t \right),\end{aligned}\ ] ] @xmath76 } - 1 \right ) \right .
\nonumber \\ \left .
+ \frac{2 \gamma \delta}{(\gamma + \delta)^2}\ , t\ , \exp{[-(\gamma + \delta ) t ] } + \frac{2
\gamma \delta}{(\gamma + \delta)^2 } t \right).\end{aligned}\ ] ] from eqs .
( [ c6vari4 ] ) , ( [ c6rplano ] ) and ( [ c6rvol ] ) we can make the following observations .
firstly , the variance can be decomposed into two contributions , one corresponding to the particle movement across the bulk and the other to the surface movement .
the dependence of the variance on the time rates parallel to the surface is linear , while in these same relations , the adsorption and desorption rates enter in a more complicated way .
the functional form obtained is similar for both movements ( which however are nonindependent ) . for a large evolution time
, the mean square distance or dispersion grows linear with time ; each diffusive process has its own slope or growing rate ( these slopes are associated with the diffusion coefficient ) .
however , this is an expected behavior due to the model we are using .
the other contributions are transient ones , that decay with a time constant @xmath77 , hence , the stronger the adsorption or desorption , the faster is the decay of these contributions .
in this section we show the theoretical results , including some special numerical procedures , and make the comparison with monte carlo simulations . in all simulations
we have fixed the following parameters : @xmath78 and @xmath79 , and have averaged over @xmath80 realizations . in fig .
[ f16 ] we present the theoretical - numerical ( that is using a computer program to calculate the inverse laplace transform @xcite ) , theoretical results obtained from eq .
( [ c6pz1 ] ) and simulation results for the temporal evolution of the probability to find the system on the surface for the bilayer case . here
we show the curves for two values of the desorption rate @xmath23 .
as is apparent from the figure , there is excellent agreement between the theoretical and simulation results .
such excellent agreement indicates that the numerical procedure for the obtention of the inverse laplace transform is a reliable tool , and that we can trust their results in those cases where analytical results are not accessible ( for instance the cases with larger number of layers that we will consider in the following ) . in fig .
[ f26 ] , and again for the bilayer case , we depict for the variance ( @xmath6 ) , both theoretical and numerical results .
again the agreement is excellent . in figs .
[ f36 ] and [ f46 ] we present the @xmath81 and the @xmath6 but now for a number of layers larger than two .
here we compare the theoretical and numerical results again .
we remark that the theoretical results were obtained fitting the inverse laplace transformation of eqs .
( [ c6g0 ] ) and ( [ c6vari3 ] ) numerically .
again we have found an excellent agreement between theoretical and simulation results .
figures [ f56 ] , [ f66 ] and [ f76 ] correspond to an analysis for the asymptotic behavior of the finite system .
figure [ f56 ] depicts the curve obtained for the @xmath11 as a function of @xmath82 ( the number of layers ) for three different and large observational times : @xmath83 .
the insert shows that for these times , the system is well inside the asymptotic region . as can be seen from the figure , there is a maximum in the motion of the system . in other words
, there is an optimal " number of layers for which the system can spread more rapidly . for a larger number of layers , the system converges to an asymptotic limit .
this is reasonable due to the fact that the finiteness of the system tends to disappear .
figure [ f66 ] depicts the same behavior as fig .
[ f56 ] but now we have fixed an observational time ( @xmath84 ) and we have used the desorption rate ( @xmath23 ) as parameter .
the figure also shows the maximum on the number of layers again but from this figure we can see that this maximum moves towards the lower layers as the desorption rate increases . the insert shows the time evolution of @xmath11 for large values of @xmath82 .
we compare this behavior with the one corresponding to the infinite case @xcite . from this figure
we can say that the finite system behaves , for the used parameters , as an infinite one when the number of layers is @xmath85 .
if the parameter @xmath23 is increased , the maximum finally disappears as can be seen in fig .
this figure shows two curves for two different @xmath23 for @xmath86 .
the insert shows the @xmath10 for these desorption rates .
finally we show a fitting of @xmath6 as a function of @xmath82 for a large evolution time . for fitting purposes we have used the following function @xmath87 where @xmath39 is a constant and @xmath88
is the fitting parameter . in fig .
[ f86 ] we depict the results obtained for @xmath88 as a function of @xmath82 .
we observe two regions : for @xmath89 the particles describe a diffusion movement . in particular this behavior was shown for the bilayer model .
the insert of the figure shows a zoom for a small number of layers , and for two different times .
the linear behavior is apparent .
a second region appears for @xmath90 where the movement results subdiffusive . for large
@xmath82 the @xmath88 parameter approaches an asymptotic value equal to @xmath91 .
this situation was predicted @xcite for infinite systems .
we have studied the evolution of particles diffusing in a volume , but have analyzed the statistical properties of their evolution on a surface .
the diffusion can be performed on the surface itself or across the bulk surrounding the surface . in this work
we focus on the particle diffusion due to the movement across the bulk , that is we have considered the _ bulk mediated surface diffusion _ of the particles .
the main feature of the bulk is its finiteness in one direction ( the axial one ) .
the other directions are unbounded .
this work complements a previous study @xcite in which we have analyzed the particle diffusion in a semi - infinite bulk .
here we have presented a theoretical model based on a set of master equations which describe the movement of particles over a simple cubic lattice .
this model is general in the sense that we include both kind of particle movement : on the surface and on the bulk .
we solved this problem by using techniques of the laplace and fourier transformation and exploiting _
dyson s formula_. it is worth remarking here that the model can describe the evolution of the system everywhere in the bulk , and the evolution of the system for all time .
we particularize to the study over a single plane , i.e. the surface .
we have obtained general solutions in the laplace space for the @xmath10 and the @xmath92 for any kind of bulk . in order to find an analytical solution in space and time
, we have particularized the problem to the bilayer case .
the expressions obtained with this assumption were compared with monte carlo simulations , finding excellent agreement .
moreover we were able to handle cases with more layers . in these cases , we have obtained the inverse laplace transform using numerical methods , and compared these results with monte carlo simulations .
the agreement between both results are excellent again .
as indicated above , the numerical procedure to obtain the inverse laplace transform is a reliable tool , and we have shown that we can trust their results in those cases where analytical results are not accessible . finally we have studied the behavior of the spreading in the asymptotic region as a function of the number of layers of the system .
we observed an optimal `` number of layers for which @xmath6 reaches its maximum value .
it is worth remarking here that this effect occurs in the limit of strong adsorption '' , that is when the relation @xmath93 is small , and disappears in the limit of weak adsorption " , that is when @xmath94 is large . a possible generalization of the present and the previous related work @xcite consists in considering the possibility of non markovian dynamics , and study the effect of such dynamics on the statistical features of the system .
this is the subject of further work .
* acknowledgments : * the authors thank v. grnfeld for a critical reading of the manuscript .
hsw acknowledges the partial support from anpcyt , argentine , and thanks the mecyd , spain , for an award within the _ sabbatical program for visiting professors _ , and to the universitat de les illes balears for the kind hospitality extended to him . | we address the dynamics of adsorbed molecules ( a fundamental issue in surface physics ) within the framework of a master equation scheme , and study the diffusion of particles in a finite cubic lattice whose boundaries are at the @xmath0 and the @xmath1 planes where @xmath2 , while the @xmath3 and @xmath4 directions are unbounded .
as we are interested in the effective diffusion process at the interface @xmath5 , we calculate analytically the conditional probability for finding the system on the @xmath0 plane as well as the surface dispersion as a function of time and compare these results with monte carlo simulations finding an excellent agreement . |
the study of periodic surfaces has recently encountered new developments in homogeneous @xmath0-spaces . in @xmath1 ,
new examples have been constructed such as for instance doubly periodic minimal surfaces by mazet , rodrguez and rosenberg @xcite or genus one constant mean curvature @xmath2 surfaces by plehnert @xcite .
other examples are involved in important results such as the resolution of alexandrov problem in a quotient space of @xmath1 by menezes @xcite or the fact the calabi - yau conjectures do not hold for embedded minimal surface in @xmath1 by rodrguez and tinaglia @xcite . in heisenberg space ,
new examples are mostly of scherk type like the jenkins - serrin theorem for compact domains obtained by pinheiro @xcite . the present paper deals with the construction of another kind of scherk type surfaces , called _ most symmetric saddle towers _ , in heisenberg space .
a saddle tower is a minimal surface that can be thought of as the desingularization of @xmath3 vertical planes , @xmath4 , intersecting along a vertical geodesic ; in particular it is a complete embedded minimal surface with @xmath5 planar ends .
historically , they were first found by scherk @xcite for @xmath6 in euclidean space @xmath7 a saddle tower with @xmath6 is usually called a _ singly periodic scherk surface _ and ( more than ) a century and half later karcher @xcite generalized the construction to any @xmath4 .
another couple of decades later , morabito and rodrguez @xcite and pyo @xcite constructed saddle towers in @xmath1 .
recently , menezes @xcite constructed the singly ( and doubly ) periodic scherk surfaces in semi - direct product spaces , including heisenberg space .
we use a classical method to construct saddle towers in heisenberg space ( see section [ sec : saddle ] ) .
we first construct a sequence of embedded minimal disks by solving plateau problems on suitable jordan curves such that it converges to an embedded minimal surface bordered by horizontal geodesic arcs , which we call the _
fundamental piece_. reflecting the fundamental piece along the geodesics in its boundary , we get the desired saddle tower : thm : saddle for any @xmath4 and @xmath8 , there exists a properly embedded singly periodic minimal surface of genus zero with @xmath5 planar ends distributed at constant angle @xmath9 .
we call it a _ most symmetric saddle tower_. since the saddle tower is composed of reflections of the fundamental piece , its ends must be equiangularly distributed to ensure the embeddedness of the surface .
note that morabito and rodrguez @xcite constructed non equiangular saddle towers with a conjugate plateau technique , which does not apply in heisenberg space .
the key step is to prove the convergence of the sequence of minimal disks , which is done by using a suitable barrier .
we construct that barrier in section [ sec : barrier ] by deforming a minimal plane so that we control the asymptotic behavior of the deformed minimal surfaces .
the deformation technique is based on the existence of an elliptic operator , called the _ compactified mean curvature operator _ , containing information on both the asymptotic behavior and the mean curvature .
this technique has already been used by the author and hauswirth @xcite for surfaces in @xmath1 of constant mean curvature @xmath2 .
it is an immediate consequence of weierstrass representation that the construction of such surfaces is not possible in euclidean space since under reasonable asumptions on the total curvature minimal ends are asymptotically rotational .
the present barrier construction highlights another difference with euclidean case : thm : disjgraphs in heisenberg space , the number of disjoint domains supporting minimal vertical graphs is unbounded .
we recall basic definitions and properties on heisenberg space and fix classic notations .
@xmath0-dimensional heisenberg space @xmath10 is a nilpotent lie group which is usually represented in @xmath11 by : @xmath12 and its lie algebra @xmath13 can be seen as the subset of @xmath14 : @xmath15 in this paper we use the parametrization of @xmath10 induced by the exponential map : @xmath16 with these notations , @xmath10 identifies with @xmath7 and the group law is : @xmath17 we call _ canonical frame _ the frame @xmath18 which is the extension by left translation of the canonical frame of @xmath7 at the origin i.e. : @xmath19 endowing @xmath10 with the following left - invariant metric : @xmath20 makes the canonical frame orthonormal .
heisenberg space is a riemannian fibration for the projection on the first two coordinates @xmath21 , which means that the isomorphism @xmath22 onto the euclidean plane is an isometry .
given this structure , vector fields spanned by @xmath23 are referred to as _ horizontal _ and the direction @xmath24 is said _
vertical_. in particular , a _ vertical graph _ in @xmath10 is a complete immersion transverse to @xmath24 . in the sequel ,
we mostly work in cylindrical coordinates @xmath25 with @xmath26 and @xmath27 such that : @xmath28 and we consider the _ cylindrical frame _ @xmath29 with : @xmath30 the cylindrical frame is also orthonormal and the levi - civita connection @xmath31 on @xmath10 in terms of @xmath29 is given by : @xmath32 the barrier constructed in section [ sec : barrier ] comes from a ( suitable ) deformation of a surface , labeled @xmath33 , to which we refer as the _ model surface_. in our model of @xmath10 , @xmath33 is nothing but the entire graph @xmath34 .
it is a complete embedded minimal surface , its tangent plane is spanned by the coordinate vector fields : @xmath35 and has unit normal : @xmath36 the end of @xmath33 is of annulus type and is vertical in the sense that the normal vector @xmath37 is asymptotically horizontal : @xmath38 when @xmath39 .
consider a point @xmath40 in @xmath10 and a unit vector @xmath41 in the tangent space @xmath42 at @xmath43 , written @xmath44 with @xmath45 , @xmath46 and @xmath47 .
the geodesic passing through @xmath43 and directed by @xmath41 at @xmath43 admits the parametrization @xmath48 by arc length with ( see @xcite for details ) : @xmath49 } { \vspace*{1ex}}\\ { \displaystyle}\hspace{22em}+ \frac{1 + \gamma^2}{2\gamma } t- \frac{r^2}{4\gamma^2 } \sin(2\gamma t ) \end{array } { \right.}.\]]if @xmath50 and @xmath51 , meaning : @xmath52 with @xmath53 , @xmath54 and @xmath55 , the parametrization writes : @xmath56}{\vspace*{1ex}}\\ { \displaystyle}x_2(t)= \rho \sin\theta+ \frac{\rho}{4 } { \left[}\sin { \left(}\frac{4t}{\sqrt{4 + \rho^2}}+ \theta { \right)}- \sin \theta { \right]}{\vspace*{1ex}}\\ { \displaystyle}x_3(t)= \frac{\rho^2}{16 } \sin { \left(}\frac{4t}{\sqrt{4 + \rho^2 } } { \right)}+ \frac{8 + \rho^2}{16 }
\frac{4t}{\sqrt{4 + \rho^2 } } { \vspace*{1ex}}\end{array } { \right.}.\ ] ] fixing @xmath57 and letting @xmath58 vary , system becomes a parametrization of the equidistant surface to @xmath33 at signed ( normal ) distance @xmath59 . and making @xmath39 , we see that asymptotically the equidistant surface grows linearly with respect to @xmath60 namely , the equidistant surface asymptotically behaves like the quadric : @xmath61 the growth of equidistant surfaces to the model surface is important in section [ sec : barrier ] .
indeed , the information we have on the asymptotic behavior of the deformations of @xmath33 is precisely the asymptotic normal signed distance to @xmath33 .
[ def : asympdist ] consider a surface @xmath62 which can be parametrized by : @xmath63 at least for @xmath60 big enough and some @xmath27 .
when it exists , the _ asymptotic horizontal signed distance _ of @xmath62 to the model surface @xmath33 in the direction @xmath64 is the quantity : @xmath65 the qualification `` horizontal '' coming from the fact that @xmath37 is asymptotically horizontal .
the classical schwarz reflection for minimal surfaces in euclidean space states that if a minimal immersion contains a straight line in its boundary , it can be smoothly extended across that line by the @xmath66-rotation about the line .
this result applies to heisenberg space and in general to homogeneous @xmath0-spaces with @xmath67-dimensional isometry group if restricted to extensions across horizontal or vertical geodesics ( see @xcite and details in @xcite ) .
moreover , in heisenberg space the geodesic reflection along a horizontal geodesic passing through the vertical axis @xmath68 is exactly the euclidean @xmath66-rotation about the geodesic .
consider a minimal surface @xmath69 bordered only by horizontal and vertical geodesic arcs .
then @xmath69 can be extended to a smooth minimal surface @xmath70 by geodesic reflections along each geodesic in the boundary of @xmath69 .
the boundary of @xmath70 also only contains horizontal and vertical geodesic arcs and the extension process can be iterated .
note that a meeting point of several geodesic arcs of the boundary of @xmath69 is a removable singularity of the extended surface @xmath70 and thus @xmath70 is smooth at this point ( see @xcite ) .
we use schwarz symmetrization in section [ sec : saddle ] to extend the fundamental piece to a complete embedded smooth minimal surface with the desired symmetry properties .
let @xmath71 be the open unit disk , @xmath72 its closure and @xmath73 the polar coordinates on @xmath74 .
the boundary @xmath75 of @xmath76 is identified with @xmath77 .
the space @xmath78 , with @xmath79 and @xmath80 , is the usual hlder space over @xmath74 and @xmath81 is the subspace of @xmath78 of functions that are zero on the boundary of @xmath76 .
finally , we consider the spaces @xmath82 endowed with the natural scalar product denoted @xmath83 and hilbert norm @xmath84 .
the model surface @xmath33 is an embedded disk . we deform it by mean of a differential operator taking into account both the asymptotic behavior and the mean curvature .
the construction of this operator is called a _
compactification _ of the mean curvature since it is based on a conformal change of the induced metric to extend it on the boundary of the parametrizing disk . to do so , we need a conformal parametrization of @xmath33 which writes ( in polar coordinates ) : @xmath85 we parametrize entire vertical graphs of @xmath10 by : @xmath86 for some map @xmath87 .
we call such a parametrization _ graph coordinates at infinity_. in the sequel , we are interested in graphs such that @xmath88 and we use graph coordinates at infinity to compactify surfaces and quantify their asymptotic behavior .
indeed , using definition [ def : asympdist ] a graph @xmath89 with @xmath88 is at asymptotic horizontal signed distance to the model surface @xmath90 .
surfaces are thus considered as _ compact surfaces with boundary _ and we can apply the method first developed by white @xcite.also note that the value @xmath91 is invariant under vertical translations . from now on , to ease the notations , we denote with indexes @xmath92 quantities related to coordinates @xmath93 respectively .
[ thm : compcourb ] for any entire graph admitting graph coordinates @xmath89 with @xmath88 , the mean curvature @xmath94 verifies : @xmath95 where @xmath96 is the determinant of the metric induced by @xmath97 , @xmath98 and @xmath99 are @xmath100 functions on @xmath74 which are real - analytic in their variables and @xmath101 is a coercive matrix on @xmath74 .
the definition of the mean curvature of @xmath89 is the following : @xmath102 where @xmath103 ( resp .
@xmath104 ) is the derivative of @xmath89 with respect to @xmath105 ( resp .
@xmath64 ) , @xmath106 is the unit normal to @xmath89 and @xmath107 is the inverse matrix of the metric @xmath108 induced by @xmath89 .
namely , we have : @xmath109 with @xmath29 the cylindrical frame .
the first fundamental form @xmath110 is : @xmath111 } , \\
g_{12}(\eta)= -\frac{32r^3}{(1- r^2)^4 } { \left[}\eta+ \frac{1 + r^2}{4r } { \left(}\eta_1- \frac{\eta \eta_2}{2r } { \right)}(1- r^2)- \frac{(1 + r^2)^2 \eta_1 \eta_2}{32r^3 } ( 1- r^2)^2 { \right]}\\ \text{and } \quad g_{22}(\eta)= \frac{16r^2(1 + r^2)^2}{(1- r^2)^4 } { \left[}1- \frac{\eta_2}{1 + r^2 } ( 1- r^2)+ \frac{\eta_2 ^ 2}{16r^2 } ( 1- r^2)^2 { \right]},\end{gathered}\ ] ] and its determinant @xmath112 writes : @xmath113 with the following expression of @xmath114 : @xmath115}^{1/ 2}.\end{gathered}\ ] ] also the expression of the unit normal @xmath106 is : @xmath116}.\end{gathered}\ ] ] the computation detailed in appendix [ sec : proof ] gives the expression with the desired regularity and : @xmath117 which shows that @xmath118 is coercive on @xmath74 .
we call such a process a _ compactification _ of the mean curvature since the quantity @xmath119 can be extended to the boundary @xmath75 .
it is also strongly linked with the compactification of the induced metric @xmath110 by the following equality : @xmath120 from now on , we denote @xmath121 the operator : @xmath122 where @xmath94 is the mean curvature of @xmath89 , and we call it the _ mean curvature operator_. using theorem [ thm : compcourb ] , we define the _ compactified mean curvature operator _ to be : @xmath123 the _ compactified jacobi operator _ is @xmath124 .
we get that : @xmath125 where @xmath126 is the jacobi operator of @xmath33 , since it is a standard fact that : @xmath127 furthermore , conducting the computation in appendix [ sec : proof ] more carefully , we get an explicit expression of @xmath128 : @xmath129 where @xmath130 stands for the usual flat laplacian . we highlight two immediate consequences .
first , the vertical coordinate @xmath131 of @xmath37 meaning @xmath132 is in the kernel of @xmath128 i.e. @xmath133 . and second , @xmath128 verifies the following green identity : @xmath134 with @xmath135 the lebesgue measure on @xmath74 .
we deduce the following lemma : [ lem : nosol ] there is no solution @xmath136 to the equation : @xmath137 by contradiction , suppose such a @xmath138 exist and apply the green identity to @xmath131 and @xmath138 : @xmath139 since : @xmath140 this is impossible .
consider the restriction @xmath141 of @xmath128 to @xmath142 and @xmath143 .
we use the inclusions @xmath144 and denote @xmath145 the orthogonal to @xmath146 in @xmath147 for the natural scalar product of @xmath148 and @xmath149 . a standard result states that the restriction @xmath141 is a fredholm operator with index zero ( see @xcite ) and furthermore @xmath150 and @xmath151 .
let @xmath152 be the operator such that @xmath153 is the harmonic function on @xmath74 ( for the flat laplacian ) with value @xmath154 on the boundary @xmath155 . in the sequel
, we decompose @xmath156 into @xmath157 , meaning that any @xmath88 is in one - to - one correspondence with a triple @xmath158 such that : @xmath159 consider @xmath160 and @xmath161 , the orthogonal projections on @xmath146 and @xmath145 respectively .
we follow white @xcite and define a suitable map to apply the implicit function theorem : let @xmath162 be the map defined by : @xmath163 then @xmath164 is an isomorphism .
we compute that @xmath165 and we know @xmath145 is the range of @xmath141 , which means @xmath164 is an isomorphism .
therefore , we can apply the implicit function theorem to @xmath166 , which states that there exist an open neighborhood @xmath167 of @xmath168 in @xmath169 and a unique smooth map @xmath170 such that : @xmath171 consequently , define smooth maps @xmath172 and @xmath173 by : @xmath174 if an entire graph admits @xmath175 as graph coordinates at infinity , we say that @xmath176 are the _ data _ of the surface and that @xmath154 is the _ value at infinity_. [ lem : eta ] the map @xmath177 has the following properties : 1 .
[ pt : zero ] @xmath178 .
[ pt : valbound ] @xmath179 .
[ pt : diff ] @xmath180 .
point [ pt : zero ] is a direct consequence of the uniqueness in the implicit function theorem . for point
[ pt : valbound ] compute : @xmath181 and for point [ pt : diff ] , we only have to show @xmath182 . to do so , compute : @xmath183 thus , @xmath184 i.e. @xmath185 .
[ rmk : lambda ] lemma [ lem : eta ] point [ pt : valbound ] shows that the value at infinity of a surface @xmath175 does not depend on @xmath186 , meaning that given a value at infinity @xmath154 there exists a @xmath187-parameter family of surfaces all with value at infinity equals to @xmath154 .
moreover , it can be shown that any two surfaces in this family are congruent only up to a vertical translation . indeed , from the half - space theorem of daniel , meeks and rosenberg @xcite , we know that the difference of heights of two entire minimal graphs diverges unless the graphs differ from each other by a vertical translation . in our case , if @xmath188 then : @xmath189 and the difference of heights is bounded .
the values of the mean curvature of deformations @xmath175 of @xmath33 are determined by @xmath190 . indeed , for @xmath191 ,
we have @xmath192 and : @xmath193 consider @xmath194 .
the fact that the parameter @xmath186 is associated to vertical translations means that we can write @xmath195 , with @xmath196 a subset of @xmath197 . furthermore , since the construction is local , we can suppose @xmath196 connected .
[ prop : local ] @xmath196 is a codimension @xmath187 smooth submanifold of @xmath197 is a subset of : @xmath198 we show that @xmath190 is a submersion at @xmath168 . from equation
compute : @xmath199 using lemma [ lem : eta ] point [ pt : diff ] and that @xmath200 .
moreover , @xmath201 is not identically zero . indeed , using : @xmath202 using corollary [ lem : nosol ] with @xmath203 deduced from lemma [ lem : eta ] point [ pt : valbound ] . since @xmath204 is continuous and non zero at @xmath168 , there exists an open neighborhood of @xmath168 in @xmath169 on which @xmath190 is a submersion . therefore , up to a restriction on @xmath196 , we can suppose @xmath190 is a submersion on @xmath205 , which implies @xmath196 is a submanifold of @xmath197 of codimension @xmath187 .
the inclusion for @xmath196 is actually equivalent to the nullity of the vertical flux of an entire graph in @xmath10 .
consider a minimal surface admitting graph coordinates at infinity @xmath89 with @xmath88 .
as shown in @xcite , the vertical flux @xmath206 of @xmath89 is for any @xmath207 : @xmath208 when taking the limit when @xmath209 .
hence , if @xmath89 is an entire graph such that @xmath210 for some data @xmath176 with @xmath211 , the flux @xmath206 is zero and so is the mean of @xmath154 .
treibergs @xcite showed that given a @xmath212 curve @xmath213 , there exists a constant mean curvature complete entire vertical graph in @xmath0-dimensional minkowski space which is asymptotically at signed distance @xmath154 from the light cone .
proposition [ prop : local ] is actually a @xmath214 local version of this result in @xmath10 : [ thm : treibergs ] consider @xmath215 small enough for the @xmath214-norm and with zero mean .
then there exists a minimal complete entire vertical graph at asymptotic horizontal signed distance @xmath154 from @xmath33 .
moreover , such a surface is unique up to vertical translations . for any @xmath154 sufficiently small in the @xmath214 norm
, we have @xmath211 and @xmath216 is a minimal entire graph admitting @xmath154 as value at infinity . and as in remark [ rmk : lambda ] , uniqueness comes from daniel , meeks and rosenberg half - space theorem @xcite . in the sequel , we fix a natural @xmath4 and a parameter @xmath8 .
to ease the writing , we denote @xmath217 and @xmath218 the horizontal geodesic directed by @xmath219 and passing through @xmath220 for any @xmath221 and @xmath222 .
we are interested in rotationally symmetric solutions to act as barriers in section [ sec : saddle ] .
they are constructed by the following result , which a corollary of theorem [ thm : treibergs ] : [ prop : symgraph ] there exist a minimal entire graph @xmath223 such that : 1 .
[ pt : reflection ] for any @xmath221 , @xmath223 contains the horizontal geodesic @xmath224 and is invariant for the geodesic reflection along that geodesic .
[ pt : positive ] the height of @xmath223 is nonnegative on the ( open ) angular sector @xmath225 . for point [ pt : reflection ] , fix @xmath226 small enough such that @xmath227 is in @xmath196 . from theorem
[ thm : treibergs ] , we know there exists a minimal entire graph @xmath223 with @xmath228 as graph coordinates at infinity , where the translation parameter @xmath186 is chosen so that @xmath223 contain the origin .
consider the map @xmath229 defined for some @xmath221 by : @xmath230 the surface @xmath231 is the image of @xmath228 under the geodesic reflection along @xmath224 .
the value at infinity of @xmath232 is : @xmath233 using remark [ rmk : lambda ] , we know that @xmath231 and @xmath228 differ by a vertical translation and since @xmath228 contains the origin which is fixed by the geodesic reflection along @xmath224 , we get @xmath234 . for point [ pt : positive ] , consider the restriction of @xmath223 to the angular sector @xmath235 , which we also denote @xmath223 in the sequel .
for any @xmath236 $ ] , we have @xmath237 and @xmath238 is positive in the interior .
hence , the height of @xmath223 is bounded from below and by vertical translation @xmath239 with positive @xmath240 , we can make @xmath241 and @xmath33 disjoint on @xmath235 .
now , suppose there exist a point inside the open angular sector at which the height of @xmath223 is negative .
translating back , we obtain a first contact point between @xmath242 and @xmath33 for some @xmath243 , which is impossible by maximum principle .
given an open subset @xmath244 of @xmath245 , @xmath246 , a minimal hypersurface in euclidean space @xmath247 is said to be _ supported _ on @xmath244 if it is the graph of a function that does not change sign over @xmath244 and is zero on the boundary @xmath248 .
a question of meeks and rosenberg @xcite is to know if the number of disjoint domains supporting minimal graphs is bounded . li and
wang @xcite proved it to be true in euclidean space for any dimension and tkachev @xcite refined the bounds . in @xmath10 , we consider open subsets of @xmath249 where @xmath249 is seen as the range of the projection @xmath66 and vertical minimal graphs .
the construction of surfaces @xmath223 shows : [ thm : disjgraphs ] in heisenberg space , the number of disjoint domains supporting minimal vertical graphs is unbounded .
in this section , we use notations introducted in section [ subsec : perioddef ] .
we build a saddle tower with @xmath5 ends distributed at constant angle @xmath250 . for any @xmath251 , consider the polygonal jordan curve @xmath252 which is the reunion of the following geodesic segments : @xmath253 note that @xmath254 are horizontal lifts of @xmath255 respectively . since the ( euclidean )
convex hull @xmath256 of @xmath252 is mean - convex , we know from meeks and yau @xcite that the plateau problem with boundary @xmath252 is solvable , meaning there exists an embedded minimal disk @xmath257 bordered by @xmath252 ( see figure [ fig : plateausol ] ) .
( 7.8 , 4.2)(-1.3 , -0.45 ) ( -1.1 , 0)(1 , 0)7.6 ( 0 , 0)(0 , 1)2 ( 6.19 , -0.3)@xmath258 ( -0.46 , 1.85)@xmath259 ( -0.84 , -0.42)(2 , 1)0.84 ( -1 , 0)(-1 , -0.25)(-0.75 , -0.375 ) ( -1.28 , -0.33)@xmath250 ( 6 , 0.75)(1.4 , 0.75)(1.2 , 1.05 ) ( 4.5 , 3)(2.5 , 2.01)(1.7 , 1.5 ) ( 3 , 1.5)(2 , 1)1.5 ( 0 , 1.5)(2 , 1)4.5 ( 0 , 0)(1 , 0)6 ( 0 , 1.5)(1 , 0)6 ( 6 , 0)(0 , 1)1.5 ( 4.5 , 2.25)(0 , 1)1.5 ( 0 , 0)(1.2 , 0.7)(1.2 , 1.05 ) ( 0 , 1.5)(1.2 , 1.4)(1.2 , 1.05 ) to ensure the convergence of a subsequence of the family @xmath260 when @xmath261 , we only need barriers from below and above and such barriers are the surfaces @xmath223 and @xmath262 respectively as constructed in proposition [ prop : symgraph ] . hence , there exists an embedded minimal surface @xmath263 bordered by @xmath264 which is the reunion of the horizontal geodesic rays : @xmath265 the surface @xmath263 is the fundamental piece we are looking for . extending @xmath263 by recursive geodesic reflections along the geodesics in its boundary .
[ thm : saddle ] for any natural @xmath4 and any @xmath8 , there exists a properly embedded minimal surface @xmath266 in @xmath10 of genus zero , invariant by the rotation of angle @xmath267 and axis @xmath68 and by the vertical translation of parameter @xmath268 .
moreover , for any @xmath221 , @xmath266 contains the geodesics @xmath224 and @xmath269 and is asymptotic away from @xmath68 to the vertical plane containing @xmath224 .
we call @xmath266 a _ most symmetric saddle tower_.
consider a map @xmath88 for some @xmath270 where @xmath156 denotes the usual hlder space over @xmath74 and let @xmath271 be the immersion given by : @xmath272 where @xmath73 are the polar coordinates on @xmath76 . denoting by indexes
@xmath92 the derivatives with respect to @xmath93 respectively , the first derivatives of @xmath89 are : @xmath109 with @xmath29 denoting the cylindrical orthonormal frame in @xmath10 and defined at generic point @xmath273 by : @xmath274 we then compute the first fundamental form @xmath108 : @xmath111 } , \\
g_{12}(\eta)= -\frac{32r^3}{(1- r^2)^4 } { \left[}\eta+ \frac{1 + r^2}{4r } { \left(}\eta_1- \frac{\eta \eta_2}{2r } { \right)}(1- r^2)- \frac{(1 + r^2)^2 \eta_1 \eta_2}{32r^3 } ( 1- r^2)^2 { \right]}\\ \text{and } \quad g_{22}(\eta)= \frac{16r^2(1 + r^2)^2}{(1- r^2)^4 } { \left[}1- \frac{\eta_2}{1 + r^2 } ( 1- r^2)+ \frac{\eta_2 ^ 2}{16r^2 } ( 1- r^2)^2 { \right]}.\end{gathered}\ ] ] and the determinant @xmath112 of the first fundamental form writes : @xmath113 with the following expression of @xmath114 : @xmath115}^{1/ 2}.\end{gathered}\ ] ] we also get the unit normal @xmath106 to @xmath89 : @xmath116}.\end{gathered}\ ] ] recall that the levi - civita connection @xmath31 on @xmath10 is given in the cylindrical frame @xmath29 by : @xmath32 we compute the conormal derivatives : @xmath275 the weingarten operator of @xmath89 is then determined by the following quantities : @xmath276}+ \frac{\eta_{11}}{w(\eta)},\end{gathered}\ ] ] with @xmath277 defined on @xmath74 , identically zero if @xmath278 and real - analytic in its variables . also :
@xmath279}+ \frac{\eta_{12}}{w(\eta)},\end{gathered}\ ] ] again with @xmath280 defined on @xmath74 , zero if @xmath278 and real - analytic in its variables , and : @xmath281}+ \frac{\eta_{22}}{w(\eta)},\end{gathered}\ ] ] with @xmath282 defined on @xmath74 , identically zero if @xmath278 and real - analytic in its variables .
we shall now compute the mean curvature itself .
denote : @xmath283 where @xmath107 is the inverse matrix of the metric @xmath110 .
we get : @xmath284}+ \frac{g^{11}(\eta)}{w(\eta ) } \eta_{11},\end{gathered}\ ] ] with @xmath285 defined on @xmath74 , identically zero if @xmath278 and real - analytic in its variables , also : @xmath286}+ \frac{g^{12}(\eta)}{w(\eta ) } \eta_{12},\end{gathered}\ ] ] with @xmath287 defined on @xmath74 , zero if @xmath278 and real - analytic in its variables , and : @xmath288}+ \frac{g^{22}(\eta)}{w(\eta ) } \eta_{22},\end{gathered}\ ] ] with @xmath289 defined on @xmath74 , zero if @xmath278 and real - analytic in its variables .
finally , we obtain a taylor expansion of the mean curvature : @xmath290 as before with @xmath291 defined on @xmath74 , identically zero if @xmath278 and real - analytic in its variables .
equation can be written : @xmath292 moreover , @xmath293 and @xmath294 are defined on @xmath74 and real - analytic in their variables , the matrix @xmath101 is coercive on @xmath74 , and @xmath99 is zero when @xmath278 .
william h. meeks , iii and shing tung yau , _ the classical plateau problem and the topology of three - dimensional manifolds .
the embedding of the solution given by douglas - morrey and an analytic proof of dehn s lemma _ , topology * 21 * ( 1982 ) , no . 4 , 409442 . | we construct most symmetric saddle towers in heisenberg space i.e. periodic minimal surfaces that can be seen as the desingularization of vertical planes intersecting equiangularly .
the key point is the construction of a suitable barrier to ensure the convergence of a family of bounded minimal disks .
such a barrier is actually a periodic deformation of a minimal plane with prescribed asymptotic behavior .
a consequence of the barrier construction is that the number of disjoint minimal graphs suppoerted on domains is not bounded in heisenberg space .
_ mathematics subject classification : _ _ 53a10 , 53c42_. |
the rate of mixing , which has not yet been observed , would provide an important window on new physics .
the contribution of box diagrams is very small due to gim suppression , and the mixing rate is believed to be dominated by long - distance processes .
these processes are themselves suppressed by @xmath7-flavor symmetry .
the mixing rate is usually expressed in terms of the parameters @xmath8 and @xmath9 , where @xmath10 and @xmath11 are the mass- and width - differences of the physical states , and @xmath12 is the average decay width .
numerous predictions for @xmath13 and @xmath14 exist @xcite ; a recent calculation gives @xmath15 @xcite .
non - standard - model ( sm ) processes are expected to enhance @xmath13 relative to @xmath14 , and measuring @xmath16 would be a strong indication of new physics @xcite .
in this paper we report a measurement of the parameter @xmath17 which is equal to @xmath14 in the limit where cp is conserved .
if cp is violated , the widths of @xmath18 and @xmath19 decays into the cp eigenstate @xmath20 differ .
these widths can be expressed as @xcite @xmath21 \\
\hat{\gamma}(\overline{d}{}^{\,0 } \ra k^+ k^- ) & = & \gamma \left[\,1 + r_m^{-1 } ( y \cos{\phi } + x \sin{\phi})\right]\,,\end{aligned}\ ] ] where @xmath22 and @xmath23 are cp - violating parameters defined in @xcite .
cp violation in mixing results in @xmath24 , and cp violation due to interference between mixed and unmixed decay amplitudes results in @xmath25 .
we investigate whether cp violation is present by measuring the quantity @xmath26 the sm predicts @xmath27 @xcite ; a large value would indicate physics beyond the sm . our measurement of @xmath28 is obtained by measuring the lifetime difference between @xmath2 and @xmath3 decays .
( throughout this paper , charge - conjugate modes are included unless noted otherwise . )
our measurement of is obtained by measuring the lifetime difference between @xmath3 and @xmath29 decays .
the analysis is based on 158 fb@xmath30 of data , and all results are preliminary . to identify the flavor of the @xmath18 and also reduce background ,
we require that all @xmath18 s originate from @xmath1 decays .
the belle detector is a large - solid - angle magnetic spectrometer that consists of a three - layer silicon vertex detector ( svd ) , a 50-layer central drift chamber ( cdc ) , an array of aerogel threshold erenkov counters ( acc ) , a barrel - like arrangement of time - of - flight scintillation counters ( tof ) , and an electromagnetic calorimeter comprised of csi(tl ) crystals ( ecl ) located inside a super - conducting solenoid coil that provides a 1.5 t magnetic field .
an iron flux - return located outside of the coil is instrumented to detect @xmath31 mesons and to identify muons ( klm ) .
the detector is described in detail elsewhere @xcite .
we select well - measured charged tracks that have impact parameters less than 1.0 cm in the radial direction and less than 2.0 cm along the beam direction ( @xmath32 ) with respect to the interaction point ( ip ) .
the @xmath18 daughter tracks are required to have at least two hits in each of the @xmath33-@xmath23 and @xmath32-measuring svd layers .
this requirement is not applied to the slow pion from @xmath34 decays . for charged particle identification , information from the cdc , tof , and acc subsystems
is combined to form an overall hadron likelihood : @xmath35 where @xmath36 stands for @xmath37 or @xmath38 .
charged particles are identified as pions or kaons using the likelihood ratios : @xmath39 kaon candidates are required to have @xmath40 .
this requirement is 88% efficient and has a pion misidentification probability of 8% .
pion candidates are required to satisfy @xmath41 .
both continuum and @xmath5 data are used in this analysis .
each event is required to contain at least one @xmath42 meson decaying to @xmath43 . to remove @xmath42 s resulting from @xmath44
meson decays , the @xmath42 is required to have a momentum in the @xmath6 center - of - mass frame ( cms ) greater than 2.5 @xmath45 .
@xmath18 meson candidates are reconstructed from @xmath2 and @xmath3 decays .
all @xmath18 candidates must satisfy @xmath46 , where @xmath47 and @xmath48 are the mean and standard deviation of a gaussian fit to the @xmath18 mass peak in data .
these parameters are calculated independently for each @xmath18 decay mode in order to help keep the @xmath49 and @xmath20 vertex resolution functions unbiased by the @xmath18 mass requirement . from the fits , @xmath48 is found to be 5.1 @xmath50 for @xmath2 and 4.5 @xmath50 for @xmath51 ( see table [ tab - datamass ] ) .
the momentum of the slow @xmath52 from @xmath34 decays is obtained by refitting using the beam position constraint , i.e. , the track must project back to the ip region .
the invariant mass of the @xmath42 candidates after slow @xmath52 refitting is required to satisfy @xmath53 .
the proper time between the production and decay of the @xmath18 candidates is determined by projecting the momentum @xmath55 and the flight length @xmath56 onto the @xmath57 plane transverse to the beam axis : @xmath58 the momentum @xmath55 is the vector sum of the @xmath38 and @xmath37 momenta , and @xmath56 is the displacement vector between the production and decay vertices .
information from the @xmath32 projections is not included due to the large uncertainty in the @xmath32 component of the ip . to improve the spatial resolution
, the vertex fit of @xmath59 decays is done in three steps .
first , the decay point is fit using only information from the @xmath18 daughter tracks
. then , the @xmath18 production vertex is found by fitting the reconstructed @xmath18 momentum vector with the ip region , whose profile is obtained from a separate study of the full data set .
finally , the @xmath18 decay vertex is corrected using information from the @xmath18 production vertex fit .
no mass constraint is applied to the @xmath18 candidates during the vertex fit to avoid mode - dependent bias that could result from a correlation between the invariant mass and a global momentum correction procedure .
the error in the position of the @xmath18 decay vertex obtained from the fit is required to be less than 150 @xmath60 m in each of the @xmath13 and @xmath14 directions .
[ sigxy ] illustrates the high efficiency of this cut .
this procedure produces typical @xmath18 decay vertex resolutions of @xmath61 m in both @xmath13 and @xmath14 .
( left plot ) and @xmath62 ( right plot ) returned by the @xmath18 vertex fit after the ip position correction .
the horizontal scale is in cm .
vertical lines indicate the position of the fit quality cut : @xmath63m.,title="fig:",scaledwidth=48.0% ] ( left plot ) and @xmath62 ( right plot ) returned by the @xmath18 vertex fit after the ip position correction .
the horizontal scale is in cm .
vertical lines indicate the position of the fit quality cut : @xmath63m.,title="fig:",scaledwidth=48.0% ] candidates ( left plot ) and @xmath3 candidates ( right plot ) .
the horizontal scale is in gev@xmath64 .
the bin size is 0.5 mev@xmath64 and the solid curves show the fit results.,title="fig:",scaledwidth=48.0% ] candidates ( left plot ) and @xmath3 candidates ( right plot ) .
the horizontal scale is in gev@xmath64 .
the bin size is 0.5 mev@xmath64 and the solid curves show the fit results.,title="fig:",scaledwidth=48.0% ] the invariant mass spectra of the selected @xmath2 and @xmath3 candidates are shown in fig .
[ fig - datamass ] .
the distributions are fit with a sum of two gaussians representing the signal and a first - order polynomial representing the background .
the results of the fit are summarized in table [ tab - datamass ] .
@c@ @c@ @c@ @c@ c@ mode & # events & main gaussian @xmath65 & main gaussian fraction & signal purity + & & ( mev/@xmath66 ) & ( % ) & ( % ) + @xmath49 & 448000 & @xmath67 & 78.4 & 99.1 + @xmath20 & 36480 & @xmath68 & 74.0 & 97.6 +
( left plot ) and @xmath3 ( right plot ) obtained from mc simulation .
the horizontal scale is in fs .
the bin size of the data points is 40 fs and the solid curves show the results of the fit .
( the fit is performed using a bin size of 10 fs . ) , title="fig:",scaledwidth=48.0% ] ( left plot ) and @xmath3 ( right plot ) obtained from mc simulation .
the horizontal scale is in fs .
the bin size of the data points is 40 fs and the solid curves show the results of the fit .
( the fit is performed using a bin size of 10 fs . ) , title="fig:",scaledwidth=48.0% ] the proper time distribution is represented by an exponential function with time constant @xmath69 convolved with the detector vertex resolution .
the properties of the detector resolution function are studied using monte carlo ( mc ) simulation of the signal decay modes .
approximately @xmath70 @xmath2 decays and @xmath71 @xmath3 decays were simulated , corresponding to two times the data sample and three times the data sample , respectively .
[ fig - res ] shows the difference between the generated and reconstructed decay times of the @xmath18 mesons .
the distributions are fit with a sum of five gaussians constrained to a common central value
. the results of the fit are summarized in table [ tab - res ] ; the confidence level ( cl ) is 22.2% . due to similar kinematics ,
the @xmath20 decay mode is fit with the same resolution function as that used for @xmath49 .
however , to account for small differences between @xmath20 and @xmath49 , the width of each gaussian component in the @xmath20 fit is multiplied by a common scale factor @xmath72 .
fitting first the @xmath49 distribution and then the @xmath20 distribution , we obtain @xmath73 .
the cl for the @xmath20 fit is 71.3% .
the common central values of the gaussians are found to be @xmath74 fs for @xmath49 and @xmath75 fs for @xmath20 . in the lifetime difference
fits for both mc and data , the relative fractions of the gaussian components of the resolution function are fixed to the values determined in the above study .
the central value of the @xmath20 resolution function is fixed to that of the @xmath49 resolution function , which is consistent with the results above .
the possible bias due to this assumption is included in the systematic error .
@c@ @c@ @c@ fit parameter & fraction ( % ) & value ( fs ) + @xmath76 & 26.1 & @xmath77 + @xmath78 & 50.4 & @xmath79 + @xmath80 & 19.8 & @xmath81 + @xmath82 & 3.1 & @xmath83 + @xmath84 & 0.6 & @xmath85 +
the @xmath18 lifetime difference is obtained from a simultaneous , binned , maximum likelihood fit to the measured proper lifetime distributions of @xmath3 and @xmath2 decays .
the consistency and quality of the fitting procedure is tested with mc events .
the lifetime distributions of @xmath86 decays were fit with an exponential function convolved with a resolution function of the form described above .
the cl of the fit is 61% .
[ fig - mcfit ] and table [ tab - mcfit ] show the results of the fit and a comparison to the parameters obtained from the resolution function study .
the parameters are seen to agree to within the quoted errors .
the resolution function scale factor @xmath72 determined from the fit is @xmath87 , also in agreement with the result from the resolution function study .
the @xmath18 lifetime obtained is @xmath88 fs , which is consistent with the generated value of 412.6 fs .
the lifetime difference obtained is @xmath89 fs , which is consistent with the generated value of exactly 0 fs .
( left plot ) and @xmath3 ( right plot ) performed on mc - generated events .
the horizontal scale is in fs .
the bin size of the data points is 40 fs and the solid curves show the results of the fit .
( the fit was done using a bin size of 10 fs . ) , title="fig:",scaledwidth=48.0% ] ( left plot ) and @xmath3 ( right plot ) performed on mc - generated events .
the horizontal scale is in fs .
the bin size of the data points is 40 fs and the solid curves show the results of the fit .
( the fit was done using a bin size of 10 fs . ) , title="fig:",scaledwidth=48.0% ] @c@ @c@ @c@ fit parameter & resolution function ( fs ) & lifetime fit ( fs ) + @xmath76 & @xmath77 & @xmath90 + @xmath78 & @xmath79 & @xmath91 + @xmath80 & @xmath81 & @xmath92 + @xmath82 & @xmath83 & @xmath93 + @xmath84 & @xmath85 & @xmath94 + @xmath95 & @xmath96 & @xmath97 + @xmath72 & @xmath98 & @xmath87 +
to fit the lifetime difference in data , one must include small contributions from background processes . to include these
, the proper time distribution of background is obtained from events in the @xmath18 mass sidebands .
the sidebands chosen begin @xmath99 mev@xmath64 from @xmath100 and cover the same total range as the signal region ( @xmath101 ) .
( left plot ) and @xmath3 ( right plot ) .
the horizontal scale is in fs .
the bin size is 40 fs and the solid curves show the results of the fit.,title="fig:",scaledwidth=48.0% ] ( left plot ) and @xmath3 ( right plot ) .
the horizontal scale is in fs .
the bin size is 40 fs and the solid curves show the results of the fit.,title="fig:",scaledwidth=48.0% ] the background proper time distributions are parameterized as the sum of an exponential convolved with a gaussian and a breit - wigner function .
the resulting spectra are shown in fig .
[ fig - bkg ] .
the systematic uncertainties due to the background fitting procedure are evaluated by varying the fitting parameters .
since the background contribution is small compared to the signal , the effect of the background magnitude and shape uncertainty is small ( approximately @xmath102 in @xmath28 ) . it was checked with mc simulation that the small fraction of signal events remaining in the mass sidebands does not affect the @xmath28 measurement above the quoted systematic error on background .
the decay time function obtained for background is included ( with normalization fixed ) in the lifetime fit to the signal region .
( left plot ) and @xmath3 ( right plot ) .
the horizontal scale is in fs .
the bin size is 40 fs and the solid curves show the results of the fit .
the dashed curves show the background component included in the fit.,title="fig:",scaledwidth=48.0% ] ( left plot ) and @xmath3 ( right plot ) .
the horizontal scale is in fs .
the bin size is 40 fs and the solid curves show the results of the fit .
the dashed curves show the background component included in the fit.,title="fig:",scaledwidth=48.0% ] @c@ @c@ @c@ fit parameter & resolution function ( fs ) & lifetime fit ( fs ) + @xmath76 & @xmath77 & @xmath103 + @xmath78 & @xmath79 & @xmath104 + @xmath80 & @xmath81 & @xmath105 + @xmath82 & @xmath83 & @xmath106 + @xmath84 & @xmath85 & @xmath107 + @xmath95 & @xmath96 & @xmath108 + @xmath72 & @xmath98 & @xmath109 + fig . [ fig - datafit ] and table [ tab - datafit ] show the results of the lifetime difference fit to the data .
the cl of the fit is 94% .
the fitted @xmath2 lifetime is @xmath110 fs , and the fitted lifetime difference between @xmath2 and @xmath3 is @xmath111 fs .
together these values give @xmath112% .
the parameters obtained by the fit for the resolution function ( see table [ tab - datafit ] ) differ from those obtained from the mc study by more than the quoted errors .
this is nominally due to vertex detector misalignment , ip mismeasurement , material unaccounted for in the mc simulation , and momentum dependence .
these effects are included when evaluating the overall systematic error on @xmath113 ( table [ tab - ycp - syst ] ) .
@l@ @c@ source & value ( % ) + @xmath114 assumption & 0.15 + @xmath72 difference on data and mc & 0.20 + bin size & 0.10 + @xmath18 mass window & 0.20 + background & 0.10 + selection requirements & 0.15 + total & 0.38 +
the cp - violating parameter in @xmath3 decays is obtained from fits to the data using the charge of the slow pion from @xmath115 or @xmath116 to distinguish @xmath18 decays from @xmath19 decays .
the @xmath18 and @xmath19 subsamples are fit separately using the resolution function parameters obtained from the full statistics fit .
background is treated as described in the previous section .
the same procedure is performed for @xmath2 decays and also for mc events in order to estimate the systematic uncertainty of the fit .
all results are listed in table [ tab - cpv ] .
the systematic uncertainty of the measurement is taken to be @xmath117 .
@c@ @c@ @c@ @c@ @xmath18 decay mode & data / mc & @xmath118 ( fs ) & ( % ) + @xmath119 & mc & @xmath120 & @xmath121 + @xmath122 & mc & @xmath123 & @xmath124 + @xmath119 & data & @xmath125 & @xmath126 + @xmath122 & data & @xmath127 & @xmath128 +
in conclusion , we report a preliminary measurement of the mixing parameter @xmath28 and the cp - violating parameter obtained by measuring lifetime differences in @xmath129 and @xmath130 decays .
the flavor of the @xmath18 or @xmath19 is identified via @xmath131 decays .
the results obtained are @xmath132 and @xmath133 .
we wish to thank the kekb accelerator group for the excellent operation of the kekb accelerator .
we acknowledge support from the ministry of education , culture , sports , science , and technology of japan and the japan society for the promotion of science ; the australian research council and the australian department of education , science and training ; the national science foundation of china under contract no .
10175071 ; the department of science and technology of india ; the bk21 program of the ministry of education of korea and the chep src program of the korea science and engineering foundation ; the polish state committee for scientific research under contract no .
2p03b 01324 ; the ministry of science and technology of the russian federation ; the ministry of education , science and sport of the republic of slovenia ; the national science council and the ministry of education of taiwan ; and the u.s .
department of energy . | we report a preliminary measurement of the mixing parameter @xmath0 and the cp - violating parameter using the decay @xmath1 followed by @xmath2 and @xmath3 .
the results are obtained from a @xmath4 data sample collected near the @xmath5 resonance with the belle detector at the kekb asymmetric energy @xmath6 collider . |
standard quantum computing @xcite is based on encoding , manipulating , and measuring quantum information encoded in the state of a collection of quantum two - level systems - qubits .
spin-@xmath0 is an ideal implementation of a qubit , since it is a natural two - level system , and every pure state of a spin-@xmath0 corresponds to a state of a qubit .
for this reason , spins have been considered as carriers of quantum information in a variety of proposals @xcite . the initial proposal @xcite called for spins in single - electron quantum dots electrically manipulated by the exchange interaction and local time - dependent zeeman fields . a variety of other encoding schemes and manipulation techniques @xcite rely upon encoded qubits . in these schemes ,
the simplicity of qubit states and minimal number of physical carriers of quantum information are traded for less stringent requirements for experimental implementations . on one hand ,
the alternative setups protect from the most common types of errors by decoupling the computational degrees of freedom from the most common sources of noise , and therefore allow for longer gating times . on the other hand , in some alternative setups
the manipulation without fast switching of the local magnetic fields becomes possible .
the optimization in the encoding and manipulation protocols is always guided by the state of the art in the experiments .
recent results suggest that spin qubits can reside in a variety of material hosts with novel properties .
quantum dots in graphene @xcite and carbon nanotubes @xcite are less susceptible to the decoherence due to nuclei and spin - orbit interaction .
spins in nanowires show very strong confinement in two spatial directions , and the gating is comparably simple @xcite . in hole systems ,
the carriers have distinct symmetry properties , and show coupling to the nuclear spins of a novel kind @xcite .
recently , the experiments in silicon @xcite have demonstrated coherent manipulation of spins similar to the one achieved in the gaas - based nanostructures . within these hosts ,
the manipulation techniques that use exchange interaction , spatially inhomogeneous time independent zeeman splitting , and nuclear hyperfine interactions are within reach . despite these developments
, gaas remains a promising route to spin qubits due to the highly advanced experimental techniques developed for this material . here , we study the implementation of the quantum gates on the encoded two - spin singlet - triplet ( st ) qubits @xcite using resources that closely resemble the ones available in the current experimental setups . there , the application of the time - dependent electric fields through the gates fabricated into a structure @xcite are preferred to time - dependent local magnetic fields .
in addition , the nuclear spins @xcite and inhomogeneous magnetic fields @xcite are possible resources for spin control . in the setups based on semiconductors , the electrons or holes in the quantum dots
are influenced by the spin orbit interaction ( soi ) , which can contribute to the control @xcite . in this work
we present a scheme for control of st - qubits which uses switching of the exchange interaction as a primary resource .
we consider the scheme that is optimized for the application of quantum gates in the network of quantum dots .
the construction of the cnot gate uses pulses of the exchange interaction as the only parameter that is time dependent .
the exchange interaction itself is not sufficient for the universal quantum computation over the st - qubits due to its high symmetry .
the additional symmetry breaking is provided by nonuniform , but static magnetic fields .
these fields describe the influence both of magnetic fields , provided by the nearby magnets , and of the coupling to the nuclear spins in the host material via hyperfine interaction . depending on the scheme used for the application of the quantum gates ,
the optimal geometry is either the one in which the magnetic fields point parallel to the axis of symmetry of the soi or perpendicular to it .
one major problem in the realization of two - qubit quantum gates ( in particular , we consider here the cnot gate based on conditional phase gates ) , is the possibility of leakage errors where the spin states defining the logical qubit leave the computational space .
these errors move the state of four spins from the @xmath1-dimensional computational space of two qubits into some other portion of the @xmath2-dimensional hilbert space of four spin-@xmath0 particles .
we consider two ways of addressing this problem .
one scheme possesses the axial symmetry due to the fact that the soi vector and magnetic fields are parallel . for this ` parallel scheme '
we are able to construct a perfect cnot gate , if we are able to control all the available parameters . having in mind 2d architectures
, we also consider the cnot gate between two qubits in the case when the soi vector and magnetic fields are perpendicular to each other . here
we can not prevent the leakage out of the computational space , however , we show that it is suppressed by a ratio between the soi and zeeman energy coming from a strong external magnetic field .
all our constructions assume that the controlled interactions are switched in time by rectangular pulses . any deviations from this form of time
dependence lead to additional corrections and affect the fidelity of the gate .
the paper is organzied as follows . in sec .
[ sec : model ] , we introduce the model for the double dots and effective hamiltonians for field gradients and exchange and spin orbit interactions . in sec .
[ sec : parallel ] , we consider the parallel geometry and derive the cnot gate via the conditional phase gates and swap gates , all based on exchange .
there we also give estimates for the switching times .
the scheme for the perpendicular geometry is then addressed in sec .
[ sec : perp ] , and we conclude in sec .
[ sec : conclusion ] .
parallel geometry : four quantum dots ( yellow discs ) aligned along the x - axis in the presence of an external magnetic field @xmath3 that is applied parallel to the soi vector @xmath4 ( red arrow ) , which must be perpendicular to the line of the dots and which we take to be the @xmath5-direction . at each dot
, there is a local magnetic field @xmath6 ( blue arrows ) , also assumed to be parallel to @xmath3 , but with alternating orientations as indicated .
the direction of @xmath3 defines the spin quantization axis .
the dots are defined electrostatically by metallic gates ( light green structures ) .
each dot contains a spin-1/2 , and the exchange ( @xmath7 ) and the soi - induced ( @xmath8 ) interactions between the spins can be controlled by changing the electrostatic potential between the dots .
the dots 1 and 2 ( from left to right ) define the first st - qubit , and the dots 3 and 4 the second st - qubit .
, width=302 ] we consider in the following singlet - triplet ( st ) qubits that are implemented by two electrons confined to a double quantum dot system @xcite , see figs .
[ fig : setupallz ] and [ fig : setup ] .
such st - qubits have been realized successfully in several labs @xcite , and single and two - qubit operations have also been demonstrated recently @xcite .
there are several schemes for the fundamental cnot gate , which can be divided into two classes , schemes which make use of exchange interaction and schemes which do not , but instead rely on coupling of dipole moments @xcite .
the latter schemes has the disadvantage to be rather slow and also to be affected by charge noise rather strongly . here ,
we focus on exchange - based schemes specifically adapted to quantum dots in iii - v semiconducting materials , that have the soi , such as in gaas or inas quantum dots .
although the soi is typically small compared to the level spacing of the dots , it needs to be taken into account in order to achieve high fidelities in gate operations .
we focus now on two such st - qubits and assume them to be encoded in four quantum dots that are arranged in a row , see figs .
[ fig : setupallz ] and [ fig : setup ] .
the external magnetic field @xmath9 is assumed to give the largest energy scale and determines the spin quantization axis @xmath5 .
the hilbert space of four spins-@xmath0 is spanned by @xmath10 basis states .
the total spin of the system is given by @xmath11 , where @xmath12 is a spin-@xmath0 operator acting on the spin in a dot @xmath13 .
due to axial symmetry , the z - component , @xmath14 , becomes a good quantum number , and the eigenstates corresponding to @xmath15 span a six - dimensional subspace .
we define the computational basis of the two st - qubits in this subspace as @xmath16 denotes the st - qubit space , ` @xmath17 ' and ` @xmath18 ' denote states of the quantum dot spins corresponding to the projection @xmath19 on the @xmath5 axis .
the remaining two states , latexmath:[\ ] ] with @xmath153 , where we assume @xmath101 [ see eq .
( [ eq : t23 ] ) ] to be even . the total gate acting on the qubits as defined by eq .
( [ phase ] ) can be written in the compact form @xmath154 where the pauli matrix @xmath155 acts on the first st - qubit ( formed by dot 1 and 2 ) and @xmath156 on the second one ( formed by dot 3 and 4 ) with @xmath157 .
choosing @xmath158 where @xmath159 is a positive integer , we obtain the @xmath56-gate , @xmath160 both eqs .
( [ eq : t23 ] ) and ( [ t_231 ] ) should be satisfied simultaneously .
for example , if @xmath161 and @xmath162 , we get @xmath163 from this we can estimate the total switching time to perform the @xmath56-gate that is a sum of the switching times at each step [ see eq .
( [ gate_consequence ] ) ] . for the scheme discussed in subsecs .
[ sec_c ] and [ sec_pi ] the slowest part is given by the swap gates @xmath60 and @xmath61 , whose switching time is limited by field gradients ( due to nuclear spins @xcite and/or micromagnets ) , @xmath164 , which corresponds to @xmath165 .
the gate can be faster if the rotations around the @xmath5-axis are performed not on the equator but more closely to the poles .
this allows us to decrease the switching time of the swap gate to @xmath166 ; however , this would require a more precise control over the pulses .
the same trick can be used to decrease the switching time of the conditional phase gate @xmath59 ( compare figs .
[ fig:23_rotation ] and [ fig:23_rotation_alt ] ) .
if field gradients larger than @xmath167 are used , the switching rates will be proportionately larger .
an alternative scheme for the conditional phase gate @xmath59 ( see also fig . [ fig:23_rotation ] ) . instead of the rotation with the pulse defined by eqs .
( [ eq : t23 ] ) and ( [ t_231 ] ) we consider a sequence of three pulses . during the first and the third pulses ( blue arcs ) the state precesses quickly acquiring the @xmath56-phase ,
( [ t_231 ] ) . during the second pulse , @xmath107 , the state precesses around the @xmath5-axis over a shorter path than the one in fig .
[ fig:23_rotation ] . as a result ,
the switching is faster.,width=302 ] using the @xmath56-gate , we construct the controlled phase flip ( cpf ) gate @xmath168 ( see footnote [ 13 ] in @xcite ) as @xmath169 finally , we obtain the cnot gate , @xmath170 by using the cpf gate and performing a basis rotation on qubit 2 ( a single - qubit rotation by @xmath171 about the @xmath172-axis ) , @xmath173 in summary , the full sequence of operations for the cnot gate , @xmath174 , is given by @xmath175 e^{-i\frac{\pi}{4}(\sigma_1^z + \sigma_2^z)}e^{-i \frac{\pi}{4 } \sigma_2^y}.\ ] ] we note again that this result has been derived under the assumption of rectangular pulse shapes .
this is certainly an idealization , and in practice we expect deviations from this shape to cause errors for the gates and to affect the gate fidelity .
the study of this issue , being very important for practical purposes , requires a separate investigation and is beyond the scope of this work .
in the previous section we have discussed the parallel geometry for which we were able to construct a perfect cnot gate under the assumption that we have a complete control over the parameters .
the cnot gate , together with single - qubit gates , allows us to simulate any other quantum gate and its implementation is a crucial step toward the realization of a quantum computer @xcite . in a next step ,
many such elementary gates need to be connected into a large network . in recent years
, the surface code @xcite has emerged as one of the most promising platforms for this goal due to its large threshold of about @xmath176 for fault tolerant error correction @xcite .
this platform requires a two - dimensional ( 2d ) geometry and can be implemented in semiconductors of the type considered here . @xcite .
schematic setup for 2d architecture .
two dots define an st - qubit ( black dotted ellipses ) .
an external magnetic field @xmath3 and local magnetic fields ( blue arrows ) are parallel . in the case of the coupling between two qubits from the same row ( red solid ellipse )
the soi vector @xmath177 is parallel to the magnetic field @xmath3 , corresponding to the parallel geometry ( see fig .
[ fig : setupallz ] ) . in the case of coupling between two qubits from two neighboring rows (
red dashed ellipse ) the soi vector @xmath178 is perpendicular to the magnetic field @xmath3 , corresponding to the perpendicular geometry ( see fig .
[ fig : setup]).,width=226 ] the basic 2d scheme is illustrated in fig .
[ fig:2d ] .
there , we show an array of quantum dots where two neighboring dots in a given row represent one st - qubit .
these quantum dots are embedded in a semiconductor where the soi is of the same type in the entire structure . as a typical example we mention rashba and/or dresselhaus soi that both depend on the momentum of the electron . as a result , the direction of the soi vector @xmath4 is always perpendicular to the line along which two quantum dots are coupled by exchange interaction @xcite .
thus , for the coupling of two such qubits in the _ same _ row , the soi vector @xmath4 is parallel to the external magnetic field @xmath3 , so we can use the scheme designed for the parallel geometry in the previous section . in contrast , if we want to couple two qubits from _ neighboring _ rows , we should also consider a ` perpendicular geometry ' ( see figs .
[ fig : setupallz ] and [ fig:2d ] ) in which the soi vector @xmath4 is perpendicular to the magnetic field @xmath3 .
this perpendicular geometry is characterized by several features . as was mentioned before , the axial symmetry in this case
is broken by the soi , leading to the coupling between computational ( @xmath179 ) and non - computational ( @xmath180 ) space .
if the magnetic field @xmath181 is sufficiently large to split the triplet levels @xmath182 far away from the computational space ( @xmath183 ) , we can neglect this leakage .
we estimate for gaas @xmath184 . at the same time
, the soi does not couple states within the computational space , so for the realization of the phase gate @xmath59 and the swaps gates @xmath60 and @xmath61 we can use the same scheme as in sec .
[ sec : parallel ] only with @xmath185 .
we have studied the implementation of the cnot gate for st - qubits in a model that is appropriate for current experiments @xcite .
the setup consists of an array of quantum dots with controlled growth direction and the relative orientation of the dots .
pairs of neighboring dots build the st - qubits , where the quantization axis is determined by an externally applied magnetic field @xmath3 .
moreover , we introduce an inhomogeneity in magnetic fields , @xmath6 , by local micromagnets or by the hyperfine field .
the resources used for time - dependent control are the exchange interaction @xmath31 , the field gradients @xmath186 , and the soi vector @xmath187 . if the soi vector @xmath4 is parallel to the external ( @xmath3 ) and local magnetic fields ( @xmath6 ) , we are able to construct a perfect scheme for the cnot gate based on the @xmath56-phase gate , @xmath57 , consisting of four basic steps .
two of the steps involve interaction of spins that belong to different qubits , and open the possibility of leakage errors . under condition of total control over system parameters , we show that the leakage can be eliminated . in the other two steps ,
the tuning of exchange interaction enables perfect swap gates even in the presence of field gradients and soi .
motivated by recent results on the surface code , we shortly comment also on the 2d architecture .
here we encounter a situation in which the soi vector @xmath4 and the magnetic fields , @xmath3 and @xmath6 , are perpendicular . in this case , the leakage to the non - computational space with @xmath188 is inevitable .
however , it can be made very small as long as @xmath189 .
depending on the pulsing scheme , the switching times for the conditional phase gate are shown to lie in the range @xmath62-@xmath190 ns for typical gaas parameters .
compared to the experimentally established decoherence times of about @xmath191 @xmath192s @xcite , this switching is sufficiently fast and shows that a cnot gate based on exchange is a promising candidate for experimental realizations .
we acknowledge discussions with andrew doherty , charles marcus , izhar neder , and mark rudner .
this work is supported by iarpa / mqco program , darpa / quest program , the swiss nsf , nccr nanoscience , and nccr qsit . | we propose a scheme for implementing the cnot gate over qubits encoded in a pair of electron spins in a double quantum dot .
the scheme is based on exchange and spin orbit interactions and on local gradients in zeeman fields .
we find that the optimal device geometry for this implementation involves effective magnetic fields that are parallel to the symmetry axis of the spin orbit interaction .
we show that the switching times for the cnot gate can be as fast as a few nanoseconds for realistic parameter values in gaas semiconductors .
guided by recent advances in surface codes , we also consider the perpendicular geometry . in this case ,
leakage errors due to spin orbit interaction occur but can be suppressed in strong magnetic fields . |
behavior of many natural systems in contact with their surroundings can be described within a stochastic picture based on langevin equations .
the basic equation of this type reads @xmath1 where @xmath2 is the deterministic `` force '' representing the internal dynamics of the system and @xmath3 is the `` noise '' describing its interaction with its complex surrounding ( heat bath ) . in many cases
this noise can be considered as white and gaussian , giving rise to the classical langevin approach used in the analysis of brownian motion .
the whiteness of the noise ( lack of temporal correlations ) corresponds to the existence of time - scale separation between the dynamics of a relevant variable of interest @xmath4 and the typical timescale of the noise .
white noise can be thus considered as a standard stochastic process that describes in the simplest fashion the effects of `` fast '' surrounding . on the other hand , the gaussian nature of the noise
is usually guaranteed by assuming the surrounding bath being composed of many practically independent subsystems and by the fact that the interaction of @xmath5 with each of these subsystems is bounded .
the first assumption allows for considering the noise as being a sum of many independent random contributions ( in thermodynamical limit infinitely many ) , which mathematically corresponds to the statement that its probability distribution is infinitely divisible and stable .
the second assumption chooses the gaussian distribution as the only one possessing finite dispersion .
however , the assumption that the perturbations in the system s dynamics due to interactions with bath are described by white gaussian noise is not always appropriate when describing real processes where each of the assumptions concerning the noise ( e.g. its whiteness or its gaussian distribution ) can be violated . in various phenomena in physics , chemistry or biology @xcite
the noise can be still interpreted as white ( i.e. , with stationary , independent increments ) and distributed according to a stable and infinitely divisible law , however , the distribution of the noise variable @xmath6 is registered as following not a gaussian , but rather a more general , lvy probability distribution .
such situations were addressed for example in refs .
the present work discusses some further properties of lvy flights in external potentials with a focus on astonishing aspects of noise - induced bifurcations and explores in more detail features of stationary states in langevin systems under the influence of asymmetric lvy noises .
lvy distributions @xmath7 correspond to a 4-parametrical family of the probability density functions characterized by their fourier - transforms ( characteristic functions of the distributions ) @xmath8 being @xcite @xmath9 \label{eq : charakt}\ ] ] for @xmath10 $ ] and @xmath11 \label{eq : charakt1}\ ] ] for @xmath12 @xcite . here
the parameter @xmath0 ( where @xmath13 $ ] ) is the stability index of the distribution describing ( for @xmath14 ) its asymptotic `` fat '' tail characteristics yielding @xmath15 for large @xmath6 .
the parameter @xmath16 characterizes a scale and @xmath17 $ ] defines a skewness ( asymmetry ) of the distribution , whereas @xmath18 denotes the location parameter . as it is clear from eq .
( [ eq : charakt ] ) , gaussian distribution corresponds to a special case of a lvy law for @xmath19 , with @xmath18 interpreted now as a mean and @xmath16 as the dispersion of the distribution .
however , this special case is somewhat degenerated : the dependence of the distribution on @xmath20 disappears due to the fact that @xmath21 , so that all gaussian distributions are symmetric ! in general cases strongly asymmetric distributions ( up to extreme , one - sided ones ) may appear , see e.g. @xcite .
such realms are discussed much less extensively than the cases of symmetric noises @xcite . in many situations ,
one is interested not in the individual properties of the trajectories @xmath4 but , instead , in the one - point probability distributions defined on ensemble of trajectories : @xmath22 at some given time @xmath23 . for stochastic system described by the langevin equation with an additive white lvy noise forcing , the distribution function @xmath24 of the variable @xmath4
fulfills the associated fractional differential fokker - planck equation .
stationary states ( if they exist ) can be then read from the asymptotic time independent solutions @xmath25 .
such solutions were discussed e.g. in @xcite where analysis of lvy flights in harmonic / superharmonic potentials has been presented . nevertheless , the discussion presented there is far from being complete , since only symmetric lvy distributions have been considered . in the present work
we pay special attention to asymmetric ones .
in this article we focus on stationary states for a particle moving in quadratic , quartic and double well potentials subjected to @xmath0-stable white noises . theoretical descriptions of such systems is based on the langevin equation and/or fokker - planck equation , which in general is of the fractional order @xcite . the model under discussion is presented in section [ sec : model ] .
section [ sec : stationary ] discusses obtained results , which are divided into three subsections regarding results for parabolic and quartic potential ( sections [ sec : parabolic ] and [ sec : quartic ] ) and double well potential model ( section [ sec : doublewell ] ) .
the paper is closed with concluding remarks ( section [ sec : summary ] ) .
additional information , regarding problem of the dimensionality of the langevin equation is included in appendix [ sec : appendix ] .
let us consider a motion of an overdamped particle in a field of a potential force , so that eq .
( [ eq : langevin0 ] ) takes the form of @xmath26 and @xmath3 denotes a lvy stable white noise process @xcite . the value of the stochastic process defined by eq .
( [ eq : langevin ] ) can be calculated as @xcite @xmath27 here , the integral @xmath28 defines a generalized wiener process @xcite that is driven by a lvy stable noise , whose increments are distributed according to a stable density with the index @xmath0 .
the lvy noise is a formal time derivative of the generalized wiener process . for the time step of integration @xmath29 , the increments of the generalized wiener process
are distributed according to the distribution @xmath30 @xcite .
we discuss the overall range of parameters @xmath31 ; \ ; \beta \in [ -1,1]$ ] excluding the case of @xmath12 with @xmath32 , for which the numerical results are unreliable due to well known numerical instabilities @xcite . putting the location parameter of the distribution to zero does not influence the generality of our results : taking location parameter @xmath18 to be nonzero is equivalent to adding a linear term to the potential ( constant drift ) .
sample @xmath0-stable probability densities are presented in fig .
[ fig : density ] .
the langevin equation ( [ eq : langevin ] ) describes evolution of a single realization of the stochastic process @xmath33 .
random numbers distributed according to a canonical form of characteristics functions given by eqs . ( [ eq : charakt])([eq : charakt1 ] ) can be generated using the janicki - weron algorithm @xcite .
more details on numerical scheme of integration of stochastic differential equations with respect to @xmath0-stable noises can be found elsewhere @xcite .
for @xmath34 , equation ( [ eq : langevin ] ) is associated with the following fractional fokker - planck equation ( ffpe ) @xcite @xmath35p(x , t ) \\ & + & \sigma^\alpha\left [ \frac{\partial^\alpha}{\partial |x|^\alpha } p(x , t ) \right . \nonumber \\ & + & \left .
\beta\tan\frac{\pi\alpha}{2}\frac{\partial}{\partial x } \frac{\partial^{\alpha-1}}{\partial |x|^{\alpha-1 } } p(x , t ) \right ] , \nonumber \label{ffpe}\end{aligned}\ ] ] where the fractional ( riesz - weyl ) derivative can be understood in the sense of the fourier transform @xcite @xmath36 the fractional derivatives in ( [ eq : ffpe ] ) originate from the form of the characteristic function , see eqs .
( [ eq : charakt ] ) and ( [ eq : charakt1 ] ) , of lvy stable variables @xcite .
the nonzero asymmetry leads to an additional , asymmetric diffusion term including an even , reflection - invariant riesz - weyl operator and an odd first derivative which changes its sign under the @xmath37 transformation . the overall order of derivatives in the diffusion terms is the same , namely @xmath0 . in the following ,
the value of the location parameter , @xmath18 , is set to zero , which guarantees that for @xmath34 lvy noise present in eq .
( [ eq : langevin ] ) is strictly stable and standard numerical methods of integration of stochastic differential equations with respect to @xmath0-stable noises ( namely the generalization of the euler scheme ) can be used @xcite .
-stable probability density functions ( pdf ) with @xmath38 ( left panel ) and @xmath39 ( right panel ) . for @xmath40 distributions are symmetric , while for @xmath41 they are asymmetric functions .
the support of pdfs for the fully asymmetric cases with @xmath41 and @xmath42 ( right panel ) assumes only negative values for @xmath43 and only positive values for @xmath44 .
note the differences in the positions of the maxima for @xmath45 and @xmath46.,width=302 ] -stable lvy type noises has been performed for a generic double well potential model ( @xmath47 ) , as well as for parabolic ( @xmath48 ) and quartic ( @xmath49 ) potentials . as discussed in the text ,
the confinement of trajectories ( observation of bounded states ) for @xmath50 is possible only if the potential slopes are steeper than for a harmonic case .
, width=302 ] the behavior of a system where a particle is subject to the additive , white , strongly non - gaussian noise could be very different from the behavior in the gaussian regime @xcite . in the gaussian case , any potential well such that @xmath51 for @xmath52 , even the piecewise linear one , is sufficient to produce bounded states , i.e. , the ones with a finite dispersion of the particle s position . on the contrary , for the lvy noises with @xmath50 , the potential which grows faster than quadratically in @xmath5
is required to produce bounded states .
furthermore , in the gaussian case stationary probability distributions for a single - well potential are unimodal , which is no always true for a lvy stable noise with the stability index @xmath50 @xcite .
qualitative and quantitative differences are caused by the fact that stable distributions are heavy - tailed and allow larger noise pulses with a higher probability than the gaussian distribution @xcite .
moreover , stationary probability distributions for the additive lvy noises ( @xmath50 ) , if they exist , are not of the boltzmann - gibbs type @xcite . in the following sections properties of stationary probability distributions for systems perturbed by the general lvy noises are discussed .
the performed simulations corroborate earlier theoretical findings @xcite .
furthermore , the influence of the nonzero asymmetry parameter @xmath20 on the shape of stationary distributions is discussed .
the stationary probability densities @xmath53 can be obtained either by analytically solving eq .
( [ eq : ffpe ] ) ( which is unfortunately possible only for a quite restricted set of special cases ) or , otherwise , numerically .
in such a case , there are two approaches possible : either using the discretization of eq .
( [ eq : ffpe ] ) @xcite or employing a monte - carlo method based on simulation of eq . ( [ eq : langevin ] ) , @xcite . for the gaussian noise
the solutions of the fokker - planck equation can be readily obtained by using shooting methods and discretization techniques @xcite .
for the general lvy case such solutions can be constructed by discretization of eq .
( [ eq : ffpe ] ) , which converts partial differential equation to a discrete markov chain @xcite . however , this approach has a drawback of slow convergence and possible instability for @xmath50 @xcite , and consequently was not used here .
thus , our data are based on monte - carlo simulations ; our method of solution of the langevin equation is based on the slightly modified standard integration scheme for stochastic equations of type ( [ eq : langevin ] ) driven by @xmath0-stable lvy type noises @xcite , see below .
stationary pdfs were extracted from ensembles of , typically , @xmath54 trajectories of a given length @xmath55 .
the value of @xmath56 was chosen by trial and comparison of numerical estimates of @xmath57 for various @xmath56 ( sufficiently long times @xmath56 are requited to let @xmath58 reach stationarity ) . a problem related to
the choice of the simulation time @xmath56 is the choice of the time step of integration @xmath29 .
the simulations have been performed with the time step of integration @xmath59 .
such a choice of @xmath29 guarantees a compromise between accuracy and the computational cost of simulations .
it is also suggested by earlier studies @xcite .
furthermore , @xmath59 makes the @xmath5-domain in which the generalized euler scheme can be used sufficiently large , see section [ sec : quartic ] . for @xmath19 ( and an arbitrary skewness parameter @xmath20 ) , the random force term in the langevin equation ( [ eq : langevin ] ) represents a gaussian white noise @xmath60 and the associated smoluchowski - fokker - planck equation governing evolution of the probability density @xmath24 reads @xmath61 with a stationary solution assuming the standard boltzmann - gibbs form @xmath62 , \label{eq : gibbspdf}\ ] ] of finite mean and variance . in contrast , lvy flights in external potentials exhibit unexpected properties @xcite and their stationary pdfs can be shown to possess finite first and second moments only , if the imposed deterministic forces are derived from steeper than the parabolic potentials , see fig .
[ fig : density ] . as an example , in fig .
[ fig : static ] stationary pdfs for a particle moving in the quartic potential subject to the white gaussian noise ( left panel ) and white cauchy noise ( right panel ) are compared .
numerical results were constructed from the ensemble of final positions reached after the long time @xmath56 obtained from the simulation of the langevin equation ( [ eq : langevin ] ) with @xmath19 , @xmath12 and @xmath63 , cf . section [ sec : quartic ] .
in forthcoming sections we are addressing this point by investigating langevin dynamics driven by asymmetric lvy white noises . ,
left panel ) and superdiffusion ( @xmath12 , right panel ) in a quartic potential @xmath49 . for @xmath19 numerical results
were constructed by discretization techniques and shooting methods @xcite .
simulation parameters : @xmath54 , @xmath59 , @xmath64 and @xmath55.,width=302 ] for @xmath65 , the fractional fokker - planck equation ( [ ffpe ] ) can be rewritten in the fourier space in the form of @xmath66 where @xmath67 is the operator giving the fourier representation of the potential , which can be found in a closed form only in the simplest cases , e.g. for polynomial potentials . in the case of asymmetric distribution
the analogous equation reads @xmath68 \hat{p}(k , t).\ ] ] the choice of the parabolic potential @xmath48 , see fig .
[ fig : density ] , results in its fourier transform @xmath69 . for symmetric @xmath0-stable noises , the corresponding equation for the stationary pdfs is @xcite @xmath70 and its solution in the fourier space reads @xmath71 i.e. , the stationary solution is a symmetric lvy distribution , see eqs .
( [ eq : charakt ] ) and ( [ eq : charakt1 ] ) .
consequently , the variance of the stationary solution is infinite .
therefore , the parabolic potential is not sufficient to produce bounded states for a particle subject to the action of a lvy noise @xcite . for potentials steeper than the parabolic
well , the confinement of superdiffusive trajectories becomes possible but , additionally , the additive lvy white noise could induce bimodality in the stationary pdf . for general @xmath0-stable driving ( @xmath34 ) the stationary solutions obey the equation @xmath72 and its solutions is @xmath73.\ ] ] for the nonzero asymmetry parameter , @xmath20 , like for the symmetric noise , the stationary probability density function is a stable law with the same stability index @xmath0 and the asymmetry parameter @xmath20 and a different scale parameter @xmath74 ( here @xmath75 ) .
the location parameter @xmath18 of the resulting distribution is zero .
the existence of these analytical results allows to use the case of parabolic potentials as a test bench for our simulation algorithms .
thus , for the testing purposes , large samples of long realizations of the stochastic process given by eq .
( [ eq : langevin ] ) were constructed . using these samples the values of the distributions parameters
have been estimated applying special software @xcite . estimated values of distributions parameters
are in good agreement with theoretical values , see tabs .
[ table4][table6 ] .
techniques of estimation of the stable distribution parameters are based on the evaluation of quantiles and characteristic functions , and on maximum likelihood methods @xcite or direct use of time series @xcite .
results obtained by quantile methods and characteristic function estimation seems to be more consistent with theoretical values than results following from maximum likelihood , see tabs .
[ table4][table6 ] .
the largest differences between theoretical and estimated values of parameters are observed for the location parameter @xmath18 . in some situations ,
marked with @xmath76 , the program used @xcite warns about numerical problems in evaluation of the distributions parameters .
to check whether results are influenced by the length of simulation results for @xmath55 and @xmath77 were compared .
estimated values of parameters for both values of @xmath56 are consistent .
therefore , only results for @xmath77 are presented , see tabs . [ table4][table6 ] .
.theoretical and estimated values of stationary pdf parameters .
in every cell numbers in bold indicate theoretical values of parameters , followings rows estimated parameters using quantile evaluation ( 2@xmath78 row ) , characteristic function evaluation ( 3@xmath79 row ) and maximum likelihood method ( 4@xmath80 row ) .
distributions parameters were estimated by use of @xcite .
simulations parameters : @xmath77 , @xmath59 .
samples contain not less than @xmath81 elements .
@xmath76 indicates cases when the software applied warned about some problems in estimation of sample parameters . [
cols="^,^,^,^",options="header " , ] for the quartic potential @xmath49 ( @xmath82 ) and symmetric @xmath0-stable noises , fractional fokker - planck equation in the fourier space has the form @xmath83 the solution of eq .
( [ eq : quartic ] ) is known for @xmath12 @xcite , @xmath84 and the corresponding stationary solution in the real space reads @xmath85 the stationary solutions ( [ eq : cauchystat ] ) of eq .
( [ eq : quartic ] ) have two main properties .
first of all the stationary pdfs are not of the boltzmann type , as typical for the stationary states for system driven by lvy white stable noises with the stability index @xmath50 @xcite . additionally , the stationary probability density function for a quartic cauchy oscillator is bimodal , with extremes located at @xmath86 @xcite .
[ fig : static ] presents stationary pdfs obtained by the simulation of the langevin equation ( [ eq : langevin ] ) along with theoretical lines for the gaussian ( left panel ) and cauchy ( right panel ) quartic oscillators .
moreover , the parabolic addition to the quartic potential @xmath87 @xmath88 , can diminish or even destroy the bimodality of the stationary pdf @xcite .
finally , for the general @xmath0-stable driving and the quartic potential the stationary density fulfills @xmath89 quartic potentials pose some additional difficulties to monte - carlo simulations making it necessary to slightly modify the standard techniques of integration of stochastic differential equations driven by @xmath0-stable lvy type noise @xcite .
due to heavy tails of stable distribution large random pulses are much more likely to occur than in the gaussian distribution leading to very long jumps from time to time putting a particle to the position where the force acting on it is very large . in this case approximating
the deterministic drift by @xmath90 is too inaccurate whatever small time step is chosen .
it can result in switch of the particle to the other side of the origin in such a way that new particle s position is more distant from the origin than the initial one leading to numerical instability and escape of the particle to infinity @xcite .
of course , this effect is weaker when smaller integration step is used .
however , taking smaller steps was proven not to give an effective solution to the problem .
our approach to it is based on separating noise and deterministic drift and integrating the last one analytically , by solving the differential equation @xmath91 to obtain @xmath92 for a given initial condition @xmath4 .
such a step ( involving the solution of an algebraic equation ) is more time - consuming than the euler integration step and is absolutely superfluous for small and moderate @xmath5 .
therefore , exact integration of the deterministic part is performed only for large @xmath5 , @xmath93 , while the noisy part is always integrated in the standard euler way . in the simulations we took @xmath94 as motivated by analytical estimates and by numerical tests . for testing purposes constructed numerical results for the cauchy noise ( @xmath12 ) were compared with the known analytical solution , see right panel of fig .
[ fig : static ] , leading to the excellent level of agreement . for @xmath47 .
@xmath95 represents threshold value of @xmath5 such that for @xmath93 deterministic part of eq .
( [ eq : langevin ] ) is integrated analytically .
the case @xmath96 corresponds simply to the standard euler scheme . for more details
see the text.,width=302 ] -stable driving with @xmath97 and various @xmath20 .
simulation parameters : @xmath54 , @xmath59 , @xmath64 , @xmath94 and @xmath55 .
stationary states for totally skewed noise , i.e. , @xmath98 are one - sided.,width=302 ] for @xmath99.,width=302 ] for @xmath100.,width=302 ] the results for the double - well potential model , see fig .
[ fig : density ] , were constructed by the numerical method described in the previous section ( section [ sec : quartic ] ) .
furthermore , we compared the influence of decreasing time step of integration and @xmath95 .
the results of comparison of both methods of reduction of number of numerical escapes are summarized in fig .
[ fig : escape ] where the influence of the time step of integration @xmath29 , duration of the simulation @xmath56 on the portion of valid ( non escaped ) trajectories are compared . stationary solutions shown in figs . [ fig : alpha05][fig : alpha18 ] are obtained for the generic double well potential model , i.e. , @xmath47 . in the simulation
the whole allowed range of @xmath0 and @xmath20 was examined , in figures only a limited choice of representative values of noise parameters is presented , namely the same as ones used in tabs .
[ table4][table6 ] . for @xmath45 with @xmath101
stable distributions are one - sided , therefore , stationary solutions for @xmath97 with @xmath98 are different from zero only on the one side of the origin . for @xmath102 stable distributions
takes all real values as manifested by nonzero probability for all @xmath5 , see fig .
[ fig : alpha05 ] .
furthermore , the symmetry of noise and the potential is reflected in the symmetry of stationary densities , i.e. , solutions for @xmath103 can be constructed by the reflection of the solutions for @xmath20 , see figs .
[ fig : alpha05][fig : alpha18 ] . for @xmath104 with any @xmath0 stationary densities are bimodal and symmetric along @xmath105 . for @xmath46 ,
the support of stable densities as well as the stationary distributions is whole real line .
consequently , even extreme values of the asymmetry parameter @xmath98 are not sufficient to switch the probability mass to the one side of the origin , see figs .
[ fig : alpha11][fig : alpha18 ] .
furthermore , it is well documented in the lower panel of fig .
[ fig : splitting ] where @xmath106 is presented .
theoretical considerations as well as the probability of being in the left state , @xmath107 , indicate that stationary densities can be one - sided only for some totally skewed @xmath0-stable noises with small @xmath0 , i.e. , @xmath98 with @xmath108 . for @xmath46 with @xmath98 two maxima of stationary pdfs
are visible . in upper panel of fig .
[ fig : splitting ] locations of the median value , which may be considered as the next measure of asymmetry of stationary probability densities , are presented .
very small values of @xmath0 ( @xmath109 ) pose special difficulties for simulations .
our simulation of eq .
( [ eq : langevin ] ) starts with the initial condition @xmath110 .
the initial transient peak of the probability density at this value is rather persistent for small @xmath0 , so that the simulation time has to be long .
on the other hand in this case simulations are prone to escape of trajectories to `` infinity '' due to too strong noise pulses and require very small @xmath29 , so that the overall quality of such results is not very good .
therefore , the results for @xmath111 are not presented here . ) of the stationary probability density @xmath53 ( upper panel ) and asymmetry of stationary distributions
measured as the fraction of the probability mass on the left hand side of the origin , i.e. , @xmath106 ( lower panel ) . for symmetric noises resulting stationary densities
are symmetric along @xmath105 and consequently @xmath112 and @xmath113 .
different symbols correspond to different values of the stability index @xmath0 : ` @xmath114 ' @xmath97 , ` @xmath115 ' @xmath99 and ` @xmath116 ' @xmath100.,width=302 ] another method to present the results for stationary distributions is the use of the effective potential . in general , the same stationary probability densities that are recorded for the double well system driven by @xmath0-stable noise , see eq .
( [ eq : langevin ] ) , can be observed in the gaussian regime in the effective potential @xmath117 .
sample effective potentials corresponding to stationary states for @xmath99 from fig .
[ fig : alpha11 ] are depicted in fig .
[ fig : alpha11_eff ] . the same stationary solutions can be observed for motion in a simple potential , like double well potentials , and @xmath0-stable stochastic driving or for potentials of the complicated form , see fig .
[ fig : alpha11_eff ] , and standard white gaussian driving .
despite the fact that stationary states are the same , other characteristics of these two processes are different . corresponding to stationary densities from fig .
[ fig : alpha11 ] . the effective potential @xmath118 together with the white gaussian noise results in stationary densities of the model ( [ eq : langevin ] ) which are presented in fig .
[ fig : alpha11].,width=302 ]
in the present work we investigated the form of stationary probability densities of a position of a particle subject to a deterministic potential force and to a lvy noise , paying special attention to the case of asymmetric stable noises . stationary density functions for system driven by @xmath0-stable lvy noises ( @xmath119 ) , if they exist , are not of the boltzmann - gibbs form . for parabolic potential the stationary density functions are @xmath0-stable laws with the same stability index @xmath0 and asymmetry parameter @xmath20 as ones of the noise .
the only difference is the scale parameter of the resulting distribution .
therefore , this case can serve as a benchmark for our simulation algorithms . by use of the special software @xcite for estimation of stable law parameters , the parameters of stationary densities
have been evaluated leading to a very good level of agreement between theoretical and estimated values of parameters .
natural consequence of the lvy type of the stationary densities for parabolic potential is divergence of the variance of the particle position . to produce bounded states , i.e. , states with finite variance of a position , potentials steeper than quadratic are necessary
thus , for the quartic potential variance of stationary densities is finite .
furthermore , for white cauchy noise analytical solutions to the stationary problem is known . here again , numerical results fully agree with earlier @xcite theoretical findings .
in the case of gaussian noise ( @xmath19 ) the symmetry of stationary density ( being the boltzmann - gibbs equilibrium distribution ) reflects symmetries of the underlying potential : the asymmetric densities correspond to asymmetric potentials , i.e. , to deterministic forces which break the symmetry of the initial problem .
the gaussian noise itself is always symmetric . for systems driven by lvy noises
( @xmath119 ) , an asymmetric stable noise together with symmetric static potential is sufficient to produce asymmetric stationary densities . in this situation
the asymmetry of stationary states originates from the asymmetry of the stochastic driving and can be controlled by changing the parameters of the noise .
our main studies have been performed for the generic symmetric double well potential model . here
the asymmetry of the stationary distribution ( as measured by the probability of being in the left / right state or location of the median ) was investigated as a function of the parameters of the noise .
the asymmetry of stationary state decreases with increasing @xmath0 , see fig .
[ fig : splitting ] . finally for @xmath19 with any value of the asymmetry parameter @xmath20 gaussian scenario is recovered and stationary density is fully symmetric .
we also checked whether stable asymmetric noise can produce unimodal stationary pdfs in the double well potential .
such situation can indeed be observed for fully asymmetric noises ( @xmath101 ) with @xmath108 , e.g. for the lvy - smirnoff noise ( @xmath97 , @xmath120 ) .
we have investigated the dynamic stochastic process modeled by the ( overdamped ) langevin equation of the form @xmath121 where : @xmath5 is a position of the particle , @xmath122 stands for a friction coefficient , @xmath123 is particle s mass , @xmath124 ( @xmath16 ) represents strength of the noise and @xmath3 is lvy stable white noise characterized by the stability index @xmath0 ( @xmath13 $ ] ) and the asymmetry parameter @xmath20 ( @xmath17 $ ] ) .
the force acting on a particle is determined by the external potential , @xmath125 .
corresponding units in eq .
( [ eq : langlwn ] ) are : @xmath126=[\mbox{length}]$ ] , @xmath127=1/[t]$ ] , @xmath128=[v'(x)]=[m]\times[\mbox{length}]/[t]^2=[\mbox{force}]$ ] , @xmath129=[\mbox{force}]\times[\mbox{length}]=[\mbox{energy}]$ ] , @xmath130=[\mbox{length}]^\alpha/[t]$ ] ( @xmath131=[\mbox{length}]/[t]^{1/\alpha}$ ] ) and @xmath132=1/[t]^{1 - 1/\alpha}$ ] .
stability index @xmath0 and asymmetry parameter @xmath20 are dimensionless . in the asymptotic limit of @xmath19
the lvy white noise is equivalent to the gaussian white noise and it has standards units , i.e. @xmath133=[\xi(t)]=1/\sqrt{[t]}$ ] . by the set of transformation : @xmath134 and @xmath135 .
( [ eq : langlwn ] ) can be transformed to the dimensionless form @xmath136 which is of the same type like eq .
( [ eq : langevin0 ] ) because the ( rescaled ) noise intensity can be incorporated to the distribution of the particle s position increments
. an alternative way of getting dimensionless form of eq .
( [ eq : langlwn ] ) can be found in a recent work @xcite . for the single - well potential @xmath137 @xmath138 ^{\frac{1}{n-2+\alpha}}.\ ] ] thus @xmath139 and @xmath140 .
therefore , for the single minima potential , the only relevant parameters are @xmath0 ( stability index ) and @xmath20 ( asymmetry parameter ) . for the generic double - well potential @xmath141 @xmath142 thus @xmath143 and @xmath144 . in consequence , for the double - well potential , the only relevant parameters in a dimensionless form of eq
( [ eq : langlwn ] ) are the lvy noise parameters @xmath0 and @xmath20 , and a ( rescaled ) noise strength . from the above analysis
, it is clear that the choice of @xmath145 , @xmath146 introduce scales to the system which are directly related to its dynamical parameters .
the research has been supported by the marie curie tok cocos grant ( 6th eu framework program under contract no .
mtkd - ct-2004 - 517186 ) .
computer simulations have been performed at the academic computer center cyfronet agh , krakw .
additionally , bd acknowledges the support from the foundation for polish science and the hospitality of the humboldt university of berlin and the niels bohr institute ( copenhagen ) .
the support by dfg within sfb 555 is also acknowledged . , _ stochastic resonant effects in systems driven by stable noises _ , phd thesis , faculty of physics , astronomy and applied computer sciences , marian smoluchowski institute of physics , jagellonian university , krakw , poland , 2005 . | properties of systems driven by white non - gaussian noises can be very different from these of systems driven by the white gaussian noise .
we investigate stationary probability densities for systems driven by @xmath0-stable lvy type noises , which provide natural extension to the gaussian noise having however a new property mainly a possibility of being asymmetric .
stationary probability densities are examined for a particle moving in parabolic , quartic and in generic double well potential models subjected to the action of @xmath0-stable noises .
relevant solutions are constructed by methods of stochastic dynamics . in situations
where analytical results are known they are compared with numerical results .
furthermore , the problem of estimation of the parameters of stationary densities is investigated . |
optical systems exhibiting comblike resonances have a wide range of scientific and technological applications , especially in spectroscopy and frequency metrology @xcite .
comblike profiles , which consist of ultrasharp resonance peaks , can be achieved _
e.g. _ in the electromagnetic ( em ) scattering by plasmonic nanoparticles or via morphology - dependent resonances in nonmetallic scatterers . indeed ,
plasmonic nanoparticles support localized surface plasmon resonances that are strongly dependent on their geometrical and material parameters , allowing for engineering the scattering response @xcite .
in addition , scatterers with a high degree of symmetry , such as spheres , spheroids and cylinders , may support morphology - dependent resonances , which are related to constructive interferences confined inside the particle by almost total internal reflections @xcite .
recently , a great deal of attention has been devoted to fano resonances in plasmonic nanostrutures due to its sensitivity to both geometry and local environment changes @xcite .
the interference between a broad bright resonance and a narrow dark resonance modes , supported by plasmonic nanostrutures , gives rise to the fano resonance , which has a characteristic , asymmetric lineshape with a narrow bandwidth @xcite . by properly choosing the materials and/or the system design , it is possible to generate _
e.g. _ multiple fano , comblike scattering resonances in plasmonic nanoparticles @xcite .
the design of these fano - comb nanoparticles may enable several applications at different spectral ranges , such as improving the resolution in comb - spectroscopy techniques @xcite , and optical tagging @xcite .
one possible way to achieve comblike resonances in em scattering is to consider multilayered core - shell nanoparticles @xcite .
the main idea is to combine localized plasmon resonances and the scattering cancellation technique for a single core - shell scatterer @xcite , whose core is dielectric and the coating is composed of multilayered semiconductor materials with a gradient of doping level .
an alternative approach to obtain fano - comb resonances is to consider nonconcentric , single - layered cylindrical nanostructures @xcite , whose fabrication is considerably simpler than multilayered nanoparticles . in the present paper ,
we propose an alternative approach , based on magnetic materials , for generating comblike scattering response in concentric , single - layered core - shell nanoparticles , which can exhibit fano resonances @xcite . instead of inducing fano - comb resonances by the geometrical symmetry breaking in single - layered core - shell nanostructures @xcite , we suggest the use of high permeability covers @xcite in center - symmetric scatterers .
indeed , in the long - wavelength limit , the presence of a ferromagnetic layer breaks the isotropy of the rayleigh scattering , and allow us to obtain multiple fano , comblike scattering resonances composed of both electric and magnetic dipole resonances in the extinction cross - sections @xcite .
these fano - like resonances , known as `` unconventional '' fano resonances @xcite , are different from the conventional fano ones @xcite , since they are polarization - independent and arise from the interference between em modes with the same multipole moment .
we explore these properties to derive analytical conditions for the occurrence of comblike resonances in coated spheres composed of a dielectric core and a high magnetic permeability shell . as the condition for the occurrence of multiple fano resonances depends explicitly on the value of the magnetic permeability of the shell
, we argue that both the position and profile of the comblike resonances could be externally tuned .
this tuning could be achieved by exploiting _
e.g. _ the properties of engineered magnetic materials , where the magnetic permeability can be modified by applying an external magnetic field and/or by varying the temperature .
this paper is organized as follows . in sec .
[ theory ] , we briefly review the lorenz - mie theory for center - symmetric coated spheres @xcite . by applying the long - wavelength approximation , we obtain in sec .
[ analytical ] an analytical condition to determine the position of the multiple fano , comblike scattering resonances as a function of the shell parameters . in sec .
[ numerical ] , we go beyond the long - wavelength limit and consider a system composed of a dielectric nanosphere coated with a magnetic thin shell whose permittivity is provided by a realistic , lossy drude model .
finally , in sec . [ conclusion ] , we summarize our main results and conclude .
here we present the main results of the analytic solution of em scattering by coated spheres , kwown as the lorenz - mie theory @xcite , that are extensively used throughout this paper .
let us assume for the incident em wave the time harmonic dependence @xmath0 , with @xmath1 being the angular frequency .
the scatterer is a center - symmetric , core - shell sphere , with inner radius @xmath2 and outer radius @xmath3 , composed of linear , spatially homogeneous , and isotropic magneto - dielectric materials .
the electric permittivities @xmath4 and magnetic permeabilities ( @xmath5 ) of core ( @xmath6 ) , shell ( @xmath7 ) and embedding medium @xmath8 are @xmath9 , @xmath10 and @xmath11 , respectively .
the scatterer geometry is depicted in fig .
[ fig1 ] . and radius @xmath2 , whereas the shell has @xmath10 and radius @xmath3 .
the surrounding medium is the vacuum @xmath11 . ] for the sake of simplicity , the medium @xmath11 is considered to be the vacuum and the relative parameters are @xmath12 and @xmath13 , with @xmath14 for the core and @xmath15 for the shell .
the scattering efficiency ( which is the respective cross - section in units of @xmath16 ) is @xmath17 where @xmath18 is the size parameter of the outer sphere ( @xmath19 being the incident wave number ) @xcite .
the extinction and absorption efficiencies are , respectively , @xmath20 and @xmath21 . here , the lorenz - mie scattering coefficients @xmath22 ( electric ) and @xmath23 ( magnetic ) are @xcite : @xmath24 where one defines the auxiliary functions @xcite @xmath25}{\widetilde{m}_2d_n(m_1x)\chi_n(m_2x)-\widetilde{m}_1\chi_n'(m_2x)}\ , \label{an}\\ b_n&=\frac{\psi_n(m_2x)\left[\widetilde{m}_2d_n(m_2x)-\widetilde{m}_1d_n(m_1x)\right]}{\widetilde{m}_2\chi_n'(m_2x)-\widetilde{m}_1d_n(m_1x)\chi_n(m_2x ) } , \label{bn}\end{aligned}\ ] ] with @xmath26 being the size parameter of the inner sphere and @xmath27/{\rm d}\rho$ ] . the functions @xmath28 , @xmath29 and @xmath30 are the riccati - bessel , riccati - neumann and riccati - hankel functions , respectively , with @xmath31 and @xmath32 being the spherical bessel and neumann functions @xcite .
the relative refractive index is @xmath33 and @xmath34 , with @xmath35 @xcite .
notice that @xmath36 is the inverse of the usual definition of relative impedance , @xmath37 , so that @xmath38 for nonmagnetic materials @xmath39 @xcite .
equations ( [ qsca])([bn ] ) are the exact expressions for core - shell spherical scatterers with arbitrary geometrical and material parameters . in the following ,
we consider approximations for both core and shell parameters regarding the incident wavelength .
we proceed our analysis by taking into account the long - wavelength regime @xmath40 . within this approximation
, the arguments @xmath41 ( core ) and @xmath42 ( shell ) of spherical bessel and neumann functions should not necessarily be small .
following tribelsky _
et al . _
@xcite , one can write @xmath43 $ ] and @xmath44 $ ] . for coated spheres ,
we calculate the new auxiliary functions in the small - particle limit ; they read @xmath45\ , \label{fn - te}\\ \mathcal{g}_n^{(a)}&\approx\frac{(2n-1)!!}{y^n}\left[\widetilde{d}_n+\frac{\widetilde{m}_2n}{y}\right]\ , \label{gn - te}\\ \mathcal{f}_n^{(b)}&\approx\frac{y^n}{(2n+1)!!}\left[\widetilde{m}_2y\widetilde{g}_n-(n+1)\right]\nonumber\\ & \ -\frac{y^{n+2}}{2(2n+3)!!}\left[\widetilde{m}_2y\widetilde{g}_n-(n+3)\right]\ , \label{fn - tm}\\ \mathcal{g}_n^{(b)}&\approx-\frac{{(2n-1)!!}}{y^n}\left[\widetilde{m}_2\widetilde{g}_n+\frac{n}{y}\right]\ .\label{gn - tm}\end{aligned}\ ] ] we find that there are electric and magnetic scattering resonances ( _ i.e. _ , constructive interferences ) in @xmath46 and @xmath47 when @xmath48 ( @xmath49 ) , respectively .
similarly , destructive interferences occur whenever @xmath50 ( @xmath51 ) .
these destructive interferences lead to very small scattering cross - sections , and characterize an em `` cloaking '' mode .
note that the relevant variables to analyze electric and magnetic resonances are indeed @xmath52 and @xmath53 , respectively @xcite . for nonmagnetic particles @xmath54 the leading term in eq .
( [ fn - tm ] ) vanishes , so that it requires higher accuracy relative to other approximations @xcite . in particular , the expressions presented in ref .
@xcite for a homogeneous sphere are retrieved as one performs the following mapping : @xmath55 [ @xmath56 , @xmath57 or @xmath58 or @xmath59 [ @xmath60 .
in addition to the small - particle limit ( @xmath62 ) , if @xmath63 ( _ i.e. _ , large permittivity and/or large permeability shell materials ) , one has @xmath64 and @xmath65 @xcite . to simplify our analysis ,
let us define the `` phase shifts '' @xmath66 where @xmath67 and @xmath68 are defined in eqs .
( [ an ] ) and ( [ bn ] ) , respectively .
the ferromagnetic limit ( @xmath61 ) is obtained from eqs .
( [ fn - te])([gn - tm ] ) by replacing @xmath69 with @xmath70 $ ] and @xmath71 with @xmath72 $ ] , and redefining the auxiliary functions .
in particular , the dipole approximation ( @xmath73 ) for the ferromagnetic small - particle limit leads to : @xmath74\right\}\ , \label{f1-te}\\ \mathcal{g}_1^{(a)}&\approx\frac{1}{y}\left\{\frac{\widetilde{m}_2}{y}-\tan\left[m_2y+\delta_1^{(a)}\right]\right\}\ , \label{g1-te}\\ \mathcal{f}_1^{(b)}&\approx-\frac{y}{3}\left\{\widetilde{m}_2y+2\cot\left[m_2y+\delta_1^{(b)}\right]\right\}\ , \label{f1-tm}\\ \mathcal{g}_1^{(b)}&\approx\frac{{1}}{y}\left\{\widetilde{m}_2-\frac{1}{y}\cot\left[m_2y+\delta_1^{(b)}\right]\right\}\ .\label{g1-tm}\end{aligned}\ ] ] in eqs .
( [ f1-te])([g1-tm ] ) , the dependence on the core parameters @xmath9 and @xmath26 is encoded in @xmath75 . regarding the shell parameters @xmath10 , we consider two reciprocal configurations that fulfill the limiting case of a shell with high refractive index ( @xmath76 ) , namely : @xmath77 large permeability shells with small permittivity ( @xmath78 ) and @xmath79 large permittivity shells with small permeability ( @xmath80 ) . by symmetry , one case can readily be obtained from the other by replacing @xmath81 with @xmath5 ; for our purposes , we focus on case @xmath77 , _
i.e. _ , high permeability materials with small permittivities such that @xmath82 .
we recall that vanishing shell parameters facilitate the scattering cancellation ( em cloaking ) .
this mechanism is based on the cancellation of the em fields due to a local negative polarizability coefficient .
the electric dipole ( @xmath73 ) is associated with the local polarization vector @xmath83\mathbf{e}(\mathbf{r},\omega)$ ] , where one considers the local fields and parameters inside the scatterer .
thereby , in case @xmath77 , one may locally induce a polarization vector out of phase to the electric field , allowing for the partial cancellation of the scattering efficiency ( @xmath84 ) .
in addition , large values of the magnetic permeability in small - particle limit are responsible for breaking the isotropy of the rayleigh scattering , favoring backscattering @xcite and em resonances inside the particle @xcite .
consequently , one could in principle engineer suitable properties to tune both cloaking and strong scattering responses in magnetic core - shell spheres .
in case @xmath77 , spherical shell materials with @xmath85 lead to high refractive ( @xmath86 ) and low impedance ( @xmath87 ) indices , since low permittivity is considered .
therefore , from eqs .
( [ f1-te ] ) and ( [ f1-tm ] ) , the electric and magnetic cloaking in the ferromagnetic limit occur for @xmath88\approx 0 $ ] and @xmath89\approx 0 $ ] , respectively .
conversely , for case @xmath79 , @xmath90 leads to @xmath86 and @xmath91 , which readily provide the electric and magnetic cloaking conditions : @xmath92\approx0 $ ] and @xmath93\approx0 $ ] .
these are the analytical conditions for em cloaking to occur in both cases @xmath77 and @xmath79 .
notice that , for this set of cloaking conditions , one can readily obtain case @xmath77 from case @xmath79 by replacing @xmath67 with @xmath68 , and vice versa . in the following , we present a special configuration where multiple electric and magnetic fano resonances in the scattering cross - section are achieved .
let us consider now a non - dispersive dielectric core @xmath9 coated with a dispersive magnetic shell @xmath96 $ ] , with @xmath97 , both of them lossless media .
fixing the size parameters @xmath26 and @xmath18 , the cloaking condition at resonance @xmath98 [ @xmath99 takes place for a certain refractive index @xmath100 and @xmath101 : @xmath102=\cot\left[\overline{m}_2y+\overline{\delta}_1^{(b)}\right]= 0\ .\end{aligned}\ ] ] since the arguments of these trigonometric functions are very large , it is convenient to consider a small frequency variation @xmath103 in the vicinity of @xmath104 , leading to a corresponding variation @xmath105 ( which is not necessarily small ) in the refractive index : @xmath106 . to simplify our analysis ,
consider that only @xmath107 depends on the frequency in this range @xmath103 , so that the impedance index remains approximately unchanged : @xmath108 .
in addition to the small - particle @xmath109 and ferromagnetic @xmath110 limits for the shell , we impose that the dielectric core satisfies the rayleigh limit : @xmath94 .
this provides @xmath111/[\varepsilon_{1{\rm r}}x-2\widetilde{m}_2\tan(m_2x)]$ ] , where we have used @xmath112 and @xmath113 is readily obtained from @xmath114 by replacing ( @xmath115 ) with @xmath116 .
consequently , @xmath117 . from the latter approximation ,
we can finally write the analytical expressions in terms of the aspect ratio @xmath118 , which is a geometric parameter that plays a crucial role on the em cloaking . imposing @xmath119 , we obtain @xmath120\\ & \approx\tan\left[\left(\overline{m}_2+\delta m\right)y + \overline{\delta}_1^{(a)}-\left(\delta m\right)x\right]\\ & = \tan\left[\overline{m}_2y+\overline{\delta}_1^{(a)}+y(1-s)\delta m\right]\\ & = \frac{\tan\left[\overline{m}_2y+\overline{\delta}_1^{(a)}\right]+\tan\left[y(1-s)\delta m\right ] } { 1-\tan\left[\overline{m}_2y+\overline{\delta}_1^{(a)}\right]\tan\left[y(1-s)\delta m\right]}\\ & = \tan\left[y(1-s)\delta m\right]\\ & \approx y(1-s)\delta m\ ; \end{aligned}\ ] ] similarly , @xmath121=-\tan[y(1-s)\delta
m]\approx -y(1-s)\delta m$ ] , since we can choose @xmath122 even if @xmath123 ( small shell thickness ) .
hence , by assuming a shell with high refractive index so that @xmath61 , the condition @xmath119 guarantees the validity of our analysis .
notice , however , if one imposes @xmath124 , this analysis is valid for every @xmath125 . in particular , for @xmath126 [ homogeneous sphere @xmath127 , we retrieve the results of ref .
@xcite . substituting these approximations into eqs .
( [ f1-te])([g1-tm ] ) and redefining the auxiliary functions , we obtain : @xmath128\ , \label{f1-a - ap}\\ \mathcal{g}_1^{(a)}&\approx\frac{\widetilde{m}_2}{y^2(1-s)}-\delta m\ , \label{g1-a - ap}\\ \mathcal{f}_1^{(b)}&\approx-\frac{y^3}{3}\left[\frac{\widetilde{m}_2}{1-s}-2\delta m\right]\ , \label{f1-b - ap}\\ \mathcal{g}_1^{(b)}&\approx\frac{\widetilde{m}_2}{1-s}+\delta m \ .\label{g1-b - ap}\end{aligned}\ ] ] from eqs .
( [ f1-a - ap ] ) and ( [ g1-a - ap ] ) , the constructive and destructive interferences in @xmath129 occur for @xmath130 $ ] and @xmath131 $ ] , respectively .
analogously , constructive and destructive interferences in @xmath132 occur for @xmath133 and @xmath134 $ ] , respectively .
in particular , we have analytically demonstrated that both @xmath129 and @xmath132 have a fano lineshape @xcite : @xmath135 , where @xmath136/[|\gamma|\sqrt{\alpha(3-\alpha)}]$ ] , @xmath137 , @xmath138 $ ] and @xmath139 .
the @xmath132 profile is obtained from @xmath129 by replacing @xmath140 with @xmath141 .
assuming that only @xmath142 depends on @xmath1 and @xmath85 , we expect a fano lineshape for both @xmath129 and @xmath132 as a function of @xmath143 .
these asymmetric dipole resonances in total cross - sections are referred to as unconventional fano resonances @xcite , which usually occur beyond the applicability of the rayleigh approximation and are intrinsically different from the conventional fano resonance that arises from the interference between different multipole orders in a specific scattering direction @xcite . here
unconventional fano resonances result from the interference of different em modes excited inside the scatterer due to its high refractive index ( hence , beyond the rayleigh limit ) with the same multipole moment @xmath73 .
moreover , note that the asymmetry parameter @xmath144 has different signs in @xmath129 and @xmath132 profiles as a function of frequency [ see the terms @xmath145 in eq .
( [ f1-a - ap])([g1-b - ap ] ) ] .
this provides a configuration in which the destructive interferences of the electric and magnetic dipoles coincide without overlapping the scattering resonances .
therefore , we have `` comblike '' electric and magnetic dipole resonances for @xmath94 , @xmath61 , and @xmath62 , with the additional condition @xmath119 if @xmath123 . the analysis for the case @xmath90 , with @xmath146 , is completely analogous .
we have demonstrated so far that it is possible to obtain , in the long - wavelength regime ( @xmath147 ) , multiple fano , comblike resonances in the scattering cross - section of concentric core - shell spheres .
the basic conditions consist of a core in the rayleigh limit @xmath148 , @xmath149 ) and a concentric shell with both large refractive index @xmath150 and aspect ratio near to unity ( @xmath151 ) . with a proper choice of parameters ,
one can obtain electric and magnetic fano - like resonances with a scattering minimum at the same frequency ( antiresonance or fano - dip ) . as we shall demonstrate , by making these approximations the fano - dips of the electric and magnetic dipole resonances overlap without any additional assumption , giving rise to the unconventional fano resonances .
however , the approximation @xmath152 ( small shell thickness ) is now mandatory .
to identify the positions in frequency where these comblike resonances occur , we set @xmath153 for @xmath154 ( ferromagnetic limit ) , leading to @xmath155 and @xmath156 . from eqs .
( [ f1-te ] ) and ( [ f1-tm ] ) , the cloaking condition becomes @xmath157=0 $ ] , _ i.e. _ , @xmath158 , with @xmath159 note that we impose @xmath119 because @xmath61 .
in particular , this cloaking condition is readily obtained from eq .
( [ f1-te ] ) , but not from eq .
( [ f1-tm ] ) . by showing that @xmath160=\pi/2-\cot^{-1}[\cot(m_2x)]=\pi/2-m_2x$ ] and using the trigonometric identity @xmath161
, one can easily demonstrate that eq .
( [ f1-tm ] ) also provides @xmath157=0 $ ] .
explicitly , we have : @xmath162&\approx\cot\left[m_2y+\pi/2-m_2x\right]\\ & = \cot\left[m_2y(1-s)+\pi/2\right]\\ & = -\tan\left[m_2y(1-s)\right]\ .\end{aligned}\ ] ] the condition @xmath163 ( _ i.e. _ , @xmath164 since @xmath61 ) leads to a homogeneous dielectric sphere @xmath9 , and , thereby , there is no cloaking .
these integer numbers @xmath165 , which discretize em cloaking for a certain set of parameters , can roughly be interpreted as indicating the `` cavity modes '' of the magnetic shell .
in fact , being @xmath166 the wavelength inside a nanoshell with thickness @xmath167 , the em cloaking occurs for @xmath168 .
note that this is the same condition for resonant modes in a fabry - prot cavity with @xmath169 being the separation distance between its mirrors .
since this analytical result is typical of resonant optical cavities , it suggests that the cloaking condition does not depend on the geometry of the particle ( optical cavity ) , but rather the magnetic material parameters , shell thickness , and irradiation schemes .
analogously , similar results are obtained for the reciprocal configuration ( high permittivity coatings ) setting @xmath170 , for @xmath171 , where the same cloaking condition is achieved . rewriting the cloaking condition
, we explicitly have a mutual dependence between @xmath172 and @xmath173 , for @xmath62 , @xmath94 , @xmath61 , and @xmath119 : @xmath174 ^ 2\ , \quad n=\pm1,\pm2,\ldots\label{cloaking}\end{aligned}\ ] ] note from eq .
( [ cloaking ] ) that for @xmath175 , large values of @xmath173 ( or @xmath172 ) are obtained for @xmath152 ( small shell thickness ) . besides , eq .
( [ cloaking ] ) does not depend on @xmath9 , and this independence on the core material parameters is valid for @xmath94 .
this is of great interest , for instance , to cloak an arbitrary dielectric particle . also , due to the electric and magnetic fano - like resonances , the cloaking frequencies given by eq .
( [ cloaking ] ) are spectrally near to a pair of scattering resonance peaks .
each of these scattering , fano - comb resonances are associated with a mode number @xmath165 and are referred to as morphology - dependent resonances of first order @xmath176 .
in the following we do not restrict ourselves to the small - particle limit , in which our previous analytical results are based , but we rather consider the exact expressions within the lorenz - mie theory , given by eqs .
( [ qsca])([bn ] ) . in particular , we consider a dielectric core with optical properties @xmath177 . for the shell material ,
we consider a lossy drude model for the relative permittivity , @xmath178 $ ] , with the same parameters as used in ref .
@xcite for aluminum - doped zinc oxide semiconductors : @xmath179 , @xmath180 thz , and @xmath181 .
the core radius is @xmath182 nm and the aspect ratio is @xmath183 ( @xmath184 ) . for the shell permeability , instead of assuming @xmath185 ( nonmagnetic material ) , we investigate the cases @xmath186 , @xmath187 , @xmath188 , and @xmath189 . for a dielectric sphere @xmath177 coated with a plasmonic , lossy magnetic shell @xmath190 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the dotted line represents the sphere without coating .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) @xmath192 for @xmath193 . the @xmath129 and @xmath132
curves refer to the electric and magnetic dipole contributions within the exact lorenz - mie theory , respectively .
the inset is calculated from the dipole approximations ( [ f1-te])([g1-tm ] ) , showing that @xmath194 is in good agreement with @xmath192 .
( b ) @xmath192 for @xmath195 . ] for a dielectric sphere @xmath177 coated with a plasmonic , lossy magnetic shell @xmath190 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the dotted line represents the sphere without coating .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) @xmath192 for @xmath193 . the @xmath129 and @xmath132
curves refer to the electric and magnetic dipole contributions within the exact lorenz - mie theory , respectively .
the inset is calculated from the dipole approximations ( [ f1-te])([g1-tm ] ) , showing that @xmath194 is in good agreement with @xmath192 .
( b ) @xmath192 for @xmath195 . ]
figure [ fig2 ] shows comblike , multiple fano resonances in the scattering efficiency @xmath192 for @xmath196>0 $ ] and @xmath1 in the vicinity of @xmath197 , with @xmath198 and @xmath199 . in fig .
[ fig2](a ) , we plot the scattering efficiency for @xmath186 ( _ i.e. _ , @xmath200 ) showing both the electric and magnetic dipole contributions with the corresponding antiresonance or fano - dip in between .
the inset of fig .
[ fig2](a ) shows the scattering efficiency @xmath201 calculated from the electric and magnetic dipole approximations , given by eqs .
( [ f1-te])([g1-tm ] ) . as can it be confirmed by comparing the inset to the main plot of fig .
[ fig2](a ) , these approximations are in good agreement with the exact lorenz - mie theory , from which the main plot in fig . [ fig2](a ) is calculated . however ,
from now on , all the calculations are performed using the exact expressions for the lorenz - mie coefficients @xcite .
increasing @xmath173 up to @xmath187 ( _ i.e. _ , @xmath202 ) , a second scattering dip appears in the same spectral frequency range , as evinced in fig .
[ fig2](b ) . a very good estimate of the frequency position of these scattering minima is given by eq .
( [ cloaking ] ) , even though the rayleigh and the small - particle limits are not entirely satisfied . for this set of parameters , very efficient cloaking is achieved for a dielectric sphere @xmath177 around @xmath203 , for @xmath204 and @xmath205 .
we emphasize , nevertheless , that our aim is not to find an optimal set of parameters to cloak a dielectric particle @xcite , but rather to achieve fano - comb scattering resonances as a function of the magnetic permeability of the coating .
coated with a plasmonic , lossy magnetic shell @xmath190 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) scattering efficiency @xmath192 and absorption efficiency @xmath206 ( inset ) for @xmath207 .
( b ) shell permittivities @xmath172 calculated from the drude model and from eq .
( [ cloaking ] ) for @xmath207 and @xmath208 , @xmath209 , and @xmath210 .
the intersection between the vertical lines and the drude curve corresponds to the localization , in frequency , of the fano - comb . ]
coated with a plasmonic , lossy magnetic shell @xmath190 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) scattering efficiency @xmath192 and absorption efficiency @xmath206 ( inset ) for @xmath207 .
( b ) shell permittivities @xmath172 calculated from the drude model and from eq .
( [ cloaking ] ) for @xmath207 and @xmath208 , @xmath209 , and @xmath210 .
the intersection between the vertical lines and the drude curve corresponds to the localization , in frequency , of the fano - comb . ] in figs .
[ fig3](a ) and [ fig3](b ) , we show the comblike scattering response for @xmath211 and compare it to the analytical prediction given by eq .
( [ cloaking ] ) , which determines the values of @xmath172 for which these multiple fano resonances occur . as it can be seen from fig .
[ fig3 ] , eq .
( [ cloaking ] ) gives an excellent prediction for the frequency position of comblike resonances even for a lossy shell with @xmath212 , @xmath213 , and for higher orders of @xmath165 .
figure [ fig3](b ) corroborates that eq .
( [ cloaking ] ) is valid even beyond the rayleigh limit for the core and is robust against absorption , provided that @xmath61 ( with @xmath214 ) and @xmath119 . in particular , note that each value of @xmath165 replicates the same scattering profile of electric and magnetic dipole resonances , but with different amplitudes , and that they are approximately equally spaced in frequency . also , the inset in fig .
[ fig3](a ) shows high absorption in these antiresonance scattering points ( fano - dips ) .
this high absorption can be associated with resonances of the em energy within the scatterer @xcite .
in fact , according to refs .
@xcite , the fano resonance in the scattering cross sections may lead to off - resonance field enhancement within the particle .
it is important to mention that the profile and the position of the comblike resonances can be externally tuned in engineered magnetic materials .
indeed , by applying an external magnetic field and/or varying the temperature , one can change the value of @xmath215 .
ferromagnetic materials , for instance , typically follow the curie - weiss law above their critical temperature @xmath216 @xcite : @xmath217 .
metamaterials made of ferroelectric compounds are examples of systems with tunable magnetic permeability at terahertz frequencies , with relatively low losses @xcite . in these materials
the tunability is achieved by varying the temperature , so that they could possibly be employed in an experimental verification of our findings .
this possibility of tuning the value of @xmath215 has influence not only on the resonance position but also on the number of resonances that compose the comblike scattering response at a given spectral range .
this can be seen , _
e.g. _ , in fig .
[ fig2 ] .
we emphasize that , to the best of our knowledge , this tunability is not found in previous proposals involving comblike resonances in core - shell nanoparticles , which are based mainly on nonmagnetic materials @xcite . in these previous systems ,
the position and profile of comblike resonances are determined _ a priori _ by the nanostructure design , either by varying the plasma frequencies with a gradient of doping level in multilayered spheres @xcite or by constructing nonconcentric core - shell nanowires @xcite .
this fact , which distinguishes our proposal from the previous ones , allows for tunable , versatile comblike scattering response in concentric , single - layered core - shell nanoparticles .
coated with a plasmonic , lossy magnetic shell @xmath218 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) scattering efficiency @xmath192 .
( b ) differential scattering efficiencies @xmath219 and @xmath220 ( inset ) . ] coated with a plasmonic , lossy magnetic shell @xmath218 $ ] .
the inner sphere has radius @xmath182 nm and aspect ratio @xmath191 .
the cloaking occurs at @xmath172 values given by eq .
( [ cloaking ] ) .
( a ) scattering efficiency @xmath192 .
( b ) differential scattering efficiencies @xmath219 and @xmath220 ( inset ) . ] in fig .
[ fig4](a ) , the shell permeability @xmath221 is one order of magnitude greater than in fig .
[ fig2](a ) , which implies that more scattering dips occur at the same spectral range .
the differential efficiencies for the back and forward scattering @xcite , respectively , @xmath222 and @xmath223 , are plotted in fig .
[ fig4](b ) . due to the fact that the shell is magnetic ( @xmath224 )
light scattering is asymmetric with a backscattering dominance ; indeed , @xmath219 has essentially the same profile as @xmath192 , as can be verified in figs .
[ fig4](a ) and [ fig4](b ) .
on one hand , the interference between electric and magnetic dipoles facilitates the suppression at the scattering dips of order @xmath165 in the backscattering direction . on the other hand , the forward scattering efficiency @xmath225 , plotted in the inset of fig .
[ fig4](b ) , shows a strong suppression in the spectral frequency region between the @xmath165 and @xmath226 dips , and additional fano - dips due to the conventional fano resonance .
this strong asymmetry between back and forward scattering may be exploited in applications .
indeed , the ultrasharp fano - comb scattering suggests applications in optical tagging and signal processing , for instance .
in contrast to previous studies on multiple fano resonances @xcite , our approach not only allows one to analytically predict , by eq .
( [ cloaking ] ) , the frequency position where these resonances are expected to occur , but also allows for the possibility of tuning their positions via the external variation of the magnetic permeability of the nanoshell .
within the lorenz - mie theory , we have investigated the possibility of achieving multiple fano resonances in the scattering response of dielectric spheres coated with a single - layer , concentric magnetic shell .
we have derived , in the long - wavelength limit , explicit analytical conditions for the occurrence of comblike resonances in core - shell spheres . in particular , we have shown that nanoshells with high magnetic permeability or high electric permittivity values induce the formation of multiple fano , comblike resonances composed of electric and magnetic dipole resonances .
these fano - comb resonances have been shown to follow a typical relation of a resonant optical cavity . as the condition for the occurrence of multiple fano , morphology - dependent resonances explicitly depends on the magnetic permeability of the shell
, we argue that both the position and profile of the comblike resonances could be tailored by an external magnetic field and/or by varying the temperature .
together with the simplicity of employing a single - layered core - shell nanoparticle , this tunability of the single scattering response distinguish our proposal from the previous ones considered so far , which are based on _
e.g. _ multilayered semiconductor materials or nonconcentric core - shell particles .
these singular scattering properties make the system proposed here a tunable , versatile optical device that may find applications in multifrequency biosensing , optical tagging and signal processing .
the authors acknowledge the brazilian agencies for support . t.j.a
. holds grants from fundao de amparo pesquisa do estado de so paulo ( fapesp ) ( 2010/10052 - 0 ) and a.s.m . holds grants from conselho nacional de desenvolvimento cientfico e tecnolgico ( cnpq ) ( 307948/2014 - 5 ) .
thanks the optoelectronics research centre and centre for photonic metamaterials , university of southampton , for the hospitality , and capes for funding his visit ( bex 1497/14 - 6 ) .
f.a.p . also acknowledges cnpq ( 303286/2013 - 0 ) for financial support . | we investigate multiple fano , comblike scattering resonances in single - layered , concentric core - shell nanoparticles composed of magnetic materials . using the lorenz - mie theory , we derive , in the long - wavelength limit , an analytical condition for the occurrence of comblike resonances in the single scattering by coated spheres .
this condition establishes that comblike scattering response uniquely depends on material parameters and thickness of the shell , provided that it is magnetic and thin compared to the scatterer radius .
we also demonstrate that comblike scattering response shows up beyond the long - wavelength limit and it is robust against absorption .
since multiple fano resonances are shown to depend explicitly on the magnetic permeability of the shell , we argue that both the position and profile of the comblike , morphology - dependent resonances could be externally tuned by exploiting the properties of engineered magnetic materials . |
since the beginning of the space age in the late 1950 s , `` space weather '' is a hot topic .
it deals with the space environment of the earth s atmosphere from approximately 90 km altitude onwards extending all the way to the sun .
strong disturbances of this space environment are primarily caused by solar flares , coronal mass ejections ( cmes ) and associated solar energetic particles ( seps ) .
owing to advances in solar research in the recent decades , today we know that cmes cause the most comprehensive spectrum of space weather disturbances , including geomagnetic storms , aurorae , geomagnetically induced currents affecting electric power lines , heating of the outer atmosphere , and many more .
cmes are huge clouds of magnetized plasma propagating from the solar corona into interplanetary space with speeds ranging from a few hundred up to @xmath53000 km s@xmath6 .
cmes often reveal a three - part structure @xcite ( fig .
[ fig:1 ] ) : the front ( leading edge ) with its shock - sheath region @xcite , the void which comprises the magnetic structure of a cme ( often with a magnetic flux rope inherited ) and the bright core consisting of filament material .
the occurrence rate of cmes is closely related to the solar cycle , and varies from 0.3 per day at solar minimum to 45 per day on average at solar maximum ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . close to the sun ,
the cme dynamics is governed by the driving lorentz forces resulting from magnetic instabilities in the fields of active regions @xcite . in interplanetary space ,
on the other hand , the cme dynamics is governed by the interaction of the cme with the surrounding solar wind flow , whereby fast cmes are decelerated and slow cmes are accelerated ( e.g. , * ? ? ?
this interaction process with the collisionless plasma in the interplanetary space is described by the magnetohydrodynamic equivalent of the aerodynamic drag force @xcite .
thus , the transit time and the arrival speed of an earth - directed cme which are both crucial parameters for space weather description and forecasting depend on both the initial speed and the state and interaction of the ambient solar wind plasma with the cme .
typical transit times range between 25 days , depending on initial speed , mass and size as well as the speed and density of the surrounding solar wind plasma .
the fastest cme events observed so far reached the earth in less than one day ( e.g. , * ? ? ?
* ; * ? ? ?
part of the solar wind energy supplied to the earth s magnetosphere is continuously deposited into the thermosphere via particle precipitation and joule heating .
this energy supply can be enormously enhanced when a fast cme with a strong southward magnetic field component arrives at the earth s atmosphere , causing a strong geomagnetic storm @xcite .
the enhanced energy input from the solar wind to the magnetosphere and its ultimate dissipation in the thermosphere causes heating and expansion of the earth s thermosphere @xcite . during periods of high cme activity ,
short timescale ( 12 days ) variations in the thermosphere are driven by magnetospheric energy input rather than by changes in the solar extreme - ultraviolet ( euv ) flux ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite have shown that in case of extreme events the geomagnetic power contributes two - thirds of the total solar power at the earth s thermosphere , with the major contribution from joule heating .
the increase in the neutral density affects earth - orbiting satellites in such a way that the drag force on the spacecraft is enhanced .
the fact that some low - earth orbiters ( leos ) are equipped with accelerometers allows the derivation of in - situ measurements of the current state of the earth s thermospheric density . to this class of satellites
belong the spacecraft of the projects challening mini satellite payload ( champ ; * ? ? ?
* ) and gravity recovery and climate experiment ( grace ; * ? ? ?
* ) . in this paper , we present a statistical study of the thermospheric response to a sample of 104 interplanetary cmes ( icmes )
; most of these events caused a significant increase in the neutral density .
the period under investigation covers seven years during the maximum and decaying phase of solar cycle 23 ( 7/2003 to 8/2010 ) , including the `` halloween storms '' .
thermospheric density changes are derived from grace accelerometer measurements and compared with icme properties measured by the advanced composition explorer ( ace ; * ? ? ?
* ) and various geomagnetic storm indices .
due to the high correlations obtained , we were able to derive regression curve parameters , which we expect to be highly valuable for the improved estimates of future cme - related thermospheric density enhancements .
for the present study we analyzed all icme events during the time range july 13 , 2003 until august 04 , 2010 as listed in the catalogue of near - earth interplanetary icmes maintained by richardson & cane ( r&c ) @xcite . during this period
the r&c list contains 106 icme events , comprising the arrival times of the shock - sheath region , the start and end times of the magnetic structure ( based on the plasma and magnetic field observations ) and the minimum value of the geomagnetic disturbance storm time index ( dst ) @xcite within the total icme duration . for our statistical study
we thoroughly analyzed the magnetic field characteristics of the icmes making use of measurements by the ace satellite located at the lagrange point l1 .
the response of the earth s thermosphere in terms of changes in neutral density are derived from acceleration measurements provided by the grace mission , which has been orbiting the earth at an altitude of @xmath7450 - 500 km since 2002 .
furthermore , we used various geomagnetic indices derived from terrestrial magnetic field measurements collected at ground - based observatories .
the icme kinematics and its associated magnetic field were derived from in - situ measurements recorded by instruments aboard the ace spacecraft in particular the solar wind electron , proton , and alpha monitor ( swepam ) @xcite and the magnetometer ( mag ) @xcite .
we used level-2 data of the proton speed ( solar wind bulk speed ) and the magnetic field ( @xmath8 component in geocentric solar ecliptic , gse , coordinates ) with a time resolution of 4min ( omni database ) @xcite . in gse
coordinates the @xmath9-component is oriented orthogonal to the ecliptic , from which we derived the southward component of the magnetic field .
we identified minimum values in the @xmath8 component separately for the shock - sheath region and the magnetic structure of the icme .
bad or missing data were linearly interpolated .
time information , such as start and end of the disturbance , were extracted from the r&c list .
observations from ace during an icme event in may 2010 .
red circles indicate the minimum values used in the further analysis .
yellow and green areas mark the shock - sheath region and magnetic structure , respectively , according to the time information provided in the r&c list.,width=302 ] figure [ fig:2 ] shows the temporal evolution of the @xmath8 component , exemplary for an icme event in may 2010 .
the shock - sheath region covers a period of strongly varying magnetic field ( fig .
[ fig:2 ] , yellow area ) , with a minimum occurring on may 28 at approximately 11:00 ut .
the second minimum occurs about 19 hours later when the magnetic structure of the icme arrived ( fig .
[ fig:2 ] , green area ) . during this period
the @xmath8 component shows a smooth behavior and changing orientation , indicative of field rotation .
the existence of a shock wave depends on the speed of the icme relative to the ambient solar wind speed . for 82 of the icme events analyzed in this study
, a shock - sheath region could be identified in the in - situ plasma and field data . in order to gain knowledge about the response of the earth
s thermosphere to icme events we analyzed accelerometer data collected by the us - german grace project . the grace satellites were launched in march 2002 and injected into a low - altitude ( initially 505 km ) , near - circular ( eccentricity @xmath10 ) , and near - polar ( inclination @xmath11 ) orbit around the earth .
the grace constellation consists of two identical spacecraft following each other in the same orbit , separated by about 220 km .
grace is a dedicated gravity field mission , and therefore both satellites are equipped with accelerometers ; the grace accelerometers were manufactured by the office national detudes et de rcherches aerospatials ( onera ) and have an accuracy of @xmath12 m s@xmath13 . as far as gravity field recovery
is concerned , accelerometer measurements are used to get rid of the non - gravitational forces acting on the satellites . here , we use these observations to compute neutral densities ( @xmath14 ) along the trajectory of the spacecraft . the transition from the along track accelerometer measurements ( @xmath15 ) to atmospheric neutral densities
can be established by @xmath16 where @xmath17 indicates the satellite mass , @xmath18 the highly complex and variable drag coefficient , @xmath19 the effective cross sectional area and @xmath20 the satellite velocity in the along track axis . for a detailed description of the technique
we refer the reader to @xcite .
due to ellipticity , the true grace orbit implies satellite altitude variations of about 50 km per revolution .
for this reason , all densities shown in this study are normalized to an average altitude of 490 km ( @xmath21 ) using the empirical thermosphere model by @xcite ( @xmath22 , @xmath23 ) according to @xmath24 we smoothed the density time series with a 5-min moving average filter to avoid single peaks biasing the results . as an example , fig .
[ fig:3 ] depicts neutral densities in a latitude - time plot covering a period of ten days in july 2004 ; during this period three icmes hit the earth .
compared to the pre event level of @xmath25 kgm@xmath26 , each icme increased the neutral density by a factor of up to five .
the perturbations started at high latitudes and thereafter propagated towards the equator within a few hours .
these phenomena are known as travelling atmospheric disturbances and occur at high latitudes due to sudden energy and momentum deposition , mostly in the form of joule heating ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . for most of the analyzed icme events
, we found the maximum geomagnetic disturbance at high latitudes near the auroral region ( with a latitude root mean square value of @xmath27 ) .
geomagnetic indices play an essential role in the space weather system , describing the response of the earth s magnetosphere to effects of the changing solar wind . in the course of the past 80 years various indices
were developed , which are monitoring the geomagnetic activity on different latitudes and time scales .
figure [ fig:4 ] gives an overview of the global distribution of geomagnetic observatories , which provide the measurements for the different geomagnetic indices .
for this study we analyzed ten different geomagnetic indices .
four of them ( a~a~ @xcite , a~m~ @xcite a~p~
@xcite and k~p~ @xcite ) are deduced from the quasi logarithmic k - index @xcite a measure for the variations in the two horizontal field components .
the hourly dst index ( derived from observations made by four near - equatorial observatories ) describes the global symmetrical geomagnetic perturbations of the ring current a current flowing westward around the earth .
further we analyzed the auroral electrojet index ( ae ) @xcite , which is based on observations in the auroral zone and monitors the horizontal field in the east west ionospheric current ; it is constructed from the au and al indices , which represent the largest and lowest deviation from a base value at each station , respectively ( @xmath28 ) .
the polar cap index ( pc ) @xcite represents geomagnetic disturbances in the polar cap regions due to ionospheric and field aligned currents @xcite ; measurements are performed by the stations thule and vostok , therefore the index can be divided into a northern ( pcn ) and southern ( pcs ) part , respectively . finally , we included the sym
h index which describes the longitudinal symmetric disturbance field in mid - latitudes for the horizontal component .
basically it is the same as the hourly dst index and describes the same axial symmetric disturbances in the horizontal component @xcite .
the main difference is in the temporal resolution of the indices ( 1h versus 1min ) , i.e. short - term changes in solar wind parameters are reflected differently in the two indices . for more information concerning geomagnetic indices
the reader is referred to @xcite .
the time period july 2003 to august 2010 was chosen because of the availability of grace calibration data @xcite . during this time span
the r&c list contains 106 icme events . for two of them ,
no grace acceleration measurements are available .
the remaining set of 104 icmes is studied in this paper . starting with the disturbance time of the icme ( taken from the r&c list ) we defined a time window of 36 hours over which we extracted the evolution of the various geomagnetic indices , the neutral densities ( from grace ) and the magnetic field measurements ( from ace ) . for every single icme event , we derived from each parameter the peak value within the predefined time window .
we thoroughly checked the data to ensure that the correct peak value is assigned to each icme event .
this is particularly important for time intervals containing more than one icme event as well as for complex icme structures .
+ to account for variations in the background thermospheric density ( caused by the non - disturbed solar activity ) we calculated the absolute density increase , i.e. the difference @xmath29 between the peak density measured during the event minus the pre event level , as well as the relative density increase in percent ; for the correlation studies the @xmath29 values were used .
exemplarily for two icme events , fig .
[ fig:5 ] shows the evolution of the individual parameters .
additionally , each panel includes the progress of the neutral density in terms of an envelope function ( red curves ) as well as the detected peak values ( blue circles ) .
the envelope function was created such that it comprises the maximum density per satellite revolution ( fig .
[ fig:5 ] , a1 , d1 ) .
figure [ fig:5 ] reveals that the temporal evolution of both the geomagnetic indices and the @xmath1 component of the magnetic field is in good agreement with the neutral density .
as far as the dst index and the sym
h index ( a3,b3 ; d3,e3 ) is concerned , the different phases of the geomagnetic storm become clearly visible : a sudden decrease of the indices during the main phase of the storm , followed by a much slower recovery phase . regarding the absolute values , over the entire range we derive slightly higher values in the sym
h index than in the dst index and in some cases a small offset between these two indices ( d3 , e3 ) .
this behaviour can be explained by the different methods in the baseline preparations for the two indices @xcite .
notice that the a indices ( row c , f ) exhibit a rather rough structure , caused by their low temporal resolution ( 3h ) .
the ae index shows disturbances throughout the whole icme events and even beyond that ( b2 ) . comparing the two polar cap indices ( b1,e1 ) we see larger differences during the most disturbed period than before and after the event , implying different conditions in the polar cap regions @xcite .
+ in general , the majority of the analyzed icme events shows a significant density increase in the atmosphere compared to the density level before the perturbation ( fig .
[ fig:6 ] ) .
the largest relative increase occurred for two events in 2005 ( may 15 and august 24 ) ; in both cases the density rose by about 750% compared to the pre
event level .
the highest absolute increases in neutral density occurred during the halloween storms in 2003 ( october 2930 and november 20 ) with @xmath30 kg m@xmath26 .
in a first step , we compare the in - situ measurements by the ace and grace spacecraft in terms of magnetic field strength ( @xmath1 in gse ) and thermospheric neutral density ( @xmath29 ) , respectively .
additionally , we use of the maximum speed ( @xmath0 ) , measured over the entire icme structure as given in the r&c list . + figure [ fig:7 ] shows scatter plots of the thermospheric density increase @xmath29 versus @xmath1 , @xmath0 , as well as two combinations of these quantities . on the one hand a proxy for the convective electric field @xcite represented by @xmath31 on the other hand a proxy for the energy input via the poynting flux into the magnetosphere by magnetic reconnection @xcite defined as @xmath32 from the regression analysis we derive the lowest correlation coefficient ( @xmath33 ) for the maximum solar wind speed during the disturbance period ( @xmath0 ) , inferring that this is not the primary factor for effective driving of geomagnetic storms .
in fact , very high impact velocities ( @xmath34 1500 km s@xmath6 ) are only present for two of the 104 events .
irrespective of whether a density response occurred or not , the majority of icmes has velocities in the range between 400 and 800 km s@xmath6 which is of the statistical average ( e.g. , * ? ? ?
+ in contrast to @xmath0 , the quantities @xmath1 , @xmath35 and @xmath36 show a much higher correlation with the thermospheric density with @xmath37 , @xmath38 and @xmath39 , respectively .
the latter correlation coefficient is particularly influenced by one extreme icme event ; excluding the gigantic halloween icme ( october 29 , 2003 ) from the calculations leads to a significantly increased correlation coefficient between the @xmath29 and @xmath36 of @xmath40 .
[ fig:7 ] ( top left ) also indicates that icme events with hardly any thermospheric response are associated with a small @xmath1 component ( @xmath4110 nt ) .
+ in the next step we analyzed the correlation between the neutral density increase @xmath29 and the different geomagnetic indices .
figure [ fig:8 ] provides an overview in terms of scatter plots .
it can clearly be seen that most of the geomagnetic indices show an excellent correlation with the thermospheric density response .
the highest correlations are derived for the indices @xmath42 ( @xmath43 ) , @xmath44 ( @xmath43 ) and dst ( @xmath45 ) together with the sym
h index ( @xmath45 ) .
the reason for the slightly lower correlation of the @xmath46 index might be due to the worse station distribution ( 13 in the northern hemisphere , 2 in the southern hemisphere ) . significantly lower correlations are present for the @xmath4 ( @xmath47 ) and ae index ( @xmath48 ) . in the first case ,
this behaviour is not surprising since it is well known that the @xmath4 index is not linearly correlated with the geomagnetic activity @xcite ; for this reason the planetary @xmath46 index was established in 1954 . between the density increase and the ae index
a non - linear relationship can be observed . analyzing fig .
[ fig:8 ] ( top , right ) we see different behaviours depending on the geomagnetic activity level .
only minor density increases are visible for ae values up to 1000 nt . above this value
, we recognize larger scattering and some indication of saturation .
the lower correlation between the ae index and the thermospheric densities is in agreement with previous findings @xcite .
these authors revealed some limitations concerning the measured quantity ( h field ) , the existence of longitudinal gaps and the small latitudinal range of the contributing observatories . compared with the other indices the correlations between the density increases and the polar cap indices ( pcn , pcs ) are smaller .
however , it should be noted that for some icme events no pcs data was available which excludes some of the strongest events ( `` halloween storms '' in october / november 2003 ) .
.correlation coefficient between the various parameters and the increase in neutral density @xmath49 ( column 2 ) as well as the @xmath1 component measured by ace ( column 3 ) . [ cols= " <
, > , > " , ] [ tab : correlation ] table [ tab : correlation ] summarizes all determined correlation coefficients with regard to the atmospheric density @xmath49 and magnetic field component @xmath1 . in general , the correlations between @xmath1 and the geomagnetic indices are very similar to those of the neutral density .
again the highest correlations are found for the @xmath42 , @xmath44 , dst and the sym h index .
likewise , the more exponential relation with the ae and @xmath4 indices and the slightly poorer correlations with the polar cap indices . + analyzing the minimum in the @xmath1 separately for the shock - sheath and the magnetic structure of the icme we may deduce some additional information . for some events , no shocks are reported , and the leading edge of the icme might be stated as the estimated arrival time of the disturbance .
see @xcite for additional information on the methods used to identify icmes . for 82 out of the 104 analyzed events it was possible to detect and
distinguish between the shock - sheath region and the magnetic structure of the specific icme .
for this subset of events we identified the peak values in the neutral density and the various geomagnetic indices separately for the shock sheath region and the magnetic structure of the icme . as a result , fig .
[ fig:9 ] illustrates the minimum @xmath1 component , either found in the shock - sheath region ( red squares ) or in the magnetic structure part ( blue squares ) of the icme , against the neutral density increase and the different geomagnetic indices , for the particular region . for the a indices and the ae index no significant differences in the regression line between the two specific regions are visible . however , large differences are present for the polar cap indices ( pcn , pcs ) , indicating that both correlate better with disturbances triggered at the time of arrival of the shock sheath region ( red regression line ) .
in contrast to that , the hourly dst index as well as the sym h index show better correlations when the minimum @xmath1 component is detected in the magnetic structure of the icme .
finally , we examine the time of occurrence of the neutral density and the the peak for the geomagnetic indices .
we find the peak times coincide significantly better with the times extracted for the minimum @xmath1 value in the magnetic structure rather than for the @xmath1 value in the shock - sheath region ( ratio : 4:1 ) .
+ the complete list of the selected peak values of all quantities discussed in this study is available as supporting information .
based on the icme catalogue maintained by r&c we have analyzed 104 icme events for the period july 2003 to august 2010 .
we studied the relationship between geomagnetic disturbances and thermospheric densities derived from grace accelerometer measurements . during periods of geomagnetic disturbances
the maximum of the signal was mainly found at high latitudes in the auroral zone ( see , e.g. , case study by @xcite ) . by correlating magnetic field characteristics of the icmes observed by the ace satellite we were able to derive linear regression parameters between the different measured quantities .
we found that the majority of these icmes , cause a significant increase in the earth s thermospheric neutral density . from the regression analysis
, it turned out that the icme parameter @xmath0 has a significantly lower correlation with the thermospheric density than the @xmath1 component or the combined parameters @xmath35 and @xmath36 , which are directly related to the energy input into the system .
it is worth noting that the highest correlation coefficient is obtained for @xmath1 ( @xmath50 ) , whereas the estimates for @xmath35 and @xmath36 do not yield improved correlations .
consequently , it is conceivable to obtain an estimate of the forthcoming maximum neutral density increase using in - situ observations of icme characteristics by appropriate in - situ measurements from satellites at the lagrange point l1 before the icme impacts earth s magnetosphere . using real - time in - situ data
, this would result in a lead time of some tens of minutes ( depending on the actual speed of the icme event ) .
furthermore , we can conform previous findings by @xcite , that all icmes causing a significantly density enhancement have shown a strong negative @xmath1 component . regarding the various geomagnetic indices we obtained high correlations with the thermospheric density as well as the @xmath1 component ; especially the sym
h index , hourly dst index , @xmath42 index and @xmath44 index showed a good accordance ( @xmath50 ) .
however , it is worth noting that , the a indices have a lower temporal resolution of three hours , which might lead to limitations in detailed space weather monitoring @xcite .
no linear relationship with the geomagnetic activity could be established for the ae and @xmath4 index .
subsequent investigations concentrated on the separation of the magnetic field observations @xmath1 in the shock sheath region and magnetic structure of the icme .
we found that the hourly dst index and the sym
h index correlate better with the @xmath1 component when the peak values are extracted from the magnetic structure part of the icme .
contrary to that , the polar cap indices ( pcn , pcs ) showed higher correlations with the @xmath1 minimum located in shock - sheath region . from a temporal view the occurrence times of the peak values in the neutral density and the geomagnetic indices coincide better ( 75% of the cases ) with the time of the minimum @xmath1 component in the magnetic structure of the icme .
however , strong perturbations from the shock - sheath region may cause an additional enhancement in the atmospheric neutral density @xcite .
finally , an important field of application of space weather is the induced satellite orbit decay due to variations in the atmospheric density . by using different parameter types ( @xmath51 , @xmath19 @xmath52 ) elaborated in the present study we determined the orbit decay rate in terms of changes in the mean semi major axis following @xcite .
figure [ fig:10 ] shows exemplarily , for the icme event on november 20 in 2003 , variations of the grace semi major axis .
the middle panel illustrates the total orbit decay over the entire event , whereas the bottom panel shows the orbit decay rate per day . comparing the evolution of the neutral density ( fig .
[ fig:10 ] , top ) and the orbit decay rate per day ( fig . [ fig:10 ] , bottom ) it becomes apparent that these quantities reveal an almost one - to - one relation . as a result ,
the distinct linear relationships derived between thermal density enhancements and the icme @xmath1 component can in principle be used to predict the orbital decay rates from the icme measurements at l1 upstream of earth .
an evaluation of this approach will be the topic of a forthcoming study .
supporting data are included as a table in an si file ; alternatively the data may be obtained from s.k .
( email : sandro.krauss[at]oeaw.ac.at ) .
a.v gratefully acknowledge the support from the austrian science fund : project nos .
fwf v195-n16 and fwf p27292-n20 .
acknowledges support from the fwf nfn subproject s11607-n16 particle / radiative interactions with upper atmospheres of planetary bodies under extreme stellar conditions. s.k .
thanks sean bruinsma for his support with the grace accelerometer calibration
. merits and limitations of the use of geomagnetic indices in solar wind - magnetosphere coupling studies .
solar wind magnetosphere coupling , edited by y. kamida nad j.a .
slavin , 315 , terra sci . ,
tokyo , 1986 .
bein , b.m .
, berkebile - stoiser , s. , veronig , a.m. , temmer , m. , muhr , n. , kienreich , i. , utz , d. , vrnak , b. impulsive acceleration of coronal mass ejections . i. statistics and coronal mass ejection source region characteristics . the astrophysical journal , 738 , 191 , doi : 10.1088/0004 - 637x/738/2/191 , 2012 .
bowman , b.r . ,
tobiska , w.k . ,
marcos , f.a . ,
huang , c.y . ,
lin , c.s . ,
burke , w.j .
, a new empirical thermospheric density model jb2008 using new solar and geomagnetic indices , aiaa / aas astrodynamics specialist conference , aiaa 2008 - 6438 , 2008b .
bruinsma , s. , forbes , j.m . , nerem , r.s .
, zhang , x. thermosphere density response to the 2021 november 2003 solar and geomagnetic storm from champ and grace accelerometer data .
j. geophys .
res . , 111 , 114 , 2006 .
chen , g. , xu , j. , wang , w. , lei , j. , burns , a.g . , a comparison of the effects if cir- and cme induced geomagnetic activity on thermospheric densities and spacecraft orbits : case studies .
j. geophys .
space physics , 117 , a08315 , doi : 10.1029/2012ja017782 , 2012 .
chen , g. , xu , j. , wang , w. , burns , a.g .
, a comparison of the effects if cir- and cme induced geomagnetic activity on thermospheric densities and spacecraft orbits : statistical studies . j. geophys .
space physics , 119 , 79287939 , doi : 10.1002/2014ja019831 , 2014 .
doornbos , e.,frster , m.,fritsche , b.,helleputte , t.v.,ijssel , j.v.d.,koppenwallner , w. , lhr , h. , rees , d. , visser , p. , air density models derived from multi - satellite drag observations , technical report , deos tu delft,317 , 2009 .
gopalswamy , n. , lara , a. , lepping , r. p. , kaiser , m. l. , berdichevsky , d. , st .
cyr , o. c. , interplanetary acceleration of coronal mass ejections .
lett . , 27 , 145 - 148 , doi : 10.1029/1999gl003639 , 2000 . gosling , j.t . ,
mccomas , d.j . ,
phillips , j.l . , bame , s.j .
, geomagnetic activity associated with earth passage of interplanetary shock disturbances and coronal mass ejections .
j. geophys .
space physics , 96 , a5 , 78317839 , doi : 10.1029/91ja00316 , 1991 .
iyemori , t. , takeda , m. , nose m. , odagi y. , toh , h. , mid - latitude geomagnetic indices `` asy '' and `` sym '' for 2009 ( provisional ) , data anal . cent . for geomagn . and space magn . , faculty of sci .
, kyoto univ . ,
kyoto , japan .
king , j.h.,papitashvili , n.e .
, solar wind spatial scales in and comparisons of hourly wind and ace plasma and magnetic field data , j. geophys .
a2 , a02209 , 10.1029/2004ja010804 , 2004 .
krauss , s. , fichtinger b. , lammer , h. , hausleitner , w. , kulikov , y.n .
, ribas , i. , shematovich , v. , bisikalo , d. , lichtnegger , h.i.m .
, zaqarashvili , t.v .
, khodachenko , m.l .
, hanslmeier , a. , solar flares as proxy for young sun : satellite observed thermosphere response to an x17.2 flare of earth s upper atmosphere
geophys . , 30 , 1129 - 1141 , doi : 10.5194 \angeo-30 - 1129 - 2012 , 2012
. krauss , s. , response of the earth s thermosphere during extreme solar events .
a contribution of satellite observations to atmospheric evolution studies .
thesis , graz university of technology , 2013 .
liu , y.d . , yang , z. , wang , r. , luhmann , j.g . , richardson , j.d . ,
lugaz , n. , sun - to - earth characteristics of two coronal mass ejections interacting near 1 au : formation of a complex ejecta and generation of a two - step geomagnetic storm .
the astrophysical journal letters , 793:l41 , doi : 10.1088/2041 - 8205/793/2/l41 , 2014 .
lukianova , r. , troshichev o. , the polar cap magnetic activity indices in the southern ( pcs ) and northern ( pcn ) polar caps : consitency and discrepancy , geophys .
lett . , 29 , 18 , 1879 , doi : 10.1029/2002gl015179 , 2002 . mccomas , d.j . ,
bame , s.j . ,
barker , p. , feldman , w.c . ,
phillips , j.l . , riley , p. , griffee , j.w .
, solar wind electron proton alpha monitor ( swepam ) for the advanced composition explorer .
space science .
, 86 , 563?612 , 1998 .
menvielle , m. , iyemori , t. , marchaudon , a. , nos , m. geomagnetic indices in geomagnetic observations and models . edited by m. mandea nad m. korte , iaga special sopron book series , vol 5 . , springer+business media 183 - 228 , isbn : 978 - 90 - 481 - 9857 - 3 , doi 10.1007/978 - 90 - 481 - 9858 - 0 , 2011 . st .
cyr , o.c . ,
howard , r.a .
, sheeley , n.r . ,
plunkett , s.p . , michels , d.j . , paswaters , s.e . , koomen , m.j . ,
simnett , g.m . ,
thompson , b.j .
, gurman , j.b . , schwenn , r.,webb , d.f . , hildner , e. , lamy , p.l . ,
properties of coronal mass ejections : soho lasco observations from january 1996 to june 1998 , j. geophys .
res . , 105 , 18169?18186 , doi:10.1029/1999ja000381 , 2000 .
stone , e.c . ,
frandsen , a.m. , mewaldt , r.a .
, christian e.r .
, margolies , d. , ormes , j.f.,snow , f. , the advanced composition explorer , space science reviews , 86 , 1 - 4 , pp .
122 , doi : 10.1023/a:1005082526237 , 1998 .
sutton , e.k .
, effects of solar disturbances on the thermosphere densities and winds from champ and grace satellite accelerometer data .
thesis , department of aerospace engineering sciences , university of colorado , 2008 .
tapley , b.d . ,
bettadpur , s. , watkins , m. , reigber , c. , the gravity recovery and climate experiment : mission overview and early results , geophysical research letters , 31 , 9 , doi:10.1029/2004gl019920 , 2004 .
temmer , m. , veronig , a.m. , vrnak , b. , rybk , j. , gmry , p. , stoiser , s. , marii , d. , acceleration in fast halo cmes and synchronized flare hxr bursts .
the astrophysical journal letters , 673 , l95 , doi:10.1086/527414 , 2008 .
thayer , j.p . ,
liu , x. , lei , j. , pilinski , m. , burns , a.g . , the impact of helium on the thermosphere mass density response to geomagnetic activity during the recent solar minimum . j. geophys .
, 117 , a07315 , doi:10.1029/2012ja017832 , 2012 .
vassiliadis , d. , angelopoulos , v. , baker , d.n . ,
klimas , a.j . , the relation between the northern polar cap and auroral electrojet geomagnetic indices in the wintertime .
geophysical research letters , 23 , 20 , 27812784 , 1996 .
vourlidas , a. , lynch , b.j . ,
howard , r.a . , li , y. how many cmes have flux ropes ? deciphering the signatures of shocks , flux ropes , and prominences in coronagraph observations of cmes .
solar physics , 284,1,179 - 201,doi : 10.1007/s11207 - 012 - 0084 - 8 , 2013 .
vrnak , b. , marii , d. , stanger , a. l. , veronig , a. m. , temmer , m. , roa , d. , acceleration phase of coronal mass ejections : i. temporal and spatial scales .
solar physics , 241 , 85 - 98 , doi:10.1007/s11207 - 006 - 0290 - 3 , 2007 .
vrnak , b. , ic , t. , vrbanec , d. , temmer , m. , rollett , t. , mstl , c. , veronig , a. , alogovi , j. , dumbovi , m. , luli , s. , moon , y .- j . ,
shanmugaraju , a. , propagation of interplanetary coronal mass ejections : the drag - based model .
solar physics , 285 , 295 - 315 , doi : 10.1007/s11207 - 012 - 0035 - 4 , 2013 .
zhang , j. , dere , k. p. , howard , r. a. , kundu , m. r. , white , s. m. , on the temporal relationship between coronal mass ejections and flares .
the astrophysical journal letters , 559 , 452 - 462 , doi:10.1086/322405 , 2001 . | for the period july 2003 to august 2010 , the interplanetary coronal mass ejection ( icme ) catalogue maintained by richardson and cane lists 106 earth - directed events , which have been measured in - situ by plasma and field instruments on board the ace satellite .
we present a statistical investigation of the earth s thermospheric neutral density response by means of accelerometer measurements collected by the grace satellites , which are available for 104 icmes in the data set , and its relation to various geomagnetic indices and characteristic icme parameters such as the impact speed ( @xmath0 ) , southward magnetic field strength ( @xmath1 ) .
the majority of icmes causes a distinct density enhancement in the thermosphere , with up to a factor of eight compared to the pre event level .
we find high correlations between icme @xmath1 and thermospheric density enhancements ( @xmath2 ) , while the correlation with the icme impact speed is somewhat smaller ( @xmath3 ) .
the geomagnetic indices revealing the highest correlations are dst and sym - h ( @xmath2 ) , the lowest correlations are obtained for @xmath4 and ae ( @xmath3 ) , which show a nonlinear relation with the thermospheric density enhancements .
separating the response for the shock sheath region and the magnetic structure of the icme , we find that the dst and sym - h reveal a tighter relation to the @xmath1 minimum in the magnetic structure of the icme , whereas the polar cap indices show higher correlations with the @xmath1 minimum in the shock sheath region .
since the strength of the @xmath1 component either in the sheath or the magnetic structure of the icme is highly correlated ( @xmath2 ) with the neutral density enhancement , we discuss the possibility of satellite orbital decay estimates based on magnetic field measurements at l1 , i.e. before the icme hits the earth magnetosphere .
these results are expected to further stimulate progress in space weather understanding and applications regarding satellite operations . |
we consider quantum error correction coding ( qecc ) models where quantum bits ( qubits ) and quantum errors are both modeled as linear combination of pauli matrices @xmath0 .
an [ @xmath1 ( block ) stabilizer code to denote quantum stabilizer codes , ( ... ) to denote classical binary codes , and ( ... ) @xmath2 to denote classical quaternary codes . ] is defined as a codeword space @xmath3 which is _ non - trivially _ stabilized by a subgroup @xmath4 of @xmath5 , where @xmath5 is length-@xmath6 tensor product of pauli - operators @xmath7 , @xmath8 , @xmath9 , @xmath10 . by definition , the independent generators of @xmath4 commute to each other , and @xmath11 .
the dichotomy between the unobservable state of a qubit and the observations we can make , lies at the heart of quantum computation including qecc . in classic error correction ,
output from the channel is observed , measured and used directly in the decoding process .
observation in quantum mechanics not only destroys the quantum state and makes restoration impossible , but the measurement does not tell the original quantum state either . the measured outcome is always one of the two basic states ,
whose probability of occurring is proportional to the power projection of the original state on these two bases .
for this reason , to decode a stabilizer code typically takes three steps : measuring the syndrome , identifying the error pattern based on the syndrome , and applying a corrective operation to reverse the error . the syndrome diagnosis and the error reversal
are well - established quantum mechanical procedures , and operate the same way regardless of the stabilizer configuration . identifying error patterns , a process analogous to syndrome - decoding of classical codes , is most challenging .
not only are the techniques used here specific to the code structure , but they generally involve tedious ( exhaustive ) search in the hilbert space . compared to stabilizer block codes , stabilizer convolutional codes enjoy the flexibility in codeword length and online encoding .
the first comprehensive description of the general formalism of quantum convolutional codes appeared in @xcite , and a variety of quantum convolutional codes has since been proposed @xcite-@xcite .
these studies focus on the code construction ( e.g. based on classical reed - solomon codes , classical low - density parity check codes , and concatenation of classical convolutional codes ) , and do not usually discuss feasible decoding algorithms .
the first decoder for quantum convolutional codes appeared in @xcite , which developed a quantum viterbi algorithm ( qva ) close in spirit to the classical viterbi algorithm .
simplification is made to this qva by attacking only a _
unitary error per block ( and ignoring all the other possible error events ) @xcite .
for quantum convolutional codes constructed from quasi - cyclic sparse matrices , a different type of decoder based on pipeline message passing is developed @xcite .
since the decoder in @xcite tackles only single errors , and that in @xcite performs well only on sparse convolutional codes , the qva in @xcite is by far the only general and ml decoding method available for stabilizer convolutional codes . however , it requires a very high complexity , caused by a very large lookup table and many rounds of the conventional viterbi algorithm . the high complexity and the induced long delay
make this algorithm very challenging to implement ( and it is for this practicality issue that @xcite proposed to simplify it at the cost of degraded performance ) .
the contribution of this paper is the development of a new decoding algorithm which works for a general quantum convolutional code and which is drastically simpler than the existing one . not using lookup tables of any kind
, the proposed algorithm exploits a simple linear - sequential - circuits based mechanism to map a length-@xmath12 ( binary ) syndrome to _ a _ length-@xmath6 ( quaternary ) candidate vector .
this candidate vector is subsequently fed to a conventional viterbi decoder to help identify the mostly likely error pattern .
the entire process is so cleverly engineered that a _ single _ candidate vector
any one among the pool of @xmath13 possible candidate vectors suffices to locate _ the _ error pattern in the ml sense .
the high efficiency of this algorithm enables us to , for the first time in quantum coding literature , simulate and present qubit error rate performance curves for a general quantum convolutional code .
a stabilizer convolutional code may be viewed as an infinite version of a stabilizer block codes with repeated structure .
the stabilizer group , @xmath14 , for an [ @xmath15 stabilizer convolutional code is given by @xcite : @xmath16 where @xmath17 is the so - called memory parameter , @xmath18 .
, in @xcite was written as @xmath19 , which is either mistaken or a typo . ] and @xmath20 s are required to be independent and to commute with each other .
like classical convolutional codes , quantum convolutional codes also have flexible and adjustable codeword lengths .
the structure of the stabilizer group generators can also be characterized by a semi - infinite matrix @xmath21 .
similar to the stabilizer block codes , each line in @xmath21 represents a generator ( some @xmath20 ) , and each column represents a qubit . as illustrated in fig .
[ fig : semi_infinite ] , @xmath21 has a block - band structure , and two neighboring blocks of generators overlap by @xmath22 qubits , representing the `` actual memory '' of the quantum convolutional code . of a quantum convolutional code.,width=316 ]
[ fig : semi_infinite ] similar to stabilizer block codes , we can label the polynomial generators consisting of pauli elements by @xmath23 polynomial vectors in @xmath24-domain .
the commutation property of stabilizers translates to : @xmath25 where @xmath24 stands for the delay element , similar to that in classical convolutional codes , and @xmath26 denotes the @xmath23 polynomial vectors of a quantum stabilizer code .
the quantum viterbi algorithm discussed in @xcite is the only known general decoder for quantum convolutional codes . a syndrome version of the classical viterbi decoder , the qva operates over pauli - operators ( non - binary input ) .
it starts with a given list of syndromes , and searches for the candidate vectors , which are analogous to the `` received sequence ( codeword ) '' in classical viterbi decoding .
the task is accomplished through table lookup , where the decoder examines the syndromes block by block .
because of the convolutional nature , the output of two neighboring syndrome blocks overlap by @xmath27 qubits , and the table lookup procedure must compare and ensure the overlapping part of the candidate vectors match .
the result is a very long list of candidate vectors all of which correspond to the given syndromes .
every one of these candidate vectors will then be fed into a classical viterbi algorithm , and among them , the most - likely sequence , namely , the error pattern , will be identified .
the overall complexity is rather prohibitive @xcite .
to avoid such a high complexity and still be able to peek into the code s performance , @xcite proposed a simplified decoder that tackles only the most frequent ( but not necessarily the most detrimental ) error patterns , namely , no more than a single error per block .
as such , the algorithm is able to inspect only two consecutive blocks at a time , and hence drastically reduces the size of the lookup table to only 9 terms , at the cost of a compromised decoder performance .
the proposed decoding algorithm roots to the classical syndrome decoding approach for linear codes .
in essence , syndrome decoding is minimum distance decoding , which is equivalent to the maximum likelihood decoding if the channel is discrete memoryless with error probability strictly less than 0.5 .
the linearity of a code allows the code space to be arranged in a special way , termed the _ standard array _ in coding jargon . for an ( @xmath28 )
classical linear block code , the standard array groups all the @xmath29 @xmath6-bit vectors into @xmath30 columns and @xmath31 rows , such that ( i ) the vectors in the same row corresponding to the same syndrome , ( ii ) the first row consists of the set of valid codewords ( starting with the all - zero codeword ) , and ( iii ) the first column consists of all the _ coset leaders _ ,
i.e. correctable error vectors with hamming weights lower than @xmath32 .
each row of the standard array is called a _
coset_. in concept , the function of a syndrome decoder is to identify the coset leader , i.e. the minimum - weight vector , corresponding to a given syndrome , but this seemingly simple function can be extremely expensive to implement in practice .
stabilizer codes in quantum coding are analogous to linear codes in classical coding , and may be tackled through a similar coset representation and syndrome decoding .
the propose decoder here is not only a _
conceptual _ quantum version of the syndrome decoder , but also a _
practical _ mechanism that provides a systematic and concrete way and practical procedure to realize the syndrome decoder with manageable complexity .
the proposed quantum syndrome decoding , which is a low - complexity ml decoding procedure , is depicted in fig .
[ fig : system_model ] .
below we explain why , how , and how efficiently the proposed algorithm works .
we start by transforming the stabilizer generators to a binary polynomial form , and find its equivalent transfer polynomial .
we then detail the step - by - step approach in fig .
[ fig : system_model ] , and demonstrate examples to show how to decode without ( exhaustive ) lookup tables .
[ ht ] here are a few notations used in the discussion .
a quantum convolutional code may be described by its _ semi - infinite stabilizer matrix @xmath21 _ , or _ stabilizer polynomial @xmath33_. the proposed decoding procedure involves the derivation of the equivalent @xmath23 or @xmath34 classical code from the stabilizer polynomial @xmath33 , which are termed the _ equivalent _ @xmath23 or @xmath34 ) _ transfer polynomial _ , and denoted by @xmath35 or @xmath36 , respectively .
the decoding procedure also makes heavy use of useful concepts in classical coding : _
syndrome former _ ( sf ) and _ inverse syndrome former _ ( isf ) . for a given @xmath37 linear code and hence a deterministic standard array
, the sf finds the syndrome associated with a given sequence ( an error pattern ) , and the isf finds _
one _ error pattern any one of the @xmath30 possibilities that associates with a given syndrome .
* decoding algorithm : * the `` syndrome decoder '' block ( dashed box ) in fig . [ fig : system_model ] illustrates the proposed quantum syndrome decoder , whose input is a binary syndrome sequence obtained from the syndrome measurement . 1 .
represent the original quantum convolutional code using its equivalent @xmath23 transfer polynomial , which leads to an equivalent classical convolutional code .
perform syndrome decoding on the equivalent classical convolutional code , through the following two substeps : 1 .
compute the @xmath23 isf from the equivalent @xmath23 transfer polynomial . use the isf to locate an arbitrary candidate vector in the coset corresponding to the given syndrome
2 . compute the @xmath23 generator polynomial @xmath38 from the equivalent @xmath23 transfer polynomial . from the generator polynomial , implement a conventional trellis decoder ( such as the viterbi algorithm or the bcjr algorithm ) , and use it to decode the candidate vector into a valid codeword .
exclusive - or ( xor ) the candidate vector and the valid codeword to get the error pattern .
the output of the decoder is a @xmath39-bit binary sequence , which can be re - assembled to an @xmath6-qubit pauli - formed error pattern . this error pattern
is then fed into the `` recovery '' block in fig .
[ fig : system_model ] and counter - acted to recover the original quantum states . * step 1 : deriving equivalent @xmath23 transfer polynomial * let @xmath33 be the stabilizer polynomial of a quantum convolutional code .
recall that a stabilizer matrix can be described either in the binary @xmath23 form , @xmath40 , where @xmath41 and @xmath42 are symplectically orthogonal , or in a quaternary @xmath34 form , where pauli operators @xmath43 are represented by @xmath34 elements @xmath44 , respectively , conforming to the zero trace inner product ( @xmath45 is a primary element in @xmath34 ) .
a stabilizer code in @xmath34 form takes a similar flavor as an @xmath34 classic code , but the syndrome takes a _ binary _ , rather than quaternary , value . let @xmath46 be a binary row in @xmath47 , and let @xmath48 be a binary error vector . without loss of generality , re - write @xmath49 and @xmath50 in @xmath34 as : @xmath51 , @xmath52 .
the syndrome corresponding to @xmath49 is computed as @xmath53 an immediate and important implication of ( [ eqn : syndrome ] ) is * lemma 1 : * @xcite an equivalent classical linear code to the quantum convolutional code with stabilizer generator @xmath54 has a binary ( @xmath23 ) transfer polynomial : @xmath55 . here equivalent is meant in terms of decoding capability promised by the code ( disregarding possible degeneracy ) .
* step 2 : syndrome decoding for convolutional codes * the problem now boils down to performing syndrome decoding for the equivalent classical convolutional codes . for an ( @xmath56 ) linear code ,
the function of the sf is to map a length-@xmath6 vector to a length-@xmath57 syndrome .
the isf performs the opposite function of sf by locating a length-@xmath6 vector in the coset associated with a given syndrome .
since each coset contains @xmath30 vectors , there are @xmath30 possible outputs an isf may produce , and the resultant vector can be any of the @xmath30 vectors depending on the specific inverse syndrome former being implemented . the choice of the isf will not affect the end - result of the proposed syndrome decoder . efficient and systematic ways exist to implement sf and isf , the simplest of which is through linear transformation or linear circuits . for a linear block code ,
the parity check matrix can serve the role of sf , and the left inverse of the parity check matrix can serve as a matching isf .
the proposed syndrome decoder makes essential use of the functions of sf and isf .
the `` syndrome measurement '' block in fig .
[ fig : system_model ] generates a measure of the syndrome , and the `` syndrome decoder '' block uses a matching isf to acquire a valid vector in the coset associated with that syndrome , and subsequently locates the coset leader after trellis decoding and an xor operation .
finally , the odd - positioned bits and the even - positioned bit in the coset leader are paired up to obtain the tensor product form @xmath58 of the error pattern , which will then be applied to the received quantum sequence ( the `` recovery '' block in fig.[fig : system_model ] , a well - established quantum mechanics procedure ) to get the correct codeword .
* lemma 2 : * the proposed syndrome decoding approach in fig.[fig : system_model ] is valid . _
proof : _ the validity of this decoder is warranted by the fact that the error pattern / coset leader is the minimum - weight vector inside each coset .
suppose that the transmitter sends a valid binary codeword @xmath59 .
the discrete memoryless channel adds some recoverable binary noise @xmath60 to @xmath59 to yield @xmath61 .
that is , @xmath62 , and hence @xmath63 , where @xmath64 denotes binary addition ( xor ) .
the quantum receiver collects @xmath61 , but instead of having a direct measurement of @xmath61 , it gets a measurement of the syndrome @xmath65 .
we show the the combination of isf and trellis decoding in fig .
[ fig : system_model ] successfully deduce @xmath60 .
inverse syndrome former maps @xmath65 to an arbitrary vector @xmath66 in that coset . in the standard array
, @xmath59 locates in the first row ( corresponding to the zero syndrome , @xmath61 , @xmath60 and @xmath66 all locate in the same row corresponding to the syndrome @xmath65 , @xmath60 is the coset leader ( because it is a recoverable error patten ) , and @xmath61 locates in the same column as @xmath59 ( because @xmath67 ) .
hence , @xmath66 is in fact some codeword corrupted by error @xmath60 , namely , there exist some valid codeword @xmath68 such that @xmath69 .
the vector @xmath66 at the output of the isf is fed to the ml trellis decoder .
if the code can support this channel , then with a probability approaching 1 , the trellis decoder is able to deduce the valid codeword @xmath68 from @xmath66 . since @xmath70 , xoring @xmath68 ( output from trellis decoder ) and @xmath66 ( output from isf ) thus produces the target error pattern @xmath60 .
@xmath71 as discussed before , the proposed syndrome decoder in fig .
[ fig : system_model ] involves two key modules , the isf and a conventional trellis decoder for the equivalent classical convolutional code .
the latter is a well - established procedure , given a generator polynomial .
lemma 1 establishes the relation between a stabilizer polynomial @xmath33 of a quantum convolutional code and the transfer polynomial @xmath35 of its equivalent classical linear convolutional code , below we focus the discussion on the derivation of the isf and the generator polynomial for a given @xmath35 .
there exists intrinsic connection between @xmath23 generator polynomial , transfer polynomial , sf and isf .
specifically , consider a binary @xmath72 convolutional code with transfer polynomial @xmath35 .
the transpose of the transfer polynomial , @xmath73 , which has dimension @xmath74 can serve as an sf .
the generator polynomial can be implemented using a linear sequential circuit specified by @xmath75 transfer function @xmath76 with rank @xmath77 satisfying @xmath78 where @xmath79 is a @xmath77-by-@xmath12 all - zero matrix .
the constraint in ( [ equ : sf2 g ] ) guarantees that all the valid codewords are associated with the length-@xmath12 all - zero syndrome @xmath80 and that a set of length-@xmath6 codewords / vectors have the same syndrome if and only if they belong to the same coset .
inverse syndrome former can be obtained by taking the left inverse of the syndrome former , and is therefore denoted by @xmath81 : @xmath82 where @xmath83 is an identity matrix with rank @xmath84 .
the inverse syndrome former may be derived using the partial gaussian elimination or the moore - penrose matrix inverse method .
the latter states that the left inverse of a rectangular matrix @xmath85 , denoted as @xmath86 , takes the form of @xmath87 .
although the moore - penrose matrix inverse was originally derived in real - valued matrix , it can be easily extended to binary matrix by with module-2 computation .
this example illustrates how to obtain the isf and the generator polynomials from a given sf .
supposed the given sf takes of the form of @xmath88 which corresponds to a rate 1/2 ( classical ) recursive convolutional code . using the moore - penrose procedure , a possible isf and generator polynomial
may be obtained : @xmath89 for a given sf , matching isfs and generator polynomials are not unique .
for instance , in the previous example , another possible choice for isf and generator polynomial are : @xmath90 it is desirable to choose a non - catastrophic generator polynomial , and an inverse syndrome former with a low complexity .
* example 2 : * this example illustrates the entire decoding procedure .
consider the quantum convolutional code in @xcite , which has a semi - infinite stabilizer matrix ( in pauli operator form ) : @xmath91 it represents a @xmath92=[3,1,1]$ ] quantum convolutional code with rate 1/3 .
following the proposed syndrome decoding algorithm , the first step is to derive the equivalent @xmath23 transfer polynomials : @xmath93 whose matrix transpose , @xmath94 , represents the syndrome former of a ( @xmath95 ) @xmath23 ( classical ) convolutional code .
the next step is to derive the isf @xmath96 and the generator polynomial @xmath38 matched to the given sf @xmath94 ( all in @xmath23 ) . following the moore - penrose procedure
, we get @xmath97 the resultant decoder diagram is shown in fig .
[ fig : qcc_decoder ] .
the binary syndrome sequence @xmath98 is separated into two streams , and feeds into the inverse syndrome former , which , in this example , is like a recursive convolutional encoder with two streams in and three streams out .
the output streams assemble to a sequence @xmath66 , passes through a bcjr decoder , and generates a sequence which is a valid codeword .
the difference between the codeword sequence and @xmath66 is the error pattern sequence @xmath99 .
@xmath99 is a binary sequence , and will be divided into 2 parts and packed back to pauli operators accordingly .
[ ht ] the proposed decoder is applicable to general quantum convolutional codes on a variety of quantum channel models characterized by memoryless independent pauli operators .
the previous discussion and examples are centered around @xmath23 implementation .
when the quantum convolutional code is derived from an @xmath34-linear codes , such as those proposed in @xcite , one may either find its equivalent binary convolutional code as discussed before , or explore syndrome decoding on its equivalent @xmath34 counterpart .
the latter is possible , because the syndromes , which are typically @xmath57-bit binary sequences , may also be represented by a length @xmath101 @xmath34 sequences in this example .
* example 3 : * consider the rate-1/3 [ @xmath102 quantum convolutional code from example 2 , whose equivalent @xmath34 convolutional code has a @xmath34 transfer polynomial @xmath36 @xmath103 the @xmath34 syndrome former is given by @xmath104
. the isf and the generator polynomial may also be derived in @xmath34 form : @xmath105 clearly , the trellis decoder ( viterbi / bcjr ) in fig .
[ fig : system_model ] can be implemented based on @xmath34 generator polynomial .
the entire syndrome decoding is carried out over a @xmath106 classical @xmath34 convolutional code .
we demonstrate the proposed algorithm by simulation on a memoryless bi - polar pauli error channel , where @xmath8 and @xmath10 occur at probability @xmath107 , and @xmath9 occurs at probability @xmath108 .
[ fig : performance ] notice that the isf is essentially a recursive convolutional code , padding bits are necessary to generate correct @xmath66 for simulations .
for an [ @xmath15 quantum convolutional code , the padding sequence should be a length-@xmath109 @xmath110 sequence .
we simulated the [ 3,1,1 ] quantum convolutional code at a frame length of 900 qubits and tested 10000 frames .
this is like a @xmath111 $ ] stabilizer block with an additional 6-bit padding ( overhead ) , so the exact code rate is @xmath112 .
the resultant performance curve is shown in fig .
[ fig : performance ] .
we have developed a new syndrome decoding approach , which is applicable to a general stabilizer convolutional code on a general pauli - error channel model .
comparing with the existing quantum viterbi decoder and its simplified version , the new algorithm enjoys significantly lower complexity ( no lookup table , and one - time viterbi ) without scarifying ml performance .
the decoding algorithm roots to the classical syndrome decoding theory , and boasts provenly optimal performance .
using this algorithm , we simulated and presented the first performance curve for a general quantum convolutional code . | a novel decoding algorithm is developed for general quantum convolutional codes . exploiting useful ideas from classical coding theory , the new decoder introduces two innovations that drastically reduce the decoding complexity compared to the existing quantum viterbi decoder .
first , the new decoder uses an efficient linear - circuits - based mechanism to map a syndrome to a candidate vector , whereas the existing algorithm relies on a non - trivial lookup table .
second , the new algorithm is cleverly engineered such that only one run of the viterbi algorithm suffices to locate the most - likely error pattern , whereas the existing algorithm must run the viterbi algorithm many times .
the efficiency of the proposed algorithm allows us to simulate and present the first performance curve of a general quantum convolutional code . |
the standard picture of thermal plasmas associated with supernova remnants ( snrs ) is that the state of the gas appears to be either an ionizing plasma ( ip ) for young or intermediate aged snrs ( e.g. kepler s snr , kinugasa and tsunemi 1999 ) , or plasma in collisional ionization equilibrium ( cie ) for old snrs ( e.g. 30 dor c , yamaguchi et al . 2009 ) .
however , recent x - ray observations have revealed several snrs are in recombination - dominated phase ( ic 443 , yamaguchi 2009 ; w49b , ozawa 2009 ; [email protected] , ohnishi 2011 ; w28 , sawada & koyama 2012 ; w44 , uchida 2012 ; [email protected] , yamauchi 2013 ) .
these recombining plasmas ( rps ) have an ionization temperature ( @xmath3 ) higher than the electron temperature ( @xmath4 ) ; @xmath3 is a useful parameter defined to be the temperature which would be required to ionize the plasma to the same degree assuming that the plasma is in cie .
the discovery of rps at snrs requires us to modify the conventional evolution scenario of snrs .
3c 391 ( g31.9 + 0.0 ) is a bright galactic snr displaying diffuse x - ray emission close the center of its radio shell and is classified as a mixed - morphology snr ( mm - snr ) ( reynolds & moffett 1993 ) .
the radio shell with a radius of has the shape of a partial circle in the northwest ( nw ) part and there is a faint extended structure in the southeast ( se ) region .
this suggests a break - out morphology into a lower density region from the main snr shell .
the detection of two oh maser spots at 1720 mhz from 3c 391 by frail ( 1996 ) indicates an interaction with molecular clouds .
they also reported a possibly extended oh emission at the edge of the nw shell .
the velocity of two maser spots ( 105 and 110 km s@xmath5 ) agrees with a previous study of h absorption velocity ( radhakrishnan 1972 ) .
h observations show that the distance to 3c 391 is between 7.2 kpc and 11.4 kpc , assuming a distance to the galactic center of 8.5 kpc .
the surrounding molecular gas of 3c 391 can be traced by co ( j @xmath6 1@xmath70 ) emission at velocities from 90 to 110 km s@xmath5 ( wilner 1998 ) .
the results of the co observation support the early expansion of 3c 391 inside a dense molecular cloud and the breakout of the blast wave to the se region .
knowledge of the astrophysical environment around snrs is of great importance in order to reveal the origin of the rps in the context of the snr evolution .
interestingly , most of the rp snrs are mm - snrs that interact with molecular clouds and are associated with bright gev gamma - ray emissions ( ic 443 , abdo 2010a ; w28 , abdo 2010b ; w49b , abdo 2010c ; w44 abdo 2010d ) .
while some young snrs show bright tev gamma - rays ( rx j1713.7 - 3946 , aharonian 2004 ) , middle - aged snrs are usually found to be predominantly emitting in the gev energy band . a significant excess ( 13@xmath8 ) of gev gamma - rays at the position of 3c 391 with fermi - lat
was reported by castro & slane ( 2010 ) .
the radio rim of the nw region and the closer oh maser spot are apart from the gev emission peak in the test statistic map .
the gev emission associated with 3c 391 probably arises from accelerated cosmic rays in the shock - heated molecular clouds .
the interaction with the molecular gas and the association of the gamma - ray emission would be the keys for understanding the morphology and spectrum of the thermal plasmas in rp snrs . in this paper , we report on an x - ray study of snr 3c 391 , taking advantages of low and stable in - orbit background and high sensitivity of the suzaku x - ray observatory ( mitsuda 2007 ) .
we construct a model for the galactic background emission to complement the poor photon statistics , then we fit several plasma models to the x - ray spectrum of 3c 391 .
our observation data is a part of the suzaku key project to search for rps in galactic snrs .
section 2 describes the observation and data reductions and section 3 presents the data analysis .
then , we interpret our results in section 4 and conclude in section 5 .
we observed snr 3c 391 for an exposure time of 100 ks with the suzaku satellite in 2010 october 2224 during the ao5 cycle .
this observation was performed as one of the key project observations aimed at systematic search for rps .
the pointing position was ( @xmath9 , @xmath10 ) = ( 282.3812 , @xmath20.9417 ) . the observation instrument we used is the x - ray imaging spectrometer ( xis ; koyama 2007 ) , x - ray charge coupled device ( ccd ) cameras on the focal plane of the x - ray telescope ( xrt ; serlemitsos 2007 ) .
xis0 and 3 are front - illuminated ( fi ) ccds , and xis1 is a back - illuminated ( bi ) ccd . the xis was operated with the normal - clocking , full - frame mode . since
in - orbit radiation damage has degraded the charge transfer efficiency of the xis , the spaced - row charge injection technique is applied for the xis .
we reprocessed the xis data with the latest calibration results instead of using the cleaned events .
we used the headas software of version 6.13 ( suzaku software of version 20 ) .
the calibration database ( caldb ) released on 2013 march was used for data processing .
we adopted standard event selection criteria for suzaku xis data processing .
we excluded events in the intervals of the south atlantic anomaly , low - earth elevation angles less than 5 degrees and the day earth elevation angle less than 20 degrees .
after the data screening , the effective exposure time is about 99.4 ks .
the data of the two editing mode , 3@xmath113 and 5@xmath115 pixel events , were combined , and then we removed flickering and hot pixels of the xis .
x - ray diffuse emission is clearly detected in the central region of the remnant and is encompassed by the radio rim in the nw .
figure [ fig : image xis3 ] shows an x - ray image of 3c 391 ( xis3 only ) , after correcting the vignetting effect and subtracting the instrumental background ( the so - called non x - ray background , or nxb ) .
the nxb was estimated by using xisnxbgen ( tawa 2008 ) .
8 pixels ( @xmath11 ) were summed and smoothed with a gaussian kernel of @xmath12 .
, width=264 ] the northwest ( nw ) region associated with the radio rim has harder x - rays than the southeast ( se ) region .
it indicates the difference of the thermal plasma or circumstellar matters between those regions .
we thus divided the image into two regions : the nw and se region .
3c 391 is located on the galactic plane , and hence the major background component is the galactic ridge x - ray emission ( grxe ) , whose x - ray flux strongly depends on the position ( e.g. uchiyama 2011 , 2013 ) . since the photon statistics of the background region ( surrounding region of 3c 391 in the same xis field ) is not good enough due to the limited exposure time and the effective area , subtracting the background data from the source data results in large statistical errors in the snr spectrum .
we therefore constructed a background model for the grxe .
we extracted the background spectrum from the region excluding 3c 391 in the same field of view , and fitted the grxe model compiled by uchiyama ( 2013 ) to the spectrum . in this background region , a few percent level of photon contamination from 3c 391 , which is due to the leakage of the xrt point spread function into the background region , can not be neglected . in order to construct the background model , we need to make phenomenological models both for the source and the background regions .
we extracted spectra from the whole 3c 391 region and the background region after subtracting the nxb .
after the vignetting correction , we subtracted the background spectrum from the 3c 391 spectrum .
the redistribution matrix file ( rmf ) and the ancillary response file ( arf ) were created by xisrmfgen and xissimarfgen , respectively ( ishisaki 2007 ) .
we used the spectral analysis software xspec version 12.8.1 ( arnaud 1996 ) .
we simultaneously fitted all the xis0 , 1 , and 3 data with a single model .
the energy band from 1.7 to 1.8 kev were ignored because of the calibration uncertainty around the si k edge .
we then modeled the spectra by using a phenomenological model : the continuum x - ray emission is described by bremsstrahlung with temperature of @xmath13 and line emissions from various ionized atoms are given by gaussian lines .
the interstellar photoelectric absorption was estimated using the cross sections by morrison and mccammon ( 1983 ) with h column density @xmath14 .
this phenomenological model successfully reproduces the data ( @xmath15 404.01/307 ) .
figure [ fig : spec phenom ] and table [ tbl : par phenom ] show the best - fit spectrum and parameters , respectively .
.parameters of the phenomenological model.@xmath16 [ cols="<,^,^,^",options="header " , ] @xmath16 the uncertainties are the 90% confidence range .
+ @xmath17 volume emission measure @xmath18 at the distance of 8 kpc , where @xmath19 , @xmath20 and @xmath21 are the electron and proton number densities ( @xmath22 ) and the emitting volume ( @xmath23 ) , respectively .
+ @xmath24 the units are 10@xmath25 photons @xmath26 s@xmath5 arcmin@xmath27 .
the uncertainties are the 1 @xmath8 confidence range .
+ [ tbl : par dist ]
the x - ray spectrum of 3c 391 is represented by two plasma models : a low temperature cie and a high temperature rp .
the cie plasma is probably emitted by the ism heated by the blast waves , because it has 1 solar abundances .
the abundances of the rp are given in figure [ fig : mass ] , together with some theoretical model of ia ( iwamoto et al .
1999 ) and core - collapse supernova ( cc sn ; woosely and weaver 1995 ) .
although the error is large , the abundance pattern except fe is similar to a cc sn of progenitor mass of @xmath2815 .
thus we regard that the rp has an ejecta origin from a cc sn of a @xmath2815 progenitor star . in section 4.1 , we discuss consistency of the observed data with this assumption .
the abundance of fe in ejecta largely depends on the mass cut of the cc sn .
if a major fraction of fe would be captured into a collapsed object ( neutron star or black hole ) in the case of 3c 391 , a small abundance of fe in the ejecta would be reasonable , unless the explosion is highly asymmetric as is demonstrated by yasumi et al .
another possibility is that only a small fraction of the fe - rich ejecta has been shocked by the reverse shock ( fe - rich ejecta are expected to be located at the very inner part of the remnant ) . since the fe abundance estimated using the fe - l lines only has a large uncertainty due to their incompleteness , we can not conclude why the abundance of fe is lower than expected .
the recombination parameter ( @xmath29 @xmath22 s ) is the largest among any rp discovered so far in mm - snrs ( yamaguchi 2009 , 2012 , ozawa 2009 , ohnishi 2011 , sawada & koyama 2012 , uchida 2012 , yamauchi 2013 ) . the spectra of the se and nw also show the rp . since the extended oh emission reported in the nw region suggests the interaction with molecular clouds , the slightly higher @xmath14 than the se would be reasonable .
on the other hand , @xmath30 and @xmath31 show no significant difference between the two regions .
we found a hint of line emission at 6.4 kev , fe k@xmath0 line at @xmath1 level from the se region , and not from the nw region where the snr interacts with molecular clouds .
however the se region is coincident with oh masers .
one possibility is that the fe k@xmath0 line is due to low energy cosmic - ray electrons , which are preferentially made in the se region .
the gyro radius of the lower energy electrons is extremely small , and hence diffusion length is far smaller than the snr size .
the absence of the fe k@xmath0 line at the nw , where cool gas is denser than the se , would be due to this limited diffusion length .
deeper observations of hard x - ray will probably detect bremsstrahlung by low energy cosmic - rays interacting with molecular clouds . in any case , the detection of fe k@xmath0 line is marginal , thus we need more detailed observations in the future .
the abundances and emission measure of rp given in table 3 are based on the h dominant plasma .
we assume that the rp originates from an ejecta of @xmath2815 progenitor ( figure [ fig : mass ] ) , which indicates super - solar abundances even for the light elements such as he , o and ne . the estimations of emission measure and abundances of heavy elements largely depend on the abundances of he , o and ne ; large abundances of these light elements enhance the continuum ( mainly bremsstrahlung ) flux , although the relative spectral shape is not affected because emission lines of he ne are absent in the relevant energy band of 1.27.0 kev .
we thus re - estimate the emission measure and abundances of heavy element assuming the light element mass distribution is the same as the progenitor star of 15 .
the abundances of mg ca relative to si in the rp component do not change from the original h dominant plasma but the averaged abundance of si fe relative to hydrogen becomes @xmath2815 solar , confirming its ejecta origin .
the emission measure becomes @xmath32 @xmath22 , where @xmath33 and @xmath34 are a filling factor and the snr distance in the unit of 8 kpc , respectively .
the ratio of the electron to the atomic hydrogen densities is @xmath35 and the number ratio of all the elements to hydrogen is 2.0 . adopting the source radii of @xmath36 pc ,
the ejecta mass is estimated to be @xmath37 . as is seen in figure 1 of chen 2004
, filling factor would be very small , hence we assume @xmath33 is in the range of 0.30.1 . then the ejecta mass is @xmath28 1025 .
this value is consistent with the initial assumption that 3c 391 is a remnant of cc sn with a progenitor mass of @xmath28 15 .
the ionization age of the rp is estimated using @xmath38 .
the electron density @xmath19 is @xmath39 @xmath22 .
then the elapsed time @xmath40 is @xmath41 ( @xmath42 ) years , which place this snr as a rather old snr .
if we assume a small filing factor , the age of @xmath43 yr estimated by chen & slane ( 2001 ) has a very similar value .
several scenarios are proposed to explain the formation of rps in snrs .
one is the rapid electron cooling by the thermal conduction from cold matters around snrs ( kawasaki 2002 ) or the adiabatic rarefaction ( itoh and masai 1989 , shimizu 2012 ) .
the other is the ionization by the suprathermal or non - thermal particles ( masai 2002 , ohnishi 2011 ) or the high energy photons ( kawasaki 2002 ) .
therefore , the presence of rp in 3c 391 means that the plasma experienced the rapid electron cooling or extra - ionization .
we first discuss the electron cooling scenario .
if the electron cooling is due to thermal conduction , a timescale of the thermal conduction is @xmath44 where @xmath45 is the temperature gradient scale length ( kawasaki et al .
this timescale is much longer than the estimated age of 3c 391 , therefore the thermal conductivity is unlikely to produce the whole rp in 3c 391 of which the radius is about 7 pc .
next scenario is rarefaction , which is the possible origin of the rp in w28 ( sawada & koyama 2012 ) .
if the progenitor of 3c 391 was a massive star with a strong stellar wind activity , there can be a dense circumstellar media ( csm ) surrounding the ejecta .
the blastwave breaking out of the shell can lead to rarefaction of the pi or near cie plasma .
thereafter the electron temperature can drop rapidly due to adiabatic expansion but the ionization states can stay more or less the same . this scenario is supported by the fact that the recombination timescale is close to the putative age of 3c 391 . while reynolds and moffett ( 1993 ) estimated that the shell break - out " occurred at about 1600 yr after the supernova , the blast wave breaks out of the csm at only @xmath28100 yr ( itoh and masai 1989 , shimizu 2012 ) .
since no significant spectral differences except the possible presence of a 6.4 kev line in the nw and a slightly larger @xmath46 in the se , we argue that the rarefaction of ejecta plasma was roughly symmetrical , unlike w49b as suggested by some models ( e.g. miceli et al 2010 ) .
our results show that the rp of 3c 391 was produced by the break - out " from the csm and evolved independently from the interstellar environment . for 3c 391
, we apply a model of an early phase break - out ( shimizu et al .
2012 ) , while zhou et al .
2011 proposed a later phase break - out for w49b .
this difference is due to their different circumstellar environments ; 3c 391 would have a denser and more compact csm compared to that of w49b .
although the extra - ionization process could be an alternative interpretation , it is still impossible to identify the origin of rp with the present data alone .
we hence encourage deeper observation of 3c 391 .
we observed the middle - aged snr 3c 391 with suzaku for 100 ks .
we generated a background model for the grxe and analyzed the x - ray spectrum of 3c 391 .
our conclusions are summarized below : * the x - ray spectrum of 3c 391 is well described by a cie+rp model .
3c 391 has the highest @xmath30 value among the rp snrs ever discovered .
this means the rp of 3c 391 approaches ionization equilibrium the most .
* estimated progenitor mass of @xmath2815 roughly agrees with the x - ray emitting mass of the rp derived from the corrected emission measure .
this means only the ejecta plasma is overionized .
* the parameters of rp have no significant difference between the nw and se regions .
thermal conduction is unlikely to be the origin of the rp at 3c 391 .
the symmetrical evolution of rp is consistent with a rarefaction of the plasma in the early phase . in order to draw a more conclusive picture
we need deeper observations with more sensitive instruments , such as those to be onboard the astro - h mission ( takahashi 2010 ) . *
our detailed analysis of the spectra of 3c 391 suggests a hint of fe k@xmath0 emission at 6.4 kev ( @xmath47 ) , which can be explained by the interaction with a molecular cloud . | recent x - ray study of middle - aged supernova remnants ( snrs ) reveals strong radiative recombination continua ( rrcs ) associated with overionized plasmas , of which the origin still remains uncertain .
we report our discovery of an rrc in the middle - aged snr 3c 391 . if the x - ray spectrum is fitted with a two - temperature plasma model in collisional ionization equilibrium ( cie ) , residuals of si ly@xmath0 line at 2.006 kev , s ly@xmath0 line at 2.623 kev and the edge of rrc of si at 2.666 kev are found .
the x - ray spectrum is better described by a composite model consisting of a cie plasma and a recombining plasma ( rp ) .
the abundance pattern suggests that the rp is associated to the ejecta from a core - collapse supernova with a 15 progenitor mass .
there is no significant difference of the recombining plasma parameters between the southeast region and the northwest region surrounded by dense molecular clouds .
we also find a hint of fe k@xmath0 line at 6.4 kev ( @xmath1 detection ) from the southeast region of the snr . |
hadron properties should be obtained from the qcd lagrangian as a function of the coupling constant @xmath0 and of the quark masses @xmath1 . in practice ,
things are made complicate by qcd being a strongly coupled theory in the low energy region . at the scale @xmath2 ,
nonperturbative effects become dominant and @xmath0 becomes large .
the nonperturbative qcd dynamics originates the confinement of quarks inside hadrons .
typical approaches include , on one hand the use of phenomenological potential models and constituent quark model descriptions , on the other hand first principles lattice simulations ( still far from the physical parameter window in many cases ) .
however , the physics of systems with a heavy quark @xmath3 allows some simplification .
the quark mass scale @xmath4 is large , bigger than @xmath2 .
then @xmath5 is small and perturbative expansions may be performed at this scale .
bound systems made of two or more heavy quarks are even more interesting @xcite .
they are nonrelativistic systems characterized by another small parameter , the heavy - quark velocity @xmath6 , and by a hierarchy of energy scales : @xmath4 ( hard ) , the relative momentum @xmath7 ( soft ) , and the binding energy @xmath8 ( ultrasoft ) . for energy scales close to @xmath2 ,
perturbation theory breaks down and one has to rely on nonperturbative methods . regardless of this , the nonrelativistic hierarchy @xmath9 persists also below the @xmath2 threshold . while the hard scale is always larger than @xmath2 , different situations may arise for the other two scales depending on the considered quarkonium system .
the soft scale , proportional to the inverse typical radius @xmath10 , may be a perturbative ( @xmath11 ) or a nonperturbative scale ( @xmath12 ) depending on the physical system .
the ultrasoft scale may still be perturbative only in the case of @xmath13 threshold states .
taking advantage of the existence of a hierarchy of scales , one can introduce nonrelativistic effective field theories ( nr efts ) @xcite to describe heavy quarkonia .
a hierarchy of efts may be constructed by systematically integrating out modes associated to high energy scales not relevant for quarkonium .
such integration is made in a matching procedure enforcing the equivalence between qcd and the eft at a given order of the expansion in @xmath6 ( @xmath14 for @xmath15 , @xmath16 for @xmath17 , @xmath18 for @xmath13 ) .
the eft realizes a factorization at the lagrangian level between the high energy contributions , encoded into the matching coefficients , and the low energy contributions , carried by the dynamical degrees of freedom .
poincar symmetry remains intact in a nonlinear realization at the level of the nr eft and imposes exact relations among the matching coefficients @xcite . by integrating out the hard modes one obtains nonrelativistic qcd @xcite .
nrqcd is making explicit at the lagrangian level the expansions in @xmath19 and @xmath20 .
it is is similar to hqet , but with a different power counting and accounts also for contact interactions between quarks and antiquark pairs ( e.g. in decay processes ) , hence having a wider set of operators . in nrqcd
soft and ultrasoft scales are dynamical and their mixing may complicate calculations , power counting and do not allow to obtain a schrdinger formulation in terms of potentials .
one can go down one step further and integrate out the soft scale in a matching procedure to the lowest energy eft that can be introduced for quarkonia , where only ultrasoft degrees of freedom remain dynamical .
such eft is called potential nonrelativistic qcd ( pnrqcd ) @xcite . in this case
the matching coefficients encode the information on the soft scale and are the potentials .
pnrqcd is making explicit at the lagrangian level the expansion in @xmath21 .
this eft is close to a schrdinger - like description of the bound state .
the bulk of the interaction is carried by potential - like terms , but non - potential interactions , associated with the propagation of low - energy degrees of freedom ( @xmath22 colour singlets , @xmath22 colour octets and low energy gluons ) , are generally present .
they start to contribute at nlo in the multipole expansion of the gluon fields and are typically related to nonperturbative effects @xcite . in this eft frame , it is important to establish when @xmath2 sets in , i.e. when we have to resort to non - perturbative methods . for low - lying resonances ,
it is reasonable to assume @xmath23 .
the system is weakly coupled and we may rely on perturbation theory , for instance , to calculate the potential . in this case , we deal with weak coupling pnrqcd .
the theoretical challenge here is performing higher - order perturbative calculations and the goal is precision physics ( see e.g. @xcite ) .
for higher resonances @xmath24 . in this case
, we deal with strongly coupled pnrqcd .
we need nonperturbative methods to calculate the potentials and one of the goal is the investigation of the qcd low energy dynamics .
pnrqcd @xcite realizes modern renormalization theory in the context of simple nonrelativistic quantum mechanics @xcite . in this framework
the schrdinger equation is exactly the equation to be solved to get the binding .
the @xmath22 potentials to be used in such equation are the wilson matching coefficients of pnrqcd obtained by integrating out from qcd all degrees of freedom but the ultrasoft ones . if the quarkonium system is small , the soft scale is perturbative and the potentials can be _ entirely _ calculated in perturbation theory @xcite .
they undergo renormalization , develop a scale dependence and satisfies renormalization group equations , which eventually allow to resum potentially large logarithms .
since the degrees of freedom that enter the schrdinger description are in this case both @xmath22 color singlet and @xmath22 color octets , both singlet and octet potentials exist . at the moment
, the static singlet @xmath22 potential is known at three loops apart from the constant term .
the first log related to ultrasoft effects arises at three loops .
such logarithm contribution at n@xmath25lo and the single logarithm contribution at n@xmath26lo may be extracted respectively from a one - loop and two - loop calculation in the eft and have been calculated in @xcite .
the singlet static energy , given by the sum of a constant , the static potential and the ultrasoft corrections , is free from ambiguities of the perturbative series ( renormalon ) . by comparing it ( at the nnll ) with lattice calculations of the static potential
one sees that the qcd perturbative series converges very nicely to and agrees with the lattice result in the short range and that no nonperturbative linear ( `` stringy '' ) contribution to the static potential exist @xcite .
the @xmath27 singlet potential is known at two loops , the @xmath28 singlet spin - dependent and spin - indepent potentials are known at one loop ( see the discussion and the original references quoted in @xcite ) . if the quarkonium system is large , the soft scale is nonperturbative and the potentials can not be entirely calculated in perturbation theory @xcite .
they come out factorized in the product of nrqcd matching coefficients and low energy nonperturbative parts given in terms of wilson loops expectation values and field strengths insertions in the wilson loop . the full expression for the qcd potentials up to order @xmath29 has been obtained in @xcite , for the @xmath30 and @xmath31 case see @xcite .
such expressions correct and generalize previous findings in the wilson loop approach @xcite that were typically missing the high energy parts of the potentials , encoded into the nrqcd matching coefficients and containing the dependence on the logarithms of @xmath4 , and some of the low energy contributions .
poincar invariance establishes exact relations between the potentials @xcite of the type of the gromes relation between spin dependent and static potentials @xcite . in this regime , _ from pnrqcd
we recover the quark potential singlet model_. however , here the potentials are calculated from qcd by nonperturbative matching .
their evaluation requires calculations on the lattice @xcite or in qcd vacuum models @xcite .
recently lattice calculations have reached a high degree of precision @xcite and well defined predictions on the behaviour of the @xmath27 and @xmath28 potentials in the confining region became possible .
such behaviour seems to deviate from a flux tube picture @xcite .
since now precise lattice data on the long distance behaviour of the potentials up to order @xmath28 are available , one should relate them to qcd vacuum model predictions on expectation values of wilson loops and field strengths wilson loop insertions .
these are gauge invariant objects containing information on the field distribution between the quarks that go well beyond the area law content .
they give us an appropriate way to `` measure '' and characterize the confinement mechanism .
for this reason it would be interesting to get predictions on the behaviour of this objects also from string theory .
when the soft scale is perturbative the energy levels are given by the expectation value of the perturbative potentials , calculated at the needed order of the expansion in @xmath0 , plus nonpotential nonperturbative contributions @xcite .
the latter start to contribute to the energy levels at order @xmath32 .
they are retardation effects and are systematically accounted for in the eft .
they enter energy levels ( and decay widths ) in the form of local or nonlocal electric and magnetic condensates .
we still lack a precise and systematic knowledge of such nonperturbative purely glue dependent objects . it would be important to have for them lattice determinations or data extractions .
inside pnrqcd it is possible to relate the leading electric and magnetic nonlocal correlators to the gluelump masses and to some existing lattice ( quenched ) determinations @xcite .
however , since the nonperturbative contributions are suppressed in the power counting , it is possible to obtain precise determinations of the masses of the lowest quarkonium resonances with purely perturbative calculations , in the cases in which the perturbative series is convergent ( i.e. after that the appropriate subtractions of renormalons have been performed ) and large logarithms are resummed . in this
framework power corrections are unambiguously defined . for
a review of the predictions on the lowest quarkonium resonances obtained in this framework see @xcite .
when the soft scale is nonperturbative the energy levels are given by the expectation values of the nonperturbative potentials described in the previous section .
_ nonpotential ( or retardation ) corrections do not exist in this case_. a full phenomenological application , taking into account both the nrqcd matching coefficients and the recent lattice evaluation of the low energy part of the potentials , has not yet been performed and would be needed . in both cases , the eft supply us with a proper and well defined quantum mechanical framework to perform systematic calculations of the quarkonium spectrum .
in particular : * there is a well defined power counting that states the terms that should be treated as ( quantum mechanical ) perturbation .
* in higher - order calculations , quantum mechanical perturbation theory requires regularization and renormalization .
pnrqcd gives us a well defined and field theory derived quantum mechanical framework to calculate perturbative corrections .
in particular , the soft uv divergences in the potential cancel against nrqcd hard matching coefficients @xcite and potential uv divergences in quantum mechanical perturbation theory cancel against nrqcd hard matching coefficients @xcite , leaving well behaved and scale independent predictions for physical quantities .
then , in this scheme no divergences arise from e.g. the iteration of the spin - spin potential in quantum mechanical perturbation theory , as it happens typically in phenomenological potential model approaches .. * spectra are function only of the standard model parameters .
conversely one can use the spectra in order to extract @xmath0 and @xmath4 @xcite .
while the real parts of the pnrqcd matching coefficients give us the potentials , the imaginary parts give us the inclusive decays widths @xcite .
also transitions may be worked out in pnrqcd . in @xcite
the m1 transition rates for the lowest quarkonia resonances has been calculated and a value for @xmath33 compatible with the experimental data has been obtained . -0.3truecm
the results on the nonperturbative potentials that we have discussed are valid away from threshold and in the case in which hybrids develop a mass gap of order @xmath2 with respect to singlet states @xcite , as the lattice indicates . for states near or above threshold a general systematic eft approach has still to be developed , while lattice results on excited resonances are just appearing .
most of the existing analyses , especially for the many new states discovered recently at the b factories , have to rely on a formalism based on potential models and coupled channels , see e.g. @xcite and refs . therein .
a. pineda , j. phys .
g * 29 * , 371 ( 2003 ) .
n. brambilla , a. pineda , j. soto and a. vairo , phys .
b * 470 * , 215 ( 1999 ) .
b. a. kniehl , a. a. penin , v. a. smirnov and m. steinhauser , nucl .
b * 635 * , 357 ( 2002 ) e. eichten and f. feinberg , phys .
d * 23 * , 2724 ( 1981 ) ; a. barchielli , n. brambilla and g. m. prosperi , nuovo cim .
a * 103 * , 59 ( 1990 ) ; n. brambilla , p. consoli and g. m. prosperi , phys .
d * 50 * , 5878 ( 1994 ) a. barchielli , e. montaldi and g. m. prosperi , nucl .
b * 296 * , 625 ( 1988 ) | we briefly review how nonrelativistic effective field theories give us a definition of the qcd potentials and a coherent field theory derived quantum mechanical scheme to calculate the properties of bound states made by two or more heavy quarks . in this framework
heavy quarkonium properties depend only on the qcd parameters ( quark masses and @xmath0 ) and nonpotential corrections are systematically accounted for .
the relation between the form of the nonperturbative potentials and the low energy qcd dynamics is also discussed . _ em _ |
many basic features of the interactions between light composite particle are now well established . in particular , a good description @xcite-@xcite , of the interactions of light nuclei such as @xmath7 , @xmath8+@xmath9he , @xmath1he+@xmath1he etc .
is obtained from a deep attractive potential with pauli - forbidden states with the addition of parity dependence , young scheme splitting and a spin - orbit interaction .
importantly , this type of interaction is now justified from microscopic considerations , both in quasi - classic picture @xcite and also in quantum mechanical shell - model framework @xcite .
these results lead to a general comprehensive understanding of the relations between various interaction models which appear , at a first glance , to contradict to each other .
nevertheless , a number of finer features of the interaction , notably the role of the dynamic polarisation of loosely bound projectiles such as @xmath0 or @xmath10li when combined with antisymmetrization effects , are not yet fully understood despite many previous efforts .
the existing problems and contradictions are illustrated by the following example .
on the one hand , it is well known ( see e.g. @xcite,@xcite-@xcite ) that antisymmetrization and coupled - channel effects in composite particle scattering unavoidably result in complex nonlocal and energy - dependent potentials , in which the breakup channels ( with their various final angular momenta ) lead to rather peculiar contributions in @xmath11-space and the specific energy dependence . on the other hand ,
for a long time it has been standard practice to describe the elastic scattering of deuterons and nuclei such as @xmath2li by the standard optical model ( via local potentials ) with a smooth energy dependence , or even by global optical models @xcite .
the low energy @xmath12he interaction presents a further example of these problems .
an attempt undertaken long ago by satchler _
et al _ , @xcite , to describe @xmath12he phase shifts for @xmath1320 mev by a standard optical potential of woods - saxon form , required a significantly energy dependence in the _ radial form _ of the potential ( i.e. of geometric parameters ) .
further studies based on @xmath14-matrix to potential inversion , @xcite , showed that the @xmath12he phase shifts can be described excellently over a wide energy range ( @xmath1365 mev ) with a gaussian - like potential with odd - even splitting and with a small energy dependence in the potential depth alone .
this odd - even splitting effect is a direct consequence of antisymmetrization , @xcite .
thus the artificial energy dependence found by satchler _ et al _ @xcite reflects only the effect of omitting the odd - even splitting in the @xmath12he potential .
similar problems are also expected to arise for @xmath0+@xmath1he and analogous systems .
moreover , many previous studies of deuteron scattering have demonstrated @xcite the strong contribution of deuteron breakup channels , in particular with @xmath15=2 , i.e. through the excitation of the @xmath3-state in n - n subsystem .
we can then ask , in what way can the above strong nonlocal contribution be incorporated into , say , the double folding potential , which , from a first glance , contains no such effects ? ( in this connection we must emphasise that the usual type of nonlocalities considered in literature @xcite differs from the energy - singular terms arising when the virtual breakup channels are excluded ) .
an important problem then arises to formulate the above nonlocal and specific energy dependent effects into a language of `` exact '' optical potentials . by this term we do not mean a standard phenomenological optical potential with a prescribed form ( and fitted parameters ) but
instead some `` exact '' potential which is reconstructed by inversion from the data or from microscopic theory .
our plan of study is as follows . as a first approach iterative perturbative ( ip ) inversions
are presented , based on phase - shift analyses of experimental data .
these inversions follow ref .
@xcite , in directly calculating an explicit majorana exchange force , instead of applying separate inversions for even and odd partial amplitudes as in previous determinations of the empirical @xmath0 + @xmath16 potential , @xcite . in this way
the even and odd partial amplitudes are treated simultaneously and the inversion process is considerably stabilised .
in the second approach , an extended rgm - study is presented for the same system in order to gain a deeper understanding of the complicated interplay between antisymmetrization and projectile breakup effects . increasing the number of coupled channels in these rgm calculations leads to three sets of microscopic models for the @xmath17he interaction : 1 . the direct @xmath17he potential without incorporation of any exchange effects and which is related to standard double folding ( df ) models ( see e.g. @xcite ) ; 2 . the one channel ( @xmath18he ) rgm model with full antisymmetrization but with a `` frozen '' deuteron and 3 . the multichannel rgm model inclusive of breakup channels .
rgm potentials are determined by inversion from the phase shifts of the last two models .
the comparison of these different potential models then yields the contributions of antisymmetrization and deuteron breakup effects in terms of the local potential .
our approach has one further advantage , since the considerable flexibility of the inversion method allows the determination of potentials , which describe the system under study not at a single energy but for a wide energy range simultaneously ( previously denoted as `` mixed case '' inversion @xcite ) .
the further comparison of our new parity dependent solutions obtained from the empirical data set and the rgm phase shifts allows _ quantitative _ deductions to be made on the validity of the df model for the deuteron projectile ( as applied in @xcite for example ) .
the final part of our study is dedicated to a reconstruction of the tensor part of the @xmath0-@xmath1he interaction . here
we study the range and depth of the @xmath0-@xmath1he tensor force and compare it with the same quantities for central and spin - orbital components of the same system .
the contents of this work are then as follows . in section
[ invs - sect ] we describe , in brief , our inversion method and discuss the importance of including odd and even partial waves simultaneously in the inversion .
section [ emp - sect ] describes the empirical input data used and presents a new solution for the empirical case .
section [ rgm - sect ] is devoted to inversion of rgm calculations for @xmath0+@xmath1he , which include coupling to the virtual breakup channels .
section [ dfm - sect ] contains a detailed discussion of validity of the df model . in section
[ tens - sect ] , we describe the calculation of a tensor @xmath0-@xmath1he potential which nicely fits the @xmath2li quadrupole moment and the tensor mixing parameter @xmath19 .
our findings are summarised in a concluding section .
the inversion method used in the next sections has been developed in moscow ( in the moscow state university ) @xcite and independently in the open university by mackintosh and coworkers , ( originally in ref .
@xcite , with further developments and references in ref .
it has been named as the iterative - perturbative ( ip ) method by the latter group and as the linearised iterative method by the moscow group .
the two approaches have only minor differences which relate to the choice of inversion basis and details of the iteration process .
the overall method is extremely flexible and convenient to use .
it enables us to reconstruct many types of potentials , central , spin - orbit and tensor , real and complex , both from phase shifts or directly from scattering observables .
the appropriate input quantities , e.g. @xmath14-matrix , differential cross sections , analysing powers etc .
are nonlinear functions of the interaction potential and so may be considered as a functional response to variations in the interaction potential . in general , excluding some special cases , a small variation of the interaction potential is directly related to a small variation of the @xmath14-matrix or other response function . the basic idea behind the approach is then a local linearization of the response function in neighbourhood of a given point in an appropriate functional space .
this local linear approximation , together with an expansion of the unknown potential in some complete basis ( for example the fourier expansion in an orthogonal basis , @xcite although gaussian functions worked well in inversions for @xmath20 + @xmath16 , @xcite ) , results in a set of linear algebraic equations to be solved to give the expansion coefficients .
the initial approximation ( i.e. initial values of the expansion coefficients or a starting reference potential ) is usually chosen based on physical considerations .
in particular , as will be shown in later sections , the results of the present paper suggest that a good choice for the initial potential would be the df potential . the initial potential is then corrected with step - by - step iterations .
the iteration procedure converges rapidly ( 5 7 iterations at most are usually required ) and gives a stable solution for the first expansion coefficients of the sought potential , where the number of reconstructed coefficients depends directly on the completeness and consistency of the input data .
the approach developed by the moscow group has also been successfully applied , @xcite to reconstruct interactions in the field of heavy ion scattering in systems such as @xmath21c@xmath22c , @xmath23o@xmath24o etc .
the algorithm developed at the open university has also been applied to a wide range of systems ; for example to @xmath20 + @xmath23o directly from experimental observables , @xcite , to mixed case @xmath25-nucleus phase shifts calculated from non - local interactions , @xcite , and to single energy empirical @xmath14-matrices , for nuclei such as @xmath23o@xmath26o , @xcite , and for @xmath27li scattering , @xcite .
the presence of well expressed low - energy resonances in the @xmath28=2 partial waves and near the @xmath14-matrix pole in the @xmath14-wave has been shown previously @xcite to render the solution of the inverse problem for the above channels so reliable that the data for even a small energy interval 0 - 5 mev is sufficient to give quite a correct reconstruction of @xmath0-@xmath1he interaction potential in even partial waves . in sharp contrast to this ,
the odd partial amplitudes are derived from the phase shift analysis with large errors due to a low sensitivity of majority of observables ( cross sections and tensor analysing powers ) to these odd - parity partial phase shifts . as a result , the odd parity solution of the inversion has large errors which become more enhanced at the higher energies @xcite .
hence , when the odd and even partial amplitudes are treated _ independently _ in the inversion @xcite , the magnitude and radial form of the majorana interaction term can not be established with sufficient accuracy . however
, the structure of this term is very important for our understanding of the role and significance of exchange effects in nucleus - nucleus interactions .
the general form of operator ( or potential ) of particle interaction can be expressed as follows : @xmath29 where @xmath30 is the wigner type interaction which includes the central , spin - orbit and tensor interaction terms , i.e. , @xmath31 @xmath32 is the majorana type interaction which includes , in general , similar terms , i.e. central , spin - orbit and tensor , and @xmath33 is the majorana exchange operator . in the case here , for the interaction of two nuclei , @xmath34 and @xmath35 , @xmath36 and @xmath37 is a relative angular momentum of the pair @xmath34 and @xmath35 .
furthermore , in the standard approach the operator ( 1 ) can be rewritten in the form : @xmath38 where the operators @xmath39 and the plus sign relates to even partial waves and minus sign to odd partial waves .
now the interaction in the odd and even partial waves are uncoupled and each one can be parametrised and determined _ independently _ from the other .
the procedure by which two inversions are applied to establish the odd and even components of interaction potentials will be denoted here as `` _ separate _ ''
this method has been previously applied to @xmath40he @xcite and @xmath0+@xmath1he @xcite . however , as discussed above , the separate determination of even and odd potential components induces significant errors in the appropriate potential if one set of partial amplitudes , viz .
odd or even , with @xmath41 or @xmath42 etc .
is determined from the phase shift analysis with considerable errors and this is quite often the situation . the alternative approach is to derive the odd and even potential components simultaneously using the original form of interaction operator ( 3 ) , i.e. to be denoted `` _ simultaneous _ '' inversion , and this procedure was first introduced in ref .
@xcite . for the inversion of @xmath40he empirical phase shifts the two approaches lead to very similar potentials .
however simultaneous inversion was necessary to provide a stable parity - dependent inversion of @xmath43o scattering observables , @xcite , because the majorana term is very small in comparison to the wigner type interaction .
on the same note , it is common practice to apply inversion to all members of the families of partial amplitudes , @xmath41 and @xmath42 simultaneously , i.e. to determine central and spin - orbit potentials directly rather than to calculate potentials for the separate spin channels with @xmath42 or @xmath41 and so on .
the general philosophy behind the simultaneous inversion approach is the following .
while the majorana components describe exchange terms in the nucleus - nucleus interaction , @xcite , these components represent only a minor effect of the exchange correction compared to that contributing to the main wigner term , which appears in the conventional simple and double folding models for nucleus - nucleus interactions .
this difference suggests that a more stabilised inversion is obtained by expanding @xmath30 and @xmath32 in separate basis sets , as in ref .
@xcite , with different scale parameters in each case , i.e. @xmath44 @xmath45 here @xmath46 and @xmath47 are basis sets with appropriate radial scale parameters .
this approach then leads to more reliable determination of the radial form and range of @xmath30 and @xmath32 compared to the potentials resulting from the separate inversion which may have significant uncertainties in the majorana component .
the inversion of empirical @xmath0+@xmath1he phase shifts , described further in the following section , well illustrates the importance of simultaneous inversion .
the improvement arises because both components are simultaneously determined from even ( well determined ) and odd ( poorly determined ) partial amplitudes on the equal footing .
if one then combines such simultaneous determination with a step - by - step extension of the basis in ( 5 ) and ( 6 ) , which is equivalent to the demand of maximal smoothness for each component , @xmath30 and @xmath32 , the inversion is stabilised .
in the next section we reconstruct the @xmath0-@xmath1he interaction potential from phase - shift analysis ( psa ) data using the algorithm developed by the moscow group .
the inversions are based on two separate sets of phase - shift data : 1 .
single energy psas of the zrich group @xcite established for the energy range 6 - 43 mev and 2 .
the extended energy - dependent psa of kuznetsova _ et al _
@xcite for 0 - 10 mev , with a special emphasis on the lowest energy region .
as in previous inversions based in this psa , tensor potential terms are excluded from all the inversion calculations presented here except those described in sect .
[ tens - sect ] , since the available data on the mixing parameters are too unreliable to produce a stable inversion .
this problem arises due to the very weak influence of the tensor interaction on the low - energy phase - shifts , e.g. @xmath48 differs from zero only in the narrow region near 6.5 mev , where the @xmath49 and @xmath50 phase shifts are close in value . for similar reasons ,
only real potentials are determined due to the limitations in the inelasticity parameters . however , in sect .
[ tens - sect ] we present a reconstruction of the @xmath1he + d tensor potential from the @xmath2li quadrupole moment and the existing psa data for the @xmath48 mixing parameter .
although the psa - solutions of zrich group contain many irregularities and some non - monotonic behaviour , especially for @xmath51 mev , the phase shifts in the energy range 6 - 14 mev @xcite are satisfactorily smooth functions of energy . on the other hand , the energy - dependent psa @xcite gives very reliable @xmath14- and @xmath3-wave phase shifts for @xmath52 mev , which strongly stabilises the inversion for both even and odd partial waves ( see sect .
[ invs - sect ] ) . therefore in this work we present results for inversions based on the phase shifts at energies up to 15 mev only .
a reconstruction of the @xmath0-@xmath1he interaction using all the zrich psa data ( up to 43 mev ) , but by separate inversion , has been presented previously , @xcite . in this section
we compare two solutions of the @xmath0+@xmath1he inverse problem , for the energy range @xmath415 mev , obtained as follows : 1 . by separate inversion of odd and even partial amplitudes , @xcite , 2 . by simultaneous inversion of both amplitudes ( present work ) . as a starting approximation
we take only one term in the expansions of each potential component , e.g. central , spin - orbit etc .
following our initial suggestions , @xcite we have used a s - wave orthonormal harmonic oscillator basis , for which the first term of the series corresponds simply to a gaussian .
then , in each subsequent stage , we add one term to each of the above components ( either to each component alternately or to all components at once ) .
such a procedure converges very rapidly and enables us to reliably control the process .
if the input information is insufficient to determine the next expansion term , either the iterations fail to converge or the uncertainties in the expansion coefficients become too large , @xmath53% , @xcite . the convergence of the method had been investigated previously @xcite and here we only give the final results of the inversion . in figs.[fig-1ab]a ,
[ fig-1ab]b and [ fig-2ab]a , [ fig-2ab]b we compare the potentials found for these two cases ( the central components are displayed on fig .
[ fig-1ab ] while the spin - orbit components are shown on fig .
[ fig-2ab ] ) . the degree to which these two potentials reproduce the phase shifts
is shown on figs .
[ fig-3 ] and [ fig-4 ] for the even and odd @xmath54-values respectively .
the apparent disagreement between the predictions of the inversion and the data of the phase shift analyses for the @xmath55-waves , ( fig .
[ fig-4 ] ) , probably arises due to the large errors in the psa data for these amplitudes , @xcite .
in particular , the results of accurate faddeev calculations for @xmath0-@xmath1he scattering @xcite show a much better agreement with our predictions than with the psa .
so our simultaneous inversion for the @xmath55-waves may be closer to the `` true '' solution than is apparent from fig .
[ fig-4 ] .
it is evident from figs .
[ fig-1ab ] and [ fig-2ab ] that the modifications to the potentials found when switching from the separate inversion to the simultaneous approach are rather small only for the even central potential , and are quite significant for the even spin - orbit and all the odd components .
this improvement is due to the stabilisation of the simultaneous inversion .
strong changes are also found in the odd potential on inclusion of the @xmath56 phase shifts in the inversion ( the dashed curve in fig .
[ fig-1ab]b ) , when compared with inversion based on the @xmath55 wave amplitudes alone ( the dotted curve in fig . [ fig-1ab]b ) .
these effects arise despite the large errors in the @xmath56-phase shifts of the psa @xcite .
clearly , the incorporation of such additional and _ independent _ input data for the inversion results in a further strong stabilisation of the inversion .
the above conclusion is in agreement with our initial general expectation that incorporation even of rather `` noisy '' but _ independent _ input data into the inversion , together with some definite additional constraint ( e.g. requiring some degree of smoothness in the potentials ) , leads to a noticeable stabilisation of the solutions .
this general feature of the ip inversion procedure then convincingly distinguishes our approach from more traditional and strict methods like gelfand - levitan or marchenko approaches . in the applications of the latter procedures , it is impossible to incorporate essentially incomplete or `` noisy '' data into the initial data set .
turning now to the spin - orbit potential , we find a further illustration of the above conclusions .
in fact , our results show that the majorana component of spin - orbit interaction can not be reliably derived from the existing psa data
. the reasons for this are partly due to the errors in the psa data , but also due to the small size of the majorana spin - orbit term .
this small majorana spin - orbit term is justified by a comparison of the wigner and majorana components for central potential ( see fig . [ fig-5 ] ) .
both the inversions based on rgm - calculations ( see fig . [ fig-2coop ] in section [ rgm - sect ] ) and those based on the psa data lead to central interactions for which the strength of majorana component is smaller than the wigner term by an order of magnitude .
this result confirms the general expectations that the role of the majorana term is then as a minor correction to the main wigner term .
evidently , if a similar relation between wigner and majorana terms exists for the spin - orbit interaction , there is no chance of determining a majorana ( i.e. exchange ) spin - orbit term from the existing psa data .
thus , summarising our findings in this section , we conclude that the effects of the odd - even splitting for @xmath57he are rather moderate .
the significance of the effects are also illustrated by the values of the volume integrals , @xmath58 for the different components @xmath59 of the inverted potential , eqs .
[ def_v]-[def_ak ] .
table [ invs - cso ] displays the magnitudes of volume integrals @xmath60 , for the potentials reported above and these values corroborate the above conclusions .
we have also estimated the value of @xmath61 for the majorana terms of the potentials determined by simultaneous inversion .
these potentials give , @xmath62 i.e. @xmath63 is about 8.5 @xmath64 of @xmath65 . hence , as was expected beforehand , the majorana term represents just a small correction to the main wigner component .
the same interrelation between the wigner and majorana components should also be valid for the spin - orbit interaction and our stabilised inversion tends to corroborate this conclusion . while the difference in @xmath66 for the spin - orbit terms obtained from separate inversion of even and odd partial waves is quite remarkable ( see fig .
[ fig-2ab ] ) , this difference disappears for the potentials obtained by the simultaneous inversion . ( strictly speaking , it is beyond the accuracy achievable with the existing psa see the table [ invs - cso ] . )
thus we may conclude that the initial assumption on which the simultaneous inversion of even and odd parity states was based , i.e. that the interaction is of mainly wigner type , is well justified by the results of the inversion .
the resonating group method ( rgm ) offers fundamental advantages for the study of the important effects contributing to @xmath57he scattering , despite the simplified approximations necessary in the method . here
contributions to the potential due to both the full antisymmetrization of all nucleons and the inclusion of breakup channels can be directly assessed in a way not possible by inversion from empirical phase shifts .
this formulation of the @xmath57he rgm calculations closely follows the work of kanada _
et al _ , @xcite , and the breakup channels are included using the pseudo - state method described therein . only @xmath67 transfer is included in the breakup calculations .
the rgm @xmath14-matrices are numerically calculated using modified forms of the codes of blge _ et al _ , @xcite , adapted to incorporate the minnesota @xmath68 force , @xcite as described in a previous work , @xcite .
the exchange mixture parameter @xmath69 is set to @xmath70 , but the potentials described below are not sensitive to this value .
the deuteron and excited @xmath71-@xmath20 pseudo - states are described by an 8 gaussian basis , @xcite , which allows coupling to 4 distortion channels of low energy deuteron excitation .
no spin - orbit force is included in these calculations since the empirical spin - orbit terms can not be determined to sufficient accuracy . tabulated @xmath14-matrix values are obtained from the rgm code for input into the ip inversion and , as in sect .
[ emp - sect ] , the inversion is confined to the energy range @xmath72 mev . with inclusion of channel coupling
the @xmath14-matrix becomes complex . at energies less than 10
mev @xmath73 is close to unity , but at higher energies , the coupling to breakup channels produces an irregular variation of @xmath73 with energy . at the higher energies the imaginary potential arising from breakup coupling has been shown to have a marked variation with energy , @xcite . the imaginary potential can not then be established to any reliable accuracy over the energy range currently considered and all inversions in this section establish real potential components only .
the theoretical rgm phase shifts can be reproduced by inversion to a far greater accuracy than the empirical @xmath5 .
however the most accurate inversions lead to irregularly shaped potentials due to an inherent , but small , energy dependence associated with the underlying nonlocality .
the energy dependence of @xmath5 can not be reproduced by simply introducing an energy dependence in the potential magnitude , but requires a variation in the _ potential form _ with energy .
since the details of such a dependence are far beyond what can be determined empirically , only smooth energy independent solutions are now considered and these lead to an adequate fit to the phase shifts .
solutions are obtained for both the two inversion methods described in sect .
[ invs - sect ] i.e. for separate inversion of even and odd @xmath54-values and for simultaneous inversion to determine @xmath30 and @xmath32 .
the choice of method is less critical in the case of theoretical @xmath5 and , with a small inversion basis , ( 2 functions per component ) the two methods lead to very similar parity dependent potentials .
the simultaneous inversions also show that a reasonable fit to the rgm @xmath5 is possible with a @xmath30 term alone , so that the addition of the @xmath32 component introduces only a second order correction .
fig [ fig-1coop ] shows the even and odd @xmath54 potentials for all four cases , with and without breakup coupling and using the two inversion approaches .
the components @xmath30 and @xmath32 for the potentials obtained by simultaneous inversion are illustrated in fig .
[ fig-2coop ] together with the rgm direct potential obtained from the double folding model for the minnesota nn force .
the breakup channels primarily change the even @xmath54 phase shifts only , @xcite , so that , although the even potential _ increases _ significantly in magnitude on inclusion of breakup coupling , there is little change in the odd component .
the restriction to energy - independent potentials leads to some uncertainties in the inversion but some interesting results are revealed .
the introduction of coupling produces both an increase in the wigner potential magnitude at small radii and a small decrease at @xmath74 fm .
however , at larger radii the solution inclusive of breakup coupling is very close to the direct potential multiplied by @xmath75=1.572 ( see also fig .
[ fig-1ab ] ) .
the large changes in the even @xmath54 phase shifts due to breakup coupling significantly decrease the parity dependence ( i.e. the amplitude of the majorana force ) for @xmath76 fm . at larger radii
there is a considerable agreement in the parity dependence for all solutions , but the strength of the @xmath32 term barely exceeds 1 mev .
we can now compare the rgm potentials with the empirical solution presented in sect .
[ emp - sect ] , which is indicated as the dotted line in fig .
[ fig-2coop ] .
the limitations of the rgm calculations and its dependence on simple interactions , preclude the possibility of a very exact agreement with empirical results .
indeed the magnitude of @xmath32 is too small at both larger radii and near the nuclear centre , although the latter problem is improved by the addition of breakup coupling .
significantly , the inclusion of breakup effects brings the majorana component closer to the empirical @xmath32 potential , although the rgm potential is still too small in magnitude for @xmath77 fm . as a consequence ,
the net effective @xmath0-@xmath1he rgm potential is close to a conventional _ local _ parity - independent potential .
a poorer agreement is found between the rgm solutions and the empirical potential resulting from separate inversion , particularly for the majorana component , which presents further evidence in favour of the simultaneous inversion .
in the last two or three decades the df - model has become extremely popular for the description of optical scattering , not only for @xmath20 , @xmath9he(@xmath9h ) , @xmath1he etc . , but also for such loosely bound particles as @xmath0 , @xmath10li or @xmath78be @xcite,@xcite-@xcite .
however , for the latter projectiles the normalisation constant @xmath75 was found to deviate considerably from unity and is anomalously low .
this problem has been explained @xcite , by the strong coupling effects of the virtual or real breakup of the loose projectile .
the accurate and fairly reliable potential for @xmath0-@xmath1he established by inversion now allows us to study the behaviour of the normalisation constant in a more quantitative manner than was possible in the simple phenomenological analyses of deuteron scattering by medium and heavy nuclei @xcite .
the most important question here is how adequate is the df potential form ?
the general success of the df model apparently argues in favour of its correctness .
nevertheless the situation here is far from trivial .
in fact , the df potential ( which corresponds to the so called no - distortion approximation ) must adequately describe only the peripheral part of inter - nuclear interaction where there is no distortion of the colliding nuclei .
moreover , to achieve this it is necessary to use an @xmath68-potential which has the correct peripheral behaviour , and the normalisation factor @xmath75 should be close to unity .
then one hopes that the short range part of the inter - nuclear interaction , where the effects of distortions and breakup are of greater importance , will be screened by a strong imaginary potential . in such circumstances
the df model will be a good approximation for the exact interaction . however , most practical applications of the df procedure are based on an effective _ scalar _ @xmath68-potential ( usually the m3y - force ) which is quite dissimilar to the true @xmath68-force .
the df potential calculated in this way must then be renormalised with some constant @xmath79 . in such an approach
the degree of similarity between the `` exact '' potential and the df potentials in both the peripheral and short - range radial regions is very interesting . below we will try to answer these important questions . in the fig .
[ fig-1ab]a a df potential , calculated with the effective @xmath68 minnesota force , is compared with the stabilised solution of the inverse scattering problem described in sect .
[ invs - sect ] .
it is evident that the _ initial _ df potential , even in the far peripheral region , is very different from the `` true '' interaction potential and the difference is rather large in the inner region .
now , following to the standard routine @xcite , we can also calculate the normalisation coefficient @xmath80 for which this df potential best reproduces the phase shifts data .
the even - parity partial amplitudes are most accurately reproduced with @xmath80=1.572 .
the quality of fit to the phase shifts reached with this value of @xmath81 is displayed in fig .
[ fig-6 ] . clearly , with the exception of the @xmath14-waves , the renormalised df - potential provides a very reasonable fit to these phase shifts .
in fact , the direct comparison of the renormalised df - potential with potential obtained by inversion ( see fig . [
fig-1ab]a ) reveals that the peripheral parts ( @xmath82 fm ) of both potentials are in reasonable agreement . in the inner region , however , there are large differences between the potentials and these differences are reflected in the @xmath14-wave phase shift behaviour . the @xmath3 wave and all higher partial phase shifts
are not influenced by these short range differences .
the role of the short range region will be further reduced by the imaginary potential which is present for @xmath83 mev and which will screen out the short range part of the potential .
thus , the rather good agreement of both df and inversion potentials at intermediate and large distances , and in the phase shift behaviour , does support the application of the df procedure for the description of deuteron scattering , even by the very light @xmath1he - nucleus .
further evidence is provided by a comparison of volume integrals , @xmath66 , for the various potentials , as shown in table [ invs - cso ] .
in fact , the volume integral of the even parity potential , as determined by simultaneous inversion , is very close ( with a difference of @xmath84 only ) to that of the df potential @xcite , calculated with the standard ( density - dependent ) ddm3y @xmath68 effective force and with @xmath80 adjusted to reproduce the even @xmath54 phase shifts .
the equivalent comparison for the odd parity phase shift shows far less agreement , and the volume integral of that m3y df potential for odd @xmath54 is closer to the volume integrals for the even parity cases .
the volume integrals for the df potentials calculated with both the minnesota and m3y - potentials and ( after multiplication by appropriate normalising factors @xmath75 ) are close to each other .
however the df model with the m3y force is expected to give a better agreement with the data than the minnesota force .
we can now turn to the role of deuteron breakup channels and antisymmetrization effects .
the basic problem here , as was emphasised earlier , is how to reconcile the apparent contradiction between the rather good fit to scattering data given by the double folding model ( with the m3y force ) , on the one hand , although highly renormalised , @xcite-@xcite,@xcite , @xcite-@xcite and the proven contributions of the large effects of the virtual deuteron breakup channels , @xcite-@xcite . in fitting the phase shifts with the df model
only the nn effective force constant is adjusted , so that a good description of data means that the _ radial form _ of deuteron polarisation potentialhe at @xmath8515 mev can be safely ignored .
] must be similar to the potential giving rise to purely elastic scattering , i.e. without any channel coupling .
thus the renormalised force constant in df potential should effectively incorporate three important effects : 1 .
coupling with virtual breakup channels ; 2 .
antisymmetrization effects ; 1 . and closely connected with the latter , 3 . the odd - even splitting effect .
our conclusions derived here together with the results of other authors can be summarised as follows : 1 .
the majorana force component , although definitely contributing , plays a rather moderate ( @xmath86 ) role and hence does not destroy the good quality of fit achieved with the pure central interaction potential of the df approach ; 2 . antisymmetrization and channel coupling effects as discussed in sect .
[ rgm - sect ] of the present work and also in refs .
@xcite are rather large when considered separately , but the effects compensate each other to a considerable degree - cluster wavefunction acts in the opposite direction , i.e. produces a `` compressing '' effect on the deuteron . ] .
thus , the resulting net effect is rather small ( see figs . [ fig-1coop ] and [ fig-2coop ] ) .
the odd parity potential shows almost no change when the deuteron breakup channels are included , whereas the even potential changes , on average , by 15 - 25@xmath64 at small radii . the reason underlying this change
is seen in fig .
[ fig-2coop ] where one can see that both the wigner term @xmath30 and the majorana term @xmath32 deepen by @xmath8710 mev near the origin on inclusion of channel coupling .
however , the interaction in the even partial waves is described by the combination @xmath88 while that for the odd partial waves is governed by the combination @xmath89 . as a result
, the channel coupling ( cc ) leads to a deepening of the even potential by @xmath8720 mev near origin and to almost no change in the odd potential component .
thus , on average , the net effect of channel coupling is about 10@xmath64 .
moreover , as we have already observed , the net effect of cc is mostly short ranged ( see the upper part of fig .
[ fig-1coop ] ) . because this short range contribution is often screened by an imaginary potential and
is effectively invisible , the long range contributions of cc can be simulated by renormalising the total nn force constant to reproduce the @xmath90 scattering cross sections .
in fact , the one - channel effective @xmath0-@xmath1he rgm potential ( see the curves 1 and 2 on the upper part of fig .
[ fig-1coop ] ) , which takes into account only the antisymmetrization effects but not the coupling to breakup channels , is very similar to the renormalised df potential ( see the dot - dashed curve on fig . [
fig-1ab]a ) .
while the df potential , as initially calculated , is somewhat different in form from the `` true '' interaction , especially in the inner region , by renormalising nn constant this difference can be minimised . in other words ,
the renormalised df potential reproduces the``true '' potential in the peripheral region ( @xmath912.5 fm ) quite well .
however , the renormalised df potential is a little deeper than the `` true '' one in the intermediate region 1.5 fm @xmath922.5 fm and also more shallow by @xmath93 than the true potential in the innermost region , @xmath941.5 fm .
consequently , when the df model is applied to deuteron scattering off light nuclei we expect to obtain quite a reasonable description of the elastic scattering cross section , but with some deviations from the data for the vector and especially the tensor analysing powers , which are sensitive to variations of the odd @xmath54 phase shifts and to interference effects . a careful comparison of the potentials determined from rgm phase shifts calculated _ without _ channel coupling effects ( solid lines in the upper part of fig .
[ fig-1coop ] ) with the renormalised df - potential @xmath95 ( the dot - dashed line on fig . [ fig-1ab]a ) shows that the deviation of the df potential from the true one is most likely to arise due to deuteron breakup channel couplingli , @xmath78be etc . @xcite . ] .
therefore one expects the df model to give a highly accurate description when applied to the description of scattering of more dense and heavily polarised projectiles such as @xmath96 particles .
recent work , @xcite , confirms this conclusion .
our main result in this section for the scattering of a loosely bound deuteron from the @xmath1he core can be formulated as follows : if the scattering is described by only a full antisymmetrization of the @xmath97he clusters without breakup contributions , then the local equivalent potential must contain a large majorana component and this scattering _ can not _
be described by a simple parity independent optical potential .
however , the addition of antisymmetrized breakup channels to the no - distortion approximation results in a strong reduction of the odd - even potential splitting and the resulting unified potential can be approximated reasonably well by a renormalised df potential .
in other words , only when both the antisymmetrization and @xmath0 breakup contributions are considered together , can use of the df model be justified .
the @xmath0+@xmath1he tensor interaction has been neglected in the preceding sections due to the lack of reliable mixing parameters in the present psa data . on the other hand ,
a tensor potential is absolutely necessary to describe the quadrupole moment of @xmath2li nucleus .
it is well known that the very small negative value of @xmath2li quadrupole moment ( @xmath98 @xmath99 ) can not be explained by the standard three - body @xmath100n model , @xcite , but probably originates through the coupling between the @xmath3-wave component of the @xmath16-particle in the closed channel @xmath101he and the @xmath3-wave component of the @xmath102 relative motion .
the negative value of @xmath103 is associated directly with the negative value of the asymptotic mixing constant @xmath104 @xcite , i.e. the sign of the @xmath3-component of the bound - state wave function ( in the asymptotic region ) must be _ opposite _ to that of the @xmath14-component .
a @xmath2li wave function of this type has been formulated , @xcite and this model does give the correct sign of quadrupole moment .
here we attempt to reconstruct a @xmath0@xmath1he potential which describes the @xmath49 and @xmath50 low - energy phase shifts and the main properties of the @xmath2li ground state , including the quadrupole moment .
consequently the data input into the inversion includes the binding energy and quadrupole moment of @xmath2li together with the phase shifts for energies up to 11 mev . to retain stability in the inversion
, we have restricted the expansions of both central and tensor parts of interactions to just one term : @xmath105 @xmath106 @xmath107 table [ ten - pot ] lists the parameters of two possible potentials satisfying the above criteria .
the properties of the @xmath2li ground state calculated from these potentials are presented in table [ ten - vals ] , together with the corresponding experimental values .
the @xmath49 and @xmath50 phase shifts and the mixing parameter @xmath19 , evaluated with the two potentials , are shown in figs .
[ fig-9 ] and [ fig-10 ] ( solid lines are for variant a , dashed line for variant b ) .
the negative values of @xmath48 can be obtained only from narrow tensor potentials .
the use of wider potentials results in both a change of sign of @xmath48 and a change of the asymptotic mixing constant of the ground state @xmath108 .
thus the range of the @xmath0-@xmath1he tensor potential is much less than that of the central potential .
this finding provides evidence for a strong @xmath0-exchange contribution to the tensor @xmath0-@xmath1he force @xcite ( @xmath109 is the pion mass ) . ] because the range of the exchange force in the exchange mechanism should be rather short due to the small r.m.s .
radius of @xmath1he and the large binding energy in the channel @xmath1he@xmath110 .
in this work we have studied some important problems relating to the interaction potential underlying the scattering of composite particles such as @xmath0+@xmath1he . in order to establish the true interaction potential
, we solved an inverse scattering problem using our effective method of linearised iterations .
the input for the inversions include both phase shifts calculated from rgm coupled channel calculations , which incorporate virtual and real breakup channels , and the results of a recent phase shift analysis . the success of the inversions presented here depended considerably on combining the very reliable even partial wave amplitudes with the `` noisy '' odd @xmath54 values within the empirical data set .
the majorana exchange interaction was _ explicitly _ evaluated in the inversion , and by this procedure the odd parity potential is determined with greater stability .
the resulting empirical potentials then represent an improvement on previous @xmath0+@xmath1he potentials for which the odd parity potential was unreliable and inaccurate .
the new stabilised empirical potential can be applied to predict improved phase shifts for the odd partial waves .
in fact , these predictions agree quite well with the rgm predictions , but only after inclusion of the breakup channel coupling , both in the wigner and the majorana components of the potential .
in fact , both the solutions ( i.e. obtained from the rgm phase shifts and from the psa data set ) contain only a small magnitude for the majorana component .
significantly , the breakup effects only contribute noticeably to the even parity potential and have little effect on the odd parity component .
this odd - even splitting is then expected to be of minor importance for other @xmath90 systems . we have also compared these `` exact '' interaction potentials with the potentials calculated using the popular df model based on two different effective nn forces : 1 .
ddm3y nn - force , and 2 .
minnesota nn - force .
this comparison shows that , despite the models producing apparently different @xmath0-@xmath1he df potentials , on the whole the general agreement is reasonable .
however , the ddm3y force , multiplied by an appropriate normalisation factor , leads to a better agreement with empirical potentials and provide a quite satisfactory reproduction of the `` experimental '' phase shifts ( see also @xcite,@xcite-@xcite ) .
nevertheless more detailed effects such as the odd - even splitting etc .
are not included in the df approach .
this agreement of the df m3y model with the empirical potential arises as a result of the interference of several effects , none of which is apparently taken into account in the df model , notably antisymmetrization and breakup channels coupling .
these two contributions mutually compensate each other in the relevant region of configuration space to the extent that the df approach plus an appropriate normalisation factor provides a reasonable approximation to the required interaction potential .
an analysis of @xmath111ca scattering at @xmath112 mev , @xcite , lead to similar results to those presented here .
in particular the interrelation between the df potential for @xmath111ca and the `` exact optical potential '' extracted from the data by a model independent analysis equates strongly to our findings for @xmath113 .
this comparison then suggests that our results may have a much wider applicability than the particular case considered here . combining the above results leads to a very powerful inversion procedure .
the modified ip inversion method , as employed widely in the present work , converges very fast and in a stable manner with a good choice of initial potential .
the df model is now confirmed as a very convenient candidate for this initial approximation .
the complete method , when invoked for inversion directly from cross - section data , @xcite , offers many opportunities to study fine details of the interactions of complicated systems such as @xmath114mg+@xmath115sn , @xmath21b+a etc . and even for unstable radioactive projectiles scattered off stable targets ( see e.g. @xcite ) .
in this paper we have also determined a @xmath0-@xmath1he tensor interaction .
the resulting potential is rather short ranged and is relatively large in amplitude ( @xmath116 mev ) . as has been previously suggested , @xcite
, this interaction may arise due to the very specific exchange effect in which the `` inner '' deuteron in the d state of the @xmath1he core is exchanged with the outer valence deuteron .
this exchange mechanism can be applied to explain both the short range character of the @xmath0-@xmath1he tensor force , through the very small r.m.s .
radius of @xmath1he , and the _ negative _ value of the @xmath2li quadrupole moment .
the second result is consistent with the fact that the s - wave and d - wave components of the total @xmath2li wavefunction , when projected onto the @xmath0-@xmath1he channel , have opposite signs in the asymptotic region .
thus the form and strength of the @xmath0-@xmath1he tensor force at both low and intermediate energies deserve detailed study .
the results presented here may be considered an initial step in this interesting direction .
many characteristic features of interaction of the @xmath0-@xmath1he system have been established in this and a series of preceding works @xcite . in these works important features such as the pauli principle manifestation and the appearance of pauli - forbidden states ,
the description of higher partial waves , channel coupling effects , the interrelation with supersymmetrical partner potentials and the general manifestation of dualism repulsion - attraction in composite particle interaction have been studied in detail .
in particular we have shown , @xcite , that the deep attractive interaction potentials in the systems @xmath1he+@xmath1he , @xmath1he+n etc .
arise as a consequence of well localised pauli forbidden states and the appropriate conditions of orthogonality for the scattering wavefunctions to these forbidden states . in turn
, the structure of the pauli forbidden states is very closely interrelated to the shell model structure of the whole unified system @xcite .
other interesting effects have been found involving the joint action of antisymmetrization and ( virtual ) breakup in @xmath2li , @xcite .
for this case calculations have established that , while the inclusion of deuteron breakup channels _ diminishes _ the @xmath18he cluster probability in the full three body @xmath2li wavefunction , the subsequent antisymmetrization of the @xmath117he wavefunction _ increases _ the @xmath18he cluster probability .
again both these important contributions produce opposing effects .
thus the present paper can be considered , in some sense , as a concluding work in the long series of our studies for @xmath18he and similar cluster systems .
the authors are grateful to prof .
v.neudatchin , dr . r.s . mackintosh and our colleagues at the moscow institute of nuclear research for useful discussions in the course of the work .
kukulin , v.g .
neudatchin , yu.f .
smirnov , i.t .
clusters as subsystems in light nuclei_. vol .
3 , in the series _ clustering phenomena in nuclei _ ,
( vieweg & s. verlag , braunschweig / wiesbaden , 1983 ) .
b.buck , c.b.dover and j.p.vary , phys.rev .
c*11 * , 1803 ( 1975 ) ; b.buck and a.a.pilt , nucl.phys . *
a280 * , 133 ( 1977 ) .
kukulin , v.g .
neudatchin , a.a .
sakharuk and v.n .
pomerantsev , phys.rev.c*45 * , 1512 ( 1992 ) ; yad.fiz .
j.nucl.phys . ) * 52 * , 402 ( 1990 ) ; phys.lett.*b255*,482 ( 1991 ) .
h. horiuchi , progr.theor.phys .
* 64 * , 184 ( 1980 ) .
y.c . tang , m. lemere and d.r .
thomson , phys .
rep . * 47 * , 167 ( 1978 ) .
s.g.cooper , phys .
c*50 * , 359 ( 1994 ) .
blokhintsev , v.i .
kukulin , d.a .
savin and a.a .
sakharuk yad.fiz .
j. nucl.phys . ) * 53 * , 693 ( 1991 ) .
h. feshbach , ann .
phys . * 5 * , 357 , ( 1958 ) .
sofianos , k.c .
panada and p.e .
hodgson , j.phys.g : nucl .
* 19 * , 1929 ( 1993 )
satchler , _ direct nuclear reactions _ ( clarendon press , oxford , 1983 ) .
r.g.lovas , a.t .
kruppa , r. beck and f. dickman , nucl.phys . *
a474 * , 451 ( 1987 ) ; k varga and r.g.lovas , phys .
c*43 * , 1201 ( 1991 ) ; y. fujiwara and y.c .
tang , phys . rev .
c*43 * , 96 ( 1991 ) ; few body systems , * 12 * , 21 ( 1992 ) : h.kanada , t.kaneko and y.c.tang , phys .
c*38 * , 2013 ( 1988 ) .
y. yahiro and m. kamimura , progr.theor.phys .
* 65 * , 2046 ; 2051 ( 1981 ) ; v.i .
kukulin and v.n .
pomerantsev , yad.fiz .
. nucl.phys . ) * 50 * , 27 ( 1989 ) .
j.cook , nucl.phys . *
a382 * , 61 ( 1982 ) .
g.r.satchler , l , w .
owen , a.j .
elwyn , g.l .
morgan , and r.l .
walter , nucl.phys .
* a112 * , 1 ( 1968 ) .
s.g.cooper and r.s.mackintosh , phys .
c*54 * , 3133 ( 1996 ) .
cooper and r.s .
mackintosh , phys .
c*43 * , 1001 ( 1991 ) .
cooper , r.s .
mackintosh , a. cst , and r.g .
lovas , phys . rev.c * 50 * , 1308 ( 1994 ) .
a.a.ioannides and r.s.mackintosh , phys .
169b * , 113 ( 1986 ) .
v.i . kukulin and v.n .
pomerantsev , yad.fiz .
physics of atomic nuclei ) * 60 * , 1228 ( 1997 ) .
g.r.satchler and w.g .
love , phys .
rep . * 55 * , 183 ( 1979 ) .
kukulin , v.n .
pomerantsev and j. horaek , phys.rev . a*42 * , 2719 ( 1990 ) .
a. adahchour , d. baye and p. descouvemont , nucl.phys .
* a579 * , 305 ( 1994 ) .
cooper , nucl .
a618 * 87 ( 1997 ) .
r.s . mackintosh and s.g .
cooper , j. phys .
g : nucl . part .
* 23 * , 565 ( 1997 ) .
cooper and r.s .
mackintosh , nucl .
* a576 * , 308 ( 1994 ) .
s.g . cooper and r.s .
mackintosh , nucl .
a582 * , 283 ( 1995 ) b.a .
robson , _ the theory of polarisation phenomena _ , ( clarendon press , oxford , 1974 ) .
b. jenny , w. grebler , v .
knig , p.a .
schmelzbach and c. schweizer , nucl.phys . *
a397 * , 61 ( 1983 ) .
kuznetsova and v.i .
kukulin yad.fiz .
j.physics of atomic nuclei ) , * 60 * , 528 ( 1997 ) ; v.m .
krasnopolsky , v.i .
kukulin , e.v .
kuznetsova , j. horaek and n.m .
queen , phys.rev .
c*43 * , 822 ( 1991 ) .
y.koike , progr . theor.phys .
* 59 * , 87 ( 1978 ) .
h. kanada , t. kaneko , s. saito and y.c .
tang , nucl . phys . *
a444 * , 209 ( 1985 ) .
g. blge , k. langanke and h .-
reusch , `` computational nuclear physics 2 '' , ed .
k. langanke , j.a .
maruhn and s.e .
koonin ( springer - verlag , new york ) ( 1993 ) .
d. r. thompson , m. lemere and y.c .
tang , nucl . phys . *
a286 * , 53 ( 1977 ) .
cooper , ou preprint oupd9707 , nucl .
a * ( to be published ) .
v.i.kukulin and g.g .
ryzhikh , prog .
nucl . phys .
* 34 * , ( 1995 ) 397 ; v.i.kukulin and v.n .
pomerantsev , sov .
. phys . * 50 * , ( 1989 ) 17 .
r.s . mackintosh and s.g .
cooper , ou preprint oupd9708 , nucl .
phys . * a * ( to be published ) .
m. kamal , v.t .
voronchev and v.i .
kukulin , vant ( problems of at .
science and technique ) , ser . fusion . * no 3 * , 37 ( 1989 ) ; j.phys.g : nucl.part.phys . *
18 * , 379 ( 1992 ) . c. samanta , s. ghosh , m. lahiri , s. ray and s.r .
banerjee , phys.rev .
c*45 * , 1757 ( 1992 ) ; t.sinha , s. ray and c. samanta , phys.rev .
c*48 * , 785 ( 1993 ) .
m.ermer , h.clement , p.grabmayr , g.j.wagner , l.friedrich and e.huttel , phys.lett .
* b188 * , 17 ( 1987 ) .
u.atzrott , p. mohr , h. abele , c. hillenmayer and g. staudt , phys.rev .
c*53 * , 1336 ( 1996 ) . h.abele and g. staudt , phys.rev .
c*42 * , 742 ( 1993 ) .
kukulin , v.n .
pomerantsev , kh.d .
razikov v.t .
voronchev and g.g .
ryzhikh , nucl.phys . *
a586 * , 151 ( 1995 ) .
n.nishioka , j.a.tostevin , r.c.johnson and k.i.kubo , nucl.phys . *
a415 * , 230 ( 1994 ) .
goncharov , a.s .
demyanova and a.a .
ogloblin , `` nucleon density distributions in @xmath118he and @xmath118b from elastic scattering by @xmath9he and @xmath9h '' , preprint inp msu 97 - 31/4482 , 1997 , moscow , inst .
moscow state univ . | improved potential solutions are presented for the inverse scattering problem for @xmath0+@xmath1he data .
the input for the inversions includes both the data of recent phase shift analyses and phase shifts from rgm coupled - channel calculations based on the nn minnesota force .
the combined calculations provide a more reliable estimate of the odd - even splitting of the potentials than previously found , suggesting a rather moderate role for this splitting in deuteron - nucleus scattering generally .
the approximate parity - independence of the deuteron optical potentials is shown to arise from the nontrivial interference between antisymmetrization and channel coupling to the deuteron breakup channels .
a further comparison of the empirical potentials established here and the double folding potential derived from the m3y effective nn force ( with the appropriate normalisation factor ) reveals strong similarities .
this result supports the application of the double folding model , combined with a small majorana component , to the description even of such a loosely bound projectile as the deuteron . in turn
, support is given for the application of iterative - perturbative inversion in combination with the double folding model to study fine details of the nucleus - nucleus potential .
a @xmath0-@xmath1he tensor potential is also derived to reproduce correctly the negative @xmath2li quadrupole moment and the @xmath3-state asymptotic constant .
= msam10 contents of the paper . 1 .
introduction 2 .
the role of parity dependence in ip inversion 3 .
stabilised inversion of the @xmath0-@xmath1he empirical data for @xmath415 mev 4 .
inversion of rgm @xmath5 with breakup channel contributions 5 .
validity of double folding model for @xmath0-@xmath1he interaction and similar systems
. 6 .
tensor @xmath0-@xmath1he potential and the @xmath6-quadrupole moment . 7
. conclusions |
after more than two decades of study , helioseismology has demonstrated the capability to probe the solar interior thanks to the seismic instruments of the soho satellite and ground - based networks @xcite . the new space missions ( sdo , picard ) are oriented toward a better understanding of the sun s internal magnetism and its impact on the earth s atmosphere . in this context , the existence of a fossil field in the sun s radiative zone is a missing bridge that motivate us to better estimate the competition between microscopic and macroscopic physics @xcite .
in addition to the international solar research community , more than 350 scientists from all over the world are working on the data from the corot and kepler spacecraft dedicated to exoplanets and astero - seismology ( see http://smsc.cnes.fr/corot for corot and http://astro.phys.au.dk/kasc/ for kepler ) .
the development of these disciplines justifies ensuring the correctness of the many fundamental properties of the plasma ( detailed abundances , reaction rates , opacity coefficients , etc . ) . the new high energy density laser facilities and the z pinch machine now enable access to the radiative properties of plasmas of stellar type in the laboratory during pico and nano second timescales ;
see @xcite , and references therein .
this review is divided into three parts : section 2 is dedicated to solar - type stars , section 3 to the more massive stars , and section 4 to the study of the @xmath0 cephei variable star case and a comparison between eight calculated iron opacity spectra .
the challenges faced by the experiments are described in section 5 .
stars produce both nuclear energy through nuclear reactions , and thermal energy due to the internal temperature that exceeds everywhere some thousand degrees .
the central plasma is heated to more than a dozen million k , and the heat is transported by convection or radiation phenomena .
the low - mass stars ( @xmath1 ) have an internal radiative zone which disappears in the region where the photons interact so strongly with the partially ionized constituents that convection becomes the most efficient way to transport the energy ; see figure 32 of @xcite or @xcite . in fact
, this boundary between radiative transport and convective transport differs from star to star depending on the star s mass , age , and composition .
hydrogen and helium are completely ionized in these radiative zones due to their high temperature and density , but in the central conditions , the heaviest elements ( iron and above ) are generally partially ionized and their contribution to the opacity coefficients is rather high , 20% in the solar case ; see equation 3 and figures in @xcite .
nevertheless , their mass fraction is of the order of 10@xmath2 . in the rest of the radiative zone , since both the temperature and density decrease with radius , by typically a factor 15 and 100 respectively , most of the heavy elements down to oxygen become partially ionized , so their relative contribution increases the rosseland mean opacity @xmath3 by a factor up to 100 at the base of the convection zone due to the multiple photon interactions on the electronic bound layers of more and more elements . as a direct consequence ,
the lifetime of a star of a given mass strongly depends on the amount of heavier elements ( carbon to uranium ) present in the star .
the total element mass fraction is generally divided into x , y , and z , with x and y corresponding to the mass fraction of hydrogen and helium , respectively , and z corresponding to the remainder .
the lifetime of a 0.8 m@xmath4 star with z=0.001 is about 14 gyr , while the lifetime of a 0.8 m@xmath4 with z=0.02 is about 22 gyr . when the heavy element contribution is small , the star contracts more quickly , the central temperature is higher , the opacity coefficients are different , and the nuclear reactions become more efficient .
this example shows the importance of verifying some detailed contributions to the total rosseland mean opacity values to properly estimate how @xmath3 changes from star to star depending on its mass and on its detailed composition .
the best known case is the sun where where our efforts to incorporate new updates of the composition highlight the important effect of the heavy elements on the sound speed profile @xcite .
the difference between the standard model sound - speed predictions and the observed sound speed is larger than the error bars coming from the knowledge of the composition and from the known difference between opal and op mean rosseland opacity tables @xcite . while for the solar conditions the rosseland mean opacities agree in general among different calculations , the detailed comparison , element by element , shows rather large differences , as mentioned by blancard et al . in turck - chize
so it is useful to compare opacity calculations of a specific element to laboratory experiments ; then the extrapolation to other stars of different composition will be more reliable .
the first experiments have been performed at the sandia z machine to check the iron contribution at a temperature and density not far from those at the base of the solar convection zone @xcite .
extensive measurements on the nif or lmj are needed for several well - chosen elements and temperatures between 193 and 400 ev to validate opacity calculations in the radiative zones of solar - like stars .
indeed the region just around the transition is clearly the most difficult region due to the interplay of microscopic and macroscopic phenomena ( turbulence , rotation and magnetic fields ) .
in fact , in contrast with the rest of the radiative zone , the boundary of the internal radiative zone depends directly on the detailed abundances of oxygen , iron , and neon and on the associated rosseland mean opacity coefficients .
it depends also indirectly on the turbulence of these layers associated with the fact that the stars are generally rotating .
so the thermal instability produces a strong horizontal shear between a differentially latitudinal rotating convective zone and a reasonably flat rotation in the radiative zone @xcite .
moreover , the magnetic field in these layers probably plays a crucial role in the solar dynamo at the beginning of the 11-year cycle . in the sub - surface layers of these stars , radiation and convection
are in competition and the heavy elements are less important because most of the elements are neutral or even in molecules . in the corona , the plasma is heated to a million degrees , so the heavy elements play a crucial role , but the very low density plasma is no longer in lte ( local thermodynamic equilibrium ) , so non - lte calculations are needed to explain the observed lines @xcite .
when the stellar mass increases , the central temperature increases and consequently carbon , oxygen , nitrogen can penetrate the coulomb barrier . in these conditions
the hydrogen burning is no longer dominated by the weak interaction but by the strong interaction where these elements act as catalysts .
the nuclear burning becomes more rapid , and the lifetime of these stars decreases from billions to millions of years , the central temperature increases strongly and the radiative diffusion has no time to take place .
the core of these stars ( @xmath5 ) becomes convective , while the envelope is radiative . in the envelope
, many elements are partially ionized , so the detailed calculation of the rosseland mean opacity is even more complex because many electronic shells are open and the bound - bound processes are extremely important .
figure 1 illustrates the case of a 10 m@xmath4 star .
all the species including iron are totally ionized in the center of this star due to high temperature , 34 mk , and a relatively low density of 8 g/@xmath6 .
the opacity is not as high ( fig 1b ) as in lighter stars , but the luminosity due to the cno cycle is so high that @xmath7 is higher than @xmath8 ( fig 1a ) . when the ratio of gradients is greater than one , the region is convective ; in the opposite case , the region is radiative
. the rest of the star is radiative except in two small regions which correspond to the partially ionized contribution of helium and hydrogen at low temperatures ( see fig 1d ) and of the iron - group elements above 100,000 k ( fig 1c ) .
figure 2 shows that the differences between op and opal are particularly important for ni , mn and cr and that the contribution of cr and ni could be very important in the case of op .
therefore , we have begun comparisons between different opacity calculations . star .
b : temperature dependence of the total rosseland mean opacity for op calculation ( solid blue line ) and opal calculation ( dashed red line ) .
c : relative contribution of the heavy elements to the total opacity .
d : relative contribution of the light elements to the total opacity.,width=340 ] two other processes are considered in the stellar equations or in stellar pulsations . both require a detailed description of the sub - surface layers : the radiative acceleration and the @xmath9 mechanism .
radiative accelerations are crucial quantities to study the diffusion of elements in stars and to understand the stellar photospheric abundances .
they strongly depend on the atomic properties of the different ions , so they largely depend on the quality of the atomic data and on the knowledge of the ionization stage of the different elements . for a detailed description of this process
, we refer to the book dedicated to g. michaud , and in particular to the paper of @xcite .
the expression for the radiative acceleration is also given in @xcite .
this process must be taken into account in the description of stellar envelopes , in addition to the gravitational settling , turbulence and sometimes magnetic activity .
for radiative accelerations , one needs to know the detailed monochromatic photon flux for all the processes ( diffusion , bound - bound , bound - free ... ) and for each element .
this information is not delivered by the opal consortium @xcite but is available for the op calculations @xcite . some specific comparisons lead to 50% differences in the radiative acceleration for the iron case in the opacity peak region mentioned in figs . 1 and 2 @xcite .
it is important to resolve this discrepancy to understand the chemically peculiar stars and the thousands of pulsators observed with corot and kepler .
the rapid variability of the above - mentioned opacities in the stellar envelope generates the so - called @xmath9 mechanism and maintains the pulsation in classical cepheids ( due to partial ionization of helium ) and for @xmath0 cephei stars ( due to the iron - group element ionization ) @xcite .
g/@xmath6 , 3.4 10@xmath2 g/@xmath6 and 2.6 10@xmath2 g/@xmath6 respectively , estimated using the op calculations .
the whole wavelength range is useful for the rosseland mean opacity determination ( @xmath10e = hc = 12394 a.kev ) and corresponds to the xuv spectrometer used for the experiment .
the ionization charge distribution for each condition is superimposed.,width=264 ] calculations for asteroseismology must predict the theoretical acoustic modes in order to identify without ambiguity the observed modes and to understand the excitation of the modes .
several papers show the diffculties for @xmath11 cephei stars @xcite and mention that an increase of opacity will improve the interpretation of the asteroseismic data .
we investigate the plasma conditions found in the envelopes of @xmath0 cephei stars for improving the interpretation of the asteroseismic observations , without neglecting previous measurements of iron ( and nickel ) opacity spectra which have successfully led to a revision of opacities in the nineties , but not always in the range useful for astrophysics @xcite . with this objective ,
we perform a comparison between different calculations of opacity spectra and an experiment to validate these calculations .
the first step consists in defining a condition compatible with a laboratory experiment .
the iron - group peak in the envelope of @xmath0 cephei stars corresponds to temperatures of 200,000 k ( 17.2 ev , 1 ev= 11,600k ) and densities around 4 10@xmath12 g/@xmath6 .
this density is too low to be investigated in a laboratory experiment .
so we have first looked for conditions which maintain the same distribution of ionization stage for iron .
= 3.4 10@xmath13 shown in figure 3 .
the hypotheses of these computations are explained briefly in the text.,width=366 ] figure 3 shows the spectral transmission of photons @xmath14 through a heated iron foil of thickness @xmath15 , @xmath16 , where @xmath17 is the opacity per mass unit in @xmath18/g .
transmission is the quantity measured in the experiment .
the mean spectrum is obtained using the op tables , for the same distribution of the iron ionization charge ( about 9 for t= 320,000 k or 27.3 ev , @xmath19 = 3.4 10@xmath2 g/@xmath6 ) .
two other spectra also are shown , for a variation of temperature of about 40% . in these cases ,
the distribution of charge is displaced by two or three units .
the figure shows that the spectra vary strongly with temperature . in the region where the @xmath9 mechanism is important ,
the rosseland mean opacity varies rapidly for this reason , so measurement at different temperatures will help to check the whole calculation of the pulsation mode excitation .
figure 4 shows the opacity spectra versus the photon energy obtained by the different calculations generally used to describe these plasmas .
clear differences can be seen in different parts of the spectral range .
the corresponding rosseland mean values show differences of up to 40% , in particular for the op value , but most of them agree within 15% despite the very different calculational approaches .
we notice also some variation of the mean ionization from one calculation to another of about 0.5 units , which is not so small if one considers the rapid variation of the opacity with the degree of ionization , but which can not explain all of the differences .
a detailed analysis of these calculations will appear in a more detailed paper .
we give here a short description of the assumptions .
based on the average atom model scaalp @xcite , the opas code combines detailed configuration and level accounting treatments to calculate radiative opacity of plasmas in lte .
the bound - bound opacity can be calculated using detailed line accounting and/or statistical methods .
the bound - free opacity is evaluated neglecting the configuration level splitting . to improve the accuracy of opacities into the complex regime of warm dense matter , where plasma and many - body effects can be important
, the free - free opacity is obtained by interpolating between the drude - like opacity and the opacity derived from the kramers formula including a gaunt factor and an electron degeneracy effect correction .
photon scattering by free electrons includes collective effects as well as relativistic corrections .
op ( opacity project ) is an on - line atomic database used in astrophysics @xcite .
topbase contains the most complete dataset of ls - coupling term energies , f - values and photoionization cross sections for the most abundant ions ( z=1 - 26 ) in the universe .
they have been computed in the close - coupling approximation by means of the r - matrix method with innovative asymptotic techniques .
topbase also contains large datasets of f - values for iron ions with configurations @xmath20 , referred to as the plus - data , computed with the atomic structure code superstructure .
the atomic physics sco code @xcite is based on the superconfiguration approach proposed by @xcite to calculate high - z element photoabsorption in lte conditions . here
calculations have been performed including non - relativistic and relativistic contributions .
sco - rcg @xcite is a hybrid opacity code which combines the statistical super - transition - array approach and fine - structure calculations .
criteria are used to select transition arrays which are removed from the super - configuration statistics , and replaced by a detailed line - by - line treatment .
the data required for the calculation of the detailed transition arrays ( slater , spin - orbit and dipolar integrals ) are obtained from the super - configuration code sco ( see above ) , providing in this way a consistent description of the plasma screening effects on the wave functions .
then , the level energies and the lines ( position and strength ) are calculated by the routine rcg of r.d .
cowan s atomic structure code @xcite .
the los alamos light element detailed configuration opacity code , ledcop @xcite , uses a basis set of ls terms based on semi - relativistic hartree - fock atomic structure calculations ( plus a mixture of ls terms and unresolved transition arrays , some of which are treated by random lines for the most complex ion stages ) to calculate opacities for elements with z @xmath21 31 .
each ion stage is considered in detail and interactions with the plasma are treated as perturbations .
calculations are done in local thermodynamic equilibrium ( lte ) and only radiative processes are included . in the sta opacity code
@xcite the total transition array of a specific single - electron transition , including all possible contributing configurations , is described by only a small number of super - transition - arrays ( stas ) .
exact analytic expressions are given for the first few moments of an sta .
the method is shown to interpolate smoothly between the average - atom ( aa ) results and the detailed configuration accounting that underlies the unresolved transition array ( uta ) method .
each sta is calculated in its own optimized potential , and the model achieves rapid convergence in the number of stas included .
cassandra is an awe ( atomic weapons establishment , aldermaston , uk ) opacity code @xcite used to model plasma in local thermal equilibrium .
cassandra s self - consistent field calculation uses the local density approximation for bound states and has a free electron contribution based upon the thomas - fermi model .
the cassandra data are british crown owned copyright 2010/mod .
flychk is a free access atomic spectroscopy code @xcite used to generate atomic level populations and charge state distributions and synthetic spectra for low - z to mid - z elements under lte or nlte conditions http://nlte.nist.gov / fly/.
the new spectral opacity measurements have been performed on luli 2000 www.luli.polytechnique.fr/ with two lasers in 2010 .
a nanosecond laser delivers an energy between 30 - 500 j in a 500 ps duration pulse .
this laser is used to irradiate a gold cavity ( hohlraum ) on which an absorbing foil of the considered element is deposited and heated . after a required delay to get the proper density and temperature , a picosecond laser interacts with a backlighter to produce x rays which interact with the formed plasma : for a description of the experimental setup , see @xcite .
we detect the transmitted opacity spectrum on a streak camera placed behind a newly designed xuv - ray spectrometer . obtaining high - quality measurements required addressing many challenges based on experience with previous measurements .
a detailed analysis of the previous experiments and their difficulties is found in @xcite .
we summarize some of these relevant for astrophysical applications : - one needs to form a plasma in lte .
this requirement supposes a good simulation of the whole experiment to probe the foil in the best conditions .
the rapid expansion of this foil during heating is limited by placing the foil between two thin samples of a low - z material ( typically carbon ) .
- we measure the foil temperature with a @xmath22dmix and limit the temperature gradient by placing the foil between two cavities .
the temperatures depend on the geometry of the experiment , the position of the foil , and the correctness of the simulation .
figure 5 shows the transmission obtained for the different calculations including a spectrometer resolution of 1 ev for the the conditions mentioned in figure 3 ; the differences must be distinguishable by the experiment .
notice the distinct behavior of the op calculation at low energy and some saturation of the spectra which illustrates the difficulties of exploring the low - energy regime .
to reduce the difficulties noted above , we measure the spectra of 3 or 4 neighboring elements ( chromium , iron , nickel , copper ) using two thicknesses for each of them .
moreover we use two cavities to heat the foil on the two faces to limit the temperature gradient . in summary
, this study shows that it would be of interest to develop complete opacity tables including photon energy spectra with the best physics available for asteroseismology .
it is important to improve the coverage of some crucial regions for interpreting the space observations . some special validations by laboratory experiments are required to validate `` the best physics '' .
@thebibliography@page= alecian , g. 2005 , eas pub .
series , 17 , 33 asplund , m. , grevesse , n. , and scott , p. 2009
, ara&a , 47 , 481 badnell , n. r. et al .
2005 , , 360 , 458 .
bahcall , j. n. , serenelli , a. m. & basu , s. 2005 , , 621 , l85 bailey , j. et al .
, 2007 , phys .
99 , 5002 bailey , j. et al . , 2009 , phys . of plasmas , 16 ,
058101 bar - shalom , a. , oreg , j. , goldstein , w. h. , schwarts , d. , zigler , z. 1989 , phys .
a , 40 , 3183 basu , s. , antia , h. m. 2008 , phys .
457 , 217 blancard , c. and faussurier , g. 2004 pre 69 , 016409 blenski , t. , grimaldi , a. and perrot , p. 1997
e 55 , r4889 blenski , t. , grimaldi , a. , perrot , f. , j. 2000 , quant .
, 65 , 91 briquet , m. et al .
2007 , mnras , 381 , 1482 chenais - popovics , c. et al .
2000 apj suplem . series .
127 , 175 chenais - popovics , c. , fajardo , m. , thais , f. et al . 2001 , j. quant .
, 71 , 249 .
chenais - popovics , c. 2002 , laser and particle beams 20 , 291 christensen - dalsgaard , j. 2004 , sol .
, 220 , 137 chung , h .- k . ,
chen , m. h. , morgan , w. l. , ralchenko , y. , lee , r. w. 2005 , hedp 1 , 3 colgan , j. , abdallah , j. , jr . , sherrill , m. e. , foster , m. , fontes , c.j . and
feldman , u. 2008 , , 689 , 585 cowan , r.d .
1981 , the theory of atomic structure and spectra , university of california press crowley , b.j.b .
2001 , jqrst , 71 , 257 daszynska - daszkiewicz , j. & walczak , p. 2010
, , 403 , 496 delahaye , f. & pinsonneault , m. 2006 apj , 625 , 563 degroote , p. et al .
2009 a&a , 506 , 111 duez , v. , mathis , s. and turck - chize , s. 2010 , , 402 , 271 dupret , m. a. 2003 , thesis , university of liege , belgium iglesias , c.a . & rogers , f. j. 1995 , apj , 443 , 469 guzik , j.a . ,
watson , l.s . &
cox , a n. 2005 , apj , 627 , 1049 loisel , g. et al .
2009 , hedp , 5 , 173 magee , n.h . , et al .
1995 , in asp conf .
78 , 51 pamyatnykh , a.a . 1999 ,
acta astronomica , 49 , 119 porcherot , q. , gilleron , f. , pain , j.c . , and blenski , t. 2010 , to be published .
seaton , m.j .
& badnell , n.r .
2004 , mnras , 354 , 457 seaton m.j .
2007 , mnras , 382 , 245 springer , p.t .
1992 , , 69 , 3735 turck - chize , s. , et al .
1993 , phys .
, 230 , 59 - 235 turck - chize , s. , et al . 2004 , , 93 , 211102 turck - chize , s. , et al . 2009 , hedp , 5 , 132 turck - chize , s. and couvidat , s. 2010 , rep .
phys . , submitted turck - chize , s. , palacios , a. , marques , j. , nghiem , p.a.p .
2010 , , 715 , 1539 turck - chize , s. , loisel , g. , gilles , d. et al .
2010c , soho24 , j. phys .
: conf . ser .
, in press winhart , g. et al .
1995 , jqsrt , 54 , 37 | the lifetime of solar - like stars , the envelope structure of more massive stars , and stellar acoustic frequencies largely depend on the radiative properties of the stellar plasma . up to now , these complex quantities have been estimated only theoretically .
the development of the powerful tools of helio- and astero- seismology has made it possible to gain insights on the interiors of stars .
consequently , increased emphasis is now placed on knowledge of the monochromatic opacity coefficients . here
we review how these radiative properties play a role , and where they are most important .
we then concentrate specifically on the envelopes of @xmath0 cephei variable stars .
we discuss the dispersion of eight different theoretical estimates of the monochromatic opacity spectrum and the challenges we need to face to check these calculations experimentally .
* published in astrophysics and space science * |
we study the capacity of the discrete - time , peak - and - average - power - limited , gaussian channel when its output is quantized using a dithered , infinite - level , uniform quantizer of step size @xmath0 and analyze its behavior in the low- and high - precision limit , where @xmath0 tends to infinity and zero , respectively .
the problem of quantization arises in communication systems where the receiver uses digital signal processing techniques , so the analog received signal must be sampled and then quantized using an analog - to - digital converter ( adc ) .
if the received signal is sampled at nyquist rate or above , and if an adc with high precision is employed , then the effects of sampling and quantization are negligible
. however , high - precision adcs may not be practical when the bandwidth of the system is large and the sampling - rate is high @xcite . in such scenarios , low - resolution adcs must be used . to better understand what communication rates can be achieved with
low - resolution adcs and nyquist sampling , various works have studied the discrete - time gaussian channel when its output is quantized using a 1-bit quantizer . at low signal - to - noise ratio ( snr ) , where communication at low spectral efficiencies takes place , it is known that a symmetric threshold quantizer if its input is above a threshold , and it produces @xmath1 if its not . a _ symmetric _
threshold quantizer is a threshold quantizer whose threshold is zero . ]
reduces capacity by a factor of @xmath2 , corresponding to a 2 db power loss @xcite , @xcite .
hence the rule of thumb that hard decisions cause a 2 db power loss . " it was recently demonstrated that this power loss can be avoided by using asymmetric threshold quantizers and asymmetric signal constellations @xcite .
however , this result requires _ flash - signaling _ input distributions ( * ? ? ?
3 ) ( see ( * ? ? ?
2 ) for a definition ) . since such inputs are known to have a poor spectral efficiency ( * ? ?
16 ) , it follows that for small yet positive spectral efficiencies , the potential power gain is significantly smaller than 2 db .
for example , at spectral efficiencies of 0.001 bits / s / hz , allowing for asymmetric quantizers with corresponding asymmetric signal constellations provides a power gain of merely 0.1 db ( * ? ? ?
v ) . in the following ,
we refer the gaussian channel with ( @xmath3-bit ) output quantization as the _
( @xmath3-bit ) quantized gaussian channel _ and to the gaussian channel without output quantization simply as the _
gaussian channel_. for the gaussian channel , binary antipodal inputs outperform flash - signaling inputs in terms of spectral efficiency ( * ? ? ?
however , for such inputs quantizing the channel output with a 1-bit quantizers incurs again a 2 db power loss at low snr , since in this case a symmetric threshold quantizer becomes asymptotically optimal as the snr tends to zero ( * ? ? ?
* prop . 2 ) .
recalling that the discrete - time gaussian channel arises from the continuous - time , bandlimited , additive white gaussian noise ( awgn ) channel by sampling the output at nyquist rate , it can be shown that , for binary antipodal signaling and a symmetric threshold quantizer , the 2 db power loss can be reduced by sampling the channel output above the nyquist rate .
for instance , it was demonstrated that , at low snr , sampling the output at twice the nyquist rate improves the power loss from 2 db for nyquist sampling to less than 1.28 db ( * ? ? ? * th . 1 ) ,
* th . 1 ) .
further results on the capacity of the 1-bit quantized gaussian channel and super - nyquist sampling include @xcite@xcite . specifically , zhang @xcite studies the generalized mutual information of this channels for a gaussian codebook ensemble and the nearest - neighbor decoding rule and demonstrates _
inter alia _ that , as the sampling rate tends to infinity , the power loss is not larger than 0.98 db .
shamai @xcite considers the noiseless case and demonstrates that the capacity is unbounded in the sampling rate .
however , it is unknown whether for a symmetric threshold quantizer the power loss can be fully avoided by letting the sampling rate tend to infinity . going beyond 1-bit quantizers
, it was shown that , at low snr , a uniform 3-bit quantizer and binary antipodal signaling achieves about 95% of the capacity of the gaussian channel , corresponding to a power loss of merely 0.223 db ( * ? ? ?
* eq . ( 3.4.21 ) ) .
the capacity of the @xmath3-bit quantized gaussian channel was studied , e.g. , in @xcite .
the numerical results obtained in @xcite suggest that , at 0 db snr , a 2-bit quantizer achieves still 95% of the capacity of the gaussian channel , while at 20 db snr , a 3-bit quantizer achieves still 85% of the capacity of gaussian channel .
however , to the best of our knowledge , there exists no closed - form expression for the capacity of the @xmath3-bit quantized gaussian channel , except for the binary case where the channel output is quantized using a symmetric threshold quantizer ( * ? ? ?
* th . 2 ) . a ubiquitous quantizer is the _ uniform quantizer _ , whose levels are equispaced , say @xmath0 apart , either with an infinite or a finite number of levels .
we refer to @xcite for a comprehensive survey of quantization theory . for finite - level uniform quantizers
, the outermost cells will be semi - infinite and the input space corresponding to these cells is referred to as the _ overload region _ @xcite .
while infinite - level uniform quantizers need an infinite number of bits to describe their output and seem therefore impractical , they have the advantage of eliminating the overload region and resulting overload distortion ( * ? ? ?
for this reason , infinite - level uniform quantizers are typically preferred in theoretical analyses , in the hope that the tail of the source to be quantized decays sufficiently fast so the overload distortion be negligible . by shannon s source coding theorem @xcite , irrespective of the number of levels
, the output of a uniform quantizer can be described by a variable - length code whose expected length is roughly the entropy of the quantizer output .
consequently , the rate of a quantizer is often measured by the entropy of its output . the step size @xmath0 of the uniform quantizer determines its precision : the smaller @xmath0 , the higher the precision .
the high - precision limit ( where @xmath4 ) was studied by gish and pierce @xcite , who showed that the difference between the entropy of the output of an infinite - level uniform quantizer and the rate distortion function converges to @xmath5 as the permitted distortion ( and hence also @xmath0 ) vanishes .
as for the low - precision limit ( where @xmath6 ) , it was shown that for exponential , laplacian , and gaussian sources the entropy of the quantizer output approaches zero with the same slope as the rate - distortion function as the allowed distortion tends to the source variance , whereas for uniform sources the slope of the entropy of the quantized output becomes infinite , in contrast to the rate - distortion function which has a finite slope @xcite@xcite . to prove their result for gaussian sources @xcite , marco and neuhoff showed that , in the low - precision limit , the entropy of the quantizer output is determined by the probabilities corresponding to the innermost cells , which is in agreement with the intuition that if the tail of the source decays sufficiently fast , then the overload distortion can be neglected ( * ? ? ?
* lemma 3 ) .
a common strategy to further simplify the theoretical analysis of uniform quantizers is _
dithering_. ( we refer again to ( * ? ? ?
v - e ) for a survey of this topic . ) in a dithered quantizer , instead of quantizing an input signal directly , one quantizes the sum of the signal and a random process ( called a _ dither _ ) that is independent of the signal .
this allows one to describe the quantization noise by additive uniform noise that is independent of the input signal . specifically , if the dither is uniformly distributed over @xmath7 $ ] , then the conditional entropy of the quantizer output given the dither is equal to the mutual information between the quantizer input and the sum of the input and independent , uniformly distributed noise ( * ? ? ?
dithered quantization was studied in numerous works .
of particular interest to us is the work by zamir and feder @xcite , which studied the rate - distortion behavior when a bandlimited stationary source is first sampled at nyquist rate or faster , then it undergoes dithered uniform quantization , and finally it is entropy - encoded .
generalizations of dithered quantization can be found , e.g. , in @xcite , @xcite .
observe that analyses of the capacity of the quantized gaussian channel are motivated by the need for low - resolution quantizers and therefore typically consider quantizers with a small number of levels .
however , the analysis of such quantizers becomes intractable as quantizer resolution and/or sampling rate increase .
in contrast , theoretical work on quantization often considers infinite - level uniform quantizers , since they allow for a simplified analysis . in this paper
, we bring together these two lines of research by studying the capacity of the gaussian channel when its output is quantized using a dithered , infinite - level , uniform quantizer of step size @xmath0 .
( we shall refer to this channel as the _ dither - quantized gaussian channel_. ) since a dithered quantizer can be described as an additive noise channel with uniform noise , the dither - quantized gaussian channel is equivalent to an additive noise channel where the noise is the sum of a gaussian and a uniform random variable .
this simplifies the analysis of its capacity . while beyond the scope of this paper
, we hope that , in the long term , studying the capacity of the dither - quantized gaussian channel will help us better understand the tradeoff in channel capacity between sampling rate and quantization resolution of the continuous - time , bandlimited , awgn channel .
the rest of this paper is organized as follows .
section [ sec : channel ] introduces the channel model and defines the capacity as well as the low - snr asymptotic capacity .
section [ sec : capacity ] presents the results ( as well as the proofs thereof ) that concern channel capacity .
section [ sec : lowsnr ] presents the results ( as well as the proofs thereof ) that concern the low - snr asymptotic capacity .
section [ sec : conclusion ] concludes the paper with a summary and a discussion of our results .
[ fig : system ] we consider the discrete - time communication system depicted in figure 1 .
a message @xmath8 , which is uniformly distributed over the set @xmath9 , is mapped by an encoder to the length-@xmath10 real sequence @xmath11 of channel inputs .
( here , @xmath12 denotes the set of real numbers . )
the channel corrupts this sequence by adding gaussian noise to produce the unquantized output sequence @xmath13 where @xmath14 is a sequence of independent and identically distributed ( i.i.d . ) gaussian random variables of mean zero and variance @xmath15 .
( here , @xmath16 denotes the set of integers . )
the unquantized sequence is then quantized using a dithered , infinite - level , uniform quantizer of step size @xmath0 .
specifically , the quantizer is a function @xmath17 that produces @xmath18 if @xmath19 , i.e. , @xmath20 where , for every @xmath21 , @xmath22 denotes the largest integer not larger than @xmath23 . , @xmath24 ,
since this choice minimizes the expected squared error .
for ease of exposition , we use the slightly simpler definition . in any case
, the actual reproduction values do not affect the achievable information rates . ] the quantizer output @xmath25 is given by @xmath26 where @xmath27 is a sequence of i.i.d .
random variables that are uniformly distributed over the interval @xmath7 $ ] , referred to as _
dither_. we assume that channel input , additive gaussian noise , and dither are independent .
the decoder observes the quantizer output @xmath28 as well as the dither @xmath29 and guesses which message was transmitted .
we impose both an average - power and a peak - power constraint on the channel inputs : for every realization of @xmath8 , the sequence @xmath30 must satisfy @xmath31 the capacity of the dither - quantized gaussian channel under the power constraints @xmath32 and @xmath33 on the channel inputs is given by ( * ? ? ?
7.3 ) @xmath34 where the supremum is over all distributions of @xmath35 satisfying @xmath36 and @xmath37 with probability one . here and throughput the paper , we omit the time indices where they are immaterial .
when the peak - power constraint is relaxed ( @xmath38 ) , we shall denote the capacity by @xmath39 . in an analogous manner , we shall denote the capacity of the gaussian channel under the power constraints @xmath32 and @xmath40 by @xmath41 , i.e. , @xmath42 where the supremum is over all distributions of @xmath35 satisfying @xmath36 and @xmath37 with probability one .
we shall omit the second argument when the peak - power constraint is relaxed , i.e. , @xmath43 . by the data processing inequality ( * ? ? ?
2.8.1 ) , @xmath44 while it is well - known that the input distribution achieving @xmath41 is discrete @xcite , to the best of our knowledge , there exists no closed - form expression for @xmath41 .
nevertheless , by relaxing the peak - power constraint , we obtain for every @xmath32 and @xmath40 @xcite @xmath45 here and throughout this paper , @xmath46 denotes the natural logarithm function .
( consequently , all rates are in nats per channel use . ) in section [ sub : main_capacity ] , we demonstrate that the inequality in becomes tight as @xmath4 and that @xmath47 tends to zero as @xmath6 .
since a dithered quantizer can be described as an additive noise channel with uniform noise @xmath48 , the dither - quantized gaussian channel is equivalent to an additive noise channel with noise @xmath49 . indeed , following the proof of theorem 1 in @xcite
, we show in appendix [ app : dither_add_noise ] that the mutual information on the right - hand side ( rhs ) of is equal to @xmath50 where the probability density function ( pdf ) of the additive noise @xmath51 is the convolution of the gaussian and the uniform pdf : @xmath52.\ ] ] here @xmath53 denotes the _ gaussian probability integral _ ( @xmath54-function )
* eq . ( 1.3 ) ) .
in addition to capacity , we also study the slope of the capacity - vs - power curve at zero when either the peak - power constraint is relaxed ( @xmath38 ) or when the peak - to - average - power ratio @xmath55 is finite and held fixed , i.e. , @xmath56 and @xmath57 we shall refer to the slope of the capacity - vs - power curve at zero as the _ low - snr asymptotic capacity_. relaxing the peak - power constraint allows for a simple expression for @xmath58 ( * ? ? ?
3 ) : @xmath59 where @xmath60 denotes relative entropy and @xmath61 denotes the conditional distribution of @xmath62 given @xmath63 .
unfortunately , @xmath58 may characterize @xmath39 only at impractically small input powers @xmath32 .
indeed , if the supremum on the rhs of is approached only as @xmath64 ( as is the case for the 1-bit quantized gaussian channel ( * ? ? ?
3 ) ) , then the input distribution that achieves the first derivative of @xmath39 at zero ( i.e. , @xmath58 ) must be flash signaling , which implies that the second derivative of @xmath39 at zero is @xmath65 @xcite . consequently , in such cases , @xmath58 describes the behavior of @xmath39 poorly , unless @xmath32 is very small . to address this problem ,
we consider also the case where the peak - to - average - power ratio @xmath66 is finite and held fixed , thereby precluding the use of flash signaling input distributions . in this case
, it was demonstrated that if the channel law satisfies a number of technical conditions , then the low - snr asymptotic capacity is given by @xcite , @xcite @xmath67 where @xmath68 denotes the fisher information @xmath69 ^ 2}{f_{z_{\delta}}(y - x ) } \d y.\ ] ] by and , and by noting that relaxing the peak - power constraint does not reduce capacity , it follows that @xmath70 in section [ sub : main_lowsnr ] , we demonstrate that the right - most inequality holds with equality irrespective of @xmath0 , while the left - most inequality holds with equality if , and only if , @xmath0 vanishes .
in this section , we study the capacity for arbitrary input powers @xmath32 in the high- and low - resolution limit , i.e. , when @xmath4 and @xmath6 , respectively .
we show that in the former case , the capacity @xmath47 converges to that of the gaussian channel , and in the latter case , it converges to zero .
[ thm : cap_highres ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , for any distribution on @xmath35 satisfying @xmath36 , @xmath71 recall that @xmath49 . to prove theorem [ thm : cap_highres ]
, it thus suffices to show that rcl _ 0 h(x+n+u _ ) & = & h(x+n ) [ eq : thm_highres1 ] + _ 0 h(n+u _ ) & = & h(n).[eq : thm_highres2 ] since @xmath72 is gaussian and @xmath35 and @xmath72 are independent , the differential entropies on the rhs of and are both finite .
furthermore , @xmath73 , @xmath74 and , by the theorem s assumption , @xmath75 . the above identities and follow therefore directly by specializing the proof of theorem 1 in @xcite to the distortion measures @xmath76 . equation holds for any input distribution satisfying the average - power constraint @xmath32 , including the capacity - achieving input - distribution .
consequently , theorem [ thm : cap_highres ] implies that the inequality in becomes tight as @xmath4 .
[ cor : highres ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , for every @xmath32 and @xmath40 , @xmath77 in view of , it suffices to show that @xmath78 where @xmath79 denotes the _ limit inferior_. to this end , we use that , by theorem [ thm : cap_highres ] , we have for any distribution of @xmath35 satisfying @xmath36 and @xmath37 with probability @xmath80 the lower bound , and hence corollary [ cor : highres ] , follows by maximizing the rhs of over all distributions of @xmath35 satisfying the power constraints @xmath32 and @xmath40 .
theorem [ thm : cap_highres ] and corollary [ cor : highres ] demonstrate that , in the high - resolution limit , the dithered quantizer incurs no loss in capacity . as we show next ,
this is in stark contrast to the low - resolution limit .
[ thm : cap_lowres ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , for every @xmath32 and @xmath40 , @xmath81 see section [ sec : cap_lowres ] .
let the signal - to - noise - and - quantization - noise - ratio ( snqnr ) of the dither - quantized gaussian channel be defined as @xmath82 theorem [ thm : cap_lowres ] is perhaps not very surprising since the snqnr tends to zero as @xmath0 tends to infinity , so it may seem plausible that also the capacity vanishes in the low - resolution limit .
however , note that the additive noise @xmath51 is non - gaussian , so it is _ prima facie _ unclear whether there is any relation between capacity and snqnr .
the weak performance of the dithered , infinite - level , uniform quantizer at low quantizer resolutions is due to the dither .
indeed , when the output of the gaussian channel is quantized using a symmetric threshold quantizer , the capacity is given by ( * ? ? ?
2 ) @xmath83 the rhs of is strictly positive , so this implies that @xmath84 .
moreover , since the concatenation of an infinite - level , uniform quantizer and a symmetric threshold quantizer results again in a threshold quantizer , it follows that the _ undithered _ uniform quantizer achieves a capacity that is at least as large as the capacity achieved by the 1-bit quantizer .
consequently , adding dither is highly detrimental in the low - resolution regime .
as we shall see , the same is also true for the low - snr asymptotic capacity , unless the peak - to - average - power ratio is unbounded .
we first note that @xmath85 has the same distribution as @xmath48 .
recalling that @xmath49 , it thus follows that @xmath86 we then prove theorem [ thm : cap_lowres ] by showing that @xmath87 where the supremum is over all distributions of @xmath35 satisfying @xmath36 and @xmath37 with probability one .
to prove , we will follow the steps that were carried out in ( * ? ? ?
ii ) to derive an upper bound on the capacity of the peak - and - average - power - limited complex gaussian channel .
specifically , we use the upper bound on the mutual information ( * ? ? ?
5.1 ) @xmath88 where @xmath53 denotes the input distribution ; @xmath89 denotes the conditional distribution of the channel output , conditioned on @xmath63 ; and @xmath90 denotes some arbitrary distribution on the output alphabet .
every choice of @xmath90 yields an upper bound on @xmath91 , and the inequality in holds with equality if @xmath90 is the actual distribution of @xmath92 induced by @xmath53 and @xmath93 . here
, we choose @xmath90 such that its pdf is @xmath94 for some @xmath95 and @xmath96 , where @xmath97 is a normalizing constant , i.e. , @xmath98 \label{eq : thm_cap_lowres_2.2}\ ] ] and @xmath99 denotes the inverse tangent function .
combining with , and using that conditioning does not increase entropy , we obtain upon substituting @xmath100 lcl i(x;(x+n)+u_1 ) & = & - h((x+n)+u_1|x ) - + & & -h((x+n)+u_1| x , n ) - + & = & - [ eq : thm_cap_lowres_2.5 ] where the last step follows because @xmath101 is independent of @xmath102 , so ( * ? ?
9.6.3 ) and the expression for the differential entropy of a uniform random variable give @xmath103 we next evaluate lcl + & = & + ( |y| > ) + [ eq : thm_cap_lowres_3 ] where @xmath104 denotes the indicator function .
when @xmath105 , then is equal to @xmath106 and give @xmath107\right ) .
\label{eq : thm_cap_lowres_3.7}\ ] ] in the following , we consider the case where @xmath108 . by the triangle inequality ,
the absolute value of @xmath109 is upper - bounded by @xmath110 .
furthermore , @xmath111 .
consequently , @xmath112 where the right - most inequality follows by chebyshev s inequality ( * ? ? ?
* ( 4.10.7 ) , p. 192 ) and because , for every @xmath35 satisfying @xmath113 , we have @xmath114 . for ease of exposition
, we define @xmath115 .
since @xmath116 , @xmath96 , applying to thus gives lcl - & & + ^2 ( ) + .
[ eq : thm_cap_lowres_5 ] to upper - bounded the last term on the rhs of , we use jensen s inequality to obtain @xmath117\right).\ ] ] by bayes law , we have @xmath118 = \frac{\e{y^2\i{|y|>\alpha}}}{\prob(|y|>\alpha ) } \leq \frac{\eps^2(\mathsf{p}+\sigma^2)+\frac{1}{12}}{\prob(|y|>\alpha)}\ ] ] where we used in the right - most inequality that @xmath119 and that , for every @xmath35 satisfying @xmath36 , the second moment of @xmath92 is upper - bounded by @xmath120 . combining with
then gives lcl & & ( |y|>)(1 + ) + & = & ( |y|>)((|y|>)+ ) + & & - ( |y|>)(|y| > ) + & & ^2()(1 + ) + _ 0<^2 ( ) || [ eq : thm_cap_lowres_8 ] where the last step follows by maximizing @xmath121 over all @xmath122 satisfying and because , by , @xmath123 . combining and with ,
we obtain for @xmath108 that lcl + & & + ^2()+ _ 0<^2 ( ) || .
[ eq : thm_cap_lowres_9 ] since the rhs of is not smaller than the rhs of , it follows that lcl + & & + ^2()+ _ 0<^2 ( ) || [ eq:40 ] where the supremum on the left - hand side ( lhs ) of is over all distribution of @xmath35 satisfying @xmath36 and @xmath37 with probability one .
since the function @xmath124 is continuous for @xmath125 and vanishes as @xmath126 , it follows that @xmath127\right ) \label{eq : thm_cap_lowres_10}\ ] ] where @xmath128 denotes the _ limit superior _ and where we have substituted @xmath97 by the rhs of .
the claim , and hence theorem [ thm : cap_lowres ] , follows from by letting first @xmath129 and then @xmath130 .
in this section , we discuss capacity at low input powers @xmath32 .
we show that , when the peak - power constraint is relaxed , the low - snr asymptotic capacity is equal to that of the gaussian channel irrespective of @xmath0 .
we further derive an expression for the low - snr asymptotic capacity for finite peak - to - average - power ratios and evaluate it in the low- and high - resolution limit .
we demonstrate that , in this case , the low - snr asymptotic capacity converges to that of the unquantized channel when @xmath0 tends to zero , and it tends to zero when @xmath0 tends to infinity .
[ thm : cue_thm1 ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , irrespective of @xmath131 , @xmath132 see section [ sec : cue_thm1 ] . theorem [ thm : cue_thm1 ] is reminiscent of theorem 2 in @xcite , which states that the low - snr asymptotic capacity of the 1-bit quantized gaussian channel equals @xmath133 , provided that we allow for flash - signaling input distributions .
as noted before , the concatenation of a uniform and a 1-bit quantizer results again in a 1-bit quantizer , so theorem [ thm : cue_thm1 ] may perhaps not be very surprising .
however , in general it is unclear how a _ dithered _ uniform quantizer compares to a 1-bit quantizer , since the dither potentially reduces capacity .
in fact , as we shall see next , for finite a peak - to - average - power ratio and as @xmath0 becomes large , the dither significantly reduces the low - snr asymptotic capacity .
[ thm : cue_thm2 ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , irrespective of @xmath66 , @xmath134 ^ 2}{q\left(\frac{y-\delta/2}{\sigma}\right)-q\left(\frac{y+\delta/2}{\sigma}\right ) } \d y.\ ] ] see section [ sec : cue_thm2 ] . observe that for a finite peak - to - average - power ratio , the low - snr asymptotic capacity depends on @xmath0 .
we next study the behavior of @xmath135 as @xmath4 and @xmath6 .
[ cor ] consider the dither - quantized gaussian channel described in section [ sec : channel ] .
then , rrcl & _ 0 _ ^()(0 ) & = & + & _ _
^()(0 ) & = & 0 . see section [ sub : proof_cor ] .
corollary [ cor ] demonstrates that , for finite peak - to - average - power ratios , the low - snr asymptotic capacity of the dither - quantized gaussian channel approaches that of the gaussian channel in the high - resolution limit and it vanishes in the low - resolution limit .
the latter result is in stark contrast to proposition 2 in @xcite ( see also @xcite,@xcite ) , which demonstrates that for a 1-bit quantizer and @xmath136 , the low - snr asymptotic capacity equals @xmath137 .
thus , for finite peak - to - average - power ratios , a low - resolution dithered quantizer performs significantly worse than a 1-bit quantizer . we shall show that @xmath138 theorem [ thm : cue_thm1 ] follows then from , , and .
let @xmath139 for some arbitrary @xmath140 . by the data processing inequality for relative entropy ( * ?
2.9 ) lcl d(p_x+z_|x = xp_x+z_|x=0 ) & & d(p_v|x = xp_v|x=0 ) [ eq : thm1_2 ] where @xmath141 denotes the conditional distribution of @xmath142 given @xmath63 .
intuitively , @xmath142 can be viewed as the output of a threshold quantizer with threshold @xmath143 and input @xmath62 .
introducing @xmath142 thus allows us to analyze the rhs of following similar steps as the ones reported in ( * ? ? ?
viii - a ) . indeed , as in (
* eq . ( 134 ) ) , we can express the relative entropy as lcl d(p_v|x = xp_v|x=0 ) & = & + & & + p_v|x(1|x ) - h_(p_v|x(1|x ) ) [ eq : thm1_3 ] where @xmath46 denotes the natural logarithm function ; @xmath144 denotes the binary entropy function ( * ? ? ?
* eq . ( 2.5 ) ) ; and @xmath145 , which can be written as @xmath146 using that @xmath147 , @xmath24 and @xmath148 , @xmath149 , can be further lower - bounded as lcl d(p_v|x = xp_v|x=0 ) & & p_v|x(1|x ) - 2.[eq : thm1_4 ] we next choose @xmath150 and lower - bound the supremum in by letting @xmath151 tend to infinity .
together with and , this yields lcl _ x0 & & _ _ 0 .
[ eq : thm1_5 ] by and the monotonicity of the @xmath54-function , we obtain @xmath152 moreover , by and the following bounds on the @xmath54-function ( * ? ? ?
19.4.2 ) @xmath153 we have for sufficiently large @xmath151 @xmath154 applying and to yields lcl + & & _ _ 0 + & = & q(- ) .
[ eq : thm1_last ] the final result , and hence theorem [ thm : cue_thm1 ] , follows from by letting @xmath155 tend to infinity . in order for to hold , for every @xmath131 , the channel law must satisfy six conditions ( * ? ? ?
ii ) : 1 . the channel can be described by a pdf @xmath156 .
the pdf @xmath157 is bounded for all @xmath158 ( for some @xmath159 ) and @xmath160 .
3 . the partial derivative @xmath161 exists for all @xmath158 and @xmath160 .
the fisher information exists and is finite for all @xmath158 . 5 .
the function @xmath162 is uniformly continuous in the mean square with respect to @xmath158 .
6 . for any @xmath163 , @xmath164 ^ 2}{f_{y|x}(y|x ) } \d y \d x = 0\ ] ] where @xmath165 note that , for the channel model described in section [ sec : channel ] , we have @xmath166 .
thus , conditions a and b follow directly by inspecting . furthermore , using that @xmath167 , we obtain from that lcl f_y|x(y|x ) & = & [ eq : thm2_1 ] which proves condition c. this also demonstrates that the fisher information exists and is given by @xmath168 ^ 2}{q\left(\frac{y - x-\delta/2}{\sigma}\right)-q\left(\frac{y - x+\delta/2}{\sigma}\right ) } \d y.\ ] ] to prove condition d , it thus remains to show that the fisher information is finite for all @xmath158 .
this , as well as conditions e and f , require slightly more involved proofs , which are presented in appendix [ app : thm2 ] . having proven conditions a - f , theorem [ thm : cue_thm2 ] follows directly by combining with . to prove part i ) , we note that , by , we have @xmath169 .
it thus remains to show that @xmath170 to this end , we use fatou s lemma ( * ? ? ?
* ( 1.6.8 ) , p. 50 ) to lower - bound as lcl _ 0 _
^()(0 ) & & _ -^ _ 0 y. we next apply lhpital s rule twice to compute the limit inside the integral . indeed , we have lcl + & = & 2 and lcl + & = & q()-q ( ) + which both tend to zero as @xmath4 .
we further have lcl ^2 & = & 2 ^ 2 + + [ eq : cor_hr_1 ] and lcl & = & + .[eq : cor_hr_2 ] noting that , as @xmath4 , and tend to @xmath171 and @xmath172 , respectively , lhpital s rule gives @xmath173 ^ 2}{q\left(\frac{y-\delta/2}{\sigma}\right)-q\left(\frac{y+\delta/2}{\sigma}\right ) } = \frac{2\frac{y^2}{\sigma^4 } e^{-\frac{y^2}{\sigma^2}}}{\sqrt{\frac{2}{\pi\sigma^2}}e^{-\frac{y^2}{2\sigma^2 } } } = \sqrt{2\pi\sigma^2}\frac{y^2}{\sigma^4}e^{-\frac{y^2}{2\sigma^2}}. \label{eq : cor_hr_3}\ ] ] integrating from @xmath65 to @xmath174 yields @xmath175 ^ 2}{q\left(\frac{y-\delta/2}{\sigma}\right)-q\left(\frac{y+\delta/2}{\sigma}\right ) } \d y = \frac{1}{2\sigma^2}\frac{1}{\sqrt{2\pi\sigma^2}}\int_{-\infty}^{\infty } \frac{y^2}{\sigma^2 } e^{-\frac{y^2}{2\sigma^2 } } \d y = \frac{1}{2\sigma^2}\ ] ] where the last step follows by identifying the terms after @xmath133 as the variance of a zero - mean , variance-@xmath15 , gaussian random variable divided by @xmath15 , which is equal to one .
this proves , which in turn proves part i ) of corollary [ cor ] . to prove part
ii ) , it suffices to show that the integral on the rhs of is bounded in @xmath0 . to this end
, we first note that the integrand in is symmetric in @xmath176 , so @xmath177 ^ 2}{q\left(\frac{y-\delta/2}{\sigma}\right)-q\left(\frac{y+\delta/2}{\sigma}\right ) } \d y = \frac{1}{4\pi\sigma^2}\int\limits_{-\infty}^{\infty } \frac{\left[e^{-\frac{(|y|-\delta/2)^2}{2\sigma^2}}-e^{-\frac{(|y|+\delta/2)^2}{2\sigma^2}}\right]^2}{q\left(\frac{|y|-\delta/2}{\sigma}\right)-q\left(\frac{|y|+\delta/2}{\sigma}\right ) } \d y.\label{eq : cor_lr_0}\ ] ] we next divide the integration region into the two regions @xmath178 for a sufficiently large @xmath179 and analyze the corresponding integrals separately . for @xmath180 , we use the monotonicity of the @xmath54-function to lower - bound @xmath181 where , for @xmath182 , the rhs of is strictly positive and tends to @xmath183 as @xmath6 .
together with the identity @xmath184 , this yields lcl + & & _ _ 1 e^- + e^- y + & & [ eq : cor_lr_y1 ] where the last inequality follows by enhancing the integration region from @xmath185 to @xmath12 .
we next consider the case where @xmath186 . by
, we have for @xmath187 lcl + & & ( 1-)e^- - e^- + & & e^- + & & e^- , y_2 [ eq : cor_lr_2 ] where @xmath188 which , for sufficiently large @xmath189 , is strictly positive and tends to @xmath190 as @xmath6 . the last inequality in
follows because @xmath191 , @xmath192 and because the function @xmath193 is monotonically decreasing in @xmath194 .
we further note that , for @xmath187 , @xmath195 by and , lcl + & & _ _ 2 e^-y + & = & e^- .
[ eq : cor_lr_y2 ] combining and with , we obtain @xmath177 ^ 2}{q\left(\frac{y-\delta/2}{\sigma}\right)-q\left(\frac{y+\delta/2}{\sigma}\right ) } \d y \leq \frac{1}{\sqrt{2\pi\sigma^2}}\left(\frac{\sqrt{2}}{q\bigl(\vartheta/\sigma\bigr)-q\bigl(\delta/(2\sigma)\bigr ) } + \frac{e^{-\frac{\vartheta^2}{2\sigma^2}}}{\mu_{\vartheta}(\delta)}\right ) .
\label{eq : cor_lr_final}\ ] ] part ii ) of corollary [ cor ] follows then by noting that , for sufficiently large @xmath189 , the rhs of is bounded in @xmath0 .
we have studied both the capacity and the low - snr asymptotic capacity of the peak - and - average - power - limited gaussian channel when its output is quantized using a dithered , infinite - level , uniform quantizer of step size @xmath0 .
we have demonstrated that the capacity of the dither - quantized channel converges to the capacity of the unquantized channel in the high - resolution limit ( @xmath4 ) , and it converges to zero in the low - resolution limit ( @xmath6 ) .
we have further demonstrated that , when the peak - power constraint is absent , the low - snr asymptotic capacity of the dither - quantized channel is equal to that of the unquantized channel irrespective of @xmath0 .
in contrast , for finite peak - to - average - power ratios , the low - snr asymptotic capacity of the dither - quantized channel depends critically on @xmath0 : as we show , it converges to the low - snr asymptotic capacity of the unquantized channel in the high - resolution limit , but it vanishes in the low - resolution limit . while dithered , infinite - level , uniform quantizers seem impractical due to the infinite number of bits required to describe their outputs , studying their behavior may help us better understand the behavior of quantizers with a small number of levels , provided that both type of quantizers have similar behaviors .
our results suggest that , with respect to channel capacity , this is the case in the high - resolution limit , but it is not the case in the low - resolution limit .
for example , the capacity of the 1-bit quantized gaussian channel ( with a symmetric threshold quantizer ) is given by , which differs not only from the capacity of the dither - quantized gaussian channel in the low - resolution limit for a given @xmath32 ( which is zero ) , but it also has a distinct asymptotic behavior as @xmath32 tends to zero .
since the concatenation of an infinite - level , uniform quantizer and a 1-bit quantizer results again in a 1-bit quantizer , we conclude that the inferior performance at low quantizer resolutions of the dithered , infinite - level , uniform quantizer is due to the dither . in other words , in the low - resolution regime , adding dither is highly detrimental . nevertheless , sampling the output of the dithered , infinite - level , uniform quantizer above nyquist rate , as studied in @xcite , may perhaps improve the performance in this regime , since such an approach reduces the quantization noise without increasing the quantizer resolution .
we shall prove by showing that lcl h(y_|u _ ) & = & h(x+z _ ) - [ eq : app_h|u ] + h(y _
|u_,x ) & = & h(z _ ) - .
[ eq : app_h|u , x ] the proof of and is almost identical to the proof of theorem 1 in @xcite . for the sake of completeness , we repeat it here .
first note that , since @xmath35 and @xmath72 are independent and @xmath72 is gaussian , it follows by ( * ? ?
* th . 4.10 ) that the distribution of the random variable @xmath196 is absolutely continuous with respect to the lebesgue measure , so its pdf , which we shall denote by @xmath197 , is defined .
furthermore , the pdf of @xmath198 relates to @xmath197 via ( * ? ? ?
* th . 4.10 ) @xmath199 where @xmath200 denotes the pdf of @xmath48 , i.e. , @xmath201 , @xmath202 .
( recall that @xmath48 is uniformly distributed over @xmath7 $ ] and @xmath49 . ) likewise , the conditional probability of @xmath203 given @xmath204 is equal to @xmath205 which together with yields @xmath206 we next use and fubini s theorem ( * ? ? ?
* ( 2.6.6 ) , p. 108 ) to express the conditions entropy of @xmath92 given @xmath48 as lcl h(y_|u _ ) & = & - _ -/2^/2 _
i=-^ p_y|u_(i |u ) p_y|u_(i |u ) u + & = & - _ i=-^ _
-/2^/2 p_y|u_(i |u ) p_y|u_(i |u ) u + & = & - - _
i=-^_-/2^/2 f_x+z_(i+/2-u ) f_x+z_(i+/2-u)u .
[ eq : app1_h|u ] by the change of variable @xmath207 , it then follows that lcl h(y_|u _ ) & = & - - _
i^(i+1 ) f_x+z _ ( ) f_x+z _ ( ) + & = & -- h(x+z _ ) .
[ eq : app1_h|u ] this proves .
the second identity follows along similar lines .
indeed , we have @xmath208 where @xmath209 denotes the pdf of @xmath72 .
furthermore , the conditional probability of @xmath203 given @xmath210 is @xmath211 which together with yields @xmath212 analog to and , we obtain from that for every @xmath24 lcl h(y_|u_,x = x ) & = & -- _
i=-^ _ i - x^(i+1)-x f_z_()f_z _ ( ) + & = & - -h(z _ ) . averaging over @xmath35 , this yields .
in this appendix , we prove the conditions stated in section [ sec : cue_thm2 ] that require more involved proofs . specifically , section [ sub : cond_d ] demonstrates that the fisher information is finite for all @xmath213 , which together with proves condition d. section [ sub : cond_e ] proves condition e and section [ sub : cond_f ] proves condition f. throughout this appendix , we shall use the following notation .
we denote the partial derivative of @xmath157 with respect to @xmath214 by @xmath215 .
we further omit the subscript of @xmath157 to keep notation compact .
finally , we define the sets @xmath216 for some arbitrary @xmath189 .
the fisher information @xmath217 is given by , namely , @xmath218
^ 2}{q\left(\frac{y - x-\delta/2}{\sigma}\right)-q\left(\frac{y - x+\delta/2}{\sigma}\right ) } \d y. \label{eq : appd_0}\ ] ] to prove that @xmath68 is finite for all @xmath213 and @xmath131 , we divide the integration region into @xmath185 and @xmath219 , for some sufficiently large @xmath179 , and show that the corresponding integrals are finite for all @xmath213 . since the @xmath54-function is continuous and @xmath185 is a closed and bounded interval , it follows from the extreme value theorem that for every @xmath180 and @xmath220 @xmath221 for some @xmath222 $ ] . together with , this yields for every @xmath180 and @xmath213 lcl f(y|x ) _ .
[ eq : appd_1 ] by the strict monotonicity of @xmath53 , it further follows that @xmath223 .
we thus have lcl + & & _ _ 1 ^2 y + & & [ eq : appd_y1 ] where the second inequality follows because @xmath224 we next consider the case where @xmath186 . to this end , we first note that the pdf @xmath225 is symmetric in @xmath226 , so it can be written as @xmath227.\ ] ] using , this can be lower - bounded as lcl f_z_(z ) & & ( 1- ) e^- - e^- + & = & e^- .
[ eq : appd_2 ] note that the term inside the square brackets on the rhs of tends to one as @xmath228 .
since by the triangle inequality @xmath229 for @xmath186 and @xmath213 , it follows that for any @xmath230 there exists a sufficiently large @xmath189 such that @xmath231 applying to , and using that the integrand is symmetric in @xmath176 , we obtain lcl + & & _ _ 2 ( |y - x|-/2)e^^2 y. + & & _ _ 2 ( |y - x|-/2)e^- y [ eq : appd_4 ] where the last step follows because , for sufficiently large @xmath189 , we have @xmath232 , which implies that @xmath233 let @xmath234 and @xmath235 . by a change of variables , can be further upper - bounded by lcl + & = & _ _ 2 ( |z|-/2)e^- z + & & _ |z|>-(|z|-/2)e^- z + & = & e^- [ eq : appd_y2 ] where the inequality follows because , by the triangle inequality , @xmath236 and because for @xmath237 and a sufficiently large @xmath189 , the term @xmath238 is nonnegative . combining and , we obtain for every @xmath213 and a sufficiently large @xmath189 @xmath239 thus , the fisher information @xmath68 is finite for all @xmath213 . by the chain rule , it follows that @xmath240 to prove condition e , we need to show that for every @xmath131 ( * ? ? ?
* eq . ( 2.3 ) ) @xmath241 ^ 2 \d y \to 0\ ] ] as @xmath242 and @xmath243 . since , for all @xmath160 , @xmath244 and @xmath245 are both bounded and continuous functions of @xmath214 and @xmath246 ( for some arbitrary @xmath159 ) , it follows that @xmath247 ^ 2 = 0 , \quad y\in\reals.\ ] ] to prove , it thus suffices to show that there exists an integrable function @xmath248 that upper - bounds @xmath249 ^ 2 \leq g(y ) , \quad y\in\reals\ ] ] for all @xmath250 and @xmath251.the claim , and hence condition e , follows then by the dominated convergence theorem ( * ? ? ?
* ( 1.6.9 ) , p. 50 ) . to prove , we follow the approach carried out in section [ sub : cond_d ] and divide the integration region into @xmath185 and @xmath219 ( for some sufficiently large @xmath179 ) and evaluate the corresponding integrals separately . for @xmath180 , we use the identity @xmath184 , , and to upper - bound lcl + & & + + & & ( ^2+^2 ) + & & [ eq : appe_y1 ] which is integrable over the bounded set @xmath185 .
we next consider the case where @xmath186 .
we first note that for any @xmath230 there exists a sufficiently large @xmath189 such that holds . using this result together with the identity @xmath184 and
, we obtain for sufficiently larger @xmath189 lcl + & & + + & & ( |y - x_1|-/2)e^^2 + & & + ( |y - x_2|-/2)e^^2 + & & ( ( |y - x_1|-/2)e^- + ( |y - x_2|-/2)e^- ) . [
eq : appe_107 ] since , by the triangle inequality , @xmath252 , it follows that for all @xmath253 and @xmath254 @xmath255 ^ 2 \leq \frac{1}{2\pi\sigma^2}\frac{\sqrt{2\pi}}{\mu_{\delta}\sigma } ( |y|+\eps-\delta/2)e^{-\frac{(|y|-\eps-\delta/2)^2}{2\sigma^2}}. \label{eq : appe_y2}\ ] ] note that the rhs of is integrable over @xmath186 . combining and
, it follows that the integrable function @xmath256 satisfies for all @xmath250 and @xmath251 .
this demonstrates that condition e is satisfied .
we upper - bound the left - hand side of by deriving an upper bound on @xmath257 ^ 2}{f(y|x ) } \d y\ ] ] that holds for sufficiently small @xmath258 and that is independent of @xmath24 and @xmath163 . to this end
, we divide the integration region into @xmath259 and @xmath260 and evaluate each integral separately .
we then show that the resulting upper bound vanishes as @xmath189 tends to infinity , thereby proving condition f. we begin by showing that for any @xmath163 and @xmath179 there exists a sufficiently small @xmath261 such that for all @xmath262 @xmath263 where @xmath264 denotes the empty set .
consequently , @xmath265 ^ 2}{f(y|x ) } \d y=0 , \quad \eps<\eps_0.\ ] ] to this end , we approximate @xmath266 for every @xmath180 by a taylor series around @xmath267 : @xmath268 for some @xmath269 , where we use the lagrange form of the remainder .
consequently , @xmath270 by , , and , it follows that @xmath271 for some @xmath223 , which implies that @xmath272 using the inequality @xmath273 it follows from , , and the monotonicity of @xmath274f that , for @xmath275 , lcl || & = & |(1+x)| + & & + & & .
[ eq : appf_7 ] the rhs of vanishes as @xmath276 , so for any fixed @xmath163 and @xmath179 , there exists an @xmath261 such that @xmath277 together with the definition of @xmath278 in , this proves .
we continue by evaluating the integral for @xmath186 . to this end
, we use that the integrand is nonnegative and that @xmath279 to upper - bound lcl _ _ , _ 2 y & & _ _ 2 y. [ eq : appf_8 ] repeating the steps in section [ sub : cond_e ] ,
the rhs of can be further upper - bounded by @xmath280 ^ 2}{f(y|x ) } \d y \leq \sqrt{\frac{2}{\pi\sigma^2 } } \frac{1}{\mu_{\delta}(\vartheta ) } e^{-\frac{(\vartheta-\eps-\delta/2)^2}{2\sigma^2 } } \label{eq : appf_y2}\ ] ] where @xmath281 is a positive function that tends to @xmath282 as @xmath283 . combining with and
, we obtain that for every @xmath163 and sufficiently large @xmath179 , there exists an @xmath261 such that @xmath257 ^ 2}{f(y|x ) } \d y \leq \sqrt{\frac{2}{\pi\sigma^2 } } \frac{1}{\mu_{\delta}(\vartheta ) } e^{-\frac{(\vartheta-\eps-\delta/2)^2}{2\sigma^2 } } , \quad \text{for all $ \eps<\eps_0$.}\ ] ] upon integrating over @xmath284 , dividing by @xmath285 , and taking the limit as @xmath276 , this yields for every @xmath163 and sufficiently large @xmath179 @xmath286 ^ 2}{f(y|x ) } \d y\d x \leq \sqrt{\frac{2}{\pi\sigma^2 } } \frac{2}{\mu_{\delta}(\vartheta ) } e^{-\frac{(\vartheta-\delta/2)^2}{2\sigma^2}}.\ ] ] condition f follows then by letting @xmath189 tend to infinity .
stimulating discussions with ram zamir are gratefully acknowledged .
j. singh , o. dabeer , and u. madhow , `` on the limits of communication with low - precision analog - to - digital conversion at the receiver , '' _ ieee trans .
_ , vol .
57 , no . 12 , pp .
36293639 , dec . 2009 .
v. v. prelov and e. c. van der meulen , `` an asymptotic expression for the information and capacity of a multidimensional channel with weak input signals , '' _ ieee trans .
inf . theory _ ,
39 , no . 5 , pp . 17281735 , sept .
1993 .
a. lapidoth and s. m. moser , `` capacity bounds via duality with applications to multiple - antenna systems on flat fading channels , '' _ ieee trans .
inf . theory _ ,
49 , no .
10 , pp . 24262467 , oct . | this paper studies the capacity of the peak - and - average - power - limited gaussian channel when its output is quantized using a dithered , infinite - level , uniform quantizer of step size @xmath0 .
it is shown that the capacity of this channel tends to that of the unquantized gaussian channel when @xmath0 tends to zero , and it tends to zero when @xmath0 tends to infinity . in the low signal - to - noise ratio ( snr ) regime , it is shown
that , when the peak - power constraint is absent , the low - snr asymptotic capacity is equal to that of the unquantized channel irrespective of @xmath0 .
furthermore , an expression for the low - snr asymptotic capacity for finite peak - to - average - power ratios is given and evaluated in the low- and high - resolution limit .
it is demonstrated that , in this case , the low - snr asymptotic capacity converges to that of the unquantized channel when @xmath0 tends to zero , and it tends to zero when @xmath0 tends to infinity . comparing these results with achievability results for ( undithered ) 1-bit quantization , it is observed that the dither reduces capacity in the low - precision limit , and it reduces the low - snr asymptotic capacity unless the peak - to - average - power ratio is unbounded .
[ multiblock footnote omitted ] |
in the process of developing experiments measuring new phenomena in physics the estimation of the sensitivity of certain configurations of experiments is of utmost importance . + a particularly active field
is the estimate of sensitivities for future neutrino oscillation experiments , see for example @xcite .
the goal of future experiments is often the measurement of a mixing angle ( denoted @xmath0 ) , the probability for a neutrino oscillation taking place being proportional to @xmath1 .
+ in sensitivity studies for neutrino oscillation experiments , `` sensitivity '' is often not defined in the same way .
furthermore uncertainties in nuisance parameters ( often sloppily called `` systematic uncertainties '' ) are often ignored ( see for example @xcite ) or included in calculations in an incomplete manner ( see discussion in @xcite ) .
little attention seems to be given to the issue of how sensitivity is defined and uncertainties are treated , despite the fact that decisions on experimental set - ups might be based on small differences in sensitivity studies .
+ in this note we try to firstly clarify the definition of `` sensitivity '' and discuss - using a toy model - potential problems which arise if instrumental uncertainties need to be considered . + the issue of sensitivity calculation has after being discussed on a recent conference on future neutrino experiments @xcite already inspired a more careful assessment of sensitivity calculation @xcite which indicates that a more formal discussion is worthwhile .
probably the most common definition of sensitivity adopted in the study of future experiments aimed at the discovery of signals of as yet undetected physics phenomena is : + _ the experiment is said to be sensitive to a given value of the parameter @xmath2 at significance level @xmath3 if the mean p - value obtained given @xmath4 is smaller than @xmath3 . _
+ here we choose ( in the spirit of neutrino oscillation experiments ) the parameter describing the new physics phenomenon to be denoted by @xmath0 .
the p - value is ( per definition ) calculated under the condition that the null hypothesis holds : @xmath5 where @xmath6 denotes the null hypothesis and @xmath7 denotes the test statistics with its observed value @xmath8 , which is distributed as the distribution function @xmath9
. we will give common definitions of @xmath7 in the next section .
+ a variation , which is the most commonly used in the context of neutrino experiments is using confidence intervals for the definition of sensitivity : + _ the experiment is said to be sensitive to a given value of the parameter @xmath2 at significance level @xmath3 if the mean @xmath10 confidence interval obtained , given @xmath4 , does not contain @xmath11 _ + both definitions can be equivalent , but are not in general , depending on the choice of test statistics and the method of confidence interval calculation .
for example , choosing to calculate upper limits would never yield detection .
+ to our knowledge , all sensitivity curves presented for neutrino oscillation experiments follow above definitions and therefore correspond to the mean observation .
this has two important implications : firstly , it should be emphasized that if the distribution of parameter estimates is gaussian , it is a well known fact ( but often ignored ) that even if @xmath12 ( i.e. the true value is at the estimated sensitivity ) , the probability for actually claiming discovery is only 50 % .
secondly , if the parameter estimates are not gaussian ( see for example @xcite ) , then the presentation of the mean result yields very little information on the actual probability for discovery .
thus , the choice to present mean experimental results is not particularly informative .
a more general definition of sensitivity therefore has to specify two probabilities : + _ the experiment is sensitive to a given value of @xmath2 if the probability of obtaining an observation n which rejects @xmath13 with at least significance @xmath3 is at least @xmath14 . _
+ where we have chosen to reformulate the definition as to be applicable to both definitions given above .
another example where an attempt is made to define sensitivity in a manner involving the detection probability can be found in @xcite
according to the neyman - pearson lemma , the uniformly most powerful test statistics that can be chosen is the likelihood ratio : @xmath15 where t denotes the test statistics and @xmath16 denotes the likelihood under the observation @xmath17 for the null hypothesis ( @xmath18 ) and the alternative hypothesis ( @xmath19 ) , respectively .
one useful property of the likelihood ratio is the fact that asymptotically : @xmath20 i.e. the distribution of t under the null hypothesis is known and the significance of the observation can be calculated from the @xmath21 distribution .
a particular common method in studies of neutrino oscillation experiments is to perform a @xmath21 fit and calculate a confidence interval from the function @xmath22 .
for example , the interval @xmath23 $ ] can be found by finding the points for which : @xmath24 where @xmath25 denotes the @xmath21 at the best fit value of @xmath0 .
this confidence interval is then often used to claim discovery by requiring @xmath26 .
+ there are two quantities which are of crucial importance in the context of calculation of confidence intervals and in the testing of hypothesis ( claiming of discovery ) .
methods to calculate confidence intervals should have _ coverage _ , defined as : + _ an algorithm is said to have the correct _ coverage _ if given a confidence level @xmath10 and a large number of repeated identical experiments , the resulting confidence intervals include the true value of the parameters to be estimated in a fraction @xmath10 of all experiments . _
+ if confidence intervals are used to claim discovery ( meaning for testing hypotheses ) , then @xmath3 is the probability for making a type i error , i.e. the probability for rejecting the null hypothesis though it is true ( often called _ significance _ ) . + the other quantity we will be interested in is the _
power_. power is the probability that the null hypothesis is rejected given that the alternative hypothesis is true .
this quantity is exactly the probability we denoted @xmath14 in the previous section .
the probability 1 - @xmath14 is the probability to make a type ii error ( accepting the null hypothesis though it is false ) .
_ nuisance parameters _ are parameters which enter the data model , but which are not of prime interest .
the probably most common example is the expected background in a poisson process .
sensitivities ( as confidence intervals ) are usually only calculated for the parameter of primary interest and it is not desired to calculate them depending on parameters which are of no physical interest and specific to the experiment .
thus , ways have to be found to marginalize the nuisance parameter .
there are two particularly common approaches : + in the first method , the probability density function ( pdf ) without uncertainty in nuisance parameters is replaced by one where there is an integration over all possible true values of the nuisance parameter _
( integration method ) _ :
@xmath27 here @xmath28 is the true value of the nuisance parameter and @xmath29 is its estimate .
since the integrated pdf is describing the probability of the true value given its estimate ( and not vice versa ) this method is bayesian .
some prior probability distribution of the true value of the nuisance parameter has to be assumed .
+ in the other common method , the pdf is replaced by one where for each @xmath30 the pdf is maximized with respect to the nuisance parameters _ ( profiling method ) _ @xmath31 with notation as above .
this method is completely frequentist , since it never treats @xmath28 as a random variable .
therefore the argument of the maximization is a likelihood function and not a pdf .
both these methods are frequently applied in high energy physics in confidence interval calculations @xcite,@xcite and references therein and to them .
+ in assessment of sensitivities of neutrino oscillation experiments , uncertainties are often included by performing a least square fit using a modified @xmath21 and use the resulting confidence interval ( see previous section ) to claim discovery .
two modifications are particularly common .
one method is to add the uncertainty in the background estimate in quadrature ( for simplicity we will restrict ourselves to background estimate uncertainties ) : @xmath32 where @xmath33 denotes the experimental result , @xmath29 the background estimate and @xmath34 the uncertainty on that estimate . under assumption of a gaussian process and
applying bayesian reasoning , @xmath35 can be viewed as equivalent to using the method illustrated in equation [ eq : integration ] .
+ the other ( probably more common ) method of inclusion is based on adding a normalization parameter to the @xmath21 and minimize the @xmath21 with respect to it , see for example @xcite @xmath36 where in addition to the parameters described above , we introduce the normalization parameter @xmath37 .
this modification is equivalent to the method represented by equation [ eq : profile ] under assumption of gaussian processes . +
a priori it can not be assumed that the modified quantities @xmath38 and @xmath35 still follow a @xmath21 distribution .
however , in general , this is the assumption employed in sensitivity calculations is obviously not modified , in our simple example : @xmath39 .
however , its distribution under the null hypothesis is not @xmath21 since @xmath29 is not constant at the true value , but a random variable . ] . in the following section we will apply above definitions to a toy model and
check the validity of the assumption using monte carlo simulations .
for simplicity , we will consider a one bin measurement , where we measure a number of events from a poisson process with background contribution and we obtain an estimate of the background from a separate measurement , which is assumed to be gaussian . in equations : @xmath40 where @xmath41 denotes a poisson process with experimental outcome @xmath17 ( number of events ) , signal parameter @xmath30 and background parameter @xmath42 and @xmath43 denotes a gaussian process with experimental outcome @xmath29 and width @xmath44 . since in the common neutrino oscillation experiment @xmath45
this toy model captures the main feature of many experiments ( though being a simplification , obviously ) .
+ using monte carlo simulations of replica of the actual experiment , we can calculate the true distribution of the test statistics defined in equation [ eq : chi11 ] and [ eq : chi21 ] under the condition that the null hypothesis is true ( @xmath46 , thus the coverage .
we can also assume @xmath47 and calculate the probability that the null distribution will be rejected given the alternative hypothesis is true , i.e. the power .
+ figures [ fig : cov ] , [ fig : cov2 ] and [ fig : pow ] exemplify the results .
figure [ fig : cov ] shows the value of the modified @xmath21s as a function of corresponding coverage .
results are shown for a true background of @xmath48 and an uncertainty in the background estimate of 20 % .
for this very simple example , it can clearly be seen that ignoring uncertainties ( in this case in the background estimate ) leads to a increased rate of false detections with respect to the one the experimenter intents .
the real false detection rate for 99 % nominal threshold for example is larger by a factor @xmath493 . the effect becomes smaller if one decides to include the additional uncertainties in one of the two ways described in equation [ eq : chi11 ] and [ eq : chi21 ] . using the latter for example the false detection rate increases by 50 % with respect to that nominally required .
+ though we are assuming a background of @xmath48 , part of the found difference could be due to the fact that we use @xmath21 statistics for a poisson process .
we therefore include the case where we assume a strictly gaussian measurement process ( see fig .
[ fig : cov ] , right panel ) .
the difference between the methods becomes less pronounced , but is still large .
+ in figure [ fig : cov2 ] we show results for smaller uncertainties in the background estimate ( 10 % ) . as intuitively expected , the impact of the method chosen to calculate the significance becomes less important .
if we consider a truely gaussian process ( right panel ) for smaller uncertainties both the method using quadratic addition and profiling give results compatible with a nominal @xmath21 distribution .
the complete measurement process consists of a measurement of background and a measurement of signal events .
a complete @xmath21 is therefore : @xmath50 and @xmath51 .
this quantity is included in the figures .
the left panel shows the results for the poisson measurement process .
the right panel assumes a gaussian process .
the uncertainties in the background estimate are assumed to be 20 % .
[ fig : cov ] in figure [ fig : pow ] we show the relative difference in power between the quadratic addition and the profile method . for large signals the power non unexpectedly approaches one , i.e. the method used to calculate the test statistics does not matter .
for low signals however one sees that the power of the profile method is up to 35 % larger than for the method of adding the uncertainty in quadrature . in our simple toy model
the set of measurements is @xmath52 and since we know the distribution of these measurements the ensemble of experiment replica is easily constructed . under more realistic experimental conditions many different both correlated and uncorrelated nuisance parameter might need to be considered .
this might become computationally very cumbersome , for example if full detector simulations need to be employed .
sometimes even uncertainties in theoretical estimates have to be considered .
it seems doubtful , though possibly the only feasible way , to treat these as random variables .
two subjects have been discussed in this note : * _ the interpretation of `` sensitivity '' _ + usually estimates of sensitivity are based on an average experimental result . for a gaussian distribution of estimates , this implies that if @xmath2 the probability for claiming discovery will be only 50 % .
this is a well known , but often ignored fact . in our experience , many physicists have the notion that if the value of the true value of the parameter is indeed equal to the sensitivity , then it should be very likely that a discovery will be made . + if the distribution of estimates is not gaussian , in general no statement about the probability of detection is made if only the average experimental result is presented .
consequently , statements about the probability for detection should be included in presentation of sensitivity estimates .
they can be calculated if the distribution of estimates is known or can be simulated .
+ * _ effect of uncertainties _
+ the results of the toy model calculation show that if uncertainties in nuisance parameters are included into the calculation of sensitivities ( and measurement results ) extra care has to be taken to make sure statistical statements ( like the significance of a discovery , or confidence level of an interval ) are still valid .
the largest mistake is not surprisingly made if the uncertainties are ignored .
the choice of method to include the systematics furthermore affects the probability of making a discovery . in addition , in presence of sizable instrumental uncertainties , the ensemble of experiments for calculating significance and power needs to be carefully defined .
+ when comparing sensitivity estimates for different experiments and experimental configurations , differences therefore certainly could arise from the way the uncertainties are included ( if at all ) in the calculations .
it seems obvious , that sensitivity curves need to be compared at the same `` real '' significance level and at the same `` real '' probability for discovery , whereas they are usually compared for the same nominal significance level and under the assumption that the probability for discovery will be always 50 % .
+ the toy model presented here is a crude simplification of the actual experimental situation where many measurement bins and different types of correlated and uncorrelated uncertainties have to be considered .
for example , a generalization of the profiling method to a more realistic experimental situation , including many bins and correlated systematic uncertainties , is given in @xcite .
the results presented in figures [ fig : cov ] , [ fig : cov2 ] and figure [ fig : pow ] should therefore rather serve as an inspiration for detailed studies of realistic experimental conditions ( see for example @xcite ) .
if possible , the statistical quantities should be studied using monte carlo simulations of many replica of the experiment under study .
application of monte carlo simulations does not only yield information on the correct false detection rates but also on the probability of detection , which is of obvious importance for assessment of sensitivities of future experiments .
this research was supported by the swedish research council under grant 621 - 2004 - 2196 .
we thank an anonymous referee for useful suggestions . 00
international scoping study , http://www.hep.ph.ic.ac.uk/iss/ p. sinervo , _ in the proceedings of phystat2003 : statistical problems in particle physics , astrophysics , and cosmology , menlo park , california , 8 - 11 sep 2003 , pp tuat004_. m. komatsu , p. migliozzi and f. terranova , j. phys .
g * 29 * , 443 ( 2003 ) [ arxiv : hep - ph/0210043 ] .
d. indumathi , m. v. n. murthy , g. rajasekaran and n. sinha , phys .
d * 74 * , 053004 ( 2006 ) [ arxiv : hep - ph/0603264 ] .
j. conrad , invited contribution to 8th international workshop on neutrino factories and superbeams ( nufact 06 ) , irvine , usa , august 2006 , http://nufact06.physics.uci.edu/workshop/program/plenary.aspx t. schwetz , phys .
b * 648 * ( 2007 ) 54 [ arxiv : hep - ph/0612223 ] .
g. punzi , _ in the proceedings of phystat2003 : statistical problems in particle physics , astrophysics , and cosmology , menlo park , california , 8 - 11 sep 2003 , pp modt002 _ [ arxiv : physics/0308063 ] .
j. conrad , _ et .
d * 67 * ( 2003 ) 012002 [ arxiv : hep - ex/0202013 ] .
w. a. rolke , a. m. lopez and j. conrad , nucl .
instrum .
a * 551 * ( 2005 ) 493 [ arxiv : physics/0403059 ] .
j. burguet - castell , d. casper , e. couce , j. j. gomez - cadenas and p. hernandez , nucl .
b * 725 * ( 2005 ) 306 [ arxiv : hep - ph/0503021 ] .
p. huber , m. lindner and w. winter , nucl .
b * 645 * ( 2002 ) 3 [ arxiv : hep - ph/0204352 ] .
v. barger , m. dierckxsens , m. diwan , p. huber , c. lewis , d. marfatia and b. viren , phys .
d * 74 * , 073004 ( 2006 ) [ arxiv : hep - ph/0607177 ]
. g. l. fogli , e. lisi , a. marrone , d. montanino and a. palazzo , phys .
d * 66 * ( 2002 ) 053010 [ arxiv : hep - ph/0206162 ] . | calculations of sensitivities of future experiments are a necessary ingredient in experimental high energy physics . especially in the context of measurements of the neutrino oscillation parameters extensive studies
are performed to arrive at the optimal configuration . in this note
we clarify the definition of sensitivity as often applied in these studies .
in addition we examine two of the most common methods to calculate sensitivity from a statistical perspective using a toy model .
the importance of inclusion of uncertainties in nuisance parameters for the interpretation of sensitivity calculations is pointed out .
sensitivity , statistical methods , neutrino oscillation experiments |
dwarf irregular galaxies ( dirrs ) are low mass , but gas rich galaxies and are found rather isolated and spread throughout the local group . all dirrs display low elemental abundances indicating that only little chemical evolution has taken place over the past 15gyr despite ongoing star formation .
the lack of strong starburst cycles in these isolated objects may be because of little or no merger interaction .
therefore , in the context the cold dark matter scenarios of hierarchical galaxy formation by merger of smaller structures @xcite , dwarf galaxies and in particular the isolated dwarf irregular galaxies could be the purest remnants of the proto - galactic fragments from the early universe .
hence , the dirrs are also one of the possible sources for the damped ly@xmath2 absorption ( dla ) systems as observed in quasar spectra over a large range of redshifts ; see e.g. @xcite for a recent compilation of dla metallicities over @xmath7 .
the dirrs being possible remnants of the early universe might allow us to study early galaxy evolution in great detail in the nearby universe , even in the local group .
in particular the early chemical evolution of galaxies is of interest here to shed light on the formation of the first generations of stars . the most powerful way to study
the chemical evolution of a galaxy is via the elemental abundances and abundance ratios of their stellar and gas content which contain a record of the star formation histories ( sfh ) of the galaxy over the last 15gyr .
the analysis of bright nebular emission lines of regions has been the most frequent approach to modeling chemical evolution of more distant galaxies to date @xcite .
so far , only a very limited number of elements can be examined and quantified when using this approach .
the chemical evolution of a galaxy depends on the contributions of all its constituents , e.g. , sne type ia and ii , high mass stars , thermal pulsing in low and intermediate mass agb stars .
thus , more elements than just those observed in nebulae need to be measured , since each have different formation sites which sample different constituents .
the @xmath2 elements like oxygen are created primarily in short - lived massive stars , while the iron - group elements are produced in sne of both high and low mass stars .
therefore , it is expected that substantially different sfhs of individual galaxies shall be represented in different present - day @xmath5 ratios as discussed e.g. in @xcite . in reverse , the accurate measurement of the absolute stellar abundances and abundance ratios allows to put constraints on the sfh of the respective galaxy .
in addition to the study of the average metallicities of a galaxy as discussed above , also the spatial distribution of abundances and abundance patterns within a galaxy contains important information about the redistribution processes of the newly produced elements and their subsequent mixing with the ism .
if the abundances of a number of stars distributed over the galaxy can be analyzed , possible chemical inhomogeneities e.g. in form of large spreads in the abundances or abundance gradients can be studied and provide additional constraints of the chemo - dynamical evolution of the galaxy .
unfortunately , most dirr systems are too distant for the detailed study and abundance analyses of e.g. their red giant branch ( rgb ) stars , which are of great importance due to their large spread in age . hence
, to date few details on the chemical evolution of the dirr galaxies have been available . in the case of the dirr sextansa studied in this work the tip of the rgb
is found at visual magnitudes of @xmath8 @xcite far beyond the capabilities of even the most efficient high - resolution spectrographs on today s 8 to 10-meter class telescopes .
however , with the same class of latest telescopes and instrumentation , the visually brightest stars of the dirr galaxies , i.e. , the blue supergiants ( @xmath9 in sextansa ) become now accessible to detailed spectroscopy , which allow us to determine the present - day abundances of @xmath2 _ and _ iron group elements .
the abundances of the @xmath2 elements derived from the rich spectra of hot massive stars can be directly compared and tied to the nebular @xmath2 element abundances because both nebulae and massive stars have comparable ages and the same formation sites .
the first abundance studies based on high - resolution spectra of a - type supergiant stars were carried out for the nearby local group dirrs ngc6822 and wlm @xcite .
@xmath2 element and iron group abundances could be derived for two stars in both dirr galaxies . for ngc6822 and wlm the [ @xmath5 ]
ratios are in good agreement with the _ solar _ ratio .
this result is somewhat surprising since both galaxies have ongoing star formation and different star formation histories as determined from their stellar populations ( e.g. , @xcite , @xcite , @xcite ) . the isolated dwarf irregular galaxy sextansa (= ddo75 ) with its distance of about @xmath11mpc @xcite
is located at a similar distance as sextansb , ngc3109 , and the antlia dwarf galaxies which form a small group of galaxies at the edge of the local group possibly not even bound to the local group which would make sextansa a member of the nearest ( sub)cluster of galaxies @xcite . the present - day chemical composition was studied from the emission - line spectroscopy of bright compact regions @xcite .
the analysis by @xcite determines an oxygen abundance of @xmath12 which corresponds to @xmath13 = -1.17 $ ] if a solar oxygen abundance of @xmath14 is adopted @xcite . in a revised analyses of the same data by @xcite ,
a higher value of @xmath15 is found , corresponding to @xmath13 = -0.95 $ ] .
the star formation history of sextansa was first studied in detail by @xcite , and more recently by @xcite and @xcite using hst vi color
magnitude diagrams ( cmd ) . a high rate of star formation has occured in the past @xmath16gyr , with very little star formation between @xmath17 and @xmath18gyrs .
a mean metallicity of @xmath19\approx-1.45\pm0.2 $ ] is derived over the measured history of the galaxy , with most of the enrichment having occurred more than 10gyr ago . only slight enrichments ( + 0.4dex ) occured between 2 and 10 gyr from fitting the red giant branch to a metallicity of @xmath19\approx-1.1 $ ] . in this paper
, we present the results from a detailed abundance analysis of three isolated and normal a supergiant stars in sextansa based on high resolution spectra obtained at the eso - vlt using uves .
we compare the stellar results to those from nebular analyses to examine the chemical homogeneity of this young galaxy .
we also use the rich stellar spectra to compare the abundances of several other elements not available in the nebular analyses to those predicted from its chemical evolution history , as well as to results from other dwarf galaxies .
a list of potential targets for high - resolution uves @xcite spectroscopy at the eso vlt was selected from the photometric catalogue by @xcite .
the visually brightest ( @xmath20 ) stars with colors ranging from @xmath21 were chosen from the corresponding cmd .
the selected targets were checked for possible crowding on hst / wfpc2 images from @xcite . for one of the targets , the star number 754 in the catalogue of @xcite ( in the following identified as `` sexa-754 '' ) , a low resolution keck / lris @xcite spectrum was obtained by @xcite , which confirmed that this star is a member of sextansa ( cf .
below ) and a normal a supergiant suited for abundance analyses .
almost 14 hours of vlt / uves spectroscopy of sexa-754 were carried out in service mode in between january and february 2002 within the specified constraints of a seeing better than @xmath22 , a dark and clear sky , and an airmass @xmath23 .
additional spectra in complementary wavelength settings in the red were obtained in a visitor mode run in april 2002 by ak , kav , cp . in parallel keck / esi
@xcite low resolution spectroscopy of other potential targets was obtained by kav in february 2002 . for the best candidates ,
the esi low - resolution spectra were complemented by high - resolution spectra taken with uves in the april 2002 visitor run .
all candidates stars were first checked for their membership to the sextansa galaxy by comparing the measured radial velocities of the stars to the heliocentric radial velocity of sextansa as determined from optical and radio spectroscopy : @xcite determined values of @xmath24km / s from the h@xmath2 emission lines of different regions , which are consistent with earlier measurements of the radial velocities of the gas , i.e. , @xmath25 @xcite and @xmath26km / s @xcite . due to the large heliocentric radial velocity of sextansa , any confusion of our stellar targets with foreground stars
is very unlikely . based on the collected esi and uves
spectra two additional stars ( sexa-1344 and sexa-1911 ) were identified for 2 times 12.5 hours of vlt / uves service mode observations , which were executed between december 2002 and february 2003 within the same constraints as given for sexa-754 before .
the candidate stars sexa-513 , sexa-983 , and sexa-1456 on the other hand were found to be not suited for our abundances analyses which are tailored for a supergiants of a narrow temperature and luminosity range and the three stars had to be discarded . a complete log of the esi and uves observations is found in tab .
[ sexa - observations ] together with some additional information on the instrumental configurations used for the individual observations and the environmental conditions during the actual observations .
the complete sample of examined sextansa targets is presented in tab .
[ sexa - sample ] together with the radial velocity and spectral type information as obtained either during the detailed analyses of the stars or for the discarded stars during the pre - selection process ; the location of the individual targets within the galaxy field is shown in fig .
[ sexa - field ] together with the positions of the four bright compact regions analyzed by @xcite .
all uves data were reduced using the dedicated uves pipeline which is described by @xcite and is available as a special package within the eso - midas data reduction system .
optimum extraction with cosmic rejection and sky subtraction was used to extract the individual stellar spectra . a typical signal - to - noise ratio ( snr ) of @xmath27 in the wavelength regime corresponding to the @xmath28 band
was obtained for a single spectrum of 4500sec exposure time .
the typically 10 individual spectra per star were then corrected for the spectrograph s response function to obtain a smooth continuum , rebinned to heliocentric velocities , and eventually combined with additional cosmic clipping .
the combined spectra were resampled and rebinned to a 2-pix resolving power of @xmath29 which was used throughout the analyses .
the reduced resolving power allows to increase the snr per resolution element without having to sacrifice line profile information .
the projected rotational velocities of the three stars were later determined to be @xmath30km / s , i.e. , all line profiles remain fully resolved at the reduced resolution of @xmath31km / s .
the snrs obtained for the combined uves spectra are listed in tab .
[ sexa - snr ] for different wavelength regions of interest .
the observations of individual stars in a distant local group galaxy as presented here are at the limit of the capabilities of today s largest telescopes and most sensitive spectrographs : despite total integration times of typically 12.5hrs per star , snrs of 30 are achieved at best for a resolving power of @xmath29 for a @xmath32mag star which is deemed just sufficient for detailed abundances analyses as will be shown in this work . in fig .
[ sexa - halpha ] the h@xmath2 profiles of the three stars are shown . to obtain the stellar line profiles ,
the contribution of the diffuse h@xmath2 emission of sextansa which was found for all stars was subtracted as part of the sky subtraction in the extraction process of the spectra .
inspecting the raw 2d spectra , the diffuse h@xmath2 emission does show spatial structure but averaged along the slit its distribution is mostly gaussian .
the diffuse emission is blueshifted by an average of @xmath33km / s with respect to the stellar spectra and consistent with the radial velocity of sextansa @xcite .
the average fwhm of the emission is @xmath34km / s corresponding to a ( @xmath35 ) velocity dispersion of @xmath36km / s .
this is in excellent agreement with a thermal velocity ( @xmath37km / s ) of a warm interstellar medium if a temperature of @xmath38k is assumed . in the sky - subtracted stellar h@xmath2 profiles in fig .
[ sexa - halpha ] no stellar line - emission is visible in h@xmath2 which indicates that the selected stars are supergiants of lower luminosity .
this was already expected from the fact that during the selection process only stars about 2 magnitudes fainter than the brightest blue stars in sextansa ( sexa-513 ) turned out to be normal a supergiants .
interestingly , no brighter , more extreme a supergiants , which are supposed to be the visually brightest stars , were found in the whole galaxy .
the same observation was made during the study of the dirr wlm @xcite .
the selection process applied to find candidate stars for abundance analyses in sextansa and wlm was not rigorous enough to exclude that this finding is a selection effect intrinsic to our procedure .
however , the lack of luminous a supergiants might as well be an evolution effect due to the low metallicity of the galaxy . similarly ,
the lack of wind emission features in h@xmath2 as seen in fig .
[ sexa - halpha ] might be due to the low metallicity suppressing the formation of stellar winds due to the reduced radiation line - driving on the metal - poor gas .
lower - luminosity supergiants as were found in sextansa are best suited for abundance analyes .
higher - luminosity stars are difficult to analyze because the line - forming regions of their atmospheres are no longer in hydrostatic equilibrium but are found in the acceleration zone of the stellar wind .
further , for lower - luminosity a supergiants , the effects due to intrinsic photospheric and wind variability on the line profiles is negligible @xcite .
equivalent width variations of weak ( @xmath39m ) photospheric lines as used for the abundance analyses in this work are found to be @xmath40% @xcite .
the atmosphere analyses of the three stars in sextansa have been carried out in line with previous analyses of a supergiant stars in the galaxy @xcite , the magellanic clouds @xcite , m31 @xcite and the dwarf irregular galaxies ngc6822 @xcite and wlm @xcite . maintaining a consistent analysis process across the different studies of a supergiant stars
is considered crucial to obtain consistent and comparable abundance results . in particular results from differential analyses
can then be considered the most reliable because the uncertainties in the determined absolute elemental abundances shall cancel out at least to some extend .
the atmosphere computations are based on plane parallel , hydrostatic , and line - blanketed atlas9 lte model atmospheres @xcite which are considered to be appropriate for the photosphere analysis of lower luminosity a supergiants @xcite . to account for the possible effects of the low metallicity of sextansa on the model atmosphere structure ,
all atmosphere models have been computed with a metallicity scaled to @xmath41 of solar and a corresponding opacity distribution function .
the value of @xmath42=-1 $ ] was chosen in agreement with the present day oxygen abundances as derived from the analysis of regions @xcite .
however , as has been shown later , the effect of atmosphere models with different metallicities on the derived elemental abundances is typically less than @xmath43dex , i.e. , negligible within the estimated systematic errors .
the analyses has been restricted to the use of weak absorption lines only to minimize the effects of the neglected spherical extension and of the neglected nlte conditions in these extended , low gravity supergiant stars on the model atmospheres .
further , weak lines are preferred because they suffer less from nlte and microturbulence effects in the line formation process .
weak lines in this context are defined as lines where a change in the microturbulence @xmath44 of @xmath45km / s results in a change in the elemental abundance of @xmath46dex . to fulfill this weak - line condition
, the equivalent width of the corresponding line has typically to be @xmath47m .
weak - line analyses are a challenge for faint and metal - poor targets as analyzed in this work , in particular if the galaxies are far away , the targets are faint and the snr of the high - resolution spectra is comparatively low as it is the case here .
the absorption lines have been measured by fitting gaussian profiles to the weak line profiles and to the stellar continuum , the latter being the main source of uncertainty in the derived equivalent widths @xmath48 .
the comparatively low snr leaves a non - negligible uncertainty of about @xmath18% in the definition of the continuum level and therefore in the measured @xmath48 in the blue part of the combined spectra .
the stellar model atmosphere parameters effective temperature @xmath49 , gravity @xmath50 , and microturbulence @xmath44 are in the following determined solely from spectral features and therefore are not affected by the ( badly - defined ) extinction .
the temperature and gravity parameters of the best fits of synthetic balmer - line profiles to the gravity - sensitive wings of the h@xmath51 profiles ( see fig . [ sexa - hgamma ] ) define a relation in the @xmath52
@xmath50 plane as shown for each star in fig .
[ sexa - tefflogg ] .
the temperature and gravity parameters for which an ionization equilibrium of and is found defines a temperature - sensitive relation in the @xmath52 @xmath50 plane ( for a detailed discussion of the nlte corrections see below ) .
the intersection of both relations defines the position in the @xmath52
@xmath50 plane for which both , the best fit to the h@xmath51 profile and the ionization equilibrium of and are achieved .
this parameter pair is adopted as the atmosphere parameters of the star . due to the low temperature of the analyzed stars and the comparatively low snr of the combined spectra ( cf .
[ sexa - snr ] ) none of the weak spectral lines of present in the observed spectral ranges could be used to determine the stellar parameters from the most reliable ionization equilibrium of and . also the strong , easy to measure @xmath53 line had to be discarded because of its known high sensitivity to nlte and microturbulence effects : nlte corrections are of the order of @xmath54dex for the temperature and gravities of the stars analyzed here ; an uncertainty of the microturbulence of @xmath55km / s results in a change of the computed abundances of the same order . instead ,
as already described above , the ionization equilibrium of and was used in this work .
this introduces an additional uncertainty to the determination of the stellar parameters because of the sensitivity of lines to nlte effects @xcite .
the nlte effects on lines are considered negligible @xcite . to quantitatively incorporate the dependencies of the nlte corrections on temperature , gravity , and metallicity on the / ionization equilibrium , the results from @xcite presented in her figs .
4 and 5 were used to derive a relation for the nlte correction to be applied to the abundances for the limited temperature range of interest in this work , i.e. , @xmath56k : @xmath57}{1000}-0.58 \nonumber\\ & & -2(\log g - 4 ) \left ( 0.04\frac{t_{\rm eff}[k]}{1000}-0.27\right ) \nonumber\\ & & -0.1 [
{ \rm m}/{\rm h } ] \nonumber\\ & & -{\it 0.28 } \nonumber\\ \end{aligned}\ ] ] the first and second terms describe the temperature and gravity dependent nlte correction as derived for solar metallicity main sequence ( @xmath58 ) stars .
the third term describes the metallicity dependence of the nlte correction as derived for metallicities in the range @xmath59=\pm0.5 $ ] . due to the considerably lower gravities and metallicities of the stars studied here
, this relation has to be used with caution . to test the relation for stars of lower gravity and metallicity as studied in this work
the relation was applied to the data set of a supergiants in the smc as analyzed by @xcite .
the stellar parameters of the smc stars are all based on the most reliable to ionization equilibrium .
the nlte corrections were determined as @xmath60 assuming that the nlte corrections for are negligible @xcite .
it is found that the above relation fits the smc data well if an additional correction of @xmath61 is added , cf .
[ sexa - feinlte ] .
it is important to note that the relation describes rather well all smc stars with to abundance differences over the whole range from @xmath62 to @xmath63dex .
this constant additional ( fourth ) term in the relation seems sufficient to absorb the additional nlte effects due to low gravity and low metallicity , at least for the narrow temperature range of @xmath64k , gravities of @xmath65 , and metallicities of @xmath59\approx-0.7 $ ] , i.e. , close to the stellar parameters of the stars studied in sextansa . in the subsequent atmosphere analysis ,
the final stellar parameters for the sextansa stars were chosen to be consistent with the nlte corrections as given by the above relation .
the adopted final corrections are @xmath66dex for sexa-754 , @xmath67dex for sexa-1344 , and @xmath68dex for sexa-1911 and are shown in fig .
[ sexa - feinlte ] , too .
it should be further noted that the remaining uncertainty of @xmath69dex in the nlte corrections has been taken into account in the overall error budget as estimated from the uncertainties of the determination of the stellar parameters ( cf .
below ) .
microturbulence @xmath44 has been determined from the line abundances of the numerous lines of different strength from the ions , , and .
a single value of @xmath44 for each star was adopted so that no trend of the computed elemental abundance is found with the equivalent width of the lines within the uncertainty of @xmath45km / s .
the estimated uncertainties in @xmath52 are dominated by the uncertainties of the derived nlte corrections but also have to account for the uncertainties of the assumption of negligible nlte effects on in the ionization equilibrium used for the temperature determination .
the effect of a change of the nlte correction of @xmath70dex ( as derived from the residuals in fig .
[ sexa - feinlte ] ) is shown in fig .
[ sexa - tefflogg ] for star sexa-754 .
if we assume an uncertainty of @xmath71dex for the lte assumption for , we obtain an estimate for the total uncertainty of @xmath69dex which translates into @xmath72k as indicated by the horizontal errorbars in fig .
[ sexa - tefflogg ] .
uncertainties in gravity are estimated to @xmath73dex from varying the fits to h@xmath51 for different @xmath74 at constant @xmath52 .
the resulting change of the synthetic profile for a variation of @xmath75dex is shown in fig .
[ sexa - hgamma ] . with the determined stellar parameters and elemental abundances ( cf .
below ) , a small section of the spectrum between @xmath76 was synthesized to determine the projected rotational velocities @xmath77 of the three stars .
this spectral region contain several stronger and isolated and lines . for a given resolving power of @xmath29 or @xmath31km / s and
the given heliocentric radial velocities of the stars the respective values of @xmath77 were obtained by least - square fitting of the synthetic line profiles to the observed spectra while keeping the elemental abundances for the metal lines fixed to the value determined before .
the resulting fits are shown in fig .
[ sexa - vsini ] and the corresponding values for @xmath78 are reported in tab .
[ sexa - analysis ] .
the spectral types of the three stars were defined according to their effective temperatures according to @xcite . it should be noted that this spectral classification does not take into account any metallicity dependence of the classification criteria but is based on stars of solar metallicities .
the luminosities of the three stars were then derived from the visual magnitudes as given in tab .
[ sexa - sample ] using the distance modulus of @xmath79 from @xcite and the ( close to zero ) bolometric corrections from @xcite corresponding to the spectral types .
the resulting spectral types and luminosities are given in tab .
[ sexa - analysis ] together with the stellar radii as computed from the luminosity and the effective temperature .
the luminosities of all three stars are consistent with the evolutionary track of a @xmath80 star @xcite , i.e. , a zams mass compatible with evolved low - luminosity a supergiants .
this finding further confirms the sextansa membership of the three analyzed supergiant stars .
the list of appropriate spectral lines for the abundance determinations and the corresponding atomic data were adopted from the previous analyses of @xcite .
atomic data and the abundances derived from the individual lines are listed in tab .
[ lines ] together with the measured equivalent widths @xmath48 .
all abundance calculations in lte / nlte and lte spectrum synthesis have been carried out using a modified version of the linfor code .
the resulting average elemental abundances are given for each ionization stage in tab .
[ sexa - analysis ] . for each abundance , two error estimates are given , first the r.m.s .
line - to - line scatter , second ( in italic ) the estimated systematic error in the elemental abundance from the uncertainties in the atmospheric parameters @xmath52 , @xmath50 , and @xmath44 .
the systematic errors of the average abundances have been estimated by the variation of the individual stellar atmosphere parameters within their estimated uncertainties while keeping the other parameters fixed .
the individual error estimates are reported in tab .
[ sexa - abuerror ] . from the error
estimates it can be seen that , , , and provide the most reliable elemental abundances .
however , abundances derived from the ion have been found to suffer from a large spread if derived from stars over a wider range in stellar parameters . the sample of 10 a supergiants in the smc as analyzed by @xcite shows a spread in the silicon abundances of @xmath81dex probably due to neglected nlte effects in the line formation of the ion .
the three stars analyzed here show very consistent silicon abundances because the stellar parameters of the stars are very close to each other .
however , since the absolute abundance probably is affected from the uncertainties that were found in other analyses , the abundances from will not be considered in the further discussion of @xmath2 element abundances .
the average elemental abundances relative to the solar abundances are shown for the three stars in fig .
[ sexa - eps2solar ] with the error bars representing the estimated combined systematic errors in the abundances .
the combined errors were computed as the quadratic sum of the individual errors from tab .
[ sexa - abuerror ] .
the abundances reported for and have been computed using nlte lineformation based on the model atom by @xcite which delivers results in good agreement with the model atom developed by @xcite .
the nlte corrections for the individual mg lines are listed in tab .
[ sexa - mg_corr ] and are found to be small and of the order of @xmath69dex .
abundances from the strong @xmath53 line blend were derived from fits to the observed line profile using spectrum synthesis in lte ; the results are reported in tab .
[ lines ] .
the nlte corrections for this line are expected to be large @xcite and very sensitive to the stellar parameters and in particular to the microturbulence @xcite .
therefore , no attempt was made to compute the nlte corrections and the line was not used for this analysis but is given for reference only . for star sexa-754 , useful upper limits for the equivalent widths of the weak lines could be given because of the higher snr of the combined spectrum of this star .
the resulting upper limits for the abundances are consistent with the abundances derived from the lines and therefore further support the stellar parameters as derived from the to ionization equilibrium . the @xmath2 element of highest interest is oxygen .
unfortunately , none of the lines in the 615 nm and 645 nm spectral regions could be measured in the combined uves spectra of moderate snr because of the weakness of the lines at the metallicity of sextansa . for @xmath82\approx-1 $ ]
, all lines in this spectral region are expected to have equivalent widths of @xmath83m , i.e. , clearly below the detection limit of the spectra .
however , the low metallicity of sextansa brings the near - infrared @xmath84-@xmath85 lines of multiplet 1 and 4 into the weak - line regime possibly suited for abundances analyses despite their known high sensitivity to nlte effects @xcite .
spectra of the respective wavelength regions are only available for one of the stars , sexa-754 . since the near - infrared lines are non - resolved multiplets , fitting of synthetic lte line profiles was used to determine lte oxygen abundances ( see fig . [ sexa-754_oi ] ) .
the oxygen abundances measured in lte are @xmath86 for @xmath87-@xmath88 and @xmath89 for @xmath90 .
detailed computations of the nlte corrections for the given stellar parameters of sexa-754 at the metallicity of sextansa will be required before accurate oxygen abundances can be derived for this star . in any case , the expected uncertainties will remain large , first because of the large required deviations from lte to bring the oxygen abundances into the reasonable range of @xmath91 ( assuming an underabundance of @xmath92=-1.0 $ ] as suggested from the other @xmath2 elements silicon and magnesium ) , second because of the low quality of the near - infrared spectra , which results in a non - negligible uncertainty in the continuum definition , which is crucial for an accurate line profile modeling .
the estimated error in the derived lte abundances of @xmath93dex is due to this uncertainty in the continuum localization .
therefore , currently no further attempt has been made to determine the nlte corrections for the near - infrared lines of sexa-754 . in the following , elemental abundances derived from
will be used instead for the discussion of @xmath2 element abundances in sextansa .
earlier studies in the smc @xcite and ngc6822 @xcite have shown that a good agreement between the @xmath2 elements oxygen and magnesium is found for different stars and so both elements may be used to represent the @xmath2 elemental abundance of the star .
the average @xmath2 element abundance of the three stars in sextansa are found to be @xmath94\right>=-1.09\pm0.02\pm{\it 0.19}$ ] which is in excellent agreement with the @xmath2 element abundances from the regions of @xmath13 = -1.17 $ ] as derived by @xcite and the recently revised value @xmath13 = -0.95 $ ] of @xcite . the iron - group elemental abundances of the three stars are in excellent agreement as derived from lines of , , , and
. the good agreement of the abundances from the two different ionization stages of chromium is a further indication for correct stellar parameter .
however , it should be noted that this equilibrium is based on a single line .
possible nlte effects on this line which were not taken into account in the abundance computation appear therefore to be small .
the agreement of the and abundances and hence the ionization equilibrium is achieved if the respective nlte corrections are applied , i.e. , @xmath66dex for sexa-754 , @xmath67dex for sexa-1344 , and @xmath68dex for sexa-1911 . as discussed before ,
the determination of the stellar parameter is based specifically on this nlte corrected ionization equilibrium .
the abundances of and have been derived from numerous lines and are regarded as the most reliable : the dependency of the abundances on the stellar parameters is small ( cf . tab .
[ sexa - abuerror ] ) and possible nlte effects on these dominant ionization stages are expected to be small .
the mean metallicity of the three stars are so found to be @xmath0\right>=-0.99\pm0.04\pm{\it 0.06}$ ] which to our knowledge makes these three stars the to date most metal - poor massive stars in any galaxy for which a detailed abundance analysis has been carried out .
titanium and scandium can be considered as either @xmath2 or iron - group elements .
the ti abundances were derived from a large number of lines of the dominant ionization stage of singly ionized titanium and are in good agreement with iron , i.e. , @xmath95\right>=0.08\pm0.05\pm{\it 0.08}$ ] while sc shows a slight underabundance of @xmath96\right>=-0.20\pm0.03\pm{\it 0.09}$ ] .
this might be due to the smaller number of available lines and higher temperature sensitivity of .
furthermore , the hyperfine - structure terms in the lines of odd elements of the iron group due the presence of a nuclear magnetic moment have been neglected . due to these uncertainties ,
no further discussion of the scandium abundances will follow .
for the first time , the present - day iron group abundances have been determined from stars in sextansa .
the mean underabundance relative to solar is @xmath97\right>=-1.03\pm0.05\pm{\it 0.11}$ ] and has been derived from the abundances which are expected to be the most reliable .
this result is well supported by abundances measured for a second iron - group element , namely with a mean underabundance of @xmath98\right>=-0.94\pm0.06\pm{\it 0.07}$ ] .
the individual iron - group abundances of the three stars from which the mean abundances were computed are in good agreement within the estimated systematic errors .
the star formation history of sextansa derived by ( * ? ? ? * and references therein ) , suggests that the metallicity reached @xmath59=-1.1 \pm0.2 $ ] only 2 gyr ago .
the metallicity spread of @xmath99dex is to account for the dispersion in the blue helium burning ( bheb ) stars .
the mean stellar metallicity found here from iron and chromium , @xmath0\right>=-0.99\pm0.04\pm{\it 0.06}$ ] , is consistent with the cmd stellar populations analysis .
the mean @xmath2 element abundance from magnesium in the three stars is @xmath3\right>=-1.09\pm0.02\pm{\it 0.19}$ ] which is in good agreement with the nebular results for oxygen , @xmath13 = -1.17 $ ] @xcite and @xmath13 = -0.95 $ ] @xcite .
it is worth noting that this independent measurement of the @xmath2 elemental abundance confirms the location of sextansa in the fundamental metallicity
luminosity relationship for dwarf galaxies @xcite .
the agreement between the @xmath2 element abundances from a supergiant stars and the nebular abundances is as expected .
these stars have ages @xmath100 myr and are expected to have formed out of the same interstellar gas as seen in the nebulae .
similar findings are reported from the abundance analyses of b - type main sequence stars in orion and the lmc @xcite , respectively , the galactic abundance gradients from b stars @xcite , and a supergiant analyses in the smc , m31 , and ngc6822 @xcite , respectively .
it should be noted here , that the pristine nebular and stellar abundances of some elements can be altered by effects of dust depletion and rotational mixing which complicates a quantitative comparison of nebular and stellar abundances @xcite .
however , in the case of the stars and nebulae studied in sextansa , the good agreement within the estimated errors suggest that the contributions due to these effects to the determined abundances are small .
only the dirr galaxy wlm shows a discrepancy between the stellar and nebular @xmath2 element abundances @xcite .
the nebulae in wlm have @xmath101=-0.89 $ ] @xcite , whereas two a supergiants have @xmath102=-0.62 $ ] , and one star suggests @xmath101=-0.21 $ ] .
these are significant differences that are currently not understood and may suggest that wlm is the first galaxy to display inhomogeneous mixing .
the spatial distribution of abundances in a galaxy keeps an important record of the re - distribution history of freshly produced elements and their mixing with the ism . as can be seen from fig .
[ sexa - field ] , the three stars analyzed in this work and the four bright compact regions examined by @xcite lie on a line across the central , visually brightest part of the galaxy . adopting d@xmath103mpc @xcite ,
no significant @xmath2 element abundance variations are found over a distance of @xmath4kpc as covered by the stars ( between sexa-754 to sexa-1344 to sexa-1911 ) , and over @xmath104kpc if the regions are included .
the same level of abundance homogeneity is found for the iron - group elements in the stars .
the high level of homogeneity of nebular abundances in dwarf irregular galaxies has been discussed as possible evidence against in situ ( `` on the spot '' ) enrichment by @xcite and therefore against the general validity of the instantaneous recycling approximation as typically used in chemical evolution models . @xcite
favour a scenario in which the newly synthesized elements remain in either a difficult to observe hot @xmath105k or an equally difficult to observe cold , dusty phase while mixing throughout the galaxy .
recent chandra spectroscopy results for ngc1569 by @xcite favor the hot phase as the reservoir for newly synthesized and expelled gas .
the @xmath5 abundance ratio is a key constraint for the chemical evolution of a galaxy .
@xmath2 elements are primarily synthesized in sneii while the iron - group elements are mainly produced in sneia but also in sneii .
a star formation burst in a galaxy will temporarily increase the @xmath5 ratio of the ism due to the massive stars that quickly enrich the ism with @xmath2 elements , while the lower mass stars will then slowly increase the content of iron - group elements resulting in a slow decline of the @xmath5 ratio of the galaxy @xcite .
@xcite demonstrated the expected differences in the @xmath5 ratios for galaxies with different star formation histories , and in particular noted that the solar @xmath5 ratio must not be regarded as an universal ratio . from the three stars in sextansa
, the @xmath5 ratio is measured as @xmath106\right>=-0.11\pm0.02\pm{\it 0.10}$ ] .
this @xmath2/fe ratio is much lower than in galactic stars of similar metallicity @xcite . at the metallicity of these sextansa stars , [ fe / h ] @xmath107 ,
the galactic thick disk and galactic halo stars overlap with a [ @xmath2/fe ] @xmath108 + 0.3 .
the lower @xmath2/fe ratios in these sextansa stars further follows the trend found from the abundance analyses of stars in two other dirrs , ngc6822 and wlm @xcite , but extends the metallicity range now sampled from @xmath109=-0.5 $ ] in those galaxies to @xmath109=-1.0 $ ] . as discussed by @xcite
, this lower metallicity now overlaps the upper metallicities sampled by red giant stars in dwarf spheroidal galaxies , which also show much lower @xmath2/fe ratios than galactic metal - poor stars @xcite . in the dwarf
spheroidal galaxies , the low [ @xmath2/fe ] ratios near [ fe / h ] = @xmath110 are most likely due to the lower star formation rates , thus slower chemical evolution .
the metallicity range of the dwarf irregular galaxies also overlaps with the damped ly@xmath2 absorption systems @xcite , which are also recognized to have low @xmath2/fe ratios .
this might suggest that low @xmath2/fe ratios are a generic effect at low metallicities .
the star formation histories for sextansa , wlm , and ngc6822 are globally similar as derived from their cmds .
a comparison can be seen in fig .
8 of @xcite .
accordingly , all three dirrs had significant star formation at ancient times , @xmath111gyr ago , with a hiatus at intermediate ages , @xmath112gyr ago , and recent star formation events in the past 1gyr .
@xcite has recently emphasized that the sfh of a galaxy affects the _ absolute _ elemental abundances , but is only of minor importance for the abundance
_ ratios _ , the latter being primarily determined by the stellar lifetimes , initial mass functions ( imf ) and the stellar nucleosynthesis .
thus , the impact of different sfhs on the @xmath5 ratio is related to the star formation efficiency , and manifests itself as a short plateau in the @xmath113 $ ] ratio at low metallicities in galaxies with low star formation rates , like irregulars ( either in bursts or continuous ) , while the plateau is maintained to higher metallicities for systems with high star formation rates like bulges and ellipticals .
thus , our finding that the [ @xmath5 ] ratios in three dirrs with similar sfhs is significantly lower than in the metal - poor galactic stars can be explained in this scenario .
the three dirrs with a metallicity range from @xmath114<-0.4 $ ] have already left the @xmath113 $ ] plateau and presently display the same lower @xmath113 $ ] ratio .
that this ratio is similar to solar suggests that the integrated stellar yields from a star formation event is consistent with the solar abundance ratios , probably due to the _ universal _ nature of stellar lifetimes , imfs , and stellar nucleosynthesis . that the high - redshift dla absortion systems also show low [ @xmath2/fe ]
ratios suggests that they are also systems with low star formation rates , but this could be either irregular galaxies or the outer parts of spirals .
a recent study of published dla system abundances in comparison with chemical evolution models of different galaxy morphologies by @xcite has identified the irregular and spiral galaxies as the possibly ideal sites to create the dla systems , while ellipticals can be ruled out .
the value of high resolution spectroscopy of the brightest blue stars in nearby galaxies for detailed , quantitative abundance studies has been further demonstrated in this work . to gain further insight into the chemical evolution of the dwarf galaxies and
hence into the early evolution of galaxies with low star formation rates , requires more detailed abundance analyses of their stars , particularly towards the lower metallicities .
we thank h.a .
kobulnicky for the low resolution keck spectrum of sexa-754 which allowed us to get this work started .
ak wants to thank the eso director general s discretionary fund ( dgdf ) for the support of this project and the institute of astronomy , university of cambridge , uk for its hospitality during a short - term visit where a major part of this work was done .
kav would like to thank the nsf for support through a career award , ast-9984073 , and the ioa cambridge for support during a one year visit .
cp wants to thank eso for the possibility of a short - term studentship at the eso facilities in vitacura , santiago de chile .
et gratefully acknowledges support from a fellowship of the royal netherlands academy of arts and sciences .
further we want to thank the referee a.j.korn for valuable comments on the manuscript .
asplund , m. , in _ cno in the universe _ , eds . c. charbonnel , d. schaerer , g. meynet ( san francisco : asp ) , in press ballester , p. , modigliani , a. , boitquin , o. , cristiani s. , hanuschik r. , kaufer , a. , wolf , s. 2000 , the eso messenger 101 , 31 becker , s.r .
1998 , in _ boulder - munich workshop ii : properties of hot , luminous stars _ ed .
howarth , asp conf .
ser . , 131 , 137 boyarchuk , a.a . ,
lyubimkov , l.s . ,
sakhibullin , n.a .
1985 , astrophysics , 21 , 203 calura , f. , matteucci , f. , vladilo , g. 2003 , , 340 , 59 cunha , k. , lambert , d.l .
1994 , , 426 , 170 dekker , h. , dodorico , s. , kaufer , a. , et al .
2000 , in _ optical and ir telescope instrumentation and detectors _ , eds .
m. iye and a.f .
moorwood , proc .
4008 , p. 534
dohm - palmer , r.c . ,
skillman , e.d . ,
saha , a. , et al .
1997 , , 114 , 2514 dohm - palmer , r.c . ,
skillman , e.d . ,
mateo , m. , et al .
2002 , , 123 , 813 dolphin , a.e . ,
saha , a. , skillman , e.d .
2003a , , 125 , 1261 dolphin , a.e . , saha , a. , skillman , e.d .
2003b , , 126 , 187 edvardsson , b. , andersen , j. , gustafsson , b. , et al .
1993 , , 275 , 101 gallart , c. , aparicio , a. , bertelli , g. , chiosi , c. , 1996 , , 112 , 1950 gigas , d. , 1986 , , 165 , 170 gigas d. , 1988 , , 192 , 264 gilmore , g. , wyse r.f.g . , 1991 , , 367 , l55 gummersbach , c.a .
, kaufer , a. , schaefer , d.r . , szeifert , t. , wolf , b. 1998 , , 338 , 881 hodge , p. , kennicutt , r.c . , stobel , n. 1994 , , 106 , 765 huchtmeier , w. k. , richter , o. g. 1986 , s , 63 , 323 kaufer , a. , stahl , o. , wolf , b. , et al .
1996 , , 305 , 887 kaufer , a. , stahl , o. , wolf , b. , et al .
1997 , , 320 , 273 kobulnicky , h.a . , skillman , e.d 1997 , , 489 , 636 kobulnicky , h.a .
1998 , priv .
korn , a.j . ,
keller , s.c .
, kaufer , a. , langer , n. , przybilla , n. , stahl , o. , wolf , b. 2002 , , 385 , 655 kurucz , r.l . , 1979 , , 40 , 1 kurucz , r.l . , 1988 , trans .
iau , vol .
20b , ed .
m. mcnally ( dordrecht : kluwer ) , 168 kurucz , r.l . , 1993 , cd - rom no.13 , cambridge , mass .
: smithsonian astrophysical observatory ledoux , c. , bergeron , j. , petitjean , p. 2002
, , 385 , 802 lejeune , t. , schaerer , d. 2001 , , 366 , 538 martin , c.m . , kobulnicky h. , heckman t. , 2002 , , 574 , 663 matteuchi , f. , tosi , m. 1985 , , 217 , 391 matteuchi , f. 2003 , , 284 , 539 mateo , m. 1998 , , 36 , 435 nissen , p.e . , chen , y.q . , asplund , m. , pettini , m. , 2003 , , in press nissen , p.e . ,
schuster , w.j .
1997 , , 326 , 751 oke , j. b. , cohen , j. g. , carr , m. , et al .
1995 , , 107 , 375 pettini , m. , ellison , s. , steidel , c.c . , bowen , d. 1999 , , 510 , 576 pilyugin , l.s .
2001 , , 374 , 412 prochaska , j.x . , gawiser , e. , wolfe , a.m. , castro , s. , djorgovski , s.g .
2003 , submitted to letters , astro - ph/0305314 przybilla , n. , butler k. , becker s.r .
, kudritzki r.p .
, venn k.a .
2000 , , 359 , 1085 przybilla , n. , butler k. , becker s.r .
, kudritzki r.p . , 2001 , , 369 , 1009 przybilla , n. 2002 , ph.d .
thesis , univ .
mnchen przybilla , n. 2003 , priv .
rentzsch - holm , i. 1996 , , 312 , 966 richer , m.g . ,
mccall , m.l .
, 1995 , 445 , 642 rolleston , w.r.j . ,
smartt , s.j . ,
dufton , p.l . ,
ryans , w.s.i .
2000 , , 363 , 537 shetrone , m.d . , ct , p. , sargent , w.l.w .
2001 , , 548 , 592 shetrone , m.d . , venn , k.a , tolstoy , e. , et al .
2003 , , 125 , 684 sheinis , a.i . , bolte m. , epps , h.w .
2002 , , 114 , 851 schmidt - kaler , th . 1982 , _ physical parameters of the stars _ in landolt - brnstein numerical data and functional relationships in science and technology , new series , group vi , volume 2b ( berlin : springer ) skillman , e. d. , terlevich , r. , teuben , p. j. , van woerden , h. , 1988 , , 198 , 33 skillman , e.d . , kennicutt , r.c . , hodge , p.w .
1989 , , 347 , 875 steinmetz , m. , navarro , j.f .
1999 , , 513 , 555 tinsley b. , 1979 , , 229 , 1046 tolstoy , e. , venn , k.a . , shetrone , m. , et al . , 2003 , , 125 , 707 tomita , a. , ohta , k. , saito , m. 1993 , , 45 , 693 van den bergh , s. 1999 , 517 , l97 van dyk , s.d . ,
puche , d. , wong , t. 1998 , 116 , 2341 venn , k.a . , 1995a , , 449 , 839 venn , k.a . ,
1995b , , 99 , 659 venn , k.a .
1999 , , 518 , 405 venn , k.a .
2000 , mccarthy , j.k . , lennon , d.j , et al .
2000 , , 541 , 610 venn , k.a . ,
kaufer , a. , mccarthy , j.m .
, et al . 2001 , , 547 , 765 venn , k.a . ,
tolstoy , e. , kaufer , a. , et al .
2003a , , in press venn , k.a . ,
tolstoy , e. , kaufer , a. , kudritzki , r.p .
, 2003b , in _ carnegie observatories astrophysics series , vol . 4 : origin and evolution of the elements _ , eds .
a. mcwilliam , m. rauch ( pasadena : carnegie observatories , ) wheeler , j.c .
, sneden , c. , truran , j. 1989 , , 27 , 279 spectrum synthesis for the the near - infrared @xmath84-@xmath115 multiplet lines .
the oxygen abundance was fitted using spectrum synthesis in lte .
the uncertainty of the lte abundances of @xmath93dex are due to the badly - defined continuum . to convert the measured lte abundances into sensible oxygen abundances detailed computation of the expectedly large nlte corrections @xcite have to be carried out ( see section [ abunds]).,width=453 ] lcccccc sexa-513 & 2002 - 02 - 13 & 08:45 & 750 & 2700 & 1.30 & 1.5 + & 2002 - 02 - 14 & 07:49 & 750 & 2700 & 1.38 & 1.6 + sexa-983 & 2002 - 02 - 13 & 09:44 & 750 & 3600 & 1.15 & 1.5 + sexa-1344 & 2002 - 02 - 14 & 09:59 & 750 & 3600 & 1.10 & 1.6 + sexa-1456 & 2002 - 02 - 13 & 11:02 & 750 & 2600 & 1.10 & 1.5 + sexa-754 & 2002 - 01 - 11 & 04:45 & 390 + 580 & 4500 & 1.39 & 0.98 + & 2002 - 01 - 13 & 05:55 & 390 + 580 & 4500 & 1.14 & 1.50 + & 2002 - 01 - 13 & 07:12 & 390 + 580 & 4500 & 1.06 & 1.20 + & 2002 - 01 - 18 & 04:55 & 390 + 580 & 4500 & 1.24 & 1.25 + & 2002 - 01 - 20 & 07:17 & 390 + 580 & 4500 & 1.06 & 0.97 + & 2002 - 01 - 23 & 05:55 & 390 + 580 & 4500 & 1.08 & 0.81 + & 2002 - 02 - 08 & 02:00 & 390 + 580 & 4500 & 1.80 & 1.19 + & 2002 - 02 - 09 & 02:32 & 390 + 580 & 4500 & 1.50 & 0.73 + & 2002 - 02 - 10 & 01:55 & 390 + 580 & 4500 & 1.77 & 0.75 + & 2002 - 02 - 11 & 02:41 & 390 + 580 & 4500 & 1.40 & 0.58 + & 2002 - 02 - 12 & 02:14 & 390 + 580 & 4500 & 1.55 & 1.13 + & 2002 - 04 - 14 & 00:58 & 437 + 840 & 4500 & 1.07 & 1.00 + & 2002 - 04 - 14 & 02:15 & 437 + 840 & 4500 & 1.08 & 1.02 + & 2002 - 04 - 15 & 00:30 & 437 + 840 & 4500 & 1.08 & 0.80 + & 2002 - 04 - 15 & 01:48 & 437 + 840 & 3600 & 1.07 & 0.67 + & 2002 - 04 - 15 & 02:50 & 437 + 840 & 3600 & 1.14 & 0.87 + & 2002 - 04 - 15 & 03:51 & 437 + 840 & 3600 & 1.33 & 1.01 + sexa-1344 & 2002 - 12 - 30 & 05:19 & 390 + 580 & 4500 & 1.48 & 1.25 + & 2002 - 12 - 30 & 06:36 & 390 + 580 & 4500 & 1.17 & 1.03 + & 2003 - 01 - 01 & 07:24 & 390 + 580 & 4500 & 1.08 & 1.42 + & 2003 - 01 - 08 & 04:39 & 390 + 580 & 4500 & 1.50 & 0.84 + & 2003 - 01 - 08 & 05:58 & 390 + 580 & 4500 & 1.18 & 1.08 + & 2003 - 01 - 08 & 07:16 & 390 + 580 & 4500 & 1.07 & 0.61 + & 2003 - 02 - 04 & 03:12 & 390 + 580 & 4500 & 1.39 & 1.00 + & 2003 - 02 - 04 & 04:29 & 390 + 580 & 4500 & 1.14 & 1.27 + & 2003 - 02 - 04 & 05:46 & 390 + 580 & 4500 & 1.06 & 1.15 + & 2003 - 02 - 05 & 04:38 & 390 + 580 & 4500 & 1.12 & 1.29 + sexa-1911 & 2003 - 01 - 12 & 04:58 & 390 + 580 & 4500 & 1.32 & 1.19 + & 2003 - 02 - 05 & 05:59 & 390 + 580 & 4500 & 1.06 & 1.02 + & 2003 - 02 - 05 & 07:19 & 390 + 580 & 4500 & 1.14 & 1.17 + & 2003 - 02 - 06 & 03:43 & 390 + 580 & 4500 & 1.23 & 0.93 + & 2003 - 02 - 07 & 04:13 & 390 + 580 & 4500 & 1.15 & 0.78 + & 2003 - 02 - 07 & 05:29 & 390 + 580 & 4500 & 1.06 & 0.95 + & 2003 - 02 - 07 & 07:49 & 390 + 580 & 4500 & 1.23 & 0.93 + & 2003 - 02 - 08 & 04:07 & 390 + 580 & 4500 & 1.15 & 1.22 + & 2003 - 02 - 08 & 06:22 & 390 + 580 & 4500 & 1.08 & 1.09 + & 2003 - 02 - 08 & 07:40 & 390 + 580 & 4500 & 1.21 & 0.94 + sexa-513 & 2002 - 04 - 13 & 23:19 & 390 + 580 & 1800 & 1.23 & 1.12 + & 2002 - 04 - 14 & 23:23 & 390 + 580 & 1800 & 1.21 & 0.91 + & 2002 - 04 - 14 & 00:12 & 390 + 580 & 1800 & 1.14 & 0.72 + sexa-983 & 2002 - 04 - 13 & 23:53 & 390 + 580 & 3600 & 1.15 & 1.13 + lrrcll sexa-754 & @xmath116 & @xmath117 & @xmath118 & a7iab supergiant & uves ( cand .
2 ) + sexa-1344 & @xmath119 & @xmath120 & @xmath121 & a7iab supergiant & esi+uves ( cand .
4 ) + sexa-1911 & @xmath122 & @xmath123 & @xmath121 & a8iab supergiant & uves ( cand .
5 ) + sexa-513 & @xmath124 & @xmath125 & @xmath126 & f hypergiant & esi+uves ( cand .
1 ) + sexa-983 & @xmath127 & @xmath128 & @xmath129 & g supergiant & esi+uves ( cand .
3 ) + sexa-1456 & @xmath130 & @xmath131 & @xmath121 & k supergiant & esi ( cand .
6 ) + lcccc @xmath52[k ] & & @xmath132 & @xmath133 & @xmath134 + @xmath50[cm / s@xmath135 & & @xmath136 & @xmath137 & @xmath136 + @xmath138[km / s ] & & @xmath139 & @xmath140 & @xmath140 + @xmath77[km / s ] & & @xmath141 & @xmath142 & @xmath143 + @xmath144 & & @xmath145 & @xmath146 & @xmath147 + @xmath148 & & @xmath149 & @xmath149 & @xmath150 + sp.type & & a7iab & a7iab & a8iab + mgi ( nlte ) & @xmath151 & @xmath152 & @xmath153 & @xmath154 + mgii ( nlte ) & @xmath151 & @xmath155 & & + siii ( lte ) & @xmath156 & @xmath157 & @xmath158 & @xmath159 + tiii ( lte ) & @xmath160 & @xmath161 & @xmath162 & @xmath163 + scii ( lte ) & @xmath164 & @xmath165 & @xmath166 & @xmath167 + cri ( lte ) & @xmath168 & @xmath169 & @xmath170 & @xmath171 + crii ( lte ) & @xmath168 & @xmath172 & @xmath173 & @xmath174 + fei ( lte ) & @xmath175 & @xmath176 & @xmath177 & @xmath178 + feii ( lte ) & @xmath175 & @xmath179 & @xmath180 & @xmath181 + lrrrrrrrrr mg i & @xmath182&@xmath183&@xmath183 & @xmath184&@xmath185&@xmath186 & @xmath187&@xmath183&@xmath188 + mg ii & @xmath189&@xmath190&@xmath191 & @xmath192&@xmath193&@xmath194 & @xmath195&@xmath196&@xmath197 + si ii & @xmath198&@xmath190&@xmath183 & @xmath196&@xmath190&@xmath199 & @xmath189&@xmath190&@xmath200 + ti ii & @xmath201&@xmath190&@xmath202 & @xmath203&@xmath193&@xmath204 & @xmath205&@xmath190&@xmath206 + sc ii & @xmath207&@xmath196&@xmath186 & @xmath182&@xmath204&@xmath186 & @xmath68&@xmath196&@xmath206 + cr i & @xmath208&@xmath183&@xmath198 & @xmath209&@xmath210&@xmath211 & @xmath208&@xmath204&@xmath212 + cr ii & @xmath213&@xmath190&@xmath193 & @xmath213&@xmath211&@xmath193 & @xmath213&@xmath190&@xmath212 + fe i & @xmath67&@xmath183&@xmath202 & @xmath214&@xmath210&@xmath204 & @xmath67&@xmath204&@xmath215 + fe ii & @xmath216&@xmath190&@xmath200 & @xmath216&@xmath211&@xmath188 & @xmath217&@xmath190&@xmath185 + ccccc cc cc cc @xmath218 & @xmath219 & @xmath55 & @xmath220 & @xmath221 & & & + @xmath218 & @xmath222 & @xmath55 & @xmath220 & @xmath223 & _ 700 _ & @xmath224 & & & & + @xmath218 & @xmath225 & @xmath55 & @xmath220 & @xmath226 & & & + @xmath218 & @xmath227 & @xmath228 & @xmath229 & @xmath230 & & & + @xmath218 & @xmath231 & @xmath228 & @xmath229 & @xmath232 & _ 385 _ & @xmath233 & & & & + @xmath218 & @xmath234 & @xmath228 & @xmath229 & @xmath235 & & & + @xmath236 & @xmath237 & @xmath238 & @xmath239 & @xmath240 & @xmath241 & @xmath242 & @xmath243 & @xmath244 & @xmath241 & @xmath245 + @xmath236 & @xmath246 & @xmath247 & @xmath239 & @xmath248 & @xmath249 & @xmath244 & @xmath250 & @xmath251 & @xmath252 & @xmath253 + @xmath236 & @xmath254 & @xmath247 & @xmath239 & @xmath255 & @xmath256 & @xmath242 & @xmath243 & @xmath257 & @xmath258 & @xmath242 + @xmath236 & @xmath259 & @xmath247 & @xmath260 & @xmath185 & @xmath261 & @xmath244 & @xmath262 & @xmath263 & @xmath261 & @xmath264 + @xmath265 & @xmath266 & @xmath267 & @xmath268 & @xmath269 & @xmath270 & @xmath271 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath265 & @xmath273 & @xmath18 & @xmath274 & @xmath275 & @xmath270 & @xmath276 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath265 & @xmath277 & @xmath278 & @xmath274 & @xmath279 & @xmath280 & @xmath281 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath265 & @xmath282 & @xmath278 & @xmath274 & @xmath283 & @xmath280 & @xmath284 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath265 & @xmath285 & @xmath228 & @xmath268 & @xmath286 & & & + @xmath265 & @xmath287 & @xmath228 & @xmath268 & @xmath288 & @xmath289 & @xmath290 & @xmath291 & @xmath292 & @xmath291 & @xmath293 + @xmath265 & @xmath294 & @xmath228 & @xmath268 & @xmath295 & & & + @xmath296 & @xmath297 & @xmath55 & @xmath298 & @xmath299 & @xmath300 & @xmath301 & @xmath300 & @xmath301 & @xmath300 & @xmath302 + @xmath296 & @xmath303 & @xmath55 & @xmath298 & @xmath304 & @xmath258 & @xmath305 & @xmath241 & @xmath306 & @xmath241 & @xmath307 + @xmath296 & @xmath308 & @xmath55 & @xmath298 & @xmath309 & @xmath310 & @xmath311 & @xmath312 & @xmath307 & @xmath312 & @xmath313 + @xmath296 & @xmath314 & @xmath238 & @xmath315 & @xmath316 & @xmath243 & @xmath317 & @xmath318 & @xmath319 & @xmath318 & @xmath320 + @xmath296 & @xmath321 & @xmath238 & @xmath315 & @xmath322 & @xmath318 & @xmath323 & @xmath249 & @xmath324 & @xmath249 & @xmath325 + @xmath296 & @xmath326 & @xmath267 & @xmath327 & @xmath328 & @xmath250 & @xmath329 & @xmath330 & @xmath331 & @xmath330 & @xmath332 + @xmath296 & @xmath333 & @xmath267 & @xmath327 & @xmath334 & @xmath300 & @xmath329 & @xmath300 & @xmath335 & @xmath300 & @xmath336 + @xmath296 & @xmath337 & @xmath247 & @xmath338 & @xmath339 & @xmath241 & @xmath340 & @xmath341 & @xmath342 & @xmath341 & @xmath305 + @xmath296 & @xmath343 & @xmath247 & @xmath338 & @xmath204 & @xmath243 & @xmath344 & @xmath345 & @xmath346 & @xmath345 & @xmath329 + @xmath347 & @xmath348 & @xmath349 & @xmath350 & @xmath232 & @xmath241 & @xmath351 & @xmath262 & @xmath352 & @xmath241 & @xmath353 + @xmath347 & @xmath354 & @xmath31 & @xmath355 & @xmath183 & @xmath249 & @xmath356 & @xmath330 & @xmath357 & @xmath243 & @xmath358 + @xmath347 & @xmath359 & @xmath31 & @xmath360 & @xmath240 & @xmath243 & @xmath361 & @xmath330 & @xmath353 & @xmath243 & @xmath362 + @xmath347 & @xmath363 & @xmath31 & @xmath364 & @xmath365 & @xmath300 & @xmath366 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath367 & @xmath351 + @xmath347 & @xmath368 & @xmath369 & @xmath355 & @xmath370 & @xmath300 & @xmath371 & @xmath372 & @xmath366 & @xmath373 & @xmath374 + @xmath347 & @xmath375 & @xmath369 & @xmath364 & @xmath376 & @xmath372 & @xmath374 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath378 & @xmath379 & @xmath380 & @xmath200 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath241 & @xmath381 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath382 & @xmath383 & @xmath384 & @xmath365 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath385 & @xmath386 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath387 & @xmath383 & @xmath388 & @xmath389 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath256 & @xmath390 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath391 & @xmath383 & @xmath384 & @xmath392 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath393 & @xmath390 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath394 & @xmath395 & @xmath295 & @xmath396 & @xmath310 & @xmath397 & @xmath398 & @xmath386 & @xmath399 & @xmath400 + @xmath377 & @xmath401 & @xmath395 & @xmath360 & @xmath402 & @xmath367 & @xmath403 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath249 & @xmath404 + @xmath377 & @xmath405 & @xmath406 & @xmath407 & @xmath408 & @xmath398 & @xmath400 & @xmath330 & @xmath409 & @xmath367 & @xmath410 + @xmath377 & @xmath411 & @xmath406 & @xmath407 & @xmath412 & @xmath373 & @xmath413 & @xmath414 & @xmath415 & @xmath243 & @xmath416 + @xmath377 & @xmath417 & @xmath418 & @xmath419 & @xmath420 & @xmath372 & @xmath421 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath372 & @xmath415 + @xmath377 & @xmath422 & @xmath423 & @xmath424 & @xmath425 & @xmath243 & @xmath426 & @xmath367 & @xmath404 & @xmath262 & @xmath427 + @xmath377 & @xmath428 & @xmath423 & @xmath429 & @xmath430 & @xmath373 & @xmath431 & @xmath367 & @xmath432 & @xmath262 & @xmath433 + @xmath377 & @xmath434 & @xmath435 & @xmath436 & @xmath437 & @xmath258 & @xmath438 & @xmath249 & @xmath386 & @xmath399 & @xmath439 + @xmath377 & @xmath440 & @xmath36 & @xmath419 & @xmath441 & @xmath399 & @xmath386 & @xmath249 & @xmath442 & @xmath310 & @xmath431 + @xmath377 & @xmath443 & @xmath435 & @xmath444 & @xmath445 & @xmath258 & @xmath404 & @xmath341 & @xmath446 & @xmath261 & @xmath447 + @xmath377 & @xmath448 & @xmath435 & @xmath436 & @xmath449 & @xmath450 & @xmath442 & @xmath318 & @xmath451 & @xmath262 & @xmath410 + @xmath377 & @xmath452 & @xmath435 & @xmath436 & @xmath453 & @xmath345 & @xmath400 & @xmath243 & @xmath454 & @xmath455 & @xmath426 + @xmath377 & @xmath456 & @xmath435 & @xmath444 & @xmath449 & @xmath450 & @xmath457 & @xmath318 & @xmath386 & @xmath262 & @xmath458 + @xmath377 & @xmath459 & @xmath435 & @xmath436 & @xmath460 & @xmath450 & @xmath461 & @xmath243 & @xmath462 & @xmath310 & @xmath426 + @xmath377 & @xmath463 & @xmath464 & @xmath465 & @xmath466 & @xmath372 & @xmath467 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath446 + @xmath377 & @xmath468 & @xmath435 & @xmath436 & @xmath469 & @xmath367 & @xmath470 & @xmath330 & @xmath471 & @xmath330 & @xmath386 + @xmath377 & @xmath472 & @xmath435 & @xmath444 & @xmath473 & @xmath372 & @xmath390 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath367 & @xmath433 + @xmath377 & @xmath474 & @xmath475 & @xmath476 & @xmath453 & @xmath330 & @xmath477 & @xmath372 & @xmath381 & @xmath367 & @xmath478 + @xmath377 & @xmath479 & @xmath480 & @xmath429 & @xmath481 & @xmath482 & @xmath447 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath483 & @xmath484 & @xmath485 & @xmath486 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath400 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath487 & @xmath488 & @xmath419 & @xmath489 & @xmath258 & @xmath457 & @xmath241 & @xmath446 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath490 & @xmath491 & @xmath492 & @xmath493 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath494 & @xmath478 & @xmath414 & @xmath495 + @xmath377 & @xmath496 & @xmath484 & @xmath492 & @xmath497 & @xmath262 & @xmath498 & @xmath318 & @xmath404 & @xmath385 & @xmath433 + @xmath377 & @xmath499 & @xmath500 & @xmath501 & @xmath502 & @xmath262 & @xmath427 & @xmath373 & @xmath503 & @xmath312 & @xmath470 + @xmath377 & @xmath504 & @xmath475 & @xmath476 & @xmath505 & @xmath372 & @xmath426 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath506 & @xmath488 & @xmath419 & @xmath507 & @xmath261 & @xmath508 & @xmath385 & @xmath509 & @xmath510 & @xmath427 + @xmath377 & @xmath511 & @xmath500 & @xmath436 & @xmath512 & @xmath330 & @xmath432 & @xmath482 & @xmath415 & @xmath367 & @xmath513 + @xmath377 & @xmath514 & @xmath515 & @xmath384 & @xmath516 & @xmath261 & @xmath517 & @xmath385 & @xmath462 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath518 & @xmath519 & @xmath380 & @xmath520 & @xmath258 & @xmath403 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath521 & @xmath522 & @xmath523 & @xmath524 & @xmath310 & @xmath409 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath525 & @xmath519 & @xmath526 & @xmath527 & @xmath310 & @xmath528 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath377 & @xmath529 & @xmath530 & @xmath476 & @xmath437 & @xmath367 & @xmath454 & @xmath372 & @xmath531 & @xmath300 & @xmath409 + @xmath377 & @xmath532 & @xmath533 & @xmath534 & @xmath535 & @xmath372 & @xmath536 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath414 & @xmath537 + @xmath377 & @xmath538 & @xmath533 & @xmath539 & @xmath540 & @xmath414 & @xmath517 & @xmath250 & @xmath503 & @xmath398 & @xmath498 + @xmath377 & @xmath541 & @xmath542 & @xmath407 & @xmath543 & @xmath330 & @xmath426 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath517 + @xmath377 & @xmath544 & @xmath542 & @xmath407 & @xmath545 & @xmath250 & @xmath439 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath546 + @xmath377 & @xmath547 & @xmath548 & @xmath380 & @xmath549 & @xmath414 & @xmath550 & @xmath330 & @xmath442 & @xmath318 & @xmath458 + @xmath377 & @xmath551 & @xmath548 & @xmath526 & @xmath453 & @xmath300 & @xmath552 & @xmath367 & @xmath381 & @xmath243 & @xmath553 + @xmath377 & @xmath554 & @xmath555 & @xmath380 & @xmath556 & @xmath494 & @xmath404 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath557 & @xmath558 & @xmath55 & @xmath226 & @xmath200 & @xmath330 & @xmath559 & @xmath372 & @xmath560 & @xmath414 & @xmath561 + @xmath557 & @xmath562 & @xmath55 & @xmath226 & @xmath563 & @xmath494 & @xmath564 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath557 & @xmath565 & @xmath349 & @xmath566 & @xmath567 & @xmath494 & @xmath559 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath569 & @xmath488 & @xmath164 & @xmath570 & @xmath372 & @xmath571 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath572 + @xmath568 & @xmath573 & @xmath574 & @xmath575 & @xmath576 & @xmath482 & @xmath577 & @xmath482 & @xmath578 & @xmath494 & @xmath579 + @xmath568 & @xmath580 & @xmath581 & @xmath582 & @xmath583 & @xmath482 & @xmath584 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath585 & @xmath515 & @xmath586 & @xmath587 & @xmath243 & @xmath588 & @xmath367 & @xmath589 & @xmath318 & @xmath590 + @xmath568 & @xmath591 & @xmath515 & @xmath592 & @xmath593 & @xmath482 & @xmath594 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath494 & @xmath572 + @xmath568 & @xmath595 & @xmath515 & @xmath592 & @xmath596 & @xmath373 & @xmath578 & @xmath414 & @xmath597 & @xmath250 & @xmath561 + @xmath568 & @xmath598 & @xmath515 & @xmath592 & @xmath599 & @xmath330 & @xmath578 & @xmath250 & @xmath600 & @xmath330 & @xmath601 + @xmath568 & @xmath602 & @xmath515 & @xmath603 & @xmath604 & @xmath482 & @xmath577 & @xmath482 & @xmath597 & @xmath372 & @xmath605 + @xmath568 & @xmath606 & @xmath607 & @xmath608 & @xmath489 & @xmath310 & @xmath577 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath609 & @xmath607 & @xmath608 & @xmath376 & @xmath399 & @xmath610 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath611 & @xmath607 & @xmath608 & @xmath441 & @xmath367 & @xmath612 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath613 & @xmath614 & @xmath586 & @xmath549 & @xmath318 & @xmath559 & @xmath243 & @xmath615 & @xmath318 & @xmath610 + @xmath568 & @xmath616 & @xmath614 & @xmath592 & @xmath617 & @xmath398 & @xmath618 & @xmath243 & @xmath578 & @xmath249 & @xmath619 + @xmath568 & @xmath620 & @xmath614 & @xmath592 & @xmath621 & @xmath373 & @xmath622 & @xmath300 & @xmath623 & @xmath330 & @xmath624 + @xmath568 & @xmath625 & @xmath614 & @xmath592 & @xmath626 & @xmath627 & @xmath588 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath568 & @xmath628 & @xmath629 & @xmath630 & @xmath449 & @xmath373 & @xmath619 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath300 & @xmath631 + @xmath568 & @xmath632 & @xmath629 & @xmath633 & @xmath453 & @xmath414 & @xmath634 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath631 + @xmath568 & @xmath635 & @xmath629 & @xmath608 & @xmath269 & @xmath482 & @xmath636 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath588 + @xmath637 & @xmath638 & @xmath418 & @xmath639 & @xmath640 & @xmath250 & @xmath641 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath367 & @xmath642 + @xmath637 & @xmath643 & @xmath36 & @xmath644 & @xmath567 & @xmath258 & @xmath645 & @xmath399 & @xmath646 & @xmath258 & @xmath642 + @xmath637 & @xmath647 & @xmath36 & @xmath648 & @xmath212 & @xmath399 & @xmath649 & @xmath450 & @xmath650 & @xmath256 & @xmath651 + @xmath637 & @xmath652 & @xmath228 & @xmath226 & @xmath653 & @xmath310 & @xmath654 & @xmath450 & @xmath655 & @xmath241 & @xmath656 + @xmath637 & @xmath657 & @xmath228 & @xmath658 & @xmath659 & @xmath372 & @xmath660 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath637 & @xmath661 & @xmath228 & @xmath43 & @xmath662 & @xmath300 & @xmath646 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath637 & @xmath663 & @xmath228 & @xmath664 & @xmath486 & @xmath372 & @xmath665 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath300 & @xmath666 + @xmath637 & @xmath667 & @xmath629 & @xmath668 & @xmath669 & @xmath300 & @xmath670 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath373 & @xmath671 + @xmath637 & @xmath672 & @xmath629 & @xmath673 & @xmath674 & @xmath385 & @xmath675 & @xmath252 & @xmath650 & @xmath385 & @xmath676 + @xmath637 & @xmath677 & @xmath629 & @xmath668 & @xmath678 & @xmath252 & @xmath654 & @xmath367 & @xmath650 & @xmath399 & @xmath642 + @xmath637 & @xmath679 & @xmath629 & @xmath680 & @xmath211 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath373 & @xmath681 & @xmath262 & @xmath682 + @xmath637 & @xmath683 & @xmath629 & @xmath680 & @xmath684 & @xmath372 & @xmath685 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath686 + @xmath637 & @xmath687 & @xmath629 & @xmath668 & @xmath688 & @xmath414 & @xmath660 & @xmath372 & @xmath689 & @xmath373 & @xmath686 + @xmath637 & @xmath690 & @xmath691 & @xmath692 & @xmath693 & @xmath627 & @xmath694 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath637 & @xmath695 & @xmath691 & @xmath696 & @xmath693 & @xmath627 & @xmath697 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath637 & @xmath698 & @xmath699 & @xmath673 & @xmath653 & @xmath414 & @xmath655 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath300 & @xmath660 + @xmath700 & @xmath701 & @xmath369 & @xmath702 & @xmath703 & @xmath243 & @xmath670 & @xmath252 & @xmath313 & @xmath252 & @xmath264 + @xmath700 & @xmath704 & @xmath705 & @xmath706 & @xmath707 & @xmath250 & @xmath346 & @xmath372 & @xmath329 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath700 & @xmath708 & @xmath709 & @xmath710 & @xmath711 & @xmath373 & @xmath302 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath700 & @xmath712 & @xmath713 & @xmath710 & @xmath714 & @xmath261 & @xmath253 & @xmath715 & @xmath325 & @xmath258 & @xmath716 + @xmath700 & @xmath717 & @xmath709 & @xmath710 & @xmath707 & @xmath241 & @xmath689 & @xmath385 & @xmath718 & @xmath385 & @xmath264 + @xmath700 & @xmath719 & @xmath709 & @xmath720 & @xmath721 & @xmath300 & @xmath263 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath330 & @xmath716 + @xmath700 & @xmath722 & @xmath713 & @xmath720 & @xmath723 & @xmath250 & @xmath697 & @xmath300 & @xmath324 & @xmath414 & @xmath681 + @xmath700 & @xmath724 & @xmath713 & @xmath720 & @xmath725 & @xmath261 & @xmath670 & @xmath258 & @xmath311 & @xmath241 & @xmath660 + @xmath700 & @xmath726 & @xmath709 & @xmath727 & @xmath728 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath372 & @xmath319 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath700 & @xmath729 & @xmath713 & @xmath727 & @xmath730 & @xmath310 & @xmath244 & @xmath399 & @xmath731 & @xmath399 & @xmath264 + @xmath700 & @xmath732 & @xmath733 & @xmath734 & @xmath735 & @xmath243 & @xmath736 & @xmath262 & @xmath718 & @xmath272@xmath272 & @xmath272@xmath272 + @xmath700 & @xmath737 & @xmath738 & @xmath739 & @xmath725 & @xmath341 & @xmath655 & @xmath341 & @xmath335 & @xmath385 & @xmath253 + @xmath700 & @xmath740 & @xmath738 & @xmath741 & @xmath742 & @xmath241 & @xmath245 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath385 & @xmath743 + @xmath700 & @xmath744 & @xmath607 & @xmath745 & @xmath746 & @xmath256 & @xmath736 & @xmath747 & @xmath329 & @xmath747 & @xmath748 + @xmath700 & @xmath749 & @xmath738 & @xmath750 & @xmath751 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath258 & @xmath257 & @xmath258 & @xmath655 + @xmath700 & @xmath752 & @xmath753 & @xmath754 & @xmath420 & @xmath300 & @xmath716 & @xmath398 & @xmath331 & @xmath373 & @xmath716 + @xmath700 & @xmath755 & @xmath753 & @xmath754 & @xmath756 & @xmath272@xmath272 & @xmath272@xmath272 & @xmath414 & @xmath306 & @xmath272@xmath272 & @xmath272@xmath272 + ccccccccccccc 3829.36 & @xmath757 & 2.71 & @xmath240 & 140 & 6.47 & 6.50 & 80 & 6.39 & 6.49 & 140 & 6.39 & 6.42 + 5167.32 & @xmath758 & 2.71 & @xmath248 & 75 & 6.40 & 6.49 & 35 & 6.41 & 6.54 & 85 & 6.33 & 6.40 + 5172.68 & @xmath758 & 2.71 & @xmath255 & 135 & 6.46 & 6.50 & 80 & 6.45 & 6.58 & 150 & 6.52 & 6.50 + 5183.60 & @xmath758 & 2.72 & @xmath185 & 160 & 6.50 & 6.49 & 100 & 6.44 & 6.55 & 160 & 6.43 & 6.42 + 3848.21 & @xmath759 & 8.86 & @xmath269 & @xmath83 & @xmath271 & @xmath271 & & & & & & + 4390.57 & @xmath760 & 10.00 & @xmath275 & @xmath83 & @xmath276 & @xmath276 & & & & & & + 7877.05 & @xmath761 & 10.00 & @xmath762 & @xmath763 & @xmath764 & @xmath281 & & & & & & + 7896.37 & @xmath761 & 10.00 & @xmath765 & @xmath763 & @xmath766 & @xmath284 & & & & & & + | we present the abundance analyses of three isolated a - type supergiant stars in the dwarf irregular galaxy (= ddo75 ) from high - resolution spectra obtained with ultraviolet - visual echelle spectrograph ( uves ) on the kueyen telescope ( ut2 ) of the eso very large telescope ( vlt ) .
detailed model atmosphere analyses have been used to determine the stellar atmospheric parameters and the elemental abundances of the stars .
the mean iron group abundance was determined from these three stars to be @xmath0\right>=-0.99\pm0.04\pm{\it 0.06}$]=\log(x/{\rm h})-\log(x/{\rm h})_\odot$ ] .
averages of the abundances of the three stars are marked with @xmath1 .
all abundances are given with two uncertainties : first the line - to - line scatter , second in italics the estimate of the systematic error due to uncertainties in the stellar atmospheric parameters . ] .
this is the first determination of the present - day iron group abundances in sextansa .
these three stars now represent the most metal - poor massive stars for which detailed abundance analyses have been carried out .
the mean stellar @xmath2 element abundance was determined from the @xmath2 element magnesium as @xmath3\right>=-1.09\pm0.02\pm{\it 0.19}$ ] .
this is in excellent agreement with the nebular @xmath2 element abundances as determined from oxygen in the regions .
these results are consistent from star - to - star with no significant spatial variations over a length of @xmath4kpc in sextansa .
this supports the nebular abundance studies of dwarf irregular galaxies , where homogeneous oxygen abundances are found throughout , and argues against in situ ( `` on the spot '' ) enrichment .
the @xmath5 abundance ratio is @xmath6\right>=-0.11\pm0.02\pm{\it 0.10}$ ] , which is slightly lower but consistent with the solar ratio .
this is consistent with the results from a - supergiant analyses in other local group dwarf irregular galaxies , ngc6822 and wlm .
the results of near solar [ @xmath2/fe ] ratios in dwarf galaxies is in stark contrast with the high [ @xmath2/fe ] results from metal - poor stars in the galaxy ( which plateau at values near + 0.4 dex ) , and is most clearly seen from these three stars in sextansa because of their lower metallicities .
the low [ @xmath2/fe ] ratios are consistent with the slow chemical evolution expected for dwarf galaxies from analyses of their stellar populations . |
the radioactive isotope @xmath0fe is believed to be synthesized through successive neutron captures on fe isotopes ( e.g. , @xmath6fe ) in a neutron - rich environment inside he burning shells in agb stars ( @xmath7fe is stored in white dwarfs and can not be ejected ) and massive stars , before or during their final evolution to core collapse supernovae ( ccsn ) .
@xmath7fe can be also synthesized in type ia sne ( woosley 1997 ) .
it is also destroyed by the @xmath7fe ( @xmath8 ) process . since its closest parent ,
@xmath9fe is unstable , the @xmath9fe(@xmath8 ) process must compete with the @xmath9fe(@xmath10 ) decay to produce an appreciate amount of @xmath7fe .
the decay chains of @xmath7fe are shown in figure 1 .
@xmath7fe firstly decays to @xmath0co , with emitting @xmath1-ray photons at 59 kev , and then decays to @xmath0ni , with emitting @xmath1-ray photons at 1173 and 1333 kev .
the gamma - ray efficiency of the 59 kev transition is only @xmath11 of those at 1173 and 1333 kev , so the gamma - ray flux at 59 kev is much lower than the fluxes of the high energy lines .
the 59 kev gamma - ray line is very difficult to be detected with present missions .
measurements of the two high energy lines have been the main scientific target to study the radioactive @xmath7fe isotope in the galaxy .
@xmath7fe has been found to be part of meteorites formed in the early solar system ( shukolyukov et al .
the inferred @xmath7fe /@xmath6fe ratio for these meteorites exceeded the interstellar - medium estimates from nucleosynthesis models , which led to suggestion that the late supernova ejection of @xmath7fe occurred before formation of the solar system ( tachibana et al .
yet , this is a proof for cosmic @xmath7fe production , accelerator - mass spectroscopy of seafloor crust material from the southern pacific ocean has revealed an @xmath7fe excess in a crust depth corresponding to an age of 2.8 myr ( knie et al .
2004 ) . from this
interesting measurement , it is concluded that a supernova explosion event near the solar system occurred about 3 myr ago , depositing some of its debris directly in the earth s atmosphere .
all these measurements based on material samples demonstrate that @xmath7fe necleosynthesis does occur in nature .
it is now interesting to search for current @xmath7fe production in the galaxy through detecting radioactive - decay @xmath1-ray lines .
fe . the mean lifetime is @xmath12 years .
the gamma - ray flux at 59 kev line is @xmath11 of those at 1173 and 1333 kev.,width=377 ]
due to its long decay time ( @xmath132.2 my ) , @xmath0fe survives to be detected after the supernova ejected it into the interstellar medium , by @xmath14-decay via @xmath0co and @xmath1 emission at @xmath15 kev and 1333 kev like other radioactive isotopes : @xmath16ti , @xmath17co , and @xmath3al .
these isotopes provide evidence that nucleosynthesis is ongoing in the galaxy ( the et al .
2006 ; diehl et al .
specially , measurements of @xmath0fe promise to provide new information about the massive star nucleosynthesis in the late pre - supernova stages
. gamma - ray signal of @xmath7fe from the sky is very weak , so there are no confident detections of @xmath7fe in the galaxy reported in the previous measurements .
recently , rhessi reported observations of the gamma - ray lines from @xmath0fe with an average flux of @xmath18 ( smith 2004 ) . now , the spectrometer aboard integral ( spi ) operates on space .
the _ inte_rnational _
g_amma-_r_ay _ a_strophysics
_ l_aboratory ( integral ) is an european ( esa ) gamma - ray observatory satellite mission for the study of cosmic gamma - ray sources in the kev to mev energy range ( winkler et al .
integral was successfully launched from baikonur cosmodrome ( kazakhstan ) on october 17 , 2002 .
the integral orbit is eccentric , with an apogee of 153 000 km , a perigee of 9000 km , and a 3 day period .
integral will continue to work until 2012 approved by esa .
spi / integral consists of 19 high purity germanium detectors which allow for high spectral resolution of @xmath19 kev at 1 mev , suitable for astrophysical studies of individual gamma - ray lines and their shapes , e.g. the 511 kev line , @xmath1-ray lines from radioactivities of @xmath16ti , @xmath20al and @xmath7fe .
the basic measurement of spi consists of event messages per photon triggering the ge detector camera .
we distinguish events which trigger a single ge detector element only ( _ single event _ , se ) , and events which trigger more than two ge detector elements nearly simultaneously ( _ multiple event _ , me ) .
we analyzed the first year of integral data to detect the @xmath1-ray lines from @xmath0fe with an average line flux of @xmath21 ( harris et al .
but the strong background lines near @xmath0fe lines still contaminate the spectra , which makes this preliminary results questionable . at present
, we use three years of integral data ( from 2003.3 - 2006.3 ) , aiming at a consolidation of the integral / spi measurement of @xmath0fe gamma - rays .
the newest results on @xmath0fe gamma - ray lines by integral / spi are presented in figure 2 .
all the fluxes given by different databases are consistent with each other .
the strong background line at 1337 kev has also been eliminated rather well .
furthermore , a superposition of the four spectra of figure 2 is shown in figure 3 . the line flux estimated from the combined spectrum
is @xmath22 .
our significance estimate for the combined spectrum is @xmath23 ( wang et al .
2007 ) .
fe from the inner galaxy : 1173 kev and 1333 kev .
we have shown the results both from se and me databases .
for the se database , we find a line flux of @xmath24 for the 1173 kev line and @xmath25 for the 1333 kev line .
for the me database , the line flux is @xmath26 for the 1173 kev line and @xmath27 for the 1333 kev line , width=529 ] fe signal in the inner galaxy , superimposing the four spectra of figure 2 . in the laboratory , the line energies are 1173.23 and 1332.49 kev ; here superimposed bins are zero at 1173 and 1333 kev .
we find a detection significance of 5@xmath28 .
the average line flux is estimated as @xmath29 .
, width=340 ]
@xmath20al is an unstable isotope with a mean lifetime of 1.04 myr .
@xmath20al can first decay into an excited state of @xmath3 mg , which de - excites into the mg ground state by emitting gamma - ray photons with the characteristic energy of 1809 kev .
@xmath20al is produced almost exclusively by proton capture on @xmath30 mg in a sufficiently hot environment .
@xmath20al origin is dominated by massive star and core - collapse supernovae , and small part of @xmath20al is attributed to agb stars and novae .
therefore , @xmath3al and @xmath0fe would share at least some of the same production sites , i.e. massive stars and supernovae .
in addition both are long - lived radioactive isotopes , so we believe their gamma - ray distributions are similar as well .
we derive the ratio of @xmath7fe /@xmath20al , which can be directly compared with theoretical predictions .
we also obtain the @xmath20al spectrum in the galaxy using three years of integral data which is shown in figure 4 .
@xmath20al flux is @xmath31 . combining the @xmath7fe result in figure 3
, we find a flux ratio of @xmath7fe /@xmath20al of @xmath32 .
al spectrum derived by integral / spi with 3 years of data .
@xmath20al flux in the galaxy is @xmath33 .
, width=340 ] .different measurements of @xmath0fe/@xmath3al flux ratio [ cols="<,^,<",options="header " , ] fe/@xmath3al from several observations , including our spi result ( also see table 1 , from wang et al .
2007 ) , with upper limits shown at 2@xmath28 for all reported values , and comparison with the recent theoretical estimates ( the upper hatched region from prantzos 2004 ; the straight line taken from timmes et al . 1995
; the lower hatched region , see limongi & chieffi 2006 ) .
our present work finds the line flux ratio to be @xmath34 .
see more details in the text .
, width=453 ] many experiments and efforts were made to measure the @xmath0fe/@xmath3al flux ratio , and we now provide the most significant detection to date ( see table1 and figure 5 ) . in the same time , different theoretical models have predicted the ratio of @xmath0fe/@xmath3al
. timmes et al .
( 1995 ) published the first detailed theoretical prediction . in their paper , they combined a model for @xmath3al and @xmath0fe nucleosynthesis in supernova explosions with a model of chemical evolution , giving a gamma - ray flux ratio @xmath35 . since 2002 , theoreticians have improved various aspects of the stellar - evolution models , including improved stellar wind models and the corresponding mass loss effects on stellar structure and evolution , of mixing effects from rotation , and also updated nuclear cross sections in the nucleosynthesis parts of the models . as a result , predicted flux ratios @xmath0fe/ @xmath3al rather fell into the range @xmath36 ( prantzos 2004 , based on , e.g. rauscher et al .
2002 , limongi & chieffi 2003 ) such high values would be inconsistent with several observational limits and our spi result .
limongi & chieffi ( 2006 ) combined their individual yields , using a standard stellar - mass distribution function , to produce an estimate of the @xmath0fe/ @xmath3al gamma - ray flux ratio expected from massive stars .
their calculations yield a lower prediction for the @xmath0fe/ @xmath3al flux ratio of @xmath37 , which is again consistent with the observational constraints .
now , we have detected both 1173 kev and 1332 kev lines of @xmath7fe in the galaxy ( near 5 @xmath28 significance ) with the 3 years of spi / integral data , which is the best results on detections of @xmath7fe in the galaxy , and confirms its existence .
the average @xmath0fe line flux from the inner galaxy region is @xmath38 . from the same observations and
analysis procedure applied to @xmath3al , we find a flux ratio of @xmath0fe/ @xmath3al of @xmath39 .
though large error bars and uncertainties exist , the original and the latest theoretical prediction of the flux ratio of @xmath0fe/ @xmath3al are consistent with our spi result .
but improvements are needed both in observations and theories . for gamma - ray astronomy , more precise measurements of gamma - ray lines in the galaxy
are required , especially for the @xmath0fe signals , which may require the more spi data and the development of next - generation gamma - ray spectrometers / telescopes .
stellar evolution models have potential for improvements in processes related to the production of @xmath0fe and @xmath3al , e.g. convective layers in the inner stars , wind models for wr and o stars and the possible effects of stellar rotation ( hirschi et al .
the nuclear physics still has serious uncertainties for the productions of @xmath3al and @xmath0fe .
for example , the cross section of @xmath40o is uncertain , which affects the prediction of both @xmath3al and @xmath0fe ; the situation of @xmath0fe is strongly influenced by the cross sections of neutron capture and @xmath14-decay which are purely theoretical : no experimental data exist for the @xmath41 and @xmath42 rates .
therefore , a concerted effort among stellar models , nucleosynthesis theory , and gamma - ray observations is required for a more satisfactory assessment of @xmath0fe synthesis in the galaxy . | gamma - ray line emission from radioactive decay of @xmath0fe provides constraints on nucleosynthesis in massive stars and supernovae .
we detect the @xmath1-ray lines from @xmath0fe decay at 1173 and 1333 kev using three years of data from the spectrometer spi on board _ integral_. the average flux per line is @xmath2 for the inner galaxy region . deriving the galactic @xmath3al gamma - ray line flux with using the same set of observations and analysis method
, we determine the flux ratio of @xmath4 gamma - rays as @xmath5 .
we discuss the implications of these results for the widely - held hypothesis that @xmath0fe is synthesized in core - collapse supernovae , and also for the closely - related question of the precise origin of @xmath3al in massive stars . |
loop formation in macromolecules plays an important role in a number of biochemical processes : stabilisation of globular proteins @xcite , transcriptional regularisation of genes @xcite as well as dna compactification in the nucleus @xcite .
the localisation of chromatin fibres to semi - compact regions known as chromosome territories is maintained among others by the topological constraints introduced by multiple loops in chromatin organisation @xcite .
numerous analytical and numerical studies have been conducted to analyse the cyclisation probability and loop size distributions in long flexible macromolecules @xcite .
the conformational properties of isolated loops ( ring polymers ) @xcite and multiple loops @xcite have been intensively studied as well . interlocking and entanglements are ubiquitous features of flexible polymers of high molecular weight
. in particular , dna can exist in the form of catenated ( bonded ) rings of various complexities @xcite .
the segments of different dna molecules can intercross through the transient breaks introduced by special enzymes ( topoisomerases ) @xcite .
numerous studies @xcite reveal the advantage of linking in stabilisation of peptide and protein oligomers . in this concern , it is worthwhile to study the conformational properties of generalised complex polymer structures , containing @xmath0 loops and @xmath1 linear chains of the same length @xmath4(see fig .
[ structura ] ) .
the properties of so - called `` rosette '' ( fig .
[ structura]a ) and `` ring - linear '' ( fig .
[ structura]b ) structures have been considered recently in numerical simulations in refs .
however , only the shape properties of very simple structures of two bonded rings ( @xmath5 ) and rings with two connected linear branches ( @xmath6 ) with excluded volume effects were analysed @xcite . in particular ,
a quite subtle difference in the conformational properties of two connected polymer rings compared with that of one isolated ring were found . on the other hand , the properties of `` star '' polymers ( fig .
[ structura]c ) have been intensively studied by now @xcite . in general
, the statistics of polymers is known to be characterised by a set of universal properties , which are independent of the details of the microscopic chemical structure @xcite . in particular , the asymptotic number of possible conformations of structure shown in fig .
[ structura]b obeys the scaling law @xcite @xmath7 here , @xmath8 is the universal critical ( flory ) exponent , governing the scaling of the size measure ( e.g. the averaged gyration radius @xmath9 ) of macromolecule according to @xmath10 @xmath11 is the set of so - called vertex exponents ( with @xmath12 ) , @xmath13 is the spatial dimension and @xmath14 is a non - universal fugacity . for an idealised gaussian ( phantom )
macromolecule without any interactions between monomers one finds : @xmath15 , @xmath16 . in the limiting cases ( @xmath17 ) and ( @xmath18 ) , respectively , we obtain the corresponding critical exponents , governing the scaling of the number of possible conformations of @xmath1-branch star and @xmath0-petal rosette polymers : @xmath19 , @xmath20 . therefore , additional topological constraints in rosette polymers lead to a considerable reduction of the number of allowed conformations , as compared with star polymers of the same molecular weight .
to compare the size measures of macromolecules of different topologies but of the same total molecular weight one can consider the universal size ratios . in particular , in the idealised gaussian case , the ratio of the gyration radii of the individual ring and open linear structures reads @xcite @xmath21 whereas comparing the size of @xmath22-branch star and a linear chain of the same total length one has @xcite @xmath23 taking into account the excluded volume effect leads to an increase of this values @xcite .
the overall shape of a typical polymer conformation is of great importance , affecting in particular the mobility and folding dynamics of proteins @xcite .
the shape of dna may be relevant for the accessibility for enzymes depending on the spatial distance between dna - segments @xcite . already in 1934
it was realised @xcite that the shape of a typical flexible polymer coil in a solvent is anisotropic and resembles that of a prolate ellipsoid .
it is convenient to characterise the asymmetry of polymer configurations in terms of rotationally invariant universal quantities @xcite constructed as combinations of the components of the gyration tensor , such as the asphericity @xmath24 .
this quantity takes on a maximum value of unity for a completely stretched , rod - like configuration , and equals zero for the spherical form , thus obeying the inequality @xmath25 . in the gaussian case , for the individual linear and circular polymer chains
one has correspondingly @xcite @xmath26 whereas the asphericity of @xmath22-branch star polymer reads @xcite : @xmath27 note that eq .
( [ adchain ] ) gives the asphericity of a trajectory of diffusive randomly walking particle .
the influence of excluded volume effects on the shape parameters of single linear and ring polymers , as well as star polymers have been analysed so far both analytically @xcite and numerically @xcite . in the present paper ,
we study the universal size and shape characteristics of complex polymer structures ( fig .
[ structura ] ) , applying the path integration method .
a special attention is paid to the analytical study of statistical properties of rosette ( fig .
[ structura]a ) and ring - linear ( fig .
[ structura]b ) structures .
the layout of the paper is as follows . in section 2 ,
we shortly describe the presentation of complex polymer system within the frames of continuous chain model .
our results are given in section 3 .
we end up by giving conclusions in section 4 .
we consider a system consisting of @xmath0 closed polymer loops ( petals ) and @xmath1 linear chains ( branches ) , all bonded together at one branching " point ( see fig .
[ structura]b ) . within the edwards continuous chain model @xcite , each of the individual branches or petals is presented as a path of length @xmath28 , parameterised by @xmath29 , where @xmath30 is varying from @xmath31 to @xmath28 ( @xmath32 ) . for simplicity
we take : @xmath33 .
the weight of the individual @xmath34th path is given by @xmath35 and the partition function of the system can thus be written as @xmath36 here , @xmath37 denotes functional path integrations over @xmath38 trajectories .
the products of @xmath39-functions describe the fact that @xmath0 trajectories are closed and that the starting point of all trajectories is fixed ( the branching " point ) .
note , that ( [ model - con ] ) is normalised in such a way that the partition function of the system consisting of @xmath38 open linear gaussian chains ( star - like structure ) is unity . exploiting the fourier - transform of the @xmath39-functions @xmath40 and rewriting in the exponent @xmath41 we evaluate the expression of partition function ( [ model - con ] ) , giving the asymptotic number of possible conformations of polymer system @xmath42^{f_1+f_2}\ ! ( 2\pi)^{-f_1d}\prod\limits_{j=1}^{f_1}\int { \rm d}\vec{q}_j\ , { \rm e}^{-\frac{q_j^2s}{2}}= ( 2\pi s)^{-df_1/2}. \end{aligned}\ ] ] comparing this relation with eq .
( 1 ) , we recover the estimate for the critical exponent @xmath43 . the average of any observable @xmath44 over an ensemble of conformations is then given by : @xmath45
the size and shape characteristics of a typical polymer conformation can be characterised @xcite in terms of the gyration tensor @xmath46 . within the framework of a continuous polymer model
the components of this tensor can be presented as @xmath47 where @xmath48 is @xmath49th component of @xmath50 ( @xmath51 ) . for the averaged radius of gyration one
has @xmath52 here and below , @xmath53 denotes averaging over an ensemble of path conformations according to ( [ modelav ] ) .
the spread in the eigenvalues @xmath54 of the gyration tensor ( [ mom ] ) describes the distribution of monomers inside the polymer coil and thus measures the asymmetry of the molecule . for a symmetric ( spherical ) configuration all the eigenvalues @xmath55 are equal , whereas for completely stretched rod - like conformation all @xmath55 are zero except one .
let @xmath56 be the mean eigenvalue of the gyration tensor .
then one may characterise the extent of asphericity of a macromolecule by the quantity @xmath57 defined as @xcite : @xmath58 with @xmath59 ( here @xmath60 is the unity matrix ) .
this universal quantity equals zero when @xmath61 , and takes a maximum value of one in the case of a rod - like configuration .
the asphericity ( [ add ] ) can be rewritten in terms of the averaged components of gyration tensor ( [ mom ] ) as follows @xcite : @xmath62 below , we give detailed evaluation of the expressions for the averaged radius of gyration ( [ rg ] ) and asphericity ( [ a ] ) of the model ( [ model - con ] ) within the framework of path integration approach . .
the solid line on a diagram is a schematic presentation of a polymer path of length @xmath4 , and wavy lines denote so - called restriction points @xmath63 and @xmath64 . with @xmath65
we denote the position of the starting `` branching '' point.,width=491 ] the radius of gyration ( [ rg ] ) can be calculated from the identity @xmath66 and evaluating @xmath67 in the path integration approach . in calculation of the contributions into @xmath67
it is convenient to use the diagrammatic presentation as given in fig .
[ rgdiagram ] . according to the general rules of diagram calculations @xcite , each segment between any two restriction points @xmath68 and @xmath69
is oriented and bears a wave vector @xmath70 given by a sum of incoming and outcoming wave vectors injected at restriction points and end points . at these points ,
the flow of wave vectors is conserved . a factor @xmath71
is associated with each segment .
an integration is to be made over all independent segment areas and over wave vectors injected at the end points .
the analytic expressions , corresponding to the diagrams ( 1)-(5 ) in fig .
[ rgdiagram ] then read @xmath72 taking the derivatives with respect to @xmath73 according to ( [ def ] ) in the expressions above and taking into account the combinatorial factors , we find for the radius of gyration @xmath74 .
\label{r12 } \end{aligned}\ ] ] the case @xmath18 corresponds to the rosette structure ( fig .
[ structura]a ) with @xmath75 at @xmath76 one recovers the gyration radius of individual ring polymer .
the case @xmath17 corresponds to star structure ( fig .
[ structura]c ) with @xmath77 at @xmath78 one receives the gyration radius of linear polymer chain . size ratio ( [ ratiors ] ) of rosette and star polymers of the same total molecular weight as function of branching parameter @xmath22
. line is guide to the eyes.,width=283 ] for the size ratio of rosette and star polymers of the same total molecular weight ( corresponding to @xmath79 ) we obtain @xmath80 note that putting @xmath81 in above relation , one restores the size ratio of the ring and open linear chain of the same molecular weight ( [ ratiorl ] ) .
the quantity ( [ ratiors ] ) decreases with increasing the branching parameter @xmath22 and in the limit @xmath82 reaches the asymptotic value @xmath83 ( see fig .
[ rrall ] ) . .
the wavy lines denote restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87 .
with @xmath65 we denote the position of the starting `` branching '' point.,title="fig:",width=491 ] .
the wavy lines denote restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87 .
with @xmath65 we denote the position of the starting `` branching '' point.,title="fig:",width=491 ] .
the wavy lines denote restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87 .
with @xmath65 we denote the position of the starting `` branching '' point.,title="fig:",width=491 ] . the wavy lines denote restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87 .
with @xmath65 we denote the position of the starting `` branching '' point.,title="fig:",width=491 ] the products of the components of the gyration tensor ( [ mom ] ) which appear in ( [ a ] ) can be calculated using the identity @xmath88 with @xmath89 again , we will use the diagrammatic presentation of contributions into @xmath90 ( see fig .
[ adiag ] ) . applying the same rules of diagram calculation , as introduced in the previous subsection and evaluating the corresponding expressions we find @xmath91 here
, @xmath92 denotes contribution of @xmath93th diagram on fig .
[ adiag ] ( see appendix for details ) .
( a ) : asphericity of rosette structure ( [ adros ] ) ( open circles ) and star - like structure ( [ adstar1 ] ) ( filled squares ) of the same total molecular weight as functions of branching parameter @xmath22 in space dimension @xmath94 .
lines are guides to the eyes .
( b ) : asphericity of the system consisting of @xmath0 polymer loops and @xmath1 linear branches ( [ adf1f2 ] ) as function of @xmath0 , @xmath1 in space dimension @xmath94.,title="fig:",width=264 ] ( a ) : asphericity of rosette structure ( [ adros ] ) ( open circles ) and star - like structure ( [ adstar1 ] ) ( filled squares ) of the same total molecular weight as functions of branching parameter @xmath22 in space dimension @xmath94 .
lines are guides to the eyes .
( b ) : asphericity of the system consisting of @xmath0 polymer loops and @xmath1 linear branches ( [ adf1f2 ] ) as function of @xmath0 , @xmath1 in space dimension @xmath94.,title="fig:",width=283 ] the resulting expression for the asphericity thus reads @xmath95\times \left[5d\left(4f_2 ^ 2(3f_2 - 2)^2+f_1 ^ 2(2f_1 + 1)^2+\right.\right.\nonumber\\ & & \left.+4f_1f_2(f_2 ^ 2(72f_2 - 238)-f_1 ^ 2(24f_1 + 113)+196)\right)+2f_1 ^ 2(8f_1 - 7)+\nonumber\\ & & \left.+16f_2 ^ 2(15f_2 - 14)+8f_1f_2(62f_1 + 34f_2 - 41)\right]^{-1}. \label{adf1f2 } \end{aligned}\ ] ] the case
@xmath18 corresponds to the rosette structure ( fig .
[ structura]a ) with @xmath96 at @xmath76 one restores the expression of the asphericity of an individual ring polymer ( [ adring ] ) .
note that for the system of two bonded rings ( @xmath97 ) we restore expression ( [ adring ] ) in accordance with the numerical results of ref .
the case @xmath17 corresponds to star structure ( fig .
[ structura]c ) with @xmath98 again , at @xmath78 , @xmath99 one restores an expression of asphericity of an individual linear polymer ( [ adchain ] ) . to compare the degree of asphericity of rosette and star polymer structures of the same molecular weight ( at @xmath79 ) , we plot the above given quantities as functions of @xmath22 at fixed @xmath94 ( see fig .
[ adall]a ) . at small @xmath22 , the star polymers are more anisotropic and extended in space than rosette structures ,
whereas both @xmath100 and @xmath101 gradually tend to zero with increasing @xmath22 .
really , in the asymptotic limit @xmath102 both the rosette and star structures can be treated as soft colloidal particles with highly symmetrical shape .
the total asphericity of the system consisting of @xmath0 closed polymer loops and @xmath1 linear branches ( eq . ( [ adf1f2 ] ) )
is plotted as function of @xmath0 , @xmath1 at fixed @xmath94 in fig .
[ adall]b .
the value of this quantity is the result of competition of two effects : decreasing the asymmetry with increasing the number of closed loops and increasing the degree of anisotropy with increasing the number of linear branches .
in the present paper , we analysed the conformational properties of polymer systems of complex topologies ( see fig .
[ structura ] ) . whereas the properties of so - called star polymers ( fig .
[ structura]c ) have been intensively studied , much less is known about the details of rosette ( fig .
[ structura]a ) and ring - linear ( fig .
[ structura]b ) structures .
multiple loop formation in polymer macromolecules plays an important role in biochemical processes such as dna compactification @xcite , which makes the rosette - like structures interesting objects to study .
note , that another possible interpretation of structures ( fig .
[ structura]a ) and ( fig .
[ structura]b ) is the following : they can be treated as projections of long flexible polymer in the bulk onto the 2-dimensional plane @xcite . restricting ourselves to the idealised gaussian case ,
when any interactions between monomers are neglected , we develop the continuous chain representation of the complex polymer model , considering each of the individual branches or petals as a path of length @xmath28 , parameterised by @xmath29 , where @xmath30 is varying from @xmath31 to @xmath28 ( @xmath32 ) .
the size and shape characteristics of a typical polymer conformation have been studied on the basis of the gyration tensor @xmath46 with the the components given by ( [ mom ] ) .
working within the framework of path integration method and making use of appropriate diagram technique , we obtained the expressions for the gyration radius @xmath103 and asphericity @xmath104 , measuring the extent of anisotropy of a typical conformation of complex polymer structures , as functions of parameters @xmath0 , @xmath1 .
in particular , our analytical results quantitatively confirm the compactification ( decrease of the effective size ) of multiple loop polymer structures as compared with structures containing linear segments [ eq . ( [ ratiors ] ) ] .
a decrease of the anisotropy of rosette polymers as compared to star - like structures of the same total molecular weight is revealed , as well [ eqs .
( [ adros ] ) , ( [ adstar1 ] ) ] .
this work was supported by the fp7 eu irses projects n269139 `` dynamics and cooperative phenomena in complex physical and biological media '' and n295302 `` statistical physics in diverse realizations '' . financial support from the academy of finland within the fidipro scheme is acknowledged .
here , we evaluate the analytical expressions , corresponding to diagram ( 11 ) on fig .
[ adiag ] .
note , that in the diagram calculations one should take into account the possible permutations of positions of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87 ( see fig .
[ 11 ] ) . with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=151 ] with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=166 ] with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=151 ] with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=151 ] with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=166 ] with various permutations of restriction points @xmath84 , @xmath85 , @xmath86 , @xmath87.,title="fig:",width=151 ]
the resulting expressions read : @xmath105 taking derivatives over the components of @xmath106 , @xmath107 in expressions above according to ( [ ozn ] ) and integrating over @xmath108 , one finally obtains the contributions @xmath109 in ( [ qfinal ] ) : @xmath110
500 h .- x .
zhou , j. phys . chem . * 104 * , 6763 ( 2001 ) .
t. cui , j. ding , and j.z.y .
chen , macromolecules * 39 * , 5540 ( 2006 ) .
toan , d. marenduzzo , p.r .
cook , and c. micheletti , phys .
lett . * 97 * , 178302 ( 2006 ) .
j. shin , a. g. cherstvy , and r. metzler , soft matter doi : 10.1039/c4sm02007c m. bishop and c. saltiel , j. chem . phys . * 85 * , 6728 ( 1986 ) .
diehl and e. eisenriegler , j. phys . a : math .
* 22 * , l87 ( 1989 ) .
o. jagodzinski , e. eisenriegler , and k. kremer , j. phys .
i france * 2 * , 2243 ( 1992 ) .
obukhov , m. rubinstein , and t. duke , phys .
lett . * 73 * , 1263 ( 1994 ) .
m. muller , j.p .
wittmer , and m.e .
cates , phys.rev.e * 61 * , 4078 ( 2000 ) .
deutsch , phys .
e * 59 * , r2539 ( 1999 ) .
shimamura and t. doguchi , phys .
e * 64 * , 020801 ( 2001 ) .
k. alim and e. frey , phys .
99 * , 198102 ( 2007 ) .
m. bohr and d.w .
heermann , j. chem .
phys . * 132 * , 044904 ( 2010 ) .
p. calabrese , a. pelissetto , and e. vicari , j. chem .
phys . * 116 * , 8191 ( 2002 ) .
t. sakaue , phys .
lett . * 106 * , 167802 ( 2011 ) .
y. jung , c. jeon , j. kim , h. jeong , s. jun , and b .- y .
ha , soft matter * 8 * , 2095 ( 2012 ) .
a. rosa , e. orlandini , l. tubiana , and c. micheletti , macromolecules * 44 * , 8668 ( 2011 ) . j. shin , a. g. cherstvy , and r. metzler , new j. phys .
* 16 * , 053047 ( 2014 ) . | multiple loop formation in polymer macromolecules is an important feature of the chromatin organisation and dna compactification in the nuclei .
we analyse the size and shape characteristics of complex polymer structures , containing in general @xmath0 loops ( petals ) and @xmath1 linear chains ( branches ) . within the frames of continuous model of gaussian macromolecule
, we apply the path integration method and obtain the estimates for gyration radius @xmath2 and asphericity @xmath3 of typical conformation as functions of parameters @xmath0 , @xmath1 . in particular ,
our results qualitatively reveal the extent of anisotropy of star - like topologies as compared to the rosette structures of the same total molecular weight . |
a great number of experimental and theoretical papers devoted to the study of cold and ultracold atomic ensembles are related to the unique physical properties inherent in these systems and their broad possible practical applications .
improvements in techniques for cooling atomic gases in atomic traps @xcite make their use very promising in metrology , frequency standards , and quantum information problems @xcite-@xcite .
most of the methods for diagnostics of the states of these ensembles are based on their interaction with electromagnetic radiation .
more importantly , the largest number of the many applications envisioned for ultracold atom ensembles have at their foundation such interactions .
the interaction of radiation with cold atomic clouds reveals a number of specific features .
first , atoms cooled down to low temperatures have very high resonance scattering cross sections , and therefore multiple scattering should be taken into account even in quite rarefied clouds .
second , atomic clouds in traps have nonequilibrium distributions over different hyperfine sublevels and inhomogeneous spatial distributions .
third , because of low mean velocities of atoms , interference effects can be observed during multiple scattering despite the absence of any ordering in the atomic cloud .
depending on the concentration @xmath0 of atoms in a cloud , the cloud can interact with light either in the weak or strong localization regimes .
the weak localization regime assumes that the medium is rarefied so that both the mean interatomic distances and the photon mean free path greatly exceed the light wavelength @xmath1 in this case , individual atoms scatter radiation independently and multiple scattering can be treated as a number of successive scatterings by free atoms . by now
, this regime has been studied in detail , both experimentally and theoretically @xcite-@xcite .
the interaction of light with dense atomic clouds has been studied far less thoroughly than with rarefied clouds .
this is mainly caused by considerable technical difficulties associated with producing such ensembles under laboratory conditions .
some problems also arise in the consistent quantum mechanical description of their interaction with the electromagnetic field . under conditions
when the mean distance between atoms becomes comparable with the light wavelength , it is necessary to consider the resonance interatomic dipole - dipole interaction . because of this interaction ,
individual atoms can not be treated as independent sources of secondary waves .
conditions are produced for the so - called dependent or cooperative scattering , and we are in fact dealing with the interaction of light with a giant quasimolecule combining all atoms in the ensemble . despite these difficulties , the strong localization regime has been attracting considerable attention recently because there is good reason to believe that interesting quantum processes are possible in this regime , such as the anderson ( strong ) localization of light @xcite-@xcite and lasing in a disordered medium @xcite-@xcite , which can have practical applications .
the aim of recent studies in this field was to obtain these effects experimentally and describe them theoretically .
the present paper is devoted to the theoretical study of light scattering by dense atomic clouds under conditions of strong localization , i.e. , to developing a theory of dependent scattering .
obviously the problem of a dense ensemble belongs to the field of macroscopic electrodynamics .
it is known that such problems can be solved by using either macroscopic or microscopic approaches . in the former case ,
description is performed by using such averaged characteristics as the field strength and atomic polarization , while the microscopic approach is based on the concept of the discrete structure of matter consisting of individual atoms or molecules .
the macroscopic approach , which we employed to solve the scattering problem , was described in detail in @xcite . in the present paper
, we develop the consistent quantum - mechanical description of the strong localization regime .
we consider the scattering of quasiresonance radiation by an ultracold disordered ensemble of atoms under conditions when the mean number of atoms @xmath2 in the volume can become comparable with unity .
we assume that probe radiation is weak and restrict ourselves to the case when the field subsystem contains no more than one photon .
atoms are assumed immobile .
a key problem in describing scattering in this case is analysis of the evolution of an atomic ensemble after its excitation by initial radiation and investigation of possible decay channels for this radiation accompanied by the creation of secondary scattered photons . in this connection ,
the given problem is closely related to that of collective spontaneous decay in a dense polyatomic system .
moreover , this problem can be reduced to that of spontaneous decay by considering the evolution of an atomic subsystem in which only one atom located far from the cloud is excited in the initial state . by calculating the dynamics of such a system ,
we automatically describe how a photon emitted by this individual atom is scattered in the cloud .
the consideration of the problem in this formulation allows us to use methods developed for solving problems of superradiance of a singly excited ensemble .
how dipole - dipole interaction influences collective spontaneous decay in a finite system was first studied for the case of two atoms in @xcite and @xcite . a consistent quantum - mechanical approach developed in these papers
was later generalized in a number of papers to polyatomic ensembles ( see @xcite-@xcite and references therein ) .
however , almost all papers on polyatomic systems used only a two - level atom model .
this prevents the correct consideration of the polarization properties of light and , as a consequence , adequate description of the resonance dipole - dipole interaction at small distances .
in addition , a real disordered atomic ensemble was replaced in calculations , as a rule , by a model of an averaged continuous medium . in this case , the role of random inhomogeneities was neglected . in our opinion , these simplifications considerably restrict the application of these theories for describing experiments with ultracold gases ; therefore , we do not use them in our paper .
another important difference of our approach is that it can be used to analyze the scattering of coherent pulsed radiation , which is also important for interpretation of experiments .
our paper contains four sections , conclusions , and an appendix . in section 2
, we solve the nonstationary schrodinger equation to calculate the amplitudes of a number of states of a combined system including atoms and electromagnetic radiation .
in section 3 , the calculated amplitudes are used to determine the intensity , spectrum , and polarization properties of scattered radiation .
section 4 contains the description and results of numerical calculations . in the conclusions ,
the main results are briefly formulated and the possibilities for further investigations in this field are discussed .
the state of a closed system consisting of atoms and electromagnetic field can be described by the wavefunction @xmath3 .
this function in the schrodinger representation satisfies the equation @xmath4
@xmath5 here @xmath6 , @xmath7 and @xmath8 are hamiltonians of free atoms not interacting with each other and the free field , and their interaction operator , respectively . in the dipole approximation used here , we have @xmath9 where @xmath10 is the electric displacement vector operator .
@xmath11 @xmath12 are the operators of positive and negative frequency components ; @xmath13 is the photon annihilation operator in a mode with wave vector @xmath14 and polarization @xmath15 @xmath8 is the quantization volume ; is the dipole moment operator of the atom @xmath16 , @xmath17 are polarization unit vectors satisfying the transversability condition ; @xmath18 we will solve equation ( [ 1 ] ) by using the general approach developed in heitler s papers and considered in detail in @xcite .
we will seek the wavefunction @xmath19 in the form of the expansion in a set of eigenstates @xmath20 of the operator @xmath21 : @xmath22 here , the subscript @xmath23 defines the state of all atoms and the field .
equation ( [ 1 ] ) gives the system of coupled equations @xmath24 for the state amplitudes @xmath25 .
here @xmath26 is the energy of the lowest state of the system of free atoms and field . system ( [ 7 ] ) contains an infinite number of equations and can not be solved in the general case . in this paper , as mentioned above , we will study the evolution of only weakly excited states of the combined system .
in addition , we will calculate all the observables of interest , namely , scattering cross sections , decay constants , levels shifts , etc .
, correct to the second order in the fine structure constant .
the higher - order radiative corrections will not be calculated . in this case , the number of @xmath27 states taken into account in calculations can be restricted . as shown in @xcite , in this approximation
it is sufficient to consider only the states @xmath28 here , @xmath29 is the transition frequency of the atom @xmath16 and the ground state energy of the atoms is set equal to zero .
note that , by choosing the states considered in the calculations , we go beyond the rotating wave approximation because we included states with energy @xmath30 which contain , along with a photon , two excited atoms @xmath16 and @xmath31 the consideration of nonresonance states proves necessary to describe correctly the dipole - dipole interaction of atoms separated by small distances less than the light wavelength . indeed , the dipole - dipole interaction can be described as the result of exchange by resonance photons .
such an exchange can occur in two ways , either via resonant or nonresonant intermediate states . in the first case
, an excited atom emits a photon , which then is absorbed by an unexcited atom . in the second case , at first
, an atom is excited and emits a photon and then the initially excited atom undergoes a transition to the ground state and absorbs a photon .
according to the energy - time uncertainty relation , during the propagation time of @xmath32 a photon between two atoms , virtual states can exist with an energy that differs from the resonance ( initial ) energy by the value @xmath33 when the interatomic distance r is large , it turns out that the second way is strongly nonresonant and therefore is usually neglected . if the interatomic distance is smaller or close to the reciprocal wave vector @xmath34 , then @xmath35 and @xmath36 .
therefore , we should take into account the second energy exchange channel as well , i.e. the channel involving the intermediate state @xmath37 .
the applicability of the rotating wave approximation in problems of the spontaneous decay of excited states of polyatomic systems is considered in more detail , for example , in @xcite -@xcite .
below , we will consider , along with the spontaneous decay and scattering of one - photon radiation , the interaction of an atomic ensemble with weak coherent radiation .
it is known that weak coherent radiation can be quite accurately represented as the superposition of one - photon and vacuum states .
therefore , we will add the state @xmath38 to group ( [ 8 ] ) .
further simplifications will be performed for a system consisting of @xmath39 atoms .
we assume that @xmath40 atoms forming a cloud are identical and can be described by a four - level system with one ground and three degenerate excited zeeman sublevels ( the @xmath41 transition ) .
one of the four - level atoms , which we will call a source , can differ in the transition frequency and the excited - state lifetime .
the general expressions obtained below will be also valid in a particular case when all @xmath39 atoms are identical and form one cloud .
consider the dynamics of a combined atom + field system , assuming that all atoms at the initial moment , except the source , are in the ground state , while the field is in the vacuum state .
the source atom is in the superposition of the ground and excited states .
this means that at the initial moment two states are excited , one of which is @xmath42 , and the second state is denoted by @xmath43 .
if @xmath44 is the initial amplitude of the @xmath43 state and @xmath45 is that of the @xmath42 state @xmath46 , then system of equations ( [ 7 ] ) for amplitudes should be supplemented by the initial conditions @xmath47 the solution obtained in this case describes the state of the atomic - field system for positive times @xmath48 .
obviously the amplitude of the state @xmath49 ( [ 8.2 ] ) in the approximations used does not change during the evolution of the system .
this is explained by the fact that transitions to this state from other states taken into account in the calculations are impossible .
the transition from @xmath42 to any of the states ( [ 8 ] ) is also impossible ( virtual transitions to a state with one photon and one excited atom , which cause lamb shifts , are taken into account by the renormalization of the transition frequency ; see the appendix ) , and therefore @xmath50 . to determine the time dependence of other amplitudes , we extend , by the heitler method , the solution of system ( [ 7 ] ) to the negative time semiaxis , assuming that @xmath51 the function @xmath52 has a jump for @xmath53 we will take this into account by adding the inhomogeneous term into the equation for @xmath52 , @xmath54 where @xmath55 is the kronecker delta .
the system obtained in this way is solved by passing to the fourier transform @xmath56 in this case , @xmath57 we will separate from ( [ 11 ] ) the equation for @xmath58 , taking into account that @xmath59 @xmath60 we introduce new variables @xmath61 @xmath62 by introducing a singular function @xmath63 , it is possible to prove in the general case @xcite that the obtained solution will satisfy the initial conditions .
functions @xmath61 satisfy the system of equations @xmath64 by solving this system , we obtain the amplitude @xmath65 @xmath66 the subscript of the excited atom is omitted for brevity in relations ( [ 15])-([14.3 ] ) , and it is also taken into account that the relation @xmath67 is fulfilled for singular functions @xmath68 . by using the first and second equations of system ( [ 15 ] ) , we obtain the closed system of equations @xmath69 u_{e^{\prime } } ( \omega ) \varsigma \left ( \hbar \omega -e_{e^{\prime } } \right ) , \;e\neq o. \label{16}\end{gathered}\ ] ] for the states with one excited atom in the cloud . note that this system ,
unlike ( [ 9 ] ) , contains a finite number of equations .
this number is determined by the number of atoms in the cloud and the number of excited states considered in the calculations .
the sums @xmath70 and @xmath71 can be calculated for an infinitely broadband field reservoir by assuming that the quantization volume is infinite ( see the appendix ) . by using expressions presented in the appendix , we rewrite system ( [ 16 ] ) in a more compact form : @xmath72 u_{e^{\prime } } ( \omega ) \varsigma \left ( \hbar \omega -e_{e^{\prime } } \right ) = \sigma _ { eo}(\omega ) \label{19}\ ] ] here , the vector @xmath73 in the right - hand side determines the excitation of different atoms in the cloud by the source radiation , while the matrix @xmath74 describes radiation transferred between atoms in the cloud . system ( [ 19 ] ) can be reduced to an integral equation by using the continuous medium approximation .
this significantly simplifies the solution of the problem for a two - level atom system @xcite-@xcite .
moreover , in this case , even an analytic solution is possible for spatially homogeneous spherical clouds .
this solution neglects , however , the important properties of real physical systems , and therefore we will solve the linear system ( [ 19 ] ) numerically .
the numerical solution is restricted by the possibilities of available computers which allow us to consider clouds containing a few thousands of atoms .
the number of atoms in quasi static traps , where the required concentrations can be achieved , can be two orders of magnitude greater ; however , by using this approach , we can correctly describe all polarization effects taking into account random inhomogeneities of the medium .
both these factors , as follows from our previous calculations @xcite , are quite important .
the formal solution of system ( [ 19 ] ) can be written in the form @xmath75 where @xmath76 ^{-1}. \label{21.1}\ ] ] by using ( [ 21 ] ) and ( [ 21.1 ] , we obtain the amplitudes of all states of interest for us : @xmath77 relations ( [ 22.3])-([26 ] ) completely describe the studied system under the above assumptions . in the next section , we will use them to obtain explicit expressions for scattered radiation parameters observed in experiments .
however , an important preliminary remark should be made .
expressions ( [ 22.3])-([26 ] ) are valid for any distance between an atom excited from the ground state and the cloud .
they describe , in particular , the modification of the spontaneous decay of the atom near the cloud or even inside it .
our main goal is to describe the scattering of light by a dense atomic cloud .
therefore , below we restrict ourselves to the case when a source atom is located far from the cloud . in this case
, we can neglect the reverse influence of the cloud on the spontaneous decay of this isolated atom taking into account only the first term in ( [ 26 ] ) by calculating @xmath78 . by including the lamb shift in the definition of frequency @xmath79
, we obtain @xmath80 here , @xmath81 is a decay constant of the atom that was initially excited
. this constant can be arbitrary .
along with simplification of the quantity @xmath78 , the expression for the vector @xmath73 describing excitation of different atoms in the cloud by the source is also simplified in this case . because the distance from the cloud to the emitting atom is large , retardation effects are important and
the polar approximation can not be used .
however , we can use the rotating wave approximation and retain only one term decreasing the least rapidly with increasing distance from the source to the cloud . in this case
@xmath82 \left ( \frac{\omega } { c}\right ) ^{2}\exp \left ( i\frac{\omega r_{eo}}{c}\right ) ; \label{27.1}\ ] ] here , @xmath83 is the vector drawn from the source o@xmath43 to the atom excited to the @xmath84 state ( hereafter for brevity , we will call it the @xmath84 atom )
. we will transform expression ( [ 27.1 ] ) taking into account that @xmath85 greatly exceeds the size of the system .
we introduce a coordinate system with a center inside the cloud .
let @xmath86 an @xmath87 be the radius vectors of the cloud atoms and the source , respectively .
then , @xmath88 and @xmath89 \exp \left ( ikr_{0}-i\mathbf{kr}_{e}\right ) ; \label{27.2}\ ] ] here , @xmath90 .
we can see from ( [ 27.2 ] ) that the source radiation in the atomic ensemble region can be treated quite accurately as a plane wave .
knowing the amplitudes of the states , we can calculate the probability of finding the system in a specified state at any instant of time . in experiments , as a rule , the parameters of electromagnetic radiation are measured at large distances from a scattering cloud .
these parameters are determined by the quantity @xmath91 .
the calculation of this quantity is the main purpose of our paper because different states @xmath92 correspond to different frequencies , polarizations , and wave vectors of the emitted photon .
relation ( [ 23 ] ) allows us to find @xmath93 at an arbitrary instant of time .
below we consider the two most important particular cases allowing a significant simplification in the analytic form .
the first case is the calculation of the probability of different field states after a long time interval when spontaneous decay has ended . in our case , these are times exceeding the excited - state decay times of the source atom and atomic ensemble .
the second case is the calculation of the transition probability per unit time when the system state still only weakly differs from the initial state .
these times are considerably shorter than the decay time ( but , of course , they greatly exceed reciprocal transition frequencies ) .
we will solve this problem taking @xmath94 @xmath95 here @xmath96 is the frequency of the emitted photon . the first term in expression ( [ 28 ] ) describes radiation incident directly on a photodetector from the initially excited atom .
the angular and spectral distributions of radiation from the cloud are determined by the square of the modulus of the second term in parentheses in ( [ 28 ] ) .
we will assume here that the photodetector is equipped with a set of apertures eliminating the contribution of the initially excited atom to the photocurrent .
then , the probability of finding a photon in the final state in a given mode will be @xmath97 the dominator in this expression is related to the efficiency of excitation of the cloud by the source radiation .
we calculated it using expression ( [ 27 ] ) . for small @xmath98 ,
only a narrow spectral group of the cloud states is efficiently excited . to determine the total excitation spectrum of the cloud in this case , it is necessary to repeat experiments many times by changing the source radiation frequency @xmath79 .
we can obtain the cloud radiation spectrum in one experiment by using a radiation source emitting a very short pulse , so that @xmath99 is large and all the characteristic collective excitation frequencies lie within the pulse spectral width @xmath99 . in another limiting case ,
when @xmath99 is much smaller than the characteristic rates of transient processes in the cloud , a quasi - stationary process will take place during which the quasi - stationary distribution of excitation over the cloud will be established .
this distribution will then change slowly ( with the decay rate @xmath100 of the excited state of the source atom ) . in this case
, we can calculate the transition probability per unit time @xmath101 . to do this
, it is necessary to perform the passage to the limit @xmath102 before passing to large times : @xmath103 we will calculate @xmath104 using the expansion of the product @xmath105 into simple fractions : @xmath106 \varsigma _ { \sigma -\gamma _ { 0}}\left ( \omega _ { o}-\omega _ { g}\right ) .$ ] here @xmath107 and @xmath98 are constants . by having them tend to zero , we obtain the corresponding generalized function ( further we will assume for definiteness that @xmath107 > @xmath98 , which does not affect the final result ) . taking into account that @xmath108 , we obtain after simple transformations @xmath109 as earlier , the first term describes the contribution of radiation emitted directly by the initially excited atom
. expressions ( [ 29])-([32 ] ) can be used to describe experiments in which the occupation of different modes of the field is detected , for example , when modes with a narrow group of wave vector directions are selected with the help of a convergent lens . to describe experiments without such a selection , for example , with point photodetectors ,
it is necessary to know the mean parameters of the field at a certain spatial point at a specified instant of time .
these mean parameters can be found from the known wave function @xmath110 for example , the mean projection @xmath111 of the vector @xmath112 at the point @xmath113 at the instant @xmath114 is @xmath115 here , the matrix @xmath116 describes the propagation of radiation from an atom excited in the state @xmath84 to the observation point , @xmath117 the explicit expression for this matrix can be obtained as it was done in the appendix upon calculation of infinite sums entering equation ( [ 16 ] ) . in the rotating wave approximation
, we obtain @xmath118 \left ( \frac{\omega } { c}\right ) ^{2}\exp \left ( i\frac{\omega \left\vert \mathbf{r - r}_{e}\right\vert } { c}\right ) . \label{36}\ ] ] here , @xmath86 is the radius vector of the atom excited to state @xmath84 and @xmath113 is the radius vector of the observation point in the chosen coordinate system .
the polarization components of the intensity correlation function can be calculated similarly : @xmath119 expressions ( [ 34 ] ) and ( [ 37 ] ) are considerably simplified in the quasistatic case . by retaining only the contribution of light scattered by the cloud and passing to the limit @xmath102
@xmath120 knowing the transition probability per unit time , we can determine the differential cross section of light scattering by an atomic cloud .
we calculate this cross section by assuming that the distance from the source to the cloud is so large that a wave coming from the source to the atomic ensemble is a plane wave with the wave vector @xmath121 with good accuracy [ see ( [ 27.2 ] ) ] . this wave is polarized .
the direction of its polarization unit vector @xmath122 is determined by the orientation of the vector @xmath14 and a zeeman sublevel to which the source atoms was excited .
we will find the photon flux density @xmath123 of this wave in the atomic ensemble region from relation ( [ 37 ] ) by retaining only the first term in it : @xmath124 here , the summation is over any two basis polarizations in a coordinate system determined by vector @xmath14 . to obtain the explicit expression for the scattering cross section , we retain in expression ( [ 32 ] ) for the probability @xmath104
only the contribution caused by the cloud , sum it over a narrow group of finite states of the field , and divide by the flux density @xmath123 of photons incident on the atomic ensemble from the source . taking into account that the radiation frequency does not change upon scattering in this case , we obtain the expression @xmath125@xmath126for the cross section of photon scattering from the mode @xmath127 , to an arbitrary mode @xmath125 .
note that we obtained the relation for the total cross section earlier @xcite by using the t - matrix formalism and showed , in particular , that the @xmath128 matrix is the projection of the resolvent of the system under study on the states with one excited atom .
the relations obtained in this section will be further used to analyze the spectral , angular , and polarization characteristics of light scattering by ultracold atomic clouds .
the main quantity necessary for the description of scattering is a resolvent @xmath129 because it determines amplitudes ( [ 22.3])-([26 ] ) of all quantum states involved in the problem and all experimental characteristics of the process under study [ see , for example , ( [ 34 ] ) , ( [ 37 ] ) and ( [ 40 ] ) ) . in our paper
, this resolvent was found numerically .
depending on convenience , we calculated either the reciprocal matrix ( [ 21.1 ] ) or directly found the numerical solution to system ( [ 19 ] ) .
the resolvent value considerably depends on the spatial configuration of the ensemble under study .
the analysis of random inhomogeneities of a medium and averaging over these inhomogeneities were performed by the monte carlo method of statistical tests .
the spatial distribution of atoms in the cloud was specified by means of pseudorandom - number generators , and then a particular observable of interest for us was calculated .
this observable was calculated repeatedly for different random configurations , and the results were averaged .
note that in most of the papers studying collective effects in dense media , the averaging procedure is performed differently by replacing the real medium by the averaged medium with a continuous density distribution , and then the observables are calculated for this averaged medium .
as follows from our analysis @xcite , in this case , some correlation effects affecting observables are described incorrectly .
the expressions obtained in the previous section describe light scattering by ensembles of ultracold atoms under various experimental conditions , in particular , for spatially inhomogeneous clouds of different shapes and arbitrary densities .
we are mainly interested in dense clouds with high atomic concentrations in which dependent scattering effects are important .
however , we will not restrict ourselves to this case only .
some results obtained in the paper concern optically dense but rarefied atomic ensembles .
the analysis of the weak localization regime as a limiting case is important , first , for verifying and substantiating the method proposed and , second , it allows us to refine some results that were earlier obtained in this regime by other methods .
below , we analyze the two main characteristics of scattering measured in experiments , namely , the angular ( spatial ) distribution of scattered light and the spectral dependence of the scattering cross section .
we restrict ourselves to the results obtained for the quasistatic case . the scattering of light by an optically dense medium in the presence of localization phenomena is considerably determined by the effects of interference and radiation trapping .
therefore , we verify first of all that the approach proposed in this paper correctly describes these effects . the basic equations ( [ 19 ] ) do not contain derivatives with respect to spatial variables . in this connection it is interesting to study how these equations describe radiation transfer in an optically dense medium , in particular , how they reproduce the bouguer - lambert law . according to this law , the intensity of the coherent component of radiation transmitted through a medium layer is attenuated exponentially .
the exponent @xmath130 ( the optical thickness of the layer ) for a homogeneous medium is determined by the concentration @xmath0 of atoms , the layer thickness @xmath131 , and the scattering cross section @xmath107 by an atom . for the j = 0 @xmath132 j=1 transition in a rarefied medium
, we have @xmath133 .
to analyze the radiation transfer process , we consider the scattering of resonance radiation by a model cylindrical atomic cloud with radius @xmath134 and length @xmath131 .
the wave vector of the incident wave is assumed parallel to the cylinder axis .
we will calculate the spatial intensity distribution @xmath135 for the coherent component [ which is determined by the square of the amplitude modulus ( [ 34 ] ) ) ] in a plane perpendicular to this axis .
the distance @xmath136 from the rear of the cylinder to the observation plane is chosen equal to a few @xmath137 to fit several fresnel zones within the end for any point of this plane .
the results of calculations for a rarefied medium with @xmath138 @xmath139 , and different @xmath131 are shown in [ f1 ] .
the obtained intensity distribution corresponds to the fresnel pattern for diffraction by a semitransparent screen . outside the shadow region , oscillations are observed typical of diffraction from the screen edge .
the intensity attenuation in the geometrical shadow region is caused by the destructive interference of the source radiation and secondary radiation emitted by the atomic ensemble .
this interference formally appears in our approach when two terms in ( [ 34 ] ) are taken into account .
note that the diffraction pattern in our approach is obtained directly upon summation of contributions from a great number of microscopic scatterers forming the medium .
the transmission coefficient @xmath140 was calculated by finding the mean light intensity in the geometrical shadow region .
these mean intensities for two thicknesses are shown in [ f1 ] by the dashed lines .
the dependence of the sample transmission on its length @xmath131 showed that the bouguer - lambert law is reproduced quite accurately in our calculations . in this case , the optical thickness increases linearly with increasing @xmath131 .
as for the dependence of optical thickness @xmath141 on concentration @xmath0 , it increases linearly only up to concentration @xmath142 ( [ f3 ] ) . the deviation from linearity at higher
@xmath0 is caused by collective effects , namely , by the interatomic resonance dipole - dipole interaction .
this interaction leads to shifts of atomic levels , so that the scattering cross section @xmath107 for resonance radiation decreases .
this decrease partially compensates the increase in concentration : the higher the atomic density , the better the compensation .
this effect , in particular , can partially explain the difficulty in experimental observation of strong localization in atomic gases .
we now consider light scattering in the case of fraunhofer diffraction .
we analyze the angular distribution of the cloud radiation at a large distance from it .
the calculation is performed using expression ( [ 40 ] ) for the differential scattering cross section .
we assume that the light frequency coincides with the transition frequency in a free atom , while the probe radiation is circularly polarized .
we will analyze , in particular , coherent backscattering ( cbs ) ; therefore , to avoid problems with sharp boundaries , the atomic cloud will be assumed inhomogeneous with a gaussian spatial density distribution .
figure [ f4 ] shows the angular distribution of scattered light for two polarization channels , with light helicity preserved ( h @xmath143 h ) or changing to the opposite ( h @xmath144 h ) .
the calculations were performed for a cloud with a gaussian radius @xmath145 .
the density at the cloud center was @xmath146 this corresponds approximately to the intermediate value between dense and rarefied ensembles . when the helicity was preserved , we observed typical fraunhofer diffraction .
the principal diffraction maximum caused by coherent forward scattering dominates .
several higher - order peaks are well distinguished , although because of a smooth decrease in the concentration at the edges of the gaussian cloud , these peaks are not as distinct as for diffraction from objects with sharp boundaries .
along with diffraction peaks , a comparatively weak background is observed , which is caused by multiple scattering in the cloud .
the background intensity weakly depends on angles , except for a narrow region near backward scattering , where a local maximum is observed .
this is the cbs cone .
radiation with orthogonal helicity ( the dashed curve in fig .
[ f4 ] ) appears due to incoherent multiple scattering ( incoherent light scattering by different atoms ) .
the intensity of this radiation in the case under study is low .
it increases monotonically for scattering angles above @xmath147 .
the cbs effect for this polarization channel and the transition under study is known to be a very weak ( see , for example , @xcite ) ) and is almost unnoticeable at this scale , which is correctly reproduced by our calculations .
thus , our microscopic calculation reproduces the interference effect upon multiple scattering in a disordered medium for both main scattering channels .
however , unlike methods used earlier for calculations of this effect and taking into account only a finite number of scattering orders , our approach takes into account all processes of multiple scattering of photons in a medium and also allows us to determine the cone shape at arbitrary atomic concentrations .
the cbs properties are shown in more detail in fig .
this figure demonstrates the concentration dependence of the cbs cone for the h @xmath143 h channel .
the visible shape of the cone changes comparatively weakly in this range of @xmath0 .
figure 4 demonstrates the qualitative change in the cbs effect for dense clouds .
it is known that all weak - localization theories proposed so far predict the maximum value of the amplification factor ( the ratio of the cone maximum to the value of the quasi - isotropic pedestal ) equal to two .
this is explained by the fact that the cone amplitude is caused by the interference of the waves scattered by the same chain of atoms but propagating along this chain in opposite directions .
these waves in rarefied media interfere only in pairs .
we see that our calculation for rarefied media ( the lower curve in fig .
[ f5 ] ) quite accurately gives an amplification factor of 2 .
the amplification of backscattering in denser media exceeds 2 . in our opinion , this is explained by two main reasons .
first , in dense media , along with the conventional mechanism resulting in weak localization , the interference of beams scattered by different but spatially close chains makes a certain contribution .
the second reason is the decrease in the fraction of light scattered outside the cone at comparatively large angles , i.e. , the decrease in the background intensity with respect to which the amplification effect is measured ( see details in @xcite ) .
we consider the spectral dependences of the differential and total scattering cross sections as another example of the application of the theory developed above .
the total cross section can be obtained by integrating the differential cross section over the entire spherical angle or , according to the optical theorem , by calculating the amplitude of light scattered at the zero angle . in this paper , we used both these methods , the optical theorem being fulfilled independently of the cloud density .
figure [ f6 ] shows the total scattering cross section spectra for a spherically homogeneous , on average , cloud with radius @xmath148 at different densities . at low concentrations ,
when the interatomic interaction is insignificant , we see a monotonic spectral dependence , which is typical of scattering by an isolated atom .
the spectrum is somewhat broadened compared to the monatomic scattering spectrum because this medium is not optically thin even at these concen trations .
as the cloud density is increased , the shape of the spectrum begins to be distorted . at the concentration @xmath149 ,
the spectral dependence is substantially nonmonotonic , and several distinct local maxima appear .
this behavior of the cross section can be explained either in the context of macroscopic electrodynamics or by using the microscopic approach .
macroscopic electrodynamics describes scattering as the result of multibeam interference ( or diffraction ) from a sphere with the specified permittivity .
such scattering in the case of spatially homogeneous spheres with arbitrary permittivities was considered in detail in the debye- mie theory ( see , for example , @xcite ) .
this theory showed , in particular , that the maximum scattering cross section for transparent spheres with the refractive index @xmath150 was observed for @xmath151 . in this case , the scattering cross section was approximately four times greater than the sphere cross section area @xmath152 .
as the radius is decreased compared to @xmath134 , the ratio of the scattering cross section @xmath107 to @xmath152 tends to zero .
as @xmath134 is increased , the ration @xmath107/@xmath153 oscillates and tends to 2 for @xmath154 the scattering cross section behaves similarly when the sphere material has a lower optical density than the environment .
however , there exists some quantitative difference .
the maximum scattering cross section is lower and is observed for smaller values of @xmath134 .
if the medium is not transparent , i.e. , the refractive index has a noticeable imaginary part ( the absorption coefficient ) , the diffraction type changes .
the oscillations mentioned above are considerably smoothed , although the maximum scattering cross section can exceed @xmath155 this behavior of the scattering cross section agrees qualitatively with results shown in fig .
[ f6 ] . in the case of an atomic cloud , the permittivity is complex .
we can introduce the real refractive index and absorption coefficient ( although note that in the case under study , a wave is not really absorbed by an atomic ensemble because the escape of photons from the incident wave - mode is accompanied by the appearance of photons in other modes ) . the microscopic approach developed in our paper
gives accurate values for the permittivity of a homogeneous atomic medium with the specified density .
however , the discussion of this issue is beyond the scope of this paper .
note that the permittivity value is determined by studying the spatial distribution of the atomic polarization taking into account that for detunings @xmath156 exceeding the atomic resonance width , the refractive index decreases with increasing @xmath156 considerably slower than the absorption coefficient . for the specified radius of the cloud @xmath157 , the condition @xmath158 our calculation of the permittivity shows that the position of the left maximum qualitatively agrees with the predictions of the mie theory .
the intensity of this maximum is smaller than the maximum possible value for transparent media , which is related to a small but finite absorption coefficient in this frequency region . as the atomic concentration in the cloud
is increased , the maximum shifts to the region of greater detuning magnitudes and its amplitude increases .
this is explained by the fact that a comparatively small value of @xmath150 in a denser medium is achieved for a large magnitude of detunings @xmath156 .
the relative influence of absorption for large detunings also decreases .
the maximum observed for positive detunings also can be explained by the debye - mie theory .
the refractive index of the atomic medium in this frequency region is smaller than unity , and therefore the right maximum intensity is lower and it is located at smaller magnitudes of @xmath156 than the left one .
the absorption coefficient for small detunings is large compared to @xmath159 ( the difference of the real refractive index from unity ) , and therefore there exists another peak related to the absorption coefficient maximum .
this peak is most distinctly observed at large concentrations when the spectral positions of all three cross section maxima are well distinguished .
figure [ f6 ] shows the concentration dependence of the shape of the total cross section spectrum for a fixed cloud size .
figure [ f7 ] illustrates the dependence of the cross section on cloud size for the specified concentration .
the calculation was performed for @xmath160 when the interatomic interaction is considerable .
we can see that , as the cloud size increases , the maxima caused by refraction shift to the region of large detunings , according to the condition @xmath161 the position of the central maximum at the specified concentration does not change .
the frequency dependence of the total cross section can be explained by using the microscopic approach . to do this , it is necessary to consider the distributions of collective atomic states over energies and lifetimes , more exactly , the distribution of the poles of the resolvent @xmath129 over the values of their real and imaginary parts ) is not normal , and its real and imaginary parts can not be simultaneously reduced to the diagonal form ) .
our analysis has shown that the maximum of the spectral dependence corresponded to the maximum density of such states taking into account their finite width .
note here that the study of poles gives the position of maxima more accurately than the debye - mie method .
this is explained by the fact that the microscopic approach correctly takes into account boundary effects . with the parameters under study ,
atoms in boundary regions are located in different conditions than atoms in central parts because they interact with a smaller number of neighbors , whereas the permittivity in the debye - mie theory is assumed the same over the entire volume of the sphere ] . in light scattering experiments ,
the differential cross section is most often studied rather than the total cross section . as a rule
, the intensity of light scattered through a specified angle is measured @xcite and its dependence on the probe light frequency is determined .
figure [ f8 ] shows this dependence calculated by us for a spherical cloud with radius @xmath157 and concentration @xmath162 on the ordinate , the total light intensity ( without selection over polarizations ) is plotted .
different curves correspond to scattering through different angles ? .
note that the behavior of these curves is substantially different .
for some scattering angles , the spectral dependence is monotonic , whereas for others , the nonmonotonic behavior is observed .
a similar difference will be manifested in the fluorescence of the cloud excited by a light pulse . in the case of a short pulse
whose spectrum overlaps the entire excitation spectrum of the atomic ensemble , fluorescence in some directions will decay monotonically , while in other directions , beats will be observed .
the spectral behavior is different not only for scattering in different directions but also for different polarization components scattered in the same direction .
this is illustrated in fig .
[ f9 ] , which shows the intensities of two orthogonal polarization components as functions of the incident light frequency .
the calculation was performed for the same conditions as in fig .
the scattering angle is @xmath163 .
the polarization of incident light , as in all other calculations , was circular .
the results demonstrate a considerable dispersion of polarization : the polarization of light scattered in the given direction can considerably depend on the incident light frequency .
we have developed a consistent quantum - mechanical theory of cooperative scattering of weak electromagnetic radiation by an optically dense macroscopic atomic ensemble . the microscopic approach proposed in the paper can be applied both to rarefied clouds , when the average number of atoms in the volume is small , and to the clouds in which this parameter exceeds unity .
the differential and total dependent scattering cross sections have been calculated for quasi stationary radiation using this approach .
the angular , spectral , and polarization properties of scattered light were determined .
the dependence of these characteristics on cloud size and atomic concentration was studied . by summing the contributions from many atomic scatterers
, we obtained , along with the known fresnel and fraunhofer diffraction phenomena , the contribution from multipole incoherent scattering and calculated the interference effect of coherent backscattering .
the parameters of the backscattering cone were determined .
note that , unlike approximate methods used earlier , in which backscattering was calculated taking into account a finite number of scattering events , our approach allows us to calculate these parameters exactly .
this effect was calculated for the first time for dense clouds where the mean free path is comparable with the light wavelength .
it was found that the complex nature of interference effects , when aside from the traditional mechanism resulting in a weak localization , it is necessary to consider the interference of beams scattered by closely located chains , leads to an amplification factor exceeding 2 .
the dependence of the scattering cross section on the incident light frequency has demonstrated a strong modification of the spectra with increasing atomic concentration .
it was shown that these spectra for dense media have a prominent nonmonotonic nature and exhibit several maxima .
it was also found that collective effects lead to a considerable difference in the spectral dependences of scattering in different directions .
the investigation of the polarization properties of scattered light revealed a considerable dispersion of polarization , which strongly depends on the incident light frequency .
our study concerns the case when the motion of atoms can be neglected .
it is obvious , however , that this motion can play an important role in some experiments .
therefore , we believe that it is interesting to consider further the influence of atomic motion on collective effects in dense atomic systems and the interaction of light with such systems . in addition , it is important for practical purposes to generalize the microscopic approach to the case when an additional control field is present . in our opinion
, the use of high density atomic systems to obtain electromagnetically induced transparency and to `` stop light '' can be quite efficient for solving problems in quantum informatics .
this work supported by the russian foundation for basic research ( project nos . 08 - 02 - 91355 and 10 - 02 - 00103 ) and the united states national science foundation ( grant no .
nsf - phy-0654226 ) .
infinite sums @xmath164 and @xmath165 appearing in eqs . ( [ 16 ] ) can be calculated for an infinitely broadband field thermostat assuming that the quantization volume is infinite .
we first consider the first sum , assuming that in state @xmath166 the atom @xmath16 is excited , while in @xmath167 atom @xmath141 is excited .
in dipole approximation ( [ 3 ] ) , we have @xmath168 here , @xmath169 .
expression ( [ a.4 ] ) was derived by calculating the sum over polarizations using completeness relation ( [ 5 ] ) , replacing the sum over @xmath14 by an integral , and using one of the representations of the singular function @xmath170 . formally ,
some of the integrals in ( [ a.4 ] ) diverge . however , this divergence is related to the behavior of integrands for large @xmath171 , which is caused by the use of the dipole approximation . by introducing , as usual , a slowly decaying exponential factor@xmath172 with a small exponent @xmath173
, we can provide the divergence of integrals . by calculating integrals and tending the parameter @xmath174 to zero in finite expressions ,
we obtain @xmath175 ; \label{a.8}\]]where @xmath176 here , @xmath177 and @xmath178 are integral cosine and sine , respectively .
similarly , we can calculate the two - atom contribution to the second sum caused by nonresonance intermediate states @xmath179 ; \label{a.101}\ ] ] here @xmath29 is the transition frequency of atoms forming the cloud .
it is easy to see that this contribution for small @xmath180 is comparable with resonance term ( [ a.8 ] ) .
expressions ( [ a.8 ] ) and ( [ a.101 ] ) can be considerably simplified in the so called polar approximation , when their value for frequency @xmath181 is replaced by their value for frequency@xmath29 of the atomic resonance .
this approximation was studied in detail in @xcite , where it was shown that it can be applied in systems where retardation effects are insignificant .
the value of @xmath134 for atomic clouds in quasistatic traps , which are studied in this paper , does not exceed hundreds of micrometers .
the shortest evolution times of the atomic system to be compared with the delay times are the lifetimes of short - lived superradiance states . for systems with size @xmath134 , these times are on the order of @xmath182 where @xmath183 is the natural excited state lifetime of a free atom . under such conditions and for typical parameters of alkali metal atoms , retardation effects can be neglected even for clouds with density @xmath184 in this case , using the polar approximation , we obtain @xmath185 @xmath186 .\end{aligned}\ ] ] note that the latter expression can be also rewritten in terms of spherical hankel functions and spherical harmonics @xcite .
the sum @xmath190 gives only the lamb shift of the ground state because it describes the spontaneous excitation of an atom located close to the excited atom , which is accompanied by spontaneous decay with absorption of a virtual photon .
we neglect the lamb shifts of the ground and excited states , assuming that they are included in the transition frequency @xmath191 g. k. campbell , a. d. ludlow , s. blatt , j. w. thomsen , m .
j. martin , m. h. de miranda , t. zelevinsky , m. m.boyd , j. ye , s. a. diddams , t. p. heavner , t. e. parkerand s. r. jefferts , metrologia 45 , 539 ( 2008 ) . | based on the developed quantum microscopic theory , the interaction of weak electromagnetic radiation with dense ultracold atomic clouds is described in detail .
the differential and total cooperative scattering cross sections are calculated for monochromatic radiation as particular examples of application of the general theory .
the angular , spectral , and polarization properties of scattered light are determined .
the dependence of these quantities on the sample size and concentration of atoms is studied and the influence of collective effects is analyzed . |
double diffusive convection ( ddc ) is the convection flow where fluid density depends on two scalar fields .
one of the most relevant application is oceanic mixing , in which the two scalars are temperature and salinity .
temperature diffuses about 100 times faster than salinity , and this huge difference in the molecular diffusivities allows for very rich dynamics in oceanic ddc flow @xcite .
for instance , an instability can occur even when the fluid is overall stably stratified @xcite .
a comprehensive review of the field can be found in the recent book by @xcite .
one of the interesting phenomena in ddc is the salt finger structure , i.e. narrow elongated vertical flows which were observed in many experiments ( e.g. by * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . for a ddc flow bounded by two reservoirs with fixed values for the two scalars ,
@xcite observed either a single layer of salt fingers , or an alternating stack of salt - finger and convective layers which resembles the thermohaline staircase observed in the ocean ( e.g. * ? ? ?
* ; * ? ? ?
three - dimensional ( 3d ) direct numerical simulations ( dns ) in a fully periodic domain produced layered structures which were very similar to the thermohaline staircase @xcite .
numerical studies also revealed that internal gravity waves can spontaneously develop through collective instability and modulate the salt - finger field @xcite . in recent ddc experiments by using electrodeposition cells , which are very close to the rayleigh - bnard ( rb ) setup , tilgner and coworkers discovered that salt fingers also occur when the fluid is unstably stratified @xcite .
this finding is very surprising and unexpected since one would assume that rb convection dominates in an unstably stratified system .
however , linear instability analysis revealed that , in this new regime , salt - finger solutions can indeed be obtained @xcite . our previous numerical results with the same flow setup well agreed with those experiments @xcite .
numerous theories and models have been developed in the past to understand the scalar fluxes produced by the salt - finger structures , e.g. see @xcite and the references therein .
recently , new models were proposed for the fully periodic ddc flow and tested against numerical results , such as the mean - field theory of @xcite and the equilibrium model of @xcite . for the finger layer bounded by two solid boundaries
, we discovered that the grossmann - lohse ( gl ) theory , which was originally developed for rb flow and showed great success @xcite , can be directly applied to ddc flow and accurately predicts the salinity flux for both our numerical results @xcite and the experimental results @xcite .
although we demonstrated the success of the gl theory in accurately predicting the salinity transfer rate , the behaviour of the flow velocity was not fully investigated in the previous work @xcite , partially due to the limited control parameters simulated therein . here , with the help of more systematic simulations which cover a much wider parameter range
, we will establish a complete description of the scaling laws for three important global responses , i.e. the salinity and heat transfer rates and the flow velocity .
moreover , we will also investigate the flow structures , such as the salt fingers and the boundary layers
. the characteristic length scales of those structures and their dependences on the flow parameters will be discussed .
the paper is organised as follows .
we first briefly describe the numerical methods and the parameter space explored ( section 2 ) . in section 3 we develop the scaling laws for the system responses . in section 4 the flow structures and further scaling relations are presented .
the applicability of the new proposed scaling strategy to other fluid system is discussed in section 5 .
the paper ends with conclusions and an outlook ( section 6 ) .
we consider incompressible flow with two different scalar fields and employ the oberbeck - boussinesq approximation , which assumes that the fluid density depends linearly on both scalars , i.e. @xmath12 $ ] . here
@xmath13 is the density with some reference value @xmath14 , and @xmath15 and @xmath16 are the temperature and salinity relative to some reference values , respectively .
@xmath17 is the positive expansion coefficient .
hereafter the subscript @xmath18 or @xmath19 denotes the quantity associated to the scalar @xmath20 .
the governing equations read [ eq : ddc ] @xmath21 in which @xmath22 with @xmath23 are three velocity components , @xmath24 is the kinematic pressure , @xmath25 is the kinematic viscosity , @xmath26 is the gravitational acceleration , and @xmath27 are diffusivities of the respective scalar components .
the continuity equation is @xmath28 .
the fluid is vertically bounded by two parallel plates separated by a distance @xmath29 . at the two horizontal plates
no - slip boundary conditions are applied to velocity , and both scalars have fixed values . in the horizontal directions we choose a domain size much larger then the size of the salt fingers , which allows us to apply periodic boundary conditions .
the flow is driven by the scalar differences between the two plates .
we define the temperature and salinity differences as @xmath30 the subscripts `` bot '' and `` top '' denote the values at the bottom and top plates , respectively . in this study
we always set @xmath31 and @xmath32 , which means that the flow is driven by the salinity difference and stabilized by the temperature difference .
the flow control parameters are the prandtl and rayleigh numbers @xmath33 with @xmath34 , or alternatively one can also use the lewis number and the density ratio @xmath35 note that @xmath2 measures the ratio between the buoyancy force induced by the two scalars .
@xmath36 corresponds to a rayleigh - bnard flow purely driven by salinity difference . as @xmath2 increases ,
the magnitude of the stabilizing buoyancy force induced by temperature difference becomes stronger when compared to the destabilising force of the salinity difference .
for otherwise fixed parameters , one can view @xmath2 also as dimensionless ( stabilising ) temperature difference between top and bottom plates . in all our simulations
the prandtl numbers are fixed at @xmath37 and @xmath38 , i.e. the typical values of seawater .
the lewis number is then @xmath39 .
the density ratio is in the range of @xmath40 , which falls into the salt - finger regime .
initially the fluid is at rest .
the temperature has a vertically linear distribution , and salinity is uniform and equal to @xmath41 , respectively .
these initial fields are similar to those in the experiments of @xcite and @xcite .
small random perturbations are superposed to both scalar fields to accelerate the development of the flow .
the parameters explored in the present work are shown in figure [ fig : param ] and more details can be found in the appendix .
the numerical method is reported in @xcite .
the typical flow structures are similar to those shown in our previous study ( e.g. see figure 1 of @xcite ) .
the bulk region is dominated by salt fingers , and thin boundary layers develop for the velocity and salinity fields adjacent to two plates .
@xmath42 plane and coloured by @xmath43 , and ( b ) on the @xmath44@xmath2 plane and coloured by @xmath45 .
the dashed lines indicate @xmath46 , i.e. the fluid layer is neutrally buoyant .
the fluid is overall unstable for the parameters below the lines and stable for those above the lines , respectively . ]
the three key responses of the system are the salinity and heat fluxes and the flow velocity , which are measured in non - dimensional form by the nusselt and reynolds numbers as @xmath47 here @xmath48 stands for the average over the entire domain and time .
@xmath49 denotes the rms value of the velocity magnitude . in figure
[ fig : rasnus ] we plot @xmath0 versus @xmath1 , both as a log - log plot and in compensated form .
we compare the numerical results to the gl theory with the updated coefficients @xcite .
note that there is no new free parameter here and the curve is fully determined by the theory . in our previous study
@xcite we have shown , for a smaller parameter range , that @xmath0 mainly depends on @xmath1 and it shows only a minor dependence on @xmath2 .
the variation of @xmath0 with @xmath1 is also close to the gl prediction . indeed , for the current dataset with a much larger parameter range
, @xmath0 still follows a single trend which is close to the gl prediction .
the small discrepancy between the numerical results and the gl theory was also found for standard rb flow when one applies the gl theory at high prandtl number , e.g. see figure 7 of @xcite . on @xmath1 in
( a ) a log - log plot and ( b ) compensated form .
symbols are coloured according to the logarithm of @xmath2 .
the dashed lines are the predictions of the gl theory . ] unlike on the salinity flux , however , the density ratio @xmath2 has strong influence on the flow velocity and on the heat transfer .
to reveal the effects of @xmath2 on the system responses , we focus on two sets of cases with @xmath1 fixed at @xmath50 and @xmath51 , respectively .
each set consists of six cases with @xmath2 ranging from @xmath52 to @xmath53 . in figure
[ fig : dreffs ] we plot the dependences of @xmath0 , @xmath3 , and @xmath4 on @xmath2 for these two sets . clearly , when @xmath1 is fixed , @xmath0 is almost constant for the range of @xmath2 considered here .
in contrast , for fixed @xmath1 both the reynolds number @xmath3 and the convective heat transfer @xmath54 decrease as @xmath2 increases from @xmath52 to @xmath53 .
the different behaviors of @xmath0 and @xmath3 versus @xmath2 suggest that for fixed salinity difference , increasing the relative strength of the stabilizing temperature difference suppresses the flow motions .
meanwhile , the flow patterns adjust themselves such that the salinity transfer only changes slightly . and ( b ) @xmath3 on @xmath2 for two fixed @xmath55 ( green triangles ) and @xmath51 ( red diamonds )
. in ( a ) the two horizontal dashed lines mark @xmath56 and @xmath57 , which are the averaged values over different @xmath2 s for each @xmath1 , respectively . in ( b )
the dashed line indicates the scaling relation @xmath58 with @xmath59 . in ( c )
the dashed line indicates the scaling relation @xmath60 with @xmath61 . ] in figure [ fig : dreffs](b ) we see the reynolds number @xmath3 as function of the density ratio @xmath2 .
note that even for the larger salinity rayleigh number @xmath62 and for small @xmath2 it is never larger than @xmath63 , far away from any turbulent behavior , but reflecting the laminarity of the flow .
also the thermal flux is very small , with @xmath64 always being smaller than @xmath65 , close to the value for pure molecular diffusion , i.e. @xmath66 . from figures [ fig : dreffs](b , c )
one also observes that the dependences of @xmath3 and @xmath54 on @xmath2 follow certain power - law scalings .
thus a possible strategy for constructing scaling laws for those two quantities would be first extracting the scaling laws of the @xmath2-dependences of @xmath3 and @xmath54 for fixed @xmath1 , and then examining their behaviours for different @xmath1 .
this strategy is applied to @xmath3 and @xmath54 as follows .
let s assume that @xmath3 scales as @xmath58 for fixed @xmath1 , where @xmath10 is a positive exponent .
the minus sign is introduced since @xmath3 decreases as @xmath2 increases .
the value of @xmath10 can be determined by the data points in figure [ fig : dreffs](b ) .
for the set of @xmath55 , the exponent given by a linear regression is @xmath67 . and for the set of @xmath62 it is @xmath7 .
the two values are very close to each other . to obtain a single value for @xmath10 , we first shift the data points vertically in figure [ fig : dreffs](b ) such that the two points at @xmath68 and @xmath69 collapse and then conduct a linear regression with all @xmath70 data points .
the final value is @xmath7 . in figure
[ fig : rasrey](a ) we plot the original values of @xmath3 compensated by @xmath71 .
the data points are scattered since @xmath3 is strongly affected by @xmath2 .
however , we can define a rescaled reynolds number @xmath72 with @xmath59 . in figure
[ fig : rasrey](b ) we plot @xmath73 against @xmath1 for the whole dataset .
all the data points collapse and the dependence @xmath74 is quite close to the gl prediction . versus @xmath1 for all data points .
( b ) dependence of the rescaled reynolds number @xmath72 with @xmath59 versus @xmath1 for all data points .
symbols are coloured according to the logarithm of @xmath2 .
the dashed lines are the predictions of the gl theory . ]
a similar analysis can be done for @xmath4 .
we assume that the convective heat flux scales as @xmath75 and calculate the value of @xmath76 from figure [ fig : dreffs](c ) by following the same procedure for @xmath10 .
the final value is @xmath8 . and a rescaled convective heat nusselt number can be defined as @xmath77 with @xmath61 . in figure [ fig : rasnut ] we plot the original value of @xmath64 and the rescaled ones @xmath78 against @xmath1 in a compensated form . again
, the original values are scattered , but the rescaled quantity @xmath78 follows a single trend versus @xmath1 .
two things should be noted from figure [ fig : rasnut ] .
first , the quantities are compensated by @xmath71 as for @xmath3 , instead of @xmath79 as for @xmath0 .
second , the dependences of @xmath78 on @xmath1 shown in figure [ fig : rasnut](b ) is very similar to that of @xmath73 shown in figure [ fig : rasrey](b ) .
a reasonable argument is that in the current flow setup , the temperature field is stabilizing rather than driving the flow .
thus the convective heat flux is generated as the temperature anomaly is `` passively '' carried by the flow motions which are sustained by the salinity difference
. therefore @xmath0 behaves similarly to that of a rb flow , while @xmath78 exhibits similar scaling as @xmath73 . versus @xmath1 .
( b ) dependence of the rescaled heat nusselt number @xmath77 with @xmath61 versus @xmath1 .
symbols are coloured according to the logarithm of @xmath2 .
note that the range of the vertical axis in ( b ) is half of that in ( a ) . ]
although the non - dimensional thermal flux is much smaller than the non - dimensional salinity flux , the density - anomaly flux associated with the temperature field may not be negligible when compared to that associated with the salinity field , because of the huge difference between the two molecular diffusivities .
this can be seen from the density flux ratio which defined as @xmath80 in figure [ fig : rasdfr](a ) we plot the variation of @xmath81 versus @xmath1 .
the density flux ratio @xmath81 has the value between @xmath52 and @xmath82 .
as the density ratio @xmath2 increases ( symbol color changes from dark blue to light yellow ) , @xmath81 also increases . in previous discussion
we showed that , for fixed @xmath1 , @xmath64 scales as @xmath60 and @xmath0 is nearly constant .
then by definition ( [ eq : rflx ] ) one expects @xmath83 for fixed @xmath1 . in figure [ fig : rasdfr](b ) we plot the rescaled density flux ratio @xmath84 , and indeed all the data points collapse . versus @xmath1 .
( b ) dependence of the rescaled density flux ratio @xmath85 with @xmath61 versus @xmath1 .
symbols are coloured according to the logarithm of @xmath2 . ]
we now turn to the flow structures . we will focus on the horizontal width of salt fingers and the thicknesses of velocity and salinity boundary layers .
then we will show the horizontal zonal flows which were observed at very high rayleigh numbers . in all our simulations
salt fingers develop in the bulk of the flow domain .
a typical flow field can be seen in figure [ fig : fg3d ] , which shows the three - dimensional volume rendering of the salinity field at @xmath86 and @xmath62 ( equivalently @xmath46 ) .
the finger layer is bounded by two thin boundary layers adjacent to both plates .
the individual salt fingers can be distinguished . and @xmath62 , or equivalently @xmath46 .
the color and opacity are both determined by salinity . ]
the thickness of the salinity boundary layer @xmath87 is defined as the distance from the plate to the location of the first peak of @xmath88 , which is the standard deviation of salinity over every horizontal plane . in figure
[ fig : nussbl ] we plot the dependence of @xmath89 on @xmath0 , and find that it perfectly scales as @xmath90 .
the scaling @xmath91 can be understood by following the argument for rb flows at high prandtl numbers @xcite .
the salinity nusselt number is related to the salinity dissipation rate by the exact relation @xmath92 ^ 2 \right\rangle_v = \kappa_s\ , ( \delta_s)^2\ , l^{-2}\ , { \mbox{\textit{nu}}}_s.\ ] ] for the salinity field with high prandtl number , the dissipation is dominated by the contribution from the boundary layers .
the volume integral of @xmath93 in the two boundary layers may be approximated as @xmath94 ^ 2 \right\rangle_{\rm bl } \sim \kappa_s \left(\frac{\delta_s}{2\lambda_s}\right)^2 \frac{2\lambda_s}{l } = \kappa_s \frac{\delta^2_s}{2\lambda_s l}.\ ] ] combining the above two equations , one readily obtains @xmath91 , which is exactly the case as shown in figure [ fig : nussbl ] . versus the salinity nusselt number @xmath0 .
the dashed line has the slope @xmath95 .
symbols are coloured according to the logarithm of @xmath2.,scaledwidth=70.0% ] the horizontal length scale of the salt fingers can be determined from the flow field on the horizontal mid - plane @xmath96 . in figures [ fig : fgreg](a , b ) we show the contours of the vertical velocity @xmath97 and the salinity @xmath16 on the horizontal plane @xmath96 for the same flow field as shown in figure [ fig : fg3d ] .
clearly , most fingers have almost circular shape in the horizontal sections .
some sheet - like links can be found , connecting different fingers , but usually they are not pronounced . and corresponding finger detection .
the control parameters are @xmath86 and @xmath62 , or equivalently @xmath46 . for this flow
the standard deviations for the vertical velocity @xmath97 is @xmath98 and for the salinity @xmath16 is @xmath99 , respectively .
( a , b ) contours of @xmath97 and @xmath16 , ( c ) the cross correlation coefficient between the variances of @xmath97 and @xmath16 , i.e. @xmath100 , and ( d ) the finger regions with @xmath101 and @xmath102 , as shown by the white patches.,scaledwidth=80.0% ] large salinity anomaly is usually transported by salt fingers , which can be clearly seen from the cross correlation between the variance of the vertically velocity @xmath103 and that of the salinity @xmath104 . the prime denotes the deviation of the quantity from its mean value over the plane . in figure
[ fig : fgreg](c ) we plot the cross correlation coefficient function @xmath105 with @xmath106 denoting the standard deviation of the quantity .
@xmath107 has large negative values at isolated regions , because the ascending ( descending ) fingers carry negative ( positive ) salinity anomaly . the global correlation coefficient , which is the average of @xmath107 , is @xmath108 .
thus @xmath103 and @xmath16 are almost perfectly negatively correlated , namely , salt fingers dominate the salinity transfer in the bulk region . from the flow fields
as shown in figures [ fig : fgreg](a , b ) , we can identify the salt - finger regions , which are the regions with @xmath101 and @xmath102 . in figure
[ fig : fgreg](d ) we present the finger regions detected by this method .
each white patch corresponds to an individual finger region .
such flow fields were stored constantly during each simulation , and the horizontal length scale of the salt finger @xmath109 is extracted from those data .
first , the total number of the finger regions @xmath110 is counted and averaged over time , and the average spatial area occupied by one finger is calculated as @xmath111 with @xmath112 being the total horizontal area of the domain .
the finger width @xmath109 is then calculated from @xmath113 by assuming that the finger has circular shape , i.e. @xmath114 . thus @xmath109 is actually the average diameter of the slender convection cells associated with salt fingers . following the argument in @xcite , we can relate the scaling behaviour of the finger width to that of the reynolds number .
since the salinity is mainly transferred by fingers , the salinity nusselt number can be approximated as @xmath115 in which @xmath116 is the characteristic velocity of the fingers and @xmath117 is the mean salinity anomaly within the fingers , respectively .
the salinity anomaly is carried from the boundary layers and the side diffusion is weak in the bulk due to the large prandtl number . then within each finger convection cell of width @xmath109 , the salinity core has similar width as the thickness of salinity boundary layer , i.e. @xmath87 , as shown by the sketch in figure [ fig : sketch ] .
the mean salinity anomaly inside a finger can then be calculated , for circular fingers , as @xcite @xmath118 then combining the above two equations , and considering that @xmath119 as proven in the previous section , one obtains @xmath120 this scaling relation can be clearly seen in figure [ fig : fgwnus ] , in which we plot @xmath121 versus @xmath0 for all cases . and
symbols are coloured according to the logarithm of @xmath2 .
the dashed line has a slope of @xmath122.,scaledwidth=60.0% ] one important consequence of ( [ eq : sfa ] ) is , to generate similar salinity flux @xmath0 , one must have @xmath123 . for
fixed @xmath1 the salinity nusselt number @xmath0 is almost constant .
then the rescaled finger width @xmath124 should also be constant . in figure
[ fig : drfgw ] we plot the rescaled finger width @xmath124 for two sets of cases with fixed @xmath55 and @xmath51 , respectively . indeed ,
when @xmath1 is fixed , @xmath124 is constant for different @xmath2 , which confirms the scaling @xmath125 .
furthermore , we have shown in the previous section that @xmath126 with @xmath59 for fixed @xmath1 , e.g. see figure [ fig : dreffs](b ) .
this implies that @xmath127 for fixed @xmath1 . in figure
[ fig : rasfgw ] we show the dependences of both the original finger width @xmath128 and the rescaled value @xmath129 on @xmath1 .
the non - dimensional finger width collapses and follows a single power - law scaling when rescaled by @xmath9 .
a liner regression gives an exponent of @xmath130 , see figure [ fig : rasfgw](b ) .
, rescaled as @xmath124 , on @xmath2 for two fixed @xmath55 and @xmath51.,scaledwidth=60.0% ] and ( b ) the rescaled finger width @xmath131 with @xmath59 on @xmath1 .
symbols are coloured according to the logarithm of @xmath2 . in
( b ) the data points collapse after the rescaling , and the dashed line represents the linear regression with a slope of @xmath132 .
] for the current flow , there is no large scale circulation in the bulk .
the vertically moving fingers induce converging or diverging flows in the horizontal directions when they move away from or towards the plates .
those horizontal motions form the velocity boundary layer .
the thickness of velocity boundary layer @xmath133 can be defined as the distance from the boundary to the location of the first peak of @xmath134 .
here @xmath134 is the standard deviation of one horizontal velocity component over the plane at the height @xmath135 .
since the velocity boundary layer is driven by the vertical motions of the fingers , it is reasonable to expect that @xmath133 scales as the finger width @xmath128 , which is confirmed by figure [ fig : fwgbls](a ) . for comparison
, we also plot the salinity boundary layer thickness @xmath89 versus @xmath128 in figure [ fig : fwgbls](a ) .
the data points are scattered and no single dependence can be found between @xmath89 and @xmath128 . and ( b ) salinity @xmath89 versus finger width @xmath128 .
symbols are coloured according to the logarithm of @xmath2 . in ( a )
the dashed line has a slope unity . ]
moreover , figure [ fig : rasubl ] displays the @xmath1-dependence of @xmath136 , which is affected by both @xmath1 and @xmath2 , as shown in figure [ fig : rasubl](a ) .
similar to the finger width @xmath128 , if we rescale @xmath136 by @xmath9 with @xmath59 , all data points collapse onto a single curve , see figure [ fig : rasubl](b ) .
the exponent calculated by a linear regression is @xmath137 , which is very close to the exponent for the rescaled finger width as shown in figure [ fig : rasfgw](b ) . and
( b ) the rescaled velocity boundary layer thickness @xmath138 with @xmath59 on @xmath1 .
symbols are coloured according to the logarithm of @xmath2 . in ( b )
the data points collapse after the rescaling , and the dashed line represents the linear regression with a slope of @xmath139 . ]
it is known that large scale oscillating waves can be excited spontaneously in a fully developed salt - finger field and modulate salt fingers , which is attributed to the collective instability @xcite .
these structures , or the internal gravity waves , were observed in simulations of unbounded domains , such as those reported in @xcite and @xcite . in the fully periodic domain ,
the phase planes of gravity waves are not necessarily perpendicular to the gravity direction , and these inclined waves cause strong fluctuations in @xmath0 as they travel in space @xcite . in the current study , the inclined waves did _ not _ appear , probably due the vertical constrain of the two plates .
instead , we observed a stack of horizontal zonal flows in alternating directions for the two cases with highest rayleigh numbers , i.e. @xmath140 and @xmath141 at a density ratio @xmath142 . in figure
[ fig : h32 ] we show such horizontal zonal flows at @xmath143 , which can be clearly seen from the contours of @xmath144 and @xmath145 on the @xmath146 mid - plane and the mean profiles in the bulk region , see panels a and b. and @xmath142 .
the mean profiles ( left ) and instantaneous contours on the vertical mid plane ( right ) are plotted for the two horizontal velocity components @xmath144 ( panel a ) and @xmath145 ( panel b ) , the vertical velocity component @xmath97 ( panel c ) and the salinity @xmath16 ( panel d ) .
the mean profiles in ( a - c ) are normalized by the corresponding volume - averaged rms values @xmath147 , @xmath148 , and @xmath149 , respectively.,title="fig:",scaledwidth=32.0% ] and @xmath142 .
the mean profiles ( left ) and instantaneous contours on the vertical mid plane ( right ) are plotted for the two horizontal velocity components @xmath144 ( panel a ) and @xmath145 ( panel b ) , the vertical velocity component @xmath97 ( panel c ) and the salinity @xmath16 ( panel d ) .
the mean profiles in ( a - c ) are normalized by the corresponding volume - averaged rms values @xmath147 , @xmath148 , and @xmath149 , respectively.,title="fig:",scaledwidth=32.0% ] + and @xmath142 . the mean profiles ( left ) and instantaneous contours on the vertical mid plane ( right ) are plotted for the two horizontal velocity components @xmath144 ( panel a ) and @xmath145 ( panel b ) , the vertical velocity component @xmath97 ( panel c ) and the salinity @xmath16 ( panel d ) .
the mean profiles in ( a - c ) are normalized by the corresponding volume - averaged rms values @xmath147 , @xmath148 , and @xmath149 , respectively.,title="fig:",scaledwidth=32.0% ] and @xmath142 .
the mean profiles ( left ) and instantaneous contours on the vertical mid plane ( right ) are plotted for the two horizontal velocity components @xmath144 ( panel a ) and @xmath145 ( panel b ) , the vertical velocity component @xmath97 ( panel c ) and the salinity @xmath16 ( panel d ) .
the mean profiles in ( a - c ) are normalized by the corresponding volume - averaged rms values @xmath147 , @xmath148 , and @xmath149 , respectively.,title="fig:",scaledwidth=32.0% ] as @xmath1 becomes large enough , i.e. for tall samples , individual salt fingers do not extend from one plate to the other . to estimate the vertical length of the salt fingers , the autocorrelation function of the vertical velocity @xmath97 is calculated in the vertical direction .
we use the flow fields on the @xmath150 mid - plane as shown in figure [ fig : h32 ] .
only the data in the range @xmath151 are used to exclude the boundary layer regions near two plates .
the autocorrelation functions are defined as @xmath152 where the overline denotes the average over @xmath153 , @xmath135 , and @xmath154 on the @xmath150 mid - plane .
the autocorrelation @xmath155 is computed for three cases with @xmath142 and @xmath156 , @xmath157 , and @xmath141 , respectively .
the curves are plotted in figure [ fig : uzcor](a ) . for the case with @xmath156
, @xmath155 never decreases to zero , and the fingers can still extend the whole height of the domain .
however , for the two cases with larger @xmath1 , @xmath155 does decrease to zero
. thus @xmath97 decorrelates at a height smaller than the domain height , implying that the average height of the salt fingers are smaller than the sample height @xmath29 . from the curves we determine the first zero point of @xmath155 at @xmath158 for @xmath140 , and @xmath159 for @xmath143 , respectively .
the autocorrelation function @xmath155 is also plotted versus the vertical separation in the viscous scale @xmath160 , see figure [ fig : uzcor](b ) .
interestingly , the two curves for @xmath140 and @xmath141 collapse with each other , but they are different from the one for @xmath156 .
note that the three cases have the same density ratio @xmath142 .
this implies that when the distance between two plates is large enough and fingers can not extend from one plate to the other , the vertical length of the salt fingers may be set by the density ratio . versus the vertical separations in ( a ) the global scale @xmath161 and ( b ) the viscous scale @xmath160 , respectively .
for all three cases the density ratio is fixed to @xmath142 .
solid line : @xmath156 , dashed line : @xmath140 , and dash - dotted line : @xmath143 . ]
we end this subsection by comparing our findings at high rayleigh numbers to those reported in literatures . in @xcite
, the gravity - wave phase was followed by a spontaneous appearance of a layered phase and staircase - like scalar profiles .
such transition was not found in our simulation .
the reason may be that @xmath162 here is much larger than that in @xcite . in our case
the layered phase may occur at higher @xmath1 .
more simulations are needed to clarify whether and when the layered phase can be realised in the current configuration . in the experiments of salt - sugar system wavy fingers were also observed both in the single finger layer occupying the entire tank or even in the finger layers bounded by two convection layers as in the staircase state @xcite .
however , they did not appear in the experiments of @xcite and @xcite .
the exact conditions for the appearance of zonal flows and wavy fingers are not clear at this stage . the present lewis number @xmath162 is smaller than those in tilgner s experiments ( @xmath163 ) but larger than those in @xcite ( @xmath164 ) , while the @xmath1 for obtaining wavy fingers in our simulations is comparable to the highest @xmath1 in tilgner s experiments and much smaller than the @xmath1 across the whole tank in @xcite .
for the scaling laws proposed in the previous two sections , one has to fit two exponents from the numerical results to collapse the data , i.e. @xmath10 for the reynolds number @xmath3 and @xmath76 for the convective heat flux @xmath64 .
the scaling laws for the flow structures , such as the finger width and the thickness of velocity boundary layers , can be deduced from the scaling of @xmath3 . in our simulations
only one combination of the prandtl numbers @xmath165 is considered .
one may expect that the values of these exponents depend on the prandtl numbers and should vary for different fluid system . in order to test the scaling laws for the different fluid systems ,
we perform the previous analysis on the experimental data of @xcite which have @xmath166 .
our previous study revealed that the gl prediction for the salinity transfer agrees with the experimental results @xcite .
here we further show that a complete description of the experimental results can be obtained by following the method described in the previous sections . in the experiments
the heat flux was not measured , and we focus on the reynolds number and the finger width .
recent studies revealed that when the density ratio is very small , the salt fingers will be replaced by large scale convection rolls and flow becomes very similar to the rb case @xcite . to confine ourselves to the salt - finger regime and be consistent with the current parameter range
, we discard the data points with @xmath167 in the dataset of @xcite . to obtain the scaling laws for
@xmath3 and @xmath128 , the only exponent we need to fit is @xmath10 . since in the experiment
it is very difficult to set the control parameters precisely , we choose ten cases within a narrow range @xmath168 , see figure [ fig : htalp ] . by a linear regression
we obtain @xmath126 with @xmath169 .
similarly , we define the rescaled reynolds number as @xmath72 . in figure
[ fig : htrey ] we plot both @xmath3 and @xmath73 against @xmath1 for the experimental results with @xmath170 in a compensated form , and compare the results with the gl theory .. ] indeed , compared to the original values the rescaled reynolds number @xmath73 collapses and is very close to the gl prediction .
on @xmath2 for the experimental data from @xcite with @xmath171 and @xmath170 .
the dashed line indicates the linear regression giving @xmath126 with @xmath172.,scaledwidth=70.0% ] versus @xmath1 for the experimental data from @xcite with @xmath170 .
( b ) the same data of panel ( a ) but rescaled as @xmath72 with @xmath172 .
symbols are coloured according to the logarithm of @xmath2 .
the dashed lines are the predictions of the gl theory . ]
a rescaled finger width can be defined accordingly as @xmath131 with the @xmath172 . in figure
[ fig : htfgw ] we plot the original and rescaled values of the finger width .
the rescaled values collapse and its dependence on @xmath1 follows a power - law scaling .
the exponent obtained by a linear regression is @xmath173 .
this value is very similar to that for @xmath165 , i.e. @xmath130 as given in subsection 4.1 .
however , based on the current results it is not clear whether this exponent is universal for different prandtl numbers .
and ( b ) the rescaled finger width @xmath131 with @xmath172 for the experimental data from @xcite with @xmath170 .
symbols are coloured according to the logarithm of @xmath2 . in ( b )
the data points collapse after the rescaling , and the dashed line represents the linear regression with slope @xmath174 . ]
it should be pointed out that our scaling relations are very close those given in @xcite .
@xcite proposed @xmath175 . from figure
[ fig : htrey](b ) one observes that @xmath176 decreases very slowly as @xmath1 increases .
in other words , the current scaling law for the reynolds number is roughly @xmath177 . for the finger width , @xcite determined @xmath178 , while our analysis gives @xmath179 .
furthermore , @xcite suggested that @xmath180 , i.e. a very weak dependence on @xmath2 and for @xmath1 an exponent very close to @xmath122 .
finally , we want to stress that the effective exponents @xmath10 and @xmath76 take different values for different fluid system , i.e. they depend on two prandtl numbers . in table [ tab : expo ] we summarise the values of @xmath10 and @xmath76 for three different combinations of @xmath181 and @xmath182 , i.e. the current data , the experimental data from @xcite , and one group of data from @xcite with @xmath183 , @xmath62 , and @xmath184 .
for the last dataset the finger regime occupies a smaller range of @xmath2 because @xmath185 is smaller compared to the other two datasets .
more simulations are needed , especially at different prandtl numbers , to fully understand the physical origin of these exponents .
.the effective exponents @xmath10 and @xmath76 for different combinations of @xmath181 and @xmath182 . [ cols="^,^,^,^,^,^ " , ]
a systematic numerical study is carried out for the ddc flow bounded by two parallel plates at the prandtl numbers @xmath186 and @xmath187 , which are similar to the values of seawater .
the salinity rayleigh number @xmath1 covers six decades in order of magnitude , i.e. from @xmath188 to @xmath141 , and the density ratio @xmath2 is between @xmath52 to @xmath53 .
the salinity nusselt number @xmath0 depends mainly on @xmath1 .
the dependence of @xmath0 on @xmath1 is well captured by the gl theory with the same coefficients as determined for rb flow . for fixed @xmath1 , as the density ratio @xmath2 increases , @xmath0 keeps constant , while both the reynolds number @xmath3 and the convective heat flux @xmath64 decrease according to certain power - law scalings , namely @xmath126 and @xmath189 .
the two exponents are calculated from the numerical results , and for @xmath165 the values are @xmath7 and @xmath8 .
then the rescaled reynolds number and convective heat flux are introduced as @xmath72 and @xmath77 .
after rescaling , both @xmath73 and @xmath78 collapse for different @xmath2 and exhibit similar dependences on @xmath1 .
we have interpreted this finding as the ( stabilising ) temperature field being advected by the velocity field like a passive scalar . for fixed @xmath1 and varying @xmath2 ,
salt fingers have different horizontal size and velocity but transfer salinity with a similar rate .
the flow fields on the horizontal mid plane indicate that salt fingers usually are circular in the horizontal sections . to generate a similar salinity flux at different density ratio ,
the horizontal width of salt fingers must scales as @xmath190 for fixed @xmath1 .
our numerical results confirm this argument , and the rescaled finger width @xmath191 follows a single power - law scaling versus @xmath1 . in the current flow ,
the velocity boundary layer is driven by the vertical motion of salt fingers , and its thickness follows the same scaling laws as the finger width .
the thickness of the salinity boundary layer , however , scales perfectly as @xmath192 , which is a natural result from the global balance between the salinity dissipation and nusselt number , and the fact that the total salinity dissipation is dominated by the contribution from the boundary layers .
the scaling laws proposed for our numerical results are also tested against the experimental data of @xcite with @xmath166 .
the exponent @xmath10 now has the value @xmath11 .
the rescaling of the reynolds number and the finger width with this new @xmath10 collapses the data points for all experimental cases with @xmath170 .
the dependence of the rescaled reynolds number on @xmath1 is very close to the gl prediction .
the two effective exponents @xmath10 and @xmath76 are also calculated for an additional set of numerical results with @xmath193 , @xmath62 , and @xmath184 .
their values are @xmath194 and @xmath195 , respectively .
all these different values of @xmath10 and @xmath76 for different @xmath196 indicate that the two exponents depend on the properties of the fluid system , and more simulations are needed to fully understand their physical origins .
when @xmath1 is high enough , a stack of horizontal zonal flows emerge in the bulk region and have alternative flow directions .
for the fluid system consider here , these zonal flow appears when @xmath197 . for these large @xmath1 corresponding to big separation between two plates the salt fingers can not extend from one plate to the opposite one , but instead have a vertical length which is smaller than the domain height .
the inclined gravity waves , which were reported for the fully periodic simulations , were not observed in our bounded flows .
this is probably due to the vertical constrain of the plates .
the current results propose various problems for future studies .
the present scaling strategy should be further validated with other combinations of prandtl numbers .
since the prandtl numbers in our simulations are similar to those of seawater , it is of great interests to test the applicability of the current scaling relations to oceanic salt - finger layers .
the same methodology may also be used to develop scaling relations for ddc in the diffusive regime , where the flow is driven by an unstable temperature difference and stabilized by a salinity difference .
finally , in our simulations even at @xmath143 , we did not observe the staircase state with alternating finger and convection layers .
determining the control parameters at which the staircase state will be realised in the current flow configuration would be very interesting and highly desired but requires more simulations .
this study is supported by foundation for fundamental research on matter , and by the netherlands center for multiscale catalytic energy conversion ( mcec ) , an nwo gravitation programme funded by the ministry of education , culture and science of the government of the netherlands .
the computing resources were provided by surfsara , and through the prace project 2014112708 .
in the following tables we provide the details of our numerical simulations . & @xmath198 & @xmath199 & @xmath200 + & @xmath201 & 0.1 & 1.104 & 9.291 & 0.3903 & 0.7248 & 11.98 & 5.750 & 10 & @xmath202 & @xmath203 + & @xmath201 & 1 & 1.029 & 9.575 & 0.2647 & 0.4965 & 7.330 & 6.014 & 5 & @xmath204 & @xmath205 + @xmath206 & @xmath201 & 10 & 1.005 & 8.635 & 0.1392 & 0.4336 & 5.609 & 7.069 & 4 & @xmath207 & @xmath208 + & @xmath209 & 0.5 & 1.063 & 11.80 & 0.4070 & 0.4588 & 6.848 & 4.784 & 5 & @xmath210 & @xmath205 + @xmath206 & @xmath209 & 5 & 1.013 & 11.11 & 0.2301 & 0.3913 & 5.160 & 5.569 & 4 & @xmath207 & @xmath208 + & @xmath211 & 0.2 & 1.154 & 15.56 & 0.7092 & 0.3895 & 6.127 & 3.584 & 4 & @xmath210 & @xmath205 + @xmath206 & @xmath211 & 2 & 1.036 & 14.78 & 0.4284 & 0.3303 & 4.617 & 4.050 & 4 & @xmath207 & @xmath205 + & @xmath212 & 1.25 & 1.060 & 17.00 & 0.5800 & 0.2994 & 4.406 & 3.451 & 4 & @xmath213 & @xmath205 + & @xmath214 & 0.1 & 1.368 & 18.39 & 1.112 & 0.4112 & 6.314 & 2.971 & 4 & @xmath215 & @xmath216 + & @xmath214 & 0.4 & 1.127 & 18.79 & 0.8295 & 0.2942 & 4.571 & 3.021 & 2 & @xmath207 & @xmath205 + @xmath206 & @xmath214 & 1 & 1.078 & 18.23 & 0.6723 & 0.2964 & 4.317 & 3.205 & 4 & @xmath213 & @xmath205 + & @xmath214 & 2 & 1.045 & 18.33 & 0.5770 & 0.2729 & 3.795 & 3.244 & 2 & @xmath217 & @xmath205 + & @xmath214 & 4 & 1.026 & 18.14 & 0.4820 & 0.2528 & 3.417 & 3.349 & 2 & @xmath217 & @xmath205 + @xmath206 & @xmath214 & 10 & 1.012 & 17.44 & 0.3591 & 0.2402 & 2.978 & 3.399 & 2 & @xmath218 & @xmath219 + @xmath206 & @xmath220 & 0.5 & 1.169 & 22.09 & 1.024 & 0.2726 & 4.164 & 2.592 & 1.6 & @xmath218 & @xmath205 + @xmath206 & @xmath220 & 5 & 1.028 & 22.43 & 0.5963 & 0.2173 & 2.756 & 2.732 & 2 & @xmath218 & @xmath219 + @xmath206 & @xmath221 & 0.2 & 1.443 & 29.01 & 1.830 & 0.2271 & 3.963 & 1.911 & 1.6 & @xmath222 & @xmath205 + @xmath206 & @xmath221 & 2 & 1.082 & 29.03 & 1.088 & 0.1960 & 2.595 & 2.094 & 2 & @xmath222 & @xmath205 + & @xmath223 & 1.25 & 1.133 & 34.01 & 1.526 & 0.1709 & 2.379 & 1.761 & 2 & @xmath224 & @xmath205 + & @xmath198 & @xmath199 & @xmath200 + @xmath206 & @xmath225 & 0.1 & 1.890 & 35.46 & 2.936 & 0.2177 & 3.676 & 1.551 & 1.6 & @xmath226 & @xmath205 + & @xmath225 & 0.4 & 1.318 & 36.66 & 2.182 & 0.1734 & 2.691 & 1.582 & 1.6 & @xmath226 & @xmath205 + @xmath206 & @xmath225 & 1 & 1.173 & 36.19 & 1.755 & 0.1692 & 2.361 & 1.651 & 2 & @xmath224 & @xmath205 + & @xmath225 & 2 & 1.100 & 36.05 & 1.459 & 0.1581 & 2.133 & 1.681 & 1.6 & @xmath227 & @xmath205 + & @xmath225 & 4 & 1.058 & 36.48 & 1.225 & 0.1478 & 1.871 & 1.672 & 1.6 & @xmath227 & @xmath205 + & @xmath225 & 10 & 1.026 & 34.84 & 0.8992 & 0.1362 & 1.631 & 1.700 & 1.2 & @xmath228 & @xmath219 + @xmath206 & @xmath229 & 0.5 & 1.396 & 42.87 & 2.646 & 0.1591 & 2.387 & 1.360 & 1 & @xmath222 & @xmath230 + & @xmath229 & 5 & 1.059 & 44.97 & 1.539 & 0.1166 & 1.488 & 1.330 & 1.2 & @xmath222 & @xmath230 + @xmath206 & @xmath231 & 0.2 & 2.026 & 56.60 & 4.869 & 0.1285 & 2.295 & 0.9862 & 1 & @xmath232 & @xmath233 + & @xmath231 & 2 & 1.168 & 58.90 & 2.853 & 0.1043 & 1.389 & 1.022 & 1.2 & @xmath234 & @xmath205 + & @xmath235 & 1.25 & 1.287 & 67.32 & 3.891 & 0.0986 & 1.335 & 0.9001 & 1.2 & @xmath236 & @xmath205 + @xmath206 & @xmath237 & 0.1 & 3.103 & 68.40 & 7.752 & 0.1142 & 2.302 & 0.7922 & 0.8 & @xmath234 & @xmath230 + & @xmath237 & 1 & 1.362 & 72.34 & 4.539 & 0.0932 & 1.303 & 0.8263 & 1.2 & @xmath234 & @xmath233 + & @xmath237 & 1.05 & 1.352 & 71.89 & 4.446 & 0.0938 & 1.322 & 0.8338 & 1.2 & @xmath238 & @xmath233 + & @xmath237 & 1.2 & 1.320 & 72.23 & 4.290 & 0.0945 & 1.286 & 0.8238 & 1 & @xmath239 & @xmath233 + & @xmath237 & 1.6 & 1.255 & 71.66 & 3.941 & 0.0934 & 1.243 & 0.8350 & 0.7 & @xmath239 & @xmath230 + & @xmath237 & 2 & 1.213 & 72.33 & 3.740 & 0.0917 & 1.184 & 0.8236 & 0.6 & @xmath239 & @xmath230 + & @xmath237 & 10 & 1.052 & 70.30 & 2.347 & 0.0768 & 0.8772 & 0.8269 & 0.8 & @xmath240 & @xmath230 + & @xmath241 & 1.05 & 1.697 & 144.3 & 11.01 & 0.0583 & 0.7615 & 0.3810 & 0.8 & @xmath242 & @xmath230 + & @xmath241 & 1.2 & 1.651 & 146.5 & 10.82 & 0.0591 & 0.7468 & 0.3794 & 0.6 & @xmath243 & @xmath233 + & @xmath241 & 1.6 & 1.530 & 145.4 & 10.11 & 0.0548 & 0.7016 & 0.4016 & 0.5 & @xmath243 & @xmath230 + & @xmath241 & 2 & 1.454 & 147.8 & 9.694 & 0.0513 & 0.6462 & 0.4087 & 0.4 & @xmath244 & @xmath245 + & @xmath246 & 1.6 & 1.920 & 297.7 & 26.12 & 0.0294 & 0.3834 & 0.1755 & 0.32 & @xmath247 & @xmath248 + & @xmath249 & 1.6 & 2.677 & 599.6 & 66.60 & 0.0152 & 0.2093 & 0.08132 & 0.16 & @xmath250 & @xmath251 + 31ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1146/annurev.fl.26.010194.001351 [ * * , ( ) ] @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) link:\doibase 10.1029/jc083ic06p02902 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0198-0149(89)90066-6 [ * * , ( ) ] link:\doibase 10.1017/s0022112003004166 [ * * , ( ) ] link:\doibase 10.1017/s002211200999098x [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1126/science.1108678 [ * * , ( ) ] link:\doibase 10.1017/jfm.2011.99 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1017/jfm.2015.93 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1017/jfm.2011.98 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.86.3316 [ * * , ( ) ] link:\doibase 10.1103/physreve.66.016305 [ * * , ( ) ] link:\doibase 10.1063/1.1807751 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1017/s0022112069001066 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) | direct numerical simulations are conducted for double diffusive convection ( ddc ) bounded by two parallel plates .
the prandtl numbers , i.e. the ratios between the viscosity and the molecular diffusivities of scalars , are similar to the values of seawater . the ddc flow is driven by an unstable salinity difference ( here across the two plates ) and stabilized at the same time by a temperature difference .
for these conditions the flow can be in the finger regime .
we develop scaling laws for three key response parameters of the system : the non - dimensional salinity flux @xmath0 mainly depends on the salinity rayleigh number @xmath1 , which measures the strength of the salinity difference , and exhibits a very weak dependence on the density ratio @xmath2 , which is the ratio of the buoyancy forces induced by two scalar differences .
the non - dimensional flow velocity @xmath3 and the non - dimensional heat flux @xmath4 are dependent on both @xmath1 and @xmath2 .
however , the rescaled reynolds number @xmath5 and the rescaled convective heat flux @xmath6 depend only on @xmath1 .
the two exponents are dependent on the fluid properties and are determined from the numerical results as @xmath7 and @xmath8 . moreover
, the behaviors of @xmath0 and @xmath5 agree with the predictions of the grossmann - lohse theory which was originally developed for the rayleigh - bnard flow .
the non - dimensional salt - finger width and the thickness of the velocity boundary layers , after being rescaled by @xmath9 , collapse and obey a similar power - law scaling relation with @xmath1 .
when @xmath1 is large enough , salt fingers do not extend from one plate to the other and horizontal zonal flows emerge in the bulk region .
we then show that the current scaling strategy can be successfully applied to the experimental results of a heat - copper - ion system ( hage and tilgner , _ phys . fluids _
, * 22 * , 076603 , 2010 ) .
the fluid has different properties and the exponent @xmath10 takes a different value @xmath11 . |
this work was supported in part by the german bundesministerium fr bildung and forschung under the contract 05ht1woa3 , by the polish state committee for scientific research grant 2 p03b 060 18 for years 2000 - 2001 and by the ec contract hprn - ct-2000 - 00148 for years 2000 - 2004 . | we discuss corrections to @xmath0 , @xmath1 and to the cp violation parameter @xmath2 in two examples of ( generalized ) minimal flavour violation models : 2hdm and mssm in the large @xmath3 limit .
we show that for @xmath4 not too heavy , @xmath5 in the mssm with heavy sparticles can be substantially smaller than in the sm due the charged higgs box contributions and in particular due to the growing like @xmath6 contribution of the double penguin diagrams involving neutral higgs boson exchanges .
hep - ph/0108226 * processes in the mssm in large limit * janusz rosiek _ physik department , technische universitt mnchen , d-85748 garching , germany _ + the determination of the ckm parameters is the hot topic in particle physics . in view of forthcoming precise results from b - factories
, it is particularly important to discuss possible effects of new physics contributions to @xmath7 processes : @xmath0 and @xmath8 mass differences and to the cp violation parameter @xmath2 . in general models of new physics
fall into the two following broad classes : + @xmath9 models in which the ckm matrix remains the unique source of flavour and cp violation - so - called minimal flavour violation ( mfv ) models @xcite and their generalizations ( gmfv models ) . in the gmfv models the non sm - like operators contribute significantly to the effective low energy hamiltonian @xcite .
+ @xmath9 models with entirely new sources of flavour and/or cp violation .
+ on the basis of the analysis @xcite we discuss here two examples of the gmfv models : the large @xmath3 limit of the 2hdm(ii ) and of the mssm , in which the ckm matrix is the only source of flavour and cp violation ( see e.g. @xcite ) .
the effective weak hamiltonian for @xmath7 transitions in the gmvf models can be written as follows [ heff ] h_eff^=2 = g_f^2m_w^216 ^ 2 _ i v^i_ckm c_i ( ) q_i . where @xmath10 are the set of 8 dimension six @xmath7 operators @xcite and @xmath11 are the appropriate ckm factors .
@xmath0,@xmath5 and @xmath2 can be expressed in terms of , respectively , three real functions @xcite f^d_tt = s_0(x_t ) 1+f_d , f^s_tt = s_0(x_t ) 1+f_s , f^_tt = s_0(x_t ) 1+f _ , @xmath12 for @xmath13 gev .
we have then @xmath14 the functions @xmath15 , @xmath16 and @xmath17 can be expressed in terms of @xmath18 as @xmath19 + { 1\over4r}c^{\rm vrr}_1(\mu_t)+ \bar p_1^{\rm lr } c^{\rm lr}_1(\mu_t ) + \bar p_2^{\rm lr } c^{\rm lr}_2(\mu_t ) \nonumber\\ & & + \bar p_1^{\rm sll}\left[c^{\rm sll}_1(\mu_t)+c^{\rm srr}_1(\mu_t)\right ] + \bar p_2^{\rm sll}\left[c^{\rm sll}_2(\mu_t)+c^{\rm srr}_2(\mu_t)\right]\end{aligned}\ ] ] where @xcite @xmath20 and the factors @xmath21 are given in @xcite .
let us consider first the 2hdm(ii ) .
at 1-loop only the charged scalar is relevant for the box diagrams contributing to and mixing . for large @xmath3 and
@xmath22 the leading terms of such contributions to the wilson coefficients @xmath18 are of the order of ( see @xcite for the complete expressions ) : @xmath23 for diagrams with @xmath24 , and @xmath25 for diagrams with @xmath26 . for large
@xmath3 the biggest contribution appears in @xmath27 and is further enhanced , compared to the sm amplitude , by the qcd renormalization effects @xcite .
it is of the opposite sign than given by the @xmath28 box diagram and can be significant only for the @xmath29-@xmath30 mixing ( so that it can compete with the sm contribution - see fig .
[ fig : bcrs9 ] ) , for which it is of the order of @xmath31 similar contributions for @xmath32-@xmath33 and transitions are suppressed by factors @xmath34 and @xmath35 , respectively , and thus very small . [ cols= " < , < " , ] unfortunately , the new cleo experimental result for the process @xmath36 set the bound @xmath37 gev @xcite .
thus , in the 2hdm(ii ) for the still allowed range of charged higgs boson mass the effects in @xmath8 can not be large . as a second realistic gmfv model we consider the mssm with the ckm as the only source of flavour / cp violation .
in the limit of heavy sparticles ( practically realized already for @xmath38 gev ) the 1-loop box diagrams involving charginos and stops are negligible .
however , for large @xmath3 even if sparticles are heavy they can still compensate the @xmath39 contribution to the @xmath40 amplitude allowing for the existence of a light , @xmath41 gev ) , charged higgs boson @xcite . from fig .
[ fig : bcrs9 ] it follows therefore that , even for @xmath42 and already at the 1-loop level the contribution of the mssm higgs sector to the @xmath43 wilson coefficient can be non - negligible . at the 2-loop level in the mssm one has to take into account the dominant 2-loop electroweak corrections , modifying the yukawa - type interactions . even for heavy sparticles these corrections can significantly modify @xcite the 2hdm(ii ) relations between the masses @xmath44 and the yukawa couplings @xmath45 , as well as induce additional , @xmath46 , terms in the charged higgs - quark couplings @xcite . for the @xmath7 processes the most important ( for non - negligible mixing of the top squarks )
is however their third effect , that is the generation of the flavour non - diagonal , @xmath3 enhanced couplings of the neutral higgs bosons to down - type quarks @xcite . additional contributions to @xmath47 , @xmath48 and @xmath49 are then generated by the double penguin diagrams involving the neutral higgs bosons exchanges .
the dominant terms obtained from such contributions are @xmath50 with @xmath51 given by ( the matrices @xmath52 and @xmath53 are defined in @xcite ) : @xmath54 where the function @xmath55 is given in @xcite and the factor @xmath56 reads as @xmath57\end{aligned}\ ] ] @xmath6 factor in eq .
( [ eqn : babucor ] ) appears because the dominant part of the effective flavour off - diagonal yukawa coupling is given @xcite by the flavour - changing chargino contribution to the @xmath58-quark self energy ( @xmath59 ) , multiplied by the tree - level yukawa coupling ( also @xmath59 ) .
two powers of the external light quark masses in ( [ eqn : babucor ] ) are canceled by the fermion propagators on internal lines . for @xmath60 , @xmath61 is close to zero @xcite , but the correction @xmath62 is proportional to @xmath63 which is not suppressed in this limit .
approximating the dimensionless factor @xmath51 by unity , it is easy to see that in the case of the mixing this correction can be for @xmath64 and @xmath65 gev as large as @xmath66 i.e. of the same order of magnitude as the sm contribution to @xmath67 .
this is illustrated in fig .
[ fig : bcrs13 ] .
an important features of the double penguin contribution are : _
i ) _ its fixed negative sign , the same as the sign of the dominant effects of the charged higgs box diagrams at large @xmath3 ; _ ii ) _ its strong dependence on the left - right mixing of the top squarks - from fig .
[ fig : bcrs13 ] it follows that large values of the stop mixing parameter @xmath68 are excluded already by the present experimental data ; _ iii ) _ it does not vanish when the mass scale of the sparticles is increased uniformly - large effects decreasing @xmath5 can be present also for the heavy sparticles provided the higgs boson masses remain low .
+ summarizing , we found that : + @xmath9 the largest effects of new contributions for large @xmath3 are seen in @xmath5 .
the corresponding contributions to @xmath0 and @xmath2 are strongly suppressed by the smallness of @xmath58-quark mass .
+ @xmath9 the dominant contributions to @xmath5 for large @xmath3 originate from double penguin diagrams involving neutral higgs particles and , to a lesser extent , in the box diagrams with charged higgs exchanges . +
* @xmath9 * the contribution of double penguins grows like @xmath6 and interferes destructively with the sm contribution , suppressing considerably @xmath5 .
it depends strongly on stop mixing , excluding large values of the mixing parameter @xmath68 . |
modern density - functional theory ( dft ) was introduced by hohenberg and kohn in a classic paper @xcite and is now the workhorse of quantum chemistry and other fields of quantum physics .
subsequently , dft was put on a mathematically firm ground by lieb using convex analysis.@xcite the central quantity of dft is the universal density functional @xmath11 , which represents the electronic energy of the system consistent with a given density @xmath12 . clearly , the success of dft hinges on the modelling of @xmath0 , an extremely complicated function of the electron density .
it is an interesting observation that , over the last two or three decades , @xmath0 has been modelled sufficiently accurately to make dft the most widely applied method of quantum chemistry , in spite of the fact that schuch and verstraete @xcite have shown how considerations from the field of computational complexity place fundamental limits on exact dft : if @xmath11 could be found efficiently , all np hard problems would be solvable in polynomial time , which is highly unlikely.@xcite from a mathematical point of view , dft is neatly formulated using convex analysis@xcite : the universal density functional @xmath11 and the ground - state energy @xmath7 are related by a conjugation operation , with the density @xmath12 and external potential @xmath13 being elements of a certain banach space @xmath14 and its dual @xmath15 , respectively .
the functionals @xmath0 and @xmath2 are equivalent in the sense that they contain the same information each can be generated exactly from the other .
the universal density functional @xmath0 is convex and lower semi - continuous but otherwise highly irregular and ill behaved .
importantly , @xmath0 is everywhere discontinuous and not differentiable in any sense that justifies taking the functional derivative in formal expressions even for the @xmath13-representable densities ,
as pointed out by lammert.@xcite for example , it is common practice to formally differentiate @xmath0 with respect to the density , interpreting the functional derivative `` @xmath16 '' as a scalar potential at @xmath17 .
however , this derivative , a gteaux derivative , does not exist .
together with the problem of @xmath13-representability , conventional dft is riddled with mathematically unfounded assumptions that are , in fact , probably false .
for example , conventional kohn sham theory assumes , in addition to differentiability of @xmath0 , that , if @xmath12 is @xmath13-representable for an interacting @xmath18-electron system , then @xmath12 is also @xmath13-representable for the corresponding noninteracting system.@xcite while providing excellent predictive results with modelled approximate density functionals , it is , from a mathematical perspective , unclear why kohn sham dft works at all .
it is the goal of this article to remedy this situation by introducing a family of regularized dfts based on a tool from convex analysis known as the _ moreau envelope _ or _
yosida regularization_. for @xmath3 , the idea is to introduce a regularized energy functional @xmath19 related to the usual ground - state energy @xmath2 by @xmath20 where @xmath21 is the usual @xmath22-norm .
the convex conjugate of @xmath23 is the moreau envelope @xmath24 of @xmath0 , from which the regularized ground - state energy can be obtained by a hohenberg - kohn minimization over densities : @xmath25 where @xmath26 .
the usual hohenberg
kohn variation principle is recovered as @xmath5 .
importantly , the moreau envelope @xmath27 is _ everywhere differentiable _ and converges pointwise from below to @xmath11 as @xmath5 .
we use the term `` regularized '' for both @xmath23 and @xmath28 , although it is @xmath28 that , as will be shown below , becomes differentiable through the procedure .
a remark regarding the banach spaces of densities and potentials is here in order . if @xmath13 is a coulomb potential , then the regularization term in eq .
becomes infinite .
moreover , the strongest results concerning the moreau yosida regularization are obtained in a reflexive setting . the usual banach spaces @xmath29 and @xmath30 for densities and potentials , respectively,@xcite are therefore abandoned , and both replaced with the hilbert space @xmath31 , where @xmath32 ^ 3 $ ] is an arbitrarily large but finite box in @xmath33 . as is well known , domain truncation represents a well - behaved approximation : as @xmath34 increases , all eigenvalues converge to the @xmath33-limit .
moreover , the continuous spectrum is approximated by an increasing number of eigenvalues whose spacing converges to zero .
we observe that , in the box , the difference @xmath35 is arbitrarily small and _
explicitly known_it does not relate to the electronic structure of the system and is easily calculated from @xmath13 .
nothing is therefore lost in the transition from @xmath1 to @xmath36 . on the contrary
, we obtain a structurally simpler theory that allows taking the derivative of expressions involving the universal functional
. moreover , the differentiability of @xmath37 implies @xmath13-representability of any @xmath12 , for noninteracting as well as interacting systems , as needed for a rigorous formulation of kohn sham theory . in this paper
, we explore the moreau envelope as applied to dft , demonstrating how every concept of standard dft has a counterpart in the moreau - regularized formulation of dft and vice versa .
the remainder of the article is organized as follows : in sec.[sec : prelim ] , we review formal dft and discuss the regularity issues of the universal functional within the nonreflexive banach - space setting of lieb.@xcite in preparation for the moreau yosida regularization , we next reformulate dft in a truncated domain , introducing the hilbert space @xmath31 as density and potential space .
the moreau
yosida regularization is a standard technique of convex analysis , applicable to any convex function such as the universal density functional .
we introduce this regularization in sec.[sec : my ] , reviewing its basic mathematical properties .
to establish notation , a review of convex analysis is given in the appendix ; for a good textbook of convex analysis in a hilbert space , with an in - depth discussion of the moreau
yosida regularization , see ref . .
following the introduction of the moreau
yosida regularization , we apply it to dft in sec.[sec : mydft ] and subsequently to kohn
sham theory in sec.[sec : ks ] .
finally , sec.[sec : conclusion ] contains some concluding remarks .
in dft , we express the born
oppenheimer ground - state problem of an @xmath18-electron system in the external electrostatic potential @xmath38 as a problem referring only to the one - electron density @xmath39 .
oppenheimer @xmath18-electron molecular hamiltonian is given by @xmath40 where @xmath41 and @xmath42 are the kinetic - energy and electron - electron repulsion operators , respectively , while @xmath43 is a multiplicative @xmath18-electron operator corresponding to the scalar potential @xmath38 .
the scalar @xmath44 is introduced to distinguish between the interacting ( @xmath45 ) and noninteracting ( @xmath46 ) systems . by levy s constrained - search argument,@xcite the ( fully interacting ) ground - state energy ,
@xmath47 can be written in the form of a hohenberg
kohn variation principle , @xmath48 where @xmath49 is the set of @xmath18-representable densities that is , @xmath50 if and only if there exists a normalized @xmath18-electron wave function with finite kinetic energy and density @xmath12 . in eq .
, the infimum extends over all properly symmetrized and normalized @xmath51 , the first - order sobolev space consisting of those functions in @xmath52 that have first - order derivatives also in @xmath52 and therefore have a finite kinetic energy .
different universal density functionals @xmath0 can be used in eq.([eq : e_of_v_dft ] ) , the only requirement of an admissible functional being that the correct ground - state energy @xmath7 is recovered .
given that @xmath53 , it follows that @xmath54 .
as demonstrated by lieb in ref . , the universal density functional @xmath0 can be chosen as a unique lower semi - continuous convex function with respect to the @xmath55 topology .
( by definition , therefore , @xmath56 for any @xmath57 ; see appendix a for remarks on extended - valued functions . ) moreover , by a sobolev inequality,@xcite we may embed the @xmath18-representable densities in the banach space @xmath58 , with norm @xmath59 and topological dual @xmath60 .
given that this banach space @xmath14 has a stronger topology than @xmath61 , a convergent sequence in @xmath14 converges also in @xmath62 . from the lower semi - continuity of @xmath0 in @xmath61 ,
we then obtain @xmath63 implying that @xmath0 is lower semi - continuous also in the topology of @xmath14 .
we note that the choice @xmath64 is not unique , but it has the virtue that all coulomb potentials are contained in @xmath15 . on the chosen banach spaces , the ( concave and continuous ) ground - state energy @xmath65 and the ( convex and lower semi - continuous ) universal density functional @xmath66 are related by the variation principles @xmath67 in the terminology of convex analysis ( see appendix a ) , @xmath68 and @xmath69 are each other s convex fenchel conjugates . to reflect the nonsymmetric relationship between @xmath2 and @xmath0 in eqs . and
, we introduce the nonstandard but useful mnemonic notation @xmath70 which is suggestive of the `` shape '' of the resulting functions : @xmath71 is concave , whereas @xmath72 is convex .
the density functional @xmath0 in eq . is an extension of the universal functional @xmath73 derived by hohenberg and kohn,@xcite the latter functional having from our perspective the problem that it is defined only for ground - state densities ( @xmath13-representable densities ) in @xmath74 , an implicitly defined set that we do not know how to characterize .
it can be shown that the functional @xmath0 defined by eq . is identical to the constrained - search functional @xcite @xmath75 , where the minimization is over all ensemble density matrices
@xmath76 corresponding to a density @xmath12 , constructed from @xmath18-electron wave functions with a finite kinetic energy . a related functional is the ( nonconvex ) levy lieb constrained search functional,@xcite @xmath77 , obtained by minimizing over pure states only . in any case , eq . defines the _ unique _ lower semi - continuous , convex universal functional such that @xmath78 .
in fact , any @xmath79 that satisfies the condition @xmath80 is an admissible density functional . in particular , @xmath81 and @xmath73
are both admissible , satisfying this requirement when extended from their domains ( @xmath49 and @xmath82 , respectively ) to all of @xmath14 by setting them equal to @xmath83 elsewhere .
the hohenberg
kohn variation principle in eq .
is appealing , reducing the @xmath18-electron problem to a problem referring only to one - electron densities .
however , as discussed in the introduction , @xmath0 is a complicated function . in particular
, here we consider its nondifferentiability .
the gteaux derivative is closely related to the notion of directional derivatives , see appendix a. a function @xmath0 is gteaux differentiable at @xmath84 if the directional derivative @xmath85 is linear and continuous in all directions @xmath86 , meaning that there exists a @xmath87 such that @xmath88 however , @xmath0 is finite only on @xmath49 . in a direction @xmath89
such that @xmath90 , @xmath91 for all @xmath92 , implying that @xmath93 and hence that @xmath0 is not continuous in the direction of @xmath94 .
the same argument shows that @xmath0 is discontinuous also in directions @xmath94 such that the density @xmath95 is negative in a volume of nonzero measure for all @xmath92 . abandoning strict gteaux differentiability for this reason , we may at the next step
investigate whether the directional derivative exists and is linear for directions that stay inside @xmath96 , the subset of @xmath14 containing all nonnegative functions that integrate to @xmath18 electrons .
after all , the discontinuity of @xmath0 in directions that change the particle number is typically dealt with using a lagrange multiplier for the particle number constraint .
however , lammert has demonstrated that , even within @xmath96 , there are , for each @xmath12 , directions such that @xmath93 , associated with short - scale but very rapid spatial oscillations in the density ( and an infinite kinetic energy).@xcite apart from lower ( upper ) semi - continuity of a convex ( concave ) function , the minimal useful regularity is not gteaux differentiability but subdifferentiability ( superdifferentiability ) , see appendix a. let @xmath97 be convex lower semi - continuous .
the subdifferential of @xmath98 at @xmath99 , @xmath100 , is by definition the collection of slopes of supporting continuous tangent functionals of @xmath98 at @xmath99 , known as the _ subgradients _ of @xmath98 at @xmath99 , see fig .
[ fig : tangents ] in appendix a. if the graph of @xmath98 has a `` kink '' at @xmath99 , then there exists more than one such subgradient .
at a given point @xmath101 , the subdifferential @xmath102 may be empty .
we denote by @xmath103 the set of points @xmath101 such that @xmath104 .
it is a fact that @xmath103 is dense in @xmath105 when @xmath98 is a proper lower semi - continuous convex function .
the superdifferential of a concave function is similarly defined . together with convexity , subdifferentiability is sufficient to characterize minima of convex functions : a convex lower semi - continuous functional @xmath106 has a global minimum at @xmath107 if and only if @xmath108 . similarly @xmath109 has a minimum if and only if @xmath110 .
subdifferentiability is a substantially weaker concept than that of gteaux ( or directional ) differentiability . clearly , if @xmath111 is gteaux differentiable at @xmath99 , then @xmath112 .
however , the converse is not true : in infinite - dimensional spaces , it is possible that @xmath113 , a singleton , while @xmath111 is not differentiable at @xmath99 .
this is so because @xmath102 being a singleton is not enough to guarantee continuity of @xmath98 . in dft
, subdifferentiability has an important interpretation .
suppose @xmath12 is an ensemble ground - state density of @xmath13 , meaning that , for all @xmath114 , we have the inequality @xmath115 then , the subdifferential of @xmath11 at @xmath12 is @xmath116 which is a restatement of the first hohenberg
kohn theorem : the potential for which @xmath12 is a ground - state density is unique up to a constant shift . on the other hand , if @xmath12 is not a ground - state density for any @xmath117 , then @xmath118 .
thus , a nonempty subdifferential is equivalent to ( ensemble ) @xmath13-representability : @xmath119 if and only if @xmath12 is @xmath13-representable . denoting the set of ensemble @xmath13-representable densities by @xmath120
, we obtain @xmath121 we note that @xmath120 is dense in @xmath96 , the subset of @xmath14 containing all nonnegative functions that integrate to @xmath18 electrons .
however , even though subdifferentiability is sufficient for many purposes , differentiability of @xmath0 would make formal manipulations easier . moreover , the characterization of @xmath13-representable @xmath122 is unknown and probably dependent on the interaction strength @xmath44 .
these observations motivate the search for a differentiable regularization of the universal functional .
let us briefly consider the superdifferential of @xmath2 , a concave continuous ( and hence upper semi - continuous ) function over @xmath15 . a fundamental theorem of convex analysis states that @xmath123 where we use the same notation for sub- and superdifferentials .
thus , the potential @xmath13 has a ground state with density @xmath12 if and only if @xmath124 ; if @xmath13 does not support a ground state , then @xmath125 is empty . denoting the set of potentials in @xmath15 that support a ground state by @xmath126
, we obtain : @xmath127 if a ground state is nondegenerate , then @xmath128 is a singleton ; together with the fact that @xmath2 is continuous , it then follows that @xmath2 is gteaux differentiable at @xmath13 . on the other hand , if the ground state is degenerate , then the subdifferential is the convex hull of @xmath129 ground - state densities : @xmath130 and @xmath2 is not differentiable at this @xmath13 unless all the @xmath131 are equal that is , if the degenerate ground states have the same density . for example , in the absence of a magnetic field , the hydrogen atom has the degenerate ground states @xmath132 and @xmath133 , with the same density .
in sec . [ sec : my ] , we outline the mathematical background for the moreau yosida regularization . many useful results , such as differentiability of the moreau envelope @xmath27 , are only available when the underlying vector space @xmath14 is reflexive or , even better , when @xmath14 is a hilbert space
. however , the banach space @xmath134 used in lieb s formulation of dft is nonreflexive . in this section ,
we truncate the full space @xmath33 to a box @xmath135 ^ 3 $ ] of finite volume @xmath136 , so large that the ground state energy of every system of interest is sufficiently close to the @xmath33 limit .
what is lost from this truncation is well compensated for by the fact that we may now formulate dft using the hilbert space @xmath137 for both potentials and densities , as we shall now demonstrate . for the spatial domain @xmath138 ,
the @xmath18-electron ground - state problem is a variational search for the lowest - energy wave function @xmath139 , the first - order sobolev space with vanishing values of the boundary of @xmath140 , the @xmath18-fold cartesian product of @xmath138 .
the search is carried out only over the subset of @xmath141 which is also normalized and properly symmetrized : for a total spin projection of @xmath142 , the corresponding subset of wavefunctions is antisymmetric in the @xmath143 first and the @xmath144 last particle coordinates separately .
any potential in the full space , @xmath145 , induces a potential @xmath146 in the truncated domain .
we remark that @xmath147 , with equivalent topologies . since the domain is bounded , the rellich kondrakov theorem@xcite states that @xmath141 is compactly embedded in @xmath148 , which in turn implies that the spectrum of the hamiltonian @xmath149 in eq . is purely discrete.@xcite thus , for any potential @xmath13 in the box , one or more ground - state wave functions @xmath150 exists .
we next observe that , if @xmath151 is a coulomb potential , then the truncated potential @xmath13 belongs to @xmath31 .
moreover , @xmath152 since @xmath138 is bounded .
it is therefore sufficient to consider the ground - state energy as a function @xmath153 regarding the continuity of @xmath154 , we note that the proof given in ref . for the continuity of @xmath2 in the @xmath155 topology
is equally valid for @xmath154 in the @xmath156 topology .
convergence in @xmath31 implies convergence in @xmath157 .
therefore , @xmath154 is continuous in the @xmath31 topology .
we remark that , as @xmath158 , @xmath159 converges to the exact , full - space ground - state energy @xmath7 . on the other hand ,
the associated eigenfunctions converge if and only if the full - space ground - state energy @xmath7 is an eigenvalue , with @xmath160 as a counterexample . invoking the usual ensemble constrained - search procedure ,
we obtain @xmath161 where @xmath162 is the set of @xmath18-representable densities : @xmath163 if and only if there exists a properly symmetrized and normalized @xmath164 such that @xmath165 .
it is straightforward to see that @xmath166 the density functional @xmath167 is completely analogous to the full - space functional @xmath0 . in particular , @xmath167 is lower semi - continuous in the @xmath168 topology by theorem 4.4 and corollary 4.5 in ref . .
we remark that @xmath169 for any @xmath163 , as seen from the fact that , if @xmath51 and @xmath170 with @xmath171 , then we must have @xmath164 .
since @xmath138 is bounded , the cauchy schwarz inequality gives for any measurable @xmath172 , @xmath173 by an argument similar to that of eq .
, @xmath167 is now seen to be lower semi - continuous also with respect to the @xmath31 topology .
note that @xmath174 so that every @xmath18-representable density is in @xmath31 . since @xmath167 is convex and lower semi - continuous on @xmath175 , we may now formulate dft in the hilbert space @xmath176 as @xmath177 given that hilbert spaces possess a richer structure than banach spaces , this formulation of dft is particularly convenient : densities and potentials are now elements of the same vector space @xmath176 and reflexivity is guaranteed .
even for the full space , @xmath178 , indicating that it is possible to avoid the use of the box .
indeed , we may restrict the ground - state energy to potentials @xmath179 : @xmath180 a concave and continuous map . invoking the theory of conjugation within this reflexive hilbert - space setting
, we have a convex lower semi - continuous universal functional @xmath181 however , coulomb potentials are not contained in @xmath182 . on the other hand , this theory is sufficient for dealing with all truncated coulomb potentials , obtained , for example , from the usual coulomb potentials by setting them equal to zero outside the box @xmath183 ; it is also sufficient when working with yukawa rather than coulomb potentials .
the optimality conditions for the hohenberg kohn and lieb variation principles in eqs . and are @xmath184 denoting the set of densities for which @xmath167 is subdifferentiable by @xmath185 ( by analogy with @xmath120 in @xmath14 ) and the set of potentials for which @xmath154 is superdifferentiable by @xmath186 ( by analogy with @xmath126 in @xmath187 ) , we obtain @xmath188 where @xmath189 is dense in the subset of @xmath176 containing all nonnegative functions that integrate to @xmath18 electrons .
the differentiability properties of @xmath167 are the same as those of @xmath0 discussed in section [ sec : nonreg ] .
to introduce differentiability , a further regularization is necessary .
in this section , we present the basic theory of moreau yosida regularization , introducing infimal convolutions in section[subsec : inf ] , moreau envelopes in section[subsec : moreau ] , proximal mappings in section[subsec : prox ] , and conjugates of moreau envelopes in section[subsec : conmoreau ] . the results are given mostly without proofs ; for these proofs , we refer to the book by bauschke and combettes,@xcite whose notation we follow closely . in preparation for the moreau
yosida regularization , we introduce the concept of infimal convolution in this section and discuss its properties on a hilbert space @xmath190 . for @xmath191 ,
the _ infimal convolution _ is the function @xmath192 given by @xmath193 in the context of convex conjugation , the infimal convolution is analogous to the standard convolution in the context of the fourier transform . here
are some basic properties of the infimal convolution for functions that do not take on the value @xmath194 : [ thm : infimal ] let @xmath191 .
then : 1 . @xmath195 ; 2 .
@xmath196 ; 3 .
@xmath197 ; 4 . if @xmath98 and @xmath129 are convex , then @xmath198 is convex .
see ref , props .
12.6 , 12.11 and 13.21 . henceforth , we restrict our attention to _ all lower semi - continuous proper convex functions _
@xmath199 , denoting the set of all such functions by @xmath200 , see appendix [ matsup ] .
we also need the concepts of coercivity and supercoercivity : a function @xmath201 is _ coercive _ if @xmath202 whenever @xmath203 and _ supercoercive _ if @xmath204 whenever @xmath203 . for example
, @xmath205 is coercive , whereas @xmath206 is not coercive . for functions in @xmath200
, we have the following stronger properties of the infimal convolution : [ thm : infimal2 ] let @xmath207 such that either @xmath129 is supercoercive or @xmath98 is bounded from below and @xmath129 is coercive
. then 1 .
@xmath208 ; 2 .
@xmath209 ; 3 .
[ item : uniqueness ] for each @xmath210 , there exists @xmath211 such that @xmath212 where @xmath213 is unique if @xmath129 is strictly convex .
point 1 follows from ref . , prop .
point 2 follows from theorem [ thm : infimal ] above .
finally , point 3 follows from the fact that strictly convex functions have unique minima ; the existence of a minimum follows from the ( super)coerciveness of the mapping @xmath214 . in the following ,
we introduce the moreau envelope of functions in @xmath200 and review its properties .
[ def : myx ] for @xmath215 and @xmath216 , the _ moreau yosida regularization _ or the _ moreau envelope _
@xmath217 is the infimal convolution of @xmath98 with @xmath218 : @xmath219 since @xmath215 and since @xmath218 is strictly convex and supercoercive , it follows from theorem[thm : infimal ] that @xmath220 .
in fact , @xmath221 is much more well behaved than a general function in @xmath200 , as the following theorem shows .
[ thm : my ] the moreau envelope @xmath221 of @xmath215 with @xmath222 satisfies the following properties : 1 .
@xmath220 with @xmath223 ; 2 .
@xmath224 for all @xmath210 and all @xmath225 ; 3 .
@xmath226 ; 4 . for all @xmath227 , @xmath228 from below as @xmath5 ( even if @xmath229 ) ; 5 .
@xmath221 is continuous ; 6 .
@xmath221 is frchet differentiable : for every @xmath210 , there exists @xmath230 such that for all @xmath231 : @xmath232 7 .
the subdifferential of @xmath221 at @xmath99 is given by @xmath233 point1 follows from theorems [ thm : infimal ] and [ thm : infimal2 ] . for points 2 and 3 ,
see ref . ,
12.9 . for point 4 , see prop .
12.32 . for points 57 ,
see props . 12.15 , 12.28 , and 12.29 . in figure
[ fig : moreau ] , the moreau envelope is illustrated for a convex function @xmath98 on the real axis .
we observe that the minimum value of @xmath111 is preserved by the moreau envelope @xmath234 and that the second argument @xmath235 to the infimal convolution removes all kinks , giving a curvature equal to that of this function . .
the function @xmath236 is plotted in thick lines , whereas @xmath111 is shown in a thinner line .
finally , for a chosen value of @xmath237 , the function @xmath238 is superposed on @xmath111 and @xmath239 using a dashed line .
[ fig : moreau ] ] from theorem[thm : infimal ] , it follows that the infimum of @xmath239 in eq .
is attained with a unique minimizer .
we make the following definitions : [ prox ] let @xmath215 and @xmath216 .
the _ proximal mapping _
@xmath240 is defined by @xmath241 where @xmath242 is _ the proximal point of @xmath98 at @xmath210_. the usefulness of the proximal mapping follows from the following theorem : [ thm : my2 ] let @xmath215 and @xmath216
. then 1 .
if @xmath243 and @xmath5 , then @xmath244 2 .
the frchet ( and gteaux ) derivative of @xmath221 at @xmath99 is given by @xmath245 3 .
for all @xmath246 , it holds that @xmath247 4 . if @xmath210 , then @xmath248 for point 1 , see the proof of prop . 12.32 in ref . .
for point 2 , see prop . 12.29 ; for point 3 , see prop .
point 4 follows from point 2 and 3 .
given that @xmath220 , there exists a concave @xmath249 such that @xmath250 and @xmath251 .
the following theorem gives the basic properties of this conjugate : [ thm : cmy ] if @xmath221 is the moreau envelope of @xmath215 , then their conjugates and the superdifferentials of these conjugates are related as @xmath252 eq .
follows from the fact that the convex conjugate of @xmath253 is @xmath254 and from theorem [ thm : infimal ] .
. follows from the fact that the superdifferential of a sum of concave functions is the sum of their superdifferentials if one of the functions is continuous at a common point in their domains , see remark 16.36 of ref . .
finally , @xmath255 .
being related in such a simple manner , @xmath256 and @xmath257 share many properties .
we note , however , that @xmath257 is strictly concave , whereas @xmath256 may be merely concave .
we remark that the moreau envelope is not defined for a _ concave _ function @xmath258 , only for convex functions .
thus , the notation @xmath259 for a @xmath260 is not to be interpreted as a moreau envelope , but as the _ concave conjugate _ of a moreau envelope , @xmath261 .
having introduced moreau yosida regularization in the preceding section , we are ready to apply it to dft on the hilbert space @xmath175 . applying eqs . and with @xmath262 and @xmath263 , we obtain the regularized lieb functional @xmath264 and ground - state energy @xmath265 , @xmath266 importantly , these functions are related to each other as conjugate functions ; just as we have already encountered for the @xmath1 and @xmath267 conjugate pairs . as such , the following hohenberg kohn and lieb variation principles hold on the hilbert space @xmath176 : @xmath268 however , unlike @xmath0 and @xmath167 , which are finite only for @xmath18-representable densities , the moreau
yosida regularized lieb functional @xmath269 is finite on the whole hilbert space : @xmath270 since , in eq .
, a finite value is always found on the right - hand side , even when @xmath271 .
a curious side effect of the regularization is therefore that the minimizing density in the regularized hohenberg
kohn variation principle in eq .
( which exists for all @xmath272 ) may not be @xmath18-representable : it may be negative in a region of finite measure or contain an incorrect number of electrons . to illustrate the behaviour of the regularized functional for nonphysical densities , consider @xmath273 when @xmath12 is @xmath18-representable and @xmath274 . from the definition of the moreau envelope in eq .
, we obtain straightforwardly that @xmath275 the regularized density functional thus depends on @xmath276 in a simple quadratic manner , with a minimum at @xmath277 .
as @xmath278 tends to zero from above , @xmath279 increases more and more rapidly with increasing @xmath280 , approaching @xmath281 more closely .
as expected , the regularized functional is differentiable in the direction that changes the number of electrons . on the face of it , the existence of minimizing ` pseudo - densities ' in the hohenberg
kohn variation principle that are not @xmath18-representable may seem to be a serious shortcoming of the moreau
yosida regularization ideally , we would like the minimizing density to arise from some @xmath18-electron wave function .
however , the appearance of nonphysical pseudo - densities is an inevitable consequence of the regularization differentiability in all directions can not be achieved without extending the effective domain of @xmath167 to all @xmath176 ; alternatively , we may retain the effective domain of @xmath18-representable densities and instead work with restricted functional derivatives , defined only in directions that conserve some properties of the density .
such an approach is straightforward for directions that change the number of electrons in the system but much more difficult for directions that lead to negative densities or to an infinite kinetic energy .
the existence of minimizing pseudo - densities that are not @xmath18-representable is less important than the fact that @xmath269 converges pointwise to @xmath0 from below as @xmath282 , even when @xmath283 .
also , we shall in the next subsection see that every @xmath284 is linked to a unique physical ground - state density @xmath285 .
it is therefore possible to regard ( and to treat ) the hohenberg
kohn minimization over pseudo - densities in @xmath176 as a minimization over physical densities in @xmath185 , as discussed below .
we also observe that @xmath286 converges pointwise to @xmath154 from below as @xmath287 .
more importantly , for any chosen @xmath216 , we may recover the exact ground - state energy @xmath154 from the regularized energy @xmath286 simply by adding the term @xmath288 , which _ does not depend on the electronic structure of the system_. indeed , this term is no more relevant for the molecular electronic system than the neglected nuclear nuclear repulsion term its purpose is merely to make the ground - state energy strictly concave and supercoercive in the external potential so that the universal density functional becomes differentiable and continuous .
indeed , no information regarding the electronic system is lost in the regularization beyond what is lost upon truncation of the domain from @xmath289 to an arbitrarily large cubic box @xmath183 , needed to make @xmath288 finite for all potentials .
according to the general theory of moreau yosida regularization , a unique minimizer , which we shall here call the _ proximal ( ground - state ) density _ , @xmath290 exists for any @xmath284 in the regularized lieb functional of eq . , which may therefore be written as @xmath291 from eq .
, we conclude that the standard lieb functional is subdifferentiable at @xmath292 and hence that @xmath292 is an ensemble @xmath13-representable ground - state density in @xmath176 : @xmath293 we also see from eq . that every @xmath284 and associated proximal ground - state density @xmath292 together satisfy the subgradient relation @xmath294 implying that @xmath295 is an external potential with ground - state density @xmath285 . in the following ,
we refer to @xmath296 as the _ proximal potential _ associated with @xmath12 .
we recall that , by the hohenberg
kohn theorem , the density determines the potential up to a constant .
the subdifferential of @xmath167 at the proximal density @xmath292 is therefore @xmath297 where @xmath296 is the proximal potential of eq ..
conversely , suppose that @xmath298 .
there then exists an external potential @xmath13 such that @xmath299 . expressing @xmath13 in the form @xmath300 for some @xmath301
, we obtain @xmath302 , which by eqs . and implies that @xmath12 is the proximal density of @xmath303 .
thus , every ensemble @xmath13-representable density @xmath298 is the proximal density of @xmath304 where @xmath13 is such that @xmath299 : @xmath305 in short , we have the important fact that the set of proximal densities in @xmath176 is _ precisely _ the set of ensemble ground - state densities @xmath306 . a density @xmath284 whose proximal density is @xmath292 is called a _ carrier density _ of @xmath292 . by the hohenberg
kohn theorem , the potential @xmath13 in eq . is unique up a constant @xmath274 .
the carrier density is therefore uniquely determined up to an additive constant .
the nonuniqueness of the carrier density also follows directly from eq.([eq : frhoc ] ) , which shows that @xmath12 and @xmath307 where @xmath284 and @xmath274 have the same proximal ground - state density @xmath285 . to summarize , even though the densities in the regularized hohenberg
kohn variation principle in eq .
are pseudo - densities ( not associated with any @xmath18-electron wave function ) , every such density @xmath284 is uniquely mapped to a ground - state density by the surjective proximal operator @xmath308 this operator performs the decomposition @xmath309 where the proximal density @xmath285 may be viewed as the ` projection ' of @xmath12 onto @xmath185 with potential @xmath310 .
we note that @xmath311 , even when @xmath298 .
the proximal operator is therefore not a true projector . for any @xmath284 ,
the proximal density @xmath292 and proximal potential @xmath296 together satisfy the usual reciprocal relations for the standard lieb functional and ground - state energy : @xmath312 see eq .
, and therefore satisfy the relation : @xmath313 thus , to every solution of the regularized hohenberg kohn variation principle with @xmath314 in eq . there corresponds a proximal solution to the standard variation principle with @xmath315 . regarding the differentiability of the regularized lieb functional , we note from theorems[thm : my ] and [ thm : my2 ] that @xmath269 is frchet differentiable so that @xmath316 with the derivative given by eq .
: @xmath317 gteaux differentiability follows from frchet differentiability : the existence of @xmath318 implies that the directional derivatives at @xmath12 exist in all directions @xmath319 and are equal to @xmath320 hence the functional derivative of @xmath269 is well defined and given by @xmath321 justifying the formal manipulations involving functional derivatives in dft , recalling that @xmath322 tends to @xmath323 pointwise from below as @xmath324 .
( however , @xmath296 need not converge to anything . )
the optimality conditions of the regularized dft variation principles in eqs . and are the reciprocal relations @xmath325 which for the regularized hohenberg kohn variation principle may now be written in the form of a stationary condition : @xmath326 in combination with eq . , we obtain @xmath327 and hence from eq . the following hohenberg
kohn stationary condition : @xmath328 suggestive of an iterative scheme with the repeated calculation of the proximal density until self - consistency .
by contrast , the lieb optimality condition @xmath329 in eq . can not be written as a stationary condition since the ground - state energy @xmath286 ( just like @xmath2 and @xmath154 ) is differentiable only when @xmath13 has a unique ground - state density . from theorem[thm : cmy ] , we obtain @xmath330 which shows that the degeneracy of the ground - state energy is preserved by the moreau yosida regularization . for any @xmath284 in eq .
, an explicit expression for the potential @xmath296 in terms of the proximal density is given in eq .
, yielding the regularized ground - state energy @xmath331 hence , _ for every @xmath284 , there exists a potential @xmath296 for which @xmath12 is the ground - state density_. stated differently , the set of ensemble @xmath13-representable pseudo - densities @xmath332 is equal to the full hilbert space : @xmath333 we recall that the proximal density @xmath292 is the exact ( standard ) ground - state energy of @xmath296 , see eq ..
in the present section , we apply moreau yosida regularization to kohn sham theory , beginning with a discussion of the adiabatic connection .
the essential point of the regularized kohn
sham theory is the existence of a common ground - state pseudo - density for the interacting and noninteracting systems , thereby solving the representability problem of kohn sham theory . in the present section ,
we simplify notation by omitting the subscript that indicates the length of the box from all quantities writing @xmath190 , for instance , rather than @xmath176 everywhere .
the presentation of moreau yosida regularized dft given in section [ sec : mydft ] was for the fully interacting electronic system , with an interaction strength @xmath334 in the hamiltonian of eq .. however , given that nothing in the development of the theory depends on the value of @xmath44 , it may be repeated without modification for @xmath335 . in particular
, we note that the set of ground - state pseudo - densities is equal to the whole hilbert space and hence is the same for all interaction strengths , see eq .. consequently , _
every @xmath336 is the ground - state pseudo - density of some @xmath337 , for each @xmath44_. to setup the adiabatic connection , we select @xmath336 . denoting by @xmath338 the regularized universal density functional at interaction strength @xmath44
, we obtain from eq .
the unique external potential @xmath339 for which the regularized ground - state energy at that interaction strength @xmath340 is given by @xmath341 as @xmath44 changes , the potential @xmath342 can be adjusted to setup an adiabatic connection of systems with the same ground - state pseudo - density @xmath12 at different interaction strengths . in the moreau
yosida regularized adiabatic connection , the pseudo - density @xmath12 has a proximal ground - state density that depends on @xmath44 : @xmath343 which is the true ground - state density in the potential @xmath342 at that interaction strength : @xmath344 in short , in the adiabatic connection , the effective potential @xmath342 has the same ground - state pseudo - density @xmath12 but different ground - state densities @xmath345 for different interaction strengths . in the next subsection ,
we shall see how this decomposition makes it possible to calculate the true ground - state energy by ( regularized ) kohn sham theory in a rigorous manner , with no approximations except those introduced by domain truncation .
consider an @xmath18-electron system with external potential @xmath346 .
we wish to calculate the ground - state energy and to determine a ground - state density of this system : @xmath347 this can be achieved by solving the interacting many - body schrdinger equation , in some approximate manner . in kohn sham theory , we proceed differently , solving instead a noninteracting problem with the same density .
we begin by transforming eq . into a regularized many - body energy , noting that the energy and superdifferential of the exact and regularized ground - state energies are related according to eqs . and as @xmath348 from these relations
, it follows that the pseudo - density @xmath349 is a ground - state density of the regularized system : @xmath350 the subscript ` c ' indicates that @xmath351 is the carrier density of both the physical ground - state of the system @xmath12 according to eq . and the ground - state density of the kohn
sham system @xmath352 : @xmath353 our task is to determine the carrier density and regularized ground - state energy by solving eq .. the solution will subsequently be transformed to yield the physical ground - state density and energy .
we observe that the carrier density @xmath354 is obtained from the physical density @xmath12 by subtracting @xmath355 with @xmath216 , see eq .
( [ eq : rhoast ] ) . in practice , @xmath356 since the external potential is the attractive coulomb potential of the nuclei .
it therefore follows that the pseudo - density is strictly positive : @xmath357 .
given that @xmath358 , there exists a kohn sham potential @xmath359 such that @xmath351 is the ground - state density of a noninteracting system in this potential : @xmath360 to determine the regularized kohn
sham potential @xmath361 , we first note that the potentials @xmath362 and @xmath361 satisfy the stationary condition in eq . : @xmath363 to proceed , we next introduce the regularized hartree exchange correlation energy and potential as @xmath364 yielding the following expression for the kohn sham potential as a function of the density : @xmath365 to solve the regularized kohn - sham problem in eq . , we first note that it is related in a simple manner to the standard kohn
sham problem : @xmath366 we then proceed in an iterative fashion .
from some trial pseudo - density @xmath367 , we iterate @xmath368 until convergence , beginning with @xmath369 and terminating when self - consistency has been established .
we emphasize that the regularized kohn sham iterations in eqs . and are _ identical to the iterations in standard kohn sham theory _ except for the use of a regularized hartree exchange
correlation potential in the construction of the kohn sham matrix and the subtraction of @xmath370 from the density generated by diagonalization of the resulting kohn
sham matrix .
having determined the ground - state carrier density @xmath351 and the corresponding kohn sham potential @xmath361 by iterating eq . and until self consistency , we calculate the interacting regularized ground - state energy as @xmath371 from which the physical ground - state energy @xmath372 is recovered by adding @xmath373 according to eq . , while the ground - state density @xmath12 is recovered by adding @xmath355 to the pseudo - density @xmath351 according to eq .. we note that the pair @xmath374 is uniquely determined to the extent that @xmath12 in eq .
is unique ; for systems with degenerate ground - state densities , several equivalent pairs @xmath374 exist . by means of moreau yosida regularization , we have thus setup kohn
sham theory in a rigorous manner , where the interacting and noninteracting ground - state densities are different ( by an amount proportional to @xmath278 ) but related by the same carrier density @xmath351 , thereby solving the noninteracting representability problem of standard kohn sham theory .
moreover , differentiability of the regularized universal density functional means that the potentials associated with this pseudo - density at different interaction strengths are well defined as the ( negative ) derivatives of the density functional . in the limit where @xmath324 , standard kohn
sham theory is approached , although the limit itself is not expected to be well behaved .
the possibility of setting up dft follows from the mathematical properties of the ground - state energy @xmath7 , which is continuous and concave in the external potential @xmath13 . by convex conjugation
, it may be exactly represented by the lower semi - continuous and convex universal density functional @xmath11 , whose properties reflect those of the ground - state energy .
unfortunately , @xmath11 depends on the density @xmath12 in a highly irregular manner , being everywhere discontinuous and nowhere differentiable .
these characteristics of @xmath0 arise in part because @xmath2 is concave but not strictly concave and not supercoercive . by modifying @xmath2 in a way that introduces strict concavity and supercoercivity without losing information about the electronic system
, we obtain an alternative dft , where the universal density functional is much more well behaved , being everywhere differentiable ( and therefore also continuous ) .
this is achieved by moreau
yosida regularization , where we apply convex conjugation not to @xmath7 itself but to the strictly concave function @xmath375 , where @xmath216 .
the resulting density functional @xmath376 is convex and differentiable .
standard dft is recovered as @xmath324 but this limit need not be taken for the theory to be exact for any chosen value of @xmath278 , we can perform dft as usual ; the exact ground - state energy is recovered as @xmath377 .
the only restriction on the exact theory is the truncation of the domain from @xmath289 to a box of finite ( but arbitrarily large ) volume ; such a domain truncation simplifies the moreau
yosida formulation of dft by introducing ( reflexive ) hilbert spaces of densities and potentials .
the densities that occur naturally in regularized dft are not physical densities since they can not be generated from an @xmath18-electron wave function in the usual manner .
nevertheless , each ` pseudo - density ' @xmath12 has a clear physical interpretation : it can be uniquely decomposed as @xmath378 , where @xmath292 is a physical ground - state density ( the ` proximal density ' ) and @xmath296 the associated potential .
this density decomposition justifies kohn sham theory : a given pseudo - density @xmath12 is uniquely decomposed as @xmath379 , at each interaction strength @xmath44 . as @xmath44 changes , the decomposition of @xmath12 changes accordingly . for the fully interacting system , @xmath380 where @xmath381 is the physical ground - state density and @xmath362 the external potential ; for the noninteracting system , @xmath382 , where @xmath383 and @xmath384 are the kohn sham density and potential , thereby solving the noninteracting representability problem of kohn sham theory .
the working equations of regularized kohn sham theory are essentially identical to those of standard kohn sham theory . here ,
we have considered standard moreau
yosida regularization .
however , we may also consider a generalized approach , in which the regularizing term @xmath385 is replaced by @xmath386 , where the operator @xmath387 is chosen based on some _ a priori _ knowledge of the desired solution .
indeed , some choices of @xmath387 result in approaches closely related to known regularization techniques , such as the zhao morrisson
parr approach @xcite to calculate the noninteracting universal density functional and the `` smoothing - norm '' regularization approach of heaton - burgess _ et .
, used both in the context of optimized effective potentials @xcite and lieb optimization methods @xcite . these and related approaches will be discussed in a forthcoming paper .
we expect such moreau
yosida techniques to be of great practical value in the implementation of procedures that attempt to determine either the ground - state energy or the universal density functional by direct optimization techniques using their derivatives , bearing in mind that both the derivatives and the objective functions are well defined in the regularized context .
this work was supported by the norwegian research council through the coe centre for theoretical and computational chemistry ( ctcc ) grant no .
179568/v30 and the grant no .
171185/v30 and through the european research council under the european union seventh framework program through the advanced grant abacus , erc grant agreement no . 267683 .
a. m. t. is also grateful for support from the royal society university research fellowship scheme .
in this section , we review some important concepts of convex analysis and the calculus of variations . suggested reading for convex analysis are van tiel s book@xcite and the classic text by ekeland and tmam.@xcite the present article relies on additional information gathered in the book by bauschke and combettes,@xcite which focuses on the hilbert - space formulation of convex analysis . for functional analysis ,
the monograph by kreyszig@xcite is recommended .
we are here concerned with extended real - valued functions @xmath388 over a banach or hilbert space @xmath389 .
note that we define @xmath390 for any @xmath391 , and @xmath392 for positive real numbers @xmath99 , but that @xmath393 is not defined .
we recall that @xmath15 , the topological dual of @xmath14 , is the set of continuous linear functionals over @xmath14 : if @xmath394 , then @xmath395 is a real - valued map , continuous and linear in @xmath107 .
we denote by @xmath396 the value of @xmath395 at @xmath99 , except in the dft setting , where the notation @xmath397 is used . for simplicity , we assume in this section that @xmath14 is reflexive so that @xmath398 . ultimately , we shall work with hilbert spaces , which are reflexive banach spaces so that @xmath399 by the riesz representation theorem of functional analysis .
let @xmath400 be an extended - valued function .
the ( effective ) domain @xmath401 is the subset of @xmath14 where @xmath98 is not @xmath83 .
the function @xmath98 is said to be proper if @xmath402 .
the function @xmath98 is convex if , for all @xmath99 and @xmath403 in @xmath14 , and for all @xmath404 , @xmath405 note that this formula also makes sense if , say , @xmath406 .
the interpretation of convexity is that a linear interpolation between two points always lays on or above the graph of @xmath98 .
we say that @xmath98 is strictly convex if strict inequality holds for @xmath407 in eq.([eq : convexity ] ) .
moreover , @xmath98 is said to be concave if the inequality is reversed in eq.([eq : convexity ] ) and strict concavity is defined similarly .
perhaps the most important property of a convex @xmath98 is that any local minimum is also a global minimum .
moreover , if @xmath98 is strictly convex , the global minimizer , if it exists , is unique .
convex optimization problems are in this sense well behaved .
the minimal useful regularity of convex functions is not continuity but lower semi - continuity . in a metric space @xmath14 ,
a function @xmath98 is said to be lower semi - continuous if , for every sequence @xmath408 converging to some @xmath107 , we have @xmath409 the importance of lower semi - continuity is that it guarantees the existence of a global minimum if @xmath410 is compact : @xmath411 for some @xmath412 .
for concave functions , upper semi - continuity is the corresponding useful notion ; @xmath98 is upper semi - continuous if @xmath413 is lower semi - continuous , by definition .
we are particularly interested in lower semi - continuous proper convex functions .
the set @xmath414 is defined as consisting of all functions that can be written in the form @xmath415 for some family @xmath416 of dual functions and some @xmath417 .
the set @xmath414 contains precisely all lower semi - continuous proper convex functions on @xmath14 and the functions identically equal to @xmath418 . in other words
, @xmath98 is lower semi - continuous proper convex or identically equal to @xmath418 if and only if it is the pointwise supremum of a set of continuous affine ( `` straight - line '' ) functions over @xmath14 .
we denote by @xmath419 all proper lower semi - continuous functions on @xmath14 : @xmath420 .
it is a fact that any @xmath421 is also _ weakly _ lower semi - continuous . on the dual space @xmath187
, we denote by @xmath422 the set of all functions that can be written in the form @xmath423 these functions are precisely the weak-@xmath424 lower semi - continuous proper convex functions on @xmath187 and the improper functions @xmath418 .
the proper functions are @xmath425 .
the unique function @xmath129 is said to be the _
convex conjugate _ of @xmath98 and is denoted by @xmath429 ; likewise , @xmath430 is the _ convex conjugate _ of @xmath129 .
a pair of functions @xmath431 and @xmath432 that are each other s convex conjugates are said to be _ dual functions_. the dual functions contain the same information , only coded differently : each property of @xmath98 is reflected , in some manner , in the properties of @xmath433 and vice versa .
we note the relations @xmath434 for functions @xmath431 and @xmath432 .
in fact , the conjugation operation is a bijective map between @xmath414 and @xmath422 , they contain precisely those functions that satisfy the biconjugation relations in eq .. a dual function @xmath394 is said to be a _ subgradient _ to @xmath98 at a point @xmath99 where @xmath111 is finite if @xmath443 meaning that the affine function @xmath444 is nowhere above the graph of @xmath98 .
the subdifferential @xmath445 is the set of all subgradients to @xmath98 at @xmath99 , see figure [ fig : tangents ] .
note that @xmath102 may be empty .
the function @xmath98 is said to be subdifferentiable at @xmath107 if @xmath446 .
a function @xmath447 has a global minimum at @xmath107 if and only if @xmath108 .
similarly @xmath109 has a minimum if and only if @xmath110 . a function @xmath431 is subdifferentiable on a dense subset of its domain @xmath105 . .
for a @xmath448 , @xmath449 is a collection of slopes of tangent functionals .
one such slope @xmath395 and its affine mapping is shown explicitly , the rest is indicated with stippled lines .
@xmath395 is not unique since the graph of @xmath98 has a `` kink '' at @xmath450.[fig : tangents ] ] in the context of dft , @xmath0 is subdifferentiable at @xmath451 if and only if @xmath12 is the ground - state density of a potential @xmath452 , @xmath453 so that @xmath454 by the hohenberg kohn theorem , we know that @xmath455 if @xmath12 is @xmath13-representable and that @xmath118 otherwise .
thus , from the point of view of convex analysis , the notion of @xmath13-representability of @xmath12 is equivalent to subdifferentiability of @xmath0 at @xmath12 .
it follows that the @xmath13-representable densities are dense in the set of @xmath18-representable densities , the effective domain of @xmath0 .
let @xmath456 .
the _ directional derivative _ of @xmath98 at @xmath99 in the direction of @xmath403 is defined by @xmath457\ ] ] if this limit exists ( @xmath83 is accepted as limit ) . for @xmath447 ,
the directional derivative @xmath458 always exists .
let @xmath107 be given . if there is a @xmath459 such that @xmath460 then @xmath98 is said to be _ gteaux differentiable _ at @xmath99 .
in other words , a function is gteaux differentiable if its various directional derivatives may be assembled into a linear functional at @xmath99 .
the gteaux derivative is the usual notion of functional derivative encountered in the calculus of variations , for which we write @xmath461 .
if @xmath98 is continuous and has a unique subgradient at @xmath99 , then it is is also gteaux differentiable at @xmath99 ; the converse statement is also true , but note that a unique subgradient alone is not enough to ensure gteaux differentiability : continuity is not implied by a unique subgradient . a stronger notion of differentiability is given by the frchet derivative .
let @xmath107 .
if there exists @xmath394 such that for all sequences @xmath462 in @xmath14 as @xmath463 , @xmath464 then @xmath98 is _ frchet differentiable _ at @xmath99 , and @xmath465 is the frchet derivative .
clearly , frchet differentiable implies gteaux differentiable , but not the other way around .
in fact , if @xmath466 exists at @xmath99 , then @xmath467 so that @xmath98 is approximated by its linearization around @xmath99 .
this is not true if @xmath98 is merely gteaux differentiable .
19ifxundefined [ 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 * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop _ _ ( , , ) | the universal density functional @xmath0 of density - functional theory is a complicated and ill - behaved function of the density in particular , @xmath0 is not differentiable , making many formal manipulations more complicated .
whilst @xmath0 has been well characterized in terms of convex analysis as forming a conjugate pair @xmath1 with the ground - state energy @xmath2 via the hohenberg kohn and lieb variation principles , @xmath0 is nondifferentiable and subdifferentiable only on a small ( but dense ) set of its domain . in this article
, we apply a tool from convex analysis , moreau yosida regularization , to construct , for any @xmath3 , pairs of conjugate functionals @xmath4 that converge to @xmath1 pointwise everywhere as @xmath5 , and such that @xmath6 is ( frchet ) differentiable . for technical reasons ,
we limit our attention to molecular electronic systems in a finite but large box .
it is noteworthy that no information is lost in the moreau
yosida regularization : the physical ground - state energy @xmath7 is exactly recoverable from the regularized ground - state energy @xmath8 in a simple way .
all concepts and results pertaining to the original @xmath1 pair have direct counterparts in results for @xmath9 .
the moreau
yosida regularization therefore allows for an exact , differentiable formulation of density - functional theory .
in particular , taking advantage of the differentiability of @xmath10 , a rigorous formulation of kohn sham theory is presented that does not suffer from the noninteracting representability problem in standard kohn sham theory . |
this research was supported by nserc - canada and by rfbr grants 98 - 02 - 17372 , 98 - 02 - 17453 and 00 - 15 - 96562 .
m.j.levine , nuovo cimento * a48 * ( 1967 ) 67 .
a.v.kuznetsov and n.v.mikheev , physics letters * b299 * ( 1993 ) 367 .
p.p.kronberg , rep .
prog . phys.*57 * ( 1994 ) 325 ; + e.asseo and h.sol , phys .
reports * 148 * ( 1987 ) 307 .
duane a. dicus , wayne w. repko , and roberto vega , hep - ph/0006264 and references therein . | we present the standard model calculation of the optical activity of a neutrino sea .
the idea that intergalactic space is a birefrigent medium for light due to the presence of a neutrino sea has been contemplated for a long time .
some thirty years ago , royer @xcite computed in @xmath0 theory an effect of order @xmath1 . later
, stodolsky noted that due to a theorem of gell - mann @xcite , there can be no such effect with massless neutrinos and a point - like coupling , and his observation was recorded by one of the present authors in a review @xcite . in the early 1980 s data on propagation of radio waves through intergalactic space put a stringent upper bound on possible optical activity of the neutrino sea @xcite and @xcite and this led to renewed estimates for the size of such effect on the assumption of a neutrino magnetic moment , which occur for a massive neutrino @xcite .
more recently @xcite an evaluation was made for a off - mass shell photons within the standard model ( sm ) .
here we note that there is a frequency dependent effect even for real photons and we also comment on the effect for virtual photons ( both within the sm ) .
thus we shall consider three possible effects : a magnetic moment effect , an effect for virtual photons , an effect for real photons .
+ if the neutrino has a magnetic moment , this moment will interact with the magnetic field of an electromagnetic wave : @xmath2 in this equation @xmath3 is the polarization vector of a photon of momenta @xmath4 , while @xmath5 is the neutrino bispinor , and @xmath6 is the neutrino magnetic moment in bohr magnetons @xmath7 .
+ with this interaction ( 1 ) the forward scattering amplitude of a photon of momentum @xmath8 from a neutrino of momentum @xmath9 is equal to @xmath10 this amplitude differs for left - handed and right - handed photons : @xmath11 and gives rise to a nonzero optical activity for a polarizes neutrino gas . + in the standard model the neutrino has no intrinsic magnetic moment , but a magnetic moment can emerge as a one loop effect if the neutrino is massive(see figure 1 ) .
+ the order of magnitude of the magnetic moment is : @xmath12 where @xmath13 and @xmath14 is the weak mixing angle ; @xmath15 .
constant @xmath16 can be obtained by direct calculation @xcite @xmath17 thus a nonzero optical activity appears in this case as a two - loop effect @xmath18 it was understood long ago that @xmath19 scattering takes place as a one - loop process even for massless neutrino .
the first estimates were performed in the fermi theory but erroneously @xcite . according to gell - mann s theorem @xcite
the amplitude should vanish for point - like interactions .
somewhat later levine @xcite evaluated the scattering mediated by w boson exchange and found a nonzero amplitude ( see figure 2 ) . due to gauge invariance
the amplitude starts with the second power of the momentum of the photon @xmath8 .
levine computed the first non - vanishing term , which in the sm looks like : @xmath20 this amplitude is even under parity , in other words it is the same for right - handed photons and left - handed photons , so it can not contribute to optical activity .
+ very recently it has been found that by expanding to the next term of order @xmath21 one finds non - vanishing p - odd terms .
the result of this calculation is @xcite : @xmath22 this amplitude vanishes for real photons but is non - vanishing for virtual photons .
the amplitude is singular with respect to the electron mass and comes entirely from the diagram 2a . + it is interesting to ask whether one can get a parity violating amplitude for real photons .
we find a positive answer to this question . in the same order of expansion over small momenta @xmath8 ( @xmath23 )
one finds the amplitude : @xmath24 where @xmath25 this does not vanish for real photons and contributes to the optical activity of neutrino gas .
the logarithmic term in equation ( 10 ) comes from diagram 2a , while the constant receives contributions from all diagrams of figure 2 .
the calculation is straightforward .
+ before we discuss the order of magnitude of the different amplitudes we note that the authors of reference @xcite have missed one diagram which contributes to off - shell photons .
this is the diagram with @xmath26 exchange shown in figure 3 . for real photons
, @xmath27 the triangle diagrams with different fermions inside the loop cancel each other , so that the standard model is anomaly free .
but for off - shell photons each diagram gives a contribution proportional to @xmath28 , where m is the mass of the fermion running inside the loop .
the main contribution comes from the electron loop and is equal to @xmath29 of the contribution from diagram 2a ( see ref .
@xcite ) .
the sum of contributions from figure 2a and figure 3 gives @xmath30 summarizing these results , we obtain the order of magnitude estimates : @xmath31 at low frequencies the scattering due to magnetic moment dominates even though it takes place at two - loop level : @xmath32 here we assume the neutrino at rest , i.e. @xmath33 .
+ as for the relative value of @xmath34 and @xmath35 @xmath36 this is sensitive to the @xmath37 value of the virtual photon . following the authors of reference @xcite
we take for intergalactic space @xmath38 and assuming @xmath39 , we obtain : @xmath40 for visible photons the on - shell activity is smaller than for off - shell photons , but in the @xmath41-ray region the reverse is the case . + in principle the contribution @xmath35 is similar to the rotation from the intergalactic magnetic fields , but the frequency dependence is very different .
a numerical estimate shows that the contribution from the neutrino sea is many order of magnitude smaller than the magnetic ( faraday ) effect , observed with radio waves @xcite .
we estimate , for radio waves and for @xmath42 neutrino sea with fermi momenta @xmath43 an effect of order @xmath44 @xmath45 which is insignificant compared to the faraday effect . at shorter wavelengths the rotatory power due to neutrino sea would increase .
these small amplitudes are matched by very small cross sections for photon - neutrino scattering @xcite .
the scattering of laser photons with high energy intensive neutrino beam is discussed in the literature ( e.g. see @xcite ) . for this case
the considered one - loop amplitude dominates over scattering due to magnetic moment : @xmath46 |
apart from a brief phase of lithium burning , brown dwarfs cool continuously .
thus any meaningful comparison with theory requires a knowledge of the age of the brown dwarf . for this reason
much effort has been devoted to finding brown dwarfs in clusters whose age is known .
the three closest clusters are the hyades ( d = 46 pc ) , coma ( d = 90 pc ) and the pleiades ( d = 130 pc ) .
the hyades and coma are old clusters with ages of 625 myr @xcite and 500 myr @xcite respectively , and were not thought to have any brown dwarfs .
more recently @xcite have found 2 brown dwarfs in the hyades and @xcite have found 13 brown dwarf candidates in the coma cluster .
the pleiades ( age 125 myr ) has some 50 known brown dwarfs @xcite , with some more recently discovered by @xcite .
thus the nearest cluster with a significant number of known brown dwarfs is the pleiades at a distance of 130 pc .
this distance , together with the intrinsic faintness of brown dwarfs , naturally makes it difficult to study cluster brown dwarfs .
by contrast , field brown dwarfs are close ( @xmath0 pc ) and easier to study but usually have unknown ages . however , some field star ages have been measured ( see for example @xcite or @xcite ) .
field brown dwarfs are found by surveys such as 2mass @xcite , denis @xcite , and the sdss @xcite .
a compilation of the known l and t dwarfs can be found in the l and t dwarf archive @xcite . one possible way of finding the ages of field brown dwarfs would be to see if they are members of a moving group .
a moving group is a group of stars with the same velocity , magnitude and direction , and the same age ( see @xcite for a recent review ) .
one of the closest moving groups is the ursa major / sirius moving group ( hereafter umsmg ) .
the core of the moving group , possibly a bound cluster , is in the direction of ursa major . indeed
the stars of the `` plough '' except @xmath1 uma are all members , as also is sirius ( see fig .
[ fig : map ] ) .
thus the sun is actually inside the umsmg .
group members can be found all around the sky , and may be very close : for example , sirius is only 2.65 pc from the sun .
the age of the umsmg has been determined as 300 myr by @xcite .
more recently , @xcite find 400 myr while @xcite find @xmath2 myr for the group age .
we will adopt an age of @xmath3 myr .
since moving group stars have a common velocity they appear to be moving towards the same place in the sky ; this is called the `` convergent point '' .
the umsmg convergent point is located at @xmath4 ( j2000 coordinates ) @xcite .
thus if a field brown dwarf has a proper motion directed towards the umsmg convergent point , it is a potential member of the umsmg .
this , coupled with two distance tests ( see below ) , allows us to identify members with considerable confidence .
the hyades is discussed in a thorough paper by @xcite .
the cluster lies at a distance of d = 46 pc , and has an extent in the sky of approximately @xmath5 ; @xcite give the position of the cluster centroid as @xmath6 .
the hyades is known to be deficient in low mass members @xcite .
these have probably evapourated from the cluster .
indeed , @xcite have identified escaped hyads and these may be thought of as part of the hyades moving group ( hmg ) .
the convergent point of the hyades is located at @xmath7 and their total space velocity is 46 km / sec @xcite .
the most recent and generally quoted hyades age is @xmath8 myr by @xcite , and we will adopt this age .
for the 70 members of the l and t dwarf archive @xcite ( available on the worldwide web at www.dwarfarchives.org ) with a measured proper motion , we first calculate the angular distance , @xmath9 , of the dwarf from the umsmg or hmg convergent point , where @xmath9 is given by @xmath10 here @xmath11 and @xmath12 are the dwarf declination and convergent point declination , and @xmath13 is the difference of their respective right ascensions .
next we find the direction , @xmath14 , from north of the convergent point , where @xmath15 a group member should have a proper motion direction equal to @xmath14 .
however , there is some velocity dispersion amongst group members , otherwise all group members would appear to be very close together , as if in a bound cluster .
for the umsmg the velocity @xmath16 is 17.98 km s@xmath17 and @xmath18 km s@xmath17 @xcite .
we adopt the same @xmath19 for the hyades recent escapers , even though this value is considerably less than the @xmath20 km s@xmath17 given by @xcite .
thus we impose the same membership conditions for both the umsmg and the hmg .
we find members have a proper motion direction within @xmath21 of @xmath14 .
this corresponds to @xmath22 , or 87% completeness , which seems reasonable .
this constraint is our first criterion for membership , and the random chance of passing this first test is clearly @xmath23 .
the extra factor of 4 is because proper motion directions are not randomly orientated ( see section 7 ) .
it may readily be shown @xcite that for a moving group the distance @xmath24 ( in parsecs ) of any member is given by @xmath25 where @xmath26 is the proper motion in arcsecs per year .
if the star is not a moving group member then the above formula does not apply .
our second test is to compare this moving cluster distance to the distance measured by parallax , @xmath27 .
once again 1.5 times the velocity dispersion leading to a 28% error compared to the parallax distance seems to cover all the members we find . as in the first test
, we estimate the random chance of a star passing this test . using the 70 dwarfs with parallaxes , minus the 4 dwarfs which we ultimately identify as umsmg members ( as discussed in the next section ) , we calculate @xmath28 , and find that 9 dwarfs have @xmath29 i.e. within 1.5 @xmath19 , or 28% .
if 9 out of 66 dwarfs pass this test by chance , the probability is @xmath30 . a similar test for the hmg yields @xmath31 .
we adopt this higher probability for both umsmg and hmg to avoid over - estimating the significance of the test outcomes .
finally we calculate the absolute magnitude at any wavelength using the parallax , and our last test is to place the objects in a colour - absolute magnitude diagram .
this third check requires that the object lies in a `` correct '' or sensible position in the colour - magnitude diagram . by
that we mean that there is some evident sequence .
we do not require that the objects fit the theoretical isochrones ( see point ( v ) under section 6 ) .
the entire l dwarf sequence for the 70 field stars is approximately 3.5 magnitudes wide , a factor 25 in intensity . allowing for binaries , an isochrone can vary in intensity at any colour by a factor 2 .
this gives @xmath32 as the random chance of passing the third test .
thus the total probability of passing all three independent tests by chance is @xmath33 , suggesting that passing all three tests gives 99.50% confidence of membership .
the dwarf archive has some 459 entries but unfortunately only 70 of these have measured proper motions .
those with proper motions also have accurate parallaxes .
of the 70 objects in the archives with proper motions we find 4 to be members of the umsmg .
three are l dwarfs and there is one t dwarf .
table [ tab : member_basic ] lists their spectral type , magnitude , and distance as determined from the moving cluster and parallax methods .
also in table [ tab : member_basic ] we give @xmath34 , the difference between the convergent point direction and measured proper motion direction .
as mentioned above , due to velocity dispersion and errors in the moving group we do not expect @xmath35 to be zero or @xmath36 .
velocity dispersion dominates over measurement errors .
as can be seen from table [ tab : member_basic ] , @xmath35 varies from @xmath37 to @xmath38 and @xmath28 differs from unity by 1% to 18% .
with these two parameters the group members effectively pick themselves . thus if @xmath35 is allowed to increase above @xmath39 to say @xmath40 , no candidates have @xmath28 close to unity .
however , in the interests of scientific integrity , we point out that a fifth star , 2m1228 - 15 , passed the first two tests and apparently passed the third .
however 2m1228 - 15 is a known near - equal mass binary @xcite .
thus its two components lie @xmath41 mag below their combined magnitude and do not fit the umsmg sequence , hence failing the third test .
this does not fit very well with our estimate of @xmath42% chance of passing the first two tests by random chance .
for the hmg we find that 5 l dwarfs and 2 t dwarfs pass all three tests . again relaxing the constraints on @xmath34 and
@xmath43 would find no further members .
however , 2m0205 - 1159 ( also known as denis - p j020529.0 - 115925 ) was found to be a binary by @xcite who measured @xmath44-band flux ratios of @xmath45 and @xmath46 .
more recently @xcite claim it is a triple system with i magnitudes of 17.30 , 18.38 and 18.80 , and spectral types l5.5 , l8 and t0 .
these parameters suggest that the primary would have more than half of the k flux and so should not be moved down 0.75 mag in the colour - magnitude diagram .
we mark 2 m 0205 - 1159 with a downward - pointing arrow in figs [ fig : mk_jk ] and [ fig : mk_hk ] ( later ) and regard its membership of the hmg as uncertain .
figure [ fig : hyades_stream ] shows the location of the hmg group in galactic radial direction ( x axis ) and perpendicular to the plane ( y axis ) .
the sun is at ( 0,0 ) .
the cluster is obvious and most of the moving group members form a stream in approximately the galactic anticentre direction , but with a few in front of the cluster .
the 7 dwarf members are shown as asterisks .
the uncertain binary member , 2m0205 , has the most negative distance perpendicular to the plane , and is thus at the extreme end of the group .
this might be considered as further evidence of its non - membership .
the stream towards the galactic centre looks very similar to , but shorter than , that away from the galactic centre .
the reason for the short length of the forward stream is no doubt because most surveys for hyads have been conducted in the general direction of the hyades .
lccccrrr2mass i d & ir spectral & m@xmath47 & m@xmath48 & m@xmath49 & d@xmath50 & d@xmath51 & @xmath34 + & type & & & & ( pc ) & ( pc ) & ( @xmath52 ) + * umsmg * + 2 m j02431371 - 2453298 & t6 & 15.381 @xmath53 0.050 & 15.137 @xmath53 0.109 & 15.216 @xmath53 0.168 & 10.7 @xmath53 0.4 & 10.6 @xmath53 0.5 & 1.64 + 2 m j03454316 + 2540233 & l1 @xmath53 1 & 13.997 @xmath53 0.027 & 13.211 @xmath53 0.030 & 12.672 @xmath53 0.024 & 27.0 @xmath53 0.4 & 31.7 @xmath53 1.2 & 5.62 + 2 m j14460061 + 0024519 & l6 & 15.894 @xmath53 0.082 & 14.514 @xmath53 0.035 & 13.935 @xmath53 0.053 & 22.0 @xmath53 1.5 & 19.7 @xmath53 1.5 & 13.54 + 2 m
j15232263 + 3014562@xmath54 & l8 & 16.056 @xmath53 0.099 & 14.928 @xmath53 0.081 & 14.348 @xmath53 0.067 & 18.6 @xmath53 0.4 & 17.1
@xmath53 1.1 & 11.70 + + * hmg * + 2 m j16241436 + 0029158 & t6 & 15.494 @xmath53 0.054 & 15.524 @xmath53 0.100 & 15.518 @xmath53 0.000 & 11.0 @xmath53 0.1 & 13.9@xmath53 0.2 & 14.95 + 2 m j0036159 + 182110 & l4 @xmath53 1 & 12.466 @xmath53 0.027 & 11.588 @xmath53 0.029 & 11.058 @xmath53 0.021 & 8.8 @xmath53 0.1 & 10.8 @xmath53 0.1 & 1.48 + 2
m j00325937 + 1410371 & l8 & 16.830 @xmath53 0.169 & 15.648 @xmath53 0.142 & 14.946 @xmath53 0.109 & 33.2 @xmath53 6.9 & 35.4 @xmath53 1.2 & 0.98 + 2 m j0205293 - 115930 & l5.5 @xmath53 2 & 14.587 @xmath53 0.030 & 13.568 @xmath53 0.037 & 12.998 @xmath53 0.030 & 19.8 @xmath53 0.6 & 20.8 @xmath53 0.2 & 5.21 + 2 m j01075242 + 0041563 & l5.5 & 15.824 @xmath53 0.058 & 14.512 @xmath53 0.039 & 13.709 @xmath53 0.044 & 15.6 @xmath53 1.2 & 15.2 @xmath53 0.3 & 1.10 + 2 m j1217110 - 031113 & t7.5 & 15.860 @xmath53 0.061 & 15.748 @xmath53 0.119 & 15.887 @xmath53 0.000 & 11.0 @xmath53 0.3 & 9.2 @xmath53 0.1 & 2.86 + 2 m j0825196 + 211552 & l6 & 15.100 @xmath53 0.034 & 13.792 @xmath53 0.032 & 13.028 @xmath53 0.026 & 10.7 @xmath53 0.1 & 8.7 @xmath53 0.1 & 6.38 + +
_ @xcite list the effective temperature of this star as @xmath55 k , with log @xmath56 in the range 4.8 - 5.1 and an age of between 0.4 - 1.7 gyr .
this age range just fits to our adopted age for the umsmg . + _ 2m1523 . _
also known as gl584c , this star , which we include as a member of the umsmg , is considered extensively by @xcite who estimate its age to be between 1.0 and 2.5 gyr . this age is the average of several methods which have a total range of 0.3 to 2.5 gyr , and thus encompasses the umsmg age of 400 myr .
+ _ 2m1624 . _ @xcite list the effective temperature of this star as @xmath57 k , with log @xmath56 in the range 5.3 - 5.4 and an age of between 4.3 - 5.8 gyr .
this age is in clear conflict with the hmg age of 625 myr .
the method used by burgasser et al is to find @xmath56 and @xmath58 from spectral indices , and compare these values with the model of @xcite which yields masses and ages directly .
they also use measured luminosities to obtain masses and radii and then the models again to find the ages .
this alternative method gives an age of 0.6 to 10 gyr , just consistent with the hyades age .
+ _ 2m0036 .
_ @xcite present a study of the magnetic properties and summarise current research on this object , citing @xmath59 k @xcite , log @xmath60 @xcite , and an inferred age of at least 1 gyr from the work of @xcite .
+ _ 2m0205 .
_ this object is a known binary @xcite ; @xcite assume an age `` greater 0.5 gyr '' but in later work , @xcite present evidence to suggest that 2m0205 is possibly a triple system and they assume an age of between 1 and 10 gyr .
+ _ 2m1217 .
_ @xcite suggest the possibility of a faint companion to this object in hubble wfpc-2 data .
however , the putative companion is close to the detection limits of the image .
[ fig : mk_jk ] plots the @xmath61 , @xmath62 colour magnitude diagram for both the umsmg and the hmg members . also shown
are the 60 other l and t dwarfs from the archive with known parallaxes ( 2m1228 - 15 a and b are plotted ) .
these show a rather scattered distribution which is to be expected since they presumably have a range of ages .
in addition we have plotted the 500 myr dusty model of @xcite and the same age cond models ( @xcite ) .
we draw the following conclusions:- 1 .
the 5 hmg l dwarfs sit on a very tight sequence .
this suggests that coeval l dwarfs , unlike field l dwarfs , form a well defined sequence .
we presume that the scattered nature of the field l dwarfs is therefore caused by their variety of ages or gravities , and possibly also metallicities .
two of the umsmg l dwarfs , if joined by a straight line , sit on a sequence that is nearly parallel and about 0.4 mag above the hmg sequence .
this is as expected since the umsmg is younger than the hmg .
any l dwarf sequence must have a turning point beyond which @xmath62 is decreasing towards the t dwarfs .
the hmg sequence reaches @xmath63 , close to the maximum @xmath64 @xcite and must therefore be close to or have reached this turning point . the umsmg l8 dwarf 2m1523 + 3014 must be on the blueward arm of the l dwarf sequence . the hmg l6 dwarf 2m0825 + 2115 might also just be on the blueward arm of the sequence .
none of the l dwarfs except the uncertain 2m0205 and discarded 2m1228 appear to be low mass ratio binaries , otherwise they would sit high on the sequence .
the only possible exception might be 2m1523 + 3014 which is alone on the blueward sequence ; however , given its locus on the lower envelope of the field stars we think it is unlikely to be a binary .
a possible reason for the lack of near - equal mass binaries is that these are of course more massive entities and therefore less likely to have been ejected from their parent clusters .
the 3 t dwarfs appear to form a short sequence that corresponds to an approximate 500 myr isochrone . at its red end
this lies below the cond model @xcite .
+ we have found 11 out of 70 field dwarfs to be members of the umsmg and hmg .
this is a high percentage , nearly 16% .
this of course implies that if we had proper motions and parallaxes for all 459 dwarfs in the dwarf archive , we would have found @xmath65 , 26 umsmg and 46 hmg members . for the hyades this would certainly increase the mass function ( see @xcite ) , but not ridiculously so given that the hyades cluster is known to be deficient in low mass stars and brown dwarfs . + that a significant percentage of field dwarfs belong to local moving groups may be due to very old ( age @xmath66 gyr ) dwarfs being very cool , faint and therefore difficult to detect and
thus under - represented in the dwarf archive .
it can be seen that the dusty 500 myr isochrone is not a good fit to either umsmg ( age 400 myr ) or the hmg ( age 625 myr ) .
this is perhaps not suprising given the theoretical difficulties of modelling dusty atmospheres .
@xcite do not give theoretical isochrones .
the cond model for t dwarfs fits the hmg at @xmath67 very well , but is not so good for the somewhat redder umsmg t dwarf .
figure [ fig : mk_hk ] shows the @xmath61 , @xmath68 colour magnitude diagram .
again both moving groups have a well defined sequence in the l regime , but the umsmg is almost vertical , whereas the hmg covers 0.3 in @xmath68 colour . of course , 0.3 is not a great range in colour and it is already well known ( e.g. @xcite ) that the field l dwarfs have a small range in @xmath68 colour . given that both moving groups have a similar age and therefore similar gravities , the most likely explanation for the difference is metallicity .
the hyades has fe / h = + 0.13 relative to the sun , while the umsmg has -0.08 @xcite .
@xcite have theoretical models of l and t dwarfs and do indeed predict that l dwarfs will be redder in @xmath62 with increased metallicity .
unfortunately they do not consider the @xmath68 dependence on metallicity .
so far we have implied that the membership of a moving group guarantees objects have the same age .
this is not necessarily true . moving groups may arise from a dispersing cluster or from a star formation event in a particular region of a molecular cloud
; in either case , the members of the group will be coeval .
alternatively a moving group may be the consequence of a dynamical process where for example the galactic bar drives some resonance to produce a group of stars with a common velocity . in this case
the stars will not be coeval ( see , for example , @xcite ) .
the pleiades moving group , sometimes called the local group , and other groups have stars of differing ages @xcite .
also , the core stars of the umsmg clearly fit a good sequence in the hr diagram and are coeval @xcite , but sirius which , on dynamical grounds is a good member , probably does not have the correct age @xcite .
resonance driven stars tend to favour particular @xmath69 ( galactic azimuthal direction ) velocities .
these favoured velocities coincide with both the umsmg and hmg @xmath69 velocities ( see for example @xcite or @xcite ) .
thus there is a greater than random chance that stars will have proper motions pointing to the convergent point of these groups .
this effect is rather difficult to quantify over the whole sky , so we have added an estimated factor of 4 in section 2 for calculating the chance of a proper motion being directed towards the umsmg or hmg s convergent point .
however if an object has the umsmg or hmg velocity by virtue of a dynamical resonance rather than from being a genuine member of the group , its moving group distance ( see section 2 ) will not be the same as its parallax distance and it will fail our second test . perhaps fortuitously
all our objects pass the second test .
of course our third test , that the dwarfs fit a ` sensible ' sequence , should select coeval objects .
however , we do not know exactly where this sequence is and the skeptic might argue that we have simply got the wrong sequence .
indeed with 2m1228 - 15 the first two tests produced an object that did not fit the umsmg sequence ( see above ) .
on the positive side the 5 hmg l dwarfs do seem to form a very good sequence , and the 2 hmg t dwarfs have very similar absolute k magnitudes .
the results for the umsmg are not so compelling , but 2 l dwarfs lie on a line parallel and above the hyades sequence as expected for a younger group .
the results described in this paper should perhaps be treated with some caution . nevertheless , the kinematic data exist and should be used . as more proper motion and
parallax data become available these and other sequences may become better established .
summarising , we can say that some moving group members are undoubtedly coeval , but membership of a moving group based on the criteria described in this paper does not guarantee coevality .
we find one t dwarf and 3 l dwarfs that belong to the umsmg , whose age is 400 myr .
we find a further 2 t dwarfs and 5 l dwarfs , one rather dubious , members of the hmg .
these stars provide preliminary empirical isochrones for these ages .
we plan to extend this technique to other moving groups .
only 70 of the 459 archived dwarfs have the proper motions and parallaxes needed to identify them with moving groups .
we believe that many more l and t dwarfs could be identified with moving groups if more proper motions and parallaxes were available .
this would allow them to be assigned an age , although it should be remembered that moving group ages can never be totally secure .
this research has benefitted from the m , l , and t dwarf compendium housed at dwarfarchives.org and maintained by chris gelino , davy kirkpatrick , and adam burgasser .
the research 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 .
we are grateful to the referee , m. bessel , for his helpful comments which improved this paper . 99 baraffe , i. , chabrier , g. , allard , f. & hauschildt , p. h. : 1998 , a&a 337 , 403 baraffe , i. , chabrier , g. , allard , f. & hauschildt , p. h. : 2002 , a&a 382 , 563 baraffe , i. , chabrier , g. , barman , t. s. , allard , f. & hauschildt , p. h. : 2003 , a&a 402 , 701 , 2002 , mnras , 334 , 193 chabrier , g. , baraffe , i. , allard , f. & hauschildt , p. h. , 2000 , apj 542 , 464 , g. , lagrange , a .- m . , dumas , c. , zuckerman , b. , mouillet , d. , song , i. , beuzit , j .- l . , lowrance , p. , 2004
, aa , 425 , l29 | we have used the moving cluster method to identify three l dwarfs and one t dwarf in the ursa major / sirius moving group ( age 400 myr ) .
five l dwarfs and two t dwarfs are found to belong to the hyades moving group ( age 625 myr ) .
these l and t dwarfs define 400- and 625- myr empirical isochrones , assuming that they have the same age . moving group membership does not guarantee coevality .
[ firstpage ] stars : kinematics stars : low mass , brown dwarfs open clusters and associations : individual : ursa major open clusters and associations : individual : hyades galaxies : star clusters |
quantum key distribution ( qkd ) @xcite allows two parties , alice and bob , to establish a common and secret key @xmath0 that is informationally secure ; see @xcite and references therein .
a widely used setup for qkd is the one suggested by bennett and brassard ( bb84 ) @xcite .
bb84 is ideally implemented by preparing and transmitting single - photon pulses .
information can be encoded in the state of one of two conjugate polarization bases , e.g. vertical / horizontal or diagonal / antidiagonal .
only those photons that were prepared by alice and detected by bob in the same basis are useful to build a sifted key , which forms @xmath0 after additional steps of information reconciliation and privacy amplification .
security follows from the inability of faithfully copying quantum information @xcite and the unavoidable information - disturbance trade - off in quantum mechanics .
nevertheless , realistic implementations of bb84 use weak coherent photon pulses that could involve many photons , avoiding the assumptions made in security analyses @xcite . such pulses could be exploited by eve , the eavesdropper , to gain access to the ( insecure ) distributed key using a so - called photon - number splitting ( pns ) attack @xcite . in a simple proposed pns attack , eve measures the number of photons in the pulse , @xmath1 . if @xmath2 , eve blocks the pulse . if @xmath3 , eve `` splits '' the pulse to obtain a copy of a single photon with the correct polarization and keeps it in her quantum memory .
eve could then obtain a full copy of @xmath0 by making measurements of her photons in the correct polarization bases , which are known after a public discussion between alice and bob . since alice and bob can not measure @xmath1
, a pns attack may go undetected .
our goal is to provide a protocol for secure qkd in the presence of pns attacks . a simple approach to overcome a pns attack considers reducing the probability of multi - photon pulses by reducing the coherent - pulse intensities .
the drawback with this approach is that the probability of creating single - photon pulses is also reduced .
then , the rate at which bits to build @xmath0 are sifted is far from optimal @xcite .
another approach is to use decoy states , that allow to detect pns attacks without a substantial reduction on the rate of sifted bits if eve is not present @xcite . in a decoy - state protocol ( dsp ) , one of several weak coherent sources
is randomly selected for each pulse .
such sources create pulses of different intensities ( mean photon numbers ) .
this gives alice and bob a means to estimate @xmath4 and @xmath5 , the number of bob s detections due to empty and single - photon pulses prepared by alice , in the same basis , respectively .
the values of @xmath4 and @xmath5 are important to determine @xmath6 , the length of the secure key .
for example , in the discussed pns attack , @xmath5 is substantially smaller than its value when eve is not present , and so is @xmath6 . in more detail , we let @xmath7 be the total number of pulses prepared by alice .
we first assume that the channel is non - adversarial , i.e. no eavesdropping attacks are present .
if the pulse has a random phase , the number of photons it contains is sampled according to the poisson distribution : @xmath8 where @xmath9 is the mean photon number for that source and @xmath10 in applications .
we let @xmath11 be the transmission / detection efficiency of the quantum channel shared by alice and bob .
if @xmath12 ( @xmath13 ) denotes the event in which bob detects a non - empty ( empty or vacuum ) pulse , @xmath14 is the probability of a detection by bob given that alice s prepared pulse contained @xmath1 photons .
@xmath15 is the so - called @xmath1-photon yield and @xmath16 due to losses in the channel . for @xmath17
, we may assume @xmath18 which is a good approximation in applications .
for @xmath19 , @xmath20 denotes bob s detector dark - count rate .
the total probability of bob detecting a pulse ( in any one cycle ) is the total yield @xmath21 @xmath22 can be estimated by alice and bob , via public discussion , from the frequency of detections after all pulses were transmitted . in qkd , we allow eve to manipulate the parameters that characterize the channel at her will .
we use the superscript @xmath23 to represent the interaction of eve with the communication .
for example , @xmath24 denotes the @xmath1-photon yield in the presence of eve . in a general intercept - resend attack ,
eve may intercept a pulse and resend a different one .
that is , each detection by bob is not guaranteed to come from the same pulse that alice prepared . in a simple pns attack ,
eve makes non - demolition measurements of @xmath1 . with this information ,
eve sets @xmath25 and @xmath26 for @xmath3 , so that @xmath27 then , if alice and bob can only estimate the total yields , a pns attack could be `` invisible '' with the right choices of @xmath28 . to increase the multi - photon yield ,
eve may use an ideal channel to resend the pulses .
( note that sophisticated pns attacks that do not change the poisson distribution are possible @xcite . )
a pns attack allows eve to have the full key @xmath0 if @xmath29 in this case , eve possesses a photon with the same polarization as that of the pulse detected by bob and no single - photon pulses are involved in creating @xmath0 . only if @xmath30 some security guarantees are possible @xcite .
such an inequality is satisfied when @xmath31 , implying a rate for sifted bits of order @xmath32 [ eq . ] .
this is undesirably small ( @xmath33 ) .
remarkably , dsps give an optimal rate of order @xmath11 with small resource overheads . a goal in a dsp
is to estimate @xmath34 and @xmath35 , which provide a lower bound on @xmath36 and @xmath37 , respectively .
empty and single - photon pulses can not be split and the information carried in their polarization can not be faithfully copied , making them useful to create a secure key . for the estimation ,
alice uses photon sources with different values of @xmath9 , but are identical otherwise . in a conventional dsp
, it is assumed that eve s pns attack treats every @xmath1-photon pulse equally and independently , regardless of its source .
that is , eve s attack is simulated by independent and identically distributed ( i.i.d . )
random variables .
the total yield in this case is , for any given @xmath9 , @xmath38 equation describes mathematically what we denote as the i.i.d . assumption .
it follows that @xmath39 |_{\mu=0 } \ ; .\end{aligned}\ ] ] then , if eve s attack satisfies @xmath40 for all @xmath9 , @xmath41 that is , by being able to estimate @xmath42 for two values of @xmath43 via public discussion , alice and bob can restrict eve s attack so that the dark - count rate and single - photon yield are almost unchanged from the non - adversarial case .
in addition , if a third source with @xmath44 is randomly invoked , an optimal key rate of order @xmath11 will be achieved . in reality , the estimation of @xmath45 and @xmath46 is subject to finite statistics and can be technically involved .
nevertheless , the i.i.d .
assumption in eq .
allows alice and bob to gain information about eve s attack by running the protocol and analyzing the ( binomial ) distributions of the detection events for each source .
however , we remark that if eve were to correlate her attacks , the i.i.d . assumption and
the corresponding security analyses would be invalid .
this is the main motivation behind our analysis . in this paper ,
we give an example that shows how the i.i.d .
assumption can be simply bypassed by eve , resulting in security parameters that are worse from those obtained under the assumption .
we then analyze the security of dsps for general pns attacks .
our main result is an estimation procedure that gives a lower bound on @xmath36 and @xmath37 , with a confidence level that is an input to the estimation procedure .
our security analysis does not use the i.i.d .
assumption and is particularly relevant when eve performs a pns attack that could correlate different pulses in one session or even different sessions .
we compare some results obtained by our estimation procedure with those obtained by using the i.i.d .
assumption , and emphasize the important of our procedure .
of high significance in cryptographic protocols is @xmath47 , the so - called security parameter .
@xmath47 measures the deviation of a real protocol implementation from an ideal one .
we use the same definition used in ref .
@xcite , that states that a real qkd protocol is @xmath47-secure if it is @xmath47-indistinguishable from a perfectly secure and ideal one .
this definition is equivalent to a statement on the trace norm of the difference between the quantum states resulting from the real and ideal protocol , respectively .
it implies that a qkd protocol that is @xmath47-secure can be safely reused order @xmath48 times without compromising its security .
usually , one fixes a value for @xmath47 and then determines the size of @xmath0 based on several protocol performance parameters .
these parameters include the number of pulses sent by alice , the number of pulses detected by bob , and the estimated bit error rates at each mean photon number . for dsps
, @xmath47 has a component @xmath49 that determines the confidence level in the estimation of a lower bound of @xmath36 and @xmath37 , due to finite statistics . a possible way to obtain such lower bounds , under the i.i.d .
assumption , is the one followed in ref @xcite . in this case , we consider a dsp with three sources , @xmath50 . the mean photon number in each pulse , for each source , is @xmath51 , @xmath52 , and @xmath53 .
each source @xmath54 randomly prepares a pulse with probability @xmath55 and we let @xmath56 be the total number of pulses for that source . @xmath56
is known to alice and bob by public discussion after all pulses are sent and @xmath57 when @xmath7 .
we write @xmath58 for the random variable that counts the number of pulses from source @xmath54 detected by bob under the presence of eve @xcite .
the exact value that @xmath58 takes in a session can also be obtained by alice and bob via public discussion after the pulses were transmitted . under the i.i.d .
assumption [ eq .
] , @xmath58 is sampled according to the binomial distribution .
then , @xmath59 is an estimator of the total yield @xmath60 $ ] , where @xmath61 $ ] denotes the mean value .
that is , for a given @xmath62 , we can establish confidence intervals @xmath63 with confidence level @xmath64 . the constant @xmath65 depends on @xmath62 and can be obtained using chernoff s bound @xcite see appendix [ app : chernoff ] . the standard deviation in this case is @xmath66 using eq . for @xmath67 , we can search for the minimum values of @xmath45 and @xmath46 that satisfy eqs .
, e.g. by executing a linear program .
both @xmath45 and @xmath46 can then be used to obtain the desired lower bounds on @xmath36 and @xmath37 , respectively , with corresponding confidence level @xmath68 .
this last step also requires using the i.i.d . assumption .
we remark that eq .
does not properly regard the problem of inferring a distribution for @xmath69 from the known @xmath70 , a problem that would require knowledge on the prior distribution of @xmath69 .
the analysis in sec .
[ sec : decoys ] used the i.i.d .
assumption that resulted in a value for @xmath71 given by eq . .
nevertheless , the actual value of @xmath71 could be much higher in more general pns attacks .
for the same confidence level , a bigger @xmath71 implies a `` wider '' confidence interval for the estimation of the yield @xmath69 ( appendix [ app : chernoff ] ) , and thus smaller lower bounds on @xmath36 and @xmath37 . the overall result in
the dsp is a secret key @xmath0 of smaller size for the same security parameter . to illustrate how eve can bypass the i.i.d .
assumption , we suggest a potential attack that results in almost no change for the total yields ( i.e. , @xmath72 ) @xcite but the variances @xmath71 are increased with respect to those of the binomial distribution [ eq . ] .
the suggested attack could be detected by alice and bob by estimating the variances directly via public discussion .
nevertheless , it still shows that a better analysis of the security of dsps is needed to make rigorous claims .
in the attack , eve first picks an integer value for @xmath73 , where @xmath74 denotes a scale for a variance or `` correlation '' of a particular distribution .
eve receives all pulses from alice and we let @xmath75 be the total number of @xmath1-photon pulses in the protocol . note that the exact value of @xmath75 is known to eve but not to alice and bob . in general , @xmath75 is sampled according to the binomial distribution @xmath76 where @xmath77 is the probability of a pulse containing @xmath1 photons : @xmath78 .
the mean and variance for such distribution are @xmath79 & = p_n k \ ; , \\
\nonumber \sigma^2_{k_n } & = p_n ( 1-p_n ) k \ ; .\end{aligned}\ ] ] given @xmath75 , eve randomly picks a value for @xmath80 , where @xmath81 is the total number of detections due to @xmath1-photon pulses prepared by alice .
in particular , we assume that eve can control @xmath82 , which determines the dark - count rate .
the distribution associated with @xmath83 has the following properties : @xmath84 & = y_n k_n \ ; , \\
\nonumber \sigma^2_{d^{\ce}_n|k_n } & = \tau^2 y_n ( 1-y_n ) k_n \ ; .\end{aligned}\ ] ] we let @xmath85 be the number of @xmath1-photon pulses , prepared by alice s @xmath54th source only , and detected by bob .
the exact value of @xmath85 is unknown to all parties . because eve does not know the source being used in the dsp , @xmath85 is sampled according to the binomial distribution when given @xmath86 : @xmath87 where @xmath88 the distribution associated with @xmath85 satisfies @xmath89 & = q_n^i d_n^{\ce } \ ; , \\
\nonumber \sigma^2_{d_n^{i,\ce}|d_n^\ce } & = q_n^i ( 1-q_n^i ) q_n^\ce \ ; .\end{aligned}\ ] ] as in sec .
[ sec : decoys ] , we let @xmath90 be the variance associated with the random variable @xmath91 , where @xmath92 and @xmath93=y^\ce(\mu^i)$ ] . an accurate estimate of @xmath94 can be obtained if we approximate @xmath95 , in the limit of large @xmath96 .
in addition , because @xmath96 is fixed , the variables @xmath75 are not independent .
however , in the large-@xmath96 limit , @xmath75 can also be approximated by its mean value .
it implies that the @xmath75 are almost independent and so are the @xmath86 and @xmath85 for different values of @xmath1 . under these approximations ,
@xmath97 in appendix [ app : attackstatistics ] we show that @xmath98 q_n^i y_n p_n k \ ; .\end{aligned}\ ] ] by inserting eq . in eq .
, we can obtain the variances as a function of @xmath99 . in fig .
[ fig : attack1 ] we compute @xmath100 and @xmath101 .
the i.i.d . assumption discussed in sec .
[ sec : decoys ] corresponds to @xmath102 see appendix [ app : attackstatistics ]
. using these results in eq .
yields wider confidence intervals for the same confidence level . and
@xmath101 for an attack in which eve correlates @xmath1-photon pulses according to the value of @xmath99 .
the channel parameters are @xmath103 , @xmath104 , @xmath105 , @xmath106 , @xmath51 , @xmath107 , @xmath108 , and @xmath109 @xcite . the results in sec .
[ sec : decoys ] are recovered for @xmath102.,width=283 ] to illustrate our point further , we consider a simple attack in which a single source @xmath110 is used to estimate the dark - count rate . here , @xmath51 and @xmath111 is known . in the non - adversarial case
, @xmath112 is sampled according to the binomial distribution with probability @xmath113 and known sample size @xmath114 .
nevertheless , for the correlated attack , we assume that eve `` receives '' the @xmath115 pulses and groups them according to blocks of size @xmath74 .
then , eve will force ( prevent ) the detection of all pulses in any one block with probability @xmath113 ( @xmath116 ) .
the random variable @xmath112 for the correlated attack satisfies @xmath117 & = y_0 k_0 \ ; , \\
\nonumber \sigma^2_{d_0^\ce } ( \tau ) & = [ y_0 ( \tau^2)^2 - ( y_0 \tau^2)^2]\frac{k_0}{\tau^2 } = \tau^2 y_0 ( 1-y_0 ) k_0 \ ; , \end{aligned}\ ] ] and @xmath102 corresponds again to the i.i.d .
assumption [ see eq . ] . in fig .
[ fig : darkcount ] ( a ) , we plot the probability that @xmath118 satisfies @xmath119 + c \sigma_{d_0^\ce } ( 1 ) \ge z^{u,\ce } \ge e[z^{u,\ce } ] - c \sigma_{d_0^\ce } ( 1 ) \ ; , \end{aligned}\ ] ] for different values of @xmath65 and @xmath99 . for @xmath102
, such a probability corresponds to the confidence level in eq . .
@xmath120=y_0 $ ] in this example . for the inverse problem , namely the estimation of @xmath113 from @xmath121 and @xmath115 , eq . may be incorrect .
we may then assume a uniform prior distribution for @xmath122 $ ] , and obtain the posterior distribution as @xmath123 which is plotted in fig .
[ fig : darkcount ] ( b ) .
our results demonstrate that , for a fixed security parameter , the accuracy in the estimation of the dark - count rate strongly depends on eve s attack and can be substantially different from the one obtained under the i.i.d .
assumption ( @xmath102 ) . , and confidence bounds , parametrized by @xmath65 .
( b ) bayesian estimation of @xmath113 , the mean dark - count rate , assuming a uniform prior and for different correlated attacks [ eq . ] .,width=294 ]
we go beyond the i.i.d . assumption and study more general and correlated pns attacks , in which eve has full control on bob s detection events .
the secure key - rate in a realistic implementation of qkd is @xcite @xmath124 which determines the size of the distributed key as @xmath125 .
@xmath126 is the total number of pulses detected by bob and prepared by alice in the same basis , that are useful for the sifted key . in bb84 , @xmath127 , where @xmath128 is the total number of detections .
@xmath129 is a lower bound on @xmath130 , the number of @xmath1-photon pulses prepared and detected in the same basis .
@xmath131 is the shannon entropy , @xmath132 and @xmath133 are coefficients due to the error correction and privacy amplification steps , ber is the total bit error rate , and @xmath134 is an upper bound to the bit error rate due to single - photon pulses only . in a dsp , we characterize a general pns attack by the distribution @xmath135 see fig .
[ fig : interceptresend1 ] for an example .
our goal is to build an estimation procedure that places confidence intervals on @xmath136 and @xmath137 from the known @xmath58 .
these intervals ultimately imply a lower bound on @xmath138 see eq . . , @xmath139 , and @xmath140
each block represents the number of pulses with @xmath141 , respectively .
the random variables @xmath75 indicate the number of @xmath1-photon pulses prepared by alice and the superscript @xmath54 denotes the source used for such pulses .
eve s attack controls the number of detections by bob , due to @xmath1-photon pulses , through @xmath86 .
, width=294 ] we assume that there are three sources satisfying @xmath142 , and @xmath143 .
nevertheless , our analysis can be easily generalized to the case in which more sources are present , where the estimation is more accurate . for each source ,
bob s detections satisfy eq . .
if a simple relationship between each @xmath85 and @xmath86 could be found , we could execute a program to solve eqs . .
such a relationship could be obtained from the binomial distribution associated with @xmath85 , when given @xmath144 [ eq . ] .
our estimation procedure uses @xmath145 to determine the confidence intervals @xmath146 the corresponding confidence level for each inequality is @xmath147 .
the upper and lower bounds are monotonic and invertible functions
. then , @xmath148 with the same confidence levels .
such confidence levels do not result from the binomial distribution as we are analyzing the inverse problem , namely the estimation of @xmath149 from the available information ( i.e. , @xmath58 and @xmath56 ) . from eqs . and
, we obtain @xmath150 see fig .
[ fig : boundsfig ] for an example . .
the yellow blocks represent the number of pulses from source @xmath151 with @xmath19 and @xmath2 .
the dark and blue blocks represent the total number of pulses with @xmath19 and @xmath2 , respectively .
the confidence level for this case is not smaller than @xmath152.,width=294 ] next , our estimation procedure executes a program to obtain @xmath153 and @xmath154 , the corresponding smallest values of @xmath112 and @xmath155 , subject to the constraints given by eqs . .
from the union bound , the confidence level in such values is @xmath156 , with @xmath157 when three sources are used .
since @xmath158 and @xmath159 are sampled according to a binomial distribution when given @xmath126 ( i.e. , the preparation and detection basis are random ) , we obtain @xmath160 where the constant @xmath161 can be obtained using eq . .
the overall confidence level for the key rate @xmath138 is @xmath68 , where the security parameter satisfies @xmath162 in the next section we obtain the confidence intervals and levels specifically for our method .
our method takes @xmath49 as input and outputs @xmath158 and @xmath159 . to satisfy eq . , we can set @xmath163 and @xmath164 [ see eqs . and ] .
next , we will find @xmath165 and @xmath166 as required by eq . .
if @xmath167 depends on @xmath85 only , the probability that @xmath86 is smaller than @xmath167 is @xmath168 with @xmath169 when given @xmath149 , the random variable @xmath85 is sampled according to eq . .
from chernoff s bound ( appendix [ app : chernoff ] ) @xmath170 and we choose the lower bound so that @xmath171 } \ ; , \end{aligned}\ ] ] with @xmath172 . the error probability satisfies
@xmath173 see appendix [ app : errors ] . a similar analysis gives the upper bound @xmath174 } \ ; , \end{aligned}\ ] ] with the same confidence level .
then , to satisfy eq . , it suffices to set @xmath175 to complete the estimation procedure , we invert eqs . and and obtain @xmath176 we can then execute a program that finds the minimum values of @xmath112 and @xmath155 subject to eqs . .
for instance , a quadratic program can be used to search @xmath177 .
such minimum values will be used in eqs . and to obtain the key rate .
a technical remark is in order . when @xmath178 , @xmath179 exponentially fast in @xmath1 .
then , the contribution of large-@xmath1 terms in eqs
. is negligible .
we can set a suitable cutoff @xmath180 in the number of photons per pulse in our analysis , to avoid unnecessary computational overheads in finding @xmath165 and @xmath166 , and with an insignificant impact in the estimated values .
we analyzed general photon - number splitting attacks and pointed out that previous security analyses on decoy - state protocols for qkd made a strong assumption on the attack .
we provided an estimation procedure that sets a lower bound on the size of the secure , distributed key , with the corresponding confidence levels .
our procedure requires executing a program to find the minimum values of the number of detections due to empty and single - photon pulses , subject to constraints that are determined by the results of the protocol and by the desired security parameter .
it results in rigorous security guarantees even if eve correlates her attack according to the number of photons in the pulse .
we emphasize that our estimation procedure is not unique : any time that a confidence interval can be set as a function of publicly available information for general attacks , then an estimation procedure is possible .
in addition , our choice of confidence intervals and @xmath181 is not essential and could be further optimized to improve the size of the secure key .
we thank jane nordholt , kevin mccabe , raymond newell , charles peterson , and stephanie wehner for discussions .
we thank the laboratory directed research and development ( ldrd ) program at los alamos national laboratory for support .
we let @xmath183 be a random variable and @xmath184 the probability distribution .
the random variable @xmath185 has the conditional distribution @xmath186 .
the probability of @xmath187 is @xmath188 .
then , it is easy to show @xmath189 + \sigma^2_{e[y|x ] } \ ; , \end{aligned}\ ] ] where @xmath190 is the variance of @xmath187 .
also , @xmath191 = \sum_{y=0}^k g(y|x ) y \ ; \end{aligned}\ ] ] is the expected value of @xmath187 when given @xmath192 , @xmath193 } = \sum_{x=0}^k f(x ) e[y|x ] - \left ( \sum_{x=0}^k f(x ) e[y|x ] \right)^2 \ ; \end{aligned}\ ] ] is the variance of @xmath194 $ ] , @xmath195 is the variance of @xmath187 when given @xmath192 , and @xmath196= \sum_{x=0}^k f(x ) \sigma^2_{y|x } \ ; \end{aligned}\ ] ] is the expected value of such a variance . in the attack discussed in sec . [
sec : attack ] , @xmath96 is fixed and the distribution of @xmath75 satisfies @xmath79=p_n k \ ; , \\ \nonumber \sigma^2_{k_n } = p_n ( 1-p_n ) k \ ; .\end{aligned}\ ] ] next , @xmath83 is chosen such that , when given @xmath75 , @xmath197 = y_n k_n \ ; , \\ \nonumber \sigma^2_{d_n^\ce|k_n } = \tau^2 y_n ( 1-y_n ) k_n \ ; .\end{aligned}\ ] ] it follows that @xmath198 = \tau^2 y_n ( 1-y_n ) p_n k \ ; , \\ \nonumber \sigma^2_{e[d_n^\ce|k_n]}= ( y_n)^2 \sigma^2_{k_n}= ( y_n)^2 p_n ( 1-p_n ) k \ ; .\end{aligned}\ ] ] then , @xmath199 when given @xmath200 , the distribution for @xmath85 satisfies @xmath89 = q_n^i d_n^{\ce } \ ; , \\ \nonumber \sigma^2_{d_n^{i,\ce}|d_n^{\ce } } = q_n^i ( 1-q_n^i ) d_n^{\ce } \ ; .\end{aligned}\ ] ] therefore , @xmath201 } = ( q_n^i)^2 \sigma^2_{d_n^{\ce } } \ ; , \\ \nonumber e[\sigma^2_{d_n^{i,\ce}|d_n^{\ce}}]= q_n^i ( 1-q_n^i ) y_n p_n k \ ; .\end{aligned}\ ] ] also , @xmath202 + \\ \nonumber & + q_n^i ( 1-q_n^i ) y_n p_n k \\ \nonumber & = [ ( \tau^2 - 1 ) q_n^i ( 1-y_n ) + ( 1-q_n^i y_n p_n ) ] q_n^i y_n p_n k \ ; .\end{aligned}\ ] ] the first term on the _ rhs _ vanishes when @xmath102 .
the second term is @xmath203 so that @xmath204 ^ 2 \ ; , \end{aligned}\ ] ] for @xmath102 .
moreover , since @xmath205 ^ 2 \ll y(\mu^i ) \ll 1 $ ] , then @xmath206 which shows that the case discussed in sec .
[ sec : decoys ] , i.e. the i.i.d .
assumption , corresponds to choosing @xmath102 in this case .
chernoff s bound @xcite sets a bound on the probabilities of `` rare '' events as a function of the standard deviation of the corresponding distribution .
more precisely , we let @xmath207 be a set of i.i.d .
random variables that satisfy @xmath208 and define @xmath209 .
a general version of chernoff s bound implies @xmath210 + c \sigma ] \le \exp \{- c^2/4 \ } \ ; , \end{aligned}\ ] ] where @xmath211 - ( e[x_j])^2}$ ] is the standard deviation . for the special case of the binomial distribution where @xmath212 with probability @xmath213 and @xmath214 otherwise , @xmath215 & = n a\ ; , \end{aligned}\ ] ] and @xmath216 & = i_{a } ( k , n - k+1 ) \\ \label{eq : chernoffboundbinomial } & \le \exp \{- ( k - n a ) ^2/(4 a ( 1-a ) n ) \ } \ ; .\end{aligned}\ ] ] here
, @xmath217 is the so - called regularized incomplete beta function . to satisfy @xmath218
\le \epsilon$ ] , it suffices to choose @xmath65 such that @xmath219 calculations of errors ---------------------- chernoff s bound for the binomial distribution [ eq . ]
implies that @xmath222 if we set , @xmath223 then @xmath224 as in eq . .
replacing @xmath225 by @xmath85 and @xmath149 by @xmath226 in eq . , and solving the resulting quadratic equation , we obtain @xmath227 / ( 2q_n^i ) \ ; .\end{aligned}\ ] ] that is , @xmath228 } } { 2 q_n^i } \ ; , \end{aligned}\ ] ] that yields eq . .
changing @xmath229 provides the upper bound without changing @xmath181 , i.e. , the confidence level . c. h. bennett and g. brassard .
quantum cryptography : public - key distribution and coin tossing . in proceedings of ieee international conference on computers , systems and signal processing , pages 175179 , 1984 .
d. mayers .
quantum key distribution and string oblivious transfer in noisy channels . in advances in cryptology
crypto 96 , volume 1109 of lecture notes in computer science , pages 343357 .
springer , 1996 .
our definition of @xmath58 does not imply that the same @xmath1-photon pulse created by alice is detected by bob , as eve may be replacing each pulse at her will .
@xmath58 only gives information about detections corresponding to those clock cycles in which alice prepared the pulse with source @xmath54 .
security will follow from the analysis on the bit error rates that will subtract those detections corrupted by eve as implied by eq . .
wassily hoeffding .
probability inequalities for sums of bounded random variables .
assoc . , 58 ( 301 ) : 13 , 1963 .
h. chernoff . a note on an inequality involving the normal distribution .
annals of probability , 9 : 533 , 1981 .
the exact values of @xmath230 are irrelevant as alice and bob can not have exact estimates of the channel parameters such as @xmath11 , in general . if eve changes these values slightly , such changes will not be noticed . | decoy - state protocols provide a way to defeat photon - number splitting attacks in quantum cryptography implemented with weak coherent pulses . we point out that previous security analyses of such protocols relied on assumptions about eavesdropping attacks that considered treating each pulse equally and independently .
we give an example to demonstrate that , without such assumptions , the security parameters of previous decoy - state implementations could be worse than the ones claimed .
next we consider more general photon - number splitting attacks , which correlate different pulses , and give an estimation procedure for the number of single photon signals with rigorous security statements .
the impact of our result is that previous analyses of the number of times a decoy - state quantum cryptographic system can be reused before it makes a weak key must be revised .
la - ur 13 - 20168 |
since the discovery of high-@xmath2 superconductivity in fe - based materials in 2008 , @xcite an enormous amount of effort has been invested to understand the microscopic behavior of these systems .
iron pnictides and chalcogenides become superconductors either by hole- or electron - doping the systems , by application of external pressure or by a combination of both . in particular , uniaxial pressure is currently being intensively discussed as a possible route towards modifying the structural , magnetic and even superconducting properties of these systems .
a regular sample below its magnetic and structural transition temperatures displays an equal number of opposite twin orthorhombic domains , effectively canceling out its anisotropic properties . to circumvent this issue and obtain a single orthorhombic domain sample ,
uniaxial tensile stress has been widely employed to detwin iron pnictides like bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 @xcite and unveil its anisotropic properties which have been argued to originate from electronic nematic degrees of freedom .
@xcite theoretically , although it is clear that in the tetragonal phase the applied uniaxial pressure acts as a conjugate field to the orthorhombic order parameter , condensing a single domain , @xcite the nature of the detwinning process deep inside the orthorhombic phase remains an open question , since different mechanisms might be at play such as twin boundary motion or reversal of the order parameter inside the domains .
@xcite besides promoting detwinning , uniaxial strain has also been shown to affect the thermodynamic properties of the iron pnictides .
recent neutron scattering experiments on bafe@xmath0as@xmath0 under compressive stress along the in - plane * b * direction reported a progressive shift to higher temperatures of the magnetic transition @xcite - a behavior also seen in @xmath3 by thermodynamic measurements .
@xcite and an apparent reduction of the magnetic moment @xcite .
moreover , blomberg _ et al .
_ observed a significant uniaxial structural distortion in bafe@xmath0as@xmath0 under tensile stress , suggesting an enhanced response to external strain .
@xcite more recently , it was found that epitaxially strained thin films of fese on a srtio@xmath4 substrate show an increase in critical superconducting temperatures up to 65 k , the highest reported @xmath2 so far .
@xcite clearly , crystal lattice strain plays a key role for the magnetic , structural and superconducting properties in fe - based superconductors and a better understanding of the microscopic origin of such behavior is desirable . in this work
we combine density functional theory ( dft ) calculations and ginzburg - landau phenomenology to analyze the effects of uniaxial compressive stress as well as uniaxial tensile stress on the magnetic , electronic and structural properties of bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 at low temperatures , deep inside the ordered phase .
stress is measured in terms of equivalent hydrostatic pressure , @xmath5 , where @xmath6 is the stress tensor matrix and positive and negative pressures correspond to applying compressive and tensile stresses respectively .
our _ ab initio_-derived estimates for the elastic constants in the orthorhombic phase agree well with experimental values .
while no sign of a true structural or magnetic phase transition is observed in the range of pressures between @xmath7 gpa and @xmath8 gpa , at a critical pressure we observe a reversal of the magnetization , _
i.e. _ exchange of ferromagnetic ( fm ) and antiferromagnetic ( afm ) directions , simultaneous to a discontinuous change in the orthorhombic order parameter @xmath9 , which also changes sign .
this behavior has important consequences on the orbital @xmath10 and @xmath11 occupancies and is also related to the shift of the magnetic ordering temperature , as we argue below .
furthermore , by employing a phenomenological ginzburg - landau model , we show that this behavior is intimately connected to the magneto - elastic coupling of the system , which by itself acts as an intrinsic conjugate field to the orthorhombic order parameter . as the applied compressive stress
is enhanced towards a critical value , it eventually overcomes the effects of the magneto - elastic coupling , rendering the zero - pressure state energetically unstable and resulting in a simultaneous reversal of the magnetization and the orthorhombic order parameter .
comparison of the dft - derived critical uniaxial pressures for cafe@xmath0as@xmath0 and bafe@xmath0as@xmath0 , combined with the ginzburg - landau result that the critical pressure is proportional to the magneto - elastic coupling , suggests that the latter is larger in cafe@xmath0as@xmath0 than in bafe@xmath0as@xmath0 .
we also propose low - temperature detwinning measurements to compare the experimental critical pressure with our _ ab initio _ estimates in order to clarify the dominant mechanism behind the detwinning process of orthorhombic iron pnictide crystals .
the paper is organized as follows : in section ii we summarize our computational methods , and in section iii we present our dft results .
section iv is devoted to the ginzburg - landau analysis and comparison with dft calculations .
section v contains the discussions and concluding remarks .
details about the _ ab initio _ method are presented in the appendix .
electronic structure calculations were performed within dft with the vienna ab initio simulations package ( vasp ) @xcite with the projector augmented wave ( paw ) basis @xcite in the generalized gradient approximation ( gga ) .
all our structural relaxations were performed under constant stress using the fast inertial relaxation engine ( fire ) .
@xcite for this purpose , we had to modify the algorithm accordingly ( see appendix a ) .
every 10 steps , we cycled through non - magnetic , ferromagnetic , antiferromagnetic - checkerboard , stripe - type antiferromagnetic ( along unit cell axis * a * ) and stripe - type antiferromagnetic ( along unit cell axis * b * ) spin configurations , and then we continued the relaxation with the lowest energy spin configuration .
as a final converged magnetic configuration in the orthorhombic phase we always found stripe - type antiferromagnetic order either along * a * or along * b * , as discussed below . the energy cutoff in the calculations was set to 300 ev and a monkhorst - pack uniform grid of @xmath12 points was used for the integration of the irreducible brillouin zone ( bz ) .
starting from the low - temperature orthorhombic structure with stripe magnetic order , we performed structure relaxations under applied uniaxial tensile and compressive stresses along * a * ( afm direction ) , * b * ( fm direction ) and the plane - diagonal * * a**+*b * direction for both bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 ( see inset of figure [ ba_latticeparameters ] ( a ) ) .
we measure stress in units of the equivalent hydrostatic pressure , @xmath5 , with @xmath6 denoting the stress tensor matrix .
we simulated pressures in the range between @xmath14 gpa and @xmath15 gpa . in the tensile stress range , below @xmath16 gpa
we observe in both systems that , for stress along * a * , a sudden expansion in @xmath17 and contraction in @xmath18 and @xmath1 axes occurs .
a similar situation arises when pulling apart along * b*. this feature signals the extreme case of absence of bonding within the material , and for this reason this pressure range will be excluded from further discussion .
figure [ ba_latticeparameters ] shows the evolution of lattice parameters for bafe@xmath0as@xmath0 as a function of uniaxial stress along * a * , * b * and * * a**+*b*. we consider both compressive stress ( positive pressure ) and tensile stress ( negative pressure ) . at @xmath19 gpa , we have @xmath17 ( afm direction ) @xmath20 @xmath18 ( fm direction ) .
bafe@xmath0as@xmath0 remains in the orthorhombic phase with nonzero increasing magnetic moment for large tensile stress ( negative pressure ) .
pulling apart ( _ i.e. _ @xmath21 ) along the ( longer ) afm direction * a * ( figure [ ba_latticeparameters ] ( a ) ) the system expands along * a * , strongly compresses along * c * and shows almost no changes along * b * ; similarly , pulling apart along the ( shorter ) fm direction * b * ( figure [ ba_latticeparameters ] ( b ) ) @xmath18 expands , @xmath1 compresses and @xmath17 shows almost no changes except at the pressure @xmath22 gpa ( figure [ ba_latticeparameters ] ( e ) ) . at this point , bafe@xmath0as@xmath0 shows a sudden jump in the orthorhombicity where * a * becomes the shorter axis and * b * becomes the longer axis .
this interchange happens with a rotation of the magnetic order by 90 degrees , _
i.e. _ the fm direction becomes parallel to the * a * axis while the afm direction becomes parallel to the * b * axis .
we will discuss this feature further below .
note that tensile stress along * * a**+*b * acts similarly on both * a * and * b * directions , which expand , while the * c * direction strongly compresses ( figures [ ba_latticeparameters ] ( c ) and ( f ) ) . under application of compressive stress ( positive pressure ) , we observe in all three cases a strong expansion along * c * and a compression along the direction of applied stress * * ( * a * , * b * or * * a**+*b * ) .
for the cases where pressure is applied along @xmath23 or @xmath24 , we observe almost no changes or a slight expansion along * b * and * a * , respectively .
since @xmath25 at zero stress , we observe the inversion of axes followed by a jump in orthorhombicity and a 90 degree rotation of the magnetization when stress is applied along * a * at @xmath26 gpa ( figures [ ba_latticeparameters ] ( a ) and ( d ) ) .
this inversion of axes , with @xmath27 for all higher compressive stress values means that the spin configuration shown in the inset of figure [ ba_latticeparameters ] ( a ) should now be turned by 90 degrees , with * b * pointing along the afm direction .
such an inversion is also observed for compressive stress along * * a**+*b * at much larger pressures of @xmath28 gpa .
figure [ ba122moments ] shows the evolution of magnetic moment , volume and as height in bafe@xmath0as@xmath0 as a function of stress .
the three quantities show a clearly monotonic behavior independent of the applied stress direction except for small jumps at the pressures @xmath22 gpa ( for stress along * a * ) and @xmath26 gpa ( for stress along * b * ) where the tetragonal condition is almost fulfilled ( @xmath17 @xmath29 @xmath18 ) ( figures [ ba122moments ] ( b ) , ( d ) , ( f ) ) .
we also note here how magnetic moments in bafe@xmath0as@xmath0 respond to different direction of pressure application .
the highest rate of suppression , of roughly 0.1@xmath30/gpa is achieved when pressure is applied within the @xmath31-plane , while application of pressure along the * c*-axis actually results in magnetic moment increase by 0.03@xmath30/gpa @xcite .
even though dft calculations overestimate the value of the ordered fe magnetic moment at @xmath19 gpa , it is to be expected that the relative changes in magnetic moment should provide a reliable description of the situation of bafe@xmath0as@xmath0 under pressure as shown in previous studies .
@xcite $ ] .
panels ( b ) , ( d ) and ( e ) show the corresponding zoom of the pressure dependence of these quantities in the range @xmath32 $ ] .
negative pressures correspond to tensile stress while positive pressures correspond to compressive stress .
[ ba122moments ] , scaledwidth=50.0% ] except for the pressures @xmath22 gpa ( for stress along * a * ) and @xmath26 gpa ( for stress along * b * ) , stress always enforces a certain degree of orthorhombicity and the system remains magnetically ordered with a decreasing ordered moment as a function of compressive stress ( figure [ ba122moments ] ( a ) ) .
moreover , since the @xmath1 axis continually expands from negative to positive pressures , @xmath33 increases accordingly as a function of stress ( figure [ ba122moments ] ( e ) ) .
these features have a direct consequence on the electronic properties of the system . as an illustration
, we show in figure [ ba_fs ] the ( non - spin polarized ) fermi surface of bafe@xmath0as@xmath0 under application of uniaxial stress @xmath34 gpa and @xmath35 gpa applied along * a * in the 1fe / unit cell equivalent brillouin zone .
we would like to note that correlation effects beyond dft as implemented in dft+dmft ( dynamical mean field theory ) , which are known to give a good agreement between the calculated fermi surfaces and angle - resolved photoemission measurements in the fe pnictides , @xcite have not been included here .
modest tensile stress of 0.07 gpa leads to the disappearance of the @xmath36 hole pocket around @xmath37 in the @xmath38 plane ( see figure 5 ( a ) in ref . ) . on the other hand , when compressive stress is applied , the hole pockets around @xmath37 significantly change in size , and additionally at a pressure of 1.7 gpa , small electron pockets , of majority @xmath36 and @xmath39 character , appear along the @xmath40 directions of the bz ( figures [ ba_fs ] ( c ) and ( e ) ) .
the increase of the @xmath36 hole pocket size with increasing uniaxial stress can be explained by the reduction of fe - fe distances along the * a * axis , leading to an increased contribution of @xmath36 - @xmath36 bonding . in fact , the effects of tensile and compressive stress on the electronic structure shown for the example of stress along * a * can be seen also in our calculations for both stress along * b * and along * * a**+*b*. in figure [ ba122dos ] we analyze the orbitally - resolved density of states at the fermi level @xmath41 . applying stress both along * a * and * b * has the same effect on the total density of states of both bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 , but there is a selective orbital order as shown in figure [ ba122dos ] .
@xmath41 is predominantly of @xmath42 character when @xmath25 and of @xmath43 character when @xmath44 .
this means that the dominant character switches from @xmath43 to @xmath42 at @xmath45 gpa , and from @xmath42 to @xmath43 at @xmath46 gpa as expected .
as@xmath0 at the fermi level @xmath41 with stress ( a ) applied along * a * , and ( b ) along * b*. lines joining the calculated points are a guide for the eye .
[ ba122dos ] , scaledwidth=45.0% ] the lattice parameters of cafe@xmath0as@xmath0 under application of compressive stress along * a * , * b * and * * a**+*b * directions show a similar overall behavior compared to bafe@xmath0as@xmath0 ( see figure [ ca122latticeparameters ] ) except for some important shifts of the pressures at which the system exchanges the fm and afm directions . when stress is applied along the * a * direction , we observe at @xmath47 gpa a jump in the orthorhombic order parameter , with a sign - change , accompanied by a reversal of the magnetic afm and fm directions .
however , analogously to the case of bafe@xmath0as@xmath0 , this is not followed by a suppression of the magnetic moments of iron .
in fact , the @xmath1 axis expands with applied stress and at @xmath47 gpa the @xmath1 lattice parameter in cafe@xmath0as@xmath0 is too large for the formation of an interlayer as - as covalent bond , necessary for a transition to a collapsed tetragonal phase and suppression of magnetic moments as observed under hydrostatic or @xmath1-axis uniaxial pressure .
@xcite for tensile stress along the ( shorter ) * b * direction , the reversal of afm and fm directions happens at @xmath48 gpa followed by a jump in the orthorhombicity .
the magnetic response of cafe@xmath0as@xmath0 is highly anisotropic as well , but contrary to the case of bafe@xmath0as@xmath0 , the magnetic moments in cafe@xmath0as@xmath0 are most effectively suppressed when pressure is applied along * c*. we measure a rate of about 0.1@xmath30/gpa @xcite .
application of pressure within the @xmath31-plane results in a suppression of the magnetic moment at a rate of 0.02@xmath30/gpa . in order to investigate the possibility of a structural and/or magnetic phase transition at higher pressures , we concentrate now on compressive stress along the diagonal of the @xmath31-plane .
we find that orthorhombicity is preserved up to 7.7 gpa , where a sharp transition to a tetragonal phase appears .
this transition is of first - order type like the orthorhombic to collapsed tetragonal phase transition under application of hydrostatic or uniaxial pressure along the * c * axis @xcite but in this case , changes of magnetic and structural properties take opposite directions ; the * c * axis undergoes a sudden expansion of about 9.5% , and * a * and * b * axes contract while the iron magnetic moments order ferromagneticaly and sharply increase in value by around 25% .
interestingly though , contrary to the application of hydrostatic and uniaxial pressure along * c * axis , the volume change here is significantly smaller , namely an expansion by about 0.9% .
these features are not observed in bafe@xmath0as@xmath0 when we compress along the diagonal of the @xmath31-plane up to pressures of 10 gpa .
using data for the response to the uniaxial stress along * a * , * b * and * c * @xcite axes we can directly evaluate the elastic constants @xmath49 in bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 corresponding to the orthorhombic deformations .
we define elastic constants to be such that @xmath50 where @xmath51 and @xmath52 are stress and strain tensor components respectively , and indices @xmath53 and @xmath54 can be @xmath55 .
strains are defined to be @xmath56 , @xmath57 and @xmath58 , where @xmath59 , @xmath60 and @xmath61 are equilibrium unit cell dimensions ( @xmath19 gpa ) .
we first directly obtain @xmath62_{ij}$ ] by performing linear fits to @xmath63 and @xmath64 is then obtained by inverting the resulting matrix . for bafe@xmath0as@xmath0 ,
the elastic constant matrix is @xmath65\mathrm{gpa}\label{elconst_ba}\ ] ] utilizing voigt and reuss averages , @xcite defined as @xmath66 it is possible to estimate the bulk modulus .
voigt and reuss averages yield @xmath67gpa and @xmath68gpa , respectively , which is in good agreement with our previous estimate @xcite and the experimental value of @xmath69gpa .
@xcite for cafe@xmath0as@xmath0 , the elastic constant matrix is given by @xmath70\mathrm{gpa}\label{elconst_ca}\ ] ] which results in bulk modulus of @xmath71gpa and @xmath72gpa using voigt and reuss averages , respectively .
both values are in good agreement with experimentally determined values of @xmath73gpa @xcite and the estimate based on fits to birch - murnaghan equation of state .
to aid the interpretation of the _ ab initio _ results , we develop a phenomenological magneto - elastic ginzburg - landau model to capture the physics of the simultaneous sign - changing jump of the orthorhombicity and reversal of the afm and fm directions . as pointed out by refs .
, the magnetic structure of the iron pnictides consists of two interpenetrating nel sublattices , with magnetizations @xmath74 and @xmath75 of equal amplitude that can point either parallel or anti - parallel to each other ( see figure [ m1m2 ] ) . and @xmath75 .
[ m1m2],scaledwidth=35.0% ] by including also the orthorhombic order parameter @xmath76 , we obtain the ginzburg - landau free energy : @xmath77 here , @xmath78 , @xmath79 , @xmath80 , and @xmath81 .
the last condition ensures that the ground state is the striped magnetic configuration ( _ i.e. _ collinear @xmath74 and @xmath75 ) .
we also must have @xmath82 in order for the magnetic free energy to be bounded .
@xmath83 is the magneto - elastic coupling and @xmath84 is the stress field conjugate to the orthorhombic order parameter .
the sign of @xmath85 is set to describe the experimental observation that ferromagnetic bonds are shorter than anti - ferromagnetic bonds .
although this model does not take into account the physics of the magnetically - driven structural transition , which comes from fluctuations beyond the ginzburg - landau analysis we perform below , @xcite it captures the main features of the _ ab initio _ results . the magnetic ground state is completely determined by the magnitude @xmath86 and the relative angle @xmath87 between @xmath74 and @xmath75 .
then , minimization of the free energy leads to three coupled equations for @xmath88 , @xmath87 , and @xmath89 : @xmath90 the last equation allows three possible solutions : @xmath91 , @xmath92 , and @xmath93 .
we focus only on the @xmath94 solutions , since they are the energy minimum at zero stress . in the ordered phase , where @xmath95 , we obtain the self - consistent equation for @xmath89 : @xmath96 for @xmath97 ,
the mean - field equations and the free energy are invariant upon changing @xmath98 and @xmath99 .
thus , we have two degenerate solutions : @xmath100 and anti - parallel @xmath74 and @xmath75 , @xmath92 , ( denoted hereafter @xmath101 ) or @xmath102 and parallel @xmath74 and @xmath75 , @xmath91 ( denoted hereafter @xmath103 ) .
the presence of a finite strain @xmath84 lifts this degeneracy . after defining :
@xmath104 the self - consistent equations for the two solutions @xmath101 and @xmath103 become simply : @xmath105 and we obtain analytic expressions for the two possible solutions : @xmath106 \end{split } \label{sol_delta}\ ] ] the interplay between the external stress field @xmath84 and the magneto - elastic coupling @xmath85 becomes evident in eqs .
( [ aux_eq_delta])-([sol_delta ] ) . for @xmath97 ,
@xmath85 acts as an external field of the same magnitude for both the @xmath101 and @xmath103 solutions , _
i.e. _ it gives rise to non - zero @xmath107 in the equations of state ( [ eq_delta ] ) , making these two solutions degenerate .
now , consider that for @xmath97 the system chooses the minimum @xmath101 ( _ i.e. _ @xmath100 and @xmath92 ) .
by increasing the external stress to a small value @xmath108 , the effective field @xmath109 is suppressed , whereas the field @xmath110 is enhanced .
although the solution @xmath103 ( _ i.e. _ @xmath102 and @xmath91 ) has a lower energy , the solution @xmath101 is still a local minimum , since the effective field @xmath109 is still finite .
this situation persists until @xmath84 increases to the point where the field @xmath109 becomes negative and large enough to make the @xmath101 solution not a local minimum . in particular , to determine when the @xmath101 solution ceases to be a local minimum , we analyze when one of the eigenvalues of the hessian matrix @xmath111 becomes negative ( with generalized coordinates @xmath112 ) .
the three eigenvalues @xmath113 are given by : @xmath114\\ & \quad\pm\frac{1}{2}\sqrt{\left[a_{s}+3u_{s}\delta^{2}-2m^{2}\left(u_{m}-g_{m}\right)\right]^{2}+4\lambda^{2}m^{2}}\\ \mu_{0 } & = \frac{m^{2}}{2}\left(g_{m}m^{2}-\lambda\delta\cos\theta\right)\label{eigenvalues } \end{split}\ ] ] for the @xmath101 ( @xmath100 , @xmath92 ) solution , the only eigenvalue that can become negative with increasing @xmath84 is @xmath115 .
we find that this happens when the condition @xmath116 is met , corresponding to an effective field @xmath117 , _ i.e. _ to the critical stress : @xmath118 at @xmath119 , the solution @xmath100 , @xmath92 is not a local minimum any longer and the system jumps to the new minimum with @xmath102 , @xmath91 , where not only the sign of the orthorhombicity is reversed , but also the angle between the magnetizations of the two sublattices ( _ i.e. _ the afm and fm directions ) .
this behavior is shown in figure [ fig_gl ] for a particular set of parameters .
( in units of @xmath120 ) as a function of the applied stress @xmath84 ( in units of @xmath121 ) .
we used parameters such that @xmath122 .
the jump happens when the @xmath100 solution is no longer a local minimum , and is accompanied by a reversal of the angle between the two sublattice - magnetizations @xmath74 and @xmath75 , _ i.e. _ a reversal of the afm and fm directions .
( b ) dft results for the strain - dependent orthorhombic order parameter @xmath123 .
the blue curve is for bafe@xmath0as@xmath0 and the red curve , for cafe@xmath0as@xmath0 .
[ fig_gl ] ] to compare with the dft results , we performed a slight modification with respect to the calculations presented in the previous section . to ensure that the external stress couples mainly to the orthorhombic mode @xmath89 and not to the longitudinal elastic mode @xmath124 , such that it does not change the volume of the system , we simultaneously applied positive ( compressive ) pressure along * a * and equal - amplitude negative ( tensile ) pressure along * b*. by doing this , we avoid terms such as @xmath125 in the free energy , rendering the comparison between the _ ab initio _ and the ginzburg - landau results more meaningful . the _ ab initio _ obtained behavior of @xmath89 as a function of @xmath84 , defined in the way described above , is shown also in figure [ fig_gl ] .
we find a qualitative agreement with the ginzburg - landau results , showing that the external stress indeed competes with the magneto - elastic coupling , helping the system to overcome the energy barrier between the @xmath101 ( @xmath100 , @xmath92 ) and @xmath103 ( @xmath102 , @xmath91 ) solutions .
a quantitative comparison becomes difficult because the dft calculations are performed deep in the ordered phase , where higher order terms in the ginzburg - landau expansion become more important .
furthermore , it is also possible that some of the magnetic parameters ( @xmath126 , @xmath127 and @xmath128 ) have themselves some implicit pressure dependence in this regime .
nevertheless , we can use eq .
( [ sigma_c ] ) as a benchmark to discuss differences in the bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 compounds .
clearly , eq . ( [ sigma_c ] ) shows that @xmath129 increases with increasing magneto - elastic coupling .
therefore , the fact that @xmath129 is three times larger for cafe@xmath0as@xmath0 than for bafe@xmath0as@xmath0 suggests that , all other parameters being equal , the magneto - elastic coupling is larger in cafe@xmath0as@xmath0 than in bafe@xmath0as@xmath0 .
this may have important impact on the coupled magnetic and structural transitions displayed by these compounds , as discussed in refs . , and
as such deserves further investigation in the future .
in this paper , we analyzed the effects of tensile and compressive stress along * a * , * b * and * * a**+*b * on bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 by means of dft calculations under constant stress conditions with the help of the fire algorithm , combined with a phenomenological ginzburg - landau model . starting from the low - temperature magnetically ordered orthorhombic phase , we found in the pressure range between @xmath7 gpa and @xmath8 gpa no real structural phase transitions in both systems except for a pronounced orthorhombicity jump accompanied by a 90 degree rotation of the magnetic order .
fm and afm directions are interchanged , as are the orbital occupations @xmath10 and @xmath11 .
this inversion of axes is a direct consequence of the interplay between the intrinsic magneto - elastic coupling and the applied stress , as revealed by our ginzburg - landau analysis .
the proportionality between the critical stress where this inversion happens and the value of the magneto - elastic coupling suggests that in cafe@xmath0as@xmath0 the magnetic and structural degrees of freedom are more strongly coupled than in bafe@xmath0as@xmath0 , which may be related to the differences observed in their magnetic and structural transitions .
@xcite we also point out that the estimates for the bulk moduli of bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 derived from our _ ab initio _ results are in good agreement with the experimental measurements .
our calculations also provide important insight on the impact of uniaxial stress on the magnetic properties of the pnictides .
[ ba122moments ] shows that the magnetic moment at zero temperature always decreases ( increases ) with compressive ( tensile ) stress , regardless of the axis that is perturbed . unlike the jump in the orthorhombicity and the reversal of the fm and afm directions ,
this is a consequence not of the magneto - elastic coupling , but of the changes in the pnictogen height promoted by the uniaxial stress .
this is an important prediction of our first - principle calculations that can be tested experimentally .
interestingly , recent neutron diffraction experiments @xcite on bafe@xmath0as@xmath0 observed that upon application of compressive stress along the * b * axis , the magnetic moment is suppressed from @xmath130 to @xmath131 .
given the small values of applied pressure , it could be that this suppression is due to a reduction of the volume fraction of the domains whose moments are oriented out of the scattering plane , as pointed out by the authors of ref . .
nevertheless , in view of our current results , it would be interesting to either apply higher pressures to completely detwin the samples at low temperatures or to apply tensile stress to make a comparison with the case of compressive stress .
we note that ref
. also found an enhancement of the magnetic transition temperature @xmath132 in the same detwinned samples .
phenomenological models @xcite attribute this effect to changes in the magnetic fluctuation spectrum of the paramagnetic phase promoted by the uniaxial stress . in this regard
, it would be interesting in future _ ab initio _ studies to systematically investigate the changes in the nesting feature of the fermi surface ( fig .
[ ba_fs ] ) as function of the uniaxial stress - specifically , changes in the @xmath133 susceptibility peak . finally , we comment on the impact of our results to the understanding of the detwinning mechanism of iron pnictide compounds . in the tetragonal phase ,
rather small uniaxial stress @xmath134 mpa is enough to completely detwin the sample , giving rise to a single domain .
@xcite this can be understood as fluctuations above the structural transition temperature giving rise to long - range order in the presence of a symmetry - breaking field .
@xcite the situation is however very different deep in the orthorhombic phase , where twin domains are already formed . experimentally , it is known that larger pressures are necessary to completely detwin the system in this case , @xcite although specific values have not been reported , to our knowledge .
one possible detwinning mechanism is the reversal of the orthorhombicity of one domain type , while the domain walls remain pinned .
this corresponds precisely to the situation studied here , where the orthorhombicity jumps at a certain critical uniaxial pressure .
our _ ab initio _ results show that such a critical pressure for bafe@xmath0as@xmath0 would be around @xmath135 mpa one order of magnitude larger than the pressure values necessary to detwin the sample in the tetragonal phase
. of course , other mechanisms can also give rise to detwinning in the ordered phase , such as domain wall motion .
therefore , we propose controlled detwinning experiments at low temperatures in bafe@xmath0as@xmath0 to measure the critical pressure necessary to form a single domain .
values comparable to the ones discussed here would be a strong indication for reversal of the order parameter inside fixed domains .
which mechanism is at play in the iron pnictides
may have important consequences for the understanding of the impact of the external stress on the anisotropic properties measured in detwinned samples particularly the in - plane resistivity anisotropy , @xcite which is likely affected by domain wall scattering . @xcite
within the fire @xcite algorithm , the energy minimization is achieved by moving the system s position @xmath136 according to the following equation of motion : @xmath137\,,\label{fire_eq}\ ] ] where @xmath138 denotes the unit vector along @xmath139 , with @xmath140 , @xmath141 is time and @xmath142 is a time - dependent friction parameter which assures that the system is moving down the energy hypersurface in an optimal manner as long as the power @xmath143 is positive . if @xmath144 becomes negative , the procedure is stopped and relaxation is reinitialized in the direction of the steepest descent . in order to use fire for relaxation of periodic systems , we change from the configuration space of @xmath145 atomic cartesian coordinates @xmath146 , @xmath147 , to an extended system of @xmath148 coordinates @xmath149 , consisting of lattice vectors which are contained in columns of the @xmath150 matrix @xmath151 , and fractional atomic positions @xmath152 within the unit cell , where cartesian and fractional positions are related by @xmath153 .
when is rewritten in terms of coordinates @xmath154 , one just needs to find the appropriate expression for forces @xmath155 , that is , derivatives of energy with respect to the coordinates @xmath154 .
since stress and strain tensors , @xmath156 and @xmath157 , can be defined through @xmath158 where @xmath159 is the identity matrix and @xmath160 is the lattice matrix after an infinitesimal deformation , it is easy to show that @xmath161 so that forces in the extended coordinates are given by @xmath162.\label{forces}\ ] ] forces @xmath163 and stresses @xmath6 are obtained from the electronic structure code , and are inserted directly into eq . , taking into account that @xmath164 , that is , total stress is the sum of internal and external stresses applied to the surface of the unit cell .
* acknowledgments.- * we would like to thank igor i. mazin for useful discussions and the deutsche forschungsgemeinschaft for financial support through grant spp 1458 , the helmholtz association for support through ha216/emmi and the centre for scientific computing ( csc , loewe - csc ) in frankfurt for computing facilities .
this research was supported in part ( rv ) by the national science foundation under grant no .
nsf phy11 - 25915 .
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 .
* 29 * , 037402 ( 2012 ) .
j. ferber , h.o .
jeschke , and r. valent , phys .
lett . * 109 * , 236403 ( 2012 ) .
p. werner , m. casula , t. miyake , f. aryasetiawan , a.j .
millis , and s. biermann , nature physics , * 8 * , 331 ( 2012 ) .
j. hu , c. setty , and s. kivelson , phys .
b * 85 * , 100507(r ) ( 2012 ) . a. kreyssig , m. a. green , y. lee , g d. samolyuk , p. zajdel , j. w. lynn , s. l. budko , m. s. torikachvili , n. ni , s. nandi , j. b. leo , s. j. poulton , d. n. argyriou , b. n. harmon , r. j. mcqueeney , phys . rev .
b * 78 * , 184517 ( 2008 ) . | starting from the orthorhombic magnetically ordered phase , we investigate the effects of uniaxial tensile and compressive stresses along * a * , * b * , and the diagonal * * a**+*b * directions in bafe@xmath0as@xmath0 and cafe@xmath0as@xmath0 in the framework of _ ab initio _ density functional theory ( dft ) and a phenomonological ginzburg - landau model .
while contrary to the application of hydrostatic or @xmath1-axis uniaxial pressure both systems remain in the orthorhombic phase with a pressure - dependent nonzero magnetic moment , we observe a sign - changing jump in the orthorhombicity at a critical uniaxial pressure , accompanied by a reversal of the orbital occupancy and a switch between the ferromagnetic and antiferromagnetic directions .
our ginzburg - landau analysis reveals that this behavior is a direct consequence of the competition between the intrinsic magneto - elastic coupling of the system and the applied compressive stress , which helps the system to overcome the energy barrier between the two possible magneto - elastic ground states .
our results shed light on the mechanisms involved in the detwinning process of an orthorhombic iron - pnictide crystal and on the changes in the magnetic properties of a system under uniaxial stress . |
it is generally accepted that large , bright galaxies reside within massive dark matter halos ; however , the radial extent of the halos is not well constrained and , hence , neither is the total mass nor the mass to light ratio of these objects . galaxy
galaxy lensing , in which the halos of foreground galaxies weakly distort the shapes of background galaxies , has recently proven to be a powerful method by which the masses and mass to light ratios of galaxies may be constrained .
galaxy lensing has been detected by a number of different groups ( see , e.g. , the review by brainerd & blandford 2003 and references therein ) and , in particular , the sloan digital sky survey ( sdss ) collaboration has obtained measurements of the galaxy galaxy lensing shear with extremely high statistical significance ( e.g. , fischer et al .
2000 ; mckay et al .
2001 ) . using weak lensing measurements of the projected mass correlation function , mckay et al.(2001 ) , hereafter sdss01 , found that @xmath14 , the mass of lens galaxies interior to a radius of @xmath1 kpc , scaled roughly linearly with the luminosities of the lens galaxies in all bandpasses except @xmath15 . since the galaxy galaxy lensing shear is small ( @xmath16% in the case of the sdss galaxies ) and is not without its own sources of error ( including the the separation of lenses from sources ) , mckay et al .
( 2002 ) performed an independent estimate of the masses of dark matter halos surrounding sdss galaxies using the dynamics of satellite galaxies .
their sample consisted of 618 host galaxies and 1225 satellites , which was considerably smaller and shallower than the sample in the weak lensing analysis due to the necessity of redshifts for all of the galaxies .
nevertheless , mckay et al . (
2002 ) , hereafter sdss02 , found that their dynamical analysis led to trends in the dependence of @xmath0 on the host galaxy luminosity that were reasonably consistent with the trends obtained from their previous weak lensing analysis .
however , the mass to light ratios found from the dynamical analysis were systematically lower than those from the lensing analysis ( @xmath17 ) . here
we perform a dynamical analysis of the masses of isolated host galaxies in the 100k public data release of the 2df galaxy redshift survey , hereafter 2dfgrs .
the 2dfgrs is a spectroscopic survey in which the objects are selected in the @xmath18 band from the apm galaxy survey ( maddox et al .
1990a , b ) , and extensions to the original survey .
ultimately , the survey will provide spectra for @xmath19 galaxies brighter than @xmath20 and will cover an area of order 2000 square degrees ( see , e.g. , colless et al .
2001 ) .
our host galaxies span a redshift range which is similar to that of the sdss02 galaxies , and our sample is of a similar size .
we select host / satellite combinations and determine dynamical masses for the host galaxies based upon the methods outlined in sdss02 in order to compare most easily to their results . in particular , we investigate the apparent lack of dependence of @xmath21 on the host luminosity found by sdss02 , and the somewhat low value of the dynamical mass to light ratio in comparison to the lensing mass to light ratio . throughout
, we adopt a flat , @xmath22dominated universe with parameters @xmath23 , @xmath24 , and @xmath25 km s@xmath9 mpc @xmath9 .
consistent with this , we take the absolute magnitude of an @xmath26 galaxy in the @xmath18 band to be @xmath27 ( norberg et al . ,
in order to compare to the results of sdss02 , we select host and satellite galaxies from the 2df survey according to the sdss02 criteria : * host galaxies must be `` isolated ''
. they must be at least twice as luminous as any other galaxy that falls within a projected radius of @xmath28 mpc , as well as within a velocity difference of @xmath29 km s@xmath9 .
* potential satellite galaxies must be at least 4 times fainter than their host , must fall within a projected radius of @xmath30 kpc of their host , and the satellite host velocity difference must be @xmath29 km s@xmath9 .
these basic selection criteria result in 864 hosts and 2340 satellites . as noted by sdss02
, however , many of the hosts have a large number of satellites around them ( in one case , a potential host in our sample has 605 satellites ) .
these are , therefore , most likely to be associated with cluster systems , rather than being truly isolated . to eliminate these objects ,
we impose a further restriction that the luminosity of the host be greater than the sum total of the luminosities of the satellites .
this , too , was done by sdss02 , and reduces our 2df sample to 859 hosts and 1693 satellites .
figure 1 .
a ) probability distribution for luminosities of the 809 host galaxies in the full sample .
luminosities are in units of @xmath26 .
b ) probability distribution for the number of satellites surrounding the host galaxies in the full sample .
c ) probability distribution for the difference in apparent magnitude between the host and satellite galaxies in the full sample .
d ) velocity dispersion of satellite galaxies in the full sample as a function of median host luminosity .
solid line shows @xmath31 , which is the best fit to the data .
dashed line shows @xmath32 .
[ fig1 ] finally , we impose two additional cuts on the host galaxies .
first , eyeball morphologies are available for the 2df galaxies with @xmath6 , and 29 of the above hosts are classified as galaxy galaxy mergers .
we delete these hosts from the sample on the basis that they are unlikely to be fully relaxed systems .
second , we delete all hosts with @xmath33 because the velocity dispersions of their satellites are poorly fit by the technique we adopt ( see below ) , and the number of interloper galaxies ( as opposed to genuine satellites ) appears to be both large ( @xmath34% ) and have a large dispersion ( @xmath35% ) .
these additional cuts leave us with a final sample of 809 host galaxies and 1556 satellites . of these , 75 are classified as ellipticals , 84 are classified as s0 , and 243 are classified as spirals . the sample of spirals is uniformly distributed in inclination angle , and there is no correlation between host luminosity and median inclination angle .
the ellipticals have a total of 171 satellites , the s0 s have a total of 303 satellites , and the spirals have a total of 478 satellites . the median redshift of the 809 host galaxies in the full sample is @xmath36 , while for the spiral hosts @xmath37 , and for the elliptical and s0 hosts @xmath38 .
figure 2 .
same as fig . 1 , but for elliptical and s0 host galaxies .
[ fig2 ] the probability distribution of the luminosities of the host galaxies , the probability distribution of the number of satellites around individual hosts , and the probability distribution of the difference in apparent @xmath18 magnitude between the hosts and their satellites are shown in panels a , b , and c of figs . 1 , 2 , and 3 . fig
. 1 shows results for the entire sample of 809 hosts , while fig .
2 shows the results for the 159 hosts classified as elliptical or s0 , and fig .
3 shows the results for the 243 hosts classified as spirals .
in order to compare with sdss02 , we adopt an analysis technique that is identical to theirs .
the radial velocity dispersions of the host galaxy halos , @xmath39 , are computed by fitting a combination of a gaussian and a constant offset to histograms of the velocity differences between the hosts and satellites .
the width of the best
fitting gaussian is a measure of @xmath39 , while the offset accounts for the fact that there will , necessarily , be some fraction of interloper galaxies that are selected as satellites when , in fact , they are not dynamically associated with the host galaxy . like sdss02
, we find that this technique provides very good fits to the velocity difference histograms , yielding values of @xmath40 per degree of freedom , @xmath41 , that are @xmath42 for hosts with @xmath43 . in the case of hosts with @xmath44 , @xmath45 and ,
hence , we do not consider these objects further .
because it is likely that more interlopers will have velocities that are greater than their hosts ( e.g. , zartisky & white 1994 ) , we determined @xmath39 for the host galaxies by fitting gaussians plus constant offsets to 3 different velocity difference histograms : ( i ) velocity differences taken to be the absolute value , @xmath46 , of the measured difference , ( ii ) negative velocity differences , and ( iii ) positive velocity differences .
we define the velocity difference to be @xmath47 , so that negative values of @xmath48 correspond to satellites which are more distant in velocity space . in all cases , the best
fitting velocity dispersions are in very good agreement amongst the 3 histograms .
in addition , we find a clear difference in the number of interlopers . in the case of the full sample of 809 hosts ,
the interloper fraction is @xmath49% for host satellite pairs with @xmath50 , @xmath51% for host satellite pairs with @xmath52 , and @xmath53% on average ( i.e. , fitting to the distribution of @xmath46 ) .
for the hosts with spiral morphologies , the mean interloper fraction is @xmath54% while for hosts with elliptical and s0 morphologies the interloper fraction is much lower , @xmath55% .
lastly , for all of our samples of host galaxies , we find that @xmath39 is independent of the radius at which it is determined . figure 3 .
same as fig . 1 , but for spiral host galaxies .
solid line in panel d shows @xmath56 km sec@xmath9 , which is the best fit to the data .
[ fig3 ] results for @xmath39 as a function of median host luminosity are shown in panel d of figs . 1 , 2 , and 3 . in the case of the full host sample and the elliptical / s0 host sample ,
the relationship between velocity dispersion and median host luminosity is fit best by linear relations , and these are shown by the solid lines in panel d of figs . 1 and 2 .
for the full host sample we find @xmath57 km s@xmath9 , and for the elliptical / s0 sample we find @xmath58 km s@xmath9 .
this is somewhat different from the results of sdss02 who found @xmath59 .
we note , however , that our data are consistent with such a relationship , and the best fitting function of the form @xmath32 is shown by the dashed lines in these figures . strikingly different from these results , however , is the relationship of velocity dispersion and median host luminosity for the spiral hosts . from panel
d of fig .
3 , it is clear that @xmath39 for the spiral hosts is independent of host luminosity , and we find @xmath60 km s@xmath9 for these objects .
to determine the masses of the dark matter halos which surround our host galaxies , we adopt the following mass estimator : @xmath61\ ] ] ( e.g. , binney & tremaine , 1987 ) . here
@xmath62 is a 3-dimensional radius , @xmath63 is the mean square radial velocity of the satellites , @xmath64 is the number density of satellites , and @xmath65 is a measure of the anisotropy in the velocity dispersion of the satellites : @xmath66 although other methods of obtaining dynamical masses using satellite galaxies have been adopted in the literature ( see , e.g. , bahcall & tremaine 1982 ; zaritsky & white 1994 ; zaritsky et al .
1997 ) , this is the method adopted by sdss02 and , therefore , we adopt it as well .
sdss02 have used the gif simulation , which incorporates semi analytic galaxy formation within a large cosmological n body simulation ( e.g. , kauffmann et al .
1999 ) , to evaluate this mass estimator .
in particular , sdss02 find that the velocity anisotropy of the satellite galaxies in the gif simulation is small ( i.e. , @xmath65 is consistent with zero at the 2@xmath67 level ) , and that the mean square line of sight velocity dispersion , @xmath68 , is consistent with the mean square radial velocity dispersion , @xmath63 , at the 1@xmath67 level . combining this with the fact that the line of sight velocity dispersion is observed to be independent of radius , the mass estimator used by sdss02 reduces to : @xmath69 we compute @xmath64 for the satellites in our sample and find @xmath70 for the satellites surrounding the hosts in the full sample , @xmath71 for the satellites surrounding elliptical and s0 hosts , and @xmath72 for the satellites surrounding spiral hosts .
these are all consistent with the results of sdss02 who find @xmath73 for their sample .
having obtained the number density of satellites as a function of radius , we now use the values of @xmath39 from figs . 1 , 2 , and 3 to determine the mass to light ratios of the host galaxies .
for consistency with sdss02 , we adopt a fiducial projected radius of @xmath74 kpc . shown in fig .
4 are the results for @xmath75 in units of @xmath76 as a function of median host luminosity .
the top panel shows results for the full sample of host galaxies , the middle panel shows results for the elliptical and s0 hosts , and the bottom panel shows results for the spiral hosts . from fig . 4 , then
, we find that @xmath11 for the 809 hosts in our full sample is fairly constant for hosts with @xmath2 and has a value of @xmath77 . for hosts with
@xmath78 there is a weak suggestion of a somewhat higher mass to light ratio .
similarly , @xmath11 for the elliptical and s0 hosts is fairly constant for hosts with @xmath2 and has a value of @xmath79 .
again , there is a slight suggestion that elliptical and s0 hosts with @xmath4 have a somewhat higher mass to light ratio .
in contrast , over the range of host luminosities explored here , @xmath11 for the spiral hosts shows a clear monotonic decrease with luminosity , and is consistent with a power law of the form @xmath80 .
figure 4 .
mass to light ratios for the host galaxies interior to a radius of @xmath1 kpc .
top panel : all 809 host galaxies , middle panel : hosts with elliptical and s0 morphologies , bottom panel : hosts with spiral morphologies .
solid lines in the top and middle panels show the mean mass to light ratio for hosts with @xmath2 .
solid line in the bottom panel show the best
fitting power law for @xmath11 .
the 2df and sdss02 host / satellite samples are of comparable depths and have similar sizes , so it is not unreasonable to make comparisons between them .
the comparison is , however , somewhat limited by the fact that the 2df galaxies are selected in @xmath18 , while the sdss02 galaxies are selected in @xmath81 , with host luminosities obtained in @xmath15 , @xmath82 , @xmath81 , @xmath83 , and @xmath84 .
( the transformation from the sdss photometry is given by @xmath85 ; norberg et al.2002 ) .
also , sdss02 did not perform separate dynamical analyses for the hosts of early and late type galaxies , so a direct comparison is not possible in this case . in all 5 sdss photometric bands ,
sdss02 find @xmath86 , so that in a given band , a single mass to light ratio characterizes the hosts . that mass to light ratio is a sharply decreasing function of the central wavelength of the bandpass ( e.g. , a factor of order 3 higher in @xmath15 than in @xmath84 ) . in @xmath82 ,
sdss02 find @xmath87 and in @xmath81 @xmath88 .
these compare well with the mass to light ratio that we obtain , @xmath89 , for the host galaxies our full sample that have luminosities of @xmath2 .
since we can not compare our @xmath21 for host galaxies of different morphologies to the results of sdss02 , we instead compare them to the weak lensing results of sdss01 .
sdss01 did not classify their galaxies according to visual morphology but , instead , used spectral features to place subsets of their lens galaxies into broad `` early '' and `` late type '' categories .
the early types represent about 40% of the total number of lens galaxies , and the late types represent another 40% of the total number of lens galaxies .
table 3 of sdss01 shows that in the bluer bands , @xmath90 is somewhat morphology dependent , with the mass to light ratio of the ellipticals exceeding that of the entire lens sample by a factor of @xmath91 in @xmath82 and by a factor of @xmath92 in @xmath81 .
again , this compares well with our results for the elliptical / s0 hosts in the 2df sample , where we find that @xmath21 for the elliptical / s0 hosts exceeds that of the full sample by a factor of @xmath93 for hosts with @xmath2 ( e.g. , fig . 4 ) .
our result that @xmath94 for the spiral hosts is in clear conflict with the results of sdss01 , who found that @xmath95 was independent of luminosity in all but the very bluest band ( @xmath15 ) .
however , our result stems from the fact that the line of sight velocity dispersion is independent of luminosity for the spiral hosts .
while this is inconsistent with the lensing results of sdss01 , it is consistent with the dynamical results of zaritsky et al .
( 1997 ) who found that the velocity difference , @xmath48 , between 69 isolated spiral galaxies ( @xmath96 ) and 115 satellites was independent of the inclination corrected h - i linewidth of the host and was , therefore , independent of the luminosity of the host ( through , e.g. , the tully fisher relation ) .
whether the conflict between the lensing and dynamical results for the halos of spiral galaxies is due to differences in sample selection or due to systematic effects in one or both of the mass estimators remains to be determined .
however , the ultimate completion of both the sdss and the 2dfgrs will aid tremendously in the resolution of this issue , and we look forward to the wealth of data that both surveys will provide in the near future .
we are pleased to thank the 2dfgrs team for making the 100k public data release available in a timely and very user friendly manner .
support under nsf contract ast-0098572 ( tgb , mjs ) is also gratefully acknowledged . | we compute @xmath0 , the dynamical mass interior to a radius of @xmath1 kpc , for a set of 809 isolated host galaxies in the 100k data release of the 2df galaxy redshift survey .
the hosts are surrounded by 1556 satellite galaxies , as defined by a set of specific selection criteria .
our mass estimator and host / satellite selection criteria are taken from those used by the sloan digital sky survey ( sdss ) collaboration for an analysis of @xmath0 for sdss galaxies and , overall , our results compare well with theirs . in particular ,
for @xmath2 we find @xmath3 , with a weak tendency for hosts with @xmath4 to have a somewhat higher @xmath5 .
additionally , we investigate @xmath5 for bright ( @xmath6 ) galaxies with elliptical , s0 , and spiral morphologies .
there are 159 hosts in the elliptical / s0 sample and , similar to the full sample , we find @xmath7 for galaxies with @xmath2 , and a weak tendency for intrinsically fainter galaxies to have a somewhat higher @xmath5 . in stark contrast to this , we find the line of sight velocity dispersion for the 243 spiral hosts to be independent of the host luminosity , with a value of @xmath8 km s@xmath9 .
thus , for spiral hosts we find that @xmath10 , where @xmath11 for a @xmath12 spiral is of order @xmath13 . |
the investigation of the nucleon structure in the framework of qcd is a major task in today s particle physics .
generalized parton distributions ( gpds ) have opened new ways of studying the complex interplay of longitudinal momentum and transverse coordinate space as well as spin and orbital angular momentum degrees of freedom in the nucleon .
the twist-2 gpds @xmath0 , @xmath1 , @xmath2 and @xmath3 of quarks in the nucleon are defined by the following parameterization of off - forward nucleon matrix elements @xmath4 for the helicity independent vector case , and @xmath5 for the helicity dependent axial vector case @xcite .
wilson lines ensuring gauge invariance of the bilocal operators are denoted by @xmath6 . here and
in the following we do not show explicitly the dependence of the gpds on the resolution scale @xmath7 .
fig.([fig : gpds ] ) shows the definition of the particle momenta and helicities .
the momentum transfer ( squared ) is given by @xmath8 ( @xmath9 ) .
using the light cone vector @xmath10 we define the longitudinal momentum transfer by @xmath11 .
the proper definition of the twist-2 tensor or quark helicity flip gpds @xmath12 , @xmath13 , @xmath14 and @xmath15 can be found in @xcite .
gpds provide a solid framework in qcd to relate many different aspects of hadron pyhsics : * the forward limit @xmath16 of certain gpds reproduces the well known parton distributions , that is @xmath17 , @xmath18 and @xmath19 . * the integral over the longitudinal momentum fraction @xmath20 of the gpds gives the dirac , pauli , axial , pseudo - scalar , tensor etc .
form factors , @xmath21 , @xmath22 , @xmath23 , @xmath24 , @xmath25 etc .
* the fourier transforms @xmath26 of the gpds @xmath0 , @xmath2 and @xmath12 at @xmath27 are coordinate space probability densities in the impact parameter @xmath28 @xcite . *
the forward limit of the @xmath29-moment of the gpd @xmath1 , @xmath30 , allows for the determination of the quark orbital angular momentum contribution to the nucleon spin , @xmath31 , where @xmath32 is the quark momentum fraction .
the fact that the impact parameter dependent quark distributions like @xmath33 have the interpretation of probability densities in the transverse plane marks a major step forward in the understanding of the nucleon structure .
a very interesting observation in this respect is that at large longitudinal momentum fraction , @xmath34 , i.e. when the active parton carries the whole nucleon momentum and the spectators give only a negligible contribution , the active parton becomes the center of momentum of the nucleon at @xmath35 .
accordingly , the impact parameter profile of the nucleon will be strongly peaked at @xmath36 .
keeping in mind that the parton distributions eventually vanish in the limit @xmath34 , we expect that @xmath37 .
transferred to momentum space , we have the striking prediction that the normalized gpd @xmath38 will be constant in the limit @xmath34@xcite .
interesting recent results on gpds using available experimental data on form factors and pdfs can be found in @xcite . on the lattice side
, we can not deal directly with matrix elements of bilocal light - cone operators .
therefore we first transform the lhs in eqs.([vectorop],[axialvectorop ] ) to mellin space by integrating over @xmath29 , i.e. @xmath39 .
this results in matrix elements of towers of local operators which are in turn parameterized in terms of so - called generalized form factors ( gffs ) @xmath40 , @xmath41 where the subtraction of traces is implicit .
the explicit parameterization for the tower of vector operators , @xmath42 , is given in @xcite in terms of the gffs @xmath43 and @xmath44 .
in the axial vector case , the corresponding expression in terms of the gffs @xmath45 and @xmath46 is shown in @xcite .
parameterizations for the tensor gpds have been derived recently in @xcite .
having computed the gffs in e.g. lattice qcd , it is fairly simple to get the moments of the corresponding gpds by using the following polynomial relations @xcite @xmath47 in the limit @xmath48 , the complete nucleon structure related to @xmath0 and @xmath2 is according to eq.([invv ] ) represented by the sets of gffs @xmath49 and @xmath50 , @xmath51 .
the lattice calculation and extraction of the gffs from nucleon two- and three - point - functions follows the standard procedure as outlined in e.g. @xcite .
our dynamical computations are currently based on a set of coupling- and quark mass - combinations with up to @xmath52 configurations and with pion masses ranging from @xmath53 down to @xmath54 gev .
the lattice spacing lies in between @xmath55 and @xmath56 fm .
since we are using non - perturbatively @xmath57 improved wilson fermions , we do not expect strong discretization effects @xcite .
a list of the parameters is given in fig.([fig : para ] ) .
we do not take into account the disconnected contributions which are numerically much harder to evaluate .
this is irrelevant as long as we concentrate on the iso - vector channel @xmath58 ( as we do here ) , for which the disconnected pieces would cancel due to iso - spin symmetry . to give an overview of our results we show in fig.([fig : first13550 ] ) the lowest three moments of the gpds @xmath59 , @xmath60 and @xmath61 as well as the lowest two moments of @xmath62 .
up to now , only the lowest moments @xmath63 which we show here are non - perturbatively renormalized , and operator mixing has to be taken into account for a proper renormalization of the @xmath64-moments for @xmath65 @xcite .
we do not expect the renormalization to affect the following general discussion and the extraction of the dipole masses below .
please note that for reasons of clarity @xmath66 has been shifted upwards by @xmath67 , as indicated by the superscript arrow in fig.([fig : first13550 ] ) , lower left panel .
all gffs have been fitted using a standard dipole ansatz , @xmath68 except for the pseudo - scalar ff @xmath69 , where we used a dipole ansatz including the pion - pole , @xmath70 the fits given by the gray bands in figs.([fig : first13550 ] ) show an overall satisfactory description of the lattice numbers .
while the statistical errors for the lowest two moments are rather small , they tend to be somewhat larger for the @xmath64-moment .
the statistics are particularly poor for @xmath71 , @xmath72 and the @xmath73 gffs , which are not shown in this work .
it turns out that the moments of @xmath0 and @xmath2 in the iso - vector channel are a factor of @xmath74 smaller than the corresponding moments of @xmath1 and @xmath3 , while @xmath75 which contributes to both @xmath0 and @xmath1 is compatible with zero .
our results indicate furthermore strong cancellations of @xmath76 and @xmath77 flavor contributions in the iso - singlet channel for the helicity independent gpd @xmath78 .
let us now take a look at the @xmath79-dependence of the gpds and the impact parameter dependent quark distribution @xmath80 .
the observation from section ( [ intro ] ) that @xmath81 translates directly to the statement that the gffs @xmath49 become constant for very large @xmath10 since increasingly more weight lies on values of @xmath29 close to one for @xmath82 . to be precise
, we expect that @xmath83 , independent of @xmath79 .
since the values for the momentum transfer squared in our computation are unevenly distributed in the interval @xmath84 gev , and since there is in particular a gap between @xmath85 and the smallest non - zero @xmath79 which can be realized on the given lattices , @xmath86 , it is not useful to perform a discrete fourier transformation ( ft ) of our results to impact parameter space .
instead , we directly use the continuous ft of the dipole ansatz eq.([dipole ] ) .
the normalized impact parameter dependent moments @xmath87 with @xmath88 are shown in fig.([fig : impact ] ) . for reasons of comparison
we show in fig.([fig : impact ] ) also the corresponding gffs , normalized to @xmath89 .
the gray bands in the impact parameter plot reflect the errors associated with the dipole masses , @xmath90 .
the results in fig.([fig : impact ] ) show very nicely the anticipated flattening of the gffs and the narrowing of the corresponding distributions in impact parameter space , going from @xmath63 to @xmath91 .
this flattening of the gpds has first been observed in @xcite .
since we have results for a large number of pion masses at hand , it is mandatory to concentrate in the following on the important issue of the @xmath92 dependence and the chiral extrapolation of our calculation . due to the use of wilson fermions , the dominant part of our computation is for pion masses above @xmath93 mev where chiral perturbation theory in its current form is most probably not applicable . in particular for the dipole masses
@xmath94 , we therefore resort to a purely phenomenologically motivated pion mass dependence of the form @xmath95 .
we have checked that a linear dependence on @xmath92 is clearly unfavored by our results . for the following analysis
we have normalized our results to @xmath96 first .
one option is then to perform dipole fits for all individual @xmath97 , @xmath98 combinations for a given moment @xmath10 and fit the pion mass dependence of the dipole masses with the given ansatz .
however to optimally utilize the information on the dipole masses from our collection of results , it is better to use a combined dipole - mass / pion - mass fit of the simple form @xmath99 depending on the two parameters @xmath100 and @xmath101 .
having done the fit , we may shift the raw numbers to a common curve given by eq.([chiraldipole ] ) at the physical pion mass , @xmath102 mev .
the results of this procedure are shown in fig.([fig : alldata1 ] ) for the gffs @xmath103 and @xmath104 , @xmath105 .
the fit parameters are found to be @xmath106 although the corresponding rather large error does not allow for a definite conclusion , it is interesting to note that the slope @xmath101 changes sign going from @xmath107 to @xmath108 .
we observe an overall nice clustering of the lattice points in fig.([fig : alldata1 ] ) much in favor of our ansatz eq.([chiraldipole ] ) .
the actual fits are given by the gray bands and show a very good @xmath109 lying below @xmath110 for the helicity independent and below @xmath111 for the helicity dependent case . in particular for @xmath104 , a small fraction of the lattice points seem to be significantly below the main line .
future investigations have to show if this is due to an unaccounted dependence on the lattice spacing , the ( finite ) lattice volume or just insufficient statistics .
an attempt to compare our results for the axial and tensor coupling , @xmath112 and @xmath113 , with chiral perturbation theory is described in @xcite .
we plan to extract the quark orbital angular momentum contribution to the nucleon spin as soon as the missing renormalization constants become available .
we have computed the lowest moments of gpds for a number of different pion masses in the range above @xmath114 mev .
the results are very well described by a combined dipole - mass / pion - mass fit .
while we observe no sign of chiral logarithms in the available range of pion masses , let us note that first promising lattice calculations towards the chiral region are currently under way using overlap @xcite as well as domain wall fermions @xcite .
the numerical calculations have been performed on the hitachi sr8000 at lrz ( munich ) , on the cray t3e at epcc ( edinburgh ) and on the apemille at nic / desy ( zeuthen ) .
this work has been supported in part by the dfg ( forschergruppe gitter - hadronen - phaenomenologie ) , the eu integrated infrastructure initiative hadron physics " as well as study of strongly interactive matter " ( in parts under contract number rii3-ct-2004 - 506078 ) .
d. mller , d. robaschik , b. geyer , f. m. dittes and j. horejsi , fortsch . phys .
* 42 * ( 1994 ) 101 [ arxiv : hep - ph/9812448 ] .
x. d. ji , phys .
d * 55 * ( 1997 ) 7114 [ arxiv : hep - ph/9609381 ] .
a. v. radyushkin , phys .
d * 56 * ( 1997 ) 5524 [ arxiv : hep - ph/9704207 ] .
m. diehl , eur .
j. c * 19 * ( 2001 ) 485 [ arxiv : hep - ph/0101335 ] .
m. burkardt , phys .
d * 62 * ( 2000 ) 071503 [ erratum - ibid .
d * 66 * ( 2002 ) 119903 ] [ arxiv : hep - ph/0005108 ] .
m. burkardt , int .
j. mod .
. a * 18 * ( 2003 ) 173 [ arxiv : hep - ph/0207047 ] .
m. burkardt , phys .
b * 595 * ( 2004 ) 245 [ arxiv : hep - ph/0401159 ] .
m. diehl , t. feldmann , r. jakob and p. kroll , arxiv : hep - ph/0408173 .
p. hoodbhoy , x. d. ji and w. lu , phys . rev .
d * 59 * ( 1999 ) 014013 [ arxiv : hep - ph/9804337 ] .
m. diehl , phys .
* 388 * ( 2003 ) 41 [ arxiv : hep - ph/0307382 ] .
hgler , phys .
b * 594 * ( 2004 ) 164 [ arxiv : hep - ph/0404138 ] .
z. chen and x. d. ji , arxiv : hep - ph/0404276 .
x. d. ji , w. melnitchouk and x. song , phys .
d * 56 * ( 1997 ) 5511 [ arxiv : hep - ph/9702379 ] .
m. gckeler , r. horsley , d. pleiter , p. e. l. rakow , a. schfer , g. schierholz and w. schroers [ qcdsf collaboration ] , phys .
* 92 * ( 2004 ) 042002 [ arxiv : hep - ph/0304249 ] .
m. gckeler _ et al .
_ , arxiv : hep - lat/0409162 . m. gckeler , r. horsley , h. perlt , p. e. l. rakow , a. schfer , g. schierholz and a. schiller , arxiv : hep - lat/0409025 ; arxiv : hep - lat/0410009 . | we present recent results on generalized parton distributions from dynamical lattice qcd calculations .
our set of twelve different combinations of couplings and quark masses allows for a preliminary study of the pion mass dependence of the transverse nucleon structure . |
the requirement of higher integration scales in electronic circuits , the onset of nanosensory applications in biomedicine , but also the fascinating capabilities of modern experimental setup with its enormous potential in polymer and surface research recently led to an increasing interest at the hybrid interface of organic and inorganic matter @xcite .
this also includes numerous detailed studies , e.g. , of polymer film wetting phenomena @xcite , pattern recognition @xcite , protein ligand binding and docking @xcite , charged adsorbed polymers @xcite as well as deposition and growth of polymers at surfaces @xcite . in most theoretical and computational studies
the polymer is anchored at the substrate with one of its ends which reduces the entropic freedom of the polymer .
these surface - grafted polymers @xcite are , e.g. , of particular interest in studies of shape transformations @xcite , e.g. , as reaction to external fields @xcite . however , in many recent experiments of organic inorganic interfaces the setup is different @xcite and is more adequately described by a polymer moving in a cavity with one adsorbing surface @xcite .
the main difference of such nongrafted polymers considered in this work is of entropic kind : in the desorbed phase the polymer can move freely within the cavity , and the polymer can fold into conformations , where the ends have no contact with the surface .
this paper is organized as follows . in sect .
[ secmod ] we describe the details of a minimalistic model for the hybrid system .
the main result , the solubility - temperature pseudo - phase diagram , is presented and discussed in sect .
[ sechyb ] .
the interpretation is consolidated by exemplified studies of fluctuations and correlations of relevant thermodynamic quantities such as numbers of contacts between monomers and monomer - substrate contacts as well as the gyration tensor , in the different phases .
the contact numbers turn out to be adequate system parameters for the description of the macrostate of the system , and therefore the free energy in dependence of these contact numbers is subject of a detailed study in sect .
[ secfree ] .
this quantity is also useful for classifying the conformational transitions between the phases which are also discussed there .
eventually , we conclude in sect .
[ secsum ] with a summary of the main results .
we employ a minimalistic simple - cubic ( sc ) lattice model @xcite which allows a systematic analysis of the conformational phases experienced by a nongrafted polymer in a cavity with one adhesive surface .
an example for the cavity model is shown in fig .
[ figmod ] .
the polymer can move between the two infinitely extended parallel planar walls , separated by a distance @xmath0 expressed in lattice units .
the substrate is short - range attractive to the monomers of the polymer chain , while the influence of the other wall is purely steric . denoting the number of nearest - neighbor , but nonadjacent monomer - monomer contacts by @xmath1 and the number of nearest - neighbor monomer - substrate contacts by @xmath2
, the energy of the hybrid system can be expressed in the simplest model as @xmath3 where @xmath4 and @xmath5 are the respective contact energy scales , which are left open in the following . for simplicity , we perform a simple rescaling and set @xmath6 and @xmath7 . here
we have introduced the overall energy scale @xmath8 and the dimensionless reciprocal solubility @xmath9 that controls the quality of the implicit solvent surrounding the polymer ( the larger the @xmath9 , the worse the solvent ) .
since contacts with the substrate usually entail a reduction of monomer - monomer contacts , there are two competing forces ( rated against each other by the energy scales ) affecting the formation of intrinsic and surface contacts .
in this paper we mainly focus on the conformational transitions the polymer experiences under different environmental conditions .
concretely , we are interested in the dependence of energetic and structural quantities on temperature @xmath10 and reciprocal solubility @xmath9 in equilibrium .
the probability ( per unit area ) for a conformation with @xmath2 surface and @xmath1 monomer - monomer contacts at temperature @xmath10 and reciprocal solubility @xmath9 is given by @xmath11 where @xmath12 is the contact density and @xmath13 the partition sum .
in this decomposition , @xmath14 stands for the density of unbound conformations , whereas @xmath15 is the density of surface and intrinsic contacts of all conformations bound to the substrate . obviously , the number of the conformations without contact to the attractive substrate , @xmath14 , depends on the distance @xmath0 between the cavity walls . for a sufficiently large distance @xmath0 from the substrate
the influence of the neutral surface on the unbound polymer is small . for @xmath16
, however , @xmath14 formally diverges .
therefore , the non - adhesive , impenetrable steric wall is necessary for regularization .
we studied polymers with up to 200 monomers by applying the contact - density chain - growth algorithm which is an improved variant of the recently developed multicanonical chain - growth sampling method @xcite .
all these methods set up on a variant of the pruned - enriched variant @xcite of rosenbluth sampling @xcite .
the main advantage of the improved method is that it directly samples the contact density @xmath17 , which is very useful for problems , where the model provides different energy scales .
this generalizes the ordinary multicanonical version @xcite which samples the density of states , i.e. , the number of states for given energy . here
we can set the two independent energy scales @xmath5 and @xmath4 or their ratio @xmath9 , respectively , _
after _ the simulation .
this allows to introduce the reciprocal solubility @xmath9 as a second environmental parameter in addition to the temperature @xmath10 .
the partition sum of the system as a function of these two parameters is simply @xmath18 and the statistical average of any function @xmath19 is given by the formula @xmath20 which is very convenient since it only requires to estimate the contact density @xmath17 in the simulation . denoting contact correlation matrix elements as @xmath21 with @xmath22
, the specific heat can be written as @xmath23 all quantities depending only on the contact numbers @xmath1 and @xmath2 can therefore simply be calculated from the estimate of the contact density @xmath17 provided by our simulation method .
although the two contact parameters are sufficient to describe the macrostate of the system and their fluctuations characterize the main pseudo - phase transition lines , it is often useful to introduce also nonenergetic quantities such as the end - to - end distance and the gyration tensor for gaining more detailed structural information of the polymer . for our specific problem at hand
it is particularly useful to study the structural anisotropy of the adsorbed polymer in the different phases . to this end
, we define the general gyration tensor for a polymer chain of @xmath24 beads with the components @xmath25 where @xmath26 , @xmath27 , is the @xmath28th cartesian coordinate of the @xmath29th monomer and @xmath30 is the center of mass with respect to the @xmath28th coordinate .
anisotropy in the polymer fluctuations is connected with the system s geometry and therefore it will be sufficient to study the components of the gyration tensor parallel ( @xmath31 , @xmath32 components ) and perpendicular ( in @xmath33 direction ) to the planar walls , @xmath34\ ] ] and @xmath35 the gyration radius is then simply the trace of the gyration tensor , @xmath36 .
the calculation of statistical averages for quantities @xmath37 that are not necessarily functions of the contact numbers @xmath2 and @xmath1 can not be performed via eq .
( [ eqav ] ) . in this case
only the more general relation @xmath38 holds , where the sum runs over all polymer conformations @xmath39 . introducing the accumulated density @xmath40 , where @xmath41 is the kronecker symbol , the expectation value can be expressed , however , in a form similar to eq .
( [ eqav ] ) : @xmath42 the quantity @xmath43 can easily be measured in simulations with the contact density chain - growth algorithm . in the following we use natural units ,
i.e. , we set @xmath44 .
for our exemplified study of the hybrid system in equilibrium we chose a polymer with 179 monomers . since this is a prime number , the polymer is unable to form perfect cuboid conformations on the sc lattice , as it is , e.g. , the case for a 100-mer .
@xcite there we found two low - temperature subphases dominated by the same @xmath45 cuboid . in one subphase it had 20 surface contacts , while in the other the cuboid was simply rotated , entailing 25 surface contacts
this is a typical example , where the exact number of monomers in the linear chain is directly connected with the occurrence of such specific pseudo - phases which are not , of course , phases in the traditional view . nonetheless , the enormous progress in high - resolution experimental structure analyses and in the technological equipment for precise polymer deposition , as well as the natural finite length of classes of polymers ( e.g. , peptides and proteins ) , explain the growing interest in pseudo - phases and the conformational transitions between them . here
we mainly focus on the expected thermodynamic phase transitions @xcite and low - temperature higher - order layering pseudo - phase transitions @xcite .
the following results were obtained from contact - density chain - growth simulations of the 179-mer in a cavity with @xmath46 ( see fig . [ figmod ] ) , choosing uniformly distributed starting points at random . in eight independent runs @xmath47 polymer conformations were generated in total .
the resulting contact density @xmath17 and accumulated densities like @xmath43 are independent of external parameters such as temperature @xmath10 and reciprocal solubility @xmath9 .
concrete values of statistical quantities for specific parameter settings are obtained by simple reweighting as in eqs .
( [ eqprob ] ) and ( [ eqavb ] ) .
= 8.8 cm discontinuities or divergences of energetic and nonenergetic fluctuations as functions of external parameters reveal typically dramatic cooperative transitions in the collective , macroscopic behavior of the system s microscopic degrees of freedom in the thermodynamic limit .
these transitions separate then thermodynamically stable phases and the transitions can uniquely be identified by certain values of the external parameters , e.g. , the transition temperature . usually , all fluctuations collapse at the same parameter sets .
but , this `` traditional view '' is only true in the thermodynamic limit .
finite - size systems usually exhibit a zoo of crossover- or pseudo - transitions , most of which disappearing in the thermodynamic limit . in special cases , e.g. , proteins , where the specific amino acid sequence is of finite length , no phase transitions in the strict sense
happen at all .
still , peaks in curves of fluctuating quantities _ can be _
signatures for `` cooperative activity '' , but this is not necessarily indicated by all fluctuations considered , and if , then typically at different parameter values .
@xcite nonetheless , in protein science , pseudo - transitions such as conformational transitions are important in the understanding of secondary structure formation and the tertiary hydrophobic - core collapse . for polymers ,
mainly the @xmath48 collapse transition , which is probably of second order , is of particular interest .
@xcite this is a real thermodynamic phase transition .
nonetheless , at least for finite systems , an additional first - order - like glassy or crystallization transition at lower temperatures is also conjectured for polymers .
@xcite all these peculiarities of finite polymer systems are also relevant for the adsorption problem we consider here . in fig .
[ figpd ] we have plotted the projection of the specific heat profile onto the solubility - temperature plane as obtained from our simulation of the 179-mer in a cavity with @xmath46 .
the color code reflects the value of the specific heat and the brighter the shading , the larger the value of @xmath49 .
black and white lines emphasize the ridges of the profile .
since we consider the specific heat as appropriate to identify pseudo - phases , these ridges mark the pseudo - phase boundaries .
as expected , the pseudo - phase diagram is divided into two main parts , the phases of adsorption and desorption .
the two desorbed pseudo - phases dc ( desorbed - compact conformations ) and de ( desorbed - expanded structures ) are separated by the collapse transition line which corresponds to the @xmath48 transition of the infinite - length polymer which is allowed to extend into the three spatial dimensions @xcite .
the region of the adsorbed pseudo - phases is much more complex , and little is known about its details , since it is relevant at lower temperatures , where conventional monte carlo methods with pivot - like updates usually tend to fail .
the presence of general phases of adsorbed - expanded ( ae ) @xcite and adsorbed - compact ( ac1 , ac2 ) conformations was postulated in adsorption studies of grafted polymers and the existence of an additional phase of surface - attached globules ( ag ) @xcite was assumed @xcite . in a recent study @xcite , it was argued that the layering transition between ac1 and ac2 is a thermodynamic phase transition .
although the polymer in our study is still relatively small , we can clearly identify pseudo - phases in fig .
[ figpd ] which can be assigned these labels , too .
those regions are separated by the black lines indicating the transitions between them .
we expect that these are transitions in the thermodynamic meaning ; only the precise location of the transition lines will still change with increasing length of the polymer .
thus , this picture confirms the previously assumed phases and it provides evidence that the ag phase is indeed there .
furthermore , we have also highlighted by white lines transitions between pseudo - phases which will probably not survive in the thermodynamic limit .
this concerns , e.g. , the higher - order layering transitions among the compact pseudo - phases ac2a@xmath50-d . in the following sections we will analyse the properties of the pseudo - phases in more detail .
the contact numbers @xmath2 and @xmath1 can be considered as system parameters appropriately describing the state of the system and are therefore useful to identify the pseudo - phases .
peaks and dips in the external - parameter dependence of self - correlations @xmath51 , @xmath52 and cross - correlations @xmath53 indicate activity in the contact - number fluctuations and , analysing the expectation values @xmath54 and @xmath55 in these active regions of the external parameters @xmath10 and @xmath9 , allow for an interpretation of the respective conformational transitions between the pseudo - phases .
= 8.8 cm in fig .
[ figc179 ] , we have plotted for the 179-mer these quantities and , for comparison , the specific heat as functions of the temperature @xmath10 at a fixed solvent parameter @xmath56 .
this example is quite illustrative as the system experiences several conformational transitions when increasing the temperature starting from @xmath57 ( see fig .
[ figpd ] ) . at temperatures very close to @xmath57 ( pseudo - phase ac1 )
all 179 monomers have contact to the substrate and 153 monomer - monomer contacts are formed .
this is the most compact contact set being possible for _ topologically two - dimensional _ , film - like conformations .
it should be noted , however , that approximately @xmath58 conformations ( self - avoiding walks ) belong to this contact set .
@xcite this high degeneracy is an artefact of the minimalistic lattice polymer model used .
it is remarkable that the conformations with the highest number of total contacts @xmath59 are film - like compact ( @xmath60 ) .
all other conformations we found possess less contacts , even the most compact contact set that dominates the five - layer pseudo - phase ac2a@xmath61 ( @xmath62 , @xmath63 , i.e. , @xmath64 ) .
the reason is that for low temperatures , those macrostates are formed which are energetically favored . entropy is not yet relevant for the @xmath56 example @xmath54 drops to 149 only up to @xmath65 .
increasing the temperature further , the situation dramatically changes , as can be seen in fig .
[ figc179 ] . in a highly cooperative process , the average number of intrinsic contacts @xmath55 significantly increases ( to @xmath66 ) at the expense of surface contacts ( @xmath54 drops to approximately 104 ) .
consequently , the strong fluctuations @xmath67 signalize a conformational transition , and the anticorrelation indicated by @xmath53 confirms that surface contacts turn into intrinsic contacts , which indirectly leads to the conclusion that the film - like structure is given up in favor of layered , spatially three - dimensional conformations .
the system has entered subphase age which is the part of the phase ag , where two - layer conformations dominate .
the subphase transition near @xmath68 from the two - layer ( age ) to the bulky regime of ag is due to the ongoing , rather unstructured expansion of the polymer into the @xmath33 direction by forming so - called surface - attached globules @xcite .
this is accompanied by a further reduction of surface contacts , while the number of intrinsic contacts changes weakly .
approaching @xmath69 , the situation is just vice versa .
intrinsic contacts dissolve and the system experiences a conformational phase transition from globular conformations in ag to random strands in ae . crossing this transition line , the system enters the good - solvent regime .
eventually , close to @xmath70 , the polymer unbinds off the substrate .
a clear signal is observed in the fluctuations of @xmath2 , i.e. , the number of average surface contacts rapidly decreases .
the expanded polymer is `` free '' and the influence of _ both _ walls is effectively steric .
this phase ( ae ) is closely related to the typical random - coil phase of entirely free and dissolved polymers in good solvent .
this example shows that a study of the contact number fluctuations is indeed sufficient to qualitatively identify and describe the conformational transitions between the pseudo - phases of the hybrid system .
for this reason , @xmath2 and @xmath1 are adequate system parameters playing a similar role as order parameters in thermodynamic phase transitions .
one of the most interesting structural quantities in studies of polymer phase transitions is the gyration tensor ( [ eqgyr ] ) . for our hybrid system
we expect that the respective components parallel ( [ eqgyrpara ] ) and perpendicular ( [ eqgyrperp ] ) to the substrate will behave differently when the polymer passes pseudo - transition lines . in order to prove this anisotropy explicitly ,
we have plotted in fig .
[ figg179 ] the expectation values , @xmath71 , and the fluctuations of these two components , @xmath72 , again for the polymer in solvent with @xmath56 . for interpreting the peaks of the fluctuations
, we have also included once more the specific heat curve for comparison .
the immediate observation is that the temperatures , where one or both gyration tensor components exhibit peaks , almost perfectly coincide with those of the specific heat . this is a strong confirmation for the phase diagram in fig .
[ figpd ] which is based on the specific heat .
obviously , even for the rather short polymer with 179 monomers , we encounter the onset of fluctuation collapse near the ( pseudo-)phase transitions .
this is very promising for future quantitative finite - size scaling analyses . at very low temperatures , i.e. , in pseudo - phase ac1
, we have argued in the previous section that the polymer - conformation is the most compact single - layer film .
this is confirmed by the behavior of @xmath73 and @xmath74 , the latter being zero in this phase .
a simple argument that the structure is indeed maximally compact is as follows .
it is well known that the most compact shape in the two - dimensional continuous space is the circle . for @xmath29 monomers residing in it , @xmath75 , where @xmath76 is the ( dimensionless ) radius of this circle .
the usual squared gyration radius is @xmath77 and therefore @xmath78 for @xmath79 .
indeed , this is close to the value @xmath80 of the ground - state conformation we identified in phase ac1 .
note that the most compact shape in the simple lattice polymer model we used in our study is a square and not a discretized circle .
@xcite = 8.8 cm near @xmath65 , the strong layering transition from ac1 to age is accompanied by an immediate decrease of @xmath73 , while @xmath74 rapidly increases from zero to about @xmath81 which is exactly the gyration radius ( perpendicular to the layers ) of a two - layer system , where both layers cover approximately the same area .
note that the single layers are still compact , but not maximally . applying the same approximation as in eq .
( [ eqcirc ] ) , the planar gyration radius for each of the two layers is now ( with @xmath82 ) @xmath83 , while we measured in this phase ( age ) @xmath84 .
this separates the subphase age from the other two - layer pseudo - phase ac2d in fig .
[ figpd ] , where the dominating conformation has perfect two - layer ( lattice ) structure with @xmath85 ( this is the same 2% difference between continuous and lattice calculation for perfect shapes as above ) .
we will discuss the conformational peculiarities in the following in more detail . the subphase transition from age to ag near @xmath68
is accompanied by a further decrease of @xmath73 whereas @xmath74 increases , i.e. , the height of the surface - adsorbed globule increases at the expense of the width .
this tendency is stopped when approaching the transition ( @xmath69 ) from the globular regime ag to the phase of expanded , but still adsorbed conformations .
while @xmath74 remains widely constant ( the fluctuation does not signalize any transition ) , the polymer strongly extends in the directions parallel to the substrate , as indicated by the peak of @xmath86 .
after unbinding from the substrate , parallel and perpendicular gyration radii behave widely isotropically ( @xmath87 ) as the influence of the isotropy - disturbing walls is weak in this regime .
it was shown in sect .
[ ssecfl ] that the contact numbers @xmath2 and @xmath1 are unique system parameters for the pseudo - phase identification of the hybrid system .
we define the restricted partition sum for a macrostate with @xmath2 surface contacts and @xmath1 monomer - monomer contacts by @xmath88 such that @xmath89 .
assuming as usual that the dominating macrostate is given by the minimum of the free energy as a function of appropriate system parameters , it is useful to define the specific contact free energy as a function of the contact numbers @xmath2 and @xmath1 , @xmath90 identifying @xmath91 as a `` micro - contact '' entropy . for given external parameters @xmath10 and @xmath9 , this relation can be used to determine the minimum of the contact free energy and therefore allows the identification of the dominant macrostate with respect to the contact numbers . in turn
, this quantity allows for an alternative representation of the pseudo - phase diagram , complementary to the one shown in fig . [ figpd ] in that it is related to the contact numbers @xmath2 and @xmath1 .
this is done by determining for ( in principle ) all values of the external parameters @xmath10 and @xmath9 the minima of the contact free energy ( [ eqrfreeb ] ) .
then the pair of values @xmath2 and @xmath1 of the minimum contact free - energy state are marked in an @xmath2-@xmath1 phase diagram .
this is shown in fig .
[ figmap179 ] , where all free - energy minima of the 179-mer for the parameter set @xmath92 $ ] and @xmath93 $ ] are included and , based on the arguments of the previous section , differently shaded according to the pseudo - phase they belong to .
the nice thing of this representation is that it allows the differentiation of continuous and discontinuous pseudo - phase transitions .
= 8.8 cm the first important observation is that the diagram is divided into two separate regions , the pseudo - phases of desorbed conformations ( dc and de ) and the remaining different phases of adsorption .
the `` space '' in between is blank , i.e. , none of these ( possible ) conformations was found to be a free - energy minimum conformation .
this shows that transitions between the adsorbed and desorbed pseudo - phases are always first - order - like .
it should be noted , that the regime of contact pairs @xmath94 lying _ above _ the shown compact phases is forbidden , i.e. , conformations with such contact numbers do not exist on the sc lattice .
the second remarkable result is that the pseudo - phases dc , de , ae , and ag are `` bulky '' , while all ac subphases are highly localized in the plot of the free - energy minima . comparing with fig .
[ figpd ] , the conclusion is that conformations in the ac phases are energetically favored ( more explicitly , for @xmath95 in ac1 and @xmath96 in the ac2 subphases ) , while the behavior in the other pseudo - phases is entropy - dominated : the number of conformations with similar contact numbers in the globular or expanded regime is higher than the rather exceptional conformations in the compact phases , i.e. , for sufficiently small @xmath97 ratios the entropic effect overcompensates the energetic contribution to the free energy .
.[tabconf ] representative minimum free - energy examples of conformations in the different pseudo - phases of a 179-mer in a cavity .
the substrate is shaded in lightgray .
[ cols="^,^,^,^",options="header " , ] the subphases ac2a@xmath50-d are strongly localized , thorn - like `` peninsulas '' standing out from the ag regime .
the discrete number and their separation leads to the conclusion that they have related structures . indeed , as can be seen in table [ tabconf ] , where we have listed representative conformations for all pseudo - phases , the few conformations dominating these subphases exhibit compact layered structures . the most compact three - dimensional conformation with 263 monomer - monomer contacts and 36 surface contacts
is favored in subphase ac2a@xmath61 and possesses five layers .
starting from this subphase and increasing the temperature , two things may happen .
a rather small change is accompanied with the transition to ac2a@xmath98 , where the number of intrinsic contacts is reduced but the global five - layer structure remains . on the other hand , passing the transition line towards ac2b , the monomers prefer to arrange in compact four - layer conformations . advancing towards ac2d ,
the typical conformations reduce layer by layer in order to increase the number of surface contacts . in ac2d
there are still two layers lying almost perfectly on top of each other .
this is similar in subphase age , where also two - layer but less compact conformations dominate . in pseudo - phase ac1
only the film - like surface layer remains .
the reason for the differentiation of the phases ac1 and ac2 of layered conformations is that the transition from single- to double - layer conformations is expected to be a real phase transition , while the transitions between the higher - layer ac2 subphases are assumed to disappear in the thermodynamic limit .
@xcite = 8.8 cm as can be seen in fig .
[ figpd ] , a transition between ac1 and the phase of adsorbed , expanded conformations , ae , is possible . since these two phases are connected in fig .
[ figmap179 ] , we expect that the transition in between is second - order - like .
indeed , this transition is strongly related with the _ two - dimensional _
@xmath48 transition since , close to the transition line , all monomers form a planar ( surface-)layer .
similarly , there is also a second - order - like transition line @xmath99 between ag and ae which separates the regions of poor ( ag : @xmath100 ) and good ( ae : @xmath101 ) solvent . also , the transition between the desorbed compact ( dc ) and expanded ( de ) conformations is second - order - like : this transition is strongly related with the well - known @xmath48 transition in three dimensions .
@xcite eventually , the transitions from the layer - phases ac2a@xmath98 , ac2b , ac2c , and age to the globular pseudo - phase ag as well as transitions between pseudo - phases dominated by the same layer type ( i.e. , between the two - layer subphases ac2d and age , and between the five - layer subphases ac2a@xmath61 and ac2a@xmath98 ) are expected to be continuous . on the other hand , the transitions among the energetically caused compact low - temperature pseudo - phases are rather first - order - like , due to their noticeable localization in the map of free - energy minima ( fig .
[ figmap179 ] ) .
the possible transitions ( see fig .
[ figpd ] ) are ac2a@xmath50ac2b , ac2b
ac2c , and ac2c ac2d , respectively .
even more interesting , however , are the transitions from the single - layer pseudo - phase ac1 to the double - layer subphases ac2d and age . in the previous sections we already discussed this transition for the special choice @xmath56 , where near @xmath65 the fluctuations of the contact numbers and the components of the gyration tensor exhibit a strong activity .
we have included into fig .
[ figmap179 ] the `` path '' of macrostates the system passes by increasing the temperature from @xmath57 to @xmath102 . at @xmath57
the system is in a film - like , single - layer state .
near @xmath65 it indeed suddenly rearranges into two layers and enters subphase age in a single step . in fig .
[ figp179](a ) we have plotted the probability distribution @xmath103 for @xmath56 and @xmath104 and it can clearly be seen that two distinguished macrostates coexist . @xcite increasing the temperature further , the system undergoes the continuous transitions from age via ag until it unfolds when entering pseudo - phase ae .
the system is still in contact with the substrate .
close to a temperature @xmath105 , however , the unbinding of the polymer off the substrate happens ( from ae to de ) .
comparing figs .
[ figmap179 ] and [ figp179](b ) , where the probability distribution at @xmath106 is shown , we see also a clear indication for a discontinuous transition .
note that we consider here the transition state , where the two minima of the free energy coincide @xcite ( see also the black dashed line in fig .
[ figpd ] ) and not the point , where the width of the distribution , i.e. , the specific heat , is maximal . since the system is finite , the transition temperature ( @xmath70 ) , as signalled by the fluctuations studied in the previous sections , deviates slightly from the transition - state temperature reported here .
in this paper , we have studied in detail the solubility - temperature ( pseudo-)phase diagram of a polymer in a cavity with an attractive substrate .
we identified the thermodynamic phases of adsorbed compact and expanded ( ac , ae ) and desorbed ( dc @xcite , de ) conformations as well as the previously not yet clearly confirmed phase of adsorbed globules ( ag ) .
although the polymer in our study possessed only @xmath107 monomers , these phases are expected to be stable also in the thermodynamic limit @xmath108 .
other noticeable phase transitions in the compact - globular adsorbed regime ( ac1-ac2d , ac1-age ) are the energetic layering transitions from film - like surface - layer to double - layer conformations which are also believed to survive the thermodynamic limit .
@xcite in addition , further subphases of higher - order layers were found in low - temperature regions and bad solvent ( ac2a@xmath50 , ac2b , and ac2c ) .
the most compact three - dimensional conformation found is cube - like and forms five layers ( in subphase ac2a@xmath61 ) . the ( pseudo-)phase diagram is based on the specific - heat profile as a function of temperature and reciprocal solubility .
although this profile allows for the identification of phases and their boundaries it does tell little about the conformational transitions between the phases . for this purpose we considered expectation values and fluctuations for the numbers of monomer - surface contacts , @xmath2 , and intrinsic monomer - monomer contacts , @xmath1 , separately .
these contact numbers turned out to be sufficient to describe the macrostate of the system and therefore are useful to describe the conformations dominating the different phases .
this view was completed by an exemplified study of the anisotropic behavior of the gyration tensor components of the polymer parallel and perpendicular to the substrate .
another central aspect was the classification of the conformational transitions between the ( pseudo-)phases . based on the contact numbers @xmath2 and @xmath1 , we defined an appropriate free energy and studied the distribution of the minima in the @xmath2-@xmath1 space . from this kind of free - energy landscape
, we found strong indications that the binding - unbinding transitions between the adsorbed and desorbed phases are first - order - like .
this was also observed for the layering transitions . on the other hand ,
the transitions across the line separating good and poor solvent , i.e. , between the compact ( or globular ) and the expanded conformations , are rather second - order - like .
this is in coincidence with the known behavior of free polymers at the @xmath48 collapse transition in two and three dimensions . since the experimental equipment and the technological capabilities
have nowadays reached an enormous standard of high single - molecular resolution , we expect that it should be possible to verify experimentally not only the existence of the described thermodynamic phases , but also the pseudo - phases being only relevant for finite polymers and specific to their lengths .
this work is partially supported by the dfg ( german science foundation ) grant under contract no .
ja 483/24 - 1 .
some simulations were performed on the supercomputer jump of the john von neumann institute for computing ( nic ) , forschungszentrum jlich , under grant no .
199 s. brown , nature biotechnol . *
15 * , 269 ( 1997 ) .
s. r. whaley , d. s. english , e. l. hu , p. f. barbara , a. m. belcher , nature * 405 * , 665 ( 2000 ) .
k. goede , p. busch , and m. grundmann , nano lett .
* 4 * , 2115 ( 2004 ) .
r. l. willett , k. w. baldwin , k. w. west , and l. n. pfeiffer , proc .
acad.sci .
( usa ) * 102 * , 7817 ( 2005 ) .
j. j. gray , curr .
opin . struct .
* 14 * , 110 ( 2004 ) .
g. reiter , phys .
lett . * 87 * , 186101 ( 2001 ) .
j. forsman and c. e. woodward , phys .
* 94 * , 118301 ( 2005 ) .
e. nakata , t. nagase , s. shinkai , and i. hamachi , j. am . chem . soc . * 126 * , 490 ( 2004 ) .
t. bogner , a. degenhard , and f. schmid , phys .
lett . * 93 * , 268108 ( 2004 ) .
e. balog , t. becker , m. oettl , r. lechner , r. daniel , j. finney , and j. c. smith , phys .
lett . * 93 * , 028103 ( 2004 ) .
m. ikeguchi , j. ueno , m. sato , and a. kidera , phys .
lett . * 94 * , 078102 ( 2005 ) .
n. gupta and a. irbck , j. chem .
phys . * 120 * , 3983 ( 2004 ) .
cheng and p .- y .
lai , phys .
e * 71 * , 060802(r ) ( 2005 ) .
g. m. foo and r. b. pandey , phys .
lett . * 80 * , 3767 ( 1998 ) ; phys .
e * 61 * , 1793 ( 2000 ) .
r. hegger and p. grassberger , j. phys .
a * 27 * , 4069 ( 1994 ) .
t. vrbov and s. g. whittington , j. phys .
a * 29 * , 6253 ( 1996 ) ; j. phys .
a * 31 * , 3989 ( 1998 ) ; t. vrbov and k. prochzka , j. phys .
a * 32 * , 5469 ( 1999 ) .
y. singh , d. giri , and s. kumar , j. phys .
a * 34 * , l67 ( 2001 ) ; r. rajesh , d. dhar , d. giri , s. kumar , and y. singh , phys .
e * 65 * , 056124 ( 2002 ) .
m. s. causo , j. chem .
phys . * 117 * , 6789 ( 2002 ) .
j. krawczyk , t. prellberg , a. l. owczarek , and a. rechnitzer , europhys . lett . * 70 * , 726 ( 2005 ) .
j .- h . huang and s .- j .
han , j. zhejiang univ .
sci . * 5 * , 699 ( 2004 ) .
m. breidenreich , r. r. netz , and r. lipowsky , europhys .
lett . * 49 * , 431 ( 2000 ) ; eur .
j. e * 5 * , 403 ( 2001 ) .
p. benetatos and e. frey , phys .
e * 70 * , 051806 ( 2004 ) .
f. celestini , t. frisch , and x. oyharcabal , phys .
e * 70 * , 012801 ( 2004 ) .
j. krawczyk , t. prellberg , a. l. owczarek , and a. rechnitzer , j. stat .
( 2004 ) , p10004 .
m. bachmann and w. janke , phys .
lett . * 95 * , 058102 ( 2005 ) .
m. bachmann and w. janke , submitted ( 2005 ) .
m. bachmann and w. janke , phys .
lett . * 91 * , 208105 ( 2003 ) ; j. chem .
* 120 * , 6779 ( 2004 ) .
a related flat - histogram chain - growth method was introduced in : t. prellberg and j. krawczyk , phys .
* 92 * , 120602 ( 2004 ) .
p. grassberger , phys .
e * 56 * , 3682 ( 1997 ) .
m. n. rosenbluth and a. w. rosenbluth , j. chem .
* 23 * , 356 ( 1955 ) .
p. d. de gennes , _ scaling concepts in polymer physics _ ( cornell university press , ithaca , 1979 ) .
f. rampf , w. paul , and k. binder , europhys .
lett . * 70 * , 628 ( 2005 ) . in very poor solvent ( i.e. , very large values of @xmath9 )
, the dc phase may degenerate , at least for the finite - length polymers , into separate globular and crystalline subphases ( see , e.g. , ref @xcite ) , but this effect is not in the focus of the present study of adsorption properties . in our recent study of a 100-mer in a cavity ,
@xcite we have still denoted ae as ae1 and ag as ae2 , because the differentiation between these two phases was still relatively diffuse due to the small size of the system . for the longer chain we are considering here ,
the identification of these phases is much clearer .
an enormous advantage of the simple - sampling algorithms based on rosenbluth sampling @xcite compared to importance - sampling methods is that they allow the approximation of the degeneracy ( `` density '' ) of states _ absolutely _ , i.e. , free energy and entropy can explicitly be determined .
this is also true for most compact three - dimensional structures which are cube - like .
therefore , globules in the lattice model are rather cubes than spheres ! c. borgs and w. janke , phys .
lett . * 68 * , 1738 ( 1992 ) ; w. janke , phys .
b * 47 * , 14757 ( 1993 ) .
w. janke , _ first - order phase transitions _
, in : computer simulations of surfaces and interfaces , nato science series , ii .
mathematics , physics and chemistry vol . * 114 * , proceedings of the nato advanced study institute , albena , bulgaria , 9 20 september 2002 , edited by b. dnweg , d. p. landau , and a. i. milchev ( kluwer , dordrecht , 2003 ) , p. 111 . strictly speaking ,
the proper definition of the coexistence point is the temperature , where the weights under the two peaks are equal , see , e.g. , refs . [ ] . | in a contact density chain - growth study we investigate the solubility - temperature pseudo - phase diagram of a lattice polymer in a cavity with an attractive surface .
in addition to the main phases of adsorbed and desorbed conformations we find numerous subphases of collapsed and expanded structures . |
there is increasing interest in two - species bose - einstein condensate ( tbec ) in last two decades since it was first studied theoretically by ho and shenoy@xcite , and was achieved experimentally by myatt , et al.@xcite .
the tbec provides an important tool to clarify the inter - species and intra - species interactions . besides
, molecules in well defined rovibrational levels with permanent electric - dipole moment can be thereby obtained , and they are valuable for studying the strongly interacting quantum gases.ospe06,webe08,ni08,bara08 . experimentally , the rb - cs mixture has been realized via a magnetic trap @xcite or an optical trap @xcite .
the rb - yb mixture has been realized via a combined magneto - optical trap nemi09 .
the k - rb , na - rb , sr - rb and the isotopic mixtures @xmath5rb-@xmath6rb , @xmath7yb-@xmath8yb , @xmath9sr-@xmath10sr , and @xmath10sr-@xmath11sr have also been realized and studied recently ( refer to the references listed in @xcite ) .
there are a number of theoretical literatures dedicated to the tbec ho96,esry97,pu98,timm98,ao98,chui99,tripp00,ribo02,chui03,luo07,xu09,shi11,gau10,nott15 , scha15,inde15,roy15,luo08,polo15 .
the ground state ( g.s . ) is found to have three phases , symmetric miscible phase , symmetric immiscible phase , and asymmetric immiscible phase , depending on the parameters .
there are a number of parameters ( the strengths of the intra- and inter - species interactions , the particle numbers , the masses of atoms , and those for the trap ) affecting the behavior of the system .
to clarify the effects of them is a main topic of study .
this paper is dedicated to obtain the phase - diagrams . how the spatial configuration of the g.s .
is affected by the parameters can be thereby visualized . for this purpose
, effort is made to divide the whole parameter - space into zones , each supports a specific spatial configuration .
based on the division , two types of phase - diagrams are plotted .
related analysis is supported by analytical formalism .
the emphasis is placed on the qualitative aspect .
the particle numbers are considered to be large ( @xmath12 ) so that the thomas - fermi approximation ( tfa ) can be applied .
the symmetric states are obtained by solving the coupled gross - pitaevskii equations ( cgp ) in an analytical way . whereas the asymmetric states are obtained by introducing a model .
both repulsive and attractive inter - species interactions are considered .
two critical cases , namely , ( i ) the repulsive inter - species interaction matches the repulsive intra - species interaction and ( ii ) the attractive inter - species interaction nearly cancels the repulsive intra - species interaction , are taken into account .
there are @xmath13 a - atoms with mass @xmath14 and interacting via @xmath15 , and @xmath16 b - atoms with @xmath17 and @xmath18 . the a- and b - atoms are interacting via @xmath19 .
their spin - degrees of freedom are considered as being frozen .
they are trapped by the isotropic parabolic potentials @xmath20 ( @xmath21 or @xmath22 ) . in the g.s . @xmath23
the atoms of the same kind will have the same wave function @xmath24 which is most advantageous to binding .
thus , @xmath25 it is well known that @xmath23 and the trap might not have the same symmetry .
when the interspecies interaction is sufficiently repulsive , the two kinds of atoms might separate from each other in an asymmetric way .
nonetheless , we consider first the case that the g.s .
is symmetric . in this case ,
let @xmath26 and @xmath27 .
we introduce a mass @xmath28 and a frequency @xmath29 . @xmath30 and @xmath31
are used as units for energy and length in this paper .
we further introduce @xmath32 and a set of parameters @xmath33 , @xmath34 , @xmath35 , @xmath36 .
this set is called the weighted strengths ( w - strengths ) . under the tfa ,
the cgp for @xmath37 and @xmath38 appear as @xmath39where @xmath40 and @xmath41 for the inter - species interaction have the same sign ( @xmath42 or @xmath43 ) while @xmath44 and @xmath45 for the intra - species interaction are considered to be positive only .
the chemical potential for the a - atoms ( b - atoms ) is equal to @xmath46 ( @xmath47 ) .
the normalization @xmath48 and @xmath49 are required . @xmath50 and @xmath51 are safely assumed .
eqs.([eq5a],[eq5b ] ) demonstrate that the combined effects of the nine dynamical parameters @xmath52 , @xmath53 , @xmath54 , @xmath55 , and @xmath56 ( @xmath21 and @xmath22 ) are sufficiently embodied by the four w - strengths .
thus an approach based on the w - strengths would lead to a simplification .
based on the w - strengths , we define @xmath57 where @xmath58 . @xmath59 and @xmath60 are crucial in constituting the solutions of the cgp as shown below .
it has been known that the spatial configuration of the symmetric g.s . has two phases .
when each kind of atoms form a core surrounding the center , then it is named miscible phase denoted as ( a , b ) ( if the b - atoms have a broader distribution ) or ( b , a ) .
when a kind of atoms form a core while the other kind form a shell away from the center , it is named immiscible phase denoted as [ a - b ] ( if the b - atoms form the shell ) or [ b - a ] .. more strictly , the miscible phase has both @xmath61 and @xmath62 being nonzero , while the immiscible phase has one of them being zero .
the analytical solutions of the symmetric g.s . under the tfa have been worked out previously @xcite . nonetheless , in terms of the w - strengths , the analytical solutions could have a considerably simpler form .
this simplification is valuable because it substantially benefits a more detailed analysis .
thus , the simplified solutions are given in the appendix . besides , the division of the whole parameter - space into zones is a crucial point in our paper and is also included in the appendix . with the division ,
various types of phase - diagrams can be plotted .
based on the analytical solutions of the cgp and the division of the parameter - space which are given in the appendix , the first type of phase - diagrams are shown in fig.1 .
let us define a critical value @xmath63 . in fig.1
the lower and upper ends of the dash - dot curve have @xmath64 and @xmath65 , respectively .
the two ends of the dash curve have @xmath66 and @xmath67 , respectively .
accordingly , @xmath68 contains three critical values @xmath69 , 1 , and @xmath70 . in fig.1a ,
when @xmath71 , the system collapses ( see below ) .
when @xmath72 , the g.s .
is in miscible phase .
there is a competition between ( a , b ) and ( b , a ) phases depending on @xmath1 . when @xmath73 is sufficiently small so that @xmath74 , then the distribution of the a - atoms will be narrower and the g.s
. will be in ( a , b ) ( refer to the appendix ) , or vice versa .
thus , from the condition @xmath75 , we can define a critical ratio @xmath76 when @xmath77 @xmath78 , the g.s .
is in ( a , b ) , or otherwise in ( b , a ) .
when @xmath79 , the distributions of the a- and b - atoms overlap completely ( refer to the appendix ) .
when @xmath80 , @xmath81 , thus the condition for the g.s . in ( a , b ) is @xmath82 ( in other words , the ( a
, b ) phase would be more probable if @xmath83 is relatively weak and/or @xmath13 is small ) . when @xmath84 , @xmath85 .
when @xmath86 increases further , the ( a , b ) phase disappears and the a - atoms begin to be distributed more outward .
when @xmath87 , there is a competition between the [ b - a ] and ( b , a ) phases depending also on @xmath1 . when @xmath73 is sufficiently small to enable @xmath88 , the a - atoms will form a shell and the g.s .
is in immiscible phase . otherwise , in miscible phase .
when @xmath89 , the zone of [ b - a ] covers nearly the whole range of @xmath1 , and thus the immiscible phase becomes dominant .
when @xmath90 , the lowest symmetric solutions of the cgp have the b - atoms being distributed more outward ( in [ a - b ] or ( a , b ) phases ) . nonetheless , when @xmath86 increases and crosses @xmath91 , a transition from [ b - a ] to [ a - b ] will occur . during this transition , the atoms in the shell and in the core interchange their roles ( i.e. , those in the core jump to the shell , and vice versa ) .
this causes a remarkable increase in energy ( as shown below ) .
accordingly , when @xmath92 , the symmetric configuration is no more stable .
for this reason , we shall more concentrate on the case @xmath86 @xmath93 . when @xmath94 , @xmath70 also @xmath95 , and the areas of the zones of [ b - a ] and [ a - b ] both become zero , and
therefore the immiscible phase does not appear ( fig.1b belongs to this case ) .
thus the difference between @xmath96 and @xmath97 is a basic requirement for the appearance of the immiscible phase . for the case @xmath98
( fig.1c belong to this case ) , a similar discussion can be performed where the a- and b - atoms interchange their roles . to conclude , in most cases , the g.s .
is in miscible phases .
there is a critical value @xmath78 to judge either ( a , b ) or ( b , a ) would win . besides , there is a critical value @xmath69 .
when @xmath99 and when @xmath13 is so small that @xmath100 , the immiscible [ b - a ] phase will emerge in the zone lying between the curve @xmath101 and the vertical line @xmath66 ( where @xmath102 ) .
note that the [ b - a ] phase requires thin this zone @xmath103 and @xmath104 ( refer to the appendix ) , it implies that the w - strengths should be appropriately given .
the condition for [ a - b ] can be similarly deduced .
thus the zones of the immiscible phase appear only by the two sides of the vertical line @xmath66 .
the area of the zones depends on how far @xmath69 deviates from 1 .
recall that @xmath69 contains the factor @xmath105 .
thus the ratio of the two frequencies is a very sensitive factor to affect the spatial configuration ( say , a small reduction of @xmath106 will reduce @xmath107 , the zone of [ b - a ] shown in fig.1a will be thereby broader , and thus favors the a - atoms to form a shell ) .
the spatial configurations , namely , the patterns of the wave functions have already been plotted in many literatures . to avoid tedious , only two particular cases are plotted here .
\(i ) for attractive @xmath2 with @xmath108 in this case both @xmath109 and @xmath60 are positive and tend to @xmath110 .
let us assume @xmath111 .
accordingly , the a - atoms have a narrower distribution .
since the radius of the inner core @xmath112(refer to the appendix ) , it tends to zero . on the other hand , the outer border @xmath113^{1/5}$ ] .
when @xmath114 , @xmath115 .
recall that @xmath116 is the limit of @xmath117 when @xmath114 .
thus , if @xmath1 is so given that it would considerably smaller than the limit , then @xmath118 would be finite while @xmath119 . in this case
, we will have a very dense core together with a thin tail extending outward formed by the b - atoms .
alternatively , if the w - strengths are so given that @xmath120 , then the tail is formed by the a - atoms . whereas , when @xmath121 , the tail disappears and all the atoms stay inside a very small core .
numerical calculations support this suggestion as shown in fig.2 , where the tail is long in 2a and 2b , but very short in 2c .
the extremely high density at the center implies that the system tends to collapse .
\(ii ) for repulsive @xmath2 with @xmath122 when @xmath123 , the two species tend to separate completely from each other ( refer to the appendix ) .
examples for the cases with @xmath124 are given in fig.3 characterized by having a shell attached to the outward border of a core .
in fig.3 @xmath16 remains unchanged , while the increase of @xmath13 pushes the shell of the b - atoms more and more outward .
when @xmath125 and @xmath126 have been known , the total energy of the system ( neglecting the kinetic energy ) is @xmath127 where @xmath128 , @xmath129 , @xmath130 , and so on .
examples of @xmath131 ( where @xmath132 ) are plotted in fig.4 . where the five curves for symmetric states goes up with @xmath86 as expected . in all cases there is a notable jump when @xmath86 increases and crosses @xmath91 ( strictly , only the case that @xmath86 tends to @xmath133 can be calculated ) .
this jump is associated with the transition from @xmath134 $ ] to @xmath135 $ ] mentioned above .
this figure demonstrates that the increase of energy in the jump can be quite large .
it implies that very high - lying collectively excited states might exist .
all the horizontal lines are for asymmetric states , they are obtained via a model shown below .
let @xmath136 and @xmath137 ( @xmath138 ) be the axial coordinates , and two parameters @xmath139 and @xmath140 are introduced .
then , it is assumed that when @xmath141 @xmath142 and when @xmath143 @xmath144 in other words , we have proposed a model in which a ball with radius @xmath140 is divided into two sides separated by the plane @xmath145 .
the a - atoms and b - atoms stay separately in the two sides .
the orientation of the z - axis is irrelevant ( because the trap is isotropic ) . from the normalization
, we have @xmath146 the two parameters @xmath140 and @xmath139 are determined by minimizing the total energy @xmath147 ( which is calculated via eq.([8 ] ) where the contribution from the kinetic energy has been neglected ) .
the resultant @xmath131 are plotted in fig.4 and they appear as horizontal lines . in various cases , due to the jump appearing at @xmath148 ,
the asymmetric states become lower in energy once @xmath149 ( i.e. , they become the g.s . ) .
this is clearly shown in the figure .
recalled that the kinetic energy has been neglected . when the particle number is huge , the ratio of the kinetic energy over the total energy is small ( say , in an estimation given in @xcite , the ratio is 6% when @xmath150 ) .
thus the neglect will not spoil the qualitative results .
the second type of phase - diagrams are given in fig.5 .
the physics extracted from fig.5 and from fig.1 are similar . therefore , we just mention a few points : ( i ) the zones for the immiscible phase appear only in the neighborhood of the horizontal line at which @xmath148 .
( ii ) for all the cases with @xmath122 , the immiscible and miscible phases are competing .
when @xmath151 is much stronger than @xmath83 , the b - atoms will form a shell and the g.s .
is in the [ a - b ] phase .. when @xmath152 is not so strong , the b - atoms will be closer to the core and the [ a - b ] becomes ( a , b ) .
when @xmath151 reduces further , the b - atoms will have a narrower distribution while the a - toms will be distributed more outward , and the ( a , b ) .
becomes ( b , a ) . when @xmath151 becomes very weak , the a - atoms will form a shell and the ( b , a ) .
becomes [ b - a ] .
( iii ) when @xmath86 decreases further from @xmath91 , immiscible phase disappears .
\(i ) due to the introduction of the w - strengths , analytical solutions of the cgp can be expressed in a considerably simpler form to facilitate related analysis .
\(ii ) a model for the asymmetric g.s . has been proposed .
the competition between the symmetric and asymmetric states has been evaluated .
a big gap in energy was found during the [ a - b ] to [ b - a ] ( or reversely ) transition which takes place when @xmath86 crosses @xmath91 . due to the gap
, the asymmetric states become the g.s . once @xmath149 .
it has been known that , in the low - density region , the overlap of the two wave functions will be remarkably underestimated under the tfa . thus the energy @xmath153 will be therefore underestimated .
this would affect the evaluation of the gap .
however , the gap appears only when a kind of atoms tend to leave completely from the other kind . in this case , the percent of @xmath153 in the total energy is extremely small .
thus , the existence of the gap , which is the difference of the two total energies , will not be spoiled by using the tfa .
\(iii ) the whole parameter - space has been divided into zones each supports a specific spatial configuration .
based on the division two types of phase - diagrams have been plotted . making use of the phase - diagrams where the boundaries have analytical forms , the effects of various parameters
can be clarified not only qualitatively but also quantitatively .
\(iv ) additional effort is made to study the feature of the g.s .
when @xmath154 and @xmath108 . in the former case the shell tends to separate from the core completely ,
while in the latter the system tends to collapse .
\(v ) we have found a critical point @xmath69 which is also useful in application .
one can tune @xmath13 and/or @xmath106 to make @xmath69 tends to @xmath65 to prohibit the formation of a shell ( i.e.,the shell can not be formed even if @xmath86 is sufficiently strong ) , or to make @xmath69 differs remarkably from @xmath65 to facilitate the formation of a shell ( i.e.,the shell can be formed even if @xmath86 is rather weak ) .
the point @xmath69 is associated with the case that a kind of atoms tend to leave completely from the center . in this case
tfa is still applicable in the qualitative sense ( comparing fig.1b and 1c of @xcite , where 1b is closer to the above case ) .
\(vi ) we have found another critical point @xmath78 which marks the complete overlap of @xmath155 and @xmath156 .
this finding has potential usage . in experiments
, one can tune the tunable parameters ( say , the trap frequency and/or the strengths ) so that the two density profiles overlap .
then , once all the other parameter are known , @xmath1 can be directly known from eq.([7 ] ) .
since the particle numbers are in general difficult to determine precisely , this approach provides an auxiliary way helpful for determining the numbers .
it was found that in the case that both @xmath155 and @xmath156 are nonzero at @xmath157 and close to each other , the tfa wave functions deviate very slightly from those beyond the tfa ( comparing fig.1a and 1b of @xcite ) .
therefore , eq.([7 ] ) is reliable because it is just associated with this case .
supported by the national natural science foundation of china under grants no.11372122 , 11274393 , 11574404 , and 11275279 ; the open project program of state key laboratory of theoretical physics , institute of theoretical physics , chinese academy of sciences , china ; and the national basic research program of china ( 2013cb933601 ) .
where @xmath109 and @xmath60 are defined in eq.([6 ] ) depending on the w - strengths directly , @xmath161^{2/5 } \\
x_{2 } & = & ( \varepsilon _ { 2}-\beta _ { 2}x_{1})/\alpha _ { 2 } \label{eq16}\end{aligned}\]]furthermore , when @xmath162 and @xmath163 are known , @xmath164 is related to them as @xmath165 . when @xmath168 , both @xmath37 and @xmath38 are zero .
thus @xmath169 and @xmath170 mark the outward borders of the a - atoms and b - atoms , respectively .
one can check directly that the above @xmath37 and @xmath38 satisfy the cgp , they are normalized , and they are continuous at the common border of the domains ( however their derivatives are not ) . obviously , the above solution would be physically meaningful only if the w - strengths are so preset that @xmath171 , @xmath172 , and @xmath173 .
the latter can be satisfied when @xmath174 is preset .
besides , in order to have both @xmath155 and @xmath156 being @xmath175 at @xmath157 , @xmath176 ( or @xmath177 , equivalently @xmath178^{2/5}\geq 2\beta _ { 2}y_{1}$ ] ) is required . due to this constraint , the phase ( a , b ) could emerge only in a specific zone inside the whole parameter - space .
this zone is bound by the surfaces @xmath179 and @xmath180 . inside this zone @xmath171 , @xmath174 , and @xmath181 hold .
referred to fig.1 and 5 . for attractive
@xmath2 , @xmath181 holds always .
thus the border @xmath180 does not appear .
accordingly , the zone for ( a , b ) is bound by the surfaces @xmath182 and @xmath183 ( where @xmath184 ) .
however , when @xmath114 or @xmath185 , both @xmath109 and @xmath60 tend to @xmath186 . hence , @xmath169 becomes extremely small and the system is highly dense at the center .
this will lead to a collapse . due to the collapse , the case @xmath187 is not considered . for ( b , a )
, the solution can be similarly obtained from the above formulae by an interchange of @xmath37 and @xmath38 together with an interchange of the indexes 1 and 2 .
thus , associated with the zone for ( a , b ) , there is a partner - zone for ( b , a ) , where the a- and b - atoms interchange their role .
since @xmath204 and @xmath205 should be positive in the second domain , @xmath171 and @xmath206 are both necessary .
in particular , the latter condition @xmath206 assures the emergence of the shell formed by the b - atoms ( because a negative @xmath60 assures that @xmath156 increases with @xmath158as shown in eq.([eq24 ] ) ) . note that , for attractive @xmath2 , we know from the definitions of @xmath109 and @xmath60 that they must have the same sign .
thus the immiscible phase does not exist when the interspecies interaction is attractive ( refer to fig.1 and 5 ) .
therefore , @xmath86 is considered as positive in the follows .
obviously , @xmath207 is further required to assure that the solution is meaningful .
when @xmath208 and with the preset @xmath171 and @xmath206 the condition @xmath209 can be satisfied if @xmath210 .
besides , from eqs.([eq19],[eq20 ] ) we have then , @xmath216 .
thus , when @xmath60 is negative and @xmath217 holds , the condition @xmath218 is satisfied . to conclude , if we can find out a @xmath193 satisfies eq.([eq21 ] ) and is lying in the range @xmath219 , then the @xmath37 and @xmath38 given above is a physically meaningful solution . by solving eq.([eq21 ] ) numerically , a unique @xmath193 ranging inside the above range let us find out the zone in the parameter - space that supports the @xmath134 $ ] solution . when @xmath220 , we have @xmath221 , @xmath222 , and @xmath223
. thus the wave functions of the @xmath134 $ ] and those of ( a , b ) with @xmath180 overlap .
recall that @xmath180 is a border of ( a , b ) .
thus the zone of @xmath134 $ ] is adjacent to the zone of ( a , b ) , and they have the common border @xmath180 . when @xmath123 ( or @xmath224 ) , we know from eq.([eq27 ] ) that @xmath225 , thus the shell tends to separate from the core completely .
examples are shown in fig.3 .
since @xmath226 is not allowed , @xmath227 should be the other border .
note that , when @xmath227 , the cgp has no physical solution .
thus , a complete separation between the shell and the core is not allowed .
to conclude , the zone of @xmath134 $ ] requires @xmath171 and @xmath206 and is bound by the borders @xmath180 and @xmath227 . the case with @xmath228
is associated with metastable states , they can be similarly discussed , if necessary .
the partner zone of @xmath134 $ ] , namely , @xmath229 $ ] , can be similarly obtained by interchanging @xmath37 and @xmath38 , and interchanging the indexes 1 and 2 . | under the thomas - fermi approximation , a relatively much simpler analytical solutions of the coupled gross - pitaevskii equations for the two - species bec have been derived .
additionally , a model for the asymmetric states has been proposed , and the competition between the symmetric and asymmetric states has been evaluated .
the whole parameter - space is divided into zones , each supports a specific phase , namely , the symmetric miscible phase , the symmetric immiscible phase , or the asymmetric phase . based on the division the phase - diagrams against any set of parameters can be plotted .
thereby , the effects of these parameters can be visualized .
there are three critical values in the inter - species interaction @xmath0 and one in the ratio of particle numbers @xmath1 .
they govern the transitions between the phases .
two cases , ( i ) the repulsive @xmath2 matches the repulsive @xmath3 , and ( ii ) the attractive @xmath2 nearly cancels the effect of the repulsive @xmath4 have been particularly taken into account .
the former leads to a complete separation of the two kinds of atoms , while the latter lead to a collapse .
finally , based on an equation derived in the paper , a convenient experimental approach is proposed to determine the ratio of particle numbers . * correspondence to [email protected] |
person detection has made tremendous progress over the last decade @xcite .
standard methods work best on pedestrians : upright people in fairly simple , predictable poses , and with minimal interaction and occlusion between the person instances .
unfortunately , people in real images are not always so well - behaved
! plenty of in - the - wild images contain multiple people close together , perhaps with their limbs intertwined , faces close , bodies partially occluded , and in a variety of poses .
a number of computer vision applications demand the ability to parse such natural images into individual people and their respective body parts for example , fashion @xcite , consumer photo analysis , predicting inter - person interactions @xcite , or as a stepping stone towards activity recognition , gesture , and pose analysis .
current methods for segmenting person instances @xcite take a top - down approach .
first they use a holistic person detector to localize each person , and then they perform pixel level segmentation . limited by the efficiency and performance of person detectors , such methods are slow when dealing with people at unknown scales and orientations .
furthermore , they suffer when presented with close or overlapping people , or people in unusual non - pedestrian - like body poses @xcite .
we propose a new approach to detangle people and their body parts in multi - person images .
reversing the traditional top - down pipeline , we pose the task as a region assembly problem and develop a bottom - up , purely region - based approach .
given an input image containing an unknown number of people , we first compute a pool of regions using both body - part semantic segmentations and object proposals .
regions in this pool are often fragmented body parts and often overlap . despite their imperfections ,
our method automatically selects the best subset and groups them into human instances .
see fig .
[ fig : teaser](a , b ) . to solve this difficult jigsaw puzzle
, we formulate an optimization problem in which parts are assigned to people , with constraints preferring small overlap , correct sizes and spatial relationships between body parts , and a low - energy association of body part regions to their person instance .
we show that this problem can be solved efficiently using decomposition and a branch and bound method .
[ fig : teaser](c ) shows an example result from the proposed method .
note that we not only estimate pixel - level body part maps , but we also indicate which part belongs to whom " , even in a crowded scene with occluding people .
experiments on three datasets show our method strongly outperforms an array of existing approaches , including bounding box detectors , cnn region proposals , and human pose detectors .
furthermore , we show the advantage of the proposed optimization scheme as compared to simpler inference techniques .
finally , we demonstrate our person detangler applied to _
proxemics recognition _
@xcite , where fine - grained estimation of body parts and body part owners is valuable to describe subtle human interactions ( e.g. , is he holding her hand or her elbow ? ) .
[ fig : teaser ] most previous methods for human instance segmentation require a person detector . in @xcite a multi - part pedestrian detector is combined with an mcmc method for walking crowd segmentation .
a pedestrian detector is used in @xcite to find people instances in bounding boxes before instance segmentation .
joint pose estimation and segmentation for single subjects have been proposed in @xcite .
multiple people instance segmentation in tv shows has been studied in @xcite using the detector crf scheme , which combines a person detector and a pixel - level crf to achieve accurate results . in @xcite ,
people detection bounding boxes are verified using face detections and then grabcut is used to refine the instance segmentation . in @xcite a pictorial structure method @xcite
is used to detect candidate human instances .
sequential assignment is used to fit the human instance masks to image data . from instance masks ,
detailed human segmentation and body part regions are further estimated using the crf . whereas existing methods largely take the strategy of first detecting individual people and then segmenting their parts
, we propose a reversal of this conventional pipeline .
in particular , we propose to start with a pool of regions that are segments or sub - regions of body parts on multiple people , and then jointly assemble them into individuated person segments .
the advantage of not depending on a holistic person detector is not only because these detectors have high computational complexity ( especially if we have to handle people with unknown orientations and scales ) , but also because it is still a difficult problem for person detectors to deal with complex human poses , inter - person interactions , and large occlusions , which our target images may include . compared to previous detector - based methods ,
our approach is more efficient and gives better results .
deep learning approaches have been studied in the joint detection and segmentation scheme @xcite , related to rcnn @xcite , though the authors target generic pascal object detection as opposed to person individuation and body part labeling .
their method starts from object region proposals such as @xcite , and each region is classified as a target , such as a human subject , by using features on both color images and binary image masks .
potentially , such a method can be scale and rotation invariant and fast .
the challenge is how to propose complete whole object regions , such as the whole mask of a person .
this is often a difficult task due to the variation of color and texture on a single subject , the thin structure of human limbs , and arbitrary human poses .
our proposed method also uses region proposals , but our method allows fragmented sub - regions and is able to reassemble the broken regions back to human body parts .
part voting approaches have been intensively studied for human or object instance segmentation . in @xcite ,
boundary shape units vote for the centers of human subjects . in @xcite , the poselets vote for the centers of people instances .
the poselets that cast the votes are then identified to obtain the object segmentation . in @xcite
the object boundary is obtained by reversely finding the activation parts used in the voting .
similar to the hough transform , such a voting approach is more suitable to targets that have relatively fixed shape .
our proposed method finds the optimal part assembly using articulation invariant constraints instead of simply voting for the person center ; it therefore can be used to segment highly articulated human subjects . our method is also related to human region parsing , in that we segment and label each person s body part regions .
human region parsing has been mostly studied for analyzing body part regions of a single person @xcite . to handle multiple people , in @xcite a pedestrian detector
is used to find the bounding box of each single person .
finding people with arbitrary poses using a bounding box detector is still a hard problem , whereas our method naturally handles multiple people with complex interactions and poses .
part segmentation has recently been used to improve semantic segmentation of animals in @xcite , but the pairwise crf method can not individuate multiple animal instances .
in contrast , our method is able to individuate tangled people with complex poses .
our work is also distantly related to human pose estimation , which has been intensively studied on depth images @xcite and on color images using pictorial structure methods @xcite and cnns @xcite .
however , unlike our approach , human pose estimation methods usually do not directly give the instance and body part region segmentation .
our method produces multiple human segmentations without extracting human poses . in summary ,
the main contributions of this paper are : ( 1 ) we tackle the new problem of multiple person instance individuation and body part segmentation from region assembly .
( 2 ) we propose a novel linear formulation . ( 3 )
we propose a lagrangian relaxation method to speed up lower bound estimation , with which we solve the optimization using fast branch and bound .
our experiments show that our method is fast and effective , outperforming an array of alternative methods , and improving the state - of - the - art on proxemics recognition .
we first overview our approach ( sec . [ sec : overview ] ) , then present the big picture formulation of region assembly as a graph labeling problem ( sec .
[ sec : labeling ] ) .
next we describe in detail how we implement the components of that formulation ( sec .
[ sec : details ] ) . and , we introduce our efficient optimization approach ( sec .
[ sec : opt ] ) .
finally we discuss parameter setting ( sec .
[ sec : params ] ) .
there are different ways to generate regions and sub - regions of human body parts . in this paper , we start with a large set of body part region proposals using three sources : a cnn trained for semantic segmentation of body parts ( details below ) , generic region proposals @xcite , and the intersection between the previous two kinds of regions
. some of the region proposals may already give body part regions of separate human instances , or more likely they are partial sub - regions of body parts .
many proposal regions do not correspond to body part regions , or may be the union of two individuals body parts .
our goal is to select a subset of regions from these proposals and reassemble them to individuate human instances and the associated body parts .
intuitively , a good configuration should have arm , leg , torso , and head regions in proportional sizes , and part regions should follow correct neighborhood relations .
we show how to search for the best jigsaw puzzle assembly meeting those criteria .
in particular , we search for the optimal set of regions so that they minimize an objective function
. let @xmath0 be the set of overlapped regions or sub - regions of different body parts .
let @xmath1 be a vector of integers that indicate a specific region in @xmath0 is assigned to a person @xmath2 from @xmath3 and @xmath4 is the number of human instance candidates determined by the algorithm during the optimization ( details below ) .
the element of @xmath1 is zero if the corresponding region candidate does not belong to any person and a natural number otherwise .
we find the optimal @xmath1 by jointly optimizing over all potential people instances : @xmath5 here @xmath6 is the cost of assigning part regions to specific human instances .
@xmath7 is a term that encourages the selected region candidates to cover corresponding body part regions .
@xmath8 is a term that enforces the assembled body regions in each detected human instance to have correct sizes .
apart from these terms , we also introduce constraint @xmath9 to limit the intersection area between the selected regions , and @xmath10 to constrain the color histogram between specific region pairs .
we also use constraint @xmath11 to enforce the total body part area of each instance person to be within an upper bound .
all these terms are defined in detail below .
[ fig : ink](a ) illustrates the region assembly problem as graph labeling .
the nodes correspond to the regions or sub - regions of different body parts .
head nodes and head - torso nodes in fig .
[ fig : ink](a ) are also denoted as human instance nodes .
the head - torso nodes represent the head - torso region combinations .
the binary edges correspond to possible region - to - human instance assignments , and the hyper edges constrain the region coupling and assignment consistency .
the binary edges and nodes have weights .
we essentially need to find an optimal node - edge labeling to minimize the total weight .
the optimization is combinatorial .
it is hard to solve due to the large number of edges , loopy structure and high order constraints . instead of directly solving the hard problem
, we decompose it into three optimizations on three simpler graphs in two stages , as shown in fig .
[ fig : ink](b ) . while satisfying different region constraints on body part assembly .
( b ) we decompose the optimization into three optimizations in two stages .
see text for details . ]
the three optimizations share a common integer program format : @xmath12 here , the vector @xmath13 includes the edge variables and the vector @xmath14 includes the human instance node variables .
the dimension of @xmath13 equals the number of torso regions in stage one or number of arm or leg regions in stage two times the number of candidate head regions .
the dimension of @xmath14 equals the number of head regions .
@xmath15 is an auxiliary variable vector .
@xmath16 are constant coefficient vectors .
@xmath17 is a constant .
@xmath18 is an all - one vector .
@xmath19 is the assignment constraint and @xmath20 represents the region coupling constraints .
how @xmath21 are determined will become clear after sec .
[ sec : details ] .
the optimization finds the pairing of nodes in each augmented bipartite graph in fig .
[ fig : ink](b ) .
the nodes on one side are the regions of torsos , arms or legs .
the nodes on the other side are the human instance representations using head regions ( stage one ) or a head - torso region combination ( stage two ) .
each part region ( arm , leg , torso ) node can be used at most once , and each human instance node may receive zero or multiple region matches .
when selecting multiple part nodes , we assemble corresponding body part regions using `` broken '' region pieces .
the nodes of part regions are coupled by the size and exclusion constraints .
now we flesh out how we instantiate the general formulation presented above .
we start from a semantic segmentation map in which each pixel is classified as one of the four part types ( arms , legs , torso and head ) or the background .
( background " = not any person " ; all person pixels are foreground " . )
the map is obtained by first computing a stack of probability maps from a cnn ( a modified alexnet @xcite ) for each part at different scales .
max - pooling is then applied to compute the body part soft semantic map .
we use graph cuts with alpha - expansion to generate the final semantic segmentation map .
please see appendix [ app : cnn ] for details .
overall , the goal is to have a large pool of part candidates with high recall , but possibly low precision ; that way , there is a high chance that we can correctly use them to assemble and separate multiple human human instances . with this in mind ,
regions and sub - regions of body parts ( torso , arm , and legs ) are generated as follows .
apart from using the connected components of body part regions from the cnn - derived semantic segmentation map , we use region proposals from @xcite to `` chop '' possibly merged part regions into smaller pieces by intersecting the region proposals with each part region .
the regions therefore include both whole body parts and fragments of body parts .
head regions are generated differently because the above method may not always be able to separate close head regions .
the head regions are circular regions whose radius is determined by the max - response scale at each head point ; the head points are detected by finding peaks in the soft semantic head map using non - maximum suppression .
( more details in appendix [ app : cnn ] . )
while our framework allows multiple head candidates with different scales at the same point , in practice we find selecting the single most likely head candidate at each point is sufficient , and so we take this simpler configuration .
the head candidate regions are further intersected with the person foreground in the semantic map .
the head detections automatically tell our system the candidate people number in the image .
now we are ready to define the energy terms @xmath6 , @xmath7 , @xmath8 and constraints @xmath9 , @xmath10 , @xmath11 in eq .
[ eq : overall ] and derive eq .
[ eq : cm ] .
we introduce a binary variable @xmath22 , the binarized version of @xmath1 in eq .
[ eq : overall ] , to label edges in fig .
[ fig : ink](b ) : @xmath23 if region @xmath2 is selected to be part of person @xmath24 , otherwise @xmath25 .
we have the following constraint on @xmath13 : @xmath26 , which means each region can only be assigned to at most one human instance .
each person instance may connect to multiple regions to handle the region splitting case .
we also introduce variable @xmath27 to indicate whether person / head candidate @xmath24 is selected .
@xmath14 is the human instance node variable .
we enforce @xmath28 . in stage two ,
the head - torso regions come from the solution of stage one , and @xmath14 is all one .
there is a cost @xmath29 to associate region @xmath2 to person instance candidate @xmath24 , and a cost @xmath30 to select instance candidate @xmath24 .
the total assignment cost is @xmath31 .
we optimize @xmath14 only in stage one . in stage two , @xmath14 is fixed to be all ones and can be removed from the optimization .
@xmath30 equals one minus the head region s peak probability on the head map , so as to emphasize costs incurred on more confident heads .
@xmath32 is a constant weight balancing the region association cost against instance selection cost . in short , the cost @xmath29 aims to associate a person instance with regions that `` look like '' part of the corresponding body parts , mostly fall in the range radius , and are close to the human instance .
we also encode the preference to use a small set of large regions to build the body parts of each human instance , to avoid trivial assemblies ( e.g. , @xcite ) .
specifically , the assignment cost @xmath33 is defined as : @xmath34 where @xmath35 equals one minus the mean probability of region @xmath2 belonging to human instance @xmath24 as part of a specific body part ( probability obtained from the soft cnn semantic maps at @xmath24 s scale ) .
the term @xmath36 is the proportion of region @xmath2 exceeding the range radius of instance @xmath24 . to maintain scale - invariance ,
the range radius for each body part is estimated using the instance / head @xmath24 s scale and a reference person height ( 150 pixels ) .
the term @xmath37 is the shortest point - to - point distance between the head region ( stage one ) or head / torso region ( stage two ) @xmath24 and part region @xmath2 , normalized based on the reference height .
the constant @xmath38 penalizes selecting many small regions to construct a body part in each human instance .
the coefficients @xmath39 , @xmath40 , @xmath41 and @xmath38 control the weights among different terms .
the setting of these coefficients are discussed later .
when composing a human instance s body part , the total area of the selected regions is limited by the body part s size : @xmath42 , where @xmath43 is the area of the region @xmath2 , @xmath44 is the scale of the human instance @xmath24 and @xmath45 is largest possible area of a body part for the reference person ( 150-pixel tall ) . in stage
one @xmath45 is the max area of the torso , and in stage two @xmath45 limits the area of arms or legs .
apart from the hard constraint , a soft one encourages the total area of a region assembly to approach a target size of the corresponding body part . we minimize @xmath46 , which can be converted to a linear form : @xmath47 . here
@xmath48 is the the average body part size of the reference person from different view points .
it corresponds to the torso in stage one and arms or legs in stage two .
we also prefer to select regions that are mostly non - overlapping to form each body part region .
thus , we introduce an exclusion constraint @xmath9 to discourage overlap .
let @xmath49 indicate whether region @xmath2 is associated to a human instance . to construct constraint @xmath9 , we let @xmath50 , where @xmath51 is the area intersection to union ratio between region @xmath52 and @xmath53 , @xmath54 is a constant .
apart from intersection exclusion , we also prefer that the color histograms should match if two regions are selected to form the same body part .
we thus enforce the constraint @xmath10 that @xmath55 , where @xmath56 is the @xmath57-distance between the normalized color histogram of region @xmath58 and @xmath59 , @xmath60 is a constant threshold .
if we simply minimize the above terms , @xmath61 will be all zero since all the coefficients in the objective are non - negative .
we introduce an extra covering term to encourage the chosen regions to cover the corresponding body part regions in the semantic segmentation map .
we maximize the total region size @xmath62 , where @xmath63 and @xmath64 is the part type of candidate @xmath2 , @xmath65 is the total area of part @xmath64 in the semantic map .
@xmath7 is proportional to the total region size .
this encourages region covering because we enforce the regions to be mostly disjoint .
combining the above terms , we have our final optimization objective : @xmath66 where @xmath17 and @xmath67 are coefficients that serve to control the weights of size term and cover term .
it is easy to verify that the formulation indeed has the format in eq .
[ eq : cm ] , if we vectorize variables @xmath68 and substitute @xmath69 by @xmath13 terms .
the optimization is a hard combinatorial problem due to the loopy structure and high order constraints .
we next propose an efficient relaxation and branch and bound method to solve the problem .
a direct linear relaxation , in which we replace the binary constraints on @xmath13 with a 0 to 1 soft constraint , gives a good lower bound .
the disadvantage of this approach is that we need a linear programming solver , and as the number of region candidates increases the complexity of solving a large linear program is high . with 1000 candidates and 2 human instances ,
the simplex method takes around 4 seconds to complete , while using the following speedup the time can be reduced to 0.1 seconds using the same cpu .
we obtain the lower bound using the lagrangian dual .
the size constraints and the exclusion constraints complicate the problem .
we move them into the objective function .
to simplify notation we use the compact format of eq .
[ eq : cm ] : @xmath70 where @xmath71 is the lagrangian multiplier vector .
we introduce an upper bound @xmath72 for @xmath15 to avoid unbounded solutions .
since the extra term in the objective is non - positive for all the feasible solutions of the original problem , the lagrangian dual gives a lower bound .
the internal part of the dual is easy to solve because it can be decomposed into three simple problems ( no p2 in stage two ) : @xmath73 : } \min_{x } ( g^t + \nu^tb)x , \mbox { s.t . } \ ; ax \le 1 , \ ; x \mbox { is binary}. \\ &
\mbox{[p2 ] : } \min_{y } ( w^t + \nu^td)y , \mbox { s.t . } \ ;
y \mbox { is binary}. \\ &
\mbox{[p3 ] : } \min_e ( \phi 1^t + \nu^tc)e , \mbox { s.t . }
0 \le e \le m.\end{aligned}\ ] ] p1 can be solved by sequential assignment : in an assignment graph such as fig .
[ fig : ink](b ) , for each body part region node , we check all the links to the human instance node and find the most negative link and let the corresponding @xmath13 variable to be 1 . if there is no negative link , no matching is made and the corresponding @xmath13 is 0 . in p2 and p3 , @xmath14 is set to 0 or 1 and @xmath15 is set to 0 or @xmath72 according to the positiveness of their coefficient .
each set of lagrangian multipliers corresponds to a lower bound of the original problem .
we are interested in the largest lower bound .
the bound with respect to the multipliers is a concave function and can be solved using the subgradient method
. the iteration alternates between solving for @xmath74 and updating @xmath71 by @xmath75 . here
@xmath76 is a small constant @xmath77 .
the initial values of these coefficients in @xmath71 are set to zero . for this problem ,
the lagrangian relaxation bound is the same as that of the linear program relaxation .
this is due to total unimodularity of the internal problem of the lagrangian dual @xcite .
* example . * fig .
[ fig : example ] shows an example of using the lagrangian relaxation to obtain the lower bound .
the max - pooling result of the initial cnn semantic body part segmentation is shown in fig .
[ fig : example](b - f ) .
we then use graph cuts with alpha - expansion to generate the hard semantic segmentation map ( middle of fig .
[ fig : example](a ) ) .
note how parts are not individuated per person in this map .
arm , leg and torso region proposals are generated from the intersection between semantic part maps and object proposals .
the lagrangian relaxation is applied to three optimizations in two stages . as shown in fig .
[ fig : example](j - l ) , the result converges quickly to the linear program relaxation result ( the red line ) in a few hundred iterations .
the relaxation assignment is illustrated in fig .
[ fig : example](g - i ) .
we see it is indeed very similar to the globally optimal solution in fig .
[ fig : example](a ) .
the complexity of finding the lower bound using the lagrangian relaxation is @xmath78 , where @xmath53 is the number of region proposals times the number of human instance candidates , and we use a fixed number of iterations in the subgradient method .
in contrast , the average complexity of a linear relaxation @xcite using the simplex method is @xmath79 .
the above dual approach can be extended to estimate the lower bound at each node of the search tree . with the lower bounds , we use the branch and bound method to find the global optimum quickly . for most problems in the experiments , where @xmath53 averages around @xmath80 , the branch and bound procedure terminates in a few seconds .
we set the thresholds @xmath81 and @xmath82 .
the weights of the unary cost terms are set to be @xmath83 and the weights for the energy terms as @xmath84 .
we fixed all parameters for all experiments after manually inspecting a few examples . with more labeled data
, we can optimize these parameters for even better performance .
( please see appendix [ app : para ] for details of how we can automatically set the parameters by maximizing the margin on positive / negative examples via a linear program . )
* overview : * in the following , we compare our approach to 1 ) simpler inference methods , to show the value added over the initial cnn body part maps ; 2 ) bounding box detector methods ; 3 ) cnn methods using region proposals ; 4 ) human pose detection based methods . having established our method s accuracy , we then demonstrate its applicability for a downstream task : proxemics recognition .
* datasets and evaluation metrics : * we evaluate the proposed method on 3 datasets : uci @xcite , which contains 589 images , 100 images from the mpii dataset @xcite that contain multiple tangled people , and buffy @xcite .
the images include complex human poses , interactions , and occlusions among subjects .
the person scales and orientations are unknown .
we manually label the human instances and four part regions in uci and mpii dataset for ground truth evaluation only ( not to train the cnn ) .
we use the area intersection to union ( iou ) ratio against the ground truth labeling to quantify the performance .
we report the iou for the human instances and mean iou over all body part labels within each instance .
to compute forward ( f ) scores , we match each ground truth segment to the best segmentation result . for the backward
( b ) scores , the matching is the other way around .
the forward score is affected by missing detections and the backward score by the false alarms .
the plots in fig .
[ fig : curve ] sweep through overlap thresholds and plot the proportion of results that have instance iou ratios higher than each threshold .
the higher the curve , the better the performance .
table [ tab1 ] shows average iou scores .
we explain the methods compared below .
[ fig : samples ] shows sample results of our method on uci and mpii . *
are our initial cnn body part maps enough ?
would a simpler inference method on top of the cnn maps be sufficient ? *
first , we stress that the cnn body part maps are not enough by definition , as they do not individuate which body part blobs go to which person .
the person and part segmentations merge when people are close .
for example , if their arms touch , that yields one connected component in the cnn output ; see fig [ fig : samples ] , second column in each set .
our cnn semantic segmentation itself is reasonable . on uci and mpii , the average foreground pixel accuracy and part pixel accuracy are @xmath85 and @xmath86 respectively .
however , this does not easily transfer to a good human instance segmentation . to confirm this quantitatively , we test 1 ) a baseline that returns connected components in the cnn map for the body part labels ( * connected * ) , and 2 ) a baseline that greedily finds the grouping of each person sequentially ( * greedy * ) . for the latter ,
after the lowest cost group is found , the regions in that group are removed and we proceed to the next one until all the head regions are exhausted . note that naive exhaustive search is extremely slow due to the huge search space .
[ fig : curve ] ( a , e ) and table [ tab1 ] show the results . our full method s strong results relative to both these baselines reveals the role of our region assembly optimization .
our efficient global optimization is necessary . * comparison with bounding box detector methods : * one widely used method ( e.g. , @xcite ) to extract human instances is to first detect people in a set of bounding boxes , and then obtain pixel - level segmentation . to test such a baseline
, we use a deformable part model ( * dpm * ) person detector @xcite and poselet ( * poselet * ) person detector @xcite , and refine the segmentation with grabcut @xcite .
we adjust the threshold of the person detectors to the lower side so that they can detect more people instances .
we also adjust the parameters of grabcut to achieve the best performance . as shown in fig .
[ fig : pff_poselet ] ( a ) , when the people have complex poses , interactions , and occlusions , the bounding boxes from person detectors are not accurate .
it is a non - trivial task for a pixel - level segmentation method to correct such errors without manual interaction .
indeed , our method gives consistently superior results to the detector based approach ( see fig .
[ fig : curve](b , f ) and in table [ tab1 ] ) .
* comparison with cnn object detectors using region proposals : * another method for human instance segmentation is first generating many region proposals and then using a classifier to extract true human instances , e.g. @xcite .
rcnn @xcite can also be modified to achieve such a function .
to compare this idea to our method , we test three kinds of region generation methods : selective search @xcite , mcg @xcite and object - independent proposals @xcite .
each rectangle image patch that encloses a region proposal is then sent to rcnn to determine the probability of the image patch containing a human instance .
for fair comparison , apart from the original dataset for training , we also include the lsp images @xcite in the refinement , which improves the baseline s human classification result .
the rcnn detection threshold is set to 0.1 .
[ fig : pff_poselet ] ( b ) shows sample results . in images with tangled people instances ,
region proposals often have a hard time to obtain full human segmentation , because the human structures are not directly used in these region proposal methods .
[ fig : curve](c , g ) and table [ tab1 ] show the quantitative comparison .
our method gives better results .
c c instance mean iou : & .
average person instance and part iou ratio comparison for the uci and mpii dataset . connected :
connected component method .
r - i : rcnn+oip , r - ii : rcnn+mcg , r - iii : rcnn+selectivesearch , cnn - d : cnn pose detector @xcite .
f : forward score .
b : backward score . in part iou table : connected component is denoted as c , greedy method as g , nbest as nb and cnn - d as cd .
the numbers are percentages . [
cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] * application for proxemics recognition : * finally , we demonstrate the utility of our human region parsing for _ proxemics _ recognition .
proxemics is the study of the spatial separation individuals naturally maintain in social situations .
the uci dataset was created to study proxemics , and is labeled for 6 classes : hand - hand ( hh ) , hand - shoulder ( hs ) , shoulder - shoulder ( ss ) , hand - torso ( ht ) , hand - elbow ( he ) and elbow - shoulder ( es ) .
we use features that include the min and max distances between each pair of upper body part regions of a person pair normalized by the average scale of the two subjects , the normalized horizontal and vertical distance of heads and the scale difference .
the data for training and testing are uniformly split at random , following the setup in @xcite . to learn the 6 proxemics classes on top of these features
, we use a random forest classifier with 100 trees and unlimited tree depth .
we repeat the experiment 10 times and report the average accuracy
. we do not use ground truth head locations .
[ fig : prox ] shows sample classifications and ap scores . our average ap score is higher than all the competing methods @xcite .
our weakness vs. @xcite on ht is likely because not only baby hugging but also other hand - on - torso images are classified as ht .
compared to the prior pose detectors , our method is more resistant to large occlusion , non - pedestrian poses , and complex interaction among people .
we identify the membership of each pixel to each human instance ; local errors , e.g. broken regions , do not cause a big problem .
we propose a novel method to segment human instances and label their body parts using region assembly .
the proposed method is able to handle complex human interactions , occlusion , difficult poses , and is rotation and scale invariant .
our branch and bound method is fast and gives reliable results .
our method s results compare favorably to a wide array of alternative methods .
this research is supported in part by u.s .
nsf 1018641 and a gift from nvidia ( hj ) and onr pecase n00014 - 15 - 1 - 2291 ( kg ) .
we train the semantic segmentation body part cnn on the lsp dataset @xcite with 27 classes ( including body joints , in - between points and background class ) using caffe @xcite . during training ,
the cnn s input size is 55@xmath8755 and the first convolution layer has stride 1 instead of 4 .
the fully connected layers before the output are replaced by simpler lenet fully connected layers and then converted to fully convolutional networks to facilitate pixel - level semantic segmentation .
the training image patches are rotated and flipped so that the trained classifier is rotation invariant . when testing , all the images are rescaled so that the longest edge is 500-pixels ; the cnn input size is the same as the input image .
the 27 classes are finally merged to 4 part types head , torso , arm and leg and a background class when obtaining the soft semantic segmentation . for each input image
, we compute the cnn soft semantic maps of the five classes ( arm , leg , torso , leg and background ) in different scales . for each scale level , the input image is downsized to a specific resolution .
the downsizing factors range from @xmath88 to @xmath89 with the step of @xmath90 .
there are totally 16 different scale levels .
the soft semantic map of each class is then generated using max - pooling .
the soft semantic maps of the five classes are finally normalized so that each value represents a probability .
the head points are localized on the merged soft head map by non - max suppression .
the non - maximum suppression uses the window of @xmath91 and the threshold of 0.2 . after each head point
is localized , we find the head s scale by going through the stack of head maps with different scales and finding the map that gives the largest value at the head point .
we record the head scale as the inverse of the corresponding image downsizing factor .
for instance , if at one head point , the semantic map with the downsizing factor of 0.5 gives the largest response in the cnn head maps , the head candidate has a scale 2 .
we compute the final semantic segmentation map using the graph cuts method , which minimizes the assignment of five labels : head , torso , arm , leg and background to each pixel in the input image .
two passes of graph cuts are used . in the first pass ,
we label each image pixel as human foreground ( body part pixel ) or background ( non - human pixel ) .
the unary term is one minus the pixel level probability of these two classes from the cnn .
the weights are 1 and 0.2 for the unary term and binary term respectively .
the binary term smooths the labeling between neighboring pixels .
each neighbor pixel pair contributes to the binary term ; the contribution equals the weight times the distance between the two neighbor pixels labels .
we define the distance between two labels to be zero if they are the same and otherwise the distance is one . in the second pass ,
we only label the foreground pixels determined from the first pass as the four body part classes : head , arm , leg or torso . we also use one minus the pixel probability from the cnn in the unary term and the same setting in the pairwise term as pass one .
the weights are also 1 and 0.2 for the unary and binary terms . to improve the result
, we include a term that penalizes pixels to be labeled as arm or torso if they fall out of the arm range radius of each detected head in the image .
the arm range radius is computed using the head s scale and a 150-pixel tall reference person s maximum head to hand distance .
the penalty is applied in the alpha - expansion procedure . for a pixel out of the range radiuses of all the detected heads ,
a penalty of 1 is included in the unary term of the alpha node if the alpha label is torso or arm , or a penalty of 1 is included in the unary term of the pixel node if the current pixel label is torso or arm .
the first pass graph cuts can be optimized using the maxflow method . in the second pass ,
20 iterations are used in the alpha - expansion .
we use a push - relabel maxflow algorithm in the implementation .
the maxflow algorithm requires integer edge capacities ; we convert the floating point capacities in the above formulation to integers by multiplying 1000 and then taking the floor . with enough labeled data
, we can automatically set the parameters in the the proposed method by maximizing the margin on positive / negative examples via a linear program .
@xmath60 can be obtained from the color statistics of the ground truth data .
we assume @xmath54 fixed for now .
we represent all the other weight parameters as a vector @xmath59 .
let @xmath92 be the values of different terms weighted by these coefficients for the positive segmentation sample of image @xmath2 , and @xmath93 be the corresponding values of the @xmath24th negative segmentation sample of image @xmath2 .
positive segmentation exemplars are from ground truth labeling .
negative exemplars can be randomly generated using region proposals , which satisfy the overlap , color exclusion and unique assignment constraints .
we maximize the energy difference between the positive and negative training examples using a similar formulation to @xcite : @xmath94 the optimization can be solved efficiently using linear programming .
@xmath54 has a small range and can be found by exhaustive search to achieve the highest segmentation performance on the training dataset . | today s person detection methods work best when people are in common upright poses and appear reasonably well spaced out in the image . however , in many real images , that s not what people do .
people often appear quite close to each other , e.g. , with limbs linked or heads touching , and their poses are often not pedestrian - like . we propose an approach to detangle people in multi - person images .
we formulate the task as a region assembly problem . starting from a large set of overlapping regions from body part semantic segmentation and generic object proposals , our optimization approach reassembles those pieces together into multiple person instances .
it enforces that the composed body part regions of each person instance obey constraints on relative sizes , mutual spatial relationships , foreground coverage , and exclusive label assignments when overlapping . since optimal region assembly
is a challenging combinatorial problem , we present a lagrangian relaxation method to accelerate the lower bound estimation , thereby enabling a fast branch and bound solution for the global optimum . as output ,
our method produces a pixel - level map indicating both 1 ) the body part labels ( arm , leg , torso , and head ) , and 2 ) which parts belong to which individual person .
our results on three challenging datasets show our method is robust to clutter , occlusion , and complex poses .
it outperforms a variety of competing methods , including existing detector crf methods and region cnn approaches .
in addition , we demonstrate its impact on a proxemics recognition task , which demands a precise representation of whose body part is where " in crowded images .
= 1 |
the recent detection of blue - shifted fe xxv / xxvi absorption lines in the x - ray spectra of several seyferts and quasars suggests the presence of highly ionized and mildly - relativistic outflows in the center of these agns ( e.g. , chartas et al .
2002 , 2003 ; pounds et al . 2003 ; dadina et al . 2005 ; markowitz et al .
2006 ; braito et al .
2007 ; cappi et al . 2009
; reeves et al . 2009 ; giustini et al .
they are possibly directly connected with accretion disk winds / outflows and the high associated outflow rate and mechanical power suggest they might be able to provide an important contribution to the expected agn cosmological feedback ( e.g. , king 2010 ; tombesi et al . 2012 ) . here
we describe the systematic analysis and characterization of these so called ultra - fast outflows ( ufos ) on a large sample of both radio - quiet and radio - loud agns .
these results are discussed in detail in tombesi et al .
( 2010a , b ; 2011a , b ) . 143 , best - fit model and background spectrum .
_ right panel : _ example of @xmath7 distribution from 1000 monte carlo simulations.,title="fig:",width=207,height=188 ] 143 , best - fit model and background spectrum .
_ right panel : _ example of @xmath7 distribution from 1000 monte carlo simulations.,title="fig:",width=207,height=151 ]
the radio - quiet sample was defined in tombesi et al .
( 2010a ) selecting all the narrow - line seyfert 1 , seyfert 1 and seyfert 2 ( with neutral absorption column density @xmath8 @xmath6 ) in the rxte all - sky slew survey catalog ( revnivtsev et al .
2004 ) and then cross - correlated with the _ xmm - newton _ catalog .
we obtained 42 sources for a total of 101 pointed _ xmm - newton _ observations .
the sources are local ( @[email protected] ) and x - ray bright .
we applied standard screening procedures and extracted the 410 kev epic pn spectra .
we carried out a uniform spectral analysis using a phenomenological baseline model composed of an absorbed power - law continuum and gaussian fe k emission lines ( see left panel of fig . 1 ) .
then , we performed a blind rearch for emission / absorption lines adding an additional narrow line to the baseline models with free positive / negative intensity and stepped the line energy between 410 kev , recording each time the associated @xmath7 deviations .
the resultant energy - intensity contour plots were inspected to select possible features .
initially , only the narrow lines ( @xmath11@xmath10100 ev ) with @xmath1299% f - test detection probability were selected and their velocity shifts were estimated using an identification as fe xxv / xxvi 1s2p/1s3p transitions . as discussed in 4 , we subsequently performed additional tests on the significance of the absorption lines at [email protected] kev using extensive monte carlo simulations and selected only those with probability @xmath1395% ( see right panel of fig . 1 ) .
thus , in tombesi et al . ( 2010a ) we detected a total of 36 absorption lines , 14 in the range [email protected] kev and 22 at [email protected] kev .
in a subsequent paper , tombesi et al .
( 2011a ) , we then performed a detailed curve of growth analysis of the fe xxv / xxvi absorption lines and a photo - ionization modeling using the _ xstar _ code ( see left panel of fig . 1 ) .
we considered an ionizing continuum equivalent to the average sed of the seyfert 1s in the sample , i.e. a @xmath15@xmath162 power - law with cut - off at e@xmath16100 kev .
we then calculated _ xstar _ grids assuming standard solar abundances and turbulent velocities of 1000 , 3000 and 5000 km / s .
given the limited energy resolution of the epic pn , only four absorption lines were resolved with width @xmath11@xmath165,000 km / s and for the other we could place only upper limits . a blind search for the best - fit _
solution(s ) was performed stepping the absorber redshift between 0.1 and @xmath170.4 , leaving the column density and ionization free , and checking for minima in the resultant @xmath18 distribution .
this allows to self - consistently take into account the possible presence of lines / edges from ions of different elements . in all cases a good fit ( p@xmath19@xmath099% )
was reached with a single _ xstar _ component and all the absorption lines were indeed consistent with blue - shifted fe xxv / xxvi transitions .
the possible degeneracy of the identification in a few cases was included in the relative larger parameter errors .
these studies allowed us to derive the distribituion of the main parameters of the ufos in the radio - quiet sample ( tombesi et al .
the detection fraction is @xmath040% , which may suggest a large covering factor of @xmath00.4 .
the lines have been found to be variable in both ew and velocity shift on time - scales even as short as @xmath1days , indicating compact absorbers .
the outflow velocities are mildly - relativistic , in the range @xmath10.030.3c , with a mean value of @xmath10.14c .
their ionization is high , in the interval log@[email protected] erg s@xmath3 cm , and the associated column densities are also large , in the range @xmath20@xmath1@xmath4@xmath5 @xmath6 . the estimated location and energetics of the ufos in the radio - quiet sample
is reported in a subsequent paper , tombesi et al .
we extented the search for ufos also in radio - loud agns and in particular to blrgs , which are the radio - loud counterparts of seyfert 1s .
they show strong relativistic radio jets , but the typical inclination of @xmath21@xmath120@xmath2240@xmath22 allows the direct observation of the inner disk in x - rays . therefore , in tombesi et al .
( 2010b ) we performed a systematic 410 kev spectral analysis of the long _ suzaku _ observations of 3c 111 , 3c 390.3 , 3c 120 , 3c 382 and 3c 445 .
we applied the same method as for the radio - quiet sample ( see 2 ) and detected blue - shifted absorption lines at e@xmath127 kev in 3/5 sources .
their significance was assessed using both f - test and extensive monte carlo simulations ( see 4 ) . through a photo - ionization modeling with _ xstar
_ we find that they are consistent with series of fe xxv / xxvi k - stell transitions blue - shifted with mildly - relativistic velocities in the range @xmath10.040.15c .
they are highly ionized , log@xmath2@xmath1646 erg s@xmath3 cm , and the associated column densities are @xmath20@xmath0@xmath4 @xmath6
. their estimated location within @xmath10.010.1 pc from the central super - massive black hole suggests a likely origin related with accretion disk outflows .
the mass outflow rate of these ufos is comparable to the accretion rate of @xmath11 @xmath23 yr@xmath3 .
their estimated kinetic power is high , in the range @xmath1@xmath24@xmath25 erg s@xmath3 , which is comparable to their typical jet power and corresponds to a significant fraction of the bolometric luminosity .
therefore , these ufos can potentially play a significant role in the expected agn feedback and can also possibly be directly linked to the jet activity .
the galaxy 3c 111 was then the subject of a follow - up study of the variability of its ufo with _ suzaku _ in september 2010 ( tombesi et al .
we obtained three pointings of @xmath160 ks spaced by @xmath17 days and observed a @xmath130% flux increase between the first and the second . a galactic absorbed power - law continuum plus a narrow fe k@xmath26 emission line at [email protected] kev
provide a good representation of the 410 kev xis spectra .
however , an additional emission line at [email protected] kev is observed in the first observation and an absorption line at [email protected] kev in the second .
the detection significance of these features is high , @xmath1299% from both f - test and monte carlo simulations .
they are also significantly variable among the observations .
the emission feature can be modeled with a highly ionized line coming from reflection off the accretion disk at @xmath120100@xmath27 from the black hole .
instead , a photo - ionization modeling of the absorption line suggests its association with a highly ionized , log@[email protected] erg s@xmath3 cm , ufo with outflow velocity @xmath10.1c and column density @xmath20@xmath16@xmath28 @xmath6 . the location of the material is constrained at @xmath29@xmath30 pc from variability .
this observation may provide the first direct evidence for an accretion disk - wind connection in an agn and is consistent with a picture in which an increased illumination of the inner disk causes an outflow to be lifted at @xmath1100 @xmath27 .
this is then possibly accelerated through radiation pressure to the observed terminal velocity of @xmath10.1c , but additional magnetic thrust can not be excluded ( e.g. , ramrez & tombesi 2012 ) .
the significance of the absorption lines detected at [email protected] kev was further investigated through extensive monte carlo simulations ( tombesi et al .
2010a , b ) . for each case
we simulated 1000 spectra assuming the baseline model without the absorption lines and calculated the @xmath7 distribution of random generated features in the 7.110 kev band .
then , we selected only the observed absorption lines with measured @xmath7 corresponding to a monte carlo detection probability @xmath1395% .
we can also exclude any contamination due to the epic pn background and calibration uncertainties and we checked that the possible presence of spectral complexities , such as reflection and warm absorption , are typically weak enough to assure a only marginal model dependency of the results .
this systematic analysis on a large sample of sources allows to estimate the global detection probability of the absorption lines of @xmath125@xmath11 and to overcame the possible publication bias claimed by vaughan & uttley ( 2008 ) .
moreover , we checked that the results are consistent using also the mos detectors and the global random probability in that case is also low , @xmath29@xmath31 . in the left panel of fig .
2 we show the distribution of the significance @xmath11 ( simply estimated from the ratio @xmath29ew@xmath12/error@xmath29ew@xmath12 ) of the blue - shifted absorption lines at [email protected] kev detected by tombesi et al .
( 2010a ) with respect to the 410 kev counts of the relative observations ( filled circles ) .
it has been claimed that the tendency of the distribution to lay close to the 3@xmath11 level and not showing a strong enhancement in the significance following an increase in counts might suggest that all the detections are fake . however , plotting also the same distribution for the absorption lines at [email protected] kev ( open triangles ) and the ionized fe k emission lines at [email protected] kev ( crosses ) , we note that they also follow this general trend and it is not peculiar of the [email protected] kev lines . in particular , we state that this does not necessarily mean that the lines are fake , given that their significance was already estimated using the f - test and extensive monte carlo simulation , but that they are intrinsically weak and variable .
we support this statement with a simple test .
we simulated 15 absorption lines assuming the same parameters as the real data , i.e. 410 kev counts in the range 3300@xmath32 and variable ew@xmath1420 , 40 , 60 ev and e@xmath146 .
8 , 9 kev .
as expected , the same trend is followed also by the simulations ( open circles ) . instead , in the right panel of fig .
2 we show the distribution of the significance of the lines detected simultaneously in the epic pn and mos . in this case the open circles refer to the narrow 6.4 kev fe k@xmath26 emission lines . we can see that the points are systematically indicating that the detection significance in the pn is higher than in the mos .
this trend is simply expected from the higher effective area and lower background of the epic pn with respect to the mos .
important improvements in the detection and characterization of these lines are expected from the higher effective area and supreme energy resolution in the fe k band offered by the micro - calorimeters on board _ astro - h _ and the proposed esa _ athena _ missions .
braito , v. , et al .
2007 , apj , 670 , 978 cappi , m. , et al .
2009 , a&a , 504 , 401 chartas , g. , brandt , w. n. , gallagher , s. c. , & garmire , g. p. 2002 , apj , 579 , 169 chartas , g. , et al .
2003 , apj , 595 , 85 dadina , m. , cappi , m. , malaguti , g. , ponti , g. , & de rosa , a. 2005 , a&a , 442 , 461 giustini , m. , cappi , m. , chartas , g. , et al .
2011 , a&a , 536 , a49 king , a. r. 2010 , mnras , 408 , l95 markowitz , a. , et al . 2006 , apj , 646 , 783 pounds , k. a. , et al .
2003 , mnras , 345 , 705 ramrez , j. m. , tombesi , f. 2012 , mnras , 419 , l64 reeves , j. n. , et al .
2009 , apj , 701 , 493 revnivtsev , m. , sazonov , s. , jahoda , k. , & gilfanov , m. 2004 , a&a , 418 , 927 tombesi , f. , cappi , m. , reeves , j. n. , et al . 2010a , a&a , 521 , a57 tombesi , f. , sambruna , r. m. , reeves , j. n. , et al .
2010b , apj , 719 , 700 tombesi , f. , cappi , m. , reeves , j. n. , et al .
2011a , apj , 742 , 44 tombesi , f. , sambruna , r. m. , reeves , j. n. , et al . 2011b , mnras , 418 , l89 tombesi , f. , cappi , m. , reeves , j. n. , braito , v. , 2012 , mnras submitted vaughan , s. , & uttley , p. 2008 ,
mnras , 390 , 421 | x - ray evidence for ultra - fast outflows ( ufos ) has been recently reported in a number of local agns through the detection of blue - shifted fe xxv / xxvi absorption lines .
we present the results of a comprehensive spectral analysis of a large sample of 42 local seyferts and 5 broad - line radio galaxies ( blrgs ) observed with _
xmm - newton _ and _ suzaku_. we detect ufos in @xmath040% of the sources .
their outflow velocities are in the range @xmath10.030.3c , with a mean value of @xmath10.14c .
the ionization is high , in the range log@xmath2@xmath136 erg s@xmath3 cm , and also the associated column densities are large , in the interval @xmath1@xmath4@xmath5 @xmath6 .
overall , these results point to the presence of highly ionized and massive outflowing material in the innermost regions of agns .
their variability and location on sub - pc scales favor a direct association with accretion disk winds / outflows .
this also suggests that ufos may potentially play a significant role in the agn cosmological feedback besides jets and their study can provide important clues on the connection between accretion disks , winds and jets . |
theoretical modeling of atmospheres of extrasolar giant planets ( egp ) is a young field , about one decade old , and as youngsters usually are , it is very active , restless , and sometimes quite unpredictable .
it had recently undergone a transition from a purely care - free stage ( no observations were available ) to a more difficult stage where there already are some observations to be fit by the theory . in this paper
, we briefly describe our recent efforts in this area .
we will try to convince the reader that despite its young age and many shortcomings , the theory is actually doing quite well .
the basic aim of a model atmosphere is to predict both the structure ( physical and chemical state ) of an atmosphere as a function of depth , as well as the theoretical emergent spectrum .
the latter is the most important result of a modeling exercise because by comparing the model predictions to actual observations one might constrain the physical conditions in planet atmospheres , and indirectly also of the whole planet .
the general problem of atmospheres of extrasolar giant planets is an extremely complex one that should generally involve a 3-dimensional radiation hydrodynamical treatment of atmospheric motions and day / night side energy redistribution , with departures from chemical equilibrium , and a self - consistent description of clouds ( particle size distribution , cloud position in the atmosphere ) , together with a self - consistent treatment of the radiation field .
such a challenge is still far beyond current computational capabilities , not withstanding the lack of , or inadequacies in , many ingredients needed for the modeling effort , such as complete and reliable atomic and molecular data and optical constants for cloud particles , to name just a few .
obviously , one has to resort to approximations .
it is fair to say that the work of different groups differs mostly by what actual approximation are adopted , and what simplifications are being made . in the absence of exact calculations
it is difficult to decide which approximations are the most restrictive .
therefore , the current and future development of this field will proceed by gradually relaxing more and more approximations , and the observational tests , whenever available , become the judge of whether one is on track .
the current standard model atmospheres of extrasolar planets are based on the following assumptions : ( i ) the atmosphere is assumed to be a plane - parallel , horizontally homogeneous layer ( which reduces the problem to 1-d ) ; ( ii ) the atmosphere is in hydrostatic equilibrium ; ( iii ) the atmosphere is in radiative equilibrium ( or radiative+convective equilibrium in convectively unstable layers ) ; ( iv ) local thermodynamic equilibrium ( lte ) holds ( therefore the opacity and emissivity are known functions of only temperature and density , and can , therefore , be pre - tabulated and not computed on the fly during model construction ) ; and ( v ) the atmosphere is in a modified chemical equilibrium ( the local chemical equilibrium which takes into account possible rainout and consequent depletion of species ) .
most groups that produce egp model atmospheres adopt the above assumptions , and use different modeling codes ( e.g. , @xcite , @xcite ) .
our group has originally used a code described in @xcite , with an additional algorithm to treat radiation scattering in clouds @xcite .
we later switched to a more powerful procedure based on adapting the general - purpose stellar atmosphere code tlusty ( @xcite ; @xcite ) , originally designed for non - lte model atmospheres of hot stars , and extended to other stellar types and accretion disks . the specific variant for the atmospheres of egp s and brown dwarfs is called cooltlusty , and has been described briefly in @xcite , @xcite , and @xcite .
the code uses the above mentioned five basic assumptions .
the chemical equilibrium and departures from it due to the rainout of the species is treated following @xcite , and the opacity tables are computed as described in @xcite .
soon after the @xcite grid of egp model atmospheres was completed , the very close - in egp , ogle - tr-56b , ( ( * ? ? ?
* konacki et al . ( 2003 ) ) ) was discovered , with a planet - star separation of a mere 0.0225 au . we set out to extend our grid to higher irradiations , and found that the computed atmospheric structure depends very sensitively on whether the tio / vo opacity is included in the opacity table or not
see @xcite .
the basic conclusions of that paper are the following : the models without tio / vo always exhibit a monotonic temperature / pressure ( @xmath0 ) profile , with temperature decreasing outward .
when the tio / vo opacity is included , the models remain essentially unchanged for larger planet - star separations , simply because tio and vo are rained out everywhere in the atmosphere , save in the deepest layers which have virtually no effect on emergent spectra .
for very close separations , the model with tio / vo exhibits a significant temperature increase toward the surface . in analogy with the solar - system planets , the region of temperature inversion may be called a stratosphere . " but the biggest surprise was that for a certain range of planetary distances , there are actually two legitimate solutions when tio / vo opacity is taken into account
one with a monotonic decrease of temperature toward surface and one exhibiting a stratosphere ! this behavior is illustrated in fig .
full lines represent models computed for the opacity table without tio / vo , while the dashed and dot - dashed lines represent models with tio / vo .
the middle model exhibits the true bifurcation the dashed and dot - dashed lines correspond to the two solutions for the atmospheric structure with exactly the same input parameters ( effective temperature , surface gravity , elemental composition , and irradiation ) .
the fact that the model without tio / vo is very close to the low - surface - temperature branch of the model with tio / vo clearly indicates that the model bifurcation , and the very existence of the outward temperature increase , arises due to the tio / vo opacity .
the physical explanation of this effect is straightforward
see @xcite .
tio / vo provide a strong opacity in the optical wavelengths , where the external irradiation ( in the case of solar - type stars ) has its maximum .
therefore , the incoming stellar radiation is efficiently absorbed in the upper atmosphere , thus leading to its heating .
the explanation of bifurcation is analogous .
an atmosphere either `` chooses '' a high surface temperature , in which case tio / vo is present , and the incoming stellar flux is efficiently absorbed ( so one indeed has , consistently , a high surface temperature ) . or , the atmosphere `` choses '' a low surface temperature , in which case tio / vo does not exist near the surface , and , therefore , there is no efficient mechanism to absorb incoming stellar radiation in the upper atmosphere to heat it .
the difference in the temperature structure is reflected in the emergent spectrum . in the right panel of fig .
[ fig1 ] , we show the emergent flux that coresponds to ogle - tr-56 , with a planet - star separation equal to 0.0225 au . the thick line is a model without tio / vo , and the thin line with tio / vo .
similar results were later obtained by ( * ? ? ?
* fortney et al .
( 2006 ) ) and ( * ? ? ?
* fortney et al .
( 2008 ) ) .
however , there is a potentially serious problem with this picture .
since the presence of tio and vo is quite natural at high-@xmath1/high-@xmath2 points , one confronts a situation in which there are two regions rich in tio / vo , separated by a middle region that is tio / vo free . in a gravitational field with a monotonic pressure profile , this gap could act like a cold trap in which the tio / vo that is transported by molecular or eddy diffusion from the upper low-@xmath2 region into the intermediate cooler region , would condense and settle out , thereby depleting the upper low-@xmath2 tio / vo - rich region .
this would eventually leave no tio / vo at altitude to provide the significant absorption that could lead to a bifurcation .
the situation is much more favorable for very close - in planets , where the temperature is so high that tio / vo is present everywhere in the atmosphere . in this case , one obtains a single solution , that with a stratosphere .
we note that the models presented in @xcite considered no day / night side heat redistribution .
consequently , these models generally tend to overestimate the atmospheric temperature .
we conclude that tio / vo reliably operate only in a limited temperature range ( @xmath3 k ) , because otherwise the cold trap effect may deplete them .
one can speculate , though , that diffusion , turbulent mixing , or a wind may actually replenish tio and vo , but no explicit calculations or even rough estimates have yet been done .
experience from the solar - system giant planets indicates that a number of unknown , optically - active compounds may exist in the upper atmosphere . with extreme irradiation , a complex non - equilibrium photochemistry will very likely take place .
consequently , there may be a number of as yet unknown opacity sources in the upper atmosphere that may build a stratosphere , even in the absence of tio / vo . in view of these uncertainties ,
one is free ( indeed , forced ) to parameterize the extra opacity ; specifically its magnitude ( @xmath4 ) , and its position in the physical and the wavelength space .
therefore , we have adopted such a philosophy .
our approach is described in detail in @xcite .
in early studies ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , the degree of the day / night side heat redistribution was described through an empirical parameter , originally denoted as @xmath5 . in @xcite
we introduced an analogous parameter , @xmath6 , as a fraction of incoming flux that is redistributed to the night side .
the underlying assumption was that the fraction @xmath6 of the incoming flux is somehow removed before the incoming radiation reached the upper boundary of the atmosphere , and deposited at the lower boundary of the night - side atmosphere .
although this treatment is useful for obtaining a rough estimate of the effects of day / night side heat redistribution , it is unsatisfactory from the physical point of view .
there are two main problems : ( i ) it is unphysical to assume that the incoming flux is removed at the top of the day - side atmosphere and injected in the bottom of the night - side atmosphere , and ( ii ) the calculated entropies in the convection zones at the day and the night side may be generally different . since convective transport is very efficient , the temperature gradient is essentially exactly adiabatic
. consequently , the whole convective core has to he isentropic , and , thus , can not exhibit any difference between the day and night sides . to overcome these problems ,
we have recently adopted an improved , albeit still parametric , approach
@xcite . out of the total incoming energy , the fraction @xmath7 is assumed to be radiated on the day side , and the fraction @xmath6 on the night side .
the irradiation energy is removed on the day side in a given depth range , parameterized by limiting pressures @xmath8 and @xmath9 .
this energy is deposited on the night side in depth range parameterized by limiting pressures @xmath10 and @xmath11 , which are generally different from @xmath8 and @xmath9 , but in actual calculations we usually take them to coincide with @xmath8 and @xmath9 . the modeling proceeds as follows : at the day side , the input parameters are @xmath12 ( @xmath13 being the surface gravity ) ; @xmath14 ( which corresponds to @xmath15 in the case of stellar atmospheres ) ; @xmath6 ; @xmath8 ; and @xmath9 .
other parameters are the distance and the radius of the parent star , and the elemental composition ( which is assumed to be the same on the night side ) .
the computed model gives the structure ( essentially , the @xmath0 profile and the self - consistent radiation intensity ) , plus the entropy in the convection zone . at the night side ,
the input parameters are : @xmath12 ( which must be the same as at the day side ) ; @xmath6 ; limiting pressures @xmath10 and @xmath11 ; and a trial @xmath16 .
the model again yields a @xmath0 profile , and the entropy at the convection zone , which is generally different from the entropy at the day side .
the intrinsic temperature @xmath16 is adjusted by trial and error so that both entropies ( day and night ) match .
we note that with strong irradiation , the convection zone may be very deep ( at @xmath17 of the order of @xmath18 to @xmath19 ) , and , therefore , the entropy matching is important for evolutionary models , and not so much for predicting emergent spectra .
for close - in egps the planet - star contrast ratios in the mid - infrared are relatively large , often exceeding @xmath20 , and such contrasts are well within reach of the infrared space telescope _
spitzer_. using its irac and mips cameras , and the irs spectrometer , one can now measure the summed light of the planet and the star in and out of the secondary eclipse , and , from the difference , determine the planet s spectrum at superior conjunction ( see several reviews in these proceedings ) .
this has led to a breakthrough in the study of extrasolar planets , for it provides a means to probe the physics , chemistry , and even meteorology of their atmospheres . before this conference , secondary - eclipse fluxes in the irac and mips
channels had been measured for five transiting planets
hd189733b , tres-1 , hd 209458b , hd 149026b , and gj 436b .
our group has performed a detailed analysis of four of them ( @xcite ) .
so far , we have not analyzed gj 436b . .
we also indicate the position where the monochromatic optical depths for the characteristic wavelengths of the four irac and the mips channel are equal to 2/3 .
( taken from burrows et al .
2007.),width=230 ] the results are displayed in the four panels of fig .
[ fig2 ] for the four transiting planets . for hd 209458b ,
we show three models with an upper - atmosphere extra absorber ( and with resulting stratospheres ) , and one traditional model without an extra absorber .
the values of the redistribution parameter @xmath6 for the models with an extra absorber are 0.3 , 0.4 , and 0.5 , and the limiting pressures for the energy sink are 0.01 and 0.1 bars .
the extra absorber , assumed to contribute between 0.4 and 1 @xmath21 m , is placed at altitude below pressures of 25 mbars .
the corresponding @xmath0 profiles for the four hd 209458b models depicted in fig .
[ fig2 ] , together with the formation depths , " defined as the position in the atmosphere where the monochromatic optical depth reaches 2/3 , are displayed in fig .
the plot explains why the flux in the irac 2 , 3 , 4 , and mips bands are higher for the model with an extra absorber ( stratosphere ) , while it is lower in the irac 1 band .
figure [ fig2 ] shows that for hd209458b the models with a stratosphere fit the irac 1 , 2 , and 4 points quite well .
however , the irac 3 point near 5.8 @xmath21 m is difficult to fit since the radiation in this band is formed at essentially the same position in the atmosphere as the irac 4 .
analogously , for hd149026b , the model with a stratosphere is clearly preferable .
the situation for hd 189733b is uncertain ; in any case a stratosphere may exist , but there is no compelling case for it . on the other hand ,
the tres-1 data are fitted perfectly with a traditional model without a stratosphere .
since the total irradiated flux is largest for hd 149026b and smallest for tres-1 see @xcite , the presence of the stratosphere seems to be very roughly correlated with the total stellar flux at the planet s surface . in @xcite , we have analyzed also the observed light curves of two non - transiting planets ( @xmath22 and b , and hd179949b ) , and found that for @xmath22 and b the presence of a stratosphere is strongly indicated in order to fit the observed data .
we have found ( @xcite ; @xcite ) that a consistent fit to all data at secondary eclipse for several strongly irradiated transiting planets ( hd 209458b , hd 149026b , and possibly hd 189733b ) , and very likely a non - transiting planet @xmath22 and b , requires that their atmospheres have temperature inversions
stratospheres at altitude .
such a thermal inversion affects : ( i ) planet / star contrast ratios at the secondary eclipse ; ( ii ) their wavelength dependences ; and ( iii ) day - night flux contrast during a planetary orbit
. moreover , the presence of the thermal inversion / stratosphere seems to roughly correlate with the total irradiated flux .
temperature inversion is caused either by tio / vo , as first suggested by @xcite , or by another , as yet unidentified , opacity sources .
these may be tholins , polyacetylenes , or various non - equilibrium compounds .
we invoke such extra absorbers , because a cold - trap effect can operate to deplete the upper atmosphere of tio / vo .
however , one may speculate that with ongoing mass loss and/or rotational shear instabilities the atmosphere may be partially replenished in tio / vo .
therefore , while tio / vo might be responsible for the formation of thermal inversions in the strongly irradiated planets , the exact nature of the absorber must be viewed as very uncertain . | we show that a consistent fit to observed secondary eclipse data for several strongly irradiated transiting planets demands a temperature inversion ( stratosphere ) at altitude .
such a thermal inversion significantly influences the planet / star contrast ratios at the secondary eclipse , their wavelength dependences , and , importantly , the day - night flux contrast during a planetary orbit .
the presence of the thermal inversion / stratosphere seems to roughly correlate with the stellar flux at the planet .
such temperature inversions might caused by an upper - atmosphere absorber whose exact nature is still uncertain . |
the _ random field ising model _ @xmath0 has been widely studied in recent years in the context of hysteresis [ 1 - 3 ] , return - point memory [ 4,5 ] and barkhausen noise [ 6 - 11 ] .
hysteresis is a common phenomenon , exhibited , for example , by most magnetic materials . in hysteresis
the magnetization lags behind the applied field and the phenomenon has received lots of attention in the past . moreover ,
if one observes carefully , the change in magnetization in systems like that in ferromagnetic alloys and amorphous ferromagnets , takes place in a series of irregular pulses as the external field is slowly varied . in experiments these pulses manifest themselves as acoustic emissions and is known as barkhausen noise .
interestingly , barkhausen signals show scale - free behavior similar to that of systems at criticality . barkhausen noise has been studied much in recent years in the context of driven disordered systems far from equilibrium example
of which include martensite transformation in shape - memory alloys [ 12 ] , eartquakes [ 13 ] , deformation of granular materials [ 14 ] and the breakdown of solids [ 15 ] .
various experiments [ 8 ] show that the distribution of size , duration and energy associated with barkhausen noise show power - law behavior .
sethna et al , in order to account for the hysteresis and barkhausen noise incorporated disorder in the form of random fields into the otherwise pure spin systems such as the ising model [ 3 ] . in the _ rfim
_ , the ising spins @xmath1 with nearest neighbor ferromagnetic interaction _ j _ are coupled to the on - site quenched random magnetic fields @xmath2 and the external field @xmath3 .
the hamiltonian of the system is @xmath4 where @xmath2 are quenched independent random variable drawn from a distribution of zero mean and variance @xmath5 .
in the study of hysteresis and barkhausen jumps the applied field is ramped up or down adiabatically ( so that the rate of spin flips is much larger than the rate of change of @xmath3 ) . in the absence of disorder ( @xmath6 ) , the model exhibits a sharp first order transition taking the magnetization @xmath7 from -1 to + 1 as @xmath8 passes through _ zj _ , _ z _ being the coordination number of the underlying lattice .
in presence of random field , the sharpness of the transition gets smeared out and the change in magnetization with field takes place in a series of `` spin - cluster '' flips and the magnetization changes in sporadic jumps as is observed in _
simulations of _ rfim _ in three - dimensions show that for large values of @xmath5 ( greater than a critical value @xmath9 ) , the spins flip in small clusters resulting in avalanches of small sizes .
the sizes of these avalanches are distributed in a power - law . at @xmath10 avalanches of all sizes
occur . for @xmath11
, the magnetization shows a first order jump ( corresponding to an infinite avalanche ) as is the case if there is no disorder .
the behavior of avalanches and avalanche size distribution at and near @xmath9 has been studied in the mean field limit [ 3 ] and the power law relating to the avalanche size distribution is known exactly in _
rfim _ in one - dimension and on bethe lattices [ 11 ] . for a linear chain and a bethe lattice of coordination number z
=3 the avalanche size ( s ) disribution decreases exponentially for large _
s_. no first order jump has been observed for any finite amount of disorder ( unbounded ) for @xmath12 .
thus the low @xmath5 behavior depends on the coordination number of the underlying lattice . for @xmath13 and for small disorder
the magnetization shows a first order discontinuity for several continuous and unimodal distributions of the random fields . the avalanche distribution _
p(s ) _ varies as @xmath14 for large s near the discontinuity .
this growth of magnetization is essentially due to the motion of an interface of a cluster of positive spins through a disordered media .
renormalization group studies of the interface ( starting from a continuum model and an interface without any backbends ) motion in disordered medium shows that the interface at the critical value of the strength of the disorder becomes rough with a roughness exponent @xmath15 [ 16 ] for a d - dimensional system .
according to this relation the upper critical dimension is five and the lower critical dimension is two . at @xmath16 , it is considered that the interface may have backbends and can encompass bubbles as it moves through the medium .
simulation results show the following generic features of the interface growth in random media [ 17 - 24 ] : ( i ) for large strengths of the disorder the interface growth is percolation-like.the fractal dimension of the magnetic - interface corresponds to that ordinary site percolation .
( ii ) at the critical value of the disorder the interface becomes self - affine with a roughness exponent which closely matches the rg result .
( iii ) for low strength of the disorder the interface growth is faceted .
we study the spin flipping dynamics in _ rfim _ in two - dimensions at zero temperature as the external field is ramped up slowly from a high negative value .
our intention is to see if there is any finite value of the disorder strength @xmath9 in two - dimension . at @xmath9 and at a particular field @xmath17 ,
the spin starts flipping from a site and does not end till the flipped spins span the entire system .
the domain wall that separates the flipped spin regions from the rest is highly tortuous ( fractal like ) .
the spin flipped region has bubbles of unflipped spins , so that @xmath7 does not jump from -1 to + 1 , instead attains a value @xmath18 irrespective of system size . in the regime when the flipped spin domain grows , the number _
m _ of flipped spins and the interface length _
i _ both grow with time _
t _ as @xmath19 . for @xmath11 , _ m
_ jumps from -1 to + 1 at a particular field and the flipped spin domain remains compact as it grows following @xmath20 . for @xmath21 , @xmath7 grows continuously with the field in jumps of different sizes .
these jumps correspond to flipping of small spin clusters at various places in the lattice .
we have studied rfim in two - dimensions on a square lattice of size _
l _ with a classical ising spin @xmath22 and a quenched random field @xmath23 on every site @xmath24 of the lattice .
each spin interacts with each of its nearest neighbor spin through a ferromagnetic interaction @xmath25 .
the random field @xmath23 is drawn from a uniform distribution @xmath26 the spins are subjected to an applied field @xmath3 which is varied .
we study the model at zero temperature .
we start with a configuration of all down spins which corresponds to a high negative value of @xmath3 .
the field @xmath3 is then slowly raised and the resulting spin flips are recorded .
the spin flipping process follows the rule of zero temperature glauber dynamics [ 25 ] , where a spin is flipped only if the flipping lowers the energy of the system as calculated according to the hamiltonian given by
eq ( 1 ) " .
we use periodic boundary conditions in our simulation . for a certain _ l _ , @xmath5 and a configuration of the random field @xmath23 we first find out the spins @xmath27 which can be most easily flipped and the corresponding value of the applied field @xmath28 required to flip the spins .
we then set @xmath17 and flip all the spins in @xmath27 simultaneously .
this constitutes one time step in our simulation .
next we check if this primary set of spin flips introduce secondary spin flips and so on , always flipping all the flippable spins simultaneously . if there is no spin that can be flipped , we increase @xmath3 to trigger next generation of spin flipping .
we continue this process till all the spins of the lattice are flipped .
we have taken results for _ l _ = 500 to 7000 and for various values @xmath5 from 2.0 to 2.8 . for a certain @xmath5 ,
our results are averaged over 50 initial random configurations of the on - site fields .
our simulation results show that there is a critical value @xmath9 of the strength of the disorder .
the growth morphology of domains of flipped spins are distinctly different for @xmath11 , @xmath21 and for @xmath10 .
we discuss the three cases below : for @xmath11 as the external field is ramped up there is a sharp first order transition ( like that happens in the absence of disorder ) at a particular field ( @xmath29)that takes the magnetization from -1 to + 1 as is shown in fig.1 this infinite avalanche nucleates from a single spin and invades the system in course of time .
the flipped spin cluster remains compact while it grows : the mass _ m _ ( the number of upturned spins ) of the cluster grows with time _
( in monte carlo steps ) as @xmath30 ( fig.2 ) . the number of spins _
i _ along the interface grows as @xmath31 ( fig.2 ) . for @xmath21
an infinite avalanche never happens .
the magnetization increases in irregular steps with the increase of external field as shown in fig.3 . in fig.4
we show a snapshot of the upturned spins at a particular field when the flipping of the spins has stopped . in this regime
we see nucleation of many domains as opposed to the nucleation of a single domain for @xmath11 . such a feature
is also observed in dilute _ rfim _ [ 10 ] . for any field , when the system has stopped evolving , a typical distribution of the random fields over the unturned spins along the interface is as shown in fig.5 .
it consists of two steps .
the higher step is for those sites that have large negative random fields that would require three upturned neighbors for its flipping .
the lower step corresponds to the random fields of those sites which can flip if two of its neighbors have already flipped .
this shows that the the interface spins are strongly pinned .
the non - existence of sites ( at the interface ) that can flip when one of its neighbor is up is responsible for the freezing of further spin flipping at the particular field . at the critical value of @xmath10
we see the growth of a single domain and a discontinuous first order jump in the magnetization at a particular value of the applied field .
however , unlike the situation for @xmath11 here the domain of flipped spins encompasses bubbles of unturned spins of various sizes at different stage of its growth and the interface has overhangs and is fractal like .
as a result magnetization does not jump from -1 to + 1 but to ( around ) 0.76 irrespective of the system size ( we have checked it for _ l _ = 500 , 800 , 2000 , 4000 , 5000 and 7000 ) . the change in magnetization with field is shown in fig.6 .
a typical up - spin cluster at a certain time of its developement is shown in fig.7 .
it shows the tortuous interface with bubbles of unturned spins ( white region ) . in this case both the mass and interface develops as @xmath32 as is shown in fig.8 . at any time during the growth of the up - spin domain the probability distribution of the random fields of the unturned spins along the interface has three steps as is shown in fig.9 .
the lowest step corresponds to the random fields on those sites that can flip if one of its neighbor is up .
this guarantees the growth of domains as along the interface one neighbor will always be up .
however when the up - spins span the system the probability distribution of the random fields along the interface corresponds to those for the bubbles only and is similar to that for @xmath21 .
this shows that the spins along the interface of the bubbles are strongly pinned .
the external field has to be continuously increased for the magnetization to change from 0.76 to + 1 which will eventually fill up the bubbles ( fig.6 ) .
we find a finite , non - zero @xmath9 for bounded distribution .
this @xmath9 demarcates a region ( @xmath21 ) where infinite avalanches never happens to that from a region where magnetization shows a jump discontinuity from -1 to + 1 .
earlier studies indicate @xmath33 for unbounded distribution .
we find at @xmath9 the interface around the flipped domains have backbends and is fractal like .
it encloses bubbles of unflipped spins in all length scales . | we study the dynamics of spin flipping at first order transitions in zero temperature two - dimensional random - field ising model driven by an external field . we find a critical value of the disorder strength
at which a discontinuous sharp jump in magnetization first occurs .
we discuss growth morphology of the flipped - spin domains at and away from criticality . |
according to the standard hierarchical structure formation theory based on the @xmath0cdm cosmology model , density perturbations grow into small haloes where the first stars and galaxies form ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these astrophysical sources produce ultraviolet ( uv ) ionizing photons that escape into the intergalactic medium ( igm ) which is eventually reionized .
the epoch when this process occurs is called the `` epoch of reionization '' ( eor ) .
the detailed astrophysics of the eor is currently poorly understood because no observation has yet probed the early and middle stages of the eor although current observations have provided fruitful information on late stage of the eor .
for example , the absorption spectra of high-@xmath1 quasars indicate that reionization was complete by @xmath2 ( e.g. * ? ? ? * ) and the number density of ly@xmath3 emitter galaxies at @xmath4 implies that the neutral hydrogen fraction increases at @xmath5 ( e.g. * ? ? ? * ) . on cosmological scales
, reionization induces thomson scattering of cmb photons off free electrons .
the optical depth of thomson scattering measured by planck is @xmath6 which corresponds , using an instantaneous reionization toy model , to a reionization at @xmath7 @xcite . to further improve our understanding of the eor , observations targeted on the cosmological redshifted 21 cm signal from the eor are on - going .
the 21 cm signal emission is due to the hyperfine structure of neutral hydrogen atoms and is expected to be a powerful tool to probe the neutral igm , yielding both astrophysical and cosmological information such as matter density fluctuations and the ionization state and thermal history of the igm at high redshift @xcite .
recently , some first - generation radio interferometers have been attempting to detect statistically the 21 cm signal from the eor , such as the murchison wide field array ( mwa ) @xcite , the low frequency array ( lofar ) @xcite and the precision array for probing the epoch of reionization ( paper ) @xcite .
these observational efforts have resulted in upper limits of the 21 cm power spectrum and on the ionized state of the igm at @xmath8 ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
furthermore , future instruments such as the square kilometre array ( ska ) @xcite and hydrogen epoch of reionization array ( hera ) @xcite are designed to detect the 21 cm signal power spectrum with higher signal to noise ratio and at higher redshift , during the cosmic dawn .
the ska should also be able to image the signal in 3d , which requires high sensitivity on small enough scale ( that is sufficient collecting area in a large enough core ) .
so we expect a wealth of 21 cm data in the near future .
then we face the fundamental question : what we can learn from the data ? from theoretical and numerical works
, we have already some insights on the process of reionization . for instance , considering galaxies in relatively massive host haloes results in larger and more uniform ionized bubbles and imprints a peak at larger scales in the 21 cm power spectrum ( e.g. * ? ? ?
* ; * ? ? ?
on the other hand , abundant and small minihaloes , which serve as absorption systems by self - shielding from ionizing photons , result in small and disjointed ionized regions ( e.g. * ? ? ? * ) . to extract information on the eor from the observed 21 cm signal
, we need to be able compute the 21 cm signal from the basic physics of reionization ( e.g. * ? ? ?
the most self - consistent method is to run numerical simulations .
there are currently two approaches .
the first is to implement the full radiation hydrodynamics ( rhd ) .
this type of simulations is relevant if small scales , where feedback effects such as photo - heating play an important role , are resolved ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the second approach is to run large scale simulations where radiative transfer ( rt ) is computed in post - processing ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this type of simulations are able to account for large scales fluctuations in the signal but they require sub - grid modelling to treat processes on galaxy scales ( feedback , ionizing photons escape fraction , etc . ) .
although numerical study based on simulations is the most consistent method , it implies a large computational cost .
this disadvantage is somewhat alleviated by using a semi - numerical approach instead of rhd or rt simulations . in most semi - numerical simulations ,
the production of ionizing photons is calculated based on the excursion set formalism , using analytically derived halo mass functions , and density fluctuations are obtained with linear perturbation theory @xcite .
however , the growth of hii regions is simply evaluated by considering the balance between the production of ionizing photons and the number of neutral hydrogen atoms @xcite .
recently , the effect of recombination is also taken into account ( e.g. * ? ? ?
the results produced by the semi - numerical approach shows good agreement with those by rt simulations on large scale ( @xmath9 mpc ) @xcite . to maximize the scientific return of the upcoming observations it is important to establish systematic procedures to derive constraints on the eor modeling parameters from the observed 21-cm data
recently , several works have studied how such constraints can be obtained by exploring the eor parameter space with semi - numerical simulations . for example , fisher analysis @xcite and bayesian parameter inference such as the markov chain monte carlo ( mcmc ) approach @xcite have been applied . in this work , we suggest a new approach for parameter reconstruction based on a machine learning method .
machine learning is one of the hot topics in data science as a method to deal with big data .
it is currently applied to many fields such as pattern recognition or search engine .
the main purpose of machine learning is to find approximate functions that , given the input produce the desired outputs .
this is achieved by learning " from training datasets with known inputs and outputs ( e.g. * ? ? ?
* ; * ? ? ?
recently , machine learning methods have been applied in the field of astronomy .
for example , learning from huge galaxy image catalogs helps with morphological classification of galaxies @xcite .
machine learning can also help with the selection and classification of transients @xcite .
applying machine learning to a large sample of spectroscopic and photometric galaxies data can improve the accuracy of estimates of photometric redshifts . using simulated gravitational wave data with noise as learning sample ,
machine learning can help us search for gravitational wave signals from noisy real data @xcite . in the context of cosmology , machine learning
is used to model galaxy formation @xcite or to make templates of nonlinear matter power spectrum @xcite .
closer to our field of interest , there is a study applying a machine learning method to estimate the escape fraction of ionizing photons during the eor @xcite . in this study , they show how machine learning can estimate the lyman continuum escape fraction by using mock spectroscopic simulation data . in our work , using a simple astrophysical parameterization of the eor , we apply artificial neural network ( ann ) , which is one of the machine learning methods , to reconstruct the parameter values from the 21 cm power spectrum data .
this paper is organised as follows . in section [ sec:21 cm ]
, we introduce the cosmological 21 cm signal and the eor parameterization we focus on . in section [ sec : ann ] , we describe our artificial neural network and test the impact of its chosen architecture . in section [ sec : result ] we show our main results , and we give a summary and discussion in section [ sec : summary ] . throughout this paper , we employ the best fit cosmological parameters obtained by .
the brightness temperature for the @xmath10 cm signal is given by ( e.g. * ? ? ?
* ) : @xmath11 .
\label{eq : brightness}\end{aligned}\ ] ] here , @xmath12 and @xmath13 represent the local spin temperature of the igm and the cmb temperature , respectively .
@xmath14 is the local optical depth in the 21 cm rest frame frequency @xmath15 , @xmath16 is the local neutral fraction of the hydrogen gas , @xmath17 is the evolved matter overdensity , @xmath18 is the local gradient of the gas velocity along the line of sight and @xmath19 is the hubble parameter .
all quantities are evaluated at redshift @xmath20 .
as we can see , the 21 cm signal includes both astrophysical and cosmological information .
thus , we can hope to use the 21 cm signal to disentangle and quantify them @xcite .
let us now introduce the power spectrum of the 21 cm fluctuations .
we define the 21 cm power spectrum as @xmath21 in our context , we use the _ dimensional _ 21 cm power spectrum , @xmath22.the 21 cm ps would describe the statistical properties of the 21 cm fluctuations perfectly if they were a gaussian random field .
however , the 21 cm fluctuations are expected to deviate from a gaussian behaviour because of astrophysical effects such as ionization and x - ray heating .
thus it is useful to compute higher order statistics and one - point statistics such as the bispectrum , the variance and the skewness to estimate non - gaussian features in the 21 cm fluctuations . in order to generate the 21 cm ps for a given set of astrophysical parameters
, we use the publicly available code * 21cmfast * @xcite .
this code is based on a semi - numerical model of cosmic reionization and thermal history of the igm .
it quickly generates maps and ps of the brightness temperature , matter density , velocity , spin temperature and ionization fraction at designated redshifts .
we performed simulations in a @xmath23 @xmath24 comoving box with @xmath25 grid cells for a wide range of eor parameters described in the next section . in our calculation , we use the 21 cm power spectrum in the range @xmath26 divided into 14 bins .
it is common to characterize eor models with parameters and then examine the effect of changing the parameters on the 21 cm signal .
we employ three key parameters which are often used .
let us briefly define these three parameters : + 1 .
@xmath27 , _ the ionizing _ _ efficiency _ : @xmath27 is the combination of several parameters related to ionizing photons escaping from high redshift galaxies and is defined as @xmath28 @xcite . here
, @xmath29 is the fraction of ionizing photons escaping from galaxies into the igm and @xmath30 is the fraction of baryons locked into stars .
these parameters are extremely uncertain at high redshift @xcite .
@xmath31 is the number of ionizing photons produced per baryon in stars and @xmath32 is the mean recombination rate per baryon . in our calculation , we explore a range of @xmath33 .
@xmath34 , _ the minimum _ _ virial _ _ temperature _ _ of _ _ haloes _ _ producing _ _ ionizing _ _ photons _ : @xmath34 parameterizes the minimum mass of haloes producing ionizing photons during the eor .
typically , @xmath35 is chosen to be @xmath36 , corresponding to the temperature above which atomic cooling becomes effective .
@xmath35 parameterizes the physics of star formation in high redshift galaxies . in haloes with virial temperature @xmath37
atomic cooling is sufficient to trigger gravothermal instability and thus star formation .
however , star formation is quenched if agn or supernovae feedback is effective and the igm is heated up .
this leads to effective minimum virial temperature larger than @xmath38[k ] . in haloes with viral temperature @xmath39 ,
hydrogen molecule cooling is necessary .
however , if stars begin to form in a halo , radiative feedback such as the photodissociation of @xmath40 by lyman - werner photons may become effective and prevent the gas from cooling @xcite .
conversely , positive feedback , such as the enhancement of @xmath40 molecules formation due to an increase in the free electrons density , tends to push the minimum virial temperature to lower value because cooling becomes more effective .
thus @xmath35 parameterizes the uncertainty in the efficiency of radiative feedback . in our work ,
we explore @xmath35 ranging from @xmath41 to @xmath42 .
@xmath43 , _ the mean free path of ionizing photons _ : the propagation of ionizing photons through the ionized igm strongly depends on the presence of absorption systems and the sizes of ionized regions are determined by the balance between sinks and sources of ionizing photons ( e.g. * ? ? ?
this process is modelled by the maximum mean free path of ionizing photons , @xmath43@xcite . physically
, the mean free path of ionizing photons corresponds to the typical distance traveled by photons within ionized regions before they are abosorbed and is determined by the number density and the optical depth of lyman - limit systems . in our calculation
, we explore @xmath43 from 10 mpc to 60 mpc .
in this section , we introduce artificial neural networks ( ann ) .
anns are one of the machine learning methods and are a mathematical model inspired by the natural neuron network in our brain .
the main purpose of anns is to construct approximate functions which associate input data with output data . in order to construct such a function
, the ann has to learn from `` _ training data _ '' .
the architecture of a simple class of ann consists of three layers : the input layer , the hidden layer and the output layer .
each of them has a number of neurons as shown in fig.[fig : fig1 ] . in a more general case
, we could choose the number of hidden layers and the number of neurons at each layer arbitrarily .
in our study , we use 1 hidden layer . note that it is mathematically proven that neural networks with only 1 hidden layer can approximate any function with any accuracy if we use a large enough number of neurons @xcite .
let us briefly describe the architecture of our ann .
the input data @xmath44 is fed to the @xmath45-th neurons in the input layer .
each neuron in the input layer is connected to the @xmath46-th neuron in the hidden layer and a weight @xmath47 is associated with the connection .
the input to the @xmath46-th neuron in the hidden layer @xmath48 is a linear combination of all the input neurons with weight @xmath47 : @xmath49 here , @xmath50 is the number of input data . in the hidden layer ,
the @xmath46-th neuron is activated by an activation function @xmath51 such as its output is @xmath52 . generally , the activation function is a nonlinear function .
we use the sigmoid function @xmath53 .
the properties of the sigmoid function are such that it saturates and returns a constant output when the absolute value of the input is large and that it is a smooth and differentiable function .
thanks to the nonlinear activation function , a trained ann can express any function . in the output layer ,
we compute linear combinations of the activated outputs of the neurons in the hidden layer with weights @xmath54 and obtain the output vector : @xmath55 here @xmath56 is the number of neurons in the hidden layer .
note that we do not activate the output value .
the aim of training the ann is to find a set of weights that ensures the output vectors produced by the ann for a set of input vectors is sufficiently close to the desired output vectors .
once we adjust the weights to reach this goal on a training sample , we can make predictions for output vectors for arbitrary input vectors outside of the training sample ( for example , new observational data ) .
a popular algorithm to compute the trained weights is the `` _ back propagation algorithm _ ''
we will describe this algorithm briefly in the following section . in this section
, we present the back propagation algorithm for the 1 hidden layer case .
we also show the back propagation algorithm for multiple hidden layers case in appendix . in order to quantify
how well the output obtained by the ann approximates the desired output for the training data set , we define the ( total ) cost function as : @xmath57 , \label{eq : cost}\ ] ] where @xmath58 is the number of training input vectors and @xmath59 is the number of neurons at the output layer . @xmath60 and @xmath61 are outputs of the ann and the ( desired ) training output data , respectively . our purpose is to find the weight set that minimises the cost function . in order to find this weights set , we need to compute the partial derivative of @xmath62 with respect to the individual weights @xmath63 and find the local minimum of @xmath62 using gradient descent .
the weights are updated by gradient descent following the formula : @xmath64 here , @xmath65 is a learning coefficient which controls how fast the weights are updated .
we used @xmath66 .
we only need to calculate the derivative of the cost function for each training input vector and then sum over all input vectors as shown in eq.[eq : gradient ] .
first , let us consider the derivative with respect to the weights between output layer and hidden layer . in this case ( @xmath67=2 ) , we can simply calculate the derivative of @xmath62 as @xmath68 next , we calculate the derivative of @xmath62 with respect to the weights between the hidden layer and the input layer . in this case
( @xmath67=1 ) , the derivative of @xmath62 is @xmath69 in the second line , we use the chain rule for derivative because @xmath62 depends on the activated neuron @xmath70 in the hidden layer only through the output neuron @xmath60 . here , @xmath71 denotes the derivative of the activation function with respect to @xmath72 .
using eqs.([eq : gradient ] ) , ( [ eq : derivative1 ] ) and ( [ eq : derivative2 ] ) , we can iterate on the gradient descent until the outputs obtained by the ann converge to the desired outputs ( minimum of the cost function ) .
the back - propagation algorithm can be summarised as follows : 1 . starting with random weights ,
compute the output of the ann using eq.([eq : hidden ] ) and eq.([eq : output ] ) for all input vectors in the training set ( _ forward propagation _ ) + 2 .
compute the cost function + 3 . compute the derivative of the cost function with respect to the weights between output layer and hidden layer with eq.([eq : derivative1 ] ) and then the derivative with respect to the weights between input layer and hidden layer with eq.([eq : derivative2 ] ) ( _ back - propagation _ ) .
update the weights with eq.([eq : gradient ] ) .
go back to ( i ) and iterate until the cost function converges to a minimum .
in our case , we prepared 70 training data sets .
each set consist of the 21 cm ps @xmath73 obtained with 21cmfast as input data and the corresponding eor parameters used in the simulation , @xmath74 as output data .
the architecture of the ann is the following : ( _ i _ ) 14 neurons in the input layer , ( _ ii _ ) 14 neurons in the hidden layer , ( _ iii _ ) 3 neurons in the output layer .
as we mentioned in section [ sec:21cmps ] , we use 14 bins for the 21 cm ps , and we use 3 eor parameters .
this is why the number of neurons in the input layer and in the output layers are 14 and 3 , respectively .
note that the number of redshifts used to train will always match the number of redshifts being fit .
to analyse in details how the ann performs we look at partial cost functions : we use the normalised root mean square error ( rmse ) defined by @xmath75 here , @xmath76 ( with @xmath77 one of @xmath78 or , @xmath79 .
@xmath80 and @xmath81 are the eor parameters evaluated by the ann and from the training data , respectively .
( red ) , @xmath27 ( green ) and @xmath35 ( blue ) as functions of the number of iterations for the learning process , for a network with 14 neurons . ]
( solid line ) , @xmath27 ( dashed line ) and @xmath35 ( dot - dashed line ) as functions of the number of neurons for @xmath82 iterations in the learning process . ]
first , we study how the rmse depends on the number of iteration for each eor parameter .
the result is plotted in fig.[fig : fig2 ] . in this figure , we fix the number of neurons in the hidden layer at 14 and perform calculation with @xmath83 iterations . after initial fluctuations ,
the rmse decreases and converges for @xmath84 iterations .
next , we show how the rmse depends on the number of neurons in the hidden layer after @xmath85 iterations .
the rmse is only weakly dependent on the number of neurons in the explored range . in the following section , we perform @xmath82 iterations for the back - propagation algorithm with 14 neurons in the hidden layer unless specifically mentioned
we apply the trained ann to 54 test datasets . as for the training datasets , these test datasets consist of the 21 cm ps and the corresponding eor parameters .
we use the 21 cm ps data as input for the ann and obtain estimated values for the eor parameters .
then , we compare the estimated values with the true values that were used to compute the ps in the 21cmfast simulation , and evaluate how the ann performs .
first , we apply the ann to the 21 cm ps at a single redshift without including the effect thermal noise or sample variance . in figs.[fig : fig4 ] and [ fig : fig5 ] , we plot the eor parameters estimated with the ann from the 21 cm ps data against the true values . in fig.[fig : fig4 ] , we show the result at @xmath86 .
as we can see , @xmath27 and @xmath35 estimated with the ann are in relatively good agreement with the true values . on the other hand , for @xmath43
, the correct values are not recovered .
this is because the 21 cm ps is not sensitive at @xmath86 to @xmath43 . on the other hand ,
the 21 cm ps becomes sensitive to @xmath43 at lower redshift when reionization advances much more ( since @xmath43 physically expresses the maximum bubble size , the effect of @xmath43 on the 21 cm ps is remarkable mostly at lower redshift ) .
thus the cost function is insensitive to changes in @xmath43 values ; the learning process of the ann is incomplete and systematic deviations remain .
note that the reason why the estimated value of @xmath43 seems to be constant at @xmath87 is that @xmath43=30 occurred more often than other values in the training datasets . in fig.[fig :
fig5 ] we show the same plots obtained from the ps at @xmath88 .
the agreement between recovered and expected values extends over a larger range than at @xmath1=12 , in particular for @xmath43 .
indeed the 21 cm ps at @xmath88 is sensitive to @xmath43 and thus the ann learning process works better . in the previous two cases , we considered the 21 cm ps without any source of noise .
we will now consider both the contribution of thermal noise and sample variance .
we model the thermal noise as a gaussian random field characterized by its power spectrum . in an annulus in fourier space with @xmath89 cells ,
the thermal noise power spectrum is ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) : @xmath90 where @xmath77 is the angle between * k * and line of sight , @xmath91 is the effective area of antenna , @xmath92 is the observed wavelength and @xmath93 are factors converting @xmath94 space units into comoving wavenumber units and are determined by cosmology .
@xmath95 is expressed by @xmath96 where @xmath97 is the total system temperature , @xmath98 is the bandwidth and @xmath99 is the effective observing time .
the spherically averaged noise can be obtained by summing over @xmath77 in a shell with radius @xmath56 shown in eq.[eq : noise_average ] .
@xmath100^{2}\right\}^{-1/2}. \label{eq : noise_average}\ ] ] we recommend to read mcquinn et al .
( 2006 ) . in our case , we assume ska specification @xcite .
we estimate the sample variance from 10 simulations using different realizations of the initial conditions . in the general case , we would include the noise at the level of the visibilities and compute the resulting noisy power spectrum . assuming that the noise is a random gaussian field , we simply generate a noisy power spectrum by using the following formula : @xmath101 where @xmath102 is the noisy power spectrum , @xmath103 is the power spectrum produced by the simulation for a given set of parameters generically labeled by the index j , and @xmath104 is _ a random draw _ from a gaussian probability distribution with variance equal to the thermal noise power spectrum or the sample variance . a different and independent draw is required for each value of @xmath56 . for the learning sample
we include only the sample variance . for each parameter set ( 70 possible values for @xmath46 ) ,
we generate 50 realizations of the noise ( labeled by @xmath105 ) .
this means that we use 3500 different ps for the learning sample .
this is necessary as we are not aware of a standard technique for directly including an uncertainty in the inputs of an ann . for the test datasets ,
we add both thermal noise and sample variance to the 21 cm ps , using the same procedure .
note that the result for the virial temperature is plotted in log scale .
it is the same for the following figures . ]
= 12 . in this case
, we include both thermal noise and sample variance . ] in fig.[fig : fig6 ] , we show the eor parameters found by the ann as functions of the values used in the simulations , using the 21 cm signal ps at @xmath88 , including both thermal noise and sample variance .
the difference between fig.[fig : fig5 ] and fig.[fig : fig6 ] is not obvious at a glance . in order to quantify the difference
, we compute the _
mean chi - square value _ , @xmath106 : @xmath107 @xmath80 is the @xmath43 , @xmath35 , @xmath27 reconstructed by the ann and @xmath108 is the value of the corresponding parameter used in the simulation . in table.[table : chi ] , we compare the @xmath106 values for each of the parameters , with and without noise .
as we can see , the @xmath106 values for the 21 cm ps with noises are worse than those without noises .
noise alters the efficiency of the learning process for the ann .
[ my - label ] [ cols="<,<,<,<,<,<",options="header " , ]
here , we show the back propagation algorithm in the case of multiple hidden layers . by analogy with eq.([eq : derivative2 ] ) , we can express the derivative of @xmath109 with respect to the weight in the @xmath67-th layer , @xmath110 @xmath111 as @xmath112 where @xmath113 .
since the changes in @xmath114 are transmitted to @xmath109 through each @xmath115 in the ( @xmath67 + 1)-th layer , the derivative of @xmath109 with respect to @xmath114 can be expressed as @xmath116 remember that @xmath115 can be expressed with the activation function @xmath51 as @xmath117 , @xmath118 here ,
we define @xmath119 , then we can re - write eq.([eq : derivativel2 ] ) as @xmath120 combining eq.([eq : delta ] ) with @xmath121 , the eq.([eq : derivativel ] ) can be re - written simply as @xmath122 this form tells us that we can easily obtain the derivative of the cost function with respect to @xmath110 , which connects the neuron @xmath45 in the ( @xmath67 - 1 ) th layer to the neuron @xmath46 in the @xmath67 th layer , as the product of @xmath123 and @xmath124 . as shown in eq.([eq : delta ] ) , we start to calculate @xmath123 from the output layer ( @xmath67=l ) to the input layer .
this is why this algorithm is called `` _ back propagation _ '' .
if we use eq.([eq : cost ] ) as the cost function , we easily calculate @xmath125 .
this work is benefited from a grant from the french anr funded project orage ( anr-14- ce33 - 0016 ) .
we thank to g.mellema , a. fialkov , s. majumdar , s.giri and k. hasegawa for their useful comments and thank to s. yoshiura for providing the thermal noise data .
abel , t. , bryan , g. l. , & norman , m. l. 2002 , science , 295 , 93 agarwal , s. , abdalla , f. b. , feldman , h. a. , lahav , o. , & thomas , s. a. 2012 , , 424 , 1409 agarwal , s. , abdalla , f. b. , feldman , h. a. , lahav , o. , & thomas , s. a. 2014 , , 439 , 2102 beardsley , a. p. , hazelton , b. j. , sullivan , i. s. , et al .
2016 , , 833 , 102 bernardi , g. , zwart , j. t. l. , price , d. , et al .
2016 , , 461 , 2847 bloom , j. s. , & richards , j. w. 2012 , advances in machine learning and data mining for astronomy , 89 bromm , v. , coppi , p. s. , & larson , r. b. 2002 , , 564 , 23 bromm , v. 2013 , reports on progress in physics , 76 , 112901 furlanetto .
p. s and briggs .
f , phys . rept .
* 433 * ( 2006 ) 181 [ astro - ph/0608032 ] .
deboer , d. r. , parsons , a. r. , aguirre , j. e. , et al .
2016 , arxiv:1606.07473 deboer , d. r. , parsons , a. r. , aguirre , j. e. , et al .
2016 , arxiv:1606.07473 fialkov , a. , barkana , r. , visbal , e. , tseliakhovich , d. , & hirata , c. m. 2013 , , 432 , 2909 fialkov , a. , cohen , a. , barkana , r. , & silk , j. 2017 , , 464 , 3498 folkes , s. r. , lahav , o. , & maddox , s. j. 1996 , , 283 , 651 greig , b. , & mesinger , a. 2015 , , 449 , 4246 greig , b. , mesinger , a. , & koopmans , l. v. e. 2015 , arxiv:1509.03312 greig , b. , mesinger , a. , & pober , j. c. 2016 , , 455 , 4295 gnedin , n. y. , kravtsov , a. v. , & chen , h .- w .
2008 , , 672 , 765 - 775 hornik , k. , stinchcombe , m . ,
white , h. , _ neural networks _ vol.2 pp359 - 366 , 1989 jacobs , d. c. , pober , j. c. , parsons , a. r. , et al .
2015 , , 801 , 51 jensen , h. , zackrisson , e. , pelckmans , k. , et al .
2016 , , 827 , 5 kamdar , h. m. , turk , m. j. , & brunner , r. j. 2016 , , 455 , 642 kamdar , h. m. , turk , m. j. , & brunner , r. j. 2016 , , 457 , 1162 kim , k. , harry , i. w. , hodge , k. a. , et al . 2015 , classical and quantum gravity , 32 , 245002 mellema , g. , koopmans , l. v. e. , abdalla , f. a. , et al . 2013 , experimental astronomy , 36 , 235 mesinger .
a and furlanetto .
s , arxiv:0704.0946 [ astro - ph ] .
. a , furlanetto .
r,2011 , mnras , 411 , 955 mesinger , a. , ferrara , a. , & spiegel , d. s. 2013 , , 431 , 621 yoshida .
n , omukai .
k , hernquist .
l , & t. abel 2006 , apj , 652 , 6 yoshiura.s , shimabukuro.h , takahashi.k , et al .
2015 , mon . not .
451 , 4785 zahn , o. , lidz , a. , mcquinn , m. , et al .
2007 , , 654 , 12 zahn , o. , mesinger , a. , mcquinn , m. , et al .
2011 , , 414 , 727 | the 21 cm signal from the epoch of reionization should be observed within the next decade .
while a simple statistical detection is expected with ska pathfinders , the ska will hopefully produce a full 3d mapping of the signal . to extract from the observed data constraints on the parameters describing the underlying astrophysical processes ,
inversion methods must be developed .
for example , the markov chain monte carlo method has been successfully applied . here
we test another possible inversion method : artificial neural networks ( ann ) .
we produce a training set which consists of 70 individual sample .
each sample is made of the 21 cm power spectrum at different redshifts produced with the 21cmfast code plus the value of three parameters used in the semi - numerical simulations that describe astrophysical processes . using this set we train the network to minimize the error between the parameter values it produces as an output and the true values .
we explore the impact of the architecture of the network on the quality of the training .
then we test the trained network on the new set of 54 test samples with different values of the parameters .
we find that the quality of the parameter reconstruction depends on the sensitivity of the power spectrum to the different parameters at a given redshift , that including thermal noise and sample variance decreases the quality of the reconstruction and that using the power spectrum at several redshifts as an input to the ann improves the quality of the reconstruction .
we conclude that anns are a viable inversion method whose main strength is that they require a sparse exploration of the parameter space and thus should be usable with full numerical simulations .
[ firstpage ] cosmology : theory intergalactic medium epoch of reionization 21 cm line |
understanding the effect of confinement on crystalline materials is crucial to many technological applications , such as lubrication , adhesion , and fabricating novel optical , catalytic materials @xcite .
for example , during film processes , which are important for optoelectronics , magnetic and electronic materials @xcite , internal stresses often arise affecting the mechanical stability and strength of the film @xcite . here
we study increasingly confined colloidal crystals to investigate how changes in their microstructure affect the microscopic stresses ultimately responsible for their bulk mechanical properties .
such colloidal systems are comprised of particles small enough to undergo brownian motions that preserve a thermodynamic ensemble , while still large enough to be optically imaged with high temporal and spatial resolutions @xcite .
as such , colloidal crystals have been widely used as a model system to elucidate many phenomena associated with confinement including crystal growth @xcite , packing @xcite , and melting mechanisms@xcite . in principle , knowing all particle positions and their interactions is sufficient to determine the suspension s structural order and stress distribution .
unfortunately , resolving the stress distribution within colloidal materials has remained experimentally challenging . in the simplest colloidal system hard - spheres
particles do not interact until contact .
the steep hard - sphere potential and experimental noise in locating particles make potential - based stress calculation ( force times relative vector ) impractical in experiment . in this work
, we use a new technique stress assessment from local structural anisotropy ( salsa ) to measure the stress distribution in confined hard - sphere colloidal polycrystals @xcite .
salsa uses the particle positions captured using confocal microscopy to calculate the orientation dependent particle collision probabilities .
these probabilities are used to determine the stress at the single particle scale . using salsa
we follow the evolving stress distribution in a polycrystal as it is compressed quasi - statically between two parallel plates .
we conduct experiments with suspensions comprised of sterically stabilized poly(methyl methacrylate ) , pmma , particles .
the particles have a diameter of @xmath0 @xmath1 m , polydispersity @xmath2 , and are fluorescently labeled with diic@xmath3(3 ) ( 1,1-dioctadecyl-3,3,3,3-tetramethylindocarbocyanine perchlorate ) for confocal imaging .
the solvent a mixture of decalin and cxb ( cyclohexylbromide ) has a refractive index and density that nearly match those of the pmma particles .
while a slight mismatch between the particle and solvent density @xmath4 0.03 g/@xmath5 is introduced to induce slow sedimentation for the bulk samples , the confined crystal samples are all density - matched .
we approximate the hard - sphere interparticle force by saturating the solvent with tetrabutyl ammonium bromide ( tbab ) at a concentration ( @xmath6 260 nm ) @xcite , and using a syringe filter to remove any excess salt granules @xcite .
the added tbab salt screens the electrostatic force and results in a debye length , @xmath6 100 nm , substantially smaller than the particle diameter@xcite .
we use a high - speed confocal laser scanning microscope ( zeiss , lsm 5 live ) to image the 3d structure of the sample .
we acquire a time series of 15 image stacks where each stack contains up to 512@xmath7512@xmath7500 voxels , corresponding to a sample volume of 71@xmath771@xmath768 @xmath1m@xmath8 .
we use our confocal rheoscope to confine our crystal samples between two plates with a separation that is uniform to within @xmath9 @xmath1 m @xcite .
the top plate , a silicon wafer ( 4 mm @xmath7 4 mm ) , is fixed to the rheoscope s kinematic mount and is static throughout the experiment .
the bottom plate , a transparent coverslip for imaging from below , is attached to a multi - axial piezo electric , and can be moved vertically ( @xmath10-axis ) to change the gap size @xcite .
the multi - axial piezo ( pi p-563.3cd ) has a travel range 300 @xmath1 m in the @xmath10 direction and an accuracy @xmath11 2 nm . using this parallel - plate setup we reduce the gap heights in a controlled manner from 38 @xmath1 m down to 6 @xmath1 m ( 24@xmath12 ) .
we process the confocal data and locate the particle positions using the crocker - grier featuring algorithm @xcite that locates particle positions with sub - pixel accuracy ( @xmath13 nm ) @xcite . from the particle positions , we employ a standard bond - order parameter method @xcite to calculate the local three dimensional ( 3d ) structure order . in the bond - order parameter calculation , we determine the normalized complex order parameter @xmath14 for each particle @xmath15 , @xmath16 , where factor @xmath17 normalizes the order parameter such that @xmath18 , @xmath19 is the spherical harmonic function of the unit vector @xmath20 pointing from particle @xmath15 to @xmath21 , and @xmath22 denotes the average of the neighbors of particle @xmath15 .
the neighboring particles are defined as those with a center - center distance within @xmath23 , which coincides with the first minimum of the radial distribution function @xmath24 . following previous protocol
, we set @xmath25 @xcite .
the number of ordered neighbors is then approximated by summing the complex inner product , @xmath26 . here , the number of ordered neighbors has a range @xmath27 . where @xmath28 denotes heaviside function @xcite .
compared to the previous definition , @xmath29 reports fractional contribution to the local ordering , and more finely resolves the crossover between crystalline and defect structures . ] from the particle positions , we can also measure the stress at the single particle scale using salsa . in our brownian hard - sphere systems ,
the force with which particles collide is related to the thermal energy @xmath30 . using a time series of featured particle positions , we deduce the thermal collision probability , and compute the stress arising from these collisions . as shown by previous work @xcite , the stress tensor @xmath31 at particle @xmath15
can be approximated by @xmath32 , where @xmath33 is the volume occupied by the particle , @xmath34 is the cutoff distance from contact . here
, @xmath35 is the time - averaged _ local structural anisotropy _ for the particle @xmath15 , @xmath36 , where @xmath37 denotes the particles that lie within a distance @xmath38 from particle @xmath15 , @xmath39 are spatial indices , and @xmath40 is the unit vector pointing from particle @xmath15 to particle @xmath21 . in granular literature , @xmath35 also denotes time - averaged fabric tensor @xcite .
while the trace of the time - averaged fabric tensor @xmath41 determines the contact particle number , the off - diagonal terms report the anisotropy of these contact particle configurations .
when averaged over time , the fabric tensor of the selected particle captures the probability of brownian collisions between it and its neighbors .
this probability is linearly proportional to the cutoff distance , or shell thickness @xmath34 when @xmath42 .
therefore , when scaling the collision probability by @xmath34 , the stress is independent of @xmath34 .
the last step of the salsa calculation scales the probability by the energy density per collision @xmath43 . in a defect - free crystal , @xmath44 is simply the system volume divided by the particle number .
however , in a crystal containing defects , the local volume occupied by each particle varies .
this variation needs to be considered to correctly measure the stress near defects .
so , we first calculate the pointwise stress , @xmath45 then perform a spatial average to obtain a macroscopic measurement at the particle - level @xcite . here ,
@xmath46 in theory , should be an infinitesimal volume . in practice , when the pointwise stresses are assigned to a 3d discrete grid , @xmath46 is the single box volume of the grid .
the continuum stress field @xmath47 is then smoothed , @xmath48 , where in our experiments @xmath49 the weighting kernel is a gaussian function @xmath50 @xcite . in particular , we set @xmath51 to remove stress features on length scales smaller than a particle .
we have tested different discrete grid sizes , and other normalized weighting kernels , and find the results insensitive to those changes .
this spatial average effectively addresses local volume variation . finally , while salsa can be modified to account for the contributions due to the confining surfaces , for simplicity , in the data presented here , we exclude boundary particles in all final presentations of local structure and stress distributions .
in the tested bulk crystal , we observe different types of defects including vacancies , dislocations , stacking faults , grain boundaries , and voids .
these defects are often close and can interact with one another .
the typical size of a single crystallite is approximately 50 @xmath1 m @xmath7 50 @xmath1 m @xmath7 50 @xmath1 m , containing on the order of @xmath52 particles .
we show a horizontal slice of a 3d confocal image in fig .
[ nfig_1_structure](a ) . in the image
, we see a crystallite in the center surrounded by several other domains .
the grain boundary of the center domain is highlighted by the dashed red contour . by performing the bond - order parameter analysis
, we determine the number of ordered neighbor particles @xmath53 , and plot its distribution in fig .
[ nfig_1_structure](b ) .
the red particles represent defect regions with lower @xmath53 ; blue particles are crystalline domains with higher @xmath53 .
to better visualize defects deeply embedded in the crystal , we remove crystalline particles with @xmath54 .
the remaining particles are shown in fig . [ nfig_1_structure](c ) . for further clarity
, we remove particles with @xmath10-positions higher than 60 @xmath1 m .
we find that the thresholded particle distribution faithfully captures the grain boundaries highlighted in the raw confocal image illustrated by fig .
[ nfig_1_structure](a ) .
[ htp ] to determine whether the crystalline order affects the stress distribution we apply salsa to our polycrystal . since the magnitudes of individual stress components depend on the orientation of the coordinate system , we focus on primary tensor invariants , including the pressure , the three principal stresses , and the von mises stress .
our polycrystal sample has a mean pressure @xmath55 approximately 39 mpa , consistent with the prediction from previous numerical simulations @xcite .
furthermore , the measured pressure as a function of height is also consistent with the estimated trend of hydrostatic pressure arising from gravitational settling due to the slight mismatch in the particle and solvent densities ( @xmath4 0.03 g/@xmath5 ) .
see sec .
[ bulk_pressure ] for detailed analyses on the bulk pressure . in fig .
[ nfig_2_pressure_vonmises](a ) we show the 3d pressure field of our polycrystal sample .
the red particles indicate regions with higher pressure and blue particles indicate regions with lower pressure .
the pressure fluctuation has a length scale @xmath56 10 particles considerably smaller than the size of a single domain indicating _ intragrain _ stress fluctuations . by comparing the pressure ( fig .
[ nfig_2_pressure_vonmises](a ) ) and @xmath53 ( fig . [ nfig_1_structure](b ) ) distributions , we find that the pressure fluctuation has a relatively random spatial distribution , and is not notably correlated with the grain boundary arrangement .
we also determine the three principal stresses , @xmath57 , @xmath58 , and @xmath59 , by calculating the eigenvalues of the measured three - by - three stress matrix , see sec.[bulk_principal_vonmises ] .
we find that the distribution of each principle stress is similar to that of the pressure .
we therefore extract the difference between them and calculate the deviatoric
von mises stress @xmath60}$ ] . similar to pressure , @xmath61 is also an invariant .
since this invariant satisfies the property that two stress configurations with equal deviatoric strain energy have the same value of @xmath61 , such a scalar is effectively a stress field signature capturing the distortion energy of a material under various loads : the corresponding von mises stress is @xmath62 .
therefore , if a material yields in uniaxial tension at @xmath63 , it will also yield under a shear strain , but now at a lower value ( @xmath64 ) . ] . in continuum
elasticity , the mean of @xmath61 is often used as a criterion in determining how close a metal or alloy sample is to the yield point when subjected to loads @xcite . in atomic - scale simulations ,
similar yielding criteria have been employed to elucidate the local yielding mechanism in glassy systems@xcite . in fig .
[ nfig_2_pressure_vonmises](b ) , we show the von mises stress field @xmath61 of our bulk polycrystal .
we find that the @xmath61 distribution is also roughly random .
counterintuitively , the distribution of the high @xmath61 particles does not follow the trend of the grain boundaries shown in fig .
[ nfig_1_structure](a ) .
in fact , many of the highly stressed regions are well within the crystalline domains ( see sec.[bulk_principal_vonmises ] and fig . [ afig_4_threshold_vonmises ] ) .
the observed uniform and random fluctuations of @xmath65 and @xmath61 are consistent with the evenly spread stress fluctuations previously reported in polycrystals of hard - sphere silica particles @xcite .
confinement is known to affect the structure of crystals @xcite . here , we use salsa to determine whether such structure modifications are accompanied by changes in the stress distribution . to create a confined polycrystal , we load our crystalline suspension volume fraction @xmath66 in a confocal rheoscope .
the rheoscope has three differential screws that allow us to finely adjust the alignment and gap @xmath67 between two parallel plates holding the sample .
we study the confined sample for seven different gap heights , starting with a bulk sample ( @xmath68 @xmath1 m @xmath69 ) .
we then gradually move the bottom plate upward , reducing the gap height down to @xmath70 @xmath1 m @xmath71 .
when the gap decreases , the parallel plates compress the sample and induce a squeeze flow that drives particles outward causing additional structural rearrangement .
an oscillatory shear flow with a small strain amplitude @xmath72 and frequency of @xmath73 hz along @xmath74 is applied for 200 cycles to prevent local jamming and speed up sample recrystallization . importantly
, this low strain amplitude does not generate large structural rearrangements .
thus , the final structure at each height @xmath67 is primarily determined by the squeeze flow and degree of confinement .
[ htp ] we show representative orthogonal slices from the confocal images for four gaps in fig .
[ fig7_confined_bond_order](a - d ) . the black regions above ( green arrow in fig .
[ fig7_confined_bond_order](b ) ) and below ( orange arrow ) the sample ( blue arrow ) correspond to the top and bottom plates , respectively .
we use confocal images to locate particle positions , and perform the bond order parameter analysis .
3d reconstructions of these particles can be seen in fig .
[ fig7_confined_bond_order](e - h ) .
the particles with high @xmath53 are in blue , and low @xmath53 are in red . by inspection
we observe that the local structural order varies non - monotonically with gap with crystals at intermediate gaps @xmath75 appearing more ordered .
in addition , the crystal structure appears to break up into smaller domains at small gaps .
we quantify these observed trends by plotting the @xmath53 histograms for the four different gap heights @xmath67 in fig .
[ fig7_confined_bond_order](i - l ) .
we find that for large gaps @xmath76 the distributions look similar with a relatively narrow width . as the sample
is confined to smaller gaps , however , we observe that the distribution becomes broader with a larger probability of having particles with lower order .
these behaviors are further quantified in fig .
[ fig8_ordered_number ] , in which the mean and standard deviation of @xmath53 for all seven gaps investigated are plotted versus @xmath77 .
these measurements indicate that reduction in structural order and increase in its variance become more pronounced for gaps smaller than ten particle diameters .
these observations are consistent with a large body of previous research indicating confinement effects become significant in samples confined to gaps containing less than ten layers @xcite .
[ htp ] we calculate the 3d pressure distributions of all tested samples , and show horizontal 2d slices ( @xmath78 at @xmath79 ) for the same four representative gap heights @xmath67 in fig . [ fig9_pressure_distribution_confinement](a - d ) .
similar to the bulk measurements , the pressure fluctuations in confined samples are evenly spread , and roughly uncorrelated with the grain boundary arrangement .
furthermore , as anticipated , the mean pressure remains approximately constant ( to within @xmath72 of the mean ) as @xmath67 decreases .
the constant pressure results from the fact that the confined sample is surrounded by a suspension reservoir residing beyond the confining plates that regulates the overall pressure of the confined zone .
we also find that the standard deviation of the pressure is independent of @xmath67 .
the uniformity of the standard deviation is illustrated by the histograms of normalized pressure @xmath80 shown in fig .
[ fig9_pressure_distribution_confinement](e ) , in which all four datasets collapse on a single gaussian distribution , consistent with the gaussian pressure distribution found in the bulk crystal samples ( see fig . [ afig2_bulk_pressure](d ) ) .
we note that while the structural order is substantially smaller at small gaps , the corresponding pressure histogram remains gaussian .
in particular , we do not see evidence of significant deviations from gaussian behavior as is typically observed in glassy systems @xcite .
although the mean and standard deviations of the pressure are unaffected by confinement , its spatial distribution substantially changes . in the bulk sample
, we find that fluctuations are long - ranged @xmath5610 particles .
however , under confinement , these fluctuations become localized , see ( fig . [ fig9_pressure_distribution_confinement](d ) ) . such a stress localization is characterized by calculating the correlation function @xmath81 . here , @xmath82 is unity at origin @xmath83 and zero at @xmath84 while @xmath85 is the variance of the pressure . for simplicity , we plot the radial part of the correlation function @xmath86 in fig .
[ fig9_pressure_distribution_confinement](f ) .
as shown , the correlation function decays faster for smaller gap size @xmath67 , indicating a more localized pressure fluctuation . at the smallest gap we explored @xmath87
the correlation length is @xmath88 about three times shorter than in bulk sample .
this reduction in correlation length tracks the decrease in grain size ( also approximately a factor of three ) as the crystal is squeezed .
[ htp ] we perform the same analysis on deviatoric stress for different gap sizes , and show 2d slices of the von mises stress @xmath89 fields for four different @xmath67 , see fig .
[ fig10_vonmises_distribution_confinement ] ( a - d ) .
consistent with the pressure distribution , the correlation between @xmath89 and grain boundary arrangement is negligible .
as shown in fig .
[ fig10_vonmises_distribution_confinement](e ) , the histogram of @xmath89 shifts to the right and broadens with decreasing @xmath67 , indicating a higher mean value and standard deviation .
these trends become more pronounced when @xmath90 . similar to our treatment of the pressure , we quantify the length scale of @xmath89 fluctuations by calculating its correlation function , @xmath91 .
we plot the radial component of the correlation @xmath92 in fig .
[ fig10_vonmises_distribution_confinement](f ) .
consistent with the pressure correlation evolution , @xmath92 also decays more rapidly with decreasing @xmath67 .
as with the pressure , this reduction in correlation length also tracks the decrease in grain size as the crystal is squeezed .
[ htp ] the pronounced change in the @xmath89 suggests a link between the local structure order and @xmath89 .
to illustrate this relationship , we plot the mean deviatoric stress @xmath93 against the ordered neighbor number averaged over sample @xmath94 in fig .
[ fig11_vonmises_order_confinement ] .
the straight dashed line fits to the data showing a clear correlation between @xmath93 and @xmath94 .
more importantly , the mean deviatoric stress approximately doubles while @xmath94 decreases only by @xmath95 . in atomic systems
higher @xmath93 indicates the system is closer to yielding suggesting a lower shear stress would be required to induce plastic deformation of the crystal .
our studies , which combine both structure and stress measurement at the single particle scale , clearly illustrate that confinement can have profound consequences for stress distributions in hard sphere crystals .
while we find that the distribution of the pressure and the von mises stress are weakly correlated with the grain boundary location , their fluctuations become more localized due to the additional grain boundaries introduced by our confining procedure .
thus , the grain boundaries effectively shield each grain from the specific spatial distribution of stresses in neighboring grains .
in addition , we find that the deviatoric stress has a much wider distribution of magnitudes and a mean value that nearly doubles as the gap is reduced from @xmath96 to @xmath97 . in our system , this reduction in gap produces smaller grain sizes .
such a grain size reduction has been suggested to have a great influence on the microstructure@xcite , defect density@xcite , and particle diffusivity @xcite in polycrystals .
more importantly , in atomic nanocrystals , grain size reduction leads to substantially lower yield stress a phenomena known as the inverse hall - petch relation @xcite .
such trends are consistent with our observed enhancement in the deviatoric stress
typically a measure of how close a system is to yielding .
further experiments in which the normal force is continuously measured or experiments in which the yield stress under shear is determined for crystals under different degrees of confinement would shed light on whether a direct link can be made between our colloidal system and films comprised of nanocrystalline grains . such studies would also help determine whether different yielding mechanisms dominate when the grain size is reduced .
for example , it has been shown in numerical simulations that in large grains dislocations penetrate the grains and entangle resulting in strain hardening . in contrast , for small grains , plastic flow induces stacking faults and twining that localize near grain boundaries , and do not contribute significantly to the flow stress .
experiments in which we can simultaneously plastically deform the crystals while measuring their order and stress evolutions would elucidate whether similar mechanisms are at play in these hard sphere colloidal crystals . more broadly , combining the bulk stress measurement and salsa provides a direct way to quantify the interplay between microscopic defect structures and macroscopic mechanical properties .
this approach opens the door to uncovering the mechanisms that underly many defect - dominated phenomena in solid mechanics including defect - assisted premelting , strain hardening , and material fatigue .
to grow a bulk crystal , we set the solvent density value ( @xmath98 g/@xmath5 ) slightly lower than the particle density ( @xmath99 g/@xmath5 ) , so the particles can sediment and form a crystal with a higher volume fraction .
we load a suspension with a volume fraction @xmath100 in a sample cell and hermetically seal it . prior to the experiment ,
the sample is placed on the microscope stage for at least 24 hours , until the sedimentation is complete .
the resulting crystal has a total thickness of approximately 280 @xmath1 m and volume fraction @xmath101 0.67 @xmath11 0.03 . during the crystal growth , the p@xmath102clect number for sedimentation is @xmath103 , where @xmath104 g/@xmath5 refers to the particle - solvent density mismatch , @xmath105 the gravitational acceleration , and @xmath30 the thermal energy .
the small p@xmath102clect number , @xmath106 , indicates that the particle sedimentation rate is much slower than its self - diffusion . as a result ,
the pmma sample can form polycrystals with far larger grain size and lower defect density than silica systems @xcite , where the particle density is significantly mismatched .
our confined crystal samples are all density - matched , and we did not observe any significant particle sedimentation during the course of the experiment .
the confined suspensions have a volume fraction @xmath107 , slightly lower than the bulk crystal value @xmath101 0.67 , and have a lower viscosity , allowing it to be loaded in a confocal rheoscope @xcite . at an imaging rate 60 frames per second ,
the acquisition time of one confocal image stack that consists of 400 slices is @xmath56 6.7 s. this time scale is comparable to a particle s relaxation time @xmath108 s where @xmath109 mpa.s is the solvent viscosity .
the final salsa stress field is averaged over 15 image stacks requiring @xmath110 s to collect .
we observe short - time stress fluctuations arising from particle random motions within their local environment over the image acquisition time .
our data is focused on these relatively short time scales rather than the stresses associated with the long time annealing of crystal grains @xcite or the glassy behavior near grain boundaries @xcite .
[ htp ] to visualize the orientations and types of crystals , we perform a bond - angle analysis distinguishing the fcc and hcp crystals @xcite .
we highlight the color of fcc crystals ( green ) and tone down the color of hcp crystals ( red ) in fig .
[ afig_1_stacking_faults](a ) . in fig .
[ afig_1_stacking_faults](b ) we reverse the color contrast .
we find the fractions of fcc ( @xmath111 ) and hcp ( @xmath112 ) crystals comparable .
this finding is consistent with previous computational studies , in which the fcc configuration is shown to be entropically favored over hcp , but only by @xmath113 per particle @xcite .
furthermore , the observed fcc fractions @xmath114 is also consistent with the values found in previous scattering @xcite and direct imaging @xcite experiments , which show that colloidal polycrystals are essentially comprised of randomly - stacked hexagonal layers with @xmath115 .
we also find that the horizontal fcc and hcp stacks ( with a crystal orientation 111 parallel to @xmath10-axis ) do not alternate consistently across the field of view , as indicated by the arrows in fig .
[ afig_1_stacking_faults](b ) .
this stacking inconsistency leads to stacking faults that can be identified by the vertical interfaces between the two crystals .
the high density of these planar defects in our nearly - equilibrium crystals arises from the small free energy cost associated with the fault formation . by observing the crystal orientation
, we also find that most of the crystal domains have a ( 111 ) plane parallel to the coverslip .
this parallel alignment is consistent with the previously reported mechanism of crystal growth from sedimentation @xcite . in this case , the first few crystal layers form simultaneously near the flat bottom plate , while further crystals grow layer by layer .
we find excellent agreement between our measured pressure value ( horizontal line in fig.[afig2_bulk_pressure](a ) ) and the value predicted by simulations ( orange line ) of hard spheres at the volume fraction found ( @xmath116 0.67 , vertical line ) in our system . to investigate how this pressure value depends on the salsa shell thickness @xmath34 , we plot @xmath117 versus @xmath34 ( black joined points ) in fig .
[ afig2_bulk_pressure ] ( b ) .
the constant @xmath117 highlighted by the green shade and horizontal red line indicates a pressure value insensitive to shell thickness between 50 nm @xmath118 180 nm . the overestimated pressure at @xmath119 50 nm arises from polydispersity and the particle overlap associated with featuring uncertainties .
the underestimated pressure at @xmath120 180 nm arises from the saturation in the collision probability once all the nearest neighbors are included .
the observed @xmath121independent pressure confirms that both our imaging and particle featuring resolutions are adequate to quantify the particle collision probability and its resulting stress . throughout all salsa analyses in this work
, we set the shell thickness @xmath122 nm .
the particle polydispersity has different influences on the normal and shear stress measurements .
a detailed discussion of this issue can be found in our previous work ( supplementary information of @xcite ) , where the shear and normal stresses near a vacancy defect are measured .
overall , since the pressure measurement relies on a more accurate identification of colliding particles , it can be influenced by the polydispersity more notably .
in contrast to the pressure , the shear component is primarily related to the angular anisotropy of the neighboring particle configuration ; hence , it is affected by the polydispersity less . in the current work , the particle polydispersity ( @xmath1233%@xmath124 nm comparable to the particle featuring error ) is readily smaller than the shell thickness @xmath125 nm that defines the colliding criterion .
therefore , we anticipate that the identification of colliding particle is primarily associated with @xmath34 rather than polydispersity .
finally , since we grow our bulk polycrystal sample from particle sedimentation , we anticipate the crystal pressure depends on the sample thickness and the mismatched density between particle and solvent .
furthermore , we anticipate that the pressure should not significantly vary with the height @xmath10 , given that the depth of view @xmath6 68 @xmath1 m is much thinner than the sample thickness @xmath126 280 @xmath1 m .
we plot the pressure averaged over the @xmath127 plane as a function of height @xmath10 in fig .
[ afig2_bulk_pressure](c ) .
overall , the pressure trend is consistent with the estimated hydrostatic pressure arising from gravity , @xmath128 ( blue dashed line ) , and we do not observe a clear decay in pressure .
the pressure variation is mainly dominated by the fluctuations from the defect distribution in the sample . in fig .
[ afig2_bulk_pressure](d ) we show the histogram of the pressure fluctuation . the histogram can be described well by a gaussian distribution ( orange line ) with a standard deviation @xmath129 mpa corresponding to 15@xmath130 of the mean .
this magnitude of pressure fluctuation is consistent with previous results found in silica systems @xcite .
we calculate the three principal stresses @xmath57 , @xmath58 , and @xmath59 and show them in fig .
[ afig_3_principal_stress ] .
the three principal stresses are the eigenvalues of the stress tensor , and are thus independent of the coordinate orientation .
the common trend between the three principal stresses corresponds to the hydrostatic pressure variation , while the difference is related to the deviatoric stress @xmath61 .
we calculate @xmath61 and visualize its distribution within the sample by excluding particles with @xmath61 < 22 mpa . the thresholded field is shown in fig . [ afig_4_threshold_vonmises ] .
we find no obvious correlation between the distribution of the particles with large @xmath61 and the grain boundaries within the polycrystal .
the authors thank marc miskin , lena bartell , brian leahy , meera ramaswamy , eric schwen , and matthew bierbaum for helpful conversations .
this work was supported by nsf dmr - cmp award no .
1507607 . c. alba - simionesco , b. coasne , g. dosseh , g. dudziak , k. gubbins , r. radhakrishnan and m. sliwinska - bartkowiak , _ journal of physics : condensed matter _ , 2006 , * 18 * , r15 a. s. dimitrov and k. nagayama , _ langmuir _ , 1996 , * 12 * , 13031311 p. jiang , j. bertone , k. hwang and v. colvin , _ chemistry of materials _ , 1999 , * 11 * , 21322140 d. t. wasan and a. d. nikolov , _ nature _ , 2003 , * 423 * , 156159 m. abkarian , j. nunes and h. a. stone , _ journal of the american chemical society _ , 2004 , * 126 * , 59785979 w. diyatmika , j. p. chu , b. t. kacha , c .- c .
yu and c .- m .
lee , _ current opinion in solid state and materials science _ , 2015 ,
* 19 * , 95106 h .- c .
tsai and d. bogy , _ journal of vacuum science & technology a _ , 1987 , * 5 * , 32873312 s .- h .
k. park , m. ryu , c .- s .
hwang , s. yang , c. byun , j .-
lee , j. shin , s. m. yoon , h. y. chu , k. i. cho _
et al . _ , sid symposium digest of technical papers , 2008 , pp .
629632 k. akimoto , s. ishizuka , m. yanagita , y. nawa , g. k. paul and t. sakurai , _ solar energy _ , 2006 , * 80 * , 715722 r. abermann and r. koch , _ thin solid films _ , 1985 , * 129 * , 7178 j. a. thornton and d. hoffman , _ thin solid films _ , 1989 , * 171 * , 531 l. b. freund and s. suresh , _ thin film materials : stress , defect formation and surface evolution _ , cambridge university press , 2004 d. j. harris , h. hu , j. c. conrad and j. a. lewis , _ physical review letters _ , 2007 , * 98 * , 148301 d. j. harris , j. c. conrad and j. a. lewis , _ philosophical transactions of the royal society of london a : mathematical , physical and engineering sciences _ , 2009 , * 367 * , 51575165 k. b. singh and m. s. tirumkudulu , _ physical review letters _ , 2007 , * 98 * , 218302 l. t. shereda , r. g. larson and m. j. solomon , _ physical review letters _ , 2008 , * 101 * , 038301 y. xu , g. k. german , a. f. mertz and e. r. dufresne , _ soft matter _ , 2013 ,
* 9 * , 37353740 r. koch , _ journal of physics : condensed matter _ , 1994 , * 6 * , 9519 d. derks , y. l. wu , a. van blaaderen and a. imhof , _ soft matter _ , 2009 , * 5 * , 10601065 a. van blaaderen , a. imhof , w. hage and a. vrij , _ langmuir _ , 1992 , * 8 * , 15141517 v. prasad , d. semwogerere and e. r. weeks , _ journal of physics : condensed matter _ , 2007 ,
* 19 * , 113102 a. van blaaderen , r. ruel , p. wiltzius _ et al .
_ , _ nature _ , 1997 , * 385 * , 321324 n. a. verhaegh , j. s. van duijneveldt , a. van blaaderen and h. n. lekkerkerker , _ the journal of chemical physics _ , 1995 , * 102 * , 14161421 j. r. royer , g. l. burton , d. l. blair and s. d. hudson , _ soft matter _ , 2015 , * 11 * , 56565665 j. c. conrad , p. p. dhillon , e. r. weeks , d. r. reichman and d. a. weitz , _ physical review letters _
, 2006 , * 97 * , 265701 e. kumacheva , r. k. golding , m. allard , e. h. sargent _
_ , _ advanced materials _ , 2002 , * 14 * , 221 a. fortini and m. dijkstra , _ journal of physics : condensed matter _ , 2006 , * 18 * , l371 k. h. nagamanasa , s. gokhale , r. ganapathy and a. sood , _ proceedings of the national academy of sciences _ , 2011 , * 108 * , 1132311326 p. pieranski , l. strzelecki and b. pansu , _ physical review letters _ , 1983 , * 50 * , 900 i. cohen , t. g. mason and d. a. weitz , _ physical review letters _ , 2004 , * 93 * , 046001 s. neser , c. bechinger , p. leiderer and t. palberg , _ physical review letters _ , 1997 , * 79 * , 2348 y. peng , z. wang , a. m. alsayed , a. g. yodh and y. han , _ physical review letters _ , 2010 , * 104 * , 205703 n. y. lin , m. bierbaum , p. schall , j. p. sethna and i. cohen , _ nature materials _ ,
2016 c. p. royall , m. e. leunissen , a .- p .
hynninen , m. dijkstra and a. van blaaderen , _ the journal of chemical physics _ , 2006 , * 124 * , 244706 a. yethiraj and a. van blaaderen , _ nature _ , 2003 , * 421 * , 513517 c. p. royall , w. c. poon and e. r. weeks , _ soft matter _ , 2013 , * 9 * , 1727 n. y. lin , s. goyal , x. cheng , r. n. zia , f. a. escobedo and i. cohen , _ physical review e _ , 2013 ,
* 88 * , 062309 n. y. lin , x. cheng and i. cohen , _ soft matter _ , 2014 ,
* 10 * , 19691976 n. y. lin , j. h. mccoy , x. cheng , b. leahy , j. n. israelachvili and i. cohen , _ review of scientific instruments _ , 2014 , * 85 * , 033905 j. c. crocker and d. g. grier , _ journal of colloid and interface science _ , 1996 , * 179 * , 298310 a. d. dinsmore , e. r. weeks , v. prasad , a. c. levitt and d. a. weitz , _ applied optics _ , 2001 ,
* 40 * , 41524159 r. besseling , e. r. weeks , a. schofield and w. poon , _ physical review letters _
, 2007 , * 99 * , 028301 i. ramsteiner , k. e. jensen , d. a. weitz and f. spaepen , _ physical review e _ , 2009 ,
* 79 * , 011403 j. hernndez - guzmn and e. r. weeks , _ proceedings of the national academy of sciences _ , 2009 , * 106 * , 1519815202 p. r.
ten wolde , m. j. ruiz - montero and d. frenkel , _ the journal of chemical physics _ , 1996 , * 104 * , 99329947 p. j. steinhardt , d. r. nelson and m. ronchetti , _ physical review b _ , 1983 , * 28 * , 784 u. gasser , e. r. weeks , a. schofield , p. pusey and d. weitz , _ science _ , 2001 , * 292 * , 258262 r. p.
dullens , d. g. aarts and w. k. kegel , _ physical review letters _ , 2006 , * 97 * , 228301 d. bi , j. zhang , b. chakraborty and r. behringer , _ nature _ , 2011 , * 480 * , 355358 e. b. tadmor and r. e. miller , _ modeling materials : continuum , atomistic and multiscale techniques _ , cambridge university press , 2011 s. pronk and d. frenkel , _ physical review letters _ , 2003 , * 90 * , 255501 c. a. schuh , t. c. hufnagel and u. ramamurty , _ acta materialia _ , 2007 , * 55 * , 40674109 c. a. schuh and a. c. lund , _ nature materials _ , 2003 , * 2 * , 449452 m. schmidt and h. lwen , _ physical review letters _ , 1996 , * 76 * , 4552 y. han , y. shokef , a. m. alsayed , p. yunker , t. c. lubensky and a. g. yodh , _ nature _ , 2008 , * 456 * , 898903 b. de nijs , s. dussi , f. smallenburg , j. d. meeldijk , d. j. groenendijk , l. filion , a. imhof , a. van blaaderen and m. dijkstra , _ nature materials _ , 2015 , * 14 * , 5660 n. v. dziomkina and g. j. vancso , _ soft matter _ , 2005 ,
* 1 * , 265279 d. bi , s. henkes , k. e. daniels and b. chakraborty , _ annu . rev .
matter phys .
_ , 2015 , * 6 * , 6383 e. i. corwin , h. m. jaeger and s. r. nagel , _ nature _ , 2005 , * 435 * , 10751078 b. p. tighe , a. r. van eerd and t. j. vlugt , _ physical review letters _ , 2008 , * 100 * , 238001 b. p. tighe , j. h. snoeijer , t. j. vlugt and m. van hecke , _ soft matter _ , 2010 , * 6 * , 29082917 v. de villeneuve , p. miedema , j. meijer and a. petukhov , _ epl ( europhysics letters ) _ , 2007 , * 79 * , 56001 m. rose , a. balogh and h. hahn , _ nuclear instruments and methods in physics research section b : beam interactions with materials and atoms _ , 1997 ,
* 127 * , 119122 b. wu , a. heidelberg and j. j. boland , _ nature materials _ , 2005 , * 4 * , 525529 l. harrison , _ transactions of the faraday society _ , 1961 , * 57 * , 11911199 j. schitz and k. w. jacobsen , _ science _
, 2003 , * 301 * , 13571359 j. schitz , _ scripta materialia _ , 2004 ,
* 51 * , 837841 a. stukowski , _ modelling and simulation in materials science and engineering _ , 2009 , * 18 * , 015012 g. ackland and a. jones , _ physical review b _ , 2006 , * 73 * , 054104 t. solomon and m. j. solomon , _ the journal of chemical physics _ , 2006 , * 124 * , 134905 s .- c .
mau and d. a. huse , _ physical review e _ , 1999 , * 59 * , 4396 d. frenkel and a. j. ladd , _ the journal of chemical physics _ , 1984 , * 81 * , 31883193 j. s. van duijneveldt , j. k. dhont and h. n. lekkerkerker , _ the journal of chemical physics _ , 1993 , * 99 * , 69416949 p. pusey , w. van megen , p. bartlett , b. ackerson , j. rarity and s. underwood , _ physical review letters _ , 1989 , * 63 * , 2753 m. elliot , b. bristol and w. poon , _ physica a : statistical mechanics and its applications _ , 1997 , * 235 * , 216223 | the mechanical properties of crystalline materials can be substantially modified under confinement .
such modified macroscopic properties are usually governed by the altered microstructures and internal stress fields . here
, we use a parallel plate geometry to apply a quasi - static squeeze flow crushing a colloidal polycrystal while simultaneously imaging it with confocal microscopy .
the confocal images are used to quantify the local structure order and , in conjunction with stress assessment from local structural anisotropy ( salsa ) , determine the stress at the single - particle scale .
we find that during compression , the crystalline regions break into small domains with different geometric packing .
these domains are characterized by a pressure and deviatoric stress that are highly localized with correlation lengths that are half those found in bulk .
furthermore , the mean deviatoric stress almost doubles , suggesting a higher brittleness in the highly - confined samples . |
convective - reactive h - combustion events are encountered when the damkhler number @xmath8 , where @xmath9 is the reaction time scale for the @xmath10 reaction and @xmath11 is the convective advection time scale @xcite . in such events
h is ingested into he - shell flash convection , and this is encountered in many instances of stellar evolution , especially at very low or zero metallicity .
examples have been found in 1-d stellar evolution simulations and include the he - core and he - shell flashes with h - ingestion events in low - mass stars with [ fe / h]@xmath12 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and in massive stars @xcite . during h - ingestion events , nuclear energy release on the convective turn - over time scale
is coupled with multi - scale turbulent mixing , and therefore results from 1-d simulations are unreliable @xcite . @xcite
presented simulations of a h - burning convective shell induced by an artificially inward shifted h - profile on top of the he - core flash convection zone . however , in their h - burning convection region the nuclear time scale of the @xmath13 reaction can be estimated from the available @xmath14 and @xmath15 information to be @xmath16 , which implies @xmath17 .
therefore , their h - burning convection zone is not convective - reactive , and therefore not a h - ingestion event in terms of our definition .
there is a small amount of h penetrating through the interface between the two convection zones , and it does not have dynamical relevance ( just as in our simulations in the first few hours , see below ) .
@xcite present h - ingestion agb shell flash convection simulations for a low - metallicity case .
they report a h - burning luminosity difference of orders of magnitude between simulations with two different grid resolutions , and an almost complete transport of protons through the entire convection zone all the way to the bottom , where the highest temperatures ( @xmath18 ) are reached and the protons are reacting with @xmath0 .
contrary to one - dimensional stellar evolution calculations , which place the location of h - burning in the he - shell flash convection zone at @xmath19 , those 3-d hydrodynamic simulations do not show any apparent hydrodynamic feedback from the energy release from h - burning .
( fig.2 * ? ? ?
shown are pressure @xmath20 and density @xmath15 in code units ( @xmath21=10^{19 } \mathrm{g\,cm^{-1}\ , s^{-2}}$ ] ) , @xmath22 = 10 ^ 3 \mathrm{g\,cm^{-3}}$ ] ) , and temperature ( @xmath23=10 ^ 8{\ensuremath{\,\mathrm k}}$ ] ) .
vertical dotted lines indicate the location of the convection boundary .
bottom : spherically averaged rms velocity averaged over the top region of the convection zone from @xmath24 to @xmath25 .
the signature of a global oscillation is evident in these averaged velocities.,title="fig:",scaledwidth=50.0% ] ( fig.2 * ? ? ?
shown are pressure @xmath20 and density @xmath15 in code units ( @xmath21=10^{19 } \mathrm{g\,cm^{-1}\ , s^{-2}}$ ] ) , @xmath22 = 10 ^ 3 \mathrm{g\,cm^{-3}}$ ] ) , and temperature ( @xmath23=10 ^ 8{\ensuremath{\,\mathrm k}}$ ] ) .
vertical dotted lines indicate the location of the convection boundary .
bottom : spherically averaged rms velocity averaged over the top region of the convection zone from @xmath24 to @xmath25 .
the signature of a global oscillation is evident in these averaged velocities.,title="fig:",scaledwidth=50.0% ] ( lr ) and the @xmath3 ( hr ) runs .
@xmath26 panel : abundance profiles of h - rich material .
bottom panel : entropy - like quantity @xmath27 , with @xmath20 pressure and @xmath28 for the monatomic ideal gas .
the bottom two panels have the same line styles .
vertical dotted lines indicate the convection boundary in the initial setup.,title="fig:",scaledwidth=50.0% ] ( lr ) and the @xmath3 ( hr ) runs .
@xmath26 panel : abundance profiles of h - rich material .
bottom panel : entropy - like quantity @xmath27 , with @xmath20 pressure and @xmath28 for the monatomic ideal gas .
the bottom two panels have the same line styles .
vertical dotted lines indicate the convection boundary in the initial setup.,title="fig:",scaledwidth=50.0% ] ( lr ) and the @xmath3 ( hr ) runs .
@xmath26 panel : abundance profiles of h - rich material .
bottom panel : entropy - like quantity @xmath27 , with @xmath20 pressure and @xmath28 for the monatomic ideal gas .
the bottom two panels have the same line styles .
vertical dotted lines indicate the convection boundary in the initial setup.,title="fig:",scaledwidth=50.0% ] ( lr ) and the @xmath3 ( hr ) runs .
@xmath26 panel : abundance profiles of h - rich material .
bottom panel : entropy - like quantity @xmath27 , with @xmath20 pressure and @xmath28 for the monatomic ideal gas .
the bottom two panels have the same line styles .
vertical dotted lines indicate the convection boundary in the initial setup.,title="fig:",scaledwidth=50.0% ] these examples demonstrate the challenge that these phases of stellar evolution present to simulation efforts
. it would be very helpful to check the simulation approach for a case for which meaningful validation data is available to check the results .
therefore , before turning our attention to the interesting problem of h - ingestion events in metal - poor stars , we perform for the first time 3-d hydrodynamic simulations of the h - ingestion flash in sakurai s object ( v4334 sgr ) , a post - agb star that experienced a very late thermal pulse ( vltp ) .
real - time observations of the light curve ( e.g. * ? ? ?
* ; * ? ? ?
* ) as well as post - flash abundance observations @xcite provide constraints on the details of the he - shell flash with h - ingestion in this case .
@xcite showed that the @xmath29 times faster rise time of sakurai s object compared to standard mixing - length theory ( mlt ) stellar evolution predictions can be accounted for if it is assumed that in h - ingestion phases the convective mixing efficiency is reduced by a factor of @xmath30 to @xmath31 , which leads to a h - combustion energy release closer to the stellar surface , where the time scale for thermal response is shorter .
@xcite showed that an early mixing split between the he - shell flash convection zone ( reduced in size ) and the newly emerging h - burning convection zone prevents standard stellar evolution models from reproducing the observed abundance features , especially the pronounced overabundance of first - peak neutron capture elements .
however , an assumed mixing permeability of the interface between the two convection zones lasting for @xmath32 after the h - ingestion event started does enable one to account for most observed abundance features . in both cases
1-d stellar evolution assumptions fail to reproduce key observables . in sect.[sec : method ] we describe the method and how the simulations were performed , while sect.[sec : results ] presents our main findings and discussion of the results .
we use the 3-d gas dynamics code and the initial setup method described in detail in @xcite .
the code features the piecewise - parabolic - boltzmann ( ppb ) moment - conserving advection scheme ( see * ? ? ?
* ) , which provides an effective resolving power of the fluid distribution that requires two to three times the grid resolution to be matched by a code based on the piecewise - parabolic method ( ppm ) without ppb .
as in previous investigations ( e.g. * ? ? ?
* ; * ? ? ?
* ) we adopt a monatomic ideal gas equation of state which represents the conditions in he - shell flash convection in the post - agb he - shell flash model for sakurai s object well ( fig.[fig : initial_star ] ) .
the nuclear burn module considers just the @xmath10 reaction @xcite with an energy release of @xmath33 .
the two - fluid setup consists of a convectively unstable shell and a stable layer below with initially the @xmath0-rich fluid conv , and a stable layer above the convection zone with the h - rich fluid h+he. both of these composite fluids have realistic mean molecular weights reflecting the abundance mix for intershell and envelope material according to the corresponding stellar evolution model .
the nuclear burn module takes appropriately into account that in each of the fluids only a certain fraction participates in the nuclear reaction .
we ignore for now the energy release of @xmath34 from the subsequent @xmath35-decay of @xmath36 with a half life of @xmath37 .
using the reaction rate length scale introduced by @xcite as a 1-d estimate of the geometric scale of the h - burning layer @xmath36 would decay after being homogenized throughout a layer approximately 20 times the thickness of the burn layer .
here we are interested in the emergence and initial evolution of any hydrodynamic response to h - ingestion and burning which we assume to depend on the localized energy release from the @xmath38 reaction . the full @xmath1 shell of the convectively unstable layer is included in the simulation , which ensures that any global or large - scale motions can be captured .
the initial setup has been constructed by using the model at @xmath39 from @xcite , which was taken from the @xmath40 post - agb stellar evolution track presented first in @xcite and further modified in @xcite .
the convection is driven by a constant heating of @xmath41 corresponding to the he - shell flash luminosity of the @xmath39 stellar evolution model .
the initial stratification ( fig.[fig : initial_star ] ) is constructed from piecewise polytropic layers that closely resemble the stellar evolution model .
a reflective spherical outer boundary condition is located at @xmath42 . for our simulation approach
we have investigated in detail the numerical resolution requirements @xcite .
in addition we have performed numerous test simulations for the sakurai s star setup , on the canadian westgrid computer orcinus with grid sizes up to @xmath2 while simulations with @xmath2 , @xmath43 and @xmath3 grids where performed on nsf s blue waters sustained petaflops computing system at ncsa .
the ppmstar code @xcite , which was originally designed to run on the los alamos roadrunner machine ( cf . * ? ? ?
* ) , performed our run at @xmath3 resolution within a single four - day period on blue waters .
it ran on @xmath44 cpu cores , roughly half the blue waters machine , at a sustained rate of @xmath45 with 64-bit precision . at this computing rate , it took about @xmath46 of real time to simulate @xmath47 of time for the star and to write to disk @xmath48 of visualization data . with @xmath49 data dumps , @xmath50 of data from this run was written , which was used to generate the movies at http://www.lcse.umn.edu/movies .
the movies were generated using our hvr volume rendering utility on the gpu nodes on blue waters .
this computational performance was possible , because our code scales extremely well on blue waters .
each node with @xmath51 cpu cores ran eight mpi processes , each with four threads updating a tiny grid brick of just @xmath52 cells .
we present the results from two runs with a @xmath2 a @xmath3 grid for a minimum duration of @xmath53 star time .
the simulations start in the same way as the entrainment simulations presented by @xcite .
an initial transient phase of less than @xmath54 quickly gives way to fully developed he - shell flash convection with dominant global modes that cause hydrodynamic shear instabilities at the top convection boundary .
downdrafts of h - rich material from above originate where large - scale upwellings and consequently coherent horizontal motions along the convective boundary converge .
h - rich clouds descend into the deeper layers and represent a large - scale inhomogeneous distribution in the region where protons start to react with @xmath0 . during this initial phase of @xmath55
the spherically and radially averaged velocities in both runs agree reasonably well ( fig.[fig : initial_star ] , bottom panel ) .
the overall entropy in the convection zone rises due to the addition of energy from nuclear reactions , and mixing throughout the convection zone is efficient enough to maintain a flat spherically averaged entropy profile .
the h - accumulation rate ( entrainment minus burning ) is @xmath56 and @xmath57 for the @xmath2 and @xmath3 runs during this initial quiet phase , about a factor four smaller than the converged entrainment rate of @xmath58 found by @xcite for slightly different runs without h burning . as discussed there , this entrainment rate is too small to cause hydrodynamic feedback from the burning of entrained h if the entrainment rate is equated with the h burn rate .
it is therefore consistent that we do not observe noticeable differences in the early quiescent phase between these simulations with h burning and the entrainment simulations of @xcite without h burning .
the quiescent phase lasts for around @xmath59 . in test simulations with a sponging boundary condition
the motions in the upper stable layer are much more pronounced in this setup .
these and other details of the simulations can alter the duration of the quiescent phase significantly .
however , none of our tests indicate that nature of the violent events to follow depend on the details of the quiescent lead - up phase .
the convection hydrodynamics changes dramatically around @xmath59 ( fig.[fig : initial_star ] ) when a major global fluid flow oscillation is about to be launched .
the overall shape of the rise and decay of the oscillation mode agrees well between the two runs , the major difference being that in the lower resolution run the event starts @xmath60 earlier .
the visualization of the 3-d flow for time @xmath61 shows the moment when the flow has already started to respond strongly to the energy input from h burning ( fig.[fig : fv_vort ] ) .
the fractional volume images from the two runs appear to differ more than at the later time ( @xmath62 ) .
however , this is mostly because we see in the @xmath2-run image a later time in the event .
overall the agreement between the two runs is good . during the lead - up to this transition at @xmath63 eventually a large enough reservoir of h
is accumulated in the upper part of the convection zone ( fig.[fig : profiles ] ) so that pockets of h - rich material advected into the burn region carries an increasing fraction of fuel . at that point
individual , particularly strong ignition events launch very pronounced , identifiable upwellings .
as these hit the stiff convective boundary above they are deflected in all directions and create enhanced levels of entrainment .
the kelvin - helmholtz instability vorticity - entrainment trains can be clearly identified in the vorticity image ( top row , fig.[fig : fv_vort ] ) .
the horizontal flows proceed along the convective boundary all the way to the antipode where their collision is forcing the h - rich entrainment layer into a downdraft .
this next fuel - enhanced advection stream causes an even more violent ignition event and subsequent launching of the next fast upwelling , which in turn causes the next circumferential flow back close to the point of origin of the first ignition .
this oscillatory back - and - forth repeats for about a dozen times and is clearly identifiable in the averaged velocity time evolution shown in the bottom panel of fig.[fig : initial_star ] .
we refer to this new phenomenon in stellar physics as the global oscillation of shell h - ingestion ( gosh ) . during the rise time of @xmath64 @xmath65 and @xmath66
are accumulated in the @xmath2 and @xmath3 runs , corresponding to an accumulation rate of @xmath67 and close to the critical entrainment rate for hydrodynamic feedback estimated in @xcite . these highly time - variable and global fluid flows
are better observed in the animated movies available at http://www.lcse.umn.edu/movies .
although the gosh proceeds beyond the initial rise time , its amplitude decreases until convective motions in the upper region of the initially unstable zone are significantly suppressed ( see vorticity image at later time and spherically averaged radial and horizontal velocities shown in fig.[fig : profiles ] ) .
this suppression follows from the entrainment of entropy that goes along with the entrainment of h - rich material from the stable layer above the convection zone ( lower panel fig.[fig : profiles ] ) . in the simulation a new upper boundary of the he - luminosity driven convection forms at @xmath68 that can be seen clearly in both runs for the later time in the fractional volume and the vorticity image .
however , the burning of entrained h - rich material continues to add entropy to the convection zone as well and causes an entropy plateau to develop in the upper region .
the @xmath3 run did not yet get to that point when we stopped it .
but the @xmath69 run with its generally higher entrainment rate shows an upper convection zone is about to form between @xmath70 and @xmath71 .
when we continued this lower - resolution run beyond @xmath72 it developed a further non - radial , violent event in which the velocity amplitude in the upper layer above the initial convection zone increases significantly .
this simulation outcome is not reliable because the outer boundary of the domain needs to be moved further out to simulate this following phase accurately . during
the gosh significant and mostly tangential motions are set in place in the stable layer above the convection zone that die down when the oscillations subside . in terms of the averaged velocities ( as those shown in fig.[fig : profiles ] ) these motions are agreeing fairly well between the two grid simulations up to @xmath62 when the gosh is ending . at later times
larger differences are evident .
this is the phase that we need to explore in more detail with higher resolution grids .
what connections can we make between these preliminary results of the 3-d hydrodynamic simulations and the constraints we have so far from comparing simulations with calibrated 1-d models ?
@xcite reproduced observed n - capture element abundances by assuming a constant entrainment rate of @xmath73 for @xmath74 to @xmath75 before terminating further neutron exposure by assuming mixing between the upper and lower region of the convection became inefficient .
our simulations so far show that compared to the very low entrainment we found in he - shell flash convection without feedback from h burning @xcite , entrainment rates are significantly enhanced by the gosh .
however , the gosh leads as well , at least initially , to a self - quenching of convective motions close to the original upper convection boundary due to the entrainment of entropy .
convection is restarted in this layer by h burning and during one ( or perhaps several ) subsequent violent outburst(s ) the entrainment rates may further increase .
these enhanced entrainment rates will have to be sustained so that the observed 2-dex enhancement of first - peak elements sr , rb , zr , y can be reproduced .
in which way this may be possible will have to be revealed by future simulations .
the present simulations show that the h - ingestion is proceeding through periodic bursts of global - mode oscillations and highly variable entrainment rates , and we will have to investigate if such entrainment histories would fundamentally allow the reproduction of the observed abundance features in sakurai s object . in our present simulations the reflective boundary condition is likely unrealistic when the flow becomes very violent in the initially stable layer above .
the location of h - ingestion energy release is related to the time - scale of the born - again light curve of vltp objects @xcite . during the long quiescent phase of our 3-d simulations h
burning takes place in deeper layers of the he - shell flash convection zone , around @xmath76 .
however , when the feedback from h burning becomes evident burning moves up to @xmath77 to @xmath78 , and here the new convective boundary forms after the gosh .
@xcite has associated the light curve time scale of sakurai s object with the mixing efficiency .
our 3-d setup ( fig.[fig : initial_star ] ) is based on this same @xmath40 1-d model used by @xcite .
the unmodified stellar evolution track shows the h - burning location at @xmath79 ( corresponding to a radius @xmath80 ) with a born - again time of @xmath81 , compared to @xmath82 ( @xmath83 ) for a modified model with reduced mixing efficiency calibrated to reproduce the observed time of @xmath84 . the entropy profile ( fig.[fig : profiles ] ) shows that the h - burning location in the 3-d hydrodynamic simulation indeed agrees very well with the lower - mixing efficiency solution of the calibrated 1-d model , and would lead to a fast born - again time as observed .
this agreement between the location of h - burning in a stellar evolution model calibrated from the light curve , and the location found in the 3-d simulation that has no free parameters provides a first validation of our simulation approach .
our simulations show highly non - radial , global and variable modes with periods of the order of a convective turn - over time .
these global modes of the h - ingestion event can not be captured in 1-d models .
however , 1-d stellar evolution simulations do agree with some aspects of our 3-d simulations , as for example the relatively early emergence of a h - burning convection zone in terms of the entrained mass of h , and the fact that energy - release from h - burning has some influence on the convective properties of the he - burning shell .
this is , for example , opposite to the finding of @xcite that h is mixed all the way to the bottom of the he - shell flash convection zone ( their fig.9 ) , a result that we do not confirm in our 3-d simulations ( fig.[fig : profiles ] ) .
however , the low - z agb thermal pulse has due to lower density a smaller gas pressure fraction ( @xmath85 ) compared to post - agb h - ingestion case where @xmath86 .
this may limit direct comparisons between their and our work .
another difference is that the transition to a h - burning driven convection zone that we report here appears after @xmath87 in our simulation , whereas the @xcite runs stop after @xmath88 .
perhaps those authors did not see a significant hydrodynamic response from h - burning energy release because they stopped the run too early .
however , as we have pointed out here , even with our much longer simulations we can be certain that we have not yet seen the full story of the h - ingestion flash .
finally , we note that our results reported here have to be considered as preliminary .
we have to postpone a more detailed analysis of the results of our simulations until we are able to carry the high resolution run further forward in time .
the computer simulations have been performed in the us on the ncsa bluewaters system under prac grant nsf / oci-0832618 and in canada on the westgrid orcinus computer .
fh acknowledges funding from an nserc discovery grant .
prw acknowledges doe support from contracts with the los alamos and sandia national laboratories , nsf cri grant cns-0708822 , and support from an nsf subcontract from the blue waters project at ncsa . | we performed 3-d simulations of proton - rich material entrainment into @xmath0-rich he - shell flash convection and the subsequent h - ingestion flash that took place in the post - agb star sakurai s object .
observations of the transient nature and anomalous abundance features are available to validate our method and assumptions , with the aim to apply them to very low metallicity stars in the future .
we include nuclear energy feedback from h burning and cover the full @xmath1 geometry of the shell .
runs on @xmath2 and @xmath3 grids agree well with each other and have been followed for @xmath4 and @xmath5 . after a @xmath6 long quiescent entrainment phase the simulations enter into a global non - spherical oscillation that is launched and sustained by individual ignition events of h - rich fluid pockets .
fast circumferential flows collide at the antipode and cause the formation and localized ignition of the next h - overabundant pocket .
the cycle repeats for more than a dozen times while its amplitude decreases . during the global oscillation
the entrainment rate increases temporarily by a factor @xmath7 .
entrained entropy quenches convective motions in the upper layer until the burning of entrained h establishes a separate convection zone .
the lower - resolution run hints at the possibility that another global oscillation , perhaps even more violent will follow .
the location of the h - burning convection zone agrees with a 1-d model in which the mixing efficiency is calibrated to reproduce the light curve .
the simulations have been performed at the nsf blue waters supercomputer at ncsa . |
let @xmath0 , and consider the linear schrdinger equation @xmath1 where @xmath2 is a self - adjoint operator on @xmath3 .
+ by the hille - yoshida theorem , the equation admits a unique solution @xmath4 . under suitable conditions on @xmath2
, this solution enjoys a local gain of regularity ( in the space variable ) : for all @xmath5 there exists @xmath6 so that @xmath7 for some weight @xmath8 and exponent @xmath9 . +
this phenomenon has been discovered by t. kato @xcite in the context of kdv equations .
for the schrdinger equation in the case @xmath10 , it has been proved by p. constantin- j .- c .
saut @xcite , p. sjlin @xcite , l. vega @xcite and k. yajima @xcite .
the variable coefficients case has been obtained by s. do @xcite .
+ the more general results are due to l. robbiano - c .
zuily @xcite for equations with obstacles and potentials .
+ let @xmath2 be a self adjoint operator on @xmath3 .
it can be represented thanks to the spectral measure by @xmath11 in the sequel we moreover assume that @xmath12 . for @xmath13 , we can then define the spectral projector @xmath14 associated to @xmath2 by @xmath15 our main result is a characterisation of the smoothing effect by the decay of the spectral projectors .
denote by @xmath16 .
[ thm1 ] + let @xmath9 and @xmath17
. then the following conditions are equivalent + ( i ) there exists @xmath18 so that for all @xmath19 @xmath20 ( ii ) there exists @xmath21 so that for all @xmath22 and @xmath19 @xmath23 + the interesting point is that we can take the same function @xmath8 and exponent @xmath9 in both statements and .
+ by the works cited in the introduction , in the case @xmath10 on @xmath24 , is known to hold with @xmath25 and @xmath26 , for any @xmath27 .
+ there is also a class of operators @xmath2 on @xmath3 for which is well understood .
let @xmath28 , and assume that for @xmath29 large enough @xmath30 and that for any @xmath31 , there exists @xmath32 so that @xmath33 .
then l. robbiano and c. zuily @xcite show that the smoothing effect holds for the operator @xmath34 , with @xmath35 and @xmath26 , for any @xmath27 .
+ we now turn to the case of dimension @xmath36 , and consider the operator @xmath37 .
we make the following assumption on @xmath38 [ assumption ] we suppose that @xmath28 , and that there exist @xmath39 so that for @xmath29 large enough + ( i ) there exists @xmath40 so that @xmath41 .
+ ( ii ) @xmath42 and @xmath43 ( iii ) $ ] for any @xmath44 , there exists @xmath32 so that @xmath33 .
for instance @xmath45 with @xmath46 satisfies assumption [ assumption ] .
+ it is well known that under assumption [ assumption ] , the operator @xmath2 has a self - adjoint extension on @xmath47 ( still denoted by @xmath2 ) and has eigenfunctions @xmath48 which form an hilbertian basis of @xmath47 and satisfy @xmath49 with @xmath50 , when @xmath51 .
+ for @xmath52 the spectral projector @xmath14 defined in can be written in the following way .
let @xmath53 , then @xmath54 observe that we then have @xmath55 + for such a potential , we can remove the spectral projectors in and deduce from theorem [ thm1 ] [ coro ] + let @xmath9 and @xmath56 .
let @xmath57 so that @xmath58 or @xmath59 satisfies assumption [ assumption ] .
then the following conditions are equivalent + ( i ) there exists @xmath18 so that for all @xmath60 @xmath61 ( ii ) there exists @xmath21 so that for all @xmath62 @xmath63 + the statements and were obtained by k. yajima & g. zhang in @xcite when @xmath8 is the indicator of a compact @xmath64 and with @xmath35 .
+ the statement holds for @xmath26 , by @xcite , but as far as we know , the bound with this @xmath8 was unknown .
+ with theorem [ thm1 ] we are also able to prove the following smoothing effect for the usual laplacian @xmath65 on @xmath66 .
[ prop ] let @xmath67 .
then there exists @xmath6 so that for all @xmath60 @xmath68 from the works cited in the introduction , we have @xmath69 for @xmath70 , for any @xmath27 .
hence proposition [ prop ] shows that we can extend the class of the weights , but we are only able to prove local integrability in time .
notation we use the notation @xmath71 if there exists a universal constant @xmath6 so that @xmath72 .
we define the self adjoint operator @xmath73 $ ] ( entire part of @xmath2 ) by @xmath74 \text{d}e_{\lambda}.\ ] ] notice that we immediately have that @xmath75 is bounded on @xmath3 .
+ the first step in the proof of theorem [ thm1 ] is to show that we can replace @xmath76 by @xmath77 in [ charac ] let @xmath9 and @xmath17
. then the following conditions are equivalent + ( i ) there exists @xmath18 so that for all @xmath19 @xmath78 ( ii ) there exists @xmath21 so that for all @xmath19 @xmath79 we assume and we prove . let @xmath19 and define @xmath80 .
this function solves the problem @xmath81 then by the duhamel formula @xmath82 therefore by and minkowski @xmath83 now use that the operator @xmath84 is bounded , and by we obtain @xmath85 which is . +
the proof of the converse implication is similar .
the proof is based on fourier analysis in time .
this idea comes from @xcite and has also been used in @xcite , but this proof was inspired by @xcite .
+ @xmath86 : to prove this implication , we use the characterisation . from and
the definition of @xmath87 , @xmath88 .
hence it suffices to replace @xmath89 with @xmath90 in and follows .
+ @xmath91 : again we will use lemma [ charac ] . we assume and we first prove that @xmath92 write @xmath93 , then @xmath94 now by parseval in time @xmath95 and by integration in the space variable and @xmath96 which yields .
+ now since the operator @xmath97 is bounded on @xmath98 and commutes with @xmath77 , we have by @xmath99 which is .
let @xmath38 satisfy assumption [ assumption ] .
then by ( * ? ? ?
* lemma 3.3 ) there exists @xmath6 such that @xmath100 for @xmath101 large enough .
this implies that @xmath102 < [ \lambda^{2}_{n+1}]$ ] for @xmath101 large enough , because @xmath103 and @xmath50 . as a consequence @xmath104 and
this yields the result .
+ we now consider @xmath58 . in this case , the eigenvalues are the integers @xmath105 , and the claim follows . with this time
fourier analysis , we can prove the following smoothing estimate for @xmath2 which satisfies assumption [ assumption ] @xmath106 where @xmath107 is defined by @xmath108 \frac1{2k}-\eta \ , \;\text{for any}\ ; \eta>0 \;\ , & \text{if } \quad q=4,\\[6pt ] \frac12-\frac{2}3(1-\frac1q)(1-\frac1k ) \quad & \text{if } \quad 4<q<\infty,\\[6pt ] \frac{4-k}{6k } \quad \quad & \text{if } \quad q=\infty .
\end{array } \right.\ ] ] this was done in @xcite with a slightly different formulation . by theorem [ thm1 ]
, we have to prove that the operator @xmath109 defined by @xmath110 is continuous from @xmath47 to @xmath47 with norm independent of @xmath22 . by the usual @xmath111 argument
, it is enough to show the result for @xmath111 .
+ the kernel of @xmath109 is @xmath112 where @xmath113 with @xmath114 and @xmath115 .
+ the kernel of @xmath111 is given by @xmath116 and by parseval and @xmath117 now , since @xmath118 and @xmath119 , we deduce that @xmath120 ( independent of @xmath22 ) , and @xmath111 is continuous for @xmath67 . | we prove the equivalence between the smoothing effect for a schrdinger operator and the decay of the associate spectral projectors .
we give two applications to the schrdinger operator in dimension one .
on donne une caractrisation de leffet rgularisant pour un oprateur de schrdinger par la dcroissance de ses projecteurs spectraux . on
en dduit deux applications loprateur de schrdinger en dimension un . |
contact problems of elasticity are widely used in many fields of science and engineering , especially in machine science , structural mechanics , geology and biomechanics .
the brief overview of existing numerical and analytical methods for the solution of contact problems can be found in @xcite .
efficient approach to the solution of multibody contact problems is the use of domain decomposition methods ( ddms ) .
ddms are well developed for the solution of linear boundary value problems , particularly for linear poisson and linear elasticity problems @xcite . the construction of ddms for unilateral contact problems , which are nonlinear , are much more complicated .
among domain decomposition methods for unilateral two - body contact problems obtained on continuous level , one should mention signorini
neumann @xcite , signorini dirichlet @xcite and signorini
signorini @xcite iterative algorithms .
all of these methods in each iteration require to solve a nonlinear one - sided contact problem with a rigid surface ( signorini problem ) for one of the bodies , and a linear elasticity problem with neumann @xcite or dirichlet @xcite boundary conditions on possible contact area for other body , or require to solve nonlinear signorini problems for both of the bodies @xcite .
moreover , to increase the convergence rate of signorini dirichlet and signorini signorini algorithms , it is recommended to perform an additional iteration , in which we have to solve linear elasticity problems with neumann boundary conditions for both of the bodies @xcite .
domain decomposition method presented in work @xcite for two - body unilateral contact problem , is also obtained on continuous level .
it is based on the use of augmented lagrangian variational formulation and uzawa block relaxation method .
this domain decomposition method in each iteration require to solve linear elasticity problems for both of the bodies . on contrary
, ddms can be constructed on discrete level , after a discretization of corresponding continuous boundary - value problem . among discrete ddms for unilateral contact problems ,
one should mark out substructuring and feti methods @xcite . in works
@xcite we have proposed on continuous level a class of penalty parallel robin
robin type domain decomposition methods for the solution of unilateral multibody contact problems of elasticity .
these methods are based on penalty method for variational inequalities and some stationary and nonstationary iterative methods for nonlinear variational equations . in each iteration of proposed ddms
we have to solve in parallel some linear variational equations in subdomains , which correspond to elasticity problems with robin boundary conditions , prescribed on some subareas of possible contact zones .
these ddms do not require the solution of nonlinear one - sided contact problems in each step .
the main result of this paper is that we have proved theorems about convergence of proposed penalty robin
robin domain decomposition methods .
the paper is organized as follows . in section 2 the classical formulation of multibody contact problem in the form of the system of second order elliptic partial differential equations with inequality and equality constrains
is given . in section 3
we consider the variational formulation of this problem in the form of convex minimization problem and in the form of elliptic variational inequality at closed convex set , and formulate theorem about a unique existence of the solution of this inequality . in section 4
we use the penalty method to reduce the variational inequality to an unconstrained minimization problem , which is equivalent to a nonlinear variational equation in the whole space .
later , we prove a theorem about a unique existence of the solution of the penalty variational equation and a theorem about strong convergence of this solution to the solution of initial variational inequality . in section 5
we consider stationary and nonstationary iterative methods for the solution of abstract nonlinear variational equations in reflexive banach spaces .
we prove theorems about convergence of these methods , and show that the convergence rate of stationary methods in some energy norm is linear .
we also prove theorem about stability of stationary iterative methods to errors which may occur in each iteration . in section 6
we present parallel stationary and nonstationary penalty robin
robin domain decomposition methods for the solution of nonlinear penalty variational equations of multibody unilateral contact problems .
we prove theorem about convergence of these methods , and show that the convergence rate of stationary robin
robin methods in some energy norm is linear . in section 7
we provide numerical investigations of proposed domain decomposition methods using finite element approximations .
the penalty parameter and mesh refinement influence on the numerical solution , as well as the dependence of the convergence rate of domain decomposition methods on iterative parameters are investigated .
in conclusion section we summarize all results presented in the paper .
introduce the cartesian coordinate system @xmath0 with basis vectors @xmath1 and consider the problem of frictionless unilateral contact of @xmath2 elastic bodies @xmath3 with piecewise smooth boundaries @xmath4 , @xmath5 ( fig . 1 ) .
denote @xmath6 .
the stress - strain state in point @xmath7 of each solid @xmath8 is descried by the displacement vector @xmath9 , the symmetric tensor of strains @xmath10 , and the tensor of stresses @xmath11 .
these quantities satisfy cauchy relations , hook s law and the equilibrium equations : @xmath12 @xmath13 @xmath14 where @xmath15 are the components of the volume forces vector @xmath16 .
the elastic coefficients @xmath17 are measurable , symmetric , and bounded with constants @xmath18 in the following sense @xcite : @xmath19 suppose that the boundary @xmath20 of each solid consists of three parts : @xmath21 , @xmath22 , @xmath23 , such that @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 .
boundary @xmath29 is the possible contact area of body @xmath30 with other bodies , @xmath31 is the possible contact area of body @xmath8 with body @xmath32 , and @xmath33 is the set of the indices of all bodies which are in contact with body @xmath34 . on each boundary @xmath20
let us introduce the local orthonormal coordinate system @xmath35 , where @xmath36 is outer unit normal to @xmath20 .
then the vectors of displacements and stresses on the boundary can be written in the following way : @xmath37
@xmath38 we assume that the surfaces @xmath39 and @xmath40 are sufficiently close @xcite
. therefore @xmath41 , where @xmath42 is the projection of point @xmath43 on the surface @xmath44 .
we denote by @xmath45 the distance in @xmath46 between bodies @xmath8 and @xmath32 before the deformation .
the sign of @xmath47 depends on a statement of the problem . on the part @xmath21 the kinematical ( dirichlet ) boundary conditions
are prescribed : @xmath48 and on the part @xmath49 we consider the static ( neumann ) boundary conditions @xmath50 where @xmath51 and @xmath52 are given boundary displacements and stresses .
further , for the simplicity of variational formulations and proofs , we assume that all bodies are rigidly fixed on the surface @xmath21 , i.e. @xmath53 on the possible contact areas @xmath54 , @xmath55 ,
@xmath56 the following unilateral contact conditions hold : absence of extension @xmath57 absence of friction @xmath58 mutual nonpenetration of the bodies @xmath59 contact alternative @xmath60 where @xmath61 .
the system of second order partial differential equations ( [ eq_pii_1 ] ) ( [ eq_pii_3 ] ) with boundary conditions ( [ eq_pii_5 ] ) ( [ eq_pii_11 ] ) is the mathematical formulation of frictionless unilateral multibody contact problem of elasticity .
note that the contact problem ( [ eq_pii_1 ] ) ( [ eq_pii_3 ] ) , ( [ eq_pii_5 ] ) ( [ eq_pii_11 ] ) is nonlinear , since the real contact areas are unknown .
let us give the weak formulation of the contact problem ( [ eq_pii_1 ] ) ( [ eq_pii_3 ] ) , ( [ eq_pii_5 ] ) ( [ eq_pii_11 ] ) in the form of variational equality and convex minimization problem .
for each body @xmath8 , @xmath5 consider sobolev space @xmath62^{3 } $ ] with scalar product @xmath63 , + @xmath64 and norm @xmath65 .
introduce the following closed subspace in @xmath66 : @xmath67 where @xmath68^{3 } $ ] is surjective , linear and continuous trace operator @xcite .
space @xmath69 is a hilbert space with the same scalar product and norm as in @xmath66 .
consider the space @xmath70 , which is the direct product of spaces @xmath69 : @xmath71 and define in it the scalar product and norm : @xmath72 , @xmath73 , @xmath74 .
note , that hilbert space @xmath75 is closed reflexive banach space .
now , let us introduce the closed convex set of all displacement vectors in @xmath70 which satisfy nonpenetration contact conditions ( [ eq_pii_10 ] ) : @xmath76 where @xmath77 is the set of all possible unordered pairs of the subscripts of bodies which are in contact with each other , and @xmath78 , @xmath79 , @xmath80 , @xmath5 .
the quantities @xmath81 , @xmath5 in ( [ eq_pii_13 ] ) have to be understood in the following way @xmath82 where @xmath83^{3 } $ ] is surjective , linear and continuous trace operator on surface @xmath84 @xcite , and @xmath85^{3 } $ ] .
note , that all equalities and inequalities in spaces @xmath86 , @xmath87 , @xmath88 and @xmath89 hold almost everywhere .
since set @xmath90 is a closed convex subset of hilbert space @xmath70 @xcite , it is weakly closed @xcite . in space
@xmath91 consider bilinear form @xmath92 , which represents the total deformation energy of the system of bodies : @xmath93 @xmath94 in space @xmath70 define linear form @xmath95 , which is equal to the external forces work : @xmath96 @xmath97 where @xmath98^{3 } $ ] , @xmath99^{3 } $ ] , @xmath5 .
* lemma 1 . * _ if @xmath100 _ , @xmath5 _ are piecewise smooth , @xmath101 _ ,
@xmath27 , @xmath102^{3 } $ ] , @xmath103^{3},$ ] _ and condition ( [ eq_pii_4 ] ) holds , then the bilinear form @xmath104 is symmetric , continuous and coercive , and the linear form @xmath105 is continuous , i.e. _ @xmath106 @xmath107 @xmath108 @xmath109 * proof . * the symmetry of bilinear form @xmath110 follows from the symmetry of elastic constants in the hook s law ( [ eq_pii_2 ] ) , and its continuity and coercivity follows from the continuity and coercivity of bilinear forms @xmath111 for each body .
the continuity of @xmath112 follows from the continuity of linear forms @xmath113 , which can be proved using trace theorems @xcite .
the detailed proof of this lemma can be found in @xcite .
@xmath114 according to @xcite , the original contact problem ( [ eq_pii_1 ] ) ( [ eq_pii_3 ] ) , ( [ eq_pii_5 ] ) ( [ eq_pii_11 ] ) has an alternative weak formulation as the convex minimization problem of the quadratic functional on the set @xmath90 : @xmath115 using the general theory of variational inequalities @xcite we have proved in @xcite the next theorem .
* theorem 1 . *
_ suppose that bilinear form @xmath104 and linear form @xmath105 satisfy properties ( [ eq_pii_18 ] ) ( [ eq_pii_21 ] ) , and set @xmath90 is convex closed subset of hilbert space @xmath70 .
then the minimization problem ( [ eq_pii_22 ] ) has a unique solution @xmath116 , and this problem is equivalent to the following variational inequality : _ @xmath117
to obtain a minimization problem in original space @xmath70 , we apply a penalty method @xcite to convex minimization problem ( [ eq_pii_22 ] ) . for the violation of nonpenetration conditions we use the penalty in the following form @xcite : @xmath118^{2}\ , ds,\ ] ] where @xmath119 is a penalty parameter , @xmath120 .
let us consider the following minimization problem with penalty in space @xmath70 : @xmath121 note , that the introduction of penalty corresponds to the introduction of conditional intermediate winkler layer between bodies with stiffness coefficient @xmath122 .
the quantity @xmath123 has a sense of the normal contact stress between bodies @xmath8 and @xmath124 , and the penalty @xmath125 represents the total normal contact stresses work .
now consider the properties of penalty term ( [ eq_pii_24 ] ) in more detail .
functional @xmath125 is nonnegative @xmath126 and gateaux differentiable in @xmath70 : @xmath127 furthermore , the gateaux differential @xmath128 is linear in @xmath129 .
* lemma 2 . * _ if surfaces @xmath31 _ , _ @xmath130 are piecewise smooth , and @xmath131 _ , _ then @xmath128 satisfy the following properties : _ @xmath132 @xmath133 @xmath134 * proof . *
it is obvious to see , that functional @xmath128 is linear in @xmath129 . at first ,
let us show the satisfaction of property ( [ eq_pii_29 ] ) .
let us write @xmath128 in the extended form : @xmath135 , where @xmath136 and @xmath137 .
taking into account the following inequality for real numbers : @xmath138 and schwarz inequality , we obtain @xmath139^{m } } ^{2},\ ] ] where @xmath140^{3 } } ^{2 } + \varepsilon _ { \alpha \beta } $ ] , @xmath141 , @xmath142 , @xmath55 . from trace theorems in @xcite , it follows , that @xmath143^{3 } } ^{2 } \le t_{\alpha \beta } ^{2 } \left\| { \bf v}_{\alpha } \right\| _ { v_{\alpha } } ^{2 } \right\}.\ ] ] substituting ( [ eq_pii_35 ] ) into ( [ eq_pii_34 ] ) , we come to inequality @xmath144 , where @xmath145 , @xmath142 , @xmath5 . similarly to this
, we obtain the same inequality for the second term of relationship ( [ eq_pii_32 ] ) .
hence @xmath146 as a result , we find @xmath147 where @xmath148 . inequality ( [ eq_pii_29 ] ) is proved .
now , we prove that condition ( [ eq_pii_30 ] ) holds . for this
we use the next inequality @xmath149z\le 0\right\},\ ] ] rewrite @xmath150 in the following way : @xmath151 where @xmath152 , @xmath153 in view of property ( [ eq_pii_36 ] ) , we obtain @xmath154 therefore , since @xmath119 , we come to inequality @xmath155 finally , let us prove the satisfaction of property ( [ eq_pii_31 ] ) . using the next inequality for real numbers
@xmath156 we obtain @xmath157 taking into account , that the components of outer unit normal to @xmath158 satisfy property @xmath159 , and using inequalities ( [ eq_pii_33 ] ) and ( [ eq_pii_39 ] ) , we find @xmath160^{3 } } ^{2 } + \left\| { \rm{tr}}_{\beta } ^{0 } ( \ , { \bf w}_{\beta } ) \right\| _ { [ l_{2 } ( s_{\alpha \beta } ) ] ^{3 } } ^{2}\right).\ ] ] now , let us write @xmath161 as follows : @xmath162 consider the term @xmath163 in more detail .
let us use inequalities ( [ eq_pii_33 ] ) , ( [ eq_pii_40 ] ) , and schwarz inequality : @xmath164 @xmath165^{3 } } ^{2 } + \left\| { \rm{tr}}_{\beta } ^{0 } ( \ , { \bf w}_{\beta } ) \right\| _ { [ l_{2 } ( s_{\alpha \beta } ) ] ^{3 } } ^{2 } \right)\ , \
, \left\| { \rm{tr}}_{\alpha } ^{0 } ( \ , { \bf v}_{\alpha } ) \right\| _ { [ l_{2 } ( s_{\alpha \beta } ) ] ^{3 } } ^{2 } .\ ] ] using inequality ( [ eq_pii_35 ] ) , we find further @xmath166 where @xmath167 , @xmath168 . in much the same way we come to inequality @xmath169 where @xmath170 .
taking into account the last two inequalities , we establish @xmath171 @xmath172 where @xmath173 . as a result
, we obtain @xmath174 @xmath175 @xmath176 where @xmath177 , @xmath178 .
@xmath114 + * theorem 2 . * _ suppose that @xmath70 is closed reflexive banach space , @xmath179 is gateaux differentiable , and conditions ( [ eq_pii_18 ] ) ( [ eq_pii_21 ] ) , ( [ eq_pii_26 ] ) , ( [ eq_pii_29 ] ) ( [ eq_pii_31 ] ) hold . then there exist a unique solution of nonquadratic minimization problem ( [ eq_pii_25 ] ) in @xmath180 and this problem is equivalent to the following nonlinear variational equation : _ @xmath181 * proof . * as shown in @xcite , due to the properties ( [ eq_pii_18 ] ) ( [ eq_pii_21 ] ) , the functional @xmath182 is strictly convex , and coercive ( @xmath183 ) , and the differential @xmath184 is linear and continuous in @xmath129 . from the property ( [ eq_pii_30 ] ) , it follows that the penalty term @xmath185 is convex in @xmath70 @xcite .
now consider the properties of functional @xmath186 this functional is gateaux differentiable in @xmath70 : @xmath187 and strictly convex , as the sum of convex functional @xmath125 and strictly convex functional @xmath182 .
in addition , since functionals @xmath188 and @xmath128 are linear and continuous in @xmath129 , it follows that @xmath189 is also linear and continuous in @xmath190 .
hence , according to @xcite , functional @xmath191 is weakly lower semicontinuous . due to the coercivity of @xmath182 and property ( 26 )
, we obtain that @xmath192 . since @xmath70 is closed reflexive banach space
, @xmath191 is weakly lower semicontinuous and coercive , then according to the theorem of existence of the minimum in reflexive banach space @xcite , there exists a solution of minimization problem ( [ eq_pii_25 ] ) in the space @xmath70 .
furthermore , the functional @xmath191 is strictly convex and gateaux differentiable in @xmath70 .
therefore , according to the theorem about necessary and sufficient conditions of the minimum in reflexive banach space @xcite , the solution of problem ( [ eq_pii_25 ] ) is unique , and this problem is equivalent to the variational equation ( [ eq_pii_41 ] ) .
@xmath114 now let us prove that the solution of penalty variational equation ( [ eq_pii_41 ] ) converges strongly to the solution of original variational inequality ( [ eq_pii_23 ] ) as @xmath193 .
let us rewrite variational equation ( [ eq_pii_41 ] ) in the following equivalent form @xmath194 where @xmath195 is the action of a functional @xmath196 on an element @xmath197 , @xmath198 is the space dual to @xmath70 , @xmath199 is gateaux derivative of functional @xmath200 , and @xmath201 .
we have proved the next lemma .
* lemma 3 . * _ suppose that properties ( [ eq_pii_30 ] ) and ( [ eq_pii_31 ] ) hold
. then operator @xmath202 in problem ( [ eq_pii_42 ] ) is a penalty operator for the kinematically allowable displacements set @xmath90 , i.e. _ 1 ) . _
@xmath203 is monotone in @xmath70 : _ @xmath204 2 ) .
_ @xmath203 satisfies the lipschitz condition in @xmath70 : _
@xmath205 3 ) .
_ the kernel of operator @xmath203 is equal to set @xmath90 : _
@xmath206 * proof .
* the monotonity of operator @xmath203 follows from the condition ( [ eq_pii_30 ] ) , and the satisfaction of lipschitz condition follows from the property ( [ eq_pii_31 ] ) . if @xmath116 , then @xmath207 , and , on the contrary , if @xmath208 , we have @xmath116 .
hence , @xmath209 . for more details see @xcite .
@xmath114 now , using the results of works @xcite , let us prove the proposition about the convergence of penalty method , applied to the variational inequality ( [ eq_pii_23 ] ) .
* theorem 3 . * _ suppose that @xmath210 is a convex closed subset of hilbert space @xmath70 _ , _ bilinear form @xmath110 _
, @xmath211 _ and linear form @xmath95 _ , @xmath212 _ satisfy properties ( [ eq_pii_18 ] ) ( [ eq_pii_21 ] ) , @xmath202 is penalty operator for set @xmath90 _ , @xmath213 _ is a unique solution of variational inequality ( [ eq_pii_23 ] ) , and @xmath214 is a unique solution of penalty variational equation ( [ eq_pii_42 ] ) with penalty parameter @xmath215 . then @xmath216 strongly in @xmath70 _ , _
i.e. @xmath217_. * proof . * in works @xcite it is proved that , if conditions ( [ eq_pii_19 ] ) ( [ eq_pii_21 ] ) hold , @xmath218 is a penalty operator for set @xmath90 and there exist solutions of problems ( [ eq_pii_23 ] ) and ( [ eq_pii_42 ] ) , then the sequence @xmath219 is bounded : @xmath220 and there exists such subsequence @xmath221 , which converges weakly in @xmath70 to some solution of variational inequality ( [ eq_pii_23 ] ) , i.e. @xmath222 moreover , in @xcite it is shown that any weakly convergent subsequence of sequence @xmath223 converges weakly in @xmath70 to some solution of variational problem ( [ eq_pii_23 ] ) . now let us assume that variational problems ( [ eq_pii_23 ] ) and ( [ eq_pii_42 ] ) have unique solutions .
then , as follows from above , the sequence @xmath224 has a unique partial weak limit @xmath213 .
since the sequence @xmath219 has a unique weak limit point , and is bounded , then according to theorem in @xcite , it is weakly convergent to this point , i.e. @xmath225 where @xmath213 is the unique solution of variational inequality ( [ eq_pii_23 ] ) .
further , let us show that @xmath219 converges strongly to @xmath213 as @xmath193 . due to ( [ eq_pii_43 ] )
, we get @xmath226 where @xmath227 is the gateaux derivative of functional @xmath228 , @xmath229 .
since @xmath230 is a solution of penalty variational equation ( [ eq_pii_42 ] ) , it is obvious that @xmath231 taking into account the monotonity of penalty operator @xmath232 and the property @xmath233 , we obtain @xmath234 in view of this property and the nonnegativity of bilinear form @xmath104 , we get the inequality @xmath235 .
hence @xmath236 passing to the limit in expression ( [ eq_pii_45 ] ) as @xmath193 , and taking into account property ( [ eq_pii_44 ] ) , we obtain @xmath237 further , in view of coercivity of bilinear form @xmath104 , it follows that @xmath238 as a result , we establish that @xmath239 . @xmath114 thus , using the penalty method we have reduced the solution of original variational inequality ( [ eq_pii_23 ] ) to the solution of the nonlinear variational equation in the whole space @xmath70 , which depends on the penalty parameter @xmath119 .
we also have proved the existence of the unique solution of penalty variational equation ( [ eq_pii_42 ] ) and its strong convergence to the solution of initial variational inequality ( [ eq_pii_23 ] ) as penalty parameter @xmath240 tends to zero .
consider an abstract nonlinear in @xmath241 variational equation in form ( [ eq_pii_41 ] ) , where @xmath70 is closed reflexive banach space , @xmath110 is bilinear form , set in @xmath242 , @xmath243 is linear functional , and the term @xmath128 is linear in @xmath129 and nonlinear in @xmath241 .
suppose that conditions ( [ eq_pii_18 ] ) ( [ eq_pii_21 ] ) , ( [ eq_pii_29 ] ) ( [ eq_pii_31 ] ) are satisfied .
hence , there exists a unique solution of the problem ( [ eq_pii_41 ] ) . for the numerical solution of the nonlinear variational equation ( [ eq_pii_41 ] )
let us use the following iterative method @xcite : @xmath244 , \,\
, k=0,1, ... \,,\ ] ] where @xmath245 is some bilinear form assigned in @xmath91 , @xmath246 , @xmath247 is the @xmath248-th approximation to the exact solution @xmath249 of the problem ( [ eq_pii_41 ] ) , @xmath250 is an initial guess , and @xmath251 is an iterative parameter . we have proved the following proposition about the convergence of the iterative method ( [ eq_pii_46 ] ) .
* theorem 4 . * _ suppose that bilinear form @xmath245 is symmetric , continuous and coercive : _ @xmath252 @xmath253 @xmath254 _ properties ( [ eq_pii_19 ] ) ( [ eq_pii_21 ] ) , ( [ eq_pii_29 ] ) ( [ eq_pii_31 ] ) are satisfied , and the iteration parameter lies in the interval @xmath255 _ , @xmath256 , @xmath257 .
_ _ _ then the sequence @xmath258 _ , _ obtained by iterative method ( [ eq_pii_46 ] ) converges strongly in @xmath70 to the exact solution @xmath249 of the variational equation ( [ eq_pii_41 ] ) , + i.e. @xmath259 _ , _ and the convergence rate in the energy norm + @xmath260 is linear : _
@xmath261 _ moreover , the maximal convergence rate reaches as @xmath262_. _ _ * proof . *
since bilinear form @xmath245 is symmetric , continuous and coercive , we may introduce a scalar product and norm @xmath263 from properties ( [ eq_pii_48 ] ) and ( [ eq_pii_49 ] ) , it follows that norms @xmath264 and @xmath265 are equivalent in space @xmath70 . in each step @xmath266 of method ( [ eq_pii_46 ] )
, we have to solve the linear variational problem : @xmath267 where @xmath268 $ ] is linear in @xmath269 , @xmath270 . using properties ( [ eq_pii_19 ] ) , ( [ eq_pii_21 ] ) , ( [ eq_pii_29 ] ) and ( [ eq_pii_48 ] )
, we obtain that @xmath271 is continuous : @xmath272 where @xmath273 , @xmath274 .
since conditions ( [ eq_pii_47 ] ) ( [ eq_pii_49 ] ) and ( [ eq_pii_52 ] ) are satisfied , we see that the problem ( [ eq_pii_51 ] ) has a unique solution @xmath275 .
now let us show , that the sequence of solutions of problems ( [ eq_pii_51 ] ) converges strongly to the solution of initial variational equation ( [ eq_pii_41 ] ) .
suppose that @xmath276 is the exact solution of problem ( [ eq_pii_41 ] ) .
introduce a notation @xmath277 , @xmath278 , and rewrite ( [ eq_pii_46 ] ) as follows : @xmath279.\ ] ] subtracting from this expression the identity @xmath280 $ ] , we obtain @xmath281.\ ] ] let us define a functional @xmath282 , @xmath283 , which is linear in @xmath269 . due to properties ( [ eq_pii_30 ] ) and ( [ eq_pii_31 ] ) ,
the following conditions hold : @xmath284 @xmath285 let us rewrite expression ( [ eq_pii_53 ] ) in the form @xmath286 , \,\ , \forall \ , { \bf v}\in v_{0 } .\ ] ] if we take @xmath287 in ( [ eq_pii_56 ] ) , we shall have @xmath288 taking into account the continuity of bilinear form ( [ eq_pii_19 ] ) , and property ( [ eq_pii_55 ] ) , we get @xmath289 further , in view of relation between norms @xmath290 we come to inequalities @xmath291 now let us take @xmath292 in expression ( [ eq_pii_56 ] )
. then @xmath293 $ ] .
this relation can be written as @xmath294 @xmath295.\ ] ] in view of properties ( [ eq_pii_19 ] ) and ( [ eq_pii_55 ] ) , we come to inequality @xmath296 suppose that @xmath297 . then , taking into account the coercitivity of bilinear form ( [ eq_pii_20 ] ) , property ( [ eq_pii_54 ] ) , and the previous inequality , we obtain @xmath298.\ ] ] in view of inequalities ( [ eq_pii_57 ] ) and ( [ eq_pii_58 ] ) , we find further @xmath299 if the following inequality @xmath300 holds , then the sequence @xmath301 will be monotonically nonincreasing : @xmath302 , and , hence , @xmath303 , where @xmath304 . passing to the limit as @xmath305 in expression ( [ eq_pii_59 ] ) , we obtain @xmath306 , i.e. @xmath307 , and , therefore @xmath308 . from inequality ( [ eq_pii_60 ] ) we establish the interval of allowable values of iteration parameter @xmath309 : @xmath310 since the norms @xmath264 and @xmath311 are equivalent , we have @xmath312 , and , hence , @xmath313 . using inequality ( [ eq_pii_59 ] ) , we find estimate @xmath314 it is not hard to show , that @xmath315 for @xmath316 .
we obtain this from the next relations : @xmath317 as follows from estimate ( [ eq_pii_61 ] ) , the convergence rate is maximal if the parameter @xmath318 is minimal , i.e. , if @xmath319 . @xmath114 * remark 1 . *
_ suppose that term @xmath128 is gateaux differentiable in @xmath241
. then conditions ( [ eq_pii_29 ] ) , ( [ eq_pii_30 ] ) in _ * _ theorem 4 _ * _ can be replaced by the following properties _
@xmath320 @xmath321 * proof .
* let us apply to @xmath322 the lagrange formula of finite increments @xcite : @xmath323 then the satisfaction of conditions ( [ eq_pii_62 ] ) and ( [ eq_pii_63 ] ) yields the satisfaction of properties ( [ eq_pii_30 ] ) and ( [ eq_pii_31 ] ) .
@xmath114 now let us investigate the stability of iterative method ( [ eq_pii_46 ] ) to errors .
suppose that conditions of * theorem 4 * are satisfied
. then for any @xmath324 , @xmath278 there exists a unique solution @xmath325 of problem ( [ eq_pii_51 ] ) .
therefore , there exists an operator @xmath326 , which maps every element @xmath327 to the solution @xmath328 of the problem ( [ eq_pii_51 ] ) , and the iterative method ( [ eq_pii_46 ] ) can be written in the following form @xmath329 now assume that in each step @xmath248 of iterative method ( [ eq_pii_64 ] ) we get some computational errors
. then this iterative method will take the form : @xmath330 @xmath331 where @xmath332 , @xmath278 is an approximate solution of problem ( [ eq_pii_51 ] ) , @xmath333 , @xmath278 is a computational error , which occurs in each step @xmath248 , and @xmath334 is an initial guess error . * corollary 1 . *
_ suppose that conditions of _ * _ theorem 4 _ * _ are satisfied , and errors which occur in each step @xmath248 of iterative method ( [ eq_pii_46 ] ) are uniformly bounded , i.e. _
@xmath335 _ then the following estimates hold : _ @xmath336 @xmath337 @xmath338 @xmath339 _ where @xmath276 is the exact solution of problem ( [ eq_pii_41 ] ) . _
the proof of this proposition follows from property ( [ eq_pii_50 ] ) .
thus , the errors , which occur in each step of iterative method ( [ eq_pii_46 ] ) , do not accumulate .
now consider the nonstationary iterative method for the solution of nonlinear variational equation ( [ eq_pii_41 ] ) , where bilinear forms @xmath340 are different in each iteration @xcite . in space
@xmath91 introduce a sequence of bilinear forms @xmath341 , @xmath278 , which satisfy the property @xmath342 for the solution of nonlinear variational equation ( [ eq_pii_41 ] ) , we have proposed the following nonstationary iterative method @xcite : @xmath343,\,\ , k=0,1, ... \,,\ ] ] where @xmath270 is the @xmath248-th approximation to the exact solution of problem ( [ eq_pii_41 ] ) , and @xmath251 is an iterative parameter . we also have proved the next proposition about the convergence of this method
. * theorem 5 . * _ suppose that conditions ( [ eq_pii_19 ] ) ( [ eq_pii_21 ] ) , ( [ eq_pii_29 ] ) ( [ eq_pii_31 ] ) hold , bilinear forms @xmath344 satisfy properties _
@xmath345 @xmath346 @xmath347 @xmath348 _ and iterative parameter @xmath309 lies in the interval @xmath349 _ , @xmath350 .
_ _ _ then the sequence @xmath258 obtained by the nonstationary iterative method ( [ eq_pii_71 ] ) converges strongly in @xmath70 to the exact solution @xmath276 of the variational equation ( [ eq_pii_41 ] ) , i.e. @xmath351_. _ _ the proof of this theorem is similar to the proof of * theorem 4*. we omit the details .
note , that in iterative methods ( [ eq_pii_46 ] ) and ( [ eq_pii_71 ] ) we can also take the parameter @xmath309 differently in each iteration .
the purpose of the nonstationary choice of @xmath309 might be the improvement of the convergence rate of iterative methods ( [ eq_pii_46 ] ) and ( [ eq_pii_71 ] ) .
note , that in the most general case iterative methods ( [ eq_pii_46 ] ) and ( [ eq_pii_71 ] ) applied to solve nonlinear penalty variational equations ( [ eq_pii_41 ] ) for multibody contact problems do not lead to domain decomposition .
therefore , we now consider such variants of these methods , which lead to domain decomposition , namely , which reduce the solution of the original multibody contact problem in @xmath352 to the solution of a sequence of separate linear variational problems in subdomains @xmath8 , @xmath5 .
let us take the bilinear form @xmath353 in the iterative method ( [ eq_pii_46 ] ) as follows @xcite : @xmath354 where @xmath355 is the next bilinear form @xcite : @xmath356 here @xmath357 are characteristic functions of some given subareas @xmath358 of possible contact zones @xmath359 , @xmath5 , @xmath142 .
the iterative method ( [ eq_pii_46 ] ) with bilinear form ( [ eq_pii_77 ] ) can be written in the following equivalent way : @xmath360 @xmath361 * lemma 4 . * _ suppose that surfaces @xmath31 _ , _ @xmath362 are piecewise smooth .
then bilinear form ( [ eq_pii_77 ] ) is symmetric , continuous and nonnegative , i.e. _
@xmath363 @xmath364 @xmath365 * proof .
* it is obvious , that conditions ( [ eq_pii_80 ] ) and ( [ eq_pii_82 ] ) hold .
thus , let us show the continuity of bilinear form ( [ eq_pii_77 ] )
. we can write @xmath366 , where @xmath367 the first term can be written in the following form : @xmath368 \left[{\bf n}_{\alpha } \cdot { \rm{tr}}_{\alpha } ^{0 } ( \ , { \bf v}_{\alpha } ) \right]\ , ds,\ ] ] taking into account , that the functions @xmath369 and the components of unit normals @xmath370 are bounded , and using inequality ( [ eq_pii_33 ] ) and schwarz inequality , we obtain @xmath371^{3 } } ^{2 } \left\| tr_{\ , \alpha } ^{0 } ( { \bf v}_{\alpha } ) \right\| _ { [ l_{2 } ( s_{\alpha \beta } ) ] ^{3 } } ^{2 } .\ ] ] in view of inequality ( [ eq_pii_35 ] ) , we find further @xmath372 similar inequality can be obtained for the second term of @xmath373 .
thus @xmath374 as a result , we establish @xmath375 @xmath376 where @xmath377 , @xmath378 .
@xmath114 from this * lemma * and * lemma 1 * , it follows that bilinear form ( [ eq_pii_76 ] ) is symmetric , continuous and coercive with constants @xmath379 and @xmath380 respectively .
in addition , due to * lemmas * * 1 * , * 2 * , and * 4 * , functionals @xmath95 , @xmath381 , and @xmath382 are linear and continuous in @xmath269 .
therefore , there exists a unique solution @xmath383 of variational problem ( [ eq_pii_78 ] ) .
thus , the conditions of * theorem 4 * are satisfied , and we obtain the next proposition
. * theorem 6 . * _ suppose that @xmath255 _ , @xmath384 .
_ then the sequance @xmath385 _ , _ obtained by iterative method ( [ eq_pii_78 ] ) ( [ eq_pii_79 ] ) , which is equivalent to iterative method ( [ eq_pii_46 ] ) with bilinear form ( [ eq_pii_76 ] ) , converges strongly in @xmath70 to the exact solution @xmath249 of nonlinear penalty variational equation ( [ eq_pii_41 ] ) for unilateral multibody contact problem , i.e. _ @xmath386 .
_ moreover , the convergence rate in norm @xmath264 is linear ( [ eq_pii_50 ] ) , where @xmath387 _ , _ and the maximal rate reaches as @xmath388 _ , @xmath257 . now let us show that iterative method ( [ eq_pii_78 ] ) ( [ eq_pii_79 ] ) leads to domain decomposition . due to relationships
@xmath389 @xmath390 method ( [ eq_pii_78 ] ) ( [ eq_pii_79 ] ) rewrites as follows : @xmath391 @xmath392 @xmath393 since the common quantities of the subdomains are known from the previous iteration , the variational equation ( [ eq_pii_83 ] ) splits into @xmath2 variational equations in separate subdomains @xmath8 .
therefore , method ( [ eq_pii_83 ] ) ( [ eq_pii_84 ] ) can be written in the following equivalent form : @xmath394 @xmath395 @xmath396 @xmath397 since bilinear forms @xmath398 , @xmath399 are symmetric , continuous and coercive , and functionals @xmath400 , @xmath401 , @xmath402 are linear and continuous in @xmath403 , it follows that there exists a unique solution @xmath404 of each variational equation ( [ eq_pii_85 ] ) .
furthermore , it is obvious to see that the unique solution @xmath405 of variational equation ( [ eq_pii_78 ] ) , i.e. ( [ eq_pii_83 ] ) , takes the form @xmath406 .
therefore , the solution of variational equation ( [ eq_pii_78 ] ) is equivalent to the solution of @xmath2 variational equations ( [ eq_pii_85 ] ) in separate subdomains , and iterative processes ( [ eq_pii_78 ] ) ( [ eq_pii_79 ] ) and ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) are equivalent .
now , consider iterative method ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) in more detail .
in each iteration of this method we have to solve in parallel @xmath2 variational equations ( [ eq_pii_85 ] ) , which correspond to some elasticity problems in subdomains with prescribed robin boundary conditions on possible contact areas : @xmath407 here @xmath408 are unknown normal stresses on possible contact areas @xmath31 .
therefore , iterative method ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) refers to parallel robin
robin type domain decomposition schemes . since domain decomposition method ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) and iterative method ( [ eq_pii_78 ] ) ( [ eq_pii_79 ] )
are equivalent , the convergence * theorem 6 * also holds for the method ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) . taking different characteristic functions @xmath409 in ( [ eq_pii_85 ] ) , i.e. different subareas @xmath410 of possible contact zones @xmath31
, we can obtain different particular cases of domain decomposition method ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) .
thus , taking @xmath411 , @xmath412 , i.e. @xmath413 , we get domain decomposition scheme with robin boundary conditions on whole possible contact areas : @xmath414 therefore , we have named this domain decomposition method as full parallel robin
robin domain decomposition scheme @xcite .
taking @xmath415 , @xmath416 , i.e. @xmath417 , we get parallel neumann
neumann domain decomposition scheme @xcite : @xmath418 @xmath419 in each step @xmath248 of this scheme we have to solve in parallel @xmath2 variational equations ( [ eq_pii_88 ] ) , which correspond to elasticity problems in subdomains with neumann boundary conditions on possible contact areas : @xmath420 note , that in the most general case , we can choose functions @xmath409 , i.e. surfaces @xmath410 , differently for each @xmath421 .
moreover , we can choose functions @xmath409 differently at each iteration @xmath248 , i.e. @xmath422 where @xmath423 , @xmath278 are some given subareas of possible contact zones @xmath31 , @xmath5 , @xmath424 .
as a result we obtain following nonstationary robin robin type domain decomposition scheme @xmath425 @xmath426 @xmath427 this domain decomposition scheme is equivalent to nonstationary iterative method ( [ eq_pii_71 ] ) with bilinear forms @xmath428 where @xmath429 bilinear forms @xmath430 are symmetric , nonnegative , and continuous with constant @xmath431 .
hence , bilinear forms ( [ eq_pii_93 ] ) satisfy properties ( [ eq_pii_72 ] ) ( [ eq_pii_74 ] ) , where @xmath379 , @xmath380 .
it is obvious to see that condition ( [ eq_pii_75 ] ) in * theorem 5 * for bilinear forms ( [ eq_pii_93 ] ) is equivalent to the following condition @xmath432 which by - turn is equivalent to the condition @xmath433 therefore , from * theorem 5 * we obtain the next proposition about convergence of nonstationary domain decomposition scheme ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] )
. * theorem 7 . * _ suppose that @xmath349 _ , @xmath434 , _ and functions @xmath435 satisfy property ( [ eq_pii_95 ] )
. then the sequence @xmath385 _ , _ obtained by nonstationary domain decomposition scheme ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] ) , converges strongly in @xmath70 to the exact solution of nonlinear penalty variational equation ( [ eq_pii_41 ] ) for unilateral multibody contact problem , i.e. @xmath386_. now let us consider a particular case of domain decomposition method ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] ) .
in each iteration @xmath248 let us choose functions @xmath435 as follows @xcite : @xmath436 where @xmath43 , @xmath437 .
then , taking into consideration that @xmath438 , we obtain the method @xcite : @xmath439 @xmath440 in each step @xmath248 of this method we have to solve in parallel @xmath2 variational equations ( [ eq_pii_97 ] ) , which correspond to elasticity problems in subdomains with prescribed displacements @xmath441 through penalty on some subareas of possible contact zones @xmath31 .
therefore , we can conventionally name this method as nonstationary parallel dirichlet
dirichlet domain decomposition scheme .
note , that in some particular cases of penalty robin robin domain decomposition method ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] ) ( for example @xmath442 , @xmath443 or @xmath444 , @xmath412 ) we can pass the limit to @xmath193 , and obtain domain decomposition schemes without penalty @xcite .
the advantages of proposed domain decomposition schemes are their simplicity , and the regularization of the original contact problem because of the use of penalty term .
these domain decomposition schemes have only one iteration loop , which deals with domain decomposition and nonlinearity of unilateral contact conditions .
the disadvantage of proposed methods is that we have to choose a penalty parameter .
finally , let us say that iterative methods ( [ eq_pii_46 ] ) and ( [ eq_pii_71 ] ) for the solution of nonlinear variational equations are rather general . from these methods , besides parallel robin
robin type domain decomposition schemes ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) and ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] ) , we also can obtain other different particular iteratve methods for the solution of penalty variational equation of unilateral multibody contact problems , which do not lead to domain decomposition . thus , taking bilinear form @xmath445 , + @xmath446 , + and iterative parameter @xmath447 in ( [ eq_pii_46 ] ) , we obtain iterative method for the solution of multibody unilateral contact problems , which can be viewed as a generalization of penalty iteration method , proposed in @xcite for the solution of crack problems with nonpenetration conditions . taking in method ( [ eq_pii_71 ] )
bilinear forms @xmath448 in each step @xmath248 as follows @xmath449 where @xmath450 @xmath451 and iterative parameter @xmath447 , we obtain iterative method , which can be viewed as active set method , i.e. semi - smooth newton method , for unilateral multibody contact problems . for one - body contact problems
active set method was introduced in @xcite .
the convergence theorem for active set method , as a variant of semi - smooth newton method , was proved in @xcite .
note , that bilinear forms @xmath452 and @xmath453 are symmetric , nonnegative and coercive . however , these two particular cases of methods ( [ eq_pii_46 ] ) and ( [ eq_pii_71 ] ) do not lead to the domain decomposition . in the next section
we investigate numerical efficiency of proposed by us penalty parallel robin robin type domain decomposition schemes ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) and ( [ eq_pii_91 ] ) ( [ eq_pii_92 ] ) .
we provide numerical investigation of proposed domain decomposition schemes for plane unilateral contact problems of two elastic bodies @xmath454 , @xmath455 . for the numerical solution of linear variational problems in subdomains
, we use finite element method ( fem ) with linear and quadratic triangular elements .
at first let us compare the convergence rates of different particular domain decomposition schemes .
consider contact problem of two transversally isotropic elastic bodies @xmath456 , @xmath455 with the plane of isotropy , parallel to the plane @xmath457 ( fig . 2 ) .
the material properties of the bodies are : @xmath458 , @xmath459 , @xmath460 , where @xmath461 , @xmath462 , and @xmath463 are the elasticity modulus , poisson s ratio , and the shear modulus for the body @xmath8 in the plane of isotropy , and @xmath464 , @xmath465 , @xmath466 are these constants in the orthogonal direction , @xmath455 .
the length and height of each body is the same , and is equal to @xmath467 .
the distance between bodies before the deformation is @xmath468 , the compression of the bodies is @xmath469 , @xmath470 , and the possible contact area is @xmath471,\ , \ ,
x_{2 } = 4\ , b\right\}$ ] .
the problem is solved by parallel robin robin domain decomposition schemes , using fem with 1595 quadratic triangular elements in each body ( 15 elements on each side of possible contact area @xmath472 ) .
we take the penalty parameter in the form @xmath473 , @xmath474 , where @xmath475 is dimensionless penalty coefficient .
the initial guesses for displacements @xmath476 , @xmath455 are chosen using the bar model @xcite : @xmath477^{- } } { \theta e'_{1 } \left(1+c\right)},\,\ , { \bf x } \in s_{12},\,\,\ , u_{2\ , n}^{0 } ( { \bf x})=\frac{4\ , bc\left[d_{12 } ( { \bf x})-\delta \right]^{- } } { \theta e'_{2 } \left(1+c\right ) } + \delta , \,\ , { \bf x } \in s_{21 } .\ ] ] we use the following stopping criterion for domain decomposition schemes @xmath478 where @xmath479^{2 } } $ ] is the discrete norm , @xmath480 are the finite element nodes on the possible contact area , and @xmath481 is the relative accuracy for displacements .
3 shows the approximations to dimensionless normal contact stress @xmath482 , @xmath483 , @xmath484 , obtained by parallel neumann
neumann scheme ( [ eq_pii_88 ] ) ( [ eq_pii_89 ] ) ( @xmath485 ) at iterations @xmath486 ( curves 14 ) for optimal iterative parameter @xmath487 and accuracy @xmath488 .
the dashed curve represents the exact solution for two half - spaces , obtained in @xcite .
hence , the real contact area is @xmath489\kern-0.15em\right]$ ] , where @xmath490\kern-0.15em\right]=\left\{{\bf x}=\left(x_{1 } , x_{2 } \right)^{{\rm t } } : \,\,\,\
, x_{1 } \in [ y;\ , \ , z],\ , \ , x_{2 } = 4\ , b\right\}$ ] .
at fig . 4 and fig . 5
the convergence rates of different particular domain decomposition schemes are compared .
the total iteration number @xmath491 dependence on iteration parameter @xmath309 for accuracy @xmath488 is shown at fig . 4 , and its dependence on logarithmic accuracy @xmath492 for optimal iteration parameter @xmath493
is shown at fig .
the first curve at these figures represents the parallel neumann
neumann scheme ( @xmath494 , @xmath495 ) , curves 2 , 3 , 4 and 5 correspond to the parallel robin robin schemes ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) with @xmath496 equal to @xmath497\kern-0.15em\right]$ ] , @xmath498\kern-0.15em\right]$ ] , @xmath499\kern-0.15em\right]$ ] , and @xmath500\kern-0.15em\right]$ ] ( @xmath501 , @xmath502 ) respectively . curve 3 also represents the nonstationary parallel dirichlet dirichlet scheme ( [ eq_pii_97 ] ) ( [ eq_pii_98 ] ) . the optimal iterative parameters @xmath503 for the schemes represented at curves 15 are @xmath487 , 0.39 , 0.72 , 0.85 and 0.92 respectively
. for @xmath319 and accuracy @xmath488 these schemes converge in 21 , 11 , 5 , 11 and 14 iterations . thus , the convergence rate of stationary robin
robin domain decomposition schemes is linear . the parallel robin robin scheme ( [ eq_pii_85 ] ) ( [ eq_pii_86 ] ) with the surfaces @xmath504 , @xmath505 most closed to the real contact area ( @xmath506\kern-0.15em\right]$ ] ) , and the nonstationary parallel dirichlet
dirichlet scheme ( [ eq_pii_97 ] ) ( [ eq_pii_98 ] ) ( @xmath507 ) , which are represented at curve 3 , have the highest convergence rates
. these two schemes also have the widest range from which the iterative parameter @xmath508 can be chosen . the convergence rate of parallel neumann
neumann scheme ( @xmath509 ) , which is represented at curve 1 , is the most slow .
now let us investigate the convergence of penalty method and its dependence on finite element discretization .
consider the unilateral contact problem of two isotropic bodies @xmath510 and @xmath511 , one of which has a groove ( fig . 6 ) .
the bodies are uniformly loaded by normal stress with intensity @xmath318 .
each body has length @xmath512 and height @xmath513 , and the grove has length @xmath514 .
the material properties of the bodies are the same : @xmath515 , @xmath516 .
the distance between bodies before the deformation is @xmath517^{+ } \right\}^{3/2 } $ ] , where @xmath518 , @xmath519 .
the possible contact area is @xmath520,\ , \ ,
x_{2 } = h\right\}$ ] .
the problem is solved by nonstationary parallel dirichlet dirichlet domain decomposition scheme ( [ eq_pii_97 ] ) ( [ eq_pii_98 ] ) with finite element approximations on triangles .
the penalty parameter is taken as follows @xmath521 where @xmath475 is dimensionless penalty coefficient .
initial guesses for displacements are chosen using the bar model @xcite : @xmath522^{- } } { \left(1+c\right ) } , \,\,\ , { \bf x } \in s_{12 } , \ ] ] @xmath523^{- } } { \left(1+c\right ) } + \delta , \,\,\ , { \bf x } \in s_{21 } , \ ] ] where @xmath524 .
we use ( [ eq_pii_99 ] ) as stopping criterion for iterative process . for iterative parameter @xmath525 $ ] , accuracy @xmath526 , penalty coefficients @xmath475 and
finite element meshes considered below , parallel dirichlet
dirichlet scheme , applied to solve this problem , converges in 215 iterations .
we investigate the dependence of the quality of numerical solution , obtained by this scheme , on penalty parameter and finite element mesh .
plots at fig .
7 represent approximations to dimensionless contact stress @xmath527 , @xmath528 for bodies with size @xmath529 and external load @xmath530 , obtained by dirichlet
dirichlet scheme for different dimensionless penalty coefficients @xmath475 at fixed finite element mesh with 64 linear triangular finite elements on each side of possible contact area @xmath472 .
curves 14 correspond to @xmath531 , @xmath532 , @xmath533 and @xmath534 respectively .
plots at fig .
8 represent approximations to @xmath535 , @xmath536 for bodies with length @xmath537 , height @xmath538 , and external load @xmath539 , obtained for differrent dimensionless penalty coefficients @xmath475 and different finite element meshes .
curves 1 and 2 at this figure correspond to @xmath540 for dimensionless penalty coefficients @xmath531 and @xmath541 respectively at finite element mesh with 32 linear triangular elements on each side of possible unilateral contact area @xmath472 . curves 3 and 4 correspond to @xmath540 for @xmath531 and @xmath541 respectively , but for finite element mesh with 64 linear triangular elements on each side of @xmath542 . dashed curve at this figure and at fig .
7 represents the exact solution , obtained in @xcite for the contact of two half - spaces .
here we see that in spite of the solution , obtained for penalty coefficient @xmath531 at mesh with 32 finite elements on each side of @xmath472 ( curve 1 at fig .
8) , the solution obtained for lower penalty coefficient @xmath541 at the same mesh become instable .
but if we refine finite element mesh twice for the penalty coefficient @xmath541 , then the influence of errors on the perturbation of initial data will decrease , and we shall obtain much better approximation to the exact solution ( curve 4 at fig .
hence , we conclude that for obtaining a nice approximation to the solution , we need to decrease penalty parameter and to refine finite element mesh simultaneously .
for the solution of unilateral multibody contact problems of elasticity we have proposed on continuous level a class of parallel robin robin type domain decomposition schemes , which are based on penalty method for variational inequalities and some stationary or nonstationary iterative methods for nonlinear variational equations . in each step of these schemes
we have to solve in parallel linear variational equations in subdomains , which correspond to some elasticity problems with robin boundary conditions on possible contact areas . numerical analysis of proposed domain decomposition schemes has been made for plane two - body contact problems using linear and quadratic finite element approximations on triangles .
convergence rates of different particular domain decomposition schemes have been compared and their dependence on iterative parameter @xmath309 has been investigated .
the penalty parameter and mesh refinement influence on the numerical solution have been examined .
numerical experiments have confirmed the theoretical results about convergence of these domain decomposition schemes .
the advantages of proposed domain decomposition schemes are their simplicity , and the regularization of the original contact problem because of the use of penalty term .
these domain decomposition schemes have only one iteration loop , which deals with domain decomposition and nonlinearity of unilateral contact conditions . in addition
, these methods are obtained on continuous level , and they do not depend on any discretization technique .
p. avery , c. farhat , the feti family of domain decomposition methods for inequality - constrained quadratic programming : application to contact problems with conforming and nonconforming interfaces , comput .
methods appl .
* 198 * ( 2009 ) 16731683 .
i. i. dyyak , i. i. prokopyshyn , convergence of the neumann parallel scheme of the domain decomposition method for problems of frictionless contact between several elastic bodies , journal of mathematical sciences * 171*(4 ) ( 2010 ) 516533 .
i. i. dyyak , i. i. prokopyshyn , domain decomposition schemes for frictionless multibody contact problems of elasticity , in : g. kreiss et al ( eds . ) , numerical mathematics and advanced applications 2009 .
proceedings of enumath 2009 , the 8th european conference on numerical mathematics and advanced applications , uppsala , july 2009 , springer , berlin heidelberg , 2010 , pp .
297305 .
i. i. prokopyshyn , penalty method based domain decomposition schemes for contact problems of elastic solids , dissertation for a candidate s degree in physics and mathematics by specialty 01.05.02 mathematical modeling and computational methods , national academy of sciences of ukraine , pidstryhach institute for applied problems of mechanics and mathematics , lviv , 2010 , url:194.44.15.230:8080/v25/resources / thesis / prokopyshyn.pdf [ in ukrainian ] .
i. i. prokopyshyn , r. m. martynyak , numerical investigation of contact interaction of two solids with a groove by domain decomposition method , problems of computational mechanics and strength of structures * 16 * ( 2011 ) 240251 [ in ukrainian ] .
grigorenko , i. i. dyyak , s. i. matysyak , i. i. prokopyshyn , domain decomposition methods applied to solve frictionless - contact problems for multilayer elastic bodies , int .
. mech . * 46*(4 ) ( 2010 ) 388399 . | the paper is devoted to penalty robin robin domain decomposition methods ( ddms ) , proposed by us for the solution of unilateral multibody contact problems of elasticity .
these ddms are based on the penalty method for variational inequalities and some stationary and nonstationary iterative methods for nonlinear variational equations .
the main result of the paper is that we give mathematical justification of proposed ddms and prove theorems about their convergence .
we also provide numerical investigations of the efficiency of these methods using finite element approximations .
* key words : * elasticity , multibody contact , variational inequalities , penalty method , iterative methods , domain decomposition * msc2010 : * 65n55 , 74s05 |
in recent years , the rapid increase in both the availability of data on networks ( of all kinds , but especially social ones ) and the demand , from many scientific areas , for analyzing such data has resulted in a surge of generative and descriptive models for network data @xcite . within statistics ,
this trend has led to a renewed interest in developing , analyzing and validating statistical models for networks @xcite . yet as networks are a nonstandard type of data , many basic properties of statistical models for networks are still unknown or have not been properly explored . in this article
we investigate the conditions under which statistical inferences drawn over a sub - network will generalize to the entire network .
it is quite rare for the data to ever actually be the _ whole _ network of relations among a given set of nodes or units ; typically , only a sub - network is available .
guided by experience of more conventional problems like regression , analysts have generally fit models to the available sub - network , and then extrapolated them to the larger true network which is of actual scientific interest , presuming that the models are , as it were , consistent under sampling .
what we show is that this is only valid for very special model specifications , and the specifications where it is _ not _ valid include some of which are currently among the most popular and scientifically appealing . in particular , we restrict ourselves to exponential random graph models ( ergms ) , undoubtedly one of the most important and popular classes of statistical models of network structure .
in addition to the general works already cited , the reader is referred to @xcite for detailed accounts of these models .
there are many reasons ergms are so prominent .
on the one hand , ergms , as the name suggests , are exponential families , and so they inherit all the familiar virtues of exponential families in general : they are analytically and inferentially convenient @xcite ; they naturally arise from considerations of maximum entropy @xcite and minimum description length @xcite , and from physically - motivated large deviations principles @xcite ; and if a generative model obeys reasonable - seeming regularity conditions while still having a finite - dimensional sufficient statistic , it must be an exponential family @xcite .
on the other hand , ergms have particular virtues as models of networks .
the sufficient statistics in these models typically count the number or density of certain `` motifs '' or small sub - graphs , such as edges themselves , triangles , @xmath0-cliques , stars , etc .
these in turn are plausibly related to different network - growth mechanisms , giving them a substantive interpretation ; see , for example , @xcite as an exemplary application of this idea , or , more briefly , section [ secergms ] below .
moreover , the important task of edge prediction is easily handled in this framework , reducing to a conditional logistic regression @xcite . since the development of ( comparatively ) computationally - efficient maximum - likelihood estimators ( based on monte carlo sampling ) , ergms have emerged as flexible and persuasive tools for modeling network data @xcite .
despite all these strengths , however , ergms are tools with a serious weakness .
as we mentioned , it is very rare to ever observe the whole network of interest . the usual procedure , then , is to fit ergms ( by maximum likelihood or pseudo - likelihood ) to the observed sub - network , and then extrapolate the same model , with the same parameters , to the whole network ; often this takes the form of interpreting the parameters as `` provid[ing ] information about the presence of structural effects observed in the network '' @xcite , page 194 , or the strength of different network - formation mechanisms ; @xcite are just a few of the more recent papers doing this .
this obviously raises the question of the statistical ( i.e. , large sample ) consistency of maximum likelihood estimation in this context .
unnoticed , however , is the logically prior question of whether it is _ probabilistically _ consistent to apply the same ergm , with the same parameters , both to the whole network and its sub - networks .
that is , whether the marginal distribution of a sub - network will be consistent with the distribution of the whole network , for all possible values of the model parameters .
the same question arises when parameters are compared across networks of different sizes ( as in , e.g. , @xcite ) .
when this form of consistency fails , then the parameter estimates obtained from a sub - network may not provide reliable estimates of , or may not even be relatable to , the parameters of the whole network , rendering the task of statistical inference based on a sub - network ill - posed .
we formalize this question using the notion of `` projective families '' from the theory of stochastic processes .
we say that a model is _ projective _ when the same parameters can be used for both the whole network and any of its sub - networks . in this article , we fully characterize projectibility of discrete exponential families and , as corollary , show that ergms are projective only for very special choices of the sufficient statistic . _
outline_. our results are not specific just to networks , but pertain more generally with exponential families of stochastic processes . in section [ secprojectiblity - and - exp - fam ] , therefore , we lay out the necessary background about projective families of distributions , projective parameters and exponential families in a somewhat more abstract setting than that of networks . in section [ secprojectibility - and - independent - increments ]
we show that a necessary and sufficient condition for an exponential family to be projective is that the sufficient statistics obey a kind of additive decomposition .
this in turn implies strong independence properties .
we also prove results about the consistency of maximum likelihood parameter estimation under these conditions ( section [ secconsistency - of - mle ] ) . in section [ secergms ] , we apply these results to ergms , showing that most popular specifications for social networks and other stochastic graphs can not be projective .
we then conclude with some discussion on possible constructive responses .
the proofs are contained in the . _ related work_. an early recognition of the fact that sub - networks may have statistical properties which differ radically from those of the whole network came in the context of studying networks with power - law ( `` scale - free '' ) degree distributions . on the one hand ,
stumpf , wiuf and may @xcite showed that `` subnets of scale - free networks are not scale - free ; '' on the other , achlioptas et al .
@xcite demonstrated that a particular , highly popular sampling scheme creates the appearance of a power - law degree distribution on nearly any network .
while the importance of network sampling schemes has been recognized since then @xcite , chapter 5 , and valuable contributions have come from , for example , @xcite , we are not aware of any work which has addressed the specific issue of consistency under projection which we tackle here . perhaps the closest approaches to our perspective are @xcite and @xcite . the former considers conditions under which infinite - dimensional families of distributions on abstract spaces have projective limits .
the latter , more concretely , addresses the consistency of maximum likelihood estimators for exponential families of dependent variables , but under assumptions ( regarding markov properties , the `` shape '' of neighborhoods , and decay of correlations in potential functions ) which are basically incomparable in strength to ours.=-1
our results about exponential random graph models are actually special cases of more general results about exponential families of dependent random variables , and are just as easy to state and prove in the general context as for graphs . setting this up
, however , requires some preliminary definitions and notation , which make precise the idea of `` seeing more data from the same source . '' in order to dispense with any measurability issues , we will implicitly assume the existence of an underlying probability measure for which the random variables under study are all measurable . furthermore , for the sake of readability we will not rely on the measure theoretic notion of filtration : though technically appropriate , it will add nothing to our results .
let @xmath1 be a collection of finite subsets of a denumerable set @xmath2 partially ordered with respect to subset inclusion . for technical reasons
, we will further assume that @xmath3 has the property of being an ideal : that is , if @xmath4 belongs to @xmath3 , then all subsets of @xmath4 are also in @xmath3 and if @xmath4 , and @xmath5 belongs to @xmath3 , then so does their union .
we may think of passing from @xmath6 to @xmath7 as taking increasingly large samples from a population , or recording increasingly long time series , or mapping data from increasing large spatial regions , or over an increasingly dense spatial grid , or looking at larger and larger sub - graphs from a single network .
accordingly , we consider the associated collection of parametric statistical models @xmath8 indexed by @xmath1 , where , for each @xmath9 , @xmath10 is a family of probability distributions indexed by points @xmath11 in a fixed open set @xmath12
. the probability distributions in @xmath13 are also assumed to be supported over the same @xmath14 , which are countable sets for each @xmath4 .
we assume that the partial order of @xmath1 is isomorphic to the partial order over @xmath15 , in the sense that @xmath16 if and only if @xmath17 . for given @xmath11 and @xmath6 , we denote with @xmath18 the random variable distributed as @xmath19 . in particular , for a given @xmath20 , we can regard the @xmath21 as finite - dimensional ( i.e. , marginal ) distributions .
for each pair @xmath22 in @xmath1 with @xmath23 , we let @xmath24 be the natural index projection given by @xmath25 . in the context of networks
, we may think of @xmath26 as the set of nodes of a possibly infinite random graph , which without loss of generality can be taken to be @xmath27 and of @xmath3 as the collection of all finite subsets of @xmath26 .
then , for some positive integers @xmath28 and @xmath29 , we may , for instance , take @xmath30 and @xmath31 , so that @xmath18 will be the induced sub - graph on the first @xmath28 nodes and @xmath32 the induced sub - graph on the first @xmath33 nodes .
the projection @xmath34 then just picks out the appropriate sub - graph from the larger graph ; see figure [ figprojective - structure - illustrated ] for a schematic example .
we will be concerned with a natural form of probabilistic consistency of the collection @xmath35 which we call _ projectibility _ , defined below .
is contained in the larger set of observables @xmath36 , @xmath18 ( on the right ) can be recovered from @xmath37 ( on the left ) through the projection @xmath34 , which simply drops the extra data . ]
the family @xmath38 is _ projective _ if , for any @xmath6 and @xmath5 in @xmath3 with @xmath39 , @xmath40 see @xcite , page 115 , for more general treatment of projectibility . in words , @xmath41 is a projective family when @xmath39 implies that @xmath42 can be recovered by marginalization over @xmath43 , for all @xmath11 . within a projective family , @xmath44 denotes the infinite - dimensional distribution , which thus exists by the kolmogorov extension theorem @xcite , theorem 6.16 , page 115 .
projectibility is automatic when the generative model calls for independent and identically distributed ( iid ) observations .
it is also generally unproblematic when the model is specified in terms of _ conditional _ distributions : one then just uses the ionescu tulcea extension theorem in place of that of kolmogorov @xcite , theorem 6.17 , page 116 .
however , many models are specified in terms of _ joint
_ distributions for various index sets , and this , as we show in theorem [ thmprojectible - iff - separable ] , can rule out projectibility . we restrict ourselves to _ exponential family _
models by assuming that , for each choice of @xmath45 and @xmath9 , @xmath19 has density with respect to the counting measure over @xmath14 given by @xmath46 where @xmath47 is the measurable function of minimal sufficient statistics , and @xmath48 is the _ partition function _ given by @xmath49 if @xmath50 , we will write @xmath51 for the random variable corresponding to the sufficient statistic . equation ( [ eqnexponential - family - density ] ) implies that @xmath52 itself has an exponential family distribution , with the same parameter @xmath11 and partition function @xmath53 @xcite , proposition 1.5 . specifically , the distribution function is @xmath54 where the term @xmath55 , which we will call the _ volume factor _ , counts the number of points in @xmath14 with the same sufficient statistics @xmath56 .
the moment generating function of @xmath52 is @xmath57 = z_{{a}}(\theta+\phi)/z_{{a}}(\theta ) .
\label{eqnmgf - of - exponential - family}\ ] ] if the sufficient statistic is completely additive , that is , if @xmath58 , then this is a model of independent ( if not necessarily iid ) data . in general , however , the choice of sufficient statistics may impose , or capture , dependence between observations . because we are considering exponential families defined on increasingly large sets of observations , it is convenient to introduce some notation related to multiple statistics .
fix @xmath59 such that @xmath23 .
then @xmath60 , and we will sometimes write this function @xmath61 , where the first argument is in @xmath14 and the second in @xmath62 . we will have frequent recourse to the increment to the sufficient statistic , @xmath63 .
the volume factor @xmath64 is defined as before , but we shall also consider , for each observable value @xmath56 of the sufficient statistics for @xmath4 and increment @xmath65 of the sufficient statistics from @xmath4 to @xmath5 , the _ joint volume factor , _
@xmath66 and the _ conditional volume factor , _ @xmath67 as we will see , these volume factors play a key role in characterizing projectibility .
in this section we characterize projectibility in terms of the increments of the vector of sufficient statistics .
in particular we show that exponential families are projective if , and only if , their sufficient statistics decompose into separate additive contributions from disjoint observations in a particularly nice way which we formalize in the following definition .
the sufficient statistics of the family @xmath35 have _ separable increments _ when , for each @xmath23 , @xmath68 , the range of possible increments @xmath65 is the same for all @xmath69 , and the conditional volume factor is constant in @xmath69 , that is , @xmath70 .
it is worth noting that the property of having separable increments is an intrinsic property of the family @xmath8 that depends only on the functional forms of the sufficient statistics @xmath71 and not on the model parameters @xmath45 .
this follows from the fact that , for any @xmath4 , the probability distributions @xmath72 have identical support @xmath73 .
thus , this property holds for all of @xmath11 or none of them .
the main result of this paper is then as follows .
the exponential family @xmath41 is projective if and only if the sufficient statistics @xmath74 have separable increments .
[ thmprojectible - iff - separable ] because projectibility implies separable increments , it also carries statistical - independence implications .
specifically , it implies that the increments to the sufficient statistics are statistically independent , and that @xmath75 and @xmath18 are conditionally independent given increments to the sufficient statistic .
interestingly , independent increments for the statistic are necessary but not quite sufficient for projectibility .
these claims are all made more specific in the propositions which follow .
we first show that projectibility implies that the sufficient statistics have independent increments .
in fact , a stronger results holds , namely that the increments of the sufficient statistics are independent of the actual sequence .
below we will write @xmath76 to signify @xmath77 .
if the exponential family @xmath78 is projective , then sufficient statistics @xmath79 have independent increments , that is , @xmath39 implies that @xmath80 under all @xmath11 .
[ propproj - to - ii ] in a projective exponential family , @xmath81 .
[ propincrement - indep - of - old - observable ] we note that independent increments for the sufficient statistics @xmath52 in no way implies independence of the actual observations @xmath18 . as a simple illustration ,
take the one - dimensional ising model , where @xmath82 , each @xmath83 , @xmath1 consists of all intervals from @xmath84 to @xmath28 , and the single sufficient statistic @xmath85 .
clearly , @xmath86 when @xmath87 , otherwise @xmath88 .
since @xmath89 , by theorem [ thmprojectible - iff - separable ] , the model is projective . by proposition [ propproj - to - ii ] , then , increments of @xmath90 should be independent , and direct calculation shows the probability of increasing the sufficient statistic by 1 is @xmath91 , no matter what @xmath92 are . while the sufficient statistic has independent increments , the random variables @xmath93 are all dependent on one another .. ] [ example : ising - model ] the previous results provide a way , and often a simple one , for checking whether projectibility fails : if the sufficient statistics do not have independent increments , then the family is not projective .
as we will see , this test covers many statistical models for networks .
it is natural to inquire into the converse to these propositions .
it is fairly straightforward ( if somewhat lengthy ) to show that independent increments for the sufficient statistics implies that the joint volume factor separates .
if an exponential family has independent increments , @xmath94 , then its joint volume factor separates , @xmath95 , and the distribution of @xmath90 is projective .
[ propii - to - volume - factor - separation ] however , independent increments for the sufficient statistics do _ not _ imply that separable increments ( hence projectibility ) , as shown by the next counter - example .
hence independent increments are a necessary but not sufficient condition for projectibility .
suppose that @xmath96 , and @xmath97 .
( thus there are 20 possible values for @xmath37 . )
let @xmath98 so that @xmath99 .
further , let @xmath100 it is not hard to verify that @xmath76 is always either @xmath101 or @xmath102 .
it is also straightforward to check that @xmath103 for all combinations of @xmath56 and @xmath65 , implying that @xmath104 , and that the joint volume factor separates . on the other hand ,
the _ conditional _ volume factors are not constant in @xmath69 , as @xmath105 while @xmath106 .
thus , the sufficient statistic has independent increments , but does not have separable increments . since projective families have separable increments ( proposition [ propproj - to - si ] ) ,
this can not be a projective family .
( this can also be checked by a direct and straightforward , if even more tedious , calculation . )
we conclude this section with a final observation .
butler , in @xcite , showed that when observations follow from an iid model with a minimal sufficient statistic , the predictive distribution for the next observation can be written entirely in terms of how different hypothetical values would change the sufficient statistic ; cf .
this predictive sufficiency property carries over to our setting .
[ thmpredictive - sufficiency ] in a projective exponential family , the distribution of @xmath107 conditional on @xmath18 depends on the data only through @xmath76 .
the main implications among our results are summarized in figure [ figimplications ] . .
probabilistic properties of the models are on the right , and algebraic / combinatorial properties of the sufficient statistic are on the left . ] _ exponential families of time series_. as the example of the ising model in section [ secindependence - properties ] ( page )
makes clear , our theorem applies whenever we need an exponential family to be projective , not just when the data are networks . in particular
, they apply to exponential families of time series , where @xmath2 is the natural or real number line ( or perhaps just its positive part ) , and the elements of @xmath1 are intervals .
an exponential family of stochastic processes on such a space has projective parameters if , and only if , its sufficient statistics have separable increments , and so only if they have independent increments .
_ transformation of parameters_. allowing the dimension of @xmath11 to be fixed , but for its components to change along with @xmath6 , does not really get out of these results . specifically ,
if @xmath11 is to be re - scaled in a way that is a function of @xmath6 alone , we can recover the case of a fixed @xmath11 by `` moving the scaling across the inner product , '' that is , by re - defining @xmath52 to incorporate the scaling . with a sample - invariant @xmath11 ,
it is this transformed @xmath90 which must have separable increments .
other transformations can either be dealt with similarly , or amount to using a nonuniform base measure ; see below .
_ statistical - mechanical interpretation_. it is interesting to consider the interpretation of our theorem , and of its proof , in terms of statistical mechanics . as is well known , the `` canonical '' distributions in statistical mechanics are exponential families ( boltzmann gibbs distributions ) , where the sufficient statistics are `` extensive '' physical observables , such as energy , volume , the number of molecules of various species , etc . , and the natural parameters are the corresponding conjugate `` intensive '' variables , such as , respectively , ( inverse ) temperature , pressure , chemical potential , etc .
equilibrium between two systems , which interact by exchanging the variables tracked by the extensive variables , is obtained if and only if they have the same values of the intensive parameters @xcite
. in our terms , of course , this is simply projectibility , the requirement that the same parameters hold for all sub - systems .
what we have shown is that for this to be true , the increments to the extensive variables must be completely unpredictable from their values on the sub - system .
furthermore , notice the important role played in both halves of the proof by the separation of the joint volume factor , @xmath108 . in terms of statistical mechanics ,
a macroscopic state is a collection of microscopic configurations with the same value of one or more macroscopic observables .
the boltzmann entropy of a macroscopic state is ( proportional to ) the logarithm of the volume of those microscopic states @xcite .
if we define our macroscopic states through the sufficient statistics , then their boltzmann entropy is just @xmath109 .
thus , the separation of the volume factor is the same as the additivity of the entropy across different parts of the system , that is , the entropy is `` extensive . ''
our results may thus be relevant to debates in statistical mechanics about the appropriateness of alternative , nonextensive entropies ; cf .
@xcite .
_ beyond exponential families_. it is not clear just how important it is that we have an exponential family , as opposed to a family admitting a finite - dimensional sufficient statistic .
as is well known , the two concepts coincide under some regularity conditions @xcite , but not quite strictly , and it would be interesting to know whether or not the exponential form of equation ( [ eqnexponential - family - density ] ) is strictly required .
we have attempted to write the proofs in a way which minimizes the use of this form ( in favor of the neyman factorization , which only uses sufficiency ) , but have not succeeded in eliminating it completely .
we return to this matter in the conclusion . _
prediction_. we have focused on the implications of projectibility for parametric inference .
exponential families are , however , often used in statistics and machine learning as generative models in applications where the only goal is prediction @xcite , and so ( to quote butler @xcite ) `` all parameters are nuisance parameters . '' but even in then , it must be possible to consistently extend the generative model s distribution for the training data to a joint distribution for training and testing data , with a single set of parameters shared by both old and new data . while this requirement may seem too trivial to mention , it is , precisely , projectibility . _ growing number of parameters_. in the proof of theorem [ thmprojectible - iff - separable ] , we used the fact that @xmath52 , and hence @xmath11 , has the same dimension for all @xmath110 .
there are , however , important classes of models where the number of parameters is allowed to grow with the size of the sample . particularly important , for networks , are models where each node is allowed a parameter ( or two ) of its own , such as its expected degree ; see , for instance , the classic @xmath111 model of @xcite , or the `` degree - corrected block models '' of @xcite
. we can formally extend theorem [ thmprojectible - iff - separable ] to cover some of these cases including those two particular specifications as follows .
assume that @xmath52 has a dimension which is strictly nondecreasing as @xmath6 grows , that is , @xmath112 whenever @xmath23 .
furthermore , assume that the set of parameters @xmath113 only grows , and that the meaning of the old parameters is not disturbed .
that is , under projectibility we should have @xmath114 for any fixed pair @xmath23 , we can accommodate this within the proof of theorem [ thmprojectible - iff - separable ] by re - defining @xmath52 to be a mapping from @xmath14 to @xmath115 , where the extra @xmath116 components of the vector are always zero .
the extra parameters in @xmath117 then have no influence on the distribution of @xmath18 and are unidentified on @xmath6 , but we have , formally , restored the fixed - parameter case .
the `` increments '' of the extra components of @xmath118 are then simply their values on @xmath37 , and , by the theorem , the range of values for these statistics , and the number of configurations on @xmath62 leading to each value , must be equal for all @xmath119 .
adapting our conditions for the asymptotic convergence of maximum likelihood estimators ( section [ secconsistency - of - mle ] ) to the growing - parameter setting is beyond our scope here .
_ nonuniform base measures_. if the exponential densities in ( [ eqnexponential - family - density ] ) are defined with respect to nonuniform base measures different from the counting measures , the sufficient statistics need not have separable increments . in the supplementary material @xcite
we address this issue and describe the modifications and additional assumptions required for our analysis to remain valid .
we thank an anonymous referee and pavel krivitsky for independently brining up this subtle point to our attention .
statistical inference in an exponential family naturally centers on the parameter @xmath11 .
as is well known , the maximum likelihood estimator @xmath120 takes a particularly simple form , obtainable using the fact [ which follows from equation ( [ eqnmgf - of - exponential - family ] ) ] that @xmath121 $ ] , @xmath122}{z_{{a}}^2(\theta ) } , \nonumber \\[-8pt ] \\[-8pt ] \nonumber t_{{a}}(x ) & = & \mathbf{e}_{{\widehat{\theta } } } [ t_{{a } } ] .\end{aligned}\ ] ] in words , the most likely value of the parameter is the one where the expected value of the sufficient statistic equals the observed value .
assume the conditions of theorem [ thmprojectible - iff - separable ] hold , so that the parameters are projective and the sufficient statistics have ( by lemma [ lemmaseparable - volume - factor - implies - projectible - increments ] ) independent increments .
define the logarithm of the partition function @xmath123
. would be the helmholtz free energy . ]
suppose that @xmath124 where @xmath125 is some positive - valued measure of the size of @xmath6 , @xmath126 a positive monotone - increasing function of it and @xmath127 is differentiable ( at least at @xmath11 ) .
then , by equation ( [ eqnmgf - of - exponential - family ] ) for the moment generating function , the cumulant generating function of @xmath52 is @xmath128 from the basic properties of cumulant generating functions , we have @xmath129 = \nabla_{\phi } \kappa _ { { a},\theta}(0 ) = r_{|{a}| } \nabla a(\theta ) .
\label{eqnexpect - of - suff - stat}\ ] ] substituting into equation ( [ eqnimplicit - exp - family - mle ] ) , @xmath130 thus to control the convergence of @xmath120 , we must control the convergence of @xmath131 .
consider a growing sequence of sets @xmath6 such that @xmath132 .
since @xmath52 has independent increments , and the cumulant generating functions for different @xmath6 are all proportional to each other , we may regard @xmath52 as a time - transformation of a lvy process @xmath133
that is , there is a continuous - time stochastic process @xmath134 with iid increments , such that @xmath135 has cumulant generating function @xmath136 , and @xmath137 .
note that @xmath52 itself does not have to have iid increments , but rather the distribution of the increment @xmath138 must only depend on @xmath139 . specifically , from lemma [ corconditional - moment - generating - function ] and equation ( [ eqnscaling - factor ] ) , the cumulant generating function of the increment must be @xmath140 $ ] .
the scaling factor homogenizes ( so to speak ) the increments of @xmath90 . writing the sufficient statistic as a transformed lvy process yields a simple proof that @xmath120 is strongly ( i.e. , almost - surely ) consistent .
since a lvy process has iid increments , by the strong law of large numbers @xmath141 converges almost surely ( @xmath44 ) to @xmath142 $ ] @xcite . since @xmath143
, it follows that @xmath144 $ ] a.s .
( @xmath44 ) as well ; but this limit is @xmath145 .
thus the mle converges on @xmath11 almost surely .
we have thus proved suppose that the model @xmath44 is projective , and that the log partition function obeys equation ( [ eqnscaling - factor ] ) for each @xmath9 .
then the maximum likelihood estimator exists and is strongly consistent .
we may extend this in a number of ways .
first , if the scaling relation equation ( [ eqnscaling - factor ] ) holds for a particular @xmath11 ( or set of @xmath11 ) , then @xmath131 will converge almost surely for that @xmath11 .
thus , strong consistency of the mle may in fact hold over certain parameter regions but not others .
second , when @xmath146 , all components of @xmath52 must be scaled by the _
factor @xmath126 . making the expectation value of one component of @xmath90
be @xmath147 while another was @xmath148 ( e.g. ) would violate equation ( [ eqnexpect - of - suff - stat ] ) and so equation ( [ eqnscaling - factor ] ) as well .
finally , while the exact scaling of equation ( [ eqnscaling - factor ] ) , together with the independence of the increments , leads to strong consistency of the mle , ordinary consistency ( convergence in probability ) holds under weaker conditions . specifically , suppose that log partition function or free energy scales in the limit as the size of the assemblage grows , @xmath149 we give examples toward the end of section [ secergms ] below .
we may then use the following theorem : suppose that an exponential family shows approximate scaling , that is , equation ( [ eqnasymptotic - scaling ] ) holds , for some @xmath11 . then , for any measurable set @xmath150 , @xmath151 where @xmath152 } , \label{eqnrate - function}\ ] ] and @xmath153 and @xmath154 are , respectively , the interior and the closure of @xmath155 .
[ thmldp ] when the limits in equations ( [ eqnldp - lower ] ) and ( [ eqnldp - upper ] ) coincide , which they will for most nice sets @xmath155 , we may say that @xmath156 since @xmath157 is minimized at 0 when @xmath158 , , by a second order taylor expansion , @xmath159 , where @xmath160 acts as the fisher information rate ; cf .
equation ( [ eqnldp - for - suff - stat ] ) holds in particular for any neighborhood of @xmath145 , and for the complement of such neighborhoods , where the infimum of @xmath161 is strictly positive .
thus @xmath131 converges in probability to @xmath145 , and @xmath162 , for all @xmath11 where equaiton ( [ eqnasymptotic - scaling ] ) holds.=-1 heuristically , when equation ( [ eqnasymptotic - scaling ] ) holds but equation ( [ eqnscaling - factor ] ) fails , we may imagine approximating the actual collection of dependent and heterogeneous random variables with an average of iid , homogenized effective variables , altering the behavior of the global sufficient statistic @xmath90 by no more than @xmath163 . in statistical - mechanical terms , this means using renormalization @xcite .
probabilistically , the existence of a limiting ( scaled ) cumulant generating function is a weak dependence condition @xcite , section v.3.2 .
while under equation ( [ eqnscaling - factor ] ) we identified the @xmath52 process with a time - transformed lvy process , now we can only use a central limit theorem to say they are close @xcite , section v.3.1 , reducing almost - sure to stochastic convergence ; see @xcite on the relation between central limit theorems and renormalization . in any event ,
asymptotic scaling of the log partition function implies @xmath120 is consistent .
as mentioned in the , our general results about projective structure in exponential families arose from questions about exponential random graph models of networks . to make the application clear
, we must fill in some details regarding ergms .
given a group of @xmath28 nodes , the network among them is represented by the binary @xmath164 _ adjacency matrix _ @xmath165 , where @xmath166 if there is a tie from @xmath167 to @xmath168 and is 0 otherwise .
( undirected graphs impose @xmath169 . )
we may also have covariates for each node , say @xmath170 .
our projective structure will in fact be that of looking at the sub - graphs among larger and larger groups of nodes .
that is , @xmath6 is the sub - network among the first @xmath28 nodes , and @xmath171 is the sub - network among the first @xmath33 nodes .
the graph or adjacency matrix itself is the stochastic process which is to have an exponential family distribution , conditional on the covariates @xmath172 ( we are only interested in the exponential - family distribution of the graph holding the covariates fixed . ) as mentioned above , the components of @xmath90 typically count the number of occurrences of various sub - graphs or motifs as edges , triangles , larger cliques , `` @xmath0-stars '' ( @xmath0 nodes connected through a central node ) , etc.perhaps interacted with values of the nodal covariates . the definition of @xmath90 may include normalizing the counts of these `` motifs '' by data - independent combinatorial factors to yield densities . a _ dyad _ consists of an unordered pair of individuals . in a dyadic independence model
, each dyad s configuration is independent of every other dyad s ( conditional on @xmath134 ) . in an ergm ,
dyadic independence is equivalent to the ( vector - valued ) statistic @xmath90 adding up over dyads , @xmath173 that is , the statistic can be written as a sum of terms over the information available for each dyad .
in particular , in _ block models _
@xcite , @xmath170 is categorical , giving the type of node @xmath167 , and the vector of sufficient statistics counts dyad configurations among pairs of nodes of given pairs of types .
dyadic independence implies projectibility : since all dyads have independent configurations , each dyad makes a separate additive contribution to @xmath90 .
going from @xmath174 to @xmath28 nodes thus adds @xmath28 terms , unconstrained by the configuration among the @xmath174 nodes .
@xmath90 thus has separable increments , implying projectibility by theorem [ thmprojectible - iff - separable ] .
( adding a new node adds only edges between the old nodes and the new , without disturbing the old counts . )
as the distribution factorizes into a product of @xmath175 terms , each of exactly the same form , the log partition function scales exactly with @xmath175 , and the conclusions of section [ secconsistency - of - mle ] imply the strong consistency of the maximum likelihood estimator .
thus hold , and these models are projective . ]
this result thus applies to the well - studied @xmath176-model @xcite .
typically , however , ergms are _ not _ dyadic independence models . in many networks , if nodes @xmath167 and @xmath168 are both linked to @xmath0 , then @xmath167 and @xmath168 are unusually likely to be directly linked .
this will of course happen if nodes of the same type are especially likely to be friends ( `` homophily '' @xcite ) , since then the posterior probability of @xmath167 and @xmath168 being of the same type is elevated .
however , it can also be modeled directly .
the direct way to do so is to introduce the number ( or density ) of triangles as a sufficient statistic , but this leads to pathological degeneracy @xcite , and modern specifications involve a large set of triangle - like motifs @xcite .
empirically , when using such specifications , one often finds a nontrivial coefficient for such `` transitivity '' or `` clustering , '' over and above homophily @xcite .
it is because of such findings that we ask whether the parameters in these models are projective .
sadly , no statistic which counts triangles , or larger motifs , can have the nice additive form of dyad counts , no matter how we decompose the network .
take , for instance , triangles .
any given edge among the first @xmath28 nodes _ could _ be part of a triangle , depending on ties to the next node . thus to determine the number of triangles among the first @xmath177 nodes , we need much more information about the sub - graph of the first @xmath28 nodes than just the number of triangles among them .
indeed , we can go further .
the range of possible increments to the number of triangles changes with the number of existing triangles .
this is quite incompatible with separable increments , so , by ( [ thmprojectible - iff - separable ] ) , the parameters can not be projective .
we remark that the nonprojectibility of markov graphs @xcite , a special instance of ergms where the sufficient statistics count edges , @xmath0-stars and triangles , was noted in @xcite .
parallel arguments apply to the count of any motif of @xmath0 nodes , @xmath178 .
any given edge ( or absence of an edge ) among the first @xmath28 nodes could be part of such a motif , depending on the edges involving the next @xmath179 nodes .
such counts are thus not nicely additive .
for the same reasons as with triangles , the range of increments for such statistics is not constant , and nonseparable increments imply nonprojective family .
while these ergms are not projective , some of them may , as a sort of consolation prize , still satisfy equation ( [ eqnasymptotic - scaling ] ) .
for instance , in models where @xmath90 has two elements , the number of edges and the ( normalized ) number of triangles or of 2-stars , the log partition function is known to scale like @xmath175 as the number of nodes @xmath180 , at least in the parameter regimes where the models behave basically like either very full or very empty erds
rnyi networks @xcite .
( we suspect , from @xcite , that similar results apply to many other ergms . )
thus , by equation ( [ eqnldp - for - suff - stat ] ) , if we fix a large number @xmath28 of nodes and generate a graph @xmath165 from @xmath181 , the probability that the mle @xmath182 will be more than @xmath183 away from @xmath11 will be exponentially small in @xmath175 and @xmath184 .
since these models are not projective , however , it is impossible to _ improve _ parameter estimates by getting more data , since parameters for smaller sub - graphs just can not be extrapolated to larger graphs ( or vice versa ) .
we thus have a near - dichotomy for ergms .
dyadic independence models have separable and independent increments to the statistics , and the resulting family is projective .
however , specifications where the sufficient statistics count larger motifs can not have separable increments , and projectibility does not hold .
such an ergm may provide a good description of a given social network on a certain set of nodes , but it can not be projected to give predictions on any larger or more global graph from which that one was drawn . if an ergm is postulated for the whole network , then inference for its parameters must explicitly treat the unobserved portions of the network as missing data ( perhaps through an expectation - maximization algorithm ) , though of course there may be considerable uncertainty about just how much data is missing .
specifications for exponential families of dependent variables in terms of joint distributions are surprisingly delicate ; the statistics must be chosen extremely carefully , in order to achieve separable increments .
( conditional specifications do not have this problem . )
this has , perhaps , been obscured in the past by the emphasis on using exponential families to model multivariate but independent cases , as iid models are always projective .
network models , one of the outstanding applications of exponential families , suffer from this problem in an acute form .
dyadic independence models are projective models , but are sociologically extremely implausible , and certainly do not manage to reproduce the data well .
more interesting specifications , involving clustering terms , never have separable increments .
we thus have an impasse which it seems can only be resolved by going to a different family of specifications .
one possibility which , however , requires more and different data is to model the evolution of networks over time @xcite . in particular , hanneke , fu and xing @xcite
consider situations where the distribution of the network at time @xmath185 conditional on the network at time @xmath56 follows an exponential family .
even when the statistics in the conditional specification include ( say ) changes in the number of triangles , the issues raised above do not apply . roughly speaking ,
the issue with the nonprojective ergm specifications , and with other nonprojective exponential families , is that the dependency structure corresponding to the statistics allows interactions between arbitrary collections of random variables .
it is not possible , with those statistics , to `` screen off '' one part of the assemblage from another by conditioning on boundary terms .
suppose our larger information set @xmath36 consists of two nonoverlapping and strictly smaller information sets , @xmath39 and @xmath186 , plus the new observation obtained by looking at both @xmath6 and @xmath187 .
( for instance , the latter might be the edges between two disjoint sets of nodes . )
then the models which work properly are ones where the sufficient statistic for @xmath36 partitions into marginal terms from @xmath6 and @xmath187 , plus the interactions strictly between them : @xmath188 . in physical language
@xcite , the energy for the whole assemblage needs to be a sum of two `` volume '' terms for its sub - assemblages , plus a `` surface '' term for their interface .
the network models with nonprojective parameters do not admit such a decomposition ; every variable , potentially , interacts with every other variable . one might try to give up the exponential family form , while keeping finite - dimensional sufficient statistics .
we suspect that this will not work , however , since lauritzen @xcite showed that whenever the sufficient statistics form a semi - group , the models must be either ordinary exponential families , or certain generalizations thereof with much the same properties .
we believe that there exists a purely algebraic characterization of the sufficient statistics compatible with projectibility , but must leave this for the future .
one reason for the trouble with ergms is that every infinite exchangeable graph distribution is actually a mixture over projective dyadic - independence distributions @xcite , though not necessarily ones with a finite - dimensional sufficient statistic . along any one sequence of sub - graphs from such an infinite graph , in fact , the densities of all motifs approach limiting values which pick out a unique projective dyadic - independence distribution @xcite ; cf .
also @xcite .
this suggests that an alternative to parametric inference would be nonparametric estimation of the limiting dyadic - independence model , by smoothing the adjacency matrix ; this , too , we pursue elsewhere .
for notation in this section , without loss of generality , fix a generic pair of subsets @xmath23 and a value of @xmath11 .
we will write a representative point @xmath189 as @xmath190 , with @xmath119 and @xmath191 . also , we abbreviate @xmath192 , for @xmath119 and @xmath191 by @xmath193 . for clarity ,
we prove the two directions separately .
first we show that projectability implies separable increments .
if the exponential family @xmath194 is projective , then the sufficient statistics @xmath195 have separable increments , that is , @xmath23 implies that @xmath70 .
[ propproj - to - si ] by projectibility , for each @xmath11 , @xmath196 which implies that , for all @xmath119 , @xmath197 re - writing the left - hand side of equation ( [ eqconstant - conditional - partition - function ] ) as a sum over the set @xmath198 of values which the increment @xmath199 to the sufficient statistic might take yields @xmath200 where the joint volume factor is defined in ( [ eqjointvolfactor ] ) . since the right - hand side of equation ( [ eqnconditional - partition - function - as - sum - over - increment - values ] ) is the same for all @xmath69 ,
so must the left - hand side .
observe that this left - hand side is the laplace transform of the function @xmath201 .
the latter is a nonnegative function which defines a measure on @xmath202 , whose support is @xmath198 .
hence , @xmath203 is the laplace transform of a discrete probability measure in @xmath202 .
but the denominator in the inner sum is just @xmath204 , no matter what @xmath69 might be . as a total and not a partial function on @xmath205 .
] so we have that for any @xmath206 , and all @xmath207 , @xmath208 since both sides of equation ( [ eqntwo - laplace - transforms ] ) are laplace transforms of probability measures on a common space , and the equality holds on all of @xmath209 , which contains an open set , we may conclude that the two measures are equal @xcite , theorem 7.3 .
this means that they have the same support , @xmath210 , and that they have the same density with respect to counting measure on @xmath211 . as they also have the same normalizing factor ( viz . , @xmath204 )
, we get that @xmath212 .
since the points @xmath69 and @xmath213 are arbitrary , this last property is precisely having separable increments .
next , we prove the reverse direction , namely that separable increments imply projectibility . this is clearer with some preliminary lemmas .
[ lemmaseparable - increments - implies - volume - factor - separation ] if the sufficient statistics have separable increments , then the joint volume factors factorize , that is , @xmath214 for all @xmath215 , @xmath56 and @xmath65 . by definition ,
@xmath216 when the statistic has separable increments , @xmath217 , so @xmath218 proving the claim .
[ lemmaseparable - volume - factor - implies - projectible - increments ] if the joint volume factor factorizes , then the sufficient statistics has independent increments , and the distribution of the sufficient static is projective . without loss of generality ,
fix a value @xmath56 for @xmath52 and @xmath65 for @xmath76 . by the law of total probability and the definition of the volume factor ,
@xmath219 if the volume factor factorizes , so that @xmath108 , then we obtain @xmath220 \biggl [ \frac{z_{a}(\theta)}{z_{b}(\theta ) } v_{{{b}\setminus{a}}}(\delta ) e^{\langle\theta,\delta\rangle } \biggr].\ ] ] it then follows that @xmath221 and thus that @xmath90 has independent increments . to establish the projectibility of the distribution of @xmath90 , sum over @xmath65 @xmath222 since @xmath223 , and both distributions must sum to 1 over @xmath56
, we can conclude that @xmath224 , and hence that the distribution of the sufficient statistic is projective .
[ lemmamixed - joint - prob - of - obs - and - stat ] if the sufficient statistics of an exponential family have separable increments , then @xmath225 abbreviate @xmath226 by @xmath56 . by the law of total probability , @xmath227
since @xmath118 is sufficient , and @xmath228 for all @xmath229 in the sum , @xmath230 by parallel reasoning , @xmath231 therefore , @xmath232 if the statistic has separable increments , then @xmath233 , and the conclusion follows .
the lemma does _ not _ follow merely from the joint volume factor separating , @xmath234 .
the conditional volume factor must also be constant in @xmath69 .
if the sufficient statistic of an exponential family has separable increments , then the family is projective .
[ propsi - to - proj ] we calculate the marginal probability of @xmath18 in @xmath235 , by integrating out the increment to the sufficient statistic .
( the set of possible increments , @xmath211 , is the same for all @xmath69 , by separability . )
once again , we abbreviate @xmath226 by @xmath56 : @xmath236 these steps use , in succession : lemma [ lemmamixed - joint - prob - of - obs - and - stat ] ; the fact that conditional probabilities sum to 1 ; the projectibility of the sufficient statistics ( via lemmas [ lemmaseparable - increments - implies - volume - factor - separation ] and [ lemmaseparable - volume - factor - implies - projectible - increments ] ) ; and the definition of @xmath237 .
proof of proposition [ propproj - to - ii ] by proposition [ propproj - to - si ] , a projective family has separable increments , and by lemma [ lemmaseparable - volume - factor - implies - projectible - increments ] , separable increments implies independent increments .
proof of proposition [ propincrement - indep - of - old - observable ] by proposition [ propproj - to - si ] , every projective exponential family has separable increments . by lemma [ lemmamixed - joint - prob - of - obs - and - stat ] , in an exponential family with separable increments , @xmath225 therefore , using projectibility , @xmath238 by the definition of @xmath239 , @xmath240 , so @xmath241 but , by lemma [ lemmaseparable - volume - factor - implies - projectible - increments ] , the sufficient statistics have a projective distribution with independent increments , implying @xmath242 therefore , @xmath243 and so @xmath244 .
proof of proposition [ propii - to - volume - factor - separation ] below we prove that if the suffiicient statistics of an exponential family have independent increments , then the volume factor separates , and the distribution of the statistic is projective . since @xmath118 is a sufficient statistic , by the neyman factorization theorem ( @xcite , theorem 2.21 , page 89 ) , @xmath245 in light of equation ( [ eqnexponential - family - density ] ) , we may take @xmath246 .
abbreviating @xmath226 by @xmath56 and @xmath193 by @xmath65 , it follows that @xmath247 by independent increments , however , @xmath248 whence it follows that , for some functions @xmath249 , @xmath250 and @xmath251 and @xmath252 to proceed , we must identify the new @xmath253 and @xmath0 functions . to this end , recalling that @xmath237 is the number of @xmath254 configurations such that @xmath255 , we have @xmath256 and , at the same time , @xmath257 clearly , then , @xmath258 while @xmath259 . since @xmath260 and @xmath261
, we need @xmath262 , and may take @xmath263 for simplicity .
this allows us to write @xmath264 which is exactly the assertion that the volume factor separates .
turning to the @xmath253 functions , we sum over @xmath65 again to obtain the marginal distribution of @xmath52 , @xmath265 now , we finally we use the exponential - family form .
specifically , we know that @xmath266 so that @xmath267 , @xmath268
. therefore , @xmath269 and normalization now forces @xmath270 as desired .
proof of theorem [ thmpredictive - sufficiency ] the conditional density of @xmath107 given @xmath18 is just the ratio of joint to marginal densities ( both with the same @xmath11 , by projectibility ) , @xmath271 which is an exponential family with parameter @xmath11 , sufficient statistic @xmath76 , and partition function @xmath272 .
proof of theorem [ thmldp ] under equation ( [ eqnasymptotic - scaling ] ) , the cumulant generating function also scales asymptotically , @xmath273 .
since @xmath274 is differentiable , the grtner
ellis theorem of large deviations theory @xcite , chapter v , implies that @xmath131 obeys a large deviations principle with rate @xmath126 , and rate function given by equation ( [ eqnrate - function ] ) , which is to say , equations ( [ eqnldp - lower ] ) and ( [ eqnldp - upper ] ) . [ corconditional - moment - generating - function ] the moment generating function of @xmath76 is @xmath275 from the proof of theorem [ thmpredictive - sufficiency ] , @xmath276 has an exponential family distribution with sufficient statistic @xmath76 .
thus we may use equation ( [ eqnmgf - of - exponential - family ] ) to find the moment generating function of @xmath76 conditional on @xmath18 , @xmath277 since , however , @xmath244 ( proposition [ propincrement - indep - of - old - observable ] ) , equation ( [ eqnconditional - moment - generating - function ] ) must also give the unconditional moment generating function .
we thank luis carvalho , aaron clauset , mark handcock , steve hanneke , brain karrer , sergey kirshner , steffen lauritzen , david lazer , john miller , martina morris , jennifer neville , mark newman , peter orbanz , andrew thomas and chris wiggins , for valuable conversations ; an anonymous referee of an earlier version for pointing out a gap in a proof ; and audiences at the boston university probability and statistics seminar , and columbia university s applied math seminar . | the growing availability of network data and of scientific interest in distributed systems has led to the rapid development of statistical models of network structure . typically , however , these are models for the entire network , while the data consists only of a sampled sub - network .
parameters for the whole network , which is what is of interest , are estimated by applying the model to the sub - network .
this assumes that the model is _ consistent under sampling _ , or , in terms of the theory of stochastic processes , that it defines a projective family .
focusing on the popular class of exponential random graph models ( ergms ) , we show that this apparently trivial condition is in fact violated by many popular and scientifically appealing models , and that satisfying it drastically limits ergm s expressive power .
these results are actually special cases of more general results about exponential families of dependent random variables , which we also prove . using such results ,
we offer easily checked conditions for the consistency of maximum likelihood estimation in ergms , and discuss some possible constructive responses . |
one of the main applications of any quantum theory of gravity should be the complete understanding of black hole evaporation .
since such a theory is not yet available , attention has been focused on simpler models whose quantization is expected to be tractable .
most notably among these are the models of two - dimensional dilaton gravity ( see , for example , @xcite for a detailed review ) .
although much insight has been gained into the evaporation process and , in particular , the back reaction of the hawking radiation onto the black hole and the problem of information loss " , a full understanding of the complete , non - perturbative , evolution remains elusive .
the reason for this is that even for such models the full quantum theory is not known . at present ,
the most popular approaches towards the quantization of gravity are superstrings and quantum general relativity . while many of the models discussed are in fact string - inspired " , a non - perturbative understanding of the full string field theory is not in sight .
effort is therefore also concentrated on a more conservative approach : the application of canonical quantization rules to the general theory of relativity .
while this fails to yield , in four spacetime dimensions , a viable theory in a perturbative sense , it might nevertheless make sense non - perturbatively .
even if this were not the case , this framework appears appropriate to exhibit the main physical problems and to suggest methods for their solution .
we shall thus restrict our investigation to this framework , too . what progress has been made towards a quantum theory of black holes within canonical quantum gravity ?
the situation of a spherically symmetric black hole without matter has been completely solved @xcite .
after the embedding variables describing the location of three - dimensional hypersurfaces in spacetime are isolated , the remaining state is a quantum mechanical wave function , @xmath0 , where @xmath1 is the mass of the hole .
the variable conjugate to @xmath1 is related to the time in the asymptotic region .
the fact that no field theoretic degrees of freedom are present reflects of course the situation of the classical theory ( birkhoff s theorem ) . in more complicated models where matter degrees of freedom are present it seems no longer possible to proceed as in @xcite and to isolate the embedding variables from the true " degrees of freedom @xcite .
we shall thus stick to the approach where all degrees of freedom are treated on an equal footing as dynamical variables . instead of finding the full solution in simple situations ,
approximation schemes have been employed in the full theory .
one example is a formal expansion of the full wave functional in powers of the gravitational constant @xmath2 @xcite .
the @xmath2-expansion scheme allows one to derive the limit of quantum field theory in a background spacetime as well as to find quantum gravitational correction terms to this limit .
one application of such correction terms for black holes was made in @xcite where it was shown that some of these terms yield non - unitary contributions which become relevant if the black hole mass approaches the planck mass .
this could be of central importance for the information loss problem .
the investigation in @xcite was , however , only heuristic , since it did not pay attention to problems of regularization .
one of the motivations for the present paper is the attempt to study the semiclassical expansion in the much simpler context of two - dimensional dilatonic gravity .
this will enable us , in particular , to understand the emergence of hawking radiation in this framework , which thus complements the standard derivation with the use of bogolubov coefficients .
furthermore , it will also be straightforward to study the validity of the limit where spacetime is semiclassical and matter fully quantum .
this question received some interest recently because it was found in @xcite that semiclassical gravity breaks down on hypersurfaces which capture both the infalling matter near the horizon and the hawking radiation .
in our paper we shall consider superpositions of different semiclassical components and investigate to which extent they become dynamically independent .
the key ingredient is thereby played by the notion of decoherence - the emergence of quasiclassical properties through the irreversible interaction with the environment ( see , for example , @xcite and the references therein ) . what can possibly play the role of the environment for the black hole ?
it is the hawking radiation itself , which may serve as the decohering agent .
although it may be weak , it possesses a large entropy capacity due to its many degrees of freedom and can thus carry all the information about the superpositions which thereby decohere . our paper is organized as follows . in section 2 we shall review the canonical formalism for the cghs model of dilatonic gravity , in particular the corresponding wheeler - dewitt equation and the quantum momentum constraint . section 3
is then devoted to a detailed discussion of the semiclassical approximation .
the highest order yields the hamilton - jacobi equation for dilatonic gravity without matter fields .
we show how the classical cghs solution can be recovered from a solution to this equation .
we also comment on a connection between this solution and the entropy of the black hole .
we then derive in section 4 the functional schrdinger equation for matter fields on the semiclassical background
. an important part of our paper is then devoted to a derivation of hawking radiation from this schrdinger equation .
we consider both the case of a collapsing geometry and of an eternal black hole .
we shall find that hawking radiation is obtained from a pure quantum state in the sense that the expectation value of the number operator for field quanta exhibits a thermal spectrum with respect to the hawking temperature .
we end this section with a brief discussion of quantum gravitational correction terms to the functional schrdinger equation . in section 5
we discuss in detail how hawking radiation can lead to decoherence for superpositions of black hole states . in particular , we study the superposition of a black hole with a white hole as well as the superposition of a black hole with no hole .
the results then justify the separate consideration of each semiclassical component , since they dynamically decouple from each other .
finally , section 6 contains a brief conclusion as well as an outlook on future work .
in this section we shall apply canonical quantum gravity to spherically symmetric black holes . from the technical point of view
, this lies between the full quantum theory and finite - dimensional minisuperspace models , since the wave functional depends now on one - dimensional _
fields_. an effective two - dimensional theory can thus be obtained by imposing a spherical symmetric ansatz for the classical four - dimensional metric and applying the standard adm procedure for the remaining degrees of freedom .
recently , however , a slightly modified two - dimensional theory has received considerable attention @xcite . apart from being string inspired " , the reason for the choice of this cghs model " is basically simplicity , giving rise to the hope that the full quantum evolution ( including the back reaction of hawking radiation on the gravitational background ) can be tackled in this framework .
for this reason we restrict our attention to this model , too , but keep in mind that this will give at best some hints of how to proceed in the full theory .
the action for the two - dimensional gravity model of cghs coupled to @xmath3 massless scalar fields reads @xmath4 where @xmath5 is the dilaton field , @xmath6 are the matter fields and @xmath2 is the ( unitless ) gravitational coupling constant , which we have introduced for later convenience .
we note that in the corresponding model resulting from dimensional reduction the term @xmath7 is replaced by @xmath8 , where @xmath9 is a relic of the curvature components orthogonal to the two - sphere , which usually is measured in terms of the planck length . in this sense
the two - dimensional cosmological constant term " is an artefact which has its origin in the real " ( four - dimensional ) gravitational constant .
alternatively , @xmath10 may be viewed as being related to the magnetic charge of four - dimensional dilatonic black holes . in order to study the canonical structure of ( [ cghs ] )
, we make the transformation @xmath11 and @xmath12 , which eliminates the kinetic term for the dilaton .
this yields @xmath13 , \label{sk}\ ] ] where we have written out only one of the scalar fields for simplicity .
following @xcite we write the metric in an adm - like parametrization : @xmath14\ , \label{met}\ ] ] where @xmath15 denotes the lapse and @xmath16 the shift function ( note the additional factor @xmath17 in the definition of the metric , which differs from the standard adm convention ) .
the action then reads , upon neglecting surface terms @xmath18 where @xmath19 here , primes and dots respectively denote a derivative with respect to @xmath20 and @xmath21 , while
@xmath22 " is used to specify the location of the constraint surface in phase space .
the momentum and hamilton ian constraints are then given by the expressions @xmath23 where @xmath24 the fields @xmath25 and @xmath26 do nt enter the momentum constraint on the same footing , since the former transforms as a density , whereas the latter transforms as a scalar .
quantization then proceeds la dirac by imposing these constraints as restrictions on physically allowed wave functionals @xmath27 $ ] and assuming the usual equal - time commutation relations , @xmath28=i\delta(x - y)$ ] etc . ,
are obeyed .
note that this may not be consistent , since @xmath26 is restricted , by the above redefinition , to positive values .
affine commutation relations would thus be more appropriate .
however , we do not expect this difference to be relevant for the present discussion which focuses on a semiclassical expansion .
note also that this quantization procedure differs from methods recently employed in this context @xcite , where the constraints are first manipulated on the classical level to isolate explicitly the embedding variables ( which have to be spacetime scalars ) , with respect to which a schrdinger equation is then obtained after quantization . since the supermetric is _ flat _ in these field variables , there is no factor ordering ambiguity in the hamiltonian constraint , and one readily obtains @xmath29 on a first glance , ( [ wdw ] ) seems to describe an almost trivial field theory , the only interaction " coming from the liouville term proportional to @xmath17 .
in particular , the field @xmath30 has totally decoupled from the gravitational fields @xmath25 and @xmath26 , a consequence of the conformal coupling in two dimensions .
however , the presence of the momentum constraints induces a correlation interaction " @xcite which prevents the formulation of solutions separating in the respective variables .
a major problem also seems to be the occurrence of anomalies which spoil the validity of ( [ mc ] ) and ( [ wdw ] ) unless a quantum modification of the hamiltonian is made @xcite .
since this may be of minor relevance for the topic of this paper , these problems will not be discussed here .
we hope , however , to return to these issues in a future publication .
since it is not yet clear how the full equations ( [ mc ] ) and ( [ wdw ] ) can be addressed rigorously , we turn here to a semiclassical expansion scheme .
such a scheme is also interesting on its own , as hawking radiation can be properly understood in that context .
this seems indeed appropriate as long as one focuses on the evolution of quantum fields on a semiclassical gravitational background @xcite .
more precisely , this expansion is of the born - oppenheimer type with respect to the gravitational constant . it is for this purpose that we have introduced the dimensionless constant @xmath2 in ( [ cghs ] ) .
alternatively , one can perform this expansion with respect to the
number @xmath3 of matter fields , where one starts from a liouville - type action in which a nonlocal redefinition of the fields in ( [ cghs ] has been preformed @xcite .
the appropriate parameter there turns out to be @xmath31 . since in @xcite
, @xmath32 takes the same place in the hamiltonian constraint that in our case is taken by @xmath2 , both schemes are formally equivalent .
we choose to perform an expansion with respect to @xmath2 .
we note , however , that this scheme is different from the one where the semiclassical einstein equations ( with the expectation value of the energy - momentum tensor on the right - hand side ) are discussed in the limit of large @xmath3 @xcite .
while this can serve to suppress graviton loops compared to matter loops , the present approximation scheme takes also into account quantum gravitational correction terms , see section 4.4 .
we thus assume an ansatz of the form @xmath33=e^{i({g}^{-1}s_0+s_1+gs_2 + \ldots)}.\ ] ] at order @xmath34 one finds @xmath35 , while at order @xmath36 we obtain : @xmath37 which is the hamilton- jacobi equation for pure gravity .
the solution of ( [ hje ] ) was found in @xcite to be @xmath38 \label{s0}\ ] ] with @xmath39 , where @xmath40 is an integration constant .
in fact , @xmath40 is a constant for the full set of constraints in the _ pure _ gravity case , since it can be shown @xcite that the spatial derivative of the functional @xmath41\equiv e^{-2\rho}\left(\frac{\pi_{\rho}^2 } { 4}-\phi^{\prime 2}\right ) + 4\lambda^2\phi \label { ob}\ ] ] is equal to a linear combination of the constraints .
note that eq.([s0 ] ) obeys @xmath42 and @xmath43 .
moreover , @xmath40 commutes with all constraints and can thus be interpreted as an observable " in fact , it is proportional to the adm mass ( see below ) .
an analogous quantity appears in the reduced models from four dimensions @xcite . in the presence of matter fields ,
the spatial derivative of the functional @xmath40 no longer vanishes on the constraint surface .
therefore , the general approach to reduction made in @xcite is no longer applicable here , see @xcite .
it is also easily checked that @xmath44 obeys the momentum constraint in this order of approximation ( since this constraint does not contain @xmath2 , its expansion is straightforward ) .
note that @xmath44 becomes imaginary if either @xmath45 or @xmath46 ( both conditions can not be satisfied simultaneously ) .
for the black hole solution discussed below , the latter condition describes the region inside the horizon .
we now recover explicitly the classical solutions for the conformal factor of the metric as well as for the dilaton , @xmath47 and @xmath48 , from the solution ( [ s0 ] ) to the hamilton - jacobi equation .
the momenta are given in ( [ mom ] ) , and using @xmath49 for @xmath50 and @xmath26 one has @xmath51 it turns out to be convenient to work in the the conformal gauge , i.e. we set @xmath52 and @xmath53 . using in addition lightcone variables with @xmath54 and squaring ( [ fd ] ) yields @xmath55 now using ( [ fd ] ) in ( [ rd ] )
gives @xmath56 leading to @xmath57 while @xmath58 and @xmath59 of eq .
( [ 1cond ] ) gives @xmath60 there are a priori two ways to solve ( [ cat ] ) , ( [ cinq ] ) , and ( [ six ] ) .
the first is to set @xmath61 and @xmath62 , corresponding to static solutions @xmath63 .
the second is to assume that the quantities in each of the parenthesis vanishes .
this is what we now do ; it will become clear that this assumption does not exclude the static solutions .
thus , @xmath64 or @xmath65 here , the functions @xmath66 and @xmath67 are non - vanishing and have opposite signs , but are otherwise arbitrary . making use of ( [ 30 ] ) , the solution for @xmath26 in ( [ 1cond ] ) may be written @xmath68 where @xmath69 is a ( negative ) constant . now inserting ( [ 33 ] ) and
( [ 30 ] ) into ( [ 1cond ] ) shows that for given @xmath66 and @xmath67 , the conformal factor takes the form of a product @xmath70 but within the conformal gauge , a change of coordinates @xmath71 and @xmath72 can always be made and the conformal factor then undergoes the transformation @xmath73 . in view of (
[ 60 ] ) , it is clear that the various @xmath66 and @xmath67 simply correspond to different choices of coordinates . in particular
, one can pick the kruskal gauge by arranging for @xmath25 to vanish . in that case ,
( [ 29 ] ) yields @xmath74 where @xmath75 s are constants .
inserting this into ( [ 1cond ] ) , one finds @xmath76 and @xmath77 . by an appropriate translation along @xmath78 and @xmath79 , @xmath80 and @xmath81
can always be set to zero .
a general solution is thus @xmath82 when reverting back to the initial variables , this is ( as expected ) the cghs @xcite black hole solution @xmath83 with @xmath84 and @xmath85 , with an adm mass @xmath86 . that this quantity is the adm mass can be understood from an analysis of the hamiltonian @xcite .
a special case is the linear dilaton vacuum solution " ( ldv ) which is obtained for @xmath87 .
clearly , a different choice of coordinates will lead to @xmath88 and @xmath89 . from ( [ 30 ] )
one recognizes that these are the static solutions which were obtained in ref .
@xcite to prove the validity of birkhoff s theorem in two - dimensional dilaton gravity . evaluating the observable ( [ ob ] ) for the cghs solution
derived above just yields the constant @xmath40 . in the following we evaluate hamilton s principal function ( [ s0 ] ) on a spacetime that is recovered from the hamilton - jacobi equation ( [ hje ] )
this is interesting because one can discuss a relation to the entropy of the black hole . in the conformal gauge we have with ( [ fd ] ) and ( [ rd ] ) @xmath90 \ , \label{s0c}\ ] ] which only depends on the dilaton field . for the cghs solution ( [ bhs ] ) ,
( [ s0c ] ) reads @xmath91 \label{s0cghs}\ ] ] interestingly , the real part of ( [ s0cghs ] ) is vanishing ( contributions from interior and exterior cancel one another separately for @xmath92 and @xmath93 ) .
the imaginary part has been interpreted in @xcite as being proportional to the entropy of the black hole .
we note from ( [ s0c ] ) that @xmath94 since this yields a non - vanishing value only if @xmath95 , i.e. , inside the horizon , one has @xmath96 how can this result be reconciled with the non - vanishing entropy of the hole ?
the contribution to @xmath97 from each horizon is @xmath98 which is one quarter of the hole s entropy .
the result in @xcite will thus be obtained if only the crossing point of the hypersurface @xmath99 with the upper branch of the future horizon is taken into account .
all this is independent of the specific hypersurface chosen .
however , crossing to the inside of the line of singularity @xmath100 should not be done , as it implies an imaginary value for the physical field @xmath5 .
thus , one should in principle keep @xmath101 . for @xmath102 , integrating for positive @xmath20 down to the line of singularity @xmath100 would give a single horizon crossing , with now @xmath103 , but of course with a non - vanishing real part .
to get an idea of how general the vanishing of the real part @xmath104 is , we consider a different model . for four - dimensional spherically
symmetric gravity , where the angular part is frozen , the corresponding two - dimensional dilaton gravity action is similar to ( [ sk ] ) , but with a modified potential @xmath105 .
the associated black hole solution of mass @xmath106 is @xcite @xmath107 \nonumber \\ \relax & = & -2 m e^{-\frac{r}{2m}}d\bar{u}d\bar{v}\ . \label{ssg}\end{aligned}\ ] ] in the last equality for the metric , we made use of the usual kruskal coordinates , @xmath108 and @xmath109 , where @xmath110 and @xmath111 and @xmath112 is the usual tortoise coordinate .
the expression for @xmath44 evaluated on a slice of constant kruskal time @xmath113 , defined from the null coordinates through @xmath114 , is ( we note that @xcite gives the solution to the hamilton - jacobi equation ( [ hje ] ) for _ arbitrary _ potentials ) @xmath115 \ , \label{s0ssg}\ ] ] in close resemblance with ( [ s0cghs ] ) .
however , the additional exponential factor makes @xmath104 non - vanishing , as can easily be checked numerically .
at the next order , @xmath116 , we have @xmath117 we now follow the four - dimensional case @xcite and set @xmath118 \chi [ \rho,\phi , f]$ ] so that ( [ g0 ] ) can be written as @xmath119=0 .
\label{dc}\end{aligned}\ ] ] this can be further simplified if @xmath120 is assumed to obey the equation @xmath121 the first term in ( [ d ] ) is formally infinite since it involves functional derivatives at the same point .
some authors have argued that , after suitable regularization , it can be ignored @xcite and that one may solve by @xmath122 as if there were no matter . however , as already mentioned above , this can at best be considered ad hoc , but for the purpose of the present paper it is not necessary to resolve this issue . we are thus left with @xmath123 where @xmath124 is the matter hamiltonian density , @xmath125 .
use of ( [ tse ] ) may appeal to either the first or second equalities in ( [ fd ] ) and ( [ rd ] ) . in the next section ,
we make use of the second equalities to discuss a difficulty in identifying the lhs of ( [ tse ] ) as a tomonaga - schwinger type time . here , we derive results for a classical background , making use of the first equalities . in the conformal gauge ( @xmath52 and @xmath53 ) , ( [ tse ] )
takes the simple form @xmath126 which can be immediately integrated to yield @xmath127 with @xmath128 .
this is the functional schrdinger equation for a free scalar field on a flat background .
the only information about the gravitational fields occurs in the definition of time through @xmath25 and @xmath26 .
note that the use of the original , physical , field variables yields an equation whose form is equivalent to that of ( [ tse2 ] ) : @xmath129 the reason for this formal similarity is of course the conformal coupling of the matter field in two dimensions .
integrating ( [ tse4 ] ) gives again the free evolution ( [ tse3 ] ) , but this time on the black hole background ( [ 43 ] ) .
as it has been shown recently @xcite , replacing the left - hand side of ( [ tse ] ) with a functional derivative @xmath130 is not consistent , since the object on the left - hand side does not commute for two different space points .
in fact , it was found that this commutator on both sides of ( [ tse ] ) amounts to the invariance of @xmath131 under spatial diffeomorphisms . in our simple two - dimensional model , this
can easily be shown explicitly .
denote the vector fields one vector field at each space point @xmath20 acting on the left - hand side of ( [ tse ] ) by @xmath132 and consider their smeared - out version @xcite @xmath133 consistency of ( [ tse ] ) then demands that @xmath134\chi=[h_m^m , h_m^n]\chi,\ ] ] where , of course , @xmath135 .
explicit calculation yields for the left - hand side @xmath134\chi = \int dx(nm'-mn')\left(\phi'\frac{\delta}{\delta\phi } + \rho'\frac{\delta}{\delta\rho}-\frac{d}{dx}\frac{\delta } { \delta\rho}\right)\chi,\ ] ] while for the right - hand side one finds @xmath136\chi = \int dx(mn'-nm')f'\frac{\delta\chi}{\delta f}.\ ] ] this , however , is nothing but the momentum constraint ( [ mom ] ) in this order of approximation . as has been emphasized in @xcite , equations like ( [ tse ] ) have to be properly interpreted in their integrated form , as a functional schrdinger equation , after a specific choice for lapse and shift has been made , so as to recover quantum field theory on a specific family of spacetimes . this does not , however , mean that the local form of the schrdinger equation ( the tomonaga - schwinger equations ) is useless , since it may be used for formal considerations , for example the discussion of the correction terms in section 4.5 .
we now show how to recover hawking radiation from solutions to ( [ tse3 ] ) .
this is complementary to the standard analysis employing the calculation of bogolubov coefficients @xcite in that the use of the schrdinger picture brings out interesting new aspects . rather than an eternal black hole ( [ 43 ] ) , we first consider the ` collapsing ' background of @xcite .
the formally analogous case of the eternal black hole ( [ 43 ] ) will be treated in the next subsection .
although ( [ tse3 ] ) was obtained in the context of the eternal black hole , it is clear from the point of view of quantum field theory on curved geometries and from conformal invariance that it also governs the evolution on an arbitrary two - dimensional background , where the variable @xmath21 stands for the time variable of the conformally flat metric .
the collapsing spacetime we are interested in can be obtained from the action ( [ cghs ] ) by assuming that a left moving shock wave of classical @xmath30 - matter is imparted ( for example ) at @xmath137 , producing a stress tensor @xmath138 , so as to form the black hole . in the kruskal gauge , where @xmath139 , one then has the background @xmath140 with @xmath141 where @xmath142 is the usual step function .
the corresponding penrose diagram is presented in figure 1 .
[ fig1 ] one might wonder how a nontrivial effect such as hawking radiation can emerge from ( [ tse3 ] ) , since this equation has the form of a free schrdinger equation .
the reason lies in the proper formulation of _ boundary conditions_. the idea is to start from the vacuum state for the scalar field in the absence of a black hole at early times ( the linear dilaton vacuum " region , ldv ) , let it then evolve according to ( [ tse3 ] ) and compare it with the vacuum solution in the presence of a black hole at late times .
the notion vacuum state " is here defined with respect to inertial " coordinates , i.e. , coordinates which exhibit explicitly the asymptotic flatness of the metric . for ( [ cghsbg ] ) ,
such coordinates are respectively @xmath143 from these lightcone coordinates , timelike and spacelike directions are defined as usual via @xmath144 and @xmath145 .
note that @xmath146 only covers the region above the shock wave , @xmath147 , and outside the horizon , @xmath148 .
the comparison of the vacuum states will be made on the ( overlapping ) slices @xmath149 . in order to obtain such overlapping slices ,
the coordinate transformation chosen in ( [ coor ] ) is slightly different from a similar one used in the literature , see @xcite . as @xmath150 on the half - line @xmath151 , we must demand the boundary condition @xmath152 as @xmath153 for the field @xmath30 because otherwise one would not expect the gaussian functionals which play the role of vacuum states ( see below ) to converge . in the coordinate @xmath154 this condition
means that @xmath30 has to _ vanish _ at the origin @xmath155 .
this turns out to be crucial for the following discussion .
it is convenient to consider the fourier transform of the fields . because of the boundary condition we have @xmath156 where @xmath157 ( reality of @xmath158 ) . in the following
we shall omit the tilde for simplicity and denote the fourier transform by @xmath159 .
it follows from ( [ fx ] ) that @xmath160 the functional schrdinger equation ( [ tse3 ] ) then reads @xmath161 we shall choose the ground state solution of this equation . in the present case
this can be represented by a gaussian functional ( see for example @xcite ) , and we have @xmath162 the ground state energy @xmath163 is of course divergent and must be regularized .
we shall choose the ground state solution ( [ gr ] ) as the _ initial condition _ for the schrdinger equation in the ` future ' variables @xmath164 .
this equation reads @xmath165 ^ 2\right)\chi_b= i \frac{\partial\chi_b}{\partial t_v } , \label{til}\ ] ] where the subscript @xmath166 " refers to the black hole region " .
the important difference to above is that @xmath167 has no restriction at @xmath168 .
therefore the fourier transform is simply given by @xmath169 which has to be contrasted with ( [ fx ] ) .
in that case , @xmath170 thus , eq . ( [ til ] ) becomes @xmath171 note that @xmath172 is _ complex _ , whereas @xmath159 is real . at @xmath149
, the space slices overlap and the states may be compared .
the ground state solution to ( [ schg ] ) at @xmath173 reads @xmath174 this is of course different from the previous ground state solution ( [ gr ] ) . to use ( [ schg ] )
we have to rewrite the ground state ( [ gr ] ) in terms of the field @xmath172 .
we thus write a bogolubov relation @xmath175 and therefore ( using ( [ fx ] ) and ( [ gk ] ) ) @xmath176 one can see by explicit calculation that this can be satisfied by the ansatz @xmath177 with @xmath178 being the coordinate transformation ( [ coor ] ) at @xmath149 .
this leads to @xmath179 in the second step of ( [ alpha ] ) , the standard approximation @xmath180 was made , picking up the dominant contribution in the neighborhood of the horizon @xcite , and 3.763.1 of @xcite was used in the last step . taking the state ( [ gr ] ) at @xmath181 and expressing it with respect to the field @xmath172 , one finds @xmath182 which is independent of the black hole mass @xmath1 .
because of the analogy to the rindler case , namely the loss " of the information on half the space slice behind the horizon , this is in accordance with a similar result found in @xcite in the context of accelerated observers and the unruh radiation . to solve ( [ schg ] ) with the initial condition ( [ coth ] ) , it is again appropriate to make a gaussian ansatz : @xmath183 inserting this into ( [ schg ] ) leads to @xmath184 the solution of this equation with the initial condition @xmath185 is simply given by @xmath186 we now consider the number operator of the mode with wave number @xmath187 , which is associated with the vacuum state ( [ vac ] ) in the presence of the hole . calculating its expectation value with respect to the state ( [ chi1 ] ) the time - developed state of the free " vacuum
one finds @xmath188 where @xmath189 ( @xmath190 ) denotes the real ( imaginary ) part of @xmath191 . at @xmath173
this is given by @xmath192 i.e. a planck distribution with the temperature @xmath193 , as expected .
employing the differential equation ( [ dom ] ) one recognizes that @xmath194 , i.e. the planck spectrum is conserved in time .
it is important to emphasize that the planck spectrum has been recovered from a _ pure _ quantum state .
this means that there exist other operators than the number operator , from which it is possible to recognize explicitly the difference to a thermal state .
note also that the density matrix of a canonical ensemble of oscillators contains some extra - term besides the @xmath195-term showing up in ( [ coth ] ) .
let us now consider how hawking radiation may arise in the eternal black hole background ( [ 43 ] ) ( which due to its time symmetry would be more properly called a black - and - white hole " ) .
the derivation in fact closely parallels the one of the collapsing hole , so a brief presentation will suffice . to describe the radiation escaping to the rhs of the black hole , we introduce the coordinates @xmath196 which cover the wedge @xmath197 of the manifold .
the metric ( [ 43 ] ) then takes the form @xmath198 which is thus asymptotically flat far from the black hole .
it is clear that the slices @xmath199 and @xmath200 coincide , so that states evolving in @xmath201 and @xmath202 may be compared there .
as before , we then have the feature that the range @xmath203 only covers the half - line @xmath204 . as a result , requiring the matter field configuration @xmath205 to vanish at spatial infinity
@xmath206 will translate into @xmath207 .
note the analogy of ( [ ttxt ] ) to the transformation between minkowski coordinates and rindler coordinates , from which it is immediately clear that the hawking temperature here is @xmath208 , independent of the mass .
this has recently also been emphasized in @xcite .
we first determine the vacuum state @xmath209 $ ] which is a solution of the schrdinger equation ( [ tse3 ] ) in @xmath201 coordinates ( @xmath159 being the fourier tranform of @xmath210 as in ( [ fx ] ) .
although the metric is not asymptotically flat in those coordinates , a vacuum state may still be defined . as we do not wish to involve arbitrarily short distance physics , we choose for initial cauchy surface a slice @xmath211 , but @xmath212 .
following essentially the same steps as before , @xmath209 $ ] is just the state ( [ gr ] ) ( with the change @xmath213 ) .
the state will then evolve up to @xmath214 .
for the evolution in terms of the @xmath202 coordinates , the transformation ( [ ttxt ] ) is conformal , so ( [ tse3 ] ) still applies and the corresponding vacuum state at @xmath199 , @xmath215 $ ] , is thus the same as ( [ vac ] ) . to determine how @xmath216 $ ] reads in terms of @xmath172 , we write a bogolubov relation as in ( [ bogor ] ) , and find the exact result @xmath217 then , following steps analogous to those leading to ( [ num ] ) from ( [ coth ] ) , we obtain that the state thus evolved according to time @xmath218 , @xmath219 $ ] , has indeed a thermal content in the sense of ( [ num ] ) .
we also note that the non - vanishing of @xmath97 ( compare ( [ bh ] ) ) is easily understood for the eternal hole .
the integral in ( [ bh ] ) has to be performed with respect to @xmath220 which ranges from @xmath221 to @xmath222 along the half - line originating from the origin in the @xmath201-diagram .
since it has only one point in common with the horizon , one directly finds the desired result ( [ i m a ] ) .
we now proceed to the next order of the semiclassical approximation , @xmath223 , to show how the functional schrdinger equation is modified by quantum gravitational corrections @xcite .
one first obtains an equation involving @xmath224 , @xmath225 as in the general , four - dimensional , case , this can be greatly simplified by rewriting @xmath226 + \eta[\rho,\phi , f]$ ] and demanding , in analogy to ( [ d ] ) , an equation for @xmath227 such that the equations simplify .
since these steps are in full analogy to the general case , we shall be very brief here . introducing a functional @xmath228 one finds for @xmath229 the corrected schrdinger equation " @xmath230 if @xmath122 , the only correction term reads @xmath231 which , of course , involves second order derivatives with respect to the gravitational degrees of freedom .
if @xmath131 is a solution to ( [ tse3 ] ) , one would expect that it separates in @xmath30 and the gravitational variables . in this case
the correction term would only yield a contribution to the phase , which should not be important .
however , as the analysis in @xcite suggests , this term should become relevant if the mass of the hole approaches the planck mass . to evaluate the correction terms properly
, one has first to employ a careful regularization , since second functional derivatives are involved .
since this is beyond the scope of this paper , it will be relegated to a future publication .
the derivation of the functional schrdinger equations ( [ tse3 ] ) was achieved through the use of a particular wkb state for the gravitational sector .
namely , the choice @xmath232 and @xmath233 was made ( although the same equation can equally be obtained with @xmath234 and @xmath235 ) .
but since the fundamental equations which are assumed here , ( [ mc ] ) and ( [ wdw ] ) , are _ linear _ , one would , however , expect that arbitrary superpositions of wkb - type states occur .
this is suggested also by the real nature of the wheeler - dewitt equation , from which it would seem artificial to choose a special complex solution .
a more natural state would thus be @xmath236 where @xmath237 is the state which evolves out of the eternal black hole geometry and reads @xmath238 the state ( [ bwh ] ) could naively be called a superposition of a black hole with a white hole " .
such states arise , for example , in the analysis of the gravitational collapse of a dust shell with a sensible boundary condition for the wave function @xcite . another possibility would be @xmath239 where the second component refers to the linear dilaton vacuum ( thus @xmath87 ) , i.e. @xmath240 one can interpret the state ( [ bnh ] ) as a superposition of a black hole with no hole .
states of this kind have in fact been discussed extensively in qed and quantum cosmology @xcite .
it has been demonstrated there that the presence of a huge number of irrelevant " degrees of freedom may cause decoherence of such states . in the case of qed
these may be charged matter states , while in quantum cosmology these may be general matter states or states describing gravitational waves . because states such as ( [ bwh ] ) are found in a born - oppenheimer approximation , the decoherence of the various components is nothing but an expression of spontaneous symmetry breaking as it happens , for example , in the case of chiral molecules @xcite .
what could possibly play the role of the decohering agent in the present case ? from the discussion in the preceding section
one would expect that hawking radiation may be able to suppress interferences between the separate wkb - states .
this is in particular suggested by its irreversible nature and the fact that it provides a huge entropy capacity . writing @xmath241 as a sum of four terms ( @xmath242 being the superposition ( [ bwh ] ) ) , the degree of decoherence between the two semiclassical components can be studied from the following component of the reduced density matrix for the gravitational sector @xmath243 = e^{2is_0 } \int { \cal d } g \ : { \cal d } g^ * \bar{\chi}_r^2[g , g^{*},\tilde{t})\equiv e^{2is_0}d_{\pm } , \label{rho}\ ] ] and its conjugate @xmath244 .
because of the quadratic dependence on the field @xmath245 , the trace in ( [ rho ] ) can immediately be evaluated , in full analogy to the qed - case discussed in @xcite .
one has , with @xmath246 given in ( [ om ] ) , @xmath247 the decoherence factor can be be evaluated by noting that the real and imaginary parts of @xmath191 are given , respectively , by the expressions @xmath248 which yields @xmath249 one thus has for the exact decoherence factor the expression @xmath250\right ) , \label{ex}\ ] ] where @xmath251 was used , and @xmath252 is a regularization length .
the important part for the amount of decoherence is its absolute value , @xmath253\right ) .
\label{ab}\ ] ] since this can not be exactly evaluated , the expansion of the exponent made in ( [ dpm ] ) appears appropriate . considering the first real term in the exponent
, one finds @xmath254\right ) , \label{dec } \end{aligned}\ ] ] where 3.986.4 in @xcite was used in the last step .
a finite value for @xmath252 is obtained , if the black hole states are put into a box of finite length , as is frequently done in gedankenexperiments . in the limit @xmath255
this goes to one as it must , since there we matched our state to the one for the dilaton vacuum which has @xmath256 . in the limit @xmath257
this would yield the simple expression @xmath258 .
one must , however , note that higher order terms in the expansion ( [ dpm ] ) become important in this limit and one has to resort to the exact expression ( [ ab ] ) . differentiating ( [ ab ] ) with respect to @xmath218 yields an expression which vanishes ( due to riemann s lemma ) in the limit of large @xmath218 .
the decoherence factor
( [ ab ] ) itself thus approaches a constant .
it is convenient to write @xmath259 where @xmath260 .
simple numerical analysis yields a limiting value for @xmath261 of about @xmath262 which is rapidly obtained for @xmath154 bigger than about @xmath263 , as depicted in figure 2 .
decoherence thus becomes effective for @xmath264 .
this can be physically interpreted as follows .
the dominating mode for the hawking temperature @xmath265 has a wavelength of the order @xmath10 .
if the black hole is put into a box with radius @xmath252 , interferences are important if @xmath252 is of the same order than this wavelength , but they become suppressed if the box is larger .
[ fig2 ] one can now calculate the analogous expression for the superposition ( [ bnh ] ) of a black hole with no hole . instead of ( [ rho ] )
one now has to calculate @xmath266 = e^{is_0^{(1)}-is_0^{(0 ) } } \int { \cal d } g \ : { \cal d } g^ * \bar{\chi}_r\bar{\chi}_0^*\equiv \
e^{is_0^{(1 ) } -is_0^{(0)}}\ \tilde { d}_{\pm } , \label{rho1}\ ] ] inserting the gaussian functionals @xmath237 and @xmath267 into this expressions , one recognizes that now @xmath268 $ ] , where @xmath269 and @xmath270 with @xmath189 and @xmath190 given by the expressions ( [ ome ] ) . in this example one then finds @xmath271 this then leads instead of ( [ f ] ) to an expression for the decoherence factor of the form @xmath272 where @xmath273 one has @xmath274 for all @xmath154 because only one component in ( [ bnh ] ) carries hawking radiation .
again , @xmath275 approaches a constant for large @xmath154 , but in contrast to @xmath261 it is not monotonic . a numerical evaluation yields , for example , the value @xmath276 ( compared to @xmath277 ) .
since the state @xmath237 does not depend on the black hole mass , there is of course no decoherence for a superposition corresponding to different masses , in contrast to the case of accelerating detectors@xcite .
this is , however , a peculiarity of the two - dimensional model . in four dimensions ,
the hawking temperature is inversely proportional to the mass , so one would expect decoherence for not - to - large masses .
heuristically , one can take this into account by replacing @xmath278 in the above expressions for the states by @xmath279 .
one would thus expect the decoherence factor in four spacetime dimensions to be of the limiting form for large times @xmath280 this expression depends now explicitly on the mass of the hole , and one recognizes that decoherence is efficient for small masses where the hawking radiation is large .
again , decoherence for a black hole in a box of dimensions @xmath281 would be efficient for the realistic case of the length @xmath252 being much bigger than the schwarzschild radius .
the central issue in our paper is the investigation of the semiclassical approximation in the context of two - dimensional dilaton gravity .
we have demonstrated how hawking radiation can be properly understood in the schrdinger representation and how the validity of the semiclassical approximation can be investigated .
we have shown , in particular , how the correlation between the hawking radiation and the black hole can decohere the latter the various semiclassical components become dynamically independent .
we want to conclude in mentioning some of the interesting open problems which are topics for future research . hawking radiation
was obtained in our framework from a _ pure _ quantum state outside the horizon . while the expectation value of the number operator for the field modes exhibits a perfect planckian spectrum , there are of course higher order operators which distinguish this state from a thermal state .
the occurrence of such a pure state can be understood from the analogy with the rindler case , where boundary conditions corresponding to the presence of a mirror at the origin can be posed .
alternatively , one can consider a quantum state on the complete manifold @xcite .
tracing out the degrees of freedom of the left wedge would then lead to the well - known thermal density matrix in the right wedge .
the derivation of the decoherence factor would proceed along the lines described in section ( [ deco ] ) .
another important problem is the possible occurrence of anomalies which could spoil the whole semiclassical limit @xcite .
this of course requires a full understanding of regularization in quantum gravity .
a proper regularization is also needed for a rigorous evaluation of the correction terms to the schrdinger equation , which have been derived in section 4.5 .
maybe one can make use of the methods developed in @xcite and @xcite where the attempt is made , in the context of a strong coupling expansion , to define the kinetic terms rigorously .
one can also try to modify the expansion scheme itself @xcite .
we also wish to comment briefly how our results are related to those presented in @xcite . there
, a vacuum matter state was freely evolved on the geometry of figure 1 for a black hole of mass @xmath1 and then compared , through an appropriate scalar product , to the same evolution on a black hole of mass @xmath282 .
it was claimed there that a value of @xmath283 of the order of @xmath284 would lead to a vanishing product , signalling a breakdown of the semiclassical approximation ( if @xmath278 is taken to be proportional to the inverse planck length ) .
first we note that considering a scalar product is essentially what was done in section 5 , although there we have not considered superpositions with different masses .
but since the matter states describing hawking radiation are identical for two different black hole masses , we would reach the opposite conclusion of an absence rather than an excess of decoherence for superpositions of different masses .
it should be remembered , however , that the foliation chosen is quite different . in our case , before reaching the horizon , our spacelike slices extend to the left spatial infinity , while for times after the shock wave , the matter evolution has support on the coordinate system @xmath146 which only covers the region outside the horizon ( analogous to the unruh effect ) . in this way , the aspect of information for half the line being ` lost ' behind the horizon is captured . in @xcite , the matter states are compared on slices that cross the shock wave and the horizon .
given the different slicing studied , further work would be needed to clarify how the two approaches are related . from a practical point of view , we note the relative simplicity of our calculation which makes use of the schrdinger picture througout . 99 a. strominger , les houches lectures on black holes " , hep - th/9501071 .
k. v. kucha , _ phys .
* d50 * ( 1994 ) 3961 . h. a. kastrup and t. thiemann , _ nucl .
_ b * 425 * ( 1994 ) 665 .
j. d. romano , spherical symmetric scalar field collapse : an example of the spacetime problem of time " , gr - qc/9501015 . c. kiefer , r. mller , and t. p. singh , _ mod .
* a9 * ( 1994 ) 2661 .
e. keski - vakkuri , g. lifschytz , s. d. mathur , and m. e. ortiz , _ phys .
_ * d51 * ( 1995 ) 1764 .
d. cangemi , r. jackiw , and b. zwiebach , physical states in matter - coupled dilaton gravity " , to appear in _ annals of physics_. s. p. de alwis and d. a. macintire , _ phys .
_ * d50 * ( 1994 ) 5164 .
d. giulini and c. kiefer , _ class .
quantum grav . _
* 12 * ( 1995 ) 403 .
s. giddings and w. nelson , _ phys .
* d46 * ( 1992 ) 2486 .
r. jackiw , in _ field theory and particle physics _ , edited by o. eboli , m. gomes , and a. santano ( world scientific , singapore , 1988 ) . i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series , and products _
( academic press , orlando , 1980 ) .
s. w. hawking , _ commun .
* 43 * ( 1975 ) 199 .
k. freese , c. t. hill , and m. mueller , _ nucl .
_ * b255 * ( 1985 ) 693 .
m. cadoni and s. mignemi , _ phys .
lett . _ * 358b * ( 1995 ) 217 .
c. kiefer and t. p. singh , _ phys .
rev . _ * d44 * ( 1991 ) 1067 .
p. hjek,_commun . math .
phys . _ * 150 * ( 1992 ) 545 . c. kiefer , _ phys .
* d46 * ( 1992 ) 1658 .
h. d. zeh , in _
stochastic evolution of quantum states in open systems and measurement processes _ , edited by l. disi and b. lukcs ( world scientific , singapore , 1994 ) .
t. horiguchi , k. maeda , and m. sakamoto , _ phys .
lett . _ * 344b * ( 1995 ) 105 .
demers , _ mod .
lett . _ * a10 * ( 1995 ) 1745 .
p. mansfield , _ nucl .
* b148 * ( 1994 ) 1130 .
s. p. kim , _ phys .
rev . _ * d52 * ( 1995 ) 3382 . | we discuss in detail the semiclassical approximation for the cghs model of two - dimensional dilatonic black holes .
this is achieved by a formal expansion of the full wheeler - dewitt equation and the momentum constraint in powers of the gravitational constant . in highest order
, the classical cghs solution is recovered .
the next order yields a functional schrdinger equation for quantum fields propagating on this background .
we show explicitly how the hawking radiation is recovered from this equation .
although described by a pure quantum state , the expectation value of the number operator exhibits a planckian distribution with respect to the hawking temperature .
we then show how this hawking radiation can lead to the decoherence of black hole superpositions .
the cases of a superposition of a black hole with a white hole , as well as of a black hole with no hole , are treated explicitly .
decoherence of black holes by hawking radiation 0.3truein jean - guy demers + + + claus kiefer + + + 0.5truein .4truein .5 in |
the central density profile of dark matter halos around galaxies and galaxy clusters is now highlighted in the dark matter cosmology .
@xcite have performed cosmological n - body simulations based on cold dark matter models and show that the equilibrium density profiles of dark matter halos have self - similar profiles . the profile ( hereafter , nfw profile )
is thought to be the final relaxed state of self - gravitating collisionless particles and thereby called a universal profile .
the nfw profile has a central cusp diverging as @xmath4 .
recent simulations with higher resolution have confirmed existence of a universal profile but show that the central density profile gets steeper and is better represented by @xmath5 @xcite although slight variation of the central density profiles from halo to halo is reported @xcite .
rotation curve measurements for dwarf galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) have shown that central density profiles are flat and a density profile with a central cusp is inconsistent with the mass distribution of dwarf galaxies . to solve the discrepancy between the theoretical prediction and the observations for dwarf galaxies , @xcite have proposed that the dark matter has a finite cross - section to allow strong self - interactions between the dark matter particles themselves .
this type of dark matter is called self - interacting dark matter ( hereafter , sidm ) .
the introduction of the self - interaction may reduce the central mass concentration found in the universal profile obtained by pure collisionless dark matter simulations .
therefore , the precise measurement of central density profiles of dark matter halos is one of the key questions to solve the nature of the dark matter . both observational and theoretical studies
have been done to constrain the sidm cross - section .
higher resolution measurements of rotation curves of dwarf galaxies ( e.g. * ? ? ?
* ) have rejected the central profile of @xmath6 although the data can not discriminate whether the central density profiles have constant density cores or @xmath7 cuspy profiles .
miralda - escude ( 2000 ) has shown that ellipticity of the dark halo core of the lensing cluster is rather high using the strong lensing data .
he has proposed to use the measured ellipticity to give an upper limit on the sidm cross - section . @xcite
have performed n - body simulations of cluster formation with various values of sidm cross - sections .
they show that a few collisions per particle per hubble time at the halo center can substantially affect the central density profile .
therefore , precise measurements of central density profiles can provide a strong constraint on the sidm cross - section .
we examine how a number ratio of radial arcs to tangential arcs depends on the central density profile of galaxy clusters .
a radial arc is a gravitationally lensed image whose major axis lies in the radial direction of a cluster - lens potential .
a tangential arc is a gravitationally lensed image whose major axis lies in the tangential direction of a cluster - lens potential .
we define ` radial arc statistics ' as a statistical average of number ratios of radial arcs to tangential arcs found in a certain cluster sample .
recently , @xcite have proposed an arc statistics for probing the central density profile of galaxy clusters . in their arc statistics , the absolute frequency of finding arcs appeared in a cluster must be used . however , the absolute frequency of finding arcs in a certain cluster sample is known to be sensitive to yet uncertain parameters e.g. evolution of background galaxies and details of mass distribution of galaxy clusters @xcite .
no model has succeeded to reproduce the observed high frequency of finding arcs yet . as shown in this paper ,
the proposed radial arc statistics is free from the systematic errors coming from above uncertainties . throughout this paper , a cosmological model of @xmath8
is assumed .
the density profile we use is @xmath9 where @xmath10 is the scale radius , @xmath11 is the critical density times the ( dimensionless ) characteristic over - density , and adopted values of @xmath12 are @xmath13 and @xmath14 .
a projected mass @xmath15 can be written as @xmath16 where @xmath17 ( @xmath18 is the line - of - sight direction . ) and @xmath19 the lens equation reads @xmath20 where @xmath21 is a source position , @xmath22 is an image position , @xmath23 is called lens parameter and @xmath24 .
@xmath25 is the angular distance from the observer to the lens , @xmath26 is that from the lens to the source galaxy , @xmath27 is that from the observer to the source galaxy , and @xmath28 is the critical surface mass density . for example
, @xmath29 corresponds to a cluster at redshift of @xmath30 with a virial temperature @xmath31 kev and a source redshift of @xmath32 ( e.g. * ? ? ? * ) .
an expected number ratio of radial arcs to tangential arcs is calculated by calculating a ratio of a cross - section forming radial arcs to that forming tangential arcs .
it is assumed that source galaxies are circular and their size is infinitesimal , i.e. lensing magnification matrix is same everywhere in the source and is represented by that at the source center . a quantity @xmath33 represents stretching - contracting factor in the radial direction of a lens potential . a quantity @xmath34 represents stretching - contracting factor in the tangential direction of a lens potential .
hereafter , we use the word ` radial arc ' for an image which satisfies @xmath35 and ` tangential arc ' for an image which satisfies @xmath36 .
the axis ratios of a radial arc @xmath37 and a tangential arc @xmath38 are given by @xmath39 respectively .
the radial and tangential cross - sections are calculated by solving the following inequalities : @xmath40 where @xmath41 is the threshold axis ratio .
we calculate ratios of radial cross - sections to tangential cross - sections for @xmath42 .
the nfw profile corresponds to @xmath43 . in figure 1
, we show the expected number ratios of radial arcs whose axis ratios are larger than the threshold value @xmath41 , to tangential arcs which have the same threshold axis ratio , against the threshold axis ratio @xmath41 .
calculations were done for @xmath44 and @xmath45 which cover almost all practical combinations of cluster redshifts and their masses , and source redshifts .
figure 1 shows that the number ratios drastically vary with @xmath12 values and variation due to variation of @xmath46 values is much smaller .
this indicates that the number ratios of radial arcs to tangential arcs in a certain cluster sample is a powerful probe for the central density profile of galaxy clusters .
biases in selection of a cluster sample may be reflected in biases of cluster masses and their redshifts , and the deepness of the observations which is equivalent to source redshifts .
changing values of these parameters only results in change of @xmath46 values .
therefore , insensitivity of the radial arc statistics to @xmath46 values shows that the selection biases of a cluster sample less affects the constraint from the radial arc statistics on the central density profile .
a dramatic increase of the radial arc number with increasing of @xmath12 from @xmath47 to @xmath48 found in figure 1 can be understood as follows .
when the central density profile is gentler than @xmath49 , a lens could have radial caustics where the stretching factor in the radial direction become infinite , and radial arcs are produced in two different ways @xcite .
when the central density profile is gentler than @xmath6 , radial arcs are produced only when source galaxies touch the radial caustics .
on the other hand , when the central density profile is steeper than @xmath6 , the width of the source image along the tangential direction is contracted .
therefore , radial arcs can be produced without significant stretching in the radial direction . as a result
, the number of radial arcs are dramatically increased when the central density profile is steeper than @xmath6 . since intrinsic source size of @xmath50 galaxies at a redshift of @xmath51 which are the dominant sources of observed arcs ,
is @xmath51 arcsec , high resolution instruments with the full - width - at - half - maximum of better than @xmath52 arcsec . , like _ hubble space telescope ( hst ) _ , _ subaru _ etc .
, is required to make a reliable sample for the radial arc statistics .
in table 1 , we list clusters for which existence of radial arcs were reported . in most of the cases ,
the radial arcs are thin and small .
special attention is necessary to find a radial arc even on the _ hst _ images
. the listed clusters in table 1 could be only a set for which existence or non - existence of radial arcs have been securely checked .
this sample is , of course , not uniform and is highly biased by the interests of the observer for each cluster .
however , as mentioned in previous section , our radial arc statistics is little affected by the selection biases .
we could safely use this sample for the comparison with theoretical results .
the hyphenated arcs in table 1 are folded or separated but nearly folded images and , therefore , each ` hyphenated arcs ' is counted as a single arc because those ( nearly ) folded images are produced from a single source .
the length of these arcs is calculated as a half ( in the case of merging of two images ) or one third ( in the case of merging of three images ) of the total length . since all of the reported radial arcs have axis ratio larger than @xmath53 , the threshold value can be set to @xmath53 .
tangential arcs with axis ratios larger than @xmath53 in the same clusters are also listed in table 1 .
table 1 shows that the observational number ratio of radial arcs to tangential arcs is @xmath54 on average .
this indicates @xmath55 .
the central density profile with @xmath56 over - produces radial arcs and can be rejected . on the other hand ,
the central density profile with @xmath57 predicts too few radial arcs compared with the observed results and also can be rejected .
therefore , the radial arc statistics supports that the central density profile of cluster is in the range of @xmath58 and dark matter is collisionless at least on the cluster scale .
we have shown that the radial arc statistics which is defined as a number ratio of radial arcs to tangential arcs appeared in a certain sample of clusters , strongly depends on the central density profile , and is almost independent of other cluster properties and source redshifts . therefore
, our radial arc statistics which uses the number ratio can provide a robust test to constrain the central density profile of galaxy clusters .
a tentative comparison with observations shows that the cluster central density profile is as steep as @xmath58 .
this result indicates that the dark matter should be collisionless at least in cluster scale .
therefore , even if the dark matter has a finite collision cross - section of the self - interaction @xmath1 , the mean free time @xmath59 of a dark matter particle ( mass @xmath3 ) should be longer than a cluster age @xmath60 .
the mean free time @xmath59 can be estimated as @xmath61 where @xmath62 km / s is a cluster 3d velocity dispersion , assuming that a characteristic density of the dark matter in clusters is represented by @xmath11 . assuming a cluster redshift of @xmath30 , its virial temperature of @xmath63 kev , and a cluster age of @xmath64 gyr , the upper limit is obtained as @xmath65 . according to the numerical simulations
@xcite , this upper limit allows weak self - interaction of the dark matter which produces a finite core in cluster mass distribution .
therefore , the above mentioned upper limit on @xmath66 should be taken as conservative limit . as shown by @xcite ,
the collision rate expected from this small cross - section in the dwarf galaxies is too small to reproduce core radii of dwarf galaxies inferred from rotation curve measurements .
therefore , the existence of cores in dwarf galaxies is still mystery .
although we have assumed infinitesimal source size , this assumption is valid only if a variation of the lensing amplification within a source is negligible . in figure 2
, we show the width of the region on the source plane which satisfies the condition of eq.(6 ) for a infinitesimal source .
the width is scaled by @xmath27 , for each threshold axis ratio using a typical value of @xmath67 and @xmath68 mpc . since size of @xmath50 galaxy
is @xmath51 arcsec at a redshift of @xmath51 which is the dominant source for arcs , figure 2 shows that the infinitesimal source approximation could provide reasonable estimations for the frequency of the tangential arcs but this assumption could not be good approximation for estimating the frequency of radial arcs for @xmath69 .
the quantitative studies of the effect of the finite source extents as well as the lens - clusters ellipticity and irregularity in mass distribution , will be done in the forthcoming paper . the cluster sample used for comparison with theoretical prediction
should be improved .
it should be mentioned , for example , that all the clusters listed in table 1 are xd clusters .
an xd cluster is a cluster which has a bright elliptical galaxy at the center of the x - ray emission .
since radial arcs are expected to appear in the very central region of a galaxy cluster and all the radial arcs listed in table 1 are found near cds , it must be checked whether a local central density increment by the mass of the cd galaxy plays an important role on forming radial arcs
. therefore much more number of clusters , including non - xd clusters , should be included in the radial arc statistics sample by future high resolution observation .
this research was partially supported by the grants - in - aid of the ministry of education , science , sports , and culture of japan ( no . 11440060 ) .
bzecourt , j. , kneib , j .-
p . , soucail , g. , & ebbels , t.m.d . , 1999 , 347 , 21 dalcanton , j.j . ,
& bernstein r.a . , 2000 , in asp conf .
197 , dynamics of galaxies : from the early universe to the present , ed .
f. combes , g. a. mamon , & v. charmandaris , 161 eke , v.r . ,
navarro , j.f . ,
frenk , c.s . , 1998 , , 503 , 569 fukushige , t. , & makino , j. , 1997 , , 477 , l9 gioia , i.m . ,
shaya , e.j . , le fvre , o. , falco , e.e . ,
luppino , g.a . , & hammer , f. , 1998 , , 497 , 573 hamana , t. , & futamase , t. , 1997 , , 286 , l7 hammer , f. , gioia , i.m . ,
shaya , e.j . ,
teyssandier , p. , le fvre , o. , & luppino , g.a . , 1997 , , 491 , 477 hattori , m. , watanabe , k. , yamashita , k. , 1997 , , 319 , 764 hattori , m. , kneib , j .-
, & makino , n. , 1999 , prog . of theor .
, suppl . , 133 , 1 jing , y.p . , & suto , y. , 2000 , , 529 , l69 miralda - escud , j. , 2000 ,
_ submitted to _
( astro - ph/0002050 ) molikawa , k. , hattori , m. , kneib , j .-
, & yamashita , k. , 1999 , , 351 , 413 moore , b. , 1994 , , 370 , 629 moore , b. , gelato , s. , jenkins , a. , pearce , f.r . , & quilis , v. , 2000 , , 535 , l21 natarayan , p. , kneib , j .-
p . , smail , i. , ellis , r.s .
, 1999 , , 499 , 600 navarro , j.f . ,
frenk , c.s . ,
white , s.d.m .
, 1997 , , 490 , 493 spergel , d.n . , &
steinhardt p.j . , 2000 , phys .
, 84 , 3760 subramanian , k. , & cowling , s.a . , 1986 , , 219 , 333 van den bosch , f.c . , & swaters , r.a .
, 2000 , _ submitted to _
( astro - ph/0006048 ) wyithe , j.s.b . ,
turner , e.l . ,
spergel , d.n .
, _ submitted to _
( astro - ph/0007354 ) yoshida , n. , springel , v. , white , s.d.m . , 2000 , _
_ _ in press _ | we show that an expected number ratio of radial arcs ( gravitationally lensed images whose major axes lie in the radial direction of a cluster - lens potential ) to tangential arcs ( gravitationally lensed images whose major axes lie in the tangential direction ) has strong dependence on the central density profile of galaxy clusters and has little dependence on other parameters , e.g. cluster temperature , background source galaxy redshift etc .. a comparison of the expected number ratios with observed ratios provides a robust test to constrain the central density profile of galaxy clusters .
a tentative comparison with the observational data shows that the central density profile of galaxy clusters is @xmath0 .
this result indicates that the dark matter is collisionless at least on cluster scale .
our result gives an upper limit on a collision cross - section of self - interaction of dark matter ( @xmath1 ) as @xmath2 where @xmath3 is a dark matter particle mass . |
as the main goal of the _ algebrodynamical _ approach @xcite , we regard the derivation of equations of physical fields ( or of fundamental physical laws , in a broader sense ) from a unique primary _ principle of purely abstract nature _ , based only on the intrinsic properties of exclusive mathematical structures ( groups , algebras , mappings etc . ) .
basic realization of this concept developed in @xcite makes use of only the _ differentiability conditions _ of functions of _ biquaternionic variable _ ,
i.e. of the cauchy - riemann ( cr ) equations generalized to the case of noncommutative ( associative ) algebras . in the framework of the version of analysis proposed in @xcite
the noncommutativity of starting algebra naturally results in the nonlinearity of generalized cr equations ( gcre ) , justifying the use of the latters as the dynamical equations of _ interacting _ fields .
infact , for the algebra of biquaternions @xmath0 ( isomorphic to the full @xmath1 complex matrix algebra ) the gcre appear to be lorentz invariant and carry natural 2-spinor and gauge structures . in the most examined case
for which a simple geometric and physical interpretation is obvious , the conditions of _
@xmath0-differentiability _ reduce to the invariant system @xcite @xmath2 for @xmath3-valued 2-spinor @xmath4 and gauge @xmath5 fields represented by @xmath6 and @xmath1 @xmath3-matrices respectively .
@xmath7 in ( 1 ) denotes usual matrix multiplication ( equivalent to that in @xmath0 ) , and @xmath8 represents a @xmath1 hermitian matrix of the increments of coordinates . as a consequence of ( 1 ) each component of the 2-spinor
@xmath9 satisfy the 4-eikonal equation @xcite . on the other hand , the _ compatibility conditions _ @xmath10 impose dynamical restrictions on the gauge field @xmath5 .
namely , the @xmath11 matrix @xmath3-valued _ connection 1-form _ @xmath12 ( @xmath13 being the pauli matrices ) by virtue of compatibility conditions * should be self - dual*. consequently , * free maxwell and yang - mills equations are satistied * identically on the solutions of ( 1 ) , for the scalar @xmath14 and the vector @xmath15 parts of the connection ( 2 ) respectively @xcite .
thus , the gcre system ( 1 ) exhibit wonderful relations to several fundamental physical equations being for the latters in some sense generating . in the approach regarded , an important role is performed by _
singular sets _ where the strengths of maxwell and ym fields turn to infinity .
such sets were found to have diverse dimensions and topology .
solutions with _
compact _ structure of singular set may be then considered as _ particle - like _ , the evolution of such singularities - particles being then governed by the system ( 1 ) itself . in this paper , we expound the process of reduction of system ( 1 ) to the equations of _ shear - free geodesic null congruences _ and , by kerr theorem , to the solution of _ algebraic _ equation @xcite . making use of this ,
we present an explicit form of the solutions to free maxwell equations , analyze the structure and evolution of their singular set and discuss the relation of these electromagnetic fields to the kerr - shild metrics and , by this , to the solutions of electrovacuum einstein - maxwell equations . in conclusion
, we consider general status and physical interpretation of `` particle - like '' singular solutions in the framework of electrogravidynamics and of unified algebraic field theory proposed .
through the elimination of the gauge field @xmath5 the system of gcre ( 1 ) may be written in a 2-spinor form @xcite @xmath16 in the gauge @xmath17 the system ( 3 ) reduces to two equations for one unknown function @xmath18 @xmath19 @xmath20 being the _ spinor coordinates_. note that as a consequence of ( 4 ) @xmath18 satisfy identically both the 4-eikonal and dalembert wave equations @xcite . assuming the eqs .
( 4 ) are solved , the components of 4-potential matrix @xmath21 may be expressed through the function @xmath18 as @xmath22 and satisfy free maxwell equations . wonderfully , eqs.(4 )
are completely identical to the equations of _ shear - free null geodesic congruences _ in the gauge for the spinor @xmath4 regarded @xcite . in accord with kerr s theorem @xcite , we then obtain the general solution of eqs.(4 ) in an implicit algebraic form @xmath23 @xmath24 being an arbitrary golomorphic function of three complex variables including two _ twistor _ ( projective ) components @xmath25 thus , a lot of solutions to free maxwell equations may be obtained through simply examining of the algebraic eq.(6 ) .
singularities of related field strengths may be then found from the _ caustic _ condition @xmath26 eliminating the only unknown function @xmath18 from two algebraic eqs.(6),(8 ) , one easily comes to the _ equation of singular set _ which determines the shape and evolution of singularities , without even taking care of explicit solving the eq.(6 ) itself .
the example of such procedure was presented in @xcite .
from the structure of twistor components ( 7 ) it s evident that stationary solutions to eq.(6 ) are exhausted by the functions @xmath27 , where @xmath28 does nt contain the time variable @xmath29 at all . in accord with the results of kerr and wilson @xcite , for stationary solutions with * compact * structure of singular set the function @xmath30
* should be at most quadratic in @xmath31 * , i.e. should have the form @xmath32 .
linear dependence on @xmath31 immediately leads to the trivial solution with zero fields . using 3-translations and 3-rotations
, the above form may be reduced to @xmath33 for which we obtain from quadratic eq.(6 ) @xmath34 for a real valued @xmath35 from ( 9 ) and the expression for potentials ( 5 ) we obtain the coulomb electric field with a point singularity and a * fixed value of the electric charge * as a consequence of nonlinear primary system of gcre ( 1 ) .
imaginary values of @xmath35 correspond to the _ ring singularity _ with radius @xmath36 and multipole structure of em fields with coulomb first main term @xcite . decomposition and separation of real parts of em fields at a distance @xmath37
gives @xmath38 @xmath39 contrary to an ambiguous value of ring s radius @xmath35 , the dimensionless electric charge is strictly fixed ( up to a sign ) by field equations , so that in absolute units it may be identified with _
elementary charge _
note that the solution ( 9 ) ( for the case @xmath41 ) and the property of charge quantization for the gcre system ( 1 ) have been obtained in a direct way in @xcite .
( recently @xcite there were some interesting attempts to explain electric charge quantization by topological reasons instead of dynamical considerations used here ) . on the other hand , solution ( 9 ) is extremely important in the framework of gtr .
indeed , the expression @xmath42 , where @xmath43 is the 2-spinor related to the function @xmath18 , defines the _ principal null congruences _ @xmath44 of a riemannian space - time endowed with a kerr - shild metric @xmath45 where @xmath46 represents the metric of auxiliary minkowski space - time .
the scalar factor @xmath47 should be then determined by the einstein vacuum or electrovacuum equations and for fundamental stationary solution ( 9 ) leads to the kerr or kerr - newman metrics respectively ( consequently , to the schwarzschild or reissner - nordstrm metrics for the case @xmath35=0 ) . in our approach
, it is of great importance that * singularities of curvature of metric ( 10 ) are fixed @xcite by the same condition ( 8) as for electromagnetic field * and define infact one unique particle - like object .
another wonderful fact is that for the kerr - newman solution of einstein - maxwell equations electromagnetic fields are just those defined by the gcre system ( apart from the property of quantization of charge for the latters ! ) and may be asymptotically presented by the eq.(10 ) .
moreover , these fields * obey maxwell equations both in flat space and in riemannian space with kerr - shild metric ( 11 ) ! * such a remarkable _ property of stability _ of electromagnetic fields under kerr - shild deformations of space - time geometry noticed in @xcite will be discussed elsewhere .
making use of correspondence between the fundamental solution ( 9 ) to the gcre system ( 1 ) and the kerr - newman solution of einstein - maxwell system , one is able to endow the solution ( 9 ) with a complete set of quantum numbers ( including mass and spin ) .
then the _ gyromagnetic ratio _ would automatically correspond to that for dirac particle , while the charge would be fixed in magnitude .
unfortunately , no natural reasons to ensure the quantization of mass could be seen nowadays .
we ll continue the discussion below .
let us consider the general quadratic form of the function @xmath24 .
when the terms bilinear in @xmath48 , @xmath49 are absent , such functions ( under the restriction on singular set to be compact ) correspond to the boosted or rotated kerr solution @xcite . on the other hand , the function
@xmath50 has been considered in detail in @xcite . the explicit expression for @xmath31 in this case
is @xmath51 where @xmath52 .
em fields correspondent to ( 12 ) are @xmath53 for real @xmath54 the fields ( 13 ) are identical to the well - known _ born solution _ for two point - like charged `` particles '' performing uniformly accelerated counter - motion . the value of electric charge for each particle does not depend on @xmath54 being * fixed and equal to the charge of fundamental solution ( 9)*. for the case of imaginary @xmath54 one has the singularity of rather exotic toroidal structure , defined by the equation @xmath55 ( see @xcite for details )
. in general case of complex - valued @xmath54 singular set manifests itself as the two rings of fixed radii performing again the oncoming hyperbolic motion along @xmath56-axis .
it may be proved that ( up to the transformations of poincare group ) the _ axisymmetric _ solutions to the eq.(6 ) ( and to the gcre system ( 1 ) respectively ! ) generated by quadratic function @xmath30 * are exhausted by the kerr - like solution ( 9 ) and the nonstationary bisingular solution ( 12 ) * together with ( toroidal or double ring - like ) modifications of the latter .
it seems , however , that the solutions to gcre with compact singularity and non - axial symmetries may be of interest too . here
we present an example of such solution which may be obtained from the generating function @xmath57 . resolving the equation @xmath58
, one comes to the following expression @xmath59 @xmath60 where @xmath61 .
the singular set for this solution is defined by the condition @xmath62 and for @xmath63 has the form of flat _ figure `` 8 '' curve _ ( fig.a ) ) .
the time evolution of this singularity is illustrated by fig.b ) .
a ) t=0 b ) t=1 [ pic1 ] em fields related to the solution ( 14 ) and being represented by the complex combination @xmath64 are : @xmath65 for each finite moment of time they decrease rapidly ( as @xmath66 ) with the distance from the centre of singularity
. the fields are neutral ( with total charge being equal to zero ) and null ( @xmath67 ) . in conclusion , we present a peculiar solution with noncompact singularity which serves as the _ analogue of electromagnetic wave _ in grce dynamics . for the solutions of wave - like type the generating function @xmath30 should depend only on one twistor component , say , @xmath48 .
then , for the equation @xmath68 the initial distribution of @xmath69 may be arbitrary fixed at @xmath70 , i.e. at the @xmath71-axis .
choosing for the latter the _ monochromatic _ dependence and resolving the equation @xmath72 , where the parameters @xmath73 are assumed to be real and positive , we find @xmath74 @xmath75 being the principal branch of the so called _ lambert _ function which is the solution of the equation @xmath76 .
the structure of singular set is simply derived after that and appears to be a neutral _ helix _ of radius @xmath77 and of lead @xmath78 propagating along @xmath71- direction with the speed of light .
electromagnetic fields are mutually orphogonal and transversal while polarization depends on the distance from the axis . in the direction perpendicular to the axis the fields fall at large distancies as @xmath79 .
as before , the fields are globally defined only up to a sign .
shear - free null geodesic congruences @xmath44 and the kerr - shild metrics ( 12 ) may be associated with the nonstationary solutions above - presented up to the scalar factor @xmath47 . at present
it s not clear if the latter may be choosed so that the einstein - maxwell electrovacuum system would be satisfied . by this , an interesting representation @xcite for the shear - free congruences ( through consideration of null cone emanated by the source moving along some curve in _ complex space _ ) as well as
the _ condition of stability _ for electromagnetic fields under the kerr - schild deformations of space - time geometry @xcite may be of great use .
well - known are the numerous problems arising in gtr and in quantum field theory in respect to the singularities of solutions of field equations ( violation of causality @xcite , divergences etc . ) . on the other hand ,
just the naked singularity of kerr - newman solution ( which appears instead of black hole solution in the case of a large angular moment ) manifests itself many remarkable properties related to that of elementary particles .
accordingly , several attempts to construct the model of electron on the base of kerr - type solutions ( kts ) have been undertaken @xcite .
however , they all dealt with the problem of physically suitable _ source _ for kts to be found which is tigthly related to the well - known _ twovaluedness _ of kerr - like geometry and electromagnetic fields in particular .
infact , the introduction of source becomes admissible only after the _ cut _ of space which restore the global uniqueness of the @xmath3-valued functions representing the fields of kts .
unfortunately , the surface of cut is quite ambigious : it may be either the _ disk _ spanning the kerr singular ring @xcite or the oblate spheroid @xcite covering the singular ring on which the kerr - newman metric turns surprisingly into the minkowski one .
consequently , one may think of the source of kts as of the `` rotating relativistic disk '' , of the `` bubble of flat geometry '' within the external kerr - newman space - time etc .
thus , we are to conclude that * there are no grounds to speak about the `` source '' of kts solutions at all * since the twovaluedness is the unavoidable feature of their internal mathematical structure . to illustrate the above statement , let us consider a simplier case of the singular `` particle - like '' solutions to free maxwell equations in flat space - time presented in this paper .
note that all of them ( apart from the coulomb and born solutions with point - like singularities ) are of the same two - valued structure being in each point defined up to a sign .
certainly , by no @xmath80- function distribution of charge and current along the singular curve one can reproduce the field distribution in the whole space .
on the other hand , such solutions are locally well defined and may be analytically continuated from the region of regularity so that the full structure of singular set is established * in a unique way*. one can not in any way change either the shape and topology of the singularity or its time evolution ( the latter property being the most important from physical point of view ) .
suppose we really hope to describe the interactions and transmutations of elementary particles by means of the solutions regarded ( which are rather to be _ multisingular _ for real physical process ) .
then we ll proceed in well - defined and unique predictions in spite of partial indefinitness of em fields and escape any divergences at all !
moreover , one may think of such solutions as of * the only possibility to explain the `` spin 1/2 '' structure * at a purely classical level and with transparent picture of space - time dynamics being preserved .
this is still more true in the framework of the algebrodynamical approach we develop , at least for two reasons .
the first one is that in respect to the internal structure of gcre system ( 1 ) , gauge ( electromagnetic plus yang - mills ) fields stand there hand by hand with the 2-spinor structure so that the latter appears naturally together with maxwell equations .
the second reason is that , apart from the right value of gyromagnetic ratio , the value of electric charge is automatically fixed by the field equations themselves . of course , the stationary kts as well as bisingular and `` figure 8 '' solutions presented here can say nothing about real dynamics of an ansamble of compact `` particle - like '' singularities .
even the problem of interaction of two kerr - newman objects is far from solution . in the framework of algebrodynamics the overdetermined structure of gcre
impose restrictions even on the initial distribution of the fields @xcite so that the scattering problem should be fully reformulated .
historically , the solution of maxwell equations with ring singularity have been obtained by appel in 1887 and revived in the works of newman and burinskii @xcite .
general study of singular solution to maxwell equations have been undertaken by bateman @xcite .
nowadays the concept of naked singularities of kts as the model for elementary particles is successfully developed , say , in the works of clement @xcite . to conclude , we argue that hostile attitude of physicists to singularities of field equations could be quite unjustified .
there exist no restrictions of principal character for the _ compact multisingular _ solutions to describe the interactions of particle - like objects in a self - consistent way .
then their transmutations could be treated as _
perestroikas _ of singularities in terms of catastrophe theory .
this programme should be implemented independently both in the framework of einstein - maxwell dynamics and of the algebraic dynamics based in particular on the gcre system ( 1 ) . | foundations of _ algebrodynamics _ based on earlier proposed equations of _ biquaternionic holomorphy _ are briefly expounded .
free maxwell and yang - mills eqs .
are satisfied identically on the solutions of primary system which is also related to the eqs . of _
shear - free null congruences _ ( sfc ) , and through them - to the einstein - maxwell electrovacuum system . _
kerr theorem _ for sfc reduces the basic system to one algebraic equation , so that with each solution of the latter some ( singular ) solution of vacuum eqs . may be associated .
we present some exact solutions of basic algebraic and of related field eqs . with _
compact _ structure of singularities of electromagnetic field , in particular having the form of _ figure `` 8 '' curve_. fundamental solution to primary system is analogous to the metric and fields of the _ kerr - newman _ solution .
in addition , in the framework of algebraic dynamics the value of _ electric charge _ for this solution is strictly fixed in magnitude and may be set equal to the _ elementary charge_. |
spatiotemporal patterns in the transverse direction of an optical field have now been widely studied theoretically and experimentally @xcite .
studies of optical pattern formation share a number of aspects and techniques with general investigations of pattern formation in other physical systems @xcite , but they also have specific features such as the role of light diffraction and the vectorial degree of freedom associated with the polarization of the light electric field amplitude . a prototype simple model which has been very useful for the understanding of pattern formation in nonlinear optical cavities is a mean field model describing a kerr medium in a cavity with flat mirrors and driven by a coherent plane - wave field@xcite .
this model was extended to take into account the polarization degrees of freedom in @xcite .
some of the basic polarization mechanisms of pattern formation in alkali vapors or other non - linear materials can be understood in terms of this simple model despite the fact that the model is too simple to give a complete description of alkali vapors .
furthermore , the relative simplicity of the model in @xcite makes it worthwhile to study it in depth as a general prototype model for the basic understanding of vectorial patterns .
a first study was undertaken in @xcite for the case of linearly polarized driving field and the positive cavity detuning . a more detailed study in which the case of elliptically polarized driving field is also considered has been presented recently in reference @xcite .
cavityless kerr media can be used as an optical phase conjugation mirror .
an ideal phase conjugate mirror should generate an output field such that the amplitude of the field , its propagation vector and its polarization unit vector are the complex conjugates of the corresponding magnitudes in the input field .
the first two properties can be achieved easily using , for example , four - wave mixing processes @xcite .
usually the third property can be obtained only when the pump waves are circularly polarized and counterrotating . in order to find a system that verifies the three properties ( vector phase conjugation ) for an arbitrarily polarized input field it is required to make use of the special tensor properties of two - photon atomic transitions in degenerate four - wave mixing processes @xcite . more precisely , vector phase conjugation can be achieved if the two - levels coupled by the two - photon transition have equal angular momenta @xmath1 with @xmath2 or @xmath3 . in this situation
the @xmath4 component of the susceptibility tensor vanishes . intuitively , as @xmath5 , the atom do not change its angular momentum either by absorbing two pump photons or by emitting a probe and a conjugate photon so that the conjugate photon must be emitted with angular momentum equal and opposite to that of the probe photon .
a detailed calculation @xcite shows that indeed this is true for the cases indicated before .
we should stress that the polarization properties of two - photon - resonant degenerate four - wave mixing processes are different from those of most other degenerate four - wave mixing processes .
in the two - photon case the underlying physical mechanism is scattering of the probe field from a spatially uniform temporally varying coherence induced by the two pump waves whereas in the other cases is scattering from a spatially varying refractive - index distribution induced by the interference between the pump and the prove beams .
experimentally , vector phase conjugation was first observed using the @xmath6 two - photon transition in sodium vapor @xcite . here we will show that for a ring cavity filled with a nonlinear kerr medium close to a two - photon - resonance ( so that @xmath7 ) and illuminated with linearly polarized input and with negative detuning a new interesting situation appears : a codimension two bifurcation in which two stationary turing - like instabilities occur simultaneously .
the first instability , if the other were not present , would originate a hexagonal pattern which is polarized linearly in the same direction as the input field .
the second instability is of pure vectorial origin and if the first instability were not present it would give rise to an elliptically polarized stripe pattern . here
we study the interplay between the two instabilities .
the codimension two bifurcation appears here in a natural way associated to the two - photon - resonant four - wave mixing nonlinearities rather that as the result of the fine tuning of two system parameters as is usually the case .
the intensity of the pump field is the single control parameter to be tuned to change the distance to the instabilities .
furthermore , the system still have another easily accessible control parameter , the detuning , which allows the system to form different patterns while remaining at the same distance to the codimension two instability threshold .
in particular we show how the detuning can be used as a tuning parameter to select the pattern .
the outline of this paper is as follows : in sec .
[ model ] we describe the model we are considering , its spatially homogeneous solution and the stability analysis of this solution . in sec .
[ weaknonlin ] using a weakly nonlinear analysis we derive the evolution equation for the patterns arising from the interacting instabilities . from these equations ,
the selected patterns for different values of the detuning are analyzed in sec .
[ patternselection ] . in sec .
[ numerical ] we describe the results from numerical simulations of the model and finally in sec .
[ conclusions ] we give some concluding remarks .
the system we consider is a ring cavity filled with an isotropic kerr medium .
the cavity is driven by an external @xmath0 polarized input field . the situation in which the polarization degree of freedom of the electromagnetic field is frozen
was first considered by l. a. lugiato and r. lefever @xcite .
geddes et al @xcite generalized the model of @xcite to allow for the vector nature of the field .
their description of this system is given by a pair of coupled equations for the evolution of the two circularly polarized components of the field envelope @xmath8 and @xmath9 , defined by @xmath10 for an isotropic medium , the equations are @xmath11 e_\pm , \label{1}\ ] ] where @xmath12 represents the components of the input field ( the right and left circularly polarized components are equal since we consider @xmath0 linearly polarized input ) , @xmath13 indicates self - focusing ( self - defocusing ) , @xmath14 is the cavity detuning , @xmath15 represents the strength of diffraction and @xmath16 is the transverse laplacian .
the parameters @xmath17 and @xmath18 are related to the nonlinear susceptibility tensor components in the following way : @xmath19 and @xmath20 @xcite .
also , for an isotropic medium we have @xmath21 . as discussed in the introduction , here we are considering two - photon transitions between levels with equal angular momenta @xmath1 where @xmath2 or @xmath3 , so that @xmath7 ( @xmath22 and @xmath23 ) .
also as we consider the self - focusing situation , in what follows we take @xmath24 .
the intensity of the input field is @xmath25 .
the steady state homogeneous solutions of eq .
( [ 1 ] ) are reference states from which transverse patterns emerge as they become unstable .
there is a symmetric ( @xmath26 ) and two asymmetric ( @xmath27 ) homogeneous solutions @xcite .
the symmetric solution corresponds to linearly polarized output light , while the asymmetric solutions to elliptically polarized output . increasing the input field ,
the asymmetric solutions appear only for values of @xmath28 larger than the instability threshold for pattern formation@xcite , so here we will only consider the symmetric solution@xcite , @xmath29 which gives an implicit formula for @xmath30 . as it is well known , eq .
( [ symmetricsol ] ) implies bistability for @xmath31 .
however here we will always consider negative detunings which are far away from the bistable regime .
an example of the symmetric solution for linearly polarized input is given in the inset of fig .
[ fig1 ] .
basic features of the stability of the steady state homogeneous symmetric solution can be analyzed by considering the evolution equations for perturbations @xmath32 defined by @xmath33 from eqs .
( [ 1 ] ) and ( [ perturba ] ) we find @xmath34\psi_{\pm } + i i_s 2(\psi_{\mp}+\psi_{\mp}^*+\vert\psi_{\mp}\vert^2)(1+\psi_{\pm } ) \,.\ ] ] it is convenient to make a change of variables to the following basis @xcite @xmath35 where @xmath36 stands for transpose . in this basis , which emphasizes the role of symmetric ( @xmath37 ) and antisymmetric ( @xmath38 ) modes , eq .
( [ pertev ] ) may be written as : @xmath39 where the linear matrix @xmath40 is a matrix with @xmath41 blocks in which the symmetric ( @xmath42 , @xmath43 ) and antisymmetric ( @xmath44 , @xmath45 ) modes are decoupled .
@xmath46 as a consequence , the linear instabilities lead to the growth of either a symmetric or an antisymmetric mode . in fourier space
we have : @xmath47 and @xmath48 where @xmath49 .
instability occurs if at least one of the eigenvalues @xmath50 of @xmath51 and @xmath52 has a positive real part . in fourier space ,
these eigenvalues are solutions of the characteristic equations : @xmath53 for @xmath54 , we have @xmath55 where @xmath56 and @xmath57 . for both eigenvalues ,
the instability occurs at @xmath58 .
the non linearities in ( [ vectevol ] ) include quadratic @xmath59 and cubic terms @xmath60 : @xmath61 and @xmath62 the structure of these terms also gives some general information on the nature of the instabilities .
in particular , if the quadratic non linearity @xmath59 does not vanish , one expects the formation of hexagonal patterns instead of stripes . as explained in ref .
@xcite when the symmetric mode becomes unstable an hexagonal pattern is expected whereas when the antisymmetric mode becomes unstable there are no relevant quadratic couplings so a stripe pattern is formed . in fig .
[ fig1 ] we plot marginal stability curves for @xmath63 as a function of @xmath64 .
the shape of the marginal stability curves is , in fact , the same for any value of the detuning @xmath14 .
this is because the eigenvalues @xmath65 given by eq .
( [ lambdasymmetric ] ) depend on @xmath66 , so a change in the value of @xmath14 is equivalent to a displacement of the origin of @xmath64 by the same amount .
the origin moves to the right if the detuning @xmath14 is increased .
the instability region i comes from the eigenvalue @xmath67 so the critical modes are symmetric and of zero frequency .
a subcritical hexagonal pattern is expected via a transcritical bifurcation .
if this were the only instability , it would correspond to the case discussed in @xcite , in which the polarization degree of freedom is not taken into account .
this instability leads to an @xmath0-polarized pattern while the @xmath68-polarized component of the field continues to be zero .
the instability region ii comes from the eigenvalue @xmath69 so the critical modes are antisymmetric and of zero frequency .
a stripe pattern is expected @xcite .
given the antisymmetric nature of the unstable mode , the @xmath0-polarized component of the field is stable and remains almost homogeneous , while the stripe pattern appears in the @xmath68-polarized component , which has zero value below the instability .
overall , the electric field displays an elliptically polarized spatial structure .
we remark that such an instability is of pure vectorial nature with no analogue when the polarization degree of freedom is frozen . in the case
considered here , @xmath70 , starting from the linearly polarized homogeneous solution , as the input field is increased , the system crosses the two instability thresholds simultaneously .
this is a codimension two bifurcation involving two sets of stationary modes .
the critical modes associated to region i are symmetric and have a critical wave number @xmath71 , while the critical modes associated to region ii are antisymmetric and have a critical wave number @xmath72 . the ratio @xmath73 can be changed easily varying the value of @xmath14 .
the eigenmodes of the linear evolution matrix @xmath40 are , in fourier space @xmath74 , @xmath75 , @xmath76 and @xmath77 , with @xmath78 where @xmath79(\vec k)$ ] denotes the fourier transform of @xmath80 .
furthermore @xmath81 where index @xmath82 stands for @xmath83 or @xmath15 and @xmath84 .
the critical modes correspond to the eigenvalues @xmath85 and @xmath86 .
note also that @xmath87 after diagonalization of the linear evolution matrix , the dynamics ( [ vectevol ] ) may be rewritten , in fourier space , as @xmath88 where @xmath89 and @xmath90 is a diagonal matrix with diagonal elements @xmath91 . @xmath92 and @xmath93 are the fourier transforms of the nonlinear terms of equation ( [ vectevol ] ) , where the @xmath94 have been replaced by the corresponding linear combinations of @xmath95 , @xmath96 , @xmath97 and @xmath98 . slightly above threshold , this dynamics may be reduced to the dynamics of the critical modes amplitudes only , through the adiabatic elimination of the stable non - critical modes .
this procedure is now standard @xcite , and we will only sketch here the main steps of its application to model ( [ vectevol ] ) , and derive evolution equations for the critical modes , up to cubic nonlinearities .
let us write the amplitudes of the critical modes as @xmath99 and @xmath100 .
their dynamics writes @xmath101 the terms in the quadratic nonlinearities @xmath102 are convolutions of products of critical and non - critical modes .
the terms involving non - critical modes only ( @xmath103 , @xmath104 , @xmath105 ) may be neglected , since they contribute , through the adiabatic elimination process , to nonlinearities of quartic or higher order .
furthermore , the structure of the cubic nonlinearities of eq .
( [ vectevol ] ) is such that the pure critical mode contribution to @xmath106 vanishes .
hence @xmath106 will be neglected .
we are thus dealing here with a non - generic case , where cubic nonlinearities are generated solely by the adiabatic elimination of stable modes from quadratic terms .
the nonlinearities @xmath107 and @xmath108 can be written at the lowest order in a non - critical modes as @xmath109\nonumber\\ & & + i_s \left [ s(\vec k ) ( \hat s_+(0 ) + \hat s_-(0 ) ) + s(-\vec
k)(\hat s_+(2\vec k ) + \hat s_-(2\vec k))\right ] \ , , \nonumber\\ ( \hat n_2\vert_{\vec k , k = k_a})_a & \simeq & i_s \int{d\vec k ' \left [ s(\vec k - \vec k ' ) a(\vec k ' ) \right . } \nonumber\\ & & \quad \quad \left .
+ s(\vec k-\vec k')\hat a_- ( \vec k ' ) \delta(|\vec k'|-k_a ) + a(\vec k-\vec k')\hat s_-(\vec k ' ) \delta(|\vec k'|-k_s ) \right ] \nonumber\\ & & + i_s a(\vec k)\left[\hat s_+(0 ) + \hat s_-(0)\right ] \ , .
\label{n2sn2a } \end{aligned}\ ] ] the non - critical modes @xmath110 , @xmath111 , @xmath112 and @xmath113 present in the eqs .
( [ n2sn2a ] ) may be expressed as an expansion in powers of critical ones using the adiabatic elimination procedure .
one has at the leading order @xmath114}\ , , \nonumber\\ \hat s_{\pm}(2\vec k_s ) & = & { i_s\over 2\lambda_{1,\pm}(2k_s)}(1 \mp 2 \beta_s ( 2k_s ) ) s(\vec k_s)s(\vec k_s ) \ , , \nonumber\\ \hat s_-(\vec k_s ) & = & { i_s\over 2\lambda_{1,-}(k_s)}\int{d\vec k'\left [ 3 s(\vec k_s-\vec k')s(\vec k')+a(\vec k_s-\vec k')a(\vec k')\right ] } \ , , \nonumber\\
\hat a_-(\vec k_a ) & = & -{i_s\over \lambda_{2,-}(k_a)}\int{d\vec k ' s(\vec k_a-\vec k')a(\vec k ' ) } \ , . \label{noncriticalmodes}\end{aligned}\ ] ] the substitution of eqs . ( [ noncriticalmodes ] ) in eqs . ( [ n2sn2a ] ) leads to the following asymptotic dynamics for the critical modes , valid close to the instability threshold @xmath115 } \nonumber\\ & & - \int{d\vec k'\int{d\vec k '' u(\vec k,\vec k',\vec k '' ) s(\vec k-\vec k')s(\vec k'-\vec k '' ) s(\vec k '' ) } } \nonumber\\ & & - \int{d\vec k'\int{d\vec k '' w(\vec k,\vec k',\vec k '' ) s(\vec k-\vec k')a(\vec k'-\vec k '' ) a(\vec k '' ) } } \nonumber\\ \dot a(\vec k)&= & ( 2i_s - 1 ) a ( \vec k)+ v_2 \int{d\vec k ' s(\vec k - \vec k')a(\vec k ' ) } \nonumber\\ & & - \int{d\vec k'\int{d\vec k ''
\bar w(\vec k,\vec k',\vec k '' ) a(\vec k-\vec k')a(\vec k'-\vec k '' ) a(\vec k '' ) } } \nonumber\\ & & - \int{d\vec k'\int{d\vec k '' \bar
u(\vec k,\vec k',\vec k '' ) a(\vec k-\vec k')s(\vec k'-\vec k '' ) s(\vec k'')}}\ , , \end{aligned}\ ] ] where @xmath116 , and @xmath117 with @xmath118 , @xmath119 , @xmath120 , @xmath121 , and @xmath122 .
it has to be noted that the @xmath123 terms come from quadratic resonances between critical and non critical symmetric modes , while the @xmath124 terms come from quadratic resonances between symmetric and antisymmetric modes . as in the case of isolated turinglike instabilities
, there is no quadratic resonance between critical antisymmetric modes only . as a result
, pattern formation is expected to strongly depend on the existence of quadratic resonances between symmetric and antisymmetric modes .
hence , for the sake of simplicity , we will consider separately the case with quadratic resonance between symmetric modes only , and the case with quadratic couplings between antisymmetric and symmetric modes , which is more intricate .
in the latter case , quadratic couplings are such that @xmath125 , with @xmath126 and @xmath127 . since @xmath128 , where @xmath129 is the angle between symmetric and antisymmetric vectors , quadratic resonance arise only for @xmath130 .
in this section we study the various patterns that may appear as asymptotic solutions of eq .
( [ noresampl ] ) , and their stability .
each of these patterns is built on an arbitrary number of critical modes pairs , and it is the nature of their nonlinear couplings that determines their stability . in the following ,
we label the modes contributing to the formation of a pattern @xmath131 , @xmath132 .
we also use the notation @xmath133 for the complex conjugate of @xmath134 . in this case
, there is no quadratic coupling between symmetric and antisymmetric modes , and there is no contribution coming form the terms with the coefficients @xmath137 , @xmath138 , @xmath139 and @xmath140 .
let us then consider separately amplitude equations for each type of modes . for patterns built on antisymmetric modes only , the amplitude equations for an arbitrary number @xmath141 of pairs of modes are : @xmath142 hence , a pattern built on @xmath141 pairs of wavevectors is marginally stable versus a @xmath143 pair of wavevectors .
so , at this level of analysis , any pattern with an arbitrary number of wavevectors is possible , including patterns of the form @xmath144 , where @xmath145 is zeroth order bessel function and @xmath146 .
the linear growth rate of the evolution of symmetric modes in the presence of such patterns is zero , so antisymmetric patterns are marginally stable versus symmetric mode patterns .
it would be necessary to go to higher orders in the amplitude equations to complete the pattern selection analysis in this case .
on the other hand , for patterns built on symmetric modes only , the amplitude equations for a triplet of such modes are : @xmath147 at the instability threshold ( @xmath148 ) , hexagonal pattern appear via a subcritical bifurcation . increasing the value of @xmath149 , stripes may also become stable for @xmath150 .
there is a small region of bistability of stripes and hexagons and for larger values of @xmath149 only the stripes remain stable .
hexagonal patterns are stable versus antisymmetric modes for @xmath151 $ ] .
hence , sufficiently close to threshold , one may expect hexagonal patterns . on increasing @xmath149 , such patterns should become unstable .
the amplitude equations for mixed structure formed by triplets of symmetric modes and an arbitrary number of antisymmetric modes are : @xmath152 \nonumber\\ \dot a_k&= & ( 2i_s - 1 ) a_k - a_k \left [ 2 u_2\sum_{j=1}^3 \vert s_j\vert^2 + 2 w_2 \sum_{l=1}^m \vert a_l\vert^2 \right ] \,.\end{aligned}\ ] ] however , these equations do not admit non trivial steady states . as a result , in these conditions ,
hexagonal or striped patterns of symmetric modes , and patterns built on an arbitrary number of antisymmetric modes may be simultaneously stable . in this case
, quadratic resonances may occur between symmetric and antisymmetric modes , and one may now expect contributions coming from the @xmath137 , @xmath138 , @xmath139 and @xmath140 in the amplitude equations , which have to be modified accordingly .
let us then consider the different types of patterns that may arise in this case , and which are built on modes belonging to the following set of critical modes ( up to an arbitrary phase angle ) ( cf .
figure [ acopl ] ) .
striped and hexagonal symmetric modes patterns are now always unstable versus antisymmetric modes .
this is due to the positive nonlinear renormalization induced by these modes in the evolution equation for their resonantly coupled antisymmetric modes ( this renormalization is associated to the terms with @xmath155 coefficients in @xmath156 ) . as in subsection
[ no - sym - antisym - coupling ] , a pattern built on @xmath141 arbitrary pairs of antisymmetric modes is marginally stable versus a @xmath143 pair of wavevectors . furthermore , since the contributions of the quadratically resonant symmetric and antisymmetric modes , in their respective amplitude equations have opposite signs ( @xmath157 and @xmath158 ) , pure antisymmetric mode patterns are also unstable versus resonantly coupled symmetric and antisymmetric ones .
recall also that a pattern built on @xmath141 arbitrary pairs of antisymmetric modes is marginally stable versus non resonant symmetric modes . as a result , when quadratic resonances between symmetric and antisymmetric modes are possible , pure steady patterns built on symmetric or antisymmetric modes only , are always unstable .
we have thus to consider the possibility for the system to develop mixed modes patterns .
let us consider the simplest case of mixed mode patterns built on one symmetric mode and two quadratically resonant antisymmetric modes ( e.g. @xmath159 , @xmath160 and @xmath161 in figure [ acopl ] ) .
their uniform amplitude equations are : @xmath162 \nonumber\\ \dot a_1&= & ( 2i_s - 1 ) a_1 + i_s s_1 \bar b_1 - a_1 \left [ 2(u_2-w_3)\vert s_1\vert^2 + ( 2w_2\vert a_1\vert^2 + ( 2w_2+w_1)\vert b_1\vert^2 ) \right ] \nonumber\\ \dot b_1&= & ( 2i_s - 1 ) b_1 + i_s s_1 \bar a_1 - b_1 \left [ 2(u_2-w_3)\vert s_1\vert^2 + ( 2w_2\vert b_1\vert^2 + ( 2w_2+w_1)\vert a_1\vert^2 ) \right ] \ , .\end{aligned}\ ] ] because of the symmetry between @xmath160 and @xmath161 we look for solutions with the same amplitude for both antisymmetric modes . defining amplitude and phase variables as @xmath163 , @xmath164 , @xmath165 , and @xmath166 ,
one has @xmath167\nonumber\\
\dot r_a & = & ( 2i_s - 1 ) r_a + i_s r_ar_s\cos \psi - r_a \left [ 2(u_2-w_3 ) r_s^2 + ( 4w_2 + 2w_1)r_a^2 \right ] \nonumber \\
\psi&= & { i_s\over r_s } [ r_a^2 - 2r_s^2]\sin\psi \ , .\end{aligned}\ ] ] a phase stable steady state corresponds thus to @xmath168 if @xmath169 , and to @xmath170 if @xmath171 . combining the steady state conditions for @xmath172 and @xmath173
, it can be seen that @xmath168 requires that @xmath174 and that @xmath175 requires that @xmath176 . as a result
, a stable steady state may only be obtained for @xmath177 , with @xmath178 .
furthermore , this condition is satisfied if the kinetic coefficients are such that @xmath179 , which is thus a necessary condition to be satisfied to obtain such mixed modes solutions .
this condition corresponds to @xmath180 or @xmath181 , with @xmath182 .
a similar analysis may be performed for a pattern formed by an hexagonal planform of symmetric modes and their quadratically resonant antisymmetric ones ( cf .
fig [ acopl ] ) .
however , in this case , the fact that the quadratic couplings between symmetric and antisymmetric modes have opposite signs does not allow the stabilization of critical patterns .
the conclusion of the analysis presented is the previous two subsections is as follows : * for @xmath183 ( @xmath184 ) , close to threshold hexagonal symmetric patterns are stable . for slightly larger values of @xmath149 symmetric stripes and hexagons
are bistable and finally only the stripes remain stable .
patterns built on an arbitrary number of antisymmetric modes are neutrally stable . as this
results come from an expansion up to cubic nonlinearities we can not conclude about the stability of the antisymmetric patterns .
it would be necessary to go to higher order terms . finally , there are no mixed stationary patterns . * for @xmath185 ( @xmath186 ) ,
pure symmetric and antisymmetric mode patterns are unstable .
also , no steady mixed mode patterns are found either .
for that reason one expects to find time dependent structures which involve both symmetric as well as antisymmetric modes . * for @xmath187 ( @xmath188 )
, one may expect steady patterns formed by the superposition of one symmetric mode and its quadratically resonant antisymmetric ones .
we should notice that the range in which this steady mixed patterns does exist may be in fact smaller than @xmath188 .
this is because our calculation was for a necessary condition for the stability of the global phase , but this is not a sufficient condition for the stability of the pattern .
* for @xmath189 ( @xmath190 ) , symmetric modes are unstable versus antisymmetric modes and no steady mixed mode solutions are found .
patterns composed of an arbitrary number of antisymmetric modes are neutrally stable , so it would be necessary to include higher order nonlinearities to conclude about their stability .
the present analysis provides the basic elements for the study of pattern formation when both symmetric and antisymmetric modes become simultaneously unstable .
as it will be discussed below , it is partially confirmed by numerical analysis of the complete dynamical model . it could be improved , on the one hand , in determining the full stability range of mixed mode patterns , and , on the other hand , in resolving the issue of finding asymptotic states in cases where no steady critical patterns are found . in such cases , as suggested by numerical analysis , one should consider the possibility of asymptotic time - dependent or non critical patterns .
the latter case would require to incorporate in the dynamics harmonics or non critical modes .
we should notice that a peculiarity of the situation considered here is that there are no cubic nonlinear terms involving only critical modes .
the cubic nonlinearities we have considered are generated solely by the adiabatic elimination of stable modes from the quadratic terms .
however for an input field which is slightly above threshold there will be a range of unstable modes around the critical one .
the cubic nonlinearity @xmath106 for these modes will be small but non zero .
as the cubic nonlinearities generated by the adiabatic elimination of stable modes from the quadratic terms are also small , it would be necessary to include in the analysis both kinds of cubic nonlinearities .
this is beyond the scope of the present paper .
we have performed several numerical integrations of eq . [ 1 ] using a numerical scheme described in detail in ref .
the method is pseudospectral and second - order accurate in time , and is similar to the so - called two - step method .
lattices of size @xmath191 were used . for @xmath192
there is not a mixed mode pattern since quadratic resonance is not possible due to the fact that @xmath193 .
an example of this situation is shown in fig .
[ t05 ] for @xmath194 . according to the weakly nonlinear analysis
stationary symmetric hexagons are stable close to threshold , however we find numerically a disordered structure for both polarization components of the field @xmath195 and @xmath196 .
as @xmath197 and @xmath198 , the structure observed in @xmath195 is formed by symmetric modes with wavevectors in the ring @xmath199 whereas the structure in @xmath196 is composed of antisymmetric modes with wavevectors @xmath200 .
the overall structure is built on an arbitrary number of symmetric and antisymmetric modes .
the weakly nonlinear analysis predicted that there were no stationary mixed patterns and , in fact , the pattern shown in fig .
[ t05 ] from numerical integration is not static , it evolves dynamically in a slow time scale . for @xmath201 , although resonance is possible , no mixed mode pattern is numerically found , as predicted analytically . as in the previous case ,
a disordered structure is obtained and no steady pattern is reached even for long time integrations .
rings of radius @xmath71 and @xmath72 in the far fields of the linearly polarized components @xmath0 and @xmath68 respectively , show that there is not a selection process of critical wave vectors , all directions become unstable . in fig .
[ t83 ] we show the field configurations for @xmath202 .
the analysis of the previous section showed that mixed mode patters may be expected for @xmath203 .
in fact , for @xmath204 a pattern formed by the superposition of a symmetric mode and the quadratically resonant antisymmetric ones is obtained , see fig .
a roll pattern is seen in the @xmath0 linearly polarized component with wave number @xmath71 , while a rectangular pattern appears in the @xmath68 component with wave number @xmath72 . as for the range of existence of mixed mode patterns , we have found numerically the existence this patterns also for @xmath205 but not for @xmath206 . for @xmath207
we find a steady pattern in which the symmetric modes are damped and a stripe pattern with wave number @xmath72 appears in the @xmath208-polarized component , while @xmath209-polarized component is almost homogeneous ( in fact , a small amplitude stripe pattern of wavevector @xmath210 is observed in @xmath195 which comes from the coupling between @xmath195 and @xmath196 ) .
this pattern made out of antisymmetric modes is the same pattern that is formed when only the antisymmetric instability is present .
overall , the electric field displays an elliptically polarized spatial structure .
similar stationary patterns are found for larger absolute values of the detuning . in fig .
[ t2 ] we show the case with @xmath211 ; this value was chosen in order to induce the formation of a square pattern , since the angle @xmath129 between symmetric and antisymmetric vectors is @xmath212 in this case , but anyway the antisymmetric stripe pattern is formed as described .
we have studied the spatial polarization structures in a mean field model for a kerr medium close to a two - photon resonance and driven by a linearly @xmath213-polarized field with negative detuning . in the self - focusing case the first pattern that is formed as the pump intensity is increased arises from the competition between two stationary instabilities which occur simultaneously .
this codimension two bifurcation appears naturally in the system and it is a consequence of the form of the nonlinearities associated to the two - photon - resonant four - wave mixing process . as this codimension two
bifurcation is not the result of the fine tuning of two parameters , as it is usually the case , it should be much more simple to be observed .
in fact there is only one parameter , the pump , which has to be tuned .
furthermore , we still have another free parameter , the detuning , which allows the system to form different patters without changing the distance to the codimension two instability threshold . near the instability threshold
, we have obtained the evolution equation for the patterns arising from the interacting instabilities using a weakly nonlinear analysis . from these analysis and from the numerical integration of the model
we have shown that we can have the following patterns : a ) dynamical structures involving an arbitrary number of symmetric and antisymmetric modes , b ) a steady state pattern formed by the superposition of a symmetric mode and the quadratically resonant antisymmetric ones and c ) an antisymmetric stripe steady pattern in @xmath196 which is the same that would appear if only the antisymmetric instability were present . as a final remark
we notice that an interesting peculiarity of the situation considered here is that the selection among these patterns can be done in a very natural way , that is , changing the value of the detuning .
helpful discussions with dr .
m. santagiustina are acknowledged .
this work is supported by the european commission through the qstruct tmr network ( project erb fmrx - ct96 - 0077 ) and by direccin general de enseanza superior e investigacin cientfica ( spain ) projects pb94 - 1167 and pb97 - 0141-c02 - 02 .
m. h. wants to acknowledge financial support from the fomec project 290 , dep .
de fsica fceyn , universidad nacional de mar del plata , argentina . | we study pattern formation associated with the polarization degree of freedom of the electric field amplitude in a mean field model describing a nonlinear kerr medium close to a two - photon resonance , placed inside a ring cavity with flat mirrors and driven by a coherent @xmath0-polarized plane - wave field . in the self - focusing case , for negative detunings
the pattern arises naturally from a codimension two bifurcation . for a critical value of the field intensity
there are two wave numbers that become unstable simultaneously , corresponding to two turing - like instabilities .
considered alone , one of the instabilities would originate a linearly polarized hexagonal pattern whereas the other instability is of pure vectorial origin and would give rise to an elliptically polarized stripe pattern .
we show that the competition between the two wavenumbers can originate different structures , being the detuning a natural selection parameter . |
bilayer graphene,@xcite two coupled hexagonal lattices of carbon atoms in the bernal stacking@xcite of graphite , is a two - dimensional zero - gap semiconductor with chiral charge carriers with berry s phase @xmath2 , having a roughly parabolic dispersion at low energies about the corners of the hexagonal first brillouin zone.@xcite these facts are testified by its unusual integer quantum hall effect@xcite ( iqhe ) , featuring a double step in the ladder of the hall conductance in a strong perpendicular magnetic field @xmath3 , observed by novoselov _
_ et al.__@xcite this double step , @xmath4 instead of the common @xmath5 for spin and valley degenerate landau levels , is due to the degeneracy of the @xmath6 landau orbitals.@xcite the gap at the integer quantum hall effect at filling factor @xmath7 has been recently measured with great accuracy,@xcite and the excitations of the iqhe states in the long wavelength limit have also been observed by infrared absorption@xcite and raman spectrocopy.@xcite further broken symmetry states have been observed@xcite in the central landau band at @xmath8 and @xmath9 , and by careful tilted - field measurements it has been shown that they arise predominantly from many - body effects , i.e. , from quantum hall ferromagnetism ( qhf).@xcite quantum hall states with broken symmetry have also been found in the @xmath10 landau level@xcite , and there is also some evidence for a fractional quantum hall plateau .
the eightfold degeneracy of the central group of landau levels is at best approximate , because the zeeman energy is unavoidably present .
while the latter is rather small on the characteristic scale of the interaction energy , a perpendicular electric field can be applied to bias the two layers,@xcite which causes an energy difference between the two valleys .
the competition of the on - site energy difference between the layers and interactions may result in interesting physics , especially at @xmath11.@xcite if the chemical potential is in the gap between landau bands , the low - energy excitations are bound particle - hole pairs,@xcite called magnetoexcitons . as the net charge of such an excitation is zero , taking appropriate linear combinations one obtains eigenstates of the total momentum . in such states
the hole and the particle are bound by the attractive coulomb interaction , forming a dipole with a separation of @xmath12 at center - of - mass wave - vector @xmath13 , where @xmath14 is the magnetic length .
these modes determine the transport gap in the @xmath15 limit . some of these modes couple to circularly polarized light,@xcite while others may be observable in inelastic light scattering experiments.@xcite for monolayer graphene yang _
_ et al.__@xcite studied the intra - landau level excitations of quantum hall ferromagnetic states , and iyengar _
et al._,@xcite bychkov and martinez,@xcite roldn _
et al._,@xcite and lozovik and sokolik@xcite discussed the inter - landau level excitations in detail both for the iqhe and qhf states .
in particular , the linear dispersion of electrons and holes in graphene makes kohn s theorem@xcite inapplicable.@xcite many - body corrections to cyclotron resonance due to the interaction with the dirac sea were calculated by shizuya.@xcite infrared absorption data by jiang _ _ et al.__@xcite and henriksen _ _
et al.__@xcite indicate a contribution from many - body effects to the inter - landau level transitions in these states .
in bilayer graphene henriksen__et al.__@xcite have found that fitting the single - body parameters does not fully explain the observed cyclotron resonance ; deacon _ _ et al.__@xcite and zou__et al.__@xcite have found a significant particle - hole asymmetry , whose origin is still debated.@xcite the wave - vector dependence of the excitations have not been observed so far . theoretically , the intra - landau level excitations in bilayer graphene are well understood . at odd integers in the central landau band ( @xmath16 )
the degeneracy of the @xmath6 landau orbitals causes fluctuations with an in - plane electric dipole character , which gives rise to unusual collective modes.@xcite at even integers in the central landau band ( @xmath17 ) orbital degeneracy does not play a similar role , and the intra - level excitations are still magnetoexcitons.@xcite for inter - ll excitations , the many - body corrections to cyclotron resonance has been calculated by renormalization@xcite including the possible particle - hole symmetry - breaking terms but using the unscreened coulomb interaction , with partial agreement with experiments.@xcite to complement these studies , here we address the issue of the inter - landau level excitations of bilayer graphene in the quantum hall regime .
we incorporate the screening of the interaction by landau level mixing , i.e. , the interaction - induced mixing of excitonic excitations between different landau level pairs .
we study the finite wave vector behavior of excitations .
our paper is organized as follows . in sec .
[ llsec ] we review the tight - binding model of bilayer graphene , and the basic facts concerning its landau levels and orbitals .
our goal is to systematically explore the range of applicability of subsequent simplified models , which neglect several parameters of the slonczewski - weiss - mcclure@xcite ( swm ) model , or account for them on the level of perturbation theory .
we intend to add a few observations to the excellent studies available in the literature.@xcite in sec .
[ kohnnote ] we briefly comment on the reasons why kohn s theorem@xcite does not apply for bilayer graphene . in sec .
[ magnetoexcitons ] we review the adaptation of the mean - field theory of magnetoexcitons to the case of bilayer graphene . in sec .
[ secfilled ] we study the excitations of the iqhe states , and in sec .
[ secpartial ] those of the qhf states .
we conclude in sec .
[ conclusion ] , with an outlook on experimental connections .
each layer of bilayer graphene consists of two sublattices , denoted @xmath18 and @xmath19 in the top layer and @xmath20 and @xmath21 in the bottom layer . in bernal stacking@xcite two sublattices , @xmath20 and @xmath19 in our notation , are exactly above / below one another , while the @xmath18 sites are above the center of the hexagons in the bottom layer , and @xmath21 sites are below the centers of hexagons in the top layer .
see fig . [ stacking ] .
( color online ) bilayer graphene in bernal stacking .
the hopping parameters of the slonczewski - weiss - mcclure model,@xcite conventionally denoted @xmath22 , @xmath23 , @xmath24 and @xmath25 , are also indicated . ] the low - energy physics of bilayer graphene can be adequately described by the tight - binding effective theories that specialize the swm model@xcite of graphite to the case of just two layers . in the vicinity of the valley centers corresponding to the @xmath26 ( @xmath27 ) and @xmath28 ( @xmath29 ) first brillouin zone corners ,
this amounts to using the hamiltonian@xcite @xmath30 where @xmath31 and @xmath32 , @xmath33 m / s is the intra - layer velocity , @xmath34 is the trigonal warping parameter , @xmath23 is the inter - layer hopping amplitude , and @xmath35 is a velocity parameter related to interlayer next - nearest neighbor hopping .
@xmath36 is the zeeman energy ( with @xmath37 being the gyromagnetic factor and @xmath38 the bohr magneton ) .
this hamiltonian acts in the basis of sublattice bloch states @xmath39 $ ] in valley @xmath26 and @xmath40 $ ] in valley @xmath28 . here
@xmath41 is the intra - layer hopping amplitude , @xmath42 is the interlayer hopping amplitude between sites above each other in the two layers .
further , @xmath43 and @xmath44 are next - nearest neighbor interlayer hopping amplitudes , as shown in fig . [ stacking ] .
@xmath45 is the on - site energy difference between the dimer sites ( @xmath46 ) and the non - dimer sites ( @xmath47 ) .
finally , @xmath48 is the potential energy difference between the layers , which may arise , e.g. , because of an applied perpendicular electric field @xmath49 .
the hamiltonian in eq .
( [ hamilton ] ) is block - diagonal in the valley index , which is conveniently described as a pseudospin . in the special case @xmath50 the system has su(2 )
pseudospin rotation symmetry . in the theoretical limit @xmath51
this is raised to su(4 ) symmetry . in this paper
we will treat @xmath52 and @xmath48 as small perturbations in comparison to the interaction energy , i.e. , we will work in the @xmath53 limit , where @xmath54 is the relative dielectric constant of the environment .
we set @xmath55 . for small momenta , @xmath56 , the two low - energy bands of the hamiltonian @xmath57 that touch each other at @xmath26 and @xmath28 in the case of vanishing magnetic field can be attributed to a @xmath58 effective hamiltonian@xcite @xmath59 where @xmath60 , and @xmath61 acts on @xmath62 $ ] in valley @xmath26 and @xmath63 $ ] in valley @xmath28 .
the landau levels ( ll s ) and landau orbitals , respectively , of the two - band hamiltonian @xmath61 are @xmath64 here @xmath65 is an integer , @xmath66 is an appropriate normalization factor , and @xmath67 are the single - particle states in the conventional two - dimensional electron gas ( 2deg ) with quadratic dispersion in the landau gauge @xmath68 , @xmath69 and @xmath70 is a hermite - polynomial . the @xmath6 orbitals are degenerate in the @xmath71 limit , and they have a layer polarization for @xmath72 . at realistic values of @xmath52 and @xmath48 ,
the @xmath6 , @xmath72 , @xmath73 states form a quasidegenerate band we will call the central landau level octet .
notice that @xmath74 and @xmath75 for @xmath71 .
we would like to determine how neglecting @xmath24 , @xmath25 , and @xmath45 changes the single - body orbitals of the four - band hamiltonian @xmath57 in eq .
( [ hamilton ] ) , and how much these differ from the simplified two - band model @xmath61 in eq .
( [ lowenergy ] ) .
the values of the swm parameters for bilayer graphene was estimated by a combination of infrared response analysis and theoretical techniques by zhang _ _ et al.__@xcite they found @xmath76 these ratios are based on @xmath77 ev . while somewhat greater values of @xmath22 are also available in the literature,@xcite we use these values for a robustness analysis .
for the particle - hole symmetry breaking term we use @xmath78 from the recent electron and hole mass measurement by zou _
et al._,@xcite which is slightly greater than the value in ref . .
( color online ) the overlap of the landau orbitals with the `` ideal '' limit , @xmath79 , as the swm parameters @xmath80 are gradually tuned from zero to their literary values [ eqs .
( [ estim ] ) to ( [ estimw4 ] ) ] for the lowest - energy landau levels . for
the effect of @xmath24 in the two - band model see ref . .
] with @xmath81 , @xmath82=1 $ ] , the hamiltonian can be expressed in terms of these landau level ladder operators .
then the eigenstates of @xmath57 can be calculated numerically .
[ overlap ] shows the overlap of the landau orbitals with the `` ideal '' limit @xmath79 as the swm parameters @xmath80 are tuned from zero to their literary values [ eqs .
( [ estim ] ) to ( [ estimw4 ] ) ] for the central ( @xmath6 ) and the two pairs of nearby ( @xmath83 ) landau levels . at small magnetic field
@xmath84 t trigonal warping alone significantly changes the orbitals from their ideal limit .
switching on @xmath25 hardly affects the central levels , but for @xmath85 it changes the electron and hole pairs ( @xmath86 ) differently , as expected from this electron - hole symmetry breaking term .
finally , the inclusion of @xmath45 hardly affects the orbitals .
these changes , however , are already small at modest fields ( @xmath87 t ) , and are further suppressed as the experimentally relevant range ( @xmath88 t ) is approached . thus neglecting the @xmath80 swm parameters
is justified in the high magnetic field range where quantum hall experiments are typically performed . as the two - band model in eq .
( [ lowenergy ] ) applies for small momenta , and the low - index landau orbitals have a small amplitude at high momenta , the two - band model is expected to be valid for the lowest few landau levels .
the landau orbitals of the two - band model have a large overlap with those of the four - band model in the `` ideal '' limit @xmath79 : 1 , 0.9995 , 0.9992 , 0.9987 for @xmath89 , respectively .
we conclude that using the landau states of two - band model in eq .
( [ lowenergy ] ) instead of these of the four - band model @xmath57 in the lowest - energy landau bands does not introduce further inaccuracy beyond the neglect of @xmath90 and @xmath45 .
therefore , we take @xmath61 as our starting point .
kohn s theorem@xcite states that interactions do not shift the cyclotron resonance in a parabolic band .
it applies equally to two- and three - dimensional systems .
it is not applicable to linear bands in monolayer graphene.@xcite we will see that it also fails for bilayer graphene , even though the bands of @xmath57 in eq .
( [ hamilton ] ) start quadratically at low energies , and those of @xmath61 in eq .
( [ lowenergy ] ) are exactly parabolic for @xmath91 . for the conventional 2deg , kohn s theorem follows because the interaction with a radiation field , @xmath92 where @xmath93 is the canonical momentum that includes vector potential of the homogeneous magnetic field , is proportional to @xmath94 , where @xmath95 and @xmath96 .
@xmath97 and @xmath98 are generators of global translations , hence they commute with any translation - invariant interaction ; moreover , @xmath99 act as ladder operators among the eigenstates of the total kinetic energy , @xmath100 . therefore , @xmath101 connects eigenstates of the total hamiltonian and conserves the interparticle interaction . @xmath102 may be described classically or quantum mechanically ; @xmath103=0 $ ] can also be checked directly . in the two - band model of bilayer graphene
the interaction with the radiation field @xmath102 is@xcite @xmath104 with @xmath105 .
it is straightforward to show that @xmath101 maps a state @xmath106 to a linear combination of @xmath107 , @xmath108 , @xmath109 and @xmath110 .
if three out of these transitions are pauli - blocked , we may end up in an eigenstate of the kinetic part of the many - body hamiltonian , but @xmath101 is no longer proportional to a linear combination of @xmath97 and @xmath98 .
in fact , it no longer commutes with them .
thus the interaction energy may differ in the electromagnetically excited many - body state and the initial state .
when the fermi energy is in a landau gap ( e.g. at filling factor @xmath111 in bilayer graphene ) , the integer quantum hall effect@xcite occurs in samples with moderate disorder.@xcite quantum hall states also occur at other integer filling factors because the exchange interaction favors symmetry - breaking ground states called quantum hall ferromagnets ; single - body terms such as the zeeman energy play a secondary role .
qhf s emerge at odd integer fillings in two - component systems even if the zeeman energy is tuned to zero .
this observation straightforwardly generalizes for su(@xmath112 ) systems.@xcite because of the clear separation of the filled and empty landau bands in the ( mean - field ) ground - state , the excitations of both classes of quantum hall systems are described in the same way .
the relevant low - energy excitations are magnetoexcitons,@xcite which are obtained by promoting an electron from a filled landau band to an empty band.@xcite these neutral excitations have a well - defined center - of - mass momentum @xmath113 .
they approach widely separated particle - hole pairs in the @xmath114 limit .
the latter limit determines the transport gap unless skyrmions form.@xcite magnetoexcitons are created from the ground - state by operators@xcite @xmath115 where @xmath116 ( @xmath117 ) specifies the landau band where the particle ( hole ) is created and @xmath18 is the area of the sample .
magnetoexcitons carry spin and pseudospin ( valley ) quantum numbers , as derived from the particle and hole landau bands involved . while the projections @xmath118 of the spin and the pseudospin are always good quantum numbers , their magnitudes @xmath119 and @xmath120 are well - defined only for ground - states that are spin or pseudospin singlets , respectively .
it is common practice@xcite to define the quantity @xmath121 and consider it the `` angular momentum quantum number '' of the exciton .
we emphasize that @xmath122 is exactly conserved by the electron - electron interaction only in the @xmath123 limit , where it is related to angular momentum .
the emergence of this quantity is best seen in the two - body problem of the negatively charged electron and the positively charged hole , as discussed in the appendix . at any finite wave vector transitions with different @xmath124
may mix . in the low magnetoexciton density limit
the interaction between magnetoexcitons is neglected . the mean - field ( hartree - fock ) hamiltonian of magnetoexcitons is well known from the literature,@xcite and so is its adaptation to spinorial orbitals:@xcite @xmath125 where @xmath116 , etc . , and @xmath126 . the first term of the r.h.s .
is the single - body energy difference of the @xmath127 and @xmath128 states , which includes the wave vector independent exchange self - energy difference of the two states . while the exchange self - energy itself is infinite for any orbital , its difference between two states
, @xmath129 is finite .
( we will define @xmath130 soon . )
this is a peculiarity of bilayer graphene .
a simple regularization procedure works for the four - band model.@xcite for monolayer graphene , a proper renormalization procedure is required.@xcite the shift @xmath131 is analogous to the lamb shift in quantum electrodynamics.@xcite the next term is the direct dynamical interaction between the electron and the hole : @xmath132 this term is diagonal both in spin and pseudospin , @xmath133 , but not in landau orbital indices . finally , the last term in eq .
( [ hmf ] ) is the exchange interaction between the electron and the hole , @xmath134 which is @xmath135 , thus couples transitions that conserve the spin @xmath136 and the valley @xmath137 of the electron and the hole individually .
sometimes we will call it the rpa contribution , as it is related to particle - hole annihilation and recreation processes .
( it is also called@xcite the depolarization term . )
notice @xmath138 vanishes in the @xmath139 limit .
we have used the notation @xmath140+\\ v^d(p)\left [ a^{(n_2')}_{\xi_2 ' } b^{(n_1)}_{\xi_1 } b^{(n_1')}_{\xi_1 ' } a^{(n_2)}_{\xi_2 } f^\ast_{|n_2||n_2'|}(\mathbf p)f_{|n_1|-2,|n_1'|-2}(\mathbf p)+ b^{(n_2')}_{\xi_2 ' } a^{(n_1)}_{\xi_1 } a^{(n_1')}_{\xi_1 ' } b^{(n_2)}_{\xi_2 } f^\ast_{|n_2|-2,|n_2'|-2,}(\mathbf p)f_{|n_1||n_1'|}(\mathbf p)\right],\end{gathered}\ ] ] @xmath141 here @xmath142 , @xmath143 , and @xmath144 is related to the fourier transform of @xmath145 in eq .
( [ etaeq ] ) , and @xmath146 is an associated laguerre polynomial .
the difference between the intralayer coulomb interaction @xmath147 and the intralayer one , @xmath148 where @xmath149 nm is the distance between the layers , is neglected as a first approximation .
the @xmath150 and @xmath151 numbers correspond to the spinorial structure of the single - body states in eq .
( [ genericorbitals ] ) : @xmath152 , @xmath153 and @xmath154 for @xmath85 .
notice that @xmath155 which follow from the similar properties of @xmath130 .
the @xmath156 limit of the magnetoexciton dispersion determines the many - body contribution to the cyclotron resonance , which may be nonvanishing in graphene systems ( c.f .
[ kohnnote ] and refs . and ) . the mean field hamiltonian matrix @xmath157 in general mixes transitions among different electron - hole pairs , restricted only by conservation laws .
landau level mixing effectively screens the interaction .
sometimes the magnetoexciton spectra are obtained using a screened model interaction instead of the bare coulomb , not letting ll transitions
mix.@xcite we believe such an approach is suitable in the @xmath158 limit , where an additional quantum number @xmath122 also restricts ll mixing , and for intra - ll modes . at finite wave vector
the mean - field theory with ll mixing removes spurious level crossings in the excitation spectra and provides insight into the orbital structure of the excitations .
technically , however , the infinite @xmath157 matrix needs to be truncated .
we first consider the states where the chemical potential is between two orbital landau bands .
this occurs at filling factor @xmath111 in bilayer graphene .
together with @xmath159 and @xmath160 , the magnitude of the spin @xmath119 and of the pseudospin @xmath120 are quantum numbers .
( in the @xmath161 limit an su(4 ) classification is also possible . ) magnetoexciton modes at the @xmath162 integer quantum hall effect in bilayer graphene .
only the optically relevant spin- and pseudospin - conserving modes are shown . in the @xmath156 limit , which is probed by purely optical experiments , the transitions in the infinite sequence at fixed @xmath163 may mix ; at @xmath164 , the excitations result from the mixing of all @xmath122 sequences .
the modes at @xmath165 are obtained by particle - hole conjugation . ] with the hole ( @xmath166 ) and the electron ( @xmath112 ) landau orbitals fixed , the sixteen possible transitions belong to four classes : ( i ) a spin singlet , pseudospin singlet state : @xmath167 ( ii ) a spin singlet , pseudospin triplet multiplet .
the @xmath168 member of this multiplet is @xmath169 ( iii ) a spin triplet , pseudospin singlet , which contains following the @xmath170 state : @xmath171 ( iv ) a nine - member multiplet that is triplet is both spin and valley .
its @xmath172 member is @xmath173 fig .
[ sketchinteger ] depicts these modes for @xmath162 .
the exchange interaction between the electron and the hole contributes only to states generated by @xmath174 . in all other excitation modes the rpa term cancels ( c.f .
the sign alternation@xcite in modes @xmath175 , @xmath176 and @xmath177 ) or is prohibited by quantum numbers . thus in the absence of accidental degeneracies , we expect a collection of nondegenerate excitations and one of fifteen - fold degenerate excitations ; the latter is decomposed as @xmath178 if @xmath179 and as @xmath180 if @xmath181 .
the extent interactions may mix transitions involving different landau level pairs depends on the interaction - to - kinetic energy ratio , parametrized by @xmath182 notice that @xmath183 in the @xmath184 limit just like for the conventional two - dimensional electron gas . realistically
( @xmath185 t , @xmath186 ) , @xmath187 ; this is by no means a small perturbation . in the conventional 2deg ll mixing
is suppressed at high fields because of the @xmath188 scaling of the relative strength of interactions , while in monolayer graphene both the interaction and the kinetic energy scale with @xmath189 , thus ll mixing is never suppressed . in bilayer
graphene the kinetic term of the hamiltonian interpolates between quadratic at small momenta and linear at high momenta ; thus ll mixing gets suppressed only for the ll s with a small index , whose orbitals are built up from low - momentum plane waves . for high - index ll s the ratio of interaction to kinetic energy is only weakly @xmath19-dependent . at fixed magnetic field and filling factor , ll mixing in bilayer graphene
is more significant than in a conventional 2deg , implemented , e.g. , in gaas quantum quantum wells , because of the smaller dielectric constant and effective mass in bilayer graphene .
one can also compare the coulomb energy scale , @xmath190 to the energy difference between adjecent landau levels , @xmath191 .
this may give the impression that ll mixing is more important at higher filling factors , but the amplitude of the undulations of the unmixed magnetoexciton dipersions also gets reduced in higher levels ; leaving the issue of the generic progress of ll mixing with increasing filling factor open .
for @xmath158 the mean - field hamiltonian mixes magnetoexcitons with different electron and hole landau levels at fixed @xmath122 , and for @xmath164 it also mixes different @xmath122 subspaces ; c.f .
[ sketchinteger ] . restricting ll mixing to a fixed @xmath122 subspace
might give the impression that ll mixing is just a quantitative correction for the long wave length part of the lowest excitation curves , resulting in increased electron - hole binding energies.@xcite however , already the lowest excitations in the different @xmath122 sectors mix strongly at finite wave vector . as the side panels of fig .
[ integer ] show , the excitations have a large projection on the @xmath122 subspaces different from their own @xmath122 in the @xmath156 limit , and may eventually be contained in one of the other subspaces for large @xmath192 ; this is an unavoidable consequence of the elimination of crossings by ll mixing .
the mixing of landau levels is especially strong in the nondegenerate excitations , which are strongly affected by the exchange interaction between the electron and the hole ( the rpa term ) in the @xmath193 region .
thus in the rest of this paper we allow the mixing of transitions restricted by @xmath194 and a maximum number @xmath195 at each fixed @xmath122 .
[ integer ] shows transitions at @xmath196 with @xmath197 and @xmath198 , while fig .
[ integer2 ] shows @xmath199 and 12 .
we will use this truncation in the spectra shown in the rest of this paper.@xcite figs .
[ integer ] and [ integer2 ] also show the kinetic energy difference between the electron and the hole for comparison .
our mean - field theory predicts an interaction shift comparable to this energy .
this prediction will be revisited with methods beyond mean - field.@xcite the @xmath156 limit of the magnetoexcitons are commonly probed by optical absorption and electronic raman scattering .
the selection rules@xcite ensure that only the @xmath200 mode and the @xmath175 , @xmath176 and @xmath177 modes of the fifteen - fold degenerate curve are active , @xmath201 is absorption and @xmath202 in raman .
particle - hole conjugation relates @xmath203 to @xmath204 ( @xmath112 integer ) with the sign of @xmath122 reversed .
( color online ) the excitations of the integer quantum hall state at @xmath199 and @xmath205 , at @xmath206 . ]
with an integer filling factor different from @xmath207 , a landau band quartet ( @xmath208 ) or octet ( @xmath209 ) is partially filled in the single electron picture .
the minimization of the interaction energy results in gapped states which break either spin rotation or pseudospin ( valley ) rotation symmetry , or both.@xcite if either the zeeman energy @xmath52 or the interlayer energy difference @xmath48 is present , they affect the order how the landau levels are filled , but exchange energy considerations are more crucial in most cases.@xcite the most convenient basis in pseudospin space may differ ; we may introduce @xmath210 with a proper choice of @xmath211 and @xmath212 , eqs .
( [ defs]-[defa ] ) include states of definite valley , bonding and antibonding states , or intervalley phase coherent states .
corresponding magnetoexciton operators are defined in an obvious manner . in particular , if @xmath213 , the @xmath11 qhf state is ferromagnetic and the choice of the pseudospin basis is irrelevant ( fig .
[ sketchnu0 ] ) . for @xmath214 ,
both the @xmath215 and @xmath216 orbital landau levels of identical spin and pseudospin are filled , where @xmath212 and @xmath211 are determined by electrostatics ( fig .
[ sketchnu-2 ] ) . for odd @xmath217 , an interlayer phase coherent ( @xmath218 ) state exists for sufficiently small @xmath48 , which yields@xcite to a layer polarized state ( @xmath219 ) at @xmath220 and @xmath221 , and to a sequence of states with partial or full orbital coherence@xcite at @xmath222 and @xmath223 .
notice that @xmath220 and @xmath223 are not related by particle - hole symmetry .
this is best understood from hund s rules@xcite : at @xmath220 only one @xmath215 orbital band is occupied , while at @xmath223 only one @xmath216 orbital band is empty . at @xmath223 ct _ _ et al.__@xcite showed that orbitally coherent states dominate the phase diagram , whose inter - ll excitations are beyond the scope of this study .
the case of @xmath220 is depicted in fig .
[ sketchnu-3 ] . beyond @xmath159 and @xmath160 ,
the magnitudes @xmath120 or @xmath119 are quantum numbers at half - filling @xmath11 .
all excitons include transitions between the non - central levels @xmath224 ; the possibility of transitions from , to , or within the central landau level octet depends on the ground state , which also resolves the transitions through the exchange self - energy differences ; see figs .
[ sketchnu0 ] , [ sketchnu-2 ] and [ sketchnu-3 ] .
the excitations are grouped by their optical signature . due to the small momenta of optical photons , valley
flipping modes are optically inactive . in the @xmath156 limit @xmath122
becomes a quantum number , and @xmath201 applies for single - photon absorption,@xcite and @xmath225 in electronic raman processes,@xcite with the @xmath202 transitions being dominant .
the angular momentum due to the helicity of the photons is transferred@xcite entirely to the orbital degree of freedom .
optically inactive modes include goldstone modes associated with the broken symmetry ( outside the scope of our study ) and generic dark modes .
it is known that two qhf ground states exist , a spin - polarized one and a valley ( layer ) polarized one@xcite .
their respective range of validity is determined by the ratio of the zeeman energy @xmath52 to the energy difference between the valleys , which in turn is related to the potential difference @xmath48 .
( in fact , layer and pseudospin can de identified within the central landau level octet . ) for concreteness , we are discussing the ferromagnetic state . here
the magnitude of the pseudospin @xmath120 is a good quantum number of the excitations .
see fig .
[ sketchnu0 ] for the transitions that span the hilbert space of the mean - field hamiltonian .
magnetoexciton modes at the @xmath11 quantum hall ferromagnetic state in bilayer graphene .
only the optically relevant spin- and pseudospin - conserving modes are shows . in the @xmath156 limit , which is probed by photon absorption or electronic raman , the transitions in the infinite sequence at fixed @xmath163 may mix ; at @xmath164 , the excitations result from the mixing of all @xmath122 sequences . ] with the electron ( @xmath112 ) and hole ( @xmath166 ) landau levels fixed , the @xmath226 transitions consist of a pseudospin triplet@xcite @xmath227 , @xmath228 , @xmath229 , and a singlet@xcite @xmath230 .
this group contains the intralevel transitions among the @xmath6 landau bands ; the goldstone modes associated with the spin rotational symmetry breaking should be in this subspace . however , our approach is not appropriate for the description of goldstone modes even at even filling factors , as we will discuss below .
the @xmath231 pseudospin triplet , @xmath232 , @xmath233 , @xmath234 , and pseudospin singlet @xmath235 , respectively , contains inter - ll transitions only .
the @xmath170 sector consists of ( i ) two triplets , @xmath236 , @xmath237 , @xmath238 , and @xmath239 , @xmath240 , @xmath241 , the rpa terms does not contribute to , and ( ii ) two singlets , @xmath242 and @xmath243 , which are mixed by the rpa term .
careful inspection reveals , however , that the two pseudospin singlets ( ii ) always appear in the mean - field hamiltonian on equal footing , e.g. , the @xmath244 transition is indistinguishable on the mean - field level from the @xmath245 transition .
this follows by @xmath246 and the following easily provable identity of the exchange self - energy cost : @xmath247 ( for @xmath216 or @xmath248 no sign change is necessary . )
( [ idph1 ] ) simply expresses particle - hole symmetry , i.e. , that the exchange self - energy cost of transitions related by particle - hole conjugation in a fixed component must be identical .
see fig .
[ phconj ] for the transitions whose comparison yields eq .
( [ idph1 ] ) .
imagine two transitions : ( i ) a particle in level @xmath166 is promoted to level @xmath112 of a valley - spin component ( left ) of which the @xmath216 landau band is filled , and ( ii ) @xmath249 of a component ( right ) of which the @xmath216 band is empty .
particle - hole conjugation of ( i ) is the promotion of a hole from landau level @xmath250 to @xmath251 , i.e. , ( ii ) ; the excitation energies must be the same .
the consequence is eq .
( [ idph1 ] ) ; the formal proof is straightforward . ]
the rpa terms are the same in each diagonal and off - diagonal position among equivalent transitions , thus they select the even and the odd linear combinations in group ( ii ) .
the even combination , @xmath252 defined in eq .
( [ m0 ] ) , gets an rpa enhancement , while the rpa cancels from the alternating sign combinations , making it energetically equivalent to the @xmath168 element of the triplets ( i ) .
thus , eventually , the @xmath253 sector contains a three - fold degenerate curve and a nondegenerate mode .
each of the four multiplets in the @xmath170 sector contains a @xmath168 mode , which is active in electronic raman or ir absorption . here
the mixing of landau levels results in more widely separated modes .
the optically active excitations are shown in fig .
[ spectrumnulla ] with ll mixing taken into account .
notice that the @xmath254 and @xmath255 transitions have an equal weight in all modes , consistent with the particle - hole symmetry at @xmath11 .
in the @xmath226 sector we find spin waves .
neglecting landau level mixing , they give rise to a gapless and a gapped intra - ll modes,@xcite and a sequence of higher inter - ll modes ; each of these is raised by the zeeman energy and split by the valley energy difference in turn .
the interaction , however , mixes these excitations , thus a clear - cut classification into intra - ll and inter - ll is no longer possible .
level repulsion unavoidably lowers the formerly gapless modes .
this effect yields apparently negative excitation energies at small wavelength .
goldstone s theorem , however , ensures that a gapless spin - wave mode is associated with the breaking of the spin rotation symmetry .
consequently , the seemingly negative energy of the lowermost excitation with a large intra - ll component is an artifact of the combination of hartree - fock mean - field theory and ll mixing .
the same anomaly occurs for monolayer graphene,@xcite but it is less apparent when the particle - hole binding energy is plotted . the @xmath214 state breaks the spin and pseudospin rotational symmetries as the ground state fills the @xmath215 and @xmath216 orbitals of the most favorable spin - pseudospin component , @xmath256 .
we restrict the discussion to spin and pseudospin preserving excitations .
see fig .
[ sketchnu-2 ] for the possible transitions .
it is easy to check that the mean - field hamiltonian matrix is identical to the one at @xmath11 . for the @xmath257 transitions ( @xmath85 integer ) this holds
because the occupancy of the central landau level octet is irrelevant as @xmath258 the octet of @xmath259 and @xmath260 transitions gives rise to two quartets of equivalent transitions by eqs .
( [ ide]-[idph1 ] ) . while at @xmath11 the @xmath256 and @xmath261 transitions of the former group bundle with the @xmath262 and @xmath263 transitions of the second group , now the @xmath256 transition of the first group bundle with the @xmath261 , @xmath262 and @xmath263 transitions of the second group
. the spectrum is still the one in fig .
[ spectrumnulla](a ) .
the orbital projection of the modes differs , c.f . fig .
[ spectrumketto ] for @xmath264 . at @xmath265 the sign of @xmath122 changes in all projections , which determines the helicity of the absorbed and inelastically scattered photons . the spin- and pseudospin - conserving magnetoexciton modes at the @xmath264 quantum hall ferromagnetic state in bilayer graphene .
the modes at @xmath265 are obtained by particle - hole conjugation . ]
( color online ) the projection of the excitations at @xmath264 on the definite @xmath122 subspaces . for degenerate curves
the weight is summed .
the spectrum at @xmath214 is identical to the one in the left panel of fig .
[ spectrumnulla ] , the letters refer to the same curves . for @xmath265 the sign of the @xmath266 and @xmath267 projections
are interchanged w.r.t .
@xmath264 . ] based on exchange energy considerations within the central landau level octet , hund s rule@xcite implies that the only occupied band is @xmath268 at @xmath220 .
the states in eq .
( [ defs ] ) progress from the layer balanced limit @xmath269 at @xmath50 to the layer polarized state @xmath219 ; this limit is achieved about @xmath270 , which is only 0.082 mev at @xmath271 t. for @xmath272 there is interlayer phase coherence.@xcite thus magnetoexcitons exist on both side of @xmath273 ; the amount electrostatics raises energy of the pseudospin - flipping modes w.r.t .
the pseudospin conserving modes saturates at @xmath274 .
both the spin and the pseudospin symmetries are broken resulting in three goldstone modes.@xcite further , barlas _ _ et al.__@xcite showed that at finite @xmath48 there is an instability to a stripe ordered phase with a rather small critical temperature .
our analysis below applies only below this temperature .
see fig .
[ sketchnu-3 ] for the transitions that span the hilbert space of the mean - field hamiltonian .
we have become aware of ref . , which derives a modified version of hund s rule considering the exchange field due to all filled levels , not just those in the central landau level octet .
while this makes no difference for even filling factors , at @xmath220 the predicted mean - field ground state fills a band of states which are equal linear combinations of @xmath216 and @xmath215 landau orbitals .
the calculation of the excitations of such ground states is delegated to future work .
the optically spin- and pseudospin - conserving magnetoexciton modes at the @xmath220 quantum hall ferromagnetic state in bilayer graphene .
notice that the @xmath262 , @xmath261 and @xmath263 transitions occur symmetrically , i.e. , the same electron and hole landau levels are allowed and the same self - energy cost is picked up from the exchange with the filled levels . ] because of the degeneracy of @xmath6 orbitals , states in central landau level octet at odd integer fillings involve fluctuations with in - plane electric dipole character.@xcite the consequent collective modes have been studied in detail by barlas _
et al_.@xcite and ct _
et al_.@xcite as we do not handle such dipolar interactions , we have omitted the predominantly intra - ll lowest curve from the spectra , and we have checked that the inter - ll excitation modes we keep contain the @xmath275 magnetoexcitons with a negligible weight .
reassuringly , we always got a weight less than 0.1% .
as the @xmath215 orbital is filled with @xmath256 electrons in the mean - field ground state , for fixed electron ( @xmath112 ) and hole ( @xmath166 ) landau levels the exchange self - energy cost of the @xmath256 transition is higher than those of the other components .
also , in the @xmath256 component an intralevel @xmath275 transition is possible , which mixes with higher @xmath256 transitions ; see fig .
[ sketchnu-3 ] for the restrictions on the possible transitions at this filling .
the other three components , on the other hand , occur symmetrically in the mean - field hamiltonian ; c.f . fig .
[ sketchnu-3 ] .
one can change basis from the excitons of type @xmath262 , @xmath261 and @xmath263 to @xmath276 the rpa contribution cancels from @xmath277 and @xmath278 , which give rise to doubly degenerate excitations .
@xmath279 has an rpa contribution .
its mixture with the distinguished @xmath256 excitations produces nondegenerate curves . in higher energy excitations , on the other hand ,
the weight of the @xmath275 transition of the @xmath256 component becomes extremely small , thus the equivalence of the four components will be approximately restored , yielding threefold quasi - degenerate and nondegenerate curves .
see fig .
[ spectrumharom ] for the dispersion of the active excitations .
the small graphs show that the projection to definite @xmath122 subspaces changes abruptly at nonzero wave vector .
notice that at higher energies @xmath280 the double degenerate curves occur in the vicinity of a nondegenerate one , indicating the approximate restoration of the equivalence of components in this limit . just like at @xmath281 ,
the state at @xmath222 is not the particle - hole conjugate of the state @xmath221 .
we do not study @xmath222 because of the relavance of orbitally coherent states.@xcite at @xmath221 the interlayer coherent and layer polarized qhf states both have magnetoexcitons .
the possible transitions in the mean - field hamiltonian are obtained trivially by raising the fermi level by four levels in fig .
[ sketchnu-3 ] .
now the @xmath262 electrons are distinguished by the possibility of an intralevel @xmath275 transition , and their higher self - energy .
the other three components occur symmetrically in the mean - field hamiltonian .
the argument is similar to the case of @xmath220 ; for the two @xmath282 transitions of spin-@xmath283 electrons and the @xmath284 transitions of the spin-@xmath285 electrons one uses @xmath286 and eq .
( [ idph1 ] ) . when subsequent @xmath201 transitions are included , the @xmath287 transition of @xmath263 is equivalent to the @xmath259 transition of @xmath256 and @xmath261 , while the @xmath287 transition of @xmath261 and @xmath256 is equivalent to the @xmath259 transition of @xmath263 .
the convenient basis change is similar to eqs .
( [ change1 ] ) to ( [ change3 ] ) , with @xmath256 replaced by @xmath263 .
there is only a slight difference between the inter - landau level excitation spectra at @xmath220 and @xmath221 : the exchange self - energy cost of the distinguished transition ( @xmath256 at @xmath220 and @xmath262 at @xmath221 ) relative to the three equivalent ones is @xmath288 at @xmath220 and @xmath289 at @xmath221 .
as this difference is already small for the lowest transitions and then decreases , we omit the @xmath221 spectrum ; its difference from fig .
[ spectrumharom](a ) is comparable to the line width .
we have calculated the inter - landau level magnetoexcitons in the integer quantum hall states as wells as the quantum hall ferromagnets at filling factor @xmath290 of bilayer graphene .
we have found that the spinorial structure of the orbitals together with the enhanced electron - hole exchange interaction effects in this multicomponent system gives rise to rather complex dispersions ; these are related both to the shape of the fourier transform of landau orbitals and to the elimination of crossings by the mixing of landau levels , which is significant because the scale of the coulomb energy is comparable to the cyclotron energies .
the @xmath156 limit of the excitation can be probed by optical absorption@xcite and electronic raman experiments.@xcite unlike for the conventional two - dimensional electron gas with a quadratic dispersion,@xcite the excitations in a quantizing magnetic field do acquire an interaction shift ; the magnitude of such a shift is one of our experimental predictions .
the wave vector dependence of the magnetoexcitons can be probed in resonant inelastic light scattering experiments,@xcite where momentum conservation breaks down mainly because of ineffectively screened charged impurities . in semiconducting samples this technique has been successfully applied , e.g. , to the study of spin - conserving and spin - flipping excitations at integers,@xcite of the long wave - length behavior of the low - lying@xcite and higher@xcite excitations of fractional quantum hall states , and of magnetoroton minuma at @xmath291 @xcite and at fractions.@xcite in graphene , the magnetophonon resonance was observed by this method.@xcite the mixing of transitions that involve different landau levels is strong in the experimentally accessible range .
this mixing smoothens the dispersion relations via level repulsion , and at finite wavelength causes a strong mixing of the modes that have different angular momenta in the zero wavevector limit . in particular
, we have found that the anticrossings due to landau level mixing result in a number of undulations in the magnetoexciton dispersions , whose van hove singularities must give a strong signal.@xcite on the theory side , we have found that the classification of magnetoexciton modes at finite wave vectors by the `` angular momentum quantum number '' is problematic , that the screening effects which we have handled via landau level mixing are significant , and that this framework is not quite suitable for studying the goldstone modes such as spin waves in the symmetry - breaking quantum hall ferromagnetic states .
this research was funded by the hungarian scientific research funds no . k105149 . c. t. was supported by the srop-4.2.2/08/1/2008 - 0011 and srop-4.2.1.b-10/2/konv-2010 - 0002 grants , and the hungarian academy of sciences .
let us consider the hamiltonian @xmath292 where @xmath293 , @xmath294 belongs to the electron and @xmath295 belongs to the hole , and @xmath296 .
( we have fixed the valley of both the electron and the hole .
a valley - independent interaction is assumed because any deviation from this is small in the ratio of the lattice constant to the magnetic length . ) introducing center - of - mass and relative coordinates @xmath297 , @xmath298 and momenta @xmath299 , @xmath300 , and separating the center - of - mass motion by the canonical transformation @xmath301 , we obtain @xmath302 where the independent harmonic oscillators @xmath303 are defined as @xmath304 in complete analogy to the case of the monolayer@xcite , the eigenstates of the kinetic part of @xmath305 are @xmath306 in terms of two - dimensional harmonic oscillator eigenstates , @xmath307 above @xmath308 is if @xmath215 or @xmath216 , else it is sgn(@xmath112 ) .
thus the kinetic energy operator clearly commutes with the operator @xmath309 which returns @xmath122 as an eigenvalue .
this operator , however , commutes with the complete @xmath305 only in the @xmath310 limit .
thus we can regard the electron - hole bound state as a two - dimensional harmonic oscillator with clockwise ( @xmath311 ) and anti - clockwise ( @xmath312 ) excitations placed in an external confinement potential . a nonzero center - of - mass motion breaks the rotational symmetry of this confinement and starts to couple the @xmath313 states . here
we closely follow iyengar _
_ et al.__@xcite the two - body hamiltonian in eq .
( [ twobody ] ) can be written is terms of center - of - mass and relative coordinates and momenta as @xmath314 where @xmath315 , and @xmath316 . with the application of the canonical transformation @xmath301
, we obtain @xmath317 eq . ( [ transformed ] ) follows by substituting eq .
( [ harmonic ] ) into eq .
( [ intermed ] ) . notice that @xmath305 in eq .
( [ transformed ] ) is independent of the transformed center - of - mass coordinates .
thus @xmath318 is conserved ; in original variables this corresponds to @xmath319 .
e. v. gorbar , v. p. gusynin , and v. a. miransky , jetp lett .
* 91 * , 314 ( 2010 ) ; phys .
b * 81 * , 155451 ( 2010 ) ; e. v. gorbar , v. p. gusynin , v. a. miransky , and i. a. shovkovy , phys . rev .
b * 85 * , 235460 ( 2012 ) .
a. bychkov , s. v. iordanskii , and g. m. eliashberg , pisma zh .
33 * , 152 ( 1981 ) [ sov .
jetp lett .
* 33 * , 143 ( 1981 ) ] ; yu .
bychkov , e. i. rashba , zh .
. fiz . * 85 * , 1826 ( 1980 ) [ sov .
jetp * 58 * , 1062 ( 1983 ) ] .
a. pinczuk , b. s. dennis , l. n. pfeiffer , k. w. west , semicond .
* 9 * , 1865 ( 1994 ) .
a. pinczuk , b. s. dennis , l. n. pfeiffer , and k. west , phys . rev
70 * , 3983 ( 1993 ) .
t. d. rhone , d. majumder , b. s. dennis , c. hirjibehedin , i. dujovne , j. g. groshaus , y. gallais , j. k. jain , s. s. mandal , aron pinczuk , l. pfeiffer , and k. west , phys .
lett . * 106 * , 096803 ( 2011 ) .
a. pinczuk , s. schmitt - rink , g. danan , j. p. valladares , l. n. pfeiffer , and k. w. west , phys . rev
. lett . * 63 * , 1633 ( 1989 ) ; m. a. eriksson , a. pinczuk , b. s. dennis , s. h. simon , l. n. pfeiffer , and k. w. west , phys .
lett . * 82 * , 2163 ( 1999 ) .
m. kang , a. pinczuk , b. s. dennis , l. n. pfeiffer , and k. w. west , phys .
lett . * 86 * , 2637 ( 2001 ) .
j. yan , s. goler , t. d. rhone , m. han , r. he , p. kim , v. pellegrini , and a. pinczuk , phys .
rev . lett . * 105 * , 227401 ( 2010 ) .
sometimes the intra - landau level excitons are called spin waves or pseudospin waves , depending on the quantum numbers that distinguish the filled and the empty levels .
the inter - ll excitons that conserve all quantum numbers are called magnetoplasmons,@xcite and those that do not are dubbed spin - flip , valley - flip , or pseudospin - flip excitations.@xcite we do not use these terms ; the class of magnetoexcitons include all of these varieties .
s. l. sondhi , a. karlhede , s. a. kivelson , and e. h. rezayi , phys .
b * 47 * , 16419 ( 1993 ) ; h. a. fertig , l. brey , r. ct , a. h. macdonald , a. karlhede , and s. l. sondhi , phys . rev .
b * 55 * , 10671 ( 1997 ) . one can easily check that the equal sign linear combination of the excitons created by the operators @xmath320 and @xmath321 is a pseudospin singlet , whereas the opposite sign linear combination is a member of a pseudospin triplet : compare @xmath322 with @xmath323 , where @xmath324 is the total pseudospin lowering operator and @xmath325 is a ground state that is invariant for pseudospin rotation . at @xmath158
we have attempted an extrapolation of the excitation energies as a function of the cutoff @xmath195 .
the energies show a decreasing tendency , but fitting a power function@xcite has proved to be impossible .
we have chosen an ad - hoc cutoff at @xmath198 , with the understanding that small quantitative deviations are possible , especially at low energies .
we have checked that using @xmath326 does not fundamentally change the spectra . at certain values of the in - plane pseudospin anisotropy , which may arise due to electron - electron and electron - phonon interactions , two more phases are possible : a canted antiferromagnet and a partially layer polarized state , c.f .
these recently proposed states are beyond the scope of our study . | if bilayer graphene is placed in a strong perpendicular magnetic field , several quantum hall plateaus are observed at low enough temperatures . of these ,
the @xmath0 sequence ( @xmath1 ) is explained by standard landau quantization , while the other integer plateaus arise due to interactions .
the low - energy excitations in both cases are magnetoexcitons , whose dispersion relation depends on single- and many - body effects in a complicated manner . analyzing the magnetoexciton modes in bilayer graphene , we find that the mixing of different landau level transitions not only renormalizes them , but essentially changes their spectra and orbital character at finite wave length .
these predictions can be probed in inelastic light scattering experiments . |
recently data on @xmath0 annihilation in the range 40 - 400 mev / c ( for the laboratory @xmath4 momentum @xmath5 ) have been produced by the obelix collaboration@xcite .
we would like to compare these with other data on @xmath1 and @xmath6-nucleus annihilation .
in particular , we are interested in : @xmath4 annihilation on nuclei from @xmath7c to @xmath8pb in the range 180 - 280 mev / c@xcite ; @xmath1 annihilation from 30 to 180 mev / c@xcite ; @xmath3d , @xmath9he and @xmath10ne at small momenta , down to 45 mev / c@xcite ; @xmath3-nucleus annihilation on intermediate nuclei at larger momenta ( over 200 mev / c , see@xcite for a recollection of these data ) ; and on the ratio between @xmath1 and @xmath2 annihilations inside nuclear targets@xcite
. we will start by comparing @xmath1 and @xmath0 data , and then we will try to correlate these with the low energy nuclear data .
one of our aims is simply to compare @xmath0 and @xmath1 annihilation cross sections .
this comparison is a delicate operation , because it requires subtraction of coulomb effects , and comparison of data coming from different experiments with antiproton and antineutron beams . due to the difficulties in calculating a 100 % reliable flux normalization , it is preferable , as far as possible , to compare data from the same experiment .
a second goal of this work is to establish which is the degree of correlation between antineutron - nucleon and antinucleon - nucleus data . indeed ,
as better discussed later , @xmath1 data below 600 mev / c can be well fitted via energy independent optical potentials , while @xmath0 data show a nontrivial energy dependence that should be reflected in the energy dependence of nuclear annihilations .
another interesting point is the role of the different isospin channels in the low energy nuclear shadowing .
it has been demonstrated that a marked nuclear shadowing characterizes low energy @xmath6-nucleus annihilation cross sections@xcite .
for this phenomenon it is possible to imagine two classes of explainations .
first , general quantum mechanical processes , as described in a later section .
mechanisms of this kind do not discriminate too much between different annihilation channels . on the other side we may imagine a role for peculiar properties of single nuclear species .
indeed , the data are rather incomplete and impossible to organize in a systematic way . in particular a different @xmath11 composition , or a different @xmath11 distribution near the nuclear surface can be relevant , since most of the theoretical models ( for reviews see e.g. @xcite ) establish peculiar properties for the different isospin channels , properties which are normally derived either from g - parity ( or c - parity@xcite ) transformations of the better known nucleon - nucleon potentials , or from peculiar quark diagrams .
so , to summarize this point , it is relevant to establish how much of the low energy nuclear shadowing of annihilation processes can be due to general mechanisms , and how much is due to the nuclear composition in terms of protons and neutrons , and to the details of the nuclear structure .
we limit ourselves to a phenomenological analysis of the data , with standard methods of nuclear physics , without entering the debate about the underlying hadronic structure mechanisms governing the annihilation process . of course , it is not always possible to separate completely the two levels of the analysis , i.e. nuclear and hadronic .
wide reviews on the debate concerning the annihilation mechanisms can be found in the references listed above .
experimental ( scattering and annihilation ) data up to 1994 are recollected in @xcite , data on antiprotonic atoms up to 1988 in @xcite , and a very recent account of the many problems related with the antinucleon experimental techniques can be found in @xcite .
we still lack a well founded method that can be properly used at low energies to link the shadowing properties ( that automatically arise from an optical potential treatment@xcite ) with the details of the nuclear structure and internal motions ( which are taken into account in impulse approximation inspired treatments ) .
the main exception in this respect is deuteron , which has been considered by some authors@xcite and where it was possible to relate clearly the overall antinucleon - nucleus processes with the antinucleon - nucleon interactions . with more complex nuclei , optical potential analyses have been carried on to study antiprotonic atoms@xcite . at momenta over 200 mev
/ c the kmt method@xcite has been used in @xcite and the glauber method@xcite in @xcite ( and probably in other works ) . in the present work
our hope is that some interesting information can be separately extracted from a pwia and an optical potential analysis . for the sake of brevity we will indicate total @xmath4 and @xmath3 annihilation cross sections with tna and tpa respectively , specifying the target .
we will indicate as @xmath12 the ratio between @xmath2 and @xmath1 annihilations in a given nucleus with atomic number @xmath13 . concerning the choice of an optical potential for discussing nucleon - antinucleon annihilation
, it is well known that many quite different optical potentials fitted successfully pre - obelix nucleon - antinucleon data@xcite and they can probably fit the data under discussion too , since these data do not represent such a strong contraint . here and in previous papers@xcite we have relied on very simple optical potentials , with woods - saxon shape , the same parameters for each spin channel and no energy dependence in the momentum range 0 - 600 mev / c ( more indications are given in section 3 ) .
although such potentials can not reproduce the full phenomenology of nucleon - antinucleon interactions , they seem sufficient for fitting the available @xmath1 low energy data . with the @xmath0
annihilation data , as described in section 3 , a higher level of sophistication is perhaps necessary .
comparison of annihilation data with different projectiles and targets is nonsense without subtracting coulomb effects at momenta below 200 mev / c .
the traditional estimation@xcite has undercome modifications@xcite in the last years .
qualitatively the effect of the coulomb forces is to focus the projectile wavefunction in the annihilation region , which is a spherical shell of thickness 0.5 - 1 fm@xcite at the target proton / nucleus surface . in a previous work@xcite we gave analytical expressions for the correcting @xmath3-nucleus enhancement factors .
e.g. , for the @xmath1 annihilation cross section the enhancement factor can be reproduced within 5 % in the range 30 - 400 mev / c by the function @xmath14 . to calculate the enhancement factor we considered some completely different optical potentials .
all of them included the electrostatic potential of a spherical charge distribution and fitted the available low energy data .
then we removed the electrostatic potential and calculated again the annihilation rates .
the obtained coulomb enhancement factor was sufficiently independent of the choice of the strong part of the optical potential , in the considered momentum range .
an observation is necessary concerning apparent `` violations '' of the isospin invariance .
isospin symmetry suggests that : ( i ) @xmath0 cross sections are equal to @xmath2 ones ; ( ii ) @xmath1 and @xmath15 are equal apart from electromagnetic effects ; ( iii ) the four cross sections that one can imagine are actually combinations of two . in principle
this is undoubtful , but in practice it does not work , and an example can clarify the point . normally , @xmath0 data come from collisions between free antineutrons and protons , while @xmath2 cross sections are extracted from deuteron targets , with two consequences : \1 ) the antiproton is attracted by the deuteron charge .
the range of action of the coulomb forces is much larger than the deuteron radius , with the result that the @xmath2 annihilation rate is almost as much `` coulomb distorted '' as the @xmath1 one , so it should be expected to be larger ( much larger at very low energies ) than the @xmath0 one .
this problem is not present at momenta @xmath16 100 mev / c , where coulomb focusing effects can be neglected .
\2 ) as already noticed@xcite , below 60 mev / c shadowing effects in nuclei are so strong that the @xmath3d annihilation cross section is smaller than the corresponding @xmath1 one in the laboratory frame , or approximately equal in the center of mass frame ( due to the different transformations relating laboratory with center of mass variables for the cases of proton and deuteron targets ) . in both cases
@xmath17 is much smaller than @xmath18 .
the previous example should clarify that , although it is very likely that one would find @xmath19 @xmath20 @xmath21 in an experiment on @xmath22 neutrons , comparisons involving different isospin channels for @xmath5 @xmath23 100
mev / c should be performed with the greatest care as far as free neutron targets , or antiproton targets , are not available .
one should therefore be aware that an isospin decomposition can not be completely free from model dependence .
recent experimental data@xcite show that at antinucleon momenta ( in the laboratory ) below 60 mev / c the nuclear shadowing effect is very strong . below 60 mev
/ c antiproton annihilation rates on deuteron and @xmath24he are smaller than on hydrogen .
the annihilation rate on @xmath25ne is larger but not that much@xcite .
this and related phenomena has been discussed by us and other authors@xcite . rather independently of the mechanism underlying the annihilation process , it had been previously demonstrated that in the framework of the multiple scattering theory@xcite , of variational methods@xcite and of optical potential treatments@xcite one can predict such shadowing effects .
it has been reported@xcite that also in the coupled - channel approach one can obtain the same result .
the fact that different methods lead to similar results suggested us to investigate the problem from a more general and qualitative , although less precise , point of view . in our work@xcite
we have shown that due to the quantum uncertainty principle the @xmath4-nucleus cross sections should be almost @xmath13-independent , apart for fluctuations due to nuclear surface effects .
consequently the @xmath3-nucleus cross sections should depend on the target because of its electric charge only .
the underlying argument is that most of the existing models ( see the suggested references @xcite ) and analyses@xcite establish that the annihilation process takes place when the centers of mass of the antinucleon and of the target nucleus are at a relative distance @xmath26 such that @xmath27 @xmath28 @xmath26 @xmath28 @xmath29 , where @xmath30 @xmath31 1 fm ( or smaller , depending on the model ) does not depend too much on the target .
so the annihilation is equivalent to a measurement of the projectile - target relative distance with uncertainty @xmath30 @xmath28 1 fm , and this measurement is incompatible with a relative momentum @xmath23 200 mev / c . to see it another way , we distinguish between two classes of nuclear reactions . on one side , inelastic reactions where the entire nucleus is involved , as in compound nucleus reactions , but the underlying projectile - nucleon processes are elastic ( e.g. neutron induced nuclear reactions ) . in this case
the characteristic reaction region coincides approximately with the target nucleus .
then the uncertainty @xmath30 coincides approximately with the nuclear radius . on the other side , we find reactions where a strong inelasticity is present at the projectile - nucleon level . in this case
reactions deep inside the nuclear volume are rare , the reaction region is a shell at the surface of the target nucleus , with thickness @xmath30 , and @xmath30 is approximately the same for all the possible targets .
the consequence of the limitations imposed by the uncertainty principle is that for antinucleon momenta @xmath5 @xmath23 @xmath32 the total reaction cross section becomes much smaller than its possible unitarity limit .
this is also established by the well known@xcite low energy limit for the phase shifts : @xmath33 @xmath34 @xmath35 for @xmath5 @xmath36 0 .
the unitarity limit is reached when a partial wave is completely absorbed in the reaction process , which means @xmath37 @xmath38 0 , i.e. @xmath39 @xmath38 @xmath40 , so the unitarity limit can not be attained at small enough @xmath5 .
uncertainty considerations suggest that for @xmath5 @xmath16 @xmath32 it is possible , for strong enough reactions , to saturate the unitarity limit , while for @xmath5 @xmath23 @xmath32 we are in the situation where @xmath33 @xmath38 @xmath41 , whatever the strength of the reaction .
a paradoxical consequence is that a smaller @xmath30 corresponds to what would be a stronger reaction at large energies , so that at low energies `` stronger '' interactions can lead to a smaller reaction rate .
this fact can be verified in optical potential treatments . on the ground that the projectile wavefunction @xmath42 is completely damped within a range @xmath30
( i.e @xmath43 is large for @xmath44 @xmath45 @xmath29 and very small for @xmath44 @xmath28 @xmath27 ) it is straightforward to demonstrate that for the scattering length @xmath46 we have ( approximately ) : @xmath47 @xmath20 @xmath48 , @xmath49 @xmath20 @xmath50 .
indeed , the @xmath42 damping requirement implies for the logarithmic derivative @xmath51 @xmath20 @xmath32 .
this is an obvious geometrical fact , but in more physical terms it is a consequence of the uncertainty principle . together with the matching condition between the logarithmic derivatives of the free motion wavefunction and of the wavefunction in the annihilation region @xmath52
@xmath38 @xmath53 this leads to the previous @xmath46-values in the limit @xmath5 @xmath36 0 .
these values of course are deduced from approximate equations , so they represent just estimates , however they suggest that the antineutron annihilation cross sections should not show a @xmath54 increase with the target mass number @xmath13 .
such an increase could be present for antiproton annihilations , but because of coulomb effects only .
when going to any specific target nucleus , non - systematic effects could be present , especially related with the structure of the nuclear surface .
an example is given in ref.@xcite .
there , anomalous behaviors are related with the non - sphericity of the nucleus .
another exception should be represented by neutron - halo nuclei , because in this case the annihilation range could be much larger than 1 fm .
also the composition of the nuclear surface in terms of protons or neutrons could be important , since all models attribute a strong isospin dependence to the antinucleon - nucleon interaction .
the exposed mechanism has an interesting consequence in the case of optical potential analyses : an increase of the strength of the imaginary part of the optical potential can lead to a decrease of the consequent reaction rate at small momenta@xcite . in the above language , an increase in the potential strength leads to a decrease in the size parameter @xmath30 , since the absorption of the projectile wavefunction takes place in a shorter range . also modification of other parameters ( radius , diffuseness , etc ) leads to consequences that are not necessarily the most obvious ones .
an example is given in the next section .
in fig . 1 we show the tna and tpa on hydrogen , together with two fits by energy - independent optical potentials . for the @xmath1 case the total interaction includes the electrostatic potential of a spherical charge distribution with radius 1.25 fm .
this charge radius is @xmath56 times the charge radius of the proton , to take into account both the proton and the antiproton extended charges .
details on the potential are given below .
a third optical potential curve shows what would be the @xmath1 cross section in absence of electrostatic interactions , within the same optical model .
what we notice first is that tna and tpa are reasonably similar for @xmath5 @xmath45 200 mev / c , but below 200 mev / c the tna falls clearly below the tpa , and this fall can not be justified in terms of coulomb interactions only , although these contribute .
we could speak of `` low energy fall '' with respect to a `` background '' which is similar to the @xmath1 case .
alternatively one can think that the tna sums two contributions : a monotonous background that is proportional to the @xmath1 annihilation rate ( coulomb effects apart ) but lower , and a broad peak ( resonance ? ) at @xmath5 @xmath31 200@xmath57300 mev / c .
this broad peak would be hidden by the isospin-0 background and by the low energy coulomb rise in the tpa case .
some of the available data at momenta @xmath45 400 mev / c ( see @xcite for a more systematic review of these data ) could support the `` peak '' idea , since at these momenta the @xmath0 annihilation rate seems to be lower than the @xmath1 one , but it is not unequivocally clear how much lower .
indeed , one @xmath0 point at 700 mev / c by @xcite is clearly below the @xmath1 annihilation , while data reported in @xcite are not very different from the corresponding @xmath1 data@xcite in the region 300 - 500 mev / c . on the other side , some data at very low energy@xcite support the `` gap '' idea , since they show a rise of the @xmath0 annihilation rate @xmath58 below 50 mev / c .
these data have been reported in the figure , and the one at 20 mev / c lies on the optical potential extrapolation of the @xmath1 data , once the electrostatic part of the potential has been removed .
the authors of ref.@xcite also report some potential predictions@xcite that support the existence of a minimum for the @xmath0 annihilation rate @xmath59 at 50 - 100 mev / c , and a set of scattering length predictions@xcite that show no great differences between @xmath1 and @xmath0 expected scattering lengths .
they also report a collection of pre - obelix @xmath4 data .
we must remark that the normalization of antineutron fluxes is a delicate matter , with the consequence that both a comparison of antineutron data coming from different experiments , and a comparison of antineutron and antiproton data , must be taken with the greatest caution .
so we will generically speak of a `` gap / peak structure '' of the @xmath0 annihilation rate compared to the @xmath1 one .
we will show in the next section that a comparison with nuclear data supports the existence of this gap / peak structure .
whatever the interpretation , there are some important remarks : \1 ) the strong oscillation of the tna with respect to the tpa ( see also fig.4 ) is an evidence of the fact that different physical mechanisms are dominating in the two cases in the considered momentum range .
if a resonance is present this is obvious .
if there is no resonance , according to the partial wave analyses presented in refs.@xcite the tna p - wave has its maximum at @xmath5 @xmath20 250
mev / c where it attains the value 120 - 140 mb . our optical potential fit ( which seems good for the tpa ) fixes the tpa p - wave maximum to about 80 mb at 150 mev / c .
semiclassical intuition fixes the corresponding impact parameters @xmath60 @xmath38 @xmath61 to 1.25 fm ( tpa ) and 0.8 fm ( tna ) .
then , purely geometrical considerations suggest a ratio , between the size of the two p - wave outcomes at their peak momentum , tna / tpa @xmath31 @xmath62 @xmath20 0.4 instead of the found value 1.6 . the most obvious conclusion would be that the p - wave interaction is much more effective in the @xmath0 case , where it acts at smaller impact parameters ( by a factor 0.6 ) .
this short range interaction does not affect the s - wave contribution , which is clearly larger in the @xmath1 case ( even after subtracting coulomb effects ) at momenta over 30 mev / c . at lower momenta ( that means longer range ) the two s - wave contributions could be equal , according to@xcite and references therein .
so the geometry and strength of the isospin-1 and isospin-0 channel interactions must be different , as far as these semiclassical considerations can be trusted .
\2 ) absurd as it may sound , the fact that at small momenta tna are sensibly smaller than tpa could mean that the interactions are stronger in the @xmath0 case than in the @xmath1 one .
this can be immediately seen in the choice of the optical model parameters .
we used a woods - saxon form , with all parameters , but one , equal for tpa and tna .
tpa : imaginary strength 8000 mev , real strength 46 mev ( attractive ) , imaginary radius 0.52 fm , real radius 1.89 fm , real and imaginary diffuseness 0.15 fm .
these values fall in the ranges used by previous authors to fit elastic data@xcite , and its imaginary part is pretty similar to other previously used optical potentials ( e.g. the one of ref.@xcite ) .
many other sets of parameters , and also different potential shapes , can lead to similar results for the available low - energy data . starting from this peculiar set , to obtain the tna curve it is sufficient to increase the imaginary radius to 0.75 fm .
alternatively , one can obtain the tna curve by leaving the imaginary radius at 0.51 fm and increasing the imaginary strength to about 12000 - 16000 mev .
so , to get a smaller cross section we need a `` stronger '' potential .
this `` inversion '' behavior is another manifestation of the general shadowing mechanism discussed in section 3 . in an optical potential model at small @xmath5 an increase ,
e.g. , of the imaginary strength @xmath63 produces an increase of the reaction cross section for small values of @xmath63 only . after a certain threshold , further increases of @xmath63
lead to a decrease of the reaction cross section@xcite . at a qualitative level
this can be explained by the uncertainty principle : an increase in the strength of the imaginary part of the potential decreases the thickness of that spherical shell , surrounding the proton / nucleus target , where annihilations are supposed to take place .
this introduces larger gradients in the projectile wavefunction .
these larger gradients imply a larger logarithmic derivative , and therefore a smaller ( imaginary part of the ) scattering length .
\3 ) in the case of @xmath1 annihilations below 600 mev / c many quite different choices of potential parameters can fit them@xcite ( including a pure imaginary potential ) without introducing a dependence of the parameters on energy .
all these fits share some common features .
they reproduce the data very well below 150 mev / c and above 300 mev / c .
the fits remain satisfactory up to 600 mev / c . in the region 150 - 300 mev / c the fits stay slightly below the data .
alternatively , it is possible to choose the parameters so as to reproduce the data from 150 to 600 mev / c , while at lower momenta the fits are above the data .
so , we could say that an energy independent optical potential can produce a `` background '' around which a slight oscillation is present ( positive for @xmath5 in the range 150 - 300 mev / c , or negative for @xmath5 @xmath28 150
mev / c ) . in the tna case
the situation is qualitatively the same , but the oscillation is much stronger and more evident .
so it seems that an energy - independent optical potential fit can be well related with the isospin-0 channel , which seemingly consists of the `` background '' only .
a background with the same shape is also present in the isospin-1 channel , but this channel also contains an oscillation that is not reproduced by our set of optical potentials .
as already remarked , we have no means to establish the nature of this oscillation .
\4 ) we have also tried to change our optical potential shape so as to produce resonances that fit the gap / peak structure which is seen in the @xmath0 annihilation .
we can not exclude that more attempts can lead to a good reproduction of the data in fig.1 ( in the described `` inversion plus resonance '' regime there is no predictable relation between the changes of the potential and the shape of the output , which makes such attempts rather frustrating ) .
we have not been able to reproduce satisfactorily these data , and perhaps more sophisticated potentials are necessary .
it was however easy to get shapes that @xmath64 looked similar .
we tried with two kinds of potential : ( i ) woods - saxon ( with attractive elastic part ) ( ii ) woods - saxon like the previous one , plus a repulsive elastic surface barreer of gaussian shape ( of the kind @xmath65 $ ] ) .
in both cases to get a broad peak where it must be it was necessary to enlarge the range of the elastic part to 2.5 - 3 fm .
this is probably due to the fact that , despite the smallness of the radius of the imaginary part , most of the annihilations take place at @xmath44 @xmath20 1 fm .
so only a much wider attractive well can produce metastable states , within such models . perhaps in a more sophisticated potential model , where different spin channels are clearly separated , it is possible to get a resonance in a spin channel where annihilation is less effective , so without the need for such a long range for the potential .
we think that these interesting data really demand for a higher level potential analysis .
because of the shadowing , a complete analysis of annihilations in nuclei below 60 mev / c is pretty complicated .
a simple comparison of different sets of data reveals interesting features , however , and we can also hope to understand something by a pwia analysis , especially at momenta over 100 mev / c . in particular , the energy dependences of the @xmath0 annihilation cross section , of the @xmath4-nucleus one , and of the ratio of @xmath2 to the @xmath1 annihilation in nuclei , show evident correlations .
of course nuclear shadowing effects are completely absent in pwia , and this must be taken into account , taking with great care the calculations below 100 mev / c . at larger momenta the main missed effect is the eclipse effect@xcite which implies an almost energy independent surface absorption factor @xmath20 0.5 .
we take this factor into account by renormalizing our pwia predictions with one datum at a specified momentum value over 200 mev / c . at small momenta however ( below @xmath31 200
mev / c ) shadowing is something more complicated than a mere eclipse effect .
the eclipse effect could never bring @xmath3 annihilation rates to decrease with increasing @xmath13 .
so we may only @xmath66 as `` shadowing '' the departures from pwia predictions , and remark that they are consistent . as we show later , anyway ,
a certain part of the smallness of the low energy @xmath3-deuteron annihilation rate with respect to the @xmath1 one is not a shadowing effect , but just due to the smallness of the @xmath2 annihilation rate below 150 mev / c . in fig.2 we show the @xmath4-nucleus data , relative to momenta between 180 and 280 mev / c , for several nuclear species from @xmath7c to @xmath8pb ( see @xcite for more details ) .
these data are consistent with an @xmath67 law , expressing dominant surface absorption , @xmath68 @xmath69 .
however the @xmath67 law found in these data can not be generalized to @xmath0 or @xmath1 ( coulomb subtracted ) data in the laboratory frame , while this generalization is only approximately possible in the center of mass frame . at 180 mev / c ,
the @xmath7c or the @xmath8 pb tna , divided by @xmath67 , give about 150 mb .
the corresponding @xmath0 and ( coulomb subtracted ) @xmath1 cross sections are @xmath20 200 mb at the same laboratory momentum . at the same center of mass momentum , which corresponds to 360 mev / c in the laboratory ,
the @xmath0 total annihilation cross section is about 130 mb , and the @xmath1 one is slightly smaller than the @xmath0 one .
so , in both cases we find deviations from the @xmath67 law at @xmath70 @xmath20 100 mev / c , but these deviations are of opposite sign if the data are compared at the same laboratory or center of mass momentum .
it is difficult to establish whether it is more proper to compare data relative to different nuclear species at the same laboratory or center of mass momentum .
impulse approximation based models suggest the same laboratory momentum , compound nucleus inspired models prefer the opposite choice .
indeed , in the former case direct momentum exchange is between the projectile and one of the target nucleons only , whose average initial momentum is zero in the laboratory . in the latter , the projectile momentum is given directly to the entire target nucleus .
an analogous , systematic comparison of all the available @xmath3-nucleus data has been presented elsewhere@xcite .
very recently a new series of @xmath4-nucleus data down to 70 mev / c has been presented@xcite , which show a respected @xmath67 law between different nuclear species .
since these data have been presented as `` preliminary '' , we ca nt discuss them in detail .
considering the energy behavior of the tna on nuclei shown in fig.2 , the most interesting feature is the systematic presence of a shoulder .
however this phenomenon seems less evident in the just quoted preliminary data .
tpa data in this region are not enough to to give us shape information . however , the optical potential calculations that we have reported in @xcite suggest that the coulomb enhancement at low energies can easily hide such details in the energy dependence of the annihilation data .
different relevant mechanisms must be taken into account in the momentum range 100 - 300 mev / c : \1 ) the tail of the low energy shadowing phenomenon .
presently we have no knowledge of tpa and tna on heavy nuclei at @xmath5 @xmath23 100 mev / c , so we do not know whether the discussed shadowing is a systematic phenomenon ( as we suggest in the previous section 2 ) or is simply peculiar of some light nucleus . and in light nuclei we do not have enough tpa data in the region 100 - 300 mev / c to understand completely the transition between large and low energy regimes .
there are several sets of data , but regretfully we do not have a satisfactorily continuous set of data relative to the same projectile on the same target ( with the exception of the @xmath1 and @xmath0 cases ) .
\3 ) the nuclear fermi motion .
the fermi momentum , which represents a rough cutoff of the nuclear momentum distribution , is @xmath20 200 mev / c .
this means that for @xmath5 @xmath28 200 mev / c the statistical distribution of the relative @xmath4-nucleon motion contains the zero - energy point , and that it is easy for a target nucleon to be faster than the projectile .
so the projectile momentum 200 mev / c represents a kind of special borderline in the physics of annihilations on nuclei .
\4 ) the @xmath0 annihilation cross section presents the above discussed gap / peak structure below 300 mev / c .
the nuclear fermi motion will soften this structure , but perhaps not enough to make it completely disappear .
a pwia calculation can give some qualitative understanding about the roles of the gap / peak structure and of the fermi motion in the nuclear tna , and also about the relative role of neutrons and protons in the target .
in fig.3 we present a pwia calculation of the tna on @xmath71ca . in fig.4
we show a pwia calculation of the @xmath12 ratio on deuteron and @xmath24he .
we recall that we have defined @xmath12 as the ratio between @xmath3 annihilations on neutrons and protons inside a nucleus with mass number @xmath13 . in figs . 5 and 6 we show pwia calculations of @xmath3 annihilations on deuteron and @xmath24he .
as a starting point we have used an equation contained in @xmath72 .
although this equation has been justified there by a long series of mathematical passages , its meaning is quite simple and can stand without demonstration : at pwia level , any event rate on a nuclear target , leading to a well defined final state , is an incoherent sum of the rates of all the possible @xmath15 and @xmath0 events leading to the same final state , weighted by the nuclear momentum distribution of the target nucleon . by `` rate '' we mean a cross section divided by the corresponding flux of colliding particles . under the simplifying assumption that different final states sum incoherently we simply have : @xmath73 where @xmath74 is the relative velocity between two colliding particles with momenta @xmath75 and @xmath76 .
this nuclear pwia average of the antinucleon - nucleon processes clearly misses any eclipse or shadowing effect ( which , e.g. , produces a tna proportional to @xmath67 at large energies ) .
so , the rise of the pwia curve in fig.3 below 150 mev / c is unreliable ( it is just proportional to the average rise of antiproton and antineutron annihilation cross sections on a proton ) . at momenta larger than 150 mev
/ c the pwia should be more reliable , apart for a slowly energy dependent eclipse factor , and could clarify the effects of the fermi motion and the different role of protons and neutrons in the target . for the @xmath71ca shells we have used harmonic oscillator eigenfunctions .
some attempts with different kinds of states have shown no qualitative differences , unless the nuclear size is changed to unphysical values .
in addition , the most external wavefunctions do not show great differences ( in the energy dependence of the outcome ) with respect to the s - wave ground state .
this is important , because in a more proper dwia ( distorted wave impulse approximation ) treatment the internal shells would be scarcely involved in the annihilation process . for deuteron and @xmath24he ( in fig.4 ) experimental single particle momentum distributions@xcite have been used . again , within this simple model we did not find a great sensitivity of the results on the peculiarities of the momentum distribution , taking into account that many details that could be present in the numerator and in the denominator are lost in the ratios of fig.4 . with deuteron , we surely missed the angular dependence of the large momentum ( d - wave dominated ) part of the distribution . to calculate the pwia curves of figure 3 we need @xmath0 and @xmath15 annihilation rates .
for fig.4 we need @xmath1 and @xmath2 annihilation rates , both subject to the effect of the deuteron or @xmath24he charge .
the @xmath15 annihilation rate has been assumed to be equal to the coulomb subtracted @xmath1 one ( i.e. the continuous curve of fig.1 ) . in fig.4
we use the same cross sections , so we neglect the effects of the nuclear electrostatic attraction . on the basis of the discussion in section 2
we assume that they do not affect the @xmath12 ratio , although they surely enhance both the @xmath77 and the @xmath78 rates .
actually a look at the data in fig.4 shows that ( with large error bars ) @xmath79 @xmath80 0.48/0.78 @xmath80 0.61 at zero energy , and 0.54/0.82 @xmath80 0.66 at about 600 mev / c . in the case of a different action of the nuclear electrostatic potential on the @xmath1 and @xmath2
cross sections , this effect would be practically absent at 600 mev / c and would depress the @xmath81 ratio with respect to the @xmath82 one at momenta @xmath23 100 mev / c .
the large error bars and data fluctuations do not allow for easy conclusions , however there are no special reasons to assume a different coulomb effect on @xmath1 and @xmath2 reactions , if both the proton and the neutron are bound in the same nucleus .
the obelix @xmath0 data cover the range 60 - 400 mev / c , and the @xmath1 ones start at 30 mev / c .
we therefore had to assume values for these cross sections for momenta outside these ranges . with @xmath1 data at momenta over 200
mev / c we have relied on the data in ref.@xcite , and at momenta below 30 mev / c on our optical potential fit ( after subtraction of coulomb forces as in fig.1 ) . for the @xmath0 case , in the region @xmath5 @xmath45 400 mev / c
the @xmath0 data have been assumed to be equal to the @xmath1 ones , as suggested by the measurements in ref.@xcite and @xcite for momenta over 300 mev / c ( see e.g. @xcite , fig.9 , for a comparison of the @xmath0 and @xmath1 data in that momentum range ) . a look at fig.4 does not help too much .
indeed , we may say that the nuclear fermi motion approximately averages the `` free '' annihilation rates within a 200 mev / c range .
then the point of view of equal `` free '' @xmath1 and @xmath2 annihilation rates at momenta in the range 300 - 600 mev / c can be supported by the absence of a clear decreasing trend in the set of deuteron points over 300 mev / c . on the contrary ,
the helium point at 600 mev / c supports the possibility of a decrease of the free @xmath83 ratio , as suggested by the measurement in ref.@xcite .
anyway , the scatter of data in the deuteron case , and the lack of data in the @xmath24he case prevent us from precise conclusions . in the limit of zero energy
, we have adopted the low energy parametrization of the @xmath0 cross section contained in references@xcite .
it implies a decrease of the @xmath12 ratios at very low energies , which is more suitable for fitting the zero - energy @xmath24he point . adopting the low energy parameters suggested by refs.@xcite would on the contrary rise the zero energy @xmath12 value , which is more coherent with the two zero - energy deuteron points .
the upper curve in fig.3 can be taken as a qualitative confirmation of the effect of the @xmath0 gap / peak structure in producing a very soft shoulder in the @xmath84nucleus annihilations at 150 - 300 mev / c in fig.3 we also show the separate contributions of the @xmath0 and @xmath15 reactions .
evidently the @xmath15 reaction alone would not interrupt the trend of a positive second derivative that characterizes expectance and data at both lower and larger momenta . within pwia
the shoulder is a direct consequence of the gap / peak structure of the @xmath0 cross section .
the fermi motion , as predictable , softens this gap / peak structure , but not completely .
we must also observe that the produced shoulder is much less pronounced than most of the ones seen in fig.2 .
our attempts with different shapes for the nuclear shells show that the shoulder evidence is enhanced by a larger radius of each nuclear shell ( i.e. a narrower momentum distribution ) , however one must arrive at slightly unrealistic nuclear dimensions in order to obtain more evident shoulders .
the strength of the rise of the pwia cross sections below 100 mev / c is probably unphysical , since we know that shadowing could be very active in that region ( although the quoted recent preliminary data @xcite confine shadowing to @xmath5 @xmath28 70 mev / c and with good approximation reproduce the predictions of fig.3 ) . in ref.@xcite some optical potential shapes possibly corresponding to low energy @xmath4-nucleus annihilations are reported , and they are almost energy - independent in the region 100 - 300 mev / c . these optical potentials were based on coulomb - subtracted @xmath1 data and on nuclear cross sections at momenta over 200 mev / c and so have no relation with that gap / peak structure that is peculiar of the @xmath0 interaction .
the data reported in @xcite confirm the ia behavior reported in fig.3 down to 70 mev / c , rather than the these optical potential predictions .
the systematic presence of the gap / peak structure in the tna / tpa ratio is much more evident in fig.4 , where we report the @xmath12 ratio for @xmath13 @xmath38 1 ( @xmath0 data coming from the obelix experiment@xcite divided by the coulomb subtracted @xmath1 annihilation cross section represented by the continuous line in fig.1 ) , for @xmath13 @xmath38 2 ( deuteron ) and @xmath13 @xmath38 4 ( @xmath24he ) .
these deuteron and @xmath24he data come from several experiments and are reviewed in ref.@xcite .
looking at fig.4 one immediately notices that all deuteron points lie above any of the @xmath24he points , and the proton target data are , in the average , clearly over the nuclear target data ( we remark that , although the proton target ratios have been calculated using the `` coulomb subtracted '' @xmath1 annihilation cross section , the effect of the coulomb subtraction is negligible above 100 mev / c ) .
the second observation concerning the data reported in fig.4 is that a certain degree of correlation between the shapes of the three groups of data is present .
since @xmath85 @xmath38 1 both in deuteron and @xmath24he , and the average value of the free @xmath86 ratio is about 1 , clearly pwia can not reproduce the average value of these nuclear @xmath12 distributions . for producing the fits contained in fig.3 we have multiplied the @xmath0 total annihilation cross section by a constant factor @xmath46 , which is 0.8 for deuteron target , and 0.6 for @xmath24he target .
at the largest energies of the considered range the pwia ratio should be reliable .
the known eclipse effect should more or less cancel between numerator and denominator .
anomalous shadowing of the kind discussed in section 3 is not strong over 100 mev / c .
so the reason for @xmath87 @xmath28 @xmath88 @xmath28 @xmath89 @xmath90 1 is unexplained .
a possibility is final state rescattering .
this should not be very effective in total annihilation cross sections , but in the @xmath12 case it can modify the number of charged particles that is used to decide whether the annihilation was on a proton or on a neutron .
it is not unlikely that this @xmath13-dependent renormalization effect has this origin , however one should analyze carefully how each experimental point has been measured .
in the case of @xmath24he the pwia fit reproduces satisfactorily the behavior of the data .
we may say that the gap / peak structure of the free @xmath91 ratio is reproduced by the @xmath81 ratio . in the deuteron case at zero - energy
the fit is not satisfactory , while over 200 mev / c some correlation seems to be present .
somehow the presence of the pwia peak at 280 mev / c ( which is a consequence of the gap / peak structure of the free ratio ) is reflected in the presence of a maximum in the data at 300 mev / c , but the large error bars and the spread of the data do not allow for more quantitative conclusions . in the last two figures we show some pwia fits of the @xmath3 annihilation cross sections on deuteron and @xmath24he . to obtain these curves the same above method has been used , but the final result has been multiplied , to take the nuclear charge into account in agreement with ref.@xcite , by the factor @xmath92 , where @xmath93 is the nuclear charge , @xmath94 is the @xmath3-nucleus relative velocity and the constant @xmath95 is characteristic of the target nucleus .
the exponent ( @xmath571.4 ) has no special physical meaning , since the above @xmath92 factor is only a fitting relation that reproduces within some percent the coulomb enhancement factor in the range 40 - 400 mev / c .
the constant @xmath95 is 0.0060 for deuteron , and 0.0040 for @xmath96 . to take surface absorption into account
, the two pwia fits have been renormalized by a constant factor , the best to reproduce the data ( or some data ) over 200 mev / c .
we remark that : ( i ) this coulomb factor is not relevant from 100 mev / c onward , ( ii ) it is not very different between light nuclei at the same @xmath3 momentum in the laboratory frame , ( iii ) in hydrogen , deuteron and @xmath24he it is near 1.5 at 50 mev / c and 2 at 25 mev / c ( laboratory frame ) . in the two figures we also report what would be the pwia distribution if the target nucleus were composed by neutrons only or by protons only ( without touching the deuteron and @xmath24he charges , which are taken into account by the @xmath92 factor in any case ) .
the first evident observation is that the effect of the @xmath0 gap / peak structure is much more evident in the deuteron case .
this is due to the compactness of the @xmath24he structure , and to the exactly opposite peculiarity of the deuteron structure .
a more compact space structure implies a broader momentum distribution .
the fermi motion tends to average out the details of the momentum dependence of the nucleon - antinucleon annihilation rates .
this mechanism is more effective with broader nuclear momentum distributions .
another observation is that , if we define shadowing as the departure between the observed data and impulse approximation predictions , this departure becomes evident in both cases below 100 mev / c , but more evident in the @xmath24he case . in the deuteron case
the shadowing phenomenon is less dramatic , but would be overestimated by not taking into account that one of the two nucleons composing this nucleus is a neutron . concluding this section
, we may say that some evident correlation exists between nuclear data and the behavior of the @xmath0 annihilation rate .
in particular , it would be difficult to justify the shape of the @xmath81 and @xmath82 ratios in absence of the gap / peak structure of the free @xmath91 ratio .
there is some chance that the shoulder found in the @xmath4-nucleus annihilation cross sections is related with the same gap / peak structure , however in this case the situation is much less clear .
still to be clarified is the fact that the energy - averaged value of the @xmath12 ratio decreases at increasing @xmath13 .
the last two figures also show that the energy dependence of the nuclear annihilation rates can be partly explained in terms of the energy dependence of the @xmath3 annihilation cross section on a free neutron .
a consistent part of this energy dependence , however , can not be explained in these terms and we may say that a strong energy - dependent shadowing is present .
we have presented an analysis comparing low energy @xmath0 annihilation data with other available data on antinucleon - nucleon and antinucleon - nucleus annihilation .
the different sets are all consistent with a gap / peak structure in the isospin-1 channel , over a regular `` background '' which can be reproduced by energy - independent optical potentials of simple form .
this structure is not too pronounced but evident , and we have also shown that it affects the nuclear annihilation rates , whose behavior can be @xmath97 explained in terms of the behavior of the antiproton - neutron annihilation rate .
we have also discussed the shadowing phenomena related with antinucleon - nucleus annihilation , and reported an interpretation in terms of the quantum uncertainty principle .
the comparison of impulse approximation predictions with the nuclear data suggests that a strong energy - dependent shadowing is present at low energies : also after including absorption corrections , the data can not be fully explained in terms of individual antinucleon - nucleon annihilations .
a.feliciello for the obelix collaboration , `` workshop on hadron spectroscopy 99 '' , march 8 - 12 1999 , lnf frascati ( italy ) , 429 ; m.agnello @xmath98 @xmath99 , nucl .
methods phys .
a399 ( 1997 ) ii ; nucl.phys .
* b * ( proc . suppl . )
56 a ( 1997 ) 227 .
a review on the compled channel model is given by i.s.shapiro and a review on quark models by a.m.green in `` antiproton - nucleon and antiproton - nucleus interactions '' , eds .
f.bradamante , j .- m.richard and r.klapish , ettore majorana international science series , plenum press 1990 , pg.81 and pg.109 .
s.frullani and j.mougey , `` single particle properties of nuclei through ( e , ep ) reactions '' , vol .
14 of advances in nuclear physics , plenum press , new york - london ; s. boffi , c. giusti and f. d. pacati , phys
. rep . * 226 * , 1 ( 1993 ) ; s.boffi , c.giusti , f.d.pacati and m.radici , `` electromagnetic response of atomic nuclei '' , vol .
20 of `` oxford studies in nuclear physics '' ( oxford university press , oxford , 1996 ) . | we compare data of antineutron and antiproton annihilation cross sections on different targets at very low energies . after subtracting coulomb effects , we observe that the ratio between the @xmath0 and @xmath1 annihilation cross sections is an oscillating function of the energy at momenta smaller 300 mev / c .
this nontrivial behavior is confirmed by the analysis of the relative number of @xmath2 and @xmath1 annihilations in nuclei .
we show that a part of the strong shadowing phenomena in @xmath3-nucleus annihilations can be explained in terms of this oscillation , while a part requires different explainations . , , , , , , , |
superconductivity in iron pnictide was discovered at the boundary of magnetically ordered and non - ordered state of fe - magnetism .
the spin - density - wave type antiferromagnetic order in rfeaso ( r = la , ce , pr , nd , sm ) and afe@xmath9as@xmath9 ( a = ca , ba , sr , eu ) could be suppressed by carrier doping@xcite or by application of chemical / externally applied pressure@xcite .
one may expect to find similar interesting results by tuning the magnetic state of cr - moments as well .
recently cr - magnetic ordering at relatively low temperature has been reported in rcrge@xmath0 system where r stands for rare - earth elements @xcite . on the other hand , ce - based ternary germanide , rtge@xmath0 ( r = rare - earth elements and t = @xmath10 elements ) system exhibits some interesting magnetic properties .
for instance , cerhge@xmath0 shows a weak kondo effect along with three magnetic phase transitions @xcite .
ce moments order antiferromagnetically below @xmath11 = 5.5 k in cenige@xmath0 [ 9 ] which exhibits pressure induced superconductivity @xcite below 0.48 k in a wide pressure range from 4 to 10 gpa .
furthermore , a complex magnetic phase diagram is found in cecoge@xmath0 [ 11 ] which also exhibit superconductivity under the application of external pressure @xcite .
the study of the anomalous physical properties close to the borderline between magnetically ordered and nonmagnetic ground states is currently of intense interest in the contemporary condensed matter physics research .
this point of instability achieved by non - thermal tuning parameters between two stable phases of matter is called quantum critical point ( qcp ) .
interestingly , a quantum critical behavior is evidenced in doped cecoge@xmath0 [ 13,14 ] .
the situation may , however , get more complicated in cecrge@xmath0 due to the presence of cr - magnetic ordering . very recently
the suppression of cr ferromagnetic ordering has been reported in lav@xmath12cr@xmath13ge@xmath0 system @xcite with a possibility for the presence of a qcp in this system .
such possibility also exists in the corresponding ce - system and this motivates us to study cecrge@xmath0 in detail .
the compounds rcrge@xmath0 are reported to crystallize in the hexagonal perovskite structure ( space group _ p6@xmath0/mmc _ ) @xcite . a preliminary investigation based on magnetization measurements of cecrge@xmath0 reports ferromagnetic ordering of cr - moments below 66
k ; much less is known about the ce - magnetism and its interaction with cr - moments in this compound .
we herein report comprehensive study of low temperature properties of cecrge@xmath0 by means of x - ray absorption spectroscopy ( xas ) , magnetic susceptibility @xmath1 , isothermal magnetization m(h ) , electrical resistivity @xmath2 , specific heat c(@xmath3 ) and thermoelectric power s(@xmath3 ) measurements . in order to explore the role of @xmath14-electron we have also measured the properties of the corresponding @xmath15-electron analog lacrge@xmath0 .
our results corroborate that cr ions order ferromagnetically below 70 k in cecrge@xmath0 and ce - ions possess a stable 3 + valance state in this compound .
our observation of a large value of sommerfeld coefficient @xmath6 , obtained from the specific heat measurement , even in the magnetically ordered state is quite remarkable in this context .
the kondo lattice type of the resistivity and the large @xmath6 tempt us to believe that cecrge@xmath0 is a moderate heavy fermion system .
polycrystalline samples of lacrge@xmath0 and cecrge@xmath0 were prepared by arc melting the high purity elements , la and ce pieces ( 99.9% , alfa - aesar ) , cr and ge chips ( 99.99% sigma aldrich ) , taken in stoichiometric ratio on water cooled copper hearth under an argon atmosphere .
the samples were flipped after each melting and were melted several times to ensure homogeneity . in order to homogenize and remove the impurity phases
the melted buttons were wrapped in ta foil and annealed in an evacuated quartz tube at 900@xmath16c for one week .
the phase purities of the annealed samples were checked by powder x - ray diffraction ( xrd ) using cu-@xmath17 radiation and metallographic examinations .
the absorption spectra of cecrge@xmath0 at the ce-@xmath18 edge ( e = 5723 ev ) have been recorded in transmission mode at the exafs beam line a1 of the hamburg synchrotron radiation facility ( hasy - lab at desy ) using the si @xmath19 double crystal monochromator . in order to perform the xas study
finely grounded / powdered sample of @xmath8 12 mg with particle size smaller than @xmath20 m was mixed with b@xmath21c and pe powder and pelletized to a 10 mm disc .
the absorption behavior has been determined in the energy range from 5500 ev up to the cr k absorption edge ( 5989 ev ) with a minimal step size of @xmath22e = 0.25 ev near the ce-@xmath18 edge ( 5723 ev ) .
the pellet was mounted in an oxford gas flow cryostat to realize temperatures down to the boiling point of liquid helium .
the spectra showed an absorption edge step of @xmath23 which represents the difference of the low and high energy background functions at the absorption edge .
the ( ce@xmath24)-reference cepo@xmath25 was measured simultaneously at ambient temperature and was used for energy calibration . evaluation of spectra , energy calibration and normalization have been performed with the athena module of the horae software package @xcite .
magnetic measurements were carried out using a commercial superconducting quantum interference device ( squid ) magnetometer ( mpms quantum design ) .
electrical resistivity measurements were carried out in the temperature range 2 - 300 k using ac transport option of physical property measurement system ( ppms , quantum design ) .
the specific heat was measured by relaxation method in ppms . the thermoelectric power ( tep )
was measured using thermal transport option of ppms .
the crushed polycrystalline samples were characterized by x - ray diffraction with cu-@xmath17 radiation to determine the single phase nature and the crystal structure .
powder x - ray data were analyzed by rietveld refinement using fullprof software@xcite .
the refinement reveals the single phase nature of the samples .
the lattice parameters for cecrge@xmath0 and lacrge@xmath0 , given in table i , are in good agreement with those reported in the literature@xcite .
the room temperature xrd pattern and rietveld refinement profile are shown in fig 1 .
compositional homogeneity was checked by performing metallographic examinations like scanning electron microscopy ( sem ) and energy dispersive x - ray spectroscopy .
( color online ) the powder x - ray diffraction pattern of cecrge@xmath0 recorded at room temperature .
the solid line through the experimental points is the rietveld refinement profile calculated for the banio@xmath0 -type hexagonal perovskite structure ( space group _ p6@xmath0/mmc _ ) .
the short vertical bars mark the bragg peak positions .
the lowermost curve represents the difference between the experimental and calculated intensities.,width=321 ] c c c c structure & + space group & _ p6@xmath0/mmc _ + + lattice parameters & lacrge@xmath0 & cecrge@xmath0 + @xmath26 ( ) & 6.198 & 6.150 + @xmath27 ( ) & 5.765 & 5.719 + @xmath28 ( @xmath29 ) & 191.79 & 187.32 + ( color online ) ce-@xmath18 edge x - ray absorption of cecrge@xmath0 at 280 k , 70 k and 5 k in comparison to the ( ce3 + ) reference spectrum of cepo@xmath25 .
( inset : un - normalised spectra).,width=321 ] to elucidate the valence state of ce in cecrge@xmath0 we performed ce-@xmath18 edge x - ray absorption study .
the spectra of cecrge@xmath0 and the ( ce@xmath24)- reference compound cepo@xmath25 show one single white line at the ce-@xmath18 edge with a typical broadening in the spectra of an intermetallic compound . neither the spectra of cecrge@xmath0 nor its derivative indicate an additional ( ce@xmath30 ) contribution which would lead to an additional absorption line at approximately 5730 ev .
the spectrum is temperature independent down to t = 5 k and consequently the ( ce@xmath24 ) valence state remains unchanged in the examined range 5 k @xmath31 t @xmath31 280 k. ( color online ) field - cooled ( fc ) dc magnetic susceptibility and inverse susceptibility of cecrge@xmath0 .
inset : magnetic susceptibility in the temperature range 2 k - 150 k. the magnetic ordering temperature is represented by the meeting point of the solid and dotted lines .
, width=321 ] the temperature dependence of the magnetic susceptibility @xmath1 of cecrge@xmath0 is shown in fig . 3 in the temperature range 2 - 300 k under an applied field of 0.1 t. at high temperature @xmath1 follows the modified curie - weiss behavior , @xmath32 where @xmath33 is temperature independent susceptibility , c is the curie constant and @xmath34@xmath35 is the weiss temperature .
3 also displays the inverse magnetic susceptibility as a function of temperature along with the fitting of @xmath36(t ) data with the modified curie - weiss behavior in the high temperature range .
the effective moment @xmath37 values calculated from the curie constants are found to be 2.66 @xmath38 and 3.18 @xmath38 for lacrge@xmath0 and cecrge@xmath0 respectively .
the observation of this increased effective moment value for cecrge@xmath0 is consistent with the trivalent state of ce - ions revealed by xas measurement .
it is to be noted in this context that these values are higher than those reported in ref .
for lacrge3 a value of 2.5 @xmath38 has been reported recently by lin et al . [ 15 ] which is nearly same as the value obtained by us .
for cecrge@xmath0 @xmath37 is higher than the calculated value for free ce@xmath24 : 2.54 @xmath38 . as it is confirmed by xas study that ce in cecrge@xmath0
possesses a stable 3 + valance state with a theoretical value of @xmath37 = 2.54 @xmath38 , ce contribution to the curie constant ( @xmath39 ) is @xmath40= 0.81 emu k / mol .
so , now by subtracting @xmath40 from @xmath39 ( = 1.26 emu k / mol ) we can determine the cr contribution to the curie constant , @xmath41 , which gives an effective moment value of @xmath37(cr)= 1.81 @xmath38[see ref .
no 18 for analysis ] .
this @xmath37(cr ) value is less than that for cr@xmath30 ( 2.8 @xmath38 ) or cr@xmath24 ( 3.8 @xmath38 ) .
the reduced cr moment suggests an itinerant character of cr magnetism .
the positive value of weiss temperature @xmath34@xmath35 suggests the presence of dominant ferromagnetic interactions in the system .
the curie temperatures @xmath42 were derived by extrapolating the maximum slope ( d@xmath43/d@xmath3 ) to zero as shown in the inset of fig .
the effective moments @xmath37 , weiss temperatures @xmath34@xmath35 and curie temperatures @xmath42 are summarized in table ii .
4 represents the magnetic isotherms m(h ) of cecrge@xmath0 measured at 1.8 k and 300 k under applied magnetic fields up to 7 t. in the ordered state m(h)at 1.8 k shows a rapid increase up to a value of @xmath8 0.6 @xmath38 at low field followed by slow increase beyond 0.2 t. this may be due to possible canted ferromagnetic ordering of cr - moments as revealed in lacrge@xmath0 by neutron diffraction study @xcite . at 1.8
k magnetization reaches a value of 0.8 @xmath38/f.u . at 7 t for cecrge@xmath0 which is lower than the value 1.3 @xmath38/f.u . at 7
t obtained for lacrge@xmath0 .
the reduced value for cecrge@xmath0 could arise due to crystal field effect / kondo effect/ interaction between the ce and cr sublattice . at 300 k m(h ) is almost linear in h as expected in the paramagnetic state .
( color online ) magnetic field dependence of isothermal magnetization data , m(h ) of cecrge@xmath0 measured at constant temperatures 1.8 k and 300 k.,width=321 ] the electrical resistivity @xmath2 of lacrge@xmath0 and cecrge@xmath0 measured in the temperature range 2 - 300 k is shown in fig .
5(a ) . while lacrge@xmath0 exhibits a typical metallic behavior with residual resistivity @xmath44 @xmath45 65 @xmath46 cm at 2 k and residual resistivity ratio rrr = @xmath47 ,
the electrical resistivity profile of cecrge@xmath0 interestingly exhibits kondo like behavior . for cecrge@xmath0 ,
the resistivity increases with decreasing temperature down to 70 k followed by a drastic fall of electrical resistivity below 70 k as a consequence of reduction of spin disorder resistivity due to cr - moment ordering .
below 30 k the resistivity again increases similar to the case of kondo insulator / semiconductor . here
, the temperature at which the resistivity anomaly occurs has been denoted by @xmath48 and is summarized in table ii .
the magnetic contribution to the resistivity @xmath49 obtained by subtracting the resistivity of lacrge@xmath0 from that of cecrge@xmath0 is depicted in fig .
the temperature variation of @xmath49 follows -ln@xmath3 behavior in two different temperature regions which we believe is associated with the presence of kondo effect as seen in many ce - based kondo lattice / heavy fermion systems .
these two regions are separated by the fm transition around 70 k. a sharp increase of @xmath49 ( ce - la ) below 86 k is an artificial effect of the subtraction process because @xmath4 = 88 k of the la - system is larger than @xmath4 = 68 k of the ce - system .
( color online ) ( a ) electrical resistivity of lacrge@xmath0 and cecrge@xmath0 , ( b ) magnetic contribution to the electrical resistivity .
solid lines are fit to the equation @xmath49= @xmath50 - c@xmath51 ln@xmath3,width=321 ] we have analyzed the data of @xmath49 using the expression : @xmath52 where @xmath50 is the spin disorder resistivity and c@xmath51 is the kondo coefficient .
the fitting to this equation at two different temperature regions are displayed by the solid lines in fig 5(b ) .
the low temperature ( 10 k - 25 k ) -ln@xmath3 behaviour represents the kondo scattering in the crystal field ground state whereas -ln@xmath3 dependence in the high temperature ( 110 k - 300 k ) part is due to kondo effect in the excited multiplets @xcite .
the high temperature ln@xmath3 dependence is one of the characteristic features of the dense kondo systems .
( color online ) ( a ) specific heat of rcrge@xmath0 as a function of @xmath3 .
inset shows the low temperature fit to the equation @xmath53 for cecrge@xmath0 .
( b ) c@xmath54 and magnetic entropy ( s@xmath54 ) as a function of temperature .
inset : c@xmath54 in the temperature range 0.4 k to 1.8 k. , title="fig:",width=321 ] ( color online ) ( a ) specific heat of rcrge@xmath0 as a function of @xmath3 .
inset shows the low temperature fit to the equation @xmath53 for cecrge@xmath0 .
( b ) c@xmath54 and magnetic entropy ( s@xmath54 ) as a function of temperature .
inset : c@xmath54 in the temperature range 0.4 k to 1.8 k. , title="fig:",width=321 ] sometimes kondo system can have large value of sommerfeld coefficient @xmath6 . in order to get the information about the linear term @xmath6 in cecrge@xmath0 system exhibiting kondo like behavior
, we have performed specific heat measurements .
6(a ) represents the temperature dependence of specific heat c(@xmath3 ) of rcrge@xmath0 , for r = la and ce .
the anomalies at @xmath8 84 k and 67 k as seen in c(@xmath3 ) data for lacrge@xmath0 and cecrge@xmath0 respectively are consistent with the anomalies observed in temperature dependence of magnetic susceptibility as well as in electrical resistivity .
the bulk nature of ferromagnetic ordering as evidenced by magnetic measurements and reported in the literature @xcite is confirmed by the anomaly in c(@xmath3 ) data at 84 k and 67 k for lacrge@xmath0 and cecrge@xmath0 respectively .
the low temperature specific heat data below 20 k can be fitted with the equation , @xmath55 where @xmath56 is the electronic contribution to the specific heat , @xmath57 is the phononic contribution to the specific heat [ inset of the fig .
the value of debye temperature @xmath58 can be estimated from @xmath59 using the relation @xcite , @xmath60 [ cols="^,^,^,^,^,^,^",options="header " , ] where r is the molar gas constant and n = 5 is the number of atoms per formula unit ( f.u . ) .
we have listed the obtained values of @xmath6 and @xmath58 for the two compounds in table ii . the sommerfeld coefficient ( @xmath6 ) for cecrge@xmath0
is found out to be 130 mj / molk@xmath7 .
this value is very large compared to the @xmath6 value 2.9 mj / molk@xmath7 obtained for lacrge@xmath0 having 4@xmath61 configuration .
this @xmath6 value of cecrge@xmath0 is comparable with that found in case of moderate heavy fermion system like cefege@xmath0 [ 21 ] , ce@xmath9rh@xmath0ge@xmath62 [ 22 ] , cecoge@xmath0 [ 11 ] , ybni@xmath9b@xmath9c [ 23 ] etc .
assuming that the quasi - particle mass renormalization in lacrge@xmath0 is negligible , the renormalization factor for quasi - particle density of states due to 4f correlations in cecrge@xmath0 can be obtained by the ratio @xmath6(cecrge@xmath0)/ @xmath63(lacrge@xmath0 ) which comes out to be @xmath8 45 .
hence , the quasi - particle mass in cecrge@xmath0 can be roughly estimated as m*@xmath64 45m@xmath65 , where m@xmath65 represents the free electron mass .
although a band structure calculation or de haas - van alphen studies on single crystal of cecrge@xmath0 is required for a better estimate of m * , the effective mass estimated in this way was found to be very close to the experimentally observed value for rrhsi@xmath0 system ( r= la and ce ) where larhsi@xmath0 has @xmath6 = 6 mj / molk@xmath7 [ 25 ] and cerhsi@xmath0 has @xmath6 = 120 mj / molk@xmath7[26 ] .
thus the ratio of @xmath6 gives m * = 20m@xmath65 which is very close to the experimentally observed m * = 19 m@xmath65 to 24 m@xmath65 in de haas - van alphen studies on cerhsi@xmath0[27,28 ] .
thus , the observation of a typical kondo lattice behavior in electrical resistivity and an enhanced @xmath6 value suggest that cecrge@xmath0 can be classified as a member of the heavy fermion system . the observation of heavy fermion behavior even in the presence of ferromagnetic order having large ordering temperature is remarkable in this system . in order to obtain the ce-@xmath14 contribution to the specific heat ( c@xmath54 )
, we have subtracted the heat capacity of lacrge@xmath0 from that of cecrge@xmath0 which is displayed in a plot of c@xmath66 versus @xmath3 , fig 6(b ) .
the first observation is that the @xmath3 dependence of the c@xmath66 value is very flat .
besides the small sharp anomaly at @xmath3 = 2.8 k there is only a weak and broad maximum at 15 k followed by a smooth decrease to higher @xmath3 , which however amount to less than 30% between 15 k and 50 k. the overall c@xmath66 value is enhanced , of the order of 120 mj / molk@xmath7 . for a stable trivalent ce@xmath24 without kondo interaction , one would expect a polarization of the ce - moment in the exchange field produced by fm ordering of cr , possibly followed by ordering of ce on its own at lower temperatures .
this would result in a clear schottky anomaly due to zeeman splitting , possibly followed by a sharp anomaly , and ending in a pronounced decrease to lowest t since the ce system is then in a fully polarized state @xcite .
this scenario does not fit at all to the experimental observations made here , namely , c@xmath66 remains large up to the lowest @xmath3 investigated , 0.4 k ( inset fig .
6(b ) ) , much more similar to the expected behavior of the heavy fermion system .
this large c@xmath66 value at lowest @xmath3 implies strong fluctuations extending down to @xmath3 = 0 k. assuming these fluctuations are due to kondo effect , c@xmath66 allows for an estimate of the kondo temperature @xmath67 using the theoretical expression derived by rajan et al @xcite @xmath68 where r is the gas constant , @xmath69 is the degeneracy of the lowest level concerned and @xmath70 is a scaling temperature which is related to the kondo temperature @xmath71 by @xmath71/@xmath70 = 1.290 .
taking ground state degeneracy @xmath69 = 2 , and ( c@xmath66)@xmath72 = 0.117 j / molk@xmath7 , @xmath71 = 48 k and taking @xmath69 = 4 we get @xmath71 = 143 k. furthermore , the @xmath14-entropy obtained from the integration of c@xmath66 increases continuously at least up to 50 k where it has reached a value close to rln2 [ fig 6(b ) ] . even above 50 k
the magnetic entropy s@xmath73 continues to increase with t , this implies that the energy splitting to first excited cef level should be smaller than 150 k. however , the absence of saturation behavior towards rln2 implies that this splitting is not well defined ( resolved ) .
the high @xmath71 obtained from the sommerfeld coefficient provides a first simple explanation , since a strong correlation that is responsible for high gamma value would lead to a broadening of the cef level of the same order of magnitude as the splitting .
however , the exchange field due to the fm ordering of cr moments provides a further mechanism , but the exchange field needed is huge , since it has to lead to zeeman splitting of the order of 150 k. inelastic neutron scattering experiments are required to have a clear idea about the cef level scheme in this system .
( color online ) thermoelectric power of lacrge@xmath0 and cecrge@xmath0 as a function of temperature.,width=340 ] as a complementary probe for characteristics of the kondo effect , we performed thermopower measurements .
fig.7 shows the thermopower s(@xmath3 ) of lacrge@xmath0 and cecrge@xmath0 between 5 and 300 k. additionally we calculated the magnetic contribution s@xmath74 to the thermopower of cecrge@xmath0 by the gorter - nordheim relation , @xmath75 where s@xmath76 and @xmath44 are the data for the isostructural la compound .
typically , ce - based kondo systems exhibit ( a ) enhanced thermopower values , which are 10 - 100 times larger than that of simple metals like cu @xcite , and ( b ) a positive sign of s(@xmath3 ) @xcite . for well - defined cef resonances and a low @xmath71
one also finds ( c ) two maxima : one around the kondo temperature @xmath71 and the second around a temperature corresponding to roughly 0.3 to 0.6 of the cef splitting @xcite .
they reflect kondo scattering on the ground state and on the excited cef states , respectively .
our data show ( a ) indeed a very high value of both s@xmath74 ( 35 @xmath77 v / k ) and s(@xmath3)(20 @xmath77 v / k ) of cecrge@xmath0 .
additionally we find ( b ) that s(@xmath3 ) is positive in the whole temperature range .
both properties indicate that cecrge@xmath0 is a kondo system . concerning ( c ) we
, however , do not see any extrema in s@xmath74 , but a monotonic increase up to room temperature .
since we expect a relatively high kondo temperature from our specific heat measurements and additionally do not find clear signatures of the cef levels in the specific heat either , we are far from the conditions mentioned for ( c ) .
indeed , it is known from theoretical calculations describing experiments e.g. on ceru@xmath9ge@xmath9 , that the temperature dependence of the thermopower can vary drastically and can even become monotonic depending on the kondo scale and the cef splitting @xcite . in the lacrge@xmath0 system
we additionally find a broad hump between 30 and 60 k , which is believed to be related to the ordering of the cr moments .
we have provided experimental evidence for the presence of kondo effect / heavy fermion behavior of the trivalent ce compound cecrge@xmath0 in presence of the ferromagnetic order due to itinerant cr - moments . while the conclusive evidence of the trivalent nature was obtained from xas , the presence of kondo behavior was revealed by electrical resistivity , thermoelectric power and heat capacity data .
the heavy fermion nature was revealed using the heat capacity measurement .
specific heat data show an enhanced @xmath6 ( 130 mj / molk@xmath7 ) value for cecrge@xmath0 compared to 2.9 mj / molk@xmath7 , obtained for lacrge@xmath0 . although magnetic order of cr was known from earlier magnetization data , our heat capacity data confirms the bulk nature of the magnetic order .
hence , the compound cecrge@xmath0 , having a high @xmath71 ( @xmath8100 k ) and a trivalent integral ce valence state at room temperature as well as at low temperature , appears to be a rare example of ce based system exhibiting heavy fermion behavior even in the magnetically ordered state .
we would like to thank dr .
v. k. anand for useful discussion .
this work is partially supported by council of scientific and industrial research ( csir ) , new delhi ( grant no .
80(0080)/12/ emr - ii ) . | physical properties of polycrystalline cecrge@xmath0 and lacrge@xmath0 have been investigated by x - ray absorption spectroscopy , magnetic susceptibility @xmath1 , isothermal magnetization m(h ) , electrical resistivity @xmath2 , specific heat c(@xmath3 ) and thermoelectric power s(@xmath3 ) measurements .
these compounds are found to crystallize in the hexagonal perovskite structure ( space group _ p6@xmath0/mmc _ ) , as previously reported .
the @xmath2 , @xmath1 and c(@xmath3 ) data confirm the bulk ferromagnetic ordering of itinerant cr moments in lacrge@xmath0 and cecrge@xmath0 with @xmath4 = 90 k and 70 k respectively .
in addition a weak anomaly is also observed near 3 k in the c(@xmath3 ) data of cecrge@xmath0 .
the t dependences of @xmath5 and finite values of sommerfeld coefficient @xmath6 obtained from the specific heat measurements confirm that both the compounds are of metallic character .
further , the @xmath3 dependence of @xmath5 of cecrge@xmath0 reflects a kondo lattice behavior .
an enhanced @xmath6 of 130 mj / molk@xmath7 together with the kondo lattice behavior inferred from the @xmath2 establish cecrge@xmath0 as a moderate heavy fermion compound with a quasi - particle mass renormalization factor of @xmath8 45 . |
reverberation mapping ( rm ) is a key technique for the study of active galactic nuclei ( agn ) .
the technique is based on the temporal fluctuations of the central continuum source , and the subsequent response of the broad line region ( blr ) emission .
the time delay between the continuum and the broad line fluctuations provides an estimate of the size of the blr , and can also be used to estimate the black hole mass @xcite .
rm is a observationally intensive , requiring observations of an agn over a period of a few tens of days @xcite . as a result , many authors have studied the data analysis techniques involved in rm , and substantial advances have been made in recent years .
the data analysis methods introduced range from those that attempt to infer the transfer function ( the distribution of lags in a single object , e.g. * ? ? ?
* ; * ? ? ?
* ) , the velocity - resolved transfer function @xcite , or the physical structure of the blr itself @xcite .
there are many subtleties and challenges involved in reverberation mapping , that we will ignore for the purposes of this letter .
we note two of them here for completeness .
firstly , the mean lag @xmath2 , where @xmath3 is the normalised transfer function , is not equal to @xmath4 times the mean radius of the blr matter distribution , @xmath5 .
secondly , the mean lag @xmath6 is not equivalent to the peak of the cross - correlation function , except in the case of very narrow transfer functions .
direct physical modelling of the blr resolves these issues @xcite .
recently , @xcite introduced an innovative approach to reverberation mapping where the results from multiple agn can be combined to yield inferences about the entire sample of agn , despite the fact that the constraints on any individual agn are poor . rather than accurately measuring @xmath0 in a single object ,
it is possible to roughly measure @xmath0 for a large number of objects ( for example , by only measuring the blr emission line flux at two epochs ) , and to infer properties about the distribution of @xmath0 values in the sample of objects ( and hence in a broader population , if the sample can be considered representative ) .
the data analysis approach of @xcite was based on an extension of the traditional cross - correlation method to @xmath7 separate objects . by `` stacking '' the cross - correlation functions ( ccfs ) from the @xmath7 objects in the sample , a peak appears which indicates a typical lag value in the sample of agn .
however , this peak is difficult to interpret .
specifically , the width of the peak may be influenced either by uncertainty ( due to the sparse data ) or due to a wide range of lags being present in the sample ( i.e. the sample of agn is diverse ) . in this letter
, we revisit the idea of `` stacked '' reverberation mapping ( i.e. using a sample of agn with sparse data to infer something about the sample ) from a bayesian perspective .
we develop a bayesian hierarchical model @xcite . which allows us to calculate the posterior distribution for hyperparameters which describe the typical lag , and the diversity of the lags .
in section [ sec : multiple ] we discuss the general principles needed to combine information from multiple poorly - measured objects . in section [ sec : single_object ] we discuss the methods used to infer the lag @xmath0 using rm data for a single agn .
we then demonstrate the technique on simulated data in section [ sec : data ] , and conclude in section [ sec : conclusions ] .
consider a sample of @xmath7 objects , each of which has parameters @xmath8 .
if the @xmath9th object is analysed on its own , the inference about its parameters @xmath8 is described by a posterior distribution @xmath10 where @xmath11 is the prior distribution and @xmath12 is the likelihood function .
if @xmath7 objects are analysed separately , inferences can be made about the diversity of @xmath8 across the sample .
however , these inferences may be incorrect due to the implicit assumption that the prior for all of the @xmath13 is independent @xcite .
a hierarchical model may be used to overcome this problem . in a hierarchical model , the prior for the parameters @xmath13
is created by introducing hyperparameters @xmath14 such that the joint prior for @xmath14 and the @xmath13 is : @xmath15 the implied prior on the @xmath13 parameters is then @xmath16 which may imply that the @xmath17 values are `` clustered '' around some typical value .
the posterior distribution for @xmath13 and the hyperparameters @xmath14 given the data @xmath18 , is then @xmath19 it is possible to consider the data for all @xmath7 objects as one big data set and infer the hyperparameters @xmath14 and the parameters of all objects @xmath13 simultaneously .
however , this can be computationally prohibitive . in some circumstances
it is tractable to analyse each object s data separately ( using a common prior @xmath11 ) and then post - process the results , reconstructing what the results of the hierarchical model would have been . in this letter
we use this latter approach .
the marginal posterior distribution for the hyperparameters @xmath14 is @xmath20\label{eq : combine}\end{aligned}\ ] ] where the expectation is taken with respect to the individual object posterior of equation [ eq : individual ] , and thus can be estimated using posterior samples from the posterior distributions for the individual objects .
this result enables us to reconstruct the posterior distribution for the hyperparameters even though the individual object inferences were made without the hierarchical structure in the prior .
this is essentially an importance sampling approximation to the full hierarchical model .
one drawback of this approach is that it implies an independent ( non - hierarchical ) prior on any per - object nuisance parameters , which can sometimes adversely affect the results ( e.g. * ? ? ?
* ) . in section [ sec
: single_object ] we define a simple model for inferring the lag of a single agn from sparse rm data .
we will use markov chain monte carlo ( mcmc ) to produce posterior samples for the lags of the @xmath7 objects , which can then be combined using equation [ eq : combine ] to yield the posterior distribution for some hyperparameters describing the distribution of lags in the sample .
if the continuum light curve of an agn is described by a function @xmath21 , then the line light curve @xmath22 is given by @xmath23 \ , d\tau\end{aligned}\ ] ] where @xmath3 is the transfer function ( assumed to be normalised ) , and @xmath24 and @xmath25 are response coefficients .
the idea of reverberation mapping is to use noisy measurements of @xmath21 and @xmath22 to infer the transfer function @xmath3 or a summary of it such as the mean lag @xmath26 . throughout this letter
we consider @xmath21 and @xmath22 in flux units , as opposed to magnitudes , and consider @xmath6 as the definition of `` the lag '' of an agn .
the posterior distribution for the lag @xmath6 of a single object @xmath9 may be obtained by fitting the following model .
we assume , for simplicity , that the transfer function is uniform between limits @xmath27 and @xmath28 , where @xmath29 : @xmath30\\ 0 , & \textnormal{otherwise } \end{array}\right.\end{aligned}\ ] ] and our goal is to measure the mean lag : @xmath31 note that inferring @xmath6 from the data requires that we marginalise over an infinite number of nuisance parameters describing the behaviour of @xmath21 at unobserved times @xcite .
the prior for the underlying time variation of the continuum emission is a continuous autoregressive process of order 1 , or a car(1 ) model .
these models have been studied extensively for agn variability ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this marginalisation can be done either analytically or inside mcmc .
we used the latter approach for simplicity .
we discretised time using ten time bins per day , and have a discrete continuum light curve @xmath32 , included in the model as a set of unknown parameters .
the prior for these parameters is : @xmath33 for @xmath34 .
this is the discrete ar(1 ) model from time series theory .
@xmath35 describes the mean level of the continuum light curve , @xmath36 controls the size of the short - term fluctuations , and @xmath37 controls the correlation timescale ( the timescale itself is given by @xmath38 ) . for the likelihood ( or sampling distribution , really the prior for the data given the parameters ) we made the conventional assumption of `` gaussian noise '' : @xmath39 we used vague priors for the parameters of the single - object model . in practice
these could be made substantially narrower .
we chose a log - uniform prior for @xmath40 ( the agn variability timescale ) between @xmath41 and @xmath42 days , a log - uniform prior for @xmath36 ( the size of the short - term fluctuations ) between @xmath43 and @xmath44 flux units , a cauchy(0 , 100 ) prior for @xmath35 ( the mean flux level of the agn ) , a log - uniform prior for @xmath28 ( the upper limit of the transfer function ) between @xmath45 and @xmath46 days , a uniform prior for @xmath27 ( the lower limit of the transfer function ) between 0 and @xmath28 , a log - uniform prior between @xmath43 and @xmath44 for the response coefficient @xmath24 and a cauchy(0 , 100 ) prior for the second response parameter @xmath25 . to implement the mcmc for a single object
, we implemented our model in the stan sampler @xcite and diffusive nested sampling @xcite . due to the large number of parameters , multiple modes , and strong correlations in the posterior distribution
, we found that diffusive nested sampling was more effective than stan .
note that our single - object model is already an improvement over standard cross - correlation techniques and is similar to the approach used by @xcite .
to test our hierarchical model , and compare it to the stacked cross - correlation function , we simulated data from a sample of 100 agn .
we assumed the continuum flux for each object fluctuates around a mean value @xmath47 in arbitrary units .
the variability timescales @xmath40 were simulated from a very broad log - uniform distribution ( i.e. a uniform distribution for @xmath48 ) with a lower limit of 10 days and an upper limit of @xmath49 days .
the variability amplitudes @xmath36 were also simulated from a log - uniform distribution with lower limit of 0.2 flux units and an upper limit of 1 flux unit .
this implies some objects are more variable than others , as is typical in reverberation mapping .
each object in the sample had its own transfer function parameters .
the upper limit @xmath28 of the transfer function was simulated using a log - normal distribution with a median value of 10 days , and the standard deviation of @xmath50 was 0.3 .
given @xmath28 , @xmath27 was simulated from a uniform distribution between @xmath51 and @xmath28 .
the resulting distribution of lags @xmath52 is very well approximated by a lognormal distribution with a median of 7.4 days and width ( quantified by the standard deviation of @xmath53 ) of 0.157 ( these two quantities are the ones we will infer using the hierarchical model ) .
the response amplitudes @xmath24 and @xmath25 were set to 0.5 and 0 respectively , for all objects .
note that the distribution of the parameters of the 100 objects is not the same as the priors used when analysing them with the single - object model .
the measured data for each of the 100 agn was continuum flux measurements , once per day , for 100 days .
the standard deviation of the measurement noise is 1 flux unit ( i.e. 2% ) .
the line data is measured on only two days , with times selected from a uniform distribution between @xmath54 and @xmath55 days ( i.e. corresponding to the latter half of the continuum data ) .
the flux of the line data is typically about half that of the continuum data ( since @xmath56 for all objects ) , and the measurement noise for the line data is 0.25 flux units , or about 1% . see figure [ fig : data ] for three illustrative data sets from the sample . for any of these objects
( although there is some information available ) . see figure [ fig : example_posteriors ] for the @xmath6 posterior distributions for these three data sets .
[ fig : data ] ] we computed the stacked cross - correlation function for the 100 data sets .
the cross - correlation function for a single object is defined as in @xcite . for a lag bin between @xmath0 and @xmath57 the amount of cross correlation
is calculated by looping over pairs of points ( one selected from the continuum data set , and another from the line data set ) , and accumulating mass when the lag between the two points is between @xmath0 and @xmath57 .
@xmath58\right ) } { \sum_{i=1}^{n_y}\sum_{j=1}^{n_l } \mathds{1}\left((t_{l_j } - t_{y_i } ) \in [ \tau , \tau + \delta]\right)}\end{aligned}\ ] ] the denominator is the number of pairs of points that fall within the lag range being considered , @xmath59 $ ] , denoted @xmath60 by @xcite .
the stacked cross - correlation function for @xmath7 objects is sum of the ccfs of the individual objects . in figure
[ fig : ccf ] , we show the stacked ccf ( on a log scale ) from our 100 simulated agn data sets . there is a clear peak around 10 days , and the full width at half maximum , while not strictly well defined , is roughly 0.25 dex .
days , but the stacked ccf has a large width .
it is difficult to interpret the cause of this width , and to determine whether the width is due to uncertainty or diversity in the sample of agn .
these difficulties motivate the hierarchical bayesian model .
[ fig : ccf ] ] as the data does not contain measurements at all possible time lags , the stacked ccf will always require some smoothing . here
, the smoothing is supplied by the choice of a bin width @xmath61 .
the stacked ccf has a substantial width .
one weakness of the stacked ccf method is that it is not clear whether this width is caused by real diversity in the sample of agn , or whether it is simply caused by the difficulty in measuring the lag @xmath6 of an object with such sparse data .
by contrast , the hierarchical bayesian approach clearly separates these two concepts . to implement the hierarchical bayesian approach on the simulated data
, we first used the single - object model on each object .
the results for three objects ( the same three objects shown in figure [ fig : data ] ) are plotted in figure [ fig : example_posteriors ] . due to the sparse data
, there is a large amount of uncertainty about the value of @xmath6 for each system .
some objects do not allow any real measurement of @xmath6 , but some do , as a result of fortuitous variability of the continuum . for three objects whose
data sets are shown in figure [ fig : data ] .
these posteriors were obtained by analysing each object separately with the single - object model . for the first object , there is a moderate `` detection '' of a lag at about 10 days ( although there is a small second mode above 100 days , which can be understood by inspecting the data in figure [ fig : data ] ) .
the second object s lag was not well constrained , and the third object s data only provided an upper limit on its lag .
[ fig : example_posteriors ] ] finally , we combined the results from the single - object model into an inference about hyperparameters @xmath62 and @xmath63 describing the central value and the diversity of @xmath6 respectively , using the result of equation [ eq : individual ] .
this requires us to know the prior @xmath64 implied by the simple - object model , since we did not assign the prior directly to @xmath6 but indirectly through @xmath27 and @xmath28 .
the prior for @xmath6 in the single - object model is very well approximated by a log - uniform distribution . since lags
@xmath6 are positive , our assumption for the prior on @xmath6 given the hyperparameters is a lognormal distribution with median @xmath62 and width ( standard deviation of @xmath53 ) @xmath63 : @xmath65 the joint posterior distribution for @xmath62 and @xmath63 , given the data from @xmath66 agn , is shown in figure [ fig : posterior ] .
the prior for @xmath62 and @xmath63 was uniform inside the rectangle shown , and the posterior distribution covers a much smaller area , indicating that the data contained a lot of information about the hyperparameters .
the true values used to generate the simulated data are denoted by the star symbol in figure [ fig : posterior ] , and is typical of the posterior distribution , as it should be . and @xmath63 , hyperparameters describing the distribution of @xmath6 values in the sample .
the prior distribution was uniform over the area shown .
the true solution @xmath67 is indicated by the white star symbol .
@xmath68 corresponds to a typical lag of 7.4 days.[fig : posterior ] ] the marginal posterior distribution for @xmath62 is shown in figure [ fig : posterior2 ] along with the posterior predictive distribution for the `` next '' @xmath6 value .
the inference about @xmath62 can be summarised as @xmath69 ( consistent with the known input value of @xmath70 ) , and the prediction about @xmath53 may be summarised as @xmath71 . the bayesian approach has clearly separated the uncertainty about @xmath62 from the diversity of the sample , described by @xmath63 . , the hyperparameter describing the center of the distribution for @xmath72 ,
is plotted as the solid line .
[ fig : posterior2 ] ] note that our method for combining individual object results into an inference about the population ( section [ sec : multiple ] ) assumes that the hierarchical prior applies to only a single parameter ( in our case @xmath6 ) .
if we incorporate additional information by using a hierarchical prior for the other model parameters , allowing us to detect ( for example ) that all of the response amplitudes @xmath24 were equal to 0.5 , we may be able to make more accurate inferences .
however , this would increase the complexity of the method and the computational expense .
reverberation mapping campaigns are observationally intensive , yet potentially very informative .
this has led to research into rm data analysis techniques , as well as less costly observing strategies .
the recent approach of @xcite demonstrated that it is possible to combine information from many sparsely - measured agn and infer properties of the sample rather than insisting on accurate results about a particular object . in this letter
we extended the approach of @xcite by implementing a bayesian hierarchical model for the data analysis .
the main advantage of this approach is a clear interpretation of the output , which is a posterior distribution for hyperparameters describing the sample of agn .
we demonstrated this approach on a simulated data set consisting of 100 agn with well - measured continuum light curves but sparsely measured ( two epochs per agn ) broad line light curves .
for the simulated data set we were able to obtain an inference on @xmath62 , a hyperparameter describing the typical value of the lag , to within 0.047 dex ( 1-@xmath63 uncertainty ) , whereas the stacked cross - correlation function had a fwhm of @xmath73 0.25 dex .
this is possible because a subset of the objects in the sample will have a measureable lag .
we hope this approach will facilitate informative studies of samples of agn .
it is a pleasure to thank tommaso treu , anna pancoast ( ucsb ) , and brandon kelly ( ucsb ) for many useful conversations about reverberation mapping .
the stan ( mc-stan.org ) team provided helpful advice about the use of their software .
bjb is partially supported by the marsden fund ( royal society of new zealand ) .
we are grateful to the referee , stephen fine , for his helpful comments . | reverberation mapping ( rm ) is an important technique in studies of active galactic nuclei ( agn ) .
the key idea of rm is to measure the time lag @xmath0 between variations in the continuum emission from the accretion disc and subsequent response of the broad line region ( blr ) .
the measurement of @xmath0 is typically used to estimate the physical size of the blr and is combined with other measurements to estimate the black hole mass @xmath1 .
a major difficulty with rm campaigns is the large amount of data needed to measure @xmath0 .
recently , @xcite introduced a new approach to rm where the blr light curve is sparsely sampled , but this is counteracted by observing a large sample of agn , rather than a single system .
the results are combined to infer properties of the sample of agn . in this letter
we implement this method using a hierarchical bayesian model and contrast this with the results from the previous stacked cross - correlation technique .
we find that our inferences are more precise and allow for more straightforward interpretation than the stacked cross - correlation results .
[ firstpage ] galaxies : active methods : data analysis |
the manipulation of microscopic particles has benefited from the high level of control and measurement provided by optical tweezers . with the technological development of fast configurable spatial light modulators ( slms ) allowing for ever more complex trapping geometries @xcite , new applications have emerged .
for example , sculpted light may have an important role in overcoming multiple light scattering issues in complex biological tissues , and such biomedical applications have only begun to be explored .
in particular the development of sculpted light patterns across the image plane , such as the generation of large trapping arrays , could have application to the in vivo trapping of larger objects , such as living cells @xcite . in degenerate quantum gases ,
the push for increasing diversification of optical trapping potentials has led to the adoption of many of the techniques from holographic optical tweezers .
slms are most often used in the fourier plane of an optical system , manipulating the phase of an input optical field to produce a configurable pattern in the conjugate trapping plane of the system @xcite .
these methods have been successfully applied to address and pattern atoms trapped in optical lattices @xcite , but demonstrations of microscopically configurable trapping potentials have been lacking , with a notable exception being the production of multiple focused spots for the confinement of single atoms @xcite .
an alternative technique to manipulating the phase of the input beam is to instead utilise direct imaging .
though somewhat rarely encountered in optical tweezers , the technique known as _ generalised phase constrast _ uses the combination of a phase - based slm and a phase - contrast filter to first create an amplitude pattern in an intermediate image plane , which is then directly reimaged to the optical tweezing plane @xcite .
the advantages of this technique are both speed and simplicity the desired amplitude pattern can be directly written to the slm without requiring the calculation of the appropriate hologram .
this technique likewise avoids the generation of phase defects and speckle in the imaged pattern that can plague slms in the fourier plane @xcite , while being adaptable to the generation of large numbers of traps @xcite .
this comes at the cost of the ability to correct wavefront aberrations , but this disadvantage can be mitigated with a well - corrected optical system , as shown here .
another drawback is that the light efficiency is proportional to the fraction of illuminated trap area to maximum trap area .
a more recent addition to the toolkit for producing arbitrary optical potentials has been the digital - micromirror device ( dmd ) .
consisting of ( up to ) millions of individually addressable mirrors in a compact package , dmds have the advantage of fast full - frame refresh rates on the order of 20 khz , @xmath1 that of comparable liquid - crystal based slms .
dmds can also be operated in a fixed fashion ( dc ) as they latch mirrors between reset pulses .
originally developed for digital light processing ( dlp ) , these devices have seen increasing use in laboratory and industrial applications @xcite .
a dmd can be considered a dynamically configurable amplitude mask , which makes it highly suitable for direct imaging applications .
dmds have been used to produce flattop beams for implementation into quantum gas experiments @xcite , incorporated into high - resolution systems for the purpose of single - site addressing in atomic quantum gas microscopes @xcite , used to produce moving lattice potentials @xcite , and have recently been utilised to produce target - shaped traps @xcite .
dmds may be used in either the fourier plane @xcite or directly imaged @xcite .
we demonstrate here the utility of direct imaging of a dmd at high numerical apertures ( na ) for optical trapping , which we apply to trapping of a bec .
our optical system has the major advantage of using commercially available optics and microscope objectives external to our glass vacuum chamber , and corrected for the relatively thin 1.25 mm thick walls . with this system
we demonstrate patterning of potentials with an upper - bound resolution of 630(10 ) nm fwhm at 532 nm illumination , within 5% of the diffraction limit for our 0.45 na objective and @xmath2 improved on recently reported results @xcite which utilised a relatively low @xmath3 na imaging system .
the high - resolution potentials appear robust to tilts and misalignments of the objective and glass walls , in contrast to other cold atom experiments @xcite .
these patterns have an image extent of @xmath4 in the atom plane , which allows nearly arbitrary sculpting of the optical potential and corresponding bec .
subsequent imaging at the 780 nm resonant wavelength of @xmath5rb achieves a submicron resolution of 960(80 ) nm fwhm , within 8% of the diffraction limit .
our dmd allows the storage of 13,889 frames and has a full - frame frame rate ranging from dc to 20 khz , enabling diverse and dynamically configurable potentials .
we utilise this rapid switching rate to introduce the use of dmds for producing time - averaged potentials , which have been previously produced from rapidly scanning beams @xcite .
we find that modulation frequencies above @xmath6 khz produce negligible heating of the atoms while allowing the production of six grayscale levels .
this technique can be combined with binary error - diffusion ( halftoning ) @xcite to increase the number of grayscale levels available .
our apparatus consists of a three - part vacuum system , comprising a two - dimensional magneto - optical trap ( 2dmot ) , three - dimensional mot ( 3dmot ) and attached science cell cuvette ( see fig .
[ fig_apparatus ] ) .
the cell is made of suprasil quartz and has 1.25 mm thick walls with an external broadband anti - reflection coating .
the 2dmot is separated from the 3dmot and science cell sections with a 12 mm long differential pumping tube with a 1.2 mm inner diameter .
this results in a pressure differential of @xmath7 , with a pressure of @xmath8 torr and @xmath9 torr on the 3dmot and 2dmot sides respectively .
the 2dmot loads @xmath10 @xmath5rb atoms into the 3dmot in 5 s. the subsequent bec production closely follows the methodology of ref . @xcite .
after a 40 ms compressed mot stage ( cmot ) @xcite , the atoms are optically pumped to the @xmath11 manifold and captured in a 100 g / cm quadrupole magnetic field . using two additional anti - helmoltz coil pairs @xcite ( see fig . [ fig_apparatus ] ) , the atoms are adiabatically transferred to the science cell in @xmath12 s @xcite .
the quadrupole field is subsequently increased to 145 g / cm , and evaporation occurs over 4 s by driving the microwave transition from @xmath13 , resulting in a cloud of @xmath14 atoms at a temperature of @xmath15k .
the field is then relaxed to 27 g / cm , loading @xmath16 atoms at @xmath17k into a red - detuned 1064 nm beam with a @xmath18 m waist , aligned @xmath19 m below the quadrupole zero , and with a potential depth of 110 @xmath20 .
we find that the cancellation of stray magnetic fields is important for efficient loading of the dipole trap .
optical evaporation is performed by reducing the beam intensity , resulting in a temperature of @xmath21 nk , with a critical temperature for condensation of @xmath22 nk .
if the optical evaporation is continued , becs of @xmath23 atoms can be produced in this trap .
instead , we load the cold thermal atoms into a second 1064 nm red - detuned beam focused into a sheet with radial and vertical waists ( @xmath24 ) = ( 1 mm , 8.5 @xmath25 m ) and a trap depth of 6.4 @xmath20 .
subsequent evaporation in the sheet - magnetic hybrid trap results in a bec of @xmath26 atoms in a highly flattened geometry , with final trapping frequencies of @xmath27 hz .
after evaporating in the sheet hybrid trap , we subsequently remove the magnetic field , while ramping up the sheet potential , resulting in all - optical confinement with @xmath28 hz .
the addition of the dmd potential to the sheet ( see below ) can dominate the weak radial harmonic confinement of this trap , producing nearly uniform atomic distributions .
in order to produce arbitrarily patterned becs , we directly image a dmd ( visitech luxbeam 4600 wuxga 1200@xmath291920 pixels ) to the atom plane by illuminating the surface with spatially - filtered 532 nm light as shown in fig . [ fig_imaging ] .
this results in approximately gaussian illumination with @xmath30 mm and maximum dmd diffraction efficiency of @xmath31 40% . an intermediate image of the dmd with @xmath32 magnification is initially produced using a matched achromat imaging system .
an aperture at the fourier plane of this system removes spurious dmd diffraction orders , which otherwise contribute stray light to the dark regions of the imaged patterns and reduce the trap lifetime .
subsequently , the dmd is imaged to the atom plane using a 500 mm achromat and nikon cfi l plan epi 20xcr infinity corrected objective , resulting in @xmath33 magnification .
the total system thus images the dmd to the atom plane with a final minimization factor of 100 ( @xmath34 magnification ) .
the objective has a numerical aperture ( na ) of 0.45 and an adjustable glass correction collar .
we have found the correction collar to be important in obtaining high - resolution patterns inside the science cell , but the system is relatively insensitive to the precise angle of the objective and only requires coarse tuning . magnification and 532 nm illumination , resulting in 650(50 ) nm fwhm ( @xmath35 nm ) .
the fwhm was determined through 100 1d fits of the image at varying angles through 180 degrees ; a single 1d fit is shown here for illustrative purposes .
( b ) a single dmd mirror imaged with @xmath36 magnification and 780 nm illumination , the imaging wavelength for @xmath5rb , resulting in 960(80 ) nm fwhm ( @xmath37 nm ) .
( c ) siemens star resolution target , as projected and reimaged with 532 nm light .
( d ) siemens star imprinted on to the atomic density and averaged over 10 runs of the experiment .
quantitative analysis of the resolution targets is presented in fig .
[ fig_mtf ] . ] as the dmd mirrors have only two states on and off the resulting image is binary .
however , given the limited spatial resolution of a typical optical system , the technique know as error - diffusion or halftoning can be used to produce intensity gradients @xcite .
we determine the resolution of our system by first turning on a single mirror of the dmd and projecting it on to the atom plane .
the mirror pitch is 10.8 @xmath25 m , so the minification factor of 100 results in an imaged mirror width of 108 nm , below the theoretical resolution limit of the top microscope system ( 605 nm fwhm at 532 nm illumination ) .
as the single mirror is not resolvable , its image at the atom plane approximates the point - spread function ( psf ) of the dmd imaging system .
we reimage this focused spot on the camera with the bottom microscope system .
accounting for the magnification factor of the reimaging system results in a 650(50 ) nm fwhm peak in the atom plane , as shown in fig .
[ psf_fig](a ) .
this value is an upper limit , as it convolves any abberations of the reimaging system into the estimate of the spot size at the atom plane .
back - propagating this resolution element to the dmd location gives a 65 @xmath25 m spot which spans a @xmath38 block of mirrors ; multiple mirrors will thus contribute to the resolution spot in the atom plane @xcite .
we therefore can use error - diffusion to control the light intensity at the atom plane , as shown in fig .
[ fig_dmd_patterns](b ) .
the right half of fig .
[ fig_imaging ] illustrates the bottom microscope imaging system . a second microscope objective ( olympus lcpln20xir ) similar to the projection objective with a 0.45 na and correction collar , but with near - infrared coatings , forms the first part of the imaging system .
collimated light at 780 nm is produced through the top objective , resulting in an imaging beam 1.1 mm in diameter .
we then incorporate a lens relay system to allow us to implement dark - field imaging in the future @xcite .
finally , an intermediate image is formed with @xmath39 magnification which is then reimaged to the ccd camera ( pixis 1024b ) with the adjustable vzm450 zoom lens ( @xmath40 @xmath41 ) .
we therefore produce absorption images of the condensate in the final trap configurations with a total zoom of ( @xmath42 @xmath36 ) . by again turning on a single mirror of the dmd , but illuminating it at 780 nm
, we can estimate the psf at this wavelength , as shown in fig .
[ psf_fig](b ) .
this results in a 960(80 ) nm fwhm peak , consistent with the measurement at 532 nm after accounting for the increased wavelength .
the resulting high resolution dmd patterns produce a repulsive potential for the atoms trapped in the sheet - magnetic hybrid .
we ramp on the intensity of the laser illuminating the dmd near the end of the final evaporation sequence in the sheet - hybrid trap to a value @xmath43 , where @xmath25 is the chemical potential of the bec in the final trapping potential . in the case of a 80 @xmath25 m @xmath29 50
@xmath25 m rectangular trapping geometry , the chemical potential calculated using the measured trap frequencies and 700,000 atoms is @xmath44 .
subsequently , the sheet power is linearly increased over 500 ms to keep the atoms trapped vertically while the magnetic field is linearly ramped to zero , increasing the vertical trapping frequency to @xmath45 hz .
the light power incident on the dmd can also be increased in this step .
however , we have observed a decrease in the bec lifetime with increasing intensity , with measured time constants of ( 18.2 , 10.9 , 9.1 , 8.2 ) s for a 80 @xmath25 m @xmath29 50 @xmath25 m rectangular dmd box potential with peaks of ( 1.2@xmath25 , 2.4@xmath25 , 4.8@xmath25 , 7.2@xmath25 ) respectively .
similar intensity - dependent lifetimes were observed for different trap geometries . as the atoms are confined to areas of low dmd light intensity , we believe residual vibrational coupling in the dmd optical path results in parametric heating of the cloud .
several high - resolution atom - trap configurations are shown in fig .
[ fig_dmd_patterns ] , emphasizing the wide range of possible patterns .
these are imaged immediately after turning off the optical trap .
alternatively , by imaging from the side after 27 ms time of flight ( tof ) expansion , we can determine atom number and bec fraction for the different distributions .
we note that we have used the same evaporation ramp for each of these configurations we expect that a near pure bec can be achieved for each with optimization .
we generally observe lower condensate fractions for smaller enclosed areas for similar optical evaporation profiles .
we briefly describe examples of the trapping potentials we have generated .
figure [ fig_dmd_patterns](a ) demonstrates a ring trap , of interest for atom interferometry @xcite and studies of phase slips and persistent currents @xcite .
figure [ fig_dmd_patterns](b ) shows a 10 @xmath25m - period optical lattice .
the binary dmd pattern was generated by applying a floyd - steinberg error diffusion algorithm @xcite to an 8-bit grayscale image of a sinusoidal lattice . as we have not included a boundary to the lattice in the dmd pattern , we leave the magnetic field on and retain a @xmath46 hz radial trapping frequency , producing a more symmetrically filled lattice .
figure [ fig_dmd_patterns](c ) demonstrates a checkerboard pattern applied to the bec , and the evaporation ramp in this case results in a bec with negligible thermal component .
this pattern emphasizes the sharp features in the atomic distribution resulting from the highly resolved features of the dmd pattern .
ring lattices have generated wide interest @xcite , and fig .
[ fig_dmd_patterns](d ) demonstrates a ring lattice of 25 sites . finally , in fig .
[ fig_dmd_patterns](e ) we project artistic impressions of bose and einstein into a nearly pure bec , resulting in a `` bose - einstein '' bose - einstein condensate and demonstrating our ability to create arbitrary potentials .
this image is an average over 5 experimental runs ( 30 s experimental cycle ) and is similar to fig .
[ psf_fig](d ) which uses a 10-run average .
these images demonstrate the repeatability of the high - resolution patterned becs .
the integrated od tof cross sections in fig .
[ fig_dmd_patterns](d - e ) deviate from gaussian fit at their center . in the case of fig .
[ fig_dmd_patterns](d ) , this is due to uneven density distribution of atoms across the ring lattice and insufficient spatial overlap , or incoherent overlap of the lattice sites after only 27 ms tof . in the case of fig .
[ fig_dmd_patterns](e ) , this deviation comes from to the high atomic densities leading to ods above @xmath47 which are below the signal to noise threshold of the horizontal - imaging . in each of these cases , the limited expansion of the cloud , when compared with the thermal component of fig .
4(b ) , along with the lack of a bimodal distribution , indicate a relatively pure bec state .
for a more quantitative characterisation of the optical systems @xcite , we produced a binary siemens star resolution target on the dmd , imaged the projected light pattern , and imaged becs loaded in to the pattern , as shown in fig .
[ psf_fig](c , d ) .
these patterns can then be analyzed to determine the modulation transfer function ( mtf ) of the total optical system .
this is accomplished using the protocol illustrated in fig .
[ fig_mtf ] .
we first determine circular paths around the siemens star , with frequency spacings @xmath48 line - pairs ( lp)/@xmath25 m , for the ( light , atom ) siemens star .
a contrast value is found for each of the 16 adjacent bright and dark spoke pairs , and the average contrast is calculated for each circular path . as the mtf is the fourier transform of the psf , which we estimate with a gaussian fit , we extract a fwhm from a gaussian fit of the measured mtf .
these quantities are then related by @xmath49 , allowing us to compare the psf fwhm extracted from the mtf with that of the single mirror image .
for the siemens star with 532 nm light , we find a mtf fwhm of @xmath50 lp/@xmath25 m , corresponding to a psf fwhm of 630(10 ) nm , in agreement with the 650(50 ) nm measurement of the single mirror image . when imaging a bec in the siemens star pattern , we find a mtf fwhm of @xmath51 lp/@xmath25 m , corresponding to a psf fwhm of 1250(20 ) nm , larger than the 960(80 ) nm single mirror image at 780 nm illumination .
we believe that this @xmath52 increase is consistent with atom diffusion due to photon recoil during the 10 @xmath25s repump and resonant imaging pulse @xcite . , with the top row representing the optical pattern at 532 nm illumination , and the bottom an average atomic density in the pattern , imaged with resonant 780 nm light .
the green circles indicate radii of equally seperated spatial frequencies used to generate the corresponding mtf plots by calculating contrast along the circular path .
right : the fwhm of the optical pattern is @xmath50 lp/@xmath25 m , corresponding to a psf fwhm of 630(10 ) nm .
the atomic density mtf fwhm is @xmath51 lp/@xmath25 m , corresponding to a psf fwhm of 1250(20 ) nm . ]
the dmd is capable of switching mirrors from dc to the specified maximum frequency of 20 khz , which we have verified through photodiode measurement .
this wide modulation range , along with the ability to store 13,889 frames on the device , enables a high level of dynamic control . at low modulation frequencies
this allows adiabatic deformation of the dmd potentials . at the other extreme
, high frequency switching can be utilised for quickly quenching the potential geometry . furthermore
, the painted - potentials technique is possible , where an average dipole force is produced through rapid modulation of the optical field @xcite .
this suggests pulse - width - modulation ( pwm ) as an alternative technique to error diffusion for producing grayscale levels , analogous to the techniques used in dlp . .
( b ) images used for analysis of the gray levels .
the grayscale image was subtracted from the average atomic background with no barriers present .
the mean density and standard deviation were calculated over the circles as indicated .
( c ) gray levels achieved through this process .
a least squares regression line ( lsrl ) is indicated .
as the optical potential of point 6 exceeds the condensate chemical potential it was excluded from the fit .
image background and untrapped atoms result in an apparent maximum density suppression of 96% .
( d ) two example pulse width modulations , corresponding to patterns 1 , the minimum pulse , and 3 , a 50% duty cycle , offset vertically for clarity .
the envelope frequency @xmath53 khz and carrier frequency @xmath54 khz divisions are indicated .
( e ) turning on and off barrier 3 only , with varying frequency , for a total modulation time of 500 cycles provides an estimate of the heating rate from mirror switching .
rates above @xmath6 khz appear to have a negligible heating effect over this modulation period . ] to explore the suitability of this system for producing time averaged potentials , we produced the dmd pattern shown in fig .
[ fig_time_avg ] , consisting of an array of six 8 @xmath25 m diameter barriers contained in a 50 @xmath25 m @xmath29 80 @xmath25 m rectangle .
pwm , with the dmd running at an envelope frequency of @xmath53 khz , was utilised , and the leading edge of each pulse was fixed .
the hexagonal array within the rectangle was modulated with varying duty cycle over six levels by subdividing the carrier pulse into six modulation divisions , with the maximum duty cycle corresponding to all divisions turned on , and the minimum corresponding to one division turned on .
the carrier frequency thus used was @xmath54 khz .
we illuminated the dmd with light corresponding to a trap depth of 1.2@xmath25 .
the corresponding relative atomic densities are shown in fig .
[ fig_time_avg](c ) . to investigate the limits of this technique , we measured the effect on temperature and bec fraction due to modulation of a single barrier ( labeled `` 3 '' in fig .
[ fig_time_avg ] ) which had a 50% duty cycle . after first forming the bec in the rectangle in the absence of any internal barrier
, we performed a modulation for 500 cycles at fixed frequencies ranging from 100 hz to 16 khz .
we observed a decreasing effect as the frequencies are increased due to net energy gain per cycle decreasing since the barrier oscillates on a smaller time scale than atom diffusion timescales ; above @xmath6 khz the effect is negligible .
dmd devices combined with commercially - available glass - corrected objectives present a powerful technique for both the microscopic patterning of quantum gases and more general optical trapping . by utilising a commercial fluorescence cell with relatively thin 1.25 mm walls ,
expensive custom - built objectives can be avoided while still achieving high resolution .
we are able to estimate a 630(10 ) nm fwhm psf for our dmd projection system at 532 nm illumination , in agreement with the single - mirror analysis .
a similar estimate of the psf at 780 nm using a single mirror gives 960(80 ) nm fwhm , while mtf analysis of an atomic resolution target gives 1250(20 ) nm fwhm .
we believe the broadening is due to diffusion of the atoms during the 10 @xmath25s repump and resonant imaging pulses .
we note that off - resonant imaging techniques such as darkground or faraday imaging @xcite might allow atomic distributions to be imaged with the full resolution .
commercial dmd devices are now well - suited for implementation in both quantum gas experiments and other optical trapping applications .
the painted - potentials technique , in combination with error diffusion methods , can generate grayscale potentials with more than one hundred levels .
the ability to easily store and trigger 13,889 frames at high speed shows great promise for the dynamic control of these microscopic potentials .
furthermore , the ability to nearly instantaneously quench the potential geometry has many potential applications , such as investigations of superfluid transport between reservoirs @xcite , and a demonstration of the superfluid fountain effect in becs @xcite .
these techniques may prove useful in the production of superfluid turbulence in becs , where , for example , the production and control of vortex pairs using moving potential barriers has recently been demonstrated @xcite .
configurable trap geometries will be necessary for designing atomtronic circuits @xcite .
as the dmd is largely wavelength insensitive ( except in terms of diffraction efficiency ) , the possibilities for use at multiple wavelengths are intriguing .
this capability could be utilised for the production of species dependent potentials @xcite .
concurrent work by the munich group has shown the usefulness of high resolution direct imaging ( 600 nm fwhm ) of a dmd in creation of a two - dimensional disordered lattice for the exploration many body localization transitions @xcite .
more generally , our results demonstrate the utility of direct imaging of an dmd . with a well - corrected optical system , performance close to the diffraction limit
can still be achieved at a non - trivial 0.45 na .
the computational simplicity of direct imaging may prove an advantage in applications that require reconfiguration in real - time ; the high operating speed of the dmd furthermore facilitates this .
these results are also applicable to optical trapping beyond quantum gases and in particular may be advantageous for the production of numerous traps for confining arrays of particles @xcite .
australian research council ( arc ) ( ce1101013 , dp160102085 ) .
we would like to acknowledge the contributions of thomas carey to the initial technical development of the bec apparatus , and alexander stilgoe for useful discussions .
s. chowdhury , a. thakur , p. svec , c. wang , w. losert , and s. k. gupta , `` automated manipulation of biological cells using gripper formations controlled by optical tweezers , '' automation science and engineering , ieee transactions on * 11 * , 338347 ( 2014 ) .
f. nogrette , h. labuhn , s. ravets , d. barredo , l. bguin , a. vernier , t. lahaye , and a. browaeys , `` single - atom trapping in holographic 2d arrays of microtraps with arbitrary geometries , '' physical review x * 4 * , 021034 ( 2014 ) .
t. fukuhara , a. kantian , m. endres , m. cheneau , p. schau , s. hild , d. bellem , u. schollwck , t. giamarchi , c. gross , i. bloch , and s. kuhr , `` quantum dynamics of a mobile spin impurity , '' nature physics * 9 * , 235241 ( 2013 ) .
p. zupancic , p. m. preiss , r. ma , a. lukin , m. e. tai , m. rispoli , r. islam , and m. greiner , `` ultra - precise holographic beam shaping for microscopic quantum control , '' arxiv preprint arxiv:1604.07653 ( 2016 ) . n. m. parry , m. baker , t. neely , t. carey , t. bell , and h. rubinsztein - dunlop , `` note : high turn density magnetic coils with improved low pressure water cooling for use in atom optics , '' review of scientific instruments * 85 * , 086103 ( 2014 ) .
j. liang , r. n. kohn jr , m. f. becker , and d. j. heinzen , `` 1.5% root - mean - square flat - intensity laser beam formed using a binary - amplitude spatial light modulator , '' applied optics * 48 * , 19551962 ( 2009 ) .
a. kumar , n. anderson , w. phillips , s. eckel , g. campbell , and s. stringari , `` minimally destructive , doppler measurement of a quantized flow in a ring - shaped bose - einstein condensate , '' new journal of physics * 18 * , 025001 ( 2016 ) .
our results represent an improvement over previously reported work on optical trapping of becs using dmds .
we note that ref .
@xcite directly imaged a dmd with a higher resolution ( @xmath55 nm at 787.65 nm illumination ) for the purposes of optical addressing of atoms trapped in an optical lattice , but not for atom trapping .
b. zimmermann , t. mueller , j. meineke , t. esslinger , and h. moritz , `` high - resolution imaging of ultracold fermions in microscopically tailored optical potentials , '' new journal of physics * 13 * , 043007 ( 2011 ) .
k. henderson , c. ryu , c. maccormick , and m. boshier , `` experimental demonstration of painting arbitrary and dynamic potentials for bose - einstein condensates , '' new journal of physics * 11 * , 043030 ( 2009 ) .
t. a. bell , j. a. p. glidden , l. humbert , m. w. j. bromley , s. a. haine , m. j. davis , t. w. neely , m. a. baker , and h. rubinsztein - dunlop , `` bose - einstein condensation in large time - averaged optical ring potentials , '' new journal of physics * 18 * , 035003 ( 2016 ) .
lin , a. perry , r. compton , i. spielman , and j. porto , `` rapid production of @xmath5rb bose - einstein condensates in a combined magnetic and optical potential , '' physical review a * 79 * , 063631 ( 2009 ) .
m. pappa , p. condylis , g. konstantinidis , v. bolpasi , a. lazoudis , o. morizot , d. sahagun , m. baker , and w. von klitzing , `` ultra - sensitive atom imaging for matter - wave optics , '' new journal of physics * 13 * , 115012 ( 2011 ) .
l. corman , l. chomaz , t. bienaim , r. desbuquois , c. weitenberg , s. nascimbene , j. dalibard , and j. beugnon , `` quench - induced supercurrents in an annular bose gas , '' physical review letters * 113 * , 135302 ( 2014 ) .
l. mathey , a. ramantahn , k. c. wright , s. r. muniz , w. d. phillips , and w. charles , `` phase fluctuations in anisotropic bose - einstein condensates : from cigars to rings , '' physical review a * 82 * , 033607 ( 2010 ) .
w. muessel , h. strobel , m. joos , e. nicklas , i. stroescu , j. tomkovi , d. b. hume , and m. k. oberthaler , `` optimized absorption imaging of mesoscopic atomic clouds , '' applied physics b * 113 * , 6973 ( 2013 ) .
k. e. wilson , e. c. samson , z. l. newman , t. w. neely , and b. p. anderson , `` experimental methods for generating two - dimensional quantum turbulence in bose - einstein condensates , '' annual review of cold atoms and molecules * 1 * , 261 ( 2012 ) .
j. catani , g. barontini , g. lamporesi , f. rabatti , g. thalhammer , f. minardi , s. stringari , and m. inguscio , `` entropy exchange in a mixture of ultracold atoms , '' physical review letters * 103 * , 140401 ( 2009 ) .
choi , s. hild , j. zeiher , p. schau , a. rubio - abadal , t. yefsah , v. khemani , d. a. huse , i. bloch , and c. gross , `` exploring the many - body localization in two dimensions , '' science * 352 * , 1547 ( 2016 ) . | programable spatial light modulators ( slms ) have significantly advanced the configurable optical trapping of particles .
typically , these devices are utilized in the fourier plane of an optical system , but direct imaging of an amplitude pattern can potentially result in increased simplicity and computational speed . here
we demonstrate high - resolution direct imaging of a digital micromirror device ( dmd ) at high numerical apertures ( na ) , which we apply to the optical trapping of a bose - einstein condensate ( bec ) .
we utilise a ( @xmath0 ) pixel dmd and commercially available 0.45 na microscope objectives , finding that atoms confined in a hybrid optical / magnetic or all - optical potential can be patterned using repulsive blue - detuned ( 532 nm ) light with 630(10 ) nm full - width at half - maximum ( fwhm ) resolution , within 5% of the diffraction limit .
the result is near arbitrary control of the density the bec without the need for expensive custom optics .
we also introduce the technique of time - averaged dmd potentials , demonstrating the ability to produce multiple grayscale levels with minimal heating of the atomic cloud , by utilising the high switching speed ( 20 khz maximum ) of the dmd .
these techniques will enable the realization and control of diverse optical potentials for superfluid dynamics and atomtronics applications with quantum gases .
the performance of this system in a direct imaging configuration has wider application for optical trapping at non - trivial nas . |
the fossil record of star formation and galaxy evolution is imprinted on the spatial distribution , ages and metallicities of galactic stellar populations .
surprisingly little is known about the old and intermediate - age stellar populations in massive galaxies outside our own milky way .
located at @xmath1780 kpc , m31 provides our best opportunity to explore the stellar populations across the face of a large disk galaxy .
furthermore , its favourable inclination means that the disk and halo components at large radii can be easily distinguished and independently studied .
most previous studies of m31 have either been based on observations of single fields ( eg .
holland et al 1996 , durrell et al 2001 ) or on large - area surveys limited to the bright disk ( eg .
walterbos & kennicutt 1988 ) .
it is generally thought that m31 resembles the milky way in many ways , although it is somewhat larger , more luminous and has a denser stellar halo ( see freeman 1999 for a review ) .
the disparity between the mean metallicity of m31 s field halo and that of the milky way ( @xmath2 factor of 10 lower ) was first recognized by mould & kristian ( 1986 ) but remains to this day poorly understood .
we searched the hst archive for all deep ( t@xmath34000s ) wfpc2 pointings towards m31 in the f555w and f814w filters .
we focus here on those fields which sample the disk and halo at large radius ( r@xmath4kpc ) .
@xmath5 ) , 47 tuc ( [ fe / h]@xmath6 ) and ngc 6553 ( [ fe / h]@xmath7 )
. also indicated is the mean v magnitude of the blue horizontal branch detected in m31 halo by holland et al ( 1996).,title="fig : " ] @xmath5 ) , 47 tuc ( [ fe / h]@xmath6 ) and ngc 6553 ( [ fe / h]@xmath7 )
. also indicated is the mean v magnitude of the blue horizontal branch detected in m31 halo by holland et al ( 1996).,title="fig : " ] figure 1 shows the wfpc2 cmd of a field at @xmath8 kpc ( or 5 disk scale lengths ) along the major axis .
based on extrapolation of the structural parameters of walterbos & kennicutt ( 1988 ) , we expect 95% of the stars in this field to belong to the disk .
the prominence and morphology of the red giant branch ( rgb ) and red clump ( rc ) indicate an old - to - intermediate age , fairly metal - rich population .
comparison of the outer disk rgb with globular cluster fiducials indicates a mean metallicity comparable to 47 tuc ( [ fe / h]@xmath6 ) .
the significant width of the rgb exceeds that of photometric errors , and is most easily explained by an intrinsic dispersion ( @xmath9 dex ) in the metallicity of the stellar population . while these properties have been noted before for the m31 halo ( eg .
holland et al 1996 ) , we find here that they also characterise the outer disk .
see ferguson & johnson ( 2001 ) for details .
the rc properties , when coupled with the mean metallicity from the rgb colour , suggest a mean age for the population of @xmath10 gyr ( cole 1999 , girardi & salaris 2001 ) .
this is also consistent with the apparent lack of asymptotic giant branch stars above the tip of the rgb ( [email protected] ) , which would represent young - to - intermediate age ( 26 gyr ) shell he - burning stars .
finally , we note the marginal detection of horizontal feature in the cmd connecting the blue plume at v@xmath11 to the rc . while this could represent the subgiant branch of a @xmath12 gyr population , it is more likely to be an old ( @xmath13 gyr ) , metal - poor ( [ fe / h]@xmath14 ) horizontal branch ( see figure 1 ) .
figure 2 shows cmds for a representative halo - dominated field and a field in which disk and halo are expected to contribute in roughly equal amounts .
apart from stellar density ( which reflects both intrinsic variations as well how much of the wf area was useable in our analysis ) , the dominant stellar population in these fields appears strikingly similar ( and indeed , both resemble the outer disk cmd in figure 1 ) .
the homogeneity of halo and outer disk populations was first remarked upon by morris et al ( 1994 ) from shallow ground - based data , but we show here the similarity holds to well below the horizontal branch .
while providing detailed information on the stellar populations , the hst pointings tell us nothing about the large - scale structure of m31 ( wfpc2 fov @xmath15 kpc@xmath16 ) .
we have therefore embarked upon a ground - based , moderate - depth panoramic imaging survey of the outskirts of m31 with the int wide field camera . to date , 58 contiguous fields ( 0.3 @xmath17 per pointing ) have been observed in the se half of the galaxy , mapping out to a projected radius of @xmath0 kpc .
an exciting first result has been the discovery of a giant stream of stars ( overdensity a factor of 2 ) in the halo of m31 near the southern minor axis ( ibata et al 2001 ) .
the stream extends beyond the limit of our imagery , or @xmath18 kpc at this position angle , and has an average v - band surface brightness of @xmath19 .
if an old coeval population is assumed , the excess stream population is found to be of similar or slightly higher metallicity than the m31 field halo and outer disk .
the stream appears to lie along a line which connects m32 and ngc 205 , the two dwarf elliptical companions of m31 .
both these satellites display rather odd properties for their morphological class including young and/or intermediate - age stars and in the case of ngc 205 , cold gas and both exhibit distorted isophotes in the outer regions , suggestive of tidal distortion and possibly disruption .
the similarity between the metallicities of these dwarfs and that of the stream suggests that one or both of them may be the origin of the feature .
we are exploring the stellar populations in the outskirts of m31 using deep hst archival pointings in combination with a ground - based panoramic imaging survey .
we summarise our first results as follows : + @xmath20 the disk of m31 appears to have a significant mean age ( @xmath21 gyr ) . for current cosmologies
, this lookback time corresponds to a redshift of @xmath22 by which about half the stellar disk was in place at 30 kpc .
this may be problematic for theories which invoke delayed disk formation as a way to circumvent the angular momentum problem seen in numerical simulations of galaxy formation ( eg .
weil et al 1998 ) .
equally puzzling is our finding that the dominant populations in the outer disk and halo are remarkably similar , suggesting these components have experienced similar formation epochs and evolutionary histories .
+ @xmath20 the discovery of a giant stellar stream in the outer halo of m31 attests to the fact that the hierarchical process of galaxy formation continues to the present - day and that at least some fraction of the m31 field halo has not formed _ in - situ _ , but has been accreted from lower mass objects .
could this interaction have polluted the outer regions of m31 to such an extent that it explains that apparent uniformity of the stellar populations in these parts ?
future observations will help us assess this possibility more throughly .
i gratefully acknowledge the input of my collaborators on the work discussed here specifically , rachel johnson and nial tanvir on the hst studies and mike irwin , rodrigo ibata , geraint lewis and nial tanvir on the int / wfc survey .
thanks also to the leids kerkhoven - bosscha fonds for financial assistance to attend this conference . | we discuss the first results from ongoing studies of the resolved stellar populations in the outskirts of our nearest large neighbour , m31 .
deep hst / wfpc2 archival observations are used to construct colour - magnitude - diagrams which reach well below the horizontal branch at selected locations in the outer disk and halo , while a panoramic ground - based imaging survey maps spatial density variations through resolved star counts to a projected radius of @xmath0kpc . |
recently @xcite we have reported the @xmath1-evolution of the non - singlet structure functions at low and high @xmath0 as the complete solution of taylor - approximated dglap equation @xcite .
the present paper reports the prediction of corresponding @xmath0-evolution .
although the conventional wisdom on dglap evolution does not favour such a feature , such a possibility was reported @xcite sometime back .
the main reason is that in this approach , the dglap equation is converted into a first order differential equation in two variables @xmath0 and @xmath1 ( @xmath2 ) instead of only @xmath1 and has @xmath0 evolution as its natural outcome .
there are , of course , various methods that have been developed recently like the brute - force @xcite , laguerre - polynomial @xcite and mellin - transformation @xcite etc . for numerical solution of dglap equation .
all these methods have good numerical accuracy and computational efficiency , particularly the mellin - transformation one .
all these solutions are very useful in the studies of a top - down scenario in order to determine the origin of ultra - high energy cosmic rays .
in addition , the matrix technique to get the numerical solution of the dglap evolution equation has also been proposed @xcite , which provides very good precision . in recent years ,
the study of parton distribution function ( pdf ) has itself become a topic of significant interest .
the standard approach to the problem is based on a choice of a specific functional form in @xmath0 with a few parameters and to obtain fit of them by comparing the exact solution of the evolution equation with global data .
based on this approach , several sets of pdf have been produced in recent years @xcite .
more recently , however , shortcoming of such an approach has been pointed out : the form of parametrization is itself a source of bias . to overcome this limitation ,
numerical method incorporating the notion of neural network @xcite has been suggested and pursued .
alternative conventional models like statistical @xcite and chiral quark model @xcite have also received cosiderable attention in this respect .
the present paper reports the analytical @xmath0-evolution of the non - singlet structure function in the formulation of ref .
@xcite which is much more general than that of earlier works @xcite and compares with the other approaches .
since such an analytical solution is in good agreement with experimental data , hence this type of study bears important significance . in sec [ secchap4:formalism ] , we develop the essential formalism while in sec [ secchap4:results ] , we will discuss the results .
[ secchap4:comments and conclusions ] will summarize a conclusion .
the dglap evolution equation for non - singlet structure function in standard form is given by @xcite @xmath3 where @xmath0 is the usual bjorken variable @xmath4 and @xmath5 is the four momentum transfer in deep inelastic scattering @xcite process .
the symbol @xmath6 stands for the usual mellin convolution in the first variable defined as @xmath7 the non - singlet kernel @xmath8 has perturbative expansion in orders of the strong coupling constant @xmath9 as @xmath10 where @xmath11 and @xmath12 are the one loop @xcite and two loop @xcite contributions to the non - singlet structure function respectively . in case of lo
, we neglect two loop contribution and so eq . ( [ eqn : ch4eqa ] ) can be rewritten as @xmath13 since the evolution of the non - singlet structure function is independent of the gluon , hence one can write from earlier equation , @xmath14 where we have @xcite @xmath15}\ ] ] where @xmath16 is given by @xmath17 and @xmath18 is defined by @xmath19 here @xmath20 is an arbitrary function and @xmath21 is the number of colours . using this explicit form of the splitting function @xmath22 , eq .
( [ eqn : ch4eqf ] ) can be written as @xcite @xmath23\end{aligned}\ ] ] here @xmath24 and the running coupling constant in leading order ( lo ) is @xmath25 with @xmath26 , @xmath27 being the number of quark flavours .
this expression can be re - written as @xmath28\end{aligned}\ ] ] here @xmath29 and @xmath30 . for small @xmath0 @xmath31
, it is justified to neglect higher derivatives of @xmath32 so that @xmath33 occured in the r.h.s .
( [ eqn : ch4eq1 ] ) can be written as @xmath34 where @xmath35 putting eq .
( [ eqn : ch4eq2 ] ) in eq .
( [ eqn : ch4eq1 ] ) and performing the u - integration yields the taylor - approximated dglap equation as @xmath36 where @xmath37 , @xmath38 and @xmath39 are explicitly mentioned in of ref . @xcite as @xmath40}{t}=\frac{p_1(x)}{t}}\ ] ] @xmath41 and @xmath42}{t}=\frac{r_1(x)}{t}}\ ] ] eq .
( [ eqn : ch4eq3 ] ) is frequently referred to as lagrange s equation @xcite .
its general solution is obtained by solving the following auxiliary systems of ordinary differential equations @xmath43 which subsequently gives @xmath44 where @xmath45 , @xmath46 and @xmath47 .
the most general form of the solution of the taylor - approximated dglap equation ( eq .
( [ eqn : ch4eq3 ] ) ) is @xcite , @xmath48 where @xmath49 is any real function of @xmath0 and @xmath1 and @xmath50,@xmath51 are two arbitrary constants . @xmath52 and @xmath53 are two explicit functions of @xmath0 and @xmath1 given by @xcite , @xmath54 and @xmath55 which are two independant solutions of eq .
( [ eqn : ch4eq3e ] ) . in order to develop our formalism , we take the numerical value of @xmath56 and use the following two physically plausible boundary conditions @xcite @xmath57 for some low @xmath58 and @xmath59 putting the boundary conditions ( [ eqn : ch4eq5 ] ) and ( [ eqn : ch4eq6 ] ) in eq . ( [ eqn : ch4eq4 ] ) , along with expressions for @xmath60 and @xmath61 from eq .
( [ eqn : ch4eq4a ] ) and eq . ( [ eqn : ch4eq4b ] ) respectively , one gets @xmath62^{n}-[x^{ns}(1)]^{n}\}}\ ] ] defining @xmath63^{n}-[x^{ns}(1)]^{n}\}}\ ] ] where @xmath64 and @xmath65 are given as @xcite @xmath66}\ ] ] and @xmath67}\ ] ] this results in @xmath68^{n}-[x^{ns}(1)]^{n}}{[x^{ns}(x_0)]^{n}-[x^{ns}(1)]^{n}}}\ ] ] as we have @xcite @xmath69 hence eq .
( [ eqn : ch4eq9 ] ) yields for @xmath70 , @xmath71^{n}}{[x^{ns}(x_0)]^{n}}}\ ] ] on the other hand for @xmath72 , the r.h.s of eq . ( [ eqn : ch4eq9 ] ) will involve inverse of @xmath73 which is singular and ill - defined , and hence excluded on physical grounds .
putting the values of @xmath64 and @xmath65 , we get @xmath0 distribution obtained from complete solution at low @xmath0 , @xmath74}.\;\;\;\ ; ( n>0)\ ] ] where @xmath75 and @xmath76 are defined as @xcite @xmath77}\ ] ] @xmath78}\ ] ] eq .
( [ eqn : ch4eq12 ] ) is also true for quark of each flavour individually i.e. @xmath79},\,\,\ , ( n>0)\ ] ] combining @xmath1-evolution for quark @xcite at low @xmath0 , viz , @xmath80 we obtain @xmath81},\,\,\ , ( n>0)\ ] ] containing @xmath1-evolution as well .
( [ eqn : ch4eq12 ] ) and eq .
( [ eqn : ch4eq16 ] ) are more general than the similar @xmath0 distribution reported recently @xcite . for high @xmath0 ,
the most general form of the solution of the taylor - approximated dglap equation is @xcite @xmath82 where @xmath83 is another function of @xmath0 and @xmath1 similar to @xmath49 in eq .
( [ eqn : ch4eq4 ] ) . as in case of low @xmath0 , we assume @xmath84 so as to yield @xmath85^{p}-[x^{{\prime}{ns}}(1)]^{p}}{[x^{{\prime}^{ns}}(x_0)]^{p}-[x^{{\prime}{ns}}(1)]^{p}}}\ ] ] where @xmath86 and @xmath87 are defined as @xcite @xmath88}\ ] ] and @xmath89}\ ] ] as @xmath90 is singular @xcite , for @xmath91 , r.h.s of eq .
( [ eqn : ch4eq18 ] ) is ill - defined and excluded on physical grounds . if on the other hand , for @xmath92 , r.h.s of eq .
( [ eqn : ch4eq18 ] ) involves @xmath93^{-1}$ ] , which vanishes identically . under this condition ( @xmath92 ) , @xmath94^{p}}{[x^{{\prime}^{ns}}(x_0)]^{p}}}\ ] ] putting the values of @xmath86 and @xmath87 , we finally get the @xmath0-distribution obtained from complete solution at high @xmath0 to be @xmath95}.\;\;\;\ ; ( p<0)\ ] ] where @xmath96 eq . ( [ eqn : ch4eq20 ] ) is also true for each quark flavour individually .
@xmath97}.\;\;\;\ ; ( p<0)\ ] ] combining @xmath1-evolution for quark @xcite at high @xmath0 , viz , @xmath98 we obtain @xmath99}.\;\;\;\ ; ( p<0)\ ] ] containing @xmath1-evolution as well . in subsection
[ subsecchap4:soln at low x ] and subsection [ subsecchap4:soln . at high x
] , we considered the numerical values of @xmath49 and @xmath83 that differentiated the low @xmath0 from the high @xmath0 regimes in case of @xmath0-evolution .
now we attempt to find out the general and complete analytical solution of dglap equation using the explicit functional forms of @xmath49 and @xmath83 and their @xmath100 and @xmath101 limits at low and high @xmath0 respectively .
now in case of low @xmath0 , considering the functional form of @xmath49 and proceeding as in subsection [ subsecchap4:soln at low x ] , we get @xmath102^{n(x , t)}-[tx^{ns}(1)]^{n(1,t)}\}}\ ] ] defining @xmath103^{n(x_0,t)}-[tx^{ns}(1)]^{n(1,t)}\}}\ ] ] we get , @xmath104^{n(x , t)}-[tx^{ns}(1)]^{n(1,t)}}{[tx^{ns}(x_0)]^{n(x_0,t)}-[tx^{ns}(1)]^{n(1,t)}}}.\ ] ] now proceeding as in subsection [ subsecchap4:soln at low x ] , we ultimately get the general @xmath0-distribution at low @xmath0 , @xmath105^{n(x , t)}}{[x^{ns}(x_0)]^{n(x_0,t)}}}.\;\;\;\ ; ( n(1,t)>0)\ ] ] instead of eq .
( [ eqn : ch4eq11 ] ) . similarly , considering the functional form of @xmath83 and proceeding as in subsection [ subsecchap4:soln . at high x ] , we ultimately get , @xmath106^{p(x , t)}}{[x^{{\prime}^{ns}}(x_0)]^{p(x_0,t)}}}.\;\;\;\ ; ( p(1,t)<0)\ ] ] instead of eq .
( [ eqn : ch4eq19 ] ) .
( [ eqn : ch4eq24 ] ) and eq .
( [ eqn : ch4eq25 ] ) can be re - expressed as @xmath107\frac{t^{n(x , t)}}{t^{n(x_0,t ) } } & & \nonumber \\
\times \frac{\exp\left[-n(x , t)\displaystyle{\int\frac{1}{p_1(x)}dx}\right]}{\exp\left[-n(x , t)\displaystyle{\int\frac{1}{p_1(x)}dx}\right]}_{x={x_0}}.\nonumber\\ \nonumber\\ ( n(1,t)>0)\end{aligned}\ ] ] and @xmath108\frac{t^{p(x , t)}}{t^{p(x_0,t ) } } & & \nonumber \\ \times \frac{\exp\left[-p(x , t)\displaystyle{\int\frac{1}{p_1^{\prime}(x)}dx}\right]}{{\exp\left[-p(x , t)\displaystyle{\int\frac{1}{p_1^{\prime}(x)}dx}\right]}_{x = x_0}}.\nonumber\\ \nonumber\\ ( p(1,t)<0)\end{aligned}\ ] ] we now put forward the generalized statement as : * the most general solution of the taylor - approximated dglap equation at low and high @xmath0 are obtainable in terms of two arbitrary real functions @xmath49 and @xmath83 such that @xmath109 is greater than zero and @xmath110 is less than zero for any @xmath1 . in case , @xmath49 and @xmath83 have only mild variation in @xmath1 , then @xmath111 and @xmath112 can be considered as the @xmath113 and @xmath114 limit of an interpolating function @xmath115 to be determined empirically from data . * alternatively , we can also determine @xmath0-evolution of @xmath116 if @xmath51 and @xmath50 of eq .
( [ eqn : ch4eq4 ] ) satisfies additional relation @xmath117 @xcite , where @xmath118 is a positive integer . in this case , for low @xmath0 , we get from eq .
( [ eqn : ch4eq4 ] ) , @xmath119^n+\alpha^{m}}\ ] ] where eq .
( [ eqn : ch4eq4a ] ) and eq .
( [ eqn : ch4eq4b ] ) have been used .
differentiating eq .
( [ eqn : ch4eq28 ] ) with respect to @xmath51 , we get @xmath120^{\frac{mn}{m-1}}}\ ] ] which forbids m=1 .
putting this value of @xmath51 in eq .
( [ eqn : ch4eq28 ] ) , we ultimately get @xmath121^{\frac{mn}{m-1}}}\ ] ] @xmath122^{\frac{mn}{m-1}}}\ ] ] we get , @xmath123^{\frac{mn}{m-1}}}{[x^{ns}(x_0)]^{\frac{mn}{m-1}}}}\ ] ] if @xmath118 is very large , @xmath124 , then eq . ( [ eqn : ch4eq32 ] ) reduces to eq .
( [ eqn : ch4eq11 ] ) and equivalently eq . ( [ eqn : ch4eq12 ] ) .
however , unlike in the previous case , there is no additional condition of @xmath70 .
similarly in case of high @xmath0 too , we get the @xmath0-distribution from the alternative way , as in low @xmath0 , similar to eq .
( [ eqn : ch4eq19 ] ) and eq .
( [ eqn : ch4eq20 ] ) , but without the additional condition on @xmath125 i.e @xmath92 .
thus unlike in the general case , the alternate solution can not distinguish between low @xmath0 and high @xmath0 regimes , which is its limitation .
we obtained an interpolating function @xmath115 from ccfr neutrino data @xcite viz .
, @xmath126}\ ] ] such that for @xmath127 and @xmath100 , it has desired behaviour of @xmath49 and @xmath83 respectively , yielding @xmath128 and @xmath129 assuming these limits as process independent , we use them to study @xmath0-distribution in figure [ fig : ch4fig1](a - e ) and figure [ fig : ch4fig2](a - e ) for low and high @xmath0 respectively using the following equations that have been redefined from eq .
( [ eqn : ch4eq11 ] ) and eq .
( [ eqn : ch4eq19 ] ) respectively as @xmath130^{h(x , t)}}{[x^{ns}(x_0)]^{h(x_0,t)}}}}\ ] ] @xmath131^{h(x , t)}}{[x^{{\prime}^{ns}}(x_0)]^{h(x_0,t)}}}}\ ] ] we compare our results with ccfr data @xcite as well as grv(98 ) @xcite and mrst lo 2001 @xcite exact results . for getting the exact results , we take the non - singlet structure function in leading order as @xcite , @xmath132 + 2xs[|v_{us}|^2+|v_{cs}|^2]-2xc}.\ ] ] where @xmath133 etc . and @xmath134 , @xmath135 etc .
are the relevant ckm matrix elements @xcite .
we have taken @xmath136 , @xmath137 , @xmath138 and @xmath139 . for low @xmath0 ( figure [ fig : ch4fig1](a - e ) ) , our preditions are found to be tallying very well ( within @xmath140 limit ) with both data and exact results in the lower part of the @xmath0 spectrum , but it does not match so well in the upper part of the @xmath0 spectrum . for high @xmath0 ( figure [ fig : ch4fig2](a - e ) ) , on the other hand , our predictions are closer to both data and exact results ( within @xmath141 limit ) in the extreme lower part of the @xmath0 spectrum . in the middle part of the @xmath0 spectrum ,
coincidence between our predictions and both data as well as exact results is not so well as expected . for extreme higher parts of the @xmath0 spectrum ,
there is excellent coincidence between our predictions and both data and exact results ( specially in the region @xmath142 ) .
the deviations of our predictions from data and exact results is presumably due to the neglect of finite @xmath0 corrections coming from the higher derivatives of @xmath32 in the taylor approximation ( neglected in eq .
( 3 ) and eq.(35 ) of ref.@xcite ) .
it may also be partly due to the neglect of explicit @xmath0 and @xmath1 dependence of @xmath49 and @xmath83 in developing the seperate formalism of low and high @xmath0 .
we redefine eq .
( [ eqn : ch4eq24 ] ) and eq .
( [ eqn : ch4eq25 ] ) as @xmath143^{h(x , t)}}{[x^{ns}(x_0)]^{h(x_0,t)}}}.\ ] ] @xmath144^{h(x , t)}}{[x^{{\prime}^{ns}}(x_0)]^{h(x_0,t)}}}.\ ] ] which are valid for low @xmath0 and high @xmath0 regions respectively . on the basis of the above equations , we study the @xmath0-distribution in figure [ fig : ch4fig1](a - e ) and figure [ fig : ch4fig2](a - e ) for low @xmath0 and high @xmath0 respectively .
the dash line in the above set of figures represents the result obtained on the basis of general solution .
+ + versus @xmath0 in bins of @xmath5 obtained from eq .
( [ eqn : ch4eq34 ] ) ( solid line ) and eq .
( [ eqn : ch4eq36 ] ) ( dashed curve ) for low @xmath0 values .
this is compared with the grv98 @xcite ( dotted curve ) and mrst2001 @xcite ( dashed dotted curve ) exact results calculated using eq .
( [ eqn : ch4eq35 ] ) .
this is also compared with ccfr data @xcite.,width=566,height=453 ] + + versus @xmath0 in bins of @xmath5 obtained from eq .
( [ eqn : ch4eq34a ] ) ( solid line ) and eq .
( [ eqn : ch4eq37 ] ) ( dashed curve ) for high @xmath0 values .
this is compared with the grv98 @xcite ( dotted curve ) and mrst2001 @xcite ( dashed dotted curve ) exact results calculated using eq .
( [ eqn : ch4eq35 ] ) .
this is also compared with ccfr data @xcite.,width=566,height=453 ] for low @xmath0 , our predictions based on the particular and the general solution does not differ appreciably in the lower part of @xmath0-spectrum , but the general solution is found to match better with both data and exact results in the upper part of @xmath0 spectrum . on the other hand , for high @xmath0 , the general solution matches better than particular solution with both data and exact results in the lower part of @xmath0 spectrum , but it is found that the opposite is true for higher part of @xmath0-spectrum .
thus we find that acceptance of explicit @xmath0 and @xmath1-dependence of @xmath49 and @xmath83 gives better result in the higher @xmath0 spectrum of low @xmath0 and lower @xmath0 spectrum of high @xmath0 .
versus @xmath0 for fixed @xmath145 @xmath146 simulated by an empirical multiplicative damping factor @xmath147 in the structure function ( general solution ) in eq .
( [ eqn : ch4eq36 ] ) for low @xmath0 values .
this is compared with the mrst2001 @xcite ( dashed dotted curve ) exact results calculated using eq .
( [ eqn : ch4eq35 ] ) as well as ccfr data @xcite .,width=566,height=453 ] versus @xmath0 for fixed @xmath145 @xmath146 simulated by an empirical multiplicative damping factor @xmath147 in the structure function ( particular solution ) in eq .
( [ eqn : ch4eq34 ] ) for low @xmath0 values .
this is compared with the mrst2001 @xcite ( dashed dotted curve ) exact results calculated using eq .
( [ eqn : ch4eq35 ] ) as well as ccfr data @xcite .,width=566,height=453 ] for low @xmath0 , we find in subsection [ subsecchap4 : general solution result . ] that predictions based on the most general solution have shoot up the ccfr neutrino data as well as exact results in the higher part of @xmath0 -spectrum .
this overshooting can be substantially reduced by the introduction of large @xmath0 effect .
such finite @xmath0 corrections , which may be considered as the conglomeration of the terms neglected while getting eq .
( [ eqn : ch4eq2 ] ) , can effectively be simulated by an empirical multiplicative damping factor @xmath147 , ( @xmath148 is positive ) in the structure function ( general solution ) in eq .
( [ eqn : ch4eq36 ] ) .
we have varied @xmath148 from 4 to 9 and found @xmath148 around 7 to yield the best fit to the data and exact result .
this is clearly shown in figure [ fig : ch4fig3 ] , where we have plotted this modified structure function at different @xmath0 for fixed @xmath145 @xmath146 . similarly , we multiply the structure function ( particular solution ) in eq .
( [ eqn : ch4eq34 ] ) by multiplicative damping factor @xmath147 and plot this modified structure function versus @xmath0 for fixed @xmath145 @xmath146 in figure [ fig : ch4fig4 ] .
this time we find the value of @xmath148 around 8 gives the best fit to the data , clearly pinpointing the supremacy of the general solution over the particular solution .
it is pertinent to note that the modified structure function damps considerably in the higher part of @xmath0 -spectrum ( of low @xmath0 ) only due to the damping factor . in the lower part of the @xmath0 spectrum ( of low @xmath0 )
, there is very little damping because @xmath149 when @xmath113 .
let us now discuss how the present work differs from that of ref.@xcite , @xcite and @xcite . in ref.@xcite ,
taylor - approximated leading order coupled dglap equation for singlet ( @xmath150 ) and gluon distribution ( @xmath151 ) were solved using the relationship @xmath152 and the particular solution @xmath153 ( @xmath51 and @xmath50 are constants ) implying @xmath154 in the general solution of the present formalism ( eq .
( [ eqn : ch4eq4 ] ) ) .
the results are then compared with deuteron data using the relation @xmath155 . in ref.@xcite
, a similar analysis was done for deuteron structure function using the additional relation @xmath156 . in ref.@xcite ,
the analysis was further extended to next - to - leading order for deuteron , proton and neutron structure function assuming the relationship @xmath157 implying that @xmath0 evolution of the formalism is not unique . in the present work ,
we have made the formalism much more general abandoning the linearity of @xmath52 but still preserving that of @xmath53 ( which is equivalent to structure function ) in the general solution .
to conclude , we have shown that the @xmath0-distribution of the non - singlet structure function obtained from leading order taylor - approximated dglap equations are compatible with data @xcite as well as exact results for low and high @xmath0 @xcite in selected parts of the @xmath0 spectrum .
we have also shown how the introduction of damping factor considerably brings the modified structure function closer to exact resuts in case of low @xmath0 .
we have also discussed how the present formalism is an improvement and generalization over those of ref.@xcite . | study of parton distribution function ( pdf ) is a topic of significant interest in qcd . to overcome the shortcomings of conventional pdfs
, several alternative methods have been suggested in recent years .
the present work reports the @xmath0-distribution of the non - singlet structure function using the complete solution of taylor - approximated dglap equation .
the formalism is more general than the similar ones reported in recent literature .
our predictions are compared with the ccfr neutrino data as well as exact results . * keywords * : non - singlet structure function , @xmath0-evolution , leading order , low and high @xmath0 * pacs nos . * : 12.38.-t,12.38.bx,13.60-hb . |
it is fairly well established that quantum electrodynamics ( qed ) , and in particular quenched qed , breaks chiral symmetry for sufficiently large couplings . this phenomenon has been observed both in lattice simulations @xcite as well as various studies based on the use of dyson - schwinger equations @xcite .
these latter calculations have generally relied on the use of a cut - off in euclidean momentum in order to regulate divergent integrals , a procedure which breaks the gauge invariance of the theory . on the other hand , continuation of gauge theories to @xmath0 dimensions
has long been used as an efficient way to regularize perturbation theory without violating gauge invariance . in nonperturbative calculations , however , the use of this method of regularization is rarely used @xcite . within the context of the dyson - schwinger equations ( dses ) only a few publications @xcite
have employed dimensional regularization instead of the usual momentum cut - off .
it is the purpose of the present paper to study dynamical chiral symmetry breaking and the chiral limit within dimensionally regularized quenched qed . we are motivated to do this by the wish to avoid some gauge ambiguities occurring in cut - off based work , which we discuss in sec .
[ sec_motivation ] . in that section
we also outline some general results which one expects to be valid for @xmath0 , independently of the particular vertex which one uses as an input to the dses .
having done this we proceed , in section [ sec : rainbow ] , with a study of chiral symmetry breaking in the popular , but gauge non - covariant , rainbow approximation .
just as in cut - off regularized work , the rainbow approximation provides a very good qualitative guide to what to expect for more realistic vertices and has the considerable advantage that , with certain additional approximations , one may obtain analytical results .
we check numerically that the additional approximations made are in fact quite justified .
indeed , it is very fortunate that it is possible to obtain this analytic insight into the pattern of chiral symmetry breaking in @xmath3 dimensions as it provides us with a well defined procedure for extracting the critical coupling of the 4 dimensional theory with more complicated vertices .
we proceed to the curtis - pennington ( cp ) vertex in section [ sec : cp ] . there
we derive , for solutions which do not break chiral symmetry , an integral representation for the exact wavefunction renormalization function @xmath4 in @xmath3 dimensions .
we also provide an approximate , but explicit , expression for this quantity .
the latter is quite useful , in the ultraviolet region , even if dynamical chiral symmetry breaking takes place as it provides a welcome check for the numerical investigation of chiral symmetry breaking with the cp vertex with which we conclude that section .
finally , in section [ sec : conclusion ] , we summarize our results and conclude .
although chiral symmetry breaking appears to be universally observed independently of the precise nature of the vertex used in dse studies , it has also been recognized for a long time that the critical couplings with almost all of these vertices show a gauge dependence which should not be present for a physical quantity . with a bare vertex
this is not surprising as this vertex ansatz breaks the ward - takahashi identity .
however , even with the curtis - pennington ( cp ) vertex , which does not violate this identity and additionally is constrained by the requirement of perturbative multiplicative renormalizability , a residual gauge dependence remains @xcite .
apart from possible deficiencies of the vertex , which we do not investigate in this paper , the use of cut - off regularization explicitly breaks the gauge symmetry even as the cut - off is taken to infinity .
this is well known in perturbation theory ( see , for example , the discussion of the axial anomaly in sect .
19.2 of ref .
@xcite ) and was pointed out by roberts and collaborators @xcite in the present context .
the latter authors proposed a prescription for dealing with this ambiguity which ensures that the regularization does not violate the ward - takahashi identity . as may be observed in fig .
[ fig : cut - off alpha_cr ] , this ambiguity has a strong effect on the value of the critical coupling of the theory .
the two curves in that figure correspond to the critical coupling @xmath5 of ref .
@xcite as well as the coupling @xmath6 one obtains by following the prescription of roberts et al .
it is straightforward to show , following the analysis of ref .
@xcite , that these couplings are related through @xmath7 also plotted in this figure are previously published numerical results @xcite obtained with both of the above prescriptions .
note that , curiously , the critical couplings obtained with the prescription of ref .
@xcite ( i.e. the calculation which restores the ward - takahashi identity ) exhibits a stronger gauge dependence , at least for the range of gauge parameters shown in fig . [
fig : cut - off alpha_cr ] .
gauge ambiguities such as the one outlined above are absent if one does not break the gauge invariance of the theory through the regularization procedure .
hence , we now turn to dimensionally regularized ( quenched ) qed .
the minkowski space fermion propagator @xmath8 is defined in the usual way through the dimensionless wavefunction renormalization function @xmath9 and the dimensionful mass function @xmath10 , i.e. @xmath11 the dependence of @xmath4 and @xmath12 on the dimensionality of the space is not explicitly indicated here .
furthermore , note that to a large extent we shall be dealing only with the regularized theory without imposing a renormalization procedure , as renormalization @xcite is inessential to our discussion .
in addition to the above , we shall consider the theory without explicit chiral symmetry breaking ( i.e. zero bare mass ) .
this theory would not contain a mass scale were it not for the usual arbitrary scale ( which we denote by @xmath13 ) introduced in @xmath14 dimensions which provides the connection between the _ dimensionful _ coupling @xmath15 and the usual _ dimensionless _ coupling constant @xmath16 : @xmath17 as @xmath13 is _ the only _ mass scale in the problem , and as the coupling always appears in the above combination with this scale , on dimensional grounds alone the mass function must be of the form @xmath18 where @xmath19 is a dimensionless function and in particular @xmath20 moreover , as @xmath21 goes to zero the @xmath13 dependence on the right hand side must disappear and hence the dynamical mass @xmath22 is either zero ( i.e. no symmetry breaking ) or goes to infinity in this limit . this situation is analogous to what happens in cut - off regularized theory , where the scale parameter is the cut - off itself and the mass is proportional it .
note that @xmath23 is not dependent on @xmath24 .
this implies immediately that there can be no non - zero critical coupling in @xmath25 dimensions : if @xmath22 is non - zero for some coupling @xmath24 then it must be non - zero for all couplings . given these general considerations ( which are of course independent of the particular ansatz for the vertex ) it behooves one to ask how this situation can be reconciled with a critical coupling @xmath5 of order 1 in four dimensions .
in order to see how this might arise , we shall extract a convenient numerical factor out of @xmath19 and suggestively re - write the dynamical mass as @xmath26 at present there is no difference in content between eq .
( [ eq : m(0 ) general form ] ) and eq .
( [ eq : m(0 ) general form 2 ] ) .
however , if we now _ define _ @xmath5 by demanding that the behaviour of @xmath22 is dominated by the factor @xmath27 as @xmath21 goes to zero , which is equivalent to demanding that @xmath28^\epsilon \
> \longrightarrow_{_{_{_{\hspace{-7 mm } { { \epsilon\rightarrow } { 0 } } } } } } % \stackrel{\epsilon \rightarrow 0 } % { \longrightarrow } \>\ > 1 \;\;\ ; , \label{eq : m_overbar def}\ ] ] then the intent becomes clear : even though @xmath22 may be nonzero for all couplings in @xmath0 dimensions , in the limit that @xmath29 goes to zero we obtain @xmath30 note that in the above we have not addressed the issue of whether or not there actually is a @xmath31 with the property of eq .
( [ eq : m_overbar def ] ) .
in fact , the numerical and analytical work in the following sections is largely concerned with finding this function and hence determining whether or not chiral symmetry is indeed broken for @xmath0 . nor @xmath31 are functions of the coupling @xmath24 , the value of @xmath5 can be determined independently of the strength @xmath24 of the self - interactions in @xmath32 dimensions . ]
notwithstanding this , as one knows from cut - off based work that there actually is a non - zero critical coupling for @xmath33 , one can at this stage already come to the conclusion that @xmath31 exists and hence that quenched qed in @xmath0 dimensions has a chiral symmetry breaking solution for all couplings . in summary ,
as the trivial solution @xmath34 always exists as well , we see that in @xmath32 dimensions the trivial and symmetry breaking solutions bifurcate at @xmath35 while for @xmath33 the point of bifurcation is at @xmath36 ; i.e. , there is a discontinuous change in the point of bifurcation . as @xmath3 approaches four ( i.e. as @xmath21 approaches @xmath37 ) the generated mass @xmath22 decreases ( grows ) roughly like @xmath38 for @xmath24 @xmath39 @xmath5 ( @xmath40 @xmath5 ) , respectively , becoming an infinite step function at @xmath36 when @xmath21 goes to zero
let us now consider an explicit vertex . to begin with , we consider the rainbow approximation to the euclidean mass function of quenched qed with zero bare mass in landau gauge .
it is given by @xmath41 note that the dirac part of the self - energy is equal to zero in the landau gauge in rainbow approximation even in @xmath0 dimensions and hence that @xmath42 for all @xmath43 .
it is of course possible to find the solution to eq .
( [ masseq ] ) numerically indeed we shall do so however it is far more instructive to first try to make some reasonable approximations in order to be able to analyze it analytically .
first , as the angular integrals involved in @xmath3-dimensional integration are standard ( see , for example , refs .
@xcite and @xcite ) we may reduce eq .
( [ masseq ] ) to a one - dimensional integral , namely @xmath44 , \label{inteq}\end{aligned}\ ] ] where @xmath45 note that for @xmath33 the mass function in eq.([inteq ] ) reduces to the standard one in qed@xmath46 . in @xmath47 dimensions the hypergeometric functions in eq .
( [ inteq ] ) preclude a solution in closed form .
however , note that these hypergeometric functions have a power expansion in @xmath21 so that for small @xmath21 one is not likely to go too far wrong by just replacing these by their @xmath48 ( i.e. @xmath33 ) limit .
after all , the reason for choosing dimensional regularization in the first place is in order to regulate the integral , and this is achieved by the factor of @xmath49 , not the hypergeometric functions .
in addition , this approximation also corresponds to just replacing the hypergeometric functions by their ir and uv limits , so that one might expect that even for larger @xmath21 that the approximation is not too bad in these regions . )
always has symmetry breaking solutions even without making this approximation of the angular integrals .
we indicate how this may be done in appendix a. ] making this replacement , i.e. @xmath50\;\;\ ; , \label{inteqxa}\ ] ] allows us to convert eq .
( [ inteq ] ) into a differential equation , namely @xmath51^\prime+\alpha \nu^{2 \epsilon}c_{\epsilon}\frac{(p^2)^{1 - \epsilon}}{p^2+m^2(p^2)}m(p^2)=0\;\;\ ; , \label{diffeq}\ ] ] with the boundary conditions @xmath52^\prime|_{p^2= \infty}=0\;\;\;.\ ] ] unfortunately , the differential equation ( [ diffeq ] ) still has no solutions in terms of known special functions . since the mass function in the denominator of eq .
( [ inteqxa ] ) serves primarily as an infrared regulator we shall make one last approximation and replace it by an infrared cut - off for the integral , which can be taken as a fixed value of @xmath53 in the infrared region ( for convenience we shall call this value the ` dynamical mass ' @xmath54 ) .
this simplifies the problem sufficiently to allow the derivation of an analytical solution . in terms of the dimensionless variables
@xmath55 , @xmath56 and @xmath57 the linearized equation becomes @xmath58\;\;\ ; ; \label{inteqx}\ ] ] for simplicity , we do not explicitly differentiate between @xmath59 and @xmath10 .
this may be written in differential form as @xmath60^\prime\ > + \
\alpha \>c_{\epsilon } \>x^{-\epsilon } m(x)=0 \;\;\ ; , \label{diffeqsimple}\ ] ] with the boundary conditions @xmath61^\prime|_{x=\infty}=0\;\;\;. \label{bc}\ ] ] this differential equation has solutions in terms of bessel functions @xmath62\;\;\ ; , \label{diffeqsol}\ ] ] where we have defined @xmath63 in order to avoid cumbersome indices on the bessel functions .
the ultraviolet boundary condition eq .
( [ bc ] ) gives @xmath64 while the infrared boundary condition leads to @xmath65_{x = a}\ > = \
> 0\;\;\;. \label{dynmasseq}\ ] ] this equation may be simplified using the relation among bessel functions @xmath66 and becomes @xmath67_{x = a}\ > = 0\ > \;\;\;. \label{eq : simple dynmasseq}\ ] ] clearly this equation is satisfied by @xmath68 , which corresponds to the trivial chirally symmetric solution @xmath69 .
however , for values of @xmath70 which are such that the argument of the bessel function in eq .
( [ eq : simple dynmasseq ] ) corresponds to one of its zeroes , the equation is also satisfied for @xmath71 , i.e. for these values of @xmath70 there exist solutions with dynamically broken chiral symmetry . if we define @xmath72 to be the smallest positive zero of eq .
( [ eq : simple dynmasseq ] ) , the dynamical mass for this solution becomes @xmath73 note that for this solution the normalization @xmath74 is not fixed by eq .
( [ inteqx ] ) as this equation is linear in @xmath59 .
later on we shall fix @xmath74 by demanding that @xmath75 , however there is no compelling reason to do this and one might alternatively fix the normalization in such a way as to approximate the true ( numerical ) solutions of eq .
( [ masseq ] ) as well as possible .
finally , note that , as expected , a dynamical symmetry breaking solution exists for any value of the coupling and that the expression for the dynamical mass is in agreement with the general form expected from dimensional considerations [ i.e. eq . ( [ eq : m(0 ) general form ] ) ] . in order to extract @xmath5 , we need to look at the behaviour of @xmath54 as @xmath21 goes to zero ( i.e. @xmath76 ) .
this may be done by noting that the positive roots of the bessel function @xmath77 have the following asymptotic behaviour ( see , for example , eq .
9.5.22 in ref .
@xcite ) : @xmath78 where @xmath79 is the @xmath80th negative zero of airy function ai(@xmath81 ) , and @xmath82 is determined @xmath83 from the equation @xmath84 for large @xmath85 the variable @xmath86 is small and so it is valid to expand @xmath81 around 1 .
writing @xmath87 we obtain @xmath88 and so @xmath89 yielding , to leading order , @xmath90 if we define @xmath91 then the leading terms in the expansion of @xmath92 are @xmath93 also , the coefficient @xmath94 appearing in eq .
( [ dynmass ] ) may be expanded @xmath95 so that for small @xmath21 the dynamical mass becomes @xmath96^{1 \over 2 \epsilon } \over ( 1+\gamma\epsilon^{2/3}-\epsilon)^{1 \over \epsilon } } \
> \left ( { \alpha \over { \pi/ 3}}\right)^{1 \over 2 \epsilon } \
> e^{1 + { d \over 2 } - \gamma\epsilon^{-1/3}}\;\;\;. \label{eq : dyn mass app}\ ] ] note that the behaviour of the first term ( for @xmath21 going to zero ) dominates over the exponential function , as required in eq .
( [ eq : m_overbar def ] ) . hence the critical coupling in four dimensions
is given by @xmath97 , as expected from cut - off based work @xcite .
returning now to the mass function itself , we may substitute the expression for the dynamical mass , i.e. , eq .
( [ dynmass ] ) , together with our choice of normalization condition @xmath98 into eq .
( [ masseq ] ) in order to eliminate @xmath74 .
one obtains @xmath99 note that the explicit dependence on @xmath13 ( and hence @xmath24 ) has been completely replaced by @xmath54 in this expression .
so far we have taken @xmath24 independent of the regularization . as we have seen
this leads to a dynamically generated mass which becomes infinite as the regulator is removed .
fomin et al .
@xcite examined ( within cut - off regularized qed ) a different limit , namely one where the mass @xmath54 is kept constant while the cut - off is removed . in our case
this limit necessitates that the coupling @xmath24 is dependent on @xmath21 through @xmath100 [ see eq .
( [ eq : dyn mass app ] ) ; note that @xmath5 is approached from above ] .
the limit may be taken analytically in eq .
( [ massfunction ] ) by making use of the known asymptotic behaviour of the bessel functions ( see eq .
9.3.23 of ref .
@xcite ) , i.e. @xmath101 as well as the asymptotic expansion of @xmath92 in eq .
( [ eq : root ] ) .
one obtains @xmath102 which agrees with the result in ref .
@xcite . to conclude this section
, we analyze the validity of the approximations made by solving eq .
( [ masseq ] ) numerically and comparing it to the bessel function solution in eq .
( [ massfunction ] ) . in fig .
[ fig : comp]a we have plotted the mass function ( divided by @xmath13 ) as a function of the dimensionless momentum @xmath103 for a moderately large coupling ( @xmath104 ) and @xmath105 .
the solid curve corresponds to the exact numerical result [ eq .
( [ masseq ] ) ] while the dashed line is a plot of eq .
( [ massfunction ] ) for these parameters . as can be seen ,
the approximation is not too bad and could actually be made significantly better by adopting a different normalization condition to that in eq .
( [ eq : norm cond ] ) however , no further insight is gained by doing this and we shall not pursue it further .
one might naively think that most of the difference between the bessel function and the exact numerical solution comes from the linearization of eq .
( [ masseq ] )
i.e. , the approximation made by going from eq .
( [ inteqxa ] ) to eq .
( [ inteqx ] ) , as the only approximation made prior to this is to replace the hypergeometric functions by unity , which is expected to be good to order @xmath21 ( i.e. in this case , 3 % ) .
this turns out to be not the case ; the dotted curve in fig .
[ fig : comp]a corresponds to the ( numerical ) solution of eq .
( [ inteqxa ] ) .
not only is the difference to the true solution essentially an order of magnitude larger than expected ( about 30 % note that fig .
[ fig : comp]a is a log - log plot ) , it is actually of opposite sign to the equivalent difference for the bessel function . in other words ,
the validity of the two approximations is roughly of the same order of magnitude and they tend to compensate .
why are the quantitative differences rather larger than expected ? on the level of the integrands the approximations are actually quite good . in fig .
[ fig : comp]b we show the integrands of eqs .
( [ masseq ] ) , ( [ inteqx ] ) and eq . ( [ inteqxa ] ) for a value of @xmath103 in the infrared ( @xmath106 @xmath107 ) . clearly the replacement of the hypergeometric functions by unity is indeed an excellent approximation , as is the linearization performed in eq .
( [ inteqx ] ) ( except in the infrared , as expected .
note that when estimating the contribution to the integral from different @xmath108 one should take into account that the x - axis in fig .
[ fig : comp]b is logarithmic ) . the real source of the ` relatively large ' differences observed for the integrals in fig .
[ fig : comp]a is the fact that these are integral equations for the function @xmath59 small differences in the integrands do not necessarily guarantee small differences in @xmath59 . to illustrate this point , consider a hypothetical ` approximation ' to eq .
( [ inteqx ] ) in which we just scale the integrands by a constant factor @xmath109 and ask the question how much this affects the solution @xmath59 . for @xmath110
the answer is rather simple : the hypothetical approximation just corresponds to a rescaling of @xmath24 by @xmath109 and as @xmath22 scales like @xmath111 we find that the solution has increased by a factor @xmath112 . in other words , even in the limit @xmath113 there remains a remnant of the ` approximation ' , namely a rescaling of @xmath22 by a factor @xmath114 !
we shall now leave the rainbow approximation and turn to the cp vertex .
the expressions for the scalar and dirac self - energies for this vertex , using dimensional regularization and in an arbitrary gauge , have already been given in ref .
before we discuss chiral symmetry breaking for this vertex we shall first examine the chirally symmetric phase .
we remind the reader that in this phase in four dimensions the wavefunction renormalization has a very simple form for this vertex @xcite , namely @xmath115 where the renormalized dirac propagator is given by @xmath116 here @xmath117 is the gauge parameter and @xmath118 is the ( dimensionless ) renormalization scale .
this power behaviour of @xmath119 is in fact demanded by multiplicative renormalizability @xcite as well as gauge covariance @xcite .
we shall derive the form of this self - energy in @xmath120 dimensions , which will provide a very useful check on the numerical results even if @xmath121 as long as @xmath122 . in the chirally symmetric phase , the unrenormalized @xmath119 corresponding to the cp vertex in @xmath3 dimensions is given by @xmath123 .
\label{eq : massless a}\ ] ] this equation may be obtained from eq .
( a6 ) of ref .
@xcite by setting @xmath124 equal to zero in that equation and by using eq .
( a8 ) of the same reference in order to eliminate the terms with coefficient @xmath125 .
the angular integral @xmath126 is defined to be @xmath127 in four dimensions the solution to eq .
( [ eq : massless a ] ) is given by a @xmath119 having a simple power behaviour while for @xmath0 this is clearly no longer the case .
nevertheless , it is possible to derive an integral representation of the solution of eq .
( [ eq : massless a ] ) by making use of the gauge covariance of this equation .
we do so in appendix b , with the result @xmath128 although this result is exact it is somewhat cumbersome to evaluate numerically because , for @xmath113 , the oscillations in the integrand become increasingly important .
for this reason we shall approximate the integrand in eq .
( [ eq : massless a ] ) by its ir and uv limits , as we did for the rainbow approximation ( as before , this approximation is good to order @xmath21 ) . using @xmath129\ ] ]
this approximation yields @xmath130\;\;\;. \label{eq : massless z approx}\ ] ] this may be converted to the differential equation @xmath131 \label{eq : z diff eq}\ ] ] where @xmath132 is defined to be @xmath133 and the appropriate boundary conditions are @xmath134 ( the ir boundary condition arises from the requirement that the integral in eq .
( [ eq : massless z approx ] ) needs to converge at its lower limit . ) in order to solve eq .
( [ eq : z diff eq ] ) , it is convenient to change variables to @xmath135 and to define @xmath136 so that the differential equation becomes @xmath137 while the boundary conditions now are @xmath138 this equation is essentially kummer s equation ( see eq .
13.1.1 of ref .
@xcite ; we use the notation of that reference in the following ) .
its general solution may be expressed in terms of confluent hypergeometric functions , i.e. @xmath139\nonumber \\ & = & e^{-a z } \
> \left \{c_1 \
> \left [ \gamma(a+1 , - a z ) \ > + \ > a z \
> \gamma(a , -a z ) \right ] \ > + \
\>[1+z]\right \}\;\;\;. \label{eq : gen_solution}\end{aligned}\ ] ] the uv boundary condition is fulfilled if @xmath140 while @xmath74 is not fixed by the boundary conditions .
although eq .
( [ eq : massless z approx ] ) is solved by eq .
( [ eq : gen_solution ] ) for arbitrary @xmath74 we shall concentrate on the solution with @xmath74=0 .
the reason for this is that the solution to the unapproximated integral [ eq . ( [ eq : res ] ) ] vanishes at @xmath110 ( see appendix b ) while the term multiplying @xmath74 in eq .
( [ eq : gen_solution ] ) diverges like @xmath141 and is therefore unlikely to provide a good approximation to eq .
( [ eq : massless a ] )
. hence we obtain @xmath142 \exp(-{\tilde c \over \epsilon } x^{-\epsilon})\;\;\;.\ ] ] finally , the renormalized function @xmath143 may be obtained from this by demanding that @xmath144 so that the renormalized wavefunction renormalization becomes @xmath145 only in the limit @xmath146 does this reduce to the usual power behaved function found in cut - off based work [ eq .
( [ eq : pow z ] ) ] while for @xmath0 it vanishes non - analytically at @xmath110 . on the other hand , note that the solution to eq .
( [ eq : massless a ] ) for finite @xmath21 only goes to zero linearly in @xmath103 . for the purpose of this paper this difference in the analytic behaviour in the infrared does not concern us as for solutions which break chiral symmetry the infrared region
is regulated by @xmath147 so that we do not expect the chirally symmetric @xmath4 to be a good approximation in this region in any case .
we shall now examine dynamical chiral symmetry breaking for the @xmath148 vertex in the absence of any explicit symmetry breaking by a nonzero bare mass , as before . even for solutions exhibiting dynamical symmetry breaking ,
it is to be expected that the analytic result derived for @xmath149 [ eq . ( [ eq : cp z analytic ] ) ] remains valid as long as @xmath103 is large compared to @xmath150 and @xmath21 is sufficiently small . that this is indeed the case
is illustrated in fig .
[ fig : a(x ) ] , where we show a typical example of @xmath151 for a solution which breaks chiral symmetry . in this figure , as well as in the rest of this section
, we shall be dealing with the renormalized @xmath119 and @xmath59 instead of the unrenormalized quantities in the previous sections .
this makes no essential difference to the physics of chiral symmetry breaking , although it of course effects the absolute scale of @xmath119 . for a discussion of the renormalization of the dimensionally regularized theory
we refer the reader to ref .
@xcite .
the comparison to the analytic result in fig .
[ fig : a(x ) ] provides a very convenient check on the numerics .
another check is provided by plotting the logarithm of @xmath22 against the logarithm of the coupling . according to eq .
( [ eq : m(0 ) general form ] ) this should be a straight line with gradient @xmath152 . as can be seen in fig .
[ fig : alpha_vs_mass ] not only does one observe chiral symmetry breaking down to couplings as small as @xmath153 , the expected linear behaviour is confirmed to quite high precision .
although the numerics in @xmath0 dimensions are clearly under control , the extraction of the critical coupling ( appropriate in four dimensions ) has proven to be numerically quite difficult . from the discussion in sections [ sec_motivation ] and [ sec :
rainbow ] , we anticipate that the logarithm of the dynamical mass has the general form @xmath154 where the last term is subleading as compared to the first as @xmath21 tends to zero .
for sufficiently small @xmath21 , therefore , @xmath5 is related to the gradient of @xmath155 plotted against @xmath156 . in fig .
[ fig : logmass_vs_inv_epsilon ] we attempt to extract @xmath5 in this way .
the logarithm of @xmath22 was evaluated for @xmath21 ranging from 0.03 down to @xmath157 for a fixed gauge @xmath158 .
the squares corresponds to a coupling constant @xmath159 , although some of the points at lower @xmath21 have actually been calculated at smaller @xmath24 and then rescaled according to eq .
( [ eq : m(0 ) general form ] ) . at present
we are unable , for these parameters , to decrease @xmath21 significantly further without a significant loss of numerical precision .
( we also note in passing that it is quite difficult numerically to move away from small values of the gauge parameter ; @xmath160 , which , judging from fig .
[ fig : cut - off alpha_cr ] , would not require a very high numerical accuracy for @xmath5 , is unfortunately not an option . )
the two fits shown in fig . [ fig : logmass_vs_inv_epsilon ] correspond to two different assumptions for the functional form of @xmath161 , which is a priori unknown .
the curves do indeed appear to be well approximated by a straight line , however we caution the reader that this does not allow an accurate determination of @xmath5 as the gradient is essentially determined by the ` trivial ' dependence on @xmath162 ( more on this below ) .
the solid line corresponds to the assumption that the leading term in @xmath163 has the same form as what we found in the rainbow approximation , i.e. @xmath164 with this form the fit parameters @xmath5 and @xmath165 are found to be @xmath166 indeed , the critical coupling is similar to what is found in cut - off based work ( see sec .
[ sec_motivation ] ; in ref .
@xcite the value was @xmath167 for @xmath158 ) . at present
it is difficult to make a more precise statement , let alone differentiate between the two curves plotted in fig .
[ fig : cut - off alpha_cr ] , as @xmath5 is quite strongly dependent on the functional form assumed in eq .
( [ eq : fit1 ] ) .
in fact , allowing an extra constant term on the right hand side of eq .
( [ eq : fit1 ] ) reduces the critical coupling to @xmath168 and the addition of yet a further term proportional to @xmath169 increases it again to @xmath170 . as these numbers appear to converge to something of the order of @xmath171 or @xmath172 one might think that @xmath5 has been determined to this precision .
however , it is not clear that the functional form suggested by the rainbow approximation should be taken quite this seriously . the dashed line in fig .
[ fig : logmass_vs_inv_epsilon ] corresponds to a fit where the power of @xmath21 of the subleading term has been left free , i.e. @xmath173 the optimum fit assuming this form for @xmath155 yields a power quite different to @xmath174 and a very much smaller @xmath5 : @xmath175 to conclude this section , let us discuss why it is that the functional form of the subleading term @xmath161 appears to be rather important even if @xmath21 is already rather small .
the reason for this is two - fold : most importantly , although the leading @xmath21 dependence of @xmath155 is indeed @xmath156 , the coefficient of this term ( leaving out the trivial @xmath24 dependence ) is @xmath176 . as @xmath5 is rather close to 1 one
therefore obtains a strong suppression of this leading term , increasing the relative importance of the subleading terms .
in addition , it appears as if the numerical results favour a subleading term which is not as strongly suppressed ( as a function of @xmath21 ) as suggested by the rainbow approximation ( i.e. the power of @xmath156 of the subleading term appears to be closer to @xmath177 rather than @xmath174 ) .
this again increases the importance of the subleading terms .
the primary purpose of this paper was to explore the phenomenon of dynamical chiral symmetry breaking through the use of dyson - schwinger equations with a regularization scheme which does not break the gauge covariance of the theory , namely dimensional regularization .
it is necessary to do this as the cut - off based work leads to ambiguous results for the critical coupling of the theory precisely because of the lack of gauge covariance in those calculations . in particular , this should be kept in mind when using the expected gauge invariance of the critical coupling as a criterion for judging the suitability of a particular vertex . to begin with ,
we have shown on dimensional grounds alone and for an arbitrary vertex , that in @xmath32 dimensions either a symmetry breaking solution does not exist at all ( in which case , however , it would also not exist in @xmath33 dimensions ) or it exists for all non zero values of the coupling ( in which case a chiral symmetry breaking solution exists in @xmath33 for @xmath178 ) . for dyson - schwinger analyses employing the rainbow and cp vertices we have shown that it is the second of these possibilities which is realized .
for these symmetry breaking solutions the limit to @xmath179 is necessarily discontinuous and so the extraction of the critical coupling of the theory ( in 4 dimensions ) is not as simple as in cut - off regularized work .
we next turned to an examination of symmetry breaking in the rainbow approximation in landau gauge , both analytically and numerically . indeed , for this vertex one could rewrite the ( linearized ) dyson - schwinger equation as a schrdinger equation in 4 dimensions and appeal to standard results from elementary quantum mechanics to explicitly show that the theory always breaks chiral symmetry if @xmath180 .
we also showed how the usual critical coupling @xmath181 may be extracted from the dimensionally regularized work .
we concluded this work with an examination of the cp vertex . by making use of the gauge covariance of the theory we derived an exact integral expression for the wavefunction renormalization function @xmath9 of the chirally symmetric solution .
furthermore we obtained a compact expression for this quantity which is an excellent approximation to the true @xmath9 even for solutions which break the chiral symmetry .
finally , we extracted the critical coupling corresponding to this vertex and found that , within errors , it agrees with the standard cut - off results . in the future
, we plan to increase the numerical precision with which we can extract this critical coupling for the cp vertex by an order of magnitude or so .
the factor limiting the precision at present is that when solving the propagator s dyson - schwinger equation with the cp vertex by iteration the rate of convergence decreases dramatically as @xmath21 is decreased below @xmath182 .
if this increase in precision can be attained it will enable one to make a meaningful comparison with the cut - off based results shown in fig .
[ fig : cut - off alpha_cr ] .
we would like to acknowledge illuminating discussions with d. atkinson , a. kizilers , v. a. miransky and m. reenders .
vpg is grateful to the members of the institute for theoretical physics of the university of groningen for hospitality during his stay there .
this work was supported by a swiss national science foundation grant ( grant no .
ceec / nis/96 - 98/7 051219 ) , by the foundation of fundamental research of ministry of sciences of ukraine ( grant no 2.5.1/003 ) and by the australian research council . in this appendix
we show that the linearized version of eq .
( [ masseq ] ) , i.e. @xmath183 has symmetry breaking solutions for all values of the coupling .
our aim here is to convert this equation to a schrdinger - like equation , which we do by introducing the function @xmath184 with this definition we have @xmath185 where @xmath186 is the @xmath3-dimensional laplacian and so @xmath187 after shifting the integration variable @xmath188 the last equation can be written in the form of a schrdinger - like equation @xmath189 where @xmath190 is the hamiltonian , @xmath191 plays the role of an energy and the potential @xmath192 given by @xmath193 where @xmath194 for @xmath195 the coefficient @xmath196 is @xmath197 while near @xmath33 it is @xmath198 .
it is well known from any standard course of quantum mechanics ( see , for example , ref.@xcite ) that potentials behaving as @xmath199 at infinity , with @xmath200 , always support bound states ( actually , an infinite number of them ) . in the present case
this can be seen by considering the schrdinger equation ( [ schrodinger ] ) for zero energy , i.e. @xmath201 .
the @xmath80-symmetric wave function then satisfies the equation @xmath202 the solution finite at the origin @xmath203 is @xmath204 the bessel function in ( [ e0function ] ) has an infinite number of zeros , which means that there is an infinite number of states with @xmath205 . returning now to eq .
( [ schrodinger ] ) , we can estimate the lowest energy eigenvalue variationally by using @xmath206 as a trial wavefunction . here
@xmath207 is related to @xmath208 by demanding that @xmath209 is normalized , i.e. @xmath210 where @xmath211 is the volume of a @xmath3-dimensional sphere .
calculating the expectation value of the `` hamiltonian '' @xmath212 on the trial wave function in eq .
( [ trialfunction ] ) we find @xmath213 \label{varenergy}\ ] ] the minimum of the `` ground state energy '' in eq .
( [ varenergy ] ) , @xmath214 , is reached at @xmath215 ( for @xmath195 the parameter @xmath208 is @xmath216 while near @xmath33 it is @xmath217^{{1 \over 2\epsilon}}$ ] ) and is given by the expression @xmath218 where the @xmath219 is the contribution from the kinetic energy while the @xmath220 corresponds to the potential energy .
for @xmath221 the potential is attractive and for @xmath222 it is always larger than the kinetic energy , so for this case we get dynamical symmetry breaking for any value of @xmath24 .
for example , for @xmath195 , one obtains @xmath223 which coincides precisely with the ground - state energy of the hydrogen atom ( not surprisingly , as we have used the ground - state hydrogen wave function as our trial function ) . in this case
the dynamical mass is @xmath224 . for @xmath3 near @xmath2 , on the other hand
, we obtain from eq .
( [ e_0 ] ) that @xmath225 this is of the general form anticipated in section [ sec_motivation ] , with @xmath226 .
indeed , for @xmath33 , the schrdinger equation ( [ schrodinger ] ) becomes an equation with the singular potential @xmath227 again , it is known from standard quantum mechanics@xcite that the spectrum of bound states for such a potential depends on the strength @xmath196 of the potential : it has an infinite number of bound states with @xmath205 if @xmath228 and bound states are absent if @xmath229 .
thus , the true critical value for the coupling is expected to be @xmath230 instead of the @xmath231 obtained with the help of the variational method ( which made use of the exponential ansatz for the wavefunction and thus only gave an upper bound for the energy eigenvalue ) . because the cp vertex in the chirally symmetric phase of qed is gauge - covariant @xcite it is possible to derive an integral representation of the wavefunction renormalization function @xmath119
[ see eq .
( [ eq : massless a ] ) ] from the landau - khalatnikov transformation @xcite .
this transformation relates the coordinate space propagator @xmath232 in one gauge to the propagator in a different gauge .
specifically , with covariant gauge fixing , we have @xmath233 } \tilde s^{\xi=0}(u)\ ] ] where @xmath234 is essentially the fourier transform of the gauge - dependent part of the photon propagator , i.e. @xmath235 specifically , we obtain @xmath236 where @xmath237 substituting the coordinate - space propagator in landau gauge , i.e. @xmath238 and carrying out the inverse fourier transform of eq .
( [ eq : lk in d dim ] ) one obtains the wavefunction renormalization function in an arbitrary gauge , namely @xmath239 note that for small @xmath103 this function vanishes : @xmath240 it may be checked explicitly that eq .
( [ eq : res2 ] ) is indeed a solution to eq .
( [ eq : massless a ] ) for arbitrary @xmath3 by making use of the expansion of eq .
( [ eq : res2 ] ) around @xmath241 . to be more precise ,
consider the rhs of eq .
( [ eq : massless a ] ) upon insertion of the power @xmath242 in the place of @xmath243 .
note that the integral converges only if @xmath244 .
after some work the result is that the r.h.s . of eq .
( [ eq : massless a ] ) becomes @xmath245 .
\label{insertdelta}\end{aligned}\ ] ] for @xmath246 this may simplified further by applying dougall s formula ( eq .
1.4.1 in @xcite ) which , in this case , reduces to @xmath247)}{1 \over \gamma(1-\delta ) \gamma(2 + \delta - 2 \epsilon)}\ ] ] using this result , eq .
( [ insertdelta ] ) becomes @xmath248 note that , as opposed to the integral representation eq .
( [ eq : massless a ] ) , this expression is defined for @xmath249 outside the range @xmath244 and so we may use it as an analytical continuation of the integral . furthermore , note that this last expression vanishes for integer @xmath250 hence we can not obtain a simple power expansion around @xmath110 for @xmath119 in this way . on the other hand , an expansion in powers of @xmath251 is possible .
if we seek a solution of the form @xmath252 we may equate the coefficients of equal powers of @xmath251 after inserting the series ( [ series ] ) into both sides of eq.([eq : massless a ] ) .
this way we obtain the recurrence relation for the coefficients @xmath253 ( @xmath254 ) as @xmath255 this may be solved leading to @xmath256^n \frac{\gamma(2-\epsilon)\gamma(1+n\epsilon)}{\gamma(2-\epsilon- n\epsilon)n!}\;\;\;,\ ] ] so that finally we obtain @xmath257^n\ ] ] as the series expansion of the solution to eq .
( [ eq : massless a ] ) .
the reader may check that this coincides precisely with the corresponding expansion of the solution obtained via the landau - khalatnikov transformations [ eq .
( [ eq : res2 ] ) ] .
the latter may be obtained by changing the variable of integration from @xmath258 to @xmath259 , expanding the exponential in the integrand and making use of the standard integral @xmath260 for a review , see c.d .
roberts and a.g .
williams , _ dyson - schwinger equations and their application to hadronic physics _ , in _ progress in particle and nuclear physics , vol .
33 _ ( pergamon press , oxford , 1994 ) , p. 477 .
recently , there has been some discussion of dimensional regularization in effective field theories ; see , for example , d. r. phillips , s. r. beane and m. c. birse , preprint no .
umd-99 - 030 , hep - th/9810049 and references therein .
l. von smekal , p. a. amundsen , and r. alkofer , nucl .
* a529 * , 633 ( 1991 ) ; m. becker , `` nichtperturbative strukturuntersuchungen der qed mittels genherter schwinger - dyson - gleichungen in dimensioneller regularisierung '' , ph.d .
dissertation , w. w. u. mnster,1995 . for a discussion of numerical renormalization of the dyson - schwinger equations ,
see f.t . hawes and a.g .
williams , phys .
d * 51 * , 3081 ( 1995 ) ; f.t .
hawes , t. sizer and a.g .
williams , phys . rev .
d * 55 * , 3866 ( 1997 ) . | in this paper we study dynamical chiral symmetry breaking in dimensionally regularized quenched qed within the context of dyson - schwinger equations . in @xmath0 dimensions
the theory has solutions which exhibit chiral symmetry breaking for all values of the coupling .
to begin with , we study this phenomenon both numerically and , with some approximations , analytically within the rainbow approximation in the landau gauge .
in particular , we discuss how to extract the critical coupling @xmath1 relevant in @xmath2 dimensions from the @xmath3 dimensional theory . we further present analytic results for the chirally symmetric solution obtained with the curtis - pennington vertex as well as numerical results for solutions exhibiting chiral symmetry breaking .
for these we demonstrate that , using dimensional regularization , the extraction of the critical coupling relevant for this vertex is feasible .
initial results for this critical coupling are in agreement with cut - off based work within the currently achievable numerical precision . |
the launch of the _ swift _ satellite @xcite has revolutionized the study of the interstellar medium of gamma - ray burst ( grb ) host galaxies and fostered new studies on the intervening intergalactic medium @xcite . while previous missions alerted the community to the existence of very bright grb afterglows ( e.g. * ? ? ?
* ) , only a few events were identified in time to allow echelle observations @xcite .
the rapid localization by _ swift _ of the hard x - ray emission to several arcminutes @xcite and the soft x - ray component to a few arcseconds @xcite has enabled observers to obtain high signal - to - noise , high - resolution spectroscopy of grbs with 10m - class telescopes . in this paper , we present a uniform dataset of echelle spectra of four grb afterglows by our grb afterglows as probes ( graasp ) collaboration . among other applications , the data presented here can be used to derive constraints on the physical conditions of the ism of grb host galaxies : the metallicity , dust - to - gas ratio , chemical abundance patterns , ionization state , the distance of the grb to the absorbing gas , etc .
future papers in this series will examine these properties in greater detail and provide comparison with analogous observations along quasar sightlines .
in addition to the data presented here , we have collected several low - resolution spectra of these and other grbs and will present that data in a companion paper @xcite along with ism abundance measurements obtained by different analysis techniques more suitable to lower resolution data @xcite .
we also present measurements of the strong systems in the grb afterglow spectra . in @xmath0 [ sec : obs ] we summarize the observations and data reduction procedures .
the methodology is briefly described in @xmath0 [ sec : colm ] and the line - profiles and column density values are given in @xmath0 [ sec : grb ] and @xmath0 [ sec : mgii ] .
the grb afterglow spectra presented here were acquired with either the high resolution echelle spectrometer ( hires ; * ? ? ?
* ) on the kecki 10 m telescope or the magellan inamori kycoera echelle spectrometer ( mike ; * ? ? ?
* ) on the magellan 6.5 m clay telescope at las campanas observatory .
table [ tab : obslog ] presents a summary of the grb events , the afterglow observations , instrumental configurations , and data quality of the spectra obtained for the four afterglows for which we have obtained high - resolution spectra : grb 050730 , grb 050820 , grb 051111 , and grb 060418 .
the data were reduced with the hires redux and mike reduxburles / www / mike/ ] pipelines @xcite .
briefly , the 2d images were bias subtracted , flat - fielded , and wavelength calibrated using standard thar and quartz lamp calibration images .
the data were optimally extracted using a non - parametric spatial profile and coadded after weighting by the global snr@xmath1 of each echelle order . a final ,
normalized 1d spectrum was produced by fitting high - order legendre polynomials to each echelle order ( hires ) or the fluxed 1d spectrum ( mike ) .
we provide additional details related to the specific observations in the following section .
following our standard practice for the analysis of interstellar lines observed in the damped ly@xmath2 systems ( quasar absorption systems with @xmath3 * ? ? ?
* ; * ? ? ?
* ) , we have measured column densities using the apparent optical depth method ( aodm ; * ? ? ? * ) .
this technique leads to non - parametric , accurate results provided the profiles are resolved . in cases where one measures multiple transitions for a single ion ( e.g. 1741 , 1751 )
, one can test for line saturation by examining the column density values derived as a function of the line - strength .
if there is no trend of increasing column density with decreasing line - strength , then the profiles are free of significant ` hidden ' saturation @xcite .
we find no indication of hidden saturation in the grb studied here .
we report column density values for all transitions detected at greater than @xmath4 statistical significance and report @xmath5 statistical uncertainty in the sections that follow . for ions or atoms where multiple transitions are measured ,
we report the weighted mean column density and combined statistical error in that mean .
while the data may formally yield errors of less than @xmath6dex , we recommend that one adopt a minimum statistical uncertainty of 5% the column density .
saturated line - profiles , i.e. , those where the minimum normalized flux of the line - profile is less than the @xmath5 statistical error , are reported as lower limits . finally , undetected or blended transitions are reported as @xmath4 upper limits assuming statistical uncertainty alone .
accounting for continuum error , these limits are more likely @xmath7 values .
we note that uncertainties in weak lines are likely to be dominated by systematics associated with the continuum placement , and therefore adopt a minimum uncertainty of @xmath8 in our own analysis .
the aodm technique can not be applied to transitions which are blended with coincident features . in cases of particular importance , we have derived line - profile fits to the data using the software package vpfit , kindly provided by r. carswell . finally , when the ly@xmath2 profile is covered by the spectroscopic observations , we have fitted this line using the _ x_fitdla _ algorithm within the xidl packagexavier / idl ] ( e.g. * ? ? ?
the following sections present velocity profiles of transitions observed along the grb sightlines .
the transitions are separated into two groups : those arising from the ground state ( termed resonance lines ) and those arising from excited levels ( ambiguously termed fine - structure lines ) . in all of the figures ,
line - blends are indicated by dotted lines . in cases where the line - blending is extremely severe ,
we have not shown the velocity profile .
the atomic data used here has been taken from recent laboratory measurements and theoretical calculations when lab measurements are not available : ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the _ swift _ bat instrument triggered on grb 050730 at ut 2005 july 30 19:58:23 and shortly after @xcite reported the detection of a bright optical afterglow ( @xmath9 at 199s after the trigger ) at ra=@xmath10 , dec=@xmath11 .
after waiting for the sun to set at las campanas observatory , we began acquiring data with the magellan telescope under good conditions .
@xcite presented a preliminary analysis of this dataset ; we present the full set of transitions observed .
figure [ fig:050730lya ] presents the ly@xmath2 and ly@xmath12profiles for the host galaxy , and the best voigt profile fit which gives @xmath13dex .
the uncertainty in this fit is dominated by the fluxing error associated with echelle spectrometers and continuum fitting . a lower resolution , well calibrated , single - order spectrum would likely give a more precise result .
figure [ fig:050730res ] shows the velocity profiles for the resonance lines associated with the ism of the host galaxy for grb 050730 .
figure [ fig:050730fine ] shows the fine - structure transitions and a resonance line for comparison .
finally , table [ tab:050730 ] lists the column density measurements for all of the transitions .
the _ swift _ bat instrument triggered on grb 050820 at ut 2005 august 20 06:34:53 and the _ swift _ team issued an xrt position within a few minutes @xcite .
@xcite reported a bright ( @xmath14 ) , optical afterglow at ra=@xmath15 , dec=@xmath16 .
we began acquiring data with the hires echelle spectrometer on the kecki telescope shortly after .
our instrumental configuration followed the standard setup for the uc planet hunters ( e.g. * ? ? ? * ) . the short slit length ( @xmath17 ) complicated sky subtraction although the seeing ( @xmath18 ) was sufficiently good to give fair results . in figure [ fig:050820lya ]
, we present the echelle orders covering the ly@xmath2 profile for the host galaxy of grb 050820 .
overplotted is a fit to the data which yields @xmath19dex .
the uncertainty in this fit is dominated by continuum error which is related to difficulties in fluxing this echelle data .
figure [ fig:050820res ] shows the velocity profiles for the resonance lines associated with the ism of the host galaxy .
figure [ fig:050820fine ] shows the fine - structure transitions and a resonance line for comparison .
finally , table [ tab:050820 ] lists the column density measurements for all of the transitions .
the _ swift _ bat instrument triggered on grb 051111 at ut 2005 november 11 05:59:39 @xcite .
owing to the moon constraint , xrt observations were delayed until 5000s after the trigger @xcite .
@xcite reported an optical afterglow at ra=@xmath20 , dec=@xmath21 .
although keck i was scheduled for engineering on this night , the observatory kindly agreed to switch instruments to hires and obtain echelle spectroscopy of grb 050820 @xcite .
figure [ fig:051111res ] shows the velocity profiles for the resonance lines associated with the ism of the host galaxy .
figure [ fig:051111fine ] shows the fine - structure transitions and a resonance line for comparison .
finally , table [ tab:051111 ] lists the column density measurements for all of the transitions .
@xcite have also presented an analysis of the same hires data acquired for this sightline .
although our central values are generally in good agreement with their results , we report significantly smaller uncertainties , perhaps because of differences in coadding the data .
the _ swift _ bat instrument triggered on grb 060418 at ut 2006 april 18 03:06:08 @xcite .
the satellite reported an immediate xrt position and @xcite reported a bright , optical afterglow at ra=@xmath22 , dec=@xmath23 .
we began observations shortly after this announcement .
figure [ fig:060418res ] shows the velocity profiles for the resonance lines associated with the ism of the host galaxy .
figure [ fig:060418fine ] shows the fine - structure transitions and a resonance line for comparison .
finally , table [ tab:060418 ] lists the column density measurements for all of the transitions .
table [ tab : summ ] presents a summary of the gas - phase abundances observed along the four sightlines . in those cases where the @xmath24 value was not measured ( grb 051111 , grb 060418 )
, we have assumed a conservative upper limit of @xmath25 .
the relative gas - phase abundances [ x / fe ] and [ x/@xmath2 ] assume solar relative abundances @xcite and do not include ionization corrections nor differential depletion corrections .
in a companion paper @xcite , we have shown that there is an enhancement of strong absorbers along grb sightlines in comparison with qso sightlines . here , we present the full set of velocity profiles and column density measurements for the absorbers along the four grb sightlines considered in that paper . in figures [ fig:050730mg1]-[fig:060418mg3 ] we present the absorption systems with @xmath27 .
the ionic column densities and equivalent widths are reported in tables [ tab:050730mgii1]-[tab:060418mgii3 ] .
we have measured the @xmath28-statistics @xcite for each profile and find @xmath29 in every case .
this suggests that a significant fraction @xmath30 ) of the absorbers are damped ly@xmath2 systems . in table
[ tab : mgsumm ] we present a summary of these quantities .
this includes the minimum @xmath31 value implied by the observed ionic column densities .
in general , the values further support the conclusion that the majority of these absorbers are dla systems .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
we acknowledge the efforts of s. vogt , g. marcy , j. wright and k. hurley in obtaining the observations of grb 050820 .
is partially supported by nasa / swift grant nng05gf55 g . ,
r. _ et al . _
2003 , in instrument design and performance for optical / infrared ground - based telescopes . edited by iye , masanori ; moorwood , alan f. m. proceedings of the spie , volume 4841 , pp .
1694 - 1704 ( 2003 ) . , 1694 .
, j. p. _ et al . _
2005 , in uv , x - ray , and gamma - ray space instrumentation for astronomy xiv .
edited by siegmund , oswald h. w. proceedings of the spie , volume 5898 , pp . 352 - 359 ( 2005 ) . ,
o. h. w. siegmund , 352 .
lcccccccccc 050730 & 19:58:23 & 3.969 & magellan / clay & mike & 12 & 2005 jul 31 00:00:04 & 2005 jul 31 00:30:04 & 1.23 & 3400 - 9400 & 11 + & & & & & 12 & 2005 jul 31 00:32:52 & 2005 jul 31 01:02:52 & 1.33 & 3400 - 9400 & 10 + & & & & & 12 & 2005 jul 31 01:05:11 & 2005 jul 31 01:35:11 & 1.48 & 3400 - 9400 & 8 + 050820 & 06:34:53 & 2.615 & kecki & hiresr & 7.5 & 2005 aug 20 07:28:59 & 2005 aug 20 07:43:59 & 1.51 & 3800 - 8000 & 14 + & & & & & 7.5 & 2005 aug 20 07:44:42 & 2005 aug 20 07:59:43 & 1.41 & 3800 - 8000 & 12 + 051111 & 05:59:39 & 1.549 & kecki & hiresr & 5 & 2005 aug 11 07:02:58 & 2005 aug 11 07:32:59 & 1.02 & 4160 - 8720 & 22 + & & & & & 5 & 2005 aug 11 07:34:23 & 2005 aug 11 08:04:23 & 1.06 & 4160 - 8720 & 18 + & & & & & 5 & 2005 aug 11 08:05:22 & 2005 aug 11 08:35:23 & 1.12 & 4160 - 8720 & 14 + 060418 & 03:06:08 & 4.048 & magellan / clay & mike & 12 & 2006 apr 18 03:34:02 & 2006 apr 18 03:54:02 & 1.61 & 3400 - 9400 & 21 + & & & & & 12 & 2006 apr 18 04:01:08 & 2006 apr 18 04:31:08 & 1.44 & 3400 - 9400 & 17 + & & & & & 12 & 2006 apr 18 04:54:31 & 2006 apr 18
05:24:31 & 1.24 & 3400 - 9400 & 9 + lccccccccc + & 0 & 0.00 & 1277.2450 & @xmath32&@xmath33$]&@xmath34&@xmath35&@xmath36 + & 0 & 0.00 & 1656.9283 & @xmath37&@xmath38$]&@xmath39&@xmath40 & + + & 1/2 & 0.00 & 1334.5323 & @xmath41&@xmath42$]&@xmath43&@xmath44&@xmath45 + & 3/2 & 63.40 & 1335.7077 & @xmath46&@xmath47$]&@xmath48&@xmath49&@xmath50 + + & & 0.00 & 1548.1950 & @xmath51&@xmath52$]&@xmath53&@xmath54&@xmath55 + & & 0.00 & 1550.7700 & @xmath56&@xmath52$]&@xmath57&@xmath58 & + + & & 0.00 & 1134.1653 & @xmath59&@xmath60$]&@xmath61&@xmath62&@xmath63 + & & 0.00 & 1134.4149 & @xmath64&@xmath60$]&@xmath65&@xmath66 & + + & & 0.00 & 1238.8210 & @xmath67&@xmath68$]&@xmath69&@xmath70&@xmath71 + & & 0.00 & 1242.8040 & @xmath72&@xmath68$]&@xmath73&@xmath74 & + + & 2 & 0.00 & 1302.1685 & @xmath75&@xmath76$]&@xmath77&@xmath78&@xmath79 + & 2 & 0.00 & 1355.5977 & @xmath80&@xmath33$]&@xmath81&@xmath82 & + & 1 & 158.26 & 1304.8576 & @xmath83&@xmath84$]&@xmath85&@xmath86&@xmath87 + & 0 & 226.98 & 1306.0286 & @xmath88&@xmath84$]&@xmath89&@xmath90&@xmath91 + + & & 0.00 & 1239.9253 & @xmath92&@xmath33$]&@xmath93&@xmath94&@xmath95 + & & 0.00 & 1240.3947 & @xmath96&@xmath33$]&@xmath97&@xmath98 & + + & & 0.00 & 1670.7874 & @xmath99&@xmath100$]&@xmath101&@xmath102&@xmath103 + + & 1/2 & 0.00 & 1260.4221 & @xmath104&@xmath105$]&@xmath106&@xmath107&@xmath108 + & 1/2 & 0.00 & 1304.3702 & @xmath109&@xmath110$]&@xmath111&@xmath112 & + & 1/2 & 0.00 & 1526.7066 & @xmath113&@xmath114$]&@xmath115&@xmath116 & + & 1/2 & 0.00 & 1808.0126 & @xmath117&@xmath118$]&@xmath119&@xmath120 & + & 3/2 & 287.24 & 1264.7377 & @xmath121&@xmath122$]&@xmath123&@xmath124&@xmath125 + & 3/2 & 287.24 & 1265.0020 & @xmath126&@xmath127$]&@xmath128&@xmath129 & + & 3/2 & 287.24 & 1309.2757 & @xmath130&@xmath131$]&@xmath132&@xmath133 & + & 3/2 & 287.24 & 1817.4512 & @xmath134&@xmath33$]&@xmath135&@xmath136 & + + & & 0.00 & 1393.7550 & @xmath137&@xmath42$]&@xmath138&@xmath139&@xmath140 + & & 0.00 & 1402.7700 & @xmath141&@xmath42$]&@xmath142&@xmath143 & + + & 1/2 & 0.00 & 1301.8740 & @xmath144&@xmath145$]&@xmath146&@xmath147&@xmath148 + + & & 0.00 & 1250.5840 & @xmath149&@xmath150$]&@xmath151&@xmath152&@xmath153 + & & 0.00 & 1253.8110 & @xmath154&@xmath150$]&@xmath155&@xmath156 & + & & 0.00 & 1259.5190 & @xmath157&@xmath150$]&@xmath158&@xmath159 & + + & & 0.00 & 1066.6600 & @xmath160&@xmath60$]&@xmath161&@xmath162&@xmath163 + + & 9/2 & 0.00 & 1608.4511 & @xmath164&@xmath47$]&@xmath165&@xmath166&@xmath167 + & 9/2 & 0.00 & 1611.2005 & @xmath168&@xmath60$]&@xmath169&@xmath170 & + & 7/2 & 384.79 & 1618.4680 & @xmath171&@xmath60$]&@xmath172&@xmath173&@xmath174 + & 7/2 & 384.79 & 1621.6856 & @xmath175&@xmath60$]&@xmath176&@xmath177 & + & 5/2 & 667.68 & 1629.1596 & @xmath178&@xmath60$]&@xmath179&@xmath180&@xmath181 + & 3/2 & 862.61 & 1634.3498 & @xmath182&@xmath60$]&@xmath183&@xmath184&@xmath185 + & 3/2 & 862.61 & 1636.3313 & @xmath186&@xmath60$]&@xmath187&@xmath188 & + & 1/2 & 977.05 & 1639.4012 & @xmath189&@xmath60$]&@xmath190&@xmath191&@xmath192 + + & & 0.00 & 1317.2170 & @xmath164&@xmath33$]&@xmath193&@xmath194&@xmath195 + & & 0.00 & 1370.1310 & @xmath196&@xmath197$]&@xmath198&@xmath199 & + & & 0.00 & 1741.5490 & @xmath200&@xmath33$]&@xmath201&@xmath202 & + lccccccccc + & 0 & 0.00 & 1277.2450 & @xmath32&@xmath33$]&@xmath203&@xmath204&@xmath205 + & 0 & 0.00 & 1560.3092 & @xmath206&@xmath33$]&@xmath207&@xmath208 & + & 0 & 0.00 & 1656.9283 & @xmath37&@xmath38$]&@xmath209&@xmath210 & + + & 1/2 & 0.00 & 1334.5323 & @xmath41&@xmath211$]&@xmath212&@xmath213&@xmath214 + & 3/2 & 63.40 & 1335.7077 & @xmath46&@xmath215$]&@xmath216&@xmath217&@xmath218 + + & & 0.00 & 1548.1950 & @xmath51&@xmath219$]&@xmath220&@xmath221&@xmath222 + & & 0.00 & 1550.7700 & @xmath56&@xmath219$]&@xmath223&@xmath224 & + + & & 0.00 & 1159.8170 & @xmath225&@xmath226$]&@xmath227&@xmath228&@xmath229 + & & 0.00 & 1160.9370 & @xmath230&@xmath226$]&@xmath146&@xmath231 & + + & 1/2 & 0.00 & 1083.9900 & @xmath232&@xmath233$]&@xmath234&@xmath235&@xmath236 + + & & 0.00 & 1238.8210 & @xmath67&@xmath226$]&@xmath237&@xmath238&@xmath239 + & & 0.00 & 1242.8040 & @xmath72&@xmath68$]&@xmath240&@xmath241 & + + & 2 & 0.00 & 1302.1685 & @xmath75&@xmath242$]&@xmath243&@xmath244&@xmath245 + & 0 & 226.98 & 1306.0286 & @xmath88&@xmath226$]&@xmath246&@xmath247&@xmath248 + + & & 0.00 & 1827.9351 & @xmath249&@xmath33$]&@xmath246&@xmath35&@xmath250 + & & 0.00 & 2026.4768 & @xmath251&@xmath252$]&@xmath207&@xmath253 & + + & & 0.00 & 1239.9253 & @xmath92&@xmath33$]&@xmath254&@xmath255&@xmath256 + & & 0.00 & 1240.3947 & @xmath96&@xmath33$]&@xmath257&@xmath258 & + + & & 0.00 & 1670.7874 & @xmath99&@xmath211$]&@xmath259&@xmath260&@xmath261 + + & & 0.00 & 1854.7164 & @xmath262&@xmath211$]&@xmath263&@xmath264&@xmath265 + & & 0.00 & 1862.7895 & @xmath266&@xmath211$]&@xmath267&@xmath268 & + + & & 0.00 & 1845.5200 & @xmath269&@xmath33$]&@xmath270&@xmath271&@xmath272 + + & 1/2 & 0.00 & 1260.4221 & @xmath104&@xmath242$]&@xmath273&@xmath274&@xmath275 + & 1/2 & 0.00 & 1304.3702 & @xmath109&@xmath242$]&@xmath276&@xmath277 & + & 1/2 & 0.00 & 1526.7066 & @xmath113&@xmath211$]&@xmath278&@xmath279 & + & 3/2 & 287.24 & 1264.7377 & @xmath121&@xmath280$]&@xmath281&@xmath282&@xmath283 + & 3/2 & 287.24 & 1265.0020 & @xmath126&@xmath252$]&@xmath284&@xmath285 & + & 3/2 & 287.24 & 1533.4312 & @xmath286&@xmath287$]&@xmath288&@xmath289 & + & 3/2 & 287.24 & 1817.4512 & @xmath134&@xmath33$]&@xmath246&@xmath290 & + + & & 0.00 & 1892.0300 & @xmath291&@xmath33$]&@xmath292&@xmath293&@xmath294 + + & & 0.00 & 1393.7550 & @xmath137&@xmath295$]&@xmath296&@xmath297&@xmath298 + & & 0.00 & 1402.7700 & @xmath141&@xmath299$]&@xmath300&@xmath301 & + + & 1/2 & 0.00 & 1152.8180 & @xmath302&@xmath303$]&@xmath304&@xmath305&@xmath306 + & 1/2 & 0.00 & 1532.5330 & @xmath307&@xmath308$]&@xmath309&@xmath310 & + + & & 0.00 & 1250.5840 & @xmath149&@xmath311$]&@xmath312&@xmath313&@xmath314 + & & 0.00 & 1253.8110 & @xmath154&@xmath311$]&@xmath315&@xmath316 & + + & & 0.00 & 1066.6600 & @xmath160&@xmath226$]&@xmath317&@xmath90&@xmath91 + + & & 0.00 & 2056.2539 & @xmath318&@xmath226$]&@xmath319&@xmath320&@xmath321 + & & 0.00 & 2062.2340 & @xmath322&@xmath323$]&@xmath324&@xmath325 & + & & 0.00 & 2066.1610 & @xmath326&@xmath33$]&@xmath327&@xmath328 & + + & 9/2 & 0.00 & 1608.4511 & @xmath164&@xmath329$]&@xmath330&@xmath331&@xmath332 + & 9/2 & 0.00 & 1611.2005 & @xmath168&@xmath333$]&@xmath334&@xmath335 & + & 7/2 & 384.79 & 1618.4680 & @xmath171&@xmath60$]&@xmath336&@xmath337&@xmath338 + & 7/2 & 384.79 & 1621.6856 & @xmath175&@xmath60$]&@xmath339&@xmath340 & + & 5/2 & 667.68 & 1629.1596 & @xmath178&@xmath60$]&@xmath246&@xmath341&@xmath342 + & 3/2 & 862.61 & 1634.3498 & @xmath182&@xmath60$]&@xmath336&@xmath343&@xmath344 + & 3/2 & 862.61 & 1636.3313 & @xmath186&@xmath60$]&@xmath284&@xmath345 & + & 1/2 & 977.05 & 1639.4012 & @xmath189&@xmath60$]&@xmath346&@xmath347&@xmath348 + + & & 0.00 & 1370.1310 & @xmath196&@xmath349$]&@xmath350&@xmath351&@xmath352 + & & 0.00 & 1703.4050 & @xmath353&@xmath226$]&@xmath354&@xmath355 & + & & 0.00 & 1709.6042 & @xmath356&@xmath226$]&@xmath357&@xmath358 & + & & 0.00 & 1741.5490 & @xmath200&@xmath33$]&@xmath359&@xmath360 & + & & 0.00 & 1751.9157 & @xmath361&@xmath226$]&@xmath362&@xmath363 & + + & & 0.00 & 2026.1360 & @xmath364&@xmath226$]&@xmath365&@xmath366&@xmath367 + & & 0.00 & 2062.6640 & @xmath368&@xmath38$]&@xmath369&@xmath370 & + lccccccccc + & 0 & 0.00 & 1656.9283 & @xmath37&@xmath33$]&@xmath371&@xmath372&@xmath373 + + & 1/2 & 0.00 & 2325.4030 & @xmath374&@xmath33$]&@xmath375&@xmath376&@xmath377 + + & & 0.00 & 1747.7937 & @xmath378&@xmath33$]&@xmath379&@xmath380&@xmath381 + & & 0.00 & 1827.9351 & @xmath249&@xmath33$]&@xmath382&@xmath383 & + & & 0.00 & 2026.4768 & @xmath251&@xmath33$]&@xmath384&@xmath385 & + + & & 0.00 & 2796.3520 & @xmath386&@xmath387$]&@xmath388&@xmath389&@xmath390 + & & 0.00 & 2803.5310 & @xmath391&@xmath387$]&@xmath392&@xmath393 & + + & & 0.00 & 1670.7874 & @xmath99&@xmath394$]&@xmath395&@xmath396&@xmath397 + + & & 0.00 & 1854.7164 & @xmath262&@xmath398$]&@xmath399&@xmath400&@xmath401 + & & 0.00 & 1862.7895 & @xmath266&@xmath398$]&@xmath402&@xmath403 & + + & & 0.00 & 1845.5200 & @xmath269&@xmath33$]&@xmath404&@xmath405&@xmath406 + & & 0.00 & 2515.0730 & @xmath407&@xmath33$]&@xmath209&@xmath408 & + + & 1/2 & 0.00 & 1808.0130 & @xmath117&@xmath122$]&@xmath409&@xmath410&@xmath411 + & 1/2 & 0.00 & 2335.1230 & @xmath412&@xmath33$]&@xmath284&@xmath413 & + & 3/2 & 287.24 & 1816.9285 & @xmath414&@xmath33$]&@xmath415&@xmath416&@xmath417 + + & & 0.00 & 1892.0300 & @xmath291&@xmath33$]&@xmath418&@xmath419&@xmath420 + + & 1/2 & 0.00 & 1910.6000 & @xmath421&@xmath33$]&@xmath422&@xmath405&@xmath406 + + & & 0.00 & 2056.2539 & @xmath318&@xmath33$]&@xmath423&@xmath424&@xmath425 + & & 0.00 & 2062.2340 & @xmath322&@xmath33$]&@xmath426&@xmath427 & + & & 0.00 & 2066.1610 & @xmath326&@xmath33$]&@xmath428&@xmath429 & + + & & 0.00 & 2576.8770 & @xmath430&@xmath33$]&@xmath431&@xmath432&@xmath433 + & & 0.00 & 2594.4990 & @xmath434&@xmath33$]&@xmath435&@xmath436 & + & & 0.00 & 2606.4620 & @xmath437&@xmath33$]&@xmath438&@xmath436 & + + & & 0.00 & 2484.0210 & @xmath439&@xmath33$]&@xmath440&@xmath441&@xmath442 + + & 9/2 & 0.00 & 2249.8768 & @xmath443&@xmath122$]&@xmath444&@xmath445&@xmath446 + & 9/2 & 0.00 & 2260.7805 & @xmath447&@xmath448$]&@xmath449&@xmath450 & + & 9/2 & 0.00 & 2344.2140 & @xmath451&@xmath452$]&@xmath453&@xmath454 & + & 9/2 & 0.00 & 2374.4612 & @xmath455&@xmath394$]&@xmath456&@xmath457 & + & 9/2 & 0.00 & 2382.7650 & @xmath458&@xmath452$]&@xmath459&@xmath460 & + & 9/2 & 0.00 & 2586.6499 & @xmath461&@xmath462$]&@xmath463&@xmath464 & + & 9/2 & 0.00 & 2600.1729 & @xmath465&@xmath466$]&@xmath467&@xmath468 & + & 7/2 & 384.79 & 2333.5160 & @xmath469&@xmath33$]&@xmath470&@xmath471&@xmath472 + & 7/2 & 384.79 & 2365.5520 & @xmath473&@xmath33$]&@xmath474&@xmath475 & + & 7/2 & 384.79 & 2383.7884 & @xmath476&@xmath33$]&@xmath477&@xmath478 & + & 7/2 & 384.79 & 2389.3582 & @xmath479&@xmath145$]&@xmath480&@xmath107 & + & 7/2 & 384.79 & 2396.3559 & @xmath481&@xmath33$]&@xmath482&@xmath483 & + & 7/2 & 384.79 & 2626.4511 & @xmath484&@xmath33$]&@xmath485&@xmath486 & + & 5/2 & 667.68 & 2328.1110 & @xmath487&@xmath33$]&@xmath488&@xmath489&@xmath490 + & 5/2 & 667.68 & 2607.8664 & @xmath491&@xmath33$]&@xmath492&@xmath489 & + & 5/2 & 667.68 & 2618.3991 & @xmath493&@xmath33$]&@xmath494&@xmath489 & + & 3/2 & 862.61 & 2338.7250 & @xmath495&@xmath33$]&@xmath496&@xmath497&@xmath498 + & 3/2 & 862.61 & 2359.8280 & @xmath499&@xmath33$]&@xmath500&@xmath501 & + & 3/2 & 862.61 & 2621.1912 & @xmath502&@xmath33$]&@xmath503&@xmath504 & + & 1/2 & 977.05 & 2345.0010 & @xmath505&@xmath506$]&@xmath507&@xmath508&@xmath509 + & 1/2 & 977.05 & 2414.0450 & @xmath510&@xmath33$]&@xmath511&@xmath512 & + & 1/2 & 977.05 & 2622.4518 & @xmath513&@xmath33$]&@xmath514&@xmath515 & + & 1/2 & 977.05 & 2629.0777 & @xmath516&@xmath33$]&@xmath517&@xmath518 & + + & & 0.00 & 1941.2852 & @xmath519&@xmath33$]&@xmath520&@xmath521&@xmath522 + + & & 0.00 & 1703.4050 & @xmath353&@xmath33$]&@xmath146&@xmath523&@xmath524 + & & 0.00 & 1709.6042 & @xmath356&@xmath33$]&@xmath525&@xmath526 & + & & 0.00 & 1741.5531 & @xmath200&@xmath33$]&@xmath527&@xmath528 & + & & 0.00 & 1751.9157 & @xmath361&@xmath33$]&@xmath529&@xmath530 & + + & & 0.00 & 2026.1360 & @xmath364&@xmath33$]&@xmath531&@xmath532&@xmath533 + & & 0.00 & 2062.6640 & @xmath368&@xmath33$]&@xmath534&@xmath535 & + lccccccccc + & 0 & 0.00 & 1656.9283 & @xmath37&@xmath226$]&@xmath536&@xmath537&@xmath538 + + & 1/2 & 0.00 & 2325.4030 & @xmath374&@xmath226$]&@xmath539&@xmath540&@xmath541 + + & & 0.00 & 1548.1950 & @xmath51&@xmath542$]&@xmath543&@xmath544&@xmath545 + & & 0.00 & 1550.7700 & @xmath56&@xmath542$]&@xmath546&@xmath221 & + + & & 0.00 & 1747.7937 & @xmath378&@xmath33$]&@xmath547&@xmath548&@xmath549 + & & 0.00 & 1827.9351 & @xmath249&@xmath550$]&@xmath551&@xmath552 & + & & 0.00 & 2026.4768 & @xmath251&@xmath553$]&@xmath554&@xmath555 & + & & 0.00 & 2852.9642 & @xmath556&@xmath557$]&@xmath558&@xmath559 & + + & & 0.00 & 2796.3520 & @xmath386&@xmath466$]&@xmath560&@xmath561&@xmath562 + & & 0.00 & 2803.5310 & @xmath391&@xmath466$]&@xmath563&@xmath564 & + + & & 0.00 & 1670.7874 & @xmath99&@xmath394$]&@xmath565&@xmath566&@xmath567 + + & & 0.00 & 1854.7164 & @xmath262&@xmath226$]&@xmath568&@xmath569&@xmath570 + & & 0.00 & 1862.7895 & @xmath266&@xmath226$]&@xmath571&@xmath572 & + + & & 0.00 & 1845.5200 & @xmath269&@xmath33$]&@xmath573&@xmath271&@xmath272 + & & 0.00 & 2515.0730 & @xmath407&@xmath33$]&@xmath574&@xmath575 & + + & 1/2 & 0.00 & 1526.7066 & @xmath113&@xmath542$]&@xmath576&@xmath464&@xmath577 + & 1/2 & 0.00 & 1808.0130 & @xmath117&@xmath226$]&@xmath578&@xmath579 & + & 1/2 & 0.00 & 2335.1230 & @xmath412&@xmath33$]&@xmath193&@xmath580 & + & 3/2 & 287.24 & 1533.4312 & @xmath286&@xmath323$]&@xmath581&@xmath582&@xmath583 + & 3/2 & 287.24 & 1816.9285 & @xmath414&@xmath33$]&@xmath547&@xmath584 & + + & & 0.00 & 1892.0300 & @xmath291&@xmath33$]&@xmath585&@xmath586&@xmath587 + + & 1/2 & 0.00 & 1910.6000 & @xmath421&@xmath33$]&@xmath503&@xmath405&@xmath406 + + & & 0.00 & 2056.2539 & @xmath318&@xmath226$]&@xmath588&@xmath589&@xmath590 + & & 0.00 & 2062.2340 & @xmath322&@xmath226$]&@xmath591&@xmath592 & + & & 0.00 & 2066.1610 & @xmath326&@xmath33$]&@xmath593&@xmath594 & + + & & 0.00 & 2576.8770 & @xmath430&@xmath226$]&@xmath595&@xmath596&@xmath597 + & & 0.00 & 2594.4990 & @xmath434&@xmath226$]&@xmath598&@xmath599 & + & & 0.00 & 2606.4620 & @xmath437&@xmath226$]&@xmath600&@xmath601 & + + & & 0.00 & 1941.2852 & @xmath519&@xmath33$]&@xmath602&@xmath603&@xmath604 + + & & 0.00 & 2484.0210 & @xmath439&@xmath33$]&@xmath605&@xmath606&@xmath607 + + & 9/2 & 0.00 & 1608.4511 & @xmath164&@xmath608$]&@xmath609&@xmath610&@xmath611 + & 9/2 & 0.00 & 1611.2005 & @xmath168&@xmath226$]&@xmath612&@xmath613 & + & 9/2 & 0.00 & 2249.8768 & @xmath443&@xmath226$]&@xmath614&@xmath615 & + & 9/2 & 0.00 & 2260.7805 & @xmath447&@xmath226$]&@xmath616&@xmath617 & + & 9/2 & 0.00 & 2344.2140 & @xmath451&@xmath618$]&@xmath619&@xmath274 & + & 9/2 & 0.00 & 2374.4612 & @xmath455&@xmath557$]&@xmath620&@xmath621 & + & 9/2 & 0.00 & 2382.7650 & @xmath458&@xmath557$]&@xmath622&@xmath623 & + & 9/2 & 0.00 & 2586.6499 & @xmath461&@xmath624$]&@xmath625&@xmath626 & + & 9/2 & 0.00 & 2600.1729 & @xmath465&@xmath627$]&@xmath628&@xmath162 & + & 7/2 & 384.79 & 2333.5160 & @xmath469&@xmath629$]&@xmath630&@xmath631&@xmath632 + & 7/2 & 384.79 & 2383.7884 & @xmath476&@xmath33$]&@xmath633&@xmath634 & + & 7/2 & 384.79 & 2389.3582 & @xmath479&@xmath226$]&@xmath635&@xmath636 & + & 7/2 & 384.79 & 2396.3559 & @xmath481&@xmath637$]&@xmath638&@xmath639 & + & 7/2 & 384.79 & 2626.4511 & @xmath484&@xmath197$]&@xmath640&@xmath641 & + & 5/2 & 667.68 & 2328.1110 & @xmath487&@xmath197$]&@xmath642&@xmath643&@xmath644 + & 5/2 & 667.68 & 2349.0220 & @xmath645&@xmath308$]&@xmath646&@xmath647 & + & 5/2 & 667.68 & 2399.9728 & @xmath648&@xmath197$]&@xmath649&@xmath650 & + & 5/2 & 667.68 & 2607.8664 & @xmath491&@xmath33$]&@xmath651&@xmath497 & + & 5/2 & 667.68 & 2618.3991 & @xmath493&@xmath33$]&@xmath652&@xmath653 & + & 3/2 & 862.61 & 2338.7250 & @xmath495&@xmath127$]&@xmath654&@xmath655&@xmath321 + & 3/2 & 862.61 & 2359.8280 & @xmath499&@xmath33$]&@xmath656&@xmath657 & + & 3/2 & 862.61 & 2411.2533 & @xmath658&@xmath33$]&@xmath659&@xmath660 & + & 3/2 & 862.61 & 2621.1912 & @xmath502&@xmath33$]&@xmath661&@xmath662 & + & 1/2 & 977.05 & 2345.0010 & @xmath505&@xmath506$]&@xmath663&@xmath664&@xmath665 + & 1/2 & 977.05 & 2414.0450 & @xmath510&@xmath33$]&@xmath666&@xmath667 & + & 1/2 & 977.05 & 2622.4518 & @xmath513&@xmath33$]&@xmath668&@xmath669 & + & 1/2 & 977.05 & 2629.0777 & @xmath516&@xmath197$]&@xmath670&@xmath671 & + + & & 0.00 & 1703.4050 & @xmath353&@xmath33$]&@xmath672&@xmath310&@xmath673 + & & 0.00 & 1709.6042 & @xmath356&@xmath33$]&@xmath674&@xmath675 & + & & 0.00 & 1741.5531 & @xmath200&@xmath226$]&@xmath676&@xmath677 & + & & 0.00 & 1751.9157 & @xmath361&@xmath33$]&@xmath678&@xmath679 & + + & & 0.00 & 2026.1360 & @xmath364&@xmath226$]&@xmath680&@xmath681&@xmath682 + & & 0.00 & 2062.6640 & @xmath368&@xmath683$]&@xmath684&@xmath685 & + lcccc @xmath686 & 3.9686 & 2.6147 & 1.5495 & 1.4900 + @xmath687 & @xmath688&@xmath689&@xmath690&@xmath690 + @xmath691m / h]@xmath692&@xmath693&@xmath694&@xmath695&@xmath696 + @xmath697&15.15&14.90&15.36&15.26 + @xmath691c / fe ] & @xmath698&@xmath699&@xmath700&@xmath700 + @xmath691n / fe ] & @xmath701&@xmath702&@xmath700&@xmath700 + @xmath691o / fe ] & @xmath703&@xmath704&@xmath700&@xmath700 + @xmath691mg / fe ] & @xmath705&@xmath706&@xmath707&@xmath708 + @xmath691al / fe ] & @xmath709&@xmath710&@xmath711&@xmath712 + @xmath691si / fe ] & @xmath713&@xmath714&@xmath715&@xmath716 + @xmath691p / fe ] & @xmath717&@xmath718&@xmath700&@xmath700 + @xmath691s / fe ] & @xmath719&@xmath720&@xmath700&@xmath700 + @xmath691ti / fe ] & @xmath700&@xmath700&@xmath721&@xmath722 + @xmath691cr / fe ] & @xmath700&@xmath723&@xmath724&@xmath725 + @xmath691mn / fe ] & @xmath700&@xmath700&@xmath726&@xmath727 + @xmath691co / fe ] & @xmath700&@xmath700&@xmath728&@xmath729 + @xmath691ni / fe ] & @xmath730&@xmath731&@xmath732&@xmath733 + @xmath691zn / fe ] & @xmath700&@xmath734&@xmath735&@xmath736 + @xmath737&@xmath87&@xmath738&@xmath739 + @xmath740&@xmath91&@xmath248&@xmath738 + @xmath741&@xmath125&@xmath283&@xmath417&@xmath583 + @xmath742&@xmath174&@xmath338&@xmath472&@xmath632 + @xmath743&@xmath181&@xmath342&@xmath490&@xmath644 + @xmath744&@xmath185&@xmath344&@xmath498&@xmath321 + @xmath745&@xmath192&@xmath348&@xmath509&@xmath665 + @xmath746&@xmath739&@xmath250&@xmath381&@xmath549 + lcccccc + & 2852.9642 & @xmath556&@xmath747$]&@xmath748&@xmath749&@xmath750 + + & 2796.3520 & @xmath386&@xmath747$]&@xmath751&@xmath752&@xmath753 + & 2803.5310 & @xmath391&@xmath747$]&@xmath754&@xmath753 & + + & 2260.7805 & @xmath447&@xmath747$]&@xmath755&@xmath756&@xmath757 + & 2344.2140 & @xmath451&@xmath747$]&@xmath758&@xmath759 & + & 2374.4612 & @xmath455&@xmath747$]&@xmath760&@xmath761 & + & 2382.7650 & @xmath458&@xmath747$]&@xmath762&@xmath763 & + & 2586.6500 & @xmath461&@xmath747$]&@xmath764&@xmath765 & + & 2600.1729 & @xmath465&@xmath747$]&@xmath766&@xmath767 & + lcccccc + & 2852.9642 & @xmath556&@xmath768$]&@xmath769&@xmath770&@xmath770 + + & 2796.3520 & @xmath386&@xmath768$]&@xmath771&@xmath772&@xmath773 + & 2803.5310 & @xmath391&@xmath774$]&@xmath775&@xmath773 & + + & 2260.7805 & @xmath447&@xmath776$]&@xmath777&@xmath778&@xmath779 + & 2374.4612 & @xmath455&@xmath776$]&@xmath780&@xmath779 & + & 2382.7650 & @xmath458&@xmath776$]&@xmath781&@xmath782 & + & 2586.6500 & @xmath461&@xmath776$]&@xmath783&@xmath163 & + & 2600.1729 & @xmath465&@xmath784$]&@xmath785&@xmath786 & + lcccccc + & 2796.3520 & @xmath386&@xmath787$]&@xmath788&@xmath789&@xmath790 + & 2803.5310 & @xmath391&@xmath787$]&@xmath791&@xmath790 & + + & 3934.7770 & @xmath792&@xmath793$]&@xmath794&@xmath795&@xmath796 + & 3969.5910 & @xmath797&@xmath793$]&@xmath798&@xmath799 & + + & 2344.2140 & @xmath451&@xmath800$]&@xmath801&@xmath802&@xmath802 + & 2382.7650 & @xmath458&@xmath787$]&@xmath803&@xmath804 & + lcccccc + & 2852.9642 & @xmath556&@xmath329$]&@xmath805&@xmath806&@xmath806 + + & 2796.3520 & @xmath386&@xmath807$]&@xmath808&@xmath809&@xmath810 + & 2803.5310 & @xmath391&@xmath811$]&@xmath812&@xmath810 & + + & 2260.7805 & @xmath447&@xmath466$]&@xmath813&@xmath814&@xmath815 + & 2344.2140 & @xmath451&@xmath329$]&@xmath816&@xmath817 & + & 2374.4612 & @xmath455&@xmath329$]&@xmath818&@xmath815 & + & 2382.7650 & @xmath458&@xmath329$]&@xmath819&@xmath820 & + & 2586.6500 & @xmath461&@xmath329$]&@xmath821&@xmath822 & + & 2600.1729 & @xmath465&@xmath329$]&@xmath823&@xmath824 & + lcccccc + & 2852.9652 & @xmath556&@xmath825$]&@xmath826&@xmath827&@xmath827 + + & 2796.3520 & @xmath386&@xmath825$]&@xmath828&@xmath829&@xmath830 + & 2803.5310 & @xmath391&@xmath825$]&@xmath831&@xmath830 & + + & 3242.9290 & @xmath832&@xmath833$]&@xmath834&@xmath835&@xmath836 + + & 2260.7805 & @xmath447&@xmath825$]&@xmath837&@xmath778&@xmath838 + & 2344.2140 & @xmath451&@xmath825$]&@xmath839&@xmath840 & + & 2374.4612 & @xmath455&@xmath825$]&@xmath841&@xmath838 & + & 2382.7650 & @xmath458&@xmath825$]&@xmath842&@xmath822 & + & 2600.1729 & @xmath465&@xmath825$]&@xmath843&@xmath844 & + + & 2026.1360 & @xmath364&@xmath833$]&@xmath845&@xmath846&@xmath846 + lcccccc + & 2852.9652 & @xmath556&@xmath847$]&@xmath848&@xmath849&@xmath849 + + & 2796.3520 & @xmath386&@xmath847$]&@xmath850&@xmath824&@xmath851 + & 2803.5310 & @xmath391&@xmath847$]&@xmath852&@xmath851 & + + & 2576.8770 & @xmath430&@xmath853$]&@xmath854&@xmath855&@xmath855 + + & 2260.7805 & @xmath447&@xmath847$]&@xmath856&@xmath857&@xmath857 + & 2344.2140 & @xmath451&@xmath847$]&@xmath858&@xmath859 & + & 2374.4612 & @xmath455&@xmath847$]&@xmath860&@xmath861 & + & 2382.7650 & @xmath458&@xmath847$]&@xmath862&@xmath863 & + & 2586.6499 & @xmath461&@xmath847$]&@xmath864&@xmath865 & + & 2600.1729 & @xmath465&@xmath847$]&@xmath866&@xmath804 & + lcccccc + & 2852.9652 & @xmath556&@xmath867$]&@xmath868&@xmath869&@xmath869 + + & 2796.3520 & @xmath386&@xmath867$]&@xmath870&@xmath871&@xmath820 + & 2803.5310 & @xmath391&@xmath867$]&@xmath872&@xmath820 & + + & 2260.7805 & @xmath447&@xmath867$]&@xmath873&@xmath874&@xmath875 + & 2344.2140 & @xmath451&@xmath867$]&@xmath876&@xmath877 & + & 2374.4612 & @xmath455&@xmath867$]&@xmath878&@xmath879 & + & 2382.7650 & @xmath458&@xmath867$]&@xmath880&@xmath881 & + & 2586.6499 & @xmath461&@xmath867$]&@xmath882&@xmath883 & + & 2600.1729 & @xmath465&@xmath867$]&@xmath884&@xmath885 & + lccccccccc + & & 0.00 & 2852.9642 & @xmath556&@xmath886$]&@xmath887&@xmath888&@xmath889 + + & & 0.00 & 2796.3520 & @xmath386&@xmath890$]&@xmath891&@xmath260&@xmath892 + & & 0.00 & 2803.5310 & @xmath391&@xmath890$]&@xmath893&@xmath894 & + + & & 0.00 & 1670.7874 & @xmath99&@xmath895$]&@xmath896&@xmath897&@xmath898 + + & & 0.00 & 1854.7164 & @xmath262&@xmath899$]&@xmath900&@xmath901&@xmath902 + & & 0.00 & 1862.7895 & @xmath266&@xmath899$]&@xmath903&@xmath904 & + + & 1/2 & 0.00 & 1808.0130 & @xmath117&@xmath905$]&@xmath906&@xmath907&@xmath908 + + & & 0.00 & 3934.7770 & @xmath792&@xmath226$]&@xmath909&@xmath910&@xmath911 + + & 1/2 & 0.00 & 3242.9290 & @xmath832&@xmath226$]&@xmath912&@xmath913&@xmath914 + & 1/2 & 0.00 & 3384.7400 & @xmath915&@xmath226$]&@xmath916&@xmath917 & + + & & 0.00 & 2056.2539 & @xmath318&@xmath226$]&@xmath918&@xmath919&@xmath920 + + & & 0.00 & 2576.8770 & @xmath430&@xmath226$]&@xmath921&@xmath922&@xmath923 + & & 0.00 & 2594.4990 & @xmath434&@xmath226$]&@xmath924&@xmath925 & + + & 9/2 & 0.00 & 2260.7805 & @xmath447&@xmath226$]&@xmath926&@xmath927&@xmath928 + & 9/2 & 0.00 & 2344.2140 & @xmath451&@xmath929$]&@xmath930&@xmath129 & + & 9/2 & 0.00 & 2374.4612 & @xmath455&@xmath929$]&@xmath931&@xmath932 & + & 9/2 & 0.00 & 2382.7650 & @xmath458&@xmath929$]&@xmath933&@xmath934 & + & 9/2 & 0.00 & 2600.1729 & @xmath465&@xmath929$]&@xmath935&@xmath623 & + + & & 0.00 & 1741.5531 & @xmath200&@xmath226$]&@xmath936&@xmath937&@xmath938 + + & & 0.00 & 2026.1360 & @xmath364&@xmath226$]&@xmath939&@xmath370&@xmath940 + & & 0.00 & 2062.6640 & @xmath368&@xmath941$]&@xmath942&@xmath943 & + lccccccc 050730 & 1.773 & 943.6&@xmath757&@xmath739&@xmath750&18.5&7.8 + 050730 & 2.253 & 886.4&@xmath779&@xmath739&@xmath770&19.0&7.2 + 050820 & 0.692 & 2991.9&@xmath802&@xmath738&19.1&7.8 + 050820 & 1.430 & 1935.7&@xmath815&@xmath739&@xmath806&18.8&8.1 + 051111 & 1.189 & 2027.2&@xmath838&@xmath846&@xmath827&19.0&8.5 + 060418 & 0.603 &
1270.4&@xmath857&@xmath739&@xmath849&20.2&8.2 + 060418 & 0.656 & 972.5&@xmath875&@xmath739&@xmath869&18.4&7.2 + 060418 & 1.107 & 1841.4&@xmath928&@xmath940&@xmath889&20.3&7.5 + | we present optical echelle spectra of four gamma - ray burst ( grb ) afterglows ( grb 050730 , grb 050820 , grb 051111 , and grb 060418 ) discovered during the first 1.5 years of operation of the _ swift _ satellite and localized by either the _ swift _ telescope or follow - up ground - based imaging .
we analyze the spectra to derive accurate column density measurements for the transitions arising in the interstellar medium ( ism ) of the grb host galaxies .
these measurements can be used to constrain the physical properties of the ism including the metallicity , dust - to - gas ratio , ionization state , and chemical abundances of the gas .
we also present measurements of the strong systems in the grb afterglow spectra . with the publication of this paper ,
we provide the first data release of echelle afterglow spectra by the graasp collaboration to the general community . |
quantum chromodynamics , qcd , is the theory of quarks , gluons and their interactions .
it is central to all modern colliders .
and , for the most part , it is what we are made of .
qcd bears a number of similarities to quantum electrodynamics ( qed ) .
just as electrons carry the qed charge , i.e. , electric charge , quarks carry the qcd charge , known as colour charge .
whereas there is only one kind of electric charge , colour charge comes in three varieties , sometimes labelled red , green and blue .
anti - quarks have corresponding anti - colour .
the gluons in qcd are a bit like the photons of qed .
but while photons are electrically neutral , gluons are not colour neutral . they can be thought of as carrying both colour charge and anti - colour charge .
there are eight possible different combinations of ( anti)colour for gluons .
another difference between qcd and qed lies in its coupling @xmath0 . in qcd
it is only moderately small , it tends to zero at high momentum scales ( asymptotic freedom , qed does the opposite ) , it blows up at small scales , and in between its evolution with scale is quite fast : at the lhc its value will range from @xmath1 at a scale of @xmath2 , to @xmath3 at a scale of @xmath4 .
these differences between qcd and qed contribute to making qcd a much richer theory . in these lectures
i will attempt to give you a feel for how qcd works at high momentum scales , and for the variety of techniques used by theorists in order to handle qcd at today s high - energy colliders .
the hope is that these basics will come in useful for day - to - day work with the qcd facets of hadron collider physics . in the fifty or so pages of these lectures
, it will be impossible to give full treatment of any of the topics we will encounter .
for that the reader is referred to any of the classic textbooks about qcd at colliders @xcite .
let us start with a brief reminder of the components of the qcd lagrangian .
this section will be rather dense , but we will return to some of the points in more detail later .
as already mentioned , quarks come in three colours .
so rather than representing them with a single spinor @xmath5 , we will need the spinor to carry also a colour index @xmath6 , which runs from @xmath7 , @xmath8 the quark part of the lagrangian ( for a single flavour ) can be written @xmath9 where the @xmath10 are the usual dirac matrices ; the @xmath11 are gluon fields , with a lorentz index @xmath12 and a colour index @xmath13 that goes from @xmath14 .
quarks are in the fundamental representation of the su(3 ) ( colour ) group , while gluons are in the adjoint representation .
each of the eight gluon fields acts on the quark colour through one of the ` generator ' matrices of the su(3 ) group , the @xmath15 factor in eq .
( [ eq : lquark ] ) .
one convention for writing the matrices is @xmath16 with @xmath17 @xmath18 by looking at the first of these , together with the @xmath19 term of @xmath20 , one can immediately get a feel for what gluons do : a gluon with ( adjoint ) colour index @xmath21 acts on quarks through the matrix @xmath22 .
that matrix takes green quarks ( @xmath23 ) and turns them into red quarks ( @xmath24 ) , and vice versa . in other words , when a gluon interacts with a quark it
_ repaints _ the colour of the quark , taking away one colour and replacing it with another .
the likelihood with which this happens is governed by the strong coupling constant @xmath25 .
note that the repainting analogy is less evident for some of the other colour matrices , but it still remains essentially correct .
the second part of the qcd lagrangian is purely gluonic @xmath26 where the gluon field tensor @xmath27 is given by @xmath28 = i f_{abc}t^c}\,,\ ] ] where the @xmath29 are the structure constants of su(3 ) ( defined through the commutators of the @xmath30 matrices ) .
note the major difference with qed here , namely the presence of a term @xmath31 with two gluon fields .
the presence of such a term is one of the major differences with qed , and , as we will discuss in more detail below , it will be responsible for the fact that gluons interact directly with gluons .
for now , note simply that it has to be there in order for the theory to be gauge invariant under local su(3 ) transformations : @xmath32 where , in the second line , we have dropped the explicit subscript @xmath33 indices , and the @xmath34 are eight arbitrary real functions of the space - time position @xmath35 .
there are two main first - principles approaches to solving qcd : lattice qcd and perturbative qcd .
the most complete approach is lattice qcd .
it involves discretizing space - time , and considering the values of the quark and gluon fields at all the vertices / edges of the resulting 4-dimensional lattice ( with imaginary time ) . through a suitable monte carlo sampling over all possible field configurations ,
one essentially determines the relative likelihood of different field configurations , and this provides a solution to qcd .
this method is particularly suited to the calculation of static quantities in qcd such as the hadron mass spectrum .
the results of such a lattice calculation are illustrated in fig .
[ fig : lattice - spectrum ] , showing very good agreement .
, title="fig : " ] + lattice methods have been successfully used in a range of contexts , for example , in recent years , in helping extract fundamental quantities such as the ckm matrix ( and limits on new physics ) from the vast array of experimental results on hadron decays and oscillations at flavour factories .
unfortunately lattice calculations are nt suitable in all contexts .
let us imagine , briefly , what would be required in order to carry out lattice calculations for lhc physics : since the centre - of - mass energy is ( will be ) @xmath36 , we need a lattice spacing of order @xmath37 to resolve everything that happens .
non - perturbative dynamics for quarks / hadrons near rest takes place on a timescale @xmath38 . but
hadrons at lhc have a boost factor of up to @xmath39 , so the extent of the lattice should be about @xmath40 .
that tells us that if we are to resolve high - momentum transfer interactions and at the same time follow the evolution of quark and gluon fields up to the point where they form hadrons , we would need about @xmath41 lattice units in each direction , of @xmath42 nodes . not to mention the problem with high particle multiplicities ( current lattice calculations seldom involve more than two or three particles ) and all the issues that relate to the use of imaginary time in lattice calculations
of course , that s not to say that it might not be possible , one day , to find clever tricks that would enable lattice calculations to deal with high - energy reactions .
however , with today s methods , any lattice calculation of the properties of lhc proton
proton scattering seems highly unlikely . for this reason
, we will not give any further discussion of lattice qcd here , but instead refer the curious reader to textbooks and reviews for more details @xcite .
perturbative qcd relies on the idea of an order - by - order expansion in a small coupling @xmath43 .
some given observable @xmath44 can then be predicted as @xmath45 where one might calculate just the first one or two terms of the series , with the understanding that remaining ones should be small .
the principal technique to calculate the coefficients @xmath46 of the above series is through the use of feynman diagrammatic ( or other related ) techniques .
the interaction vertices of the qcd feynman rules are shown in fig .
[ fig : feynman - interactions ] ( in some gauges one also needs to consider ghosts , but they will be irrelevant for our discussions here ) . , left ) and triple - gluon ( @xmath29 , right ) vertices of qcd .
these interpretations are only sensible insofar as one imagines that the number of colours in qcd , @xmath47 , is large .
, title="fig : " ] , left ) and triple - gluon ( @xmath29 , right ) vertices of qcd .
these interpretations are only sensible insofar as one imagines that the number of colours in qcd , @xmath47 , is large .
, title="fig : " ] the @xmath48 interaction in fig .
[ fig : feynman - interactions ] comes from the @xmath49 term of the lagrangian .
we have already discussed how the @xmath50 factor affects the colour of the quark , and this is represented in fig . [
fig : colour - flow](left ) , with the gluon taking away one colour and replacing it with another .
the triple - gluon vertex in fig .
[ fig : feynman - interactions ] comes from the @xmath51 part of the lagrangian , via the product of a @xmath52 term in one @xmath53 factor with the @xmath54 term in the other .
it is the fact that gluons carry colour charge that means that they must interact with other gluons . in terms of colour flows
, we have the repetition of the idea that the emission of a gluon can be seen as taking away the colour from the gluon ( or anti - colour ) and replacing it with a different one .
because of the double colour / anti - colour charge of a gluon , one can anticipate that it will interact with ( or emit ) other gluons twice as strongly as does a quark . before coming to mathematical formulation of that statement ,
let s comment also on the 4-gluon vertex of fig .
[ fig : feynman - interactions ] . this comes from the product of two @xmath54 type terms in @xmath55 and is order @xmath56 whereas the two other interactions are order @xmath25 .
though fig .
[ fig : colour - flow ] gives some idea of how the colour factors @xmath50 and @xmath29 in the feynman rules are to be understood , it is useful to see also how they arise in calculations .
after squaring an amplitude and summing over colours of incoming and outgoing particles , they often appear in one or other of the following combinations : [ eq : colour ] @xmath57{0.3\linewidth } \includegraphics[scale=0.7]{busstepp - plots / qcd - tr.eps } \end{minipage } \\
\label{eq : colour : q2qg } \sum_a t^a_{ab }
t^a_{bc } = c_f \delta_{ac}\,,\quad { c_f = \displaystyle\frac{{n_c}^2 - 1}{2{n_c } } = \frac43 } & \qquad \begin{minipage}[b]{0.3\linewidth } \includegraphics[scale=0.7]{busstepp - plots / qcd - cf.eps } \end{minipage } \\
\label{eq : colour : g2gg } \sum_{c , d } f^{acd}f^{bcd } = { c_a}\delta^{ab}\,,\quad { { c_a}= { n_c}= 3 } & \qquad \begin{minipage}[c]{0.3\linewidth } \includegraphics[scale=0.7]{busstepp - plots / qcd - ca.eps } \end{minipage } \\ \label{eq : colour : fierz } t^a_{ab } t^a_{cd } = \displaystyle\frac12 \delta_{bc } \delta_{ad } - \frac1{2{n_c } } \delta_{ab}\delta_{cd}\ , \text { ( fierz ) } & \qquad \begin{minipage}[c]{0.3\linewidth } \includegraphics[scale=0.7]{busstepp - plots / qcd - fierz.eps } \end{minipage } \end{aligned}\ ] ] where @xmath58 is the number of colours in qcd and it is useful to express the results for general numbers of colours ( because it is sometimes useful to consider how results depend on @xmath59 , especially in the limit @xmath60 .
each mathematical combination of colour factors has a diagrammatic interpretation .
equation ( [ eq : colour : g2qq ] ) corresponds to a gluon splitting into @xmath61 which then join back into a gluon ; or , the sum over colours in the squared amplitude for @xmath62 .
equation ( [ eq : colour : q2qg ] ) corresponds to the square of gluon emission from a quark .
equation ( [ eq : colour : g2gg ] ) arises as the square of gluon emission from a gluon .
one sees that there is almost a factor of 2 between eqs .
( [ eq : colour : q2qg ] ) and ( [ eq : colour : g2gg ] ) ( modulo corrections terms @xmath63 ) , which is the mathematical counterpart of our statement above that gluons emit twice as strongly as quarks .
finally the approximate colour - flow interpretation that we had in fig .
[ fig : colour - flow](left ) can be stated exactly in terms of the fierz identity , eq .
( [ eq : colour : fierz ] ) .
most higher - order qcd calculations are carried out with dimensional regularization ( the use of @xmath64 dimensions ) in order to handle the ultraviolet divergences that appear in loop diagrams . in the process of going from @xmath65 to @xmath64 dimensions
, one needs to introduce an arbitrary ` renormalization ' scale , generally called @xmath12 , in order to keep consistent dimensions ( units ) for all quantities .
divergences , is usually carried out in the modified minimal subtraction ( @xmath66 ) scheme ( see , e.g. , section 11.4 of ref .
@xcite ) , by far the most widespread scheme in qcd . ] the value of the qcd coupling , @xmath67 , depends on the scale @xmath12 at which it is evaluated . that dependence can be expressed in terms of a renormalization group equation
@xmath68 where @xmath69 with @xmath70 being the number of ` light ' quark flavours , those whose mass is lower than @xmath12 . the negative sign in eq .
( [ eq : renorm - group ] ) is the origin of asymptotic freedom , the fact that the coupling becomes weaker at high momentum scales , i.e. , the theory almost becomes a free theory , in which quarks and gluons do nt interact .
conversely at low momentum scales the coupling grows strong , causing quarks and gluons to be tightly bound into hadrons .
the importance of the discovery of these features was recognized in the 2004 nobel prize to gross , politzer and wilczek .
why does the qcd @xmath71-function have the opposite sign of that in qed ?
the fact that the vector particles ( gluons ) of the theory carry colour charge is central to the result .
however , while there have been various attempts to give simple but accurate explanations for the negative sign @xcite , in practice they all end up being quite involved .
so , for the purpose of these lectures , let us just accept the results .
if we ignore all terms on the right of eq .
( [ eq : renorm - group ] ) other than @xmath72 , and also ignore the subtlety that the number of ` light ' flavours @xmath70 depends on @xmath12 , then there is a simple solution for @xmath73 : @xmath74 where one can either express the result in terms of the value of the coupling at a reference scale @xmath75 , or in terms of a non - perturbative constant @xmath76 ( also called @xmath77 ) , the scale at which the coupling diverges . only for scales
@xmath78 , corresponding to @xmath79 , is perturbation theory valid .
note that @xmath76 , since it is essentially a non - perturbative quantity , is not too well defined : for a given @xmath80 , its value depends on whether we used just @xmath72 in eq .
( [ eq : renorm - group ] ) or also @xmath81 , etc . however , its order of magnitude , @xmath82 , is physically meaningful insofar as it is closely connected with the scale of hadron masses .
one question that often arises is how @xmath12 , the renormalization scale , should relate to the physical scale of the process .
we will discuss this in detail later ( section [ sec : fixed - order ] ) , but for now the following simple statement is good enough : the strength of the qcd interaction for a process involving a momentum transfer @xmath83 is given by @xmath84 with @xmath85 .
one can measure the strength of that interaction in a range of processes , at various scales , and fig . [
fig : running - coupling ] @xcite shows a compilation of such measurements , together with the running of an average over many measurements , @xmath86 , illustrating the good consistency of the measurements with the expected running . , together with the band obtained by running the world average for @xmath0 within its uncertainties .
figure taken from ref .
@xcite . ]
colliders like the tevatron and the lhc are mainly geared to investigating phenomena involving high - momentum transfers ( more precisely large transverse - momenta ) , say in the range @xmath87 to @xmath2 . there , the qcd coupling is certainly small and we would hope to be able to apply perturbation theory .
yet , the initial state involves protons , at whose mass scale , @xmath88 , the coupling is certainly not weak .
and the final states of collider events consist of lots of hadrons . those are nt perturbative either .
and there are lots of them tens to hundreds .
even if we wanted to try , somehow , to treat them perturbatively , we would be faced with calculations to some very high order in @xmath0 , at least as high as the particle multiplicity , which is far beyond what we can calculate exactly : depending on how you count , at hadron colliders , the best available complete calculation ( i.e. , all diagrams at a given order ) , does nt go beyond @xmath89 or @xmath90 .
certain subsets of diagrams ( e.g. , those without loops ) can be calculated up @xmath91 roughly .
so we are faced with a problem .
exact lattice methods ca nt deal with the high momentum scales that matter , exact perturbative methods ca nt deal with low momentum scales that inevitably enter the problem , nor the high multiplicities that events have in practice . yet
, it turns out that we are reasonably successful in making predictions for collider events .
these lectures will try to give you an understanding of the methods and approximations that are used .
one simple context in which qcd has been extensively studied over the past 30 years is that of @xmath93 annihilation to hadrons .
this process has the theoretical advantage that only the final state involves qcd .
additionally , huge quantities of data have been collected at quite a number of colliders , including millions of events at the @xmath94 mass at lep and slc .
we therefore start our investigation of the properties of qcd by considering this process .
there is one qcd approximation that we will repeatedly make use of , and that is the soft and collinear approximation .
` soft ' implies that an emitted gluon has very little energy compared to the parton ( quark or gluon ) that emitted it . `
collinear ' means that it is emitted very close in angle to another parton in the event . by considering gluons that are soft and/or collinear one
can drastically simplify certain qcd calculations , while still retaining much of the physics .
the soft and collinear approximation is sufficiently important that it s worth , at least once , carrying out a calculation with it , and we ll do that in the context of the emission of a gluon from @xmath95 events .
though there are quite a few equations in the page that follows , the manipulations are all quite simple !
we re interested in the hadronic side of the @xmath96 amplitude , so let s first write the qed matrix element for a virtual photon @xmath97 ( we can always put back the @xmath98 and the photon propagator parts later if we need to which we wo nt ) : @xmath99{0.15{\textwidth } } \includegraphics[width={\textwidth}]{newfigs / qcd - gamma2qq.eps } \end{minipage}\,,\ ] ] where the diagram illustrates the momentum labelling . here @xmath100 and @xmath101 are the spinors for the outgoing quark and anti - quark ( taken massless ) , @xmath102 is the quark s electric charge and the @xmath103 are the dirac matrices . in what follows
we shall drop the @xmath104 quark colour indices for compactness and reintroduce them only at the end .
the corresponding amplitude including the emission of a gluon with momentum @xmath105 and polarization vector @xmath106 is [ eq : mqqg - start ] @xmath107{0.15\linewidth } \includegraphics[width={\textwidth}]{newfigs / qcd - gamma2qqg01.eps } \end{minipage } \quad+ \quad \begin{minipage}[c]{0.15\linewidth } \includegraphics[width={\textwidth}]{newfigs / qcd - gamma2qqg02.eps } \end{minipage } \\ & = -\bar u(p_1 ) ig_s { \slashii{\epsilon}}t^a \frac{i({\slashii{p}}_1+{\slashi{k}})}{(p_1+k)^2 } i e_q \gamma_\mu v(p_2 ) + \bar u(p_1 ) i e_q \gamma_\mu \frac{i({\slashii{p}}_2+{\slashi{k}})}{(p_2+k)^2 } ig_s { \slashii{\epsilon}}t^a v(p_2)\ , , \end{aligned}\ ] ] with one term for emission from the quark and the other for emission from the anti - quark and use of the notation @xmath108 .
let s concentrate on the first term , collecting the factors of @xmath109 , and using the anti - commutation relation of the @xmath110-matrices , @xmath111 , to write @xmath112}{(p_1+k)^2 } e_q \gamma_\mu t^a v(p_2)\ , , \\ & \simeq i g_s\frac { p_1.\epsilon } { p_1.k}\ , \bar u(p_1 ) e_q \gamma_\mu t^a v(p_2)\ , , \end{aligned}\ ] ] where to obtain the second line we have made use of the fact that @xmath113 , @xmath114 , and taken the soft approximation @xmath115 , which allows us to neglect the terms in the numerator that are proportional to @xmath105 rather than @xmath116 . the answer including both terms in eq .
( [ eq : mqqg - start ] ) is @xmath117 where the first factor has the lorentz structure of the @xmath118 amplitude , i.e. , apart from the colour matrix @xmath119 , @xmath118 is simply proportional to the @xmath120 result .
we actually need the squared amplitude , summed over polarizations and colour states , @xmath121 we have now explicitly written the quark colour indices @xmath104 again . to obtain the second line we have made use of the result that @xmath122 [ cf .
( [ eq : colour : q2qg ] ) ] , whereas for @xmath123 we have @xmath124 . to carry out the sum over gluon polarizations we have exploited the fact that @xmath125 , plus terms proportional to @xmath126 and @xmath127 that disappear when dotted with the amplitude and its complex conjugate
. one main point of the result here is that in the soft limit , the @xmath128 squared matrix element _ factorizes _ into two terms : the @xmath129 matrix element and a piece with a rather simple dependence on the gluon momentum .
the next ingredient that we need is the the phase space for the @xmath130 system , @xmath131 . in the soft approximation , we can write this @xmath132 where @xmath133 is the energy of the gluon @xmath105 .
we see that the phase space also factorizes .
thus we can write the full differential cross section for @xmath61 production plus soft gluon emission as the @xmath61 production matrix element and phase space , @xmath134 , multiplied by a soft gluon emission probability @xmath135 , @xmath136 with @xmath137 where we have used @xmath138 , expressing the result in terms of the polar ( @xmath139 ) and azimuthal ( @xmath140 ) angles of the gluon with respect to the quark ( which itself is back - to - back with the antiquark , since we work in the centre - of - mass frame and there is negligible recoil from the soft gluon ) . with a little more algebra , we get our final result for the probability of soft gluon emission from the @xmath61 system @xmath141 this result has two types of non - integrable divergence : one , called the soft ( or infrared ) divergence when @xmath142 and the other , a collinear divergence , when @xmath143 ( or @xmath144 ) , i.e. , the gluon becomes collinear with the quark ( or antiquark ) direction .
though derived here in the specific context of @xmath145 production , these soft and collinear divergences are a very general property of qcd and appear whenever a gluon is emitted from a quark , regardless of the process . if we want to calculate the @xmath146 corrections to the total cross section , the diagrams included in eq .
( [ eq : mqqg - start ] ) are not sufficient .
we also need to include a one - loop correction ( ` virtual ' ) , specifically , the interference between one - loop @xmath97 diagrams and the tree - level @xmath147 amplitude , for example a contribution such as @xmath148{newfigs / qcd - qq-1loop}\ ] ] which has the same perturbative order ( number of @xmath25 factors ) as the square of eq .
( [ eq : mqqg - start ] ) . the total cross section for
the production of hadrons must be finite .
the integral over the gluon - emission correction has two non - integrable , logarithmic divergences .
these divergences must therefore somehow be cancelled by corresponding divergences in the virtual term .
this is the requirement of _ unitarity _ , which is basically the statement that probability of anything happening must add up to @xmath149 . the most straightforward way of doing the full calculation for the total cross section is to use dimensional regularization in the phase space integral for the real emission diagram and for the integration over the loop momentum in the virtual diagram .
however , in order just to visualize what is happening one can also write @xmath150 where the first term , @xmath149 , is the ` born ' term , i.e. , the production of just @xmath61 , the second term is the real emission term , and the third term is the loop correction .
since we need to integrate gluon emission beyond the soft and collinear region , we have introduced a function @xmath151 , which parametrizes the deviation of the matrix - element from its soft limit when @xmath152 , with @xmath83 the centre - of - mass energy of the process .
@xmath153 has the property @xmath154 .
we have written the virtual term in a similar form , using @xmath155 to parametrize its structure ( we cheat a bit , since the loop momentum integral includes regions of phase space where the gluon is offshell ; this wo nt matter though for us here ) .
the statement that real and virtual divergences cancel means that @xmath156 should be identical to @xmath153 in the soft or collinear limits @xmath157 thus the corrections to the total cross section come from the region of hard ( @xmath158 ) , large - angle gluons ( for which perturbation theory is valid )
. there s a good reason for this : soft and collinear emission takes place on a time - scale @xmath159 that is long compared to that , @xmath160 , for the production of the @xmath61 pair from the virtual photon .
anything that happens long after the production of the @xmath61 pair can not change the fact that there will be a qcd final state ( though it can change the properties of that final state ) , and so it does not affect the total cross section . similarly , whatever dynamics is involved in effecting the transition between partons and hadrons is also expected to occur on a long time - scale ( @xmath161 ) and so should not modify the total cross section .
this is important because it allows us to neglect the issue that we can not directly compute the properties of hadron production .
the fact that the corrections to the total cross section are dominated by a region of hard gluon emission is reflected in a reasonable behaviour for the perturbative series @xmath162 where we have expressed the result in terms of @xmath0 evaluated at a renormalization @xmath163 and the coefficients correspond to @xmath164 ( for all known terms in the series , including electroweak effects , see refs .
@xcite as well as references therein ) . as a first step towards asking questions about the final state
, our next exercise is to attempt to determine the mean number of gluons that are emitted from a quark with energy @xmath165 .
if the emission probability is small ( @xmath166 ) then to first order in the coupling the mean number of emitted gluons is equal to the probability of emitting one gluon @xmath167 the integral diverges for @xmath168 and @xmath143 , however , we can reasonably argue that the divergent structure only holds as long as perturbation theory is valid .
this motivates us to cut the divergences off at a scale @xmath169 , because below that scale the language of quarks and gluons loses its meaning . that immediately tells us that we should have @xmath170 , but it s not so immediately clear how the @xmath139 integral will be cut off .
it turns out , for reasons to do with invariance of the small - angle emission pattern as one boosts the quark in the ( longitudinal ) direction of its motion , that the correct variable to cut on is _ transverse momentum _ ,
we therefore find , to first order in the coupling , @xmath172 where we have explicitly kept track only of the term with the largest number of logarithms .
if we take @xmath173 , how big is this result ?
we have to decide on the scale for @xmath0 .
being hopelessly optimistic , i.e. , taking @xmath174 gives us @xmath175 which , numerically , corresponds to @xmath176 .
this is neither small numerically , nor parametrically ( @xmath177 ) .
does this render perturbation completely useless for anything other than total cross sections ?
we can follow two possible avenues to help answer this question .
one approach is to calculate the next order , and see what structure it has .
alternatively we can ask whether there are final - state observables that have a better - behaved perturbative series than ` the mean number of gluons ' .
once one gluon has been emitted , it can itself emit further gluons .
to understand what the gluon multiplicity might look like to higher orders , it s useful to write down the general pattern of emission of a soft gluon both from a quark and from a gluon , which is essentially independent of the process that produced the ` emitter ' : [ eq : fs - universal - splitting ] @xmath178{0.34\linewidth } \includegraphics[width=0.8{\textwidth}]{figs / qcd - div - quark } \end{minipage } & \simeq\quad \frac{2{\alpha_\mathrm{s}}{c_f}}{\pi } \frac{de}{e } \frac{d\theta}{\theta}\ , , \\
\label{eq : fs - gluon - splitting } \begin{minipage}[c]{0.34\linewidth } \includegraphics[width=0.8{\textwidth}]{figs / qcd - div - gluon } \end{minipage } & \simeq\quad \frac{2{\alpha_\mathrm{s}}{c_a}}{\pi } \frac{de}{e } \frac{d\theta}{\theta}\ , .
\end{aligned}\ ] ] these expressions are valid when the emitted gluon is much lower in energy than the emitter , @xmath179 , and when the emission angle @xmath139 is much smaller than the angle between the emitter and any other parton in the event ( this is known as the condition of angular ordering @xcite ) .
the structure of emission of a soft gluon is almost identical from a quark and from a gluon , except for the substitution of the @xmath180 colour factor in the quark case with the @xmath181 colour factor in the gluon case .
a ) event , with first a single gluon ( a ) , then multiple emissions of gluons both from the @xmath61 pair and from the previously emitted gluons ( b ) , followed by some process , ` hadronization ' , that causes hadrons to be produced from the gluons , giving an event ( c ) , that structurally resembles a real event ( d ) ( @xmath182hadrons at lep in the opal detector),title="fig : " ] b ) event , with first a single gluon ( a ) , then multiple emissions of gluons both from the @xmath61 pair and from the previously emitted gluons ( b ) , followed by some process , ` hadronization ' , that causes hadrons to be produced from the gluons , giving an event ( c ) , that structurally resembles a real event ( d ) ( @xmath182hadrons at lep in the opal detector),title="fig : " ] c ) event , with first a single gluon ( a ) , then multiple emissions of gluons both from the @xmath61 pair and from the previously emitted gluons ( b ) , followed by some process , ` hadronization ' , that causes hadrons to be produced from the gluons , giving an event ( c ) , that structurally resembles a real event ( d ) ( @xmath182hadrons at lep in the opal detector),title="fig : " ] d ) event , with first a single gluon ( a ) , then multiple emissions of gluons both from the @xmath61 pair and from the previously emitted gluons ( b ) , followed by some process , ` hadronization ' , that causes hadrons to be produced from the gluons , giving an event ( c ) , that structurally resembles a real event ( d ) ( @xmath182hadrons at lep in the opal detector),title="fig : " ] since quarks and gluons emit in similar ways , every gluon that is emitted from the quark can itself emit further gluons , and so forth
. most of the emissions will either be in almost the same direction as the original quark ( due to the collinear divergence ) and/or be soft .
this is represented in figs .
[ fig : soft - coll - emission - v - event](a ) and ( b ) ( for simplicity we ve not shown gluons splitting to @xmath61 pairs , which also occurs , with just a collinear divergence ) .
this still only gives a description of events in terms of quarks and gluons , whereas real events consist of hadrons .
though hadronization , the transition from quarks and gluon to hadrons is not something that we know how to calculate from first principles , one idea that has had some success is local parton hadron duality ( lphd ) ( see , e.g. , ref .
it states that after accounting for all gluon and quark production down to scales @xmath183 , the transition from partons to hadrons is essentially local in phase space .
thus the hadron directions and momenta will be closely related to the partons , and the hadron multiplicity will reflect the parton multiplicity too .
this is illustrated in fig .
[ fig : soft - coll - emission - v - event](c ) , comparing it also to the picture of a real event , fig .
[ fig : soft - coll - emission - v - event](d ) .
the latter illustrates how the hadrons do tend to have the same collimated angular distribution as is predicted for gluons , with the small number of exceptions having low energy ( i.e. , soft ) as can be seen from the larger curvature in the experiment s magnetic field .
this comparison with a single event is suggestive that our picture of gluon emission and hadronization might be reasonable .
a more quantitative test can be obtained by calculating the number of emitted gluons .
this requires the extension of eqs .
( [ eq : ng - integral])([eq : ng - result - running ] ) to multiple gluon emission .
the full calculation does nt fit into the space available for these lectures ( see instead textbook discussions in refs .
@xcite ) , but the basic idea is that there are terms @xmath184 for all orders @xmath185 and that one can calculate their coefficients analytically .
the structure of the result is @xmath186 where we ve neglected to write the prefactor in front of the exponential , and we ve also not given the subleading terms @xcite .
hadrons events , comparing the experimental data at a range of centre - of - mass energies @xmath83 , with the qcd prediction using a fitted normalisation and non - perturbative scale @xmath76 .
figure adapted from ref .
@xcite.,title="fig : " ] + how is eq .
( [ eq : ng - all - orders ] ) to be related to the hadron multiplicity ?
the simplest assumption is that each final parton gives some ( unknown ) fixed number of hadrons which must be fitted to data .
equation ( [ eq : ng - all - orders ] ) then predicts not the total hadron multiplicity but its energy dependence .
that prediction is illustrated in fig .
[ fig : hadron - multiplicity ] and shows remarkable agreement with data over a range of energies , providing strong evidence that the picture outlined above is a fair reflection of ` reality ' .
the above approach can be extended to calculate other properties of events such as the energy spectrum of hadrons , the fluctuations in the number of hadrons , and even correlations between hadrons , generally with reasonable success . however , as one probes more detailed features of events , the analytical calculations become significantly more complicated and one also becomes increasingly sensitive to the oversimplicity of the lphd concept . having
said that , the same ideas that we are using , i.e. , the importance of multiple soft and collinear splitting together with a transition from partons to hadrons , are , in numerical form , at the base of widely used monte carlo parton - shower event generators like pythia , herwig and sherpa .
we will discuss them in more detail in section [ sec : mc ] .
it is heartening that the above soft - collinear discussion gave such a good description of the data .
however , it did involve the application of perturbation theory to kinematic regions where its validity is questionable , the need to calculate dominant contributions at all orders in @xmath0 , and the introduction of a free parameter to ` fudge ' the fact that we do nt understand the non - perturbative physics .
a natural question is therefore whether one can formulate final - state observables for which these problems are not present .
the answer is that one can . for an observable to be
calculate based on just one or two orders of perturbation theory it should be infrared and collinear ( irc ) safe . in the words of ref .
@xcite : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for an observable s distribution to be calculable in [ fixed - order ] perturbation theory , the observable should be infra - red safe , i.e. insensitive to the emission of soft or collinear gluons
. in particular if @xmath187 is any momentum occurring in its definition , it must be invariant under the branching @xmath188 whenever @xmath189 and @xmath190 are parallel [ collinear ] or one of them is small [ infrared ] .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for example , the multiplicity of gluons is not irc safe , because it is modified by soft and collinear splitting .
the energy of the hardest particle in an event is not irc safe , because it is modified by collinear splitting .
however , the total energy flowing into a given cone is irc safe , because soft emissions do nt modify the energy flow , and collinear emissions do nt modify its direction .
this last example comes from sterman and weinberg @xcite , who defined an @xmath93 event as having 2 ` jets ' if at least a fraction @xmath191 of the event s energy is contained in two cones of half - angle @xmath192 . if we take @xmath192 to be @xmath193 and @xmath194 , then fig .
[ fig : soft - coll - emission - v - event](d ) is an example of such a 2-jet event .
we can adapt our expression for the total cross section , eq .
( [ eq : tot - xsct - series ] ) , to give us the 2-jet cross section as follows @xmath195 \right)\,.\ ] ] for small @xmath196 or @xmath139 this is just like the total cross section , with full cancellation of divergences between real and virtual terms [ cf .
( [ eq : vr - properties ] ) ] . for large @xmath196 and large @xmath139 a _ finite _ piece of real - emission cross section
is cut out by the factor @xmath197 and it corresponds to scales with @xmath158 and large angles , for which perturbation theory is valid .
this then gives @xmath198 where @xmath199 , @xmath200 , etc .
are all of order @xmath149 ( as long as @xmath106 and @xmath192 were not taken too small ) .
similarly one could define a 3-jet cross section by requiring that it not be a 2-jet event and that all but a fraction @xmath106 of the energy be contained in 3 cones of half angle @xmath192 .
this would give a cross section of the form @xmath201 where , again , the coefficients are all @xmath202 .
so whereas the cross section for getting an extra gluon is divergent , the cross section for an extra jet is finite and small , @xmath146 .
one difficulty with the extension of the sterman
weinberg definition to multiple jets is to know how to place the cones .
since jet - finding is a well - developed subject in its own right , we will return to in detail in section [ sec : jets ] . the sterman
weinberg jet cross section gives a discrete classification of events : an event either has two jets , or more .
an alternative class of infrared and collinear safe observables is that of event shapes , which give a continuous classification of events .
the most widely studied example is the thrust , @xmath203 , @xmath204 where the sum runs over all particles , and one chooses the thrust axis @xmath205 ( a 3-dimensional unit vector ) so as to maximize the projection in the numerator . for a perfectly collimated @xmath206-jet event
, the thrust axis aligns with the jet axes and the thrust is @xmath149 . for a ` mercedes ' type event with three identical collimated jets ,
the thrust axis will be aligned with any one of the three jets and the thrust will be @xmath207 .
intermediate events will have intermediate values of the thrust .
one application of the thrust variable is given in fig .
[ fig : vector - scalar - thrust ] .
it shows data for the thrust distribution from lep , compared to @xmath146 calculations of the thrust distribution in qcd and in a variant of qcd in which the gluon is a scalar particle rather than a vector particle .
the scalar gluon case does not have a divergence for soft emission ( the collinear divergence is still there ) , with the result is that the distribution diverges less strongly in the 2-jet limit than for vector gluons .
the data clearly prefer the vector - gluon case , though they do also show the need for higher - order corrections at thrust values close to @xmath207 and to @xmath149 .
more generally , event shapes like the thrust have seen broad use in measurements of the qcd coupling , tuning of monte carlo event generators ( see section [ sec : mc ] ) , studies of the hadronization process , and also as an event - topology discriminant in searches for decays of particles beyond the standard model . the @xmath208 process has allowed us to examine many of the basic ideas of perturbative qcd : soft and collinear divergences , the question of which observables are perturbatively calculable or not ( related to infrared and collinear safety ) and even what happens if one takes perturbation theory seriously outside its strict domain of applicability ( one acquires a rough understanding of the collimated , high - multiplicity structure of real events ) .
having considered processes that involve hadrons in the final state , let us now examine what happens when they are present in the initial state .
the importance of understanding initial - state hadrons is obvious at the lhc . within the ` parton model ' , we write , for example , the hadron collider cross section to produce a @xmath94 and a higgs boson as @xmath209{0.35\linewidth } \includegraphics[width={\textwidth}]{busstepp - plots / hh - collisions01.eps } \end{minipage } \ ] ] where @xmath210 is the squared @xmath211 centre - of - mass energy
, @xmath212 is the number density of quarks of type @xmath213 carrying a fraction @xmath214 of the momentum of the first proton , and similarly with @xmath215 for the other proton .
the @xmath216 functions are known as ` parton distribution functions ' ( pdfs ) .
they multiply the ` hard ' ( here , electroweak ) cross section , @xmath217 for the process @xmath218 , a function of the squared partonic ( @xmath61 ) centre - of - mass energy , @xmath219 . after integrating over @xmath214 and @xmath220 ( and summing over quark species ) ,
one obtains the total cross section for @xmath221 .
the above form seems quite intuitive , but still leaves a number of questions : for example , how do we determine the momentum distributions of quarks and gluons inside the proton ?
how does the ` factorization ' into pdfs and a hard part stand up to qcd corrections ?
understanding these questions is crucial if , one day , we are to take measured cross sections for @xmath222 and interpret them , for example , in terms of the coupling of the higgs to the @xmath94 . and
they re just as crucial for practically any other physics analysis we might want to carry out at the lhc .
the parton distribution functions are properties of the ( non - perturbative ) proton .
a natural first question is whether we can calculate the pdfs based on lattice qcd . in principle , yes , and there is ongoing work in this direction ( see , e.g. , ref .
@xcite ) , however , currently lattice qcd has not reached an accuracy in these calculations that is competitive with the combination of experimental measurements and perturbative qcd analyses that we discuss below .
the process where we have learnt the most about pdfs is deep inelastic scattering ( dis ) , i.e. , lepton
proton scattering in which the photon that is exchanged between lepton and proton has a large virtuality .
the kinematics of the `` quark - parton - model '' dis process is represented in fig .
[ fig : dis ] ( left ) and an event from the h1 detector at hera is shown on the right .
kinematic variables that are usually defined are @xmath223 where @xmath224 is the photon virtuality , @xmath35 is the longitudinal momentum fraction of the struck quark in the proton , and @xmath225 is the momentum fraction lost by the electron ( in the proton rest frame ) . to zeroth order in @xmath0 ( the ` quark parton model ' ) , the dis cross section can be written as @xmath226 written in terms of the @xmath227 structure function , which , to zeroth order in @xmath0 is given by @xmath228 ( the `` @xmath229 '' superscript is a reminder that we re only considering the electromagnetic part here ) .
given the sum over all flavours in eq .
( [ eq : f2em ] ) , disentangling the information about individual flavours might seem like quite a daunting task . we can attempt to see where the information comes from by starting off with the assumption that the proton consists just of up and down quarks , in which case @xmath230 where we have introduced the shorthand @xmath231 , etc .
( later we will drop the `` @xmath116 '' subscript altogether ) . in eq .
( [ eq : f2ud ] ) we now have a linear combination of just two functions . the next step is to use isospin symmetry , the fact that the neutron is essentially just a proton with @xmath232 , i.e. , @xmath233 ( ignoring small electromagnetic effects ) , so that @xmath234 appropriate linear combination of @xmath235 and @xmath236 ( in practice one uses deuterons , or nuclei as a source of neutrons ) therefore provides separate information on @xmath237 and @xmath238 .
[ br][br]@xmath239``@xmath240 '' [ br][br]@xmath241``@xmath242 '' data @xcite for protons and deuterons so as to obtain @xmath243 and @xmath244 , assuming only @xmath245 and @xmath246 quarks inside the proton , together with the expectations from the same linear combination based on the cteq6d pdf parametrizations @xcite .
right : results for different valence and sea quark distributions from cteq6d at @xmath247 .
, title="fig : " ] data @xcite for protons and deuterons so as to obtain @xmath243 and @xmath244 , assuming only @xmath245 and @xmath246 quarks inside the proton , together with the expectations from the same linear combination based on the cteq6d pdf parametrizations @xcite .
right : results for different valence and sea quark distributions from cteq6d at @xmath247 . ]
the results of this exercise are shown in fig .
[ fig : nmc - data ] ( left ) .
as expected we see more up quarks than down quarks .
but there s also a problem : let s try to extract the total number of up quarks , @xmath248 .
we see that the data increase towards small @xmath35 , and a parametrization ( cteq6d @xcite ) of these and other data even seems to diverge as @xmath249 , @xmath250 .
is common to essentially all parametrizations . ] given that the plot is supposed to be for @xmath251 and that we need to integrate @xmath252 to get the total number of up quarks in the proton , it looks like we ll obtain an infinite number of up quarks , which is hardly consistent with expectations from the picture of a proton as being a @xmath253 state .
one thing we ve ` neglected ' is that there can also be anti - up and anti - down quarks in the proton , because the proton wavefunction can fluctuate , creating @xmath254 and @xmath255 pairs , ` sea quarks ' , and so give rise to @xmath256 and @xmath257 distributions .
therefore instead of eq .
( [ eq : f2ud ] ) , we should have written @xmath258 since quarks and antiquarks have identical squared charges .
so what we called `` @xmath251 '' in fig .
[ fig : nmc - data](left ) was actually @xmath259 ( with some admixture of strange and charm quarks too ) .
the infinite number of quarks and antiquarks can then just be interpreted as saying that fluctuations within the proton create infinite numbers of @xmath61 pairs , mostly carrying a small fraction of the proton s momentum .
returning to the statement that the proton has 2 up and 1 down quark , what we mean is that the net number of up minus anti - up quarks is 2 , @xmath260 where @xmath261 is also called the valence quark distribution @xmath262 .
how can we measure the difference between quarks and antiquarks ?
the answer is through charged - current processes ( e.g. , neutrino scattering ) , since a @xmath263 interacts only with @xmath246 and @xmath264 , but not with @xmath265 or @xmath245 .
we could imagine forming linear combinations of data sets with proton and nuclear targets , with photon and @xmath266 exchange , etc .
, in order to obtain the different quark - flavour pdfs . in practice
it is simpler to introduce parametrizations for each flavour , deduce predictions for dis , neutrino - scattering and other cross sections , and then fit the parameters so as to obtain agreement with the experimental data .
this is known as a ` global fit ' .
we will return to such fits below , but for now let s just look at the results for different quark distributions , shown in fig .
[ fig : nmc - data ] ( right ) .
we see that the valence quark distributions are mainly situated at moderate @xmath35 values . for @xmath267
they fall off as a moderate power of @xmath268 , roughly as @xmath269 .
this fall - off comes about because to have one @xmath245 quark carrying most of the proton momentum implies that the other @xmath245 and the @xmath246 quark must both carry only a small fraction , and the phase space for that to occur vanishes as @xmath270 ( the proper treatment for this is through ` quark - counting rules ' ) . for @xmath271 , @xmath272 , which is related to something called regge physics .
the sea - quark distributions ( @xmath273 , etc . )
are concentrated at small @xmath35 values .
they fall off much more steeply at large @xmath35 : for a @xmath274 quark to have a large momentum fraction , 3 @xmath245-quarks and one @xmath246-quark must have correspondingly small-@xmath35 , so the phase space is even more restricted than in the case of valence quarks . for @xmath249 , the exact behaviour is not entirely simple , but as a rough rule of thumb @xmath275 with @xmath276 0.20.4 .
of course we re still missing the pdfs for one kind of parton : gluons .
well known fact that if we evaluate the fraction of the proton s momentum carried by all the quarks , @xmath277 the result is about @xmath278 .
it s fair to suspect that the gluon is responsible for the other half , but how are we to establish its shape given that it s not directly probed by photons or @xmath266 ? to answer that question
we must go beyond the ` naive ' leading - order partonic picture of the proton s quarks interacting with probes , and bring in qcd splitting . in eq .
( [ eq : fs - quark - splitting ] ) we wrote the universal form for the final - state ` splitting ' of a quark into a quark and a soft gluon .
let s rewrite it with different kinematic variables , considering a hard process @xmath279 with cross section @xmath280 , and examining the cross section for @xmath279 with an extra gluon in the final state , @xmath281 .
we have @xmath282{0.21{\textwidth } } \includegraphics[width={\textwidth}]{newfigs / qcd - div - final - lab.eps } \end{minipage } \qquad\qquad \sigma_{h+g } \simeq \sigma_{h } \frac{{\alpha_\mathrm{s}}{c_f}}{\pi } \frac{dz}{1-z } \frac{dk_t^2}{k_t^2}\,,\ ] ] where @xmath196 in eq .
( [ eq : fs - quark - splitting ] ) corresponds to @xmath283 and we ve introduced @xmath284 . if we avoid distinguishing a collinear @xmath285 pair from a plain quark ( measurements with irc safe observables ) then , as we argued before , the divergent part of the gluon emission contribution always cancels with a related virtual correction @xmath282{0.21{\textwidth } } \includegraphics[width={\textwidth}]{busstepp - pdf / qcd - div - final - virt.eps } \end{minipage } \qquad\qquad \sigma_{h+v } \simeq -\sigma_{h } \frac{{\alpha_\mathrm{s}}{c_f}}{\pi } \frac{dz}{1-z } \frac{dk_t^2}{k_t^2}\,.\ ] ] now let us examine what happens for initial - state splitting , where the hard process occurs _ after _ the splitting and the momentum entering the hard process is modified @xmath286 : @xmath282{0.21{\textwidth } } \includegraphics[width={\textwidth}]{busstepp - pdf / qcd - div - init - real.eps } \end{minipage } \qquad\qquad \sigma_{g+h}(p ) \simeq \sigma_{h}(zp ) \frac{{\alpha_\mathrm{s}}{c_f}}{\pi } \frac{dz}{1-z } \frac{dk_t^2}{k_t^2}\,,\ ] ] where we have made explicit the hard process s dependence on the incoming momentum , and we assume that @xmath280 involves momentum transfers @xmath287 , so that we can ignore the extra transverse momentum entering @xmath280 . for virtual terms , the momentum entering the process is unchanged , so we have @xmath282{0.21{\textwidth } } \includegraphics[width={\textwidth}]{busstepp - pdf / qcd - div - init - virt.eps } \end{minipage } \qquad\qquad \sigma_{g+h}(p ) \simeq -\sigma_{h}(p ) \frac{{\alpha_\mathrm{s}}{c_f}}{\pi } \frac{dz}{1-z } \frac{dk_t^2}{k_t^2}\,,\ ] ] the total cross section then gets contributions with two different hard cross sections : @xmath288}_{\text{finite}}\,.\ ] ] note the limits on the integrals , in particular the @xmath224 upper limit on the transverse - momentum integration : the approximations we re using are valid as long as the transverse momentum emitted in the initial state is much smaller than the momentum transfers @xmath83 that are present in the hard process . of the two integrations in eq .
( [ eq : total - two - different - hard ] ) , the one over @xmath289 is finite , because in the region of the soft divergence , @xmath290 , the difference of hard cross sections , @xmath291 $ ] , tends to zero . in contrast , the @xmath292 integral diverges in the collinear limit : the cross section with an incoming parton ( and virtual corrections ) appears not to be collinear safe .
this is a general feature of processes with incoming partons : so how are we then to carry out calculations with initial - state hadrons ? in section [ sec : gluon - multiplicity ] , when trying to make sense of final - state divergences , we introduced a ( non - perturbative ) cutoff .
let s do something similar here , with a cutoff , @xmath293 , called a factorization scale ( which will usually be taken at perturbative scales ) . the main idea in using this cutoff is that any emissions that occur with @xmath294 are absorbed ( ` factorized ' ) into the pdf itself . thus the pdfs become a function of @xmath293 .
we can write the lowest order term for the cross section , @xmath295 , and the correction with one initial - state emission or virtual loop , @xmath296 , as follows + [ eq : sigma - with - factorization ] @xmath297\ ; q({x } , { { \mu_\text{\sc f}}^2})}_{\mathsf{finite}}\ , , \end{aligned}\ ] ] + where we have now included also the integral over the longitudinal momentum fraction @xmath35 of the parton extracted from the proton .
the emissions and virtual corrections with @xmath294 are now implicitly included inside the @xmath298 pdf factor that appears in the @xmath295 contribution , and only those with @xmath299 arise explicitly in the @xmath146 term .
this term ( whose real - emission part is represented in the diagram to the left ) is now finite , albeit potentially large if @xmath300 .
this situation of having a non - integrable divergence that somehow needs to be regularized and absorbed with a scale choice into some ` constant ' of the theory ( here the pdfs ) , is reminiscent of renormalization for the coupling constant .
the differences are that here we are faced with a divergence in the ( collinear ) infrared rather than one in the ultraviolet . and that unlike the coupling , the pdfs are not fundamental parameters of the theory , but rather quantities that we could calculate if only we had sufficiently sophisticated theoretical ` technology ' . nevertheless , as for the coupling , the freedom in choosing the scale that enters the regularization , called the factorization scale , implies the presence of a renormalization group equation for the pdfs , the dokshitzer
gribov lipatov
altarelli parisi ( dglap ) equation . to see what form the dglap equation takes ,
let us fix the longitudinal momentum of the quark entering the hard process to be @xmath301 ( whereas above we d fixed the momentum for the quark extracted from the proton ) .
next we examine the effect on the pdfs of integrating over a small region of transverse momentum @xmath302 , [ eq : dglap - picture+eq ] @xmath303{0.45\linewidth } \includegraphics[width={\textwidth}]{newfigs / qcd - dglap - epsilon.eps } \end{minipage}\right ) \\ & = \frac{{\alpha_\mathrm{s}}}{2\pi } \int_x^1 dz \ , { p_{qq}(z)}\ , \frac{q(x / z,{\mu_\text{\sc f}}^2)}{z } \;-\ ; \frac{{\alpha_\mathrm{s}}}{2\pi } \int_0 ^ 1 dz \ , { p_{qq}(z)}\ , q(x,{\mu_\text{\sc f}}^2 ) \end{aligned}\ ] ] where @xmath304 is the real part of the ` splitting kernel ' for a quark splitting to a quark plus a gluon , @xmath305 until now , we had concentrated on the soft limit , which was equivalent to approximating @xmath306 .
what eq .
( [ eq : dglap - picture+eq ] ) tells us is that as we increase the factorization scale , we get extra partons with longitudinal momentum fraction @xmath35 that come from the branching of partons in the proton at lower factorization scales but larger momentum fractions @xmath307 ( @xmath308 ) .
there are also loop contributions ( second term on the rhs ) to the parton density at a fixed @xmath35 value , which are negative contributions to the evolution .
the way to think about these is that when a parton with momentum fraction @xmath35 branches to partons with lower momentum fractions , the original parton is lost and the loop diagram accounts for that .
it s a bit awkward to write the real and virtual parts separately in eq .
( [ eq : dglap - picture+eq ] ) , especially if one wants to explicitly see the cancellation of the divergences for @xmath309 .
it s therefore standard to use the more compact notation @xmath310 where the subscript plus , known as the ` plus ' prescription , is to be understood as follows : @xmath311_+ } \,f(z ) & = \int_x^1 dz \ , { g(z ) } \ , { f(z ) } - \int_0 ^ 1 dz \ , { g(z)}\ , { f(1 ) } \\ & = \int_x^1 dz \ , { g(z)}\ , ( f(z ) - f(1 ) ) - \int_0^x dz \ , { g(z)}\ , { f(1)}\end{aligned}\ ] ]
so that the factor @xmath312 , which goes to zero at @xmath313 , kills the divergence due the singular behaviour of @xmath314 for @xmath309 .
equation ( [ eq : dglap - qq - compact ] ) involves just quarks , but the proton contains both quarks and gluons , so the full dglap equations are actually coupled evolution equations .
schematically , for just a single quark flavour , they read @xmath315 and more generally they span all quark flavours and anti - flavours . in labelling the different flavour entries , we ve included arrows ( usually not shown ) , e.g. @xmath316 , so as to emphasize that we have evolution from the right - hand parton type to the left - hand parton type .
the splitting functions other than @xmath317 are given by [ eq : splitting - fns ] @xmath318,\qquad\quad p_{gq}(z ) = { c_f}\left [ \frac{1+(1-z)^2}{z } \right]\,,\\ p_{gg}(z ) & = 2{c_a}\left [ \frac{z}{(1-z)_+}+ \frac{1-z}{z}+z(1-z)\right ] + \delta(1-z)\frac{(11{c_a}-4n_f{t_r})}{6 } \ , .
\end{aligned}\ ] ] additionally , @xmath319 and , to this first order in the coupling , @xmath320 and @xmath321 are both zero .
several features of the splitting functions are worth noting : @xmath322 and @xmath323 are both symmetric in @xmath324 ( except for the virtual part ) . @xmath317 and @xmath323 diverge for @xmath309 , which corresponds to soft - gluon emission . and @xmath323 and @xmath325 both diverge for @xmath326 ( corresponding to a soft gluon entering the ` hard ' process ) .
this last point implies that pdfs @xmath327 and @xmath328 must grow at least as fast as @xmath329 for @xmath249 : even if such a divergence is absent in the non - perturbative ` initial conditions ' for the quark and gluon distributions at low scales @xmath293 , dglap evolution inevitably introduces it into the result for @xmath330 and @xmath331 at higher scales . ; ( b ) the result of evolution to @xmath332 ; ( c ) a purely gluonic initial condition at @xmath333 ; and ( d ) the result of its evolution to @xmath332.,title="fig : " ] ; ( b ) the result of evolution to @xmath332 ; ( c ) a purely gluonic initial condition at @xmath333 ; and ( d ) the result of its evolution to @xmath332.,title="fig : " ] ; ( b ) the result of evolution to @xmath332 ; ( c ) a purely gluonic initial condition at @xmath333 ; and ( d ) the result of its evolution to @xmath332.,title="fig : " ] ; ( b ) the result of evolution to @xmath332 ; ( c ) a purely gluonic initial condition at @xmath333 ; and ( d ) the result of its evolution to @xmath332.,title="fig : " ] pictorially , the effect of dglap evolution is illustrated in fig .
[ fig : dglap ] .
a more quantitative view is given in fig .
[ fig : dglap - evolution ] , which shows the effect of dglap evolution with an initial condition that is pure quark ( two left plots ) or pure gluon ( two right plots ) . in both cases
one sees that evolution generates some amount of the missing parton type ; one also sees how it depletes the parton distributions at large @xmath35 , and increases them at small @xmath35 ( especially in the case of the gluon ) .
the attentive reader may have observed that the figure labels the scale of the pdfs as @xmath224 rather than @xmath334 : this is because it is standard to take @xmath335 so as to minimize the size of the the @xmath146 term of eq .
( [ eq : sigma - with - factorization ] ) which arises , roughly , from the integral over transverse momenta from @xmath334 to @xmath224 .
i.e. , one usually chooses to factorize essentially _ all _ initial - state radiation into the pdfs and so into the lo cross section .
( left - most plot ) .
if one evolves that initial condition assuming a gluon distribution that is zero at low @xmath224 , then agreement with high - scale data is poor ( central plot ) ; whereas with a significant low - scale gluon component ( taken from the cteq6d parametrization ) , agreement becomes good at high scales ( right - most plot ) .
[ fig : gluon - dglap - impact ] , title="fig : " ] ( left - most plot ) .
if one evolves that initial condition assuming a gluon distribution that is zero at low @xmath224 , then agreement with high - scale data is poor ( central plot ) ; whereas with a significant low - scale gluon component ( taken from the cteq6d parametrization ) , agreement becomes good at high scales ( right - most plot ) .
[ fig : gluon - dglap - impact ] , title="fig : " ] ( left - most plot ) .
if one evolves that initial condition assuming a gluon distribution that is zero at low @xmath224 , then agreement with high - scale data is poor ( central plot ) ; whereas with a significant low - scale gluon component ( taken from the cteq6d parametrization ) , agreement becomes good at high scales ( right - most plot ) .
[ fig : gluon - dglap - impact ] , title="fig : " ] since , as we ve see in fig .
[ fig : dglap - evolution ] , the presence of a gluon distribution helps drive quark evolution , we can use the experimentally observed pattern of quark evolution to help constrain the gluon .
the left - hand plot of fig .
[ fig : gluon - dglap - impact ] shows data from zeus @xcite and nmc @xcite on @xmath336 at some low but still perturbative scale @xmath337 .
the data are compared to the expectations based on the cteq6d pdfs quark content at that scale , illustrating the good agreement .
since these are data for @xmath227 , they have no direct sensitivity to the gluon distribution .
the middle plot shows data for @xmath338 , together with the results of dglap evolution from @xmath339 , assuming that the gluon distribution was zero at @xmath340 .
there s a clear discrepancy . in the right - hand plot ,
the comparison is made with evolution whose initial condition at @xmath340 contained a quite large gluon component ( exactly that in the cteq6d distributions ) , causing the quark distribution at small @xmath35 values to increase faster with @xmath224 than would otherwise be the case , bringing the evolution into agreement with the data . ]
it s interesting to ask just how much of a gluon distribution is needed in order to get the agreement shown in fig .
[ fig : gluon - dglap - impact ] .
the answer is given in fig .
[ fig : quark - gluon - dists - cteq6d ] and one sees that the gluon distribution is _ enormous _ , especially at small values of @xmath35 . it is fair to ask whether we can trust a result such as fig .
[ fig : quark - gluon - dists - cteq6d ] , so in this section we will examine some of ingredients and issues that are relevant to the ` global fits ' that inform our knowledge of pdfs . as a function of @xmath224 for many different @xmath35 values , compared to the results of a global fit by the zeus collaboration . ] as a function of @xmath224 for many different @xmath35 values , compared to the results of a global fit by the zeus collaboration . ]
figure [ fig : kin - regions - f2 ] ( left ) illustrates the kinematical regions in the @xmath35 and @xmath224 plane covered by the experimental data sets typically used in global fits .
everything below the diagonal line corresponds to dis data , and the right - hand plot shows the comparison between a fit ( by zeus ) and the bulk of the dis data , illustrating the excellent consistency between fit and data .
agreement with such a broad data set is already a non - trivial achievement .
figure [ fig : kin - regions - f2](left ) also shows shaded regions that span the diagonal line .
these correspond to hadron - collider jets data , which provide valuable direct information on the gluon distribution in global fits , as we will discuss below .
the other topics that we will address here relate to the accuracy of our determinations of pdfs .
perhaps the most convincing cross - check of pdf extractions comes from tevatron jet data ( there are also important jet data from hera ) .
the process of factorizing initial - state radiation into the pdf at a given scale is equally valid in dis and @xmath341 ( or @xmath211 ) collisions , as illustrated in the following pictorial representation of the two cases : @xmath282{0.6\linewidth } \includegraphics[width={\textwidth}]{college - pdf / dis+hh - factorization02.eps } \end{minipage}\ ] ] given factorization and a determination of pdfs in dis , one can simply take the expression eq .
( [ eq : parton - model - pp ] ) for a generic cross section in hadron
hadron collisions , and rewrite it with explicit factorization scales : @xmath342 such a formula , with @xmath343 replaced by @xmath344 ( summing also over processes with gluons , etc . ) can be used to obtain predictions for the differential inclusive jet spectrum at the tevatron .
figure [ fig : inclusive - jets ] shows comparisons of data from d and from cdf with predictions from cteq6.5 @xcite and mrst2004 @xcite pdf parametrizations , illustrating excellent agreement .
the two right - hand plots show how different incoming partonic scattering channels contribute to the cross section , highlighting the significant contribution from gluons . , taken from ref .
, title="fig : " ] , taken from ref .
, title="fig : " ] , taken from ref . @xcite .
, title="fig : " ] it is perhaps misleading to use the word ` prediction ' about fig .
[ fig : inclusive - jets ] : most advanced fully global pdf fits actually make use of data such as that in fig . [
fig : inclusive - jets ] as part of the fit .
still , it is a powerful consistency check that it is possible to obtain agreement both with the jet data , which is sensitive directly to quark and to gluon distributions , as well as with dis data , which is directly sensitive to the quarks and indirectly to the gluons , through the scaling violations .
one technical comment is due concerning factorization .
while our discussion has been limited to leading order , many of the figures that we have shown also involve higher orders ( to which we will return ) .
when using pdfs with predictions beyond lo , it is necessary to specify the ` factorization scheme ' , i.e. , the specific procedure by which one separates emissions below and above @xmath293 .
the figures in sections [ sec : dis ] and [ sec : dglap ] made use of the ` dis ' scheme ( hence cteq6__d _ _ ) , defined such that @xmath227 is given by eq .
( [ eq : f2em ] ) , free of any @xmath146 ( or higher ) corrections . while that scheme has the benefit of pedagogical simplicity , in real calculations ( and all plots in this section ) it is usually more convenient to use the ` @xmath345 ' factorization scheme , based on dimensional regularization .
an important part of the activity on global fits in recent years has been geared to the estimation of the uncertainties on pdfs .
figure [ fig : pdf - uncertainties ] shows the uncertainties on recent pdf sets from the two most established pdf fitting groups , mstw and cteq , illustrating uncertainties that are in the couple of per cent to ten per cent range .
figure [ fig : pdf - uncertainties - impact ] illustrates the impact of the uncertainties on predictions for a range of cross sections at the tevatron and the lhc .
( figure taken from ref .
@xcite),title="fig : " ] ( figure taken from ref . @xcite),title="fig : " ] ( figure taken from ref . @xcite),title="fig : " ] estimating pdf uncertainties is something of an art : for example , one must parametrize the pdfs at some input scale and there is freedom in how flexible a parametrization one uses : too rigid ( or with too many theorist s assumptions ) and the global fit may not have the flexibility to describe the data or may appear to have inappropriately small uncertainties in regions where there are no data ; with too flexible a parametrization the fits may develop artefacts that adapt to statistical fluctuations of the data .
other issues include reconciling barely compatible data sets and deciding what values of @xmath346 variations are reasonable to estimate errors given the incompatible data sets .
in addition to mstw and cteq ( and several other groups ) , a recent entrant to pdf fitting is the nnpdf collaboration , which is developing procedures that attempt to minimize theoretical bias in questions such as the parametrizations .
first global fit results including @xmath211 data have been given in ref .
@xcite , though for fully accurate treatment of the hera data , one should await their inclusion of heavy - quark effects .
their results so far tend to be similar to those of mstw and cteq , except in the regions of small-@xmath35 near the edge of the available hera phasespace and for strange quark distributions , where they find somewhat larger uncertainties .
figure [ fig : pdfs - to - lhc ] ( left ) illustrates the kinematic region in the @xmath35 and @xmath224 plane that is covered by the lhc ( with @xmath347 ) , compared to that for hera and fixed - target experiments .
the lhc region is labelled with a grid corresponding to mass ( @xmath348 , and one takes @xmath349 ) and rapidity ( @xmath225 ) of the object that is being produced .
these are related to the incoming momentum fractions @xmath214 and @xmath220 and the @xmath211 squared centre - of - mass energy through @xmath350 an object produced at a rapidity @xmath225 involves @xmath351 and @xmath352 .
one feature that s immediately visible from the plot is that much of the lhc kinematic plane covers regions where pdfs have not been measured directly .
we will therefore rely heavily on dglap evolution for our predictions .
the right - hand plot in fig .
[ fig : pdfs - to - lhc ] illustrates just how much , in that it gives the factor by which the gluon distribution evolves in going from @xmath353 to @xmath354 .
depending on the region in @xmath35 , this can be a large factor , @xmath355 . when compared to the experimental uncertainties on pdfs as shown in figs .
[ fig : pdf - uncertainties ] and [ fig : pdf - uncertainties - impact ] , we clearly have to ask how well we know the evolution . to @xmath356 using cteq61 distributions as a fixed input at the low scale , and carrying out lo dglap evolution with hoppet @xcite with @xmath357 ( see text for further details).,title="fig : " ] ( -2.9,3.5 ) ( -1.5,5.0 ) to @xmath356 using cteq61 distributions as a fixed input at the low scale , and carrying out lo dglap evolution with hoppet @xcite with @xmath357 ( see text for further details ) .
] in section [ sec : fixed - order ] we will discuss in detail how uncertainties are estimated in theoretical predictions . for now , essentially , there is freedom in eq .
( [ eq : matrix - dglap ] ) to choose a renormalization scale @xmath358 for @xmath0 that differs from @xmath334 . a conventional way of estimating the uncertainties is to choose @xmath359 , varying @xmath360 in the range @xmath361 ( actually one often just takes three values , @xmath357 ) . starting from a fixed input at @xmath362 and evolving with different choices for @xmath360 gives the width of the band shown in fig .
[ fig : pdfs - to - lhc ] ( right ) . that width is much larger than the uncertainties that we see in figs .
[ fig : pdf - uncertainties ] and [ fig : pdf - uncertainties - impact ] . up to @xmath356 , for lo , nlo and nnlo evolution .
the bands correspond to the envelope of the results with three scale choices @xmath363 and what is plotted is the ratio to the result at scale @xmath356 , as obtained with nnlo evolution using @xmath364 . ]
fortunately we are not limited to leading - order ( lo ) dglap evolution , i.e. , just the @xmath146 term in eq .
( [ eq : matrix - dglap ] ) .
the order @xmath89 ( next - to - leading order nlo ) corrections to the dglap equation were calculated around 1980 @xcite and in 2004 the calculation of the nnlo corrections was completed @xcite .
to give an idea of the complexity of that calculation , rather than taking a couple of lines as in eq .
( [ eq : splitting - fns ] ) at lo , the nnlo results take @xmath355 pages to write down ! the impact of including higher orders on the evolution uncertainties is illustrated in fig .
[ fig : evolution - uncertainties ] . the same cteq61 input distribution is evolved from @xmath362 to @xmath356 with lo , nlo and nnlo splitting kernels , using three scale choices , @xmath357 .
the figure then shows the results for the gluon distribution at scale @xmath356 , normalized to gluon distribution obtained with nnlo evolution and @xmath364 .
one sees how the uncertainty comes down substantially as higher orders are included : @xmath365 in some regions at lo , @xmath366 over most of the @xmath35 range at nlo and @xmath367 at nnlo . in the nnlo case ,
the uncertainty is usually smaller than the experimental uncertainties for the pdfs that were shown in fig .
[ fig : pdf - uncertainties ] . is to be interpreted with some caution .
we have taken a fit carried out with @xmath364 , and then evolved it to high scales with @xmath368 . however , for example , the gluon distribution is partially determined from the evolution of @xmath227 , so if the fit itself were carried out with @xmath369 , the fit result would probably change , introducing additional dependence of the @xmath370 results on the choice of @xmath360 .
this could conceivably cancel some of the dependence seen in fig .
[ fig : evolution - uncertainties ] . ] here are some of the points to retain from this section .
firstly , the proton really is what we expect it to be , i.e. , a @xmath253 state ; however , fluctuations of the proton introduce many extra @xmath371 pairs ( ` sea ' ) , as well as a substantial amount of ` glue ' , carrying @xmath372 of the proton s momentum .
the sea and gluon distributions diverge at small momentum fractions @xmath35 .
determination of the proton s pdfs involves fitting data from a range of experiments , with direct sensitivity to quarks ( e.g. , dis ) , indirect sensitivity to the gluon ( dis @xmath224 evolution ) , and direct sensitivity to quarks and gluons ( jet data ) .
one of the keys to being able to measure consistent pdfs from different experiments , thinking about them in perturbative qcd and then applying them to predict results at new experiments is ` factorization ' : initial - state radiation , though collinear divergent , is process - independent ; the divergent part can be absorbed into the definition of the pdfs , and then a universal set of pdfs , evolved between different scales with the dglap equations , can be used for any process .
finally , the accuracy with which we know pdfs is quite remarkable : both from the experimental side and the theoretical side , in the best cases we know the pdfs to within a few per cent
. this will be important in interpreting future signals of new particles , for example in higgs - boson production at the lhc when we want to deduce its electroweak couplings given a measurement of its cross section . if you need to use pdfs yourself , the best place to get them is from the lhapdf library @xcite .
in this section we will look at some of the different classes of technique can be used to make qcd predictions at lhc . among the topics that we ll touch on are leading order ( lo ) , next - to - leading order ( nlo ) and next - to - next - leading order ( nnlo ) calculations , parton - shower monte carlos , and then methods to combine the two types of calculation . )
pair and the subsequent decay of each gluino through a squark ( @xmath373 ) to quarks and a neutralino ( @xmath374 ) ; the experimental signature involves four jets and missing transverse energy ( @xmath375 ) from the unobserved neutralino .
right : a background that mimics this signature , with missing energy coming from the production of a @xmath94-boson that decays to neutrinos . ] ) pair and the subsequent decay of each gluino through a squark ( @xmath373 ) to quarks and a neutralino ( @xmath374 ) ; the experimental signature involves four jets and missing transverse energy ( @xmath375 ) from the unobserved neutralino .
right : a background that mimics this signature , with missing energy coming from the production of a @xmath94-boson that decays to neutrinos . ] ) pair and the subsequent decay of each gluino through a squark ( @xmath373 ) to quarks and a neutralino ( @xmath374 ) ; the experimental signature involves four jets and missing transverse energy ( @xmath375 ) from the unobserved neutralino .
right : a background that mimics this signature , with missing energy coming from the production of a @xmath94-boson that decays to neutrinos . ] ) pair and the subsequent decay of each gluino through a squark ( @xmath373 ) to quarks and a neutralino ( @xmath374 ) ; the experimental signature involves four jets and missing transverse energy ( @xmath375 ) from the unobserved neutralino .
right : a background that mimics this signature , with missing energy coming from the production of a @xmath94-boson that decays to neutrinos . ] many of the examples that we ll use will involve @xmath94 ( and sometimes @xmath376 ) production at hadron colliders .
one reason is that @xmath94 and @xmath376 bosons decay to leptons and neutrinos ( missing energy ) , both of which are easily - taggable handles that are characteristic of signals in many new - physics scenarios .
an illustration is given in fig .
[ fig : signal - v - background - diagrams ] , which depicts supersymmetric production of a gluino pair and subsequent decay to four jets and missing transverse energy from the unobserved neutralinos .
because of the complexity of the decays and the fact that the missing energy is the sum of that from two neutralinos , it can be difficult to extract clear kinematic structures ( such as an invariant mass peak ) that make a signal emerge unambiguously over the background .
in such cases the contribution from the signal may just be to give a cross section that is larger than background expectations over a broad kinematic range .
but that will only be a ` signal ' if we understand what the backgrounds are .
the extent to which we will want to ( or have to ) rely on qcd predictions of backgrounds in deciding whether there are signals of new physics at the lhc is a subject that deserves in - depth consideration ( for a nice discussion of it , see ref .
but qcd predictions will come into play in many other ways too .
monte carlo parton shower programs , which simulate the full hadronic final state , are crucial in evaluating detector acceptances and response . and knowing qcd predictions ( both for backgrounds and possible signals ) is crucial in the design of methods to search for new physics , as well as for extracting meaning from the data ( about couplings , spins , etc . )
when , it is to be hoped , we finally see signals of something new .
fixed - order predictions , which involve the first couple of terms in the qcd perturbative expansion for a given cross section , are conceptually quite simple : it is easy to state which contributions are included , and as one includes further orders in the expansion one can reasonably hope to see systematic improvement in the accuracy of one s predictions .
we ll first look at a couple of examples of fixed - order predictions , in order to develop a feel for how the perturbative expansion behaves , and how one estimates its accuracy .
we will then examine more generally what theoretical inputs are needed for predictions for a given process , and what practical forms the predictive tools take . in eq .
( [ eq : tot - xsct - series ] ) , we wrote the total cross section for @xmath378 as a perturbative series expansion in @xmath0 that multiplied the born cross section @xmath145 .
the expansion was formulated in terms of the coupling evaluated at a renormalization scale @xmath379 equal to the centre - of - mass energy @xmath83 , i.e. , @xmath380 .
that choice is , however , arbitrary : for example , the most energetic gluon that could be produced in @xmath381 would be one with @xmath382 , so maybe we should be choosing @xmath383 .
and in loop diagrams , one integrates over gluon energies that go beyond @xmath83 , so maybe @xmath384 would be just as reasonable .
total cross section , together with an indication of the conventional choice of scale - variation range , title="fig : " ] total cross section , together with an indication of the conventional choice of scale - variation range , title="fig : " ] because of this arbitrariness , a convention has emerged whereby one calculates a ` central value ' for the prediction by setting the renormalization scale equal to the main physical scale for the process ( e.g. , the centre - of - mass energy at an @xmath93 collider ; for hadron - collider processes the choice may be less obvious ) .
the uncertainty is then estimated by varying the scale by a factor of two in either direction from the central value , i.e. , taking @xmath385 .
this is illustrated in fig .
[ fig : ee2hadrons - scale - dependent ] ( left ) , which plots @xmath386 as a function of @xmath379 , showing how the @xmath379-dependence translates into an uncertainty ; note that @xmath199 can be read from eq .
( [ eq : tot - xsct - series ] ) . given an expansion of the running coupling ( i.e. , of the middle result of eq .
( [ eq : renorm - group-1loop - soln ] ) , @xmath387 ) , we can rewrite eq .
( [ eq : ee - nlo ] ) as @xmath388 this tells us that as we vary the renormalization scale for a prediction up to @xmath146 ( nlo ) , we effectively introduce @xmath389 ( nnlo ) pieces into the calculation : by generating some fake set of nnlo terms , we are probing the uncertainty of the cross section associated with the missing full nnlo correction .
if we calculate the actual nnlo cross section for general @xmath379 , it will have a form @xmath390 observe that the @xmath200 coefficient now depends on @xmath379 .
this is necessary because the second - order coefficient must cancel the @xmath389 ambiguity due to the scale choice in eq .
( [ eq : ee - nlo - explicit - scale ] ) .
this constrains how @xmath391 depends on @xmath379 : @xmath392 where @xmath393 can again be read from eq .
( [ eq : tot - xsct - series ] ) .
if we now express @xmath394 in terms of @xmath395 , we will find that the residual dependence on @xmath379 appears entirely at @xmath146 , i.e. , one order further than in eq .
( [ eq : ee - nlo - explicit - scale ] ) .
this is reflected in the right - hand plot of fig .
[ fig : ee2hadrons - scale - dependent ] , which illustrates how the impact of the scale variation at nnlo is significantly reduced , since we are now probing the impact of missing @xmath90 terms , rather than @xmath89 terms .
if we had an arbitrarily large number of terms in the @xmath0 expansion , the scale dependence would disappear exactly .
the fact it does nt in the presence of a fixed number of terms may initially seem like a drawback , but in some respects it s a blessing in disguise because it provides a useful handle on the uncertainties .
this is why scale variation has become a standard procedure .
it s worth bearing in mind that it is nt a failsafe mechanism : a trivial example comes from the lo curve in fig .
[ fig : ee2hadrons - scale - dependent ] .
it does nt have any scale variation because they do nt depend on @xmath0 , yet it differs significantly from the higher - order results . at lo
the @xmath396 cross section involves a single underlying hard partonic process , namely @xmath397 , which is purely electroweak . to go from the @xmath397
squared matrix element to the @xmath398 result , one must integrate over the quark distributions @xmath399 for which one must choose a factorization scale @xmath293 .
a natural choice for this scale is @xmath400 , but as with the renormalization scale it is conventional to vary it by a factor of two either side of the central choice in order to obtain a measure of the uncertainties in the prediction . adding nlo and nnlo terms , the structure becomes @xmath401 + \ , { \alpha_\mathrm{s}}^2({\mu_\text{\sc r } } ) \hat\sigma_{2,ij\to z\!+\!x } ( x_1 , x_2 , { \mu_\text{\sc f } } , { \mu_\text{\sc r } } ) \bigg].\end{gathered}\ ] ] we now have a sum over the flavours @xmath109 and @xmath402 of the initial partons , because starting from nlo there are contributions from ( say ) gluon - quark scattering [ cf . fig .
[ fig : z - xsct](left ) ] .
the cross section is written as being for @xmath403 , where the @xmath404 means that we allow anything ( e.g. , quarks , gluons ) to be produced in addition to the @xmath94-boson . at @xmath146 the @xmath293 dependence of the @xmath405 coefficient
partially cancels the dependence present at @xmath406 coming from the @xmath293 dependence of the pdfs .
that dependence is further cancelled at @xmath389 , as is part of the @xmath379 dependence that is introduced in the @xmath407 term . the plot on the right of fig .
[ fig : z - xsct ] shows the @xmath94-boson cross section as a function of its rapidity @xcite .
the bands indicate the uncertainty due to scale variation ( taking @xmath408 ) and @xmath293 simultaneously , though common , is not the only possible procedure .
an attractive alternative is to vary both independently around a central scale , with the additional requirement that @xmath409 @xcite . ] and show how this uncertainty undergoes important reductions going from lo to nlo to nnlo .
one of the interesting features that comes out of fig .
[ fig : z - xsct ] is that the lo prediction is only good to within a factor of @xmath410 to @xmath206 , despite the fact that @xmath411 would imply @xmath412 accuracy .
this is because the @xmath146 corrections come with large coefficients .
this is not uncommon in hadron - collider cross sections .
furthermore the lo uncertainty band seems not to provide a faithful measure of the true uncertainty .
other aspects of the perturbative expansion do seem to behave as one would expect : the size of the uncertainty band decreases significantly going from lo to nlo ( 1020@xmath413 ) to nnlo ( a few per cent ) . and
the actual shift in the central value in going from nlo to nnlo is substantially smaller than that from nlo to lo .
are these characteristics representative of the ` typical ' situation for collider observables ?
we only have predictions up to nnlo in a handful of cases ( see below ) and in those it is . in cases where we just have nlo predictions ,
the features of large ` k - factors ' ( nlo / lo enhancements ) with a reduced nlo uncertainty band are not uncommon , suggesting that beyond nlo corrections should be small .
exceptions are known to arise in two types of case : those where new enhanced partonic scattering channels open up at nlo ( or beyond ) ; and that involve two disparate physical scales .
for example , if you ask for the @xmath94-boson to have a transverse momentum @xmath414 that is much smaller than @xmath415 , then each power of @xmath0 in the expansion of the cross section will be accompanied by up to two powers of @xmath416 , leading to large coefficients at all orders in the perturbative expansion .
these are due to incomplete cancellation between real and virtual ( loop ) divergences : loop corrections do not affect the @xmath94-boson @xmath414 and so are fully integrated over , whereas real emissions do affect the z @xmath414 and so are only partially integrated over . at lo and at nlo .
right : cross section at the lhc for the @xmath94-boson , differential in rapidity , at lo , nlo and nnlo , as taken from ref .
@xcite . ] at lo and at nlo .
right : cross section at the lhc for the @xmath94-boson , differential in rapidity , at lo , nlo and nnlo , as taken from ref .
@xcite . ] as an example of a more complex process , consider the production of a @xmath94-boson plus a jet .
the leading order cross section requires the calculation of the @xmath146 squared diagrams for @xmath417 , @xmath418 and @xmath419 .
the nlo cross section additionally requires all @xmath389 contributions with a @xmath94 boson and at least one jet , as illustrated in fig .
[ fig : loop+legs - static ] , i.e. , the squared tree - level diagram for @xmath420 and the interference of the 1-loop and tree - level diagrams for @xmath421 . more generally , fig .
[ fig : loop+legs - static ] allows you to read off the contributions that you will need for an n@xmath422lo calculation of @xmath423 : just take all entries in fig .
[ fig : loop+legs - static ] with at least @xmath185 partons , up to order @xmath424 . entries in black are known and have already been used to obtain predictions for lhc and tevatron . the one entry in grey , the 2-loop @xmath425 contribution ,
is known but has yet to be used in any hadron - collider prediction , for reasons that we will discuss below .
entries that are absent ( e.g. , @xmath426 partons at two loops ) have so far proven to be too complicated to calculate .
the classes of contributions calculated for @xmath427 provide a representative view of the situation for other processes as well , with tree - level diagrams calculated up to quite high final - state multiplicities , @xmath428 , 1-loop diagrams having been used for processes with up to 3 or sometimes 4 final - state particles , and 2-loop diagrams available and used only for @xmath429 type processes , essentially @xmath430 , @xmath431 and @xmath432 . , where @xmath109 and @xmath402 are arbitrary incoming partons , according to the number of outgoing partons , the number of loops and the number of powers of the coupling . an ` x ' represents a squared tree - level diagram , an ` o ' represents the interference of a 1-loop diagram with a tree - level diagram , and a ` ' represents the interference of a two - loop diagram with a tree - level diagram or the square of a 1-loop diagram .
entries in black are known and used ; entries in grey are known but have not been used .
the entries in the shaded ellipses are those that are relevant for the nlo calculation of the cross section for the production of a @xmath94-boson with a jet . ]
it s natural to ask in what form these various calculations are available . for certain very simple quantities , for example ,
the total cross section for @xmath433 production , or for @xmath376 , @xmath94 or higgs production , the result of the perturbative calculation can be written as in eq .
( [ eq : pp2z - nnlo ] ) , @xmath434 where the @xmath435 are functions whose analytical expressions can be found in the relevant papers ( refs .
@xcite for @xmath376 and @xmath94 , and refs .
@xcite for higgs - boson production ) . to obtain a prediction
, one just has to type them into a computer program and then integrate over @xmath214 and @xmath220 . in most cases
, however , one wants to calculate a cross section that incorporates experimental cuts , such as lepton acceptances , transverse momentum cuts on jets , etc . in these cases
the type of tool that should be used depends on the order to which you want the answer .
[ [ lo - predictions ] ] lo predictions + + + + + + + + + + + + + + as long as one is dealing with infrared safe observables , then for a lo prediction one need only include tree - level diagrams , in kinematic regions in which their contributions are finite .
the simplest approach therefore is to carry out monte carlo integration over phase - space points , have a subroutine that determines whether a given phase - space point passes the cuts , and if it does calculate the squared matrix elements and pdf factors for each possible partonic subprocesses .
quite a number of tools enable you to do this : alpgen @xcite , comix / sherpa @xcite , comphep @xcite , helac / phegas @xcite and madgraph @xcite .
they allow you to calculate cross sections for a broad range of @xmath436 scattering processes with @xmath185 up to 68 ( or in some cases even beyond ) . some of these ( comphep , madgraph ) use formulae obtained from direct evaluations of feynman diagrams .
this gives them significant flexibility in terms of the range or processes they support ( e.g. , with easy inclusion of many new physics models ) , though they suffer at large @xmath185 because the number of diagrams to evaluate grows rapidly as @xmath185 increases ( cf .
table [ tab : ndiags ] ) .
others ( alpgen , helac / phegas and comix / sherpa ) use methods designed to be particularly efficient at high multiplicities , such as berends giele recursion @xcite , which builds up amplitudes for complex processes by recursively reusing simpler ones ( a nice technical review of the technique is given in ref .
@xcite ) .
.[tab : ndiags ] the number of feynman diagrams for tree level @xmath437 scattering @xcite [ cols="<,^,^,^,^,^,^,^",options="header " , ] [ [ nlo - predictions ] ] nlo predictions + + + + + + + + + + + + + + + when , in sections [ sec : total - xsct ] and [ sec : ir - safe - obs ] , we looked at the cancellation between divergences in real and loop diagrams , we wrote the loop diagram with an explicit integral over phase space so as to be able to match the divergences between real and loop diagrams and cancel them easily .
a subtlety that we ignored is that in practical evaluations of loop diagrams , the integral over loop momenta is carried out in @xmath64 dimensions rather than 4 dimensions , in order to regularize the divergences that appear and obtain an answer whose finite part is known independently of any related tree - level diagrams . on the other - hand ,
experimental cuts are defined in four dimensions , so the real tree - level diagrams must be integrated in four dimensions , which implies divergent results if the real diagrams are taken alone .
this mismatch between the ways loop and tree - level diagrams are handled is one of the main difficulties in carrying out calculations that include experimental cuts beyond lo .
for the calculation of a process with @xmath185 partons at lo , the standard technique nowadays to deal with the problem is to introduce a @xmath438-parton _ counterterm _ where the @xmath439 parton is always soft and collinear so that it does nt affect an ir safe observable .
it is subtracted from the @xmath438-parton real diagram in four dimensions and designed so as to cancel all of its soft and collinear divergences .
it is also designed such that the kinematics of its @xmath439 parton can be integrated analytically in @xmath64 dimensions so that the result can be easily added to the loop diagram and cancel _ its _ divergences . since the counterterm is subtracted once and added once , its net impact is null , and
can just be thought of as a way of reshuffling divergences .
this is known as a ` subtraction ' procedure and the variant most widely used in current nlo computer codes is ` dipole ' subtraction @xcite ; other methods that have seen numerous applications include fks @xcite and antenna @xcite subtraction .
scattering events in the calculation of the dijet cross section , shown as a function of the transverse momentum of the harder jet .
right : weights at nlo from real @xmath440 events , subtraction counterterm events and loop events ( after addition of the integrated counterterm ) , again as a function of @xmath414 of the hardest jet .
results obtained with nlojet`++ ` @xcite . ] scattering events in the calculation of the dijet cross section , shown as a function of the transverse momentum of the harder jet .
right : weights at nlo from real @xmath440 events , subtraction counterterm events and loop events ( after addition of the integrated counterterm ) , again as a function of @xmath414 of the hardest jet .
results obtained with nlojet`++ ` @xcite . ]
an illustration of how subtraction works in practice is given in fig .
[ fig : nlojet - weights ] for dijet production . in the left - hand plot
we see weights for the lo process as a function of the jet @xmath414 .
though there is some dispersion in the weights , there is a clear upper bound as a function of @xmath414 , reflecting the finiteness of the matrix - elements .
the right - hand plot shows weights at nlo .
here there is no clear upper bound to the weights ; however we do see that unusually large ( nearly divergent ) weights come in clusters : usually one positive ( red , ` + ' ) weight , accompanied by one or more negative ( green , ` @xmath441 ' ) weights with identical jet transverse momenta , so that each event in the cluster contributes to the same bin of the cross section and their weights sum to a finite result . technically , one main consideration has so far limited the range of processes for which nlo results exist : the availability of the loop amplitude . until recently loop amplitudes were usually calculated semi - manually for each process .
the complexity of the calculations increased significantly with the number of outgoing legs , limiting available results to those with at most three outgoing partons .
many nlo results for @xmath442 and @xmath443 processes are incorporated into programs such as nlojet`++ ` for jet production @xcite , mcfm for processes with heavy quarks and/or heavy electroweak bosons @xcite , vbfnlo for vector - boson fusion processes @xcite , and the phox family @xcite for processes with photons in the final state . in the past couple of years ,
techniques have come to fruition that offer the prospect of automated calculation of arbitrary loop diagrams . though full automation is not here yet , a number of @xmath444 processes have now been calculated at nlo thanks to these advances , including @xmath445 @xcite , @xmath446 @xcite ( where @xmath402 represents a jet ) and @xmath447 @xcite and @xmath448 @xcite .
process have also appeared @xcite . ]
it should be said that nlo calculations are very computing intensive : for some observables it is not unusual to have to devote several years of cpu time in order to get adequate numerical convergence of the monte carlo integration .
[ [ nnlo - predictions ] ] nnlo predictions + + + + + + + + + + + + + + + + nnlo predictions suffer from the same problem of cancelling divergences between real and virtual corrections that is present at nlo , with the complication that instead of having one soft and one collinear divergence , there are now two of each , greatly complicating the task of figuring out counterterms to allow experimental cuts to be implemented in four dimensions . as a result
, the only general subtraction type approaches that exist currently are for processes without incoming hadrons , notably @xmath449 @xcite . for hadron collider
processes it is only @xmath429 processes that are available , specifically vector - boson ( fewz @xcite and dynnlo @xcite ) and higgs - boson ( fehip @xcite and hnnlo @xcite ) production , using methods that are not so easily generalizable to more complicated processes .
the programs we ve discussed so far , known as ` matrix element monte carlos ' provide a powerful combination of accuracy and flexibility as long as you want to calculate ir and collinear safe observables ( jets , @xmath376 s , @xmath94 s , but not pions , kaons , etc . ) , do nt mind dealing with wildly fluctuating positive and negative weights , and do nt need to study regions of phase space that involve disparate physical scales .
all these defects are essentially related to the presence of soft and collinear divergences .
yet we know that real life does not diverge .
so it is natural to wonder whether we can reinterpret the divergences of perturbation theory physically .
it turns out that the right kind of question to ask is `` what is the probability of _ not _ radiating a gluon above some ( transverse momentum ) scale @xmath292 '' . starting from a @xmath61 system , using the results of section [ sec : ee2hadrons ] , we know that to @xmath146 the answer in the soft and collinear limit goes as @xmath450 it so happens that in the soft and collinear limit , this result is easy to work out not just at first order , but at all orders , giving simply the exponential of the first order result @xmath451\,.\ ] ] whereas eq .
( [ eq : sudakov - expansion ] ) had a ` bare ' infinity if one took @xmath452 , eq .
( [ eq : sudakov - exponentiated ] ) is simply bounded to be between @xmath453 and @xmath149 .
the quantity @xmath454 is known as a sudakov form factor .
we ve been very approximate in the way we ve calculated it , neglecting for example the running of the coupling ( @xmath0 should be placed inside the integral and evaluated at the scale @xmath455 ) and the treatment of hard collinear radiation ( the @xmath456 integral should be replaced with the full collinear splitting function ) , but these are just technical details .
the importance of the sudakov form factor is that it allows us to easily calculate the distribution in transverse momentum @xmath457 of the gluon with largest transverse momentum in an event : @xmath458 this distribution is easy to generate by monte carlo methods : take a random number @xmath459 from a distribution that s uniform in the range @xmath460 and find the @xmath457 that solves @xmath461 .
given @xmath457 we also need to generate the energy for the gluon , but that s trivial . if we started from a @xmath61 system ( with some randomly generated orientation ) , then this gives us a @xmath462 system . as a next step one can work out the sudakov form factor in the soft / collinear limit for there to be no emission from the @xmath462 system as a whole above some scale @xmath463 ( @xmath464 ) and
use this to generate a second gluon .
the procedure is then repeated over and over again until you find that the next gluon you would generate is below some non - perturbative cutoff scale @xmath465 , at which point you stop .
this gives you one ` parton shower ' event .
this is essentially the procedure that s present in the shower of pythia 8 @xcite and the @xmath414 ordered option of pythia 6.4 @xcite , as well as ariadne @xcite and sherpa 1.2 ( the sherpa reference @xcite describes an earlier version ) .
it is also possible to choose other ordering variables : the original pythia shower@xcite is based on virtuality ordering ( plus an angular veto ) .
this is still the most widely used shower which works well on a range of data ( though there are theoretical issues in some formulations of virtuality ordering ) . in the herwig family of programs
@xcite it is angular ordering that is used .
the above shower descriptions hold for final - state branching . with initial - state hadrons
, one also needs to be careful with the treatment of the pdfs , since the collinear splitting that is accounted for in the parton shower is also connected with the way the pdf is built up at the scale of the hard scattering .
the sequence of steps for the generation of a parton - shower event in @xmath211 collisions is illustrated in fig .
[ fig : showering ] .
\a ) + event in herwig : ( a ) specification of the colliding beams and their energy , ( b ) generation of the kinematics and partonic flavour of the hard subprocess , @xmath466 , and ( c ) generation of the initial- and final - state parton showers , title="fig : " ] + b ) + event in herwig : ( a ) specification of the colliding beams and their energy , ( b ) generation of the kinematics and partonic flavour of the hard subprocess , @xmath466 , and ( c ) generation of the initial- and final - state parton showers , title="fig : " ] \c ) + event in herwig : ( a ) specification of the colliding beams and their energy , ( b ) generation of the kinematics and partonic flavour of the hard subprocess , @xmath466 , and ( c ) generation of the initial- and final - state parton showers , title="fig : " ] real events consist not of partons but of hadrons .
since we have no idea how to calculate the transition between partons and hadrons , monte carlo event generators resort to ` hadronization ' models .
one widely - used model involves stretching a colour ` string ' across quarks and gluons , and breaking it up into hadrons @xcite . for a discussion of the implementation of this ` lund ' model in the mc program pythia , with further improvements and extensions , ref .
@xcite and references therein provide many details .
another model breaks each gluon into a @xmath61 pair and then groups quarks and anti - quarks into colourless ` clusters ' , which then give the hadrons .
this cluster type hadronization is implemented in the herwig event generator @xcite and recent versions of sherpa .
both approaches are illustrated in fig .
[ fig : hadronization ] data for the particle ( scaled ) momentum distribution ( left ) and the thrust event shape distribution ( right ) with two tunes of the pythia event generator @xcite.,title="fig : " ] data for the particle ( scaled ) momentum distribution ( left ) and the thrust event shape distribution ( right ) with two tunes of the pythia event generator @xcite.,title="fig : " ] hadronization models involve a number of ` non - perturbative ' parameters .
the parton - shower itself involves the non - perturbative cutoff @xmath465 .
these different parameters are usually tuned to data from the lep experiments .
the quality of the description of the data that results is illustrated in fig .
[ fig : tuning ] . a final point in our brief description of monte carlo event generators concerns the ` underlying event ' in @xmath211 and @xmath467 collisions .
in addition to the hard interaction that is generated by the monte carlo simulation , it is also necessary to account for the interactions between the incoming proton ( or photon ) remnants .
this is usually modelled through multiple extra @xmath442 scattering , occurring at a scale of a few gev , and known as multiple parton interactions .
this modelling of the underlying event is crucial in order to give an accurate reproduction of the ( quite noisy ) energy flow that accompanies hard scatterings in hadron - collider events .
our description here of monte carlo event generators has been fairly brief . for a more complete discussion ,
a good starting point is the lectures notes by sjstrand @xcite from the 2006 school .
parton - shower monte carlo programs do a good job of describing most of the features of common events , including the hadron - level detail that is essential for the correct simulation of detector effects on event reconstruction .
another nice feature of theirs is that events have equal weight , just as with real data .
a drawback of parton - shower monte carlos is that , because they rely on the soft and collinear approximation , they do not necessarily generate the correct pattern of hard large - angle radiation .
this can be important , e.g. , if you re simulating backgrounds to new - physics processes , for which often the rare , hard multi - jet configurations are of most interest . in contrast
, fixed - order programs do predict these configurations correctly .
the purpose of this section is to give two examples of comparisons between parton - shower predictions and fixed - order predictions , in order to help illustrate their relative strengths . in plain jet production ,
parton shower monte carlos start from a hard event that consists of parton parton scattering diagrams like @xmath468 , @xmath469 , etc . , and then rely on the showering to generate extra radiation .
while the showering is only correct in the soft and/or collinear limit , it does sometimes generate extra hard radiation . in fig .
[ fig:3rd - jet ] we can see distribution in transverse momentum of the 3rd hardest jet , in herwig events in which we have imposed a cut on the hardest jet , @xmath470 .
that is compared to the nlo ( nlojet`++ ` ) prediction for the same distribution , together with its uncertainty band from scale variation .
distribution for the jet with the third largest transverse momentum , @xmath471 , in @xmath36 pp events where the hardest jet has transverse momentum @xmath470 .
the solid histogram is the result from herwig 6.5 @xcite .
the band corresponds to the ratio of the nlo 3-jet cross section , differential in @xmath471 ( with the @xmath472 cut ) , to the nlo cross section for events to pass the @xmath472 cut .
the width of the band corresponds to the scale uncertainty and the results have been obtained with nlojet`++ ` @xcite . ]
distribution for the jet with the third largest transverse momentum , @xmath471 , in @xmath36 pp events where the hardest jet has transverse momentum @xmath470 .
the solid histogram is the result from herwig 6.5 @xcite .
the band corresponds to the ratio of the nlo 3-jet cross section , differential in @xmath471 ( with the @xmath472 cut ) , to the nlo cross section for events to pass the @xmath472 cut .
the width of the band corresponds to the scale uncertainty and the results have been obtained with nlojet`++ ` @xcite . ] in much of the range , one observes relatively good agreement between the two distributions : the 2030@xmath413 differences that are visible around @xmath473 are no larger than the uncertainties that would be obtained from a lo calculation , despite the fact that in this region herwig does not even include the exact lo @xmath440 matrix element .
of course , it is hard to be sure whether the good agreement is meaningful generally , or instead just a coincidence specific to our particular choice of observable and the only way to be sure is , for each new observable , to also generate the nlo prediction .
the nlo prediction is not without its limitations though : at low @xmath471 , the uncertainty band on the nlo prediction blows up .
this is a manifestation of large higher - order corrections , which compromise the convergence of the perturbative series .
they arise because we have a large ratio of scales between @xmath474 ) and @xmath471 ( a few tens of gev ) .
such large scale ratios translate into nlo corrections whose size relative to the lo contribution go as @xmath475 and @xmath476 .
the picture seen above of good agreement between parton shower monte carlo and fixed - order predictions does not always hold .
events with vector bosons are among those that parton shower programs have the greatest difficulty reproducing .
this is illustrated in fig .
[ fig : mangano - z - multijet ] ( left ) , which shows the ` integrated @xmath477 spectrum for the @xmath478 jet ' in events with a @xmath94-boson , i.e. , the cross section for the @xmath478 jet to have a transverse energy above @xmath477 . results are given both from herwig and from alpgen , which provides an exact lo ( tree - level ) prediction for each jet s spectrum .
jet to have a transverse energy above a given @xmath477 , in ( 14 tev ) lhc events with a @xmath94-boson , as calculated with herwig and a tree - level ( lo ) prediction from alpgen . figure taken from ref .
right : kinematic configurations contributing to @xmath479jet and @xmath426jet events . ]
@xmath480 : jet to have a transverse energy above a given @xmath477 , in ( 14 tev ) lhc events with a @xmath94-boson , as calculated with herwig and a tree - level ( lo ) prediction from alpgen .
figure taken from ref .
right : kinematic configurations contributing to @xmath479jet and @xmath426jet events . ]
@xmath481 : jet to have a transverse energy above a given @xmath477 , in ( 14 tev ) lhc events with a @xmath94-boson , as calculated with herwig and a tree - level ( lo ) prediction from alpgen .
figure taken from ref .
right : kinematic configurations contributing to @xmath479jet and @xmath426jet events .
, title="fig : " ] jet to have a transverse energy above a given @xmath477 , in ( 14 tev ) lhc events with a @xmath94-boson , as calculated with herwig and a tree - level ( lo ) prediction from alpgen . figure taken from ref .
right : kinematic configurations contributing to @xmath479jet and @xmath426jet events .
, title="fig : " ] the distribution for the first jet is fine : this is by construction , since herwig ( like pythia ) includes the full matrix element for @xmath482 production .
what is shocking is the result for the second ( and higher ) jets , for which the @xmath477 spectra are in complete disagreement between herwig and alpgen . at first sight
it is mysterious how herwig could be doing such a good job for pure jet production , fig .
[ fig:3rd - jet ] , yet such a poor job when there s also a @xmath94-boson in the event , fig .
[ fig : mangano - z - multijet ] . the diagrams in the right - hand part of fig .
[ fig : mangano - z - multijet ] help explain what s going on .
herwig generates hard configurations like those in the upper line , labelled z+j .
events with two jets are generated in herwig by emission of a gluon off ( say ) the high-@xmath414 quark .
however , there are also events ( bottom right ) which look like a dijet event with a @xmath94-boson radiated off a quark line .
since at high @xmath414 the @xmath94-boson s mass becomes irrelevant in its production , such diagrams acquire soft and collinear enhancements ( just like gluon radiation ) .
however , today s parton - shower monte carlos only include qcd showering , not electroweak showering and therefore they are never in a position to start from a dijet event and radiate a @xmath94-boson from it .
therefore they miss a very large part of the cross section .
this example helps illustrate a general feature of the use of monte carlos : if you are to trust the results , it is crucial that you know what you have asked the monte carlo to generate and whether the observable you are interested in is truly likely to be dominated by what the monte carlo can generate . in the above subsections we saw various strengths and weaknesses of different predictive techniques :
nlo codes give predictions with well controlled normalizations , for a reasonable range of processes , as long as one is nt faced with observables that involve disparate scales .
tree - level ( lo ) predictions can be generated up to quite high multiplicities for a broad range of processes , though without good control of the normalization ( i.e. , often no better than a factor of two ) . and
parton shower monte carlos provide reliable behaviour in soft - collinear regions , giving a fully exclusive final state , though they have normalizations which at best are no better than lo normalizations and sometimes they do dramatically badly in reproducing multi - jet structure .
it is natural to ask whether one can develop tools that combine ( or merge ) the advantages of all three .
this is an active research topic , and here we will just outline the ideas behind two well - established merging tasks : combination of different multiplicity lo matrix elements with parton showers ; and combination of nlo and parton shower predictions .
suppose you ask for @xmath94+jet production as the initial hard process in pythia or herwig .
as we saw above , these programs contain the correct matrix element ( me ) for @xmath94+parton production , but do a very bad job of estimating @xmath94 + 2jet production .
+ parton and @xmath94 + 2-parton events ] one naive solution to this problem would be to generate @xmath94 + 2-parton events with alpgen , madgraph , or some other preferred lo me tool and then ask herwig or pythia to shower those configurations .
however , if one showers both @xmath94+parton and @xmath94 + 2-parton events , then one is faced with a double - counting issue , as illustrated in fig .
[ fig : me+ps - double - counting ] . in some events ( left )
the showering off the @xmath94+parton configuration just leads to soft and collinear emissions . similarly off the @xmath94 + 2-parton events ( middle ) . however , sometimes ( right ) the showering off the @xmath94+parton configuration leads to the production of a relatively hard , large - angle gluon that is in the same phase - space region that is already covered in the @xmath94 + 2-parton event sample .
two main methods exist to avoid this double counting : ckkw matching @xcite and mlm matching @xcite .
the latter , named after its inventor , m. l. mangano , is the one we will describe here ( it is the simpler of the two ) .
let s examine the basics of how it proceeds : * introduce a ( dimensionful ) transverse momentum cutoff scale @xmath483 and a ( dimensionless ) angular cutoff scale @xmath484 for matrix element generation .
* generate tree - level events for @xmath94 + 1-parton , @xmath94 + 2-partons , up to @xmath94+@xmath485-partons , where all partons must have @xmath486 and be separated from other partons by an angle greater than @xmath484 ( we will discuss the definition of this ` angle ' later in section [ sec : jets ] ) .
the numbers of events that one generates in the different samples are in proportion to their cross sections with these cuts . * for each tree - level event from these samples , say one with @xmath185 partons , shower it with your favourite parton - shower program . *
apply a jet algorithm to the showered event ( choose the algorithm s angular reach @xmath153 to be @xmath487 ) and identify all jets with @xmath488 , where the merging scale @xmath489 is to be taken @xmath490 .
* if each jet corresponds to one of the partons ( i.e. , is nearby in angle ) and there are no extra jets above scale @xmath489 , then accept the event .
( for the sample with @xmath491 , the condition is that there should be no extra jets with @xmath492 . ) * otherwise , reject the event . , with the @xmath489 @xmath414 cutoff represented by the dashed line ] the action of the mlm procedure on the events of fig .
[ fig : me+ps - double - counting ] is illustrated in fig .
[ fig : mlm - in - action ] , showing which events would be accepted and which ones rejected .
one immediately sees how the double - counting issue disappears : in the rightmost event , the showering of a @xmath94+parton event leads to an extra jet above @xmath489 ; since this event now has more jets above @xmath489 than it had partons , it is rejected .
in contrast the middle event , which also has two jets above @xmath489 , was generated from a @xmath94 + 2-parton event and is accepted .
so is the leftmost event with only one jet , starting from a @xmath94 + 1-parton event . by providing a remedy for the double - counting issue , ensuring that the hard jets always come just from the matrix element
, the mlm procedure also ensures that hard jets above @xmath489 have distributions given by the tree - level matrix - element calculations .
the rejection of extra jets also plays another important role : when there are big differences in scales between the jets and @xmath489 ( or between different jets ) , the monte carlo showering would want to ` fill up ' that phase space with emissions . however , whenever it does so , the event gets rejected by the matching procedure . as long as the monte carlo is carrying out a reasonable showering of these multi - parton events ,
then the procedure is equivalent to the introduction of a sudakov form factor encoding the probability of not emitting radiation .
the above ` mlm ' merging of matrix - elements and parton showers ( meps ) is the main procedure available with alpgen , for use with both herwig and pythia . it is also provided ( in a variant form ) within madgraph .
the sherpa monte carlo also has its own matrix - element generator(s ) and provides ` ckkw ' meps matching @xcite , which instead of the veto steps of mlm matching , uses an analytical calculation of the sudakov form factors . these and other matrix - element / parton - shower merging schemes are discussed in detail in ref .
they all share the feature of a matching scale to separate the region under the ` responsibility ' of matrix elements and that delegated to the parton shower . in all cases physics predictions well above the matching scale
should be independent of the scale .
additionally , distributions at scales around the matching scale should be reasonably smooth , as long as the matching scale has been chosen in a region where the parton shower can be expected to give a reasonable description of emission ( for caveats , see ref .
@xcite ; for a method that avoids the need for a matching scale , see ref .
@xcite ) .
+ 2-jet events at the tevatron , differential in the transverse momentum of the second jet , as measured by d @xcite .
comparisons are shown to lo and nlo predictions ( from mcfm ) , parton - shower predictions ( pythia , herwig ) and merged matrix - element + parton - shower predictions ( alpgen+pythia , sherpa ) . ]
meps predictions ( as well as other predictive methods ) are compared to experimental results for @xmath94 + 2-jet production in fig .
[ fig : meps - in - action ] ( bottom right ) .
the meps results show good agreement for the shape of the observable in question , the @xmath414 distribution of the second jet , and they are much more successful than plain parton - shower predictions .
since their normalizations are based on lo calculations , they do , however , suffer from substantial scale dependence , which is much larger than the scale uncertainty one would obtain at nlo ( bottom left ) . as a result of their considerable success in describing the shapes of experimental distributions ,
meps predictions have become one of the main tools for a broad range of tevatron and lhc analyses , especially those involving complex final states .
we ve seen how to obtain a parton shower structure with lo accuracy for a range of different jet multiplicities .
however , given the large uncertainty on the normalizations of the predictions , there would be considerable advantages to obtaining nlo accuracy .
one might think that since we ve had nlo predictions and parton shower predictions for a couple of decades , such a task should not be too hard .
the main difficulty comes from the fact that nlo predictions involve divergent event weights ( cf .
[ fig : nlojet - weights ] ) , which are nt even positive definite .
two approaches are in use to get around this problem , the mc@nlo @xcite and powheg @xcite methods .
the idea behind the mc@nlo approach is to ` expand ' the monte carlo parton shower to first order in @xmath0 .
i.e. , the monte carlo s parton showers already contain some ( partially wrong ) estimate of the true nlo corrections and the aim is to figure out what that estimate is .
this requires a deep understanding of the monte carlo program . as a next step ,
one calculates the difference between the true nlo contributions and the partial ones included in the monte carlo .
one of the keys to the mc@nlo method is that as long as the monte carlo gives the correct pattern of soft and collinear splitting ( which it is supposed to ) , then the differences between true nlo and monte carlo partial nlo should be _
then one can generate partonic configurations with phase - space distributions proportional to those finite differences and shower them .
symbolically , if we imagine a problem with one phase - space variable , say energy @xmath196 , then we can write the ` expansion ' of the monte carlo cross section as @xmath493 where @xmath494 is the coefficient of @xmath0 for real emission of a gluon with energy @xmath196 in the monte carlo and @xmath495 is the ( divergent ) coefficient of the virtual corrections . the mc@nlo prediction
is then given by @xmath496 where @xmath497 and @xmath498 are the true nlo real and virtual coefficients .
each term in ( small ) brackets in eq .
( [ eq : mc@nlo ] ) should separately be finite and corresponds to a given topology of event to be showered : a lo topology for the `` @xmath149 '' and @xmath499 terms , and a nlo real topology for the @xmath500 term .
( a more complete and clearer ` toy ' explanation is given in the original mc@nlo paper@xcite , which makes for very instructive reading . )
the mc@nlo approach has the practical characteristic that all event weights are @xmath501 .
quite a range of processes are available in the mc@nlo approach for herwig , including the production of a higgs boson , single and double vector - bosons , a heavy - quark pair , various single - top processes , and @xmath502 and @xmath503 .
the characteristic of these processes is that they are nearly all free of light jets at lo ( except for one of the single - top processes ) , because this simplifies the set of divergences that need to be dealt with .
very recently one pythia @xcite and a couple of herwig++ processes ( ref .
@xcite and references therein ) were also interfaced within the mc@nlo approach .
an alternative to mc@nlo is powheg @xcite .
it aims to avoid the ( small fraction of ) negative weights that are present in mc@nlo and also seeks to be less tightly interconnected with a specific monte carlo program .
the principle of the powheg approach is to write a simplified monte carlo that generates just one emission beyond lo .
this single - emission monte carlo is designed in such a way as to be sufficient to give the correct nlo result .
it essentially works by introducing a sudakov form factor such as eq .
( [ eq : sudakov - exponentiated ] ) in which the contents of the square bracket are replaced by the integral over the exact real radiation probability above @xmath292 ( plus a constant term that accounts for the finite part of the 1-loop correction )
. then emissions with transverse momenta below the scale of the first emission are left to the default monte carlo program , for example herwig or pythia ( implementing a transverse momentum veto to ensure that nothing is generated above the scale of the first , powheg , emission ) .
the range of processes available within powheg is gradually catching up with that for mc@nlo , and it is hoped that this will be helped by the availability of systematic tools to help with the development of new processes @xcite .
pairs in the mc@nlo approach , compared to the plain herwig result ( rescaled by the @xmath504 @xmath505-factor ) and to the nlo calculation .
shown for a @xmath36 lhc .
figure taken from ref .
@xcite . ] an illustration of a result with mc@nlo is given in fig .
[ fig : mcatnlo - result ] for the transverse - momentum distribution of a @xmath506 pair .
it illustrates how mc@nlo reproduces the herwig shape in the low-@xmath414 region ( with nlo normalization ) , the nlo distribution at high @xmath414 , and that neither herwig nor plain nlo are able to properly describe both regions .
in this section , we have seen quite a range of different predictive methods for qcd at hadron colliders .
two big classes of predictive methods exist : partonic fixed - order calculations , which have well controlled accuracy , but wildly fluctuating positive and negative event weights ; and monte carlo parton shower tools , which give a much more complete description of events ( and uniform event weights ) .
development is active on both sets of tools individually .
on one hand we ve mentioned the challenge of having a broader range of processes known at nlo and a handful at nnlo . and
though we have not really touched on it , there is also a very active programme to develop parton shower monte carlos in c++ as replacements for venerable but ageing fortran codes like pythia 6.4 and herwig 6.5 .
in addition there are methods with the advantages of both fixed - order and parton - shower programs .
it is widespread nowadays to merge different lo ` tree - level ' predictions ( e.g. , @xmath94+parton , @xmath94 + 2-partons , @xmath94 + 3-partons , etc . ) together with parton showers . and for simple processes , those with at most one light parton in the final state , it is possible to combine nlo accuracy with the full parton - shower description .
ultimately the hope is to be able to combine @xmath94+jet , @xmath94 + 2-jets , @xmath94 + 3-jets all at nlo accuracy also merging them with parton showers , so as to obtain accurate descriptions for the whole range of processes that are relevant at the lhc , both as backgrounds and as signals of new particles and new physics .
the concept of a jet has already arisen in various contexts , so in this final section we will examine jet - related ideas in more detail .
event that can be interpreted as having a 2-jet , @xmath61-like structure ; middle : an event that can be interpreted as having a 3-jet , @xmath462 , structure ; right : the same event reinterpreted as having a 4-jet structure , @xmath507.,title="fig : " ] event that can be interpreted as having a 2-jet , @xmath61-like structure ; middle : an event that can be interpreted as having a 3-jet , @xmath462 , structure ; right : the same event reinterpreted as having a 4-jet structure , @xmath507.,title="fig : " ] event that can be interpreted as having a 2-jet , @xmath61-like structure ; middle : an event that can be interpreted as having a 3-jet , @xmath462 , structure ; right : the same event reinterpreted as having a 4-jet structure , @xmath507.,title="fig : " ] consider the three events of fig .
[ fig : jet - ambiguities ] . in the left - hand one ,
one interpretation is that we re seeing an @xmath95 event , in which there has been soft and collinear showering followed by a transition to hadrons .
this is a classic picture of a ` 2-jet ' event .
the middle event is more complex : energy flow is not limited to two cones .
one interpretation of the event is that a @xmath61 pair has emitted a hard gluon @xmath508 , and all three have undergone soft and collinear showering .
however , the same event can also be interpreted ( right ) as a @xmath509 event , with further soft and collinear showering .
deciding between these two interpretations means choosing just how hard and separated in angle an emission has to be in order for it to be considered a separate jet ( cf . the angular and energy parameters , @xmath192 and @xmath106 , in our discussion of the 2-jet cross section in section [ sec : ir - safe - obs ] ) .
in making this choice , it would be highly painful to visually inspect each of the @xmath510 events written to disk every year at the lhc .
instead one uses a set of rules , a ` jet definition ' , by which a computer can take a list of particle momenta for an event ( be they quark and gluons , or hadrons , or even calorimeter deposits ) , and return a list of jets .
if one modifies an event just through soft and collinear emission , then the set of jets should not change , i.e. , the result of applying the jet definition should be insensitive to the most common effects of showering and hadronization , as illustrated in fig . [
fig : jet - projections ] .
jets are central to collider physics : both theory and experimental results are often presented in terms of jet cross sections , and thus jets provide the meeting point between the two . as we saw in section
[ sec : matrix - elements - with - parton - showers ] , jets are also used to assemble together different kinds of theory predictions . and jets are an input to almost all physics analyses : to new physics searches ( since new particles may decay to quarks or gluons , giving jets ) , in higgs searches , top physics , monte carlo validation , fits of pdfs , etc .
the construction of a jet involves different considerations : * which particles are grouped together into a common jet ?
the set of rules that one follows for deciding this is usually known as a jet algorithm , and it comes with parameters that govern its exact behaviour .
a common parameter is @xmath153 which determines the angular reach of the jet algorithm .
* how do you combine the momenta of particles inside a jet ?
one needs to specify a ` recombination scheme ' .
the most common is to simply add the 4-vectors of the particles ( the ` e - scheme ' ) .
this gives jets that are massive ( so jets can not be thought of as a direct stand - in for partons , which are massless ) .
taken together , the algorithm , its parameters and the recombination scheme specify a ` jet definition ' .
two broad classes of jet definition are in common use : cone algorithms , which take a top - down approach , and sequential recombination algorithms , based on a bottom - up approach
. below we ll give a brief discussion of each of kind of algorithm , referring the reader to ref .
@xcite for a more complete description of all the variants that exist .
there are many types of cone algorithm , but all rely on the idea that soft and collinear branching does nt modify the basic direction of energy flow
. one of the simpler cone algorithms ( we ll call it ic - pr , for iterative cone with progressive removal of particles ) is that used by the cms experiment during much of their preparation for lhc running .
one first sorts all particles according to their transverse momentum , and identifies the one with largest transverse momentum .
this is referred to as a seed particle , @xmath511 .
one draws a cone of radius @xmath153 around the seed in hadron - collider variables this means identifying all particles with @xmath512 , where @xmath513 is the rapidity of particle @xmath109 , @xmath514 its azimuth , and @xmath515 and @xmath516 the rapidity and azimuth of the seed .
one then identifies the direction of the sum of the momenta of those particles .
if it does nt coincides with the seed direction then one uses that sum as a new seed direction , and iterates until the sum of the cone contents coincides with the previous seed .
this is referred to as finding a stable cone .
the particles inside that stable cone make a jet , and they re removed from the list of particles in the event .
the procedure is then repeated on the particles that remain , removing particles each time one finds a stable cone ( @xmath517 jet ) , until no particles remain and one has the complete set of jets . of these jets one retains only those above some transverse - momentum threshold @xmath518 . for all particles here ) . left : expectations for the behaviour of a collinear - safe jet algorithm , where the jets found by the algorithm should be independent on the collinear splitting of the hardest particles .
right : in a collinear unsafe algorithm such as the ic - pr type cms cone , the splitting of the central hard particle causes the leftmost particle to become the hardest in the event , leading to a two - jet rather than a one - jet event . ]
there is one major drawback to the above procedure : the use of the particles @xmath414 s to decide which one to take as the first seed .
this is problematic , because particle @xmath414 s are not collinear safe quantities . as illustrated in the two right - hand events of fig .
[ fig : coll - unsafety ] , in an ic - pr algorithm , if the hardest particle undergoes a collinear splitting then this can result in another particle in the event becoming the ` new ' hardest particle , giving a different set of final jets as compared to events without the splitting . thus in the example of fig .
[ fig : coll - unsafety ] there is a divergent ( real , positive ) contribution to the 2-jet cross section and a separate divergent ( 1-loop virtual , negative ) contribution to the 1-jet cross section .
in contrast , for a collinear - safe algorithm ( two leftmost events ) , the collinear - splitting of the hardest particle does not change the set of final jets .
then the real and virtual divergences both contribute to the 1-jet cross section and so cancel each other .
collinear unsafety means that certain cross sections can not be calculated at nlo ( or sometimes nnlo ) one will only obtain nonsensical infinite answers .
furthermore , even if one is working at some low perturbative order which is not divergent ( e.g. , lo ) , the fact that higher orders diverge means that the convergence of the whole perturbative series becomes questionable , compromising the usefulness even of the low orders . over the past two decades
there has been significant discussion of such problems .
there are many other variants of cone algorithm , and nearly all suffer from problems either of collinear safety , or infrared safety .
one class that has been widely used at the tevatron avoids the ordering of initial seeds , and instead obtains stable cones using all possible particles as seeds : stable cones are not immediately converted into jets , but instead , once one has the list of the stable cones found by iterating from all possible seeds one then uses a ` split merge ' procedure to decide how particles that appear in multiple stable cones should be unambiguously assigned to a single jet . and two hard partons .
the addition of a soft gluon converts the event from having two jets to just one jet .
in contrast to fig . [
fig : coll - unsafety ] , here the explicit angular structure is shown ( rather than @xmath414 as a function of rapidity ) . ]
this procedure avoids the collinear - safety issue ( the order of particles @xmath414 s no longer matters ) , however , it turns out that it instead introduces an infrared - safety issue : adding an extra soft particle creates a new seed , which can lead to an extra stable cone being found , which feeds through the split merge procedure , altering the final set of ( hard ) jets .
this tends to happen when two hard particles are separated by a distance @xmath519 that satisfies @xmath520 ( so that a cone centred on either fails to capture the other and each hard particle leads to its own jet ) and one adds a soft particle in between the two ( so that a cone centred on the soft particle includes both hard ones , which then end up in a single jet ) , as illustrated in fig . [
fig : ir - unsafety ] .
a partial fix for the problem was given in ref .
@xcite ( adopted by the tevatron in ref .
@xcite ) , which involves adding extra seeds at the midpoints between all pairs of stable cones and uses those as new starting points for finding additional stable cones before starting the split merge step .
a full fix involves a non - seed - based approach to exhaustively finding all possible stable cones in an event , in an algorithm known as the seedless infrared safe cone ( siscone ) @xcite .
sequential - recombination jet algorithms take a bottom - up approach to constructing jets , as if they were inverting the sequence of splittings of the parton shower . of course
that sequence does nt really exist unambiguously , since gluon emission is a quantum - mechanical process involving coherent emission from all colour sources in an event .
however , for collinear emissions the picture that there is a single ` emitter ' is not a poor approximation .
the most widely used sequential recombination algorithm to date is the @xmath292 algorithm , originally formulated for @xmath93 events @xcite .
recall , from eq .
( [ eq : fs - universal - splitting ] ) , that the soft and collinear limit of the gluon - emission probability for @xmath521 is @xmath522 where @xmath523 is @xmath524 ( @xmath525 ) if @xmath6 is a gluon ( quark ) , and where we ve written @xmath526 in the denominator to avoid specifying which of @xmath109 and @xmath402 is the soft particle .
the essential idea of the @xmath292 algorithm is to define a ` distance measure ' @xmath527 between every pair of particles @xmath528 , @xmath529 in the collinear limit , @xmath527 reduces to @xmath530 , which is the relative transverse momentum between particles @xmath109 and @xmath402 ( hence the name @xmath292 algorithm ) , normalized to the total visible ( or sometimes centre - of - mass ) energy @xmath83 .
apart from the normalization , this is just what appears in the denominator of the splitting probability , eq .
( [ eq : splitting - again ] ) , so that pairs of particles that would arise from a splitting with a strong divergence are considered very close together .
the algorithm works by identifying the pair of particles that has the smallest @xmath527 and recombining them into a single particle ( also called a ` pseudojet ' ) .
it then recalculates all @xmath527 distances taking into account the new particle , and again recombines the pair that s closest .
the procedure is repeated until all @xmath527 distances are above some threshold @xmath531 , at which point the pseudojets that remain become the event s jets . for hadron collisions and deep - inelastic scattering , the version of the @xmath292 algorithm that is most commonly used reads as follows @xcite . for every pair of particles
define a ( dimensionful ) inter - particle distance @xmath532 , @xmath533 where @xmath153 is a parameter whose role is similar to that of @xmath153 in cone algorithms . also define a beam distance for every particle , @xmath534 the algorithm proceeds by searching for the smallest of the @xmath532 and the @xmath535 .
if it is a @xmath532 then particles @xmath109 and @xmath402 are recombined into a single new particle .
if it is a @xmath535 then @xmath109 is removed from the list of particles , and called a jet .
this is repeated until no particles remain .
note that the distance in eq .
( [ eq : dij - kt ] ) just reduces to that of eq .
( [ eq : yij - kt - alg ] ) in the collinear limit ( modulo @xmath224 normalization ) .
so one is still dealing with the relative transverse momentum between pairs of particles . as with cone algorithms , in this ` inclusive longitudinally invariant @xmath292 algorithm , ' arbitrarily soft particles can form jets .
it is therefore standard to place a @xmath536 cutoff on the jets one uses for ` hard ' physics .
one can verify that @xmath153 in the @xmath292 algorithm plays a similar role to @xmath153 in cone algorithms , using the following observations : if two particles @xmath109 and @xmath402 are within @xmath537 of each other , i.e. , @xmath538 , then @xmath539 and so @xmath109 and @xmath402 will prefer to recombine rather than forming separate jets . if a particle @xmath109 is separated by more than @xmath153 from all other particles in the event then it will have @xmath540 for all @xmath402 and so it will form a jet on its own .
@xmath140 plane covered by jets in an illustrative ( simulated ) hadron - collider event with an ic - pr type cone algorithm ( left ) and the inclusive longitudinally - invariant @xmath292 algorithm ( right ) .
the jet finding was carried out with the fastjet package @xcite , with the addition of very soft ghost particles to trace the extent of the jets @xcite .
, title="fig : " ] @xmath140 plane covered by jets in an illustrative ( simulated ) hadron - collider event with an ic - pr type cone algorithm ( left ) and the inclusive longitudinally - invariant @xmath292 algorithm ( right ) .
the jet finding was carried out with the fastjet package @xcite , with the addition of very soft ghost particles to trace the extent of the jets @xcite .
, title="fig : " ] despite this similarity to the behaviour of cone algorithms for pairs of particles , the @xmath292 algorithm gives jets that ` look ' somewhat different .
figure [ fig : shapes ] illustrates what happens when one clusters a simulated hadron - collider event with an ic - pr type cone algorithm and with the @xmath292 algorithm . in both cases the figure shows ( in a given colour ) the calorimeter cells that are included in each jet . for the ic - pr algorithm
the hardest jets all appear circular , as expected given the use of cones in the definition of the algorithm ( in cone algorithms with split merge steps , the jets are often not circular , because of the less trivial mapping from stable cones to jets ) . for the @xmath292 algorithm ,
the jets have quite irregular ( or jagged ) edges , because many of the soft particles cluster together early in the recombination sequence ( owing to their small @xmath414 and hence @xmath532 ) in patterns that are determined by the random distributions of those particles in @xmath414 and rapidity and azimuth .
the irregularity of the jet boundaries has often been held against the @xmath292 algorithm .
one reason is that it makes it harder to calculate acceptance corrections : for example , if you know that some part of a detector is misbehaving , it is not obvious how far a @xmath292 jet must be from that part of the detector in order not to be affected by it .
another reason relates to the linearity of the dependence of the jet momentum on soft - particle momenta : in the ic - pr algorithm , the hard core of the jet essentially determines the jet boundary , and the algorithm depends linearly on the momenta of any soft particles within the boundary , and is independent of particles outside it . in the @xmath292 algorithm , varying the momentum of one soft particle in the vicinity of the jet core can affect whether it and other soft particles get clustered with that core or not .
this can complicate energy calibrations for the jet algorithm , though techniques exist to correct for this to some extent ( jet - area - based subtraction @xcite , which leaves just a residual term known as back - reaction @xcite ) .
a feature of the @xmath292 algorithm that is attractive is that it not only gives you jets , but also assigns a clustering sequence to the particles within the jet .
one can therefore undo the clustering and look inside the jet .
this has been exploited in a range of qcd studies ( e.g. , ref .
@xcite ) , and also in discussions of searches of hadronic decays of boosted massive particles such as @xmath376 , @xmath541 , or @xmath94 bosons , top quarks , or new particles ( early examples include refs . @xcite ; for more recent examples , see the reviews in refs .
jet substructure studies are also often carried out with the cambridge / aachen ( c / a ) algorithm @xcite , which essentially replaces @xmath542 in eqs .
( [ eq : dij - kt],[eq : dib - kt ] ) but is otherwise like the @xmath292 algorithm .
it turns out that it is possible to design a sequential - recombination algorithm with many of the nice properties of cone algorithms via a simple modification of the @xmath292 algorithm s distance measures @xcite : @xmath543 the correspondence with the divergences of eq .
( [ eq : splitting - again ] ) is partially lost : objects that are close in angle still prefer to cluster early , but that clustering tends to occur with a hard particle ( rather than necessarily involving soft particles ) .
this means that jets ` grow ' in concentric circles out from a hard core , until they reach a radius @xmath153 , giving circular jets just as with the ic - pr cone , as shown in fig .
[ fig : antikt ] .
however , unlike the ic - pr cone , this ` anti-@xmath292 ' algorithm is collinear ( and infrared ) safe , meaning that it is safe to use with fixed - order qcd predictions .
this , combined with the fact that it has been implemented efficiently in the fastjet jet - finding code @xcite , has led to it being adopted as the default jet algorithm by both the atlas and cms collaborations . , but shown for the anti-@xmath292 algorithm ] note that the anti-@xmath292 algorithm does not provide useful information on jet substructure : if a jet contains two hard cores , then the @xmath292 ( or c / a ) algorithms first reconstruct those hard cores and merge the resulting two subjets
. the anti-@xmath292 algorithm will often first cluster the harder of the two cores and then gradually agglomerate the contents of the second hard core . because of the intricacies of calibrating jets , past experiments have tended to concentrate their efforts on just one or two jet definitions , which are used across the board for qcd studies , top - quark studies , higgs and new physics searches . typically the choices have been for cone - type algorithms with @xmath153 values in the range 0.40.7 . at the lhc there are prospects of experimental methods ( for example topoclusters at atlas and particle flow at atlas ) that make it easier to use a broad range of jet definitions .
we ve already mentioned , above , the use of jet substructure in new physics searches .
these methods will definitely benefit from experiments flexibility in choosing different jet definitions ( for example , many of the proposed searches use quite a large jet radius , @xmath544 11.5 ) . more generally , when using jets , it is important to establish what you are trying to get out of the jet algorithm , what it is that you re trying to optimize .
different studies will want to optimize different things .
for example in qcd studies , such as the inclusive jet - spectrum measurement that goes into pdf fits , one criterion might be to choose a jet definition that minimizes the non - perturbative corrections that need to be applied when comparing perturbative predictions with data .
resonance decaying to @xmath61 ( top ) and a @xmath545 resonance decaying to @xmath546 ( bottom ) .
the @xmath547 indicates the width of the region that contains @xmath548 of the resonance events ( corresponding to about @xmath549 of the resonance events that pass basic selection cuts ) .
smaller numbers imply better reconstruction .
figure taken from ref .
many further plots available from ref .
, title="fig : " ] + resonance decaying to @xmath61 ( top ) and a @xmath545 resonance decaying to @xmath546 ( bottom ) .
the @xmath547 indicates the width of the region that contains @xmath548 of the resonance events ( corresponding to about @xmath549 of the resonance events that pass basic selection cuts ) .
smaller numbers imply better reconstruction .
figure taken from ref .
many further plots available from ref .
, title="fig : " ] in searches for new physics , for example when looking for a resonance in the dijet invariant mass spectrum , the criterion might be the following : given a narrow resonance , what jet definition will lead to the narrowest peak in the dijet mass spectrum ?
it turns out that the answer depends significantly on properties of the resonance you re trying to reconstruct .
figure [ fig : quality ] illustrates how siscone with @xmath550 does well for reconstructing a @xmath356 resonance that decays to @xmath61 . while for a @xmath545 resonance decaying to @xmath546 , you re better off with an algorithm ( such as siscone , or others supplemented with the trick of filtering and/or variants @xcite ) with a substantially larger radius , @xmath551 @xcite . and
if you find a resonance , you might then want to know which jet definition will allow you to measure its mass most accurately , which may not be the same choice that allowed you to find the resonance most easily .
understanding analytically which jet choices work well is the subject of ongoing theoretical work @xcite , which involves figuring out how perturbative radiation , hadronization , the underlying event and pileup all come together to affect jet momenta .
in these lectures we have covered material that ranges from the qcd lagrangian through to @xmath93 scattering , soft and collinear divergences , pdfs , fixed - order calculations , parton showers and jets . it is my hope that the material here will have given you enough background information to at least understand the basics of these different concepts .
the aspects of qcd that we have examined will play a role in nearly all analyses at the lhc . whether because monte carlo parton - shower programs are used for estimating detector effects on almost every measurement ; or because in searches for new physics one might look for an excess with respect to qcd predictions of backgrounds ; or perhaps because an understanding of qcd will make it possible to more clearly pull out a signal of new physics that involves qcd final states .
if you want to find out more about the topics discussed here ( or those left out due to time constraints , notably heavy quarks ) , good places to look include the textbooks @xcite mentioned at the beginning of these lectures , the references given throughout , and transparencies from the various summer schools dedicated to qcd , for example , the cteq ( and mcnet ) schools @xcite .
i wish to thank the organizers of the 2009 european school of high - energy physics for their warm hospitality during the school itself and for their kind patience in waiting for these lecture notes to be completed .
i am also grateful to the physics department of princeton university for hospitality while finishing these lecture notes .
this work was supported by the french agence nationale de la recherche , under grant anr-09-blan-0060 .
r. k. ellis , w. j. stirling and b. r. webber , camb .
monogr . part .
cosmol .
* 8 * ( 1996 ) 1 .
g. dissertori , i. g. knowles and m. schmelling , high - energy experiments and theory ( clarendon , oxford , 2003 ) .
j. m. maldacena , adv .
* 2 * ( 1998 )
231 [ int . j. theor .
* 38 * ( 1999 ) 1113 ] [ arxiv : hep - th/9711200 ] .
s. s. gubser , i. r. klebanov and a. m. polyakov , phys .
* b428 * ( 1998 ) 105 [ arxiv : hep - th/9802109 ] .
e. witten , adv .
* 2 * ( 1998 ) 253 [ arxiv : hep - th/9802150 ] .
j. erlich , e. katz , d. t. son and m. a. stephanov , phys .
* 95 * ( 2005 )
261602 [ arxiv : hep - ph/0501128 ] . l. da rold and a. pomarol , nucl .
phys . * b721 * ( 2005 ) 79 [ arxiv : hep - ph/0501218 ] .
s. durr _ et al .
_ , science * 322 * ( 2008 ) 1224 .
j. smit , introduction to quantum fields on a lattice : a robust mate , cambridge lect .
notes phys .
* 15 * ( 2002 ) 1 .
t. degrand and c. e. detar , lattice methods for quantum chromodynamics ( world scientific , singapore , 2006 ) .
v. n. gribov , lectures at the 12th winter school of the leningrad nuclear physics institute , slac - trans-176 ( 1977 ) ; s. d. drell , _ in _ a festschrift for maurice goldhaber , trans .
ny acad .
, series ii , * 40 * ( 1980 ) 76 .
n. k. nielsen , am .
j. phys .
* 49 * ( 1981 ) 1171 ; r. j. hughes , phys .
* b97 * ( 1980 ) 246 ; _ idem _ nucl .
* b186 * ( 1981 ) 376 ; k. johnson , _ in _ asymptotic realms of physics , edited by a. guth , k. huang and r. l. jaffe ( mit press , cambridge , ma , 1983 ) .
s. bethke , eur .
j. * c64 * ( 2009 ) 689 [ arxiv:0908.1135 [ hep - ph ] ] .
s. g. gorishnii , a. l. kataev and s. a. larin , phys .
b * 259 * , 144 ( 1991 ) .
l. r. surguladze and m. a. samuel , phys .
lett . * 66 * , 560 ( 1991 ) [ erratum - ibid .
* 66 * , 2416 ( 1991 ) ] .
t. hebbeker , m. martinez , g. passarino and g. quast , phys . lett .
b * 331 * , 165 ( 1994 ) .
k. g. chetyrkin , j. h. kuhn and a. kwiatkowski , phys .
* 277 * ( 1996 ) 189 .
a. baikov , k. g. chetyrkin and j. h. kuhn , phys . rev . lett .
* 101 * ( 2008 ) 012002 [ arxiv:0801.1821 [ hep - ph ] ] .
a. baikov , k. g. chetyrkin and j. h. kuhn , arxiv:0906.2987 [ hep - ph ] .
v. s. fadin , yad .
* 37 * ( 1983 ) 408 ; b. i. ermolaev and v. s. fadin , jetp lett .
* 33 * ( 1981 ) 269 [ pisma zh .
fiz . * 33 * ( 1981 ) 285 ] ; a. h. mueller , phys .
* b104 * , 161 ( 1981 ) ; y. l. dokshitzer , v. s. fadin and v. a. khoze , z. phys .
* c15 * ( 1982 ) 325 ; a. bassetto , m. ciafaloni and g. marchesini , phys .
* 100 * ( 1983 ) 201 .
m. arneodo _ et al . _
[ new muon collaboration ] , nucl .
* b483 * ( 1997 ) 3 [ arxiv : hep - ph/9610231 ] .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky and w. k. tung , jhep * 0207 * ( 2002 ) 012 [ arxiv : hep - ph/0201195 ] . s. chekanov _ et al .
_ [ zeus collaboration ] , eur .
j. * c21 * ( 2001 ) 443 [ arxiv : hep - ex/0105090 ]
. w. k. tung , h. l. lai , a. belyaev , j. pumplin , d. stump and c. p. yuan , jhep * 0702 * ( 2007 ) 053 [ arxiv : hep - ph/0611254 ] .
a. d. martin , r. g. roberts , w. j. stirling and r. s. thorne , phys .
* b604 * ( 2004 ) 61 [ arxiv : hep - ph/0410230 ] .
v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
* 101 * ( 2008 ) 062001 [ arxiv:0802.2400 [ hep - ex ] ] .
a. abulencia _ et al .
_ [ cdf - run ii collaboration ] , phys . rev . *
d75 * ( 2007 ) 092006 [ erratum-_ibid .
_ * d75 * ( 2007 ) 119901 ] [ arxiv : hep - ex/0701051 ] .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , eur .
j. * c64 * , 653 ( 2009 ) [ arxiv:0905.3531 [ hep - ph ] ] .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , eur .
j. * c63 * ( 2009 ) 189 [ arxiv:0901.0002 [ hep - ph ] ] .
p. m. nadolsky _ et al .
_ , phys .
* d78 * ( 2008 ) 013004 [ arxiv:0802.0007 [ hep - ph ] ] .
r. d. ball , l. del debbio , s. forte , a. guffanti , j. i. latorre , j. rojo and m. ubiali , arxiv:1002.4407 [ hep - ph ] . g. p. salam and j. rojo , comput .
commun . * 180 * , 120 ( 2009 ) [ arxiv:0804.3755 [ hep - ph ] ] .
w. furmanski and r. petronzio , phys .
* b97 * ( 1980 ) 437 .
g. curci , w. furmanski and r. petronzio , nucl .
* b175 * ( 1980 ) 27 .
a. vogt , s. moch and j. a. m. vermaseren , nucl .
* b691 * ( 2004 ) 129 [ arxiv : hep - ph/0404111 ] .
s. moch , j. a. m. vermaseren and a. vogt , nucl .
phys . * b688 * ( 2004 ) 101 [ arxiv : hep - ph/0403192 ] .
lhapdf , http://projects.hepforge.org/lhapdf/ m. l. mangano , eur .
j. * c59 * ( 2009 ) 373 [ arxiv:0809.1567 [ hep - ph ] ] . c. anastasiou , l. j. dixon , k. melnikov and f. petriello , phys .
* d69 * ( 2004 ) 094008 [ arxiv : hep - ph/0312266 ] .
m. cacciari , s. frixione , m. l. mangano , p. nason and g. ridolfi , jhep * 0404 * ( 2004 ) 068 [ arxiv : hep - ph/0303085 ] .
r. hamberg , w. l. van neerven and t. matsuura , nucl .
phys . *
b359 * ( 1991 ) 343 [ erratum-_ibid_. * b644 * ( 2002 ) 403 ] .
r. v. harlander and w. b. kilgore , phys .
* 88 * ( 2002 ) 201801 [ arxiv : hep - ph/0201206 ] . c. anastasiou and k. melnikov , nucl .
* b646 * ( 2002 ) 220 [ arxiv : hep - ph/0207004 ] .
v. ravindran , j. smith and w. l. van neerven , nucl .
phys . *
b665 * ( 2003 ) 325 [ arxiv : hep - ph/0302135 ] .
r. v. harlander and k. j. ozeren , arxiv:0909.3420 [ hep - ph ] .
m. l. mangano , m. moretti , f. piccinini , r. pittau and a. d. polosa , jhep * 0307 * ( 2003 ) 001 [ arxiv : hep - ph/0206293 ] ; http://cern.ch / mlm / alpgen/. t. gleisberg and s. hoche , jhep * 0812 * ( 2008 ) 039 [ arxiv:0808.3674 [ hep - ph ] ] .
e. boos _ et al .
_ [ comphep collaboration ] , nucl .
instrum .
* a534 * ( 2004 ) 250 [ arxiv : hep - ph/0403113 ] ; http://comphep.sinp.msu.ru/. f. a. berends and w. t. giele , nucl .
* b306 * ( 1988 ) 759 .
l. j. dixon , arxiv : hep - ph/9601359 .
s. catani and m. h. seymour , nucl .
phys . *
b485 * ( 1997 ) 291 [ erratum-_ibid .
_ * b510 * ( 1998 ) 503 ] [ arxiv : hep - ph/9605323 ] .
s. frixione , z. kunszt and a. signer , nucl . phys .
* b467 * ( 1996 ) 399 [ arxiv : hep - ph/9512328 ] .
d. a. kosower , phys .
* d57 * ( 1998 ) 5410 [ arxiv : hep - ph/9710213 ] ; j. m. campbell , m. a. cullen and e. w. n. glover , eur . phys .
j. * c9 * ( 1999 ) 245 [ arxiv : hep - ph/9809429 ] ; d. a. kosower , phys .
* d71 * ( 2005 ) 045016 [ arxiv : hep - ph/0311272 ] .
z. nagy , phys .
* d68 * ( 2003 ) 094002 [ arxiv : hep - ph/0307268 ] ; + http://cern.ch/nagyz/site/nlojet++.html . a. bredenstein , a. denner , s. dittmaier and s. pozzorini , arxiv:0905.0110 [ hep - ph ] . g. bevilacqua , m. czakon , c. g. papadopoulos , r. pittau and m. worek , arxiv:0907.4723 [ hep - ph ] .
g. bevilacqua , m. czakon , c. g. papadopoulos and m. worek , arxiv:1002.4009 [ hep - ph ] . c. f. berger _
_ , arxiv:0907.1984 [ hep - ph ] .
a. gehrmann - de ridder , t. gehrmann , e. w. n. glover and g. heinrich , jhep * 0712 * ( 2007 ) 094 [ arxiv:0711.4711 [ hep - ph ] ] .
s. weinzierl , phys .
* 101 * ( 2008 ) 162001 [ arxiv:0807.3241 [ hep - ph ] ] ; jhep * 0906 * ( 2009 ) 041 [ arxiv:0904.1077 [ hep - ph ] ] .
k. melnikov and f. petriello , phys .
* d74 * ( 2006 ) 114017 [ arxiv : hep - ph/0609070 ] .
s. catani , l. cieri , g. ferrera , d. de florian and m. grazzini , arxiv:0903.2120 [ hep - ph ] . c. anastasiou , k. melnikov and f. petriello , nucl .
* b724 * ( 2005 ) 197 [ arxiv : hep - ph/0501130 ] .
s. catani and m. grazzini , phys .
* 98 * ( 2007 ) 222002 [ arxiv : hep - ph/0703012 ] . t. sjostrand , s. mrenna and p. skands , comput . phys . commun .
* 178 * ( 2008 ) 852 [ arxiv:0710.3820 [ hep - ph ] ] ; http://home.thep.lu.se/~torbjorn/pythia.html .
l. lonnblad , comput .
commun .
* 71 * ( 1992 ) 15 .
t. gleisberg , s. hoche , f. krauss , m. schonherr , s. schumann , f. siegert and j. winter , jhep * 0902 * ( 2009 ) 007 [ arxiv:0811.4622 [ hep - ph ] ] ; + http://projects.hepforge.org / sherpa/. b. andersson , g. gustafson , g. ingelman and t. sjostrand , phys . rep .
* 97 * ( 1983 ) 31 .
t. sjostrand , nucl .
phys . *
b248 * ( 1984 ) 469 .
b. r. webber , nucl .
phys . * b238 * ( 1984 ) 492 .
a. buckley , h. hoeth , h. lacker , h. schulz and j. e. von seggern , eur .
j. * c65 * ( 2010 ) 331 [ arxiv:0907.2973 [ hep - ph ] ] .
t. sjostrand , monte carlo generators , arxiv : hep - ph/061124 , 2006 european school of high - energy physics , aronsborg , sweden , ed .
r. fleischer , cern-2007 - 005 ( cern , geneva , switzerland , 2007 ) , p. 51 .
s. catani , f. krauss , r. kuhn and b. r. webber , jhep * 0111 * ( 2001 )
063 [ arxiv : hep - ph/0109231 ] .
j. alwall _ et al .
_ , eur .
j. * c53 * ( 2008 ) 473 [ arxiv:0706.2569 [ hep - ph ] ] . n. lavesson and l. lonnblad , jhep * 0804 * ( 2008 ) 085 [ arxiv:0712.2966 [ hep - ph ] ] .
w. t. giele , d. a. kosower and p. z. skands , phys .
* d78 * ( 2008 ) 014026 [ arxiv:0707.3652 [ hep - ph ] ] . v.
m. abazov _ et al . _
[ d0 collaboration ] , phys .
* b678 * ( 2009 ) 45 [ arxiv:0903.1748 [ hep - ex ] ] .
s. frixione and b. r. webber , jhep * 0206 * ( 2002 ) 029 [ arxiv : hep - ph/0204244 ] .
p. nason , jhep * 0411 * ( 2004 ) 040 [ arxiv : hep - ph/0409146 ] .
p. torrielli and s. frixione , arxiv:1002.4293 [ hep - ph ] .
o. latunde - dada , jhep * 0905 * ( 2009 ) 112 [ arxiv:0903.4135 [ hep - ph ] ] .
s. frixione , f. stoeckli , p. torrielli and b. r. webber , arxiv:1010.0568 [ hep - ph ] .
s. d. ellis , private communication to the opal collaboration ; d. e. soper and h .- c .
yang , private communication to the opal collaboration ; l. a. del pozo , university of cambridge phd thesis , ralt002 , 1993 ; r. akers _ et al . _
[ opal collaboration ] , z. phys . * c63 * , 197 ( 1994 ) ; g. c. blazey _ et al .
_ , hep - ex/0005012 .
s. catani , y. l. dokshitzer , m. olsson , g. turnock and b. r. webber , phys .
* b269 * , 432 ( 1991 ) ; s. catani , y. l. dokshitzer , m. h. seymour and b. r. webber , nucl .
* b406 * , 187 ( 1993 ) . s. d. ellis and d. e. soper , phys .
rev . * d48 * , 3160 ( 1993 ) [ hep - ph/9305266 ] .
m. cacciari and g. p. salam , phys .
* b641 * ( 2006 ) 57 [ arxiv : hep - ph/0512210 ] .
m. cacciari , g. p. salam and g. soyez , http://fastjet.fr/. v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
rev . * d65 * , 052008 ( 2002 ) [ hep - ex/0108054 ] .
m. h. seymour , z. phys .
* c62 * ( 1994 ) 127 .
j. m. butterworth , b. e. cox and j. r. forshaw , phys . rev .
* d65 * ( 2002 ) 096014 [ arxiv : hep - ph/0201098 ] . j. m. butterworth _ et al .
_ , arxiv:1003.1643 [ hep - ph ] .
m. cacciari , g. p. salam and g. soyez , jhep * 0804 * ( 2008 ) 063 [ arxiv:0802.1189 [ hep - ph ] ] .
m. cacciari , j. rojo , g. p. salam and g. soyez , jhep * 0812 * ( 2008 ) 032 [ arxiv:0810.1304 [ hep - ph ] ] . m. cacciari , j. rojo , g. p. salam and g. soyez , http://quality.fastjet.fr .
j. m. butterworth , a. r. davison , m. rubin and g. p. salam , phys .
* 100 * ( 2008 ) 242001 [ arxiv:0802.2470 [ hep - ph ] ] .
d. krohn , j. thaler and l. t. wang , arxiv:0912.1342 [ hep - ph ] . c. buttar _
_ , arxiv:0803.0678 [ hep - ph ] .
d. krohn , j. thaler and l. t. wang , arxiv:0903.0392 [ hep - ph ] .
m. dasgupta , l. magnea and g. p. salam , jhep * 0802 * ( 2008 ) 055 [ arxiv:0712.3014 [ hep - ph ] ] .
m. rubin , arxiv:1002.4557 [ hep - ph ] .
2008 cteq - mcnet school , debrecen , hungary , august 2008 , http://conference.ippp.dur.ac.uk/conferenceotherviews.py?view=ippp&confid=156 . | the aim of these lectures is to provide students with an introduction to some of the core concepts and methods of qcd that are relevant in an lhc context . |
the study of superfluidity of either bosons or fermions has been a central topic in the field of ultracold atomic gases , starting from seminal experimental studies on the bose - einstein condensation of bosons @xcite and continuing with investigations of the bcs - bec crossover @xcite .
more recent experiments with two - component fermi gases have addressed the case of a finite population imbalanced , both in three spatial dimensions @xcite and in one dimension @xcite . among the goals of these experiments
is the search for the transition from a fully paired superfluid phase to the normal state and , in particular , for competing pairing states that survive a finite polarization .
these include the sarma state @xcite or the fulde - ferrell - larkin - ovchinnikov ( fflo ) state @xcite . with the advances optical lattice engineering @xcite , it has also become possible to investigate fermionic pairing states within the framework of the so - called hubbard model . for repulsive onsite interactions
the hope is to reach sufficiently low temperatures to search for pairing states away from half filling @xcite , whereas in the case of attractive interactions , there is a natural tendency to pair formation . besides experiments with homonuclear mixtures , where the pseudo - spin degree of freedom arises from preparing atoms in different hyper - fine states , there is also the possibility of working with heteronuclear mixtures such as , for instance , the @xmath0k-@xmath1li system @xcite . in that case , one deals both finite mass imbalance and population imbalance . on the other hand , bosonic gases in optical lattices
have provided the first example of a quantum phase transition from a mott - insulator to a superfluid state in the context of ultra cold atoms @xcite , which was subsequently observed in one dimensional setups @xcite as well . in analogy with the fermionic case , both homonuclear and heteronuclear bosonic mixtures
have been realized @xcite ; from a theoretical viewpoint , the two - species bose - hubbard model encompasses a remarkably rich physics , ranging from super - counterflow and antiferromagnetic phases in case of interspecies repulsion @xcite to pair superfluidity and density - wave instabilities in the attractive regime @xcite . from a theoretical point of view , it is of advantage to consider the one - dimensional case , for which both powerful analytical and numerical methods are available that can provide us with practically exact answers in many regimes . in ultra - cold atomic gases ,
it is also perfectly possible to realize one - dimensional systems experimentally , both for bosons @xcite and fermions @xcite , adding strong motivation to study this case .
moreover one - dimensional systems have been demonstrated to harbor unconventional superfluid phases which are the central topic of this work . in our work , we are interested in three types of superfluid states of two - component systems and their existence in the attractive , asymmetric hubbard model : ( i ) the more conventional fully paired phase , which can be considered as a fluid of cooper pairs , ( ii ) superfluids of larger composite objects such as trimers , and ( iii ) the fflo state in which a superfluid of pairs with finite center - of - mass momentum appears . the first two cases , i.e. , the equal density ( or dimer fluid ) and the trimer fluid , can be realized in both fermi - fermi mixtures @xcite and bose - bose mixtures in one - dimension @xcite while the fflo state exists only in the case of two - component fermi mixtures ( see @xcite and references therein )
. in our study we use numerically exact methods density matrix renormalization group ( dmrg)@xcite and stochastic series expansion quantum monte carlo @xcite to probe the stability of the above cited phases at a finite population and mass imbalance in the presence of a harmonic trapping potential .
our study complements previous works that mostly focussed on the homogeneous case @xcite .
moreover , we incorporate aspects that are typical of many experiments , namely the possibility of tuning the ratio of the effective masses through the depth of the optical lattice ; the fact that in a mass imbalanced system the two components usually experience a different trapping potential ; the difficulty in controlling the population of the two components to an arbitrary degree of precision , resulting typically in a non - zero global polarization .
finally , we also study the effect of a small but finite temperature on the shell structure and coherence properties of trapped mass - imbalanced two - component gases . here
we summarize our main results . in the case of fermions and for the bulk system
it is well - known that the fully paired phase is the ground state of a two - component fermi gas away from half filling @xcite . in the presence of a harmonic trap and at a finite population imbalance , the fully paired phase can only survive in the wings of the particle cloud at small polarizations @xcite , as recently observed in ref . .
here we show that adding mass imbalance to the system stabilizes the fully paired phase in the trap : if the light particles are the majority species , the fully paired phase occupies the center of the trapped system , in agreement with the grand - canonical phase diagram of the asymmetric hubbard model at a finite spin polarization @xcite .
we further show that a small temperature does not destroy the equal density phase .
the main conclusion is therefore that it is not necessary to enforce the condition of a perfectly balanced gas to observe and study properties of a dimer fluid .
this result applies to both fermions and hard - core bosons .
a particular feature of mass - imbalanced systems in one dimension is that they allow for the formation of bound states of more than two components , which in the two - species hubbard model with equal masses is forbidden .
this can happen for both fermions @xcite and bosons @xcite .
the existence of these states requires commensurate ratios of the densities of heavy particles ( @xmath2 ) versus light ones ( @xmath3 ) .
for instance , a trimer fluid can only be stable if @xmath4 @xcite , where @xmath5 is the density of the pseudo - spin @xmath6 component in a system of length @xmath7 with @xmath8 particles of species @xmath9 .
for the homogeneous system , this sets the global polarization to @xmath10 ( @xmath11 ) and it is in fact possible to stabilize this phase in a harmonic trap by enforcing this condition globally as was shown in ref . at sufficiently low densities .
here we investigate the fate of the trimer fluid upon deviating from the global polarization of @xmath12 , both for the case of fermions and bosons , finding that the trimer fluid rapidly gets pushed into the wings of the gas and subsequently disappears even for very small deviations from eq .
[ eq : comm ] .
therefore , one needs to fine - tune the global polarization to precisely @xmath12 to ensure that a large fraction of particles participates in this state . a difference in the trapping potential for the two components , however , has little effect on the trimer phase . for the case of bosons ,
we demonstrate that the formation of the trimer fluid leaves clear fingerprints in the momentum distribution function , a quantity that is easily accessible in experiments .
finally , concerning the fflo state , we show that it survives the addition of the harmonic trap and the mass imbalance .
similar to the case of a system with mass imbalance only @xcite , this phase occupies the center of the trap in a large parameter regime , corresponding in particular to the case in which heavy particles are the majority ones .
these numerically exact observations are consistent with the known phase diagrams for homogeneous system via a local - density approximation @xcite .
the plan of the paper is the following .
first , in sec .
[ sec : model ] we discuss typical experimental settings and conditions for two examples : ( i ) a heteronuclear fermi - fermi mixture such as @xmath0k and @xmath1li and ( ii ) a homonuclear bose - bose mixture , where the mass - imbalance arises due to a spin - dependent optical lattice .
this discussion serves to guide our numerical study . in section [ sec : fermions ] , we provide an extensive analysis of the density profiles of a mass - imbalanced fermi mixture and we present a state diagram for the various shell structures . in sec .
[ sec : bosons ] , we study both hardcore and softcore bosons and probe the stability of the dimer and trimer fluid . we conclude each of these section with a detailed summary of the results , while the main aspects are summarized again in sec .
[ sec : summary ] .
the purpose of this section is to describe the experimental set - ups to realize mass - imbalanced systems described by the asymmetric hubbard model . in this context
, we discuss heteronuclear mixtures , focussing on the example of the fermi - fermi system @xmath0k and @xmath1li .
the case of a homonuclear fermi - fermi mixture for which the mass - imbalance is induced by spin - dependent optical lattices has been described in @xcite .
then we turn to homonuclear bose mixtures in a spin - dependent optical lattice .
our numerical analysis will be based on the attractive , asymmetric 1d hubbard model : @xmath13 where @xmath14 creates a fermion with a pseudo - spin @xmath6 at site @xmath15 , @xmath16 , @xmath17 is the local density , @xmath18 is the hopping parameter with an explicit dependence on the pseudo - spin index , and @xmath19 is the attractive onsite interaction energy .
we set the lattice spacing to unity and impose open boundary conditions .
we add a harmonic confining potential parameterized by constants @xmath20 . without loss of generality we will consider @xmath21 .
a positive ( negative ) polarization @xmath22 corresponds to a majority of light ( heavy ) particles .
we also introduce the _ local _ magnetization through @xmath23 we further define the ratio of the trapping potentials as @xmath24 the proposed experimental setup is schematically shown in fig . [
fig : setup ] .
two pairs of counterpropagating laser beams with orthogonal linear polarization are orthogonally intersected .
this creates a 2d array of individual atom traps in the shape of one - dimensional tubes . in this example
we choose the wavelength to be @xmath25 nm , far red detuned from the atomic transitions .
considering a two - species fermionic mixture of @xmath26 and @xmath27 , we take into account the transition wavelengths for the d - lines ( @xmath28 nm for lithium and @xmath29 nm / @xmath30 nm for potassium ) . in order to realize one - dimensional optical lattices we superimpose a pair counterpropagating laser beams perpendicular trapping beams and blue detuned with respect to the wavelengths of the d lines . in the following subsections , we will discuss the resulting trapping potentials @xmath20 , the hopping parameters @xmath18 and the onsite interaction @xmath31 . an array of independent one - dimensional optical dipole traps is created by intersecting two perpendicular standing waves ( red beams ) .
a third orthogonal optical standing wave ( blue beams ) generates a repulsive lattice potential along the traps .
the density profiles of trapped atoms can be observed by an imaging setup from a transverse direction.,width=302 ] the ratio of the optical trapping potentials @xmath32 ( @xmath33li , k ) is a fixed number that amounts for the given wavelengths to @xmath34 note that @xmath35 and @xmath36 . due to the difference in the optical potentials and the different masses the respective trapping frequencies are generally different : @xmath37 where @xmath38 is the waist of the trapping beams and @xmath39 is the trap depth in the center of the crossed beam setup . in general
the wavelength for the trap can be chosen such that the difference in the optical potentials for the two atomic species compensates the mass difference such that equal trapping frequencies are obtained @xcite .
this so called _ magic _ wavelength , @xmath40 , occurs for the mixture of @xmath26- @xmath27 at @xmath41 nm .
the normalized lattice depth : at the magic wavelength ( @xmath42 nm ) the lattice depths are equal for @xmath26 and @xmath27 ( solid line ) .
two nanometer above ( dotted ) and below ( dashed ) the lattice depths differ , while the ratio is inverted at the magic wavelength . using a beam waist of @xmath43
less than one watt of laser power is sufficient to achieve relevant lattice depths.,width=321 ] the one - dimensional optical lattice is realized by superimposing a pair of counterpropagating laser beams perpendicular to the beams that form the trapping array . for a given intensity of the lasers creating the optical lattice , potassium and lithium will experience optical lattice potentials of a different depths as illustrated in fig .
[ fig : vo ] .
the experimentally controllable parameters are the amplitudes of the standing wave forming the optical lattice , _ i.e. _ , the lattice depths @xmath44 .
these , and the masses determine the respective frequencies : @xmath45 it is useful to introduce the atomic recoil energies as @xmath46 where @xmath47 is the wave vector of the standing wave , @xmath48 is the laser wavelength , and @xmath49 is the lattice spacing .
this allows us to express the frequencies as : @xmath50 for wavelengths of the lattice laser blue detuned to the atomic transitions there exists a single _ magic _ wavelength , @xmath51 , at which the lattice depths @xmath44 normalized to the respective recoil energies @xmath52 are equal : @xmath53 for the @xmath26-@xmath27 mixture this occurs at @xmath54 nm . in table
[ tab : magicwave ] we summarize the magic wavelengths and the ratios of the optical trapping potentials at @xmath25 nm for different combinations of alkaline atomic species . as is obvious from the table , in all cases , the trapping _
potentials _ differ .
we assume a separable three - dimensional lattice potential of the form @xmath55 . \nonumber\end{aligned}\ ] ] this allows us to calculate the hopping matrix elements from the one - dimensional mathieu equation .
the result is shown in fig .
[ fig : tlik ] ( solid lines ) . in the limit of a deep lattice ,
one can obtain an analytical expression @xcite : @xmath56 with @xmath57 and recoil energies @xmath52 .
this result is also included in fig .
[ fig : tlik ] ( dashed lines ) , and it agrees with the exact solution for large @xmath58 or respectively laser power @xmath59 . the ratio of the tunneling energies of @xmath26 and @xmath27 is shown in fig .
[ fig : tlitk ] .
while the bare mass ratio is @xmath60 , we deduce that the ratio of the actual hopping matrix elements further depends on the laser power @xmath61 through the lattice depth @xmath44 .
this is illustrated in fig .
[ fig : tlitk ] thus in general , the ratio @xmath62 can be tuned and is not solely given by the bare mass ratio . in our numerical analysis of a two - component fermi mixture , we will work at @xmath63 , corresponding to @xmath64 .
this requires to detune the wavelength slightly above the magic wavelength @xmath65 ( compare fig .
[ fig : tlitk ] . )
the tunneling energies at the magic wavelength are the same for @xmath26 and @xmath27 ( solid line ) .
the asymptotic expression ( dashed line ) only describes the exact solution of the mathieu equation at sufficiently high lattice depths ( see eq . ) .
the maximum at low lattice depths is not physical.,width=321 ] ratio of the tunneling energies for @xmath26 and @xmath27 for different lattice wavelengths . at the magic wavelength
the ratio is constant at the inverse mass ratio ( solid line ) .
two nanometers above and below the magic wave - length the ratio of the tunneling energies differs from the bare mass ratio ( dashed lines ) . by increasing the lattice power the ratio of the tunneling energies can be tuned over a wide range that can be set by the choice of the wavelength .
the dmrg and sses simulations presented in this work are done for a ratio of the tunneling energies @xmath66 .
, width=321 ] the onsite repulsion @xmath31 between two atoms of different species is given by @xcite : @xmath67 in the expression , @xmath68 are the wannier functions , @xmath69 is the reduced mass , and @xmath70 is the scattering length that could be different for two atomic species , hyperfine flavors , or bose - fermi mixtures . for the evaluation of @xmath31 we assume a deep lattice @xmath71 and that the particles occupy the lowest band .
we can then approximate the wannier functions by the ground state functions of a 1d harmonic oscillator : @xmath72 with @xmath73 notice that for _ soft - core bosons _ , one needs to consider excited states .
a simple and straightforward gaussian integration yields : @xmath74 this expression can be simplified if we work at the magic wavelength for the lattice , since for this case we have : @xmath75 expressing the energies in units of one of the recoil energies , we obtain @xmath76 in the extreme 1d limit , this entire expression can be simplified even further , since the wannier functions become dirac deltas in the transverse direction : @xmath77 these equations indicate the recipe for controlling the interactions , which can be done by either changing the lattice depth with lasers , or by tuning the scattering length using feshbach resonances . notice that if the lattice is too deep , multi - band processes can occur , but the validity of the one - band approximation is generally satisfied in practice .
.[tab : magicwave ] magic wavelengths resulting in identical trapping frequencies or lattice potentials for different combinations of alkaline atoms .
the calculation takes both d lines for each species into account .
[ cols= " < , > , > , > " , ] in the case of bosons the hubbard hamiltonian contains as well intraspecies interactions : @xmath78 \nonumber\\ & + & u \sum\limits^l_{i=1 } n_{i\uparrow } n_{i\downarrow } + \sum_{i , \sigma } v_{\sigma}(i - l/2)^2n_i .
\label{eq : one}\end{aligned}\ ] ] here @xmath79 creates a boson of type @xmath6 on site @xmath80 . in this case , mass imbalance can be realized either in heteronuclear mixtures ( such as @xmath81rb/@xmath82k @xcite ) or in hyperfine mixtures in spin - dependent optical lattices ( such as in the case of @xmath81rb as realized in a number of recent experiments @xcite .
we will focus the following short discussion on the latter example , in which the hubbard parameters can be tuned by the lattice depth , by the optical lattice wavelength , and by an interspecies feshbach resonance .
we consider @xmath81rb in the hyperfine mixture of @xmath83 and @xmath84 hyperfine states , which are collisionally stable and for which several feshbach resonances have been identified @xcite .
state - dependent optical lattices can be easily realized when the optical lattice wavelength is close to the d1 and d2 lines 795 and 780 nm respectively @xcite .
we consider a highly anisotropic optical lattice , with wavelength @xmath85 nm and depth @xmath86 for the transverse components defining 1d tubes , and wavelength @xmath87 nm and variable depth @xmath88 for the longitudinal component along the tubes . here
@xmath89 is the recoil energy , where @xmath90 for the transverse components and @xmath91 for the longitudinal one . as in the previous section ,
we use the solution of the mathieu equation to determine the hoppings @xmath92 and @xmath93 , and we calculate the intraspecies couplings @xmath94 and @xmath95 in the gaussian approximation @xcite .
e.g _ that the hopping ratio @xmath96 spans the interval @xmath97 when @xmath88 goes from @xmath98 to @xmath99 ; in the same parameter range , the ratios @xmath100 and @xmath101 are above 10 , suppressing double occupancy of the same species : in this case an appropriate description of the system is provided by a model of hardcore bosons , on which we will focus our attention later in the paper .
the interspecies coupling @xmath31 , which would naturally be very strong in this parameter interval ( @xmath102 ) , can be suppressed by exploiting the above cited feshbach resonances .
in this section we study the density profiles and correlations of a mass- and population imbalanced two - component fermi gas with attractive interactions in a harmonic trap .
this extends previous studies of population imbalanced mixtures on the one hand and mass - imbalanced systems on the other hand .
we shall briefly review the known results .
the phase diagram of a spin - imbalanced two - component fermi gas with equal masses and attractive interactions can be obtained from the bethe ansatz , both in the continuum @xcite and in lattice models @xcite . only recently
, it was rigorously shown that the partially polarized phase is the one - dimensional analogue of the fflo state by means of exact numerical methods @xcite , confirming predictions from mean - field theory @xcite and bosonization @xcite . in 1d
, this means that the pair - pair correlation functions are modulated with @xmath103 and decay as a power law , slower than any competing correlation in the two - particle channel ( @xmath104 is the fermi momentum ) .
the computation of correlations using the bethe ansatz is generally very difficult , and therefore , the exponents of correlations were first calculated by numerically solving the corresponding the bethe ansatz equations @xcite .
very recently , first analytical results for the fflo correlations from bethe ansatz have been presented in @xcite . in more general models that incorporate the coupling of fermions to a molecular channel @xcite ,
the partially polarized phase is of the fflo type on the bcs side only @xcite .
the fate of the fflo state in coupled one - dimensional systems has been recently addressed @xcite . for spin - imbalanced systems in one dimension ,
it is by now well - established that there is a two - shell structure in a harmonic trap @xcite .
the central shell is always occupied by the partially polarized phase , while in the wings there is a fully paired phase at low polarizations and a fully polarized phase at large polarizations .
thus , there is a critical polarization @xmath105 at which the shell structure changes ( exactly at @xmath105 , the whole system is in a partially polarized phase ) .
this was first predicted by applying the local density approximation to the exactly known phase diagram @xcite and then confirmed in exact numerical simulations @xcite .
the fflo - type correlations in the partially polarized phase are stable against the presence of a harmonic trap @xcite .
the effect of temperature on the shell structure was studied using both bethe ansatz methods combined with lda @xcite and quantum monte carlo simulations @xcite . in a recent experiment with a 3d array of one - dimensional tubes @xcite
, the theoretical predictions for the shell structure were quantitatively verified @xcite . for more details , see the review @xcite .
a mass - imbalanced system of attractively interacting fermions is no longer integrable .
the phase diagram for the population and mass - imbalanced case was therefore obtained by using field theory and dmrg calculations @xcite . in ref .
@xcite , the main focus was on the partially polarized phase which shrinks as the mass - imbalance increases due to the instability of a strongly mass imbalanced systems against a collapse for @xmath106 and phase separation for @xmath107 @xcite .
the partially polarized phase is , similar to the case of a spin - imbalanced mixture , of the 1d fflo type @xcite .
a complete phase diagram for the case of a majority of heavy particles was presented in ref .
@xcite , where in particular , the existence of trimer - fluid phases was demonstrated .
these trimer phases can exist if the densities of the two components obey a certain ratio , namely @xmath108 this is a _ necessary _ condition only ; in ref .
@xcite it was shown that the trimer gap goes to zero above a critical density @xmath109 . moreover , in the trimer - fluid phase , the trimer correlations decay algebraically @xcite , while the ( s - wave ) pairing correlations decay exponentially @xcite .
note that there are in fact many more stable multimer bound states at other rational ratios of @xmath3 and @xmath2 @xcite , which we will not study in this work . as a consequence of the presence of these trimer phases ( and phases of liquids of even larger compounds ) , the partially polarized phase is separated from the vacuum by an extended line in the @xmath110-@xmath111 phase diagram @xcite , in contrast to the population imbalanced system , where this is a critical point @xcite .
another main difference is the broken particle - hole symmetry of the the mass- and spin - imbalanced mixture with respect to the spin - imbalanced case only .
this has immediate consequences for the shell structure in a harmonic trap . to illustrate this ,
let us resort to the local density approximation ( lda ) , valid if the trapping potential is sufficiently smooth , as it is the case with parabolic potentials whose harmonic oscillator length is much larger than the lattice spacing or the interparticle distance . in the case of a single species and within lda
, the local properties of the trapped system can be quantitatively related to those of a bulk system whose chemical potential equals the local chemical potential in the trap , namely @xmath112 . in the case of a two - component mixture
the chemical potential is generally species dependent , and one can parametrize the two chemical potentials as @xmath113 and @xmath114 .
the lda assigns to local regions of the trap the behavior of a bulk system with local average chemical potential @xmath115 , but with fixed chemical potential difference ( or magnetic field ) @xmath116 . in general , to induce a finite polarization into the system the magnetic field @xmath111 has to overcome ( in modulus ) either one of the following gaps : @xmath117 where @xmath118 is the ground state energy of a system with a given @xmath119 and @xmath120 in the bulk case .
the gaps @xmath121 are related to the spin ( pairing ) gap @xmath122 as @xmath123 . in other words , to obtain @xmath124 , one needs either @xmath125 ( @xmath107 ) or @xmath126 ( @xmath106 ) . in the case of equal masses
the polarization gaps are both equal to @xmath127 , and hence they are known to be finite through the bethe ansatz solution of the model in the case of attractive interactions @xmath128 @xcite . in particular , as depicted in fig .
[ f.spingap ] , the bethe ansatz solution predicts that the @xmath122 is a _ decreasing _ function of @xmath110 below half filling .
therefore , in this situation , applying the lda to a trapped system a larger spin gap is found in the trap wings where the density is lower : @xmath111 being constant across the trap , this means that a polarization @xmath22 induced by @xmath111 will appear first at the trap center only , moving gradually to the trap wings as @xmath111 ( or @xmath22 ) increases .
is the total particle density . adapted from ref .
@xcite , width=340 ] the situation changes radically in the case of mass imbalance .
for a two - component fermi gas the gaps @xmath121 were calculated via dmrg in ref .
@xcite ( these results carry over to the case of a two - component gas of hard - core bosons to be discussed later ) .
as could be anticipated , the two polarization gaps become unequal , @xmath129 , and , most importantly , they are no longer particle - hole symmetric functions of the chemical potential ( but the spin gap @xmath122 is ) .
in particular @xmath130 becomes an _ increasing _ function of the chemical potential / the density , which is schematically depicted in fig . [ f.spingap ] .
this implies that , in the mass imbalanced case , an increasing ( positive ) magnetic field @xmath111 will first polarize the trap wings , while the trap center remains unpolarized and exhibits full pairing .
based on the above considerations and the phase diagram of ref .
@xcite valid for @xmath131 , one can expect a variety of shell structures in a harmonic trap using the local density approximation . for @xmath106
, there will be three cases : ( i ) a partially polarized core plus fully paired wings for @xmath132 , ( ii ) an extended region with only a partially polarized phase at @xmath133 , and ( iii ) a partially polarized core and fully polarized wings at @xmath134 . in ref .
@xcite , it was demonstrated that at @xmath12 there is only a partially polarized phase where the commensurability condition for the existence of trimers is fulfilled locally at every point in the trap , confirming the predictions of ref .
@xcite . in the case of @xmath107
, we expect only one critical polarization @xmath135 , separating a three - shell structure from a two shell structure . at small @xmath22 ,
the equal - density phase occupies the core of the system , followed by an intermediate shell that is partially polarized , and a fully polarized region in the wings . at large @xmath136 ,
the core is partially polarized , surrounded by a fully polarized shell .
one goal of our work is to confirm these predictions by using numerically exact dmrg simulations @xcite .
we first compute the density and spin - density profiles at zero temperature using dmrg to obtain the state diagram of such a system and show that this is consistent with the qualitative expectations drawn from the grand - canonical phase diagram @xcite .
we devote particular attention to the stability of phases with commensurate densities , i.e. , the fully paired phase and the trimer phase against the presence of the trap .
in addition , we calculate the pairing correlations in the presence of the trap and show that they are of the fflo type in the partially polarized phase .
next , we investigate the effect of varying @xmath31 on the critical polarization separating different shell structures at @xmath107
. then , we study the behavior in _ spin-_dependent traps , motivated by the discussion from sec .
[ sec : fermi_exp ] .
finally , we use quantum monte carlo simulations to analyze the effect of a finite temperature on the shell structure , focussing on the stability of the fully paired phase at @xmath107 .
let us first make the connection between the symbols used here and the experimental set - up more transparent .
we envision a @xmath1li-@xmath0k mixture in an optical lattice .
we associate @xmath137 we will work with @xmath138 , which implies to go to a wave - length larger than @xmath139 ( compare fig . [ fig : tlitk ] ) .
we avoid the regime of very strong mass imbalance @xmath140 since in that regime , the system is unstable against phase separation ( @xmath107 ) or a collapse ( @xmath106 ) @xcite . since the bare mass ratio is @xmath141 , it is therefore advantageous to tune the system to an intermediate ratio of effective masses .
for most of the simulations we will work with @xmath142 ( implying @xmath143 ) .
furthermore we will consider the ratio of the trapping frequencies , parameterized by @xmath144 as a free parameter , and we will first discuss @xmath145 .
the discussion from sec .
[ sec : fermi_exp ] suggests that for the @xmath1li-@xmath0k mixture , @xmath146 and we will explore this in our dmrg analysis as well .
: ( a ) at low @xmath22 , there are two shells : ed in the core and pp and fp in the wings .
( b ) at large @xmath22 , pp sits in the core and fp sits in the wings . parameters : @xmath147 , @xmath142 , @xmath148 , @xmath149 and ( a ) @xmath150 , ( b ) @xmath151 .
( solid lines : @xmath152 , dashed lines : local population difference @xmath153).,width=321 ] : ( a ) at low @xmath22 , there are two shells : pp in the core and fp in the wings .
( b ) at large @xmath22 , pp sits in the core and fp sits in the wings .
parameters : @xmath147 , @xmath142 , @xmath148 , @xmath154 and ( a ) @xmath155 , ( b ) @xmath156 .
( solid lines : @xmath152 , dashed lines : local population difference @xmath153).,width=321 ] we investigate the density profiles at @xmath157 using dmrg .
our dmrg simulations are done on a chain with @xmath158 sites , and using between @xmath159 and @xmath160 states .
we consider variable trapping potentials , confining the atomic cloud well within the simulation box . in order to label the emergent shell structures ,
we introduce a set of acronyms : pp ( partially polarized ) , ed ( equal density ) , fp ( fully polarized ) , tp ( trimer phase ) .
then , we use strings of these acronyms that describe the shell structure , going from the core to the wings . as an example , the label pp - ed stands for a system with a pp phase in the core and an ed phase in the wings . in this section , we consider the case of @xmath145 , i.e. , equal trapping potentials for both components .
typical density and spin density profiles obtained from dmrg simulations are shown in figs .
[ fig : p_neg ] and [ fig : p_pos ] for @xmath107 and @xmath106 , respectively .
these results confirm our qualitative expectations discussed in sec .
[ sec : fermi_over ] , namely that at @xmath107 , there are three shells for @xmath162 ( ed - pp - fp ) and two at large polarizations @xmath136 ( pp - fp , compare fig . [
fig : p_neg ] ) . at @xmath106 , there are two critical polarizations separating the different shell structures from each other : ( i ) pp - ed for @xmath163 , ( ii ) pp for @xmath164 , and ( iii ) pp - fp for @xmath165 ( compare fig .
[ fig : p_pos ] ) . , density profiles that match the trimer condition @xmath166 ( solid lines : @xmath152 , dashed lines : local population difference @xmath167 ) .
( a ) @xmath12 : the trimer condition is fulfilled in the entire trap ( first shown in ref .
( b ) @xmath168 , ( c ) @xmath169 : pp in the core , the trimer condition is fulfilled in a thin outer shell .
parameters : @xmath147 , @xmath142 , @xmath170 , @xmath171 . ,
title="fig:",width=321 ] , density profiles that match the trimer condition @xmath166 ( solid lines : @xmath152 , dashed lines : local population difference @xmath167 ) .
( a ) @xmath12 : the trimer condition is fulfilled in the entire trap ( first shown in ref .
( b ) @xmath168 , ( c ) @xmath169 : pp in the core , the trimer condition is fulfilled in a thin outer shell .
parameters : @xmath147 , @xmath142 , @xmath170 , @xmath171 .
, title="fig:",width=321 ] the particular case of @xmath172 is addressed in fig .
[ fig : p_trim ] .
right at @xmath12 and at sufficiently low density , the trimer condition @xmath173 is fulfilled in the entire trapped cloud ( see fig . [
fig : p_trim](a ) where we reproduce the results from ref .
@xcite ) . in this parameter regime , the results of refs .
@xcite and @xcite suggest that the ground state is a trimer fluid . upon deviating only slightly from @xmath12 ( which ensures the trimer condition _ globally _ ) ,
the trimer condition is no longer fulfilled _ locally _ , i.e. @xmath174 .
this demonstrates that in order to stabilize the trimer fluid in a harmonic trap one needs to fine - tune the global polarization to @xmath12 .
while our main interest is in the low density regime , we here also include examples of typical density profiles at large densities , i.e. , where one of the two components has density @xmath175 .
these are presented in fig .
[ fig : bis ] .
the first two examples are for @xmath106 and have @xmath176 in the core of the system , followed by a thin partially polarized shell . at small @xmath106 ,
the outer wings are fully paired [ see fig . [
fig : bis](a ) ] while at large @xmath106 , they are partially polarized [ see fig .
[ fig : bis](b ) ] . in the regime of @xmath107 ,
first the light atoms form a band insulator @xmath177 . in the particular example shown in fig .
[ fig : bis](c ) , the core is surrounded by a thin pp phase and a broad fully polarized wing .
qualitatively , one - species band insulators with @xmath178 form faster at @xmath106 than at @xmath107 , i.e. , by increasing the total particle number or by making the trap tighter .
not surprisingly it is energetically favorable to displace the light atoms ( @xmath179 ) into the outer regions . :
emergence of single species band insulators at large densities .
typical density profiles at ( a ) @xmath180 , ( b ) @xmath181 , ( c ) @xmath182 ( @xmath183 , @xmath142 , @xmath147).,width=321 ] our results for the shell structure of a mass- and spin - imbalanced system at @xmath184 and @xmath142 are summarized in the state diagram fig .
[ fig : state ] .
we focus on those regions of polarization @xmath22 and effective density @xmath185 in which no component has formed a band insulator yet ( i.e. , we restrict ourselves to @xmath186 ) .
evidently , in most of the state diagram the partially polarized phase sits in the core with the exception of the region @xmath187 , where the equal density phase occupies the central region of the trap .
it is interesting to emphasize that this does not happen for a purely spin - imbalanced system : there , any arbitrarily small polarization pushes the equal density phase to the outer wings @xcite . in that sense , the mass - imbalance stabilizes the fully paired / equal density phase . in particular ,
if one is interested in this phase , it is therefore not necessary to fine - tune the _ global _ polarization to @xmath188 .
this is a consequence of the broken particle - hole symmetry in a mass - imbalanced system as discussed in sec .
[ sec : fermi_over ] .
( @xmath142 , @xmath147 ) .
, width=321 ] the trimer phase , realized at global polarization @xmath12 , extends up to @xmath189 ( triangles up in fig .
[ fig : state ] ) . to determine this point
we have followed the density in the center of the trap and compared it to the critical density of the bulk system beyond which the trimer gap closes @xcite , which is at @xmath190 for @xmath142 . in ref .
@xcite , it was further shown that at sufficiently low density and at @xmath12 , the trimer correlations in the trap follow a power law .
one possibility to observe the formation of trimers in an experiment would be to open the trap and to let the gas expand in 1d in the optical lattice ( such an experiment was performed in 2d and 3d optical lattices with balanced two - component fermi gases @xcite ) . due to the formation of heavy objects , i.e. , the trimers , that are protected by the associated excitation gap @xcite
, the expansion should be substantially slower at @xmath12 compared to other polarizations .
a similar behavior was seen in numerical simulations for the expansion of two - component fermi gases with a high density of doublons @xcite .
our suggestion therefore is to monitor the expansion velocity as a function of both polarization ( at low densities @xmath191 ) and of effective density at @xmath12 . regarding the critical polarizations @xmath192 ( @xmath193 ) , we observe a weak dependence on effective density .
most notably , going to small @xmath194 further stabilizes the ed - pp - fp regime .
this is consistent with the usual argument that in one dimension , pairing at equal densities is more robust at low densities @xcite .
it is further instructive to discuss the spatial extent of the atomic cloud , and of the clouds of each individual component .
this follows the analysis of ref .
@xcite for the one - dimensional spin - imbalanced fermi gas , where it was suggested that the critical polarization can be read off from the polarization dependence of the radii.in fact , a similar analysis was then used in the experimental work on density profiles of one - dimensional spin - imbalanced fermi gases @xcite .
we now take a cut at a fixed @xmath185 through the state diagram and we estimate the extent @xmath195 ( @xmath196 ) as the region with a finite particle density , where @xmath197 is the cloud radius and @xmath198 are the radii for @xmath6 .
typical results are shown in fig .
[ fig : radii ] . several aspects deserve being mentioned .
first , the cloud is the smallest in the region @xmath199 , where only a partially polarized phase is present .
secondly , for @xmath200 ( where @xmath201 is the critical polarization separating pp from pp - fp ) , one observes @xmath202 , while for @xmath203 , one has @xmath204 . at @xmath205 and @xmath135 , however , there are no clear features in the dependence of the radii on polarization ( we have to keep in mind , though , that the particle numbers are fairly small ) . while we have mostly focussed on @xmath142 , it is also interesting to study the effect of @xmath31 on the stability of the ed phase .
the dependence of @xmath135 on @xmath31 is shown in fig . [
fig : pcrit_u ] . increasing @xmath206
makes the regime in which the ed phase occupies the core of the system larger , as expected . , triangles ) , the light ( @xmath207 , squares ) , and the heavy fermions ( @xmath208 , circles ) for @xmath209 and @xmath210 ( @xmath142 , @xmath147 ) .
the horizontal , dashed lines mark the critical polarizations @xmath201 , @xmath205 , and @xmath135 ( from left to right).,width=321 ] separating ed - fp - pp from pp - fp at @xmath107 as a function of @xmath31 for @xmath161 ( @xmath147 , @xmath210 , @xmath211).,width=321 ] we now show that in the pp phase the fflo correlations are robust to the presence of the trap .
we focus on the region @xmath106 , since at @xmath107 , the pp phase sits in the core only for @xmath136 . in that regime , however , the spin density is typically strongly varying across the trap , which disfavors clear signatures of the fflo state . in a harmonic trap ,
the signatures of the fflo state are the cleanest whenever the polarization varies slowly with the local chemical potential @xcite . in order to demonstrate the presence of fflo correlations we compute the fourier transform of the pair - pair correlations @xmath212\ , \rho_{lm}\ , .
\label{nkpair}\ ] ] where @xmath213 here , and throughout the paper , we adopt the same discretization of the @xmath214 vectors as in a homogeneous system , namely @xmath215 expresses the number of particles with momentum @xmath214 in the interval @xmath216 for a system of size @xmath7 .
our dmrg results are displayed in fig .
[ fig : fflo ] .
clearly , in the curves with @xmath217 , we observe a maximum in @xmath218 at some incommensurate momentum @xmath219 , indicative of oscillating pair correlations . in a homogeneous system in one dimension and in the fflo state ,
the modulation q is given by @xmath220 i.e. , @xmath221 . in a harmonic trap ,
first of all , not all majority fermions participate in the fflo state since all particles in the fp regions have to be excluded .
second , the length entering in eq . is the one that is actually occupied by the quasi - condensate .
therefore , @xmath219 is _ not _ simply proportional to the global polarization . if one accounts for that , following the procedure described in ref .
@xcite then one obtains @xmath222 where @xmath223 and @xmath224 .
@xmath225 is the region occupied by the quasi - condensate , and all @xmath226 are obtained by integrating @xmath227 over that region .
equation should properly describe the scaling of the position @xmath219 of the maximum in @xmath218 if the spin - density is approximately constant in the pp shell .
we plot @xmath219 vs both the total polarization @xmath22 ( circles ) and @xmath228 ( squares ) in the inset of fig .
[ fig : fflo ] .
@xmath219 deviates from eq . by up to 10%
this can partially be explained by taking into account the uncertainties of determining @xmath219 and @xmath229 from the finite - size data , and by finite - size effects . for @xmath230 , @xmath142 ( @xmath147 . ) .
we display data for @xmath231 , with the peak in @xmath218 shifting from left to right , respectively ( we do not see a clear signature for @xmath232 ) .
inset : position @xmath219 of the maximum in @xmath218 vs @xmath22 ( squares ) and vs @xmath228 ( circles , see text in sec .
[ sec : fflo ] ) .
the dashed line is @xmath233 , i.e. , eq . .
, width=321 ] to complete the discussion , one would need to show that the fflo correlations decay algebraically along the trap ( see the discussion in ref .
@xcite for the spin - imbalanced case ) . for a homogeneous system with mass- and spin - imbalance ,
the momentum distribution functions were discussed in refs .
@xcite , while in refs .
@xcite , it was shown that away from @xmath12 , the ( s - wave ) paring correlations decay algebraically , modulated with @xmath234 .
we now turn to the case of unequal trapping potentials , @xmath236 .
@xmath237 implies that the light atoms are squeezed into the core of the trapped gas whereas for @xmath238 , they are being pushed out .
_ zero global polarization
_ in the case of @xmath239 , a partially polarized phases can be induced if the trapping potentials are sufficiently different form each other ( results not shown here ) . however , drastically different trapping potentials are necessary even at small @xmath240 and the partially polarized phase has a strongly varying spin density .
this disfavors this set - up as a way of realizing the fflo state , which leaves clearer fingerprints whenever the spin density varies slowly with the chemical potential @xcite . _
@xmath107 , stability of the ed phase
_ next we address the question whether unequal trapping potentials drastically alter the state diagram or not . as an example , we consider @xmath241 , shown in figs .
[ fig : v_unequal](a)-(c ) for @xmath242 and @xmath243 , respectively . at @xmath145 ,
the ed phase occupies the core of the system .
only a very large @xmath244 achieves enough compression of the light particles as to wash out completely the ed phase . at @xmath245 on the other hand , the central ed shell grows bigger than for @xmath246 , as more excess @xmath247-particles migrate towards the wings . to render this observation more quantitative ,
we plot the critical polarization @xmath135 as a function of @xmath244 in the inset of fig .
[ fig : v_unequal](b ) . @xmath135 monotonously decreases with increasing @xmath244 , with a very weak dependence on @xmath244 beyond @xmath248 .
it is important to stress that for the experimental parameters sketched in sec .
[ sec : fermi_exp ] , the equal density shell is actually stabilized since @xmath249 .
the trimer phase , realized at global polarization @xmath12 in the entire trap at sufficiently small densities , is remarkably stable against varying @xmath244 .
this is demonstrated in fig .
[ fig : trimer_eta ] , where , starting from the parameters of fig .
[ fig : p_trim](a ) , we vary @xmath244 from @xmath250 to @xmath251 . by compressing the trap for the light fermions ( @xmath237 ) the cloud size shrinks drastically , thus increasing the density in the center of the trap .
eventually , the trimer condition eq . is no longer fulfilled along the entire trap : the heavy fermions prefer to stay in the wings [ see the example of @xmath252 in fig . [
fig : trimer_eta](a ) ] .
one also drives the core of the system out of the regime in which the trimer gap is finite ( compare refs . @xcite ) . in the opposite regime , @xmath238 ,
the cloud expands , yet down to @xmath253 , we still observe a perfect match of @xmath254 , equivalent to @xmath255 . ,
@xmath256 : typical density profiles at ( a ) @xmath145 , ( b ) @xmath257 , ( c ) @xmath243 ( @xmath183 , @xmath142 , @xmath147 ) .
inset in ( b ) : critical polarization @xmath105 separating ed - pp - fp from pp - fp as a function of @xmath244 for @xmath142 , @xmath147 , @xmath210.,width=321 ] , @xmath258 : typical density profiles at @xmath259 ( @xmath260 , @xmath142 , @xmath147 . ) at very large @xmath237
, the trimer conditions is not fulfilled in the center of the trap , i.e. , @xmath261 .
, width=321 ] ( color online ) ( a ) finite - temperature state diagram of a mass imbalanced mixture with @xmath262 , @xmath263 , and @xmath264 .
the polarization scan is obtained by fixing @xmath265 and varying @xmath120 .
( b ) trajectory in the @xmath266 plane ( compare fig . [
fig : state ] ) described with the above parameters .
, width=340 ] as seen in the previous sections , the formation of an ed dimer liquid in the trap center is very robust to the presence of a finite , positive polarization @xmath22 for a large window of characteristic densities @xmath194 .
here we probe the robustness of this phenomenon to the further effect of finite temperatures , by making use of quantum monte carlo ( qmc ) simulations based on the stochastic series expansion ( sse ) algorithm @xcite . in our present study
we use a canonical formulation based on double directed - loop updates @xcite .
[ fig : finite - t ] refers to a path through the state diagram of fig .
[ fig : state ] .
this corresponds to an the experimentally meaningful situation , in which @xmath267 is kept fixed , while @xmath120 is varied , scanning the @xmath22 axis at variable @xmath194 . the actual trajectory in the @xmath266 plane is indicated in the figure .
here @xmath262 and @xmath268 as in the previous sections , while @xmath269 . as a criterion for dimer liquid formation in the trap center
we require that the local magnetization vanishes over the 10 central sites .
this criterion is met at @xmath270 ( actually @xmath271 , ensuring the elimination of thermal effects ) over a sizable polarization range @xmath272 , as already seen in fig .
[ fig : state ] . as @xmath273
is increased above zero , the polarization range featuring an unpolarized dimer liquid ( i.e. , the ed phase ) in the trap core shrinks , but it remains sizable up to temperatures @xmath274 , proving the robustness of ed pairing to realistic conditions .
indeed , even though pairing correlations in one - dimensional systems become short - ranged as soon as @xmath273 becomes finite , the finite gaps @xmath121 that were defined in eq . prevent the thermally excited majority ( @xmath247 ) particles from flowing from the trap wings into the center .
the parameter region featuring an unpolarized core in fig .
[ fig : finite - t ] shrinks asymmetrically from the side of positive polarizations : indeed at higher @xmath22 the excess @xmath247 particles sitting in the trap wings have a higher potential energy , and therefore they need a smaller thermal energy to overcome the gaps @xmath121 and to flow into the trap center .
so far we have studied the properties of three pairing states of a spin- and population imbalanced fermi gas in a harmonic trap , namely the equal density ( or dimer fluid ) phase , the trimer fluid phase and the fflo state .
our main result for the ed phase is that it occupies the core of a trapped system over a wide range of positive polarizations .
unequal trapping potentials which are typical for a heteronuclear system destabilize the ed phase if the confinement for the light particles is much stronger than for the heavy ones , whereas in the opposite regime , the ed phase is favored .
this phase is stable against thermal fluctuations as long as they do not overcome the polarization gaps @xmath121 .
while fermionic mixtures have been shown to support three different types of superfluid phases , fflo , ed and trimer phase , which are stable even in the presence of a trapping potential , the question naturally arises whether such pairing instabilities are indeed possible when considering bosonic binary mixtures .
the main difference between fermionic and bosonic statistics resides in the fact that bosons can ( quasi-)condense , resulting in a sharp peak at wavevector @xmath275 in the momentum distribution @xmath276 where @xmath277 is the one - body density matrix ( obdm ) , and @xmath278 are bosonic operators . strictly speaking ,
the distinctive feature of phases dominated by one - body coherence is the power - law decay of the obdm , eq . - which may or may not result in a divergent @xmath279 peak depending on the exponent of the power law . yet in a trapped system such a power - law behavior is hard to extract , due to the inhomogeneity imposed by the trap ; and , more importantly , from an experimental point of view the most accessible observable is the momentum distribution .
hence in the following we will concentrate on the @xmath215 ( or condensate ) peak , and in particular on its relative changes in height and width as one - body coherence is suppressed or enhanced .
the model hamiltonian in eq.[h_bos ] has been investigated in a series of regimes .
if the densities , the masses and the interaction parameters are the same for the two species , a paired phase appears above a critical value of the interaction ratio @xmath280 @xcite , that is , one needs sufficiently strong intraspecies repulsion in order to open a spin gap which suppresses one - body coherence .
the resulting paired phase has an exponentially decaying obdm , and a power - law decaying pair - correlation function , @xmath281 a number of numerical and analytical calculations support such a picture , which fully recovers the fermionic case in the tonks - girardeau ( or hardcore ) limit @xmath282 . in the presence of a finite mass imbalance
, the ed paired phase may evolve into a charge - density - wave phase and then into a crystalline phase for specific filling fractions , as explicitly shown in the hardcore case in ref .
finally , a collapsed phase appears in the strongly attractive regime @xmath283 , since the intraspecies repulsion is not sufficiently strong to prevent very large single - site occupancies . on the other hand , in the absence of a trap an arbitrary imbalance in the populations will lead to a revival of one - body coherence , which becomes algebraically decaying ( the same property applies to fermions ) . in the case of bosons , finite - momentum pairing is not observed in the presence of population imbalance , so that population imbalanced bosonic mixtures with attractive interactions do not feature qualitatively different signatures with respect to mixtures of decoupled species .
this issue is quite relevant experimentally , given that having perfectly balanced populations is essentially impossible : even in the case of homonuclear mixtures , fluctuations of order @xmath284 are generally expected in the population of both species .
is then bosonic pairing impossible to observe experimentally in one - dimensional gases ?
luckily the answer is no " , if one adds the two further ingredients which are the central topic of this paper , namely mass imbalance _ and _ a trapping potential . in the hard - core case
, one can extend the lda argument presented in sec .
[ sec : fermi_over ] and conclude that , in presence of mass imbalance , a dimer fluid is stable in the center of the trap up to a finite polarization , depending on @xmath96 and @xmath31 .
( color online ) polarization dependence of core density and momentum distribution in a mixture with @xmath267 and varying @xmath120 .
the system size @xmath158 enters into the definition of @xmath215 .
other parameters as in fig .
[ fig : finite - t ] . ( a ) core density ; ( b ) height of the condensate peak ; ( c ) full width at half maximum ( fwhm ) of the condensate peak ( in units of the inverse lattice spacing ) ;
( d ) trajectory in the @xmath266 plane.,width=264 ] this picture is fully confirmed by a direct calculation of trapped hardcore bosons , shown in fig . [
fig : t0nup24 ] .
there we simulate an experimentally meaningful situation of a trapped hardcore boson mixture , in which @xmath267 is kept fixed , while @xmath120 is varied , thereby varying the polarization @xmath22 continuously .
we monitor the evolution of the core density @xmath285 ( averaged over the 10 core sites ) revealing that a finite window of _ positive _ polarizations exists for which the core remains unpolarized , so that @xmath286 .
in this situation ed pairing is robust at the trap center , and this has strong signatures in the _ global _ coherence properties of the cloud , captured by the momentum distribution . indeed in the polarization window in which the core densities are equal
, the peak height @xmath287 is strongly suppressed , and it hardly changes as @xmath22 decreases , even though @xmath120 is increasing .
the peak height @xmath288 shows two kinks at the two boundaries of the polarization window in question .
in particular the point @xmath188 marks a sharp kink - like minimum associated with full pairing of all particles across the trap .
similar features are revealed in the full width at half maximum ( fwhm ) of the @xmath289 distributions : the polarization window with equal core densities shows a strong enhancement of the fwhm for the @xmath290-particles , and its boundaries are marked by two kinks in the fwhm for the @xmath247-particles .
hence bosonic pairing at the trap center is robust to the presence of a finite positive polarization , and it bears strong signatures at the level of the momentum distribution . a detailed calculation of the polarization gap @xmath130 is still missing for the softcore boson case .
nonetheless in the following we will show that for realistic parameters a softcore boson mixture in a trap exhibits a similar behavior as well .
( color online ) evolution of core density and momentum distribution in a mixture with @xmath291 and varying @xmath120 ; here @xmath292 , and the other parameters are as in fig .
[ fig : t0nup24 ] . ( a ) core density ; ( b ) height of the condensate peak ; ( c ) full width at half maximum ( fwhm ) of the condensate peak ; ( d ) trajectory in the @xmath266 plane.,width=264 ] in this and all of the following sections , we will focus on the case of further population imbalance , and in particular , the case @xmath293 , in which , as discussed earlier , ed pairing is always absent in the trap center . in this regime trimer formation can instead appear when @xmath12 . such a case of strongly imbalanced population has not been extensively studied yet ; in particular , the emergence of trimer formation has up to now been discussed only in systems of fermionic atoms or dipolar molecules @xcite .
the key point here is to qualitatively and quantitatively understand the competition between possible trimer instabilities and the emergence of either phase separation or single species superfluidity as a function of the hopping imbalance , which is expected to play a prominent role as in the fermionic case .
we begin our discussion with the case of hardcore bosons , moving then to the case of softcore bosons in the next sections . ref . has shown that , in the grand - canonical ensemble , a mixture of hardcore bosons or fermions with mass imbalance @xmath294 shows a direct transition from a trimer liquid phase with fixed polarization @xmath12 to a vacuum phase as the chemical potential is lowered .
hence , within the lda approximation , a trapped mixture can not realize a trimer liquid phase in the trap center unless its polarization is exactly at @xmath12 : indeed , if the polarization differs from @xmath295 , the extra particles of either species can not be accommodated in the trap wings , given that a trimer liquid in the trap center can only be flanked by an empty region .
this condition seems to put a very serious limitation to the possibility of observing a trimer liquid phase in a realistic trapped system , given that , as mentioned above , a fine tuning in the population of the two species is experimentally very hard .
nonetheless , strong signatures of the formation of a trimer liquid phase at @xmath12 can be seen in a broader range of polarizations , and specifically in the momentum distribution .
[ fig : trimerformation - nk0 ] shows a polarization scan at very low temperature in a mixture containing @xmath296 particles and a variable number @xmath120 of @xmath290-particles with @xmath297 .
we chose a very weak trapping potential ( @xmath298 ) , leading to low densities in the system , for which trimer formation is a most robust phenomenon @xcite .
we observe that , as expected from lda , the polarization @xmath295 is achieved in the trap center ( @xmath299 ) only when the global polarization @xmath22 is exactly at the same value .
when the trimer condition on the polarization is satisfied , a trimer liquid appears to form , as shown by an anomaly in the one - body coherence properties , namely a weak suppression in the height of the condensate peak @xmath215 as well as a slight enhancement of its width .
but the most notable feature is that the formation of trimers at polarization @xmath12 influences the whole evolution of the one - body coherence properties for nearby values of @xmath22 .
indeed a sharp kink appears in the height of the condensate peak @xmath287 at @xmath12 , marking a net change of slope in the dependence of this quantity on @xmath22 .
in fact a similar kink is observed for @xmath188 , at which the ed paired phase spreads throughout the trap . that kink is associated with the fact that @xmath290-particles added to the system to give @xmath106
will not form dimers with @xmath247-particles , giving rise to a strong enhancement of @xmath290-particle coherence . on the other hand ,
the further enhancement of coherence for @xmath300 shows that for @xmath301 the @xmath290-particles added to the system have the tendency to form trimer bound states with the @xmath247-particles , and it is only when the trimer formation has been saturated at @xmath12 that the one - body coherence grows faster with increasing @xmath120 , whence the kink at @xmath12 . in principle
the extra @xmath290-particles added to give @xmath302 might form quartets ( 3 @xmath290 s and 1 @xmath247 ) , which would lead to an another kink in the condensate peak with p becomes lower than @xmath303 ( saturation of quartets ) , and the same for bound states formed from even more particles . on the other hand the binding energy of these complexes , if finite , will be increasingly low as the size of the complex increases , so that the kinks in the @xmath287 curve will be increasingly weak . in the present section we focus on softcore bosons , and we address the experimentally relevant case of @xmath81rb mixtures confined in species - dependent optical lattices . as discussed in sec .
[ sec : model ] , this case is an archetypical case aimed at underpinning the interplay between bosonic soft - coreness and trimer pairing .
such settings provide intermediate hopping asymmetry and sufficiently large intraspecies repulsion , as required to avoid phase separated regimes for large values of @xmath304
. we will start our discussion by considering the effect of a finite intraspecies interaction in the low - density limit , identifying the more favorable parameter setup to observe trimer physics in a finite , inhomogeneous system , which will then be investigated by means of dmrg and qmc simulations .
understanding the three - body pairing mechanism in the low - density limit sheds light onto the more complicated many - body picture : in fact , a finite trimer binding energy with magnitude comparable to the hopping rates @xmath18 is a good starting point to observe the tp phase at the many - body level , as has been noticed in the fermionic case @xcite . at the few - body level , the main difference between the fermionic case and the bosonic one stems from the finite intraspecies repulsion , which , in a @xmath305 perturbation picture , allows for additional exchange processes in the bosonic case . at a qualitative level , these terms soften the repulsion between a bound @xmath306 pair and a @xmath290-particle , thus potentially increasing the trimer binding energy by a factor of order @xmath307 . in order to get quantitative information
, we evaluate the trimer binding energy via dmrg simulations , by considering the quantity @xmath308 where @xmath309 is the ground state energy of a system of length @xmath7 and population @xmath310 respectively . the trimer binding energy , which corresponds to the trimer gap in the zero - density limit @xcite ,
is then defined as @xmath311 , and can be extrapolated using a fourth - order algebraic fit of simulation data with @xmath312 .
typical results for the parameter range @xmath313 $ ] are shown in fig .
[ f : et ] ( upper panel ) , where , as a comparison mark , we also present results for the hard - core ( hc , @xmath314 ) case .
the first notable feature is that the trimer binding energy of soft - core ( sc ) bosons is always larger than the hc one , thus confirming the raw qualitative picture described above ; the difference between the two notably shrinks at larger lattice depths , where the intraspecies repulsion becomes larger ( @xmath315 for @xmath316 ) . moreover , deeper lattices induce larger mass imbalance , which by itself is reflected in a larger binding energy with respect to the hopping rate @xmath92 .
rb mixtures as a function of the lattice depth @xmath317 ( in units of the recoil energy @xmath318 .
note that here , the recoil energy is identical for the two components as it only depends on the bare mass ) .
panel ( a ) : @xmath319 in units of @xmath320 . in this case ,
deeper lattices lead to larger mass / ratio , and such a larger @xmath319 .
panel ( b ) : @xmath319 in nk units as considered in sec .
[ sec : model ] .
deeper lattices strongly reduce the temperature scale , as evident from the sharp decrease of the binding energy in such scale . for all interaction strengths considered , the soft - core case has typical energy scales always larger than the hard - core one.,title="fig:",width=245 ] rb mixtures as a function of the lattice depth @xmath317 ( in units of the recoil energy @xmath318 .
note that here , the recoil energy is identical for the two components as it only depends on the bare mass ) .
panel ( a ) : @xmath319 in units of @xmath320 . in this case ,
deeper lattices lead to larger mass / ratio , and such a larger @xmath319 .
panel ( b ) : @xmath319 in nk units as considered in sec .
[ sec : model ] .
deeper lattices strongly reduce the temperature scale , as evident from the sharp decrease of the binding energy in such scale . for all interaction strengths considered ,
the soft - core case has typical energy scales always larger than the hard - core one.,title="fig:",width=245 ] in order to establish the optimal experimental setting to observe trimer physics , one has to take into consideration that a finite temperature @xmath273 in ultracold gases may indeed prevent any relevant observation of paired phases with associated binding energy smaller that @xmath321 . in the lower panel of fig .
[ f : et ] , we show the binding energy in nk units in the same parameter regime as in the upper panel : in fact , large values of @xmath317 significantly decrease the absolute value of @xmath319 , making the effect of thermal fluctuations more and more relevant . combining the aforementioned arguments ,
we conclude that the best setting where trimer physics may indeed be observable is determined in our case by the condition @xmath322 ; moreover , since in this case @xmath323 , we will choose in the following @xmath324 , as larger interspecies attraction may indeed lead to a collapse in an inhomogeneous setup . in the following , we employ dmrg calculations in order to elucidate the presence and stability of the trimer phase in trapped systems , keeping up to 400 states per block and applying up to 10 sweeps , with a truncation error in the final dmrg step usually smaller than @xmath325 .
we consider equal trapping potentials , @xmath326 , starting with a commensurate polarization @xmath12 and then study the effects of incommensurate polarizations .
case for attractive @xmath81rb bosonic mixtures . here ,
@xmath327.,width=245 ] as we have seen in sec .
[ sec : density ] , fermionic gases with exactly commensurate polarization can stabilize a trimer phase over the entire system , as shown from the density profiles in fig . [
fig : p_trim ] .
since the trimer instability is favored in the low - density regime , we focused on population regimes where the typical total density in the middle of the trap fluctuates around @xmath328 . in fig .
[ f : bos_bal_dens ] , a typical density profile is shown for @xmath327 particles in a @xmath158 site system : as in the fermionic case , @xmath152 and @xmath329 coincide within numerical errors all over the system , thus providing a first signature of trimer formation .
another independent evidence of trimer formation is then obtained by looking at the correlation function decay in the central part of the system ; in particular , in the trimer phase , both @xmath330 and pairing correlations @xmath331 are expected to decay exponential with @xmath332 , whilst trimer correlations @xmath333 are expected to decay algebraically . in fig .
[ f : bos_bal_corr ] , the decay of correlation functions in the middle of the system for the same parameter set of fig .
[ f : bos_bal_dens ] clearly shows that , while both single particle and dimer correlations decay exponentially , the trimer correlations do not , further confirming the stability of the trimer phase in the inhomogeneous setup ; moreover , @xmath334 does not show any oscillating behavior with @xmath332 , in accordance with the bosonic nature of the three - body bound state .
case as a function of the distance with respect to the trap center : all correlations decay exponentially except for @xmath273 , pointing to a dominant trimer instability . here , @xmath335 denotes single particle correlation function for the @xmath336 species.,width=245 ] as we have seen in sec .
[ sec : density ] for the case of fermions and in sec .
[ s : trimer_onebcoherence ] for hardcore bosons , even a very small deviation from commensurate densities has a drastic effect on trimer phases .
the consequences of @xmath337 may be even more drastic in the bosonic case , which are not subject to the pauli principle and can thus form higher density regions in the middle of the trap .
we have investigated the same parameter regime as in sec .
[ sec : fermions ] in the presence of a minimal density imbalance , @xmath338 , which does also represent the typical experimental threshold of population control .
similarly to the fermionic case , the trimer liquid is fragile with respect to incommensurability . at first
, one sees that the condition @xmath339 is not fulfilled in the middle of the trap : as shown in fig .
[ f : bos_unbal_dens ] , there is a significant departure from commensurability even for the smallest density imbalance considered .
secondly , correlation functions are also strongly affected : in particular , single - particle correlations change significantly from exponential to algebraic decay , signally the emergence of quasi - condensation . in fig .
[ f : bos_unbal_corr ] , we compare the superfluid correlation of the heavy species at different density imbalance for different values of @xmath22 : in both incommensurate cases , the decay in the center of the trap is algebraic . case for attractive @xmath81rb bosonic mixtures .
the total number of particles in the system is @xmath327.,width=245 ] despite the fragility of trimer formation to slight departures from the @xmath12 condition , a strong signature of this phenomenon is still observed in the evolution of the one - body coherence when @xmath22 is changed around the value @xmath295 , as discussed in sec .
[ s : trimer_onebcoherence ] .
here we discuss the robustness of this effect for finite temperatures , and for the softcore case .
[ fig : trimerformation - nk0-t ] and [ fig : trimerformation - nk0-softt ] show the height of the condensate peak with a varying polarization @xmath22 for a mixture containing @xmath291 atoms and a variable number @xmath120 of @xmath290-atoms .
[ fig : trimerformation - nk0-t ] refers to hardcore bosons at increasing temperatures , while fig .
[ fig : trimerformation - nk0-softt ] shows the case of softcore bosons with the same parameters as for @xmath81rb with @xmath340 , and with @xmath341 . in both cases
the trapping potential is very weak , @xmath342 .
we observe that the distinct signatures of formation of dimer and trimer liquids the kinks in @xmath215 at @xmath188 and @xmath12 respectively are still observed in the softcore case , albeit less clearly than in the hardcore case , particularly for what concerns the trimer kink . both for the softcore and the hardcore case
the kinks appear to be robust at fairly low temperatures , although the trimer kink is quickly rounded off as t reaches values in the range @xmath343 .
this shows clearly that the observation of trimer formation with the current experimental diagnostics of time - of - flight measurements is indeed possible as far as the hamiltonian parameters are concerned ( _ e.g. _ for @xmath81rb mixtures ) , but it requires extreme conditions of very low densities and very low temperatures .
the two latter requirements are actually compatible , as very low temperatures can be in principle achieved by further evaporative cooling or algorithmic cooling @xcite of the atoms already loaded in the optical lattice .
( color online ) polarization dependence of the condensate peak at various temperatures for a trapped mixture of hardcore bosons with @xmath291 and increasing @xmath120 .
other parameters are as in fig .
[ fig : trimerformation - nk0].,width=340 ] ( color online ) polarization dependence of the condensate peak at various temperatures for a trapped mixture of softcore bosons with @xmath291 and increasing @xmath120 . here
@xmath344 and @xmath345 ; all other parameters are as in fig . [
fig : trimerformation - nk0-t ] .
, width=340 ] binary bosonic mixtures represent a valuable setup where both dimer and trimer liquid physics in presence of mass imbalance leaves strong signatures in the momentum distribution accessible to experiments . in case of hard - core intraspecies repulsion , equal density pairing in the central part of the trap
is relatively stable with respect to thermal effects , and may be observed in a wide range of polarizations as long as the lighter species density is larger , that is @xmath107 .
trimer physics is instead more delicate : why in general soft - core interactions increase the trimer binding energy at fixed mass imbalance , a very small departure from the commensurate condition @xmath12 drastically changes both correlation functions and real space distribution of trapped systems .
nevertheless , sufficiently low temperatures and densities may indeed allow for the observation of such exotic liquids in a trapped gas by looking at the momentum distribution of the heavier component in both hard - core and soft - core mixtures , a sharp signature being the strong suppression of zero - momentum contribution as a function of @xmath22 .
ultracold atomic gases represent ideal setups to explore unconventional superfluid states of quantum low dimensional systems such as fflo states and superfluids of composite particles . in this work ,
we have investigated binary bosonic and fermionic mixtures in the presence of both finite spin- and mass - imbalance , taking into consideration typical experimental features such as the inhomogeneity induced by trapping potentials and a finite temperature . in the first part
, we have shown how one can modify the effective mass imbalance by considering a proper tuning of the underlying optical lattice potential , inducing different tunneling rates for the different species . in the second part , taking full advantage of the detailed microscopic hamiltonian study of sec .
[ sec : model ] , we have carried out a combined dmrg and qmc numerical study of both fermi - fermi and bose - bose mixtures for realistic experimental parameters . in the fermionic case
, a finite mass - imbalance helps in stabilizing exactly paired phases in presence of spin - imbalance : in sharp contrast with the equal mass case , a majority of light particles stabilizes an equal density region in the middle of the trap in a broad regime of total densities and polarizations .
this phase is also robust to finite temperature effects , as corroborated by qmc simulations .
the partially polarized phase is of the fflo type , with a clear signature in the momentum distribution of pairs .
fflo should be easier to see on the @xmath106 side , where the heavy fermions are the majority species .
we can thus conclude that mass - imbalanced fermionic mixtures such as @xmath346li-@xmath347k represent a valuable setup to observe equal - density pairing and fflo superfluidity in 1d systems under realistic experimental conditions .
large mass imbalance is known to lead to a richer showcase of superfluid states known as composite liquids : as a case study , we have investigated the presence and stability of trimer liquids in both bosonic and fermionic mass - imbalanced mixtures .
while the typical binding energy of trimer bound states is smaller than standard pairs , in the case of exactly commensurate densities trimer liquids are indeed robust in inhomogeneous setups even in the presence of slightly different trapping potentials . in the bosonic case , an experimental signature of such strongly correlated composite liquids
is provided by the single particle momentum distribution , which displays a sharp kink in its polarization dependence even at finite ( albeit small ) temperature . however , once one departs from @xmath12 , the trimer liquid is not stable , as signaled by both the density distribution and the correlation function decay in the central part of the trap . in summary ,
the experimental realization of trimer liquids in binary , mass - imbalanced mixtures represents a challenging task , as both low temperature and densities , combined with an accurate control on spin - imbalance , will be necessary .
nevertheless , such phases may indeed leave appreciable experimental signatures even on easily accessible observables such as momentum distribution .
we thank r. lutchyn and g. orso for fruitful discussions . f.h .-
, k.d . and u.s .
acknowledge support form the deutsche forschungsgemeinschaft through for 801 .
m.d . acknowledges support by the european commission via the integrated project aqute .
k.d . was supported by the national research foundation and the ministry of education , singapore .
m . and u. s. thank the kitp at uc santa barbara , where parts of this research were carried out , for its hospitality .
this research was supported in part by the nsf under grant .
nsf phy05 - 51164 .
thanks nsf for support through grant no .
dmr-0955707 .
m. h. anderson , j. r. ensher , m. r. matthews , c. e. wieman , and e. a. cornell , science * 269 * , 198 ( 1995 ) .
k. b. davis , m .- o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , phys .
. lett . * 75 * , 3969 ( 1995 ) . c. c. bradley , c. a. sackett , j. j. tollett , and r. g. hulet , phys .
lett . * * 75 , 1687 ( 1995 ) .
m. bartenstein , a. altmeyer , s. riedl , s. jochim , c. chin , j. h. denschlag , and r. grimm , phys .
lett . * 92 * , 120401 ( 2004 ) .
t. bourdel , l. khaykovich , j. cubizolles , j. zhang , f. chevy , m. teichmann , l. tarruell , s. j. j. m. f. kokkelmans , and c. salomon , phys .
lett . * 93 * , 050401 ( 2004 ) . c. a. regal , m. greiner , and d. s. jin , phys .
* 92 * , 040403 ( 2004 ) .
m. w. zwierlein , c. a. stan , c. h. schunck , s. m. f. raupach , a. j. kerman , and w. ketterle , phys .
* 92 * , 120403 ( 2004 ) . m. w. zwierlein , c. h. schunck , a. schirotzek , and w. ket- terle , nature * 442 * , 54 ( 2006 ) .
m. w. zwierlein , a. schirotzek , c. h. schunck , and w. ket- terle , science * 311 * , 492 ( 2006 ) .
g. b. partridge , w. li , r. i. kamar , y. an liao , and r. g. hulet , science * 311 * , 503 ( 2006 ) . g. b. partridge , w. li , y. a. liao , r. g. hulet , m. haque , and h. t. c. stoof , phys .
* 97 * , 190407 ( 2006 ) .
liao , a. s. c. rittner , t. paprotta , w. li , g. b. partridge , r. g. hulet , s. k. baur , and e. j. mueller , nature * 467 * , 567 ( 2010 ) .
g. sarma , phys .
solids * 24 * , 1029 ( 1963 ) .
p. fulde and a. ferrell , phys
. rev . * 135 * , a550 ( 1964 ) .
a. larkin and y. ovchinnikov , zh .
fiz * 47 * , 1136 ( 1964 ) .
i. bloch , j. dalibard , and w. zwerger , rev .
* 80 * , 885 ( 2008 ) .
m. khl , h. moritz , t. stferle , k. gnter , and t. esslinger , phys .
* 94 * , 080403 ( 2005 ) .
jrdens , n .
strohmaier , k .
gnter , h .
moritz , and t. esslinger , nature ( london ) * 455 * , 204 ( 2008 ) .
u. schneider , l. hackermller , s. will , t. best , i. bloch , t. a. costi , r. w. helmes , d. rasch , and a. rosch , science * 322 * , 1520 ( 2008 ) . m. taglieber , a .- c .
voigt , t. aoki , t. w. hnsch , and k. dieckmann , phys .
lett . * 100 * , 010401 ( 2008 ) .
et al . _ ,
phys . rev .
lett . * 100 * , 053201 ( 2008 ) .
voigt , m. taglieber , l. costa , t. aoki , w. wieser , t. w. hnsch , and k. dieckmann , phys .
102 * 020405 ( 2009 ) .
a. trenkwalder , c. kohstall , m. zaccanti , d. naik , a. i. sidorov , f. schreck , and r. grimm , phys .
106 * , 115304 ( 2011 ) .
m. greiner , o. mandel , t. hnsch , and i. bloch , nature ( london ) * 419 * , 51 ( 2002 ) .
t. stferle , h. moritz , c. schori , m. khl , and t. esslinger , phys .
* 92 * , 130403 ( 2004 ) .
e. haller , r. hart , m. mark , j. danzl , l. reichsllner , m. gustavsson , m. dalmonte , g. pupillo , and h .- c .
ngerl , nature * 456 * , 597 ( 2010 )
. j. catani , l. de sarlo , g. barontini , f. minardi , and m. inguscio , phys .
a * 77 * , 011603 ( 2008 ) .
d. mckay and b. demarco , new j. phys .
* 12 * , 055013 ( 2010 ) .
b. gadway , d. pertot , j. reeves , m. vogt , and d. schneble , phys .
lett . * 107 * , 145306 ( 2011 ) . a. b. kuklov and b. v. svistunov , phys .
* 90 * , 100401 ( 2003 ) .
e. altman _
_ , new j. phys . * 5 * , 113 ( 2003 ) .
m. guglielmino , v. penna , and b. capogrosso - sansone , phys .
a * 84 * , 031603 ( 2011 ) .
l. mathey , phys .
b * 75 * , 144510 ( 2007 ) .
a. hu , l. mathey , i. danshita , e. tiesinga , c. j. williams , and c. w. clark , phys .
a * 80 * , 023619 ( 2009 ) .
t. roscilde , c. degli esposti boschi and m. dalmonte , epl * 97 * , 23002 ( 2012 ) .
g. orso , e. burovski , and t. jolicoeur , phys .
lett . * 104 * , 065301 ( 2010 ) .
e. burovski , g. orso , and t. jolicoeur , phys .
* 103 * , 215301 ( 2009 ) .
a. feiguin , f. heidrich - meisner , g. orso , and w. zwerger , lect . not
. phys . * 836 * , 503 ( 2011 ) .
s. r. white , phys .
lett . * 69 * , 2863 ( 1992 ) .
u. schollwck , rev .
phys . * 77 * , 259 ( 2005 ) .
u. schollwck , ann .
( ny ) * 326 * , 96 ( 2011 ) .
o. f. syljusen and a. w. sandvik , phys .
e * 66 * , 046701 ( 2002 ) .
g. g. batrouni , m. j. wolak , f. hebert , and v. g. rousseau , epl * 86 * , 47006 ( 2009 ) .
b. wang , h .-
chen , and s. das sarma , phys .
rev . a * 79 * , 051604(r ) ( 2009 ) .
g. roux , e. burovski , and t. jolicoeur , phys .
a * 83 * , 053618 ( 2011 ) .
f. essler , h. frahm , f. ghmann , a. klmper , and v. e. kore- pin , _ the one - dimensional hubbard model _
( cambridge univer- sity press , 2005 ) . g. orso , phys . rev .
lett . * 98 * , 070402 ( 2007 ) .
h. hu , x .- j . liu , and p. d. drummond , phys .
lett . * 98 * , 070403 ( 2007 ) .
m. casula , d. m. ceperley , and e. j. mueller , phys .
a * 78 * , 033607 ( 2008 ) . f. heidrich - meisner , g. orso , and a. e. feiguin , phys .
a * 81 * , 053602 ( 2010 ) .
a. e. feiguin and f. heidrich - meisner , phys .
b * 76 * , 220508(r ) ( 2007 ) . w. v. liu , f. wilczek , and p. zoller , phys .
a * 70 * , 033603 ( 2004 ) .
, s. ospelkaus , phd thesis , university of hamburg ( 2006 ) .
a. albus , f. illuminati , and j. eisert , phys .
a * 68 * , 023606 ( 2003 ) .
j. catani , g. barontini , g. lamporesi , f. rabatti , g. thalhammer , f. minardi , s. stringari , and m. inguscio , phys .
* 103 * , 140401 ( 2009 ) .
lett . * 89 * , 283202 ( 2002 ) .
g. g. batrouni , m. h. huntley , v. g. rousseau , and r. t. scalettar , phys .
lett . * 100 * , 116405 ( 2008 ) .
m. tezuka and m. ueda , phys .
lett . * 100 * , 110403 ( 2008 ) .
m. rizzi , m. polini , m. a. cazalilla , m. r. bakhtiari , m. p. tosi , and r. fazio , phys .
b * 77 * , 245105 ( 2008 ) . a. i. buzdin and v. v. tugushev , zh .
. fiz . * 85 * , 735 ( 1983 ) . k. machida and h. nakanishi , phys .
b * 30 * , 122 ( 1984 ) .
k. yang , phys .
b * 63 * , 140511 ( 2001 ) .
a. lscher , r. m. noack , and a. m. luchli , phys . rev . a * 78 * , 013637 ( 2008 ) . j. lee and x. guan , nucl .
b * 853 * , 125 ( 2011 ) .
j. n. fuchs , a. recati , and w. zwerger , phys .
lett . * 93 * , 090408 ( 2004 ) .
i. v. tokatly , phys .
lett . * 93 * , 090405 ( 2004 ) .
a. recati , j. n. fuchs , and w. zwerger , phys .
a * 71 * , 033630 ( 2005 ) .
f. heidrich - meisner , a. e. feiguin , u. schollwck , and w. zwerger , phys .
a * 81 * , 023629 ( 2010 ) .
s. k. baur , j. shumway , and e. j. mueller , phys .
a * 81 * , 033628 ( 2010 ) .
m. m. parish , s. k. baur , e. j. mueller , and d. a. huse , phys .
99 * , 250403 ( 2007 ) .
e. zhao and w. v. liu , phys .
a * 78 * , 063605 ( 2008 ) .
a. e. feiguin and f. heidrich - meisner , phys .
* 102 * , 076403 ( 2009 ) .
r. m. lutchyn , m. dzero , and v. m. yakovenko , phys .
rev . a * 84 * , 033609 ( 2011 ) .
p. kakashvili and c. j. bolech , phys .
a * 79 * , 041603 ( 2009 ) .
m. j. wolak , v. g. rousseau , c. miniatura , b. gremaud , r. t. scalettar , and g. g. batrouni , phys .
a * 82 * , 013614 ( 2010 ) .
g. orso , e. burovski , and t. jolicoeur , cras ( paris ) physique * 12 * , 39 ( 2011 ) .
u. schneider , l. hackermller , j. p. ronzheimer , s. will , s. braun , t. best , i. bloch , e. demler , s. mandt and d. rasch and a. rosch , nature phys . * 8 * , 213 ( 2012 ) . f. heidrich - meisner , s. r. manmana , m. rigol , a. muramatsu , a. e. feiguin , and e. dagotto , phys .
a * 80 * , 041603 ( 2009 ) .
s. langer , m. schuetz , i. mcculloch , u. schollwck , and f. heidrich - meisner , arxiv:1109.4364 ( unpublished ) .
t. roscilde , phys .
a * 77 * , 063605 ( 2008 ) .
m. rizzi and a. imambekov , phys .
a * 77 * , 023621 ( 2008 ) .
d. s. petrov , g. e. astrakharchik , d. j. papoular , c. salomon , and g. v. shlyapnikov , phys .
lett . * 99 * , 130407 ( 2007 ) .
m. dalmonte , p. zoller , and g. pupillo , phys .
lett . * 107 * , 163202 ( 2011 ) .
m. popp , j .- j .
garcia - ripoll , k. g. vollbrecht , and j. i. cirac , phys .
a * 74 * , 013622 ( 2006 ) . | we present a systematic investigation of attractive binary mixtures in presence of both spin- and mass - imbalance in one dimensional setups described by the hubbard model . after discussing typical cold atomic experimental realizations and the relation between microscopic and effective parameters , we study several many - body features of trapped fermi - fermi and bose - bose mixtures such as density profiles , momentum distributions and correlation functions by means of numerical density - matrix - renormalization - group and quantum monte carlo simulations .
in particular , we focus on the stability of fulde - ferrell - larkin - ovchinnikov , dimer and trimer fluids in inhomogeneous situations , as typically realized in cold gas experiments due to the harmonic confinement .
we finally consider possible experimental signatures of these phases both in the presence of a finite polarization and of a finite temperature . |
let @xmath3 be @xmath1 distinct points in the space @xmath0 , called vertices , and positive numbers @xmath4 , called weights .
the function @xmath5 defined by @xmath6 is called the weber function , where @xmath7 denotes the euclidean norm .
it is well - known that this function is not differentiable at the vertices , and strictly convex if the vertices are not collinear ( we will assume this hypothesis from now on ) .
the weber problem ( also known as the fermat - weber problem ) is to find a point in @xmath0 that minimizes the weighted sum of euclidean distances from the @xmath1 given points , that is , we have to find the solution of the following unconstrained optimization problem : @xmath8 this problem has a unique solution @xmath9 in @xmath0 .
the problem was also stated as a pure mathematical problem by fermat @xcite , cavalieri @xcite , steiner @xcite , fasbender @xcite and many others .
several solutions , based on geometrical arguments , were proposed by torricelli and simpson . in @xcite historical details and geometric aspects
were presented by kupitz and martini . in @xcite weber
formulated the problem ( [ equation : unconstrained weber problem ] ) from an economical point of view .
the vertices represent customers or demands , the solution to the problem denotes the location of a new facility , and the weights are costs associated with the interactions between the new facility and the customers . among several schemes to solve the weber location problem ( see @xcite ) ,
one of the most popular methods was presented by weiszfeld in @xcite .
the weiszfeld algorithm is an iterative method based on the first - order necessary conditions for a stationary point of the objective function . if we define @xmath10 by : @xmath11 the weiszfeld algorithm is : @xmath12 where @xmath13 is a starting point .
the weiszfeld algorithm ( [ equation : weiszfeld algorithm ] ) , despite of its simplicity , has a serious problem if some @xmath14 lands accidentally in a vertex @xmath15 , because the algorithm gets stuck at @xmath15 , even when @xmath15 is not the solution of ( [ equation : unconstrained weber problem ] ) .
many authors studied the set of initial points for which the sequence generated by the weiszfeld algorithm yields in a vertex ( see @xcite ) .
vardi and zhang @xcite derived a simple but nontrivial modification of the weiszfeld algorithm in which they solved the problem of landing in a vertex . generalizations and new techniques for the fermat - weber location problem have been developed in recent years . in @xcite eckhardt applied the weiszfeld algorithm to generalized weber problems in banach spaces .
an exact algorithm for a weber problem with attraction and repulsion was presented by chen et al . in @xcite .
kaplan and yang @xcite proved a duality theorem which includes as special cases a great variety of choices of norms in the terms of the fermat - weber sum . in @xcite carrizosa et al .
studied the so called regional weber problem , which allows the demand not to be concentrated onto a finite set of points , but follows an arbitrary probability measure . in @xcite drezner and
wesolowsky studied the case where different @xmath16 norms are used for each demand point .
in @xcite the so called complementary problem ( the weber problem with one negative weight ) was studied by jalal and krarup , and geometrical solutions were given .
in @xcite drezner presented a weiszfeld - like iterative procedure and convergence is proved if appropriate conditions hold . in some practical problems
it is necessary to consider barriers ( forbidden regions ) .
barriers were first introduced to location modeling by katz and cooper @xcite .
there exist several heuristic and iterative algorithms for single - facility location problems for distance computations in the presence of barriers ( see @xcite ) . in @xcite pfeiffer and
klamroth presented a unified formulation for problems with barriers and network location problems . a complete reference to barriers in location problems
can be found in @xcite .
barriers can be applied to model real life problems where regions like lakes and mountains are forbidden . on the other hand
, there are location problems whose solution needs to lie within a closed set .
for example , see @xcite for a discussion of the case when the solution is constrained to be within a maximum distance of each demand point .
drezner and wesolowsky @xcite studied the problem of locating an obnoxious facility with rectangular distances ( @xmath17 norm ) , where the facility must lie within some prespecified region ( linear constraints ) .
a primal - dual algorithm to deal with the constrained fermat - weber problem using mixed norms was developed in @xcite by idrissi et al .. in @xcite hansen et al .
presented an algorithm for solving the weber problem when the set of feasible locations is the union of a finite number of convex polygons . in @xcite pilotta and torres considered a weber location problem with box constraints .
constrained weber problems arise when we require that the solution is in an area ( feasible region ) determined by , for example , environmental and/or political reasons .
it could be the case for a facility producing dangerous materials that must be installed in a restricted ( constrained ) area .
another example could be the location of a plant in an industrial zone or of a hospital in a non - polluted area . in this paper
a constrained location problem is considered .
an algorithm is proposed to solve the following problem : @xmath18 where @xmath19 is a closed and convex set , generalizing the problem formulated in @xcite .
problem ( [ equation : constrained weber problem ] ) could be seen as a nonlinear programming problem and solved by standard solvers , but they may fail since the weber function is not differentiable at the vertices .
it can be proved that problem ( [ equation : constrained weber problem ] ) has a unique solution @xmath20 , since the function @xmath21 is strictly convex and @xmath19 is a closed and convex set . on the other hand ,
it is well - known that the convex hull of the given vertices @xmath3 contains the solution @xmath9 of the unconstrained weber problem ( see for instance @xcite ) .
if @xmath19 contains the convex hull , both solutions @xmath20 and @xmath9 agree .
in other cases , the solution @xmath20 is not necessarily a projection of @xmath9 over @xmath19 ( see @xcite ) .
the algorithm is based basically on a slight variation of an orthogonal projection of the weiszfeld algorithm presented in @xcite , that is well defined even when an iterate coincides with a vertex .
properties of the sequence generated by the proposed algorithm related with the minimization problem [ equation : constrained weber problem ] will be proved in the following sections .
the paper is structured as follows : section [ section : the modified weiszfeld algorithm ] describes the results in @xcite in which a modified weiszfeld algorithm is presented and some notation is introduced . in section [ section : the proposed algorithm ] the proposed algorithm is defined .
section [ section : some definitions and technical results ] is dedicated to definitions and technical lemmas . in section [ section : convergence results ]
the main results about convergence to optimality are presented .
numerical experiments are considered in section [ section : numerical experiments ] .
finally , conclusions are given in section [ section : conclusions ] .
some words about notation .
as it was mentioned , we will call @xmath9 the solution of problem ( [ equation : unconstrained weber problem ] ) and @xmath20 the solution of problem ( [ equation : constrained weber problem ] ) . the symbols
@xmath22 and @xmath23 will refer to the standard euclidean norm and standard inner product in @xmath0 , respectively . for a function @xmath24 we will denote by @xmath25 the left - hand side derivative at @xmath26 , and by @xmath27 the right - hand side derivative at @xmath26 .
this section reviews the main results presented in @xcite in which the authors generalize the weiszfeld algorithm for the case that an iterate lands on a vertex . from now on ,
this algorithm will be referred to as the modified weiszfeld algorithm . in order to make notation easier ,
we define the function @xmath28 by : @xmath29 notice that @xmath30 for all @xmath31 . in @xcite ,
the number @xmath32 was called @xmath33 . a generalization for the iteration function @xmath34 , defined in ( [ equation : weiszfeld iteration function ] ) , is given by @xmath35 defined as follows : @xmath36 notice that @xmath37 coincides with @xmath38 in @xmath39 .
let @xmath40 and @xmath41 be : @xmath42 the function @xmath43 generalizes the negative gradient of the weber function since , for all @xmath44 , @xmath45 the following lemma is very easy to prove ( see ( * ? ? ?
* equation ( 14 ) ) ) , and it relates the functionals @xmath37 and @xmath43 .
[ lemma : widetilder equal to 2a(widetildet - x ) ] for all @xmath31 we have @xmath46 $ ] .
if we define @xmath47 by : @xmath48 we can see that @xmath49 for all @xmath31 . the modified weiszfeld algorithm presented in @xcite
is defined by : @xmath50 where @xmath13 and @xmath51 is given by : @xmath52 where @xmath53 is defined by @xmath54 .
[ remark : properties of beta ] 1 .
if @xmath44 , then @xmath55 because @xmath56 .
so , we can deduce that @xmath57 .
notice that this fact implies that the functional @xmath58 is continuous in @xmath39 .
it can be seen that if @xmath59 , then @xmath60 ( see @xcite ) .
3 . from equation ( [ equation : iteration function of the modified algorithm with beta ] )
we obtain that @xmath61 for @xmath31 .
the main result in @xcite is : [ theorem : main result of modified weiszfeld algorithm ]
the following propositions are equivalent : 1 . @xmath62 .
2 . @xmath63 .
@xmath64 . where @xmath65
this section is dedicated to describe the proposed algorithm , introducing some definitions and remarks .
first of all , we can notice that problem ( [ equation : constrained weber problem ] ) has a unique solution , due to the fact that @xmath21 is a non - negative , strictly convex , and continuous function , @xmath66 and @xmath19 is closed and convex . in order to define the proposed algorithm at the vertices
, we will need to determine which points of the segment that joins @xmath15 and @xmath67 are in the feasible set @xmath19 . if @xmath68 , let the set @xmath69 be defined by : @xmath70 : ( 1 - \lambda ) t(a^k ) + \lambda a^k \in \omega \right\}.\ ] ] notice that @xmath69 could be equal to the empty set in case that @xmath15 and @xmath67 do not belong to @xmath19 . on the other hand , if @xmath71 , then @xmath72 , which means that @xmath73 .
thus , we can define : @xmath74 in case a vertex @xmath15 is not in @xmath19 , there is no need to define the number @xmath75 . in the following lemma ,
a set of basic properties of @xmath75 are listed : [ lemma : properties of lambda(a^k ) ] if @xmath68 and @xmath71 then : 1 .
@xmath76 $ ] .
2 . if @xmath77 then @xmath78 .
3 . if @xmath79 then @xmath80 $ ] .
the proof of ( a ) follows from the definition of @xmath69 .
if @xmath77 , then @xmath81 , so @xmath82 , and this proves ( b ) . finally , for item ( c ) , let us consider that @xmath79 . since @xmath19 is a closed set , there is an entire ball centered at @xmath67 that does not intersect @xmath19 , which implies that there exists @xmath83 such that @xmath84 for all @xmath85 $ ] .
thus , @xmath86 $ ] and this concludes the proof .
let us call @xmath87 the orthogonal projection over @xmath19 .
since @xmath19 is a nonempty , closed and convex set , the operator @xmath88 is a continuous function @xcite .
we define the iteration function @xmath89 by : @xmath90 there will be no need to define @xmath2 outside @xmath19 since the proposed algorithm generates a sequence of feasible points .
the iteration function @xmath2 at @xmath91 coincides with the orthogonal projection of @xmath92 over the feasible set when @xmath93 is different from the vertices . only when @xmath93 is a vertex @xmath15 belonging to @xmath19 , @xmath94 is defined as the farthest possible feasible point of the segment that joins @xmath93 with @xmath92 .
the following remark states some basic properties of the iteration function of the proposed algorithm .
[ remark : properties of q ] if @xmath71 and @xmath77 , then @xmath95 .
if @xmath71 , it can be seen that : @xmath96 the functional @xmath2 is continuous in @xmath39 .
the proofs of ( a ) and ( b ) are straightforward . for ( c ) , since @xmath88 is continuous in @xmath0 ( see @xcite ) and @xmath58 is continuous in @xmath39 ( see remark [ remark : properties of beta ] ) , we have that @xmath2 is continuous in @xmath39 .
the proposed algorithm is described below .
[ algorithm : proposed algorithm ] let @xmath97 be a closed and convex set . assume that @xmath98 is an initial approximation such that @xmath99 for all @xmath100 and @xmath101 .
given @xmath102 a tolerance and @xmath103 , do the following steps to compute @xmath14 : * step 1 : * compute : @xmath104 * step 2 : * stop the execution if @xmath105 and declare @xmath14 as solution to the problem ( [ equation : constrained weber problem ] ) .
otherwise return to step 1 . from the definition of @xmath2
it follows that algorithm [ algorithm : proposed algorithm ] generates a sequence of feasible iterates .
also notice that if there are vertices in the feasible set , @xmath106 can be one of them , for example , a vertex @xmath107 such that @xmath108 for all @xmath101 . on the other hand ,
if there are no vertices in the feasible set , @xmath106 can be chosen as the projection over @xmath19 of the null vector .
the purpose of this section is to define some entities and prove technical lemmas that will be important in the proof of the main results .
first of all , we will define some useful operators for making notation easier . if @xmath109 , then we define @xmath110 and @xmath111 by : @xmath112 notice that , when @xmath113 , @xmath114 is not necessarily a norm and @xmath115 is not necessarily an inner product . according to this definition ,
if @xmath116 and @xmath117 are sets such that @xmath118 and @xmath119 , it can be seen that : @xmath120 for @xmath91 , let us define the following sets of indices : @xmath121 notice that for all @xmath31 we have that @xmath122 and @xmath123 .
let @xmath124 be the following function : * if @xmath44 : @xmath125.\ ] ] * if @xmath126 for some @xmath68 : @xmath127.\ ] ] it can be seen that the function @xmath128 is related to the iteration function @xmath2 of the proposed algorithm , and the iteration function @xmath58 of the modified algorithm .
[ lemma : properties of alpha ] if @xmath91 , then @xmath129.$ ] if @xmath44 , then : @xmath130 = \sum_{j=1}^m \frac { w_j q(x ) } { \| x - a^j \| } - \sum_{j=1}^m \frac { w_j a^j } { \| x - a^j \| } \\ & & \\ & = & \left ( \sum_{j=1}^m \frac { w_j } { \| x - a^j \| } \right ) \left [ q(x ) - \frac { \sum_{j=1}^m \frac { w_j a^j } { \| x - a^j \| } } { \sum_{j=1}^m \frac { w_j } { \| x - a^j \| } } \right ] = 2 a(x ) \left [ q(x ) - \widetilde{t}(x ) \right ] \\ & & \\ & = & 2 a(x ) \left [ q(x ) - t(x ) \right].\end{aligned}\ ] ] where in the last equalities we have used the definition of @xmath37 as in ( [ equation : generalized weiszfeld iteration function ] ) , and the fact that @xmath131 due to remark [ remark : properties of beta ] .
if @xmath132 for some @xmath68 , we follow a similar procedure than in the previous case .
now , we will define auxiliary functions that take into account the projection @xmath88 in order to prove a descent property of @xmath21 ( see next sections ) .
if @xmath91 , we define : 1 .
@xmath133 , where : @xmath134 2 .
if @xmath135 define @xmath136 where : @xmath137 a useful property of @xmath138 , that follows from the definition , is pointed out in the following remark . [
remark : ex o q = q ] if @xmath91 then @xmath139 .
the iteration function @xmath2 inherits an important property from the orthogonal projection @xmath88 .
[ lemma : property of q ] if @xmath91 we have that @xmath140 .
if @xmath44 , then @xmath141 . by a property of the orthogonal projection @xcite we have that @xmath142 .
if @xmath132 for some @xmath68 , remark [ remark : properties of q ] and lemma [ lemma : properties of lambda(a^k ) ] imply : @xmath143 and this concludes the proof .
the next technical lemma will help us to save computations in other lemmas .
[ lemma : norm of q - aj ] if @xmath91 , @xmath116 is a subset of @xmath144 and @xmath100 , then : @xmath145 if @xmath91 , we have : @xmath146 if @xmath91 , let us define @xmath147 by : @xmath148 the values that @xmath149 assumes at @xmath93 and @xmath94 will play an important role in the proof of a property of the objective function @xmath21 .
[ lemma : property of gx(x ) ] let @xmath91 be . 1 . if @xmath44 then : @xmath150 2 .
if @xmath132 for some @xmath68 , then @xmath151 . let us suppose that @xmath44 . by property ( [ equation : property norma ] ) and
( [ equation : definition of ex ] ) , we have for @xmath152 : @xmath153 using lemma [ lemma : norm of q - aj ] , we can see that : @xmath154.\end{aligned}\ ] ] due to ( [ equation : property norma ] ) , the definition of the weber function @xmath21 , the definition of @xmath155 as in ( [ equation : weight function ] ) , the property ( [ equation : property cinnerproda ] ) and the definition of @xmath128 as in ( [ equation : definition of alpha1 ] ) , we obtain : @xmath156 by lemma [ lemma : properties of alpha ] , the fact that @xmath157 for all @xmath158 and ( [ equation : property innerproda ] ) , we get : @xmath159 which concludes the proof of ( a ) .
now , let us assume that @xmath132 for some @xmath68 .
then : @xmath160 this concludes the proof of ( b ) .
the number @xmath161 can be computed in the next lemma .
[ lemma : property of gx(qx ) ] let @xmath91 be . 1 .
if @xmath44 then : @xmath162 2 . if @xmath132 for some @xmath68 , then @xmath163 first , let us consider @xmath44 .
due to remark [ remark : ex o q = q ] we have : @xmath164 by lemma [ lemma : norm of q - aj ] we obtain : @xmath165.\end{aligned}\ ] ] due to the definition of the weber function @xmath21 , the definition of @xmath155 as in ( [ equation : weight function ] ) and the definition of @xmath128 as in ( [ equation : definition of alpha1 ] ) , we deduce that : @xmath166 by lemma [ lemma : properties of alpha ] we get : @xmath167 concluding the proof of ( a ) .
now , consider @xmath132 for some @xmath68 .
due to ( [ equation : definition of gx ] ) we have : @xmath168 by lemma [ lemma : norm of q - aj ] , the definition of the weber function @xmath21 and the definition of @xmath155 as in ( [ equation : weight function ] ) we obtain : @xmath169 \right\rangle \\ & + & w_k \| q(a^k ) - a^k \|.\end{aligned}\ ] ] manipulating algebraically , @xmath170 due to the definition of @xmath128 ( see ( [ equation : definition of alpha2 ] ) ) and the definition of @xmath43 ( see ( [ equation : definition of widetilder ] ) ) we get : @xmath171 by lemma [ lemma : widetilder equal to 2a(widetildet - x ) ] and lemma [ lemma : properties of alpha ] we have : @xmath172 which concludes the proof . the next lemma deals with the last two terms of @xmath173 .
[ lemma : property of something equal to zero ] if @xmath71 for some @xmath68 , the number @xmath174 is equal to zero . if @xmath175 the result is true
so , from now on , let us consider that @xmath176 .
first , let us check that @xmath59 . in case that @xmath177 , then @xmath178 by theorem [ theorem : main result of modified weiszfeld algorithm ] .
since @xmath71 , then @xmath77 . by remark [ remark : properties of q ]
we have that @xmath179 which is a contradiction . by remark [ remark : properties of beta ]
, we have that @xmath180 ( since @xmath59 ) and : @xmath181 extracting common factors , using remarks [ remark : properties of beta ] and [ remark : properties of q ] , the fact that @xmath182 ( if @xmath178 then @xmath177 by theorem [ theorem : main result of modified weiszfeld algorithm ] ) , and the fact that @xmath183 ( if @xmath184 then @xmath178 by definition ( [ equation : iteration function of the modified algorithm with beta ] ) ) we get that : @xmath185.\end{aligned}\ ] ] simplifying and using the definition of @xmath186 we have that : @xmath187 = 0,\ ] ] which concludes the proof .
the purpose of the next two lemmas is to determine a strict inequality between the functions @xmath149 and @xmath21 at suitable points .
first of all , we have to prove the following result .
[ lemma : comparison between gxqx and gxx ] let @xmath91 be such that @xmath188 . 1 . if @xmath44 , then @xmath189 .
besides that , if @xmath190 and @xmath191 , then @xmath192 .
2 . if @xmath132 for some @xmath68 , then @xmath193 . if @xmath44 , then @xmath194 , by lemma [ lemma : property of gx(x ) ] and lemma [ lemma : property of gx(qx ) ] . besides that , if @xmath190 and @xmath191 we deduce that @xmath195 . thus , @xmath192 .
if @xmath132 for some @xmath68 , by lemmas [ lemma : property of gx(x ) ] , [ lemma : property of gx(qx ) ] and [ lemma : property of something equal to zero ] we have : @xmath196 due to lemma [ lemma : property of q ] and the fact that @xmath197 we obtain : @xmath198 and the proof is finished .
[ lemma : property gxqx < fx/2 ] let @xmath91 be such that @xmath188
. then @xmath199 .
let us consider the case when @xmath44 . by lemmas [ lemma : property of q ] , [ lemma : property of gx(x ) ] and [ lemma : comparison between gxqx and gxx ] we have that
: @xmath200 if @xmath201 , then @xmath202 .
therefore : @xmath203 if @xmath190 and @xmath204 , then there exists an index @xmath205 such that @xmath206 since @xmath188 .
thus , @xmath207 , which implies : @xmath208 if @xmath190 and @xmath191 , due to lemmas [ lemma : property of q ] , [ lemma : property of gx(x ) ] and [ lemma : comparison between gxqx and gxx ] , we have that : @xmath209 now , when @xmath132 for some @xmath68 , @xmath210 due to lemma [ lemma : property of gx(x ) ] and lemma [ lemma : comparison between gxqx and gxx ] .
the next lemma states an equality that relates the weber function and @xmath149 at appropriate points
when @xmath44 .
besides that , this result will be crucial in the next section .
[ lemma : equality of gxqx ] let @xmath44 be such that @xmath91 and @xmath188 . then : @xmath211 due to the definition of @xmath149 as in ( [ equation : definition of gx ] ) and remark [ remark : ex o q = q ] we get that : @xmath212 adding and subtracting @xmath213 we have : @xmath214 ^ 2 \\
& = & \frac{1}{2 } \sum_{j=1}^m w_j \left\| x - a^j \right\| + \sum_{j=1}^m w_j \left ( \left\| q(x ) - a^j \right\| - \left\| x - a^j \right\| \right ) \\ & + & \sum_{j=1}^m \frac { w_j } { 2 \| x - a^j \| } \left ( \left\| q(x ) - a^j \right\| - \left\| x - a^j \right\| \right)^2.\end{aligned}\ ] ] notice that the first term of the last equality is the weber function ( divided by two ) , and the last term is a non - negative number , so we will define it as @xmath215 .
so , using the definition of the weber function in the middle term we obtain : @xmath216
this section states the main results about convergence of the sequence @xmath217 generated by algorithm [ algorithm : proposed algorithm ] .
the next theorem establishes that if a point @xmath91 is not a fixed point of the iteration function , then the function @xmath21 strictly decreases at the next iterate .
[ theorem : descent property ] let @xmath91 be such that @xmath188
. then @xmath218 .
let us consider that @xmath44 .
by lemma [ lemma : property gxqx < fx/2 ] , we have that : @xmath219 by lemma [ lemma : equality of gxqx ] we get that : @xmath220 simplifying the last expression we obtain : @xmath221 finally , @xmath222 therefore , @xmath218 . now , consider that @xmath132 for some @xmath68 .
following a reasoning similar than in @xcite , using lemma [ lemma : comparison between gxqx and gxx ] we have that : @xmath223 by definition of @xmath224 we know that : @xmath225 using the fact that @xmath226 for @xmath227 and @xmath228 we obtain that : @xmath229 rearranging terms we deduce that : @xmath230 and the proof is complete .
[ corollary : not increasing sequence ] let @xmath217 be the sequence generated by algorithm [ algorithm : proposed algorithm ] .
then the sequence @xmath231 is not increasing .
even more , each time @xmath232 the sequence strictly decreases at the next iterate .
if the sequence @xmath217 generated by algorithm [ algorithm : proposed algorithm ] were not bounded , then we could choose a subsequence @xmath233 such that @xmath234 .
but this implies that @xmath235 , which is a contradiction since the sequence @xmath231 is not increasing .
the sequence @xmath217 generated by algorithm [ algorithm : proposed algorithm ] is bounded .
so , there exists a subsequence convergent to a point @xmath236 .
hence , @xmath20 is a feasible point . due to the nondifferentiability of @xmath21 at the vertices @xmath3 , we can not use the kkt optimality conditions at @xmath15 .
therefore , if @xmath15 and @xmath237 are in @xmath19 , let us define @xmath238 \rightarrow \mathbbm{r}$ ] by : @xmath239 if @xmath71 , @xmath240 , @xmath241 $ ] and @xmath19 convex , we have that @xmath242 . notice that the right - hand side derivative @xmath243 ( or the directional derivative of @xmath21 in the direction of @xmath237 ) exists ( see @xcite ) .
besides that , @xmath244 the next lemma shows that if we are in a vertex @xmath15 , the directional derivative of @xmath21 at @xmath15 in the direction of @xmath245 is a descent direction .
[ lemma : property of a^k as a minimum ] let @xmath71 be such that @xmath79 . then : @xmath246,\ ] ] where : @xmath247 a(a^k )
\left\| \widetilde{t}(a^k ) - a^k \right\| \left\| q(a^k ) - a^k \right\|.\ ] ] if @xmath178 then @xmath77 , which is a contradiction . besides that ,
if @xmath184 , we would have that @xmath178 because of ( [ equation : iteration function of the modified algorithm with beta ] ) , and again it would be a contradiction .
so , we will consider @xmath183 and @xmath182 for the rest of the proof .
since @xmath182 , then @xmath180 ( see remark [ remark : properties of beta ] and theorem [ theorem : main result of modified weiszfeld algorithm ] ) .
let us prove equation ( [ equation : property of a^k as a mininum 2 ] ) first .
now , by ( [ equation : property of gakzp ] ) we can see that : @xmath248 notice that if @xmath175 , equation ( [ equation : property of a^k as a mininum 2 ] ) holds .
so , let us consider from now on that @xmath176 . by using lemma [ lemma : widetilder equal to 2a(widetildet - x ) ]
we replace @xmath249 and get : @xmath250 extracting common factors and using the definition of @xmath251 when it belongs to @xmath252 we obtain : @xmath253.\end{aligned}\ ] ] by remarks [ remark : properties of beta ] and [ remark : properties of q ] the vectors @xmath254 and @xmath255 are parallel , so : @xmath256.\ ] ] which is equivalent to ( [ equation : property of a^k as a mininum 2 ] ) .
now , let us prove ( [ equation : property of a^k as a mininum 1 ] ) .
if @xmath257 then @xmath258 for all @xmath241 $ ] , thus @xmath259 , and therefore the inequality ( [ equation : property of a^k as a mininum 1 ] ) holds .
so , let us assume that @xmath260 for the rest of the proof . using ( [ equation : property of gakzp ] ) and due to lemma [ lemma : widetilder equal to 2a(widetildet - x ) ] to replace @xmath249 : @xmath261 extracting common factors : @xmath262.\end{aligned}\ ] ] using the expression for @xmath180 we obtain : @xmath263.\end{aligned}\ ] ] if @xmath237 belongs to the segment that joins @xmath15 and @xmath67 we have that @xmath264 and @xmath255 are parallel vectors , then : @xmath265 a(a^k ) \left\| z - a^k \right\| \left\| \widetilde{t}(a^k ) - a^k \right\|.\ ] ] we can write @xmath266 where @xmath267 $ ] . therefore : @xmath265 a(a^k ) \left ( 1 - \lambda \right ) \left\| t(a^k ) - a^k \right\| \left\| \widetilde{t}(a^k ) - a^k \right\|.\ ] ] for all @xmath267 $ ] .
the minimum value of the right - hand side of the last expression happens when @xmath268 , so : @xmath265 a(a^k ) \left ( 1 - \lambda ( a^k ) \right ) \left\| t(a^k ) - a^k \right\| \left\| \widetilde{t}(a^k ) - a^k \right\|.\ ] ] using remark [ remark : properties of q ] we conclude that : @xmath265 a(a^k ) \left\| q(a^k ) - a^k \right\| \left\| \widetilde{t}(a^k ) - a^k \right\|.\ ] ] now we will prove an equivalence that characterizes the solution of ( [ equation : constrained weber problem ] ) in terms of the iteration function @xmath2 . moreover , if @xmath20 is a regular point that is not a vertex , then @xmath20 is a kkt point . from now on
, let us consider that @xmath269 where @xmath270 is a convex function and @xmath271 is an affine function .
[ theorem : equivalence minimum fixed point ] let @xmath19 be defined as in ( [ equation : characterization of omega ] ) and @xmath91 . consider the following propositions : 1 .
@xmath93 is a kkt point .
@xmath93 is the minimizer of the problem ( [ equation : constrained weber problem ] ) .
@xmath272 .
if @xmath44 , @xmath273 and @xmath274 are continuously differentiable , and @xmath93 is a regular point , then ( a ) , ( b ) and ( c ) are equivalent . if @xmath132 for some @xmath68 , then ( b ) implies ( c ) .
let @xmath44 be . since @xmath21 is strictly convex and @xmath19 is convex , the kkt optimality conditions are necessary and sufficient .
therefore , it holds that ( a ) is equivalent to ( b ) .
now we will prove that ( b ) implies ( c ) .
let us suppose that @xmath93 is the minimizer of the problem ( [ equation : constrained weber problem ] ) . if @xmath93 were not a fixed point of the iteration function @xmath2 , we would have that @xmath188 , which means that @xmath218 by theorem [ theorem : descent property ] .
this contradicts the hypothesis . to demonstrate that ( c ) implies ( a ) , we will assume that @xmath93 is a fixed point of @xmath2 , that is , @xmath275 .
since @xmath141 , @xmath93 is the solution of : @xmath276 since @xmath277 and @xmath273 are convex , @xmath274 is affine , and @xmath93 is a regular point , the kkt optimality conditions hold at @xmath93 .
that is , there exist multipliers @xmath278 and @xmath279 such that ( see @xcite ) : @xmath280 multiplying these equations by @xmath281 , using equation ( [ equation : nabla equal to minus widetilder ] ) , lemma [ lemma : widetilder equal to 2a(widetildet - x ) ] and remark [ remark : properties of beta ] , we obtain : @xmath282 where @xmath283 and @xmath284 are multipliers .
therefore , @xmath93 is a kkt point of the problem ( [ equation : constrained weber problem ] ) ( see @xcite ) . now , let us suppose that @xmath132 for some @xmath68 .
as before , if @xmath93 is a minimizer of the problem ( [ equation : constrained weber problem ] ) , then @xmath175 , otherwise @xmath285 , which would be a contradiction . numerical experiments .
[ section : numerical experiments ] the purpose of this section is to discuss the efficiency and robustness of the proposed algorithm versus a solver for nonlinear programming problems . a prototype code of algorithm [ algorithm : proposed algorithm ] was programmed in matlab ( version r2011a ) and executed in a pc running linux os , intel(r ) core(tm ) i7 cpu q720 , 1.60ghz .
we have considered a closed and convex set @xmath286 defined by the set @xmath287 , where @xmath273 is given by : @xmath288 \displaystyle \frac { 4 } { 5 } x + y - \frac { 59 } { 10 } \\[3 mm ] \displaystyle x - \frac { 11 } { 2 } \\[3 mm ] \displaystyle \frac { 3 } { 2 } x - y - \frac { 35 } { 4 } \\[3 mm ] \displaystyle x - y - \frac { 13 } { 2 } \\[3 mm ] \displaystyle - 4 + \frac { 1 } { 8 } ( x - 1 ) + \frac { 1 } { 16 } ( x - 1 ) ^2 + \frac { 1 } { 32 } ( x - 1 ) ^2 ( x - 3 ) - y \\[3 mm ] \displaystyle - \frac { 1 } { 3 } x - y - \frac { 11 } { 3 } \\[3 mm ] \displaystyle - \frac { 2 } { 3 } x - y - \frac { 13 } { 3 } \\[3 mm ] \displaystyle - 4 x + y - 19 \end{array } \right].\ ] ] the feasible set is defined by linear and nonlinear constraints , as it can be seen in figure [ figure : feasible set ] . ,
width=434 ] we have built @xmath289 different experiments where for each one : * the number of vertices was @xmath290 .
* the vertices were normally distributed random vectors , with mean equal to @xmath291 and standard deviation equal to @xmath292 . * the weights
were uniformly distributed random positive numbers between @xmath291 and @xmath292 .
* tolerance was set to @xmath293 .
on one hand , each experiment was solved using algorithm [ algorithm : proposed algorithm ] and , on the other hand , it was considered as a nonlinear programming problem and solved using function @xmath294 ( see @xcite and references therein ) .
since the weber function ( [ equation : weber function ] ) is not differentiable at the vertices , nonlinear programming solvers may fail .
let @xmath295 be the solution of ( [ equation : constrained weber problem ] ) obtained by @xmath294 in experiment @xmath296 , and @xmath297 .
analogously , let @xmath298 be the solution of ( [ equation : constrained weber problem ] ) obtained by algorithm [ algorithm : proposed algorithm ] in experiment @xmath296 , and @xmath299 .
figure [ figure : functionalvalues ] shows the difference between the arrays @xmath300 and @xmath301 . both methods finished succesfully in all cases , however , algorithm [ algorithm : proposed algorithm ] found equal or better results for all experiments .
for example , the difference @xmath302 was greater than @xmath303 in @xmath304 experiments
( the maximum difference ocurred in experiment @xmath305 ) . and @xmath294.,width=434 ]
feasibility of the solutions @xmath298 can be checked computing @xmath306 .
results can be seen in figure [ figure : feasibility ] obtained by algorithm [ algorithm : proposed algorithm].,width=434 ]
this paper proposes a weiszfeld - like algorithm for solving the weber problem constrained to a closed and convex set , and it is well defined even when an iterate is a vertex .
the algorithm consists of two stages : first , iterate using the fixed point modified weiszfeld iteration ( [ equation : iteration function of the modified algorithm with beta ] ) , and second , either project onto the set @xmath19 when the iterate is different from the vertices , or , if the iterate is a vertex @xmath15 , take the point belonging to the line that joins @xmath67 with @xmath15 as defined in ( [ equation : iteration function of the proposed algorithm ] ) .
it is proved that the constrained problem ( [ equation : constrained weber problem ] ) has a unique solution .
besides that , the definition of the iteration function @xmath2 allows us to demonstrate that the proposed algorithm produces a sequence @xmath217 of feasible iterates .
moreover , the sequence @xmath231 is not increasing , and when @xmath307 , the sequence decreases at the next iterate .
it can be seen that if a point @xmath20 is the solution of the problem ( [ equation : constrained weber problem ] ) then @xmath20 is a fixed point of the iteration function @xmath2 .
even more , if @xmath20 is different from the vertices , the fact of being @xmath20 a fixed point of @xmath2 is equivalent to the fact that @xmath20 satisfies the kkt optimality conditions , and equivalent to the fact that @xmath20 is the solution of the problem ( [ equation : constrained weber problem ] ) .
these properties allows us to connect the proposed algorithm with the minimization problem .
numerical experiments showed that the proposed algorithm found equal or better solutions than a well - known standard solver , in a practical example with 1000 random choices of vertices and weights . that is due to the fact that the proposed algorithm does not use of the existence of derivatives at the vertices , because the weber function is not differentiable at the vertices .
44 natexlab#1#1[2]#2 , , , , , .
, , , ( ) . , , ( ) .
, , , , ( ) .
, , , ( ) . , , ( ) .
, , , ( ) . , , , , ( ) .
, , , , ( ) . , , , ( ) .
, , , , ( ) . , , , , , ( ) . , , , , , . .
, , , ( ) . , , , ( ) .
, , , , ( ) .
, , , , ( ) . , , , .
, , , ( ) . , , , ( ) .
, , , ( ) . ,
, springer series in operations research , , , .
, , , ( ) . . , , in : , , , , pp . . , , ( ) . , , , in : , volume of _ _ , , , , pp . .
, , , international series in operations research & management science , 116 , , , edition , . , , .
, , , ( ) . , , ( ) .
, , , ( ) . , , , ( ) . , , , , .
, , , , volume , , .
, , , ( ) . , , , ( ) .
, , , , . . , , ( )
, , ( ) . , , ( ) | the weber problem consists of finding a point in @xmath0 that minimizes the weighted sum of distances from @xmath1 points in @xmath0 that are not collinear .
an application that motivated this problem is the optimal location of facilities in the 2-dimensional case .
a classical method to solve the weber problem , proposed by weiszfeld in 1937 , is based on a fixed point iteration . in this work
a weber problem constrained to a closed and convex set is considered .
a weiszfeld - like algorithm , well defined even when an iterate is a vertex , is presented .
the iteration function @xmath2 that defines the proposed algorithm , is based mainly on an orthogonal projection over the feasible set , combined with the iteration function of a modified weiszfeld algorithm presented by vardi and zhang in 2001 .
it can be seen that the proposed algorithm generates a sequence of feasible iterates that have descent properties . under certain hypotheses ,
the limit of this sequence satisfies the kkt optimality conditions , is a fixed point of the iteration function that defines the algorithm , and is the solution of the constrained minimization problem .
numerical experiments confirmed the theoretical results .
location , weber problem , weiszfeld algorithm , fixed point iteration 90b85 , 90c25 , 90c30 |
in order to calculate the derivative @xmath135 one should calculate the corresponding derivatives of the function @xmath136 .
so one obtains @xmath137 and @xmath138 @xmath139 we took into account that for the equilibrium value of @xmath32 , the condition @xmath140 holds , from which follows that @xmath141 .
similarly @xmath142 is also equal to zero .
so , one obtains : @xmath143 for integration over @xmath28 in eq .
( 6 ) , the following relations are also used : @xmath144 where we used the right hand side of eq .
( 3 ) for @xmath145 at @xmath33 .
widest - label y. ishibashi , _ incommensurate phases in dielectrics _ , edited by r. blinc and a.p .
levanyuk , modern problems in condensed matter sciences , vol.*14.2 * , north - holland ( 1986 ) .
a.m. arutyunyan , b. brezina , s.kh .
esayan , v.v .
lemanov , sov .
, solid state , * 24 * , 814 , ( 1982 ) . g. dolino , j.p .
bachheimer , phys.stat.sol .
( a ) , * 41 * , 673 , ( 1977 ) .
alexandrov , a.n .
vtyurin , v.f .
shabanov , sov .
phys . , pisma jetp , * 28 * , 153 , ( 1978 ) .
golovko , a.p .
levanyuk , sov .
phys . , jetp , * 50 * , 780 , ( 1979 ) .
m. iizumi , j.d .
axe , g. shirane and k. shimaoka , phys . rev . * b 15 * , 4392 ( 1977 ) .
sannikov , a.p .
levanyuk , sov .
phys . , solid state , * 4 * , 1005 , ( 1978 ) .
i.e. dzyaloshinski , sov .
jetp , * 47 * , 336 , 992 , ( 1964 ) .
golovko , sov .
, solid state , * 22 * , 2960 , ( 1980 ) . | we show that a ferroelectric phase transition takes place in the incommensurate phase of the @xmath0 crystal .
the ferroelectric character of the ic phase explains the second - harmonic generation observed in the corresponding temperature range .
the crystals of the @xmath1 family ( about 20 crystals ) @xcite undergo two successive phase transitions on cooling : an incommensurate ( ic ) transition at temperature @xmath2 with a modulation vector near the @xmath3 point in the brillouin zone , and upon further cooling at temperature @xmath4 a commensurate lock - in transition to the triple - period ferroelectric phase with the modulation vector @xmath3 .
the most striking peculiarity of the @xmath0 crystal , which belongs to the @xmath1 family , is the second - harmonic generation ( shg ) in the ic phase,@xcite of the same intensity magnitude as that in the @xmath3- phase .
low temperature @xmath3- phases of the @xmath1 type crystals are improper ferroelectrics , and therefore the shg is a normal manifestation of this feature .
however , so far there was neither pointed out a plausible reason for the shg in the ic phase of @xmath0 , nor explained why it takes place not in all the crystals of the @xmath1 family .
it is worth noting that the shg was observed also in the ic phase of quartz by dolino and bachheimer ( 1977)@xcite and in the ic phase of the ammonium fluoroberyllate @xmath5 by alexandrov et al .
( 1978)@xcite , and both observations also can be explained using the theory developed below .
the attempt to explain the shg in the ic phase of @xmath5 was carried out by golovko and levanyuk@xcite based on the spatial dispersion of the dielectric constant .
however , the expected effect appeared to be small , and besides , it is not explained why in such a case the shg is not observed in all the crystals with the ic phases . in the present paper
we show that within the ic phase of the @xmath0 crystal , necessarily a transition to the ferroelectric ic phase takes place .
we analyze the exact solution in frame of the landau theory .
the physical reason for such a ferroelectric transition is the coupling between the crystal s polar symmetry vibrational mode @xmath6 and some displacements of the ic domain walls .
we demonstrate that relative displacements of the domain walls in the domainlike ic structure induce polarization along the @xmath7 direction , and can be viewed as an additional phonon branch of a polar symmetry .
this mode is not explicitly introduced in the equations below .
it is a component of the ic phase @xmath8 function , and is taken into account implicitly , in the integral form .
the coupling of the two modes renormalizes the frequency of the lower mode down to zero within the ic range , which induces a ferroelectric transition .
structure of the ic modulation of @xmath0 was defined in the neutron diffraction by iizumi _
et al_.@xcite the thermodynamic potential describing the ic phase of @xmath9 was detailed studied by sannikov and levanyuk.@xcite for the two - component order parameter with components @xmath10 and @xmath11 it can be written as : @xmath12 where @xmath6 is polarization vector in the @xmath7 direction , @xmath13 is electric field in the @xmath7 direction . in case of @xmath14
this potential would describe a phase transition from the symmetric high - temperature phase to the triple - period commensurate phase at the temperature @xmath15 .
for @xmath16 the equilibrium values of the modulation phase @xmath17 are @xmath18 ( otherwise @xmath19 ) . however , in virtue of the lifshitz invariant ( @xmath20 ) the phase transition takes place from the high - temperature phase to the ic phase at @xmath21 with @xmath22 and @xmath23 .
substituting @xmath24 in potential ( 1 ) one obtains for the @xmath25- and @xmath6- dependent part of potential ( 1 ) the following expression : @xmath26 minimization of this expression with respect to @xmath25 gives the equation : @xmath27 after single integration over @xmath28 and taking for the integration constant the form @xmath29 one obtains : @xmath30 where @xmath31 . the integration constant @xmath32 , as well as the polarization vector @xmath6 ,
can be defined via minimization of the potential with respect to these parameters after substituting the solution of eq .
( 3 ) in the potential ( 2 ) .
prior to considering the ferroelectric transition in this system let us discuss the solutions of eq .
( 3 ) for the case of @xmath33 .
the solution to this kind of equations for the ic structures was analyzed in the pioneer work by dzyaloshinski @xcite ( it was also detailed studied by golovko@xcite ) , and for eq . ( 3 ) , it has the form : @xmath34 where @xmath35 is the _ elliptic amplitude function _ and @xmath36 is the ic vector at the temperature of the ic transition .
( in the present paper the notations of the wolfram research , `` mathematica '' program for the elliptic functions are used . )
substituting solution ( 4 ) in the potential ( 2 ) and minimizing over the parameter @xmath32 one obtains : @xmath37 is the _ complete elliptic integral of the second kind _ , which has the asymptotic + @xmath38 eq .
( 5a ) can be numerically solved with respect to @xmath32 for given coefficients @xmath39 , and one obtains that the parameter @xmath32 monotonously decreases from @xmath40 at @xmath41 ( i.e. , at the ic transition @xmath2 ) down to @xmath42 at the continuous lock - in transition @xmath43 . as it follows from eq .
( 5a ) , at the lock - in transition point the ic amplitude is tending to @xmath44 which gives that @xmath45 , if the remaining coefficients are of the same order of magnitude . in fig .
1(a ) the solutions given by eq . ( 4 ) for @xmath46 , @xmath47 and @xmath33 for the three values of the ic amplitude are plotted . at the ic transition @xmath2 ( @xmath48 )
the corresponding graph must look like a straight line with slope @xmath36 ( @xmath49 , i.e. , sinusoidal modulation ) , and on cooling it acquires the domain - like features . due to the symmetry , each point @xmath28 on the curves of fig .
1(a ) is associated with a nonzero polarization , which is spatially distributed as @xmath50 with zero - integral effect .
meanwhile , each domain there has a nonzero - integral polarization . in the areas of any of two neighboring domains @xmath51 and @xmath52
respectively , giving rise to the @xmath6-s of opposite signs . in fig .
1(b ) the transformation of the curves of fig .
1(a ) under the applied electric field @xmath53 is shown .
the field @xmath53 is taken into account via replacing in eq .
( 3 ) of the polarization vector @xmath6 by @xmath54 .
an identical transformation of the domains takes place in the presence of nonzero polarization @xmath6 in crystal ( when @xmath55 ) .
corresponding solutions were found both analytically and numerically ( using the `` mathematica '' software ) with the same result .
note that the integral polarization vector contributed by the nonequal domains in fig .
1(b ) is not zero .
now we show , that the above described transformation of the domain structure from that shown in fig .
1(a ) to fig .
1(b ) takes place spontaneously , i.e. , without applying an external electric field . in other words
, there takes place a ferroelectric phase transition inside the ic phase .
let us calculate the dielectric susceptibility of the crystal in the ic phase : @xmath56 differentiation of potential ( 2 ) gives : @xmath57 @xmath58 is the ic period , given by eq .
( 10 ) below . in order to calculate the derivative @xmath59 , we introduce notation @xmath60 integration of eq .
( 3 ) gives in this notation , @xmath61 , from which follows that @xmath62 .
the latter allows one to calculate the derivative @xmath63 via calculation of the corresponding derivatives of the function @xmath64 in the integral form ( see the appendix ) . after substituting of the calculated derivative in eq .
( 6 ) , one obtains : @xmath65 where @xmath66 is a _
hypergeometric function_. this hypergeometric function has the asymptotic @xmath67 which well fits the function in all the range @xmath68 , i.e. , not only at @xmath69 . as one can see from eq .
( 10 ) , the ic period has an asymptotic : @xmath69 , @xmath70 ( as shown below , on approaching the lock - in temperature @xmath43 the coefficient @xmath32 is tending to zero as @xmath71 ) .
so , for the susceptibility @xmath72 one obtains : @xmath73 and the stability coefficient @xmath74 necessarily vanishes ( since the @xmath32 -dependent term in eq .
( 8) is negative and diverging with @xmath69 ) , inducing ferroelectric phase transition at the temperature @xmath75 .
a loss of stability necessarily takes place even for the case of large @xmath76 ( i.e. , even for a not soft polar mode of the high temperature phase ) , and it is expected in the medium temperature range of the ic phase ( @xmath77 )
. however , some alternative situations are possible near the loss of stability @xmath75 . as it follows from eq .
( 3 ) , the polarization @xmath6 which appears below the transition should be sufficiently small : @xmath78 otherwise , the expression under the root in eq .
( 3 ) takes negative values for some @xmath17 , what means that the crystal does not fall within the ic phase , but within the commensurate lock - in phase instead , skipping the ferroelectric ic phase . in terms of fig .
1(b ) , it is equivalent to the infinite increasing of the one domain s size , which covers all of the crystal area . the right hand side of eq .
( 9 ) is very small , since @xmath79 in eq .
( 9 ) is a small parameter , and @xmath32 tends to zero in the low - temperature range of the ic phase ( see also below ) . in the case of a first - order ferroelectric phase transition giving rise to a large @xmath6 ( in landau theory ,
it means a negative fourth - order term in the potential s @xmath6- expansion ) , inequality ( 9 ) does not hold , and the transition transfers the crystal directly to the lock - in phase , skipping the ferroelectric ic phase .
even for some order - disorder transitions , which are very close to the second - order type ( the polarization vector @xmath6 below the order - disorder transitions is larger , due to the small curie constant compared to that for the displacement type ) the expected ferroelectric transition is more likely to turn into the transition to the lock - in phase , skipping the ferroelectric ic phase .
note that among the crystals of the @xmath1 family only @xmath0 is of a displacement type , and therefore for this crystal the ferroelectric phase transition ( from the ic to the ferroelectric ic phase ) can be considered as a most likely .
in other words , the ic structure of @xmath0 becomes ferroelectric with the modulation , spatially distributed like that in fig .
1(b ) .
one should expect that the ferroelectric transition takes place in the beginning of the domainlike modulation formation , i.e. , at temperatures near the curve i in fig .
1(a ) , when the @xmath32 coefficient falls below unity .
as it follows from eqs.(5a , 5b ) , for estimates of inequality ( 9 ) , the asymptotic of the parameter @xmath69 can be used : @xmath80 near the lock - in point we presented @xmath81 , where @xmath82 is some constant .
so , eq . ( 9 ) can be presented as @xmath83 , when @xmath84 ( we neglected a logarithmically diverging term @xmath85 , which does not change order of magnitude of the estimated values , since , as discussed below , @xmath86 can not be sufficiently small for it ) . for obtaining of the inharmonic fourth order term in the @xmath6 expansion of the potential one should calculate the derivative @xmath87
we shall introduce here only the result of this calculation .
in addition to the @xmath88 term in eq .
( 2 ) , one obtains a negative , strongly diverging as @xmath89 with @xmath69 term .
if the sum of the inharmonic @xmath90 terms becomes negative earlier than @xmath75 , then takes place a first - order transition .
the latter makes more probable the first order transition , even if the coefficient @xmath91 is not small .
now we shall discuss some features of the ic phase for @xmath33 case in order to compare it with behavior of the ferroelectric ic phase .
to study the temperature evolving of the ferroelectric ic structure upon cooling one should minimize potential ( 2 ) also over the ic amplitude @xmath92 .
however , we imply in the present paper only increasing of the ic amplitude @xmath92 upon cooling , and do not discuss the character of the lock - in transition .
first we discuss some structure peculiarities of the non - ferroelectric ic phase of @xmath0 .
the length of each plateau in fig .
1(a ) is six times shorter than the ic wavelength @xmath93 , where @xmath94 is the ic wavevector [ @xmath94 is equal to the average slope of the given curve in fig .
1(a ) ] . from this geometrical fact
it follows , that the ic modulation argument @xmath31 can be presented as @xmath95 , where @xmath96 is a periodic function with period @xmath97 . as it follows from eq .
( 4 ) , roughly @xmath96 can be approximated by @xmath98 .
the ic satellite reflections observed in the x - ray diffraction are induced by the fourier - components of the modulation function @xmath99 $ ] .
the ic satellites observed for such a modulation should appear in the points @xmath100 ; @xmath101 near the main bragg- reflections @xmath102 .
in other words , the manifestation of the domain - like ic structure in @xmath0 in diffraction should be only the simultaneous observation of the ic satellites of about the same intensity ( asymmetrically disposed with respect to the @xmath103 point ) in the points @xmath104 and @xmath105 near @xmath106 .
intensive ic satellites in the points @xmath107 , which are observed in diffraction , are not contributed by the domain features of the modulation , and can not be viewed as a manifestation of a domainlike ic structure .
the satellites @xmath104 and @xmath105 were not observed for @xmath0 , which proves that the domainlike structure is not well developed in @xmath0 , and the ferroelectric phase exists only in the medium range of the ic phase .
so , as it follows from the exact solution of eq .
( 3 ) , any crystal of the @xmath1 type with well developed ic domain structure ( i.e. , sufficiently small @xmath32 ) should be ferroelectric .
the ferroelectric transition does not change the translation symmetry of the ic phase , and therefore the satellites disposition for the ferroelectric ic phase should not essentially differ from that for the @xmath33 case .
we also discuss the character of increasing of the ic modulation period on cooling , which is logarithmically diverging , as it follows from eq .
( 4 ) .
the ic period for the modulation given by eq .
( 4 ) is @xmath108 where @xmath109 is the _ complete elliptic integral of the first kind_. with @xmath69 the function @xmath110 has the asymptotic @xmath111 and @xmath112 which is logarithmically diverging .
the temperature @xmath113 can approach the lock - in temperature @xmath43 not closer than the temperature fluctuation @xmath114 , i.e. , @xmath115 where @xmath116 is the crystal volume and @xmath117 is the heat capacity .
so , for any crystal of a finite size ( @xmath118 ) , the period @xmath119 can increase only about 10 times compared to @xmath120 , in virtue of the logarithmical character of the divergence ( it is also confirmed by our numerical simulations ) . besides , in the real experiment the temperature @xmath113 can not approach the lock - in temperature @xmath43 even closer than the instrumental temperature resolution @xmath121 , which does not allow to observe any diverging ic period , and the lock - in transition always should be observed as a discontinuous phase transition with jump in the wavevector @xmath94 down to @xmath122 . for calculation of the ferroelectric ic modulation s period [ the @xmath28-length of the six neighboring domains in fig .
1(b ) ] , one should integrate eq .
( 3 ) from @xmath123 through @xmath124 , which gives : @xmath125 @xmath126\;\,\end{aligned}\ ] ] where @xmath127 is the _
elliptic integral _ , and @xmath128 , i.e. , the condition @xmath129 coincides with eq . ( 9 ) , and therefore it always holds in the ic phase
. the parameter @xmath130 decreases down to zero with increasing of the polarization vector @xmath6 on cooling .
integral ( 11 ) was calculated exactly , and after being simplified to the form of eq .
( 12 ) for @xmath131 .
the first term in eq .
( 12 ) , which gives the size of the larger domains in fig .
1(b ) , is diverging with @xmath132 as @xmath133 .
meanwhile , the size of the smaller domains is given by the second term in eq .
( 12 ) , and it is always finite , though it also increases with cooling .
summarizing , we showed that in the ic phase of the @xmath9 crystal takes place ferroelectric phase transition , which explains the shg in the ic phase of this crystal .
formation of the ferroelectric ic phase is most likely near the displacement - type phase transitions , though it can occur also in those of the order - disorder type .
the ic period increases on cooling towards the lock - in temperature @xmath43 as @xmath134 , though the size of the domains , polarized opposite to the prevailing polarization , remains finite . |
the slac and kek b factories are asymmetric e@xmath2e@xmath3 colliders which run at centre - of - mass energies at , and close to , the @xmath4 resonance . at these energies , around 10.58 gev
, the cross section for tau - pair production , at 0.89 nb , is close to the 1.05 nb cross section for b@xmath5 production .
these machines are therefore also tau factories , and the experiments babar and belle have between them so far recorded over @xmath6 tau - pair events . in comparison , the four lep experiments each recorded only about @xmath7 tau - pair events , mainly running on the z resonance , while cleo - iii at cesr took about @xmath8 tau - pair events .
however , it should be noted that experimental conditions for tau physics at lep were particularly clean , resulting in relatively small systematic errors . for channels with large branching fractions , where systematic errors dominate
, the lep results will remain competitive for some time .
measuring static properties of the tau , such as mass , lifetime and electroweak couplings , allow tests of lepton universality and cpt invariance .
searches for lepton - flavour violation ( lfv ) in tau decays are particularly topical . in the standard model ( sm ) ,
minimally augmented with massive neutrinos and neutrino oscillations , lfv decays of the tau are allowed , but at an unobservably low level @xcite . however , various extensions to the sm predict lfv rates accessible to the b factories , with branching fractions of order @xmath9 and higher @xcite . belle has used 3-prong tau decays from the process @xmath10 , in a sample of 414 fb@xmath11 of data , to measure the tau mass and to put an upper limit on the relative mass difference of the two charge states of the tau @xcite .
they construct the pseudomass @xcite , a kinematic quantity which , in the absence of initial and final state radiation , displays a sharp edge at the tau mass . by fitting the edge ( see fig . [
fig : massdiff ] ) , they obtain a mass @xmath12 mev/@xmath13 and a relative mass difference @xmath14 at the 90% confidence level ( cl ) . the measurement of the relative mass difference represents a significant improvement over the previous limit on this quantity .
babar has made a preliminary measurement of the tau lifetime and the relative lifetime difference of the two charge states , using an 80 fb@xmath11 data sample .
they measure a lifetime of @xmath15 fs and a relative difference , @xmath16 , of @xmath17% ( the systematic error has still to be evaluated ) .
all of the measurements to date are consistent with cpt invariance and with lepton universality .
belle has recently published results @xcite@xcite on searches for the lfv decays of the tau to a lepton plus two pseudoscalar mesons ( @xmath18 or k@xmath19 ) , including the cases where the mesons are decay products of a k@xmath20 , @xmath21 , k@xmath22 or @xmath23 .
the most promising channel among these may be @xmath24 , where the minimal supersymmetric standard model ( mssm ) with small @xmath25 could give a branching fraction as large as @xmath26 .
these decay modes may also be sensitive to new physics models with heavy dirac neutrinos and to models with dimension - six effective fermionic operators that induce @xmath27 mixing . with a data sample of 158 fb@xmath11 , belle set 90% cl upper limits that vary from @xmath28 for @xmath29 to @xmath30 for @xmath31 . using 281 fb@xmath11
they obtain limits of @xmath32 for e@xmath3k@xmath20 and @xmath33 for @xmath34k@xmath20 belle has also made a search @xcite for the decay @xmath35 putting an upper limit at @xmath36 at 90% cl .
the belle results improve on previous lep and cdf exclusion limits in the @xmath25 versus @xmath37 plane .
babar has recently used a data sample of 233 fb@xmath11 to search for the decay @xmath38 .
in mssm with the seesaw mechanism , the branching fraction for this process is directly related to the value of @xmath25 and the mass of lightest chargino .
babar excludes this process @xcite with an upper limit on the branching fraction of @xmath39 at 90% cl .
the non - strange hadronic tau decays are dominated by vector and axial hadronic states , with the vector part of the so - called spectral function being particularly important as input to the calculation of the muon magnetic moment anomaly , @xmath40 .
the largest uncertainty in this calculation comes from the lowest order hadronic correction to the photon propagator , whose value is inferred from measurements of the cross section for the process e@xmath2e@xmath41 at low energies .
almost three - quarters of this correction is due to the @xmath42 final state , dominated by the @xmath43 .
the e@xmath2e@xmath3 annihilation data can be related , by isospin invariance ( conserved vector current ) , to the decay @xmath44 for which much higher statistics measurements are available . in principle , therefore , the tau decay data could provide more precise input for @xmath45 ; however there remains a discrepancy between the e@xmath2e@xmath3 and @xmath0 data which currently is not understood .
a recent review can be found in ref . .
studies of the strange spectral functions in hadronic tau decays are the route to the world - best measurements of the ckm matrix element @xmath1 and the mass of the strange quark .
all measurements to date of the strange tau decays are limited by statistics , and so the b - factory data have a very important role to play .
already , the babar and belle preliminary measurements of the @xmath46 modes are more precise than the previous world averages .
belle has used a sample of 72 fb@xmath11 to make a first study of their @xmath48 mass spectrum from @xmath0 decay @xcite .
they fit to the phenomenological model of gounaris and sakurai @xcite which includes a set of three interfering @xmath21 breit - wigner amplitudes , and they find the need for a @xmath49 state in addition to the two well - known @xmath50 and @xmath51 resonances ( see fig .
[ fig : belle2pi ] ) .
belle reports a branching fraction bf@xmath52 .
the result for @xmath53 is in agreement with previous tau measurements from aleph @xcite and cleo @xcite and in disagreement with the numbers deduced from the e@xmath2e@xmath3 data @xcite . in a recent babar analysis @xcite ,
a data sample of 232 fb@xmath11 has been used to study the process @xmath55 .
the branching fraction is measured as @xmath56 , representing a significant improvement on the previous world average .
the mass spectrum of the hadronic system , with all tracks taken as pions , is shown in fig .
[ fig:5pifigure1 ] , where it is compared to the output of the tauola monte carlo model @xcite .
[ fig:5pifigure2 ] shows the inclusive @xmath42 mass spectrum , indicating a significant contribution from @xmath43 .
the @xmath57 spectrum ( fig . [ fig:5pifigure3 ] ) shows a signal from the f@xmath58 meson , from which the branching fraction is measured to be bf@xmath59 .
these data will be used to improve the implementation of these channels in the tauola monte carlo .
babar has looked for evidence for decays of the @xmath0 to seven or more pions @xcite , which would be highly suppressed by phase space . in principle
, the rate of such decays would be sensitive to the mass of the @xmath0 neutrino , but recent results on the neutrino mass differences has somewhat reduced the interest in this , since indications are that the neutrino masses are too small to be probed by high - multiplicity @xmath0 decays . using 232 fb@xmath11 of data
, babar has placed the following upper limits on branching fractions at 90% cl : @xmath62 for @xmath63 ; @xmath64 for @xmath65 ; @xmath66 for @xmath67 ; and @xmath68 for @xmath69 .
babar has made a preliminary measurement of @xmath70 with 124 fb@xmath11 of data , obtaining a branching fraction of @xmath71 .
for the complementary mode , @xmath72 , belle uses 351 fb@xmath11 to measure a branching fraction of @xmath73 .
the two measurements are consistent . in preliminary fits to their k@xmath74 mass spectrum , belle finds some evidence for the presence of the k@xmath75 together with a scalar state ( the so - called @xmath76 ) .
each of babar and belle expects to accumulate samples of close to , or greater than , 1000 fb@xmath11 by the end of 2008 .
thus there is the potential for much more @xmath0 physics in the coming years .
limits on lfv could be reduced down to @xmath77 for some channels , with even the possibility of the emergence of evidence for new physics . in semihadronic decays , measurements of the strange spectral functions
should provide an improved measurement of the ckm matrix element @xmath1 and perhaps also of the mass of the strange quark .
measurements of the non - strange channels will include the important vector spectral functions , which should continue to aid understanding of the non - perturbative , hadronic contribution to the anomalous magnetic moment of the muon .
many thanks go to the babar and belle collaborations , whose work i have reported .
i would also thank the organizers of charm 2006 for what was a most enjoyable and productive workshop . | some recent @xmath0-physics results are presented from the babar and belle experiments at the slac and kek b factories , which produce copious numbers of @xmath0-lepton pairs .
measurements of the tau mass and lifetime allow to test lepton universality and cpt invariance , while searches for lepton - flavour violation in tau decays are powerful ways to look for physics beyond the standard model .
in semihadronic , non - strange tau decays , the vector hadronic final state is particularly important in helping determine the hadronic corrections to the anomalous magnetic moment of the muon , while studies of strange final states are the best available ways to measure the ckm matrix element @xmath1 and the mass of the strange quark .
presented at charm 2006 , international workshop on tau - charm physics , june 05 - 07 2006 , beijing , china |
we study bose - einstein correlations ( bec ) in hadronic @xmath4 decay using data collected by the l3 detector at an e@xmath5e@xmath6 center - of - mass energy of @xmath7 .
approximately 36 million like - sign pairs of well - measured charged tracks from about 0.8 million hadronic z decays are used @xcite .
events are classified as two- or three - jet events using calorimeter clusters with the durham jet algorithm with a jet resolution parameter @xmath8 , yielding about 0.5 million two - jet events .
there are few events with more than three jets , and they are included in the three - jet sample . to determine the thrust axis of the event we also use calorimeter clusters .
the two - particle correlation function of two particles with four - momenta @xmath9 and @xmath10 is given by the ratio of the two - particle number density , @xmath11 , to the product of the two single - particle densities , @xmath12 .
since we are interested only in bec , @xmath12 is replaced by @xmath13 , the two - particle density that would occur in the absence of bec , yielding @xmath14 .
an event mixing technique is used to construct @xmath15 .
since the mass of the identical particles of the pair is fixed , @xmath16 is defined in six - dimensional momentum space , which is often reduced to a single dimension , @xmath17 .
but there is no reason to expect the hadron source to be spherically symmetric in jet fragmentation .
in fact , the source is found to be elongated along the jet axis @xcite , but only by about 25% , which suggests that a parametrization in terms of the single variable @xmath18 , may be a good approximation .
this is also suggested by studies of various decompositions of @xmath18 @xcite .
with a few assumptions , @xmath16 is related to the fourier transform , @xmath19 , of the ( configuration space ) density distribution of the source , @xmath20 : @xmath21 \left(1 + \delta q \right ) \;.\ ] ] the parameter @xmath22 and the @xmath23 term are introduced to parametrize possible long - range correlations not adequately accounted for in @xmath15 , and @xmath24 to account for several factors , such as lack of complete incoherence of particle production and presence of long - lived resonance decays .
model - independent ways to study deviations from the gaussian are to expand @xmath19 about a gaussian using the edgeworth expansion or to replace @xmath25 by a symmetric lvy distribution . both the symmetric lvy ( fig .
1a ) and the edgeworth parametrizations do a fair job @xcite of describing the region @xmath26 , but fail at higher @xmath18 , particularly the region 0.61.5where @xmath16 dips below unity , indicative of an anti - correlation .
this is clearly seen in fig . 1a . for two - jet events .
the curves correspond to fits of the parametrizations : ( a ) the symmetric lvy ; ( b ) the simplified @xmath0-model , eq.([eq : asymlevr2 ] ) ; and ( c ) the full @xmath0-model , eq.([eq : levyr2a ] ) . also plotted
is @xmath27 , the difference between the fit and the data .
the dashed line represents the long - range part of the fit , _
i.e. _ , @xmath28 .
the dotted line in ( a ) represents a linear fit in the region @xmath29 .
[ fig : a_levy_2jet ] , title="fig : " ] for two - jet events .
the curves correspond to fits of the parametrizations : ( a ) the symmetric lvy ; ( b ) the simplified @xmath0-model , eq.([eq : asymlevr2 ] ) ; and ( c ) the full @xmath0-model , eq.([eq : levyr2a ] ) .
also plotted is @xmath27 , the difference between the fit and the data .
the dashed line represents the long - range part of the fit , _
i.e. _ , @xmath28 . the dotted line in ( a ) represents a linear fit in the region @xmath29 .
[ fig : a_levy_2jet ] , title="fig : " ] for two - jet events .
the curves correspond to fits of the parametrizations : ( a ) the symmetric lvy ; ( b ) the simplified @xmath0-model , eq.([eq : asymlevr2 ] ) ; and ( c ) the full @xmath0-model , eq.([eq : levyr2a ] ) .
also plotted is @xmath27 , the difference between the fit and the data .
the dashed line represents the long - range part of the fit , _
i.e. _ , @xmath28 . the dotted line in ( a ) represents a linear fit in the region @xmath29 . [ fig : a_levy_2jet ] , title="fig : " ] we have seen that bec depend , at least approximately , only on @xmath18 , not on its components separately and that @xmath16 in the region 0.61.5 dips below its values at higher @xmath18 . a model which predicts such @xmath18-dependence as well as a dependence on the transverse mass , @xmath30 , which has been previously observed @xcite , is the @xmath0-model @xcite , in which it is assumed that in the overall center - of - mass system the average production point @xmath31 , of particles with a given four - momentum @xmath32 is given by @xmath33
. in the case of two - jet events ,
@xmath34 , and @xmath35 is the longitudinal proper time ; for the case of three - jet events the relation is more complicated .
the second assumption is that the distribution of @xmath36 about its average is narrower than the proper - time distribution , @xmath37
. then @xmath16 is found @xcite to depend only on @xmath18 , the values of @xmath38 of the two pions , and the fourier transform of @xmath37 . since there is no particle production before the onset of the collision
, @xmath37 should be a one - sided distribution .
we choose a one - sided lvy distribution , which is characterized by three parameters : the index of stability @xmath39 , the proper time of the start of particle emission @xmath40 , and @xmath41 , which is a measure of the width of @xmath37
. then @xcite @xmath42 \right . \\ & \left .
\quad \cdot \exp \left[-\left(\frac{\delta\tau { q^2}}{2}\right)^{\!\alpha}\frac{a_1^\alpha+a_2^\alpha}{2 } \right ] \right\ } \left(1+\epsilon q\right ) \ ; . \end{split}\ ] ] note that the cosine factor generates oscillations corresponding to alternating correlated and anti - correlated regions , a feature clearly seen in the data ( fig . 1 ) .
note also that since @xmath34 for two - jet events , the @xmath0-model predicts a decreasing effective source size with increasing @xmath43 . before proceeding to fits of eq.([eq : levyr2a ] ) , we first consider a simplification of the equation obtained by assuming ( a ) that particle production starts immediately , _
i.e. _ , @xmath44 , and ( b ) an average @xmath38-dependence , which is implemented by introducing an effective radius , @xmath45 , defined by @xmath46 this results in @xmath47 ( 1 + \epsilon q ) \;,\ ] ] where @xmath48 is related to @xmath45 by @xmath49 fits of eq.([eq : asymlevr2 ] ) are first performed with @xmath48 as a free parameter .
the fits for both two- and three - jet events have acceptable confidence levels ( cl ) , and describe well the dip in the 0.61.5 region , as well as the peak at low values of @xmath18 .
the estimates of some fit parameters are rather highly correlated .
for example , for two - jet events the estimated correlation coefficients from the fit for @xmath39 , @xmath45 and @xmath50 are @xmath51 , @xmath52 , and @xmath53 . taking the correlations into account ,
the fit parameters satisfy eq.([eq : asymlevrar ] ) , the difference between the left- and right - hand sides of the equation being less than 1 standard deviation for two - jet events and about 1.5 standard deviations for three - jet events .
no significant long - range correlation is observed : @xmath54 is zero within 1 standard deviation , and fits with @xmath54 fixed to zero find the same values , within 1 standard deviation , of the parameters as the fits with @xmath54 free .
thus the method to remove non - bose - einstein correlations in the reference sample is apparently successful .
fits are also performed imposing eq.([eq : asymlevrar ] ) .
the result for two - jet events is shown in fig .
1b . for two - jet events ,
the values of the parameters are comparable to those with @xmath50 free . for three - jet events , the imposition of eq.([eq : asymlevrar ] ) results in values of @xmath39 and @xmath45 closer to those for two - jet events , but the @xmath55 is noticeably worse , though acceptable , than with @xmath50 free . also , the values of @xmath54 differ somewhat from zero , which could indicate a slight deficiency in the description of bec similar to , but on a much smaller scale , the edgeworth and symmetric lvy fits . for two - jet events , @xmath34 , while for three - jet events the situation is more complicated .
we therefore limit fits of eq.([eq : levyr2a ] ) to the two - jet data . for each bin in @xmath18
the average values of @xmath56 and @xmath57 are calculated , where @xmath56 and @xmath57 are the transverse masses of the two particles making up a pair , requiring @xmath58 .
using these averages , eq.([eq : levyr2a ] ) is fit to @xmath59 .
the fit yields @xmath60 fm , and we have re - fit with @xmath40 fixed to zero .
the results are shown in fig .
the parameters @xmath39 , @xmath41 and @xmath24 are highly correlated with estimated correlaton coefficients @xmath61 , @xmath62 , and @xmath63 .
since the @xmath0-model describes the @xmath43 dependence of @xmath16 , its parameters , @xmath39 , @xmath41 , and @xmath40 , should not depend on @xmath43 .
however , @xmath24 , which is not a parameter of the @xmath0-model , but rather a measure of the strength of the bec , can depend on @xmath43 .
the large correlation between the fit estimates of @xmath24 , @xmath39 , and @xmath41 complicate the testing of @xmath43-independence .
we perform fits in various regions of the @xmath56-@xmath57 plane keeping @xmath39 and @xmath41 fixed at the values obtained in the fit to the entire @xmath43 plane .
the regions are chosen such that the numbers of pairs of particles in the regions are comparable .
the @xmath43 regions and the cls of the fits are shown in fig.[fig : mtmt ] .
the cls are reasonably uniformly distributed between 0 and 1 .
the data are thus in agreement with the hypothesis of @xmath43-independence of the parameters of the @xmath0-model .
r0.4 ( 454,444)(-50,0 ) ( 0,14)(1,0)414 ( 14,0)(0,1)414 ( 0,0 ) ( 14,14)(14,22)(26,22)(26,14 ) ( 0,0 ) ( 14,22)(14,30)(34,30)(34,22 ) ( 0,0)(14,30)(46,30)(46,42)(14,42 ) ( 0,0 ) ( 14,42)(66,42)(66,414)(14,414 ) ( 0,0 ) ( 26,14)(26,22)(42,22)(42,14 ) ( 0,0 ) ( 34,22)(46,22)(46,30)(34,30 ) ( 0,0 ) ( 46,30)(58,30)(58,42)(46,42 ) ( 0,0 ) ( 66,42)(86,42)(86,414)(66,414 ) ( 0,0 ) ( 42,14)(62,14)(62,22)(42,22 ) ( 0,0 ) ( 46,22)(70,22)(70,30)(46,30 ) ( 0,0 ) ( 58,30)(82,30)(82,42)(58,42 ) ( 0,0 ) ( 86,42)(122,42)(122,414)(86,414 ) ( 0,0 ) ( 70,22)(414,22)(414,30)(70,30 ) ( 0,0 ) ( 82,30)(82,42)(414,42)(414,30 ) ( 100,14)(100,0)4(0,1)8 ( -21,90)1 ( -21,190)2 ( -50,150)@xmath64 ( -21,290)3 ( -21,390)4 ( 90,-15)1 ( 190,-15)2 ( 130,-30)@xmath65 ( 290,-15)3 ( 390,-15)4 ( 410,-15 ) ( 4,100)(0,100)4(1,0)8 + ( 110,20)(-10,-10 ) ( 0,0)(1,0)100 ( 0,0)(10,0)11(0,-1)1 ( -2,-6)0 ( 98,-6)1 ( 45,-8)cl ( 10,0)(0,1)5 ( 48,0)(0,1)5 ( 74,0)(0,1)5 ( 13,0)(0,1)5 ( 22,0)(0,1)5 ( 33,0)(0,1)5 ( 34,0)(0,1)5 ( 66,0)(0,1)5 ( 17,0)(0,1)5 ( 55,0)(0,1)5 ( 59,0)(0,1)5 ( 24,0)(0,1)5 ( 4,0)(0,1)5 ( 11,0)(0,1)5 in this section we return to the question of a possible elongation of the region of homogeneity , as viewed in the rest frame of the pion pair , and the possible dependence of the bec correlation function on components of @xmath18 .
the @xmath0-model predicts that the two - particle bec correlation function @xmath16 depends on the two - particle momentum difference only through @xmath18 , not through components of @xmath18 .
however , previous studies @xcite have found that @xmath16 does depend on components of @xmath18 and indeed that the shape of the region of homogeneity is elongated along the event ( thrust ) axis .
the question is whether this is an artifact of the edgeworth or gaussian parametrizations used in these studies or shows a defect of the @xmath0-model . in the studies of elongation ,
@xmath66 is split into three components in the longitudinal center of mass system ( lcms ) of the pair : @xmath67 the lcms frame is defined as the frame , obtained by a lorentz boost along the event axis , where the sum of the three - momenta of the two pions ( @xmath68 ) is perpendicular to the event axis .
assuming azimuthal symmetry about the event axis suggests that the region of homogeneity have an ellipsoidal shape with one axis , the longitudinal axis , along the event axis . in the lcms frame the event axis
is referred to as the longitudinal direction , the direction of @xmath68 as the out direction , and the direction perpendicular to these as the side direction .
the components of @xmath69 along these directions are denoted @xmath70 , @xmath71 , and @xmath72 , respectively .
@xmath73 is then replaced in the gaussian or edgeworth parametrizations by @xmath74 @xmath75 and @xmath76 are measures of the longitudinal and transverse size of the source , respectively , whereas the value of @xmath77 reflects both the transverse and temporal sizes . in eq.([eq - rqlcms ] ) is usually denoted @xmath78 .
we prefer to use @xmath79 to emphasize that , unlike @xmath75 and @xmath76 , @xmath77 contains a dependence on @xmath80 . ]
since the present analysis uses a somewhat different event sample and mixing algorithm , we have first checked that we find values of @xmath81 consistent with our previous analysis @xcite .
we next investigate whether the @xmath0-model parametrizations could accommodate an elongation by incorporating eq.([eq - rqlcms ] ) in eq.([eq : asymlevr2 ] ) .
if eq.([eq : asymlevrar ] ) is imposed , this results in @xmath82 \left(1 + { \ensuremath{\epsilon_\mathrm{l}}}{\ensuremath{q_\mathrm{l}}}+ { \ensuremath{\epsilon_\mathrm{side}}}{\ensuremath{q_\mathrm{side}}}+ { \ensuremath{\epsilon_\mathrm{out}}}{\ensuremath{q_\mathrm{out}}}\right ) \;,\ ] ] where @xmath83 a fit of eq.([eq - asymlevr2bp ] ) for two - jet events for @xmath84 has a cl of 66% , and a clear preference for elongation is found : @xmath85 .
the value of @xmath45 found in the @xmath0-model fit shown in fig .
1c lies between the values of @xmath86 and @xmath87 found here .
furthermore , the value of @xmath81 agrees well with that found using the edgeworth parametrization . to directly test the hypothesis of no separate dependence on components of @xmath18
, we investigate two decompositions of @xmath18 : @xmath88 the first , eq.([eq - q2dec - el ] ) , corresponds to the lcms frame where the longitudinal and energy terms are combined ; its three components of @xmath18 are invariant with respect to lorentz boosts along the event axis .
the second , eq.([eq - q2dec - q ] ) , corresponds to the lcms frame boosted to the rest frame of the pair ; its three components are invariant under lorentz boosts along the out direction .
the test then consists of replacing @xmath89@xmath66 by @xmath89 times one of the above equations and comparing a fit where the coefficients of all three terms are constrained to be equal to a fit where each coefficient is a free parameter . for both parametrizations
the fit allowing separate dependence on the components of @xmath18 has a significantly better cl than the fit which does not .
the fit using eq.([eq - q2dec - el ] ) attains a cl of 2% when no separate dependence is allowed . while this is not poor enough to reject , by itself , the hypothesis of separate dependence , the fit allowing it is , with cl = 38% , much better .
when eq.([eq - q2dec - q ] ) is used the fit not allowing separate dependence is rejected by a cl of @xmath90 , while allowing it results in the marginally acceptable cl of 2% . for both decompositions the differences in @xmath55 between the fit allowing separate dependence and that not allowing it is huge , 296 and 464 , respectively , while a difference of only 2 is expected if there is no separate dependence .
this provides very strong evidence in favor of separate dependence on the components of @xmath18 .
moreover , the radii corresponding to @xmath72 and @xmath91 are found to be unequal by about a factor 2 , which violates the assumption of azimuthal symmetry assumed in the lcms analysis .
the usual parametrizations of bec of pion pairs in terms of @xmath18 , such as the gaussian , edgeworth , or symmetric lvy parametrizations , are found to be incapable of describing the bec correlation function @xmath16 , particularly in the anti - correlation region , @xmath92 .
the @xmath0-model assumes a very high degree of correlation between momentum space and space - time and introduces a time dependence explicitly .
the two - particle bec correlation function @xmath16 is then dependent on one two - particle variable , @xmath18 , and on @xmath38 , a parameter in the correlation between momentum space and space - time .
the time - dependence is assumed to be given by an asymmetric lvy distribution , @xmath37 .
the parameter @xmath38 can be absorbed into an effective radius to obtain eq.([eq : asymlevr2 ] ) , which successfully describes bec for both two- and three - jet events . for two - jet events @xmath34 and the introduction of an effective radius is unnecessary .
the bec correlation function is then a function not only of @xmath18 but also of the transverse masses of the pions .
this description of @xmath93 also successfully describes the data .
nevertheless , the @xmath0-model description breaks down when confronted with data expressed in components of @xmath18 .
the @xmath0-model predicts that the only two - particle variable entering @xmath16 is @xmath18 .
there is no dependence on components of @xmath18 separately .
this implies a spherical shape of the pion region of homogeneity , whereas previous analyses using static parametrizations have found a shape elongated along the event axis , as evidenced by @xmath76 being smaller than @xmath75 in the lcms .
assuming azimuthal symmetry about the event axis , @xmath76 is the transverse radius of the region of homogeneity . accordingly , the @xmath0-model equations for @xmath16 have been modified _
ad hoc _ to allow an elongation .
the elongation found agrees with that found in previous analyses .
however , in the rest frame of the pair the radius parameter in the out direction is found to be approximately twice that in the side direction .
thus the assumption of azimuthal symmetry about the event axis is found not to hold .
we note that elongation is expected @xcite for a hadronizing string in the lund model , where the transverse and longitudinal components of a particle s momentum arise from different mechanisms .
perhaps the absence of azimuthal symmetry arises from gluon radiation , which is represented by kinks in the lund string .
therefore a possible improvement of the @xmath0-model could be to allow the longitudinal and transverse components to have different degrees of correlation between a particle s momentum and its space - time production point .
van dalen , in proc .
8@xmath96 int .
workshop on multiparticle production `` correlations and fluctuations 98 : from qcd to particle interferometry '' , ed .
t. csrg _ et al .
_ , ( world scientific , singapore , 1999 ) , p. 37 . | bose - einstein correlations of pairs of identical charged pions produced in hadronic z decays are analyzed in terms of various parametrizations .
a good description is achieved using a lvy stable distribution in conjunction with a model where a particle s momentum is highly correlated with its space - time point of production , the @xmath0-model . however
, a small but significant elongation of the particle emission region is observed in the longitudinal center of mass frame , which is not accommodated in the @xmath0-model .
this is investigated using an _ ad hoc _ modification of the @xmath0-model .
bose - einstein correlations and the tau model + w.j .
metzger@xmath1 , t. novk@xmath2 , t. csrg@xmath3 , w. kittel@xmath1 + for the l3 collaboration + @xmath1radboud university , p.o .
box 9044 , 6500 kd nijmegen , netherlands + @xmath2kroly rbert college , 3200 gyngys , hungary + @xmath3mta kfki rmki , 1525 budapest 114 , hungary , and harvard university , + 17 oxford st . , cambridge , ma 02138 , u.s.a . |
currently , direct dark matter detection experiments searching for weakly interacting massive particles ( wimps ) are one of the promising methods for understanding the nature of dark matter and identifying them among new particles produced at colliders as well as reconstructing the ( sub)structure of our galactic halo @xcite . in order to determine the mass of halo wimps _ without _ making any assumptions about their density near the earth or their velocity distribution _ nor _ knowing their scattering cross section on nucleus , a model independent method by combining two experimental data sets with two different target nuclei has been developed @xcite .
this method builds on the earlier work on the reconstruction of the ( moments of the ) one dimensional velocity distribution function of halo wimps , @xmath10 , by using data from direct detection experiments @xcite . in the analysis of reconstructing @xmath10 ,
the moments of the wimp velocity distribution function can be determined from experimental data directly with an _ unique _
input information about the wimp mass @xmath11 .
hence , one can simply require that the values of a given moment of @xmath10 determined by two experiments agree .
this leads to a simple analytic expression for determining @xmath11 @xcite , where each moment can in principle be used .
additionally , under the assumptions that the spin independent ( si ) wimp nucleus interaction dominates over the spin
dependent ( sd ) one and the si wimp coupling on protons is approximately the same as that on neutrons , a second analytic expression for determining @xmath11 has been derived @xcite .
finally , by combining the first estimators for different moments with each other and with the second estimator , one can yield the best fit wimp mass as well as minimize its statistical uncertainty . in the work on the development of the model
independent data analysis procedure for the determination of the wimp mass , it was assumed that the analyzed data sets are background
free , i.e. , all events are wimp signals .
active background discrimination techniques should make this condition possible .
for example , the ratio of the ionization to recoil energy , the so called `` ionization yield '' , used in the cdms - ii experiment provides an event by event rejection of electron recoil events to be better than @xmath12 misidentification @xcite . by combining the `` phonon pulse timing parameter '' ,
the rejection ability of the misidentified electron recoils ( most of them are `` surface events '' with sufficiently reduced ionization energies ) can be improved to be @xmath13 for electron recoils @xcite . moreover , as demonstrated by the cresst collaboration @xcite , by means of inserting a scintillating foil , which causes some additional scintillation light for events induced by @xmath14-decay of @xmath15 and thus shifts the pulse shapes of these events faster than pulses induced by wimp interactions in the crystal , the pulse shape discrimination ( psd ) technique can then easily distinguish wimp induced nuclear recoils from those induced by backgrounds . however , as the most important issue in all underground experiments , the signal identification ability and possible residue background events which pass all discrimination criteria and then mix with other real wimp induced events in our data sets should also be considered . therefore ,
in this article , as a more realistic study , we take into account different fractions of residue background events mixed in experimental data sets and want to study how well the model
independent method could reconstruct the input wimp mass by using these `` impure '' data sets and how `` dirty '' these data sets could be to be still useful
. the remainder of this article is organized as follows . in sec . 2 we review the recoil spectrum of elastic wimp nucleus scattering and introduce two kinds of background spectrum used in our simulations .
3 we first review briefly the model independent method for the determination of the wimp mass
. then we show numerical results of the reconstructed wimp mass by using mixed data sets with different fractions of residue background events based on monte carlo simulations .
we conclude in sec .
4 . some technical details will be given in an appendix .
in this section we first review the recoil spectrum of elastic wimp nucleus scattering .
then we introduce two forms of background spectrum which will be used in our simulations .
some numerical results of the measured energy spectrum superposed by the wimp scattering and background spectra will also be discussed .
the basic expression for the differential event rate for elastic wimp nucleus scattering is given by @xcite : = _ ^ dv .
[ eqn : drdq ] here @xmath16 is the direct detection event rate , i.e. , the number of events per unit time and unit mass of detector material , @xmath17 is the energy deposited in the detector , @xmath18 is the elastic nuclear form factor , @xmath10 is the one dimensional velocity distribution function of the wimps impinging on the detector , @xmath19 is the absolute value of the wimp velocity in the laboratory frame .
the constant coefficient @xmath20 is defined as , [ eqn : cala ] where @xmath21 is the wimp density near the earth and @xmath22 is the total cross section ignoring the form factor suppression .
the reduced mass @xmath23 is defined by , [ eqn : mrn ] where @xmath11 is the wimp mass and @xmath24 that of the target nucleus . finally , @xmath25 is the minimal incoming velocity of incident wimps that can deposit the energy @xmath17 in the detector : = [ eqn : vmin ] with the transformation constant , [ eqn : alpha ] and @xmath26 is the maximal wimp velocity in the earth s reference frame , which is related to the escape velocity from our galaxy at the position of the solar system , @xmath27 km / s .
note that , as will be shown below , the earth s velocity relative to the galactic halo is time
dependent , and considering the random motion of wimps in the galaxy , the relation between the one dimensional cut off @xmath26 and the three
dimensional one @xmath28 is thus rather complicated .
nevertheless , it is unlike to affect the event rate as well as the results shown in this article significantly . in the literature , for simplicity and practical uses ,
@xmath26 is often set as @xmath29 ( e.g. , @xcite ) . the simplest semi realistic model halo is a spherical isothermal maxwellian halo .
more realistically , one has to take into account the orbital motion of the solar system around the galaxy as well as that of the earth around the sun
. the one dimensional velocity distribution function of this shifted maxwellian halo can be expressed as @xcite f_1 , ( v ) = .
[ eqn : f1v_sh ] here @xmath30 is the orbital velocity of the sun in the galactic frame , and @xmath31 is the earth s velocity in the galactic frame @xcite : v_e(t ) = v_0 ; [ eqn : ve ] @xmath32 june 2nd is the date on which the velocity of the earth relative to the wimp halo is maximal . substituting eq .
( [ eqn : f1v_sh ] ) into eq .
( [ eqn : drdq ] ) , an analytic form of the integral over the velocity distribution function can be given as @xcite _ ^ dv = + .
[ eqn : int_f1v_sh ] here @xmath33 is the error function , defined as ( x ) = _ 0^x e^-t^2 dt . on the other hand , for practical , numerical uses ,
an approximate form of the integral over @xmath10 was introduced as @xcite _ ^ dv = c_0 e^-^2 q / c_1 v_0 ^ 2 , [ eqn : int_f1v_cc ] where @xmath34 and @xmath35 are two fitting parameters of order unity .
not surprisingly , their values depend on the galactic orbital and escape velocities , the target nucleus , the threshold energy of the experiment , as well as on the mass of incident wimps .
note that , the characteristic energy @xmath36 and thus the shape of the recoil spectrum depend highly on the wimp mass : for light wimps ( @xmath37 ) , @xmath38 and the recoil spectrum drops sharply with increasing recoil energy , while for heavy wimps ( @xmath39 ) , @xmath40 const . and the spectrum becomes flatter . in most theoretical models , the spin
independent ( si ) wimp nucleus interaction with an atomic mass number @xmath41 dominates over the spin dependent ( sd ) one @xcite .
additionally , for the lightest supersymmetric neutralino which is perhaps the best motivated wimp candidate @xcite , and for all wimps which interact primarily through higgs exchange , the si scalar coupling is approximately the same on both protons p and neutrons n , the `` pointlike '' cross section @xmath22 in eq .
( [ eqn : cala ] ) can thus be written as _ 0 = a^2 ^2 ,
[ eqn : sigma0si ] where = m_r , p^2 @xmath42 is the effective @xmath43 four point coupling , @xmath44 is the atomic mass number of the target nucleus , and @xmath45 is the reduced mass of the wimp mass @xmath11 and the proton mass @xmath46 . for the si wimp nucleus cross section , an analytic form for the elastic nuclear form factor , inspired by the woods
saxon nuclear density profile , has been suggested by engel as @xcite f_ws^2(q ) = ^2 e^-(q s)^2 .
[ eqn : fq_si_ws ] here @xmath47 is a spherical bessel function , @xmath48 is the transferred 3-momentum , given as a function of the recoil energy transferred from the incident wimp to the target nucleus , @xmath17 , and the mass of the target nucleus , @xmath24 ; @xmath49 is the effective nuclear radius with @xmath50 and the nuclear skin thickness @xmath51 . for our simulations with residue background events ,
two forms of background spectrum are considered .
the simplest choice for the background spectrum is the constant spectrum : _ bg , const = 1 .
[ eqn : drdq_bg_const ] more realistically , inspired by ref .
@xcite , we introduce a _ target
dependent exponential _ spectrum given by _
bg , ex = .
[ eqn : drdq_bg_ex ] here @xmath17 is the recoil energy , @xmath44 is the atomic mass number of the target nucleus .
the power index of @xmath44 , 0.6 , is an empirical constant , which has been chosen so that the exponential background spectrum is somehow _ similar to _ , but still _ different from _ the expected recoil spectrum of the target nuclei ; otherwise , there is in practice no difference between the wimp scattering and background spectra .
note that , among different possible choices ( e.g. , the exponential form used in ref .
@xcite ) , we use in our simulations the atomic mass number @xmath44 as the simplest , unique characteristic parameter in the general analytic form ( [ eqn : drdq_bg_ex ] ) for defining the residue background spectrum for _ different _ target nuclei .
however , it does _ not _ mean that the ( superposition of the real ) background spectra would depend simply / primarily on @xmath44 or on the mass of the target nucleus , @xmath24 .
in other words , it is practically equivalent to use expression ( [ eqn : drdq_bg_ex ] ) or @xmath52 directly for a @xmath53 target .
note also that , firstly , two forms of background spectrum given in eqs .
( [ eqn : drdq_bg_const ] ) and ( [ eqn : drdq_bg_ex ] ) are rather naive ; however , since we consider here _ only a few residue _ background events induced by perhaps _ two or more _ different sources , pass all discrimination criteria , and then mix with other wimp induced events in our data sets of @xmath0 _ total _ events , exact forms of different background spectra are actually not very important and these two spectra , in particular , the exponential one , should practically not be unrealistic . secondly , for using the maximum likelihood analysis to determine the wimp mass , as described in refs .
@xcite , a prior knowledge about the wimp scattering spectrum and eventually about the background spectrum is essential @xcite .
in contrast , as demonstrated in ref .
@xcite and will be reviewed in the next section , the model
independent data analysis procedure requires only measured recoil energies ( induced mostly by wimps and occasionally by background sources ) from two experimental data sets with different target nuclei .
therefore , for applying this method to future real data from direct detection experiments , the prior knowledge about ( different ) background source(s ) is _ not required at all_. target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details . , title="fig:",width=370 ] target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods
saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details .
, title="fig:",width=370 ] + target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details . , title="fig:",width=370 ] target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details .
, title="fig:",width=370 ] + target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods
saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details .
, title="fig:",width=370 ] target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution and the woods
saxon elastic nuclear form factor ; whereas the dashed green curves are the exponential background spectra normalized to fit to the chosen background ratio , which has been set as 20% here .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
the background windows have been assumed to be the same as the experimental possible energy ranges .
5,000 experiments with 500 total events on average in each experiment have been simulated .
see the text for further details .
, title="fig:",width=370 ] + , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details .
, title="fig:",width=370 ] , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details . ,
title="fig:",width=370 ] + , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details .
, title="fig:",width=370 ] , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details .
, title="fig:",width=370 ] + , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details .
, title="fig:",width=370 ] , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . see the text for further details .
, title="fig:",width=370 ] + in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] we show measured energy spectra ( solid red histograms ) for a @xmath53 target with six different wimp masses : 10 , 25 , 50 , 100 , 250 , and 500 gev based on monte carlo simulations .
the dotted blue curves are the elastic wimp nucleus scattering spectra for the shifted maxwellian velocity distribution given in eq .
( [ eqn : f1v_sh ] ) with @xmath54 km / s , @xmath55 , in eq .
( [ eqn : ve ] ) , has been ignored . ] and @xmath56 km / s and the woods
saxon elastic nuclear form factor in eq .
( [ eqn : fq_si_ws ] ) .
the dashed green curves are the exponential background spectra given in eq .
( [ eqn : drdq_bg_ex ] ) , which have been normalized so that the ratios of the areas under these background spectra to those under the ( dotted blue ) wimp scattering spectra are equal to the background signal ratio in the whole data sets ( i.e. , 20% backgrounds to 80% signals shown in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] ) .
the experimental threshold energy has been assumed to be negligible and the maximal cut off energy is set as 100 kev .
5,000 experiments with 500 total events on average in each experiment have been simulated .
the measured energy spectra ( solid red histograms ) shown in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] are averaged over the simulated experiments .
five bins with linearly increased bin widths have been used for binning generated signal and background events . as argued in ref .
@xcite , for reconstructing the one dimensional wimp velocity distribution function , this unusual , particular binning has been chosen in order to accumulate more events in high energy ranges and thus to reduce the statistical uncertainties in high velocity ranges .
however , as we will show later , for the determination of the wimp mass , one needs either events in the _ first _ energy bin or _ all _ events in the whole data set .
hence , there is in practice no difference between using an equal bin width for all bins or the ( linearly ) increased bin widths .
note here that , firstly , the possible energy ranges in which residue background events exist ( the background windows ) have been assumed to be the same as the entire experimental possible energy ranges ( e.g. , between 0 and 100 kev for simulations shown in figs . [
fig : drdq - bg - ex - ge-000 - 100 - 20 ] ) .
secondly , the actual numbers of signal and background events in each simulated experiment are poisson distributed around their expectation values _ independently_.
this means that , for example , for simulations shown in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] we generate 400 ( 100 ) events on average for wimp signals ( backgrounds ) and the total event number recorded in one experiment is then the sum of these two numbers .
thirdly , for the simulations demonstrated here as well as in the next section , we assumed that all experimental systematic uncertainties as well as the uncertainty on the measurement of the recoil energy could be ignored .
the energy resolution of most existing detectors is so good that its error can be neglected compared to the statistical uncertainty for the foreseeable future with pretty few events . in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] it can be found that , as mentioned earlier , the shape of the wimp scattering spectrum depends highly on the wimp mass : for light wimps ( @xmath57 gev ) , the recoil spectra drop sharply with increasing recoil energies , while for heavy wimps ( @xmath58 gev ) , the spectra become flatter .
in contrast , the exponential background spectra shown here depend only on the target mass and are rather _ flatter _ ( _ sharper _ ) for _ light _ ( _ heavy _ ) wimp masses compared to the wimp scattering spectra .
this means that , once input wimps are _ light _
( _ heavy _ ) , background events would contribute relatively more to _ high _ ( _ low _ ) energy ranges , and , consequently , the measured energy spectra would mimic scattering spectra induced by _ heavier _
( _ lighter _ ) wimps . as a comparison , in figs .
[ fig : drdq - bg - const - ge-000 - 100 - 20 ] we generate background events with the constant spectrum given in eq .
( [ eqn : drdq_bg_const ] ) .
it can be seen clearly that , since the background spectrum now is flatter for all wimp masses , background events contribute always relatively more to _ high _ energy ranges , and , therefore , the measured energy spectra would always mimic scattering spectra induced by _ heavier _ wimps .
in this section we first review the model independent method for determining the wimp mass introduced in refs .
. then we demonstrate some numerical results of the reconstructed wimp mass by using mixed data sets from wimp signals and background events based on monte carlo simulations . here
we review briefly the model independent data analysis procedure for the determination of the wimp mass by using two experimental data sets with different target nuclei .
detailed derivations and discussions can be found in refs .
@xcite . in the earlier work @xcite
, it was found that the normalized one dimensional velocity distribution function of incident wimps can be solved from eq .
( [ eqn : drdq ] ) directly and , consequently , its generalized moments can be estimated by @xcite ( v ( ) , v ( ) ) = _ v()^v ( ) v^n f_1(v ) ~ dv + = ^n .
[ eqn : moments ] here @xmath59 , @xmath60 are the experimental minimal and maximal cut off energies ,
r ( ) _ expt , q = [ eqn : rmin ] is an estimated value of the _ measured _ recoil spectrum @xmath61 ( _ before _ the normalization by the exposure @xmath62 ) at @xmath63 , and @xmath64 can be estimated through the sum : i_n ( , ) = _ a , [ eqn : in_sum ] where the sum runs over all events in the data set that satisfy @xmath65 $ ] . by requiring that the values of a given moment of @xmath10 estimated by eq .
( [ eqn : moments ] ) from two experiments with different target nuclei , @xmath66 and @xmath67 , agree , @xmath11 appearing in the prefactor @xmath68 on the right hand side of eq .
( [ eqn : moments ] ) has been solved as @xcite : .
@xmath69 can be defined analogously .
here @xmath70 , @xmath71 and @xmath72 are the masses and the form factors of the nucleus @xmath66 and @xmath67 , respectively , and @xmath73 refer to the counting rates for detectors @xmath66 and @xmath67 at the respective lowest recoil energies included in the analysis . note that , firstly , the general expression ( [ eqn : mchi_rn ] ) can be used either for spin independent or for spin dependent scattering , one only needs to choose different form factors under different assumptions . secondly , the form factors in the estimate of @xmath74 and @xmath75 using eq .
( [ eqn : in_sum ] ) are also different . on the other hand , by using the theoretical prediction that the si wimp nucleus cross section dominates , and the fact that the integral over the one dimensional wimp velocity distribution on the right hand side of eq .
( [ eqn : drdq ] ) is the minus
first moment of this distribution , which can be estimated by eq .
( [ eqn : moments ] ) with @xmath76 , one can easily find that @xcite _ 0 |f_p|^2 = .
[ eqn : rho0_fp2 ] note that the exposure of the experiment , @xmath77 , appears in the denominator . since the unknown factor @xmath78 on the left hand side above is identical for different targets , it leads to a second expression for determining @xmath11 @xcite : . | _ = .
[ eqn : mchi_rsigma ] here @xmath79 has been assumed , _ , x , [ eqn : rsigmax_min ] and similarly for @xmath80 . in order to yield the best
fit wimp mass as well as to minimize its statistical uncertainty by combining the estimators for different @xmath82 in eq .
( [ eqn : mchi_rn ] ) with each other and with the estimator in eq .
( [ eqn : mchi_rsigma ] ) , a @xmath81 function has been introduced as @xcite ^2 ( ) = _ i , j ^-1_ij , [ eqn : chi2 ] where f_i , x _ x^i ^i , [ eqn : fixa ] for @xmath83 , and f_n_max+1 , x _ x ; [ eqn : fixb ] the other @xmath84 functions @xmath85 can be defined analogously . here @xmath86 determines the highest moment of @xmath10 that is included in the fit .
the @xmath87 are normalized such that they are dimensionless and very roughly of order unity in order to alleviate numerical problems associated with the inversion of their covariance matrix .
note that the first @xmath88 fit functions depend on @xmath11 only through the overall factor @xmath14 and that @xmath11 in eqs .
( [ eqn : fixa ] ) and ( [ eqn : fixb ] ) is now a fit parameter , which may differ from the true value of the wimp mass .
finally , @xmath89 in eq .
( [ eqn : chi2 ] ) is the total covariance matrix .
since the @xmath66 and @xmath67 quantities are statistically completely independent , @xmath89 can be written as a sum of two terms : _
ij = cov + cov .
[ eqn : cij ] the basic requirement of the expressions for determining @xmath11 given in eqs .
( [ eqn : mchi_rn ] ) and ( [ eqn : mchi_rsigma ] ) is that , from two experiments with different target nuclei , the values of a given moment of the wimp velocity distribution estimated by eq .
( [ eqn : moments ] ) should agree .
this means that the upper cuts on @xmath10 in two data sets should be ( approximately ) equal . since @xmath90
, it requires that @xcite q_max , y = ^2 q_max , x .
[ eqn : match ] note that @xmath14 defined in eq .
( [ eqn : alpha ] ) is a function of the true wimp mass .
thus this relation for matching optimal cut off energies can be used only if @xmath11 is already known .
one possibility to overcome this problem is to fix the cut off energy of the experiment with the heavier target ,
minimize the @xmath91 function defined in eq .
( [ eqn : chi2 ] ) , and then estimate the cut off energy for the lighter nucleus by eq .
( [ eqn : match ] ) algorithmically @xcite . in this subsection
we show some numerical results of the reconstruction of the wimp mass with mixed data sets from wimp induced and background events by means of the model
independent method described in the previous subsection .
the upper and lower bounds on the reconstructed wimp mass are estimated from the requirement that @xmath81 exceeds its minimum by 1 . as in ref .
@xcite , @xmath92 and @xmath53 have been chosen as two target nuclei .
the scattering cross section @xmath22 in eq .
( [ eqn : cala ] ) has been assumed to be dominated by the spin
independent wimp
nucleus interaction .
the experimental threshold energies of two experiments have been assumed to be negligible and the maximal cut
off energies are set the same as 100 kev .
2 @xmath93 5,000 experiments have been simulated . in order to avoid large contributions from very few events in high energy ranges to the higher moments
@xcite , only the moments up to @xmath94 were included in the @xmath81 fit .
[ fig : mchi - sige - ex-000 - 100 - 050 ] shows the reconstructed wimp mass and the lower and upper bounds of the 1@xmath2 statistical uncertainty with mixed data sets from wimp induced and background events as functions of the input wimp mass . as in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] , the exponential background spectrum has been used and the background windows are set as the same as the experimental possible energy ranges , i.e. , between 0 and 100 kev for both experiments .
the background ratios shown here are no background ( dashed green curves ) , 10% ( long dotted blue curves ) , 20% ( solid red curves ) , and 40% ( dash dotted cyan curves ) background events in the whole data sets .
each experiment contains 50 _ total _ events on average before cuts on @xmath95 for the experiments with the si target .
remind that _ all _ events recorded in our data sets are treated as wimp signals in the analysis , although statistically we know that a fraction of these events could be backgrounds . statistical uncertainty with mixed data sets from wimp induced and background events as functions of the input wimp mass .
@xmath92 and @xmath53 have been chosen as two target nuclei .
the background ratios shown here are no background ( dashed green curves ) , 10% ( long dotted blue curves ) , 20% ( solid red curves ) , and 40% ( dash dotted cyan curves ) background events in the whole data sets in the experimental energy ranges between 0 and 100 kev .
each experiment contains 50 _ total _ events on average before cuts on @xmath95 for the experiments with the si target ; _ all _ of these events are treated as wimp signals .
other parameters are as in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] .
see the text for further details .
, title="fig:",width=566 ] + it can be seen clearly that , for _ light _ wimp masses ( @xmath96 gev ) , the larger the fraction of background events in the data sets , the _ heavier _ the reconstructed wimp masses as well as the statistical uncertainty intervals .
this is caused directly by the background contribution to _ high _ energy ranges shown in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] . as discussed in sec .
2.3 , the background spectrum is relatively _ flatter _
compared to the scattering spectrum induced by _ light _
wimps , and the energy spectrum of all recorded events would thus mimic a scattering spectrum induced by wimps with a _
relatively heavier _ mass .
not surprisingly , the larger the background ratio , the more the background contribution to high energy ranges , and , consequently , the more strongly _ overestimated _ the reconstructed wimp masses as well as the statistical uncertainty intervals .
in contrast , for _ heavy _ wimp masses ( @xmath58 gev ) , fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] does not show very clearly but a tendency ) events discussed later . ]
that the larger the fraction of background events , the _
lighter _ the reconstructed wimp masses as well as the statistical uncertainty intervals .
this is now caused by the background contribution to _ low _ energy ranges shown in figs .
[ fig : drdq - bg - ex - ge-000 - 100 - 20 ] .
as discussed in the previous section , the background spectrum is relatively _ sharper _ compared to the scattering spectrum induced by _ heavy _
wimps , and the energy spectrum of all recorded events would thus mimic a scattering spectrum induced by wimps with a _
relatively lighter _ mass .
moreover , the larger the background ratio , the more the background contribution to low energy ranges , and , consequently , the more strongly _ underestimated _ the reconstructed wimp masses as well as the statistical uncertainty intervals . , except that the background window in each experiment have been set as 0 50 kev ( upper ) and 50 100 kev ( lower ) . note that the background ratios shown here are 20% ( solid red curves ) and 40% ( dash dotted cyan curves ) in the upper frame , whereas 5% ( dotted magenta curves ) and 10% ( long dotted blue curves ) in the lower frame .
, title="fig:",width=566 ] + , except that the background window in each experiment have been set as 0 50 kev ( upper ) and 50 100 kev ( lower ) .
note that the background ratios shown here are 20% ( solid red curves ) and 40% ( dash dotted cyan curves ) in the upper frame , whereas 5% ( dotted magenta curves ) and 10% ( long dotted blue curves ) in the lower frame . , title="fig:",width=566 ] + nevertheless , from fig . [ fig : mchi - sige - ex-000 - 100 - 050 ] it can be found that , with @xmath1 20% residue background events in the analyzed data sets , the true values of the wimp mass can still fall in the middle of the 1@xmath2 statistical uncertainty band and one could thus in principle reconstruct the wimp mass pretty well ; if wimps are light ( @xmath97 gev ) , the maximal acceptable fraction of residue background events could even be as large as @xmath1 40% .
for a wimp mass of 100 gev with 20% background events in the data sets , the reconstructed wimp mass and the statistical uncertainty are @xmath3 , compared to @xmath4 for background free data sets ; for a lighter wimp mass of 50 gev , the reconstructed wimp mass and the statistical uncertainty change from @xmath98 ( background free ) , to @xmath99 ( 20% background ) , and @xmath100 ( 40% background ) .
on the other hand , considering different efficiencies of discrimination ability against different background sources in different energy ranges in different experiments , in figs .
[ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] we shrink the background window in each experiment to a relatively lower range between 0 and 50 kev ( upper ) and a relatively higher range between 50 and 100 kev ( lower ) . since our background spectrum is exponential , for the case shown in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] , only very few background events could be observed in the energy range between 50 and 100 kev .
hence , for the case with the background window only in the _ low _ energy range , not surprisingly , the results of the reconstructed wimp mass shown in the upper frame of figs . [ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] should not differ very much from those shown in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ]
. however , due to the little bit more contribution to the _ low _ energy range from background events , _ all _ the reconstructed wimp masses shown here are somehow _ lighter _ than those shown in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] .
hence , with @xmath1 20% residue background events in low experimental possible energy ranges , one could in principle reconstruct the wimp mass with a 1@xmath2 statistical uncertainty as @xmath101 ( for a wimp mass of 100 gev ) or @xmath102 ( for a wimp mass of 50 gev ) .
in contrast , since the wimp scattering spectrum is in principle approximately exponential and thus only ( very ) few wimp induced events could be observed in high energy ranges , if we have background windows in only high experimental possible energy ranges , the ( pretty large ) contributions from background events could cause ( strong ) _ overestimates _ of the reconstructed wimp masses .
it is even worse for large wimp masses ( @xmath58 gev ) gev ) should be caused by our setup for the upper cut off of the reconstructed wimp mass of 3000 gev in the simulations . ] .
nevertheless , as shown in the lower frame of figs .
[ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] , with @xmath1 5% residue background events observed only in high energy ranges , one could in principle still estimate the wimp mass with a 1@xmath2 statistical uncertainty as @xmath103 ( for an input wimp mass of 100 gev ) or @xmath104 ( for an input wimp mass of 50 gev ) .
our results shown in figs . [ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] indicate that a small fraction of background events in _ low _ energy ranges might not affect the reconstructed wimp mass significantly
. however , the wimp mass could be ( strongly ) _ overestimated _ once the same ( or even smaller ) amount of background events exists in _ high _ energy ranges . in practice one simple way to reduce the overestimate induced by an excess of background events in high energy ranges might be checking the shape of measured recoil spectrum . however , considering some suggested modifications of the standard shifted maxwellian velocity distribution , e.g. , contributions from discrete `` streams '' with ( nearly ) fixed velocities @xcite or the `` late infall '' component in the velocity distribution with a velocity @xmath105 @xcite , it should at least be very careful to reject any recoil event observed in high energy ranges artificially .
, except that the expected number of total events in each experiment has been set as 500 .
, title="fig:",width=566 ] + in fig .
[ fig : mchi - sige - ex-000 - 100 - 500 ] we rise the expected number of total events in each experiment by a factor of 10 , to 500 events on average before cuts for the case that residue background events exist in the entire experimental possible energy ranges . as shown here , all statistical uncertainties shrink by a factor @xmath106 compared to the results shown in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] .
in addition , the underestimate of the reconstructed values of heavy input wimp masses caused perhaps by the use of pretty few ( @xmath107 ) events has been reduced with larger data sets ; and , the tendency of the underestimate of the reconstructed wimp mass for heavy wimp masses ( @xmath58 gev ) becomes more clearly .
finally , fig .
[ fig : mchi - sige - ex-000 - 100 - 500 ] shows that , for the determination of the wimp mass by using data sets of @xmath108(500 ) total events , the maximal acceptable background ratio could be @xmath1 10% ( i.e. , @xmath108(50 ) background events ) or even @xmath1 20% , if wimps have a mass of @xmath108(100 gev ) . as discussed in ref .
@xcite , the statistical fluctuation of the reconstructed wimp mass by the algorithmic procedure in the simulated experiments seems to be pretty problematic , in particular for heavier input wimp masses .
moreover , as mentioned in the previous subsection , with only @xmath1 50 total events in each experiment , the tendency of the underestimate of the reconstructed wimp mass for heavier wimp masses ( @xmath58 gev ) seems not to be very clear .
hence , as done in ref .
@xcite , in order to study the statistical fluctuation of the reconstructed wimp mass with different background ratios in our data sets , we consider in this subsection the estimator @xmath109 introduced in ref .
@xcite :
m = \ { l c l 1 + , & & if m _ , in m _ , lo1 ; + & & + , & & if m_,lo1 <
m _ , in < m _ , rec ; + & & + , & & if m_,rec <
m _ , in < m _ , hi1 ; + & & + - 1 , & & if m _ , in m_,hi1 . .
[ eqn : deltam ] here @xmath110 is the true ( input ) wimp mass , @xmath111 its reconstructed value , @xmath112 are the @xmath113 lower bounds satisfying @xmath114 , and @xmath115 are the corresponding @xmath113 upper bounds . the estimator @xmath109 defined above indicates basically the _ strength of the deviation _ of the reconstructed wimp mass from the true ( input ) value .
if the reconstructed 1@xmath2 lower and upper bounds on the wimp mass in one simulated experiment cover the true value : @xmath116 , @xmath109 is determined as the deviation of the `` reconstructed wimp mass '' from the true one in units of the difference between the reconstructed value and the 1@xmath2 lower ( upper ) bound , once the reconstructed value is overestimated ( underestimated ) . however , if the true wimp mass lies _ outside _ of the experimental 1@xmath2 bounds ( the reconstructed value is more strongly over-/underestimated ) , @xmath109 is determined as the deviation of the `` 1@xmath2 lower ( upper ) bound '' from the true wimp mass in units of the difference between the 1@xmath2 and 2@xmath2 lower ( upper ) bounds . note
that , it has been found in ref .
@xcite as well as in the results presented in the previous subsection that the uncertainty intervals of the median reconstructed wimp mass are quite asymmetric ; similarly , the distance between the 1@xmath2 and 2@xmath2 bounds can be quite different from the distance between the reconstructed value and the 1@xmath2 bound @xcite . the definition of @xmath109 in eq.([eqn : deltam ] )
takes these differences into account , and also keeps track of the sign of the deviation : if the reconstructed wimp mass is overestimated ( underestimated ) , @xmath109 is positive ( negative ) .
moreover , @xmath117 if and only if the true wimp mass lies between the experimental @xmath113 bounds . defined in eq .
( [ eqn : deltam ] ) for an input wimp mass of 200 gev with 50 ( upper ) and 500 ( lower ) total events on average before cuts in each experiment .
parameters and notations are as in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] .
note that the bins at @xmath118 are overflow bins , i.e. , they also contain all experiments with @xmath119 .
see the text for further details .
, title="fig:",width=566 ] + defined in eq .
( [ eqn : deltam ] ) for an input wimp mass of 200 gev with 50 ( upper ) and 500 ( lower ) total events on average before cuts in each experiment .
parameters and notations are as in fig .
[ fig : mchi - sige - ex-000 - 100 - 050 ] .
note that the bins at @xmath118 are overflow bins , i.e. , they also contain all experiments with @xmath119 .
see the text for further details .
, title="fig:",width=566 ] + in figs .
[ fig : mchi - dev - sige - ex-000 - 100 ] we show the normalized distributions of the estimator @xmath109 defined in eq . (
[ eqn : deltam ] ) for a rather heavy input wimp mass of 200 gev with 50 ( upper ) and 500 ( lower ) total events on average before cuts in each experiment . as discussed in ref .
@xcite , the deviation of the reconstructed wimp mass in the simulated experiments looks asymmetric and non gaussian .
however , it can be seen here clearly that , the more the background events in our analyzed data sets , the more concentrated the @xmath109 value in the range between @xmath120 and 0 as well as between 0 and @xmath121 . moreover , for the case with rather larger data sets of 500 total events , by increasing the background ratio the distribution becomes to be more symmetric and gaussian like , although the central value of @xmath109 seems to fall at @xmath122 because of the underestimate of the reconstructed wimp mass . in ref .
@xcite it has been mentioned that with increasing number of total events the distribution of the estimator @xmath109 becomes slowly gaussian .
[ fig : mchi - dev - sige - ex-000 - 100 ] here ( and fig .
[ fig : mchi - dev - sige - const-000 - 100 - 050 ] shown later also ) indicate that with a _
larger _ background ratio in the analyzed data sets the distribution of @xmath109 approaches to be gaussian _
more fast_. this interesting observation might be able to offer some new ideas for improving the algorithmic procedure for the reconstruction of the wimp mass with a higher statistical certainty . in order to check the need of a prior knowledge about an ( exact ) form of the residue background spectrum , we consider briefly in this subsection a rather extrem case , i.e. , the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) . , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used . note that the background ratios shown here are 5% ( dotted magenta curves ) and 10% ( long dotted blue curves ) .
, title="fig:",width=566 ] + in fig .
[ fig : mchi - sige - const-000 - 100 - 050 ] we show the reconstructed wimp mass and the lower and upper bounds of the 1@xmath2 statistical uncertainty with mixed data sets as functions of the input wimp mass . as in figs .
[ fig : drdq - bg - const - ge-000 - 100 - 20 ] , the windows of the constant background spectrum are set as the same as the experimental possible energy ranges , i.e. , between 0 and 100 kev for both experiments .
the background ratios shown here are no background ( dashed green curves ) , 5% ( dotted magenta curves ) , and 10% ( long dotted blue curves ) background events in the whole data sets .
each experiment contains again 50 total events on average before cuts ; all of these events are treated as wimp signals in the analysis .
it can be seen clearly that , as discussed above , since the constant background spectrum has relatively _ flatter _ shape compared to the wimp scattering spectrum for not only light , but also heavy wimp masses , and the measured energy spectrum should thus always mimic a scattering spectrum induced by _ heavier _ wimps , the reconstructed wimp masses are therefore _ overestimated _ for all input wimp masses , especially for the heavier masses . actually , the result shown here looks more likely that shown in the lower frame of figs .
[ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] , since in both cases residue background events contribute significantly more ( compared to the exponential like wimp scattering spectrum ) in high energy ranges .
not surprisingly , the larger the background ratio , the more strongly overestimated the reconstructed wimp masses , in particular for the heavier input wimp masses .
nevertheless , for ( approximately ) constant residue backgrounds with a fraction of @xmath1 5% in background windows as the entire experimental possible ranges , one could in principle still estimate the wimp mass with a 1@xmath2 statistical uncertainty as ( for 100 gev wimps ) or @xmath123 ( for 50 gev wimps ) .
once wimps are light ( @xmath124(25 gev ) ) , the maximal acceptable background ratio could even be @xmath1 10% .
, except that the background window in each experiment has been set as 0 50 kev .
, title="fig:",width=566 ] + , except that the constant background spectrum in eq .
( [ eqn : drdq_bg_const ] ) has been used .
, title="fig:",width=566 ] + , except that the expected number of total events in both experiments has been set as 500 .
note that the background ratios shown here are no background ( dashed green curves ) , 1% ( long dotted blue curves ) , and 2% ( solid red curves ) , i.e. , a factor of 10 smaller than the ratios used before .
see the text for further details .
, title="fig:",width=566 ] + moreover , as done in sec . 3.2.1 , in fig .
[ fig : mchi - sige - const-000 - 050 - 050 ] we shrink the background window in each experiment to the lower energy range between 0 and 50 kev .
not surprisingly , while for light wimps ( @xmath125 gev ) , relatively more background events still contribute to high energy ranges ; for heavy wimps ( @xmath126 gev ) , relatively more background events contribute now to low energy ranges and , consequently , the reconstructed wimp masses are therefore _ underestimated _ for heavy wimps . on the other hand , as in sec . 3.2.2 , in fig .
[ fig : mchi - dev - sige - const-000 - 100 - 050 ] we check the normalized distributions of the estimator @xmath109 for an input wimp mass of 200 gev with 50 total events on average before cuts in each experiment .
it can be seen very clearly that , with increasing background ratio the value of @xmath109 concentrates more and more strongly to 2 .
this means that , due to the contribution from residue background events , the reconstructed wimp mass is most possibly @xmath1 2@xmath2 overestimated .
moreover , compared to the non gaussian form of the distributions for the case with the exponential background spectrum shown in the upper frame of figs .
[ fig : mchi - dev - sige - ex-000 - 100 ] , the distributions with the constant spectrum look more likely gaussian , despite of the asymmetry due to the overestimate of the wimp mass
. nevertheless , figs .
[ fig : mchi - dev - sige - ex-000 - 100 ] and fig .
[ fig : mchi - dev - sige - const-000 - 100 - 050 ] indicate that background events seem to let the distribution of the deviation of the reconstructed wimp mass be more symmetric and gaussian , no matter what kind of energy spectrum they would have . finally , in fig .
[ fig : mchi - sige - const-000 - 100 - 500 ] we rise the expected number of total events in each experiment by a factor of 10 , to 500 events on average before cuts for the case that residue background events exist in the entire experimental possible energy ranges .
note that the background ratios shown here are no background ( dashed green curves ) , 1% ( long dotted blue curves ) , and 2% ( solid red curves ) , i.e. , a factor of 10 smaller than the ratios used before . in the lower frame of figs .
[ fig : mchi - sige - ex-000 - 050 - 100 - 050 ] and in fig .
[ fig : mchi - sige - const-000 - 100 - 050 ] , we found that once @xmath1 5% 10% events in our analyzed data sets are residue backgrounds and ( most of ) these events are recorded in high energy ranges , no matter what kind of spectrum shape they would have , the reconstructed wimp mass could be ( strongly ) _
overestimated_. however , fig .
[ fig : mchi - sige - ex-000 - 100 - 500 ] and fig .
[ fig : mchi - sige - const-000 - 100 - 500 ] here show that , by increasing the event number and decreasing the background ratio , one could in principle determine the wimp mass ( pretty ) precisely _ without _ knowing the ( exact ) form of the spectrum of residue background events .
in this paper we reexamine the model independent data analysis method introduced in refs .
@xcite for the determination of the mass of weakly interacting massive particles from data ( measured recoil energies ) of direct dark matter detection experiments directly by taking into account a fraction of residue background events , which pass all discrimination criteria and then mix with other real wimp induced events in the analyzed data sets .
differ from the maximum likelihood analysis described in refs .
@xcite , our method requires _ neither _ prior knowledge about the wimp scattering spectrum _ nor _ about different possible background spectra ; the unique needed information is the recoil energies recorded in _ two _ direct detection experiments with _ two different _ target nuclei . in sec . 2 we considered first the measured energy spectrum for different wimp masses with two forms of possible residue background spectrum : the _ target dependent exponential _ spectrum and the _ constant _ spectrum
the exponential background spectrum contributes relatively more events to _ high _ energy ranges once wimps are _ light _ ( @xmath96 gev ) , and to _ low _ energy ranges for _ heavy _ wimp masses ( @xmath58 gev ) ; whereas the constant background spectrum contributes always relatively more events to _ high _ energy ranges . as the consequence , the energy spectrum of all observed events looks more likely to be a scattering spectrum induced by _
heavier _ wimps , once the spectrum of residue background events ( induced perhaps by two or more different sources ) is either exponential like ( and wimps are light ) or approximately constant ( for all wimp masses ) ; while if wimps are heavy and the residue background spectrum is approximately exponential , the measured energy spectrum would look more likely to be a scattering spectrum induced by _
lighter _ wimps . in sec .
3.2 the data sets generated in sec . 2 have been analyzed for reconstructing the mass of incident wimps by using the model independent method . with the exponential background spectrum ,
the input wimp mass would be _ overestimated _ once wimps are light ( @xmath96 gev ) , or , in contrast , would be _ underestimated _ for heavy wimps ( @xmath58 gev ) .
our simulations show that , for background windows in the _ entire or low _ experimental possible energy ranges , one could in principle reconstruct the wimp mass with a maximal fraction of @xmath1 20% of residue background events in the analyzed data sets ; whereas for background windows in _ high _ energy ranges , the maximal acceptable fraction of residue backgrounds is only @xmath1 10% .
moreover , in order to check the need of a prior knowledge about an ( exact ) form of the residue background spectrum , we considered also the case with the constant background spectrum . in this rather extrem case ,
the wimp mass would always be _ overestimated _ , especially for heavy wimps ( @xmath58 gev ) .
our simulations give then a maximal acceptable fraction of of residue background events in the data sets for background windows in the _ entire or low _ experimental possible energy ranges .
nevertheless , we found also that , by means of increased number of observed ( wimp induced ) events and improved background discrimination techniques @xcite , the wimp mass could in principle be determined ( pretty ) precisely , no matter what kind of energy spectrum residue background events would have . on the other hand , in order to check the statistical fluctuation of the reconstructed wimp mass with increased background ratio , we considered also the distribution of the deviation of the reconstructed wimp mass from the true value .
it was found in ref .
@xcite that , for a rather heavy wimp mass of 200 gev , the distribution of the deviation of the reconstructed wimp mass is asymmetric and non gaussian , either with data sets of only a few ( @xmath108(50 ) ) events or with larger date sets ( of @xmath108(500 ) events ) .
however , our simulations with different background ratios show that , firstly , for both used ( exponential and constant ) background spectra , with increasing background ratio the distribution of the deviation of the reconstructed wimp mass becomes more and more concentrated , although still asymmetric and non gaussian . secondly ,
for the more realistic exponential background spectrum and using data sets with a larger number of total events , with increasing background ratio the distribution of the deviation becomes somehow more symmetric and gaussian .
this observation might be able to offer some new ideas for improving the algorithmic procedure for the reconstruction of the wimp mass with a higher statistical certainty . in summary ,
our study of the effects of residue background events in direct dark matter detection experiments on the determination of the wimp mass shows that , with currently running and projected experiments using detectors with @xmath127 to @xmath128 pb sensitivities @xcite and @xmath13 background rejection ability @xcite , once two or more experiments with different target nuclei could accumulate a few tens events ( in one experiment ) , we could in principle already estimate the mass of dark matter particle with a reasonable precision , even though there might be some background events mixed in our data sets for the analysis .
moreover , two forms for background spectrum and three windows for residue background events considered in this work are rather naive .
nevertheless , one should be able to extend our observations / discussions to predict the effects of possible background events in their own experiment .
hopefully , this will encourage our experimental colleagues to present their ( future ) results not only in form of the `` exclusion limit(s ) '' , but also of the `` most possible area(s ) '' on the cross section versus mass plan .
the authors would like to thank the physikalisches institut der universitt tbingen for the technical support of the computational work demonstrated in this article .
cls would also like to thank the friendly hospitality of the max planck
institut fr kernphysik in heidelberg where part of this work was completed .
this work was partially supported by the national science council of r.o.c . under contracts no .
nsc-96 - 2112-n-009 - 023-my3 and no .
nsc-98 - 2811-m-006 - 044 as well as by the lhc physics focus group , national center of theoretical sciences , r.o.c ..
here we list all formulae needed for the model independent data analyses procedure used in sec . 3 .
detailed derivations and discussions can be found in refs .
@xcite .
[ [ estimating - rqmin - i_nqmin - qmax - and - their - statistical - errors ] ] estimating @xmath129 , @xmath64 , and their statistical errors ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ firstly , consider experimental data described by , i = 1 , 2 , , n_n , n = 1 , 2 , , b. [ eqn : qni ] here the entire experimental possible energy range between @xmath130 and @xmath95 has been divided into @xmath131 bins with central points @xmath132 and widths @xmath133 . in each bin , @xmath134 events will be recorded .
since the recoil spectrum @xmath135 is expected to be approximately exponential , the following ansatz for the _ measured _ recoil spectrum ( _ before _
normalized by the exposure @xmath77 ) in the @xmath82th bin has been introduced @xcite : _ expt , ~ n _ expt , ~ q q_n ~ e^k_n ( q - q_s , n ) .
[ eqn : drdqn ] here @xmath136 is the standard estimator for @xmath61 at @xmath137 : r_n = , [ eqn : rn ] @xmath138 is the logarithmic slope of the recoil spectrum in the @xmath82th @xmath139bin , which can be computed numerically from the average value of the measured recoil energies in this bin : = - , [ eqn : bqn ] where ^ .
[ eqn : bqn_lambda ] the error on the logarithmic slope @xmath138 can be estimated from eq .
( [ eqn : bqn ] ) directly as ^2(k_n ) = k_n^4 ^-2 ^2 , [ eqn : sigma_kn ] with ^2 = .
[ eqn : sigma_bqn ] @xmath140 in the ansatz ( [ eqn : drdqn ] ) is the shifted point at which the leading systematic error due to the ansatz is minimal @xcite , q_s , n = q_n + . [ eqn : qsn ] note that @xmath140 differs from the central point of the @xmath82th bin , @xmath132 . from the ansatz ( [ eqn : drdqn ] ) , the counting rate at @xmath63 can be calculated by r ( ) = r_1 e^k_1 ( - q_s , 1 ) , [ eqn : rmin_eq ] and its statistical error can be expressed as ^2(r ( ) ) = r^2 ( ) , [ eqn : sigma_rmin ] since ^2(r_n ) = .
[ eqn : sigma_rn ] finally , since all @xmath141 are determined from the same data , they are correlated with ( i_n , i_m ) = _ a , [ eqn : cov_in ] where the sum runs again over all events with recoil energy between @xmath130 and @xmath95 . and the correlation between the errors on @xmath129 , which is calculated entirely from the events in the first bin , and on @xmath141 is given by ( r ( ) , i_n ) + = r ( ) ~ i_n ( , + b_1 ) + + ; [ eqn : cov_rmin_in ] note that the sums @xmath142 here only count in the first bin , which ends at @xmath143 .
[ [ statistical-errors-on-mchi-given-in-eqs.eqnmchi_rn-and-eqnmchi_rsigma ] ] statistical errors on @xmath11 given in eqs .
( [ eqn : mchi_rn ] ) and ( [ eqn : mchi_rsigma ] ) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ by using the standard gaussian error propagation , a lengthy expression for the statistical error on @xmath144 given in eq .
( [ eqn : mchi_rn ] ) can be obtained as .
( ) | _ = + ^1 / 2 .
[ eqn : sigma_mchi_rn ] here a short hand notation for the six quantities on which the estimate of @xmath11 depends has been introduced : c_1 , x = i_n , x , c_2 , x = i_0 , x , c_3 , x = r_x ( ) ; [ eqn : cix ] and similarly for the @xmath145 .
estimators for @xmath146 have been given in eqs .
( [ eqn : cov_in ] ) and ( [ eqn : cov_rmin_in ] ) .
explicit expressions for the derivatives of @xmath147 with respect to @xmath148 are : = _ n , x , [ eqn : drnx_dinx ] = - _ n , x , [ eqn : drnx_dizx ] and = + _ n , x ; [ eqn : drnx_drminx ] explicit expressions for the derivatives of @xmath69 with respect to @xmath145 can be given analogously .
note that , firstly , factors @xmath149 appear in all these expressions , which can practically be cancelled by the prefactors in the bracket in eq .
( [ eqn : sigma_mchi_rn ] ) . secondly , all the @xmath150 and @xmath151 should be understood to be computed according to eq .
( [ eqn : in_sum ] ) with integration limits @xmath130 and @xmath95 specific for that target .
similar to the analogy between eqs .
( [ eqn : mchi_rn ] ) and ( [ eqn : mchi_rsigma ] ) , the statistical error on @xmath152 given in eq .
( [ eqn : mchi_rsigma ] ) can be expressed as .
_ = + ^1 / 2 , [ eqn : sigma_mchi_rsigma ] where we have again used the short hand notation in eq .
( [ eqn : cix ] ) ; note that @xmath153 does not appear here .
expressions for the derivatives of @xmath154 can be computed from eq .
( [ eqn : rsigmax_min ] ) as = _ , x , [ eqn : drsigmax_dizx ] = _ , x ; [ eqn : drsigmax_drminx ] and similarly for the derivatives of @xmath80 .
[ [ covariance - of - f_i - defined - in - eqs.eqnfixa - and - eqnfixb ] ] covariance of @xmath87 defined in eqs .
( [ eqn : fixa ] ) and ( [ eqn : fixb ] ) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the entries of the @xmath89 matrix given in eq .
( [ eqn : cij ] ) involving basically only the moments of the wimp velocity distribution can be read off eq .
( 82 ) of ref .
@xcite , with an slight modification due to the normalization factor in eq .
( [ eqn : fixa ] ) defined in eq .
( [ eqn : fixb ] ) can be computed from the same basic quantities , i.e. , the counting rates at @xmath130 and the integrals @xmath155 , it can directly be included in the covariance matrix . ] : = _ m^2 + - ^j ( j+1 ) f_i cov(i_0 , i_j ) - ^i ( i+1 ) f_j cov(i_0 , i_i ) .
+ + d_i d_j ^2(r ( ) ) - ( r ( ) , i_0 ) . + . + [ eqn : cov_fi ] here we have defined _ m , [ eqn : calnm ] , [ eqn : td_alpha ] and d_i = , [ eqn : dia ] for @xmath83 ; and d_n_max+1 = .
[ eqn : dib ] 99 cdms collab .
, z. ahmed _ et al .
_ , arxiv:0912.3592 [ astro-ph.co ] ( 2009 ) .
cresst collab .
, r. f. lang _ et al .
_ , _ astropart .
phys . _ * 33 * , 60 ( 2010 ) ; cresst collab . , j. schmaler _ et al .
_ , _ aip conf .
proc . _ * 1185 * , 631 ( 2009 ) .
edelweiss collab .
, a. broniatowski _ et al .
_ , _ phys .
_ * b 681 * , 305 ( 2009 ) ; edelweiss collab . , e. armengaud _ et al .
_ * b 687 * , 294 ( 2010 ) . c. l. shan , _ aip conf .
proc . _ * 1200 * , 1031 ( 2010 ) ; arxiv:0910.1971 [ astro-ph.im ] ( 2009 ) .
p. sikivie and j. r. ipser , _ phys .
_ * b 291 * , 288 ( 1992 ) ; p. sikivie , _ phys
* b 567 * , 1 ( 2003 ) .
a. natarajan and p. sikivie , _ phys .
_ * d 73 * , 023510 ( 2006 ) ; _ phys .
_ * d 76 * , 023505 ( 2007 ) .
cogent collab .
, c. e. aalseth _
_ , arxiv:1002.4703 [ astro-ph.co ] ( 2010 ) .
d. hooper , j. i. collar , j. hall and d. mckinsey , arxiv:1007.1005 [ hep - ph ] ( 2010 ) . | in the earlier work on the development of a model independent data analysis method for determining the mass of weakly interacting massive particles ( wimps ) by using measured recoil energies from direct dark matter detection experiments directly , it was assumed that the analyzed data sets are background
free , i.e. , all events are wimp signals . in this article , as a more realistic study , we take into account a fraction of possible residue background events , which pass all discrimination criteria and then mix with other real wimp induced events in our data sets .
our simulations show that , for the determination of the wimp mass , the maximal acceptable fraction of residue background events in the analyzed data sets of @xmath0 total events is @xmath1 20% , for background windows of the entire experimental possible energy ranges , or in low energy ranges ; while , for background windows in relatively higher energy ranges , this maximal acceptable fraction of residue background events can not be larger than @xmath1 10% . for a wimp mass of 100 gev with 20% background events in the windows of the entire experimental possible energy ranges ,
the reconstructed wimp mass and the 1@xmath2 statistical uncertainty are @xmath3 ( @xmath4 for background free data sets ) .
march 2010 * effects of residue background events in direct dark matter + detection experiments on the determination of the wimp mass * + yu - ting chou@xmath5 and chung - lin shan@xmath6 + @xmath7 _ institute of physics , national chiao tung university + no . 1001 , university road , hsinchu city 30010 , taiwan , r.o.c . _ + _ e - mail : _ [email protected] + + @xmath8 _ department of physics , national cheng kung university + no . 1 , university road , tainan city 70101 , taiwan , r.o.c . _ + _ e - mail : _
[email protected] + + @xmath9 _ physics division , national center for theoretical sciences + no . 101 , sec . 2 , kuang - fu road , hsinchu city 30013 , taiwan , r.o.c . _ |
the drell - yan pair production remains up to now the theoretically cleanest way to access sea - quark distributions in hadrons .
necessary stage of such analysis is the measurement both with protons and neutrons , in which sea - quarks with different flavors are probed .
as the construction of a target containing free neutrons is not feasible , nuclear targets have to be employed . in the experiment that used the simplest nuclear target
the deuteron , the flavor asymmetry of the sea - quark distributions in the nucleon has been observed @xcite . at the same time
, analysis of the deep inelastic scattering off different nuclei shows that nucleon valence quark structure changes beyond mass - shell @xcite .
recent studies within the bethe - salpeter formalism point to the relativistic nature of this phenomenon .
relativistic effects should be manifested as modification of the sea quark distribution in the bound nucleon and , therefore , will change the contribution of the sea quark component in the nucleon structure extracted from the deuteron data .
this , in turn , affects estimations of the anti - quark component made by analyzing the deuteron and proton data . to clarify the role of the relativistic effects in the drell - yan ( dy ) pair production
, we study this process within the bethe - salpeter approach .
since the amplitude for the drell - yan scattering is defined by the imaginary part of the amplitude for the forward scattering @xmath2 in the cross - channel , we can obtain it within the bethe - salpeter formalism developed for dis @xcite . within this approach
the matrix element is defined by the nucleon space - time distribution and vacuum average of the t - product of the nucleon fields and nucleon electromagnetic current ( see ref .
@xcite ) . following this method
we get the following expression of the cross - section for the drell - yan scattering off deuteron : @xmath3 this expression gives the cross - section of the dy pair production off the deuteron in terms of the off - mass - shell dy nucleon cross - section @xmath4 and the nucleon distribution function @xmath5 .
this distribution function is defined by the bethe - salpeter amplitudes , and together with the denominator it composes the four - dimensional momentum distribution of the struck nucleon inside the deuteron carrying the total momentum @xmath6 . in that way eq .
( [ fermi4d ] ) expresses the nucleon blurring that results from the four - dimensional distribution of the nucleon inside the nucleus . by analogy with the non - relativistic @xmath7d momentum distribution
we call it four - dimensional fermi motion . due to the four - dimensional integration in eq .
( [ fermi4d ] ) actual calculations require information about nucleon cross - section @xmath4 in the kinematical region of the off - mass - shell values of the nucleon energy @xmath8 ( @xmath9 ) . the off - mass - shell behavior of @xmath4 is unobservable , since then explicit microscopic calculations of the cross - section have no experimental reference and strongly model dependent .
thus , eq . ( [ fermi4d ] ) has to be rewritten in terms of measurable quantities such as the cross - section for dy scattering off the physical nucleon .
the simplest solution of this problem is provided by the integration in eq .
( [ fermi4d ] ) with respect to @xmath8 .
analytical properties of the integrand in eq .
( [ fermi4d ] ) give a way to do it explicitly . within the assumption that @xmath10 and @xmath11 are regular with respect to @xmath8 the integrand contains a second order pole corresponding to the struck nucleon and a first order pole corresponding to the spectator .
these poles lie in the different half - planes of the complex plane @xmath8 .
so , we can choose one of the singularities to perform the integration . to express the nuclear hadron tensor in terms of the nucleon structure functions
it is necessary to choose the second order pole ( @xmath12 ) that corresponds to the struck nucleon .
the result of the contour integration in vicinity of the second order pole in eq.([fermi4d ] ) is defined by the derivative of the pole residue with respect to @xmath8 at the point @xmath13 .
doing the integration and using the relation in the bjorken limit @xmath14 we get the following expression for the cross section of the dy pair production off the deuteron : @xmath15 \frac{f^{n / d}(m_d,{\bf p})}{8m_{d}e_{\rm n}^2{\delta^{\rm n}_{\rm d}}^2},\label{s2a}\ ] ] where @xmath16 is the bjorken @xmath0 of the struck nucleon , @xmath17 is the nucleon on - shell energy , @xmath18 is equivalent to the total energy shift of the struck nucleon due to the binding and fermi motion .
the function @xmath19 together with the denominator composes the three dimensional momentum distribution of the struck nucleon inside the deuteron . according to the normalization condition for the bethe - salpeter vertex function this distribution satisfies the baryon and momentum sum rules @xcite and coincides with the usual nuclear momentum distribution .
thus , the first term in eq .
( [ s2a ] ) , which results from the derivative of the propagator of the nuclear spectator , expresses contribution of the conventional @xmath7d - fermi motion . the second term with @xmath20 results from the nucleon dis amplitude derivative : @xmath21 this term is responsible for the deviation from unity of the ratio of the cross sections .
its contribution is proportional to the coefficient @xmath22 ; which , therefore , characterizes contribution of the four - dimensional fermi - motion .
it is important to note that the struck nucleon pole gives the factor @xmath23 , since then the contribution from all other singularities ( for example nucleon self - energy cut or anti - nucleon pole ) are suppressed at least as @xmath24 @xcite .
since the mean value of the energy shift is small ( @xmath25 ) , eq.([s2a ] ) provide deuteron drell - yan cross - section with accuracy up to terms of order @xmath26 .
using the gluck - reya - voght ( grv ) parameterization @xcite for the @xmath4 and solution of the bethe - salpeter equation with the graz - ii separable kernel @xcite as @xmath19 we make numerical calculation fo the ratio @xmath27 .
it is important to note that the full grv parameterization is obtained within the approximation @xmath28 , therefore it already implies isospin asymmetry of the sea - quark distribution in the nucleon in order to describe the dy data . to check influence of the relativistic nuclear effects on the deviations of the ratio @xmath27 from unity we use the grv parameterization without quark - sea asymmetry ( @xmath29 ) .
the results of the calculation are presented in fig.[fig1 ] .
the dashed curve represents the calculation within the approximation with @xmath28 , the solid curve represent calculation with the four - dimensional fermi - motion taken into account .
it is clear from the comparison with the new fnal data @xcite , that the relativistic kinematics can reproduce the deviation of the cross - section ration without the flavor asymmetry of the nucleon sea .
it is important to note , that the @xmath30-dependence of the deviation is defined by the coefficient @xmath31 in front of the cross - section derivative same as in case of dis .
thus , we can conclude that observed flavor asymmetry results from the relativistic kinematics of bound nucleons ; the bound nucleon structure change have the same nature in the dis and dy processes ; the effect has the same @xmath30-dependence as emc - effect .
a more detailed study of the @xmath30-dependence of the effect and high-@xmath0 data can shed more light on the nature of the sea - quark flavour asymmetry in the nucleon .
i would like to thank g. smirnov , u. mosel , h. toki and o.lynnik for useful discussions .
i also thank research center for nuclear physics of the osaka university for the support and warm hospitality . the work is made under partial support of the fenu innovation educational programme in the framework of the national project `` education '' . | the study presented in this paper shows that accounting for the relativistic structure of the deuteron allows to explain the ratio of the drell - yan pair production cross - section at the low bjorken @xmath0 off the deuteron and the proton .
thus , the sea quark distributions in the nucleon should be studied with accounting for the effects of the relativistic structure of the deuteron .
the suggested approach reduces theoretical uncertainty in extracting the ratio @xmath1 from the data and it is important for the clarification of the nature of the sea quark asymmetry in the nucleon . |
in the quasi - elastic regime ( qe ) , the doubly - differential cross section , in which a neutrino carrying initial four - momentum @xmath2 scatters off a nuclear target to a state of four - momentum @xmath3 can be written in born approximation as follows : [ elem ] = l _ w_a^ , where @xmath4 is the fermi constant and @xmath5 is the ckm matrix element coupling @xmath6 and @xmath7 quarks .
the leptonic tensor is completely determined by lepton kinematics , whereas the nuclear tensor @xmath8 , containing all the information on strong interactions dynamics , describes the response of the target nucleus , @xmath9 where @xmath10 and @xmath11 are the initial and final hadronic states carrying four momenta @xmath12 and @xmath13 and @xmath14 is the nuclear electroweak current operator ; the sum includes all hadronic final states . in the impulse approximation ( ia ) scheme ,
the nuclear current can be written as a sum of one - body currents , i.e. @xmath15 , while the final state reduces to the direct product of the hadronic state produced at the weak vertex ( with momentum @xmath16 ) and that describing the @xmath17-nucleon residual system , with momentum @xmath18 : @xmath19 .
following ref .
@xcite , the final expression of the hadronic tensor can be cast in the following form : w_a^&= & d^3pde p(*p*,e ) w^(p , q ) , [ hadtensor ] where @xmath20 and the function @xmath21 is the target _ spectral function _
i.e. the probability distribution of finding a nucleon with momentum @xmath22 and removal energy @xmath23 in the target nucleus .
the quantity @xmath24 is the tensor describing the weak interactions of the @xmath25-th nucleon in free space ; the effect of nuclear binding of the struck nucleon is accounted for by the replacement @xmath26 with @xmath27 .
the second argument in the hadronic tensor is @xmath28 .
the rfg @xcite model , widely used in montecarlo simulations , provides the simplest form of the spectral function , @xmath29 where @xmath30 is the fermi momentum and @xmath31 is the average binding energy , introduced to account for nuclear binding .
thus , in this model @xmath30 and @xmath31 are two parameters that are _ adjusted _ to reproduce the experimental data . for oxygen , the analysis of electron scattering data yields @xmath32 mev and @xmath33 mev . within the relativistic mean field approximation we refer to the model described in @xcite , where , like in the previous cases , the nuclear current is written as a sum of single - nucleon currents
the wave functions for the target and the residual nuclei are described in terms of an independent - particle model .
then , the transition matrix elements can be cast in the following form : @xmath34 where @xmath35 and @xmath36 are the wave functions for initial bound and final outgoing nucleons , respectively , and @xmath37 is the relativistic current operator . in particular
, the relativistic bound - state wave functions ( for both initial and outgoing nucleons ) are obtained as a solution of the dirac equation , in the presence of the same relativistic nuclear mean field potential , derived from a lagrangian containing @xmath38 , @xmath39 and @xmath40 mesons .
the last model we want to take into account has been introduced in @xcite , where the hadronic tensor is expressed in terms of the nuclear response functions treated in the random phase approximation ( rpa ) .
the response functions are related to the imaginary part of the corresponding full polarization propagators and the introduction of the rpa approximation means that the polarization propagators are the solutions of integral equations involving the bare propagators and the effective interaction between particle - hole excitations . within this formalism
, the authors of @xcite were able to show that multinucleon terms sizably increase the genuine charged current qe cross section in such a way to reproduce the miniboone results @xcite .
the mechanism responsible for the enhancement that brings the theoretical cross section into agreement with the data is multi nucleon knock out , leading to two particle - two hole ( 2p2h ) nuclear final states . in the following
, we will refer to this `` generalized '' qe cross section as rpa-2p2h whereas we adopt the short rpa for the genuine qe cross section . to summarize this section , we present in fig.[fig : xsect ] a comparison of the five total qe cross sections for the @xmath41 process ( left panel ) and @xmath42 ( right panel ) , in the energy range @xmath43 $ ] gev .
the curves have been computed using the dipole structure of the form factors and , in particular , a value of the axial mass close to @xmath44 gev .
_ total charged current qe cross sections for the @xmath45 ( left panel ) and @xmath46 ( right panel ) processes in the energy range @xmath43 $ ] gev._,title="fig : " ] _ total charged current qe cross sections for the @xmath45 ( left panel ) and @xmath46 ( right panel ) processes in the energy range @xmath43 $ ] gev._,title="fig : " ] as it can be easily seen in the left panel , the rfg prediction sizably overestimates the sf , rmf and rpa results by roughly @xmath47 , a fact that is well known to happen also for many other models with a more accurate description of the nuclear dynamics than the rfg approach . on the other hand ,
the inclusion of 2p2h contributions largely enhances the qe cross section in the rpa approximation for energies above @xmath48 gev , although it is smaller than the rfg at smaller energies . for antineutrinos ( right panel )
the observed pattern is almost the same .
to estimate the impact of different models of the cross section on the measurement of @xmath0 and leptonic cp violation we choose a @xmath49 @xmath50-beam facility as a representative example because the neutrino flux from such a facility spans up to @xmath51 gev with the peak around @xmath52 gev and is thus mostly sensitive to the quasielastic region explored here .
details of the detector response can be found in ref .
it is important to notice that we are only using the quasielastic contribution to the neutrino cross section depicted in fig.[fig : xsect ] . as an illustration we have focused on the dependence on the nuclear model adopted of two different observables , namely the _ cp and @xmath0 discovery potentials _ , defined as the values of the cp - violating phase @xmath53 and @xmath0 for which respectively the hypothesis of cp conservation @xmath54 or @xmath55
can be excluded at 3@xmath38 after marginalizing over all other parameters .
the cp discovery potential is shown in the left panel of fig.[fig : disc ] , where we superimposed the results obtained using the rfg cross section and the sf , rmf and rpa calculations .
_ left panel : cp discovery potential in the @xmath56 plane .
solid lines refer to the sf model , dotted lines to the rfg , short - dashed lines to the rmf , dot - dashed to rpa and long - dashed to rpa-2p2h .
right panel : @xmath0 discovery potential._,title="fig : " ] _ left panel : cp discovery potential in the @xmath56 plane .
solid lines refer to the sf model , dotted lines to the rfg , short - dashed lines to the rmf , dot - dashed to rpa and long - dashed to rpa-2p2h .
right panel : @xmath0 discovery potential._,title="fig : " ] we clearly see that , for @xmath57 ( where the sensitivity is maximal ) the rfg model gives a prediction which is around a factor 2 better than the sf , rmf and rpa models ( which , as expected , behave almost in the same way ) in @xmath58 and around a @xmath59 better than the rpa-2p2h .
this is not surprising because the @xmath50-beam facility used in our simulations mainly probes energies smaller than @xmath60 gev , where the rfg is still larger than any other model ( see fig.[fig : xsect ] ) . for the other points in the parameter space , the difference is less evident but still significant .
the difference in the sensitivity to @xmath0 can be seen in the right panel of fig .
[ fig : disc ] where we show the results in the @xmath61-plane . in this case
the predictions of the sf , rmf and rpa models differ by up to a factor of @xmath62 compared to the rfg model for @xmath57 while the difference is less pronounced for @xmath63 .
the rpa-2p2h results are more similar to the rfg for @xmath63 and to the other models for @xmath64 .
this is also evident in the right panel where we show the cp fraction . finally , it is interesting to comment on the effect of using different nuclear models also in the simultaneous determination of @xmath0 and @xmath53 .
we have performed such an analysis for the input value @xmath65 ( not including degenerate solutions coming from our ignorance of the octant of @xmath66 @xcite and in the neutrino mass ordering ) finding that , using the rfg and rpa-2p2h models , we are able to reconstruct the true values of @xmath0 and @xmath53 within reasonable uncertainties , whereas with the other models we can only measure two distinct disconnected regions ( the fake one around the value of @xmath0 and @xmath64 ) , which worsen the global sensitivity on those parameters . in conclusion
, we have analyzed the impact of five different theoretical models for the neutrino - nucleus charged current qe cross sections on the forecasted sensitivity of the future neutrino facilities to the parameters @xmath0 and @xmath53 .
we have found that the sensitivities computed with the fg are better , by up to a factor of 2 , with respect to the other models .
we acknowledge miur ( italy ) , for financial support under the contract prin08 .
9 o. benhar , n. farina , h. nakamura , m. sakuda and r. seki , phys .
d * 72 * , 053005 ( 2005 ) [ arxiv : hep - ph/0506116 ] ; o. benhar and d. meloni , nucl .
a * 789 * , 379 ( 2007 ) [ arxiv : hep - ph/0610403 ] ; o. benhar and d. meloni , phys .
d * 80 * , 073003 ( 2009 ) [ arxiv:0903.2329 [ hep - ph ] ] . c. maieron , m. c. martinez , j. a. caballero and j. m. udias , phys .
c * 68 * , 048501 ( 2003 ) [ arxiv : nucl - th/0303075 ] ; m. c. martinez , p. lava , n. jachowicz , j. ryckebusch , k. vantournhout and j. m. udias , phys .
c * 73 * , 024607 ( 2006 ) [ arxiv : nucl - th/0505008 ] . m. martini , m. ericson , g. chanfray and j. marteau , phys .
c * 80 * , 065501 ( 2009 ) [ arxiv:0910.2622 [ nucl - th ] ] ; m. martini , m. ericson , g. chanfray and j. marteau , phys .
c * 81 * ( 2010 ) 045502 [ arxiv:1002.4538 [ hep - ph ] ] .
j. burguet - castell , d. casper , e. couce , j. j. gomez - cadenas and p. hernandez , nucl .
b * 725 * , 306 ( 2005 ) [ arxiv : hep - ph/0503021 ] .
d. meloni , phys .
lett . * b664 * , 279 - 284 ( 2008 ) .
[ arxiv:0802.0086 [ hep - ph ] ] . | in recent years the experimental study of neutrino oscillations has much contributed to our knowledge of particle physics by establishing non vanishing neutrino masses and by measuring or constraining the corresponding mixing angles . within the domain of neutrino oscillations , the main goal of the next generation of facilities
is the measurement of the mixing angle @xmath0 and the observation of leptonic cp violation , for which we have no hints at the moment .
we discuss more in detail how various models of neutrino - nucleus cross section affect the forecasted precision measurement of @xmath0 and the cp violating phase @xmath1 @xcite .
+ _ contribution to nufact 11 , xiiith international workshop on neutrino factories , super beams and beta beams , 1 - 6 august 2011 , cern and university of geneva ( submitted to iop conference series ) _ |
[ cols= " < , < " , ] after fixing compensator structure , coefficients @xmath0 are obtained by applying straightforward least squares optimization .
+ we should emphasize here that fitting has to be done for both real and imaginary part of @xmath1 $ ] , thus the actual compensator structure is twice that depicted in figure [ fig : compensator_structure ] .
+ simulation parameters for system * s * are as follows : symbol rate @xmath2 , carrier frequency @xmath3 , with 64qam input symbol sequence .
nonlinear distortion subsystem * f * of * s * , used in simulation , is defined in , where the delays @xmath4 are given by the vector @xmath5 $ ] , with @xmath6 .
digital simulation of the continuous part of * s * was done by representing continuous signals by their discrete counterparts , obtained by sampling with high sampling rate @xmath7 .
as input to * s * , we assume periodic 64qam symbol sequence , with period @xmath8 .
this period length is used for generating input / output data for fitting coefficients @xmath0 , as well as generating input / output data for performance validation .
+ in figure [ fig : evm_comparison ] we present evm obtained for different compensator structures , as well as output evm with no compensation , and case with ideal compensator @xmath9 .
as can be seen from figure [ fig : evm_comparison ] , compensator fitted using the proposed structure in figure [ fig : compensator_structure ] outperforms other compensators , and gives output evm almost identical to the ideal compensator .
this result was to be expected , since model in figure [ fig : compensator_structure ] approximates the original system * s * very closely , and thus is capable of approximating system @xmath10 closely as well .
this is not the case for compensators modeled with simple volterra series , due to inherently long ( or more precisely infinite ) memory introduced by the lti part of * s*. even if we use noncausal volterra series model ( i.e. @xmath11 ) , which is expected to capture true dynamics better , we are still unable to get good fitting of the system * s * , and consequently of the compensator @xmath12 .
+ advantage of the proposed compensator structure is not only in better compensation performance , but also in that it achieves better performance with much more efficient strucuture .
that is , we need far less coefficients in order to represent nonlinear part of the compensator , in both least squares optimization and actual implementation ( table [ table ] ) . in table
[ table ] we can see a comparison in the number of coefficients between different compensator structures , for nonlinear subsystem parameter value @xmath13 .
data in the first column is number of coefficients ( i.e. basis elements ) needed for general volterra model , i.e. coefficients which are optimized by least squares .
the second column shows actual number of coefficients used to build compensator .
least squares optimization yields many nonzero coefficients , but only subset of those are considered significant and thus used in actual compensator implementation .
coefficient is considered significant if its value falls above a certain treshold @xmath14 , where @xmath14 is chosen such that increase in evm after zeroing nonsignificant coefficients is not larger than 1% of the best achievable evm ( i.e. when all basis elements are used for building compensator ) . from table
[ table ] we can see that for case 3 volterra structure , 10 times more coefficients are needed in order to implement compensator , than in the case of our proposed structure . and
even when such a large number of coefficients is used , its performance is still below the one achieved by this new compensator model .
in this paper , we propose a novel explicit expression of the equivalent baseband model , under assumption that the passband nonlinearity can be described by a volterra series model with the fixed degree and memory depth .
this result suggests a new , non - obvious , analytically motivated structure of digital precompensation of passband nonlinear distortions caused by power amplifiers , in digital communication systems .
it has been shown that the baseband equivalent model can be written as a series connection of a fixed degree and low memory volterra model , and a long memory discrete lti system , called reconstruction filter .
frequency response of the reconstruction filter is shown to be smooth , hence well aproximable by low order polynomials .
parameters of such a model ( and accordingly of the predistorter ) can be obtained by applying simple least squares optimization to the input / output data measured from the system , thus implying low implementation complexity .
state of the art implementations of dpd , have long memory requirements in the nonlinear subsystem , but structure of our baseband equivalent model suggests that the long memory requirements are shifted from the nonlinear part , to the lti part , which consists of fir filters and is easy to implement in digital circuits , giving it advantage of much lower complexity .
we also argued that this baseband model , and its corresponding dpd structure , can be readily extended to ofdm modulation .
simulation results have shown that by using this new dpd structure , significant reduction in nonlinear distortion caused by the rf pa can be achieved , while utilizing full frequency band , and thus effectively using maximal input symbol rate .
the authors are grateful to dr .
yehuda avniel for bringing researchers from vastly different backgrounds to work together on the tasks that led to the writing of this paper .
l. ding , g. t. zhou , d. r. morgan , z. ma , j. s. kenney , j. kim , and c. r. giardina , `` a robust digital baseband predistorter constructed using memory polynomials , '' _ ieee trans .
1 , pp.159 - 165 , january 2004 .
d. r. morgan , z. ma , j. kim , m. zierdt , and j. pastalan , `` a generalized memory polynomial model for digital predistortion of rf power amplifiers , '' _ ieee trans . signal process .
3852 - 3860 , october 2006 .
a. zhu , j. c. pedro , and t. j. brazil , `` dynamic deviation reduction - based volterra behavioral modeling of rf power amplifiers , '' _ ieee trans .
theory techn .
4323 - 4332 .
, december 2006 .
a. zhu , p. j. draxler , j. j. yan , t. j. brazil , d. f. kimball , and p. m. asbeck , `` open - loop digital predistorter for rf power amplifiers using dynamic deviation reduction - based volterra series , '' _ ieee trans .
theory techn .
56 , no . 7 , pp . 1524 - 1534 . ,
july 2008 .
m. rawat , k. rawat , and f. m. ghannouchi , `` adaptive digital predistortion of wireless power amplifiers / transmitters using dynamic real - valued focused time - delay line neural networks , '' _ ieee trans .
theory techn .
95 - 104 , january 2010 .
m. rawat , k. rawat , f. m. ghannouchi , s. bhattacharjee , and h. leung , `` generalized rational functions for reduced - complexity behavioral modeling and digital predistortion of broadband wireless transmitters , '' _ ieee trans .
485 - 498 , february 2014 . c. yu , l. guan , and a. zhu , `` band - limited volterra series - based digital predistortion for wideband rf power amplifiers , '' _ ieee trans .
theory techn .
4198 - 4208 , december 2012 .
f. wang ; d.f .
kimball , d.y .
lie , p.m. asbeck , and l. e. larson , `` a monolithic high - efficiency 2.4-ghz 20-dbm sige bicmos envelope - tracking ofdm power amplifier , '' _ ieee j. solid - state circuits _ , vol.42 , no.6 , pp.1271,1281 , june 2007 .
j. reina - tosina , m. allegue - martinez , c. crespo - cadenas , c. yu , and s. cruces , `` behavioral modeling and predistortion of power amplifiers under sparsity hypothesis , '' _ ieee trans .
theory techn .
745 - 753 , february 2015 .
o. tanovic , a. megretski , y. li , v. stojanovic , and m. osqui , `` discrete - time models resulting from dynamic continuous - time perturbations in phase - amplitude modulation - demodulation schemes , '' _ proceedings of the european control conference 2015 _ , pp . 3571 - 3576 , july 15 - 17 , 2015 . | we consider discrete - time ( dt ) systems s in which a dt input is first transformed to a continuous - time ( ct ) format by phase - amplitude modulation , then modified by a non - linear ct dynamical transformation f , and finally converted back to dt output using an ideal de - modulation scheme . assuming that f belongs to a special class of ct volterra series models with fixed degree and memory depth
, we provide a complete characterization of s as a series connection of a dt volterra series model of fixed degree and memory depth , and an lti system with special properties .
the result suggests a new , non - obvious , analytically motivated structure of digital compensation of analog nonlinear distortions ( for example , those caused by power amplifiers ) in digital communication systems .
we also argue that this baseband model , and its corresponding digital compensation structure , can be readily extended to ofdm modulation .
results from a matlab simulation are used to demonstrate effectiveness of the new compensation scheme , as compared to the standard volterra series approach .
* key words : * communication system nonlinearities , nonlinear systems , modeling , phase modulation , amplitude modulation |
the frequency spectrum of observations of the integrated brightness of the sun is often referred to as the solar acoustic spectrum @xcite , as it is mainly used to study pressure or sound waves inside the sun .
these sound waves can be observed on the surface of the sun as oscillation pattens which can be represented as spherical harmonics of the eigenfunctions of the sound waves inside the sun .
these sound waves lead to a series of distinct peaks in the solar acoustic spectrum ( see fig . 1 ) and by studying the frequencies , amplitudes and line - widths of these peaks we can through helioseismology obtain knowledge of the physics that takes place inside the sun ( see e.g. * ? ? ?
* ) , but more information are hidden in the solar acoustic spectrum . at lower frequency
the solar acoustic spectrum contains signals from spots , granulation and faculae .
in fact the acoustic signal from granulation and faculae continues up to the frequencies of the p - mode oscillations and even to higher frequencies and therefore this signal is often referred to as the acoustic background .
studies of sun - like stars have shown that other stars show a similar acoustic background ( see e.g. * ? ? ?
* ; * ? ? ?
@xcite could be the first to suggest that the solar acoustic background was modelled with a number of exponentially decaying functions .
the idea is that the signal from e.g. granulation is understood as a random signal with some memory and thus its autocorrelation function should be given by an exponentially decaying function defined by a characteristic time - scale and an amplitude @xcite .
long uninterrupted observations from the south pole and from the @xmath1 satellite have later revealed that the solar acoustic background contains signals from many more phenomena than just granulation .
@xcite identified a total of 6 difference components in the acoustic spectrum from observations of the ca ii k line in the solar atmosphere from the south pole .
the identified components were , from lower to higher frequency : non - periodic fluctuations attributed to active region evolution ; non - periodic fluctuations attributed to granulation overshoot ; non - periodic fluctuations attributed to chromospheric bright points ; periodic fluctuations attributed to the photosphere ( i.e. p - mode oscillations ) and periodic fluctuations attributed to the chromosphere ( i.e. chromospheric oscillations ) . whereas the signals from granulation and p - mode oscillations are reported in all studies ,
the signals associated with the chromosphere are only reported in the study by @xcite , as this is the only study that analyzes the acoustic background in observations of the ca ii lines .
the feature between the granulation signal and the p - mode oscillations has on the other hand also been reported in most studies , especially studies based on intensity measurements , but an agreement on the origin of this features has not been reached .
@xcite analyzed observations of solar irradiance from the virgo instrument on @xmath1 and identified not only a granulation component in the frequency spectrum , but also components from active regions , super- and meso - granulations and bright points .
the characteristic time - scale of the component that @xcite identified as bright points is similar to the characteristic time - scale of the component that @xcite identified as chromospheric bright points .
identified the same component in integrated light observations from virgo , but speculated that this component might be due to a second granulation population as shall be showen later this is not likely . bright points and faculae are related phenomena and they are both related to the magnetic network on the solar surface .
the difference is that whereas bright points are mainly seen in the inter - granular lanes , faculae are seen inside the granules and that whereas bright points can maintain their brightness over several tens of minutes @xcite , faculae change significantly on granular evolution time - scales @xcite .
for this reason it is more likely that the component between the granulation and p - mode oscillation components is caused by faculae rather than bright points . in the picture of @xcite faculae
are deffined as `` _ the edges of granules _ seen through the ' ' forest `` of the magnetic field in plages and the network '' .
the faculae does in other words occur in a localized region of the granules and it is thus clear that the characteristic time - scale of faculae will be a bit shorter than that of granulation . there are no quantitative studies that directly relates the time - scale of granulation to the time - scale of faculae based on high - resolution observations . a qualitative argument for that the characteristic time - scale of faculae is shorter than that of granulation is given in the small movie in fig . 9 of @xcite .
this small movie of a 3d radiative magneto convective simulation shows 3 distinct granules that live throughout the full length of the movie ( 330 sec ) and a number of faculae on these granules .
if one look e.g. at the facular at ( 0@xmath23,1@xmath24 ) it is clearly visible at @xmath3 sec , but completely gone at @xmath4 sec . at @xmath5 sec a new facular starts to evolve just to the right of where the other was located and at @xmath6 sec , this facular has its peak brightness . in other words , though the evolution of faculae follows the evolution of the granules it is clear from the movie that the same granule can , during its lifetime , contain a number of faculae
i.e. the characteristic time - scale of faculae is shorter than that of granulation .
a few studies have used hydrodynamical models of stellar atmospheres to simulate acoustic spectra of sun - like stars , but so far no such studies have been made using _
magneto_hydrodynamical models , which is needed in order to reproduce the facular component . in this paper we will use observations from the green channel on the virgo instrument on @xmath1 @xcite to analyze the acoustic signature of faculae .
we will show how the characteristic time - scale and amplitude that we measure of this component in the acoustic background are consistent with measurements based on high - resolution observations of the solar surface .
we also analyze any temporal variations in the different components in the solar acoustic background including the facular component .
the paper is arranged as follows . in section 2
we give a short theoretical description of the acoustic background and present the model that we use to model the granulation , the faculae and the p - mode oscillations components in the acoustic spectrum .
this model is used in the analysis in section 3 , where we also describe a statistical test of the significant of the facular component in the acoustic background .
section 4 presents the results of the analysis and concluding remarks are found in section 5 .
the paper also includes an appendix that describe a test based on monte carlo simulations of the statistical method developed in section 3 .
the original formulation of the solar acoustic background by @xcite is the following : @xmath7 where @xmath8 is the power density at frequency @xmath9 , @xmath10 is the amplitude of the signal and @xmath11 is the characteristic time - scale .
1 is normalized to parseval s theorem so that the variance of the signal in the time domain equals the sum of @xmath8 over all frequencies : @xmath12 which means that @xmath10 in eq .
1 does in fact reflects the variance of the signal in the time domain @xcite . note that every component in the acoustic background will contribute with the signal given by eq . 1 to the spectrum
. the spectrum will thus be a sum of individual signals from i.e. granulation and faculae each defined by unique values of @xmath10 and @xmath11 .
unfortunately , the original model by @xcite fails to reproduce the observed solar acoustic background for frequencies higher than the atmospheric acoustic cut - off frequency .
the reason for this is that granulation can not be modelled with turbulent cascades as it is done in the drift model by @xcite .
turbulence shows a distribution with a slope of around @xmath13 in power , convection on the other hand has a lower limit in the time domain on which changes can take place .
this means that on small time - scales ( or at high frequency ) convection is not noisy whereas turbulence is and therefore the acoustic background decays with a slope smaller than @xmath13 . in the sun the decay rate turns out to be around @xmath14 at frequencies higher than the atmospheric acoustic cut - off frequency . taking this into account the harvey model extends to : @xmath15 @xcite used this model to successful describe the granulation and facular components in the solar acoustic background between 400 @xmath0hz and up to 8000 @xmath0hz , which is close to the nyquist frequency .
@xcite used another approach to solve the problem with the fast decline at high frequency and considered the slope as a free parameter . when this is done the decline
will often be faster then @xmath13 and this will solve the problems at high frequency . in this way @xcite found slopes of @xmath16 and @xmath17 for the granulation and the bright point components , respectively .
the slope gives the decay rate , which calibrates the amount of memory in the process responsible for the noise @xcite and there is no reason that this value should be exactly @xmath13 , or that is should be the same for the different components . on the other hand , there is a physical argument why the slope should not be the same at low and high frequency ( which were given above ) , so allowing the slope to be a free parameter only partly solves the problem .
we have therefore chosen another approach and modelled only the lower frequency part of the spectrum ( up to 3200 @xmath0hz ) .
this means that no extra component is needed in order to model the high - frequency part .
the following model is thus used to model the granulation and facular signals in the solar acoustic background between 100 and 3200 @xmath0hz : @xmath18 the envelope of the p - mode oscillations shows variability on a weekly time - scale with an amplitude of 6.2% , which is properly mainly due to changes in the damping rates of the oscillation modes @xcite .
the envelope of the p - mode oscillations therefore needs to be modelled along with the acoustic background in order to disentangle changes in the damping rates of the oscillation modes from changes in the acoustic background .
a gaussian envelope is therefore also included in the model in order to model the envelope of the p - mode oscillations , but as the high frequency part of the spectrum is not analyzed it is not necessary to include a white noise background .
of course the background will have some effect on the amplitudes that are measured of the granulation and facular components , but as the white noise level in the virgo observations is generally around three orders of magnitudes smaller than the granulation and facular amplitudes this effect is negligible . a gaussian has here been chosen for the envelope of the p - mode oscillations , but one or two lorentzians could also have been chosen .
this would how ever not affect the final results significantly and it is not clear which functional form that agrees best with the behavior of the sun and other sun - like stars in general . figs .
2 & 3 show examples of fits to the solar acoustic background using the models described above either containing or not containing a facular component .
the observations and the calculation of the spectrum are described in section 4 .
the data analysis consists of three steps ; calculating the power density spectra , modelling the power density spectra and evaluating the significance of the different measured parameters and the goodness of the different models . for the analysis we use a 13-year time series from the green channel on the virgo instrument of @xmath1 .
there are some gaps in this time series , especially around the @xmath1 vacation and recent measurements do suffer from increased noise level . in the analysis performed here we have divided the full 13-year time series into substrings of one month each . in order to get the lowest probability for artefacts such as gaps and increased noise levels to affect the data analysis we have performed a discrimination of both bad data points and bad substrings .
bad data points were identified by calculating a logarithmic running box variance log @xmath19 : @xmath20 ^ 2}\right),\ ] ] where @xmath21 is the @xmath22th observation , @xmath0 is the mean value of the observations between @xmath23 and @xmath24 and @xmath25 is the width of the running box which was set equal to 10 points ( @xmath26 = 21 ) .
using this formulation all data points with a logarithmic running box variance that deviated more than 5@xmath10 from the mean were then removed .
this removed around 5% of the data points .
the reason for using the logarithmic variance is that many data points show artificially low scatter . after removing the bad data
points all monthly substrings with a duty cycle lower than 95% were also removed .
this left us with 119 monthly substrings out of 147 possible .
we did test that the relative high demands for selecting good data points did not affect the results .
the test showed slightly divagating results when using either the 119 good monthly substrings or all the 147 substrings , but the differences were insignificant .
the power density spectra were calculated using least - squares .
each spectrum was normalized by the effective observation length given as the reciprocal of the area under the window function in order to convert the spectra into power density spectra .
acoustic backgrounds have earlier been modelled by smoothing the spectra with a gaussian running mean , with a width of a few hundred @xmath0hz , and then assuming that the differences between the smoothed observed spectrum and the model followed a normal distribution , which allowed a comparison utilising least squares ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
though this works in practice , there are no mathematically formal arguments for that the differences between the smoothed observed spectrum and the model should follow a normal distribution . a method for modelling the observed spectra that is statistically valid and that does not require the spectra to be smoothed with a gaussian running mean
is therefore presented below . in order to model the solar acoustic background
we make use of maximum likelihood estimators ( see e.g. * ? ? ?
* and references herein ) .
the purpose is to calculate a logarithmic likelihood function @xmath27 between @xmath26 independent measurements @xmath28 i.e. power density at a given frequency , and a model @xmath29 given by a set of parameters @xmath30 . adopting the notation of and the logarithmic likelihood function
can be calculated as : @xmath31 where @xmath32 is the probability density function .
the position of the minimum of @xmath27 in @xmath30 space will give us the most likely value of @xmath30 .
the so - called formal error bars can be calculated as the diagonal elements of the inverse of the hessian matrix @xmath33 : @xmath34 in order to calculate the probability density function we use the formulation by where a binned version of the original spectrum is compared to the model .
the formulation by can be used on any model @xmath29 and not just for calculating the significance of peaks in the spectrum .
we thus calculate the binned spectrum @xmath35 by summing @xmath28 over @xmath36 bins : @xmath37 following the probability density function is then given as : @xmath38=\frac{\mu_i^{\nu_i}}{\gamma(\nu_i)}\mathcal{s}_i(x , n)^{\nu_i-1}e^{\mathcal{s}_i(x , n)},\ ] ] where @xmath39 is the gamma function and @xmath40 and @xmath41 are given as : @xmath42 and @xmath43 ^ 2}{\displaystyle\sum\limits_{k = i - n/2}^{k = i+n/2}{f_k^2({\bf \lambda})}}.\ ] ] @xmath36 appears in @xmath0 and @xmath9 in order to get the right amplitudes . using the formalism above the model in eq .
4 , either containing or not containing a facular component , can be compared to an observed spectrum .
the model containing a facular component is based on a total of 9 free parameters
i.e. three from granulation , three from faculae and three from the p - mode oscillations , whereas the model not containing a facular component is based on only 6 free parameters . having the maximum likelihood of each model from eq .
6 we are then able to calculate the statistical significance of the facular component in the acoustic spectrum by calculating the logarithmic likelihood ratio @xmath44 : @xmath45 in our case @xmath46 equals 6 and @xmath47 equals 3 . following @xcite we can then compare the value of @xmath13ln@xmath44 to a @xmath48 distribution with @xmath47 degrees of freedom and in this way calculate the significance of the facular component in the acoustic spectrum .
.acoustic background parameters [ cols="<,^,^",options="header " , ] [ tab1 ]
two kinds of results are presented : one set of results on the average spectrum of the 119 one months substrings and one set of results on the temporal variability in the spectra of the individual substrings .
the analysis is made by binning 100 frequency bins ( @xmath49 ) , which gives a frequency resolution in the binned spectra of around 39 @xmath0hz .
error bars for all parameters were calculated as the diagonal elements of the inverse of the hessian matrix ( see eq . 7 ) the average spectrum from the 119 one month substrings is modelled with a model containing a facular component and a model not containing a facular component .
the results from the modelling are given in table 1 and the average observed spectra with the models overlaid are shown in figs . 2 & 3 . as the decay rates were treated as free parameters in both models the decay rates for the granulation component are different in the two models .
this leads to differences in the models at high frequency as the obtained decay rate @xmath50 of @xmath51 @xmath52 0.1 for the granulation component in the model without a facular component is too small to describe the acoustic background at high frequency .
a number of studies have used high - resolution observations of the solar surface to measure the amplitude and time - scale of granulation and faculae , but in general there is little agreement between the different results .
the reason for this is probably differences in the way the amplitude and the time - scale are measured , but also differences in the resolution of the observations , differences in the solar activity level at the time of the observations and whether the observations were taken close to the disk centre or close to the limb .
the high - resolution observations can therefore only be used as a rough comparison to the seismic results . if a typical size of granules ( or cell area ) of 1.2 mm@xmath53 is assumed , then around 2.5 million granules are found on the visible solar surface .
the contrast of a typical granule is around 20% .
the typical contrast of faculae is around 50% at the peak distance from the disk center ( @xmath54 ) @xcite and the relative fraction of granules that show faculae is around 1.8% @xcite .
if each granule on the solar surface is assumed to cause a signal of relative amplitude @xmath55 at a random time and the entire solar surface is covered with granules , then this will cause a time series of integrated sun light with a variance of : @xmath56 where @xmath57 is the number of granules present on the visible solar surface at any given time .
the amplitudes of the granulation component were measured to be @xmath58 and @xmath59 ppm for the model containing a facular component and the model not containing a facular component , respectively .
this is consistent with the theoretical estimate of around 126 ppm that is found by dividing 20% with the square root of 2.5 million granules .
the amplitude of the intensity signal from faculae follows that of granules , except that it can not be assumed that the entire solar surface is covered with faculae . in the most simple approach the amplitude
should therefore be scaled with the relative coverage of faculae @xmath60 : @xmath61 the amplitude of the facular component was measured to be @xmath62 ppm .
this agrees roughly with the theoretic estimate of around 42 ppm obtained by multiplying @xmath6350% with the square root of @xmath631.8% and dividing it with the square root of 2.5 million granules .
the characteristic time - scales of convection were measured to be @xmath64 and @xmath65 sec for the model containing a facular component and the model not containing a facular component , respectively .
this is consistent with the coherence time - scale between 156 and 246 sec found by del moro ( 2004 ) . for faculae
a characteristic time - scale of @xmath66 sec was measured which is consistent with the fact that it should be shorter than the granulation time - scale .
the amplitude and characteristic time - scale of the granulation and facular components in the acoustic spectra are consistent with the high - resolution measurements .
this indicates that the interpretation that the second bump in the acoustic spectrum originates from faculae is consistent though the comparison does not make the argument alone .
the interpretation is also based on the relative relations between the amplitude and characteristic time - scale of granulation , faculae and p - mode oscillations .
13 shows why the component in the acoustic spectrum between the granulation and p - mode oscillation components does not originate from a second granulation population as suggested by .
this suggestion is based on the work by who finds , using high - resolution images of quiet granulation , a second population of granules with a shorter coherence time , but also smaller brightness contrasts .
the fact that this second population of granules is both more numerous and have smaller brightness contrasts than ordinary granules means that it should not be visible in the acoustic spectrum as the numerator of eq .
13 would be smaller and the denominator would be larger than for ordinary granulation .
not fwhm ) of 0.4 year . ]
the significance of the facular component in the average observed spectrum can be calculated from the likelihood ratio between the two different models .
the significance was measured to be 83% .
this means that though the model with the facular component and a total of 9 free parameters does describe the average observed spectrum better than the model without the facular component and 6 free parameters , it does not describe it significantly better .
visually , it is clear that the fit in fig . 2 is better than the fit in fig . 3 , but given the statistic of the observed spectra and the number of free parameters it is not significant better .
in order to analyze why the facular component is not significant in the average observed spectrum we calculate the significance of the facular component in the spectrum of each individual one month substring .
the decay rates and characteristic time - scales for the granulation and facular components and the frequency of maximum power for the p - mode oscillation component had to be fixed to the values given in the first column of table 1 .
if this was not done the location and slope of the exponential decaying function would change in order to make up for changes in amplitude .
this means that the models used for modelling the 119 individual one month substrings contained 3 and 4 free parameters , respectively .
the logarithmic likelihood ratios between the models with and without a facular component for the 119 individual one month substrings are shown in fig .
it is seen that most of the logarithmic likelihood ratios are between 0 and 10 with a few scatter points below and above .
the figure does reveal a few points with logarithmic likelihood ratios below 0 .
a negative logarithmic likelihood ratio means that in these months the model without the facular component describes the observed spectrum better than the model with the facular component , or in other words , that adding more free parameters to the model makes a worse agreement between the observed spectrum and the model for these months .
the reason for this is that the decay rates of the granulation components are not the same in the two different models and for some months the slow decay rate in the model without the facular component describes the observed spectrum better than the fast decay rate in the model with the facular component .
these scatter points with logarithmic likelihood ratios below 0 could be the reason for the facular component not being significant in the average observed spectrum .
clearly , a lot of variability is seen in fig . 4 , but no trend known to follow the visibility or amplitude of faculae is seen .
in particular no clear variability that follows the solar cycle is seen ( see fig . 5 for comparison ) .
the reason for this could either be that the measured ratios are not accurate enough to show intrinsic facular variability or that the acoustic signal from faculae depends stronger on the number of available granules than on the fraction or contrast of faculae ( see eq .
6 shows the distribution of the significances of the facular signal measured in the 119 one month substrings assuming that the logarithmic likelihood ratios follow a @xmath48 distribution with one degree of freedom .
this distribution shows a clear increase towards a significance of one that falls of rapidly down to a significance of around 0.97 and then have a long tail down to smaller significances ( also smaller significances than what is shown on the figure ) .
the conclusion from the figure is that the acoustic signal from faculae is significantly present in the observed solar acoustic background 47% of the time .
this figure support our impression that the acoustic signal from faculae is present on the sun most of the time and that the reason that the average spectrum does not shows a significant facular signal is because of a few months where the significance is really low
i.e. where the model without the facular component describes the observed spectrum better .
the same effect is seen in the spectrum calculated from the full 13 years of observations
i.e. that the acoustic signal from faculae in not significantly present in the spectrum .
again , the reason for this is most likely that the signal is not present all the time , which reduces the significance of the signal when analyzing the full 13 years of observations .
distribution with one degree of freedom .
it is seen that 56 months ( 47% ) have a significances higher than 0.99 , while 75 ( 63% ) have a significance higher than 0.95 . ] both the frequencies , amplitudes and line - widths of the individual p - mode oscillation modes in the acoustic spectrum are known to change with the solar cycle ( see e.g. : * ? ? ?
* ; * ? ? ?
* ) and we therefore also expect the average shape and position of the p - mode oscillation envelope to change with the solar cycle . the temporal variability of the height and width of the p - mode oscillations envelope is shown in fig
. 7 for the model with and without a facular component ( the position of the envelope is kept fixed in both models ) .
there is a good agreement between the results from the two different models with a linear pearson correlation coefficient of 0.95 ( 0.97 for the smoothed curves ) for the height and 0.83 ( 0.70 for the smoothed curve ) for the width .
a clear anti - correlation is seen between the height of the envelope and the solar cycle ( see fig . 5 for comparison ) so that the height has minimum around 2000 , which coincides with the peak of the solar cycle .
the anti - correlation between the height of the envelope and the solar cycles is consistent with other studies ( i.e. * ? ? ?
* ; * ? ? ?
* ) and can be explained by the fact that the life - time of the p - mode oscillation is known to decrease with increasing solar activity @xcite .
the envelope width does not show any correlation with the solar cycle .
the position of the envelope , which is not analyzed here , as it was fixed to the average value , does only show small changes compared to the hight and width @xcite .
the uncertainties on the parameters measured in the average spectrum are comparable in size to the uncertainties measured in the individual one month substrings .
this reflects that the scatter seen in i.e. figs .
7 & 8 is much larger than the uncertainties on the individual measurements in other words it reflects that the parameters does show intrinsic variability on a monthly time - scale .
the temporal variability of the amplitude of the granulation component does not show any clear correlation with the solar cycle , neither for the model with or without the facular component .
the amplitudes of the granulation component show an offset between the two different models and only some agreement is seen in the temporal variability with a linear pearson correlation coefficient of 0.74 ( 0.84 for the smoothed curves ) .
instead some signs of a quasi - annual modulation is seen , especially in the results from the model with the facular component .
the ratio between the variance of the measurements ( the scatter ) and the mean value of the uncertainties is 5.12 for the model with and 4.97 for the model without the faculae component , indicating that the variability is not due to noise .
a quasi - annual modulation of both the granulation characteristic time - scale and amplitude was also seen in the analysis of observations from the golf instrument on @xmath1 by in phase with the variability seen here .
explained the modulation by the fact that the observation altitude ( or depth ) of the golf instrument in the solar photosphere changes with the observation wavelength which again changes with the orbital velocity of @xmath1 , which changes with a period of one year .
as the virgo observations analyzed here are intensity observations , the effect of a wavelength change should not be prominent , but the quasi - annual variability is still seen , especially around solar activity maximum .
though the facular component is not significantly present in the acoustic spectra of all the one month substrings the amplitude of the component can still be measured in all the substrings .
these amplitudes are plotted in fig .
the uncertainties are relatively large for these measurements , and the measured amplitudes show only little scatter within these uncertainties .
the amplitude of the scatter is around a few ppm .
the gaussian running mean of the measurements ( represented with the black line ) show a quasi - biennial variability with maximum around 2001 , 2003 , 2005 and 2007 .
the amplitude of this variability is around one ppm
i.e. significantly lower than the scatter of the individual points .
it is therefore not clear if the quasi - biennial variability is intrinsic to the sun or if it is just random noise .
a similar quasi - biennial variability was seen in the residuals of the p - mode oscillations frequency shifts in the study by @xcite . in fig .
10 we therefore compare the residuals of the measured amplitudes of the facular component with the residuals of the p - mode oscillations frequency shifts from @xcite . in order to do that we have subtracted a 3rd order polynomial from the measurements in fig . 9 and
binned the one month measurements to the temporal resolution of the residuals of the p - mode oscillations frequency shifts ( 182.5 days ) .
the two sets of measurements show a correlation with a linear pearson correlation coefficient of 0.34 .
the reason that no strong correlation is seen could however be the relative large uncertainties on both sets of measurements and it can thus not been ruled out that the faculae and the p - mode oscillations frequency shifts have the same origin .
@xcite speculate that the quasi - biennial variability seen in the residuals of the p - mode oscillations frequency shifts originates from a second solar dynamo located not at the so - called tachocline at the base of the convection zone , but in the strong shear layer just below the surface .
this idea of a near surface dynamo , which has also been suggested by @xcite , could also explain the short periodic magnetic variability that has recently observed in other sun - like stars @xcite .
if the presents of faculae and p - mode oscillation frequency shifts are indeed correlated ( which we can however not claimed based on the present measurements ) then this could suggests that the presence of faculae on the surface of the sun is controlled not only by the 11-years solar cycles , which is expected to be located in the solar tachocline , but also by a quasi - biennial dynamo located in the strong radial shear layer just below the solar surface @xcite , which suggest that the variability in the amplitude of the facular component could be caused by a second dynamo .
in general it has to be noted that the quasi - annual and quasi - biennial periods , seen in the granulation and facular amplitudes , are very peculiar
. varying instrumental effects such as altitude , pointing , temperature or the like could be the culprit .
one of the things that were done in order to test this was to redo the analysis using only the monthly substrings where the facular signal had a significance larger than 95 and 99% and this did not change any of the conclusions on the facular signal - i.e. the same kind of periodicity were still seen .
13 years of observations of the integrated brightness of the sun from the virgo instrument on @xmath1 have been analyzed .
three distinct features are seen in the frequency spectrum of these observations . by comparing the measured characteristic time - scale and amplitude of these phenomena to high - resolution observations of the solar surface
it is shown that the three features can convincingly be explained as granulation , faculae and p - mode oscillations .
the feature in the acoustic background related to faculae is the weakest and it is not significantly present in the average spectrum made from 119 individual one month spectra .
the feature is on the other hand present in 56 ( 47% ) out of the 119 one month spectra with a significance larger than 0.99 and in 75 ( 63% ) with a significance larger than 0.95 .
the temporal variability of the amplitudes of the granulation , facular and p - mode oscillations components in the solar acoustic background have also been analyzed .
the p - mode oscillations show a clear anti - correlation with the solar cycle and the amplitude of the granulation component show some signs of an quasi - annual variability , which is also seen in velocity observations , but no correlation with the solar cycle .
the amplitude of the facular component does also not show any correlation with the solar cycle , but instead some signs of a quasi - biennial variability is seen .
the residuals of the temporal variability of the amplitude of the facular component show a correlation with a linear pearson correlation coefficient of 0.34 with the residuals of the p - mode oscillations frequency shifts from the study by @xcite .
i would like to thank w.j .
chaplin and h. kjeldsen for useful discussions , j. christensen - dalsgaard for suggesting that i wrote this paper and t. appourchaux for patiently explaining to me how the significance of the facular signal could be tested .
i also thank a.b .
broomhall for providing me with the residuals of the p - mode oscillations frequency shifts from bison .
the carlsberg foundation is acknowledged for financial support .
99 aigrain , s. , favata , f. , & gilmore , g. 2004 , stellar structure and habitable planet finding , 538 , 215 anderson , e. r. , duvall , t. l. , jr . , & jefferies , s. m. 1990 , apj , 364 , 699 appourchaux , t. 2003 , a&a , 412 , 903 appourchaux , t. 2004 , a&a , 428 , 1039 appourchaux , t. , gizon , l. , & rabello - soares , m .- c . 1998 , a&as , 132 , 10 baudin , f. , samadi , r. , appourchaux , t. , & michel , e. 2007 , arxiv:0710.3378 berger , t. e. , schrijver , c. j. , shine , r. a. , tarbell , t. d. , title , a. m. , & scharmer , g. 1995 , apj , 454 , 531 berger , t. e. , rouppe van der voort , l. , lfdahl , m. g. 2007 , apj , 661 , 1272 brandenburg , a. 2005 , apj , 625 , 539 bhm - vitense , e. 2007 , apj , 657 , 486 chaplin , w. j. , elsworth , y. , isaak , g. r. , miller , b. a. , & new , r. 2000 , mnras , 313 , 32 chaplin , w. j. , elsworth , y. , isaak , g. r. , et al .
2003 , apj , 582 , l115 chaplin , w. j. , et al .
2010 , apj , 713 , l169 de pontieu , b. , carlsson , m. , stein , r. , rouppe van der voort , l. , lfdahl , m. , van noort , m. , nordlund , . , & scharmer , g. 2006 , apj , 646 , 1405 del moro , d. 2004 , a&a , 428 , 1007 fletcher , s. t. , broomhall , a .- m . , salabert , d. , basu , s. , chaplin , w. j. , elsworth , y. , garcia , r. a. , & new , r. 2010 , apj , 718 , l19 frhlich , c. , et al .
1997 , sol .
phy . , 170 , 1 frhlich , c. 2009 , a&a , 501 , l27 handberg , r. , & campante , t. l. 2011 , a&a , 527 , a56 harvey , j. 1985 , future missions in solar , heliospheric and space plasma physicenvelopes , 235 , 199 harvey , j. w. , duvall , t. l. , jr . , jefferies , s. m. , & pomerantz , m. a. 1993 , gong 1992 . seismic investigation of the sun and stars , 42 , 111 houdek , g. , balmforth , n. j. , christensen - dalsgaard , j. , & gough , d. o. 1999 , a&a , 351 , 582 garca , r. a. , mathur , s. , salabert , d. , ballot , j. , rgulo , c. , metcalfe , t. s. , & baglin , a. 2010 , science , 329 , 1032 gough , d. o. , et al . 1996 , science , 272 , 1296 iben , i. , jr .
, & mahaffy , j. 1976 , apj , 209 , l39 karoff , c. 2008 , phd dissertation , aarhus university kjeldsen , h. , et al .
2008 , apj , 682 , 1370 lefebvre , s. , garca , r. a. , jimnez - reyes , s. j. , turck - chize , s. , & mathur , s. 2008 , a&a , 490 , 1143 libbrecht , k. g. , & woodard , m. f. 1990 , nature , 345 , 779 lomb , n. r. 1976 , ap&ss , 39 , 447 ludwig , h .-
2006 , a&a , 445 , 661 ludwig , h .-
, et al .
2009 , a&a , 506 , 167 metcalfe , t. s. , basu , s. , henry , t. j. , soderblom , d. r. , judge , p. g. , knlker , m. , mathur , s. , & rempel , m. 2010 , apj , 723 , l213 michel , e. , et al .
2008 , science , 322 , 558 nordlund , a. , spruit , h. c. , ludwig , h .-
g . , & trampedach , r. 1997 , a&a , 328 , 229 toutain , t. , & appourchaux , t. 1994 , a&a , 289 , 649 trampedach , r. , christensen - dalsgaard , j. , nordlund , a. , & stein , r. f. 1998 , the first mons workshop : science with a small space telescope , 59 uitenbroek , h. , tritschler , a. , & rimmele , t. 2007 , apj , 668 , 586 vzquez rami , h. , rgulo , c. , & roca corts , t. 2005 , a&a , 443 , l11 wedemeyer - bhm , s. , & rouppe van der voort , l. 2009 , a&a , 503 , 225 wilks , s. s. 1938 , ann .
stat , 9 , 60
in order to test the robustness of the statical analysis developed in section 3 , a number of monte carlo simulations were made . in the simulations
it was assumed that the acoustic spectrum followed the model given in eq . 4 and an acoustic spectrum was simulated for the parameters given in table 1 . using an inverse fourier transform
the acoustic spectrum was convented from the frequency to the time domain , where realistic random noise were added at each time - step .
hereafter the acoustic spectrum was convented back into the frequency domain , where it was analyzed the background parameters and the significance of the facular signal measured .
the two parameters that were changed in the simulation were the amplitude of the facular component and the noise level all other parameters were fixed to the values given in table 1 .
this means that for each set of facular amplitude and noise level the simulation returned not only measured facular amplitude and noise level , but also measured values for all the other parameters in table 1 , which could be used to evaluate the precision of the procedure .
the amplitude of the facular component was changed between 45 and 70 ppm and the noise level from 0 to 80 ppm per measurement ( equivalent to the point - to - point scatter in the time series ) . the main result of this monte carlo simulations is shown in fig .
a1 , which shows the relation between the amplitude of the facular component and the logarithmic likelihood ration between the model with and without the facular component .
it is seen that there is no linear correlation between the two .
this is a bit surprising as a spectrum with a hight amplitude facular component is expected to have a higher likelihood with a model with a facular component than a spectrum with a low amplitude facular component would have .
this is also the case , but the problem is that there is no linear relation between the likelihood and the amplitude of the facular component for the model that does not include the facular component . in other words , is the reason why no linear correlation is seen that though the likelihood between the model and the observations increases linear as a function of facular amplitude for the model with the facular component , the same ( or more correctly the opposite ) is not true for the model without the facular component .
this does first of all explain why some sings of variability can be seen in fig .
9 , which shows the measured facular amplitude as a function of time , while not in fig . 4 , which shows the likelihood ration as a function of time .
it does also explain why we can measure the amplitude and characteristic time - scale of the facular component and the uncertainties on these parameters even in months where the component is not significant .
the reason is that in order to measure the amplitude and characteristic time - scale we only use the model that includes the facular component , while in order to measure the likelihood ratio we also uses the model that does not include the facular component .
no trends were seen between the measured parameters and the noise level in the simulations , except for increased scatter .
for some of the simulations with the highest noise levels it was not possible to get the model to convert to the observed spectrum .
the monte carlo simulations have also been used to test the uncertainties on the different parameters - especially the parameters related to the facular component .
this is done by plotting a histogram of the difference between the parameter values used in the simulation and the measured parameter values divided by the uncertainty on the measured parameter values ( @xmath67 ) .
if the errors on the parameters were random distributed around a mean value we would expect that these histograms would all be gaussian function with a variance of one .
this is also the case for the parameters in table 1 ( the parameters related to the facular component , is shown in fig .
a2 ) , though the variance might be a bit larger than one , this is not unexpected as the uncertainties are only formal uncertainties . in order to get more reliable uncertainties one would eventual have to preform a full bayesian analysis as it is sometimes done when modeling the p - modes using markov chain monte carlo . | the integrated brightness of the sun shows variability on time - scales from minutes to decades .
this variability is mainly caused by pressure mode oscillations , by granulation and by dark spots and bright faculae on the surface of the sun . by analyzing the frequency spectrum of the integrated brightness
we can obtain greater knowledge about these phenomena .
it is shown how the frequency spectrum of the integrated brightness of the sun in the frequency range from 100 to 3200 @xmath0hz shows clear signs of both granulation , faculae and p - mode oscillations and that the measured characteristic time - scales and amplitudes of the acoustic signals from granulation and faculae are consistent with high - resolution observations of the solar surface . using 13 years of observations of the sun s integrated brightness from the virgo instrument on the @xmath1 satellite
it is shown that the significance of the facular component varies with time and that it has a significance above 0.99 around half the time .
furthermore , an analysis of the temporal variability in the measured amplitudes of both the granulation , faculae and p - mode oscillation components in the frequency spectrum reveals that the amplitude of the p - mode oscillation component shows variability that follows the solar cycles , while the amplitudes of the granulation and facular components show signs of quasi - annual and quasi - biennial variability , respectively .
[ firstpage ] sun : faculae , plages sun : granulation sun : helioseismology sun : oscillations hz ) , the facular component ( around 1800 @xmath0hz ) and the p - mode oscillations component ( around 3000 @xmath0hz ) are seen as changes in the slope of the spectrum .
the different components are more clearly marked in fig .
2 .
it is also seen that the spectrum declines faster at frequencies higher than the p - mode oscillations or the atmospheric acoustic cut - off frequency .
the spectrum was calculated from 13 years of observations of the sun s integrated brightness from the virgo instrument on the @xmath1 satellite and the peak at 5555 @xmath0hz is an artefact . ] |
understanding the ground state properties of spin systems on a lattice is of central importance in many - body physics , but at the same time , it is a highly challenging problem in many scenarios .
an important step in understanding its difficulty has been the insight that computing e.g. the ground state energy of a classical spin system is , in general , an complete problem @xcite : while the energy of any given spin configuration can be easily computed , finding the configuration with minimal energy is in general a difficult task it can be as hard as any problem in , i.e. , any problem whose solution can be efficiently verified . for quantum spin systems , an additional difficulty arises : generally , we can not even expect to have an efficient description of the ground state .
thus , it seems that the only statement we can make about the difficulty of the problem is that given a quantum register with the ground state , we will be able to efficiently estimate its energy using a quantum computer .
indeed , it has been shown that this is the best we can say , as the problem of estimating the ground state energy of a quantum system is a complete problem for the class @xcite , the quantum analogue of : it contains all problems which have a quantum solution which can be efficiently checked on a quantum computer , and thus , determining the ground state energy of a quantum spin system is as hard as any of these problems ; in fact , the problem retains its hardness even when restricted to two - dimensional lattices of qubits with nearest - neighbor interactions @xcite or one - dimensional chains @xcite .
it is an interesting question to understand the reasons underlying the additional complexity of quantum spin systems as compared to classical systems . to this end ,
restricted version of the problem which lie between classical and general quantum spin hamiltonians have been studied : for instance , it has been shown that so - called _ stoquastic _ hamiltonians , where all off - diagonal elements are negative , have a complexity which lies in between and , as those systems can be related to classical random processes @xcite ; in fact , these are exactly the hamiltonians which allow for quantum monte carlo simulations as they do not exhibit a sign problem .
another restricted class of hamiltonians are _ commuting hamiltonians _ , that is , hamiltonians which can be written as a sum of mutually commuting few - body terms . for those systems
, all terms can be simultaneously diagonalized , just as for classical systems ; however , the corresponding eigenbasis can be highly entangled , making it unclear whether a useful classical description of the ground state can be provided .
in fact , commuting hamiltonians encompass systems which exhibit rich non - classical physics , in particular models with topological order and even anyonic excitations , such as kitaev s toric code and quantum double models @xcite , or levin and wen s string net models @xcite .
commuting hamiltonians are also of interest since the fixed points of renormalization flows in gapped phases are expected to be commuting hamiltonians , and thus understanding their structure might give insight into the structure of gapped quantum phases . finally , understanding the complexity of commuting hamiltonians is of interest in quantum complexity , as it might be a step towards a quantum pcp theorem , which would assess how the difficulty of estimating the ground state energy is related to the desired accuracy which is integer for commuting projectors . what is know about the complexity of finding the ground state energy of commuting hamiltonians , or rather , of determining whether the ground state minimizes each term in the hamiltonian individually the commuting hamiltonian problem ? for lattices in two and higher dimensions , commuting hamiltonian is an hard problem , as it e.g. encompasses the ising model on a planar graph @xcite . on the other hand ,
it is not clear whether the general commuting hamiltonian problem is _ inside _ , since it is not clear in general how to provide an efficiently checkable description of the ground state . for two - local ( i.e. , two - body ) hamiltonians , bravyi and vyalyi @xcite
have shown that the problem is in by using @xmath0algebraic techniques ( their result also implies that one - dimensional commuting hamiltonians are efficiently solvable ) ; subsequently , aharonov and eldar @xcite have proven containment in for hamiltonians with three - body interactions both for qubits on arbitrary graphs , and qutrits on nearly - euclidean interaction graphs .
in both of these works , the classical certificates do not only prove the problem to be in , but can in fact be used to construct constant depth quantum circuits which _ prepare _ the ground state .
this , in particular , implies that the corresponding hamiltonians including qutrits with three - body interactions can not exhibit topological order @xcite . on the other hand , kitaev s toric code , which is the ground state of a commuting hamiltonian with four - body interactions of qubits ,
does have topological order , and thus , we can not expect any approach which yields a low - depth circuit to work beyond three qutrits . in this paper
, we study the commuting hamiltonian problem on a square lattice of qubits with plaquette - wise interactions , and prove that it is in .
that is , we consider a square lattice of qubits , with a hamiltonian with mutually commuting terms acting on the four qubits adjacent to each plaquette , and show that the problem of deciding whether its ground state minimizes the energy of each local term in the hamiltonian is in : i.e. , in case the ground state has this property , a classical certificate exists which can be checked efficiently by a quantum computer .
our approach differs considerably from the aforementioned approaches in that the certificate can not be used to devise a quantum circuit for preparing the ground state , and is thus also applicable to systems with topological order ; it should be noted that the same holds true for the proof in ref .
@xcite that commuting hamiltonian with factorizing projectors is in .
we will consider a 2d square lattice with spins on the vertices , and either open or periodic boundary conditions .
the hamiltonian @xmath1 consists of plaquette terms @xmath2 which act on the four spins adjacent to the plaquette @xmath3 , and we impose that all the terms in the hamiltonian commute , @xmath4=0\ \forall\,p , q$ ] .
as the hamiltonian terms commute , there is a basis of eigenstates of @xmath5 which are simultaneously eigenstates of all the @xmath2 .
we would like to know the computational difficulty of the following problem , called commuting hamiltonian : is there an eigenstate @xmath6 of @xmath5 which minimizes the energies for all @xmath2 individually , i.e. , are the ground states of @xmath5 also ground states of each @xmath2 ? in the following , we will show that in the case of qubits , the existence of such a state can be proven within , i.e. , there is a classical certificate which proves the existence of such a @xmath6 , and which can be checked efficiently classically .
note that on the other hand , it is clear that the problem is hard , as it e.g. encompasses classical ising spin glasses in a field which are known to be hard even for two - level systems @xcite . for the following
, it will be useful to reformulate commuting hamiltonian as follows : define the _ local ground state projectors _
@xmath7 as the projectors onto the ground state subspace of @xmath2 ; the @xmath7 commute again , @xmath8=0 $ ]
. then , @xmath9 is the projector onto the subspace spanned by the states which are ground states of all @xmath2 . since commuting
hamiltonian asks whether such states exist , it is equivalent to asking whether @xmath10 .
we start by coloring the plaquettes of the square lattice black and white in a checkerboard pattern , and denote the set of black and white plaquettes by @xmath11 and @xmath12 , respectively .
let @xmath13 be the projectors onto the joint ground state space of the black and white @xmath2 , respectively ; then , commuting hamiltonian corresponds to determining whether @xmath14 .
this is equivalent to asking whether @xmath15\ne 0\ ] ] ( this can be seen using eigendecompositions of @xmath16 and @xmath17 ) , and we will consider this formulation of the problem from now on ; to prove commuting hamiltonian is contained in , we therefore need to show that a classical certificate for the validity of ( [ eq : commham - trace - formulation ] ) can be provided .
a helpful example to keep in mind in the first part of our discussion is kitaev s toric code @xcite : there , @xmath18 for @xmath19 , and @xmath20 for @xmath21 , with @xmath22 and @xmath23 the pauli matrices . in the following ,
let us study the structure of each layer individually ( we will w.l.o.g .
choose black ) . to this end , we will use a result of bravyi and vyalyi based on @xmath0-algebraic techniques @xcite ; a detailed explanation of those techniques can also be found in @xcite . the basic insight from ref .
@xcite is the structure of two commuting projectors .
consider two projectors @xmath24 and @xmath25 acting on a space @xmath26 ; the two operators overlap on @xmath27 . now consider the schmidt decompositions @xmath28 i.e. @xmath29=0 $ ] for @xmath30 , and similarly for @xmath31 , @xmath32 , and @xmath33 .
then , @xmath34=0 $ ] implies that @xmath35=0 $ ] for all @xmath36 , and thus , @xmath31 and @xmath32 span commuting @xmath0-algebras , cf .
@xcite . using the standard form of finite @xmath0-algebras
, it follows that the space @xmath27 has a canonical decomposition @xmath37 where the @xmath31 span the full matrix algebra on @xmath38 while acting trivially on the rest , and correspondingly for @xmath32 and @xmath39 .
this shows that the space @xmath27 can be cut into direct sum `` slices '' ( the @xmath40 ) such that in each slice , @xmath41 and @xmath42 act on independent subsystems .
more formally , there exists a decomposition @xmath43 of @xmath44 , with @xmath45 the projectors onto @xmath46 , such that @xmath47=0 \mbox { and } [ \pi_\alpha , r]=0\ ] ] and thus @xmath48 where @xmath49 and @xmath50 act on different subsystems @xmath38 and @xmath51 , i.e. , factorize .
note that the above decomposition allows to compute the @xmath45 and thus the @xmath52 and @xmath53 efficiently .
each vertex in the black sublattice is acted upon by exactly two commuting projectors @xmath7 ; thus , we can apply the preceding argument to all vertices to find decompositions @xmath54 , @xmath55 , of the hilbert space at each vertex @xmath56 , such that @xmath17 projected onto the slice @xmath57 factorizes , @xmath58 ( this implicitly defines the @xmath59 ) , and @xmath17 can be written as @xmath60 note that in the sum on the right hand side , we implicitly regard the tensor products as being canonically embedded into the full hilbert space , and we will use this convention in the following . the analogous decomposition can be performed for the white sublattice , yielding a ( in general different ! ) decomposition @xmath61 note that in order to distinguish the decomposition for the the black and the white layer , we strictly use labels @xmath62 and @xmath63 , respectively ; moreover , we denote the projectors decomposing the white layer by @xmath64 .
e.g. , for kitaev s toric code the @xmath65 are projectors onto the @xmath23 eigenstates , and the @xmath66 onto the @xmath22 eigenstates . using the stucture of @xmath17 and @xmath16 , eqs .
( [ eq : black - decomposition ] ) and ( [ eq : white - decomposition ] ) , we can rewrite the commuting hamiltonian problem , eq .
( [ eq : commham - trace - formulation ] ) , as @xmath67\ ; \ ] ] recall that we regard the tensor products as being canonically embedded into the full hilbert space .
since each of the traces is positive ( as it is the trace of the product of two positive operators ) , the above is equivalent to the existence of @xmath62 and @xmath63 such that @xmath68 , where @xmath69\ .\ ] ] thus , we can ask the prover to provide us as a certificate with some @xmath62 and @xmath63 for which @xmath70 ; if we can moreover show that @xmath71 can be computed efficiently ( or rather in ) , this will prove that commuting hamiltonian is in .
note that @xmath71 can be interpreted as the overlap of the ( unnormalized ) states @xmath72 and @xmath73 , both of which are tensor products of states supported on individual plaquettes , but with different partitions in the two layers .
computing such an overlap can in general be as hard as contracting projected entangled pair states ( peps ) @xcite , i.e. , -hard @xcite : any peps can be expressed this way by using one layer for the bonds and the other for the peps projections . of course , the fact that these states arise from two commuting layers @xmath17 and @xmath16 yields additional constraints , and we will show in the following that those constraints allow for the efficient evaluation of @xmath71 in the case of qubits . for kitaev s toric code ,
e.g. , we could choose @xmath74 : this yields @xmath75 = 2^{-n}\ne 0\ , \ ] ] proving the existence of a zero - energy ground state ; note that this certificate does not carry any information on how to prepare the state .
let us now show that for a lattice of qubits , the overlap @xmath76 can be computed efficiently . to this end
, we will show that the computation of the overlap can be decomposed into a product of overlaps of one - dimensional structures which can be computed efficiently .
let us first consider the black layer . for each vertex @xmath56 , the decomposition ( [ eq : cstardec ] ) of the local hilbert space can either be trivial ( no direct sum ) or non - trivial , @xmath77 . in the former case
, this implies that at most one of the adjacent plaquette terms @xmath7 acts non - trivially of vertex @xmath56 ; in the latter case , the sum consists of exactly two _ one - dimensional _
projectors @xmath65 , making use of the fact that the hilbert space at each site is a qubit , i.e. , two - dimensional .
we will denote the set of vertices with a non - trivial decomposition in the black layer by @xmath78 , and in the white layer by @xmath79 ( in which the one - dimensional projectors are labelled @xmath66 ) . in @xmath71 ,
all vertices in @xmath80 contribute only a one - dimensional subspace and thus can be traced out : that is , all vertices in @xmath81 can be removed ( taking care whether the overlap of the one - dimensional projectors is non - vanishing ) , while for vertices where only one layer has a one - dimensional decomposition , this yields new effective plaquette terms @xmath82 in the other layer by projecting the original plaquette terms @xmath7 onto that one - dimensional subspace ; thus , the problem of checking whether @xmath71 is non - zero reduces to computing the overlap of the new effective plaquette terms @xmath82 .
formally , this reads @xmath83 \\[2ex ] & \stackrel{\llap{\mbox{\scriptsize(a)}}}{= } { \mathrm{tr}}\bigg[\ , \underbrace { \big(\!\prod_{w\in { \mathcal{f}}_w}\!\!\bar\pi^w_{\beta_w } \bigotimes_{p\in { \mathcal{p}}_b}\!\ ! \pi_p^{\vec\alpha } \prod_{w\in { \mathcal{f}}_w}\!\!\bar\pi^w_{\beta_w}\big)}_{(*)}\ ; \big(\!\!\!\prod_{v\in { \mathcal{f}}_b\backslash { \mathcal{f}}_w } \!\!\!\!\!\!\!\pi^v_{\alpha_v } \bigotimes_{p\in { \mathcal{p}}_w}\!\ ! \pi_p^{\vec\beta } \prod_{v\in { \mathcal{f}}_b\backslash{\mathcal{f}}_w } \!\!\!\!\!\!\!\pi^v_{\alpha_v } \big)\ ; \,\bigg]\nonumber\\[2ex ] & \stackrel{\llap{\mbox{\scriptsize(b)}}}{= } \prod_{v\in{\mathcal{f}}_b\cap { \mathcal{f}}_w}\!\!\!\!\ ! { \mathrm{tr}}\big[\pi^v_{\alpha_v}\bar\pi^v_{\beta_v}\big]\ \times\ { \mathrm{tr}}\bigg [ \,\,\big(\!\!\bigotimes_{p\in { \mathcal{p}}_b } \rho_p\,\big ) \big(\!\!\bigotimes_{p\in { \mathcal{p}}_w } \rho_p\,\big ) \bigg]\ .
\label{eq : afterb}\end{aligned}\ ] ] here , we have used in step ( a ) that for all @xmath84 , @xmath85 and anologously for the black layer ; in step ( b ) , we have defined new effective plaquette terms @xmath82 by virtue of @xmath86\ , \ ] ] and correspondingly for the white plaquettes ; the first factor in ( [ eq : afterb ] ) takes care of the terms in @xmath81 in @xmath87 .
note that the @xmath82 are now only supported on those vertices not in @xmath80 , as those have been traced out .
the task of checking whether @xmath88 has thus been reduced to checking this for ( [ eq : afterb ] ) : for the first term , this can be clearly done efficiently , and so , it remains to prove that the overlap @xmath89\ ] ] can be computed efficiently .
note that the @xmath82 are now supported on plaquettes of a square lattice with vertices missing .
moreover , while the @xmath82 do no longer commute , in each layer at most one @xmath82 acts non - trivially on each vertex ; we will make use of this fact repeatedly in the following .
the situation encountered in computing the overlap @xmath90 is depicted in fig .
[ fig:1d-2d - chains ] . here
, the dots in each plaquette denote the vertices on which @xmath82 acts non - trivially ( the lines just connect the vertices involved in @xmath82 ) .
if the @xmath82 on adjacent plaquettes act non - trivially on the same qubit ( we will say they `` overlap '' ) , they form connected structures which we need to contract in order to evaluate @xmath90 . for one - dimensional structures as the one on in fig .
[ fig:1d-2d - chains]a , this contraction can be carried out efficiently : one starts from one plaquette and proceeds along one direction of the one - dimensional chain , always tracing out the degrees of freedom on the inside .
this way , at every point in the computation only the state at the boundary ( which has fixed size ) needs to be stored , and thus , the contraction can be carried out efficiently . on the other hand , branching structures like the one in fig .
[ fig:1d-2d - chains]b can in general not be contracted efficiently , since the size of the boundary is a priori not bounded ; in fact , e.g. quantum circuits , and even postselected quantum circuits , can be encoded this way , making such contractions in general a computationally hard task . however , as we will show in the following , the structures formed by the @xmath82 in @xmath90 , eq .
( [ eq : theta ] ) , will always be one - dimensional , and thus @xmath90 can be computed efficiently .
to this end , we will consider the state @xmath91 on a plaquette @xmath92 ( the `` central '' plaquette ) , and show that it can overlap non - trivially with the states @xmath82 of at most two of the adjacent plaquettes , thus ruling out branching structures as the one on the right of fig .
[ fig:1d-2d - chains ] .
we will make intensive use of the fact that in each layer , at most one plaquette term @xmath82 can act non - trivially on any given vertex ; in the graphical notation of fig .
[ fig:1d-2d - chains ] , we will highlight this fact by placing a cross opposite of any dot : @xmath93{dot - and - cross}}\ ] ] here , the dot indicates that the corresponding @xmath82 acts non - trivially on a vertex qubit ( the diamond in the center ) , while the cross indicates that the corresponding @xmath82 does act trivially . note that this in particular implies that @xmath91 can at most overlap non - trivially with the four horizontally and vertically adjacent plaquettes from the other layer , but not with diagonally adjacent plaquettes .
the simplest case is when the state @xmath91 on the central plaquette @xmath92 involves only two vertices non - trivially , for instance @xmath94{pair1}\ ] ] now , both qubits @xmath95 and @xmath96 can be acted upon non - trivially by at most one white plaquette the other has to be empty , following the rule ( [ eq : dot - cross ] ) that opposite of any dot there has to be a cross ; this way , only one - dimensional structures can arise : @xmath94{pair - conn}\ ] ] this clearly holds for any possible @xmath91 which acts non - trivially on only two qubits , and for any configuration of the adjacent plaquettes ; it follows that only one - dimensional structures can emerge this way . in order to understand the cases
where @xmath91 acts non - trivially on three or four qubits , let us first analyze the following situation : @xmath97{central - step - left}\ ] ] here , @xmath91 acts non - trivially at least on qubits @xmath95 and @xmath98 , and @xmath99 acts non - trivially on qubit @xmath95 which implies that @xmath100 acts trivially on it . in the following , we will show that this implies that @xmath100 also has to act trivially on qubit @xmath98 .
we will prove this by contradiction , so assume that @xmath100 acts non - trivially on qubit @xmath98 .
since @xmath100 is obtained from the original projector @xmath101 on that plaquette by a partial projection on some of the other vertices , this implies that @xmath101 acts non - trivially on qubit @xmath98 ( where it spans the whole @xmath0algebra , since we have traced out all vertices where this was not the case ) . on the other hand , @xmath99 and
thus @xmath102 acts non - trivially on qubit @xmath95 , and thus spans the whole @xmath0algebra on it .
since @xmath102 and @xmath101 commute , this means that @xmath101 acts trivially on qubit @xmath95 ; that is , @xmath101 and @xmath103 need to commute on qubit @xmath98 alone : however , since @xmath101 spans the whole @xmath0algebra on qubit @xmath98 , this would imply that @xmath103 and thus @xmath91 has to act trivially on qubit @xmath98 , giving a contradiction .
thus , we have the following `` lemma '' : @xmath104{central - step - left } } \qquad{\bm\longrightarrow}\qquad \raisebox{-0.5cm}{\includegraphics[height=1.25cm]{central - step - right}}\ ] ] let us now consider the case where the state on the central plaquette involves all four qubits non - trivially , and let us start by assuming w.l.o.g . that @xmath99 acts non - trivially on qubit @xmath95 : @xmath94{four - start}\ ] ] this implies that qubit @xmath95 is not acted upon by @xmath100 .
using eq .
( [ eq : central - lemma - qubits ] ) , we infer that @xmath100 can not involve qubit @xmath98 either , @xmath94{four - tnot2}\ ] ] eq .
( [ eq : central - lemma - qubits ] ) also shows that @xmath105 has to act trivially on qubit @xmath106 otherwise , @xmath99 would act trivially on qubit @xmath106 and thus qubit @xmath95 , which it does nt : @xmath94{four - stage3-v2}\ ] ] in order to obtain a branching structure , both @xmath105 and @xmath107 need to act non - trivially on some of the qubits .
however , if @xmath105 acts non - trivially on qubit @xmath96 , eq .
( [ eq : central - lemma - qubits ] ) implies that @xmath107 has to act trivially on qubits @xmath98 and @xmath96 : @xmath94{four - stage4-v2}\ ] ] this shows that for @xmath91 acting non - trivially on all four qubits , the central plaquette can only couple to at most two adjacent plaquettes , forming one - dimensional structures .
it remains to study what happens in the case of tripartite entanglement on the central plaquette .
we start from the following configuration : @xmath94{three - prestart-1}\ ] ] ( note that we do nt make any assumptions on how the hilbert space of qubit @xmath98 decomposes . ) clearly , in order to obtain a branching structure , either @xmath100 has to act non - trivially on qubit @xmath95 , or @xmath107 has to act non - trivially on qubit @xmath96 .
we consider w.l.o.g .
the first possibility and infer from eq .
( [ eq : central - lemma - qubits ] ) that @xmath99 has to act trivially on both qubits @xmath95 and @xmath106 : @xmath94{three - start-1}\ ] ] in order to obtain a branching structure , we now have to get both @xmath107 and @xmath105 involved .
however , if @xmath107 acts non - trivially on qubit @xmath96 , eq . ( [ eq : central - lemma - qubits ] ) implies that @xmath105 has to act trivially on both qubits @xmath96 and @xmath106 , and thus , the structure formed around @xmath91 will again be one - dimensional .
together , this shows that the overlap @xmath90 , eq .
( [ eq : theta ] ) , decays into one - dimensional structures for which the overlap can be computed efficiently . in turn
, this implies that for given @xmath62 and @xmath63 , @xmath71 can be computed efficiently , and thus , the commuting hamiltonian problem on a square lattice of qubits with plaquette interactions is in . in the preceding proof , we have assumed infinite accuracy , but as we will now show , our argument still applies if we compute with finite accuracy . to this end , let @xmath108 denote the number of qubits in the system ; we will need to show that the computation time scales as @xmath109 .
we assume that the hamiltonian terms are given exactly and can be represented with @xmath109 digits .
first , note that the trace in eq .
( [ eq : commham - trace - formulation ] ) , which equals the sum in eq .
( [ eq : sum_alphabeta_omega ] ) , evaluates to an integer , and thus , there exists at least one pair @xmath110 such that @xmath111 . if we request this particular @xmath110 as a certificate , it is sufficient if we can determine @xmath71 to @xmath112 digits .
( this is crucial , since there can be @xmath110 for which @xmath71 is arbitrarily small . )
@xmath71 is obtained from contracting a polynomial number of terms which are either @xmath7 ( which are known exactly ) or @xmath65 and @xmath66 , and the latter can be determined to @xmath109 accuracy from the @xmath0decomposition ( [ eq : cstardec ] ) , which is the solution to a ( fixed - size ) eigenvalue problem .
it follows that @xmath71 can be computed to the required @xmath109 accuracy in @xmath109 time , and our proof still applies .
we have studied the commuting hamiltonian problem on a square lattice of qubits with plaquette - wise interaction and shown that the problem is complete . differently speaking
, we have shown that there exists a classical certificate for the fact that the ground state of the system minimizes each term locally which can be checked efficiently on a classical computer .
the central idea for our proof has been to split the system into two layers in each of which the commuting terms overlap on individual sites , and to argue that the existence of a state minimizing all local terms is equivalent to the existence of a pair of ground states for the two layers with non - zero overlap .
each layer could be decomposed using the @xmath0algebraic techniques introduced to the problem in @xcite , allowing to find an efficient description of its ground state subspace .
finally , we showed that the overlap of ground states of two layers can be computed efficiently , by showing that it gives rise to of one - dimensional structures only .
a somewhat surprising feature of our approach is that while it certifies the existence of a ground state , it can not ( to our knowledge ) be used to devise a way how to prepare the ground state ; in fact , due to the possibility of having topological order in such systems , any circuit preparing their ground states would need to have at least logarithmic depth , or linear depth if it was local @xcite .
our method does , in principle , also apply beyond qubits : we can still split the system into two layers , decompose both of them into direct sum slices @xmath62 and @xmath63 , and ask the prover to provide labels @xmath62 and @xmath63 with non - zero overlap @xmath71 . while we can not make sure any more that @xmath71 can be computed efficiently , we can always ask the prover to also provide us with an instruction on how to efficiently contract the states , in case there is a way to do so , e.g.by providing the optimal contraction order .
in particular , this applies to the case where the decomposition in the direct sum gives one - dimensional spaces , such as in kitaev s toric code or quantum double models ; as well as to cases where the @xmath82 are separable states .
our idea also applies to any other graph which can be split into two layers in such a way that the @xmath0technique of @xcite can be applied to each of them , and in fact to any type of hamiltonian which is composed of two layers with eigenbases for the zero - energy subspace whose overlap can be computed efficiently , such as for product bases .
note that on the other hand , a decomposition into three layers can not be used for our purposes , since for three positive operators @xmath113 , @xmath114 , and @xmath92 , @xmath115 $ ] can have both real and imaginary parts of either sign , so that eq .
( [ eq : sum_alphabeta_omega ] ) is no longer equivalent to eq .
( [ eq : def - omega ] ) being non - zero ( as the @xmath59 and @xmath116 do not commute any more ) . an interesting open question relating to the present approach to the problem is whether it can be generalized beyond qubits . for four - level systems and beyond ,
this is likely not the case , since the local hilbert space can decompose into two qubits , and thus operators commuting on a single spin can both act non - trivially on it , i.e. , eq .
( [ eq : dot - cross ] ) does not hold any more . on the other hand , for qutrits
this is not the case once we have fixed a slice in the direct sum ; yet , it is not clear how to establish a version of eq .
( [ eq : central - lemma - qubits ] ) .
in particular , the non - trivial projections @xmath54 can now have both rank @xmath95 and @xmath98 , and in the latter case we can not simply trace out the corresponding degree of freedom ; it is however not clear that this does rule out an analogue to eq .
( [ eq : central - lemma - qubits ] ) .
we acknowledge helpful conversations with dorit aharonov , sergey bravyi , lior eldar , tobias osborne , and volkher scholz . this work has been supported by the gordon and betty moore foundation through caltech s center for the physics of information , the nsf grant , and the aro grant no . w911nf-09 - 1 - 0442 . | we consider the computational complexity of hamiltonians which are sums of commuting terms acting on plaquettes in a square lattice of qubits , and we show that deciding whether the ground state minimizes the energy of each local term individually is in the complexity class .
that is , if the ground states has this property , this can be proven using a classical certificate which can be efficiently verified on a classical computer .
different to previous results on commuting hamiltonians , our certificate proves the existence of such a state without giving instructions on how to prepare it . |
with the increasing availability of radioactive ion - beams @xcite the study of structure and reactions of exotic nuclei has become feasible @xcite . in particular , detailed investigations of the fusion process
are crucial for the prediction of heavy - element formation and will lead to a better understanding of the interplay among the strong , coulomb , and weak interactions as well as the enhanced correlations present in these many - body systems .
recently , particular attention has been given to the @xmath0ni+@xmath1sn system , where a large sub - barrier fusion enhancement was observed @xcite .
for this system fusion cross sections were measured in the energy range 142 mev @xmath2 195 mev .
in particular , it was found that fission is negligible for @xmath3 160 mev and therefore the evaporation residue cross sections have been taken as fusion cross sections .
the enhancement of sub - barrier fusion was originally deduced from comparison with a barrier penetration calculation , using a phenomenological woods - saxon interaction potential whose parameters were fitted to reproduce the evaporation residue cross sections for the @xmath0ni+@xmath4sn system @xcite .
similarly , early coupled - channel calculations , which are known to enhance the fusion cross sections by considering coupling to various excitation channels and neutron transfer , have significantly underestimated the low energy fusion cross sections for the @xmath0ni+@xmath1sn system @xcite .
subsequently , more sophisticated coupled - channel calculations lead to an improvement for the description of the lower energy data . finally , the inclusion of the neutron transfer channels with positive @xmath5 value in addition to inelastic excitations resulted in the best description to date @xcite .
the theoretical analysis of the fusion data generally involves determination of a phenomenological ion - ion potential such as the bass model @xcite , the proximity potential @xcite , or potentials obtained via the double - folding method @xcite .
subsequently , the actual fusion cross section is calculated by either using barrier penetration models @xcite , or the coupled - channel method @xcite .
the latter includes various excitations of the target and/or projectile using the coupled - channel formalism @xcite , as well as the inclusion of neutron transfer , and can be consistently applied at energies above and below the barrier @xcite .
effectively , the inclusion of each additional excitation leads to a modification of the original inert core ion - ion potential , resulting in a series of effective barriers .
one common physical assumption used in many of these calculations is the use of the frozen density or the sudden approximation . in this approximation
the nuclear densities are unchanged during the computation of the ion - ion potential as a function of the internuclear distance .
furthermore , the effects included in channel couplings are usually based on the static properties of the participating nuclei , which may accurately represent the early stages of the collision process , but are expected to change as the two ions strongly interact . while these methods provide a useful and productive means for quantifying multitudinous reaction data it is desirable to include dynamical effects and make contact with the microscopic theories of nuclear structure and reactions .
recently , we have developed a new approach for calculating heavy - ion interaction potentials which incorporates all of the dynamical entrance channel effects included in the time - dependent hartree - fock ( tdhf ) description of the collision process @xcite .
these effects include the neck formation , particle exchange , internal excitations , and deformation effects to all order , as well as the effect of nuclear alignment for deformed systems .
the method is based on the tdhf evolution of the nuclear system coupled with density - constrained hartree - fock calculations to obtain the ion - ion interaction potential .
preliminary calculations for the @xmath0ni+@xmath1sn system highlighted the importance of dynamical deformation effects @xcite .
here we give a completed study of fusion cross sections using this formalism . in the next section
we will summarize some theoretical aspects of the density constrained tdhf theory along with methods to incorporate dynamical alignment into our calculations , as well as the method used the calculate cross sections from the resulting barriers . in section [ sec : results ] we present interesting aspects of the reaction dynamics and compare our results with experiment and other calculations .
in this subsection we give a qualitative description of the density constrained tdhf method which is used to obtain the dynamical barriers for the @xmath0ni+@xmath1sn system .
further details of the method can be found in ref .
@xcite .
the _ density constraint _ is a novel numerical method that was developed in the mid 1980 s @xcite and was used to provide a microscopic description of the formation of shape resonances in light systems @xcite . in this approach
the tdhf time - evolution takes place with no restrictions . at certain times during the evolution the instantaneous density is used to perform a static hartree - fock minimization while holding the total density constrained to be the instantaneous tdhf density .
in essence , this provides us with the tdhf dynamical path in relation to the multi - dimensional static energy surface of the combined nuclear system .
since we are constraining the total density all moments are simultaneously constrained . the numerical procedure for implementing this constraint and the method for steering the solution to @xmath6
is discussed in refs .
the convergence property is as good if not better than in the traditional constrained hartree - fock calculations with a constraint on a single collective degree of freedom . in ref .
@xcite we have shown that the ion - ion interaction potential is simply given by @xmath7 where @xmath8 is the density constrained energy at the instantaneous separation @xmath9 , while @xmath10 and @xmath11 are the binding energies of the two nuclei obtained with the same effective interaction .
we would like to emphasize again that this procedure does not affect the tdhf time - evolution and contains no _ free parameters _ or _ normalization_. in practice , tdhf runs are initialized with energies above the coulomb barrier and in ref .
@xcite we have shown that there is no appreciable energy dependence to the barriers obtained via the density constrained tdhf method .
a detailed description of our new three - dimensional unrestricted tdhf code has recently been published in ref . @xcite . for the effective interaction
we have used the skyrme sly5 force @xcite including all of the time - odd terms . in general , the fusion cross sections depend on the interaction potential and form factors in the vicinity of the coulomb barrier .
these are expected to be modified during the collision due to dynamical effects .
in addition , experiments on sub - barrier fusion have demonstrated a strong dependence of the total fusion cross section on nuclear deformation @xcite .
the dependence on nuclear orientation has received particular attention for the formation of heavy and superheavy elements @xcite and various entrance channel models have been developed to predict its role in enhancing or diminishing the probability for fusion @xcite .
recently , we have developed a new approach for calculating the effect of nuclear alignment for deformed systems @xcite .
in essence , the procedure for incorporating alignment into the evolution of the heavy - ion collision dynamics is done in two separate steps : a ) a dynamical coulomb alignment calculation to determine the probability that a given nuclear orientation occurs at the distance @xmath12 , where the tdhf run is initialized .
the alignment generally results from multiple _ e2_/_e4 _ coulomb excitation of the ground state rotational band .
the distance @xmath12 is chosen such that the nuclei only interact via the coulomb interaction .
b ) a tdhf calculation , starting at this finite internuclear distance @xmath12 , for a fixed initial orientation of the deformed nucleus . since the experiments are usually done with unpolarized beams , in a full quantum mechanical calculation one would have to average over discrete quantum mechanical rotational bands . in the classical limit , this corresponds to averaging over orientation angles .
a general study of taking the classical limit of the relative nuclear motion during a heavy - ion collision which includes inelastic excitations of one of the heavy ions in the entrance channel has been given in ref .
@xcite .
the heavy - ion interaction potential between two deformed nuclei depends on the distance vector between their centers - of - mass , @xmath13 , and on the relative orientation of their intrinsic principal axis systems which may be described in terms of three euler angles @xmath14 per nucleus , i.e. in the most general case we have @xmath15 explicit expressions for this interaction potential within the double - folding method are given in ref .
the expression for @xmath16 can be simplified if the intrinsic nuclear density distributions are axially symmetric ; in this case , the potential does not depend on the euler angles @xmath17 which describe rotations about the symmetry axes . if we put , for convenience , the distance vector in @xmath18-direction , @xmath19 , the potential between two deformed axially symmetric nuclei has the structure @xmath20 finally , if one of the nuclei is spherical , e.g. nucleus ( 1 ) , the potential is simply given by @xmath21 where the euler angle @xmath22 describes the direction of the nuclear symmetry axis relative to the internuclear distance vector .
details of the dynamic alignment formalism are presented in @xcite .
we give here a brief summary : for a given incident energy @xmath23 we carry out a semiclassical coulomb excitation calculation of the dominant collective levels of the deformed nucleus .
the energy levels and @xmath24-transition matrix elements for @xmath0ni are taken from experimental data @xcite : @xmath25 mev , @xmath26 mev and @xmath27@xmath28 ( oblate deformation ) .
the coulomb excitation calculation starts at very large internuclear distances ( about @xmath29 fm ) when both nuclei may be presumed to be in their respective ground states and stops at the ion - ion separation distance @xmath12 ( about @xmath30 fm ) .
the coulomb excitation amplitudes determine the probability distribution of initial orientations . in fig .
[ fig : align_ni_sn ] we show the differential alignment probability as a function of the euler angle @xmath31 used in our calculations . dynamic alignment due to coulomb excitation of @xmath0ni .
shown is the orientation probability as a function of the euler angle @xmath31 in a central collision at internuclear distances @xmath32 fm ( dashed curve ) and at @xmath33 fm ( solid curve ) . ] in the case of one spherical nucleus ( @xmath1sn ) and one deformed reaction partner ( @xmath0ni ) , the total fusion cross section is given by an integral over all orientation ( euler ) angles , with solid angle element @xmath34 @xmath35 where @xmath36 represents the alignment probability and @xmath37 is the fusion cross section associated with a particular alignment . for a consistent calculation of fusion cross sections at above and below the barrier energies
we have adopted the commonly used _ incoming wave boundary condition _ ( iwbc ) method @xcite .
once the ion - ion potential for a particular orientation , @xmath38 , is calculated the two - body schrdinger equation becomes @xmath39 \psi_l(r,\beta)=0\;,\ ] ] where @xmath40 is the reduced mass and @xmath41 denotes orbital angular momentum .
iwbc assumes that once the minimum of the potential is reached fusion will occur , consequently no imaginary part of the potential is needed . in practice ,
the schrdinger equation is integrated from the potential minimum , @xmath42 , where only an incoming wave is assumed , to a large asymptotic distance , where it is matched to incoming and outgoing coulomb wavefunctions to obtain the penetration factor , @xmath43 , as the ratio of the incoming flux at @xmath42 to the incoming coulomb flux at large distance .
the total cross section is given by @xmath44 with @xmath45 .
for the numerical implementation we have followed the procedure for the coupled - channel code ccful described in ref .
@xcite and exactly reproduced their results for an inert - core potential .
we have carried out a number of tdhf calculations with accompanying density constraint calculations to compute @xmath38 given by eq .
( [ eq : vr ] ) .
a detailed description of our new three - dimensional unrestricted tdhf code has recently been published in ref @xcite . for the effective interaction
we have used the skyrme sly5 force @xcite including all of the time - odd terms .
the code was modified to self - consistently generate initial states for @xmath0ni with different orientations .
all of our tdhf calculations were done at an initial energy of @xmath46 mev and separation @xmath47 fm . as we have reported in ref .
@xcite the potential barriers obtained from the density constrained tdhf method are not sensitive to the initial energy ( above the barrier ) .
we have tested this again by running a few orientations at @xmath48 mev and @xmath49 mev and did not observe any appreciable difference . in a tdhf collision leading to fusion the relative kinetic energy in the entrance channel
is entirely converted into internal excitations of a single well defined compound nucleus . in tdhf theory the dissipation of the relative kinetic energy into internal excitations is due to the collisions of the nucleons with the `` walls '' of the self - consistent mean - field potential .
tdhf studies demonstrate that the randomization of the single - particle motion occurs through repeated exchange of nucleons from one nucleus into the other .
consequently , the equilibration of excitations is very slow , and it is sensitive to the details of the shape evolution of the composite system .
this is in contrast to most classical pictures of nuclear fusion , which generally assume near instantaneous , isotropic equilibration .
time - dependence of the isovector dipole and isoscalar quadrupole moments for the head - on collision of @xmath0ni+@xmath1sn system at @xmath46 mev and @xmath50 . ]
recently , the importance of transfer of neutrons with positive @xmath5 value in fusion has been emphasized @xcite . in tdhf
this effect manifests itself as the excitation of the pre - compound collective dipole mode , which is likely when ions have significantly different @xmath51 ratio , and is a reflection of dynamical charge equilibration . in fig .
[ fig : qt ] we show the time evolution of the isovector dipole and isoscalar quadrupole moments for the head - on collision of @xmath0ni+@xmath1sn system at @xmath46 mev and @xmath50
. the fourier transform of these oscillations show a @xmath52 mev isovector dipole peak as well as @xmath53 mev and @xmath54 mev isoscalar quadrupole peaks .
dipole gamma yield for the head - on collision of @xmath0ni+@xmath1sn system at @xmath46 mev and @xmath50 . ] for a time - dependent charge distribution it is possible to calculate the @xmath55-yield as a function of gamma energy @xcite , @xmath56 .
the asymptotic yield integrated over a spherical surface is given by @xmath57 where the amplitudes @xmath58 @xmath59^{1/2 } \left[\frac{\hbar \omega}{\hbar c}\right]^{l+2}m_{l}(\hbar \omega ) \;,\ ] ] are given in terms of the fourier transform of the moments of the density @xmath60 computation of the yield given by eq .
[ eq : yield ] shows that the dominant contribution is from the dipole mode , which is shown in fig .
[ fig : yield ] .
( color online ) change in neutron ( blue curve ) and proton ( red curve ) numbers of the left fragment ( originally @xmath0ni ) as a function of the ion - ion separation coordinate , @xmath61 .
the dashed lines are there to emphasize the asymptotic values . ]
while the above analysis provides a picture of particle exchange dynamics over time it is also possible to examine particle exchange during the initial stages of the collision , namely from the well separated nuclei to the time when the minimum separation is reached . in fig .
[ fig : left ] we show the neutron and proton transfer to the left fragment , which in our case is initially @xmath0ni . as we can see in fig .
[ fig : left ] the number of neutrons and protons increase for the left fragment as the ions come into contact . as expected the transfer of neutrons starts earlier since they are dominant in the surface region . for small @xmath61 values
the increase is simply due to charge equilibration as the two nuclei have a substantial overlap .
however , at a separation of @xmath62 fm , which corresponds approximately to the top of the potential barrier as discussed in the next subsection , we have about two neutrons transferred , and at @xmath63 fm we have as many as six neutrons transferred .
it is also interesting to observe that the large transfers in the early stages of the collision happen quickly , indicated by the sudden jumps in neutron number .
naturally , the right fragment undergoes just the opposite of these trends . in this subsection
we discuss the barriers obtained via the density constrained tdhf method , and other related quantities , such as the effective mass and the reduced mass .
we have performed calculations of @xmath38 in @xmath64 intervals from @xmath65 to @xmath50 . in fig .
[ fig : vrb ] we show all of these barriers .
the lowest barrier corresponds to @xmath50 orientation of the symmetry axis of the oblate @xmath0ni nucleus with respect to the collision axis .
each subsequent barrier is reduced by @xmath66 up to the highest barrier at @xmath65 orientation .
( color online ) potential barriers , @xmath38 , obtained from density constrained tdhf calculations for the @xmath0ni+@xmath1sn system .
angle @xmath31 indicates different orientations of the deformed @xmath0ni nucleus in @xmath64 intervals . ] for the case of @xmath65 orientation the calculated barrier is almost exactly the same as the empirical barrier used in ref .
@xcite without channel couplings , having a barrier height of @xmath67 mev and location of the barrier peak at approximately @xmath68 fm .
the difference for smaller @xmath61 values is due to the use of the point coulomb interaction in the model calculation , which is unphysical when nuclei overlap . as seen in fig .
[ fig : vrb ] for higher @xmath31 values the barrier is lowered and the barrier peak moves to larger @xmath61 values , with the lowest barrier having a height of @xmath69 mev and peaking around @xmath70 fm .
the physical picture which emerges from these calculations is that for all energies above @xmath69 mev , which is the peak of the lowest barrier , the fusion cross section will be dominated by lower barriers as the cross section above the barrier is substantially larger than sub - barrier cross sections . in other words
the only experimental data point that appears to be truly sub - barrier is the lowest energy point at @xmath71 mev .
effective mass calculated from eq .
( [ eq : eff ] ) for the head - on collision of @xmath0ni+@xmath1sn system at @xmath46 mev and @xmath50 . ] in addition to all of the dynamical effects included in the potentials @xmath38 it is also possible to construct effects due to dynamical mass . since in the density constrained tdhf method the potential is obtained from the tdhf evolution by essentially extracting the internal excitation energy from instantaneous tdhf solutions via the density constraint , the energy conservation becomes @xmath72 for a particular initial orientation at asymptotic energy @xmath23 we obtain the collective velocity @xmath73 directly from the tdhf evolution and the potential @xmath38 from the density constraint calculations .
thus , the effective mass is given by @xmath74}{{\dot{r}}^{2}}\ ] ] in fig .
[ fig : eff ] we show the effective mass as a function of the ion - ion separation coordinate @xmath61 for energy @xmath46 mev and @xmath50 .
the solid curve is the effective mass obtained from eq .
( [ eq : eff ] ) scaled with the constant reduced mass @xmath40 for the two ions .
the ratio starts from unity at large distances and increases as @xmath61 gets smaller and as the ions pass the top of the barrier as indicated by the dotted line in fig .
[ fig : eff ] .
this is due to the fact that the relative velocity , @xmath73 , becomes small as the ions begin to overlap substantially , while the numerator of eq .
[ eq : eff ] remains non - zero since the energy is above the barrier .
after the initial slowdown the ions accelerate once more before reaching the composite system for which @xmath75 .
a similar behavior is observed for all values of @xmath31 .
traditionally , the effective mass is obtained from constrained hartree - fock ( chf ) or adiabatic time - dependent hartree - fock ( atdhf ) calculations @xcite and shows a strikingly similar behavior to the ones obtained through our method .
( color online ) reduced mass calculated for the head - on collision of @xmath0ni+@xmath1sn system at @xmath46 mev and @xmath50 . ]
in addition to the effective mass it also possible to calculate the dynamical reduced mass .
it is well known that the naive formula for reduced mass given by @xmath76 does not correctly represent the dynamical behavior of this quantity , monotonically increasing from its asymptotic value to the predictable final value when @xmath77 , i.e. a composite has been formed and half the mass is in the right half and other half on the left half of the numerical box . an alternative way to calculate the dynamical reduced mass
is given in ref .
@xcite . here
the dynamical centers and momenta for the left and right halves are calculated via @xmath78 where the index @xmath79 denotes left and right halves .
the left and right masses are then calculated by @xmath80 leading to the dynamical reduced mass @xmath81 with @xmath82 .
this quantity is plotted in fig .
[ fig : mur ] along with the naive expression , @xmath83 .
while we observe interesting structure in @xmath84 the overall change in the magnitude around the barrier region is not large enough to substantially alter the fusion cross sections .
this is also shown in fig .
[ fig : eff ] where we have also plotted the ratio of the effective mass to the dynamical reduced mass ( dashed curve ) .
therefore , in practice we have not included @xmath84 in our cross section calculations .
we have calculated the total fusion cross section as a function of energy using the alignment averaged fusion formula given in eq .
( [ eq : fusion ] ) .
the total fusion cross section using the potential barriers obtained from density constrained tdhf calculations directly ( with constant reduced mass ) is shown by the blue curve in fig .
[ fig : xsection ] . also shown is the the latest coupled - channel calculation @xcite including inelastic excitation of @xmath0ni to first @xmath85 and @xmath86 states and @xmath1sn to the first @xmath85 state , as well as two - neutron transfer ( red dashed curve ) .
we have also included the dynamical effective mass by making the well known coordinate scale transformation @xcite @xmath87 as a result of this transformation all of the effects of the dynamical effective mass are transferred to the scaled potential while the reduced mass @xmath40 remains constant at its asymptotic value .
this is convenient for a number of reasons , one being that the we can use our fusion code without any modifications .
the resulting cross sections are shown by the black curve in fig .
[ fig : xsection ] . as we see the effect of the dynamical mass is to raise the cross section at higher energies . with the exception of the lowest energy data point the calculated cross sections are in very good agreement with data and the extended coupled - channel calculation .
we emphasize again that our density constrained tdhf calculations contain no adjustable parameters or normalization factors . in the region around @xmath88
mev our calculations over predict the experimental cross section while the coupled - channel one slightly under predict .
the agreement at lower and higher energies are excellent .
the question about the lowest energy data point is still an open one .
it is our understanding that a new experiment is planned to measure this cross section @xcite .
( color online ) total fusion cross section as a function of @xmath23 . shown
are the experimental data ( filled circles ) , the latest coupled - channel calculation @xcite ( red dashed curve ) , and our density constrained tdhf cross sections with dynamical effective mass ( black curve ) and with constant reduced mass ( blue curve ) . ]
if this enhancement is confirmed it would be a challenge since all of the existing theories underestimate the cross section by a few orders of magnitude .
we have shown in our barrier calculations that this is the only point which is truly sub - barrier fusion .
on the other hand such a large value for the cross section is more consistent with an above the barrier or very close to the top of the barrier energy .
this would indicate that there is a mode of @xmath0ni nucleus which is not well described by the current microscopic interactions , like the skyrme force .
one possibility may be triaxiality since @xmath0ni is experimentally found to be gamma soft .
we have performed density constrained tdhf calculations of fusion cross sections for the @xmath0ni+@xmath1sn system .
our results agree very well with the measured data and with a state of the art coupled - channel calculation despite having no adjustable parameters .
this indicates that many of the reaction dynamics are included in the tdhf description of the collision .
as we investigate fusion reactions involving neutron rich and deformed nuclei it is apparent that an understanding of the structure of these nuclei is crucial to the description of the reaction dynamics . for these nuclei various effects , such as inelastic excitations , particle transfer , and other dynamical effects
lead to substantial modification of the naive potential barrier calculations which assume an inert core and no dynamics .
consequently , the definition of _ sub - barrier _ fusion becomes ambiguous since it is difficult to determine the barrier a priori .
present calculations indicate that the lowest energy data point may be the only one truly at a sub - barrier energy .
if a new measurement confirms this data point it would be of great interest since none of the theories can reproduce this cross section , thus indicating a fundamental property that is not included in any of the calculations . , doe / nsf nuclear science advisory committee , april 2002 , www.sc.doe.gov/production/henp/np/nsac/nsac.html , ( u.s . department of energy , washington , dc , 2002 ) . c. l. jiang _ et al .
93 * , 012701 ( 2004 ) .
j. f. liang _
et al . _ ,
lett . * 91 * , 152701 ( 2003 ) ; * 96 * , 029903(e ) ( 2006 ) .
j. f. liang and c. signorini , int .
e * 14 * , 1121 ( 2005 ) .
j. f. liang _
et al . _ , physical review c ( in press ) .
r. bass , nucl .
* a231 * , 45 ( 1974 ) .
r. bass , _ nuclear reactions with heavy ions _ , ( springer verlag , new york , 1980 ) .
j. blocki , j. randrup , w. j. swiatecki , and c. f. tsang , ann .
( n.y . ) * 105 * , 427 ( 1977 ) .
j. randrup and j. s. vaagen , phys . lett . *
b77 * , 170 ( 1978 ) .
m. seiwert , w. greiner , v. e. oberacker , and m. j. rhoades - brown , phys .
c * 29 * , 477 ( 1984 ) .
j. r. birkelund and j. r. huizenga , phys .
c * 17 * , 126 ( 1978 ) .
g. r. satchler and w. g. love , phys . rep . * 55 * , 183 ( 1979 ) .
g. bertsch , j. borysowicz , h. mcmanus , and w. g. love , nucl . phys .
* a284 * , 399 ( 1977 ) . m. j. rhoades - brown and v. e. oberacker , phys .
lett . * 50 * , 1435 ( 1983 ) .
m. j. rhoades - brown , v. e. oberacker , m. seiwert , and w. greiner , z. phys .
a * 310 * , 287 ( 1983 ) .
n. takigawa and g. f. bertsch , phys .
c * 29 * , 2358 ( 1984 ) .
a. b. balantekin and n. takigawa , rev .
* 70 * , 77 ( 1998 ) .
s. landowne and s. c. pieper , phys .
c * 29 * , 1352 ( 1984 ) .
m. j. rhoades - brown and m. prakash , phys .
lett . * 53 * , 333 ( 1984 ) .
k. hagino , n. rowley , and a. t. kruppa , comp .
. comm . * 123 * , 143 ( 1999 ) .
h. esbensen , prog .
* 154 * , 11 ( 2004 ) .
h. esbensen , phys .
c * 72 * , 054607 ( 2005 ) .
a. s. umar and v. e. oberacker , phys .
c * 74 * , 021601(r ) ( 2006 ) .
a. s. umar and v. e. oberacker , phys .
c * 74 * , 061601(r ) ( 2006 ) .
r. y. cusson , p. -g .
reinhard , m. r. strayer , j. a. maruhn , and w. greiner , z. phys .
a * 320 * , 475 ( 1985 ) .
a. s. umar , m. r. strayer , r. y. cusson , p. -g .
reinhard , and d. a. bromley , phys .
c * 32 * , 172 ( 1985 ) .
a. s. umar and v. e. oberacker , phys .
c * 73 * , 054607 ( 2006 ) .
e. chabanat , p. bonche , p. haensel , j. meyer and r. schaeffer , nucl . phys . *
a635 * , 231 ( 1998 ) ; nucl . phys . *
a643 * , 441(e ) ( 1998 ) .
a. s. umar , m. r. strayer , d. j. ernst , k. r. s. devi , phys . rev .
c * 30 * , 1934 ( 1984 ) .
r. g. stokstad , y. eisen , s. kaplanis , d. pelte , u. smilansky , and i. tserruya , phys .
lett . * 41 * , 465 ( 1978 ) .
k. nishio , h. ikezoe , s. mitsuoka , k. satou , and s. c. jeong , phys .
c * 63 * , 044619 ( 2001 ) . c. simenel , ph
chomaz , and g. de france , phys .
lett . * 93 * , 102701 ( 2004 ) .
a. s. umar and v. e. oberacker , phys
c * 74 * , 024606 ( 2006 ) . evaluated nuclear structure data file ( ensdf ) , national nuclear data center , brookhaven national laboratory , http://www.nndc.bnl.gov/ensdf .
g. h. rawitscher , phys .
135 , 605 ( 1964 ) .
dasso , s. landowne , and a. winther , nucl . phys .
a407 * , 221 ( 1983 ) . v. i. zagrebaev , v. v. samarin , and w. greiner , phys .
c * 75 * , 035809 ( 2007 ) .
k. goeke , f. grmmer , and p. -g .
reinhard , ann .
* 150 * , 504 ( 1983 ) .
denis lacroix , http://arxiv.org/pdf/nucl-th/0202063 .
j. f. liang _ et al . _ , ( private communication ) . | we study fusion reactions of the @xmath0ni+@xmath1sn system using the recently introduced density constrained time - dependent hartree - fock formalism . in this formalism
the fusion barriers are directly obtained from tdhf dynamics .
in addition , we incorporate the entrance channel alignment of the deformed ( oblate ) @xmath0ni nucleus due to dynamical coulomb excitation .
we discuss the influence of particle transfer and other dynamical effects on the fusion cross sections . calculated cross sections are in very good agreement with data and other calculations .
= 10000 = 10000 = 10000 |
the problems of entropic and spectral rigidity of riemannian manifolds have been widely studied , beginning with the surfaces @xcite .
the works treat riemannian metrics on compact surfaces @xcite , @xcite , @xcite , on higher dimension manifolds @xcite , @xcite , @xcite , @xcite , or on surfaces with singularities @xcite .
the related problem of boundary rigidity of a riemannian metric features many results @xcite , @xcite , @xcite .
the rigidity of an absolutely continuous flow conjugacy persists in some way with the presence of a magnetic field on a compact surface @xcite , and so do entropic rigidity in this case @xcite .
the topological entropy of the magnetic flow in higher dimension has also been studied @xcite , @xcite . unlike the geodesic flow ,
a conjugacy being only continuous ( in fact hlder - continuous ) between two magnetic flows on a surface had not been treated .
when the surface is compact and the jacobi endomorphism @xcite of the magnetic flow is negative , this flow has got the anosov property @xcite ; two such flows have got the same marked length spectrum of periodic orbits if and only if they are @xmath0-conjugated @xcite .
the main result presented here is that the flow associated with a riemannian metric and a non zero magnetic field on a closed surface , if it has got a negative jacobi endomorphism , can not have the same marked length spectrum as the geodesic flow associated with another riemannian metric having a negative curvature and the same total volume .
the assumption on the equality of the total volumes is essential @xcite . *
[ thm : spm ] * let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface
. let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 , .
let @xmath7 be a @xmath4-magnetic field over @xmath1 .
the magnetic flow @xmath8 is supposed to have a negative jacobi endomorphism . if the magnetic flow @xmath9 and the geodesic flow @xmath10 have the same marked length spectrum , and if the surface @xmath1 has the same total volume for the two metrics , then the two metrics are isotopic , which means that one is the image of the other by a diffeomorphism @xmath11 of @xmath1 homotopic to the identity , and the magnetic field @xmath7 is zero .
the proof consists in coming back to the known case where there exists an absolutely continuous conjugacy between the two flows @xcite .
the proof of the regularity of the conjugacy is inspired by u. hamenstdt s methods @xcite .
we construct linearizations of the universal covering of the surface , compatible with the stable spaces of the flow .
this is useful to proof that the lyapounoff exponents of the periodic orbits are preserved ( theorem [ thm : reguconj ] ) , which ensures that the conjugacy is smooth @xcite .
the regularity of the conjugacy used to proof the theorem [ thm : spm ] is valid in general for two magnetic flows ( we denote @xmath12 the unit tangent bundle of @xmath13 ) : * [ cor : reguconj ] * let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface .
let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath15 are supposed to have negative jacobi endomorphisms .
if the two magnetic flows have the same marked length spectrum , then they are conjugated by a @xmath4-diffeomorphism @xmath16 from @xmath17 onto @xmath18 .
a uniformization of a surface equipped with a metric with negative curvature has already been constructed @xcite ; it applies to an anosov flow on a @xmath19-manifold , but with the condition that the stable spaces be of @xmath20-class , which is unlikely for the magnetic flow @xcite .
the uniformly quasiconformal diffeomorphisms present another example of uniform structures on stable spaces @xcite .
it seems legitimate to ask if the construction presented here is practicable for other flows whose stable spaces are not necessarily of @xmath20-class , and particularly to which extent a @xmath0-conjugacy between two such flows could be differentiable . given a manifold @xmath1 , diffeomorphic to @xmath21 , with a magnetic flow having a negative jacobi endomorphism and having the gradient of centre - stable and centre - unstable spaces uniformly bounded , and given a point @xmath22 , and a unitary vector @xmath23 , the _ linearization _
@xmath24 ( defined in section [ sec : linea ] ) sends @xmath1 onto @xmath25 ; the geodesic directed by @xmath26 onto the straight line @xmath27 ; and the horocycles associated with the centre - stable manifold of @xmath26 onto the straight lines orthogonal to @xmath26 .
this linearization , used as is , presents a little rigidity . * [ thm : rigir ] * let @xmath1 be an oriented surface diffeomorphic to @xmath21 , equipped with two @xmath4-riemannian metrics @xmath2 and @xmath3 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath28 are supposed to have pinched negative jacobi endomorphisms ( the @xmath20-norms of @xmath7 and @xmath14 are thus bounded ) , and the gradient of the centre - stable @xmath29 spaces and the gradient of the centre - unstable @xmath30 spaces for @xmath6 are supposed to be uniformly bounded . if there exist a diffeomorphism @xmath31 and a point @xmath22 satisfying @xmath32 then the two metrics are images one of each other by @xmath11 , and so are the two magnetic fields : @xmath33 . when metrics and magnetic fields are invariants under a cocompact group , the rigidity of the linearization is stronger in some way . *
[ thm : rigicomp ] * let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface .
let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath34 are supposed to have negative jacobi endomorphisms .
if there exist two vectors @xmath35 , @xmath36 and a @xmath20-diffeomorphism @xmath31 homotopic to the identity , of which a lift @xmath37 over @xmath38 satisfies @xmath39 , then the two metrics are isotopic , transported by @xmath11 , and so are the two magnetic fields : @xmath33 . using the tools of the construction of the linearization , we also get a result of constancy of ( future ) lyapounoff exponents along the centre - stable manifolds ( theorem [ thm : lyap ] ) .
in the following , @xmath1 denotes a complete , connected , oriented surface , equipped with a @xmath4-riemannian metric whose curvature is negatively pinched : @xmath40 .
the cartan - hadamard theorem ( @xcite , p.138 ) implies that the universal cover @xmath41 is diffeomorphic to @xmath21 , with cover mapping @xmath42 . within sections [ sec :
courbure ] , [ sec : fluc ] , [ sec : transport ] , [ sec : linea ] and [ sec : regu ] , the surface @xmath1 is simply connected , thus equal to @xmath38 . the projection of @xmath43 and @xmath44 on @xmath1 is written down @xmath45 .
the cover mapping is @xmath42 .
the surface @xmath1 is said _ closed _ if it is compact ( without boundary ) .
let @xmath46 be the rotation of angle @xmath47 in the tangent space @xmath43 . for a curve @xmath48 ,
the equation of the magnetic flow @xmath49 associated with a magnetic field @xmath50 is @xcite : @xmath51 the flow is a one - parameter group of diffeomorphisms acting on @xmath44 .
the magnetic field @xmath52 is supposed to be smooth .
the _ jacobi endomorphism _ associated with this second order differential equation @xcite is the application @xcite : @xmath53 when the surface @xmath1 is compact ( closed ) , saying that the jacobi endomorphism is negative is equivalent to saying that it is pinched between two strictly negative constants . in the following ,
the real function @xmath50 is a magnetic field such that the associated jacobi endomorphism @xmath54 satisfies the pinching condition , which means that there exist two positive constants
@xmath55 et @xmath56 verifying : @xmath57 * @xcite * with the assumption , to a vector @xmath58 are associated the stable @xmath59 and unstable @xmath60 jacobi fields along the orbit of @xmath26 , with components @xmath61 and @xmath62 in the base @xmath63 satisfying @xmath64 the stable and unstable spaces are determined by the mappings : @xmath65 @xmath66 the tangential component of the stable space at @xmath58 satisfies the relation @xcite : @xmath67 writing @xmath68 yields as in @xcite , section @xmath69 : @xmath70 let @xmath71 , @xmath72 , @xmath73 and @xmath74 be respectively the stable , centre - stable , unstable , centre - unstable manifolds associated with the unitary vector @xmath26 .
the stable horocycle of the magnetic flow associated with @xmath26 is @xmath75 .
the stable horocycle associated with @xmath76 is @xmath77 .
the busemann function associated with @xmath26 is the mapping @xmath78 such that @xmath79 . under the assumption ,
the centre - stable and centre - instable spaces identified to the ricatti applications @xmath80 and @xmath81 are of @xmath20-class over @xmath44 and @xmath82 ; the horizontal ( orthogonal ) component of the stable ( @xmath83 ) and instable ( @xmath84 ) jacobi fields are of @xmath20-class over @xmath85 and @xmath86 ; the circle at infinity @xmath87 admits also a differential structure of @xmath20-class ( @xcite , section @xmath88 ) .
let @xmath89 be the point at infinity corresponding to the future orbit of @xmath90 ( and identified with its centre - stable manifold ) .
given two distinct points @xmath91 and @xmath92 , we denote @xmath93 as the unitary vector tangent to @xmath1 at @xmath94 , directing the unique curve solution joining @xmath94 to @xmath95 ( in this order ) .
let @xmath1 be a complete connected oriented surface , equipped with a riemannian metric @xmath96 of @xmath4-class with pinched negative curvature @xmath97 , and with a uniformly bounded magnetic field @xmath52 , of @xmath4-class , with its jacobi endomorphism satisfying the pinching condition . using the method of the second order differential equations of foulon @xcite , let @xmath98 be the generating field of the magnetic flow , @xmath99 be the horizontal field , @xmath100 be the vertical field ; together ,
they constitute a basis field of the bundle @xmath101 , tangent to the unitary tangent bundle .
let @xmath102 be the dual basis field .
let @xmath26 be in @xmath44 and @xmath103 for @xmath6 be two jacobi fields , with @xmath104 ; they are associated to the tangent vectors @xmath105 at every point of the orbit of @xmath26 . the wronskian form @xmath106 applied to the pair of jacobi fields yields @xmath107
this is an invariant @xmath108-form under the magnetic flow on @xmath44 .
it gives a symplectic form on the space of orbits of the magnetic flow .
its absolute value is equal to the liouville measure on the space of orbits @xcite .
we have : @xmath109 for every @xmath110 .
this quantity lays between @xmath111 and @xmath112 .
[ thm : courbure ] on a complete , connected , simply connected , oriented surface @xmath1 , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable @xmath113 ( respectively centre - unstable @xmath114 ) spaces is uniformly bounded , the geodesic curvature of the stable ( respectively unstable ) horocycles of the magnetic flow is uniformly bounded .
[ fig : un ] the unstable case is similar to the stable case ; we only consider this last one .
let : \i ) @xmath26 be in @xmath44 , \ii ) @xmath115 be a smooth curve from @xmath116-\varepsilon , \varepsilon [ $ ] to @xmath117 such that @xmath118 for every @xmath119-\varepsilon , \varepsilon [ $ ] , \iii ) @xmath120 , \iv ) @xmath121 , \v ) @xmath122 , @xmath123 , vi ) @xmath124 , \vii ) @xmath125 , \viii ) @xmath126 , \ix ) @xmath127 , \x ) @xmath128 be the length of the curve @xmath129 on the interval @xmath130 $ ] .
we have @xmath131 for all @xmath132 and @xmath133 the equations , , imply @xmath134 thus we have for all @xmath135 and all @xmath132 : @xmath136 @xmath137 with the assumption @xmath138 for all @xmath119-\varepsilon , \varepsilon [ $ ] , there comes : @xmath139 thus @xmath140 we have also : @xmath141 from the relations and comes : @xmath142 v_s + ( u_-+\kappa w_-)w_- n(v_s ) .\ ] ] in order to estimate the curvature of @xmath143 , the norm of @xmath144 should be controlled .
the equation gives : @xmath145 the upper estimate implies : @xmath146 with the equation , this yields for all @xmath147 $ ] : @xmath148 let this last constant be written down @xmath149 .
the geodesic curvature of the horocycle @xmath150 is @xmath151 from the equalities and we deduce @xmath152 thus @xmath153 in conclusion , the geodesic curvature of the stable horocycles is uniformly bounded , and so is the geodesic curvature of the unstable horocycles .
the proof of the following result is left to the reader as an exercise .
[ cor : courbure ] with the assumptions of the theorem [ thm : courbure ] , there exists a mapping @xmath154 , continuous at @xmath155 and which annulates at @xmath155 , such that for all horocycle @xmath156 of the magnetic flow , every diffeomorphism @xmath157 and all @xmath158 , we have @xmath159
[ thm : fluc ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable space @xmath113 is uniformly bounded . for @xmath160 and @xmath161 , the family of mappings @xmath162 converges when @xmath163 grows to @xmath164 , uniformly on the compacts of @xmath165 , to a mapping @xmath166 , continuous on @xmath167 , and which admits a partial derivative with respect to @xmath168 in the direction @xmath169 , continuous on @xmath167 .
let : \i ) @xmath170 belong to @xmath171 , \ii ) @xmath172 belong to @xmath1 , \iii ) @xmath150 be a smooth parametrization of the horocycle @xmath173 such that @xmath174 and @xmath175 , \iv ) @xmath176 the smooth parametrization of @xmath1 by @xmath21 which follows from it for all @xmath160 , \v ) @xmath177 , @xmath178 , @xmath179 , \vi ) @xmath180 , \vii ) @xmath181 , \viii ) @xmath182 , \ix ) @xmath183 be the length of the curve @xmath184 on the interval @xmath185 $ ] .
for all real numbers @xmath186 we have : @xmath187 the mapping @xmath188 is continuous on @xmath189 and admits partial derivatives with respect to @xmath186 , continuous on @xmath189 .
studying the uniform convergence of @xmath190 when @xmath163 tends to @xmath164 is thus sufficient to proof the derivability of @xmath166 with respect to @xmath191 .
we have @xmath192 for all @xmath193 , and @xmath194 for all @xmath195 ; thus @xmath196 for all @xmath197 .
the relation @xmath198 and the equation imply @xmath199 thus we have for all @xmath135 and all @xmath132 : @xmath200 @xmath201 @xmath202 integrating the last inequality with respect to @xmath132 gives @xmath203 because of @xmath204 , we obtain @xmath205 even by changing the horocycle , there comes for all @xmath135 and all @xmath206 : @xmath207 @xmath208 we get for all @xmath209 and all @xmath147 $ ] : @xmath210 @xmath211 @xmath212 .\ ] ] from the upper estimate @xmath213 results the existence of a constant @xmath214 depending only from @xmath215 , @xmath216 , @xmath217 , @xmath56 and @xmath55 such that for all @xmath218 positive and @xmath147 $ ] we have @xmath219 from the corollary [ cor : courbure ] , the equations and imply : @xmath220 @xmath221 the uniform cauchy criterion on the compacts of @xmath167 implies the convergence when @xmath163 goes to @xmath164 of @xmath222 to a fonction @xmath223 continuous on @xmath167 .
the equation ensures the existence and continuity of the partial derivative of @xmath223 with respect to @xmath224 .
the equation implies the upper estimate for @xmath209 : @xmath225 thus @xmath226 which ensures the convergence , uniform over the compacts of @xmath227 , of the family of mappings @xmath228 , when @xmath163 goes to @xmath164 , to a continuous mapping of @xmath229 .
the relations , and @xmath230 imply the uniform convergence , over the compacts of @xmath231 , of the family of mappings @xmath232 , when @xmath163 goes to @xmath164 , to a continuous mapping .
this gives the condition of derivability with respect to @xmath132 for the function @xmath223 , thus the derivability with respect to @xmath191 anounced for the function @xmath166 .
the trivial relation @xmath233 gives the derivability with respect to @xmath172 .
[ def : transfert ] with the assumptions of the theorem [ thm : fluc ] , for @xmath58 , @xmath234 , the limit mapping calculated in the theorem [ thm : fluc ] is called _ stable transfer _ from @xmath235 to @xmath26 and is written down : @xmath236 the _ extended stable transfer _ from @xmath235 to @xmath26 is the mapping which to @xmath237 associates @xmath238 [ cor : regutransfert ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable space @xmath113 is uniformly bounded .
the stable transfer and extended stable transfer are of @xmath20-class on a given centre - stable manifold .
this follows from the derivability of the stable transfer stated in the theorem [ thm : fluc ] . the symplectic structure of the space of geodesics ( section [ sec : symp ] ) leads to the following result .
[ thm : croisefluc ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable @xmath113 ( respectively centre - unstable @xmath114 ) spaces is uniformly bounded . for a given point
@xmath170 at infinity , and unitary vectors @xmath26 , @xmath235 belonging to the centre - stable manifold @xmath239 determined by @xmath170 , the family of mappings @xmath240 converges uniformly when @xmath163 goes to @xmath164 to a continuous mapping .
the limit mapping admits a partial derivative with respect to @xmath235 ( in the direction @xmath239 ) , continuous with respect to @xmath241 .
following the formula , for @xmath242 , @xmath243 , we have @xmath244 because the gradient of @xmath245 is uniformly bounded , the quotient @xmath246 tends to one uniformly over the compacts , when @xmath163 goes to @xmath164 , this fact implying the uniform convergence over the compacts : @xmath247 the regularity of the limit results from the theorem [ thm : fluc ] .
[ def : transfu ] with the assumptions of the theorem [ thm : croisefluc ] , for @xmath58 , @xmath234 , the limit mapping calculated in the theorem [ thm : croisefluc ] is called _ unstable transfer _ from @xmath235 to @xmath26 and is written down : @xmath248 the _ extended unstable transfer _ from @xmath235 to @xmath26 is the mapping which to @xmath237 associates @xmath249 the following result is immediate .
[ cor : regutransu ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable @xmath113 ( respectively centre - unstable @xmath114 ) spaces is uniformly bounded .
the unstable transfer and the extended unstable transfer are of @xmath20-class on a given centre - stable manifold .
[ thm : lyap ] let @xmath1 be a compact , connected , oriented , surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 with jacobi endomorphism satisfying the pinching condition . if the ( future ) lyapounoff exponents of the magnetic flow are defined at an element @xmath26 of @xmath44 , then they are defined and constant along its centre - stable manifold @xmath72 .
the compacity insures the uniform bounds over the gradients of @xmath52 , @xmath114 and @xmath113 .
it is sufficient to pass to the universal cover of @xmath1 and to apply the theorems [ thm : fluc ] and [ thm : croisefluc ] .
in this section are collected some tools for the section [ sec : linea ] ; the notations and the assumptions are those of the section [ sec : fluc ] , particularly of the proof of the theorem [ thm : fluc ] .
let @xmath250 be the parallel transport along the curve @xmath251 which sends @xmath252 onto @xmath253 .
let @xmath254 be the angle between the vectors @xmath255 and @xmath256 .
composing @xmath257 with the rotation of angle @xmath258 yields the isometry @xmath259 which sends the direct orthonormal basis @xmath260 onto the direct orthonormal basis @xmath261 .
the mapping @xmath262 is called _
horocyclic transport_. the horocyclic transport is continuous on @xmath263 . in this section
the control of @xmath264 is precised in different ways .
[ lem : transport1 ] with the above notations , we have for all @xmath160 , @xmath147 $ ] , @xmath243 : @xmath265 for all @xmath160 , @xmath110 , @xmath266-\varepsilon , \varepsilon [ $ ] , @xmath267 , @xmath268 , we have @xmath269 the field @xmath270 is the solution @xmath271 of @xmath272 thus , due to the upper estimate , it satisfies the differential inequality @xmath273 therefore @xmath274 we have @xmath275 @xmath276 thus @xmath277 @xmath278 @xmath279 , \ ] ] and @xmath280 @xmath281 @xmath282 .\ ] ] we deduce from this @xmath283 @xmath284 the formula implies : @xmath285 following , due to the upper estimate , the rotation satisfies the differential inequality @xmath286 @xmath287 the definition of @xmath264 permits to conclude .
we have also the upper estimate @xmath288 @xmath289 from which follows the : [ lem : transport2 ] with the preceding notations , we have for all @xmath160 , @xmath147 $ ] , @xmath243 : @xmath290 the horocyclic transport presents some uniformity .
[ lem : transport3 ] let @xmath291 be a field of linear endomorphisms of class @xmath20 over @xmath44 , bounded in @xmath20-norm . with the preceding notations , for all @xmath160 ,
@xmath147 $ ] and @xmath292 , we have @xmath293 @xmath294 we have @xmath295 @xmath296 which is bounded from above , following the lemma [ lem : transport2 ] , by @xmath297 @xmath298 the covariant derivative with respect to @xmath132 of @xmath299 admits the same first term as above for upper bound .
the quantity that we aim to estimate in the lemma annulates at @xmath300 , the upper estimate of the statement is obtained by integrating the expression @xmath301 over @xmath132 from @xmath155 .
[ def : poussee ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable space @xmath113 is uniformly bounded . for @xmath58 , @xmath110 , the _
stable push _ is the mapping @xmath302 which to every @xmath22 associates @xmath303 .
[ thm : linea ] with the assumptions of the definition [ def : poussee ] , for all unitary vector @xmath58 , there exists a unique mapping @xmath24 from @xmath1 into @xmath304 such that : @xmath305 writing down @xmath306 .
moreover , the mapping @xmath307 is continuous from @xmath308 into @xmath43 .
the conditions @xmath309 , @xmath310 and @xmath311 ensure naturally the uniqueness of the solution .
let @xmath26 be in @xmath44 .
let us define @xmath312 , @xmath313 .
taking the notations of the sections [ sec : fluc ] and [ sec : transport ] , we have @xmath314 and we define : @xmath315 @xmath316 we have @xmath317 thus @xmath318 @xmath319 the equation implies for @xmath209 : @xmath320 } \hskip -20 pt c_3 l_{-\varepsilon } ^\varepsilon ( c ) e^{-q_1 t } \left| \left\langle \frac{dc}{ds}(\sigma ) , n_\theta ( \sigma , 0 ) \right\rangle \right|d\sigma \leq c_3 l_{-\varepsilon } ^\varepsilon ( c)^2 e^{-q_1 t } .\ ] ] by taking @xmath321 and with the foregoing upper estimate and the corollary [ cor : courbure ] we get : @xmath322 for every compact @xmath323 of @xmath308 , the family of continuous mappings @xmath324 satisfies the uniform cauchy condition over @xmath323 when @xmath163 tends to @xmath164 .
it converges thus towards a mapping continuous over @xmath308 .
establishing the formulas @xmath309 and @xmath310 is immediate .
there remains to proof the formula @xmath311 . for @xmath325 , @xmath326 ,
@xmath327 , if @xmath328 is the _ geodesic _ jacobi field along the geodesic curve directed by @xmath329 , such that @xmath330 and @xmath331 , the linear mapping tangent to the exponential satisfies ( @xcite , 3.46 p.117 ) : @xmath332 from the bounds on the gauss cuvature results the existence of a constant @xmath333 such that for all @xmath334 $ ] : @xmath335 the derivative of the exponential in every zero vector is the identity of the tangent vector space , and @xmath336 is the identity of @xmath253 . even by situating in a chart in the neighbourhood of @xmath337 , there exists a constant @xmath338 such that : @xmath339 \ [ \leq \int_{\sigma = 0}^s \left\vert \frac{\partial } { \partial s } \exp_{\pi \psi ^\kappa _ tv}^{-1 } p_\theta ( t,\sigma ) - \chi ( \theta , \sigma , t)\frac{\partial p_\theta } { \partial s } ( t,\sigma ) \
d\sigma \right\vert \ ] \ [ \leq c_5 d\left ( \pi \psi ^\kappa _ tv , p_\theta ( t , s ) \right ) ^2 + c_1 e^{-q_1 t } \int_{\sigma ' = 0}^s\int_{\sigma = 0}^{\sigma ' } \left\vert \frac{\partial p_\theta } { \partial s } ( t,\sigma ) \right\vert \left\vert \frac{dc } { ds } ( \sigma ' ) \right\vert \ d\sigma d\sigma ' \ ] \ [ \leq c_5 d\left ( \pi \psi ^\kappa _ tv , p_\theta ( t , s ) \right ) ^2 + c_5 l_0^\varepsilon ( p_\theta ( t , . ) ) e^{-q_1 t } l_{-\varepsilon } ^\varepsilon ( c).\ ] ] the first term of the last member comes from the effect of the bounds over the gauss curvature on the exponential , and the second term comes from the lemma [ lem : transport2 ] by carrying out two successive integrations .
we deduce the inequality : @xmath340 @xmath341 \int_{\sigma \in [ 0,s ] } \hskip -20 pt \frac{y_-(v(p_\theta ( t,\sigma ) , \theta ) , t)}{y_-(v , t ) } \left\vert \frac{dc}{ds}(\sigma ) \right\vert \ d\sigma .\ ] ] the quantity @xmath342 is bounded from above by @xmath343 which is inferior or equal to @xmath344 .
according to the theorem [ thm : fluc ] , there exists a constant @xmath345 ( depending on @xmath346 ) such that the last integral above is bounded by @xmath347 independently of @xmath163 , from which we get the upper estimate : @xmath348 the limit stated in the formula @xmath311 of the theorem is thus obtained .
the linearization cooperates to some extent with the magnetic flow .
[ cor : linea ] for @xmath58 , @xmath22 , @xmath349 , we have @xmath350 this immediately results from the construction of the theorem [ thm : linea ] .
[ thm : regulin ] let @xmath1 be a complete , connected , oriented , simply connected surface , equipped with a @xmath4-riemannian metric @xmath96 whose curvature is negatively pinched : @xmath40 , and with a @xmath4-magnetic field @xmath52 whose @xmath20-norm is bounded , with jacobi endomorphism satisfying the pinching condition , and such that the gradient of the centre - stable @xmath113 spaces is uniformly bounded . for @xmath58 and @xmath351 , the mapping @xmath24 is of @xmath352 over @xmath1 , and with the extended stable transfer @xmath353 coming from the definition [ def : transfert ] , its derivative at @xmath354 is : @xmath355 with the notations of the sections [ sec : fluc ] and [ sec : linea ] , we have @xmath356 which converges when @xmath163 tends to @xmath164 towards @xmath357 the longitudinal component of the derivative of @xmath24 is @xmath26 , and the continuous derivability of the derivative of @xmath24 results from the corollary [ cor : regutransfert ] .
[ cor : diffeo ] with the same assumptions as in the theorem [ thm : regulin ] , for all @xmath58 , the mapping @xmath24 is a @xmath352-diffeomorphism .
it is clear that the mapping @xmath24 is surjective .
it is a local @xmath352-diffeomorphism according to the theorem [ thm : regulin ] , it is therefore a cover of @xmath304 ( which is isomorphic to @xmath21 ) , and consequently it is a diffeomorphism . we may notice that the mapping @xmath358 with @xmath172 fixed can not be absolutely continuous .
let @xmath26 and @xmath329 be two unitary tangent vectors belonging to a same unstable manifold , such that @xmath172 belongs to the curve directed by @xmath26 .
the vectors @xmath76 and @xmath359 are arbitrarily close when @xmath163 tends to @xmath360 .
the orthogonal component of @xmath361 is zero and the norm of the orthogonal component of @xmath362 tends to @xmath164 when @xmath163 tends to @xmath360 .
[ thm : reguconj ] let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface .
let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath15 are supposed to have negative jacobi endomorphisms ( thus satisfying the pinching condition ) .
if the two magnetic flows have the same marked length spectrum , then they are conjugated by a bi - hlder - continuous homeomorphism @xmath16 from @xmath17 onto @xmath18 which preserves the lyapounoff exponents of periodic orbits .
the gradient of the centre - stables spaces @xmath113 and of the centre - instables spaces @xmath114 are uniformly bounded because the unitary tangent bundles are compact .
the jacobi endomorphisms satisfie the pinching condition for the same reason .
the existence of the bi - hlder - continuous conjugacy @xmath16 from @xmath17 onto @xmath18 is well - known @xcite .
let @xmath26 be a @xmath363-periodic vector for @xmath364 ; its conjugate @xmath365 is @xmath363-periodic for @xmath366 and the conjugacy @xmath16 maps @xmath71 onto @xmath367 .
the linearizations associated to @xmath368 are written down @xmath369 for @xmath6 .
let @xmath370 be the bijection , restricted to the orthogonal spaces identified to the real line , defined as follows : @xmath371 the lyapounoff exponents are written down : @xmath372 let us denote : @xmath373 according to the corollary [ cor : linea ] , for each of the two flows , every @xmath363-periodic vector @xmath26 satisfies : @xmath374 for all @xmath375 and @xmath376 we therefore have : @xmath377 because the conjugacy @xmath16 is bi - hlder - continuous and because the linearizations are of @xmath352-class , there exist two constants @xmath378 and @xmath3790,1]$ ] such that for all @xmath380 $ ] we have : @xmath381 thus for all natural integer @xmath382 we have : @xmath383 which implies for all @xmath384 : @xmath385 by making @xmath382 tend to @xmath164 there comes @xmath386 from which comes @xmath387 thus the lyapounoff exponents coincide on the periodic orbits .
[ thm : spm ] let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface .
let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 , .
let @xmath7 be a @xmath4-magnetic field over @xmath1 .
the magnetic flow @xmath8 is supposed to have a negative jacobi endomorphism ( thus satisfying the pinching condition ) . if the magnetic flow @xmath9 and the geodesic flow @xmath10 have the same marked length spectrum , and if the surface @xmath1 has the same total volume for the two metrics , then the two metrics are isotopic , which means that one is the image of the other by a diffeomorphism @xmath11 of @xmath1 , homotopic to the identity , and the magnetic field @xmath7 is zero .
this results from the corollary [ cor : reguconj ] : if the two volumes are equal , and the flows @xmath364 and @xmath388 are conjugate by an absolutely continuous homeomorphism , the result is known @xcite .
[ thm : rigir ] let @xmath1 be an oriented surface diffeomorphic to @xmath21 , equipped with two @xmath4-riemannian metrics @xmath2 and @xmath3 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath28 are supposed to have negative jacobi endomorphisms satisfying the pinching condition ( the @xmath20-norms of @xmath7 and @xmath14 are thus bounded ) , and the gradient of the centre - stable @xmath29 spaces and the gradient of the centre - unstable @xmath30 spaces for @xmath6 are supposed to be uniformly bounded .
if there exist a diffeomorphism @xmath31 and a point @xmath22 satisfying @xmath32 then the two metrics are images one of each other by @xmath11 , and so are the two magnetic fields : @xmath33 .
let @xmath329 be in @xmath17 , @xmath389 and @xmath390 .
let @xmath391 be in @xmath392 such that @xmath393 .
we have necessarily @xmath394 : thus the mapping @xmath11 is an isometry .
its differential conjugates the flows , therefore it preserves the geodesic curvature of the orbits , which means the magnetic fields .
[ thm : rigicomp ] let @xmath1 be a closed ( compact without boundary ) , connected , oriented surface .
let @xmath2 and @xmath3 be two @xmath4-riemannian metrics over @xmath1 whose curvatures are negatively pinched : @xmath5 for @xmath6 .
let @xmath7 , @xmath14 be two @xmath4-magnetic fields over @xmath1 .
the two magnetic flows @xmath8 and @xmath34 are supposed to have negative jacobi endomorphisms ( thus satisfying the pinching condition ) . if there exist two vectors @xmath35 , @xmath36 and a @xmath20-diffeomorphism @xmath31 homotopic to the identity , of which a lift @xmath37 over @xmath38 satisfies @xmath39 , then the two metrics are isotopic , transported by @xmath11 , and so are the two magnetic fields : @xmath33 .
let @xmath329 be in @xmath72 , @xmath389 .
we have necessarily @xmath395 .
since @xmath329 is chosen arbitrarily in the centre - stable manifold of @xmath26 , we may chose a vector @xmath329 whose projection on @xmath17 has a dense orbit under @xmath364 when the time tends to @xmath360 ; we may also replace @xmath329 by every element of its orbit .
writing down @xmath396 the covering of @xmath38 onto @xmath1 , we obtain @xmath397 .
the differential of @xmath11 thus preserves the norm on a dense subset of the unitary tangent bundle ; because it is assumed continuous , the mapping @xmath11 is an isometry .
the differential of @xmath37 preserves the geodesic curvature on the whole orbit of @xmath329 , thus by projecting and by a density argument , we deduce that @xmath398 is the composed of @xmath399 by @xmath11 . | the main result presented here is that the flow associated with a riemannian metric and a non zero magnetic field on a compact oriented surface without boundary , under assumptions of hyperbolic type , can not have the same length spectrum of topologically corresponding periodic orbits as the geodesic flow associated with another riemannian metric having a negative curvature and the same total volume .
the main tool is a regularization inspired by u. hamenstdt s methods
. 30 pt universit de nantes , dpartement de mathmatiques , laboratoire jean leray u. m. r. 6629 , 2 , rue de la houssinire , bp 92208 , f-44322 nantes cedex 03 . [email protected] |
current power grid is being modernized into smart grid by enabling the integrations of advanced information and communication technologies relating power consumption and storage @xcite . integrating new technologies to power grid
enables us to control the grid in a more flexible and efficient manner .
for example , two - way communication and information technologies allow the use of demand response and automatic meter reading , respectively . another important feature is to merge variable renewable energy sources such as wind , solar , and wave energy generators to the grid .
it is well known that renewable energy attracts more attentions due to its cleanness and profitability .
worldwide 144 countries now have their own political targets for increasing shares of renewable energy generations and especially european union targets 20% of renewable energy shares by 2020 @xcite
. however , variable renewable integration imposes challenges to operation and management of power grid . as pointed out in @xcite , renewable energy can not be fully predicted .
there are many wind forecasting methods which are well surveyed in @xcite , but forecasting errors could be one source for uncertainties and become dominant as time horizon increases .
another source of uncertainties is the increasing use of intermittent renewable energy sources in a distributed way @xcite .
even more it fluctuates fast and largely @xcite .
for example , wind power generation contains @xmath0 high frequency components @xcite .
fast changing generation of renewable energy with unknown amount of power threatens the stability of the power grid @xcite .
furthermore , it is difficult for a synchronous generator to track fast changing signals because it has a high inertia . in this paper , we consider distributed frequency regulation problems in smart power grid with integration of wind power energy , demand response , and battery energy storage system ( bess ) .
designed distributed controllers are expected to regulate angular frequencies to a desired common constant under fast changing and unknown wind generations . in particular
we study the following two problems .
the first problem is _ distributed robust frequency regulation _ with known frequencies of wind power generations .
the known frequency assumption is relaxed in the second problem called _ distributed robust adaptive frequency regulation_. we propose the distributed controllers with ( adaptive ) internal model . under the proposed distributed controllers , mechanical generation and elastic load are able to completely reject unknown wind power generations , regulating network wide angular frequencies to a desired constant . to our best knowledge
, there is no existing controller which achieves this performance when frequencies of wind power generations are unknown .
moreover , this paper is the first time to investigate distributed adaptive internal model control .
furthermore , we match the trackable frequencies of wind generations with different grid components to balance the power generations and demand .
this idea is inspired by the fact that each controllers have different inertias and thus they have different ranges of trackable frequencies of signals .
we first split wind power generation signals into low , medium , and high frequency parts , which is based on van der hoven s horizontal wind speed spectrum model @xcite and trigonometric function formulas . at each bus ,
a bess is interconnected with wind power generations to filter out the high frequency parts .
the low and medium components are asymptotically reconstructed via a local internal model . at generator bus ,
the reconstructed signals are tracked by two different controllers .
mechanical power generations track the low frequency parts and elastic demands are driven by real - time prices to track the medium frequency parts . at load bus ,
elastic demands track both the low and medium frequency parts of wind power .
the remaining part of our solution of each problem is described as follows .
_ solution 1 : robust frequency regulation .
_ we first convert the global frequency regulation problem into a global stabilization problem .
the augmented system at each generator bus consists of synchronous generator , demand response , and internal model .
the augmented system at each load bus consists of demand response , and internal model .
they are in a lower triangular structure . in the same coordinate transformation , redundant terms inspired by back - stepping
are added to stabilize the system from the outer state to inner state .
the state of each augmented system becomes input to state stable ( iss ) with respect to the states of neighboring augmented systems .
we choose control gains such that state transition matrix be hurwitz to stabilize the networked system .
it has been shown that , even when each control authorities can access to their local parameters only , the stabilizing controllers can be designed always .
in this case , the control gains are designed such that local iss gain functions are contraction mappings .
a newly derived distributed constrained small gain theorem ensures that the collection of local contraction mappings are sufficient for the network stability .
consequently , the proposed distributed controllers with the local internal models ensure that all the states are exponentially stable with respect to the desired manifolds .
_ solution 2 : robust adaptive frequency regulation .
_ designing an internal model requires a constant matrix which depends on frequencies of wind power generations .
the frequencies of wind generations , however , are unknown and thus it is required to estimate the matrix .
we design a projected adaptive parameter estimator for this purpose .
we again solve the problem by converting the frequency regulation problem into a global stabilization problem with back - stepping inspired coordinate transformation terms to make the augmented system iss with respect to the neighboring states .
the proposed distributed controllers with adaptive internal model guarantee asymptotic convergence of all the states to their manifolds instead of exponential stability _
solution 1 _ can guarantee .
moreover , it is assumed that the synchronous generator model can be slightly simplified .
this paper is enriched from our preliminary results @xcite .
first of all , we relax the known frequency assumption ; i.e. , _ solution 2_. secondly , we relax the assumption on the bus .
previously , we assume that each bus must have a generator and load . to make the problem more realistic , in this paper
, we consider more types of buses as shown in figure [ type_table ] .
thirdly , simulations are applied to a more complex and realistic topology .
lastly , there are detailed proofs which were omitted from the preliminary results . distributed control methods on power system
have been studied extensively . primary control
( droop - control ) scheme adjusts mechanical power generations to stabilize the frequencies via governors in a decentralized way @xcite . to improve restoration of angular frequency , secondary control
is introduced .
secondary control ( automatic generation control ) tunes the set - point of governors to achieve nominal frequencies in a centralized way @xcite .
our control problems consider the primary and secondary controls at the same time in a distributed way under the existence of fast changing and unknown signals .
there have been attempts attenuating the impact of the uncertainties induced by renewable energy , power demand , and other sources .
robust control techniques include riccati equation @xcite , @xmath1 control @xcite , lmis @xcite , etc .
also , it was shown that oscillations induced by load disturbances can be effectively alleviated by a bess @xcite .
another way to address the uncertainties is to reject disturbances and/or asymptotically track reference signals .
internal model has been widely used to achieve the goal @xcite . on power systems , recent paper @xcite
uses internal model principle to address uncertainty induced by time varying demands .
different from our work , the paper @xcite only deals with secondary control on a simpler ( second order ) model with voltage dynamics , and guarantees regional asymptotic convergence rate assuming that frequencies of demand signals are known . only with the known frequency assumption ,
our first solution guarantees global exponential convergence of the fourth order dynamics .
we further extend that , with unknown frequencies , our second solution guarantees global asymptotic convergence of the third order dynamics . on top of this
, we split unknown signals into three parts according to their frequency ranges to be tracked and filtered by two different controllers and bess .
traditional approach to balance power demand and generation is to adjust power generations .
some of recent studies focus on achieving the power balance via demand controlling such as optimal load control @xcite , load switch @xcite , etc .
the features and overview of demand side control is well introduced in @xcite .
it has been proposed that load side control can be faster and have a large capacity through implementing a switching device so that such control scheme is used to assist conventional mechanical generation system @xcite . in this paper ,
real - time demand response is used to track medium ( and low ) frequency parts of wind power generations at generator bus ( at load bus ) .
this paper is organized as follows .
we start by introducing all the system models in section [ sec : problem - formulation ] . based on the illustrated models , the two problems ; i.e.
, _ robust frequency regulation _ and _ robust adaptive frequency regulation _ problems are described in section [ sec : controller ] .
we propose solutions for the problems in section [ sec : sol1 ] and [ sec : sol2 ] , respectively .
stability is formally analyzed .
lastly , section [ sec : sim ] presents simulations of frequency regulation on the ieee 68-bus test system to illustrate the performance of our controllers .
we define @xmath2 } \triangleq \sup_{t_1 \leq t \leq t_2 } \|x(t)\|$ ] .
matrix @xmath3 denotes the identity matrix with dimension @xmath4 .
@xmath5 denotes a block matrix having @xmath6 to @xmath7 as main diagonal blocks and the off - diagonal blocks are all zero matrices .
notation @xmath8 indicates the @xmath9 dimensional zero row vector . for a set @xmath10
, its cardinality is described by @xmath11 . for a positive constant @xmath12 and a vector @xmath13^t$ ] ,
function @xmath14 is defined by @xmath15^t$ ] where , for @xmath16 , @xmath17 and @xmath18 if @xmath19 , @xmath20 , and @xmath21 , respectively .
frobenius norm is a vector norm defined by @xmath22 where @xmath23 is the @xmath24 row and @xmath25 column of matrix @xmath26 .
we use @xmath27 norm @xmath28 if it is not specified .
the following definition of input - to - state stability is adopted from @xcite .
the system @xmath29 is input - to - state stable ( iss ) if there are a class @xmath30 function @xmath31 and a class @xmath32 function @xmath33 such that for all @xmath34 , @xmath35})\end{aligned}\ ] ] for any initial state @xmath36 and any bounded input @xmath37 .
the present section introduces system models on which our problem based .
we illustrate the models of power network , synchronous power generator , wind power , bess , and demand response . to avoid notational confusion
, we begin by introducing system variables and parameters as follows . @xmath38 power network is described by an undirected graph @xmath39 where @xmath40 denotes the set of buses in the network and @xmath41 denotes the set of transmission lines between buses .
the set @xmath42 denotes the time invariant set of neighboring buses of @xmath43 ; i.e. , @xmath44 .
each bus is either a generator bus @xmath45 , load bus with nonzero load @xmath46 or load bus with zero load @xmath47 where @xmath48 , @xmath49 , and @xmath50 denote the set of corresponding buses .
buses @xmath51 belong to a local control authority .
the generator bus consists of mechanical generation @xmath52 , wind power generation @xmath53 , and load @xmath54 .
the control authority @xmath45 can control mechanical generation via reference input and elastic load via power pricing .
the load bus @xmath46 consists of wind power generation @xmath53 , and load @xmath54 .
the control authority @xmath46 can control elastic load via power pricing .
the load bus @xmath47 does not belong to any control authority and its local system can not be controlled . integrating consumer as a part of grid component via real time pricing
can reduce the need for some infrastructures such as new power plants while maintaining the quality and reliability of electricity @xcite .
integration of demand response as an auxiliary control method is suggested in @xcite .
we adopt the fourth order model of synchronous power generator from @xcite with integrating wind power generation signals @xmath55 as a negative load or a part of generation shown in @xcite .
consider , for @xmath45 , @xmath56 where @xmath57 and @xmath58 .
the first equation in indicates the evolution of phase angle @xmath59 .
the second equation is referred to as swing dynamics , indicating frequency fluctuations due to power imbalances .
a general control objective is to regulate all the angular frequencies @xmath60 to a constant set point @xmath61 . to achieve this
, it is required to balance the power generations and demand , otherwise frequency @xmath60 changes .
for example , frequency @xmath60 increases if the aggregate of power generation surpasses power demand @xmath62 , and decreases with a reversed inequality .
the third and forth equations denote turbine governor dynamics with reference input @xmath63 .
governor regulates torque of the rotor and eventually regulates angular frequency @xmath60 to a desired set point @xmath64 .
the fourth order synchronous generator model can be simplified to the third order model @xcite : @xmath65 where @xmath66 is the input of this simplified system . in the case of load bus @xmath46
, the dynamic equation model can be expressed by the first and second differential equation of with no mechanical generation @xmath67 as proposed in @xcite : @xmath68 where , in this load bus model , values @xmath69 and @xmath70 are effective moment and damping constants of a postulated load model . the dynamic equation model of bus @xmath47 is also described by with @xmath71 . as @xcite
, consider the wind power generation system of control authority @xmath43 described by @xmath72 where @xmath73 denotes turbine s power coefficient and each wind power generator might have different coefficient .
it is pointed out in @xcite that , due to unknown fluctuations of wind speed @xmath74 , wind power generation @xmath75 can not be fully predicted .
therefore , it is difficult to track the power imbalances caused by wind power fluctuations . to deal with this problem ,
we first divide wind power generations into three parts according their frequency ranges : @xmath76 where @xmath77 , @xmath78 , and @xmath79 denote low , medium , and high frequency parts of wind generations .
the frequency partitioning is based on van der hoven s horizontal wind speed spectrum model @xcite and trigonometric function formulas .
there are many studies including @xcite based on this wind speed spectrum model .
according to the wind speed spectrum model , wind speed can be approximated well by a combination of sinusoidal functions . by applying trigonometric function formulas to the first equation in
, we can also approximate wind power generations as a combination of sinusoidal functions . considering dominant frequencies of wind speed : @xmath80 and @xmath81
, we suggest that approximate ranges are @xmath82 for low , @xmath83 for medium , and all the higher ranges for high frequency generations . by van der
hoven s wind speed spectrum model @xcite , we assume that @xmath77 and @xmath78 are generated by a partially known exosystem which is described by @xmath84 where states @xmath85 @xmath86^t\in{\mathds{r}}^{2\ell}$ ] denote exogenous signals .
each state represents a sinusoidal signal .
values @xmath87 and @xmath88 denote the maximum number of low / medium frequency parts of wind power generations , and @xmath89 .
matrix @xmath90 is defined by @xmath91 where @xmath92 , \ \
\phi_{i , l}^m \triangleq \left [ \begin{array}{cc } 0 & 1 \\ -(\rho^{m}_{i , l})^2 & 0 \\ \end{array } \right].\end{aligned}\ ] ] the values @xmath93 for @xmath94 and @xmath95 for @xmath96 denote low and medium frequencies , respectively .
the dependency of @xmath97 on @xmath98 is emphasized where @xmath98 is the set of all @xmath93 and @xmath95 . in exosystem , @xmath55 is a linear combination of sinusoidal functions with frequencies @xmath93 and @xmath95 .
frequencies could be zero , i.e. , @xmath99 for some @xmath100 .
for nonzero frequencies , it always holds that @xmath101 for @xmath102 and @xmath103 such that @xmath104 . for example , assume @xmath105 and @xmath106 $ ] .
exosystem generates @xmath107 and @xmath108 where @xmath109 , @xmath110 , @xmath111 , and @xmath112 are constants decided by initial conditions of
. then wind power generations are given by a combination of sinusoidal functions : @xmath113 .
the control authority @xmath114 can not measure exosystem states @xmath115 , and initial conditions @xmath116 .
frequencies @xmath117 are known in the first problem , and unknown in the second problem .
it implies that amplitudes and phase shifts of sinusoidal functions are unknown because they are determined by unknown initial conditions of .
however , the structure of the exosystem and @xmath118 are known to the control authorities .
we have proposed that wind power generation signals can be divided into low , medium and high frequency parts . however , synchronous generators and demand response can not track fast changing high frequency wind power generations . to increase generation quality
, it is proposed that a bess can be used to smooth renewable energy fluctuations in @xcite . to filter out @xmath119
, we place the bess at the output of wind power generations .
the bess acts as a low pass filter ; i.e. , the signal looses its high frequency components @xmath120 by going through the bess .
consider the transfer function of the bess , as @xcite , @xmath121 where @xmath122 and @xmath123 denote the laplace transforms of the input and output power of the bess , respectively .
the effectiveness of transfer function is determined by the time constant @xmath124 .
therefore , it is important to choose a proper constant @xmath124 such that the transfer function sufficiently suppresses the high frequency wind power generation @xmath79 . with input ,
the output of the bess is @xmath125 .
local power demand @xmath54 of bus @xmath51 includes all the local power consumptions of consumers , instantaneous and aggregate demands .
demand @xmath54 can be separated into elastic demand @xmath126 and inelastic demand @xmath127 as shown in @xcite ; i.e. , @xmath128 . under the assumption that marginal benefit function @xmath129 of consumer is negatively proportional to the power consumption
, elastic demand @xmath126 is governed by the following dynamics : @xmath130 where @xmath131 parameters @xmath132 and @xmath133 denote power consumer benefits and @xmath134 is a consumer response constant .
real - time electricity price @xmath135 can be used as a controller . in
, the power demand would increase when marginal benefit is greater than the given price @xmath135 . the way to choose parameters @xmath132 and @xmath133 , and constant @xmath134 is introduced in @xcite . for notational simplicity ,
we introduce pricing controller @xmath136 instead of @xmath135 .
we assume that time varying inelastic power demand @xmath127 , and constants @xmath137 , @xmath138 are known to control authority @xmath114 .
all the power generations and loads are coupled via swing dynamics , the second equation of .
the schematic connections are described in figure [ gl_bus ] .
at generator bus , reference power @xmath63 and pricing control @xmath136 are control inputs available to the local control authority .
all the states in the diagram are used to determine inputs ( state feedback control ) and the control authorities objective is to regulate all the states to desired manifolds . at load bus @xmath46
, available control input is pricing control @xmath136 only . in our system setting ,
wind power generations are allowed to be zero for any bus .
therefore , each bus has a freedom to remove the wind power system with the bess .
this implies that the system without wind power generators is a special case of the wind power integrated system .
moreover , if medium frequency parts of wind power generations at a generator bus are missing , then demand response can also be removed at the same bus because the role of demand response at the generator bus is only to follow medium frequency wind generation .
we consider two frequency regulation problems . both problems aim to regulate angular frequencies to a constant and phase angle differences between neighboring buses to specified constants . in the first problem illustrated in section [ sec
: formulation ] , we assume that frequencies of wind power generation signals are known to control authorities .
this assumption is relaxed in the second problem in section [ sec : formulation2 ] . in the second problem
, it is only assumed that the maximum number of frequencies is known . as a preliminary step to solve the two problems ,
a local internal model is designed in section [ sec : internal ] to reconstruct unknown wind power generations .
our objective is to regulate frequency @xmath60 to a desired constant @xmath61 ; e.g. , @xmath139 , and @xmath59 to @xmath140 under unpredictable wind generations .
constant phase shift @xmath141 could be different from @xmath142 for @xmath143 . in the generator bus , to achieve this goal
, @xmath52 is expected to track low frequency wind power generation @xmath77 via input @xmath63 and @xmath126 is expected to track medium frequency wind power generation @xmath78 via pricing control @xmath136 . by examining all the system models described in section [ sec : problem - formulation ] , we have desired states and inputs which are called manifolds : @xmath144 where @xmath145 .
it is empasized that manifolds @xmath146 @xmath147^t$ ] and @xmath148 $ ] are functions of unknown wind power energy @xmath77 and @xmath78 . to ensure the existence of the manifold
, we have assumptions [ asmp1 ] and [ asmp2 ] .
inelastic power demand @xmath127 is a twice differentiable function for @xmath149 .
[ asmp1 ] for the set of desired power flow @xmath150 for @xmath151 , there exists a set of solutions @xmath141 and @xmath142 for equation @xmath145 .
for all the buses @xmath47 , desired incoming and outgoing power flows are balanced ; i.e. , @xmath152 for @xmath47 .
[ asmp2 ] we desire to design distributed controllers which regulate the system states
@xmath153 @xmath154^t \triangleq [ \theta_i(t ) , w_i(t ) , p_{m_i}(t ) , p_{v_i}(t ) , $ ] @xmath147^t$ ] and inputs @xmath155^t$ ] to their manifolds @xmath156 and @xmath157 exponentially for @xmath158 .
however , it is challenged by the fact that exosystem states @xmath115 , and wind power generations @xmath77 and @xmath78 can not be directly measured .
hence , manifolds are not measurable neither . to address the challenge
, we will design local internal models to reconstruct via state and input measurements . in this problem
, we assume that all the frequencies of wind power generation signals are known to control authorities .
all the frequencies @xmath117 for @xmath159 of wind power generations @xmath55 are known to the local control authority @xmath114 .
[ atsm1 ] in the load bus @xmath46 , elastic demand @xmath160 is expected to track both low and medium frequency wind power generation and stabilize the frequency .
therefore , the manifolds for states @xmath59 and @xmath60 are identical to those of but the manifold for the elastic demand @xmath126 is @xmath161 in .
we desire to design distributed controllers which regulate the system states @xmath162^t \triangleq [ \theta_i(t ) , w_i(t ) , p_{l_i}^e(t)]^t$ ] and inputs @xmath163 to their manifolds @xmath164^t$ ] and @xmath165 exponentially for @xmath166 .
the objective of _ problem 2 _ remains the same as that of _ problem 1 _ but there are a couple of distinctions . first , control authorities do not know the frequencies of wind power generation signals in this problem ; i.e. , assumption [ atsm1 ] is relaxed .
secondly , to solve this problem , we use the simplified synchronous generator model and suppose that the following is true : the control authorities know the maximum number @xmath167 of low / medium frequency parts of wind power generations .
[ atsm3 ] moreover , we assume that maximum frequency @xmath168 such that @xmath169 for @xmath102 is known because it can be estimated by the chosen time constant @xmath124 . in the generator bus , with the third order synchronous generator model , we have the same manifolds with excluding @xmath170 .
distributed controllers are expected to regulate the system states @xmath171^t \triangleq [ \theta_i(t ) , w_i(t ) , p_{m_i}(t),$ ] @xmath147^t$ ] and inputs @xmath172^t$ ] to their manifolds @xmath173 and @xmath157 asymptotically for @xmath158 . in the load bus @xmath46 , we desire to design distributed controllers which regulate the system states @xmath162^t \triangleq [ \theta_i(t ) , w_i(t ) , p_{l_i}^e(t)]^t$ ] and inputs @xmath163 to their manifolds @xmath174^t$ ] and @xmath165 asymptotically for @xmath166 .
to design a local internal model , we combine the third and fourth equations in : @xmath175 the pair @xmath176 is observable for @xmath149 .
[ asmp3 ] under assumption [ asmp3 ] , for any controllable pair @xmath177 with @xmath178 being hurwitz for @xmath43 and @xmath179 , there exists a unique non - singular matrix @xmath180 for the sylvester equation @xcite : @xmath181 with @xmath182 , becomes @xmath183 now we have a local internal model candidate : @xmath184 where @xmath185 .
internal model acts as an estimator and its states @xmath186 are expected to asymptotically track unmeasurable states @xmath187 .
the manifolds of internal model states @xmath188 are @xmath189 in this case .
suitable controllers are able to stabilize error dynamics @xmath190 . according to certainty equivalence principle @xcite , internal model states @xmath186
are used to replace @xmath187 in the manifolds so that the manifolds can be used in feedback control .
we want to estimate @xmath77 and @xmath78 separately .
it is possible because @xmath77 and @xmath78 do not share the common frequencies @xmath117 and/or exosystem states @xmath191 .
we define @xmath192 $ ] and @xmath193 $ ] where @xmath194 $ ] .
then @xmath195 , @xmath196 and @xmath197 . for notional simplicity
, we will use states with internal model states @xmath198^t$ ] and manifolds @xmath199^t$ ] .
in this section , we present a solution of the _ robust frequency regulation _ problem described in section [ sec : formulation ] . under assumptions
[ asmp1 ] , [ asmp2 ] , [ atsm1 ] , and [ asmp3 ] , we design distributed controllers such that all the states are exponentially stable with respect to their manifolds . in section [ sec : coo1 ] , we convert the frequency regulation problem into a stabilization problem .
the resulting system is iss with respect to neighboring states . in section [ sec : stab1 ] , we formally analyze the global exponential stability .
the purpose of the coordinate transformation is to transform the global frequency regulation problem to manifolds into the global stabilization problem .
this can be done by subtracting manifolds from the state and input variables . to ensure the stability of the augmented system , we make use of its special lower triangular structural .
it is possible to progressively stabilize from the outer state to the inner state .
therefore , we add more terms to the coordinate transformation inspired by backstepping @xcite to guarantee stability . since internal model states @xmath188
are expected to track @xmath200 exponentially for @xmath149 , the estimation errors @xmath201 for @xmath202 converge to zero exponentially . by certainty
equivalent principle , we use the known term @xmath203 to replace unknown @xmath204 when subtracting the manifolds from the states .
consider the coordinate transformation described by @xmath205 where @xmath206^t$ ] for @xmath158 , @xmath207^t$ ] for @xmath166 , @xmath208^t$ ] for @xmath47 and @xmath209 for arbitrary positive constants @xmath210 for @xmath211 . under coordinate transformation ,
our frequency regulation problem becomes the stabilization problem of all the transformed states .
this is because states @xmath212 converge to their manifolds if the system states @xmath213 and internal model state @xmath214 are stable .
note that additional coordinate transformation term @xmath215 does not change the origin .
moreover , we choose a controller @xmath216 where @xmath217^t$ ] for @xmath158 , @xmath218 for @xmath166 and @xmath219 for a positive constant @xmath220 . through
coordinate transformation and input , system and and internal model become @xmath221 where @xmath222 , and @xmath223 are corresponding constant matrices .
the exponential stability of @xmath213 implies that the original system states and internal model @xmath212 , and inputs @xmath224 in , , and exponentially follow their manifolds and @xmath187 .
this is because the additional terms described in are identical to zero when states @xmath213 are zero .
the following theorem summarizes the exponential stability of system with respect to its manifold .
consider algorithm [ algo ] with @xmath225 choose @xmath226 ; @xmath227 ; choose a controllable pair @xmath228 ; @xmath229 , @xmath230 ; find the solution @xmath231 of sylvester equation .
@xmath232 ; @xmath233 @xmath234 ; ( for @xmath45 ) @xmath235 @xmath236 ; @xmath237 . under assumptions
[ asmp1 ] , [ asmp2 ] , [ atsm1 ] , [ asmp3 ] and that @xmath238 is controllable for @xmath239 , we apply distributed controllers with internal model to system .
then , states with internal model @xmath212 , and inputs @xmath224 described in , , and for @xmath43 are globally exponentially stable with respect to their manifolds and @xmath187 for any initial states @xmath240 if and only if control gains @xmath241 and matrices @xmath242 and @xmath243 are chosen such that matrix @xmath244 is hurwitz where @xmath245 and @xmath246^t$ ] .
moreover , such control gains @xmath210 and matrices @xmath242 and @xmath243 always exist if @xmath247 for @xmath248 .
[ attheo1 ] * sufficiency : * assume that @xmath244 is hurwitz . then , linear time invariance system @xmath245 in is exponentially stable .
if @xmath249 is exponentially stable , then @xmath250 in is exponentially stable with respect to their manifolds and @xmath187 because coordinate transformation @xmath251 in does not change the origin .
* necessity : * this can be proven by reversing the argument steps above .
now we prove the existence of control gains and matrices by construction .
especially , one can always choose them via algorithm [ algo ] .
we first show that algorithm [ algo ] always generates a set of control gains @xmath210 and matrices @xmath242 , @xmath243 . in the generator bus , since the selections of @xmath252 and @xmath253 do not depend on @xmath254 , @xmath255 , @xmath256 , @xmath257 , or @xmath220 , we can freely choose @xmath252 and @xmath253 through step 3 and 4 in algorithm [ algo ] .
after that , we can choose @xmath254 because the selection of @xmath254 does not depend on @xmath255 , @xmath256 , @xmath257 , or @xmath220 .
likewise , @xmath255 , @xmath256 , @xmath257 , and @xmath220 can be chosen sequentially . for the load bus @xmath46
, it can be proven with the same argument of the generator bus .
this completes the proof .
we prove that matrix @xmath244 is hurwitz by showing equivalent condition that is the exponential stability of states and internal model states .
assume that we apply the distributed controllers and choose a set of control gains and matrices through algorithm [ algo ] , and @xmath258 for @xmath248 .
first consider any @xmath51 and choose quadratic lyapunov function candidates as described by @xmath259 the lie derivative of each lyapunov function candidate along the trajectories of system becomes @xmath260 note that lyapunov candidate @xmath261 is iss lyapunov function @xcite and there exists an iss lyapunov function if and only if the system is iss(theorem 1 in @xcite ) .
therefore , this is equivalent to iss with respect to neighboring states : @xmath262 } \}\ } \label{el1}\end{aligned}\ ] ] for @xmath263 and @xmath264 .
we omit the expression of the iss gains @xmath265 .
please refer lemma 2.14 in @xcite to get them .
it is not difficult to show that the chosen control gains satisfy @xmath266 ; iss contraction gain mappings .
now consider any bus @xmath47 . with @xmath267 , eigenvalues of local system
@xmath47 are the solutions of @xmath268 this implies that all the eigenvalues @xmath269 have negative real parts because @xmath270 and @xmath271 ; the local system is iss with respect to neighboring states .
its iss gain functions are , for @xmath272 , @xmath273 where @xmath274 but @xmath275 ; iss contraction gain mappings . all the iss gains @xmath276 are contraction mappings . by distributed constrained small gain theorem [ ap - the3 ] in appendix , we can ensure the stability of the given systems . moreover , by remark [ remark : last ] , ensures the exponential stability of @xmath213 .
this implies that @xmath244 is hurwitz . in theorem [ attheo1 ] , we show the sufficient and necessary condition of distributed controllers which ensure the global exponential stability of @xmath213 as well as the global exponential stability of @xmath212 with respect to their manifolds .
the stability of @xmath212 implies that inputs @xmath224 converge to their manifolds .
however , to guarantee matrix @xmath244 is hurwitz , all the network - wide system information need to be known .
moreover , all of the new controllers should be redesigned whenever there is a new integration , change or detachment of a grid component .
algorithm [ algo ] shows another extreme ; when only local system information is available to each control authorities , still it is always possible for them to design distributed controllers such that matrix @xmath244 is hurwitz , or equivalently , the system with internal model is exponentially stable .
it is not difficult to check that algorithm [ algo ] requires local information only .
if the control gains and matrices are chosen through algorithm [ algo ] , only local control gains should be modified when there is parameter changes because control gains and matrices depend on their local information only .
it should be emphasized that there is a trade off between parameter information and conservativeness .
algorithm [ algo ] requires local information only , but it might result in conservative control gains and matrices . the controller design in this section bases on assumption [ atsm1 ] ; i.e. , all the frequencies of wind power generation signals are known .
if this is the case , control authorities can directly access the solution @xmath277 of sylvester equation and thus , controllers can use the terms @xmath278 and @xmath279 to estimate unknown terms .
however , if this is not the case , the control authorities can not access the term @xmath280 and thus term @xmath281 can not be part of manifolds or coordinate transformation .
we deal with this problem in section [ sec : sol2 ] .
in this section , we consider the case that control authority @xmath114 can not directly access true value @xmath282 which depends on unknown frequencies @xmath98 . for this reason
, we adopt an adaptive technique for @xmath283 to asymptotically reconstruct the true value @xmath284 where @xmath285 is an estimator matrix of @xmath286 .
the local estimation law is in conjunction with local internal model illustrated in section [ sec : internal ] .
as we did before , we convert the problem into a global stabilization problem in section [ sec : coo2 ] . to guarantee the stability of transformed system , we add one more coordinate transformation term inspired also by backstepping in the same section .
an adaptive law is proposed in section [ sec : adaptive2 ] and the stability of the system is formally analyzed in section [ sec : stab2 ] . like section [ sec : coo1 ]
, the coordinate transformation is to convert the global regulation problem into a global stabilization problem .
additional term is also inspired by backstepping and makes the system iss with respect to neighboring states .
consider the coordinate transformation @xmath287 where @xmath288^t$ ] for @xmath158 , @xmath289^t$ ] for @xmath166 , @xmath208^t$ ] for @xmath47 and @xmath290 under coordinate transformation , the frequency regulation problem is transformed to a stabilization problem for the same reason as .
the difference is that we use estimator matrix @xmath285 instead of true matrix @xmath286 .
the origin of the system does not change under the additional term @xmath215 in the coordinate transformation .
consider input @xmath291 where @xmath292^t$ ] for @xmath45 , @xmath293 for @xmath46 , and @xmath294 through coordinate transformation and input , system and , and internal model become @xmath295 where @xmath296 , and @xmath223 are corresponding constant matrices and @xmath297 . analyzing the stability of dynamics
has potential difficulties because unknown constant matrix @xmath286 and @xmath298 appear in @xmath299 , @xmath300 , @xmath301 , and @xmath302 . in this section ,
we introduce parameter estimator dynamics of @xmath285 . for notational simplicity ,
@xmath303 $ ] , @xmath304 $ ] , and @xmath305 @xmath306 $ ] denote the @xmath307 row vector of @xmath286 , @xmath285 , and @xmath298 respectively .
moreover , we define @xmath308 as the @xmath307 element of @xmath309 $ ] .
we design parameter estimator @xmath285 as follows : @xmath310 where @xmath311 for @xmath45 , @xmath312 for @xmath46 , @xmath313 , and @xmath314 is an arbitrary constant .
the adaptive law @xmath315 can also be implemented instead of . however , the extra term @xmath316 in speeds up the convergence rate by restricting the parameter estimator s size within @xmath317 . to see this ,
assume we choose initial conditions @xmath318 for @xmath319 .
moreover , suppose that @xmath320 is @xmath321 for some @xmath322 , then its derivative is @xmath323 where @xmath324^t$ ] . because @xmath325 , state @xmath326 remains @xmath327 for @xmath328 .
this implies @xmath329 as well as @xmath317 for @xmath330 .
the following theorem summarizes the asymptotic stability of transformed system .
choose @xmath226 ; @xmath331 ; choose a controllable pair @xmath228 ; @xmath332 ; @xmath230 ; @xmath333 ; @xmath334 ; ( for @xmath45 ) @xmath335 ; [ atalgo1 ] under assumptions [ asmp1 ] , [ asmp2 ] , [ atsm3 ] , [ asmp3 ] , and that @xmath238 is controllable for @xmath239 , we apply distributed controllers with adaptive internal model and .
then , states with internal model @xmath212 , and inputs @xmath224 described in , , and for @xmath43 are asymptotically convergent to their manifolds and @xmath187 for any initial states @xmath240 , and @xmath336 if @xmath337 is hurwitz where @xmath338 and @xmath339^t$ ] , @xmath246^t$ ] , @xmath340 @xmath341 @xmath342^t$ ] .
moreover , such control gains @xmath210 and matrices @xmath242 and @xmath243 always exist if @xmath247 for @xmath248 .
[ attheo2 ] we first prove that the networked system is stable if @xmath337 is hurwitz .
assume @xmath337 is hurwitz .
we choose lyapunov candidate @xmath343
\left [ \begin{array}{c } \hat{\textbf{x}}(t)\\ \hat{\textbf{t}}(t)\\ \end{array } \right ] .
\label{ate34}\end{aligned}\ ] ] the lie derivative of @xmath344 is @xmath345 we formulate the proofs into several claim statements . * claim a * : @xmath346 .
+ * proof .
* by taking post - multiply @xmath347 and taking frobenius norm on both sides of sylvester equation , @xmath348 by definition , @xmath349 becomes @xmath350 apply and the norm relation @xmath351 ( lemma 10 in @xcite ) to .
then we have @xmath352 note that @xmath353 when its argument is a vector .
claim a implies @xmath354 for @xmath355 .
* claim b * : @xmath356 . + * proof . *
consider adaptive law .
if @xmath357 , then @xmath358 .
let @xmath359 denote a set of indices @xmath360 such that @xmath361 .
likewise , @xmath362 . then , @xmath363 where @xmath364 is applied .
apply the properties of claim b to to get @xmath365 where @xmath337 is symmetric and @xmath366 by assumption . multiply @xmath18 and take integral from @xmath367 to @xmath368 on the both sides of to have @xmath369 where @xmath370 because @xmath371 is non - increasing as shown in . therefore by barbalat s lemma ( lemma 8.2 in @xcite ) , and the uniform continuity of @xmath372 , value @xmath372 asymptotically converges to zero .
this implies the asymptotic convergence of @xmath213 to zero , as well as the asymptotic convergence of @xmath212 , and @xmath224 to their manifolds .
the uniform continuity of @xmath372 is proven in claim c. * claim c * : @xmath372 is a uniformly continuous function . + * proof . * the value @xmath373 is uniformly bounded by some constant @xmath374 because @xmath375 .
consider @xmath376 the term @xmath377 is given by @xmath378 . by uniform boundedness of all @xmath379 , for @xmath380 and
any @xmath381 , @xmath382 where @xmath383 , @xmath384 , @xmath385 , and @xmath386 are positive constants and @xmath387^t$ ] is a vector with a positive constant @xmath388 .
therefore , @xmath389 where this is a strictly increasing function of @xmath390 and @xmath391 as @xmath392 . by applying these relations to
, we can prove the uniform continuity of @xmath393 ; i.e. , for any given @xmath394 , there always exists @xmath226 such that for all @xmath395 , @xmath396 .
claim c completes the proof for the stability part .
now we proceed to prove that there always exists a set of control gains and matrices such that @xmath337 is negative definite by construction .
especially , algorithm [ atalgo1 ] generates such set . to prove this ,
assume we choose a set of control gains and matrices through algorithm [ atalgo1 ] .
it is not difficult to show that the chosen control gains and matrices satisfy @xmath397 where @xmath398 with @xmath399 , @xmath400 , @xmath401 , @xmath402 , @xmath403 , and @xmath404 .
now , again consider lyapunov candidate and its lie derivative for @xmath51 . in theorem [ attheo1
] , it has been proven that system for @xmath47 is iss contraction mappings with respect to neighboring states .
thus , there exists iss lyapunov function for @xmath47 ( theorem 1 in @xcite ) .
we apply some properties to equation .
the first one is @xmath405^t\| = \alpha_i \sqrt{2\ell}$ ] .
the second one is @xmath406 coming from claim a. the last one is @xmath407 for any positive constant @xmath31 . by applying these properties to , it becomes , for any positive constants @xmath408 , @xmath409 , @xmath410 , and @xmath411 , @xmath412 where @xmath413 and @xmath414 is negative definite . by plugging into , we have @xmath415 where matrix @xmath416 is negative definite . therefore , matrix @xmath337 is also hurwitz .
theorem [ attheo2 ] states a sufficient condition of system stability .
the lie derivative of lyapunov candidate can only be negative semidefinite because the parameter adaptive error terms are sign indefinite .
the existence of stabilizing control gains and matrices is guaranteed by algorithm [ atalgo1 ] .
again , the algorithm shows another extreme ; it is always possible to choose a set of stabilizing gains and matrices even when only local information are available , but is might be conservative .
it is obvious that algorithm [ atalgo1 ] can be followed only with local system information .
applying the approach used in this section to the fourth order system has potential difficulties .
manifold @xmath417 depends on @xmath285 because we want @xmath52 to cancels @xmath418 .
this term affects manifold @xmath419 which should be a function of @xmath420 .
likewise , @xmath421 will be a function of @xmath422 which depends on unmeasurable terms @xmath286 and @xmath187 .
in this section , we present simulations to show the performance of the proposed controllers .
most of parameters are adopted from @xcite .
we , however , choose relatively high frequencies of wind generations to inspect evolutions of @xmath52 and @xmath126 in a short time . by van der
hoven s spectrum model @xcite , proper frequencies are around @xmath423 and @xmath424 but higher frequencies @xmath425 , and @xmath426 are chosen instead in this simulation .
the controllers are applied to the single line diagram of the ieee 68-bus test system topology shown in @xcite .
the topology includes 16 generators ( @xmath427 ) , 24 load buses ( @xmath428 ) and 28 load bus with zero load ( @xmath429 ) .
we assume that each generator / load bus @xmath51 has a local wind turbine . for @xmath158 , each wind turbine generates @xmath430 , and , for @xmath166 , each wind turbine generates @xmath431 . on the test system topology , we simulate both the _ robust ( adaptive ) frequency regulation _ problems proposed in section [ sec : sol1 ] and [ sec : sol2 ] .
moreover , we conduct one more simulation to compare the effect of connection and disconnection of wind power generations with other controllers .
the parameters are adopted from page 598 in @xcite : @xmath432 , and @xmath433 for @xmath149 .
load damping constants are @xmath434 for @xmath51 and @xmath435 for @xmath47 , and the manifolds as @xmath436 . demand response parameters @xmath437 and @xmath438 and @xmath439 are borrowed from @xcite .
we set the manifolds as @xmath436 and inelastic demands are @xmath440 for @xmath166 .
wind power generators with @xmath441 and @xmath442 $ ] are attached to generator buses @xmath45 .
load buses @xmath46 have wind power generators with @xmath443 , @xmath444 , @xmath445 $ ] and @xmath446 $ ] .
note that one can choose maximum frequency of wind power generation signals by choosing a proper time constant @xmath124 in the bess . for the _ robust frequency regulation _ problem
, we choose the following set of control gains and matrices for @xmath45 and @xmath46 , respectively : @xmath447^t$ ] , @xmath448^t$ ] and matrices @xmath449 , m_{i}= \left [ \begin{array}{cccc } -2&0&0&0\\ 11&-2.5&0&0\\ 4&-1&-3&0\\ -6&5&1&-3.5\\ \end{array } \right ] \label{matrix_m}\end{aligned}\ ] ] with @xmath450^t$ ] , @xmath451^t$ ] .
such sets of the gains and matrices satisfy that matrix @xmath244 is hurwitz . for the _ robust adaptive frequency regulation _ problem , we choose the following set of control gains and matrices for @xmath45 and @xmath46 , respectively : @xmath452^t$ ] , @xmath453^t$ ] and matrices with @xmath450^t$ ] , @xmath454^t$ ] .
such sets of the gains and matrices satisfy that matrix @xmath337 is hurwitz .
to regulate angular frequency to @xmath139 , we aim that each demand response @xmath126 in @xmath46 tracks low and medium frequency parts of local wind power generations , and mechanical generation @xmath52 in @xmath45 tracks low frequency parts of local wind power generations .
the results are shown in figures [ 68bus_p1_new ] . in each subfigures
, the horizontal axis represents time and the vertical axis represents corresponding values .
the first subfigure shows that angular frequency of bus @xmath455 converges to @xmath139 exponentially .
all the states also converge to their manifolds as shown in the second subfigure ; i.e. , the designed distributed controllers achieve the objective and eventually regulate network - wide frequencies to @xmath456 .
mechanical power and elastic demand are used to balance the power generation and usage by tracking wind power generations .
especially , the trajectory of elastic demand @xmath457 of bus @xmath455 tracks desired signals , low and medium frequency parts of wind power generations and stabilizing signals ; i.e. , @xmath458 as shown in the third subfigure .
the last subfigure presents the trajectory of mechanical generation @xmath459 of bus @xmath460 tracks low frequency parts of wind power generations and stabilizing signals ; i.e. , @xmath461 .
we intentionally choose very low frequency @xmath462 of inelastic demand signal @xmath127 to see whether @xmath126 and @xmath52 follows sinusoidal wind power generations @xmath463 and @xmath55 , respectively . as shown in the figure [ 68bus_p2_new ] , angular frequency
is regulated to @xmath139 .
the second subfigure shows that network - wide state errors with respect to their manifolds converge to zero ; i.e. , the controllers achieve frequency regulation .
the angular frequency regulation is achieved by the fact that each @xmath457 and @xmath459 follow the desired manifolds , balancing power where @xmath455 and @xmath460 ; low and medium frequency parts and low frequency parts .
we conduct one more simulation of _ robust frequency regulation _ problem . however at this time , wind power generations are connected at time @xmath464 and disconnected at time @xmath465 . furthermore , the frequency fluctuation of the proposed controllers is compared with regular distributed controllers without internal model . for a fair comparison , synchronous generators in generator buses and demand response in load buses are both used to stabilize the frequencies in the both simulations .
regular distributed controllers suppress the frequency fluctuation and its amplitude will decrease as its control gains increase , but can not completely reject the fluctuation . unlike the regular controllers ,
when wind power generators are connected to the system , the proposed distributed controllers with internal model are able to completely reject the disturbances induced by wind power generations . with the disconnection of wind power generator , the frequencies also converge to desired point @xmath139 exponentially because the system without wind power generators is a special case of the system which we are dealing with in the paper .
we have proposed distributed controllers for _ robust ( adaptive ) frequency regulation _ problems under fast changing and unknown wind generations .
we adopt local internal model ( and adaptive technique ) to cope with uncertainties . to overcome fast changing natures of uncertain signals , two of new grid components are integrated . at each bus with wind power generations , a bess filters out high frequency parts of unknown signals . at generator bus ,
demand response tracks its medium frequency parts and synchronous generator tracks low frequency parts and stabilizes the system . at load bus ,
demand response tracks both low and medium frequency components and stabilize the local system .
distributed constrained small gain theorem is introduced in this appendix .
the theorem is an extension of constrained small - gain theorem in @xcite to a network set - up .
the system is input - to - state stable with respect to neighboring states .
equivalently , there exist a class @xmath468 function @xmath469 and class @xmath470 functions @xmath471 and @xmath472 such that for @xmath473 and @xmath149 , [ ap - asm1 ] @xmath474 } ) , { \nonumber}\\ & \max_{j \in { \mathcal{n}}_i}\{\gamma_{ij}(\|x_j\|_{[t_0,t]})\}\}. \label{ap - e12}\end{aligned}\ ] ] * ( distributed constrained small gain theorem ) * under assumptions [ ap - asm1 ] and [ ap - asm2 ] , the system is iss with respect to @xmath478 .
equivalently , there exists class @xmath468 function @xmath31 and class @xmath470 function @xmath471 such that for all @xmath479 and @xmath480 , the solution of exists and for @xmath473 , @xmath481 } ) . \label{theoremeq1}\end{aligned}\ ] ] moreover , the function @xmath482 @xmath483 is a class @xmath30 function candidate of @xmath484 in where @xmath485 is a constant .
[ ap - the3 ] * claim 1 : * we will prove by induction that the following hold for @xmath490 : @xmath491}\leq \max\{\beta_i(\|x_i(t_0)\|,0),{\nonumber}\\ & \max_{\substack{(i , i_1,\cdots , i_\kappa)\in\mathcal{p}_{i i_\kappa}\\ i_1,\cdots , i_\kappa\in s_{\ell}}}\gamma_{ii_1}\circ\cdots \circ\gamma_{i_{\kappa-1}i_{\kappa}}\circ\gamma_{i_{\kappa}d } ( \|d_{i_{\kappa}}\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\in s_{\ell}\setminus\{i\}}\max_{\substack{(j , i_\kappa,\cdots , i)\in\mathcal{p}_{ji}\\ i_1,\cdots , i_{\kappa}\in s_{\ell}}}{\nonumber}\\ & \gamma_{ii_1}\circ\gamma_{i_1i_2}\circ\cdots\circ\gamma_{i_\kappa j}\circ \beta_j(\|x_j(t_0)\|,0),{\nonumber}\\ & \max_{j\notin s_{\ell}}\max_{\substack{(i , i_1,\cdots , i_\kappa , j)\in\mathcal{p}_{ij}\\ i_1,\cdots , i_\kappa\in s_{\ell}}}\gamma_{ii_1}\circ\cdots\circ\gamma_{i_{\kappa}j}(\|x_j\|_{[t_0,t]})\}. \label{ap - e14}\end{aligned}\ ] ] + * proof . * by , one can see that @xmath492}\leq & \max\{\beta_1(\|x_1(t_0)\|,0),\gamma_{1d}(\|d_1\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\neq 1}\{\gamma_{1j}(\|x_j\|_{[t_0,t]})\}\},\label{ap - e15}\end{aligned}\ ] ] and @xmath493}\leq & \max\{\beta_2(\|x_2(t_0)\|,0),\gamma_{2d}(\|d_2\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\neq 2}\{\gamma_{2j}(\|x_j\|_{[t_0,t]})\}\}.\label{ap - e16}\end{aligned}\ ] ] substitute into , and it renders the following : @xmath492}\leq & \max\{\beta_1(\|x_1(t_0)\|,0),\gamma_{1d}(\|d_1\|_{[t_0,t]}),{\nonumber}\\ & \gamma_{12}\circ\beta_2(\|x_2(t_0)\|,0),\gamma_{12 } \circ\gamma_{2d}(\|d_2\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\neq 2}\{\gamma_{12}\circ\gamma_{2j}(\|x_j\|_{[t_0,t]})\},{\nonumber}\\ & \max_{j\notin\{1,2\}}\{\gamma_{1j}(\|x_j\|_{[t_0,t]})\}\}.\label{ap - e22}\end{aligned}\ ] ] since @xmath494 is a contraction mapping , it follows from that @xmath492}\leq & \max\{\beta_1(\|x_1(t_0)\|,0),\gamma_{1d}(\|d_1\|_{[t_0,t]}),{\nonumber}\\ & \gamma_{12}\circ\beta_2(\|x_2(t_0)\|,0),\gamma_{12 } \circ\gamma_{2d}(\|d_2\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\notin\{1,2\}}\{\max\{\gamma_{1j } , \gamma_{12}\circ\gamma_{2j}\}(\|x_j\|_{[t_0,t]})\}\}.\label{ap - e17}\end{aligned}\ ] ] by symmetry , one can show a similar property to for @xmath495}$ ] .
so holds for the case of @xmath496 .
now assume that holds for some @xmath497 .
similar to , we have @xmath498}\leq \max\{\beta_{\ell+1}(\|x_{\ell+1}(t_0)\|,0),{\nonumber}\\ & \gamma_{(\ell+1)d}(\|d_{\ell+1}\|_{[t_0,t]}),\max_{j\neq ( \ell+1)}\{\gamma_{(\ell+1)j}(\|x_j\|_{[t_0,t]})\}\}.\label{ap - e18}\end{aligned}\ ] ] following analogous steps above , one can show that holds for @xmath499 . by induction
, we complete the proof .
* claim 2 : * we show by contradiction the existence and boundedness of the solution to . + * proof . *
a direct result of claim 1 is that the following holds for all @xmath500 : @xmath491}\leq \max\{\beta_i(\|x_i(t_0)\|,0),{\nonumber}\\ & \max_{\substack{(i , i_1,\cdots , i_\kappa)\in\mathcal{p}_{i i_\kappa}\\ i_1,\cdots , i_\kappa\in { \mathcal{v}}}}\gamma_{ii_1}\circ\cdots \circ\gamma_{i_{\kappa-1}i_{\kappa}}\circ\gamma_{i_{\kappa}d } ( \|d_{i_{\kappa}}\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\neq i}\max_{\substack{(j , i_\kappa,\cdots , i)\in\mathcal{p}_{ji}\\ i_1,\cdots ,
i_{\kappa}\in { \mathcal{v}}}}{\nonumber}\\ & \gamma_{ii_1}\circ\gamma_{i_1i_2}\circ\cdots\circ\gamma_{i_\kappa j}\circ \beta_j(\|x_j(t_0)\|,0)\}.\label{ap - e19}\end{aligned}\ ] ] since all the gain functions @xmath472 are contraction mappings , renders the following : @xmath491}\leq \max\{\beta_i(\|x_i(t_0)\|,0),{\nonumber}\\ & \max_{j\in{\mathcal{n}}_i}\gamma_{ij}\circ\gamma_{jd}(\|d_j\|_{[t_0,t]}),\max_{j\in { \mathcal{n}}_i}\beta_j(\|x_j(t_0)\|,0)\}.\label{ap - e20}\end{aligned}\ ] ] because of the choice of @xmath240 and the bound on @xmath478 , the relation holds for any @xmath124 .
it implies that @xmath501 for all @xmath502 and thus is uniformly bounded .
it completes the proof .
* claim 3 : * we show the iss property of .
we will prove by induction that the following holds for all @xmath503 : @xmath504}_i(\|x\|_{\infty},t - t_0 ) , \gamma_i^{[\ell-1]}(\|d\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\notin s_{\ell}}\max_{\substack{(i , i_1,\cdots , i_\kappa , j)\in\mathcal{p}_{ij}\\ i_1,\cdots , i_\kappa\in s_{\ell}}}\gamma_{ii_1}\circ\cdots\circ\gamma_{i_{\kappa}j}(\|x_j\|_{[t_0,t]})\ } , \label{ap - e8}\end{aligned}\ ] ] for some class @xmath468 function @xmath505}_i$ ] where @xmath506\}$ ] .
+ * proof . *
let @xmath496 .
note that for any constant @xmath485 , @xmath507}),\max_{j\neq1}\gamma_{1j}(\|x_j\|_{[t_0+\frac{2{\mathcal{l}}-1}{2{\mathcal{l}}}t , t_0+t]})\}{\nonumber}\\ & \leq \max\{\beta_1(\|x\|_{\infty},\frac{1}{2{\mathcal{l}}}t),\gamma_{1d}(\|d_1\|_{[t_0+\frac{2{\mathcal{l}}-1}{2{\mathcal{l}}}t , t_0+t]}),{\nonumber}\\ & \max_{j\neq1}\{\gamma_{1j}(\|x_j\|_{[t_0+\frac{2{\mathcal{l}}-1}{2{\mathcal{l}}}t , t_0+t]})\}\}. \label{ap - e2}\end{aligned}\ ] ] for any @xmath508 $ ] , it holds that @xmath509}),{\nonumber}\\ & \max_{j\neq2}\{\gamma_{2j}(\|x_j\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+\tau]})\}\}{\nonumber}\\ & \leq \max\{\beta_2(\|x\|_{\infty},\frac{1}{{\mathcal{l}}}t ) , \gamma_{2d}(\|d_2\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]}),{\nonumber}\\ & \max_{j\neq2}\{\gamma_{2j}(\|x_j\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]})\}\}. \label{ap - e3}\end{aligned}\ ] ] so implies that @xmath510}\leq\max\{\beta_2(\|x\|_{\infty},\frac{1}{{\mathcal{l}}}t),{\nonumber}\\ & \gamma_{2d}(\|d_2\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]}),{\nonumber}\\ & \max_{j\neq2}\{\gamma_{2j}(\|x_j\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]})\}\}. \label{ap - e4}\end{aligned}\ ] ] substitute into , and we have @xmath511}),{\nonumber}\\ & \gamma_{12}\circ\beta_2(\|x\|_{\infty},\frac{1}{{\mathcal{l}}}t),\gamma_{12}\circ\gamma_{2d}(\|d_2\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]}),{\nonumber}\\ & \gamma_{12}\circ\gamma_{21}(\|x_1\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]}),{\nonumber}\\ & \max_{j\notin s_2}\max\{\gamma_{1j},\gamma_{12}\circ\gamma_{2j}\ } ( \|x_j\|_{[t_0+\frac{2{\mathcal{l}}-2}{2{\mathcal{l}}}t , t_0+t]})\}.\label{ap - e5}\end{aligned}\ ] ] since @xmath512 is a contraction mapping , there is class @xmath468 function @xmath513 such that @xmath514}),\gamma_{12}\circ\gamma_{2d}(\|d_2\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\notin s_2}\max\{\gamma_{1j},\gamma_{12}\circ\gamma_{2j}\}(\|x_j\|_{[t_0,t]})\}. \label{ap - e6}\end{aligned}\ ] ] by symmetry , there is class @xmath468 function @xmath515 such that @xmath516 } ) , \gamma_{21}\circ\gamma_{1d}(\|d_1\|_{[t_0,t]}),{\nonumber}\\ & \max_{j\notin s_2}\max\{\gamma_{2j},\gamma_{21}\circ\gamma_{1j}\ } ( \|x_j\|_{[t_0,t]})\}.\label{ap - e7}\end{aligned}\ ] ] hence , we have shown that holds for @xmath496 . now assume holds for some @xmath517 . recall that @xmath518 } ) , \max_{j\neq \ell+1}\gamma_{ij}(\|x_j\|_{[t_0,t]})\}.\label{ap - e9}\end{aligned}\ ] ] by using similar arguments towards the case of @xmath496 , one can show holds for @xmath499 . now we proceed to find a relation between @xmath519 and @xmath520 . because @xmath521 , note that @xmath522}),\\ & \max_{j\neq i}\{\gamma_{ij}(\|x_j\|_{\infty})\}\}.\end{aligned}\ ] ] similar to , one can show by induction that there are class @xmath470 functions @xmath98 and @xmath523 such that @xmath524})\}. \label{ap - e10}\end{aligned}\ ] ] the combination of and achieves the desired result .
now proceed with the proof that function @xmath525 is a candidate of class @xmath30 function @xmath31 in .
we first find candidates of functions @xmath526}$ ] in and @xmath98 in and then combine them together .
note that by substituting into , we have equation .
consider class @xmath30 functions in equation : @xmath527 this implies that , in , @xmath528 is a @xmath30 function candidate .
likewise , in , @xmath529}(x , t)=\sum_{k=1}^{\ell}\beta_k(x,\frac{t}{(2{\mathcal{l}})^{\ell-1 } } ) \label{cooc1}\end{aligned}\ ] ] is a @xmath30 function candidate for @xmath530 because we conduct @xmath531 times of the substitutions . in a similar way
, one can show that , in , @xmath532 is a class @xmath32 function candidate for @xmath530 .
now we proceed to find a relation between @xmath526}$ ] and @xmath98 when @xmath533 . with equation , @xmath534})\}. \label{coo2}\end{aligned}\ ] ] by combining and , @xmath535}_i ( |{\mathcal{v}}|\max_{k \in { \mathcal{v}}}\rho_k(\|x(t_0)\|),t - t_0),{\nonumber}\\ & \ \ \ \ \ \ \ \ \ \gamma_i^{[\ell-1]}(\|d\|_{[t_0,t]})\}.\end{aligned}\ ] ] this implies that @xmath536}_i(|{\mathcal{v}}|\max_{k \in { \mathcal{v}}}\rho_k(x),t ) \label{cooc3}\end{aligned}\ ] ] is one of the class @xmath30 function candidates . by applying and to , we have the result .
if functions @xmath537 in for @xmath149 are @xmath538 , then @xmath484 in is also in the same form : @xmath539 where @xmath540 are constants , @xmath541 are increasing functions without bound and @xmath542 are class @xmath32 functions . [
remark : last ] l. guo , y. zhang , and c. s. wang . a new battery energy storage system control method based on soc and variable filter time constant . in _ innovative smart grid technologies ( isgt ) , 2012 ieee pes _ , pages 17 .
ieee , 2012 .
keep , f.e .
sifuentes , d.m .
auslander , and d.s .
using load switches to control aggregated electricity demand for load following and regulation . in _ power and energy society general meeting , 2011 ieee _ , pages 17 .
ieee , 2011 .
lim , y. wang , and r. zhou . robust decentralised load - frequency control of multi - area power systems . in _ generation ,
transmission and distribution , iee proceedings- _ , volume 143 , pages 377386 .
iet , 1996 .
n. lu and d.j .
design considerations for frequency responsive grid friendly tm appliances . in _
transmission and distribution conference and exhibition , 2005/2006 ieee pes _ , pages 647652 .
ieee , 2006 .
z. qin , m. liserre , f. blaabjerg , and h. wang .
energy storage system by means of improved thermal performance of a 3 mw grid side wind power converter . in _ industrial electronics society ,
iecon 2013 - 39th annual conference of the ieee _ , pages 736742 .
ieee , 2013 .
k. yoshimoto , t. nanahara , and g. koshimizu .
new control method for regulating state - of - charge of a battery in hybrid wind power / battery energy storage system . in _
power systems conference and exposition , 2006 .
2006 ieee pes _ , pages 12441251 .
ieee , 2006 . | this paper investigates distributed angular frequency regulation of multi - machine power systems with wind generations when wind power signals frequencies are known ( and unknown ) .
the challenge is to regulate angular frequencies under fast changing and unpredictable wind generations . to address the issue
, we split unknown signals into three parts according to their frequency ranges and design an ( adaptive ) internal model to reconstruct the low and medium frequency parts .
we integrate heterogeneous grid components to deal with fast changing signals . at each bus with wind power generations , a battery energy storage system is used to filter out high frequency components of wind generations . at generator bus ,
demand response is utilized to deal with their medium frequency components and a synchronous generator is exploited to handle their low frequency components . at load bus ,
demand response handles low and medium frequency components .
the proposed controllers ensure exponential stability ( and asymptotic convergence ) of system states with respect to their desired signals .
simulations on the ieee 68-bus test system are conducted to demonstrate the effectiveness of the proposed controllers . and |
the discrete nature of vorticity in quantum fluids ( superfluid @xmath0he , @xmath1he and atomic bose
einstein condensates ) makes it possible to give physical meaning to vortex reconnections .
the importance of vortex reconnections in turbulence can not be overestimated .
reconnections randomise the velocity field thus sustaining the turbulent vortex tangle in a steady state @xcite and changing the flow s topology @xcite . by triggering a kelvin wave cascade @xcite ,
reconnections are responsible for the decay of turbulent kinetic energy at very low temperatures @xcite .
the dependence of the vortex reconnection frequency on the observed vortex line density is therefore an important quantity @xcite . in quantum fluids ,
the existence of vortex reconnections was first conjectured by schwarz @xcite ; it was then theoretically demonstrated by koplik and levine @xcite , and experimentally observed at the university of maryland @xcite using micron - size solid hydrogen tracers trapped in superfluid vortex lines .
more precisely , the maryland investigators inferred the existence of reconnections by the sudden movement of trapped particles . in doing
so they relied on numerical calculations by de waele and aarts @xcite , who reported that the distance between two vortices , @xmath2 , varies as @xmath3 where @xmath4 is the quantum of circulation , @xmath5 is time and @xmath6 is the time of reconnection .
this @xmath7 scaling agrees with the approximate analytic solution of the nonlinear schroedinger equation ( also called the gross pitaevskii equation , or gpe for short ) found by nazarenko and west @xcite .
de waele and aarts also claimed that , on their way toward a reconnection , the vortices form a universal pyramidal cusp structure which is independent of the initial condition . the numerical calculations of de waele and aarts @xcite were based on the vortex filament model pioneered by schwarz @xcite , which assumes that the vortex core is much smaller than any other length scale in the problem .
the vortex filament model thus breaks down in the vicinity of a reconnection .
it is therefore instructive to repeat the calculation seeking full numerical solutions of the gpe , which does not suffer such limitation and is a well - known model of superfluidity .
this is indeed what we plan to do in this paper , together with revisiting the claim of universality of the reconnecting pyramidal cusp geometry .
the single - particle complex wavefunction @xmath8 for @xmath9 bosons of mass @xmath10 obeys the 3-dimensional time dependent gpe @xmath11 where @xmath12 is the position , @xmath13 is the strength of the delta function interaction between the bosons , and @xmath14 is the single particle energy .
the wavefunction is normalised by the condition that @xmath15 . if we let @xmath16 be the unit of time , @xmath17 the unit of length , and @xmath18 the unit of @xmath19 , we obtain the following dimensionless gpe @xmath20 we solve eq .
[ eq : gpe ] in an @xmath21 periodic box , using fourth order centred differences and runge kutta fehlberg adaptive time stepping .
+ he typical numerical resolution is @xmath22 for example for the first calculation described in sec .
[ sec:3 ] .
more details of the numerical technique and the method used to set up initial conditions will be described elsewhere @xcite .
in a first set of numerical experiments , we consider the same planar vortex rings configuration of de waele and aarts @xcite .
two vortex rings of ( dimensionless ) radius @xmath23 are located on the @xmath24 plane , initially separated by the ( dimensionless ) distance @xmath25 , and moving in the positive @xmath26 direction .
[ fig:1 ] shows that , during the evolution , the rings turn into each other .
the pyramidal cusp which forms ( and which eventually will lead to a vortex reconnection ) trails behind the rest of the moving vortex configuration .
( colour online ) .
ring reconnection .
isosurfaces of @xmath27 ( dimensionless units ) showing two rings moving along the positive @xmath26 direction at ( dimensionless )
times @xmath28 ( left ) and @xmath29 right .
note the formation of the reconnecting cusp.,title="fig : " ] ( colour online ) .
ring reconnection .
isosurfaces of @xmath27 ( dimensionless units ) showing two rings moving along the positive @xmath26 direction at ( dimensionless ) times @xmath28 ( left ) and @xmath29 right .
note the formation of the reconnecting cusp.,title="fig : " ] by symmetry , the reconnection is on the @xmath30 plane , which facilitates the computation of the minimum distance between the vortex rings , @xmath31 , as a function of time @xmath5 , until , at @xmath32 , the rings reconnect .
we repeat the calculation for different initial separations @xmath33 , @xmath34 , @xmath35 and @xmath36 . fig .
[ fig:2 ] shows a log - log plot of @xmath31 vs @xmath5 and confirms that indeed @xmath37 , in agreement with eq .
[ eq : aarts ] .
we fit our @xmath31 vs @xmath5 data to the form @xmath38 , \label{eq : fit}\ ] ] where @xmath39 and @xmath40 are parameters .
we find that @xmath41 , and that @xmath39 increases and @xmath40 decreases for increasing @xmath42 .
more precisely we obtain @xmath43 , @xmath44 , @xmath45 , @xmath46 and @xmath47 , and @xmath48 , @xmath49 , @xmath50 , @xmath51 and @xmath52 , respectively for @xmath25 , @xmath53 , @xmath34 , @xmath35 and @xmath36 .
the errors are approximately @xmath54 for @xmath39 and @xmath55 for @xmath40 , and arise mainly from the inaccuracy in localising the axes of the vortex lines .
( colour online ) .
pyramidal reconnection cusp and definition of the angle @xmath56 ( which a vortex makes with itself ) and the angle @xmath57 ( which a vortex makes with the other vortex ) . ,
scaledwidth=50.0% ] the maryland group reported their results in the ( dimensional ) form @xcite @xmath58 , \label{eq : um}\ ] ] ( where we introduced the subscript `` d '' to stress that a quantity is dimensional ) , where @xmath59 and @xmath60 are their fitting parameters .
the numerical results of de waele and aarts @xcite correspond to @xmath61 . since dimensional and dimensionless length , time and circulation
are related by @xmath62 , @xmath63 and @xmath64 ( where @xmath65 and @xmath66 are respectively the dimensional and dimensionless circulations ) , our ( dimensionless ) results agree with maryland s ( dimensional ) results if @xmath67 and @xmath68 , which is indeed the case ( the value of @xmath14 needed to get @xmath69 to compare @xmath40 with @xmath70 can be estimated from the definition of the healing length @xmath71 and the fact that @xmath72 ) . following de waele and aarts @xcite , we define the intravortex angle @xmath56 and the intervortex angle @xmath57 of the pyramidal cusp formed by the two vortices as they approach the reconnection .
these angles are computed from the dot products of the vector @xmath73 with the vectors @xmath74 and @xmath75 , where @xmath76 , @xmath77 , and @xmath78 as shown in fig .
[ fig:3 ] . if the total angle @xmath79 is equal to @xmath80 , then the reconnection is `` flat . ''
if @xmath81 the vortices form a pyramid .
[ fig:4 ] shows that reconnecting angles are approximately constant , @xmath82 and @xmath83 , independently of the initial distance between the vortex rings , again in agreement with the finding of de waele and aarts @xcite , although their numerical values are slightly different : their @xmath56 increases from @xmath84 to @xmath85 over a change of two orders of magnitude of the initial distance . whereas @xmath86 .
initial vortex ring
line configuration to define the angle @xmath87 .
, scaledwidth=30.0% ]
in a second set of numerical experiments , we consider the interaction of a vortex line ( set in the middle of the computational box and aligned in the @xmath88 direction ) with a vortex ring ( located in the @xmath89 plane , with centre at @xmath90 ) which moves in the positive @xmath26 direction ) .
the geometry of the initial configuration can be parametrised by the angle @xmath87 between the vortex ring and the vortex line , see fig .
[ fig:5 ] ; @xmath87 ranges from @xmath91 ( the vortex ring passes to the left of the line ) to @xmath92 ( the vortex ring passes to the right of the line ) .
[ fig:6 ] shows that , for ring
line reconnections , the distance between the vortices is very similar to what is determined for ring
ring reconnections .
fitting as in eq .
[ eq : fit ] we have @xmath93 , @xmath94 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 and @xmath99 , and @xmath100 , @xmath101 , @xmath102 , @xmath101 , @xmath103 , @xmath104 , @xmath105 and @xmath101 respectively for initial angles @xmath106 , @xmath107 , @xmath108 , @xmath109 , @xmath110 , @xmath85 , @xmath111 and @xmath112 .
in contrast to what happens in ring
ring reconnections , there is less evidence for a universal pyramidal cusp .
fig [ fig:7 ] shows that @xmath56 and @xmath57 depend on @xmath87 , that is to say , on the initial geometry .
fig [ fig:8 ] shows that the total angle @xmath113 ( which would be @xmath80 for a `` flat '' reconnection ) , is not constant .
it is interesting to note that , if @xmath114 , when the vortex ring is sufficiently close to the vortex line , the part of the vortex ring which is near the vortex and the part of the vortex line which is near the ring are almost parallel ; since they have the same circulation , they tend to rotate about each other .
this motion perturbs and deflects the vortex ring , which passes past the ( perturbed ) vortex line without any reconnection . because of this curious effect , shown in fig .
[ fig:9 ] , we have no data for @xmath114 in figs .
[ fig:7 ] and [ fig:8 ] .
[ plottop ] +
by numerically finding nonlinear solutions of the governing gpe , we have confirmed that , as vortex filaments approach a reconnection , their distance scales as @xmath115 , as predicted by nazarenko and west @xcite . by fitting our @xmath2 vs @xmath5 data to eq .
[ eq : fit ] , we find better quantitative agreement with maryland s observations @xcite than obtained with the vortex filament method @xcite .
finally , we find that reconnections involve the formation of a pyramidal cusp .
however , on the length scales which we can explore using the gpe ( which are different from length scales described by the vortex filament method ) , the angles of this cusp are not universal as previously claimed @xcite . | we present numerical solutions of the gross pitaevskii equation corresponding to reconnecting vortex lines .
we determine the separation of vortices as a function of time during the approach to reconnection , and study the formation of pyramidal vortex structures .
results are compared with analytical work and numerical studies based on the vortex filament method . |
the perturbed two - body problem is one of the central areas of interest of celestial mechanics .
a wide class of perturbations of the dynamics of a binary system are described by the generic brumberg force [ @xcite ] .
another type of perturbation is provided by general relativistic modifications of gravitational dynamics .
such measurable effects arise in planetary motion ( like the general relativistic contribution to the precession of the perihelion of the planet mercury ) , however more important are the modifications in the strong - field regime , typically appearing in compact binaries composed of neutron stars and / or black holes .
a well - known example is the hulse - taylor double pulsar [ @xcite ] , the orbital period of which is continuously decreasing as gravitational radiation escapes from the system [ @xcite ] ( for a recent review see [ @xcite ] ) .
efforts to directly detect gravitational waves with earth - based interferometric detectors ( ligo[@xcite ] , virgo [ @xcite ] , geo [ @xcite ] and tama [ @xcite ] ) are currently under way .
gravitational waves from distant coalescing galactic black hole binaries will be searched for by the forthcoming laser interferometer space antenna ( lisa ) [ @xcite ] . with improving detector sensitivity , upper limits for the signal emitted by different configurations of gravitational wave sources
were already set ( [ @xcite ] , [ @xcite ] , [ @xcite ] ) .
capturing gravitational waves in the noisy background requires a 3.5 post - newtonian ( pn ) accuracy both in the orbital phase [ @xcite ] and in the energy loss [ @xcite ] occurring in a compact binary .
the backreaction of the escaping gravitational radiation on the orbit decreases the orbital period of the compact binary .
although higher order corrections to this effect are too small to be directly detected in the early stages of the inspiral , they are important in the latter stages , when the post - newtonian parameter increases up to @xmath0 ( at @xmath1 gravitational radii separation ) and higher order contributions become quite significant . therefore the contributions to the energy loss @xmath2 and angular momentum loss @xmath3 have been worked out up to 2pn order accuracy in the wave generation ( or equivalently , to 4.5 pn order following the keplerian picture ) , in the case of eccentric orbits , and up to even higher orders for circular orbits .
( here @xmath4 denotes the time - average over one period , e.g. a secular contribution . )
the corrections arise at each pn order from both general relativistic modifications of the dynamics and/or as a manifestation of various physical characteristics of the system , like the spins , mass quadrupole momenta and magnetic dipole momenta of the constituents .
the final aim of gravitational wave astronomy is to incorporate all of these physical characteristics into accurate templates to be used in data processing .
recovering the physical characteristics of astronomical objects from a simultaneous analysis of data collected in both the electromagnetic spectrum and gravitational wave bandwidth would be a major breakthrough in modern astronomy . driven by this motivation
, the interest in the perturbed two - body problem has been renewed . in this context , in [ @xcite ] a generic perturbed two - body problem with radial equation @xmath5 was considered .
the point mass @xmath6 ( representing the reduced mass of the binary ) is orbiting a fixed gravitational centre @xmath7 ( the total mass of the system ) at distance @xmath8 ( the separation of the components ) .
the energy @xmath9 and the magnitude of the orbital angular momentum @xmath10 are constants of the _ perturbed _ motion and @xmath10 from the rest of the coefficients of @xmath11 and @xmath12 in the radial equation is advantageous , as it will enable us to express all results in terms of constants with physical interpretation . ] .
the expression @xmath13 ( the subscript @xmath14 denoting newtonian ) has the familiary newtonian functional form , however @xmath9 and @xmath10 characterize the perturbed motion .
the perturbing terms contain the small coefficients @xmath15 , assumed _ constants_. as will be shown in detail in section 3 , both the generic perturbing brumberg force , the first post - newtonian order ( 1pn ) general relativistic corrections [ @xcite ] and the spin - orbit ( so ) interaction for compact binaries [ @xcite ] fit into this scheme .
the generic equation ( [ radial ] ) defines a radial period @xmath16 and turning points @xmath17 . with these assumptions the integrals of the type @xmath18 ( with constant @xmath19 and @xmath20 arbitrary integer )
could be easily evaluated by use of the _ residue theorem_. important steps in the proof of [ @xcite ] were the introduction of suitable generalized true and eccentric anomaly parametrizations of the radial motion , @xmath21 and @xmath22 as @xmath23 together with their complex counterpart @xmath24 and @xmath25 , respectively .
the advantage of this method over other methods for computing the secular integrals ( [ integral ] ) is its striking simplicity .
as proven in [ @xcite ] , in the majority of physically relevant cases the only pole is in the origin .
extremely rarely , a second pole ( also given ) occurs .
the method was successfully applied in [ @xcite ] for the evaluation of the averaged energy and angular momentum losses due to the so interaction for compact binaries on elliptic orbit , and the results were in agreement with those of [ @xcite ] , obtained by a different method .
the more accurate study of the inspiral phase of compact binaries however reveals other important contributions to be taken into account .
these are the perturbations due to the ( a ) spin - spin ( ss ) interaction [ @xcite ] ; ( b ) the interaction of the magnetic dipole moments for neutron stars with huge magnetic fields ( dd ) [ @xcite ] and ( c ) the contribution arising from the motion of one of the binary components ( treated as a monopole ) in the quadrupolar field of the other ( qm ) [ @xcite ] .
the corresponding secular contributions to the energy and angular momentum losses could be computed [ @xcite ] , [ @xcite ] , and [ @xcite ] by the method presented in [ @xcite ] , in spite of the radial equation being _ different _ from eq .
( [ radial ] ) , which rendered these computations outside the domain of validity of theorems 1 and 2 of [ @xcite ] .
therefore the question arose , whether the results presented in [ @xcite ] would hold more generically .
in all above mentioned cases the radial equation was a generalization of eq .
( [ radial ] ) with the coefficients @xmath15 being _ periodic functions _ of the true anomaly @xmath26 .
thus in this paper we raise the question , whether the results of [ @xcite ] would hold for _ generic _ periodic functions @xmath27 .
the answer we find is that some restrictions apply , and we derive the class of admissible periodic functions @xmath15 .
section 2 contains the announcement and proof of the _ generalized _ theorems 1 and 2 , as compared to [ @xcite ] .
starting from the explicit expressions for the respective interactions , we show in section 3 , that the ss , qm and dd perturbations fit into the class of admissible perturbing functions @xmath15 . in section 4 , as a simple application of the presented method , we give the self - interaction spin contributions to the luminosity of a compact binary consisting of black holes , in the case when @xmath28 .
the turning points of the radial motion ( [ radial ] ) , with @xmath29 are the loci @xmath30 , given by : @xmath31 ^{i-2}\ .
\label{rminmax}\ ] ] here @xmath32 represents the magnitude of the laplace - runge - lenz vector characterizing a keplerian motion with @xmath9 and @xmath33 and we have introduced the notations @xmath34 unless in the case presented in [ @xcite ] , here @xmath35 .
the expression of the turning points allow to write up the parametrizations ( [ true ] ) and ( [ eccentric ] ) in detail , whenever needed .
the integrals ( [ integral ] ) could be formally rewritten in terms of the integration variable @xmath26 as @xmath36 for @xmath37 the factor @xmath38 in the denominator of the integrand in eq .
( [ int1 ] ) has the binomial expansion @xmath39 ^{n}=\sum_{k=0}^{n}\left ( _ { k}^{n}\right ) \!\left ( \!\frac{1\!+\!\cos \chi } { r_{min}}\right ) ^{k}\!\left ( \frac{1\!-\!\cos \chi } { r_{max}}\right ) ^{n - k}\ . \label{binomial}\ ] ] by applying the chain rule @xmath40 , with @xmath41 derived from the true anomaly parametrization ( [ true ] ) , and inserting @xmath42 as the series expansion of the radial equation ( [ radial ] ) , we find for the second factor of the integrand ( [ int1 ] ) @xmath43\ .\end{aligned}\ ] ] the second term of the expression ( [ dtdchi ] ) contains @xmath44 in the denominator , and apparently becomes singular at @xmath45 ( @xmath46 any integer ) .
the key point of the proof of theorem 1 of [ @xcite ] was that a factor of @xmath44 could be also separated from @xmath47 , so that the integrand of @xmath48 became regular everywhere .
the proof has also showed that there is no other @xmath26-dependence left in the denominator .
we would like to see whether this convenient property of @xmath47 holds for generic periodic functions @xmath49 .
for this we decompose @xmath49 as follows @xmath50 the coefficients @xmath51 and @xmath52 being constants
. then the condition @xmath53 follows .
( [ series_phi ] ) is equivalent with a decomposition into a generic fourier series , however better suited for our purposes .
its advantage lies in the property of the terms with @xmath54 , of manifestly containing the factor @xmath44 .
therefore singular integrands in ( [ int1 ] ) could arise only from the terms containing @xmath55 , @xmath56 , @xmath57 and @xmath58 .
we will discuss these one by one . as @xmath55 is the constant part of @xmath49 , the contributions containing @xmath55 are the constant perturbing functions already dealt with in [ @xcite ] .
thus the terms containing @xmath55 are regular .
a similar algebra as for the terms with @xmath55 applies to the terms containing @xmath56 . by employing the binomial expansion
( [ binomial ] ) and the expression of the turning points ( [ rminmax ] ) , those contributions to @xmath47 which contain the factor @xmath59 sum up to @xmath60 \,,\ ] ] the expressions @xmath61 are defined by eqs .
( 29)-(31 ) of [ @xcite ] , and they are proportional to @xmath44 .
this proves that the terms with @xmath56 do not give singular contributions to the integrand of @xmath48 . what remains to check are the @xmath62 terms of @xmath47 , given by @xmath63 ( we have used that , according to eq .
( [ fi0gi0 ] ) , the expressions @xmath64 receive no contribution from @xmath65 and @xmath66 ) as the coefficients @xmath67 are already of first order , the keplerian true anomaly parametrization @xmath68 can be employed in eq .
( [ gamma1 ] ) . a tedious but straightforward algebra , based on ( a ) the binomial expansion of @xmath69 ; ( b ) the replacement cos@xmath70 , whenever possible ; ( c ) a second binomial expansion of @xmath71 , whenever applicable and ( d ) grouping together the terms without a @xmath72 factor , into terms proportional to @xmath73 and @xmath74 , leads to the following two conditions on the coefficients @xmath57 and @xmath58 : @xmath75 which have to be satisfied in order @xmath76 to be regular for every @xmath77 .
the last question to be addressed is what are the conditions to be imposed on @xmath19 , in order that @xmath48 stays regular .
the answer is simple : @xmath19 can be any periodic function of @xmath26 , which is regular .
such functions will not affect the proof presented above .
with these conditions satisfied , all factors in the integrand of @xmath78 become simple polynomials in @xmath79 and @xmath80 , provided @xmath81 .
this renders the only pole in the origin , and we have proven : * theorem 1 * : _ for all perturbed keplerian motions characterized by the radial equation ( [ radial ] ) , with periodic perturbing functions _ @xmath15 _ _ obeying the conditions ( [ series_phi ] ) and ( [ condition1 ] ) , and for arbitrary periodic functions _ _ @xmath19 _ _ the integral _ _
@xmath82 _ is given by the residue in the origin of the complex true - anomaly parameter plane . _ in order to cover the complementary case of @xmath83 , we employ the eccentric anomaly parametrization ( [ eccentric ] ) .
the integrals @xmath84 then are evaluated as @xmath85 and for any @xmath86 we apply the binomial expansion @xmath87 which is a polynomial in @xmath88 . from the eccentric anomaly parametrization ( [ eccentric ] ) and the radial equation ( [ radial ] ) ( with @xmath89 in place of @xmath90 ) we find @xmath91\ .\end{aligned}\ ] ] here @xmath92 and @xmath93 are identical with those defined in terms of the true anomaly ( this is , because at @xmath94both angles @xmath26 and @xmath95 take the values @xmath96 and @xmath97 , respectively)@xmath98 considerations completely analogous to those in the proof of theorem 1 , together with the relation between the two parametrizations @xmath99 after lengthy computations lead to the following result . provided the eqs .
( [ condition1 ] ) hold , a factor of @xmath100 separates out from @xmath101 and cancels out the @xmath100 factor of the denominator in eq .
( [ dtdxi ] ) .
as main difference with respect to the true anomaly parametrization , the terms @xmath102 in the expression ( [ iota ] ) could lead to additional factors @xmath103 in the denominator of the integrand .
therefore , the second pole @xmath104 could appear in the process of passing to a complex eccentric anomaly parameter .
we can now enounce our * theorem 2 * : _ for all perturbed keplerian motions characterized by the radial equation ( [ radial ] ) , with periodic perturbing functions _ @xmath15 _
_ obeying the conditions ( [ series_phi ] ) and ( [ condition1 ] ) , and for arbitrary periodic functions _
_ @xmath19 , _ the integral _
@xmath105 _ is given by the sum of the residues in the origin of the complex eccentric - anomaly parameter plane and ( for any _
@xmath51 _ _ with _ _ @xmath106 _ or _
@xmath52 _ _ with _
_ @xmath107 _ _ in _ _ @xmath108 _ _ ) at _ _ @xmath109__. _ _ our theorems contain as special cases the theorems presented in [ @xcite ] . there ,
the conditions ( [ condition1 ] ) are automatically satisfied , as @xmath15 are constants .
it is remarkable , that the constancy of @xmath15 can be relaxed in a quite generic way , in which the infinite number of coefficients in the expansion of @xmath110 should obey only the two contraints ( [ condition1 ] ) .
in this section we present two important classes of dynamical systems , for which our theorems apply .
the generic perturbing brumberg force [ @xcite ] , [ @xcite ] , includes a wide range of perturbations employed in celestial mechanics .
it can be derived from the lagrangian @xmath111 where @xmath112 and @xmath113 are perturbation parameters .
the radial equation derived in [ @xcite ] from this lagrangian is of the form ( [ radial ] ) , with the small coefficients @xmath114 as all @xmath115 are constants , these perturbations fit into the domain of validiy of the methods discussed in [ @xcite ] .
another important example is provided by compact binary systems , consisting of black holes / neutron stars .
their dynamics is conservative up to the 2pn order , the first dissipative effects due to gravitational radiation occuring at 2.5 pn order .
the conservative dynamics including all leading order contributions , due to either various physical characteristics , or to the general relativistic modifications of the motion is described by the lagrangian @xmath116 with the various contributions derived first in [ @xcite ] ( pn ) , [ @xcite ] , ( so ) , [ @xcite ] ( ss ) , [ @xcite ] ( qm ) and [ @xcite ] ( dd ) read and @xmath117 .
the angle subtended by them is @xmath118 .
the total spin is @xmath119 and @xmath120 .
the magnitude and direction of the magnetic dipole moments @xmath121 are denoted as @xmath122 and @xmath123 .
they subtend the angle @xmath124 with each other . in a coordinate systems @xmath125 with the axes @xmath126 , where @xmath127 is the unit vector in the @xmath128 direction ,
the polar angles @xmath129 and @xmath130 of the spins are defined as @xmath131 ( see [ @xcite ] ) . in the coordinate system @xmath132 with the axes @xmath133 , where @xmath134 are the unit vectors in the @xmath135 directions , respectively , the polar angles @xmath136 and @xmath137 of the the magnetic dipole moments @xmath138are @xmath139 ( see [ @xcite ] ) . the quadrupolar parameters ( see [ @xcite ] ) are defined as @xmath140 , where @xmath141 is the quadrupole - moment scalar [ @xcite ] of the @xmath142 axially symmetric binary component with symmetry axis @xmath143 . ] : @xmath144 \ , \nonumber \\ \mathcal{l}_{so } & = & \frac{g\mu } { 2c^{2}r^{3}}\mathbf{v}\cdot \lbrack \mathbf { \ r}\times ( 4\mathbf{s}+3{\mbox{\boldmath $ \sigma$ } } ) ] \ , \nonumber \\ \mathcal{l}_{ss } & = & \frac{g}{c^{2}r^{3}}\left [ \left ( \mathbf{s}_{\mathbf{1 } } \cdot \mathbf{s}_{\mathbf{2}}\right ) -\frac{3}{r^{2}}\left ( \mathbf{r\cdot s } _ { \mathbf{1}}\right )
\left ( \mathbf{r\cdot s}_{\mathbf{2}}\right ) \right ] \ , \nonumber \\ \mathcal{l}_{qm } & = & \frac{g\mu m^{3}}{2r^{5}}\sum_{i=1}^{2}p_{i}\left [ 3\left ( \mathbf{\hat{s}}_{\mathbf{i}}\cdot \mathbf{r}\right ) ^{2}-r^{2 } \right ] \ , \nonumber \\ \mathcal{l}_{dd } & = & \frac{1}{r^{3}}\left [ 3(\mathbf{n\cdot d_{1}})(\mathbf{\ n\cdot { d}_{2}})-\mathbf{d_{1}\cdot { d}_{2}}\right ] \ .
\label{lag}\end{aligned}\ ] ] note that the 1pn accurate relativistic corrections to the keplerian motion characterized by @xmath145 is the particular case of the perturbing brumberg lagrangian @xmath146 , with the specifications @xmath147 ( where @xmath148 ) .
therefore the pn perturbations fit into the domain of validiy of the methods discussed in [ @xcite ] .
the same holds for the so perturbations , where the only non - vanishing @xmath149 in the radial equation is @xmath150 this is a constant , as the angles @xmath129 span by the spins with the orbital angular momentum @xmath151 are constants ( to leading order ) . nor the radial equation with the perturbing function ( [ phiso ] ) characterize uniquely the so - perturbation .
this is , because a spin - supplementary condition ( ssc ) should be additionally imposed .
we have followed here the ssc of [ @xcite ] and [ @xcite ] .
there are other known ssc - s , including the covariant ssc employed in [ @xcite ] , which gives the non - vanishing constant perturbations @xmath152and @xmath153 .
these can again be expressed in terms of the angles @xmath154 , which are cons@xmath155ts in the leading order , required here . ]
the situation is entirely different for the ss , qm and dd contributions to the dynamics of a compact binary . in all these cases
the energy @xmath9 is a constant of motion , while ( in contrast with the 1pn and so contributions ) the magnitude @xmath33 of the orbital angular momentum is not .
indeed , with the notable exception of the pn contribution , it can be derived from the respective lagrangians that the orbital angular momentum @xmath156 is not conserved:@xmath157 .
while for so perturbations this means merely a precessional change about the vector @xmath158 ( so that @xmath10 is conserved ) , the situation is more complicated in the ss , qm and dd cases , in which @xmath10 is a function of the orbital position @xmath159 . in all these cases
however an angular average @xmath160 over the radial motion can be computed , which is conserved to the accuracy of linear effects we consider here .
the reason for this is that up to 2pn , the dynamics is conservative .
a description in terms of the constants of motion @xmath9 and @xmath161 is then in order , as described in detail in ( [ @xcite ] , [ @xcite ] and [ @xcite ] . the radial equation in all these cases can be expressed in the form ( [ radial ] ) , with @xmath160 in place of @xmath10 while the perturbing terms contain the following explicit @xmath162-dependences : @xmath163 \nonumber \\ & & + ( 3gm\mu \!+\!2\overline{a}\cos \!\chi ) \cos ( 2\chi \!+\delta ) \}\ , \nonumber \\
_ { 3}^{ss } & = & \frac{g\mu } { c^{2}}s_{1}s_{2}[3\cos \kappa _ { 1}\cos \kappa _ { 2}-\cos \gamma \nonumber \\ & & -3\sin \kappa _ { 1}\sin \kappa _ { 2}\cos ( 2\chi \!+\delta ) ] \ , , \nonumber \\ \varphi _ { 2}^{qm } & = & -\frac{gm^{3}\mu ^{3}}{2\overline{l}^{2 } } \!\!\sum_{i=1}^{2}p_{i}\sin ^{2}\!\kappa _ { i}\{2\overline{a}\cos \!\left ( \chi \!+\!\delta _ { i}\right ) \nonumber \\ & & + \!\left ( 3g\mu m\!+\!2\overline{a}\cos \!\chi \right ) \cos ( \!2\!\chi \!+\!\delta _ { i})\}\ , , \nonumber \\ \varphi _ { 3}^{qm } & = & -g\mu ^{2}m^{3}\sum_{i=1}^{2}p_{i}\left [ 1-\!3\!\sin ^{2}\!\kappa _ { i}\cos ^{2}\!\left ( \!\chi \!+\frac{\delta _ { i}}{2}\!\right ) \right ] \ , , \nonumber \\ \varphi _ { 2}^{dd } & = & -\frac{\mu ^{2}d_{1}d_{2}}{\overline{l}^{2}}\left [ ( 3gm\mu + 4\overline{a}\cos \chi ) \mathcal{b}_{2}(\chi ) -\overline{a}\sin \chi \mathcal{b}_{2}^{\prime } ( \chi ) \right ] \ , , \nonumber \\ \varphi _ { 3}^{dd } & = & -\mu d_{1}d_{2}\left [ \mathcal{a}_{0}-3\mathcal{b } _ { 2}(\chi ) \right ] \ , , \label{phivar}\end{aligned}\ ] ] where @xmath164 is the angle between the periastron and the node line [ @xcite ] , @xmath165 represents an average , @xmath166 and @xmath167 differences in the azimuthal angles .
the quantty @xmath168 is the magnitude of the laplace - runge - lenz vector characterizing a keplerian motion with @xmath9 and @xmath169 .
finally , @xmath170 and @xmath171 denote @xmath172 where @xmath173 , @xmath174 .
note that for the ss , qm , and dd interactions there are no contributions @xmath108 for any @xmath175 .
it is straightforward to transform the expressions ( [ phivar ] ) into the form ( [ series_phi ] ) , and to identify the coefficients @xmath65 and @xmath176 . from among the various emerging contributions ,
those entering in the constraint ( [ condition1 ] ) are @xmath177 , @xmath178 , @xmath179 and @xmath180 , with @xmath181 a complicated expression depending on the particulars of the ss , dd or qm contributions @xmath182\ \
.\end{aligned}\ ] ] it is easy to verify that the condition ( [ condition1 ] ) holds in all cases , irrespective of the particular form of @xmath181 .
this is why the method developed originally for constant @xmath15-s could be employed for these cases as well
gravitational radiation produces dissipative effects in the orbital evolution .
while the energy @xmath9 , and the total orbital momentum @xmath183 are conserved up to 2pn order accuracy ( including the conservative perturbations considered in this paper and the 2pn relativistic perturbation ) , they start to evolve due to escaping gravitational radiation .
this dissipative evolution can be conveniently computed with our method , presented in section 2 .
the secular losses due to gravitational radiation are exactly integrals of the type ( [ integral ] ) , in which the condition of periodicity of @xmath19 is obeyed for all losses of ss , dd and qm type .
as already argued in section 3.b , the radial equation written in terms of @xmath160 is also of the form ( [ radial ] ) in all of these cases .
therefore our theorems can be applied in computing secular radiative changes .
such secular effects were discussed in ( [ @xcite ] , [ @xcite ] and [ @xcite ] ) for the ss , dd and qm contributions , respectively . in this section
we would like to apply the method for computing the so - called self - spin contribution to the gravitational luminosity ( energy loss ) of compact binaries composed of black holes .
for a black hole the quadrupole moment scalar is directly related to the spin : @xmath184[@xcite ]
. therefore the qm losses are effectively self - spin contributions and add to those discussed in [ @xcite ] , [ @xcite ] .
such contributions become important whenever the mass ratio of the compact binary is at least @xmath185 , which is frequently the case for colliding galactic black holes , which are the most important source types to be searched for by lisa . in such cases , as @xmath186
, we can safely ignore @xmath187 .
therefore the only ss interaction is the self - spin contribution .
coefficients @xmath188 in the self - spin contribution to the luminosity .
[ cols="^ , > , > , > , > " , ] the instantaneous energy loss can be found by taking the appropriate derivatives of the mass and current quadrupole tensors of the mass @xmath6 , as described for example in [ @xcite ] .
the self - spin contribution to the instantaneous energy loss computed in this way is a complicated expression of @xmath189 and @xmath26 . by employing the parametrization ( [ true ] ) ,
we obtain @xmath190 as function of @xmath26 alone . the secular contribution to
the energy loss arises by integration of @xmath190 over one radial orbit .
this integration is carried out by computing the residues enclosed in the circle @xmath191 . due to theorem 1 ,
we are assured that there is only one pole , at @xmath192 .
the summed - up self - spin contribution to the luminosity is @xmath193 \sin ^{2}\widetilde{\kappa } \right\ } \ , \label{luminosity}\ ] ] where @xmath194 the angles @xmath195 and @xmath196 are shown in fig[fig1 ] .
the coefficients @xmath188 are enlisted in table [ table ] .
( [ luminosity])-([z ] ) contain the totality of second order contributions in the spins to the luminosity , provided @xmath197 , and as such , represent the first correction to the lense - thirring approach , discussed in [ @xcite ] .
is the spin of the supermassive black hole with mass @xmath198 , while the spin of @xmath199 is negligible .
as there is either no @xmath200 or @xmath201-contribution to the orbital angular momentum [ @xcite , @xcite ] , @xmath202 is the newtonian orbital angular momentum .
@xmath195 is the angle between @xmath203 and @xmath151 .
the angle between the node line and projection of the spin in the plane of the orbit is denoted as @xmath204 . ]
perturbed keplerian motions can be parametrized in many different ways .
( for a review of these parametrizations see [ @xcite ] . )
the method of computing secular effects presented in [ @xcite ] relies on novel complex parametrizations of the radial part of the motion and its advantage over previous approaches [ @xcite ] , [ @xcite ] is in its overwhelming simplicity .
other methods can of course be successfully applied , like the use of the laplace second integrals for the legendre polynomials ( [ @xcite ] , [ @xcite ] ) . in the present work we have generalized both the class of integrands and the class of perturbing forces for which the simple method relying on the use of complex true and eccentric anomaly parametrizations applies . in a generic perturbed two - body problem , it is easy to see whether a particular secular effect to be computed is of the type ( [ integral ] ) , with @xmath19 periodic , and then to check whether the periodic perturbing functions @xmath15 obey the conditions ( [ condition1 ] ) . then , depending on the power @xmath20 , either theorems 1 or 2 can be applied .
we have also shown explicitly , that for several physically interesting linear perturbations in the conservative dynamics these theorems apply .
whether the so , ss , qm or dd perturbations are of 1pn order or higher , depends on the specific system considered . for comparable mass compact binaries for example
the so contribution is of 1.5pn order , while the ss , qm , dd contributions are all of order 2pn .
by contrast , when one of the masses of the compact binary dominates over the other , the ss contribution becomes negligible , however the so - called self - spin contribution of the dominant spin becomes important .
driven by this remark , as an application to the presented method , we have computed the self - spin contribution to the gravitational luninosity .
this work was supported by otka grants no .
t046939 and ts044665 .
l..g . wishes to thank the jnos bolyai scholarship for support . | binary systems subject to generic perturbations evolve on quasiperiodic orbits .
we derive the most generic class of perturbations , which allow to evaluate secular effects via generalized complex true and eccentric anomaly parameters , by use of the residue theorem .
such perturbations include both the generic brumberg force and various linear contributions to the conservative dynamics of compact binaries , up to the second post - newtonian order . as an example , we compute the self - spin contribution to the luminosity due to gravitational radiation of a compact binary consisting of galactic black holes , the most important type of source for lisa .
this translates to simply calculate the residue of the instantaneous energy loss in the origin of the complex parameter plane , which illustrates the power of the presented method . |
in this section we will consider a large class of systems with a generic number of degrees of freedom @xmath5 .
let us define the operator @xmath6 which is the column vector composed by @xmath5 generalized coordinates together with @xmath5 canonical conjugate momenta .
it is possible to express the canonical commutation relations involving coordinates and momenta compactly as @xmath7 = i \hbar \ , \mathsf j_{jk}$ ] with @xmath8 the elements of the symplectic matrix @xmath9 here @xmath10 and @xmath11 are the @xmath5 dimensional identity and zero matrix , respectively . in the remainder of these notes
, we will be dealing with quadratically coupled harmonic oscillators . in this scenario ,
the use of first and second moments of @xmath12 provides a powerful tool for the description of the physically relevant quantities involved in the evolution of the system .
we thus introduce the mean value ( mv ) vector @xmath13 $ ] and the covariance matrix ( cm ) @xmath14 of elements @xmath15.\ ] ] the focus of our work will be the study of a nearest neighbor - coupled harmonic chain whose @xmath5 elements are in contact with ( individual ) local reservoirs at finite temperature .
the evolution of the chain can thus be described , in general , through the lindblad master equation @xmath16 - \frac{1}{2\hbar } \ !
\sum_{m } ( \ { \hat{l}_m^\dag \hat{l}_m , \hat\rho \ } - 2 \hat{l}_m \hat\rho \hat{l}_m^\dag ) \ ] ] with the following general form of a quadratic hamiltonian and linear lindblad operators @xmath17 where @xmath18 is the adjacency matrix of the hamiltonian , @xmath19 is a column vector encompassing possible position and momentum displacements , @xmath20 represents a possible energy offset , and @xmath21 contains the coupling strengths between a given element of the chain and the respective reservoir . finally , @xmath22 are constants @xcite .
using such a general description of the coherent and incoherent part of the evolution , we can straightforwardly work out the dynamical equations of motion for both @xmath23 and the cm by calculating their time derivative and using the state evolution provided by eq .
( [ lindblad ] ) as @xcite @xmath24 where we have introduced @xmath25 and @xmath26 which are defined in terms of the _ decoherence _ matrix @xmath27 . by definition
, we have @xmath28 and @xmath29 . for time independent problems ,
( [ mvcm ] ) can be solved exactly as @xmath30 with @xmath31 and @xmath32 the mv and cm of the initial state , respectively .
the steady - state ( or fixed - point ) solutions of such equations can be found by imposing @xmath33 , which are equivalent to the conditions ( in what follows , the subscript @xmath34 will be used to indicate steady - state values ) @xmath35 the equation satisfied by @xmath36 is of the stationary lyapunov form @xcite that , under the conditions above , admits a unique positive - definite solution iff the eigenvalues of @xmath37 have positive real parts . in this case
, we find @xmath38 for - time dependent @xmath39 , @xmath40 , @xmath41 , and @xmath42 , the form of @xmath43 in eq .
( [ mvcmss ] ) is no longer valid and the conditions over the lyapunov equation for @xmath36 must hold at each instant of time .
note that , in order to deduce eqs .
( [ mvcm ] ) , ( [ mvcmsol ] ) , and ( [ cmss ] ) , we did not need to make any assumption on the initial state of the system but only use the quadratic and linear structure of eq .
( [ ham - lind ] ) and ( [ lindblad ] ) , respectively .
the rest of our analysis will focus on the dynamics of thermodynamically relevant quantities such as currents and energy .
we can now start analyzing explicitly the system that we have in mind .
we consider the chain of oscillators depicted in fig .
[ fig1chain ] , each interacting with its own thermal reservoir at temperature @xmath44 .
the hamiltonian of the chain arises from the application of the rotating - wave approximation on a nearest - neighbour hooke - like coupling model , which gives us @xmath45 where @xmath46 and @xmath47 are the frequency of the oscillators and their mutual coupling rate respectively , and @xmath48 is the creation operator of the @xmath49 oscillator . using the notation introduced before , @xmath50 can be written as in eq .
( [ ham - lind ] ) with @xmath51 and the adjacency matrix @xmath52 , where @xmath53 and @xmath54 is the kronecker symbol .
the coupling between a given oscillator and the respective thermal reservoir is described by the lindblad operators @xcite @xmath55 where @xmath56 is the bath - oscillator coupling and @xmath57 is the mean occupation number of the reservoirs at temperature @xmath58 .
this choice allows us to make the identifications @xmath59 we now make the explicit assumption that the first and last oscillator in the chain are also affected by two additional reservoirs , which we label @xmath60 and @xmath61 , having temperatures @xmath62 .
this allows us to approximate @xmath63 for @xmath64 .
therefore , such baths contribute with @xmath65 we are now in a position to give some motivations for the specific choice of the system to study . in a realistic scenario ,
the impossibility to achieve full isolation leads one to take into account the external and uncontrollable influences from the environment over the evolution of a system . in our case
such disturbances are represented by the @xmath5 thermal reservoirs attached to each oscillator of the chain . on the other hand , as we aim at studying heat transport across the system , we need to set a temperature gradient , which is imposed , in our setting , by the external end - chain baths . as such gradient
is supposed to be the leading mechanism for the transport process , it is reasonable to assume that @xmath66 are the largest temperatures across the system .
we can now go back to the formal description of the system and write the decoherence matrix as @xmath67 with @xmath68 as the contribution given by the reservoirs @xmath60 and @xmath61 to the dynamics is all in the matrix @xmath69 of eq .
( [ dynmat ] ) , we refer to them as _ all - diffusive_. in order to simplify our analysis without affecting its generality , we now take @xmath70 . from eq .
( [ hamsys2 ] ) and the expression found for @xmath71 , we rewrite ( [ dynmat ] ) as @xmath72 with @xmath73 .
this allows us to achieve the following expression for the cm using ( [ mvcmsol ] ) @xmath74 with @xmath75 and @xmath76 defined in ( [ matrixu ] ) . following the lines
given in the appendix and integrating @xmath77 by parts , it is then possible to show that @xmath78 with @xmath79 the symbol for a hadamard matrix product @xcite and the matrices @xmath80 and @xmath81 having the elements @xmath82 t } } { \zeta + i ( { \nu_j -\nu_k } ) } \end{aligned}\ ] ] with @xmath83 ( _ cf . _
the appendix ) .
the steady - state form of such solution can be found as illustrated in the previous section , which yields @xmath84 with @xmath85 $ ] .
remarkably , eq . is the cm of a vacuum state corrected by terms whose origin is entirely ascribed to the presence of the reservoirs .
furthermore , the nullity of the diagonal elements of @xmath86 guarantees that , in the long - time limit , each oscillator is in a thermal state . if the reservoirs connected to the elements of the chain have all the same temperature ( so that @xmath87 ) , eq .
can be cast into the form @xmath88 with @xmath89 .
this is the cm of a thermal equilibrium state at temperature @xmath90 for the @xmath5 oscillators plus corrections due to the all - diffusive reservoirs . in the appendix
, we analyze in details some aspects of the structure of the cm in ( [ cmsssys ] ) and ( [ cmsssys4 ] ) .
in a thermodynamical system ruled by hamiltonian @xmath50 and described by the density matrix @xmath91 , the variation of the internal energy is associated with work and heat currents .
in fact @xmath92 while the first term in the right - hand side is associated with the work performed on / by the system in light of the time - dependence of its hamiltonian , the second term accounts for heat flowing into / out of the system itself .
as the hamiltonian of our problem is time - independent , any change in the mean energy of the chain should be ascribed to the in - flow / out - flow of heat . by inserting the right - hand side of eq . in place of @xmath93
above , we find @xmath94 with @xmath95 the heat current induced by the @xmath96 lindblad operator . physically speaking , eq .
( [ totcurr1 ] ) shows that the total heat current in the system is formed by the net result of currents due to each reservoir .
this largely enriches the phenomenology of heat propagation and thermalization in our system , especially compared to usual previous settings @xcite .
this expression can be specialized to the case of the system addressed in sec .
[ tsaid ] to give ( _ cf . _
the appendix ) @xmath97 .
\nonumber \end{aligned}\ ] ] summing over all lindblad operators , the total current reads @xmath98 % - { \rm tr } \left [ { \bf h } \left ( { \bf v } + \langle \hat x \rangle_{t } \langle \hat x \rangle_{t}^\top \right ) \mathsf j \ , { \rm im}{\bf \upsilon } \right].\ ] ] the first term is the diffusive part of the current and is constant in time if the set of @xmath99 s does not depend on time explicitly .
the system s internal energy can be easily worked out to take the general form @xmath100\\ % & = \frac{{\rm e}^{-\zeta t } } { 2 } \ !
{ \rm tr } ( \mathbf h { \bf v}_{\!0 } \!+ \ ! % { \bf h } \langle \hat x \rangle_0 \langle \hat x \rangle_0^\top)\ ! + \ !
% \frac{1}{2\zeta } \left ( 1 \!-\ ! { \rm e}^{-\zeta t } \right)\ ! % { \rm tr } ( { \bf h d } ) \\ % & = \frac{1}{2}{\rm e}^{-\zeta t } \ , { \rm tr } ( \mathbf h { \bf v}_{\!0 } ) + \frac{1}{2}{\rm e}^{-\zeta t } \langle \hat x \rangle_0 \cdot { \bf h } \langle \hat x \rangle_0 \\ & + \hbar\omega \left ( \frac{n}{2 } + \!\sum_{k=1}^n \bar n_k \ ! + \ ! \frac{\zeta_{\rm a}}{\zeta } \bar n_{\rm a } + \frac{\zeta_{\rm b}}{\zeta } \bar n_{\rm b } \right ) \left ( 1 - { \rm e}^{-\zeta t } \right ) .
\end{aligned}\ ] ] at the steady state , we can write @xmath101 showing that the mean energy of the system does not depend on the coupling strength between the oscillators and is fully determined by the the interactions with the reservoirs .
the contribution that each oscillator gives to the equilibrium energy in eq .
( [ sshsys1 ] ) is not uniform across the chain , as can be seen from fig .
[ fig2therm ] where we plot the mean occupation number of each oscillator @xmath102_{kk } -1/2.\ ] ] despite the individual contribution of each bath for the mean energy in ( [ sshsys1 ] ) , the state of the chain is described by the cm ( [ cmsssys4 ] ) , which encompasses the collective effects of all the reservoirs resulting from the mixing process effectively implemented by the inter oscillator coupling . with @xmath103
. the ( blue ) square points are for @xmath104 with @xmath103 .
finally , the ( violet ) dots are for @xmath104 with @xmath105 .
the remaining parameters are @xmath106 , @xmath107 , @xmath108 .
for all the three plots k = 1, .. ,n.,width=302 ] with @xmath103 .
the ( blue ) square points are for @xmath104 with @xmath103 .
finally , the ( violet ) dots are for @xmath104 with @xmath105 .
the remaining parameters are as in fig .
[ fig2therm ] . , width=302 ] the general behavior of both quantities against the length of the chain can be seen from fig .
[ fig4thercur1 ] . for this range of @xmath5 ( chain lengths )
the local occupation number decreases with @xmath5 while the current does the opposite .
as far as we could check numerically , for @xmath121 , both quantities become independent of the length of the chain .
however , due to the lack of an analytical proof we can not assure that this is really the case in the thermodynamic limit .
obviously , the relation eq .
( [ currssintsys ] ) among these quantities is valid for any chain length .
furthermore , any oscillator in the bulk of the chain , _
i.e. _ , any element identified by a label @xmath122 , has the same occupation number of the reservoir attached to it .
this is due to the fact that , by eq .
( [ matrixol ] ) , @xmath123 and @xmath124 . moreover , the individual currents , _
i.e. _ , the currents due to the local standard thermal baths , are null according to eq .
( [ currssintsys ] ) .
this implies that in this limit the local baths play no major role in the thermalization of the bulk oscillators . in this respect ,
our results share some similarity with the classical approach of the problem , where a chain of harmonic oscillators is attached to just two thermal baths at its ends .
in fact , in the work by reider _
et al . _
@xcite , the bulk oscillators attains a constant temperature as in our case when the limit @xmath125 is considered . and the 10@xmath126 ( @xmath127 ) ] , the midpoint ( bulk ) @xmath128 , and the last oscillator @xmath129 .
we consider @xmath103 ; the remaining parameters are the same as in fig .
[ fig2therm].,width=321 ]
in this section we use the formalism and results illustrated so far to analyze the transport of heat in a few paradigmatic examples , all encompassed by the general treatment of the problem provided above . in particular , we want to study the effects caused by the presence of end - chain all - diffusive reservoirs , something not yet explored in this context before . to this end
, we now consider different variations on the basic setup described on fig .
[ fig1chain ] .
for example , we vary the distribution of reservoirs , the coupling mechanism among the oscillators and include dephasing . in order to establish a benchmark to evaluate the role played by the diffusive baths , we start the analysis by taking @xmath130 in ( [ lindexbath ] ) .
the corresponding steady - state cm is as in eq .
( [ cmsssys ] ) with @xmath131 while the steady - state energy and the current are given by eq .
( [ sshsys1 ] ) and eq .
( [ currsys ] ) , respectively . in order to remain as close as possible to the system discussed in the previous section , we take @xmath132 for @xmath133 . as all reservoirs are of the ordinary type , the approximation in eq .
( [ lindexbath ] ) does not hold for the end - chain baths .
however , as one can see from fig .
[ fig5therm ] , the same pattern for the mean excitation number displayed in fig .
[ fig2therm ] is found . in fig .
[ fig6curr ] , we then plot the currents , given in eq .
( [ currssintsys ] ) , generated by the attachment to the reservoirs with temperatures @xmath58 , @xmath134 .
they are all negative , as in fig .
[ fig3curr ] , and sum up to @xmath135 however , the actual value of the sum of the currents depends on the number of oscillators since @xmath136 and @xmath137 depends on the length of the system .
again , the individual currents are the difference between the energy stored in each oscillator and the mean energy occupation of the respective reservoir . as the number of oscillators in the chain increases , the current and mean energy behave very much like those in fig .
[ fig4thercur1 ] . at this point
, one might wonder about the reason for the negativity of the internal currents .
actually , it turns out that this is a simple consequence of the structure of eq .
( [ currssintsys ] ) when eq .
( [ zerocond ] ) is taken into account : numerical explorations shows that the oscillators attached to the highest temperature reservoirs will have @xmath138 , while the other oscillators will not as their occupation number will also be determined by their own lower - temperature reservoir and the contributions coming from the higher - temperature ones .
oscillators and temperatures of the baths given in terms of @xmath139 and @xmath140 ; squares ( blue ) : chain with @xmath104 oscillators and @xmath139 and @xmath140 ; circles ( violet ) : @xmath104 and @xmath141 and @xmath140 .
the remaining parameters are the same as in fig .
[ fig2therm ] except for @xmath142 and @xmath143 , which taken to be zero and in all cases @xmath144.,width=302 ] , although we have taken @xmath145 , _
i.e. _ , we have excluded the positive currents from the first and last reservoir.,width=302 ] in fig .
[ fig7curr ] , we plot the results valid for a different configuration , where one internal reservoir has the highest temperature .
notwithstanding the differences with respect to the patterns shown in figs .
[ fig5therm ] and [ fig6curr ] , the currents and energy of this configuration follow the same chain - length dependence discussed above
. oscillators with @xmath146 and @xmath147 .
the remaining parameters are the same as in fig .
[ fig2therm ] .
inset : mean occupation number of the oscillators in the chain . ,
width=302 ] let us consider now a chain of oscillators connected only to the all - diffusive reservoirs @xmath148 and @xmath149 .
when @xmath150 for @xmath151 the resulting dynamics is not stable ( _ cf . _
appendix ) .
however , the evolution of the system can be deduced from eq .
( [ cmsyst ] ) by taking @xmath152 to give @xmath153 with @xmath154 .
the diagonal elements of @xmath155 are obtained taking the limit @xmath156 and are given by @xmath157 . note that the ordering of the two limits above does not commute .
as in the case of eq.([cmsssys ] ) , the matrix elements @xmath158 .
consequently , the reduced cm of each oscillator is diagonal meaning that it is a thermal state for any time instant . in fig .
[ fig8therm ] , we show the mean occupation number of each oscillator , @xmath159 , at some instants of time for the evolution in eq .
( [ cmsyst2 ] ) and an initial vacuum state .
the process of excitation of the elements of the chain starts from its ends to then progressively move towards its center .
the mean occupation number of the oscillators increases on average linearly in time , _
i.e. _ , they oscillate by the effect of the orthogonal matrices @xmath160 around the linear rate given by @xmath161_{kk}$ ] [ _ cf . _
( [ cmsyst2 ] ) ] . as an interesting remark
, we observe that the distribution corresponding to the case of @xmath162 displays oscillators having occupation numbers larger than those of the oscillators in touch with the all - diffusive reservoirs .
this is an effect of the competition between linear time increase of @xmath163 and the oscillatory behavior induced by the actual absence of a steady state .
the total current for this system is obtained as @xmath164,\ ] ] and is thus a constant and that helps us in determining the mean energy of the system @xmath165 the indefinite growth of energy supplied by the current can be seen as a signature of instability of the system . in this context , the relative temperature of the two reservoirs is irrelevant since both contributes additively for the energy [ eq .
( [ meanshtsys2 ] ) ] and for the current [ eq . ( [ currsys3 ] ) ] .
oscillators attached only to two end - chain diffusive reservoirs .
we sample the dynamics at the instants of time @xmath166 . the chain is initially prepared in its vacuum state .
we have taken @xmath167 and @xmath168 .
as for the other parameters of the system , we have @xmath106 and @xmath108.,width=321 ] a standard thermal reservoir , as those distributed across the chain in fig . [ fig1chain ] ,
exchanges energy with the system in two distinct ways : diffusion , described by the matrix @xmath169 in eq .
( [ dynmat ] ) , which is responsible of the enhancement of energy ; and dissipation , described by @xmath37 , which extracts energy from the system . the balanced competition of these two effects drives the system to equilibrium at the steady state .
so far , the ordinary reservoirs and the all - diffusive have not been treated on equal footing , and it will be interesting to understand the behavior of a chain when a balanced competition of environmental mechanisms is considered . to this end ,
let us consider the system depicted in fig .
[ fig1chain ] with @xmath170 for @xmath171 , _
i.e. _ , while the reservoirs of the bulk chain are detached , those at the end of it ( having temperatures @xmath172 , @xmath173 , @xmath174 , and @xmath175 ) are still operative . in this case , the matrices ( [ dynmat ] ) are given by @xmath176 with @xmath177 and @xmath178the stability of this system is independent of the diffusive reservoirs since they did not contribute to @xmath179 in eq .
( [ dynsys2 ] ) .
furthermore , the action of the all - diffusive reservoirs is only to enhance the mean occupation number of the standard ones , that is , the solution to this problem is equivalent to take a chain with only two end - system standard reservoirs with mean occupation numbers @xmath180 and @xmath181 .
the methods employed in ref .
@xcite , which addressed the problem embodied by eq .
( [ dynsys2 ] ) with @xmath182 , will be useful to find a solution to this case . at the steady state attained by eq .
( [ dynsys2 ] ) , the sum of all currents is null ( as expected ) while the mean occupation number for each oscillator is given by the expressions @xmath183 all the internal oscillators have the same occupation number , as in ref .
the mean energy at the steady state can be evaluated from eq .
( [ meanshtsys ] ) @xmath184 the currents through the two oscillators are given by eq .
( [ currdissys ] ) , _ i.e. _ , @xmath185 with @xmath186 and are equal to @xmath187 this perfect balance implies that the currents at the stationary state for the two standard reservoirs are constrained to sum up @xmath188 which is the sum of the currents due to the all - diffusive ones .
this is analogous to what we have witnessed in sec .
[ tsaid ] .
since @xmath189^{-1 } $ ] , where @xmath190 is the inverse temperature , one can see that , in the classical limit @xmath191 , the current [ eq . ( [ sscur2 ] ) ] and the temperature of each reservoir in the bulk [ eq . ( [ ocn ] ) ]
behave as in the classical case @xcite .
that means that the currents are proportional to the temperature difference and the bulk oscillators thermalize at the mean value temperature of the reservoirs . in ref .
@xcite , assadian _
considered a chain of oscillators attached to two standard thermal reservoirs at the ends , which is the same configuration described in sec .
[ caiii ] but with @xmath182 . under these restrictions , the results in eq .
( [ sshsys3 ] ) and ( [ sscur2 ] ) remain valid and can be extracted from their work .
besides this example , they consider also the presence of @xmath5 purely dephasing reservoirs , each one attached to each oscillator of the chain .
the contribution of the dephasing mechanisms to the dynamics is modelled adding the following lindblad operators to the master equation regulating the dynamics of the system @xmath192 the special form of these reservoirs is such that they do not introduce new currents in the system .
this can be verified by calculating the individual currents to find that @xmath193 . on the other hand
, their presence drastically changes the behavior of the mean occupation value of each oscillator . as it can be seen from eq .
( [ meansssys ] ) , these are now given by @xcite @xmath194 ( \bar n_1 - \bar n_n ) } { 8\omega^2 + 2\zeta^2 + 2(n-1)\gamma \zeta } , \\
% \bar n^{(k)}_\star & = \tfrac{1}{2}(\bar n_1 + \bar n_n ) + \frac { ( n-2k+1 ) \gamma \zeta ( \bar n_1 - \bar n_n ) } { 8\omega^2 + 2\zeta^2 + 2(n-1)\gamma \zeta } , \\
% \bar n^{(n)}_\star & = \tfrac{1}{2}(\bar n_1 + \bar n_n ) - \frac { [ \zeta^2 + ( n-1)\gamma \zeta ] ( \bar n_1 - \bar n_n ) } { 8\omega^2 + 2\zeta^2 + 2(n-1)\gamma \zeta } \end{aligned}\ ] ] for @xmath195 . note that if the temperature of standard end - chain reservoirs are equal , @xmath196 , all the oscillators thermalize with the standard reservoirs having the same mean occupation number @xmath197 .
the behavior of the mean occupation number is plotted in fig . [ fig9therm ] .
dephasing reservoirs . diamonds ( green ) : chain with @xmath198 oscillators and bath mean occupation numbers @xmath139 . squares ( blue ) : chain with @xmath104 oscillators and @xmath139 . circles ( violet ) : @xmath104 and @xmath199 .
all other parameters are as in fig .
[ fig2therm].,width=302 ] as already commented , the dephasing reservoirs do not contribute to the currents . at the stationary state , we have @xmath200 this result is remarkable , as it shows that for @xmath201 , which is trivially satisfied for a large enough chain , a fourier - like dependence on the size of the system is recovered @xcite .
the classical version of the same problem has been studied in ref .
@xcite .
we now describe the modifications induced by the dephasing reservoirs when they are attached , one by one , to the system .
the lindblad operator in eq .
( [ linddeph ] ) can be rewritten as ( see also the appendix ) @xmath202 with the @xmath203 real matrix @xmath204_{jk } = \sqrt{\gamma_m } ( \delta_{jm } \delta_{mk } + \delta_{j+n\ , m } \delta_{m\,k+n}).\ ] ] without dephasing reservoirs ( @xmath205 ) , the currents in the system are the ones described by ( [ sscur2 ] ) with the substitutions @xmath206 and @xmath207 . following the prescriptions in the appendix
, we solve numerically the system with @xmath208 and @xmath209 , @xmath210 and calculate the current as function of the number of oscillators @xmath5 .
adding progressively more dephasing reservoirs until @xmath211 [ in this situation the current is given by eq .
( [ currsssysiii ] ) ] and calculating the current allows us to show in fig .
[ fig10curr ] the smooth transition from the situation described by eq .
( [ sscur2 ] ) to that associated to eq .
( [ currsssysiii ] ) . and the number of dephasing reservoirs @xmath212 .
we highlight the curves describing the extreme cases @xmath213 and @xmath129 , which are the functions reported , respectively , in eq .
( [ sscur2 ] ) with @xmath214 and in eq .
( [ currsssysiii ] ) with @xmath215 .
the dephasing coupling is @xmath216 while the remaining parameters are the same as in fig .
[ fig2therm ] .
, width=302 ] classically , size - dependent currents in chains of oscillators arise under the presence of anharmonicity or disorder @xcite , which can be realized in various ways .
one can , for example , introduce different frequencies and/or couplings across the chain . for the sake of definiteness
, we consider the system of fig .
[ fig1chain ] , now ruled by the hamiltonian @xmath217 the structure of eq .
( [ cmsssys ] ) remains the same with @xmath160 being replaced by the matrix that diagonalizes the adjacency matrix @xmath218 and @xmath219 ( which appears in the definition for @xmath220 ) being its eigenvalues and both can be calculated numerically for a given set of couplings @xmath221 . to introduce disorder ,
we arbitrary choose a set of coupling constants @xmath222 with magnitudes similar to what has been considered so far . in figs .
[ fig11therm ] and [ fig12curr ] , respectively , we plot the mean occupation number and the current for a set of distinct coupling constants .
structurally speaking , the behavior of the system does not change with the introduction of disorder .
this can be seen when comparing these figures , respectively , with fig .
[ fig2therm ] and fig .
[ fig3curr ] .
observe the similarities on the profile of the curves , the first , and last oscillators and , mainly , the behavior of the bulk .
furthermore , for the quantities shown in figs .
[ fig11therm ] and [ fig12curr ] , we also have analyzed different sets of inhomogeneous constant couplings @xmath223 . in the units of the paper , each @xmath222
is bounded by @xmath224 $ ] in our simulations in order to keep a fair comparison of the results . from these simulations ,
results obtained using different sets differ , but the general trends shown in those plots are preserved , especially the behavior of the bulk oscillators .
.,width=302 ] .,width=302 ] it is also interesting to see what happens with the individual currents and occupation numbers near the thermodynamical limit .
these are plotted in fig .
[ fig13thercur2 ] . as before ,
currents and mean occupation numbers for the oscillators in the bulk become independent on the length for long - enough chains .
however , it is interesting to remark that the last oscillator is strongly influenced by the disorder while the behavior of the remaining oscillators is essentially the same as in the case of fig .
[ fig4thercur1 ] . and the 10@xmath126 @xmath225 , the midpoint ( bulk ) @xmath226 , and the last oscillator @xmath129 for a chain with distinct constant couplings .
bottom panel : set of coupling constants used in the simulation .
we consider @xmath227 .
the remaining parameters are the same as in fig .
[ fig2therm].,width=321 ] until now , we have worked in the regime of the rotating - wave approximation ( rwa ) , which enables the explicit analytical form of the cm of the chain [ _ cf .
_ eq.([cmsyst ] ) ] .
this contrasts significantly with any classical approach to the transmission of heat across a harmonic chain , which is in general performed assuming the standard spring - mass coupling ( smc ) @xcite . in this subsection
we will thus briefly address the case of an smc - like coupling to make a more faithful comparison with the classical case .
the hamiltonian of a chain of @xmath5 oscillators coupled by the standard smc coupling has the adjacency matrix @xmath228 , where @xmath229 as this matrix is almost of the toeplitz form , a procedure similar to the one used in the appendix can be used to find the covariance matrix of the system , which is given by @xmath230 { \bf o } _ { { \bf \gamma}'}^{-\top}.\ ] ] here @xmath231_{jk } = -1/(\nu_j ' - \nu_k ' ) $ ] with @xmath232},\ ] ] and @xmath233 defined by @xmath234 we can now calculate the mean occupation number and the current for each oscillator , whose behavior is shown in fig .
[ fig14therm ] and fig [ fig15curr ] , respectively . with @xmath103 ; squares ( blue ) : chain of @xmath104 with @xmath103 ; circles ( violet ) : @xmath104 with @xmath105 .
the remaining parameters are @xmath235 , @xmath107 , and @xmath108.,width=302 ] . , width=302 ] as one can see , the results for the smc case are structurally similar to those found in the rwa
. such similarities extend also to the situations where either the diffusive baths are not considered or the distribution of coupling constants across the chain is not uniform .
we have investigated heat transport in quantum harmonic chains connected to different types of heat baths .
we have obtained the exact expression for the currents across the system highlighting the crucial role played by the properties of the environment .
such detailed analysis was instrumental to the study of a few paradigmatic configurations .
in particular , just like in ref .
@xcite , we have found that the fourier law is not predicted by the models considered here unless dephasing is included , destroying the ballistic behavior .
this is akin to a substrate external potential in the classical version of the problem addressed here . in our findings ,
fourier law is still not observed in the presence of disorder as long as the coupling with the bath is weak , allowing a description using a markovian master equations , as addressed in this manuscript .
our results are consistent with those known for unidimensional classical harmonic chains and shed new light on the interplay between transport properties of spatially extended quantum media and the nature of the environmental systems interacting with it .
fn , fls and mp are supported by the cnpq `` cincia sem fronteiras '' programme through the `` pesquisador visitante especial '' initiative ( grant nr .
401265/2012 - 9 ) .
mp acknowledges financial support from the uk epsrc ( ep / g004579/1 ) .
mp and af are supported by the john templeton foundation ( grant i d 43467 ) , and the eu collaborative project thermiq ( grant agreement 618074 ) . ai and mp gratefully acknowledge support from the cost action mp1209 thermodynamics in the quantum regime " .
ai is supported by the danish natural science research council .
fls is a member of the brazilian national institute of science and technology of quantum information ( inct - iq ) and acknowledges partial support from cnpq ( grant nr .
308948/2011 - 4 ) .
in this appendix we provide additional details on the mathematical approach to the problems addressed in the main body of the paper . here
we address the derivation of the expressions for the currents and mean energy for a system evolving according to eq .
( [ lindblad ] ) when a quadratic hamiltonian as the one in eq .
( [ ham - lind ] ) and quadratic lindblad operators are considered .
specifically , we assume the form @xmath236 where @xmath237 is a @xmath203 real matrix . taking the derivative of @xmath238 and @xmath239 defined in eq .
( [ cmdef ] ) , using the master equation ( [ lindblad ] ) , and recalling the commutation relation @xmath7 = i \hbar \ , \mathsf j_{jk}$ ] , we get the dynamical equations @xmath240 where @xmath241 both @xmath37 and @xmath169 are defined in eq .
( [ dynmat ] ) . by inserting eq .
( [ ham - lind ] ) and eq .
( [ alind ] ) into the definition of the individual @xmath242 s in eq .
( [ totcurr1 ] ) , we get @xmath243 \nonumber \\ & + & \tfrac{\hbar}{2 } { \rm tr } \left [ { \bf h } ( \mathsf j \delta_m)^2 \left ( { \bf v } + \langle \hat x \rangle_{t } \langle \hat x \rangle_{t}^\top \right ) \right ] \nonumber \\ & - & { \rm tr } \left [ { \bf h}\ , { \rm i m } \left(\lambda_m \lambda_m^\dag \right ) \mathsf j \left ( { \bf v } + \langle \hat x \rangle_{t } \langle \hat x \rangle_{t}^\top \right)\right ] \nonumber \\ & + & \mathsf j\xi \cdot \left[{\rm
i m } \left(\lambda _ m \lambda_m^\dag \right ) \mathsf j - \hbar \mathsf j { \delta}_m { \bf v } \ , { \delta}_m \mathsf j \right ] \langle \hat x \rangle_t \nonumber \\ & + & { \rm i m } ( \mu_m^\ast \lambda{_m } ) \cdot ( \mathsf j\xi - \mathbf h \langle \hat x \rangle_t ) .
\end{aligned}\ ] ] summing over all lindblad operators , see eq .
( [ totcurr1 ] ) , the total current has the following form @xmath244 + { \rm tr } \left [ { \bf h } \ , { \tilde{\bf \gamma } } \left ( { \bf v } + \langle \hat x \rangle_{t } \langle \hat x \rangle_{t}^\top \right ) \right ] \nonumber \\ & & + \ , ( \xi - \eta ) \cdot \mathbf h \langle \hat x \rangle_t - \mathsf j\xi \cdot \tilde{\bf \gamma}\langle \hat x \rangle_t , \end{aligned}\ ] ] which is zero for a possible steady state , _
i.e. _ , the solution of ( [ amvcm2 ] ) with @xmath245 .
the internal energy of the system is easily worked out as @xmath246 .
we get @xmath247 + \xi \cdot \mathsf j \langle \hat x \rangle_t + h_0.\ ] ] taking the derivative of this equation and rearranging the expressions one also finds eq .
( [ atotcurr2 ] ) . the matrix @xmath248 given by eqs .
( [ hamsys2 ] ) and appearing in ( [ dynsys ] ) is tridiagonal and symmetric toeplitz , and can thus be diagonalized by a simple ( symmetric ) orthogonal transformation @xcite as @xmath249 , where @xmath250 the matrix that diagonalizes @xmath37 in ( [ dynsys ] ) can then be constructed as @xmath251 with @xmath252 which gives us @xmath253 as the spectrum in eq .
( [ gammaeig ] ) has positive real part , @xmath254 is stable . moreover , in light of the fact that @xmath69 in eq .
( [ dynsys ] ) is a positive definite matrix , the system allows for a steady state , whose moments are given by ( [ mvcmss ] ) . by integrating the expression for @xmath77 by parts ,
we get @xmath255 - \zeta { \bf i } = { \rm e}^{-i { \pmb h } t } { \pmb d } { \rm e}^{i { \pmb h } t } { \rm e}^{-\zeta t } - { \pmb d}.\ ] ] by diagonalizing @xmath256 with the help of eq . ( [ toep ] ) and introducing the matrices @xmath257 , @xmath258 , and @xmath259 given in eq .
, we find that @xmath260 .
starting from this , one can straightforwardly show that @xmath261 { \bf o}.\ ] ] in turn , the dynamical solution in eq .
( [ cmsolt ] ) can be obtained by using this result and noticing that @xmath262 we now analyze some structural details of the blocks of the cm in eq .
( [ cmsssys ] ) , which with the help of ( [ matrixi ] ) becomes @xmath263 where @xmath264 is a @xmath265 hermitian matrix .
actually , we want to demonstrate that @xmath266 satisfies @xmath267 which also proves the assertion on eq .
( [ zerocond ] ) .
we start by defining the function @xmath268 with @xmath269 , in such a way that from eq .
( [ matrixi ] ) one can write @xmath270 the desired result , eq .
( [ result ] ) , follows from proper reorganization of the indexes involved in the summation , as outlined bellow .
first consider the case @xmath5 _ even_. in this situation , we rewrite eq .
( [ auxdef1 ] ) as @xmath271 & \end{aligned}\ ] ] with @xmath272 and @xmath273 . from the definitions in eq .
( [ matrixol ] ) , one can show straightforwardly that @xmath274 and eq .
( [ sum2 ] ) becomes @xmath275 , \end{aligned}\ ] ] where the indexes of the sum span the same set as in ( [ sum2 ] ) .
finally , if @xmath276 is odd ( even ) , @xmath277 is odd ( even ) , and the above sum is purely imaginary ( real ) , which proves the statement ( [ result ] ) for the case @xmath5 _
even_. the proof for @xmath5 _ odd _ follows the same steps but with ( [ sum1 ] ) rewritten as @xmath278 where @xmath279 is the sum in ( [ sum3 ] ) with the index running over the sets @xmath280 and @xmath281 . also , @xmath282 .
\nonumber \end{aligned}\ ] ] now , we should give a closer look to the terms in eq .
( [ sum4 ] ) . from ( [ auxdef1 ] ) and eq .
( [ matrixol ] ) , it is straightforward to show that @xmath283 and real for all @xmath284 . for @xmath285 odd ,
either @xmath286 or @xmath212 must be even , causing this function to be zero for any value of @xmath284 .
one can rewrite ( [ sum4.1 ] ) as @xmath288 . \\
\nonumber\end{aligned}\ ] ] notice that terms like @xmath289 appearing in ( [ sum4.1 ] ) gave rise to @xmath290 in ( [ term3 ] ) , since the contribuiting @xmath284 s are necessarilly odd according to ( [ term2 ] ) .
also , according to ( [ term2 ] ) , one can see that for @xmath276 odd ( even ) , the above sum is purely imaginary ( real ) . the conclusions from ( [ sum3 ] ) are also applied to @xmath279 in ( [ sum4 ] ) .
this completes the proof for @xmath5 _
odd_. 99 a. dhar , advances in physics * 57 * , 457 ( 2008 ) .
s. lepri , r. livi , and a. politi , phys . rep . * 377 * , 1 ( 2003 ) . h.c .
fogedby , and a. imparato , _ j. stat .
( 2012 ) p04005 .
fogedby , and a. imparato , _ j. stat
. mech . _ ( 2014 ) p11011 .
z. rieder , j.l .
lebowitz , and e. lieb , j. math .
* 8 * , 1073 ( 1967 ) .
landi , and m.j .
de oliveira phys .
e * 89 * , 022105 ( 2014 ) .
martinez , and j.p .
paz , phys .
rev . lett . * 110 * , 130406 ( 2013 ) .
n. freitas and j.p .
paz , phys .
e. * 90 * , 042128 ( 2014 ) .
a. asadian , d. manzano , m. tiersch , and h.j .
briegel , phys . rev .
e * 87 * , 012109 ( 2013 ) .
f. nicacio , r.n.p .
maia , f. toscano , and r.o .
vallejos , phys .
a * 374 * , 4385 ( 2010 ) .
horn , and c.r .
johnson , _ topics in matrix analysis _
( cambridge university press , new york , 1994 ) .
d. kulkarni , d. schmidt , and s .- k .
tsui , linear algebra and its applications * 297 * , 63 ( 1999 ) .
to avoid misunderstandings we stress that @xmath42 and @xmath291 are , respectively , a @xmath292 vector and a @xmath292 scalar of a set indexed by the labels of the @xmath292 lindblad operator . note that the sum on eq.([lindblad ] ) now encompasses the primed and the unprimed lindblad operators . | we address the problem of heat transport in a chain of coupled quantum harmonic oscillators , exposed to the influences of local environments of various nature , stressing the effects that the specific nature of the environment has on the phenomenology of the transport process .
we study in detail the behavior of thermodynamically relevant quantities such as heat currents and mean energies of the oscillators , establishing rigorous analytical conditions for the existence of a steady state , whose features we analyze carefully .
in particular , we assess the conditions that should be faced to recover trends reminiscent of the classical fourier law of heat conduction and highlight how such a possibility depends on the environment linked to our system .
understanding the transport properties in open systems in contact with several energy or particle baths represents a challenge for nonequilibrium physics . ideally , one would like to characterize and even calculate explicitly the statistics of the energy and particle currents , similarly to what can be done with observables in ensembles at equilibrium
. however the properties of the currents in out - of - equilibrium systems depends strongly on the bath properties , and on the characteristics of the system - bath coupling . in this context
, chains of oscillators have been extensively used as microscopic models for heat conduction and , in general , for out - of - equilibrium systems @xcite to investigate the behavior of the thermal conductivity for different interaction potentials between the oscillators , different bath properties , or different system - bath couplings @xcite .
the fourier law of heat conduction implies that the heat current @xmath0 flowing throughout a system under a temperature gradient scales as the inverse of the system size @xmath1 , _
i.e. _ , @xmath2 . in the classical case
, it is known that this law is violated in 1d homogeneous harmonic systems @xcite where heat is carried by freely propagating elastic waves , while the current scales with the system size in presence of anharmonicity or disorder either in mass or in the coupling constant .
although the transport is anomalous ( @xmath3 , with @xmath4 ) in these cases , the fourier law is finally restored only in presence of a external substrate potential @xcite or in the presence of a locally attached energy - conserving reservoirs for each oscillator @xcite .
quantum mechanically , a realistic description of a quantum medium for the transport of heat would imply the use of an explicitly open - system formalism and the introduction of system - environment interactions . in this context
, it is interesting to identify the conditions , if any , under which heat transport across a given quantum system can be framed into the paradigm of fourier law .
finding a satisfactory answer to this question is certainly not trivial , in particular in light of the ambiguities that the validity of fourier law has encountered even in the classical scenario @xcite . in the quantum scenario ,
significant studies are embodied in the work by martinez and paz @xcite , who show the emergence of the three laws of thermodynamics in arbitrary networks evolving under a quantum brownian master equation .
this has been applied in ref .
@xcite to show that the heat transport in this system is anomalous .
assadian _
et al . _
@xcite , on the other hand , have addressed a chain of oscillators described by a lindblad master equation , finding that a fourier - like dependence on the system size can be observed for very long harmonic chains in the presence of dephasing .
our work is also concerned with heat transfer in harmonic chains but the physical environment surrounding the chain is what differentiates our work from the ones previously cited .
basically , we include the possibility of establishing a temperature gradient using purely diffusive reservoirs , something not yet considered in the literature .
additionally , we further explore the effect of having the chain members locally attached to regular thermal baths at different temperatures . in this paper , we contribute to such research efforts by studying a general quadratic model for the dynamics of the system that , in turn , are affected by individual thermal reservoirs and exposed to the temperature gradient generated by all - diffusive environments .
our approach is able to pinpoint the origins of the specific energy distributions observed by varying the operating conditions of the system and thus identify the role played , respectively , by the diffusive and thermal reservoirs in the process of heat transport .
we find working configurations that deviate substantially from the expectations arising from fourier law and single out scenarios that are strictly adherent to such a paradigm , thus remarking the critical role played by the nature of the environment affecting the medium in the establishment of the actual mechanism for heat transport .
the remainder of this paper is organized as follows . in sec . [ tan ]
we introduce the formalism used to address the dynamics of the system . the general scenario addressed in our investigation
is described in sec .
[ tsaid ] , while the thermodynamic properties and phenomenology of heat currents are analyzed in sec .
[ aohcatc ] .
section [ atpc ] is devoted to the analysis of a few significant cases that help us addressing the deviations from ( and adherences to ) fourier law .
finally , sec .
[ conc ] is devoted to the conclusions . |
the measurement of the electric polarizability @xmath10 of an atom is a difficult experiment : this quantity can not be measured by spectroscopy , which can access only to polarizability differences , and one should rely either on macroscopic quantity measurements such as the electric permittivity ( or the index of refraction ) or on electric deflection of an atomic beam . for a review on polarizability measurements , we refer the reader to the book by kresin and bonin @xcite . for alkali atoms ,
all the accurate experiments were based on the deflection of an atomic beam by an inhomogeneous electric field and , in the case of lithium , the most accurate previous measurement was done in 1974 by bederson and co - workers @xcite , with the following result @xmath11 m@xmath2 .
however , in 2003 , amini and gould , using an atomic fountain @xcite , have measured the polarizability of cesium atom with a @xmath12% relative uncertainty , which is presently the smallest uncertainty on the electric polarisability of an alkali atom .
atom interferometry , which can measure any weak modification of the atom propagation , is perfectly adapted to measure the electric polarizability of an atom : this was demonstrated in 1995 by d. pritchard and co - workers @xcite with an experiment on sodium atom and they obtained a very high accuracy , with a statistical and systematic uncertainties both equal to @xmath13% .
this experiment was and remains difficult because an electric field must be applied on only one of the two interfering beams : one must use a capacitor with a thin electrode , a septum , which can be inserted between the two atomic beams . using our lithium atom interferometer @xcite ,
we have made an experiment very similar to the one of d. pritchard @xcite and we have measured the electric polarizability of lithium with a @xmath4% uncertainty , limited by the uncertainty on the mean atom velocity and not by the atom interferometric measurement itself @xcite . in the present paper
, we are going to describe in detail our experiment with emphasis on the improvements with respect to the experiments of d. pritchard s group @xcite : we have designed a capacitor with an analytically calculable electric field ; we have obtained a considerably larger phase sensitivity , thanks to a large atomic flux and an excellent fringe visibility ; finally our interferometer , which uses laser diffraction , is species selective : the contribution of any impurity ( heavier alkali atoms , lithium dimers ) to the signal can be neglected .
we may recall that several experiments using atom interferometers have exhibited a sensitivity to an applied electric field @xcite but these experiments were not aimed at an accurate measurement of the electric polarizability .
two other atom interferometry experiments @xcite using an inelastic diffraction process , so that the two interfering beams are not in the same internal state , have measured the difference of polarizability between these two internal states . finally , two experiments @xcite have measured the aharonov - casher phase @xcite : this phase , which results from the application of an electric field on an atom with an oriented magnetic moment , is proportional to the electric field .
this paper is organized as follows .
we briefly recall the principle of the experiment in part ii .
we then describe our electric capacitor in part iii and the experiment in part iv .
the analysis of the experimental data is done in part v and we discuss the polarizability result in part vi . a conclusion and two appendices complete the paper .
if we apply an electric field @xmath14 on an atom , the energy of its ground state decreases by the polarizability term : @xmath15 when an atom enters a region with a non vanishing electric field , its kinetic energy increases by @xmath16 and its wave vector @xmath17 becomes @xmath18 , with @xmath19 given by @xmath20 .
the resulting phase shift @xmath21 of the atomic wave is given by : @xmath22 where we have introduced the atom velocity @xmath23 and taken into account the spatial dependence of the electric field along the atomic path following the @xmath24-axis .
this phase shift is inversely proportional to the atom velocity and this dependence will be included in our analysis of the results .
the principle of the experiment , illustrated on figure [ interferometer ] , is to measure this phase shift by applying an electric field on one of the two interfering beams in an atom interferometer @xcite .
this is possible only if the two beams are spatially separated so that a septum can be inserted between the two beams .
this requirement could be suppressed by using an electric field with a gradient as in reference @xcite but it seems difficult to use this arrangement for a high accuracy measurement , because an accurate knowledge of the values of the field and of its gradient at the location of the atomic beams would be needed .
to make an accurate measurement , we must know precisely the electric field along the atomic path and guard electrodes are needed so that the length of the capacitor is well defined , as discussed by d. pritchard and co - workers @xcite .
it would probably be better to have guard electrodes on both electrodes , but it seems very difficult to draw guard electrodes on the septum and to put them in place very accurately .
therefore , as in reference @xcite , we have guard electrodes only on the massive electrodes .
however , we have chosen to put our guard electrodes in the plane of the high voltage electrode . with this choice
, the calculation of the electric field can be done analytically .
figure [ capacitor ] presents two schematic drawings of the capacitor and defines our notations , while an artist s view is presented in figure [ artistview ] .
like in reference @xcite , our capacitor is as symmetric as possible with respect to the septum plane , but , for a given experiment , only one half of the capacitor is used , the other part creating no electric field with @xmath25 everywhere .
if a dielectric slab with a permittivity @xmath26 is introduced in a plane capacitor , the field lines are distorted and concentrated towards the slab . because our capacitor contains dielectric spacers
, one could fear a similar effect but this effect does not exist when the dielectric slab fills completely the gap between the electrodes . following figure [ capacitor ] , the spacers , with a dielectric constant @xmath27 completely fill the space @xmath28 , with vacuum in the rest of capacitor , @xmath29 .
let @xmath30 ( respectively @xmath31 ) be the potential with @xmath32 ( @xmath28 ) . in the @xmath33 planes , the continuity equations give :
@xmath34 as shown below , a * 2d * solution of the laplace equation of the form @xmath35 exists for the capacitor .
then clearly , if we take @xmath36 , this solution fulfills both continuity equations on the dielectric borders in the @xmath33 planes and we know that the solution is unique .
we consider only one half ( with @xmath37 ) of the capacitor represented on figure [ capacitor ] .
we know the potential on the borders of the capacitor , with @xmath38 and @xmath39 if @xmath40 and @xmath41 if @xmath42 .
to get the potential everywhere , we start by calculating the fourier transform @xmath43 of @xmath44 : @xmath45 then , for a separable harmonic function of @xmath46 and @xmath24 , with a @xmath24-dependence of the @xmath47 form , its @xmath46-dependence is necessarily given by a linear combination of the two functions @xmath48 .
using this result and the conditions on the borders at @xmath49 and at @xmath50 , we get the value of @xmath35 everywhere : @xmath51 from which we can deduce the electric field everywhere . on the septum surface @xmath52 , the electric field is parallel to the @xmath46-axis and we can calculate exactly the integral of @xmath53 on this surface ( see appendix a ) .
we need the capacitor effective length @xmath54 which is defined by : @xmath55 where @xmath56 is the electric field of an infinite plane capacitor with the same electrode spacing @xmath57 . using equation ( [ a51 ] ) , we get the exact value of @xmath54 : @xmath58 \approx 2a - \frac{2h}{\pi}\ ] ] where exponentially small corrections of the order of @xmath59 $ ] have been neglected in the approximate result . however , the atoms do not sample the electric field on the septum surface but at a small distance , of the order of @xmath60 @xmath61 m in our experiment , and what we need is the integral of @xmath53 along their mean trajectory .
this mean trajectory is not exactly parallel to the septum , but it is easier to calculate the integral along a constant @xmath46 line .
we may use either the potential given by equation ( [ n5 ] ) or maxwell s equations to relate the field components near the septum surface to its value on the surface .
the calculation , given in appendix a , proves that the first correction to the effective length is proportional to @xmath62 : @xmath63 the @xmath62 correction is a fraction of the main term @xmath64 equal to @xmath65 and with our dimensions ( @xmath66 @xmath61 m , @xmath67 mm and @xmath68 mm ) , this correction is close to @xmath69 .
this correction is negligible at the present level of accuracy and we will use the value of @xmath54 given by the approximate form of equation ( [ n91 ] ) . more precisely , we will write : @xmath70\ ] ] let us describe how we build this capacitor .
the external electrodes are made of glass plates ( @xmath71 mm long in the @xmath24 direction , @xmath72 mm high in the @xmath73 direction , @xmath74 mm thick in the @xmath46 direction ) covered by an evaporated aluminium layer . to separate the guard electrodes from the high voltage electrode ,
a gap is made in the aluminium layer by laser evaporation @xcite .
we found that @xmath75 @xmath61 m wide gaps give a sufficient insulation under vacuum to operate the capacitor up to @xmath76 v. these gaps are separated by a distance @xmath77 mm , so that the two guard electrodes are @xmath78 mm long .
the glass spacers are @xmath79 mm thick plates of float glass ( @xmath74 mm @xmath80 @xmath71 mm ) used without further polishing .
the distance @xmath81 from the spacer inner edge to the atomic beam axis is equal to @xmath82 mm .
we have found that a float glass plate is flat within @xmath83 @xmath61 m over the needed surface .
this accuracy appeared to be sufficient for a first construction , as the main geometrical defects are due to the way we assemble the various parts by gluing them together and by an imperfect stretching of the septum .
we use a double - faced tape arclad 7418 ( from adhesive research ) to assemble the spacers on the external electrodes .
the septum is made of @xmath84 @xmath61 m thick mylar from goodfellow covered with aluminium on both faces . in a first step , the mylar sheet is glued on circular metal support .
it is then covered by a thin layer of dish soap diluted in water and the mylar is heated near @xmath85c with a hot air gun .
then , we clean the mylar surface with water and let it dry .
after this operation , the mylar is well stretched and its surface is very flat . we have measured the resonance frequencies of the drum thus formed , from which we deduced a surface tension of the order of @xmath60 n / m ( this value is only indicative as this experiment was made with another mylar film which was @xmath86 @xmath61 m thick ) . once stretched , the mylar film is glued on one electrode - spacer assembly with an epoxy glue epotek 301 ( from epoxy technology ) , chosen for its very low viscosity , and then it is cut with a scalpel . in a final step ,
a second electrode - spacer assembly is glued on the other face of the mylar . finally ,
as shown in figure [ artistview ] , wires are connected to the various electrodes using an electrically conductive adhesive epotek ee129 - 4 ( also from epoxy technology ) .
we are going to discuss the various points by which the real capacitor differs from our model .
we have shown that the potential @xmath30 is reduced to a * 2d * function when an homogeneous dielectric slab fills completely the gap between the electrodes , with the border between the vacuum and the slab being a plane perpendicular to the electrodes .
the real dielectric slab is the superposition of a tape , a glass spacer and a glue film , each material having a different permittivity @xmath87 .
the differences in permittivity perturb the potential which should take a * 3d * character extending on a distance comparable to the tape or glue film thicknesses .
this perturbation seems negligible , because the tape and the glue films are very thin and also because these three dielectric materials have not very large @xmath87 values .
our calculation assumes that the potential is known everywhere on the border .
but , on the high - voltage electrode , we may fear that the potential is not well defined in the @xmath75 @xmath61 m wide dielectric gaps separating the high voltage and guard electrodes as these gaps might get charged in an uncontrolled way .
this is not likely if the volume resistivity @xmath88 of the pyrex glass used is not too large : more precisely , the time constant for the charge equilibration on the gap surface is given by @xmath89 ( within numerical factors of the order of one ) and this time constant remains below @xmath90 second if @xmath91 @xmath92.m .
we have found several values of the resistivity of pyrex glass at ordinary temperature , in the range @xmath93 @xmath92.m and with such a conductivity , this time constant is below @xmath94 s. our calculation neglects the surface conductivity , due to the adsorbed impurities , which should further reduce this time constant .
therefore , we think that it is an excellent approximation to assume that the potential @xmath30 makes a smooth transition from @xmath25 to @xmath95 in the gaps . then , using equation ( [ a3 ] ) , it is clear that the detailed shape of the transition has no consequence as these details are smoothed out by the convolution of @xmath96 by the function @xmath97 which has a full width at half maximum equal to @xmath98 .
we can use equation ( [ n91 ] ) to calculate the effective length , provided that we add to the length of the high voltage electrode the mean width of the two gaps . in the present work
, we have taken one gap width , @xmath75 @xmath61 m , as a conservative error bar on the effective length .
a superiority of our capacitor design is that these gaps are very narrow , thus minimizing the corresponding uncertainty on the capacitor effective length and we hope to be able to further reduce this uncertainty .
the thickness @xmath57 of the capacitor , which is the sum of the thickness of the spacers , the tape and the glue film , is not perfectly constant . using a mitutoyo litematic machine ,
we have measured , with a @xmath99 @xmath61 m uncertainty , the capacitor thickness as a function of @xmath24 , in the center line of the two spacers , at @xmath100 mm .
the average of these two measurements gives the thickness @xmath101 in the @xmath102 plane around which the atom sample the electric field .
the thickness @xmath57 is not perfectly constant but it is well represented by a linear function of @xmath24 , given by @xmath103 , the maximum deviation @xmath104 being considerably smaller than the mean @xmath57 value noted @xmath105 .
as these deviations are very small ( see below ) , it seems reasonable to use use equation ( [ n14 ] ) provided that terms involving powers of @xmath57 are replaced by their correct averages .
the first term in @xmath106 corresponds to the integral of @xmath53 over the capacitor length , from @xmath107 to @xmath108 and we must take the average value of @xmath106 over this region . neglecting higher order terms ,
this average is given by : @xmath109\ ] ] in equation ( [ n14 ] ) , the second term in @xmath110 , corresponds to end effects and this quantity must be replaced by the following two - point average : @xmath111\ ] ] both corrections involve the same factor @xmath112 $ ] .
although the capacitor is as symmetric as possible , this symmetry is only approximate and we give the parameters for the half we have used for the set of measurements described below .
the length of the high voltage electrode , including one gap width is @xmath113 mm , the error bar being taken equal to one gap width , as discussed above .
the distance @xmath57 between the electrodes gap width is described by @xmath114 mm and @xmath115 mm .
the correction term @xmath116 is completely negligible .
in this part , we are going to recall the main features of our lithium atom interferometer , to give the values of various parameters used for the present study , to present the data acquisition procedure and the way we extract the phases from the data .
our atom interferometer is a mach - zehnder interferometer using bragg diffraction on laser standing waves .
its design is inspired by the sodium interferometer of d. pritchard and co - workers @xcite and by the metastable neon interferometer of siu au lee and co - workers @xcite .
a complete description has been published @xcite .
the lithium atomic beam is a supersonic beam seeded in argon and , for the present experiment , we have worked with a low argon pressure in the oven @xmath117 mbar , because the detected lithium signal increases when the argon pressure decreases .
the oven body temperature is equal to @xmath118 k , fixing the vapor pressure of lithium at @xmath119 mbar and the nozzle temperature is equal to @xmath120 k. with these source conditions , following our detailed analysis @xcite , the argon and lithium velocity distributions are described by a parallel speed ratio for argon equal to @xmath121 and a parallel speed ratio for lithium equal to @xmath122 ( the parallel speed ratio is defined by equation ( [ m4 ] ) below ) .
we use bragg diffraction on laser standing waves at @xmath123 nm : the laser is detuned by about @xmath124 ghz on the blue side of the @xmath125 - @xmath126 transition of the @xmath0li isotope , the signal is almost purely due to this isotope , which has a natural abundance equal to @xmath127% , and not to the other isotope @xmath128 .
moreover , any other species present in the beam , for instance heavier alkali atoms or lithium dimers , is not diffracted and does not contribute to the signal .
the case of lithium dimers deserves a special discussion because they are surely present in the beam and the lithium dimer has an absorption band system due its @xmath129 transition with many lines around @xmath130 nm : for most rovibrational levels of the @xmath131 state , the absorption transition which is closest to the laser frequency has a very large detuning of the order of hundreds to thousands of ghz and the intensity of this resonance transition is also weaker than the resonance transition of lithium atom , because of the franck - condon factor .
therefore , the lithium dimers have a negligible probability of diffraction and do not contribute to the interferometer signals .
the interference signals in a mach - zehnder interferometer are given by : @xmath132\ ] ] where the phase @xmath133 of the interference fringes can be written : @xmath134 the first term of @xmath133 is particular to three - grating interferometers : the diffracted beam of order @xmath135 by grating @xmath136 has a phase dependent on the grating position @xmath137 . in our case of laser diffraction ,
the grating position @xmath137 is given by the mirror position @xmath138 and @xmath139 is the grating wavevector , where @xmath140 is the laser wavevector .
this phase term is very interesting because it is non dispersive and it is commonly used to observe interference fringes . in our case , we scan the position @xmath141 of mirror @xmath142 by a piezoelectric translation stage .
the second term @xmath21 represent any phase difference between the two beams and in particular , it will represent the phase shift due to the application of an electric field on one of the two paths . in the present work ,
we have only used the diffraction order @xmath143 so that the center of the two beams are separated by about @xmath144 @xmath61 m at the location of the capacitor , which is located just before the second laser standing wave .
the capacitor is attached to the top of the vacuum chamber and not to the rail supporting the laser standing wave mirrors @xmath138 : in this way , we do not increase the vibrations of the mirror positions @xmath137 the capacitor is held by a translation stage along the @xmath46-direction , which can be adjusted manually thanks to a vacuum feedthrough and a double stage kinematic mount built in our laboratory .
the first stage , operated with screws , can be used only when the experiment is at atmospheric pressure while the second stage , actuated by low - voltage piezo - translators , can be adjusted under vacuum . when the septum is inserted between the two atomic paths ,
the atom propagation is almost not affected by its presence and , as shown in figure [ fringes84 ] , we have observed a fringe visibility equal @xmath145 % and a negligible reduction of the atomic flux . to optimize the phase sensitivity ( see a discussion in reference @xcite ) , we have opened the collimation slit @xmath146 and the detection slit @xmath147 ( see reference @xcite ) with widths @xmath148 @xmath61 m and @xmath149 @xmath61 m , thus increasing the mean flux up to @xmath150 counts / s and slightly reducing the fringe visibility down to @xmath151% ( see figure [ signals ] ) .
we have made a series of recordings , labelled by an index @xmath136 from @xmath90 to @xmath152 , with @xmath153 when @xmath136 is odd , and with @xmath154 , when @xmath136 is even with @xmath155 volts .
for each recording , we apply the same linear ramp on the piezo - drive of mirror @xmath142 in order to observe interference fringes and @xmath156 data points are recorded with a counting time per channel equal to @xmath157 s. figure [ signals ] presents a pair of consecutive recordings .
the high voltage power supply has stability close to @xmath158 and the applied voltage is measured by a hp model @xmath159 voltmeter with a relative accuracy better than @xmath160 . for each recording ,
the data points @xmath161 have been fitted by a function : @xmath162 \nonumber \\
\mbox{with } \psi_i ( n)&= & a_i + b_i n
+ c_i n^2\end{aligned}\ ] ] where @xmath163 labels the channel number , @xmath164 represents the initial phase of the pattern , @xmath165 an ideal linear ramp and @xmath166 the non - linearity of the piezo - drive . for the @xmath153 recordings , @xmath164 , @xmath165 and @xmath166
have been adjusted as well as the mean intensity @xmath167 and the visibility @xmath168 . for the @xmath169 recording ,
we have fitted only @xmath164 , @xmath167 and @xmath168 , while fixing @xmath165 and @xmath166 to their value @xmath170 and @xmath171 from the previous @xmath25 recording .
we think that our best phase measurements are given by the mean phase @xmath172 obtained by averaging @xmath173 over the @xmath156 channels .
the @xmath174 error bar of these mean phases are of the order of @xmath175 mrad , increasing with the applied voltage up to @xmath176 mrad because the visibility is considerably lower when the applied voltage @xmath177 is large ( see figure [ visibility ] ) .
this rapid decrease of the visibility is due to the velocity dependence of the phase and to the velocity distribution of the lithium atoms .
the mean phase values @xmath172 values of the @xmath153 recordings are plotted in figure [ phasezero ] : they present a drift equal to @xmath178 mrad / minute and some scatter around this regular drift .
the most natural explanation for this drift is a change of the phase @xmath133 resulting from a variation of the mirror positions @xmath137 : @xmath133 changes by @xmath90 radian for a variation of @xmath179 equal to @xmath180 nm .
we have verified that the observed drift has the right order of magnitude to be due to the differential thermal expansion of the structure supporting the three mirrors : its temperature was steadily drifting at @xmath181 k / minute during the experiment and the support of mirror @xmath142 differs from the other supports , as it includes a piezo translation stage , which is replaced by aluminium alloy for the mirrors @xmath182 and @xmath183 .
presently , we have no explanation of the phase scatter , which presents a quasi - periodic structure as a function of time : its rms value is equal to @xmath184 milliradian and , unfortunately , this scatter gives the dominant contribution to our phase uncertainty .
the phase shift @xmath185 due to the polarizability effect ( the average @xmath186 recalls that our experiment makes an average over the velocity distribution , as discussed below ) is taken equal to @xmath187 where the recording @xmath136 corresponds to the applied voltage @xmath177 : the average of the mean phase of the two @xmath153 recordings done just before and after is our best estimator of the mean phase of the interference signal in zero field . in figure
[ phaseshifts ] , we have plotted the phase shift @xmath188 as a function of the applied voltage @xmath177 .
we have chosen for the error bar on @xmath188 the quadratic sum of the @xmath174 error bar given by the fit of @xmath189 and the @xmath184 milliradians rms deviation of the @xmath153 phase measurements .
to interpret the experimental data , we must take into account the velocity distribution of the lithium atoms .
we assume that this velocity distribution is given by : @xmath190\ ] ] where @xmath191 is the most probable velocity and @xmath192 is the parallel speed ratio .
with respect to the usual form of the velocity distribution for supersonic beams , we have omitted a @xmath193 factor which is traditionally introduced @xcite but when the parallel speed ratio @xmath192 is large enough , this @xmath193 has little effect and the main consequence of its omission is a slight modification of the values of @xmath191 and @xmath192 . here , what we consider is the atoms contributing to the interferometer signal and their velocity distribution may differ slightly from the velocity of the incident beam , as bragg diffraction is velocity selective . the experimental signal can be written : @xmath194 \nonumber \\ & = & i_0 \left[1 + \left<{\mathcal{v}}\right > \cos\left ( \psi + \left<\phi\right > \right ) \right]\end{aligned}\ ] ] where @xmath195 is the value of the phase @xmath196 for the velocity @xmath197 . if we introduce @xmath198 and expand @xmath199 in powers of @xmath200 up to second order , the integral can be taken exactly , as discussed in appendix b. however , the accuracy of this approximation is not good enough when @xmath201 is large and we have used direct numerical integration to fit our data .
using equations ( [ m4 ] ) and ( [ m5 ] ) , we have fitted the measured phase @xmath196 and visibility @xmath202 as a function of the applied voltage @xmath30 .
the phase measurements received a weight inversely proportional to the square of their estimated uncertainty and we have adjusted two parameters , the value of @xmath203 and the parallel speed ratio @xmath192 .
the results of the fits are presented in figures [ phaseshifts ] and [ visibility ] .
the agreement is excellent , in particular for the phase data , and we deduce a very accurate value of @xmath204 : @xmath205 where the error bar is equal to @xmath174 .
the relative uncertainty on @xmath206 is very small , @xmath207% , which proves the quality of our phase measurements .
we also get an accurate determination of the parallel speed ratio : @xmath208 this value of the parallel speed ratio is larger than the predicted value for our lithium beam , @xmath209 ( see above ) and this difference can be explained by the velocity selective character of bragg diffraction .
the lithium electric polarizability is related to the value of @xmath210 by : @xmath211 \left[1 + \frac{h_1 ^ 2}{h_0 ^ 2}\right]\ ] ] all the geometrical parameters @xmath212 , @xmath105 and @xmath104 describing the capacitor are known with a good accuracy and we deduce a still very accurate value of the ratio of the electric polarizability @xmath10 divided by the mean atom velocity @xmath191 : @xmath213 we have measured the mean atom velocity @xmath191 by various techniques we have made measurements by doppler effect , measured either on the laser induced fluorescence signals or on the intensity of the atomic beam which is reduced by atomic deflection due to photon recoil . in the first case , the laser was making an angle close to @xmath214 with the atomic beam and it is difficult to measure this angle with sufficient accuracy . in the second case , we have used a laser beam almost contra - propagating with the atoms , so that the uncertainty on the cosine of the angle is negligible .
the signal appears as an intensity loss on the atomic beam and , the loss is not very large because we have used only one laser , so that the atoms are rapidly pumped in the other hyperfine state . the experimental signal is shown in figure [ atomrecoil ] . from a fit of this data
, we get a value of the mean velocity @xmath215 m / s .
we have also recorded the diffraction probability as a function of the bragg angle , by tilting the mirror forming a standing wave .
this experiment is similar to the one described inour paper @xcite ( see figure 3 ) but it is made with a lower power density so that only the first order diffraction appears .
the diffraction is detected by measuring the intensity of the zero - order atomic beam , as shown in figure [ diffraction ] . using an independent calibration of the mirror rotation as a function of the applied voltage on the piezo - actuator
, we get a measurement of the bragg angle @xmath216 @xmath61rad corresponding to @xmath217 m / s .
these two values are very coherent and we may combine them to give our best estimate of the most probable velocity @xmath191 : @xmath218 we can compare this measurement with the theoretical prediction for supersonic expansions . for a pure argon beam , in the limit of an infinite parallel speed ratio , theory predicts @xmath219 m / s where @xmath220 is the nozzle temperature and @xmath221 the argon atomic mass . with @xmath222 k , we get @xmath223 m / s .
three small corrections must be made .
we must correct for the finite value of the argon parallel speed ratio @xmath224 , estimated to be @xmath225 , using the semi - empirical relation of beijerinck and verster @xcite and the associated correction reduces the most probable velocity @xmath191 by a fraction equal to @xmath226% ( this correction is calculated in the limit of a vanishing perpendicular temperature @xcite ) .
we must replace the argon atomic mass by a weighted mean of the lithium and argon atomic masses and , with @xmath227 mbar of lithium in @xmath228 mbar of argon , this correction increases the velocity by @xmath229% .
finally , we must take into account the velocity slip effect : the light lithium atoms go slightly faster than the argon atoms .
this difference has been calculated by numerical simulation by p. a. skovorodko @xcite and this quantity is expected to scale like @xmath230 , so that the correction in our case is estimated to be @xmath231% .
we thus predict a most probable velocity @xmath232 m / s , where the uncertainty comes solely from the uncertainty on the temperature @xmath220 .
this value is in satisfactory agreement with our measurements . using the measured value of the most probable velocity @xmath191
, we get the lithium electric polarizability of @xmath0li : @xmath233 the final uncertainty bar is equal to @xmath4% , resulting from the quadratic sum of the @xmath234% uncertainty on the most probable velocity @xmath191 , the @xmath235% uncertainty on the effective length of the capacitor , the @xmath236% uncertainty due to the capacitor spacing @xmath105 and the @xmath237% uncertainty on the interferometric measurement .
unexpectedly , the atom interferometry result has the smallest uncertainty !
our measurement is a mean of the polarizability of the two hyperfine sublevels @xmath238 and @xmath239 of @xmath0li .
these two levels have not exactly the same polarizability .
the difference @xmath240 has been measured with great accuracy by mowat @xcite .
we can express this result as a fraction of the mean polarisability , @xmath241 .
this difference is fully negligible with our present accuracy .
our result is in excellent agreement with those of the previous measurements of @xmath10 which are considered as being reliable .
the 1934 measurement of scheffers and stark @xcite , which gave @xmath242 m@xmath2 , is generally considered to be incorrect .
the first reliable measurement is due to bederson and co - workers @xcite in 1961 , who obtained @xmath243 m@xmath2 , by using the e - h gradient balance method . in 1963 ,
chamberlain and zorn @xcite obtained @xmath244 m@xmath2 by measuring the deflection of an atomic beam .
finally , in 1974 , a second experiment was done by bederson and co - workers @xcite , using the same e - h gradient balance method improved by the calibration on the polarizability of helium atom in the @xmath245 metastable state , and they obtained the value @xmath11 m@xmath2 .
we may also compare our measurement with theoretical results .
many calculations of @xmath10 have been published and one can find a very complete review with 35 quotations in table 17 of the paper of king @xcite published in 1997 . here is a very brief discussion of the most important results : * the first calculation of @xmath10 was made in 1959 by dalgarno and kingston @xcite and gave @xmath246 a.u .. this calculation was rapidly followed by several other works . * if we forget hartree - fock results near @xmath247 a.u .
, a large majority of the published values are in the @xmath248 a.u . range . * in 1994 , kassimi and thakkar @xcite have made a detailed study with two important results .
they obtained a fully converged hartree - fock value , @xmath249 a.u . and
this result , far from the experimental values , proves the importance of electron correlation .
they also made a series of nth - order mller - plesset calculations with @xmath250 , @xmath124 and @xmath251 , from which they extract their best estimate with an error bar , @xmath252 a.u .. * in 1996 , yan et al .
@xcite have made an hylleraas calculation , with the final value @xmath253 a.u .
, this value and its error bar resulting from a convergence study .
our result is extremely close to these two very accurate calculations of @xmath10 .
two minor effects have not been taken into account in these calculations , namely relativistic correction and finite nuclear mass correction , but these two effects are quite small .
the relativistic correction on the polarizability has been studied by lim et al .
they have made different calculations ( hartree - fock , second order mller - plesset , coupled cluster ccsd and ccsd(t ) ) and in all cases , their relativistic result is lower than the non - relativistic result with a difference in the @xmath254 a.u . range .
we do not quote here their results : even the hartree - fock value is lower by @xmath255 a.u .
than the one of reference @xcite , because the chosen basis set is too small .
as far as we know , no calculation of @xmath10 has been made taking into account the finite nuclear mass .
an order of magnitude of the associated correction should be given by the hydrogenic approximation : @xmath256 ^ 3 $ ] and in this approximation , the polarisability of @xmath0li should be larger than its @xmath257 value by @xmath258 a.u .
but one should not expect this approximation to predict even the sign of the correction .
a high accuracy calculation of the finite mass effect is surely feasible , following the hylleraas calculations of yan and drake , who have already evaluated the finite mass effect on some energies @xcite and oscillator strengths @xcite of lithium atom .
we have made a measurement of the electric polarizability of lithium atom @xmath0li by atom interferometry and we have obtained @xmath259 a.u . with a @xmath4% uncertainty .
our measurement is in excellent agreement with the most accurate experimental value obtained by bederson and coworkers @xcite in 1974 and we have reduced the uncertainty by a factor three .
our result is also in excellent agreement with the best theoretical estimates of this quantity due to kassimi and thakkar @xcite and to yan et al . @xcite .
the neglected corrections ( relativistic effect , finite nuclear mass effect ) should be at least ten times smaller than our present error bar .
our measurement is the second measurement of an electric polarizability by atom interferometry , the previous experiment being done on sodium atom by d. pritchard and co - workers @xcite in 1995 ( see also @xcite and @xcite ) .
this long delay is explained by the difficulty of running an atom interferometer with spatially separated beams . using a similar experiment , j. p. toennies and co -
workers have compared the polarizabilities of helium atom and helium dimer but this work is still unpublished @xcite .
we want now to insist on the improvements we have done with respect to the other measurement of an electric polarizability by atom interferometry , due to d. pritchard and co - workers @xcite : * the design of our capacitor permits an analytical calculation of the @xmath53 integral along the atomic path .
this property is important for a better understanding of the influence of small geometrical defects of the real capacitor . in the present experiment ,
the uncertainty on the @xmath53 integral is equal to @xmath157% and we think that it is possible to reduce this uncertainty near @xmath260 % with an improved construction . * we have obtained a very good phase sensitivity of our atom interferometer : from our recordings , we estimate this phase sensitivity near @xmath5 mrad/@xmath6 .
the accuracy achieved on phase measurement has been limited by the lack of reproducibility of the phase between consecutive recordings .
we will stabilize the temperature of the rail supporting the three mirrors , hoping thus to improve the phase stability .
even if we have not been able to fully use our phase sensitivity , we have obtained set of phase - shifts measurements exhibiting an excellent consistency and accuracy , as shown by the quality of the fit of figure [ phaseshifts ] and by the accuracy , @xmath261% , of the measurement of the quantity @xmath262 . * in his thesis @xcite , t. d. roberts reanalyzes the measurement of the electric polarizability of sodium atom made by c. r. ekstrom et al .
@xcite : he estimates that a weak contribution of sodium dimers to the interference signals can be present as material gratings diffract sodium dimers as well as sodium atoms and he estimates that , in the worst case , this molecular signal might have introduced a systematic error as large as @xmath263% on the sodium polarizability result .
our interferometer is species selective thanks to the use of laser diffraction and this type of error does not exist in our experiment .
only lithium atoms are diffracted and even , with our choice of laser wavelength , only the @xmath0li isotope contributes to the signal .
the main limitation on the present measurement of the electric polarisability of lithium @xmath0li comes from the uncertainty on the most probable atom velocity @xmath191 . with some exceptions , like the aharonov - casher phase shift @xcite which is independent of the atom velocity ,
a phase - shift induced by a perturbation is inversely proportional to the atom velocity , at least when a static perturbation is applied on one interfering beam .
this is a fundamental property of atom interferometry and clever techniques are needed to overcome this difficulty : * t. d. roberts et al .
@xcite have developed a way of correcting the velocity dependence of the phase shift by adding another phase shift with an opposite velocity dependence .
they were thus able to observe fringes with a good visibility up to very large phase shift values . *
our present results prove that a very accurate measurement can be made in the presence of an important velocity dispersion without any compensation of the associated phase dispersion , but by taking into account the velocity distribution in the data analysis . in these two cases
, one must know very accurately a velocity , the most probable velocity in our case and the velocity for which the correction phase cancels in the case of reference @xcite
. the uncertainty on this velocity may finally be the limiting factor for high precision measurements .
obviously other techniques can be used to solve this difficulty .
finally , we think that our experiment illustrates well two very important properties of atom interferometry : * the sensitivity of atom interferometry is a natural consequence of the well - known sensitivity of interferometry in general , which is further enhanced in the case of atom by the extremely small value of the de broglie wavelength .
our phase measurement is in fact a direct mesaurement of the increase @xmath7 of the atom velocity @xmath8 when entering the electric field .
@xmath7 is very simply related to the observed phase shift : + @xmath264 + this variation is extremely small , with @xmath265 for the largest electric field used in this experiment , corresponding to @xmath266 rad .
our ultimate sensitivity corresponds to a phase @xmath267 milliradian which means that we can detect a variation @xmath9 , whereas the velocity distribution has a fwhm width equal to @xmath268% ! * when an atom propagates in the capacitor placed in our atom interferometer , its wavefunction samples two regions of space separated by a distance @xmath269 @xmath61 m with a macroscopic object , the septum , lying in between and this situation extends over @xmath270 microsecond duration , without inducing any loss of coherence .
this consequence of quantum mechanics remains very fascinating !
we thank cnrs spm and rgion midi pyrnes for financial support .
we thank p. a. skovorodko and j. l. heully for helpful information .
the first step is to calculate the fourier transform @xmath43 of @xmath271 defined by equation ( [ n4 ] ) : @xmath272 the potential @xmath35 given by equation ( [ n5 ] ) can then be calculated and , from @xmath35 , we can deduce the electric field everywhere and in particular , on the septum surface where it is parallel to the @xmath46-axis : @xmath273 with @xmath274 .
the electric field @xmath275 is given by the inverse fourier transform of the product of two functions @xmath43 and @xmath276 .
therefore , the field @xmath277 is the convolution of their inverse fourier transforms which are @xmath44 and @xmath97 : @xmath278 using reference @xcite ( equation 6 of paragraph 4.111 , p. 511 ) , we get an explicit form of @xmath97 : @xmath279 this result proves that the field decreases asymptotically like @xmath280 , when @xmath281 .
we can also get a closed form expression of the electric field @xmath277 but we may get the integral of @xmath53 without this result , simply by using the parseval - plancherel theorem and reference @xcite ( equation 4 of paragraph 3.986 , p. 506 ) : @xmath282\ ] ] the atoms sample the electric field at a small distance of the septum surface and we need the integral of @xmath53 along their mean trajectory .
this mean trajectory is not parallel to the septum , but it is easier to calculate this integral along a constant @xmath46 line . from maxwell
s equations , one gets the first correction terms to the field when @xmath46 does not vanish : @xmath283 where the derivatives are calculated for @xmath49 . after an integration by parts ,
one gets : @xmath284 dz \nonumber \\ & = & \int_{-\infty}^{+\infty}\left [ e_{x}^2(x=0,z ) + 2 x^2 \left[\frac{\partial e_x}{\partial z}\right]^2\right ] dz\end{aligned}\ ] ] where we have kept only the first non vanishing correction term in @xmath62 .
the calculation of the integral is also done with the parseval - plancherel theorem and , after some algebra , we get : @xmath285 ^ 2 dz & = & x^2 \frac{v_0 ^ 2}{h^3 } \left[\frac{2}{3 } - \frac{\frac{\pi a}{h}\coth\left(\frac{\pi a}{h } \right ) - 1 } { \sinh^2\left(\frac{\pi a}{h}\right)}\right ] \nonumber \\ & \approx & x^2 \frac { 2 v_0 ^ 2}{3 h^3}\end{aligned}\ ] ] where the approximate result is obtained by neglecting the exponentially small terms of the order of @xmath286 $ ] .
we want to calculate : @xmath290 \\
\label{b42 } \left<{\mathcal{v}}\right > & = & { \mathcal{v}}_0 \frac{s_{\|}}{\left[s_{\|}^4 + \phi_m^2\right]^{1/4 } } \exp\left[-\frac{\phi_m^2 s_{\|}^2}{4\left ( s_{\|}^4 + \phi_m^2\right ) } \right ] \\
\label{b43 } \left<\phi \right > & = & \phi_m + \frac{1}{2 } \arctan\left[\frac{\phi_m } { s_{\|}^2 } \right ] - \frac{\phi_m^3 } { 4\left ( s_{\|}^4 + \phi_m^2\right)}\end{aligned}\ ] ] we have tested this approximation by comparing this approximate formula with the result of a computer program , for a parallel speed ratio @xmath291 , corresponding to our experimental case . as shown on figure [ visibilityaverage ] , the agreement is very satisfactory , at least with our present accuracy on visibility values , with differences of the order of @xmath90% .
we have also studied the difference @xmath292 as a function of @xmath201 and the results are presented in figure [ phaseaverage ] .
this difference can reach large values , for instance @xmath293 rad when @xmath294 rad .
one may remark that , as obvious on equation ( [ b43 ] ) , the difference @xmath295 is linear in @xmath201 and positive , when @xmath201 is small , and becomes negative and roughly cubic in @xmath201 for larger @xmath201 values .
if the parallel speed ratio @xmath192 is large , @xmath296 , as long as the linear term is dominant , the velocity averaged phase is given by : and not by @xmath201 . with @xmath291 , the approximate and numerical results are almost equal as long as @xmath298 , but their difference increases rapidly for larger @xmath201 values , being close to @xmath299 radian when @xmath300 : even if this difference is a very small fraction of @xmath201 , this difference is not fully negligible and we have decided not to use the approximate analytical results ( [ b42 ] ) and ( [ b43 ] ) to fit the data .
j. schmiedmayer , m. s. chapman , c. r. ekstrom , t. d. hammond , d. a. kokorowski , a. lenef , r. a. rubinstein , e. t. smith and d. e. pritchard , in atom interferometry edited by p. r. berman ( academic press 1997 ) , p 1 | using an atom interferometer , we have measured the static electric polarizability of @xmath0li @xmath1 m@xmath2 @xmath3 atomic units with a @xmath4% uncertainty .
our experiment , which is similar to an experiment done on sodium in 1995 by d. pritchard and co - workers , consists in applying an electric field on one of the two interfering beams and measuring the resulting phase - shift . with respect to d. pritchard s experiment ,
we have made several improvements which are described in detail in this paper : the capacitor design is such that the electric field can be calculated analytically ; the phase sensitivity of our interferometer is substantially better , near @xmath5 mrad/@xmath6 ; finally our interferometer is species selective so that impurities present in our atomic beam ( other alkali atoms or lithium dimers ) do not perturb our measurement .
the extreme sensitivity of atom interferometry is well illustrated by our experiment : our measurement amounts to measuring a slight increase @xmath7 of the atom velocity @xmath8 when it enters the electric field region and our present sensitivity is sufficient to detect a variation @xmath9 . |
holmberg ii ( hereafter abbreviated as hoii ) is one of the most famous examples of a dwarf galaxy exhibiting a violent , disrupted ism , as traced by its appearance in the 21cm hi line @xcite .
hi holes are present across the entire galaxy even well beyond hoii s stellar body .
@xcite compiled a catalog of individual hi shells .
the measured radial expansion velocities ( 10 - 25@xmath2 ) and diameters ( 1002000pc ) indicate that most of the holes are still expanding and have ages of @xmath3 to @xmath4yr .
hoii has been studied at many other wavelengths as well , such as radio continuum wavelengths @xcite , optical @xcite and fuv @xcite . besides
, it has been the topic of studies to determine its mass distribution and dark matter content @xcite and a popular galaxy to compare numerical simulations / models of expanding hi shells against @xcite .
recent hi observations , including those of hoii , show that hi holes are older and reach larger sizes in dwarf galaxies as compared to similar structures in more massive , spiral galaxies .
some of this can be attributed to the fact that dwarf galaxies have shallower gravitational potentials which leads , for the same observed hi velocity dispersion , to a puffed up disk .
hence , hi holes can grow to larger sizes before breaking out of the disk .
moreover , dwarf galaxies lack in general differential rotation or spiral density waves ( e.g. , @xcite ) . in other words , once formed , structures like hi holes in the ism are not strongly modified and/or destroyed by large scale streaming motions . the most straightforward way to interpret
the presence of the hi cavities is to attribute them to past stellar activity within their centers ( the standard picture ) : strong winds of o and b stars can create individual bubbles of pc size whereas subsequent supernova events create cavities of a few tens to hundreds of parsecs extent . in this picture ,
the expansion of the hi holes is powered by the overpressure created within the coronal gas resulting from the sn explosions , e.g. @xcite .
although the standard picture of the creation of the hi holes is widely accepted , it is not without its critics .
for example @xcite , analyzed broad band bvr ( 4@xmath5 detection limit : b=23mag ) and narrow band h@xmath6 images to search for star clusters in the centers of the hi holes in hoii .
based on the results derived by @xcite for the age and energy needed to create the hi hole , and assuming that the stellar cluster remained a coherent structure during its evolution and using conventional initial mass functions ( imfs ) , they argue that these star clusters should still be detectable . as their observations revealed far less remnant clusters than expected , @xcite ruled out clusters as the origin for the hi holes in all but 6 out of 48 cases .
certainly especially the giant holes outside the optical disk of hoii are difficult to explain since star formation currently does nt seem to play a dominating role at large galactocentric radii of hoii .
x ray observations of dwarf galaxies can be used to test the validity of the standard picture .
coronal gas within the interiors of the hi holes might be detectable in x
rays , whereas objects related to the end points of stellar evolution such as x - ray binaries are expected to show up as hard x - ray point sources .
low mass x - ray binaries are prominent x - ray sources detectable even at the distance of hoii . in the optical regime
however , these objects are very faint and difficult to detect even with large telescopes . x ray observations seem the best way forward now to further check the validity of the standard picture .
@xcite as well as @xcite have reported data on the strongest x ray source in hoii only . here , we perform an analysis of all sources related to hoii within the entire field observed with the pspc .
sec . 2 deals with the data reduction and analysis of three pointed pspc observations of hoii which we retrieved for that purpose .
3 we introduce the x ray hardness ratio diagram as a tool to classify the x ray spectrum for faint sources .
our results are presented in sec . 4 , which are then discussed in sec . 5 .
we summarize our conclusions in sec .
lrr 600140p & 04/92 & 7258 + 600431p & 10/92 & 11607 + 600431p-1 & 03/93 & 3701 + we retrieved 3 pointed pspc observations towards hoii from the archive ( see table [ obstable ] for details ) .
the pspc data were analyzed using the exsas software package provided by the `` max planck institut fr extraterrestrische physik '' in garching @xcite .
all three observations were merged into a single photon event file and re centered on the mean position : @xmath7 .
the net integration time of the merged pspc data is 22566 sec , making hoii one of deepest studied dwarf galaxies by .
the photon events were binned into the standard 1/4 kev ( also denoted as c - band ) , 3/4 kev ( m - band ) , and 1.5 kev ( j - band ) energy bands and a total energy band ( corresponding to the pulse height invariant ( pi ) channels channels 1141 , 5290 , 91201 and 11201 ) .
we calculated a merged exposure map for each individual x ray image to overcome the problem of radially decreasing sensitivity and vignetting of the pspc .
the radius of the rosat pspc point spread function ( psf ) depends on the x ray photon energy and the off axis angle with respect to the optical axis , resulting in a varying angular resolution between @xmath8 and @xmath9 within an off axis radius of @xmath10 from the optical axis .
we determined the x ray background ( xrb ) intensity level outside the hi distribution of hoii ( @xmath11 ) , which we subsequently used as the `` off '' intensity value .
we integrated the x
ray photons of each individual source within a circular area with a diameter equal to 3.5 times the diameter of the psf ( @xmath12 ) for each individual energy band .
this area covers 99% of the x ray source photons . from this total number of photons we subtracted the contribution of the xrb , and evaluated the residual x
ray intensity level in the immediate surrounding of the x ray source .
we classified an x ray source as detected if the number of net counts exceeded the 3@xmath5 threshold above the noise set by the background level .
the results are shown in fig .
[ fig : hi - xray ] top panel which shows the pspc 1/4kev map as contours superimposed on a grey - scale rendering of the neutral hydrogen distribution ( @xcite ) .
the bottom panel shows the x ray intensity distribution averaged over the entire 0.12.1kev energy range .
[ fig : complementary ] displays this latter map in the top panel as contours superimposed on a 4.86ghz radio continuum image @xcite whereas the contours in the bottom panel are overlayed on a map of the h@xmath6 emission .
the extragalactic xrb is due to the superposed emission of individual objects , mostly active galactic nuclei ( agns ) and quasi stellar objects ( qsos ) @xcite .
unfortunately , the angular resolution of the pspc does not allow one to resolve faint individual xrb sources , i.e. , the pspc is confusion limited . to constrain the contribution of the diffuse extragalactic xrb radiation
, one has to evaluate the overall spectral characteristics of this x ray emission component .
@xcite showed that the extragalactic xrb emission can be approximated by a power law with @xmath13 and @xmath14 .
they determined an intensity of the extragalactic xrb at 1 kev of about 9.6 @xmath15 .
taking this value for the xrb and assuming that the extragalactic xrb is only attenuated by the interstellar medium ( ism ) of the milky way , we derive within the area of an individual x ray source towards hoii ( @xmath16 ) , an x ray flux for the extragalactic xrb of about @xmath17@xmath18 .
so far , we have been focusing on the overall extragalactic xrb flux level . however , we have to take into account the log(@xmath19 ) vs. log(@xmath20 ) relation @xcite to evaluate the frequency of extragalactic xrb sources at a certain flux level .
unfortunately , there is no straightforward way to discriminate between unrelated background sources and those belonging to hoii .
however , we can reduce this ambiguity by studying the x ray colors of a source and by searching for counterparts at other wavelengths using complementary data ( which is what we will do in the following ) .
extragalactic sources unrelated to hoii and absorbed by its gas distribution
will appear as hard x - ray sources , detectable preferentially in the m and j band .
the 3@xmath5 x ray flux level across the entire energy band is @xmath21@xmath18 , above the diffuse extragalactic background level . applying the log(@xmath19 ) vs.
log(@xmath20 ) relation derived by @xcite to the area of hoii hi gas distribution , we expect to detect about 7 significant x - ray sources unrelated to hoii . assuming a distance to this object of 3.2 mpc and a galactic x ray attenuating column density of @xmath22 @xmath23 @xcite
, we derive a detection luminosity threshold of @xmath24@xmath1for sources associated with hoii ( neglecting for the moment absorption within this galaxy ) . only very young supernova remnants or accreting x
ray binaries typically exceed this threshold .
the pspc x
ray data presented here therefore trace only two extreme populations in hoii : the actual ( @xmath25 yr , i.e. , supernova remnants ) or a population older than @xmath26yr , such as pulsars or accreting x ray binaries .
the intermediate age stellar population can be traced via h@xmath6 and far - ultraviolet ( fuv ) observations @xcite .
combining all data provides a unique picture of the star formation history of hoii over the last 100 myr .
in this section we will introduce a useful tool to constrain the spectral properties of faint x - ray sources . using this tool it becomes possible to estimate whether the spectrum of a ( faint ) x ray source is predominantly characterized by a thermal emission spectrum or rather by a power law .
we will first discuss briefly the characteristics of those sources likely to be encountered in our object and then proceed to explain what we call the hardness ratio diagram and where these different sources are likely to be encountered in these diagnostic diagrams .
supernova remnants are characterized by their thermal plasma radiation .
below a temperature of @xmath27k , emission lines dominate the x ray spectrum . above this temperature , bremsstrahlung is the most important cooling mechanism of the hot plasma .
the cooling time of the plasma can be approximated via @xmath28 where @xmath19 is the total number of particles , @xmath29 is the hot plasma temperature in [ k ] , @xmath30 the ion density , @xmath31 the electron density , both in [ @xmath32 and @xmath33 the normalized cooling rate @xcite .
this leads to @xmath34yr , assuming a temperature of @xmath27k and a typical electron density of about @xmath35 .
the expected x
ray luminosity of a single young sn is @xmath36 @xmath1@xcite .
core collapse sn are correlated in space and time , their progenitor having been formed in groups , ob associations or even larger conglomerates . expanding individual supernova remnants are therefore likely to merge , creating expanding shells referred to in the literature as super - bubbles or supergiant shells which are supposedly filled with tenuous , high temperature gas .
these shells show up in hi maps as holes in the neutral hydrogen distribution . in the case of hoii
the largest holes have diameters roughly twice the @xmath37 scale height of the hi disk of 600pc , according to @xcite .
the cooling time of the coronal gas within the interior of the shells is much longer than the ages of the hi holes which should make them detectable in x rays .
however , the largest super - shells , with diameters exceeding 1kpc , likely broke out of the disk , at which moment the hot gas interior is vented in to the halo , the hot plasma is lost and the hole becomes invisible in x
rays . in the case of the smaller , still confined holes , the interior x ray emission should be detectable if the hi column density of the approaching side of the hi shell does not exceed values of @xmath38@xmath39 , above which value the soft x ray emission ( @xmath40 kev ) of the hot gas is absorbed _ in situ _ by the shell . pulsars ,
cooling neutron stars as well as accreting x ray binaries are the very end points of stellar evolution and may be observable within the interior of the holes , under the assumption that the origin of the holes is caused by star forming activity and subsequent rapid stellar evolution of the most massive stars .
their x ray luminosities typically range between @xmath41@xmath42 . to identify the emission mechanism of an x ray source the standard procedure is to extract a spectrum from the x ray data .
the brightest source associated with hoii has previously been studied in this way by @xcite . however , due to the large number of free parameters ,
even the analysis of this high signal to noise x ray spectrum is not without its ambiguities . in the case of faint x
ray sources ( @xmath43 @xmath18 ) , which are discussed here , the insufficient signal to noise ratio does not allow us to extract much information from the pspc spectra .
however , the count rates within the individual broad energy bands contain significant information about the x ray source spectrum .
we therefore decided to study the hardness ratios to constrain the x ray emission process .
for this aim , we subtracted the contribution of the extragalactic x - ray background emission from the count rates of the individual x - ray sources . in fig .
[ colourcolour ] we plot the hardness ratio 1 ( hr1 ) ( @xmath44 ) versus the hardness ratio 2 ( hr2 ) ( @xmath45 )
. the stronger the photoelectric absorption the harder the resulting x ray spectrum .
the lines in the figure indicate the locus where one can expect individual sources to fall in the hardness ratio diagram , depending on the type of x ray emission , either thermal or power law , over a range of temperatures and power law indices , respectively , and over a range of foreground absorptions .
this latter effect is quite comparable to the reddening caused by intervening dust in an optical color color diagram .
the relations displayed in fig . [ colourcolour ] , both for the power law ( solid lines ) as well as the thermal plasma spectra assuming solar metal abundances ( dashed lines ) , are plotted for increasing absorbing foreground column density ( from left to right @xmath46 and @xmath47 ) .
the leftmost relation of both the power law or the thermal case represents the un - absorbed situation .
the numbers along the tracks for @xmath48 ( which corresponds to the absorbing foreground column density corresponding to the milky way ; @xcite give the spectral slope for the power
law spectra ( solid lines ) and log(@xmath29[k ] ) in case of the thermal plasma spectra . obviously ,
ray sources will predominantly be found in the lower left part of the x ray hardness ratio diagram , whereas highly absorbed or intrinsically hard x ray sources populate the upper right part of the diagram .
in total we detected 31 significant x ray sources within the extent of the hi distribution of hoii .
as mentioned earlier , only 7 x - ray sources are expected , based on the log(@xmath19 ) vs. log(@xmath20 ) relation derived by @xcite . to avoid any confusion with unrelated extragalactic xrb sources , we decided to study only those x
ray sources which are identified in at least one additional frequency range .
this complementary information also gives us a better handle on the nature of the x ray source in question .
this left us with 13 sources , all believed to be related to hoii .
the properties of these 13 sources ( positions , fluxes , luminosities , complementary data ) are listed in table [ xraysources ] .
we mark those sources with an `` x '' which emit significant soft ( 1/4 kev ) x ray photons , implying that they are most likely associated with hoii .
because a background x - ray source will be attenuated by the entire amount of ism belonging to hoii , this leads to strong photoelectric absorption , reducing especially the soft part of the x - ray spectrum .
we tried to constrain the x ray emission process further by studying the spectral properties of the x ray sources via the hardness ratio diagram ( sec .
the 1/4 kev map ( fig .
[ fig : hi - xray ] ) shows the emission which is most likely associated with x ray plasma radiation .
because of the fairly high luminosity limit of the observations ( @xmath0@xmath1 ) , supernova remnants are the most likely candidates to be encountered .
the x ray sources # 6 , # 7 , # 9 , # 10 and # 13 are associated with hi holes cataloged by @xcite their centers are located within the extent of the hi holes ( for details see table [ xraysources ] ) .
although these soft x ray sources are located within the low volume density cavities of hoii , one can not exclude a chance association with an extragalactic x
ray background source .
the x ray sources which most probably belong to hoii should be detectable simultaneously in the radio and/or the optical regime .
sources # 7 , # 8 , and # 13 are positionally coincident with radio continuum emission ( see fig.[fig : complementary](top ) and also @xcite ) .
according to @xcite the x ray source # 7 ( hi hole no .
36 ) coincides with a steep spectrum non thermal radio source . in fig .
[ diagramme ] we plot all x ray sources brighter than @xmath49 @xmath18 in a hardness ratio diagram similar to the one presented in fig.[colourcolour ] .
this suggests for source # 7 a steep x ray energy power law spectrum ( @xmath50 ) , implying an accreting x
ray binary system as a likely source of the radiation @xcite .
@xcite favor an extragalactic origin , unrelated to hoii , because they interpret the steep radio continuum spectrum as an indication for spectral aging of the electron population which is frequently discussed in the framework of evolutionary models of radio galaxies .
however , the x ray as well as the positionally coincident radio continuum source fits very well within the hi hole no .
36 of the @xcite catalog .
moreover , close to x ray source # 7 a small hii region is located ( fig .
[ fig : complementary ] ) .
one may speculate that this hii region and associated h@xmath6 emission is due to secondary star formation , on the rim of hi hole no . 36 ( puche et al .
1992 ) .
the x ray sources # 6 , # 9 and # 10 are not exactly coincident with h@xmath6 or radio continuum emission .
however , h@xmath6 and radio continuum radiation is present within the @xmath51 diameter discussed earlier which is why we included these sources in our analysis .
close to the southern boundary of # 6 ( hi holes no . 22 and no .
23 ) , h@xmath6 and thermal radio continuum emission indicates the presence of a patchy star forming region .
the fuv data of @xcite suggest a young star forming region with an age of 2.5 3.5 myr .
inspection of table 2 of @xcite supports the view that x ray source # 6 ( positionally coincident with hi hole no .
22 ) is associated with a stellar cluster . the x ray sources # 9 ( hi hole no .
29 and no .
32 ) and # 10 ( hi hole no .
21 ) are associated with faint h@xmath6 and fuv emission at their rims .
the fuv data implies an age of @xmath52myr @xcite .
this is consistent with the findings based on the x ray hardness ratio diagram ( fig .
[ diagramme ] ) and suggests power
law spectra for both x ray emitting sources .
their x ray luminosities of a few @xmath53@xmath1classify both as likely candidates of accreting binary systems @xcite .
sources # 1 , # 5 , # 8 and # 11 are located outside hi holes , towards high column density regions of hoii .
the accidental association with an extragalactic xrb source becomes much less likely , because of the strong attenuation of a possible unrelated extragalactic soft x
ray source by the significant amounts of x ray absorbing matter within hoii ( in addition to the roughly constant absorption due to the milky way ) .
source # 1 is associated with bright h@xmath6 and fuv emission , however it is not detected in radio continuum maps .
this can be interpreted as an indication for a developed star cluster with evolved stars , which already partly left the main sequence .
the fuv data implies an age of @xmath52myr .
the x ray emission is too faint to set constraints on the most likely x ray emission process , even when employing the hardness ratio diagram .
so , no firm conclusions as to the nature of the source can be derived from the available data .
source # 5 , which shows x ray emission at the 3@xmath5 level , coincides with a radio continuum source . @xcite derived a thermal radio continuum spectrum for that source from their multi frequency analysis .
this is consistent with the h@xmath6 map ( bottom panel of fig.[fig : complementary ] ) which shows a compact hii region at this position .
therefore the most likely conclusion regarding the nature of this source is that of a young star forming region .
source # 8 emits thermal radio continuum emission as well .
this is consistent with the h@xmath6 map of hoii ( fig .
[ fig : complementary ] ) which reveals a very prominent hii region at this location .
the linear extent of the hii region is of order 100 pc .
the x ray hardness ratio diagram also implies a thermal origin of the detected x ray emission ( log(@xmath29[k ] ) @xmath54 7 ) .
a single luminous o star has a luminosity of about @xmath55@xmath1 in the x ray energy window @xcite .
therefore , we would require a contribution by several hundred young stars in order to account for the observed x
ray emission .
because the radio continuum spectrum is thermal in nature , the region has to be very young as otherwise supernovae and their remnants would start to dominate the radio continuum spectrum with a non - thermal component .
taking into consideration that the h@xmath6 ridge related to source # 8 is located between the boundaries of two hi holes ( no .
29 and no .
36 ) , very recent induced star formation may be taking place here .
@xcite derived an age of 3.5 4.5 myr for that particular region , consistent with the presence of a young massive star forming region .
additional evidence for a stellar cluster is provided by the optical search for stellar emission within the hi holes performed by @xcite who report that a stellar cluster may underlie the nearby hi hole no .
x ray source # 11 is of interest because the analysis of @xcite implies a young star forming region , with an age of 2.5 3.5 myr , whereas this region is not detected in the h@xmath6 map .
its location in the x ray hardness ratio diagram implies a power law x ray spectrum , indicating an accreting compact object .
both the absence of h@xmath6 radiation and the x ray data suggest that towards this high column density region we observe more likely the end point of stellar evolution , like an x ray binary .
sources # 2 , # 3 , # 4 and # 12 are located far beyond the stellar distribution .
sources # 2 and # 4 are associated with faint radio continuum emission .
both sources are prominent within the 3/4 kev and the 1.5 kev energy band , whereas in the 1/4 kev map they are only seen at the @xmath56 level .
this indicates that both sources have a hard intrinsic x ray spectrum or , alternatively , are highly absorbed x ray sources .
the hardness ratio diagram implies in the case of source # 2 , that its emission may be associated with thermal plasma radiation . in the case of source # 4
it is not possible to differentiate between a power law and a thermal plasma x ray spectrum .
source # 3 is located in between both sources .
these three sources appear in the x ray map as a coherent structure .
source # 3 is visible at high signal - to - noise level in the 1/4 kev map ( fig . [
fig : hi - xray ] ) .
together with the fact that it is situated towards high hi column densities suggests that this source is associated with hoii and not located beyond its @xmath57 column density distribution .
this is the reason why we decided to include this source in table [ xraysources ] .
however , the complementary data do not reveal any obvious counterpart for source # 3 .
finally , source # 12 is located towards a high hi column density region of hoii .
neither in the radio continuum nor in the h@xmath6 map ( fig .
[ fig : complementary ] ) is there any emission associated with this particular source .
the fuv map of @xcite , however , revealed excess emission positionally coincident with this particular source . quite comparable to the case of source # 11 ( see discussion above ) the analysis of @xcite argues for a young star forming region of 2.5 3.5 myr , whereas the x ray data does not exclude the possibility that an end point of stellar evolution may account for the observed x ray emission .
this hypothesis is furthermore supported by the absence of h@xmath6 emission .
the x ray luminosity of source # 12 , with @xmath58 @xmath1 , is compatible with both a supernova remnant or an accreting binary system .
the brightest x ray source , labeled # 13 in this paper , has been the subject of papers by @xcite and @xcite .
this source dominates the entire x - ray emission of hoii , because 87% of all detected photons emerge from this unique object . without its contribution the total x
ray luminosity of hoii would only be about @xmath59 , which is equivalent to the added x - ray luminosities of about six high mass x - ray binaries . in this respect hoii
is not a spectacular object in x - rays .
here we reanalyze this object using the x ray hardness ratio diagram .
this source can be very well approximated by a single component thermal x ray spectrum with a temperature of log(@xmath29[k ] ) = 7.0 and an absorbing column density of @xmath60 ( the dashed line to the very right ) .
the neutral hydrogen column density of @xmath61 ( dashed line with the log(@xmath29[k ] ) values ) does not fit the hardness ratios of source # 13 within the uncertainties .
this implies that source # 13 is located within a low density cavity within the interstellar medium of hoii , as at least @xmath62 is due to galactic foreground extinction , leaving at most @xmath63 to any residual column density of gas belonging to the gaseous body of hoii along the line of sight to source # 13 . comparing the plasma temperature based on the x ray hardness ratio diagram with the results of @xcite
, our temperature value falls in between the extreme temperatures derived by them .
the x ray attenuating column density implied by the x ray hardness ratio diagram , however , is significantly lower than that derived by these authors . if one wanted to explain the hardness ratios of source # 13 with a power law spectrum one would have to assume that the x ray spectrum of this source is highly absorbed ( see fig .
[ colourcolour ] ) .
none of the four plotted tracks representing the power law x ray spectra fit the position of source # 13 in the hardness ratio diagram .
inspecting table 2 of @xcite shows that a power law x ray spectrum would require an absorbing column density of about @xmath64 , implying that according to the @xmath57 of @xcite , source # 13 would have to be located on the far side of hoii or deeply embedded within a molecular cloud . to constrain the nature of source # 13 further ,
we consider the x ray luminosities of likely x ray emitting candidates .
accreting binaries have typical luminosities of @xmath65@xmath1 .
source # 13 however has @xmath66@xmath1 . on the other hand ,
several supernova events are observed with reported x ray luminosities of @xmath66@xmath1 @xcite .
the supernova hypothesis appears to be a more likely explanation than a highly unusual x ray binary system , which exceeds the eddington limit .
the supernova hypothesis is further supported by the radio continuum data @xcite .
they classified a positionally coincident radio continuum source as a supernova remnant .
partial support for this interpretation comes from @xcite who applied a wobble correction to the hri image of hoii and claim that 25% of the emission from this source is due to an extended component with a diameter of order @xmath67 .
their _ asca _ data reveal , in addition to confirming a soft component which they attribute to ( multiple ) supernovae , a power law component accounting for most of the flux density beyond 2.5 kev , implying the presence of an intermediate mass black hole .
a more definitive description of the nature of source # 13 will have to await or _ xmm newton_.
a casual inspection of fig . [ diagramme ] shows that most of the x ray sources of interest lie above the track with @xmath68 which is in excellent agreement with what is expected from foreground extinction due to the milky way @xcite .
this is an important result as it implies that the x ray sources are not deeply embedded within the interstellar medium of hoii .
of the five x ray sources located within the interiors of hi holes cataloged by @xcite , two of them appear to be associated with supernova remnants ( # 7 and # 13 ) , one can be identified with a young star - forming region ( # 6 ) and the remaining two are most likely associated with the end points of stellar evolution , more specifically accreting x ray binaries ( # 9 and # 10 ) .
four x ray sources are located towards high column density regions of hoii . because of their location in the soft part of the x - ray hardness ratio diagram , and given that they would have been fully absorbed if they were located beyond the high column density regions , they must be in front of most of the hi and hence it is likely that they are sources related to hoii .
one may be associated with an evolved stellar cluster ( # 1 ) , two with young star forming regions ( # 5 and # 8) and one with an accreting x ray binary system ( # 11 ) .
we find four x ray sources which are located far beyond the stellar body and close to the edges of the hi column density distribution of hoii .
all sources are either supernova remnants or x ray binaries .
the importance here is that even though there is no evidence based on optical imaging @xcite of past star formation , this implies that star formation has occurred some @xmath4yr ago at large galactocentric radii .
this suggests that in general the supergiant hi holes may indeed be residuals from energetic events associated with past star forming activity of hoii , on timescales @xmath69yr .
as most sources are too weak to derive an x ray spectrum , we developed a classification based solely on the hardness ratio diagram .
the faintest source analyzed in this way has an x ray flux of only @xmath70 @xmath18 .
although not conclusive , the hardness ratio diagram allows a separation between sources with thermal and power law spectra . to reach a firmer conclusion as to the nature of the x
ray sources we combined all available information ranging from the x ray to the low frequency radio regime .
all but one of the x ray sources , # 2 , are associated with fuv emission , as presented by @xcite , confirming their spatial correlation with regions of recent , massive star formation .
some of the x ray emission of hoii is associated with hii regions .
this correlation remains somewhat tentative , however , as we have to take into account that the x ray detection threshold is high , strongly selecting only the brightest star forming regions . more sensitive and
observations will overcome this instrumental limitation .
in the case of x ray source # 8 , the h@xmath6 , radio continuum and x ray data indicate independently the presence of a massive star forming region . because of its location in between two prominent hi holes , we speculate that one of the holes might be re - heated by a recent supernova event ( x ray source # 7 ) ; one may then further speculate that this is evidence for induced star formation .
the observations reported here did not reveal any hot gas associated with the hi holes . at first sight
this might seem remarkable .
for example , in the milky way several hot
plasma filled hi cavities are known .
the most prominent one is the local bubble which encloses the sun @xcite .
its x ray emission is very soft ( log(@xmath29[k ] ) @xmath71 6.0 ) and faint .
however , its pressure of @xmath72 is a factor of 2 to 5 higher than that of the neutral clouds located within the cavity .
hence , this local cavity presumably is a low volume density region produced by several supernova events , the latest supernova event dating back to @xmath73 years .
another interesting galactic feature is the x ray bright super - shell known as loop i or north polar spur @xcite .
loop i has a diameter of about @xmath74 300pc .
its temperature is log(@xmath75[k ] ) = 6.7 , much higher than that of the local x ray plasma .
it is highly over pressured ( @xmath76 ) and still expanding
. observed with from a hypothetical vantage point , such as an external galaxy , both x ray features would have remained undetected because the photoelectric absorption of the enclosing thin neutral shell with @xmath77 is sufficiently high for the x
ray photons originating within the interior of the shell to be absorbed .
the only extragalactic detections of hot gas in super - bubbles claimed thus far are restricted to dwarf galaxies .
examples are the supergiant shell lmc4 @xcite , and the super - bubbles n44 and n11 @xcite , all three situated in the lmc .
other examples are the supergiant shell sgs2 in ngc4449 @xcite and the possible super - shell near holmberg ix @xcite .
recently @xcite detected marginally extended x ray emission coinciding with a supergiant shell in ic2574 in a pointed observation .
they found an x ray luminosity ( @xmath78 = 3.2mpc ) of @xmath79 ergs@xmath80 . assuming , as usual , a raymond
smith model @xcite they derived a plasma temperature of log(@xmath81)\,=\,6.8\pm0.3 $ ] and an internal density of @xmath82@xmath83 .
the internal pressure of @xmath84k@xmath83 is again much higher than the pressure of the ambient warm neutral and ionized medium ( @xmath85k@xmath83 ) which led them to suggest that it is probably this hot gas which is still driving the expansion of the shell ( see e.g. , @xcite ) .
however , follow
up observations with indicate that the source is more compact . in short , only few , possibly pathological holes , are detected in x
. we should not be too surprised as several authors have pointed out that the x ray luminosity of the thermal gas expected to fill the super - bubbles is well below the detection limit @xcite the expected x
ray luminosity can be written according to @xcite ( their eq .
6 ) see also @xcite ) @xmath86 where l@xmath87 is the mechanical energy expressed in units of @xmath88 ergs@xmath80 , @xmath89 the ambient density in atom@xmath83 , t@xmath90 the age of the super - bubble in myr and @xmath91 a dimensionless scaling factor for the classical conductivity ( @xmath92 ) .
the metalicity @xmath93 also influences the luminosity in the x - ray domain . according to @xcite the dependence of @xmath94 on @xmath95
can be approximated by @xmath96 .
in example , in a low metalicity interstellar medium with @xmath97 the expected x - ray luminosity of an x - ray plasma will only be reduced by a factor of three , not by an order of magnitude . in the case of hoii ( metalicity : about 0.25@xmath98 , @xcite ) the expected x - ray luminosity will only be marginally reduced .
the `` standard supernova '' described by eq .
[ luminosity ] will expand into an ambient neutral medium , with a typical density ranging between @xmath99 .
the age of an expanding structure can be estimated from hi @xcite and fuv data @xcite to a few tens of @xmath100 years .
taking @xcite values of the total mechanical energy needed to create the holes of @xmath101 , we can calculate the range of expected x - ray luminosity from eq .
[ luminosity ] . assuming a conductivity of at least @xmath102 gives a range of @xmath103 .
according to @xcite and @xcite the mechanical energy ranges between @xmath104 .
this leads to predicted values of l@xmath105 .
a supernova remnant evolving within a high volume density environment has the highest probability to be detected in x - rays with _ rosat_. as an example , source # 7 and hi hole no .
36 are positional coincident , using @xmath106 @xcite and @xmath107 @xcite gives @xmath108 . with @xmath109 and @xmath102
we derive @xmath110 .
given the overall uncertainties this is in good agreement with the observed value ( @xmath111 , see table [ xraysources ] ) .
in a more typical low density environment @xmath112 the x - ray luminosity is only of order @xmath113 , i.e. , below the sensitivity of our _ rosat _ data . what does this imply ?
simply put , the vast majority of super - bubbles are far too faint to have been detected by _
rosat_. those which have been are in some sense peculiar .
one method to boost the x ray luminosity , and increase the x ray emitting lifetime , would be to invoke mass loading @xcite .
besides , there remains the possibility that these cases are associated with x ray binaries rather than plasma filled super - bubbles . for example , in the case of ic2574 , due to the limited angular resolution of the _ rosat _ pspc , it was not possible to differentiate between the x - ray emission originating from hot gas and the thermal emission from an unresolved individual x - ray source , as now seems likely on the basis of our data of ic2574 .
these unresolved sources do have the same luminosities , of order @xmath88ergs@xmath80 , as super - shells .
alternatively , there could be a contribution to a thermal plasma by supernovae developing in a dense environment , such as sn1988z which reportedly reached an x ray luminosity of @xmath114ergs@xmath80 @xcite .
moreover , it might be difficult to detect coronal gas from expanding super - bubbles as there could be a conspiracy at work .
perhaps only ( super)bubbles in the making , which are still actively being powered and hence over - pressured , would be sufficiently bright to be seen by , for example , _
rosat_. however , these objects tend to be fairly small , young , and surrounded by a dense hi shell which would absorb the x
rays , especially the softer ones .
once a bubble has reached its final size , after some @xmath4yr , the interior has cooled down to below @xmath100k and no x
rays will be detected .
it is only with the new generation of satellites , such as and that we can hope to settle this issue once and for all .
we analyzed pspc data of the dwarf galaxy holmberg ii .
the data provide information on faint x
ray emission features down to a @xmath56 unabsorbed limiting flux of @xmath115 @xmath18 .
this corresponds to a luminosity threshold of @xmath116@xmath1 . in total
we detected 31 significant x ray sources located within the hi column density distribution of hoii . using the log(@xmath19 ) vs. log(@xmath20 ) relation of @xcite
we expect to detect about 7 xrb sources within the area of hoii hi gas disk . to avoid confusion with unrelated extragalactic xrb sources , we investigated further only those x
ray sources which have counterparts in fuv , h@xmath6 or radio continuum emission .
this additional selection criterion is fulfilled by 13 x ray sources . in order to determine the nature of faint x
ray sources for which no high signal to noise spectral information can be obtained , we applied the technique of the hardness ratio diagram . for this purpose
, we evaluated the intensities in the broad 1/4 kev , 3/4 kev and 1.5 kev energy band and calculated the corresponding x
ray hardness ratios . depending on the amount of photoelectric attenuation ,
an x ray source will fall within a specific zone in this diagram . in analogy with the optical regime
, one can consider the variation of the x ray color due to photoelectric absorption as the counterpart for reddening of stellar light by intervening dust .
the x ray hardness ratio diagram offers a very sensitive tool to constrain the x ray emitting process even for faint sources .
the faintest source analyzed in this paper has an x ray flux of only @xmath117 @xmath18(#1 ) .
of the 13 sources detected with confidence and for which counterparts at other wavelengths have been found , we speculate that of order 3 are young snrs , 3 are associated with regions of active star formation , 1 might be the superposed contribution of the coronal x ray emission of several hundred o stars and of order 5 are most likely x ray binaries .
the strongest source , # 13 , was already studied previously by @xcite . whereas they claim a power law spectrum and relatively high absorbing column density of about @xmath118 , and suggest some source of accretion on to a compact object , we prefer to interpret this source as having a softer intrinsic spectrum , and an absorbing column density of only @xmath119 .
our analysis leads us to propose that this source is most likely a snr .
this is supported by the analysis published by @xcite who find that some 27% of the x ray emission comes from an extended source and that the spectrum below 2.5kev is best fit by a thermal plasma .
however , their _ asca _ observations show that above 2.5kev this source shows a power law spectrum which they ascribe to accretion on to an intermediate mass black hole .
no extended emission which could be attributed to hot gas filling the hi super - bubbles was detected above our luminosity limit of @xmath0@xmath1 .
this is not too surprising and in agreement with theoretical estimates which predict luminosities of l@xmath120 .
we like to thank an anonymous referee for careful reading of the manuscript and a number of good comments which improved the presentation of the paper .
jk was supported by the deutsches zentrum fr luft und raumfahrt under grant no .
50 or 0103 .
fw acknowledges national sciene foundation grant ast 96 - 13717 , eb was supported by a research grant from conacyt ( no .
27606e ) .
lllrrrc 1 & 08:19:12 & 70:45:59 & @xmath121 & @xmath122 & fuv & x + 2 & 08:18:34 & 70:48:40 & @xmath123 & @xmath124 & rcs / therm .
x ray & + 3 & 08:18:35 & 70:47:20 & @xmath125 & @xmath126 & & + 4 & 08:18:43 & 70:46:03 & @xmath127 & @xmath128 & fuv , rcs , x ray therm .
o. powl & + 5 & 08:18:53 & 70:44:45 & @xmath129 & @xmath130 & fuv , therm .
rcs , h@xmath6 & x + 6 & 08:19:03 & 70:43:03 & @xmath129 & @xmath130 & fuv , therm .
rcs , h@xmath6 & x + 7 & 08:19:20 & 70:43:30 & @xmath131 & @xmath132 & fuv , snr rcs , x ray powl . , h@xmath6 & x + 8 & 08:19:17 & 70:42:56 & @xmath133 & @xmath134 & fuv , therm .
rcs , x ray therm .
, h@xmath6 & x + 9 & 08:19:11 & 70:42:01 & @xmath135 & @xmath136 & fuv , h@xmath6 & x + 10 & 08:19:00 & 70:41:27 & @xmath137 & @xmath138 & fuv , h@xmath6 & x + 11 & 08:19:18 & 70:40:40 & @xmath139 & @xmath140 & fuv & x + 12 & 08:19:10 & 70:39:39 & @xmath141 & @xmath142 & fuv , rcs , x ray therm .
o. powl &
+ 13 & 08:19:33 & 70:42:18 & @xmath143 & @xmath144 & fuv , snr rcs , x ray therm .
, h@xmath6 & x + arthur , s.j . , henney , w.j .
, 1996 , , 457 , 752 bomans , d.j . ,
chu , y.h . , hopp , u. , 1997 , , 113 , 1678 bomans , d.j .
, dennerl , k. , krster , m. , 1994 , 283 , l21 bureau , m. , carignan , c. , 1997 , `` dwarf galaxies : probes for galaxy formation and evolution '' , abstract book ( xxiiird iau general assembly , kyoto ) , ed .
takahira , s. , p. 36
chu , y.h . ,
chang , h.w .
, su , y.l . ,
mac low , m.m . , 1995 , , 450 , 157 chu , y.h .
, mac low , m.m .
, 1990 , 365 , 510 chu , y.h . , mac low , m.m . , garcia
segura , g. , wakker , b.p .
, kennicutt , r.c . , 1993 , , 414 , 213 egger , r.j . ,
aschenbach , b. , 1995 , , 294 , l25 fabbiano , g. , 1989 , , 27 , 87 fabbiano , g. , 1996 , `` x - ray emission of galaxies '' in
rntgenstrahlung from the universe , eds .
zimmermann u. et al .
fabian , a.c . ,
terlevich , r. , 1996 , , 280 , l5 gendreau , k.c . ,
mushotzky , r. , fabian , a.c .
et al . , 1996 , pasj , 47 , 5 hartmann , d. , burton , w.b . , 1997 , `` atlas of galactic neutral hydrogen ''
( cambridge university press : cambridge ) hasinger , g. , burg , r. , giacconi , r. et al .
, 1998 , , 329 , 482 hunter , d.a . ,
gallagher , j.s . , 1985 , , 58 , 533 kudritzki , r.p . ,
palsa , r. , feldmeier , a. , puls , j. , pauldrach , a.w.a . , 1995 , `` rntgenstrahlung from the universe '' mpe - report 263 , eds .
zimmermann , h.u . ,
trmper , j.e . , yorke , h. , p. 9
mac low , m.m . ,
chang , t.h .
, chu , y.h .
, points , s.d . ,
smith , r.c .
, wakker , b.p .
, 1998 , , 493 , 260 martin , c.l . ,
kennicutt , r.c . , 1995 , , 447 , 171 mashchenko , s.ya . ,
silich , s.a . , 1995 ,
arep 39 , 587 miller , b.w . , 1995 , , 446 , l75 miyaji t. , lehmann i. , hasniger g. , 2001 astro - ph/0102500 mushotzky , r.f . , cowie , l.l . , barger , a.j . ,
arnaud , k.a . , 2000 , , 404 , 459 oey , m.s . , clarke , c.j . , 1997 , , 289 , 570 puche , d. , westpfahl , d. , brinks , e. , roy , j .-
, 1992 , , 103 , 1841 raymond , j.c . ,
smith , b.w . , 1977 , , 35 , 419 rhode , k.l . ,
salzer , j.j . ,
westpfahl , d.j . ,
radice , l.a . , 1999 , , 118 , 323 schulz , n.s .
, 1999 , , 511 , 304 sfeir , d.m . , lallement , r. , crifo , f. , welsh , b.y . , 1999 , , 346 , 785 silich , s.a . , tenorio tagle , g. , terlevich , r. , terlevich , e. , netzer , h. , 2000 , ( accepted ) stewart , s.g . , fanelli , m.n . , byrd , g.g .
et al . , 2000 , ,
529 , 201 sutherland r.s .
, dopita m.a . , 1993 , , 88 , 253 tongue , t.d . ,
westpfahl , d.j . , 1995 , , 109 , 2462 walter , f. , brinks , e. , 1999 , , 118 , 273 walter , f. , kerp , j. , duric , n. , brinks , e. , klein , u. , 1998 , , 502 , l143 weaver , r. , mccray , r. , castor , j. , shapiro , p. , moore r. , 1977 , , 218 , 377 white , n.e . , swank , j.h . ,
holt , s.s . , 1983 , , 270 , 711 zezas , a.l . ,
georgantopoulos , i. , ward , m.j . , 1999 , , 308 , 302 zimmermann , u. , boese , g. , becker , w. , et al . , 1998 , mpe report | we present a study of the irregular dwarf galaxy holmberg ii based on pspc observations ( total exposure time : 22ksec ) .
holmberg ii is a nearby ( 3.2mpc ) , well studied dwarf irregular galaxy .
it is famous for its interstellar medium which is dominated by expanding structures such as hi holes and shells .
we search for x ray emission from point sources as well as for diffuse emission , down to the detection limit of the data . using x
ray hardness ratio diagrams we differentiate between thermal plasma and power law x ray spectra which helps to determine the nature of the individual sources . correlating the x ray data with complementary observations ranging from the far ultraviolet to the radio regime we increase the probability of correctly identifying sources belonging to holmberg ii .
we did not detect soft x
ray emission originating from hot gas within supergiant hi shells above our luminosity sensitivity limit of ( @xmath0@xmath1 ) .
this finding can probably be attributed to blow out in the case of the largest holes and insufficient sensitivity ( due to strong photoelectric absorption ) in case of the smaller hi holes . however we find faint x
ray sources well beyond the stellar body but within the hi distribution of holmberg ii , which suggests the presence of x ray binaries .
this indicates that star formation has taken place across the entire gaseous disk of holmberg ii in the past , some of which may have created the structures seen in the ism at large galactocentric radii . |
one of the important processes that will be studied at high precision at ilc with and without beam polarization is w - pair production .
phenomelonogical studies of this process within the standard model ( sm ) have been carried out in great detail @xcite .
since properties of the weak gauge bosons are closely linked to electroweak symmetry breaking ( ewsb ) and the structure of the gauge sector in general , detailed study of w physics will throw light on what lies beyond the sm .
the study of mechanisms of ewsb is one of the main concerns of particle physics today .
the standard higgs mechanism is less than satisfactory , and faces difficulties such as the hierarchy problem .
looking beyond the sm , the newly proposed little higgs scenarios @xcite provide a dynamical way to generate the ewsb , in contrast to the _ ad hoc _ introduction of the elementary scalar sector .
apart from this aesthetically appealing feature , little higgs models provides rich phenomenology with predictions that could be vidicated or ruled out at future colliders such as the lhc and the ilc .
one major feature of such models is the presence of additional gauge bosons in the physical spectrum .
these influence processes like w - pair production in @xmath0 collisions , firstly directly through their exchange in the process , and secondly through the change of standard couplings through mixing with other gauge bosons .
although these additional gauge bosons are typically too heavy to be produced at reference ilc energies of 500 and 800 gev which we use in the present study , their effects manifest themselves as stated above .
recently it was pointed out , in a preliminary study , that for one such model known as the littlest higgs model ( lhm ) , the fraction of longitudinally and transversely polarized of one of the w s could be significantly different from the corresponding fraction in the sm @xcite . in this work ,
we consider a refined treatment of the lhm to be described below , and extend the prior work to polarization fractions and total cross - sections of the w s , energy and angular distributions of decay leptons , as well as to observables like the forward - backward and left - right asymmetries , which are more sensitive to the effects of the lhm compared to the cross - sections .
the ilc is expected to have large beam polarizations which will significantly enhance the sensitivity to new physics , for a review , see ref .
we consider different beam polarizations with the aim of improving the sensitivity of the observables considered here .
this article is organized as follows : in section 2 we , very briefly , introduce the lhm , and describe its particle spectrum and couplings relevant to @xmath1 . in section 3
we present our analysis of the total cross section and w - polarization fractions in the lhm and compare with the sm case . in section 4
we take up the task of probing the model by considering decay of one of the @xmath2 to a lepton pair .
we consider the energy and angular distributions for the cases of sm and lhm .
we also discuss the left - right as well as forward - backward asymmetry in this section .
finally we summarize our study and present our conclusions in section 5 .
note that we have included beam polarization effects in this study in each of the relevant sections .
in little higgs models @xcite a non - linear realization of some global symmetry @xmath4 broken down to @xmath5 is considered .
the nambu - goldstone bosons ( ngb ) of the symmetry breaking are candidate higgs fields . in a specific model , called lhm @xcite @xmath6 is broken down to @xmath7 via a vacuum expectation value ( vev ) of order @xmath8 .
interactions of ngb s are described by a non - linear sigma model , which is an effective theory valid below the cut off @xmath9 . in the version of the lhm @xcite we will consider in this report ,
@xmath10 is gauged , which is broken down to the sm gauge group @xmath11 . under this , the 14 ngb s transform as a real singlet , a real triplet , a complex doublet and a complex triplet .
the real fields become the longitudinal degrees of the heavy gauge bosons , while the sm gauge bosons , @xmath12 and @xmath13 remain massless at this stage .
the doublet ngb field has the correct quantum numbers to be identified as the standard higgs doublet . at tree level
, they have only derivative couplings , but quantum corrections at one - loop level generate a coleman - weinberg potential with quadratic and quartic terms , consequently breaking electroweak symmetry .
gauge symmetry is constructed such that , in the absence of any one ( original ) gauge interaction the higgs is massless to all orders .
this also ensures that quadratically divergent contributions to the mass - square term at one - loop level are cancelled between the gauge bosons from the two sectors .
logarithmically divergent terms contribute to the potential . in order to avoid a quadratic divergence due to a top - quark loop ,
a pair of ( weak - singlet ) weyl quarks @xmath14 is introduced , which mix with the ordinary left- and right- quarks to give mass eigenstates . here
again , it is so arranged such that the quadratic divergence coming from the standard top - quark is cancelled by its heavy counterpart , and the logarithmically divergent part is added to the coleman - weinberg potential .
the model achieves ewsb , at the same time , protecting the higgs mass from accquiring quadratically divergent corrections at one loop .
it is therefore to be expected that these models will have a rich phenomenology with distinct signatures that can be probed at upcoming collider experiments .
an incomplete list of phenomenological studies of different variations of the little higgs model scenarios is ref .
@xcite .
our interest here is the effective theory below the cut - off @xmath15 . for the process
@xmath16 , we have an @xmath17channel process with the exchange of the heavy neutral gauge boson , @xmath18 , in addition to the standard channels as shown in fig .
[ fig : fd ] . in the lhm.,width=10,height=94 ]
apart from the contribution due to the additional @xmath19-channel process , the lhm also changes the sm couplings .
the relevant couplings in terms of the parameters of the lhm is given below in terms of the global symmetry breaking scale @xmath8 , the parameter @xmath20 , which represents mixing between the two gauge sectors of the lhm and the vev .
the three - point gauge couplings involving @xmath21 are given by : @xmath22,\ ] ] where all the momenta are considered outflowing , and @xmath23 .
individual @xmath24 are given by : @xmath25 fermion couplings are given by : @xmath26~ \gamma^{\mu}~(1-\gamma^5)\\ & & g_{eev}=i\gamma^{\mu}(c^{v}_v - c^{a}_v\gamma^{5}),\end{aligned}\ ] ] where @xmath27\\ & & c^{a}_{z}= -\frac{e}{\sin2\theta_w}\left[-\frac{1}{2}-\frac{v^{2}}{f^{2 } } \frac{\sin4\theta\cot\theta}{2}\right]\\ & & c^{v}_{z_{h}}=c^{a}_{z_{h}}=-\left(\frac{e\cot\theta}{4\sin\theta_w}\right)\end{aligned}\ ] ] following ref .
@xcite we consider the measured values of the fermi coupling constant , @xmath28 , the @xmath29-boson mass , @xmath30 , and the fine structure constant , @xmath31 as the standard model input parameters .
the weak mixing angle is obtained from the relation : @xmath32 . the bare weak mixing angle
, @xmath33 is related to the measured weak coupling angle , @xmath34 through the following relation : @xmath35\ ] ] the weak coupling constant expressed in terms of the other parameters becomes @xmath36.\ ] ] the triplet vev , @xmath37 is related to the generated quartic coupling , and the guage couplings through @xmath38 which is further constrained to @xmath39 @xcite . in our numerical analysis we consider the approximate relation , @xmath40 .
note that our numerical results are not very sensitive to this choice .
thus we are left with two free parameters @xmath8 and @xmath41 . as argued by @xcite ,
precision electroweak measurements restrict the parameters to be @xmath42 tev and @xmath43 . in our numerical analysis
we consider some representative values satisfying these restrictions .
in this section we present the results of our numerical analysis to probe the lhm through the process @xmath44 at the ilc .
we compute the total cross section incorporating beam polarization using the helicity amplitudes given in ref .
@xcite with the new couplings and with the added contribution due to the exchange of @xmath45 . with beam polarization , in general
, the polarized cross section may be expressed as @xcite : @xmath46,\end{aligned}\ ] ] where @xmath47 with @xmath48 representing the left- and right - polarized electrons ( and positrons ) , respectively .
the degree of polarization is defined as : @xmath49 , where @xmath50 denote the number of left - polarized and right - polarized electrons ( and positrons ) , respectively .
more than 80% of electron beam polarization and large positron beam polarization are expected to be achieved at ilc . in our analysis
we consider the ideal possibility of 100% polarization of the beams .
our results are presented in the figures below . in fig .
[ fig : cs ] we present the total production cross - section for a typical choice of parameters of the lhm and in the sm of the case of unpolarized and polarized beams with a specific choice of beam polarization .
it may be seen that in the case of unpolarized beams the cross section of lhm does not deviate much from that of the sm for energies up to 1 tev .
the presence of beam polarization changes this situation significantly .
the combination , @xmath51 and @xmath52 is seen to provide the largest deviation , which is the case we have chosen to display .
however , for this configuration a reduced number of @xmath3-pairs is produced , as the dominant @xmath53-channel is cut off .
notice also that there is no contribution due to the @xmath45 exchange in this case , as the @xmath18 couples only to the left - handed electrons .
the effect , therefore is purely due to the deviation of the standard model couplings .
we will return to some more properties of this in the next sub - section .
@xmath54 production in an @xmath55 collision for sm [ blue - solid ] and lhm [ red - dashed ] with [ a ] unpolarized beams [ b ] polarized beams with @xmath56=1 and @xmath57=0 .
the parametres for lhm are @xmath8=1 tev and @xmath58=0.3 , title="fig:",width=6,height=6 ] @xmath54 production in an @xmath55 collision for sm [ blue - solid ] and lhm [ red - dashed ] with [ a ] unpolarized beams [ b ] polarized beams with @xmath56=1 and @xmath57=0 .
the parametres for lhm are @xmath8=1 tev and @xmath58=0.3 , title="fig:",width=6,height=6 ] here we explore the sensitivity of the ilc to the lhm when we consider polarization fractions of the w bosons .
such measurements have been considered in past experiments for precision studies of the w boson properties at lep , which has measured the fractional cross section of the polarized @xmath3 s @xcite . at ilc
higher precision is expected to be reached .
we define the polarization fractions as @xmath59 @xmath60 where @xmath61 stands for longitudinal polarization , and @xmath62 stands for transverse polarizations .
the three polarization fractions are studied as a function of @xmath63 which are plotted in fig .
[ fig : f ] .
it is readily observed that there is a significant deviation in the case of @xmath64 and @xmath65 , while @xmath66 is largely unaffected . since these fractions depend on various couplings in a complicated way , we do not attempt to explain the effects in terms of the changes in the couplings . in table [ tab : sig_f ] we present firstly the ratio of the cross section in the lhm to that in the sm for typical parameter choices , as well as the ratio of @xmath64 in lhm to that in the sm at @xmath67 gev and 800 gev for the two illustrative parameter space points . at @xmath63=500 gev for @xmath8=1 tev and @xmath58=0.3 the deviation is about 65% with unpolarized beams .
this is improved marginally to about 66% for @xmath68 .
while for the slightly larger value of , @xmath69 tev , the effect is not as dramatic , we still have significant deviation of about 25 - 30% in @xmath64 . on the other hand ,
the configuration with purely right handed electrons and unpolarized positrons has the effect of completely washing out the effect of the lhm in the polarization fractions .
therefore , beam polarization has the dramatic effect of disentangling the effects of the new physics .
crucially , there is an interplay between the pure left - handed coupling nature of @xmath18 and the corrections to the couplings of the sm z - boson coming from the parameters of the lhm which makes this possible . considering the precision at which @xmath64 could be measured at ilc ,
such effects are very interesting . ,
[ a ] longitudinal ( @xmath64 ) , and [ b ] transverse ( @xmath70 ) , with unpolarised beams for sm [ blue - solid ] and lhm [ red - dashed ] .
the parametres in this case are @xmath8=1 tev and @xmath58=0.3,title="fig:",width=6,height=6 ] , [ a ] longitudinal ( @xmath64 ) , and [ b ] transverse ( @xmath70 ) , with unpolarised beams for sm [ blue - solid ] and lhm [ red - dashed ] .
the parametres in this case are @xmath8=1 tev and @xmath58=0.3,title="fig:",width=6,height=6 ] .ratios of the cross section and w polarization fractions in the lhm to those in the sm for different beam polarizations , and for some illustrative values of @xmath8 and @xmath58 .
[ cols="^,^,<,^,^,^,^ , > " , ] another useful observable related to the angular asymmetry is the forward - backward asymmetry defined as : @xmath71 in table [ tab : f_back ] @xmath72 is tabulated for different parameter values at two different collider energies . deviation of about 5% is observed in the asymmetry for @xmath73 tev and @xmath74 .
the energy spectrum of the secondary leptons are sensitive to the @xmath75 helicities .
the energy distribution in the centre of mass frame may be written in terms of the polarization fractions of the @xmath3 s in the following form @xcite : @xmath76 .
& & \end{aligned}\ ] ] in fig .
[ fig : edist ] we present the energy spectrum of the charged decay lepton for @xmath77 gev .
we notice that a slightly larger fraction of hard leptons are produced in the case of lhm compared to the case of the sm .
the effect is much smaller in the case of @xmath67 gev . while it is true that the effect is not dramatic , given the fact that the lepton energy spectrum could be obtained easily and with high efficiency , this observable might be useful in probing the lhm .
= 1.50 truein the new gauge boson @xmath45 in the lhm has the peculiar property of coupling only to left handed fermions as mentioned earlier . on the other hand , the sm @xmath29 couples to both left- and right- handed fermions , but the corrections to the @xmath78 coupling is such that only the left - handed electron coupling is affected .
thus , one would expect appreciable change in the asymmetry between the left- and right - polarized cross sectionos .
we define the left - right asymmetry in the differential cross section as : @xmath79 where @xmath41 is the @xmath3 scattering angle .
[ fig : lr - asym ] shows the lr asymmetry for two energies . even at low energies
the deviation becomes apparent . and lhm [ red - dashed ] at [ a]@xmath63=500gev [ b]@xmath63=800gev . the parametres considered are @xmath73 tev and @xmath74.,title="fig:",width=6,height=5 ] and lhm [ red - dashed ] at [ a]@xmath63=500gev [ b]@xmath63=800gev . the parametres considered are @xmath73 tev and @xmath74.,title="fig:",width=6,height=5 ] model&@xmath8 ( tev)&@xmath58&@xmath63 = 500gev&@xmath63 = 800gev + + + & 0.992 & 0.995 + & 1 & 0.3 & 0.988 & 0.986 + lhm&1.5&0.3 & 0.990 & 0.992 + & 1.5&0.5 & 0.991 & 0.993 + + + & 0.985 & 0.987 + & 1 & 0.3 & 0.978 & 0.967 + lhm&1.5&0.3 & 0.982 & 0.979 + & 1.5&0.5 & 0.983 & 0.981 + + + & 0.974 & 0.976 + & 1 & 0.3 & 0.962 & 0.945 + lhm&1.5&0.3 & 0.969 & 0.964 + & 1.5&0.5 & 0.970 & 0.967 + + + & 0.960 & 0.962 + & 1 & 0.3 & 0.945 & 0.920 + lhm&1.5&0.3 & 0.953 & 0.945
+ & 1.5&0.5 & 0.955 & 0.949 + + + & 0.944 & 0.946 + & 1 & 0.3 & 0.927 & 0.897 + lhm&1.5&0.3 & 0.937 & 0.926 + & 1.5&0.5 & 0.939 & 0.931 + we may go one step further by considering an integral version of this asymmetry as better efficiency may be obtained this way , by integrating each of the differential cross sections from an opening angle @xmath34 up to an angle @xmath80 , for various realistic values of @xmath34 to which the data can be integrated without difficulty .
we define the integrated left - right asymmetry as : @xmath81 where @xmath82 stands for @xmath83 .
this asymmetry , for different parameter values at @xmath67 gev and at @xmath77 gev is tabulated in table [ tab : asym ] .
we see that the asymmetry is not affected in any significant way at @xmath67 gev or at @xmath77 gev for the parameter combinations considered , when no cut on angle is applied .
interesting patterns may be observed from this table .
an interplay between the differential asymmetry plotted above , and the fact that the bulk of the contribution to the cross section comes from the forward region where the asymmetry itself is not appreciable leads to more sigificant results when the cut - off angle is larger .
in other words , as the cut - off angle increases , the region of deviation between the lhm and the sm is weighted more efficiently . the case of @xmath84
is worthy of note , as there is a cross - over in the asymmetry for @xmath77 gev and the effect is completely is wiped out .
understanding the phenomenon of electroweak symmetry breaking is central to the study of elementary particle physics . among the viable alternatives to the standard higgs mechanism is the little higgs scenarios , which provides a natural way to dynamically generate electroweak symmetry breaking . in this report
we have considered the lhm which is one simple version of this scenario .
such a model predicts the existence of additional gauge bosons with masses in the tev region .
we have considered the process @xmath85 to probe this model .
presence of a heavy neutral gauge boson in addition to the sm gauge bosons , and the change of couplings of the sm particles affect this process .
we have studied the total cross section and the polarization fraction of the @xmath3 s produced for typical parameter values .
we conclude that for suitable choice of beam polarizations , there can be more than 50% deviation in the cross section for @xmath63 around and above 500 gev .
the polarization fractions are found to be more sensitive to the new effects , which can be two or three times the sm value at @xmath77 gev .
the fact that the @xmath3 polarization fractions can be very precisely measured at ilc shows that study of these observables can effectively probe the lhm .
study of secondary lepton distributions can be carried out efficiently at the ilc .
our study of the energy and angular distributions in the laboratory frame shows that significant deviation from the sm expectation is possible for parameter sets of lhm allowed by the present experimental constraints .
we see that the angular distribution is better suited in the present case with fraction of secondary leptons emitted in the backward direction deviates from the sm value by significant amounts .
we have introduced a lr asymmetry for the differential as well as the integrated cases , where the integration is performed over an opening angle given by @xmath34 .
it is shown that a judicious choice of @xmath34 can provide a window for observing striking deviations from the sm and provide a discriminating tool for the lhm .
while a more complete study including , for instance , detector efficiencies , and larger parameter space scan is pending , our study with representative parameter points illustrates that @xmath3 pair production in @xmath0 collisions at ilc energies can probe the lhm very effectively . the expression for energy - angle correlation for the unpolarised beams
is given in ref .
@xcite , for the sm and for the bess models respectively , which may be easily adapted for the lhm .
the generalization to include arbitary beam polarization for each of these models can also be done in a straightforward manner . here
we give it for the lhm .
the expressions result from straightforward dirac trace techniques to include longitudinal beam polarization , by evalauting the electronic part of the feynman diagrams given in fig .
1 . in particular , for @xmath87 and @xmath88 the requisite traces which produce the beam polarization dependence associated with them are analogous to those that have been explicitly discussed in ref . @xcite .
the beam polarization dependence in @xmath89 and @xmath90 which we have presented explicitly here , reflect the maximal parity violating nature of the couplings of the ws to the electrons and positrons in the t - channel diagram . for arbitrary beam
we have @xmath91 with @xmath92 a^{'}_{s}&=&2\left(1+\frac{1}{4\tau}-2x-2\frac{\tau}{x}\right)\cos\theta_{l},\nonumber\\[2 mm ] a_{int}&=&-2\tau+\frac{x}{\tau}-2+\frac{x}{2\tau}(1-x)\left(1+\frac{1}{2\tau}\right ) + \left(1+\frac{1}{2\tau}-\frac{2\tau}{x}-2x\right)\cos\theta_{l}\nonumber\\ & & -\left(1-\frac{1}{2\tau}\right)\left(1-\frac{x(1-x)}{2\tau}\right)\cos^{2}\theta_{l}\nonumber\\ & & -rx^{2}\left(2+(\cos\theta_{l}-\beta\cos\theta ) \times\left(2-\left(1+\frac{1}{\tau}\right)\beta\cos\theta+\cos\theta_{l}\right)\right),\nonumber \\[2 mm ] a_{t}&=&\left(-2+\frac{2x}{\tau}+\frac{x(1-x)}{4\tau^{2}}\right)+\frac{\cos\theta_{l}}{2\tau } + \left(1-(1-x)\frac{x}{2\tau}\right)\frac{\cos^{2}\theta_{l}}{2\tau}\nonumber\\ & & -\frac{2}{\tau}x^{2}r(\beta\cos\theta-\cos\theta_{l})\beta\cos\theta + 2x^{2}ar^{3}(\beta\cos\theta-\cos\theta_{l})^{2}. \nonumber\end{aligned}\ ] ] here @xmath93^{-\frac{1}{2}}\nonumber\\[2 mm ] & & a=2\tau-1+\beta\cos\theta\cos\theta_{l } , \nonumber\end{aligned}\ ] ] where @xmath94 is the scattering angle of @xmath95 and @xmath96 is the velocity of @xmath95 , both in the centre of mass frame .
the coefficients , @xmath97 s are given below for arbitrary beam polarizations with @xmath98 and @xmath99 denoting the degrees of electron and positron beams , respectively .
in addition to the standard channels ( as given in ref .
@xcite ) , these coefficients include the contribution due the @xmath18 exchange .
@xmath100\right.+\nonumber \\ & & s^{2}_{z}~g^{2}_{zww}~\left[(c^{v}_{z})^{2}+(c^{a}_{z})^{2}-p~(2 c^{a}_{z}c^{v}_{z})\right]+\nonumber \\ & & s^{2}_{z_{h}}~g^{2}_{z_{h}ww}~\left[(c^{v}_{z_{h}})^{2}+(c^{a}_{z_{h}})^{2}-p~(2 c^{a}_{z_{h}}c^{v}_{z_{h}})\right]+ \nonumber \\ & & 2~ s_{z}~g_{\gamma ww}~g_{zww}~\left[c^{v}_{\gamma}c^{v}_{z}+c^{a}_{\gamma}c^{a}_{z}-p~(c^{a}_{\gamma}c^{v}_{z}+c^{a}_{z}c^{v}_{\gamma})\right]+\nonumber \\ & & 2~s_{z_{h}}~g_{\gamma ww}~g_{z_{h}ww}~\left[c^{v}_{\gamma}c^{v}_{z_{h}}+c^{a}_{\gamma}c^{a}_{z_{h}}-p~(c^{a}_{\gamma}c^{v}_{z_{h}}+c^{a}_{z_{h}}c^{v}_{\gamma})\right]+\nonumber \\ & & 2~s_{z}~s_{z_{h}}~g_{zww}~g_{z_{h}ww}~\left.\left[c^{v}_{z}c^{v}_{z_{h}}+c^{a}_{z}c^{a}_{z_{h}}-p~(c^{a}_{z}c^{v}_{z_{h}}+c^{a}_{z_{h}}c^{v}_{z})\right]\right)\nonumber\\[2 mm ] c^{'}_{s}&=&(1-p_{e^{-}}p_{e^{+}})~\left(g^{2}_{\gamma ww}~\left[2c^{a}_{\gamma}c^{v}_{\gamma}-p~((c^{v}_{\gamma})^{2}+(c^{a}_{\gamma})^{2})\right]\right.+\nonumber \\ & & s^{2}_{z}~g^{2}_{zww}~\left[2c^{a}_{z}c^{v}_{z}-p((c^{v}_{z})^{2}+(c^{a}_{z})^{2})\right]+ \nonumber \\ & & s^{2}_{z_{h}}~g^{2}_{z_{h}ww}~\left[2c^{a}_{z_{h}}c^{v}_{z_{h}}-p~((c^{v}_{z_{h}})^{2}+(c^{a}_{z_{h}})^{2})\right]+\nonumber \\ & & 2~s_{z}~g_{\gamma ww}~g_{zww}~\left[(c^{a}_{z}c^{v}_{\gamma}+c^{a}_{\gamma}c^{v}_{z})-p~(c^{v}_{\gamma}c^{v}_{z}+c^{a}_{\gamma}c^{a}_{z})\right]+\nonumber \\ & & 2~s_{z_{h}}~g_{\gamma ww}~g_{z_{h}ww}~\left[(c^{a}_{z_{h}}c^{v}_{\gamma}+c^{a}_{\gamma}c^{v}_{z_{h}})-p~(c^{v}_{\gamma}c^{v}_{z_{h}}+c^{a}_{\gamma}c^{a}_{z_{h}})\right]+ \nonumber \\ & & 2~s_{z}~s_{z_{h}}~g_{z_{h}ww}~g_{zww}~\left.\left[(c^{a}_{z_{h}}c^{v}_{z}+c^{a}_{z}c^{v}_{z_{h}})-p~(c^{v}_{z}c^{v}_{z_{h}}+c^{a}_{z_{h}}c^{a}_{z})\right]\right)\nonumber\\[2 mm ] c_{int}&=&g^{2}_{e\nu w}~\left(g_{\gamma ww}~(c^{v}_{\gamma}+c^{a}_{\gamma})+s_{z}~g_{zww}~(c^{v}_{z}+c^{a}_{z})+s_{z_{h}}~g_{z_{h}ww}~(c^{v}_{z_{h}}+c^{a}_{z_{h}})\right ) \nonumber \\ & & \times(1-p_{e^{-}})(1+p_{e^{+}})\nonumber\\[2 mm ] c_{t}&=&\frac{g^{4}_{e\nu w}}{2}\times(1-p_{e^{-}})(1+p_{e^{+}})\nonumber\end{aligned}\ ] ] here , the effective polarization , @xmath101 , and the propagator factors are defined as : @xmath102 where @xmath103 is the mass of the corresponding gauge boson , @xmath104 .
we have assumed that the centre of mass energy is sufficiently far away from the threshold regions of the gauge bosons involved .
k. hagiwara , r. d. peccei , d. zeppenfeld and k. hikasa , nucl .
b * 282 * , 253 ( 1987 ) .
g. gounaris , j. layssac , g. moultaka and f. m. renard , int .
j. mod .
phys . a * 8 * , 3285 ( 1993 ) .
n. arkani - hamed , a. g. cohen , t. gregoire and j. g. wacker , jhep * 0208 * , 020 ( 2002 ) [ arxiv : hep - ph/0202089 ] . n. arkani - hamed , a. g. cohen , e. katz , a. e. nelson , t. gregoire and j. g. wacker , jhep * 0208 * , 021 ( 2002 ) [ arxiv : hep - ph/0206020 ] . n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) [ arxiv : hep - ph/0206021 ] . m. schmaltz and d. tucker - smith , ann .
rev . nucl . part .
* 55 * , 229 ( 2005 ) [ arxiv : hep - ph/0502182 ] .
m. perelstein , prog .
phys . * 58 * , 247 ( 2007 ) [ arxiv : hep - ph/0512128 ] .
p. poulose , pramana * 69 * , 909 ( 2007 ) .
g. a. moortgat - pick _
et al . _ ,
rept . * 460 * , 131 ( 2008 ) [ arxiv : hep - ph/0507011 ] .
m. perelstein , m. e. peskin and a. pierce , phys .
d * 69 * , 075002 ( 2004 ) [ arxiv : hep - ph/0310039 ] . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys .
d * 68 * , 035009 ( 2003 ) [ arxiv : hep - ph/0303236 ] .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys .
d * 67 * , 095004 ( 2003 ) [ arxiv : hep - ph/0301040 ] .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys . lett .
b * 563 * , 191 ( 2003 ) [ erratum - ibid .
b * 603 * , 257 ( 2004 ) ] [ arxiv : hep - ph/0302188 ] .
t. han , h. e. logan and l. t. wang , jhep * 0601 * , 099 ( 2006 ) [ arxiv : hep - ph/0506313 ]
. j. a. conley , j. l. hewett and m. p. le , phys .
d * 72 * , 115014 ( 2005 ) [ arxiv : hep - ph/0507198 ] .
y. b. liu , l. l. du and x. l. wang , j. phys .
g * 33 * , 577 ( 2007 ) [ arxiv : hep - ph/0609208 ] .
w. kilian , d. rainwater and j. reuter , phys .
d * 74 * , 095003 ( 2006 ) [ erratum - ibid .
d * 74 * , 099905 ( 2006 ) ] [ arxiv : hep - ph/0609119 ] .
e. asakawa _ et al .
_ , arxiv:0901.1081 [ hep - ph ] .
s. yang , arxiv:0904.1646 [ hep - ph ] .
a. dobado , l. tabares - cheluci , s. penaranda and j. r. laguna , arxiv:0907.1483 [ hep - ph ] . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys .
d * 67 * , 115002 ( 2003 ) [ arxiv : hep - ph/0211124 ] .
a. dobado , l. tabares - cheluci and s. penaranda , phys .
d * 75 * , 083527 ( 2007 ) [ arxiv : hep - ph/0612131 ] .
a. a. pankov and n. paver , phys . lett .
b * 324 * , 224 ( 1994 ) .
m. acciarri _ et al . _
[ l3 collaboration ] , phys .
b * 474 * , 194 ( 2000 ) [ arxiv : hep - ex/0001016 ] .
v. a. kovalchuk , m. p. rekalo and i. v. stoletnyi , z. phys .
c * 23 * , 111 ( 1984 ) .
p. poulose , s. d. rindani and l. m. sehgal , phys .
b * 525 * , 71 ( 2002 ) [ arxiv : hep - ph/0111134 ] .
b. ananthanarayan and s. d. rindani , phys .
lett . * 73 * , 1215 ( 1994 ) [ arxiv : hep - ph/9310312 ] .
b. ananthanarayan and s. d. rindani , phys .
d * 50 * , 4447 ( 1994 ) [ arxiv : hep - ph/9403346 ] .
b. ananthanarayan and s. d. rindani , phys .
d * 51 * , 5996 ( 1995 ) [ arxiv : hep - ph/9411399 ] . | we study the possibility of using w pair production and leptonic decay of one of the w s at the ilc with polarized beams as a probe of the littlest higgs model .
we consider cross - sections , polarization fractions of the w s , leptonic decay energy and angular distributions , and left - right polarization asymmetry as probes of the model .
with parameter values allowed by present experimental constraints detectable effects on these observables at typical ilc energies of 500 gev and 800 gev will be present .
beam polarization is further found to enhance the sensitivity . * w physics at the ilc with polarized beams as a probe of the littlest higgs model * + centre for high energy physics + indian institute of science + bangalore 560 012 , india + + department of physics + indian institute of technology guwahati + guwahati 781 039 , india + |
this work dedicated to the study of the so - called multilayer saint - venant system , which arises as an approximate model for the propagation of waves in the ocean or atmosphere , when density stratification can not be neglected .
we will refer to as _ free surface system _ the following first - order , quasilinear system of @xmath0 coupled evolution equations : @xmath1 here , the unknowns @xmath2 and @xmath3 represent respectively the deformation of the @xmath4 interface and the layer - averaged horizontal velocity in the @xmath4 layer , at time @xmath5 and horizontal position @xmath6 where @xmath7 ; see figure [ f.sketchofthedomain ] . if @xmath8 , then we denote @xmath9 and @xmath10 .
we denote by @xmath11 the mass density of the homogeneous fluid in the @xmath4 layer , whereas @xmath12 is the gravitational acceleration .
finally , @xmath13 is the depth of the @xmath4 layer . by convention
, we set @xmath14 ( above the upper free surface is vacuum ) , and @xmath15 ( the bottom is flat ) .
we restrict ourselves to the setting of stable stratification , namely @xmath16 we may rescale the variables so as to replace the factor @xmath17 more precisely , we use the following nondimensionalization : @xmath18 where @xmath19 is a characteristic horizontal length ( say the wavelength of the flow ) , @xmath20 is a characteristic vertical length ( say the typical depth of one layer at rest ) , and @xmath21 is a velocity .
such that @xmath22 , then @xmath23 measures the typical velocity of propagation of the baroclinic modes ; see appendix [ s.spectral ] . ] although @xmath24 does not appear in our system , it plays a very important role as the saint - venant system may be seen as the first - order asymptotic model obtained from the full multilayer water - wave system in the limit @xmath25 ; see @xcite in the one - layer case and @xcite in the bilayer ( albeit irrotational ) setting .
it may also be formally obtained using the hydrostatic and columnar motion assumptions ; see @xcite . in this work ,
we ask + * qn : * _ what is the behaviour of the solutions to in the limit @xmath26 ? _
the first observation is that the velocity evolution equations become singular , as @xmath27 since @xmath28 by convention , so that even the existence of solutions on a non - trivial time interval is far from straightforward . at the linear level ,
it is known @xcite ( see also appendix [ s.spectral ] ) that the flow may be decomposed into @xmath29 modes , propagating as linear wave equations with distinct velocities . in our setting , the first mode , _
i.e. _ the barotropic mode , propagates much faster than the other , baroclinic modes , in the sense that the former is typically of size @xmath30 while the latter are uniformly bounded when @xmath31 ; hence the singularity .
while such a decomposition is exact only in the linear setting , we show in this work that the flow behaves in a similar way in the weak density contrast limit even when strong nonlinearities are present , provided that the initial surface deformation is small , the depth of each layer remains positive and shear velocities are not too large . roughly speaking ,
we show that one may then approximate the flow as the superposition of rapidly propagating acoustic waves and a non - singular `` slow '' mode with non - trivial dynamics .
let us be more precise .
asymptotically , the fast mode describes the propagation of the free surface , @xmath32 , and total volume flux , namely @xmath33 where @xmath34 and @xmath35 is the orthogonal projection onto irrotational vector fields .
one has at first order , in dimension @xmath8 and provided that @xmath32 and @xmath36 are initially balanced so that @xmath37 , the linear acoustic system @xmath38 the slow component contains all the baroclinic modes which interact strongly one with each other in the nonlinear setting .
we show that it is asymptotically described by the _ rigid - lid model _ , which is obtained from the free - surface system by setting @xmath39 in the mass conservation equations , and replacing @xmath40 with @xmath41 in the velocity evolution equations ( see @xcite ) .
in addition , we apply the so - called _ boussinesq approximation _ ( see _ e.g. _
@xcite ) to the limit system , that is we set @xmath42 in the velocity evolution equations while @xmath43 remains fixed and non - trivial .
this yields @xmath44 with the same notation as before , except @xmath45 and @xmath46 .
in particular , the first equation is not an evolution equation but a constraint ( conservation of total mass ) , namely @xmath47 physically speaking , @xmath48 is ( up to a constant ) the pressure at the flat , rigid lid . from a mathematical viewpoint
, @xmath49 is the lagrange multiplier associated with the above divergence - free constraint .
it may be reconstructed from the knowledge of @xmath50 by solving the poisson equation @xmath51 we thus offer a rigorous justification ( the formal justification is generally attributed to armi @xcite ) of the widely used rigid - lid and boussinesq approximations for free surface multilayer shallow - water flows with a small density contrast . before stating our main results ,
let us recall for convenience the system of equations at stake : @xmath52 where @xmath53 with convention @xmath54 and @xmath55 .
in the following , we fix parameters @xmath56 ; and denote @xmath57 we can then reconstruct @xmath58 with @xmath59 ( @xmath60 ) and @xmath61 where @xmath62 is the only parameter allowed to vary , and by assumption @xmath63 it is also convenient to denote ( notice the @xmath64 prefactor ; see remark [ r.small-surface ] below ) @xmath65 so that the control of @xmath66 in sobolev space @xmath67 ( see appendix [ s.notations ] for notations ) yields , when @xmath8 , @xmath68 let us now state the main results of this work .
[ t.wp ] let @xmath69 and @xmath70 be such that @xmath71 where @xmath72 , and we recall the convention @xmath54
. one can set @xmath73 such that if @xmath74 satisfies additionally @xmath75 then there exists @xmath76 and a unique @xmath77 strong solution to and @xmath78 , and satisfying , with @xmath79 and @xmath80 for any @xmath81 .
moreover , one has @xmath82 uniformly with respect to @xmath83 .
[ t.convergence ]
let @xmath8 , @xmath69 , and @xmath84 as above .
then there exists @xmath76 with @xmath85 and , for @xmath62 sufficiently small , * @xmath86 a unique strong solution to and @xmath78 ; * @xmath87 a unique strong solution to with initial data @xmath88 where @xmath89 is the total depth , @xmath90 with convention @xmath91 and @xmath35 the orthogonal projection onto irrotational vector fields . *
@xmath92 a unique strong solution to with @xmath93 moreover , one has for any @xmath94 , @xmath95 where @xmath96 .
[ r.small-surface ] as mentioned in the introduction , our hypotheses contain a smallness assumption on the initial deformation of the surface , namely @xmath97 .
this assumptions is natural so as to balance the contributions in the ( preserved in time ) energy : @xmath98 without this assumption , the flow possesses a strongly nonlinear barotropic component , and energy methods yield a well - posedness theory over a small time - domain , @xmath81 , @xmath99 ; see proposition [ p.wp-naive ] and remark [ r.naive ] , below . on this timescale ,
the baroclinic component do not evolve , so that all the dynamics is described by the barotropic component ( asymptotically as @xmath100 ) .
the requirement @xmath83 does not lose in generality in theorem [ t.wp ] : the case of non - small @xmath62 follows from the standard well - posedness theory of quasilinear systems , as proved by monjarret in @xcite and stated in proposition [ p.wp-naive ] , below . our proof does not rely on , but rather provides , the existence and uniqueness of strong solutions of the limit ( rigid - lid ) system . in that respect ,
one may see the free - surface system as a penalized model for relaxing the rigid - lid constraint .
sharper well - posedness results for the rigid - lid system in the two - layer case and without the boussinesq approximation are provided in @xcite .
theorem [ t.convergence ] is restricted to @xmath8 because we use dispersive decay estimates on rapidly propagating acoustic waves in order to control nonlinear coupling effects between the fast and slow modes . in the case of dimension @xmath101 , and provided that the initial data is sufficiently localized in space , we justified in @xcite a similar mode decomposition of the flow , by making use of the different spatial support of each mode after small time .
proposition 4.4 therein , together with proposition [ p.energy-estimate-hs ] in the present work , offer a convergence between the exact and the approximate solution with rate @xmath102 .
the same convergence rate holds in the case of dimension @xmath8 and well - prepared initial data , in the sense of proposition [ p.well-prepared initial data ] . in the following , for the sake of simplicity , we limit our study to the case of dimension @xmath8 , although we find it more telling to keep the notation @xmath103 .
the proof of theorem [ t.wp ] is easily adapted to the case of dimension @xmath101 .
one could add , without any additional difficulty , a uniformly bounded and order - zero term to the system , so as to take into account for instance the coriolis force , atmospheric pressure variations , or bottom topography .
notice however that these terms should be of size @xmath102 in the evolution equation for @xmath32 ; in particular , only small topography may be dealt with using directly our strategy .
similarly , except in the one - layer case where the the component due to coriolis effect is an anti - symmetric perturbation of a symmetric system , one can not allow a rapid rotation such as in the quasi - geostrophic regime , which would correspond here to @xmath104 where @xmath105 is the rossby number ; see @xcite .
our results are valid for arbitrary @xmath29 , but not uniformly
. in particular , we can not control the dependence of @xmath106 as @xmath29 grows .
thus our strategy can not be adapted to study the system in the limit @xmath107 , corresponding to the physically relevant situation of continuous stratification .
a similar shortcoming was already noticed and discussed by ripa in @xcite .
a well - posedness result on system is stated and proved in @xcite .
it follows from a standard analysis on quasilinear systems since a symbolic symmetrizer may be exhibited ( see section [ s.wp-naive ] below ) .
however , due to the presence of singular components in the system , the _ a priori _ maximal time of existence of the solutions may be bounded from below only as @xmath108 using this method . such a result is unsuitable for our purpose as the time interval shrinks to zero in the considered limit , and is inconsistent with oceanographic observations of large amplitude internal waves propagating over long distances ; see _ e.g. _ @xcite and references therein . in order to go beyond this analysis and provide a time of existence of solutions uniformly bounded from below with respect to @xmath62 small , we need to take advantage of some additional structural properties of the system . this structure is put to light by a suitable change of variable , which we describe below .
let us introduce the shear velocities , @xmath109 , and the total horizontal momentum , @xmath110 , as follows : @xmath111 so that , conversely , for any @xmath112 , @xmath113 one may rewrite system using the new variables as follows : @xmath114 where @xmath115 , and @xmath116 are meant as the expressions in terms of @xmath117 given in .
the above change of variables may be seen as an approximate normal form allowing to decouple the slow and fast components of the flow . indeed , since @xmath118 one sees immediately that the singular terms appear only as linear components on the evolution equations for @xmath119 and @xmath110 or more precisely @xmath33 and involve only @xmath119 and @xmath33 .
in other words , the leading - order terms form a system of rapidly propagating acoustic waves in @xmath120 : @xmath121 the remainder contains quasilinear components depending on both the fast ( @xmath122 ) and slow ( @xmath123 ) variables , so we need to consider the full system of equations in order to obtain the desired uniform energy estimates .
let us now , for the sake of simplicity , restrict our discussion to the case of dimension @xmath101 . from the above , we may rewrite the system as @xmath124\partial_x{{\sf v}}=\partial_t{{\sf v}}+ \frac1{\varrho}{{\sf l}}\partial_x{{\sf v}}+ { { \sf c}}[{{\sf v}}]\partial_x{{\sf v}}=0\ ] ] where @xmath125 , so that @xmath126 represents the above acoustic wave system for the fast variables , while @xmath127 ; and @xmath128 $ ] contains lower - order ( in terms of @xmath62 ) and coupling terms .
we shall make use of the fact that one can construct a `` good '' symmetrizer of the system under the form , namely we exhibit real , positive - definite matrices @xmath129 $ ] such that @xmath129=({{\sf t}}[{{\sf v}}])^\top$ ] and @xmath129{{\sf b}}[{{\sf v}}]=({{\sf t}}[{{\sf v}}]{{\sf b}}[{{\sf v}}])^\top$ ] , and satisfying the decomposition @xmath130={{\sf t}}_{(0)}+{{\sf
t}}_{(1)}[{({{\sf id}}-{\sf \pi}_{\rm f})}{{\sf v}}]{({{\sf id}}-{\sf \pi}_{\rm f})}+{\mathcal{o}}({\varrho}),\ ] ] where @xmath131 is the orthogonal projection onto @xmath132 , the slow variables
. indeed , one obtains an energy estimate by taking the @xmath133 inner - product of the equation with @xmath129{{\sf v}}$ ] , which only requires to estimate @xmath134{{\sf b}}[{{\sf v } } ] ) } _ { l^\infty}+\norm { \partial_t ( { { \sf t}}[{{\sf v } } ] ) } _ { l^\infty}.\ ] ] using that @xmath135 and @xmath136 are constant operators and that @xmath137\partial_x{{\sf v}}={\mathcal{o}}(1)$ ] , we see that the above are estimated uniformly with respect to @xmath62 small ; thus we have a uniform control of the @xmath133 norm .
the corresponding @xmath138 estimate with @xmath69 does not bring additional difficulties , using that @xmath135 commutes with the fourier multiplier @xmath139 .
there remains to understand why such symmetrizer exists for our system .
one could check , after tedious calculations , that the explicit one provided by monjarret in @xcite ( after applying the congruent transformation associated with the change of variables ) satisfies the necessary hypotheses , but we offer in appendix [ s.spectral ] an alternative and more robust construction .
we show that , provided that @xmath140 satisfies , , then @xmath141 $ ] has @xmath142 real and distinct eigenvalues .
two of them are asymptotically equivalent to @xmath143 as @xmath100 while the other ones are uniformly bounded with respect to @xmath62 small .
the spectral projection corresponding to the former converge towards the projections onto the eigenspaces corresponding the two non - trivial eigenvalues of @xmath135 . using the scale separation between the eigenvalues
, one shows that the spectral projections corresponding to the latter are uniformly bounded , and that they converge as @xmath144 towards independent , rank - one projections onto subspaces of @xmath132 .
our symmetrizer is then , classically , @xmath130{\stackrel{\rm def}{=}}\sum_{j=1}^{2n } ( { { \sf p}}_j[{{\sf v}}])^\top { { \sf p}}_j[{{\sf v}}]\ ] ] where @xmath145 $ ] is the spectral projection onto the @xmath146 eigenspace of @xmath147 $ ] . that @xmath129 $ ] enjoys the desired properties follows , using standard perturbation theory @xcite , from the fact that @xmath135 is constant , @xmath148 and the strong scale separation between @xmath149 and uniformly bounded eigenvalues .
an additional difficulty arises in the situation of horizontal dimension @xmath8 , due to the fact that the symmetrizer of the system which is constructed from the symmetrizer in dimension @xmath101 and a rotational invariance property is only a symbolic symmetrizer , as opposed to symmetrizers in the sense of friedrichs .
thus we rely on para - differential calculus , but extra care must be given to `` lower order terms '' in the sense of regularity , which may effectively hurt our energy estimates if they are not uniformly bounded with respect to @xmath62 . as a matter of fact , we use that one can construct an explicit operator ( defined as a fourier multiplier ) which symmetrizes the linear , singular contributions of the system , and use para - differential calculus only on the next order components in terms of @xmath62 .
given the uniform ( with respect to @xmath62 small ) energy estimates , the large time well - posedness ( theorem [ t.wp ] ) follows from the standard theory for quasilinear hyperbolic systems .
the convergence results ( theorem [ t.convergence ] as well as additional assertions in section [ s.convergence ] ) proceed from rather standard techniques in the study of singular systems ; see references below .
in @xcite , the author studied the so - called inviscid bilayer saint - venant ( or shallow water ) system in the limit of small density contrast .
the change of variables allowing for uniform energy estimates was exhibited therein , and convergence towards a solution of the rigid - lid limit , as well as a second - order approximation , was deduced in the case of well - prepared initial data .
this work is therefore an extension of these results to the situation of ( horizontal ) dimension @xmath8 , ill - prepared initial data as well as arbitrary number of layers .
as already noticed in the aforementioned work , our problem has many similarities with the ( two - dimensional ) incompressible limit for euler equations , as studied initially in @xcite . as a matter of fact ,
if we consider only one layer of fluid , then our problem corresponds exactly to a special case of the isentropic incompressible limit , and we recover the results of ukai @xcite and asano @xcite .
we will not detail the very rich history of results concerning this problem ( we let the reader to @xcite for comprehensive reviews ) but rather aim at pointing out similarities and differences of our situation .
let us first recall the two - dimensional isentropic euler equations for inviscid , barotropic fluids : @xmath150 where @xmath151 is a given pressure law , @xmath152 is the density , @xmath153 the velocity , and @xmath154 the dimensionless mach number . as claimed above , one recognizes exactly in the one layer setting ( @xmath155 ) , by setting @xmath156 , and identifying @xmath157 of course , the difficulty in our case is that , as one considers additional layers of fluids , these equations are coupled with additional equations on additional unknowns , so as to produce a full quasilinear system . since these additional equations are non - singular with respect to the small parameter , it is tempting to compare our situation with the incompressible limit for the non - isentropic euler equations , where is coupled with an additional evolution equation for the entropy , @xmath158 : @xmath159 and @xmath160 .
our situation , however , is quite different .
this can be seen from the fact that , contrarily to the non - isentropic euler equations , the linearized system is balanced , in the sense that a small perturbation of the `` slow '' component of the reference state induces only a small deviation for the solution . in other words , using the notation of the above discussion , the symmetrizer of the non - isentropic euler equations does not satisfy @xmath129={{\sf t}}_{(0)}+{{\sf t}}_{(1)}[{({{\sf id}}-{\sf \pi}_{\rm f})}{{\sf v}}]{({{\sf id}}-{\sf \pi}_{\rm f})}+{\mathcal{o}}({\varrho})$ ] but only @xmath129={{\sf t}}_{(0)}+{{\sf t}}_{(1)}[{({{\sf id}}-{\sf \pi}_{\rm f})}{{\sf v}}]+{\mathcal{o}}({\varrho})$ ] ; see discussion in @xcite .
this additional property in our situation allows in particular to straightforwardly deduce @xmath138 energy estimates from the corresponding @xmath133 energy estimate ; and to obtain the strong convergence result of theorem [ t.convergence ] simply from dispersive estimates on the `` acoustic '' component of the flow , as originally carried out by ukai @xcite and asano @xcite in the isentropic case . in order to deal with this situation , mtivier and schochet @xcite
( see also @xcite ) rely on the fact that their system enjoys a diagonal block structure and that the symmetrizer commutes exactly with the singular operator , denoted @xmath161 in the above discussion
. roughly speaking , this means that we may control the compressible and isentropic component of the flow independently of the acoustic component by simply projecting onto @xmath132 .
since such assumptions are only approximately satisfied in our situation , our system is rather related to the `` @xmath62-balanced '' ( and not `` @xmath62-diagonal '' ) systems studied by klainerman and majda @xcite , although we do not restrict ourselves to well - prepared initial data . in this spirit ,
our proofs rely as little as possible on explicit calculations , thus we expect that the general strategy may be successfully applied to other situations and other frameworks , such as the ones presented in @xcite .
this is why we use mostly in the following the terminology of `` fast _ vs _ slow '' mode / component instead of `` barotropic _ vs _ baroclinic '' which is more relevant to our initial oceanographic motivation ( see @xcite ) ; or `` acoustic _ vs _ incompressible '' associated with euler equations . in section [ s.wp-naive ]
, we recall that our quasilinear system admits an explicit ( symbolic ) symmetrizer , which yields immediately a well - posedness theory for , for any fixed @xmath162 . in section [ s.normal-form ] , we exhibit the structural properties enjoyed by our system , after the change of variables , , and which allow the uniform ( with respect to @xmath62 small ) energy estimates provided in section [ s.energy-estimates-l2 ] ( proposition [ p.energy-estimate-l2 ] ) and section [ s.energy-estimates-hs ] ( proposition [ p.energy-estimate-hs ] ) .
we deduce in section [ s.wp ] the large - time well - posedness result of theorem [ t.wp ] .
additionally , we show in proposition [ p.well-prepared initial data ] that an assumption of well - prepared initial data is propagated by the flow for positive times .
section [ s.convergence ] is dedicated to convergence results .
we first state in proposition [ p.weak-convergence ] a weak convergence result for the solutions of the free - surface system as @xmath100 . as in the incompressible limit for euler equations ,
the convergence can not be strong uniformly in time , due to the rapidly propagating fast mode .
however , we show in proposition [ p.convergence-wp ] that this small initial layer in time vanishes in the case of well - prepared initial data , and then characterize the defect for general initial data in section [ s.convergence-strong ] , yielding theorem [ t.convergence ] .
appendix [ s.notations ] contains a description of some notations used throughout the text , as well as a short review of standard results concerning product and commutator estimates in sobolev spaces ( section [ s.product ] ) and bony s paradifferential calculus ( section [ s.para ] ) .
finally , appendix [ s.spectral ] is dedicated to some results on the eigenstructure of our system , which are used in section [ s.normal-form ] .
in this section , we fix @xmath162 and construct an explicit ( symbolic ) symmetrizer of .
this offers a well - posedness result similar to theorem [ t.wp ] , although non - uniformly with respect to @xmath62 small .
this analysis has been provided by monjarret in @xcite ; we recall it here for the sake of completeness and because the objects defined therein will be of later use .
let us first rewrite the free - surface system with a matricial , compact formulation .
provided that @xmath163 with @xmath164 , one can rewrite equivalently as @xmath165\partial_x { { \sf u}}+\frac1{\varrho}{{\sf a}}^y[{{\sf u}}]\partial_y { { \sf u}}= { \mathbf{0}},\ ] ] with @xmath166{\stackrel{\rm def}{=}}\left(\begin{mat}(b){c : c : c } { \sf m}(u^x ) & { \sf h}(\zeta ) & { \sf 0}_n\\ : { \sf r } & { { \sf d}}(u^x)&{\sf 0}_n \\ : { \sf 0}_n & { \sf 0}_n & { { \sf d}}(u^x ) \\
\end{mat}\right ) , \qquad \frac1{\varrho}{{\sf a}}^y[{{\sf u}}]{\stackrel{\rm def}{=}}\left(\begin{mat}(b){c : c : c } { \sf m}(u^y ) & { \sf 0}_n & { \sf h}(\zeta ) \\ : { \sf 0}_n & { { \sf d}}(u^y ) & { \sf 0}_n\\ : { \sf r } & { \sf 0}_n & { { \sf d}}(u^y ) \\ \end{mat}\right).\ ] ] here and thereafter , we heavily make use of the block structure of @xmath0-by-@xmath0 matrices .
we denote by @xmath167 the @xmath29-by-@xmath29 matrix with only zero entries , and for @xmath168 , @xmath169 .
moreover , @xmath170 are upper - triangular and @xmath171 is lower - triangular and are defined by @xmath172 [ p.wp-naive ] let @xmath69 and @xmath70 be such that @xmath173 where we recall that @xmath174 by convention .
one can set @xmath175 such that if @xmath74 satisfies additionally @xmath176 then there exists a unique @xmath177 and @xmath178 , maximal solution to and @xmath78 .
moreover , if @xmath179 , then @xmath180 ( @xmath181 ) or one of the hyperbolicity conditions , ceases to be true .
[ r.naive ] naively following the above strategy and keeping track of the dependency of constants with respect to the parameter @xmath62 would yield a disappointing lower bound on the maximal time of existence , namely @xmath182 , even when the initial surface deformation is assumed small as in theorem [ t.wp ] .
it is the main result of this work that , in this case , the well - posedness theory and uniform energy estimates can be extended to a non - shrinking time domain as @xmath100 . the conditions ,
ensure that the symmetrizer , defined in below , is coercive .
it is therefore a sufficient condition for hyperbolicity . except in very specific cases
, one has very few information on the domain of hyperbolicity ; see discussion in appendix [ s.spectral ] .
the fact that requires a control on the shear velocities , @xmath183 rather than on the velocities themselves is allowed by some freedom in the choice of the symmetrizer .
notice in particular that the hessian of the conserved energy yields a natural symmetrizer for our system of conservation laws in the case of irrotational flows , but it does not enjoy the desired property .
the ability to construct symmetrizers depending strongly on the shear velocities but only weakly on a background velocity ( or on the total volume flux ) is also essential for us to obtain results outside of the scope of well - prepared initial data ; see @xcite for an analysis where this property is not used .
let us introduce the symbol of : @xmath184{\stackrel{\rm def}{=}}\xi^x { { \sf a}}^x[{{\sf u}}]+\xi^y { { \sf a}}^y[{{\sf u}}].\ ] ] an important property of our system is that is satisfies _ rotational invariance _
* section 1.2 ) .
more precisely , one can easily check that @xmath185={{\sf q}}(\xi)^{-1}{{\sf a}}^x[{{\sf q}}(\xi){{\sf u}}]{{\sf q}}(\xi)|\xi|,\ ] ] where @xmath186 where @xmath187 is the @xmath29-by-@xmath29 identity matrix and @xmath188 .
obviously , @xmath189 is homogeneous of degree @xmath190 in @xmath191 , with entries in @xmath192 and is orthogonal : @xmath193 .
this allows to construct a ( symbolic ) symmetrizer of the system from a ( friedrichs ) symmetrizer of @xmath194 alone .
more precisely , define @xmath195= \left(\begin{mat}(b){c : c : c } { \sf d}({\tilde}r ) & { \sf l}^\top & { \sf 0}_n \\ : { \sf l } & { \sf d(\gamma)}{\sf d}(h ) & { \sf 0}_n \\ : { \sf 0}_n & { \sf 0}_n & { \sf d(\gamma)}{\sf d}(h ) \\ \end{mat}\right)\ ] ] with @xmath196 and @xmath197{\sf \delta}{\sf d}[{\textbf{e}}_{\varrho } ] $ ] where @xmath198 , @xmath199 and the parameter @xmath200 may be chosen freely in @xmath201 . that @xmath202{{\sf a}}^x[{{\sf u}}]$ ] is symmetric is easily checked once we notice the following identities in : @xmath203 , @xmath204 and @xmath205 .
it follows that @xmath206{\stackrel{\rm def}{=}}{{\sf q}}(\xi)^{-1}{{\sf s}}^x[{{\sf q}}(\xi){{\sf u}}]{{\sf q}}(\xi)$ ] is a symbolic symmetrizer of , since @xmath207{{\sf a}}[{{\sf u}},\xi]={{\sf q}}(\xi)^\top{{\sf s}}^x[{{\sf q}}(\xi){{\sf u}}]{{\sf a}}^x[{{\sf q}}(\xi){{\sf u}}]{{\sf q}}(\xi)|\xi|\ ] ] is obviously symmetric .
finally , one may choose @xmath208 for instance , so that as soon as @xmath66 satisfies , with @xmath209 sufficiently small , then @xmath210 $ ] is strictly diagonally dominant with positive diagonal entries , and therefore definite positive , uniformly for any @xmath211 .
proposition [ p.wp-naive ] is now a direct consequence of the standard theory of first - order hyperbolic quasilinear systems ; see @xcite for instance .
in this section , we establish uniform ( with respect to @xmath62 small ) energy estimates , which are the essential ingredients in the proof of our main results .
we first exhibit in section [ s.normal-form ] some properties of the system obtained after the change of variables - , and which allow the @xmath133 energy estimate of proposition [ p.energy-estimate-l2 ] ( section [ s.energy-estimates-l2 ] ) and in turn the @xmath138 energy estimate of proposition [ p.energy-estimate-hs ] ( section [ s.energy-estimates-hs ] ) . in
what follows , we fix @xmath212 and always assume that @xmath213 ( recall the convention : @xmath174 ) .
more precisely , we work with @xmath214 defined by @xmath215 consequently , the change of variables - define self - homeomorphisms between @xmath216 and we denote @xmath217 consider the jacobian matrix associated to @xmath218 : @xmath219{\stackrel{\rm def}{=}}\left(\begin{mat}(b){c : c : c } { \sf i}_n & { \sf 0}_n & { \sf 0}_n \\ : { \sf c}(u^x ) & { \sf \delta}_h(\zeta ) & { \sf 0}_n \\ : { \sf c}(u^y ) & { \sf 0}_n & { \sf \delta}_h(\zeta ) \\ \end{mat}\right)\ ] ] where @xmath220 from the inverse function theorem , one has @xmath221= ( { { \sf j}}^{f^{-1}}[f({{\sf v}})])^{-1}\ ] ] and @xmath222)^{-1 } = \left(\begin{mat}(b){c : c : c } { \sf i}_n & { \sf 0}_n & { \sf 0}_n \\ : -{\sf \delta}_h(\zeta)^{-1 } { \sf c}(u^x ) & { \sf \delta}_h(\zeta)^{-1 } & { \sf 0}_n \\ : -{\sf \delta}_h(\zeta)^{-1 } { \sf c}(u^y ) & { \sf 0}_n & { \sf \delta}_h(\zeta)^{-1 } \\ \end{mat}\right)\ ] ] where ( as is easier seen directly from ) , @xmath223 applying the change of variables @xmath224 in yields , for sufficiently regular functions ( see lemma [ l.utov ] below ) , @xmath225\partial_x { { \sf v}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_x { { \sf v}}={\mathbf{0}},\ ] ] which we can identify with , and where @xmath226,{{\sf b}}^y[{{\sf v}}]$ ] are explicitly given in terms of @xmath227,{{\sf a}}^y[{{\sf v}}]$ ] ( displayed in ) and @xmath228 $ ] through @xmath229 = ( { { \sf j}}^f[{{\sf v}}])^{-1}{{\sf a}}^x[f({{\sf v } } ) ] { { \sf j}}^f[{{\sf v } } ] \quad \text { and } \quad { { \sf b}}^y[{{\sf v } } ] = ( { { \sf j}}^f[{{\sf v}}])^{-1}{{\sf a}}^y[f({{\sf v } } ) ] { { \sf j}}^f[{{\sf v } } ] .\ ] ] finally , we denote by @xmath230 the orthogonal projection onto the `` fast '' variables , namely @xmath231 with @xmath232 and @xmath233 .
the following result is now straightforward .
[ l.j ] let @xmath234 , and recall the definition of @xmath189 in
. then @xmath235= { { \sf j}}^f[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi).\ ] ] moreover , @xmath228,({{\sf j}}^f[{{\sf v}}])^{-1}:{\mathbb{v}_{h_0}}\to { \mathcal{m}}_{n(1+d)}({\mathbb{r}})$ ] are well - defined and smooth and satisfy [ j - estimates ] @xmath236 } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \quad ; \quad \norm{({{\sf j}}^f[{{\sf v}}])^{-1 } } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ; \\\label{j - estimate-3 } \norm{{{\sf j}}^f[{{\sf v}}]-{{\sf j}}^f[{({{\sf id}}-{\sf \pi}_{\rm f})}{{\sf v } } ] } \leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \norm{{{\sf v}}}.\end{gathered}\ ] ] only the last estimate requires an explanation .
remark that the first variable of @xmath140 contributes to @xmath237 only through @xmath238 .
thanks to the @xmath62 prefactor , one deduces that @xmath239 where we denoted @xmath240 .
similarly , by , one has @xmath241 and , by definition , @xmath242 , so that @xmath243 where @xmath244 represents the velocity variables of @xmath245 , as given by .
the result is now clear .
[ l.b ] the functions @xmath246 are well - defined and smooth and satisfy the rotational invariance @xmath247{\stackrel{\rm def}{=}}\xi^x { { \sf b}}^x[{{\sf v}}]+\xi^y { { \sf b}}^y[{{\sf v}}]={{\sf q}}(\xi)^{-1}{{\sf b}}^x[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi)|\xi|;\ ] ] as well as the following estimates : [ b - estimates ] @xmath248}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) , \quad \norm{{({{\sf id}}-{\sf \pi}_{\rm f}^x)}{{\sf b}}^x[{{\sf v}}]}\leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ; \label{b - estimates-1}\\ \norm{{{\sf b}}^x[{{\sf v}}_1]-{{\sf b}}^x[{{\sf v}}_2]}\leq { \varrho}\ c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}_1},\norm{{{\sf v}}_2})\norm{{{\sf v}}_1-{{\sf v}}_2}. \label{b - estimates-4}\end{gathered}\ ] ] identity is deduced from , and : @xmath249+\xi^y { { \sf b}}^y[{{\sf v}}]&= ( { { \sf j}}^f[{{\sf v}}])^{-1 } ( \xi^x { { \sf a}}^x[f({{\sf v } } ) ] + \xi^y { { \sf a}}^y[f({{\sf v } } ) ] ) { { \sf j}}^f[{{\sf v } } ] \\ & = ( { { \sf j}}^f[{{\sf v}}])^{-1 } { { \sf q}}(\xi)^{-1}{{\sf a}}^x[{{\sf q}}(\xi)f({{\sf v } } ) ] { { \sf q}}(\xi ) { { \sf j}}^f[{{\sf v}}]|\xi|\\ & = { { \sf q}}(\xi)^{-1}({{\sf j}}^f[{{\sf q}}(\xi){{\sf
v}}])^{-1}{{\sf a}}^x[f({{\sf q}}(\xi ) { { \sf v } } ) ] { { \sf j}}^f[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi)|\xi|\\ & = { { \sf q}}(\xi)^{-1}{{\sf b}}^x[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi)|\xi| . \end{aligned}\ ] ] estimates may be deduced from identity and the explicit expressions of @xmath250 $ ] and @xmath228 , \
( { { \sf j}}^f[{{\sf v}}])^{-1}$ ] given in , , .
they are also apparent when identifying with .
in particular , one immediately sees that the evolution equations for @xmath251 are non - singular ( for @xmath62 small ) , and the singular term on the evolution equation for @xmath252 involves only @xmath253 , so that @xmath254}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}})\ ] ] and @xmath255-\frac1{\varrho}{({{\sf id}}-{\sf \pi}_{\rm f}^x)}{{\sf b}}^x[{{\sf v}}_2]}\leq \ c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}_1},\norm{{{\sf v}}_2})\norm{{{\sf v}}_1-{{\sf v}}_2}.\ ] ] the only singular terms ( for @xmath62 small ) arise from the first and last equations , which read @xmath256\partial_x{{\sf v}}+r_1^y[{{\sf v}}]\partial_y{{\sf v}},\ ] ] where @xmath257 are smooth and enjoy the same estimates as @xmath258 above .
we now only need to remark that @xmath259 ( by definition ) and , since @xmath242 , @xmath260 where , again , @xmath261 is smooth and uniformly estimated as above .
lemma [ l.b ] is now straightforward .
finally , let us provide some estimates on the symbolic symmetrizer of the system .
[ l.t ] there exists @xmath262 a smooth function such that @xmath263)^\top={{\sf t}}^x[{{\sf v}}]\quad \text { and } \quad ( { { \sf t}}^x[{{\sf v}}]{{\sf b}}^x[{{\sf v}}])^\top={{\sf t}}^x[{{\sf v}}]{{\sf b}}^x[{{\sf v}}].\ ] ] moreover , one can set @xmath264 such that if @xmath83 and @xmath265 satisfy @xmath266 then one has the following estimates : [ t - estimates ] @xmath267 , \quad \norm{{{\sf t}}^x[{{\sf v}}]}\leq c_0 , \quad \text { with } c_0^{-1}=n(1+d),\ c_0=c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ; \label{t - estimates-1}\\ \norm{{{\sf t}}^x[{{\sf v}}_1]-{{\sf t}}^x[{{\sf v}}_2]}+\frac1{\varrho}\norm{({{\sf t}}^x[{{\sf v}}_1]-{{\sf t}}^x[{{\sf v}}_2]){{\sf \pi}_{\rm f}^x}}\leq c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}_1},\norm{{{\sf v}}_2 } ) \norm{{{\sf v}}_1-{{\sf v}}_2 } ; \label{t - estimates-2}\\ \norm{{{\sf t}}^x[{{\sf v}}_1]-{{\sf t}}^x[{{\sf v}}_2]}\leq c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}_1},\norm{{{\sf v}}_2 } ) \left(\norm{{({{\sf id}}-{\sf \pi}_{\rm f})}({{\sf v}}_1-{{\sf v}}_2)}+{\varrho}\norm{{{\sf v}}_1-{{\sf v}}_2}\right ) .
\label{t - estimates-3}\end{gathered}\ ] ] one could define the symmetrizer as @xmath268= ( { { \sf j}}^f[{{\sf v}}])^\top { { \sf s}}^x[f({{\sf v } } ) ] { { \sf j}}^f[{{\sf v } } ] , $ ] where @xmath269 $ ] is the symmetrizer associated with @xmath250 $ ] and has been displayed in , and check the properties directly on this explicit symmetrizer . some of the estimates , however , rely on delicate cancellations which have no obvious explanation using this method .
this is why we find it more instructive to construct our symmetrizer using the spectral properties of our system , whose study we postpone to appendix [ s.spectral ] for the sake of readability .
it is proved in lemmas [ l.b - trivialeigen ] and [ l.b - eigen ] that @xmath226 $ ] has only real and semisimple eigenvalues : @xmath270 = \sum_{n=1}^n \mu_{n}[{{\sf v } } ] { { \sf p } } _ { n}[{{\sf v}}]+ \mu_{-n}[{{\sf v } } ] { { \sf p}}_{-n}[{{\sf v } } ] + u^x_n[{{\sf v } } ] { { \sf p}}^0_n[{{\sf v}}],\ ] ] where @xmath271 , u^x_n[{{\sf v } } ] \in { \mathbb{r}}$ ] , and @xmath272 , { { \sf p}}^0_n[{{\sf v}}]$ ] are rank - one spectral projections .
we now define @xmath273 = \sum_{n=1}^n ( { { \sf p } } _ { n}[{{\sf v}}])^\top { { \sf p } } _ { n}[{{\sf v}}]+ ( { { \sf p } } _ { -n}[{{\sf v}}])^\top { { \sf p}}_{-n}[{{\sf v}}]+({{\sf p}}^0_n[{{\sf v}}])^\top { { \sf p}}^0_n[{{\sf v}}].\ ] ] indeed , @xmath274 $ ] is obviously symmetric , and so is @xmath275{{\sf b}}^x[{{\sf v}}]= \sum_{n=1}^n \mu_{n}[{{\sf v } } ] ( { { \sf p } } _ { n}[{{\sf v}}])^\top { { \sf p } } _ { n}[{{\sf v}}]+ \mu_{-n}[{{\sf v } } ] ( { { \sf p } } _ { -n}[{{\sf v}}])^\top { { \sf p}}_{-n}[{{\sf v } } ] + u^x_n[{{\sf v } } ] ( { { \sf p}}^0_n[{{\sf v}}])^\top { { \sf p}}^0_n[{{\sf v}}],\ ] ] and one has for any @xmath276 , @xmath277 { { \sf w}},{{\sf w}}\big)=\sum_{n=1}^n \norm { { { \sf p } } _ { n}[{{\sf v}}]{{\sf w}}}^2+\norm { { { \sf p } } _ { -n}[{{\sf v}}]{{\sf w}}}^2+\norm { { { \sf p}}^0_n[{{\sf v}}]{{\sf w}}}^2\geq \frac1{(n(1+d))^2 } \norm{{{\sf w}}}^2.\ ] ] the upper bound in follows from @xmath278}+\norm{{{\sf p } } _ { -n}[{{\sf v}}]}+\norm{{{\sf p}}^0 _ { n}[{{\sf v } } ] } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}}),\ ] ] which is given by in lemma [ l.j ] with in lemma [ l.b - trivialeigen ] , and in lemma [ l.b0-eigen ] with - in lemma [ l.b - eigen ] .
this proves .
similarly , and are easily deduced from the explicit expression for @xmath279 in , and the estimates , , as well as .
we conclude this section by collecting the above information on the symbols of our operators : [ c.prop ] let @xmath280 satisfying and @xmath281 , define @xmath282{\stackrel{\rm def}{=}}\xi^x { { \sf b}}^x[{{\sf v}}]+\xi^y { { \sf b}}^y[{{\sf v}}]={{\sf q}}(\xi)^{-1}{{\sf b}}^x[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi)|\xi| ; \\ { { \sf t}}[{{\sf v}},\xi]{\stackrel{\rm def}{=}}{{\sf q}}(\xi)^{-1}{{\sf t}}^x[{{\sf q}}(\xi){{\sf v}}]{{\sf q}}(\xi ) \quad \text { and } \quad { { \sf p}}_{\rm f}(\xi){\stackrel{\rm def}{=}}{{\sf q}}(\xi)^{-1 } { { \sf \pi}_{\rm f}^x}{{\sf q}}(\xi ) .
\end{gathered}\ ] ] then one has @xmath283)^\top={{\sf
t}}[{{\sf v}},\xi]$ ] , @xmath283{{\sf b}}[{{\sf v}},\xi])^\top={{\sf t}}[{{\sf v}},\xi]{{\sf b}}[{{\sf v}},\xi]$ ] and the following estimates : @xmath284 } \leq c_0\ |\xi| \quad \text { and } \quad \norm{({{\sf id}}-{{\sf p}}_{\rm f}(\xi)){{\sf b}}[{{\sf v}},\xi]}\leq { \varrho}\ c_0\ |\xi| ; \label{c - estimates-1}\\ c_0 { { \sf id}}\leq { { \sf t}}[{{\sf v}},\xi ] \quad \text { and } \quad \norm{{{\sf t}}[{{\sf v}},\xi]}\leq c_0 ; \label{c - estimates-2 } \end{gathered}\ ] ] with @xmath285 , @xmath286 ; and for any @xmath287 satisfying , one has @xmath288-{{\sf b}}[{{\sf v}}_2,\xi ] } & \leq { \varrho}\ c_1\ \norm{{{\sf v}}_1-{{\sf v}}_2 } \ |\xi| ; \label{c - estimates-3 } \\ \norm{{{\sf t}}[{{\sf v}}_1,\xi]{{\sf b}}[{{\sf v}}_1,\xi]-{{\sf t}}[{{\sf v}}_2,\xi]{{\sf b}}[{{\sf v}}_2,\xi]}&\leq { \varrho}\ c_1\ \norm{{{\sf v}}_1-{{\sf v}}_2}\ |\xi| ; \label{c - estimates-4}\\ \norm{{{\sf t}}[{{\sf v}}_1,\xi]-{{\sf t}}[{{\sf v}}_2,\xi]}&\leq c_1\ \left(\norm{{({{\sf id}}-{\sf \pi}_{\rm f})}({{\sf v}}_1-{{\sf v}}_2)}+{\varrho}\norm{{{\sf v}}_1-{{\sf v}}_2}\right ) ; \label{c - estimates-5}\\ \norm{({{\sf t}}[{{\sf v}}_1,\xi]-{{\sf t}}[{{\sf v}}_2,\xi]){{\sf p}}_{\rm f}(\xi)}&\leq { \varrho}\ c_1\ \norm{{{\sf v}}_1-{{\sf v}}_2}\ .
\label{c - estimates-6}\end{aligned}\ ] ] with @xmath289 .
the following proposition shows that , thanks to the structure of the system exhibited in the previous section , one is able to control the ( @xmath133 ) energy of solutions , uniformly on a time interval independent of @xmath62 small .
this is the key ingredient in the proof of our main results .
[ p.energy-estimate-l2 ] let @xmath290 satisfying , with @xmath291 , and @xmath292 , @xmath293 be such that for any @xmath81 , one has @xmath294 \partial_x { { \sf w}}\ + \
\frac1{\varrho}{{\sf b}}^y[{{\sf v } } ] \partial_y { { \sf w}}\ = \ { { \sf r}}.\ ] ] one can set @xmath295 such that if @xmath83 and @xmath296 , then @xmath297 with @xmath298 , @xmath299 and @xmath300 this section is dedicated to the proof of this result .
the main ingredients are the properties of the symbol of system as well as its symmetrizer , collected in corollary [ c.prop ] .
energy estimates for such symmetrizable systems can be obtained thanks to bony s paradifferential calculus associated with these symbols ; see @xcite .
we shall however be cautious as paradifferential calculus typically provides estimates `` up to lower order operators '' : while this is sufficient for regularity aspects , this could induce order - zero but large ( namely non uniformly bounded with respect to @xmath62 small ) remainder terms , preventing the desired uniform energy control stated in .
this is why we decompose the symbols into a first order contribution which admits a natural quantization as a fourier multiplier , whereas only second - order contributions will be paradifferentialized .
let us be more specific .
define @xmath301 |\xi| { \stackrel{\rm def}{=}}\chi(\xi ) \big({{\sf
b}}^x[{{\sf q}}(\xi){{\sf v}}(t , x ) ] - { { \sf b}}^x[{\mathbf{0 } } ] \big)|\xi|\ ] ] where @xmath302 is a smooth non - negative cut - off function ( @xmath303 for @xmath304 and @xmath305 for @xmath306 ) ; and let @xmath307 be the associated paradifferential operator ( see definition [ d.paradiff ] ) .
similarly , define @xmath308 { \stackrel{\rm def}{=}}\chi(\xi ) \big({{\sf
t}}^x[{{\sf q}}(\xi){{\sf v}}(t,{\mathbf{x } } ) ] - { { \sf t}}^x[{\mathbf{0 } } ] \big)\ ] ] and @xmath309 the associated paradifferential operator ; and @xmath310 + \frac12 \left ( { \mathcal{t}}_{\delta{{\sf t } } } + { \mathcal{t}}_{\delta{{\sf t}}}^\star\right ) + \lambda { \sf \lambda}_{\varrho}^{-1 } \right ) { { \sf q}}(d ) , \ ] ] where @xmath311 and @xmath312 will be determined later on .
we claim in lemma [ l.s ] , below , that the properties on @xmath313 $ ] given in corollary [ c.prop ] are sufficient to show that @xmath314 is a uniformly bounded and coercive operator , and show a precise estimate on its time derivative , @xmath315 . in lemma
[ l.paralinearization ] , we then rewrite as a symmetric , paradifferential equation , from which the energy estimate is easily deduced .
[ l.s ] let @xmath140 be as in proposition [ p.energy-estimate-l2 ] and fix @xmath316
. then @xmath317 , defined by , is self - adjoint and one can set @xmath318 and @xmath319 , such that if @xmath320 and @xmath321 , then one has @xmath322 where @xmath323 .
moreover , @xmath324 is well - defined and one has @xmath325 where @xmath326 . that @xmath314 is self - adjoint is obvious ( recall in particular that @xmath189 is orthogonal ) .
let us now decompose @xmath327 + \frac12 \left ( { \mathcal{t}}_{\delta{{\sf t}}_{(1 ) } } + { \mathcal{t}}_{\delta{{\sf t}}_{(1)}}^\star\right ) + \lambda{\sf \lambda}_{\varrho}^{-1 } \\ & \hspace{2cm}+ \frac12 \left ( { \mathcal{t}}_{\delta{{\sf t}}_{(2 ) } } + { \mathcal{t}}_{\delta{{\sf t}}_{(2)}}^\star\right ) - \chi(d)^{1/2 } ( { { \sf t}}^x[{{\sf z}}]-{{\sf t}}^x[{\mathbf{0}}])\chi(d)^{1/2 } \\ & \hspace{2cm}+ \chi(d)^{1/2 } ( { { \sf t}}^x[{{\sf z}}]-{{\sf t}}^x[{\mathbf{0}}])\chi(d)^{1/2 } - ( { { \sf t}}^x[{{\sf z}}]-{{\sf t}}^x[{\mathbf{0 } } ] ) \\ & { \stackrel{\rm def}{=}}{{\sf t}}^x[{{\sf z } } ] + { \mathcal{s}}_{(1)}+\lambda{\sf \lambda}_{\varrho}^{-1}+{\mathcal{s}}_{(2)}+{\mathcal{s}}_{({\rm r } ) } , \end{aligned}\ ] ] where @xmath328 is obtained from @xmath140 by setting to zero all @xmath4 entries with @xmath329 , and ( notice @xmath330 ) @xmath331 -{{\sf t}}^x[{{\sf q}}(\xi){{\sf z } } ] ) , \\ \delta{{\sf t}}_{(2)}(t,{\mathbf{x}},\xi ) & { \stackrel{\rm def}{=}}\chi(\xi ) ( { { \sf t}}^x[{{\sf z}}]-{{\sf t}}^x[{\mathbf{0 } } ] ) .
\end{aligned}\ ] ] using in corollary [ c.prop ] , we can define @xmath332 such that @xmath333 { { \sf w}},{{\sf w}}\big)_{l^2 } \quad \text { and } \quad \norm { { { \sf t}}^x[{{\sf z } } ] { { \sf w}}}_{l^2}\leq \frac12 c_0\norm{{{\sf w}}}_{l^2 } .\ ] ] by in corollary [ c.prop ] and since @xmath140 satisfies , one has for any @xmath334 , @xmath335 -{{\sf t}}^x[{{\sf q}}(\xi){{\sf z}}(t,{\mathbf{x } } ) ] } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}(t,{\mathbf{x } } ) } ) ( \nu^{-1}+{\varrho}\norm{{{\sf v}}(t,{\mathbf{x}})}).\ ] ] similarly , the lipschitz estimate yields for any @xmath336 and @xmath337 , @xmath338 so that the contribution from @xmath339 is estimated thanks to proposition [ p.paradiff-estimates ] ( item i ) as follows : @xmath340 similarly , using in corollary [ c.prop ] yields , uniformly in @xmath81 , @xmath341-{{\sf t}}^x[{\mathbf{0}}]}_{w^{1,\infty}}+\frac1{\varrho}\norm{({{\sf t}}^x[{{\sf z}}]-{{\sf t}}^x[{\mathbf{0}}]){{\sf \pi}_{\rm f}^x}}_{w^{1,\infty}}\leq c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}}_{l^\infty } ) \norm{{{\sf v}}}_{w^{1,\infty}};\ ] ] and one obtains identically the corresponding estimates for derivatives with respect to @xmath191 .
thus propositions [ p.paradiff-estimates ] and [ p.paradiff-particular cases ] yield for any @xmath342 , @xmath343 one easily checks that the last contribution satisfies the same estimate : @xmath344 here , we used that for scalar functions @xmath345 and
@xmath346 , one has @xmath347 where @xmath348 is a universal constant ( by duality , since for any @xmath349 , one has @xmath350 and @xmath351 ) .
thanks to the above estimates , one may choose @xmath352 sufficiently large so that one has @xmath353 it is now clear ( using that @xmath189 is orthogonal for @xmath211 ) that one can restrict @xmath354 as in the statement so that holds . as for the second part of the statement , by definition of the quantization @xmath355 ( see definition [ d.paradiff ] ) and
since @xmath356 commutes with constant operators and fourier multipliers , one has @xmath357 moreover , using in corollary [ c.prop ] , one has @xmath358 ) } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}(t,{\mathbf{x } } ) } ) \left(\norm{{({{\sf id}}-{\sf \pi}_{\rm f})}\partial_t{{\sf v}}(t,{\mathbf{x}})}+{\varrho}\norm{\partial_t{{\sf v}}(t,{\mathbf{x}})}\right).\ ] ] again , one easily obtains the corresponding estimates for derivatives with respect to @xmath191 and proposition [ p.paradiff-estimates ] ( item i ) yields @xmath359 estimate is proved , and the proof of lemma [ l.s ] is complete .
[ l.paralinearization ] let @xmath360 be as in proposition [ p.energy-estimate-l2 ] and @xmath361 as in lemma [ l.s ] .
then one has @xmath362 { { \sf b}}^x[{\mathbf{0 } } ] |d| + { \mathcal{t}}_{i{\sf \sigma } } \right ) { { \sf q}}(d ) { { \sf w}}= { \mathcal{r}}[{{\sf v}},{{\sf w}},{{\sf r}}],\ ] ] where @xmath363{{\sf b}}^x[{{\sf q}}(\xi){{\sf v}}(t,{\mathbf{x } } ) ] - { { \sf t}}^x[{\mathbf{0}}]{{\sf b}}^x[{\mathbf{0 } } ] \big)|\xi| , \ ] ] and @xmath364\in l^1(0,t;l^2({\mathbb{r}}^d))^{n(1+d)}$ ] satisfies @xmath365 } _ { l^2 } ( t)\leq c_0 \norm{{{\sf r}}}_{l^2}(t ) + c_1 \norm{{{\sf w}}}_{l^2}(t)\ ] ] with @xmath366 .
moreover , one has @xmath367 notice first that , using the rotational invariance property , one has @xmath368\partial_x{{\sf
w}}+{{\sf b}}^y[{\mathbf{0}}]\partial_y { { \sf w}}= i ( { { \sf q}}(d))^{-1 } { { \sf b}}^x[{\mathbf{0 } } ] { { \sf q}}(d)|d|{{\sf w}}.\ ] ] thus applying the operator @xmath369 to yields with @xmath370{\stackrel{\rm def}{=}}{\mathcal{s}}{{\sf r}}\ + \ ( { { \sf q}}(d))^{-1 } { \mathcal{t}}_{i{\sf \sigma } } { { \sf q}}(d ) { { \sf w}}\\ -({{\sf q}}(d))^{-1 } \left ( \frac12 \left ( { \mathcal{t}}_{\delta{{\sf t } } } + { \mathcal{t}}_{\delta{{\sf t}}}^\star\right ) + \lambda { \sf \lambda}_{\varrho}^{-1}(d)\right ) { { \sf q}}(d)\left(\frac1{\varrho}{{\sf b}}^x[{{\sf v}}]\partial_x{{\sf w}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y { { \sf w}}\right ) \\ -({{\sf q}}(d))^{-1 } { { \sf t}}^x[{\mathbf{0 } } ] { { \sf q}}(d ) \left(\frac1{\varrho}\big({{\sf b}}^x[{{\sf v}}]-{{\sf b}}^x[{\mathbf{0}}]\big)\partial_x{{\sf w}}+\frac1{\varrho}\big({{\sf b}}^y[{{\sf v}}]-{{\sf b}}^y[{\mathbf{0}}]\big)\partial_y { { \sf w}}\right ) .\end{gathered}\ ] ]
we first note that , using estimate in corollary [ c.prop ] yields immediately @xmath371\partial_x{{\sf w}}+{{\sf b}}^y[{\mathbf{0}}]\partial_y { { \sf w}}\right)}_{l^2}=\frac1{\varrho}\norm { { \sf \lambda}_{\varrho}^{-1}(d){{\sf b}}^x[{\mathbf{0 } } ] { { \sf q}}(d ) |d|{{\sf w}}}_{l^2}\leq c({\mathfrak{m}},h_0^{-1},\lambda ) \norm{{{\sf w}}}_{l^2}.\ ] ] we then deduce from estimate in corollary [ c.prop ] that @xmath372-{{\sf b}}^x[{\mathbf{0}}]}_{w^{1,\infty}}+\norm{{{\sf b}}^y[{{\sf v}}]-{{\sf b}}^y[{\mathbf{0}}]}_{w^{1,\infty}}\leq { \varrho}\ c(h_0^{-1 } , { \mathfrak{m}},\norm{{{\sf v}}}_{l^\infty } ) \norm{{{\sf v}}}_{w^{1,\infty}};\ ] ] and in turn , by , @xmath373\partial_x{{\sf w}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y { { \sf w}}\right)}_{l^2}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}}_{l^\infty } ) \norm{{{\sf v}}}_{w^{1,\infty } } \norm{{{\sf w}}}_{l^2}.\ ] ] thus there only remains to estimate @xmath374&{\stackrel{\rm def}{=}}({{\sf q}}(d))^{-1 } { \mathcal{t}}_{i{\sf \sigma}_{(1)}}{{\sf q}}(d){{\sf w}}- \frac12 ( { { \sf q}}(d))^{-1}\left ( { \mathcal{t}}_{\delta{{\sf t } } } + { \mathcal{t}}_{\delta{{\sf t}}}^\star\right ) { { \sf q}}(d ) \left(\frac1{\varrho}{{\sf
b}}^x[{{\sf v}}]\partial_x{{\sf w}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y { { \sf w}}\right ) , \\ { \mathcal{r}}_{(2)}[{{\sf
v}},{{\sf w}}]&{\stackrel{\rm def}{=}}({{\sf q}}(d))^{-1 } { \mathcal{t}}_{i{\sf \sigma}_{(2)}}{{\sf q}}(d){{\sf w}}\\ & \hspace{2 cm } -({{\sf q}}(d))^{-1 } { { \sf t}}^x[{\mathbf{0 } } ] { { \sf q}}(d )
\left(\frac1{\varrho}\big({{\sf b}}^x[{{\sf v}}]-{{\sf b}}^x[{\mathbf{0}}]\big)\partial_x{{\sf w}}+\frac1{\varrho}\big({{\sf b}}^y[{{\sf v}}]-{{\sf b}}^y[{\mathbf{0}}]\big)\partial_y { { \sf w}}\right ) , \end{aligned}\ ] ] where @xmath375 \quad \text { and } \quad { \sf \sigma}_{(2)}(t,{\mathbf{x}},\xi ) { \stackrel{\rm def}{=}}\frac1{\varrho}{{\sf t}}^x[{\mathbf{0}}]\delta{{\sf b}}(t,{\mathbf{x}},\xi ) .\ ] ] these terms are estimated exactly as in the proof of lemma [ l.s ] , _ i.e. _ using the paradifferential calculus of propositions [ p.paradiff-estimates ] and [ p.paradiff-particular cases ] together with the estimates of corollary [ c.prop ] , thus we do not detail .
let us just indicate why these contributions are uniformly bounded with respect to @xmath62 small .
the case of @xmath376 $ ] is quickly settled by in corollary [ c.prop ] . as for @xmath377
$ ] , we decompose as above @xmath270 = \frac1{\varrho}{{\sf b}}^x[{\mathbf{0 } } ] + \frac1{\varrho}\big ( { { \sf b}}^x[{{\sf v } } ] -{{\sf b}}^x[{\mathbf{0}}]\big).\ ] ] the contribution from the second component is uniformly bounded with respect to @xmath62 small , again thanks to in corollary [ c.prop ] . the contribution from the first component
may also be uniformly bounded by remarking that @xmath378 } \leq \norm{\big(\delta{{\sf t}}(t,{\mathbf{x}},\xi ) { { \sf \pi}_{\rm f}}\big ) { { \sf b}}^x[{\mathbf{0 } } ] } + \norm{\delta{{\sf t}}(t,{\mathbf{x}},\xi)\big({({{\sf id}}-{\sf \pi}_{\rm f})}{{\sf b}}^x[{\mathbf{0 } } ] \big ) } \leq { \varrho}\times c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}})\norm{{{\sf v}}},\ ] ] where we used , and in corollary [ c.prop ] . altogether , one estimates @xmath379 in as desired , namely @xmath365 } _ { l^2 } \leq c_0 \norm{{{\sf r}}}_{l^2 } + c_1 \norm{{{\sf w}}}_{l^2 } , \ ] ] with @xmath366 .
there remains to estimate @xmath380 . by in corollary [ c.prop ] , one has @xmath381 and @xmath382 is symmetric .
proposition [ p.paradiff-estimates ] ( items ii . and
iii . ) as well as proposition [ p.paradiff-particular cases ] yield @xmath367 the proof of lemma [ l.paralinearization ] is complete . [ [ completion - of - the - proof . ] ] completion of the proof .
+ + + + + + + + + + + + + + + + + + + + + + + + assume @xmath383 is sufficiently regular , say @xmath384 , so that all the calculations below are well - defined .
we compute the @xmath133 inner - product of identity with @xmath383 : it follows @xmath385 { { \sf w}},{{\sf w}}\big)_{l^2}\big|+\frac12 \big|\re\big ( i \frac1{\varrho}({{\sf q}}(d))^{-1 } { { \sf t}}^x[{\mathbf{0 } } ] { { \sf b}}^x[{\mathbf{0 } } ] { { \sf q}}(d)|d| { { \sf w } } , { { \sf w}}\big)\big|\\+\big|\re\big({\mathcal{t}}_{i{\sf \sigma } } { { \sf w}},{{\sf w}}\big)_{l^2 } \big|+\big| \big({\mathcal{r}}[{{\sf r}},{{\sf v}},{{\sf w}}],{{\sf w}}\big)_{l^2}\big|.\end{gathered}\ ] ] the second term on the right - hand - side is identically zero since @xmath386 { { \sf b}}^x[{\mathbf{0 } } ] $ ] is symmetric . the other terms are estimated thanks to cauchy - schwarz inequality and lemmata [ l.s ] and [ l.paralinearization ] . altogether , this shows that provided we restrict @xmath83 and @xmath296 as in lemma [ l.s ] , one has @xmath387}_{l^2}\norm{{{\sf w}}}_{l^2},\ ] ] with @xmath388 .
estimate follows from gronwall s lemma and the coercivity of @xmath369 , _
i.e. _ in lemma [ l.s ] , together with the control of @xmath389 provided in lemma [ l.paralinearization ] .
the fact that the same estimate holds for general @xmath292 solution to may be obtained by density and thanks to a standard regularization process ; see ( * ? ? ?
* theorem 7.1.11 ) .
proposition [ p.energy-estimate-l2 ] is proved .
the @xmath133 energy estimate derived in proposition [ p.energy-estimate-l2 ] quickly induces a similar @xmath138 estimate , for any @xmath69 , by using once more the specific structure of our system of equations , namely that singular terms appear only as linear components of the system .
[ p.energy-estimate-hs ] let @xmath69 and @xmath390 be such that @xmath140 satisfies , with @xmath291 and @xmath391 , and @xmath392 \partial_x { { \sf w}}\ + \
\frac1{\varrho}{{\sf b}}^y[{{\sf v } } ] \partial_y { { \sf w}}\ = \ { { \sf r}},\ ] ] with @xmath393 .
assume that @xmath83 and @xmath296 with @xmath106 as in proposition [ p.energy-estimate-l2 ] .
then one has , for any @xmath316 , @xmath394 with @xmath395 , @xmath396 and @xmath397 denote @xmath398 , with @xmath399 .
then one has @xmath400\partial_x{{\sf
w}}^s+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y{{\sf w}}^s=\lambda^s{{\sf r}}+\big[\lambda^s,\frac1{\varrho}{{\sf b}}^x[{{\sf v}}]\big]\partial_x{{\sf w}}+\big[\lambda^s,\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\big]\partial_x{{\sf w}}.\ ] ] we shall apply proposition [ p.energy-estimate-l2 ] to the above system , thanks to the standard tools on sobolev spaces recalled in section [ s.product ] .
notice first that since since @xmath69 , sobolev embedding yields @xmath401 so that we only need to estimate the commutator on the right - hand - side of to apply proposition [ p.energy-estimate-l2 ]
. since @xmath402 commutes with @xmath403 $ ] , one has @xmath404\big]\partial_x{{\sf
w}}\ = \ \frac1{\varrho}\big[\lambda^s,{{\sf b}}^x[{{\sf v}}]-{{\sf b}}^x[{\mathbf{0}}]\big]\partial_x{{\sf w}}.\ ] ] now , thanks to the product and commutator estimates recalled in section [ s.product ] , and following the proof of lemma [ l.b ] , one easily checks that , for any @xmath81 , @xmath405\big]\partial_x{{\sf w}}}_{l^2 } = \frac1{\varrho}\norm { \big[\lambda^s,{{\sf b}}^x[{{\sf v}}]-{{\sf b}}^x[{\mathbf{0}}]\big]\partial_x{{\sf w}}}_{l^2}\leq c({\mathfrak{m}},h_0^{-1 } , \norm{{{\sf v}}}_{h^s } ) \norm{{{\sf v}}}_{h^s}\norm{\partial_x { { \sf w}}}_{h^{s-1}}.\ ] ] obviously , since @xmath406={{\sf q}}((0,1))^{-1}{{\sf b}}^x[{{\sf q}}((0,1)){{\sf v}}]{{\sf q}}((0,1))$ ] ( by ) , one has @xmath407\big]\partial_y{{\sf w}}}_{l^2 } \leq c({\mathfrak{m}},h_0^{-1 } , \norm{{{\sf v}}}_{h^s } ) \norm{{{\sf v}}}_{h^s}\norm{\partial_y { { \sf w}}}_{h^{s-1}}.\ ] ] one could apply the @xmath133 estimate of proposition [ p.energy-estimate-l2 ] to @xmath408 satisfying , but one obtains a slightly stronger result by stepping back and using directly the differential inequality of the proof , namely : @xmath409}_{l^2}\norm{{{\sf w}}^s}_{l^2},\ ] ] with @xmath388 , and @xmath410}_{l^2}\leq c_0 \norm{{{\sf r}}}_{h^s } + c_s \norm{{{\sf w}}}_{h^s } , \ ] ] where @xmath411 , and @xmath412 .
estimate follows from gronwall s lemma and the coercivity of @xmath369 , _
i.e. _ in lemma [ l.s ] . again
, this estimate is proved for sufficiently regular @xmath413 , but may be extended to general @xmath414 solution to by density and thanks to a standard regularization process .
this concludes the proof of proposition [ p.energy-estimate-hs ] .
in this section , we collect the information gathered in the previous sections , which quickly yield theorem [ t.wp ] , as well as the propagation of well - prepared initial data ( see proposition [ p.well-prepared initial data ] ) .
let us first recall that proposition [ p.wp-naive ] offers the existence and uniqueness of a ( maximal ) solution to , for sufficiently regular initial data .
our results give additional information on the large time behaviour of these solutions , using in particular the energy estimates of proposition [ p.energy-estimate-l2 ] and [ p.energy-estimate-hs ] .
these estimates , however , are based on a different formulation of the equations , namely , which we claimed to be equivalent .
let us precisely state below in which sense .
[ l.utov ] let @xmath163 with @xmath69 , satisfying .
then @xmath415 , defined by , satisfies @xmath416 and @xmath417 conversely , if @xmath416 satisfies , then the change of variables defines @xmath86 and @xmath418 moreover , if @xmath66 as above is a strong solution to ( or , equivalently , ) , then @xmath140 is a strong solution to ( or , equivalently , ) ; and conversely .
notice that @xmath86 implies immediately @xmath416 since @xmath419 is an algebra for any @xmath420 .
the converse is also true since the multiplication by @xmath421 is continuous from @xmath419 to @xmath419 ; see section [ s.product ] .
it follows , by continuous sobolev embedding , that all the terms in and as well as in the calculations below are well - defined in @xmath422 .
the evolution equations for @xmath423 in are straightforwardly deduced from the ones for @xmath424 in .
the evolution equation for @xmath110 follows from @xmath425 plugging the expression for @xmath426 in , and using @xmath427 , yields immediately the desired expression of the evolution equation for @xmath110 in .
the corresponding result concerning the compact matricial formulation of the systems , and in particular , is obvious by the chain rule .
this proves the first part of the statement .
the second part is identical since all these calculations are reversible : the jacobian associated to the change of variables is invertible ; see . by proposition [ p.wp-naive ]
, one can set @xmath428 such that if holds with @xmath209 , then there exists a unique @xmath429 strong solution to and @xmath78 . by lemma [ l.utov ]
, the change of variable defines @xmath430 and @xmath140 satisfies @xmath431\partial_x{{\sf v}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y{{\sf v}}={\mathbf{0}},\ ] ] and @xmath432 .
let us denote @xmath433 we restrict our discussion below to @xmath434 , so that ( by continuity ) @xmath435 . using the system satisfied by @xmath140 , in corollary [ c.prop ] as well as sobolev embeddings , one checks @xmath436 for any @xmath437 ; recall @xmath438 is defined in .
in particular , one has for any @xmath437 , @xmath439}\times { \mathbb{r}}^d}{\varrho}\norm{\partial_t({\varrho}^{-1}\zeta_1)}+\sum_{n=2}^n \norm{\partial_t\zeta_n}+\norm{\partial_t v_n^x}+\norm{\partial_t v_n^y}\leq \bnorm{{{\sf v}}}_s\leq c({\mathfrak{m}},h_0^{-1},m)m,\ ] ] from which we deduce @xmath440 and , similarly , @xmath441 it follows that , for given @xmath442 , one can set @xmath443 such that if @xmath83 and holds with @xmath209 , then @xmath140 satisfies the assumptions of proposition [ p.energy-estimate-hs ] for any @xmath444 with @xmath445 we thus deduce the energy estimate @xmath446 with @xmath447 and @xmath448 . in particular
this shows that one can choose @xmath449 such that @xmath450 going back to the original variables through and by lemma [ l.utov ] , this shows that one can restrict @xmath451 and the time interval @xmath452 , with @xmath453 bounded as above , so that @xmath454 is uniformly bounded and - remain satisfied . by continuity , uniqueness of the maximal solution and thanks to the blow - up criteria in proposition [ p.wp-naive ]
, we deduce that one can set @xmath455 this concludes the proof of theorem [ t.wp ] .
[ p.well-prepared initial data ] let @xmath69 and @xmath84 satisfying - as in theorem [ t.wp ] , and denote @xmath456 the solution to and @xmath78 .
if @xmath74 satisfies initially @xmath457 then there exists @xmath458 such that @xmath459 uniformly with respect to @xmath83 .
let us denote by @xmath460 the solution to and @xmath78 defined by theorem [ t.wp ] ; and by @xmath461 the associated solution to , namely @xmath431\partial_x{{\sf v}}+\frac1{\varrho}{{\sf
b}}^y[{{\sf v}}]\partial_y{{\sf v}}={\mathbf{0}},\ ] ] obtained through the change of variables ( see lemma [ l.utov ] ) .
finally we denote @xmath462 . using the above equation and the product estimates recalled in section [ s.product ] , one has immediately @xmath463 . differentiating the above system of equations yields @xmath464\partial_x{{\sf w}}+\frac1{\varrho}{{\sf b}}^y[{{\sf v}}]\partial_y{{\sf w}}=
-\frac1{\varrho}\partial_t({{\sf b}}^x[{{\sf v}}])\partial_x{{\sf v}}-\frac1{\varrho}\partial_t({{\sf b}}^y[{{\sf v}}])\partial_y{{\sf v}}.\ ] ] by construction ( see the proof of theorem [ t.wp ] above ) , one can restrict @xmath465 so that @xmath140 satisfies the assumptions of proposition [ p.energy-estimate-l2 ] , namely - for @xmath81 ; and one has @xmath466 by estimate in corollary [ c.prop ] , one has @xmath467
what is more , the additional smallness assumption on @xmath468 yields @xmath469 uniformly for @xmath83 . finally , by in corollary [ c.prop ]
, one has @xmath470)\partial_x{{\sf v}}}_{l^2}+\norm { \partial_t({{\sf b}}^y[{{\sf v}}])\partial_y{{\sf v}}}_{l^2}\leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}}_{l^{\infty } } ) \norm{{{\sf v}}}_{w^{1,\infty}}\norm{\partial_t { { \sf v}}}_{l^2}.\ ] ] altogether , after applying proposition [ p.energy-estimate-l2 ] with @xmath462 , one deduces @xmath471 with @xmath472 . applying gronwall s lemma to @xmath473 yields @xmath474 for @xmath472 sufficiently large . using again the system of equations satisfies by @xmath140 , namely , estimates and yield @xmath475 for @xmath472 and uniformly with @xmath83 . proposition [ p.well-prepared initial data ] is proved .
in this section , we investigate the asymptotic behaviour of the previously obtained solutions in the limit @xmath100 .
we first show that the solutions of the free - surface system converge weakly towards solutions of the rigid - lid system .
strong convergence results are then obtained , first by assuming that the initial data is well - prepared , and then by approaching the oscillatory `` defect '' through rapidly propagating acoustic waves .
our first ( weak ) convergence result is the following .
[ p.weak-convergence ] as @xmath100 , let @xmath476 satisfying - . denote , for @xmath62 sufficiently small , @xmath477 the solution to with @xmath478 . then , as @xmath144 , @xmath479 converges weakly ( in the sense of distributions and up to a subsequence ) towards @xmath480 a solution of the rigid - lid system , with initial data @xmath481 where @xmath89 , @xmath90 with convention @xmath91 and @xmath35 is the orthogonal projection onto irrotational vector fields . restricting @xmath83 if necessary
, @xmath482 satisfies the hypotheses of theorem [ t.wp ] .
thus one can define @xmath483 from the solution to with initial data @xmath484 ; and @xmath485 with @xmath486 .
thus , by banach - alaoglu theorem on @xmath487 , we can extract a weakly converging subsequence ( in the sense of distributions ) , that we still denote @xmath479 : @xmath488 with @xmath489 .
let us first notice that since @xmath479 satisfies one has , uniformly for @xmath490 , @xmath491 thus , since the embedding of @xmath492 in @xmath67 is locally compact and by aubin - lions lemma and cantor s diagonal argument , one has ( again up to the extraction of a subsequence ) @xmath493 strongly in @xmath494 .
it follows by the logarithmic convexity of sobolev norms , that for any @xmath495 , @xmath496 thereafter , we fix @xmath497 so that @xmath498 and is an algebra . for
the same reasons as above , we have also @xmath499 and , using @xmath500 , @xmath501 let us now define @xmath502 notice now that all the ( quadratic ) nonlinear terms are strongly convergent , so that @xmath503 and @xmath504 with convention @xmath505 .
notice also that passing to the ( weak ) limit in the first equation of yields @xmath506 we now define @xmath507 notice the identity @xmath508 so that @xmath509 it follows , using the formula with @xmath510 and @xmath511 , and since all the components converge strongly in @xmath512 , that @xmath513 converges strongly as well . by virtue of uniqueness of the weak limit
, one has @xmath514 we conclude by plugging the decomposition @xmath515 into the evolution equations for velocities in .
notice the identity , using that @xmath516 is irrotational , @xmath517 thus the only ( quadratic ) term which does not involve at least one strongly convergent factor turns out to be an exact gradient and independent of @xmath518 ; and so is the unbounded ( linear ) component of the equation , namely @xmath519 .
this shows , passing to the weak limit all the other terms in the equation , that there exists @xmath41 , independent of @xmath518 , such that @xmath520 it is straightforward to pass to the limit in the conservation of mass equations , so that @xmath521 and we have already seen that @xmath522 . thus @xmath523 is a solution to , and one checks immediately that @xmath524 .
this concludes the proof of proposition [ p.weak-convergence ] .
[ p.convergence-wp ] using the notations of proposition [ p.weak-convergence ] , let @xmath476 as @xmath100 , with @xmath525 well - prepared as in proposition [ p.well-prepared initial data ] . then ( up to extracting a subsequence ) @xmath526 strongly in @xmath527 , for all @xmath495 , as @xmath100 . moreover , @xmath528 and @xmath529 , where @xmath530 is the pressure associated with @xmath531 solution to .
we follow the proof of proposition [ p.weak-convergence ] . however
, we may use additionally that , thanks to proposition [ p.well-prepared initial data ] ( and using the system of equations to control time derivatives ) @xmath532 it follows ( up to the extraction of a subsequence ) @xmath533 and @xmath534 strongly in @xmath535 , and therefore in @xmath512 for @xmath495 ; and @xmath536 ( in the sense of distributions ) . by banach - alaoglu theorem
, @xmath537 has a weak limit in @xmath538 when @xmath144 , and passing to the weak limit in the velocity evolution equations shows that this limit is @xmath530 .
the proof of theorem [ t.convergence ] is divided in three parts .
we first construct a `` slow mode '' approximate solution , thanks to the appropriate rigid - lid solution .
we then construct a `` fast mode '' approximate solution , satisfying an acoustic wave equation with appropriate initial data .
finally , we show that , thanks to dispersive estimates on the fast mode , the coupling effects between the two modes are small , so that the superposition of the two components solves approximately the free - surface system with appropriate initial data .
the energy estimate of proposition [ p.energy-estimate-l2 ] , applied to the difference between the exact and the approximate solution , allows to conclude .
_ construction of the slow mode .
_ using proposition [ p.convergence-wp ] with well - prepared initial data @xmath539 and artificially setting @xmath100 , one obtains in the limit @xmath540 a solution to with initial data @xmath481 where we recall that @xmath89 is the total depth , @xmath90 with convention @xmath91 and @xmath35 the orthogonal projection onto irrotational vector fields .
moreover , the corresponding pressure satisfies @xmath541 , and one has @xmath542 ( since @xmath479 , the solutions of with @xmath543 from which @xmath531 is constructed , satisfy the same estimate by theorem [ t.wp ] ) .
let us prove that one may ( uniquely ) choose @xmath544 .
the regularity of @xmath531 strong solution to is sufficient to claim that @xmath541 is uniquely defined by @xmath545 where @xmath546 with convention @xmath547 .
notice now that@xmath548 and @xmath549 it follows , using that @xmath550 is an algebra , that there exists @xmath551 such that @xmath552 thus one may define a unique solution @xmath553 by fourier analysis . from the product estimates in sobolev spaces ( see section [ s.product ] ) and estimate , one has @xmath554 we denote @xmath555 . if @xmath62 is chosen sufficiently small , then @xmath556 satisfies and therefore the change of variables defines @xmath557 satisfying ( see lemma [ l.utov ] ) @xmath558 it is easy to check , using @xmath59 and since @xmath559 , that @xmath557 satisfies up to a remainder term denoted @xmath560 ; and @xmath561 _ construction of the fast mode .
_ we constructed above an approximate solution of ( in the sense of consistency ) , but which does not fulfil the required initial condition .
we correct this defect through an explicit `` fast mode '' .
define @xmath562 where @xmath563 and @xmath564 solve @xmath565 with initial condition @xmath566 and @xmath567 , so that @xmath568 where @xmath569 is defined from @xmath74 after the change of variables .
the above is an acoustic wave equation and is well understood .
the following results may be found in @xcite for instance .
there exists a unique solution @xmath570 .
it satisfies @xmath571 for any @xmath572 , and @xmath573 what is more , since @xmath8 , one has the strichartz estimates @xmath574 where @xmath575 are admissible , namely @xmath576 and @xmath577 and @xmath578 .
set for instance @xmath579 and @xmath580 . by a scaling argument and
differentiating once the system , one has @xmath581 it follows that we control quadratic nonlinearities as @xmath582 where we used hlder inequality , then sobolev embedding and - , with the restriction on the time interval , @xmath583 we deduce that @xmath584 satisfies the equations up to a remainder @xmath585 such that @xmath586 _ control of the coupling terms .
_ let us denote @xmath587 .
let us first check that @xmath588 is an approximate solution to , in the sense of consistency . from the above
, we have that @xmath589 \partial_x{{\sf v}}^{\rm app } + \frac1{\varrho}{{\sf b}}^y[{{\sf v}}^{\rm app } ] \partial_y{{\sf v}}^{\rm app } = { { \sf r}}^{\rm slow}+{{\sf r}}^{\rm fast}+{{\sf r}}^{\rm coupl}\ ] ] where @xmath590 and @xmath591 have been defined and estimated previously ; and @xmath592 - { { \sf b}}^x[{{\sf v}}^{\rm fast } ] \big)\partial_x{{\sf v}}^{\rm fast } + \frac1{\varrho}\big({{\sf b}}^y[{{\sf v}}^{\rm app } ] - { { \sf b}}^y[{{\sf v}}^{\rm fast } ] \big)\partial_y{{\sf v}}^{\rm fast } \\
+ \frac1{\varrho}\big({{\sf b}}^x[{{\sf v}}^{\rm app } ] - { { \sf b}}^x[{{\sf v}}^{\rm slow } ] \big)\partial_x{{\sf v}}^{\rm slow } + \frac1{\varrho}\big({{\sf b}}^y[{{\sf v}}^{\rm app } ] - { { \sf b}}^y[{{\sf v}}^{\rm slow } ] \big)\partial_y{{\sf v}}^{\rm slow }
. \end{gathered}\ ] ] by estimate in corollary [ c.prop ] , one has @xmath593 we deduce from , and , proceeding as for , @xmath594 let us now denote @xmath460 the strong solution to with initial data @xmath78 as defined by theorem [ t.wp ] ; and @xmath140 the corresponding solution to defined by the change of variables . by theorem [ t.wp ] and lemma [ l.utov ]
, one has @xmath595 proceeding as above , we find that the difference between the exact and the approximate solution , @xmath596 , satisfies @xmath597
\partial_x{{\sf w}}+ \frac1{\varrho}{{\sf b}}^y[{{\sf v } } ] \partial_y { { \sf w}}={{\sf r}}^{\rm slow}+{{\sf r}}^{\rm fast}+{{\sf r}}^{\rm coupl}+{{\sf r}},\ ] ] with @xmath598 - { { \sf b}}^x[{{\sf v } } ] \big)\partial_x{{\sf v}}^{\rm app } + \frac1{\varrho}\big({{\sf b}}^y[{{\sf v}}^{\rm app } ] - { { \sf b}}^y[{{\sf v } } ] \big)\partial_y{{\sf v}}^{\rm app } .\ ] ] again , by estimate in corollary [ c.prop ] , one has @xmath599 we now apply proposition [ p.energy-estimate-l2 ] with @xmath600 , and deduce @xmath601 with @xmath298 , @xmath602 . using the above control on the initial data and remainder terms , , , and gronwall s lemma
, we obtain @xmath603 by the logarithmic convexity of sobolev norms and estimates , , , it follows @xmath604 for any @xmath495 .
this concludes the proof of theorem [ t.convergence ] , with the exception of the uniqueness of the strong solution of the rigid - lid system .
however , given two solutions @xmath605 with same initial data , we may construct as above ( _ i.e. _ multiplying the pressure with a @xmath62 prefactor ) @xmath606 , two families of approximate solutions of the free - surface system , for arbitrarily small @xmath62 . applying proposition [ p.energy-estimate-l2 ] with the equation satisfied by the difference between the two solutions ( after the change of variable ) , using , and taking the limit @xmath100 , yields @xmath607 .
we denote by @xmath608 a non - negative constant depending on the parameters @xmath609 , @xmath610, and whose dependence on @xmath611 is always assumed to be non - decreasing .
it may also depend without acknowledgment on the horizontal dimension , @xmath103 ; number of layers , @xmath29 ; sobolev index at stake , @xmath612 . given @xmath613 a topological vector space
, @xmath614 denotes its continuous dual , endowed with the strong topology .
given @xmath615 a hilbert space and @xmath616 a continuous linear operator , we denote @xmath617 its adjoint . for @xmath618 ,
we denote @xmath619 the standard lebesgue spaces associated with the norm
@xmath620 the space @xmath621 consists of all essentially bounded , lebesgue - measurable functions @xmath622 endowed with the norm @xmath623 for @xmath624 , we denote by @xmath625 ( endowed with its canonical norm ) where we use the standard multi - index notation for @xmath626-differentiation .
we denote by @xmath627 the space of continuous functions on @xmath628 with continuous derivatives up to the order @xmath629 , endowed with the same norm . the real inner product of any functions @xmath630 and @xmath631 in the hilbert space @xmath632 is denoted by @xmath633 for any real constant @xmath634 , @xmath635 denotes the sobolev space of all tempered distributions , @xmath636 , such that @xmath637 , where @xmath638 is the fourier multiplier @xmath639 for any function @xmath640 defined on @xmath641 with @xmath76 , and any of the previously defined functional spaces , @xmath613 , we denote @xmath642 the space of functions such that @xmath643 is controlled in @xmath613 , uniformly for @xmath644 , and use double bar symbol for the associated norm : @xmath645 for @xmath624 , @xmath646 denotes the space of @xmath613-valued continuous functions on @xmath647 with continuous derivatives up to the order @xmath629 .
finally , we denote , in order to avoid confusions , @xmath648 we denote by @xmath649 and @xmath650 the euclidean inner product and norm on vector space @xmath651 or @xmath652 and @xmath653 the corresponding induced norm on @xmath654 , the space of @xmath29-by-@xmath29 matrices with real entries ( the choice of the norms has little significance ) . if the entries belong to a banach algebra @xmath613 ( _ e.g. _ @xmath655 or @xmath656 with @xmath657 ) , then we denote @xmath658 , @xmath659 and @xmath660 the corresponding inner product , vector and matrix norms .
we quickly recall the standard product , schauder and kato - ponce estimates in @xmath661 , @xmath420 .
proofs or references concerning the following results may be found for instance in @xcite .
consider scalar functions @xmath662 . by sobolev embedding
, one has @xmath663 and @xmath664 it follows that the product is well - defined and continuous .
moreover , one has @xmath665 and @xmath666 moreover , if @xmath667 with @xmath624 , @xmath668 and satisfies @xmath669 , then @xmath670 and @xmath671 a consequence of the above is of particular importance in our setting .
let @xmath672 be such that @xmath673 .
then for any @xmath674 , @xmath675 and @xmath676 ( it suffices to remark @xmath677 , and apply the schauder estimate with @xmath678 a smooth function such that @xmath679 if @xmath680 and @xmath681 if @xmath682 ) . finally , we have the celebrated kato - ponce estimate for commutators : @xmath683 \partial_x g}_{l^2 } \leq c \norm{f}_{h^{\sigma}}\norm{\partial_x g}_{h^{\sigma-1}}\leq c \norm{f}_{h^{\sigma}}\norm{g}_{h^\sigma}.\ ] ] in this section , we recall some results concerning bony s paradifferential calculus .
we follow the definition and most of the notations of @xcite , although the latter is restricted to scalar functions and operators . the generalizations to ( finite dimensional ) vector spaces brings no additional difficulty since each operation
may be reduced to a linear combination of entrywise scalar operations .
[ d.symbols ] let @xmath624 and @xmath684 .
we denote @xmath685 the space of locally bounded functions @xmath686 which are @xmath687 with respect to @xmath191 and such that for any @xmath688 , @xmath689 belongs to @xmath690 and there exists a constant @xmath691 such that @xmath692 where we use the standard multi - index notation for @xmath626-differentiation . for @xmath693 ,
we denote @xmath694 given a symbol @xmath693 , one can associate a suitable paradifferential operator .
[ d.paradiff ] let @xmath693
. then we define the paradifferential operator @xmath695 , with symbol @xmath696 , by @xmath697 where @xmath698 is defined with @xmath699 where @xmath700 is the ( component - by - component ) fourier transform of @xmath701 with respect to @xmath702 , and @xmath703 is an _
admissible cut - off _ in the sense of ( * ? ? ?
* definition 5.1.4 ) , and whose expression does not need to be precised . we now give some results used in this work .
[ p.paradiff-estimates ] let @xmath704 and assume @xmath705 . 1
if @xmath706 , then for any @xmath634 , @xmath707 is bounded , and @xmath708 2 . if @xmath709 and @xmath710 , then @xmath711 is bounded , and @xmath712 3 . if @xmath709 , then @xmath713 is bounded , and @xmath714 we will also make use of the particular cases of fourier multipliers and paraproducts : [ p.paradiff-particular cases ] let @xmath693 with @xmath715 1 .
if @xmath696 depends only on @xmath716 : @xmath717 , then @xmath718 where @xmath719 is the fourier multiplier associated with @xmath696 , _
i.e. _ @xmath720 2 .
if @xmath696 depends only on @xmath702 : @xmath721 with @xmath624 , then @xmath722
in this section , we give some information on the eigenstructure of the operators at stake in the multilayer shallow water model , namely @xmath226 $ ] defined in ; the translation in terms of the initial formulation , or @xmath723 $ ] , is immediate through the similarity transform .
more precisely , we show that the above matrices are semisimple provided that each layer s depth is positive and the shear velocities are sufficiently small .
this provides the complete eigenstructure of the full symbol @xmath724+\xi^y{{\sf b}}^y[{{\sf v}}]$ ] thanks to the rotational invariance property ( see lemma [ l.b ] ) .
we use this result in order to construct a symmetrizer of the system with the desired properties described in section [ s.normal-form ] ( lemma [ l.t ] ) , but we believe that such information is of independent interest . indeed , despite numerous works on the subject , the available information on the domain of hyperbolicity and eigenstructure of the multilayer shallow - water model is very sparse outside of the one or two - layer situation . the one - layer case is very classical and there is no need to discuss the subject here .
pioneer work on the two - layer case , in the limit @xmath725 and dimension @xmath101 setting , include @xcite for the free - surface case and @xcite for the rigid - lid situation .
we let the reader refer to @xcite and references therein for the case of @xmath726 .
additional , numerical information may be found in @xcite .
sufficient criteria for the hyperbolicity of the bi - fluidic shallow water model in the general situation of dimension @xmath8 and free - surface situation are provided in @xcite , while the rigid - lid setting is treated in @xcite .
starting from @xmath727 layers , explicit results become out of reach except for very specific situations ; see @xcite . in the general case of @xmath29 layers , the author provides in
@xcite a non - explicit sufficient condition for strict hyperbolicity in the case of dimension @xmath101 , and for @xmath728 , using that the eigenproblem , in absence of shear velocities , reduces to a finite - difference analogue of the sturm - liouville problem ( a similar tridiagonal reduction already appeared much earlier in the literature ; see @xcite and references therein ) . in @xcite , monjarret provides a sufficient criterion for hyperbolicity , for @xmath101 or @xmath8 , by exhibiting the symmetrizer recalled in section [ s.wp-naive ] .
very precise information concerning the eigenstructure are also provided in an asymptotic limit which does not fit in our situation when @xmath729 , since it requires a sharp scale separation for densities between each layer ; see ( * ? ? ?
* ( 4.1 ) ) . in this section
, we build upon these works , by showing that the strategy of @xcite extends to dimension @xmath8 and arbitrarily small density contrast @xmath62 ; under hypotheses on the flow which are fully consistent ( although , again , non - explicit ) with the criteria given in @xcite . roughly speaking , we show that provided that shear velocities are not too large , the nonlinear evolution of the flow maintains @xmath29 modes of propagation .
one of them is the barotropic mode , and is responsible for the singular time oscillations of the system in the limit @xmath100 . in our scaling , the wave speed of the @xmath730 baroclinic modes are uniformly bounded from above and from below , and remain isolated if the shear velocities are sufficiently small .
[ l.b - trivialeigen ] let @xmath280 .
there are @xmath29 linearly independent eigenvectors of @xmath732 $ ] , with corresponding eigenvalue @xmath733 , as given by , associated with rank - one eigenprojections @xmath734 { \stackrel{\rm def}{=}}({{\sf j}}^f[{{\sf v}}])^{-1 } { \sf \pi}_{2n+n } { { \sf j}}^f[{{\sf v } } ] , \ ] ] where @xmath735 is the orthogonal projection onto the @xmath4 variable .
the result is straightforward when using the block formulation of @xmath736 $ ] , , and the similarity transformation , _
i.e. _ @xmath226 = ( { { \sf j}}^f[{{\sf v}}])^{-1}{{\sf a}}^x[f({{\sf v } } ) ] { { \sf j}}^f[{{\sf v}}]$ ] . [ l.b0-eigen ] let @xmath737 .
then for any @xmath162 , @xmath738 $ ] has @xmath142 distinct , real , non - zero eigenvalues , @xmath739 , @xmath740 , with @xmath741 moreover , one can set @xmath742 such that if @xmath83 , then the associated eigenprojections satisfy ( recall the definition of @xmath744 in ) @xmath745 } + \norm { { { \sf p}}_{\pm n}[{{\sf z } } ] } & \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf z } } } ) \qquad ( n\geq 2),\\ \label{pz-2 } \norm { { { \sf p}}_{\pm 1}[{{\sf z } } ] { ( { { \sf id}}-{\sf \pi}_{\rm f}^x)}}+ \norm { { ( { { \sf id}}-{\sf \pi}_{\rm f}^x)}{{\sf p}}_{\pm 1}[{{\sf z } } ] } & \leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf z}}}),\\ \label{pz-3 } \norm { { { \sf p}}_{\pm n}[{{\sf z } } ] { { \sf \pi}_{\rm f}^x}}+ \norm { { { \sf \pi}_{\rm f}^x}{{\sf p}}_{\pm n}[{{\sf z } } ] } & \leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf z}}})\ \qquad ( n\geq 2 ) .\end{aligned}\ ] ] all these objects are smooth with respect to @xmath83 and @xmath737 . the eigenvalue problem concerning @xmath738 $ ] is related by to the one for @xmath746 $ ] , which given the simple block - structure exhibited in may be reduced , for @xmath747 , to the eigenvalue problem for the following tridiagonal matrix @xmath748 ( recall notations and identities in the proof of proposition [ p.wp-naive ] ) .
equivalently , we consider the eigenvalue problem for @xmath749 since @xmath750 is a real , symmetric tridiagonal matrix with non - zero ( positive ) off - diagonals entries @xcite , there exists @xmath751 and @xmath752 an orthonormal basis of @xmath651 such that @xmath753 since @xmath754 , one may check that @xmath755 where @xmath756 is the determinant of the @xmath518-by-@xmath518 upper - left submatrix ( _ i.e. _ leading principal minor ) of @xmath750 , from which we deduce @xmath757 . now , let us consider @xmath758 _ i.e. _ the matrix obtained from @xmath750 by setting to zero the first row s and column s entries .
using the above analysis on the leading principal minor of order 1 , one obtains immediately that there exists @xmath759 and @xmath760 , such that @xmath761 ( @xmath112 ) .
the above eigenvalues and eigenvectors depend continuously on the parameters which are , by assumption , bounded in a compact set ; in particular there exists @xmath762 , for a given @xmath29 , such that @xmath763 again by continuity , one can set @xmath764 such that if @xmath83 , then @xmath765 satisfy the above ( replacing @xmath766 with @xmath767 ) .
furthermore , we have ( augmenting @xmath766 if necessary ) @xmath768 notice now that the above analysis is not restricted to @xmath162 . indeed ,
the formula for @xmath750 above defines a ( complex ) tridiagonal matrix for any @xmath771 and is single - valued , as least for @xmath62 sufficiently small . since the off - diagonal entries do not vanish , @xmath750 , and therefore @xmath738 $ ] has @xmath142 distinct and non - zero eigenvalues for any @xmath772 .
moreover , we are in the situation of ( * ? ? ?
* theorem ii.1.10 ) , namely @xmath750 is normal for a sequence @xmath773 ( simply restricting to @xmath774 ) , and therefore @xmath775 and the associated eigenprojection are holomorphic around @xmath776 . the extra information concerning the corresponding eigenprojections in the limit of vanishing @xmath62 are obtained thanks to standard perturbation theory ( * ? ? ? * chapter ii.1 ) . indeed , from ( see also the proof of lemma [ l.b ] ) ,
one can write for any @xmath771 , @xmath777 = { { \sf l}}+ { \varrho}\ \delta{{\sf b}}[{{\sf z}}]\ ] ] with @xmath778 and @xmath779 $ ] is smooth with respect to @xmath328 and holomorphic with respect to @xmath62 , and satisfies @xmath780}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf z } } } ) .\ ] ] it is obvious that @xmath135 has only two non - zero eigenvalues : @xmath781 with @xmath782 . using that @xmath783 ( resp .
@xmath784 ) is a simple eigenvalue of @xmath785 $ ] ( resp .
@xmath135 ) , we introduce the dunford - taylor integral for the eigenprojection @xmath786=\frac{-1}{2\pi i}\int_{\gamma_{\pm1}}\sum_{k=0}^\infty ( { { \sf l}}-\eta{{\sf id}})^{-1}(-{\varrho}\delta{{\sf b}}[{{\sf z } } ] ( { { \sf l}}-\eta{{\sf id}})^{-1})^k\ d\eta,\ ] ] where @xmath787 is a positively oriented closed curve enclosing @xmath783 as well as @xmath784 , but excluding the other eigenvalues of @xmath785 $ ] and @xmath135 , namely @xmath190 and @xmath788 ( @xmath789 ) .
one can restrict @xmath790 , @xmath764 such that @xmath791 } \max_{\eta\in \gamma_{\pm}}\norm { ( { { \sf l}}-\eta{{\sf id}})^{-1}}\leq 1/2 $ ] , and therefore the series in the dunford - taylor integral is uniformly convergent . in particular , @xmath786 = \frac{-1}{2\pi i}\int_{\gamma_{\pm1 } } ( { { \sf l}}-\eta{{\sf id}})^{-1 } \ d\eta+\frac{{\varrho}}{2\pi i}\int_{\gamma_{\pm1}}({{\sf l}}-\eta{{\sf id}})^{-1 } \sum_{k=1}^\infty { \varrho}^{k-1}\left(- \delta{{\sf b}}[{{\sf z } } ] ( { { \sf l}}-\eta{{\sf id}})^{-1}\right)^k\ d\eta.\ ] ] the first term on the right hand side is exactly the eigenprojection onto @xmath792 , and the series in the second term is immediately estimated
. we deduce @xmath793 } \leq c(h_0^{-1},{\varrho},\norm{{{\sf z } } } ) , \quad \norm { { ( { { \sf id}}-{\sf \pi}_{\rm f}^x)}{{\sf p}}_{\pm 1}[{{\sf z } } ] } + \norm { { { \sf p}}_{\pm 1}[{{\sf z } } ] { ( { { \sf id}}-{\sf \pi}_{\rm f}^x ) } } \leq { \varrho}\ c(h_0^{-1},{\varrho},\norm{{{\sf z}}}),\ ] ] thus the first part of and is proved .
one can not directly use the same technique for the other eigenvalues , as they correspond to an _ exceptional _ point of @xmath785={{\sf l}}+{\varrho}\delta{{\sf b}}[{{\sf z}}]$ ] , namely @xmath190 is an eigenvalue of @xmath135 with algebraic multiplicity @xmath794 , and splits for @xmath772 in @xmath795 distinct eigenvalues , @xmath796 ( completed with the @xmath797 linearly independent eigenvectors given in lemma [ l.b - trivialeigen ] which remain in the kernel ) . as a consequence , the dunford - taylor integral around this group of eigenvalues only yields a control of the total projection , namely the sum of the corresponding eigenprojections ; or , if one integrates around a single eigenvalue , one can not in general ensure that the series is convergent , even for @xmath62 small .
however , we have seen that @xmath775 is holomorphic in @xmath62 and converges towards @xmath798 as @xmath100 , so that @xmath799 is holomorphic near @xmath776 .
this means ( * ? ? ?
* theorem ii.1.8 ) that the exceptional point is not a _ branch _ point , and therefore @xmath800 $ ] is single - valued , but still may have a pole at @xmath776 .
using now that @xmath190 is a semisimple eigenvalue of the unperturbed operator , @xmath135 , and that we have shown that @xmath801 , the eigenvalues of @xmath785 $ ] splitting from @xmath190 are simple ( as they are one - dimensional ) , we may use the so - called reduction process , and deduce ( * ? ? ?
* theorem ii.2.3 ) that the associated spectral projections , @xmath800 $ ] , are actually holomorphic at @xmath776 .
this shows , and by continuity that @xmath800{\left\vert_{_{{\varrho}=0}}\right.}$ ] is a projection onto the kernel of @xmath135 .
thus @xmath802{\left\vert_{_{{\varrho}=0}}\right.}){{\sf \pi}_{\rm
f}^x}={{\sf \pi}_{\rm f}^x}({{\sf p}}_{\pm n}[{{\sf z}}]{\left\vert_{_{{\varrho}=0}}\right.})={\sf 0}_{n(1+d)}$ ] , and follows .
lemma [ l.b0-eigen ] is proved .
[ l.b - eigen ] let @xmath280 .
then one can set @xmath804 such that if @xmath83 and @xmath140 satisfies additionally @xmath805 then the matrix @xmath732 $ ] is diagonalizable .
in addition to the @xmath29 `` trivial '' eigenvectors described in lemma [ l.b - trivialeigen ] , @xmath732 $ ] has @xmath142 eigenvectors corresponding to distinct and real eigenvalues , @xmath806 , such that @xmath807 moreover , there exists @xmath286 such that the associated spectral projections are smooth with respect to @xmath280 ; and , for any @xmath809 , @xmath810 - { { \sf p}}_{\pm 1}[{{\sf v}}_2 ] } & \leq c_1\ { \varrho}\norm{{{\sf v}}_1-{{\sf v}}_2},\\ \label{pv-2 } \norm { ( { { \sf p}}_{\pm n}[{{\sf v}}_1 ] - { { \sf p}}_{\pm n}[{{\sf v}}_2]){{\sf \pi}_{\rm f}^x } } & \leq c_1\ { \varrho}\norm{{{\sf v}}_1-{{\sf v}}_2 } \qquad ( n\geq 2),\\ \label{pv-3 } \norm { { { \sf p}}_{\pm n}[{{\sf v}}_1 ] - { { \sf p}}_{\pm n}[{{\sf v}}_2 ] } & \leq \ c_1 \big(\norm{{({{\sf id}}-{\sf \pi}_{\rm f})}({{\sf v}}_1-{{\sf v}}_2)}+{\varrho}\norm{{{\sf v}}_1-{{\sf v}}_2}\big ) \qquad ( n\geq 2),\end{aligned}\ ] ] where @xmath811 .
we shall use perturbation arguments , starting from the knowledge that , thanks to lemma [ l.b0-eigen ] , the non - zero eigenvalues of @xmath785 $ ] , for any @xmath737 , are simple . in
what follows , we denote @xmath737 the vector obtained from setting to zero all @xmath4 entries of @xmath140 with @xmath329 . identifying with , we may write ( improving the description provided in lemma [ l.b ] ) @xmath812 = { { \sf b}}^x[{{\sf z } } ] + { \varrho}k^x { { \sf id}}+ { \varrho}\delta{{\sf b}}_{\rm f}[{{\sf v}}]+ { \varrho}\delta{{\sf b}}_{\rm s}[{{\sf v}}],\ ] ] where one can choose @xmath813 ( for example @xmath814 ) such that @xmath815 contains ( at first order in @xmath62 ) only contributions from shear velocities @xmath816 , while @xmath817 contains a leading - order contribution in @xmath818 , but only on `` fast variable '' entries
. more precisely , one has @xmath819 } + \norm { \delta{{\sf b}}_{\rm f}[{{\sf v } } ] } & \leq \
c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \norm{{{\sf v } } } , \\
\label{est - deltabs } \norm { \delta{{\sf b}}_{\rm s}[{{\sf v } } ] } & \leq \ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ( \nu^{-1}+{\varrho}\norm{{{\sf v } } } ) , \\ \label{est - deltabf } \norm { \delta{{\sf b}}_{\rm f}[{{\sf v } } ] { ( { { \sf id}}-{\sf \pi}_{\rm f}^x ) } } & \leq \ { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \norm{{{\sf v } } } . \end{aligned}\ ] ] since all the non - trivial eigenvalues of @xmath785 $ ] ( and therefore the ones of @xmath820+{\varrho}k^x{{\sf id}}$ ] ) are simple , we may use the dunford - taylor integral @xmath821=\frac{-1}{2\pi i}\int_{\gamma_{\pm n}}\sum_{k=0}^\infty { { \sf r}}(\eta)\big(-{\varrho}(\delta{{\sf b}}_{\rm s}+\delta{{\sf b}}_{\rm f } ) { { \sf r}}(\eta)\big)^k\ d\eta,\ ] ] where @xmath822 is a positively oriented closed curve enclosing the eigenvalue @xmath823 , but excluding the other eigenvalues of @xmath820+{\varrho}k^x{{\sf id}}$ ] , and @xmath824+{\varrho}k^x{{\sf id}}-\eta{{\sf id}})^{-1}= \sum_{n'=1}^n \frac { { { \sf p}}_{n'}[{{\sf z } } ] } { { \varrho}\mu_{n'}({{\sf z}})+{\varrho}k^x-\eta}+\frac { { { \sf p}}_{-n'}[{{\sf z}}]}{{\varrho}\mu_{-n'}({{\sf z}})+{\varrho}k^x-\eta}+\frac{{{\sf p}}^0_{n'}[{{\sf z}}]}{{\varrho}k^x-\eta},\ ] ] with @xmath825,{{\sf p}}^0_n[{{\sf z}}]$ ] have been defined in lemmata [ l.b - trivialeigen ] and [ l.b0-eigen ] .
let us first consider the case @xmath826 . from lemma [ l.b0-eigen ] ,
there exists @xmath762 such that if @xmath826 , one can choose @xmath827 as the circle of center @xmath828 and of radius @xmath829 . using and
, one can restrict @xmath83 with @xmath830 such that the series in is immediately convergent and @xmath831-{{\sf p}}_{\pm 1}[{{\sf z } } ] } \leq \frac{1}{\pi } \sum_{n=1}^\infty { \varrho}^n \big(\sup_{\gamma_{\pm n}}\norm { { { \sf r}}(\eta)}\big)^{1+n } \big(\norm { \delta{{\sf b}}_{\rm s}}+\norm{\delta{{\sf b}}_{\rm f}}\big)^n\leq { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \norm{{{\sf v}}},\ ] ] since @xmath832 = \frac{-1}{2\pi i}\int_{\gamma_{\pm n}}{{\sf r}}(\eta)\ d\eta$ ] is the first term of the series .
since @xmath833,{{\sf p}}_{\pm 1}[{{\sf v}}]$ ] are rank - one , one has @xmath834 { { \sf p}}_{\pm1}[{{\sf v}}]\big)\ = \ { \varrho}\mu_{\pm 1}({{\sf z } } ) \ + \ \operatorname{tr}\big ( { { \sf b}}^x[{{\sf v } } ] ( { { \sf p}}_{\pm1}[{{\sf v}}]- { { \sf p}}_{\pm1}[{{\sf z}}])\big)\ + \
\operatorname{tr}\big ( ( { { \sf b}}^x[{{\sf v } } ] -{{\sf b}}^x[{{\sf z } } ] ) { { \sf p}}_{\pm1}[{{\sf z}}]\big ) , \ ] ] and the upper and lower bound on @xmath835 in follows from the ones on @xmath836 given in , and the above estimate with in lemma [ l.b ] .
we now turn to the case @xmath837 . in this case
, we set @xmath822 as the circle of center @xmath828 and of radius @xmath838 ( with @xmath766 as in ) , thus we may only ensure @xmath839 as a consequence , we need the precised estimates - as well as the decomposition into partial fraction to ensure that the series converge .
indeed , thanks to , one may augment @xmath840 and lower @xmath841 in order to ensure @xmath842}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ( { \varrho}\norm{{{\sf v}}}+\nu^{-1 } ) \leq \frac12.\ ] ] now , we use that for any @xmath843 , @xmath844 { { \sf p}}_{\pm n'}[{{\sf z } } ] = \big ( \delta{{\sf b}}_{\rm f}[{{\sf v } } ] { ( { { \sf id}}-{\sf \pi}_{\rm f}^x)}\big ) { { \sf p}}_{\pm n'}[{{\sf z } } ] + \delta{{\sf b}}_{\rm f } [ { { \sf v } } ] \big ( { { \sf \pi}_{\rm f}^x}{{\sf p}}_{\pm n'}[{{\sf z } } ] \big ) .\ ] ] thus by and , one has @xmath845{{\sf
p}}_{\pm n'}[{{\sf z } } ] } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \ { \varrho}\norm{{{\sf v } } } \qquad ( n'\geq 2).\ ] ] the contribution from @xmath846 $ ] is straightforwardly estimated , and the contribution from @xmath847 with @xmath848 contains no difficulty , since , @xmath849 is uniformly bounded .
altogether , one has @xmath850{{\sf r}}(\eta ) } \leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \ { \varrho}\norm{{{\sf v}}}.\ ] ] restricting @xmath83 if necessary , the series in converges and @xmath851-{{\sf p}}_{\pm n}[{{\sf z } } ] } \leq \ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) ( { \varrho}\norm{{{\sf v}}}+\nu^{-1 } ) .\ ] ] in the same way , and using , one easily sees that @xmath852 and therefore @xmath853-{{\sf p}}_{\pm n}[{{\sf z } } ] ) { { \sf \pi}_{\rm f}^x } } \leq \ { \varrho}\ c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v } } } ) \norm{{{\sf v } } } .\ ] ] estimates , , are obtained identically as above , using the decomposition @xmath855={{\sf
b}}^x[{{\sf v}}_1]+{\varrho}(k^x_2 -k^x_1 ) { { \sf id}}+\delta{{\sf b}}[{{\sf v}}_1,{{\sf v}}_2]\ ] ] where @xmath856 \ { \stackrel{\rm def}{=}}\ { { \sf b}}^x[{{\sf z}}_2]-{{\sf b}}^x[{{\sf z}}_1 ] + { \varrho}(\delta{{\sf b}}_{\rm f}[{{\sf v}}_2]-\delta{{\sf b}}_{\rm f}[{{\sf v}}_1])+ { \varrho}(\delta{{\sf b}}_{\rm s}[{{\sf v}}_2]-\delta{{\sf b}}_{\rm s}[{{\sf v}}_1 ] ) , \ ] ] and remarking that @xmath857-\delta{{\sf b}}_{\rm f}[{{\sf v}}_1 ] } + \frac1{\varrho}\norm { ( \delta{{\sf b}}_{\rm f}[{{\sf v}}_2]-\delta{{\sf b}}_{\rm f}[{{\sf v}}_1 ] ) { ( { { \sf id}}-{\sf \pi}_{\rm f}^x ) } } & \leq c_1 \norm{{{\sf v}}_1-{{\sf v}}_2}\ , \end{aligned}\ ] ] with @xmath811 . this concludes the proof of lemma [ l.b - eigen ]
. the proof of lemma [ l.b - eigen ] is somewhat cumbersome and rely on delicate properties of @xmath226 $ ] , namely with and , because we wanted to be as precise as possible as for the hyperbolicity conditions ( see remark [ r.naive ] ) .
the proof is considerably shortened and appears more robust if one replaces the assumption of lemma [ l.b - eigen ] with the more stringent @xmath858 , as one may then simply use @xmath226={{\sf b}}^x[{{\sf z } } ] + { \varrho}\delta{{\sf b}}[{{\sf v}}]$ ] with @xmath859}\leq c({\mathfrak{m}},h_0^{-1},\norm{{{\sf v}}})\norm{{{\sf v}}}$ ] in lieu of .
n. masmoudi .
examples of singular limits in hydrodynamics . in _ handbook of differential equations : evolutionary equations .
equ . , pages 195275 .
elsevier / north - holland , amsterdam , 2007 .
j. schijf and j. schnfled .
theoretical considerations on the motion of salt and fresh water . in _
minnesota international hydraulics convention , a joint meeting of international association for hydraulic research and hydraulics division , american society civil engineers _ , pages 321333 , 1953 . | we study the inviscid multilayer saint - venant ( or shallow - water ) system in the limit of small density contrast .
we show that , under reasonable hyperbolicity conditions on the flow and a smallness assumption on the initial surface deformation , the system is well - posed on a large time interval , despite the singular limit . by studying the asymptotic limit
, we provide a rigorous justification of the widely used rigid - lid and boussinesq approximations for multilayered shallow water flows .
the asymptotic behaviour is similar to that of the incompressible limit for euler equations , in the sense that there exists a small initial layer in time for ill - prepared initial data , accounting for rapidly propagating `` acoustic '' waves ( here , the so - called barotropic mode ) which interact only weakly with the `` incompressible '' component ( here , baroclinic ) . |
_ topological recursion _
developed by eynard , orantin and chekhov produces invariants of a riemann surface @xmath3 equipped with two meromorphic functions @xmath4 and a bidifferential @xmath5 for @xmath6 @xcite .
we require the zeros of @xmath7 to be simple and refer to the data @xmath8 as a _
spectral curve_. for integers @xmath9 and @xmath10 , the invariant @xmath11 is a multidifferential on @xmath3 or , in other words , a tensor product of meromorphic 1-forms on @xmath12 . in this paper ,
all spectral curves will have underlying riemann surface @xmath13 and bidifferential @xmath14 . in that case
, we may specify the spectral curve parametrically via the meromorphic functions @xmath15 and @xmath16 .
we call a spectral curve _ regular _ if it is non - singular at the zeros of @xmath7 for example , if the curve is non - singular .
see section [ sec : eo ] for precise definitions .
the invariants @xmath17 of the _ airy curve _
@xmath2 are ( total derivatives of ) the following generating functions for intersection numbers of chern classes of the tautological line bundles @xmath18 on the moduli space of stable curves @xmath19 @xcite .
@xmath20 a regular spectral curve locally resembles the airy curve @xmath2 near zeros of @xmath7 , which are assumed to be simple .
this leads to universality in the behaviour of topological recursion on regular spectral curves the invariants are related to intersection theory on @xmath19 .
three progressively more refined statements of this relationship are as follows . 1 .
eynard and orantin @xcite proved that the invariants @xmath17 behave asymptotically near a regular zero of @xmath7 like the invariants @xmath17 of the airy curve @xmath2 at the origin .
hence , they store the intersection numbers appearing in equation .
eynard @xcite pushed this further , proving that the lower order asymptotic terms of @xmath17 on a regular spectral curve also encode intersection numbers .
these come in the form of explicit combinations of hodge integrals on @xmath19 and a generalisation @xmath21 , which he calls the moduli space of @xmath22-coloured stable curves .
3 . for a special class of regular spectral curves ,
dunin barkowski , orantin , shadrin and
spitz @xcite extended the results of eynard , by proving that the multidifferentials @xmath17 encode ancestor invariants in a cohomological field theory , which is fundamentally related to intersection theory on @xmath19 . in this paper
, we consider irregular spectral curves that locally resemble the curve @xmath1 near some zeros of @xmath7 .
( in section [ sec : irr ] , we show that any other local irregular behaviour is ill - behaved . ) for such curves , the local behaviour of the invariants @xmath17 is no longer determined by the intersection numbers of equation .
an analogue of statement ( 1 ) above holds , although we do not currently have an analogue of equation to relate the invariants of the spectral curve @xmath1 to a moduli space .
instead , we consider a specific problem the enumeration of dessins denfant which is governed by an irregular spectral curve .
we rely on this concrete example to shed light on the local behaviour of all irregular spectral curves . to achieve this ,
we first show that the enumeration of dessins denfant satisfies topological recursion on the irregular spectral curve @xmath0 .
we then prove a three - term recursion for its 1-point invariants , and use this to determine an exact formula for the 1-point invariants of the spectral curve @xmath1 .
dessin denfant _ is a bicoloured graph embedded in a connected orientable surface , such that the complement is a union of disks .
the term _ bicoloured _ means that the vertices are coloured black and white such that each edge is adjacent to one vertex of each colour .
consequently , the underlying graph of a dessin denfant is necessarily bipartite .
one can interpret a dessin denfant as a branched cover @xmath23 unramified over @xmath24 , often referred to as a _
belyi map_. the bicoloured graph is given by @xmath25)\subset\sigma$ ] , with the points @xmath26 representing black vertices and the points @xmath27 representing white vertices .
let @xmath28 be the set of all genus @xmath29 belyi maps @xmath23 with ramification divisor over @xmath30 given by @xmath31 , where the points over @xmath30 are labelled @xmath32 .
two belyi maps @xmath33 and @xmath34 are isomorphic if there exists a homeomorphism @xmath35 that covers the identity on @xmath36 and preserves the labelling over @xmath30 .
equivalently , one can interpret @xmath28 as the set of connected genus @xmath29 dessins denfant with @xmath37 labelled boundary components of lengths @xmath38 . by a boundary component of a dessin denfant , we mean a cycle in the underlying graph corresponding to the boundary of one of the labelled disks in @xmath39 .
we require an isomorphism between two dessins denfant to preserve the labelling on their boundary components . for any @xmath40 , define @xmath41 where @xmath42 denotes the automorphism group of the dessin denfant @xmath43 . for integers @xmath9 and @xmath10 ,
define the generating function @xmath44 if we let @xmath45 and define @xmath46 for @xmath47 , then we may observe that this generating function is well - behaved with respect to @xmath48 .
in particular , theorem [ th : main ] below implies that @xmath49 is a rational function of @xmath48 for @xmath50 , with poles only at @xmath51 of certain orders . the derivative @xmath52 is also rational and together with @xmath53 defines a plane curve known as the spectral curve .
more precisely , theorem [ th : main ] shows that the ( total derivatives of ) @xmath49 satisfy topological recursion on the spectral curve @xmath54 , given parametrically by @xmath55 furthermore , the topological recursion determines the generating functions @xmath49 uniquely .
[ th : main ] for @xmath56 , the multidifferential @xmath57 is the analytic expansion of the invariant @xmath17 of the spectral curve at the point @xmath58 . from the universality property described above
, the asymptotic behaviour of the generating function near its pole at @xmath59 is given by @xmath60 .
more precisely , we have @xmath61.\ ] ] note that the spectral curve given by equation is irregular .
the local behaviour near its poles @xmath62 is our main interest .
one immediate consequence of irregularity is the novel feature that the genus 0 generating functions @xmath63 are analytic at @xmath64 .
more generally , the orders of poles of @xmath49 at @xmath64 are independent of @xmath37 .
this is in contrast to @xmath65 having poles of order @xmath66 , which is the case for most of the spectral curves that appear in the literature .
properties of rational functions on the curve yield a structure theorem for @xmath67 see theorem [ th : struc ] as well as explicit formulae .
for example , we have @xmath68 another consequence of theorem [ th : main ] is a general property of the invariants @xmath17 , known as the _ dilaton equation_. for the spectral curve of interest , it implies that @xmath69 where one can make sense of evaluation at @xmath70 using the structure theorem for @xmath67 .
moreover , we prove in proposition [ th : pointed ] that for @xmath37 positive , @xmath71 has a combinatorial meaning it enumerates dessins denfant with @xmath72 black vertices labelled . the enumerative problem in this paper and the associated spectral curve given by equation are closely related to others in the literature .
topological recursion on rational spectral curves with @xmath73 describe enumeration of discrete surfaces @xcite , which includes the special case of lattice points in moduli spaces of curves @xcite , a more refined version of dessin enumeration @xcite , and the gromov witten invariants of @xmath36 @xcite .
however , note that each of these examples is governed by a regular spectral curve
. a _ quantum curve _ of a spectral curve @xmath74 is a schrdinger - type equation @xmath75 , where @xmath76 is a non - commutative quantisation of the spectral curve with @xmath77 and @xmath78 .
this differential operator annihilates a wave function @xmath79 , which is a formal series in @xmath80 associated to the spectral curve .
the path from the quantum curve to the spectral curve is well - defined in the semi - classical limit @xmath81 , the differential operator reduces to a multiplication operator that vanishes precisely on the spectral curve . on the other hand ,
constructing the quantum curve from the spectral curve is not canonical .
the main issues lie in the construction of the wave function and the ambiguity in ordering the non - commuting operators @xmath82 and @xmath83 .
one remedy for these issues is a conjectural construction of the wave function @xmath84 from the invariants @xmath17 of the spectral curve , suggested for example by gukov and sukowski @xcite .
we prove this conjecture for the spectral curve @xmath0 .
define the wave function as follows .
@xmath85\ ] ] [ thm : qua ] the quantum curve of @xmath0 is given by @xmath86 .
strictly speaking , to make sense of the action of a differential operator on a formal series in @xmath80 , it is necessary to know that all sums are finite . in section [ sec :
qua ] , we give a more precise statement of theorem [ thm : qua ] , in terms of a differential operator annihilating the formal series @xmath87[[x^{-1}]]$ ] .
+ one of the main purposes of the present paper is to understand the universality exhibited by the invariants associated to the irregular spectral curve @xmath88 at irregular zeros of @xmath7 , this plays the role of the airy curve at regular zeros of @xmath7 .
we expect many results relating the airy curve to invariants of spectral curves to have analogues in this setting .
in particular , we expect the invariants of our curve to be related to a new moduli space . in section
[ sec : asym ] , we apply topological recursion directly to the spectral curve given by equation .
we calculate putative volumes of these unidentified moduli spaces and dually , intersection numbers on them .
the invariants are non - zero only for positive genus , much like enumeration of branched covers of a torus or volumes of spaces of holomorphic differentials .
indirectly , we use the asymptotic behaviour of @xmath49 defined in equation near its pole @xmath64 to study the spectral curve .
more generally , the type of enumerative problem governed by necessarily has no contribution in genus 0 . to make this idea clearer ,
consider for the moment the enumeration of non - bipartite fatgraphs with bipartite boundary components in other words , boundary components of even lengths .
for example , the square graph pictured below left is bipartite and may be considered the boundary of the non - bipartite genus 1 fatgraph pictured below right .
( 0,0 ) circle(2 mm ) ; ( 0,2 ) circle(2 mm ) ; ( 2,2 ) circle(2 mm ) ; ( 2,0 ) circle(2 mm ) ; ( 0,0)(2,0)(2,2)(0,2)cycle ; ( 0,0 ) circle(10 mm ) ; ( 1,1 ) circle(10 mm ) ; ( 0,1 ) circle(2 mm ) ; define @xmath89 to be the weighted count of connected genus @xmath29 non - bipartite fatgraphs with labelled bipartite boundaries of lengths @xmath38 and such that the vertices are required to have valency greater than or equal to two see equation
. the valency condition on the vertices reduces the growth in @xmath90 from exponential to polynomial .
in particular , this invariant vanishes in genus zero since bipartite boundary components implies that the graph is bipartite for simple homological reasons .
one can show that @xmath89 is quasi - polynomial in @xmath91 modulo 2 .
an interesting invariant is obtained by measuring its failure to be polynomial , which we do in the following way .
recall that if @xmath92 is a quasi - polynomial modulo 2 , then it has a natural decomposition @xmath93 , where @xmath94 are polynomials . for the analogous decomposition of a quasi - polynomial @xmath95 in severable variables , it is the coefficient of @xmath96 that interests us .
@xmath97\ ] ] for example , @xmath98 where @xmath99 $ ] . from these expressions
, one can extract the non - polynomial parts @xmath100 and @xmath101 , which determine the invariants @xmath102 and @xmath103 of the spectral curve @xmath1 . more generally , the top degree part of @xmath104 is equivalent to the invariant @xmath11 of @xmath1 .
returning to the enumeration of dessins denfant , consider the three - term recursion @xmath105 which is proven in section [ sec:3term ] .
we remark that it has a rather different character to the topological recursion of theorem [ th : main ] .
in particular , it enables one to calculate @xmath106 recursively from @xmath107 for @xmath108 , without requiring @xmath109 for @xmath110 .
it is analogous to the three - term recursion for fatgraphs with one face of harer and zagier @xcite .
our three - term recursion implies a recursion satisfied by the 1-point invariants of the spectral curve @xmath1 , which then leads to the following exact formula .
[ th:1point ] the 1-point invariants of the spectral curve @xmath1 , given parametrically by @xmath111 and @xmath112 , are @xmath113 we would hope to recognise some type of intersection number in the formula of theorem [ th:1point ] analogous to the 1-point invariants of the airy curve @xmath2 , which are given by @xmath114 the spectral curve @xmath0 bears a clear resemblance to the regular spectral curve @xmath115 , which has been well - studied in the literature , since it arises from a matrix model with pure gaussian potential .
the invariants of the curve @xmath115 take the form @xmath116 , where @xmath117 here , @xmath118 denotes the set of connected genus @xmath29 fatgraphs graphs embedded in a connected orientable surface such that the complement is a union of disks with @xmath37 labelled boundary components of lengths @xmath119 .
again , we require an isomorphism between two fatgraphs to preserve the labelling on their boundary components .
note that each bipartite fatgraph can be bicoloured in two distinct ways , thereby producing two dessins denfant , so we obtain @xmath120 in fact , equality occurs when @xmath121 , although no such explicit relation exists in higher genus .
one can obtain the spectral curves @xmath115 and @xmath0 from the stieltjes transforms @xmath122 of the probability densities @xmath123}\ \qquad \text{and } \qquad \lambda(t)=\frac{1}{2\pi}\sqrt{\frac{4-t}{t}}\cdot\mathbbm{1}_{[0,4]}.\ ] ] these are known as the wigner semicircle distribution and the marchenko
pastur distribution , respectively .
it is elementary to show that the catalan numbers @xmath124 arise as moments of these probability densities .
@xmath125 regular behaviour of spectral curves arise from so - called _ soft edge _
statistics , while the irregular behaviour of the spectral curve given by equation arises from so - called _ hard edge _ statistics .
these terms refer to the behaviour of the associated probability densities at the endpoints of the interval of support .
topological recursion takes as input a _
spectral curve _
@xmath8 consisting of a compact riemann surface @xmath3 , a bidifferential @xmath126 on @xmath3 , and meromorphic functions @xmath127 .
we furthermore require that the zeros of @xmath7 are simple and disjoint from the zeros of @xmath128 @xcite .
a more general setup allows _ local spectral curves _ ,
in which @xmath3 is an open subset of a compact riemann surface . in this paper
, we deal exclusively with the case when @xmath3 is the riemann sphere @xmath13 , with global rational parameter @xmath129 , endowed with the bidifferential @xmath130 .
the two main examples that we consider take the pair of meromorphic functions to be @xmath131 and @xmath132 .
they are mild variants of the usual setup , since in both cases , @xmath128 has a pole at a zero of @xmath7 .
nevertheless , as we will see below , topological recursion is well - defined in this case and retains many of the desired properties , while losing some others . for integers @xmath133 and @xmath10 , topological recursion outputs multidifferentials @xmath134 on @xmath3
in other words , a tensor product of meromorphic 1-forms on the product @xmath12 , where @xmath135 . when @xmath50 , @xmath134 is defined recursively in terms of local information around the poles of @xmath136 for @xmath137 .
since each zero @xmath138 of @xmath7 is assumed to be simple , for any point @xmath139 close to @xmath138 , there is a unique point @xmath140 close to @xmath138 such that @xmath141 .
the recursive definition of @xmath134 uses only local information around zeros of @xmath7 and makes use of the well - defined map @xmath142 there .
the invariants are defined as follows , for @xmath143 and @xmath130 with @xmath144 .
start with the base cases @xmath145 for @xmath50 and @xmath146 , define @xmath147,\ ] ] where the outer summation is over the zeros @xmath138 of @xmath7 and the @xmath148 over the inner summation means that we exclude terms that involve @xmath149 .
we define @xmath150 by the following formula @xmath151 \ , { \mathrm{d}}x(z)}=\frac{1}{2[y(\hat{z})-y(z ) ] \ , x'(z)}\left ( \frac{1}{z - z_1}- \frac{1}{\hat{z}-z_1}\right)\frac{{\mathrm{d}}z_1}{{\mathrm{d}}z},\ ] ] which is well - defined in the vicinity of each zero of @xmath7 .
note that the quotient of a differential by the differential @xmath152 is a meromorphic function .
the recursion is well - defined even when @xmath153 has poles at the zeros @xmath138 of @xmath7 .
it does not use all of the information in the pair @xmath154 , but depends only on the meromorphic differential @xmath155 and the local involutions @xmath142 . for @xmath50 ,
the multidifferential @xmath17 is symmetric , with poles only at the zeros of @xmath7 and vanishing residues . in the case
@xmath132 , the differential @xmath156 is analytic and non - vanishing at the zero @xmath157 of @xmath7 .
this leads to the vanishing of its genus zero invariants , since the kernel @xmath158 has no pole at @xmath157 and by induction , @xmath159 has no pole at @xmath157 .
interesting invariants arise via @xmath102 , since a pole at @xmath157 occurs in the expression @xmath160 . non - triviality of @xmath102 leads to non - triviality of @xmath17 for all @xmath161 . for @xmath50 ,
the invariants @xmath17 of regular spectral curves satisfy the following _ string equations _ for @xmath162 @xcite .
@xmath163 they also satisfy the dilaton equation @xcite @xmath164 where the summation is over the zeros @xmath138 of @xmath7 and @xmath165 is an arbitrary antiderivative .
the dilaton equation enables the definition of the so - called _ symplectic invariants _ @xmath166
the dilaton equation still holds for irregular spectral curves whereas the string equations no longer hold .
the failure of the string equations can be explicitly observed for the curve @xmath1 . one can classify the local behaviour of a spectral curve near a zero of @xmath7 into four types one of these is regular and the other three are irregular . in all four cases , one can define multidifferentials @xmath11 using equation .
if @xmath138 is a zero of @xmath7 , then one of the following four cases must occur . 1 .
* regular .
* the form @xmath128 is analytic and @xmath167 .
+ equivalently , @xmath138 is a regular zero of @xmath7 if it is a smooth point of @xmath3 . in this case
, there is a pole of @xmath17 at @xmath138 of order @xmath66 @xcite .
2 . * irregular .
the form @xmath128 is analytic at @xmath138 and @xmath168 . +
this case is ill - behaved since @xmath169 loses the key property of symmetry under permutations of @xmath48 .
( note that the symmetry of @xmath17 is not a priori apparent , since the recursion of equation treats @xmath170 as special . ) for example , if we consider the rational spectral curve given parametrically by @xmath111 and @xmath171 , then topological recursion yields @xmath172.\ ] ] 2 .
the meromorphic function @xmath153 has a pole at @xmath138 of order greater than one .
+ in this case , the kernel @xmath173 defined above has no pole at @xmath138 due to the pole of @xmath153 that appears in the denominator .
the residue at @xmath138 in equation therefore vanishes and one obtains no contribution from a neighbourhood of @xmath138 .
the invariants in this case match those of the local spectral curve obtained by removing the point @xmath138 .
the meromorphic function @xmath153 has a simple pole at @xmath138 . +
this case is the main concern of the present paper .
again , the kernel @xmath173 defined above has no pole at @xmath138 , but a pole of @xmath174 at @xmath138 allows non - zero invariants to survive .
the invariants enjoy many of the properties of the invariants for regular curves , such as symmetry of @xmath169 under permutations of @xmath48 .
the pole of @xmath17 at @xmath138 is now of order @xmath175 , which follows from the local analysis in section [ sec : asym ] .
we conclude that the only interesting cases are ( 1 ) and ( 2c ) , which involve regular zeros of @xmath7 or a zero of @xmath7 at which @xmath153 has a simple pole .
if case ( 2a ) is to prove interesting , then one would probably need to adjust the definition of topological recursion in order to recover the symmetry of the invariants .
kazarian and zograf @xcite prove that @xmath176 satisfies the recursion @xmath177\ ] ] for @xmath178 and the base case @xmath179 .
the proof uses an elementary cut - and - join argument that is a variation of the tutte recursion @xcite .
note that @xmath180 is a catalan number .
this is due to the fact that the recursion of equation in the case @xmath181 reproduces the catalan recursion and initial condition @xmath182 the recursion is equivalent to the fact that the generating functions @xmath183 satisfy loop equations @xmath184+\frac{\delta_{g,0 } \
, \delta_{n,1}}{x_1}. \nonumber \ ] ] the solution of the loop equations for @xmath185 defines the spectral curve via the equation @xmath186 * the coefficient of @xmath187 in @xmath188 is @xmath189 . *
the coefficient of @xmath187 in @xmath190 is @xmath191 . *
the coefficient of @xmath187 in @xmath192 is @xmath193 . *
the coefficient of @xmath187 in @xmath194 $ ] is @xmath195 .
this observation uses the fact that @xmath196 the loop equations are almost a special case of the loop equations appearing in the work of eynard and orantin ( * ? ? ?
* theorem 7.2 ) . using their notation , equation would correspond to @xmath197 and @xmath198 would not be polynomial in @xmath199 .
note that such a choice of @xmath200 and @xmath201 has no meaning there .
define @xmath202 to be the set of genus @xmath29 dessins without vertices of valence 1 and with @xmath37 labelled boundary components of lengths @xmath38 .
we refer to such dessins without vertices of valence 1 as _ pruned_. the notion of pruned structures has found applications for various other problems @xcite and will allow us to prove polynomiality for the dessin enumeration here .
the main idea is that a dessin can be created from a pruned dessin by gluing planar trees to the boundary components .
the bicolouring of the vertices extends in a unique way to the additional trees .
conversely , one obtains a unique pruned dessin from a dessin via the process of pruning in other words , repeatedly removing degree one vertices and their incident edges until no more exist . by gluing planar trees
, the count @xmath210 contributes to @xmath211 for all non - negative integers @xmath212 .
the contribution is equal to @xmath210 multiplied by the number of ways to glue planar trees with a total of @xmath213 edges to boundary component 1 , multiplied by the number of ways to glue planar trees with a total of @xmath214 edges to boundary component 2 , and so on .
the number of ways to glue @xmath215 edges to a boundary component of length @xmath216 can be computed as follows .
it is simply the number of ways to pick rooted planar trees @xmath217 with @xmath215 edges in total .
there are @xmath218 rooted planar trees with @xmath219 edges , where @xmath220 is the sequence of catalan numbers .
so the number of ways to choose rooted planar trees @xmath217 with @xmath215 edges in total is simply the @xmath221 coefficient of @xmath222 .
if we call this number @xmath223 , we obtain the following formula .. ] @xmath224 the next proposition shows that @xmath228 satisfies a _ functional recursion _
, in the sense that it involves only terms with simpler @xmath229 complexity on the right hand side . this will be useful in understanding the pole structure of the generating function .
such a recursion is in contrast with equation in the non - pruned case , which includes @xmath229 terms on both sides of the equation .
[ th : recprune ] the pruned dessin enumeration satisfies the following recursion for @xmath230 .
@xmath231 here , @xmath232 and we set @xmath233 for @xmath234 . the word _
stable _ over the summation indicates that we exclude all terms that involve @xmath235 or @xmath236 .
we count dessins in @xmath237 with a marked edge . the most obvious way to count such objects is to choose a dessin in the set , which can be accomplished in @xmath238 ways , and then to choose a suitable edge , which can be accomplished in @xmath239 ways . so the total number of such objects is @xmath240 , which forms the left hand side of the recursion .
to form the right hand side of the recursion , we count the same objects in the following way . for a dessin in @xmath237 with a marked edge , remove the marked edge and repeatedly remove degree 1 vertices and their incident edges to obtain a pruned dessin .
one of the following three cases must arise . * * case 1 . *
the marked edge is adjacent to face @xmath219 on both sides and its removal leaves a connected dessin .
+ suppose that @xmath241 edges are removed in total they necessarily form a path .
the resulting pruned dessin must lie in the set @xmath242 , where @xmath243 .
+ conversely , there are @xmath244 ways to reconstruct a marked pruned dessin in @xmath237 from a pruned dessin in the set @xmath242 , where @xmath243 , by adding a path of @xmath241 edges .
the factor @xmath241 accounts for choosing a marked edge along the path .
the factors @xmath245 and @xmath246 account for the choice of endpoints of the path . * * case 2 . *
the marked edge is adjacent to face @xmath219 on both sides and its removal leaves the disjoint union of two dessins .
+ suppose that @xmath241 edges are removed in total they necessarily form a path .
suppose that the two components have faces @xmath247 and @xmath248 .
the resulting pruned dessins must lie in the sets @xmath249 and @xmath250 , where @xmath243 .
furthermore , we can not obtain a pruned dessin of type @xmath251 in this manner .
+ conversely , there are @xmath244 ways to reconstruct a marked pruned dessin in @xmath237 from two pruned dessins in the sets @xmath249 and @xmath250 , where @xmath243 , by adding a path of @xmath241 edges . the factor @xmath241 accounts for choosing a marked edge along the path .
the factors @xmath245 and @xmath246 account for the choice of endpoints of the path . * * case 3 . *
the marked edge is adjacent to faces @xmath219 and @xmath252 , for @xmath253 .
+ suppose when walking along the marked edge from the white vertex to the black vertex that face @xmath219 lies on the left and face @xmath252 on the right .
suppose that @xmath246 edges are removed in total they necessarily form a path .
the resulting pruned dessin must lie in the set @xmath254 , where @xmath255 .
+ conversely , there are @xmath256 ways to reconstruct a marked pruned dessin in @xmath237 from a pruned dessin in the set @xmath254 , where @xmath255 , by adding a path of @xmath246 edges .
the factor @xmath246 accounts for choosing an edge along the path .
the factor @xmath245 arises from choosing where to glue one of the ends of the path .
note that there is a unique choice to glue in the other end to create a face of perimeter @xmath90 on the left and a face of perimeter @xmath257 on the right .
there is a crucial subtlety that arises in the third case , which we now address .
one can discern the issue by considering the sequence of diagrams below , in which @xmath90 increases from left to right , relative to @xmath257 .
the third case actually contributes to diagrams like the one on the far right , in which face @xmath219 completely surrounds face @xmath252 , or vice versa .
in fact , the edge that we remove can lie anywhere along the dashed path in the schematic diagram .
note that this contributes to the second case , in which the marked edge is adjacent to the face labelled @xmath219 on both sides and its removal leaves the disjoint union of two connected graphs .
however , observe that this surplus contribution is precisely equal to the terms from the second case that involve @xmath236 , so one can compensate simply by excluding such terms .
given that we have already witnessed that @xmath258 , we can restrict to the so - called _ stable _ terms in the second case , which are precisely those that do not involve @xmath235 or @xmath236 . * in the first case for all values of @xmath219 and @xmath243 ; * in the second case for all _ stable _ values of @xmath219 , @xmath243 , @xmath259 , and @xmath260 ; and * in the third case for all values of @xmath219 , @xmath252 , and @xmath255 .
calculation of @xmath261 builds dessins from loops of circumference @xmath245 .
@xmath262 note that this is precisely the recursion of proposition [ th : recprune ] , using the fact that @xmath263 .
taking this definition , one can also apply the recursion in the case @xmath264 .
finally we obtain @xmath265 calculation of @xmath228 for small values of @xmath29 and @xmath37 indicates that it is a quasi - polynomial modulo 2 , although this structure does not follow immediately from the recursion above . in order to prove it , we use the following asymmetric version of the recursion . [ prop : precursion ]
the pruned dessin enumeration satisfies the following recursion for @xmath230 . @xmath266 \\ & + \sum_{i \in s } \bigg [ \sum_{p+q=\mu_1+\mu_i } pq \left.b_{g , n-1}({\boldsymbol{\mu}}_s)\right|_{\mu_i = p } + \mathrm{sign}(\mu_1-\mu_i ) \sum_{p+q=|\mu_1-\mu_i| } pq \left.b_{g , n-1}({\boldsymbol{\mu}}_s)\right|_{\mu_i = p } \bigg]\end{aligned}\ ] ] here , @xmath146 and for @xmath234 , we set @xmath233 .
use the fact that both the symmetric recursion given by proposition [ th : recprune ] and the asymmetric recursion here uniquely determine all @xmath238 from the base cases @xmath267 and @xmath268 .
so it suffices to show that the symmetric version follows from the asymmetric version , and this can be seen by symmetrising .
* for all non - negative integers @xmath22 and @xmath216 , the functions @xmath273 are odd quasi - polynomials modulo 2 in @xmath274 of degree @xmath275 .
* for all non - negative integers @xmath22 , the functions @xmath276 are quasi - polynomials modulo 2 that are odd in @xmath277 and even in @xmath278 of degree @xmath279 .
by the previous example , the base cases @xmath280 and @xmath261 are indeed even quasi - polynomials modulo 2 of degrees 0 and 1 , respectively . now consider @xmath228 satisfying @xmath281 and suppose that the proposition is true for all @xmath228 of lesser complexity .
then the recursion of proposition [ prop : precursion ] expresses @xmath282 as a finite linear combination of terms of the form @xmath283 where @xmath22 and @xmath216 are non - negative integers and @xmath284 . upon dividing by @xmath277 , we find that @xmath205 is a quasi - polynomial modulo 2 that is even in @xmath119 .
furthermore , one can check that it is of the correct degree .
therefore , we have proven the proposition by induction on @xmath285 . | we study topological recursion on the irregular spectral curve @xmath0 , which produces a weighted count of dessins denfant .
this analysis is then applied to topological recursion on the spectral curve @xmath1 , which takes the place of the airy curve @xmath2 to describe asymptotic behaviour of enumerative problems associated to irregular spectral curves . in particular , we calculate all one - point invariants of the spectral curve @xmath1 via a new three - term recursion for the number of dessins denfant with one face . |
in the @xmath3 region of the rare - earth nuclei , three bands are observed to interact with each other at @xmath4 .
the bands are g- , s- and t - bands . the g- and
s - bands are well - known rotational bands of the ground state and rotation - aligned configurations , respectively . on the other hand ,
the t - band means `` tilted - rotating '' band which was proposed by frauendorf @xcite . in @xmath5os ,
the t - band has a @xmath6 band head , while @xmath1os has @xmath7 .
these high-@xmath0 states are assigned to two quasi - particle excitations in the neutron i@xmath8 orbits . because the neutron fermi levels in these nuclei are in the high-@xmath9 orbits in the i@xmath8
, it is expected that high-@xmath0 states are excited favourably and appear near the yrast line .
these three bands , i.e. , g- and s - bands ( low-@xmath0 ) and t - band ( high-@xmath0 ) , seem to interact with each other near @xmath10 to cause a backbend in the yrast line .
an interesting phenomenon here is that the t - bands split into two sequences : one of them consists of even-@xmath11 members while the other of odd-@xmath11 , which is energetically lower than the former .
( see fig.[os182 ] . )
therefore , it seems that so - called `` signature splitting '' and `` signature inversion '' are induced by these multi - band crossings .
these phenomena are observed systematically for even - even nuclei in the @xmath3 region such as @xmath12w and @xmath13os .
to explain these phenomena , i.e. , a coupling between high- and low-@xmath0 states ( in other words , break - down of the @xmath0-selection rule ) , we propose the _
wobbling model_. the term `` wobbling motion '' was originally introduced into nuclear structure physics by bohr and mottelson @xcite , in order to explain excited rotational structures of triaxial nuclei .
the wobbling motion in this context is considered as the small deviation or fluctuation of angular momentum from the typical collective rotation around the @xmath14-axis .
the motion is quantized and presents phonon - like structures above the ground state band .
on the other hand , our wobbling motion is based on the theory by kerman and onishi @xcite .
it is based on the time - dependent variational model ( tdvm ) and attempts to handle general three - dimensional nuclear rotations . in our present model , we describe the wobbling state , @xmath15 , as a superposition of different tilted - axis - cranked hfb states , @xmath16 , where @xmath17 and @xmath18 are defined in the constraints of angular momentum . @xmath19 .
( note that the tilt angle @xmath17 is measured from the @xmath14-axis to see the deviation of the rotation axis from the @xmath14-axis . )
the wobbling states are written , @xmath20 where @xmath21 is the gcm wave function and the variation is made with respect to this function .
the pairing - plus - q@xmath22q force is employed as an residual interaction for the spherical parts ( from the nilsson model ) in the hamiltonian .
quantization is made through the angular momentum projection technique ( @xmath23 ) . as a result
, our wobbling motion can describe not only small but also large amplitude wobbling motion around any types of rotations ( with any kinds of deformations ) .
in other words , this model describes a dynamical picture of nuclear excited modes such as rotations and vibrations .
fig.[wobble ] shows schematically how the wobbling motion occurs .
as a nucleus rotates rapidly , the rotation - alignment occurs , in both of the g- and t - bands . as a result ,
the yrast band takes on the s - band character ( left panel ) and the t - band has some amount of fluctuations in @xmath0-distribution as well ( center panel ) .
as the nucleus rotates more rapidly , the @xmath0 fluctuation around high- and low-@xmath0 components spreads and finally there is an overlap between them ( right panel ) . as a result , the corresponding state has very large fluctuation in @xmath0 .
semi - classically , this state coupling low-@xmath0 states with high-@xmath0 can be pictured as the wobbling motion .
we performed numerical calculations and confirmed that the wobbling model can describe the main features of the multi - bands crossing phenomena .
particularly , our calculations can reproduce the coupling of low- and high-@xmath0 components in the yrast and excited bands through the crossing .
( see fig.[calc ] . )
then , we propose an interpretation for the mechanism of the signature inversion through the wobbling model : first , as a result of the rotation - alignment , the yrast line takes on the s - band character ; then , the t - band approaches to the yrast line and inter - band interactions start between even-@xmath11 members of the t - band and members in the s - band , while odd-@xmath11 members in the t - band are unperturbed because there are no counterparts in the yrast band .
thus , the even-@xmath11 members are pushed up from the original t - band sequence while the odd-@xmath11s keep staying in the sequence .
this mechanism can explain the signature inversion ( and splitting ) phenomena .
at the same time , the yrast line is pushed down through the inter - band interaction , and as a consequence backbending can be enhanced .
the detailed results and discussions are presented in our previous paper @xcite .
in @xmath1os , there is an isomer with @xmath24 , and its rotational band seems to be isolated .
thus , it is expected to have a pure high-@xmath0 ( @xmath25 ) structure ( at least , no low-@xmath0 component ) .
podolyk et al .
@xcite have measured the effect of low-@xmath0 components in the t - band ( @xmath7 ) by the comparison of the transition rates between these two high-@xmath0 bands ( i.e.,@xmath7 and @xmath26 ) .
furthermore , it is hoped to identify the effect of high-@xmath0 components ( @xmath27 ) in the yrast line at high spin ( @xmath28 ) , which influences the transition probability from the @xmath29 isomer ( having a very short half life , @xmath30130 ns ) to the yrast line .
( see fig.[exp1 ] . )
we have investigated a mechanism of signature inversion and splitting in the @xmath3 region by means of _ gcm - after - amp _ on the tilted - axis - cranked hfb states . with this method
, we have qualitatively reproduced the main features of the excited level structure in the high-@xmath0 t - band .
we interpret this result from a point of view of an inter - band interaction between the low-@xmath0 ( s- ) and high-@xmath0 ( t- ) bands .
we have shown that the perturbed states have the character of wobbling motion , that is , a dynamical mode coupling low-@xmath0 par ( principal - axis rotation ) and high-@xmath0 tar ( tilted - axis - rotation ) states . in terms of this wobbling model
, we have discussed an enhancement of backbending in the @xmath3 region , in which the typical rotation - alignment is somewhat suppressed due to the location of the neutron fermi level
. 1 s. frauendorf , nucl . phys . *
a557 * , 469c ( 1993 ) .
a. bohr and b. mottelson , _
`` nuclear structure ( vol.ii)''_,(benjamin , 1975 ) .
a.k.kerman , and n.onishi , nucl.phys.*a361 * , 179 ( 1981 ) .
m. oi et al . , phys . lett . *
b480 * , 53 ( 2000 ) .
podolyk et al . ,
private communications .
p. chowdhury et al . , nucl . phys . *
a485 * , 136 ( 1988 ) . | we analyze a mechanism of coupling of high- and low-@xmath0 bands in terms of a dynamical treatment for nuclear rotations , i.e.,wobbling motion .
the wobbling states are produced through the generator coordinate method after angular momentum projection ( _ gcm - after - amp _ ) , in which the intrinsic states are constructed through fully self - consistent calculations by the 2d - cranked ( or tilted - axis - cranked ) hfb method . in particular , the phenomena of `` signature inversion '' and `` signature splitting '' in the t - band ( tilted rotational band ) are explained in terms of the wobbling model .
our calculations will be compared with new data for in - band e2 transition rates in @xmath1os , which may shed light on the mechanism of the anomalous @xmath2 isomer decay , directly to the yrast band . |
recent bottom - up @xcite and top - down @xcite demonstrations of si channels with diameters of only a few nanometers shape the roadmap towards the ultimate limits of scaling . silicon nanowires ( sinws ) , in particular , can be easily integrated with existing semiconductor technologies . as experimental results
remain difficult to obtain and analyze at this lengthscale , modelling has a crucial role to play in assessing this technology .
owing to computational and methodological advances , great progress has been achieved in atomic - scale transport calculations , first in molecular systems @xcite and later in semiconductor nanowires @xcite .
nevertheless , semiconductor physics is most easily captured within the effective mass theory and many device simulators are built based on this approximation @xcite .
this implies that simple models describing the basic mechanisms are required @xcite .
recently , using an effective mass model we were able to determine the important parameters in scattering from neutral p - dopants in sinws and reproduce results from a first - principles approach within dft @xcite . here , in a similar fashion
, we take a first step in explaining scattering from si atoms at the wire surface promoted to different oxidation states ( typically si@xmath0 and si@xmath1 ) . due to oxidation , this type of oxygen - related defects are unavoidable at the surface of a pure semiconductor wire core even in surface treated sinws and can be considered as a source of roughness on the atomic scale @xcite .
using an effectively one - dimensional model of scattering from a well we fit the conductance of ultrathin ( @xmath2 3 nm in diameter ) sinws as obtained from a combined dft - green s function formalism .
an atomic - scale analysis based on calculations from first - principles will be presented elsewhere @xcite .
this approach allows us to identify the scattering mechanism as intra - subband scattering from a shallow potential well . in section [ sec2 ]
, we summarize the equations of the effective mass model solved using a quantum transport scattering approach . in section [ sec3 ]
, we present a discussion of our results for various typical scenarios of local surface oxidation .
we conclude with section [ sec4 ] . -oriented hydrogenated sinws with diameter w = 1.15 nm as measured from the distance between furthest si atoms in the slice.,title="fig:",width=151][fig1 ]
the general hamiltonian that governs the three - dimensional electron motion reads @xmath3 here , @xmath4 is the effective mass tensor in the coordinate system of the device ( i.e. , x - axis is oriented along the channel ) , @xmath5 is the momentum operator for a fully spatially quantized electron motion and @xmath6 is the single - particle operator for the electrostatic potential that the electron feels when injected in the channel . in what follows we focus on hole transport and take that the effective mass tensor is diagonal within our sinw coordinate system .
this leads to an effective mass hamiltonian [ eq1_c1 ] for the nanostructure reading @xcite @xmath7 we also assume that the effective mass varies only along the channel , i.e. , the x - direction , allowing us to use an ansatz of the form @xmath8 applying this ansatz and using the orthonormality property for the function @xmath9 together with a smooth disorder assumption along transport , i.e. very slow variation , then it is eligible to assume that the function @xmath9 is independent of @xmath10 .
this approximation has also been used in ref . @xcite . under these assumptions we end up with an one - dimensional effective mass equation governing transport within the nanowire @xmath11 where @xmath12 is determined by the solution of @xmath13 the quantum mechanical scattering problem corresponding to equation [ eq4_c1 ]
is then solved to yield the transmission coefficient , @xmath14 , for each channel and , hence , the landauer conductance @xmath15 .
for definitiveness , we consider sinws grown along the [ 110]-direction and fix the diameter to @xmath16 nm .
similar results are obtained for the lowest subbands in wires with larger w. figure 1 shows the cross - section of the wire channel .
the atomistic model for the promotion of si atoms to the states si@xmath0 and si@xmath1 consists of starting from the neutral oxidation state si@xmath17 at the surface of the hydrogenated wires and introducing oxygens as shown in figure 2 .
the resulting si - o - h and si - o - si bonds are the simplest cases of oxygen bridge- and back- bonds . and si@xmath1 oxidation state , respectively.,title="fig:",width=151 ] and si@xmath1 oxidation state , respectively.,title="fig:",width=151][fig2 ] before performing a transport calculation using eq .
[ eq4_c1 ] it is instructive to compare the electronic structure of the pristine hydrogenated wire with the bands of a large enough supercell that includes an oxygen - derived defect .
this is done in figure 3 , where we plot the bands derived from a density functional tight - binding ( dftb ) approach .
the latter is an efficient approximation to the kohn - sham dft scheme .
we first note that the effective mass as determined by the curvature of the bands does not change by the introduction of the oxygen . in general , impurities cause variations of the effective mass @xmath18 along the channel . but here this effect can be disregarded . -oriented si nanowire with cross - section as in figure 1 .
the supercell lattice parameter a is 1.92nm.,title="fig:",width=264][fig3 ] a second observation is that alignment of the bands in a pristine - oxidised - pristine wire arrangement would yield a potential well of the order of a few mev for hole propagation in both subbands . for the effective mass model
, we take this into account as a phenomenological parameter in eq .
[ eq4_c1 ] through the x - dependence of w. the conductance for a potential well extending over a region of length @xmath19 is plotted in figure 4 . also in figure 4
, we show for comparison the result of a combined dftb - green s function approach as well as the ideal hole conductance of a pristine hydrogenated wire .
it is evident that for both cases of an si - o - h bond and an si - o - si back - bond the simple model of considering the region around the oxidized si site as a shallow potential well yields a remarkable quantitative agreement .
this picture is consistent with ballistic transport and the positive oxidation state of si .
we note that within the effective mass model only intrasubband scattering is considered , thereby , implying that this is the dominant scattering mechanism . as an aside we confirmed this by independent analysis of the quantum mechanical s - matrix in the atomic - scale model . [ fig4 ]
in conclusion , effective mass models are useful in analyzing measurements , model building and for practical calculations of charge carrier transport in semiconductor wires . to this end
, we derived a simple picture within this framework to explain results of hole - transport calculations in locally oxidized silicon nanowires grown in the [ 110 ] direction using atomic - scale models in conjuction with density functional theory ( dft ) .
this appoach points to an intra - subband scattering mechanism from a potential well and may further apply in other impurity systems .
p. drouvelis acknowledges financial support from the irish research council for science engineering and technology while at tyndall national institute where part of this work was performed .
the research of g. fagas is supported by science foundation ireland . | we present a simple model to describe the lowest - subbands surface scattering in locally oxidized silicon nanowires grown in the [ 110 ] direction . to this end
, we employ an atomistically scaled effective mass model projected from a three - dimensional effective mass equation and apply a quantum transport formalism to calculate the conductance for typical potential profiles .
comparison of our results with hole - transport calculations using atomistic models in conjuction with density functional theory ( dft ) points to an intra - subband scattering mechanism from a potential well .
drouvelis and fagas : scattering model for locally oxidized si nanowires intraband scattering , silicon nanowires , oxidation |
the generation of a sustained spin polarized current constitutes a key requirement in practical spintronics . in two - dimensional electron gas ( 2deg )
systems with rashba spin - orbit coupling ( soc ) , spin precession about the rashba field and the accompanying zitterbewegung - like motion would usually result in zero net transverse spin current on average.@xcite however , a net spin polarization can be generated in the transverse direction via the spin hall effect .
based on the time - gauge field @xcite or the semi - classical geometric force @xcite descriptions , it can be shown that a vertical ( out - of - plane ) spin polarization can be sustained under adiabatic conditions , albeit under ballistic limit only .
other methods of generating spin current which have been proposed include the application of a non - uniform external magnetic field in a soc medium @xcite and the use of optical excitation.@xcite recently , an all - electrical method has been also investigated by utilizing a time - varying gate voltage.@xcite the idea is based on the fact that the rashba soc can be tuned by changing the applied gate voltage.@xcite however , it is pointed out that , in the rashba system , only the in - plane polarized spin currents are generated.@xcite therefore , it would be interesting if the out - of - plane polarized component of the spin current can be generated by applying this method . in this article , we will present a theoretical study of the spin current generation in 2deg system that combines both time - dependent rashba and dresselhauss socs . in frame work of non - abelian gauge field
, we will point out that time - modulated gate voltage can indeed induce a spin current polarized in the out - of - plane direction of the 2deg .
this spin current can be maintained even under non - ballistic situation , i.e. , in the presence of impurities .
moreover , we will show that such combined soc systems can be utilised as spin pumping sources to generate spin current without applying bias .
we first consider a system with dresselhaus and rashba soc , with the rashba coupling being time - dependent ( shown schematically in fig .
[ fig.1 . ] ) .
the hamiltonian of the system is : @xmath1 where @xmath2is the effective electron mass , @xmath3 is the dresselhaus soc , and @xmath4 is the rashba soc , with the rashba coupling consisting of static and time - varying components , i.e. , @xmath5 . the last term in eq .
is the potential energy which may include the external applied electric field and the impurity field . to the first order in soc couplings , eq
can be rewritten in the form of the yang - mills hamiltonian @xmath6 , where @xmath7 is the gauge field given by @xmath8 and the corresponding curvature is @xmath9.\ ] ] the physical fields extracted from the curvature tensor can be considered as the effective magnetic and electric fields , i.e. , @xmath10 where @xmath11 , @xmath12 is the unit vector in the @xmath13-direction .
these yang - mills fields exert a spin - dependent force on the electron , with the lorentz - like magnetic field acting in the direction transverse to the electron motion as @xmath14 , where @xmath15 is the velocity .
note that the lorentz force term is proportional to the spin current polarized along the @xmath13-direction , which is given by @xmath16 .
therefore , one can arrive at the expression for the quantum mechanical force : @xmath17 from the above , we see the interplay between spin force , spin current and the time modulation of the rashba soc .
the spin current is dependent on either the spin force or the time - modulated rsoc , and would still be present if either one of the two terms vanishes .
one can directly evaluate the spin current from its definition by solving the spin dynamics of the system.@xcite however , this method may be quite involved if the hamiltonian is non - trivially time - dependent .
alternatively , the spin current may be obtained by evaluating the force by some other means , and substituting its expression into eq . .
in the following , we evaluate the force by classical consideration , and subsequently apply the force expression to determine the spin current . , scaledwidth=30.0% ] we consider the instantaneous eigenstate and eigen - energy of the time - dependent soc hamiltonian @xmath18 , given by @xmath19 in which @xmath20 , @xmath21 with @xmath22 . the expectation value of an operator at time @xmath23 is defined as @xmath24 , with @xmath25 is the fermi distribution function .
the velocity operator can be formally derived from the hamiltonian as @xmath26 , and the force is @xmath27 . at some instant @xmath23
, the average force can be decomposed into two parts as @xmath28 in which @xmath29 where , @xmath30 is the velocity of electron belonging to the @xmath31th branch , and the integral is taken over momentum space .
the first force term @xmath32 is related to the evolution of the velocity over time due to the applied electric field and the modulating rashba coupling , while the second component @xmath33 is related to the explicit time - dependence of the eigen - state and will vanish if the hamiltonian is time - independent .
_ constant rashba coupling_ we first consider the static case when @xmath34 . the second term on the right hand side of eq .
vanishes , hence the spin hall current ( shc ) is just dependent on the total force acting on the electrons .
since the eigen - energy and eigen - state are time - independent , its force component @xmath33 vanishes . for simplicity
, we consider an electric field applied along the @xmath35-direction , following which the longitudinal force in eq .
is readilyf found to be @xmath36 where @xmath37 . comparing the above force expression with that obtained by the gauge formalism , i.e. eq .
, we arrive at the formula of the spin current for the two eigenbranches : @xmath38 the total shc at zero temperature is then found by integrating over all occupied states in the momentum space : @xmath39 where , @xmath40 and @xmath41 are the fermi momenta corresponding to the two eigenbranches , and the difference between them is given by : @xmath42 with this , the shc expression simplifies to @xmath43 the above result is consistent with the well - known universal spin hall conductivity of @xmath44 obtained previously by sinova _
et al._,@xcite and by shen,@xcite the latter by considering linear response ( kubo ) transport theory and the berry phase of the system . up to this point , we have explicitly shown the relation between the spin hall current and the spin force in a time - independent system . in ref .
@xcite , the spin force concept is introduced to describe the spin dynamics in a rashba soc system . specifically , the study focused on the _ transverse _ component of the spin force and its effect on the zitterbewegung dynamics of the electrons in the system .
however , the connection between this spin force and the well - known spin hall current was not established . in the above analysis
, we have shown that it is the _ longitudinal _ component of the spin force , whose magnitude is proportional to the driving electric field , which gives rise to the spin hall current with the universal conductivity of @xmath45 .
the longitudinal spin force is related to the lorentz force of the yang - mills ( non - abelian ) effective fields arising from the rashba and dresselhaus soc effects .
our physical picture of the spin hall effect is consistent with previous descriptions of the spin hall effect based on different theoretical models.@xcite
we now consider the effect on the spin current of impurities , which can be modeled as some randomly placed delta potentials : @xmath46 .
as has been pointed out previously,@xcite the spin hall current would be suppressed in an infinitely large disordered system . in the steady state ,
the total force is zero , i.e. , @xmath47 .
additionally , the external electric field which is the driving force for the spin current , is also effectively canceled by the effect of impurities.@xcite thus from eq . , the spin current vanishes if the rashba coupling is constant in time , as expected . however , in the presence of a time - modulated rashba soc strength , the @xmath48 term in eq .
is non - zero , and this leads to a spin current of @xmath49 in which @xmath50 is the spin density , with @xmath51 denoting the expectation values taken over all momentum and spin spaces , and impurity configurations .
we see that the transverse spin current in any one direction depends on the spin polarization along that direction ; in general the latter may be finite and hence the spin current can be sustained . on the other hand , the spin current can also be interpreted as the response to the effective electric field which is given by eq . .
thus , we have @xmath52 , with the `` spin conductivity '' of @xmath53 .
we consider a special case in which the rashba and dresselhaus coupling strengths are approximately the same , but with the former having an additional ( small ) time - dependent variation , i.e. , the rashba coupling is given by @xmath54 with @xmath55 .
then to the first order in @xmath56 , we can express the spin current in eq .
as : @xmath57 thus , the spin current is approximately proportional to the modulation frequency , but is independent of the modulating amplitude @xmath56 .
normally , when an electron with its spin aligned , say , vertically , is passed through the rashba ( dresselhauss ) 2deg system , the spin will rotate about the in - plane rashba field.@xcite this spin precession is associated with zitterbewegung - like motion of the electron.@xcite under the influence of an electric field , the precessional dynamics is modified giving rise to a net spin polarization and hence a spin hall current .
since the spin precession motion originates from the soc interaction , this suggests that we induce a net spin polarization by modulating the soc to control the spin precession without the need to apply any electric field .
thus , we will study the possibility to generate a sustained spin current under the influence of time - modulated soc but in the absence of any dc electric field .
first , we separate the hamiltonian in eq . into two parts : time - independent hamiltonian @xmath58 , and time - dependent hamiltonian @xmath59 .
the evolution of electron spin state can be found by means of the first - order time - dependent perturbation theory , in which @xmath60 is treated as the perturbation .
we assume that the modulation is started at @xmath61 ; meanwhile the applied electric field is switched on adiabatically from the past as @xmath62.@xcite if the initial electron spin state is represented by the spinor @xmath63 , then its subsequent evolution up to time @xmath23 under the full hamiltonian is given by @xmath64 in which @xmath65 and @xmath66 are eigen - states and eigen - energies of the unperturbed hamiltonian . to the first order in both the dynamic rashba component @xmath56 and the electric field , the time - dependent coefficients @xmath67 satisfy the following differential equations :
@xmath68 the above can be readily solved once the initial conditions are specified , and subsequently the spin polarization @xmath69 can be evaluated .
we first consider the system of electrons , which are initially in an admixture of the two eigen - states .
the initial spin polarization vector is @xmath70 , corresponding to the two eigen - branches . to the first order in @xmath71 , the spin polarization vector at a subsequent time @xmath23
is easily found to be @xmath72,\ ] ] where @xmath73 with @xmath74 being the response functions and @xmath75 .
the two terms @xmath76 and @xmath77 are the contributions to the spin polarization due to the applied electric field and time modulation of the rashba coupling , respectively .
to the first order in both @xmath78 and @xmath56 , these two factors can be considered as independent driving forces which induce the spin polarization .
we focus on @xmath79 , the spin polarization arising from the time - modulation in the rashba soc .
it is interesting to point out that in the absence of dresselhauss soc ( i.e. , @xmath80 ) , the phase factor is simply @xmath81 , and therefore , from eq .
, the spin polarization @xmath79 becomes zero .
hence , the time - modulation of the rashba soc will only induce a spin polarization if both types of socs , i.e. , rashba and dresselhauss , are present in the system .
in addition , we find that @xmath79 is an odd function of the momentum vector @xmath82 , so that summing over all occupied states in momentum space would yield a zero net @xmath13-spin polarization . however , we will show below that the spin current may still be finite . .
the inset shows the dependence of the spin current amplitude on the modulation frequency @xmath83 obtained numerically ( solid ) . for comparison
, we plot the analytical results for the maximum amplitude given by eq .
( horizontal dashed ) and the off - resonant amplitude given by eq .
( bold dashed ) .
the maximum amplitude occurs at @xmath84.,scaledwidth=45.0% ] as discussed in previous papers,@xcite the modification to the electron spin due to an applied electric field can give rise to a spin hall current ( even if the net spin polarization is zero ) .
this can be readily seen by evaluating the expression @xmath85 , which would recover the result of eq . .
we apply the same expression to evaluate the spin current arising from the spin polarization @xmath86 arising from the time - modulation of the rashba soc , i.e. , @xmath87 , and average the result over the momentum space to yield the net spin current .
[ fig2 ] illustrates the ac spin current generated in the system when the rashba coupling is modulated by a sinusoidal gate voltage , i.e. @xmath88 , assuming typical parameter values : @xmath89 , @xmath90 , @xmath91 , @xmath92 .
note that when the gate voltage is changed , the fermi energy and effective mass will also be modified as a consequence @xcite .
however , for simplicity , we assume a small change in the gate voltage , so that the fermi energy and effective mass can be treated as constant . we see that the amplitude of the spin current is maximum around the resonant region @xmath84 , where @xmath93 the larmor frequency of electron near the fermi surface ( see inset of fig . [ fig2 ] ) . for the special case of @xmath94 ,
this maximum amplitude can be estimated as @xmath95 while the off - resonant amplitude is given by @xmath96 in the above , we have assumed that @xmath97 , where @xmath98 being the energy scale associated with the spin - orbit couplings , @xmath99 the fermi energy .
the analytical results of eqs . and are in approximate agreement with the numerical results ( see inset of fig . [ fig2 ] ) .
in this paper , we have applied the spin force picture to a general time - dependent soc system . by considering the lorentz force arising from the non - abelian gauge field of the soc effects
we establish the relation between the spin current and spin force in the system .
previously the spin force in a rashba system was invoked to explain the zitterbewegung motion of electrons , but no link was made to the spin hall effect .
we applied our gauge field method to show that the _ longitudinal _ component of the spin force induces the spin hall effect .
this not only provides a physical picture of the underlying origin of the spin hall effect , but also recovers the universal spin hall conductivity in a constant ( time invariant ) rashba system .
we also showed that the spin current vanishes under strong impurity scattering , a result consistent with previous findings .
however , under time modulation of the rashba soc a sustained spin current is obtained , the amplitude of which is proportional to the in - plane spin density and the modulation frequency .
finally , we evaluated the ac spin current generated by a rashba soc with a sinusoidal time variation in the absence of any dc electric field .
for the special case of equal rashba and dresselhaus coupling , we derived the analytical expression for the magnitude of the spin current , which approximately agrees with the numerical results .
+ 21ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | the generation of spin current and spin polarization in a 2deg structure is studied in the presence of dresselhaus and rashba spin - orbit couplings ( soc ) , the strength of the latter being modulated in time by an ac gate voltage . by means of the non - abelian gauge field approach , we established the relation between the lorentz spin force and the spin current in the soc system , and showed that the longitudinal component of the spin force induces a transverse spin current . for a constant ( time - invariant ) rashba system ,
we recover the universal spin hall conductivity of @xmath0 , derived previously via the berry phase and semiclassical methods . in the case of a time - dependent soc system ,
the spin current is sustained even under strong impurity scattering .
we evaluated the ac spin current generated by a time - modulated rashba soc in the absence of any dc electric field .
the magnitude of the spin current reaches a maximum when the modulation frequency matches the larmor frequency of the electrons . |
forty years after the discovery of pulsars it is fair to say that an immense amount of progress has been made in the quest to understand how neutron stars ( ns ) form and evolve in different stellar systems .
our understanding of these processes has benefited most importantly by the study of binary pulsars or x - ray binaries harboring ns with either stellar or degenerate companions ( white dwarfs or other ns ) .
progress has been made both by studying specific observed systems as well as whole populations in the galactic field or in globular clusters.these analyses and results have always coupled strongly to efforts in the areas of core - collapse and supernova ( sn ) explosions of massive stars , thermonuclear explosions of accreting white dwarfs , and nuclear evolution of massive stars . through this diverse body of work (
for which it is impossible to include here all important references , but relevant overviews may be found in @xcite ) a generally accepted understanding seemed to have emerged anchored on the following crucial elements most relevant to the discussion presented in this paper : ( i ) ns form at the end of stellar nuclear evolution as remnants of sn explosions of massive stars of @xmath2m@xmath3 up to @xmath4m@xmath3 ; ( ii ) ns in close binaries typically have helium - rich progenitors at the time of core collapse ; these progenitors are the helium - rich cores ( typically more massive than 2m@xmath3 ) of original massive binary companions which have lost their hydrogen - rich envelopes through binary interactions involving ( stable or unstable ) mass transfer between binary components and/or mass loss from the binary ; ( iii ) at the time of their formation ns acquire recoil kicks due to asymmetries that develop during the core - collapse process ; typical kick magnitudes inferred from studies of pulsar population kinematics and the formation of binaries with ns are in the range of 100300kms@xmath0 ; ( iv ) ns are born with birth spin periods of @xmath5ms and spin down over a time span of @xmath2myr ; if in binaries they can spin up and reach spin periods of 10ms or shorter later in their life through accretion of material from their companion and accompanied reduction in magnetic - field strength . in recent years a growing number of ns studies have challenged different aspects of the `` standard picture '' as described above .
although each one of these in isolation may not have been powerful enough to cause a paradigm shift , the current set of inter - linked arguments presented in the literature in the past @xmath6 years has certainly attracted attention and has opened the door to potential changes in our current understanding of how ns form .
in fact , as we will discuss in what follows , these recent developments do not actually erase or substitute our earlier understanding ; instead they bring forward new possibilities that had not received much attention or had not even been raised previously . with a broad - brush look ,
two main physical processes lie at the core of a lot of the issues raised in connection to ns formation and evolution : ( i ) _ electron - capture supernovae _ ( ecs ) , and ( ii ) _ hypercritical accretion onto ns _ in common envelope phases ( dynamically unstable mass transfer leading to common - envelope formation and ejection through ns inspiral towards the core of a massive companion ) .
the possibility of ecs was originally put forward as a potential channel for ns formation that does not involve the `` usual '' collapse of the iron core of a massive star at the end of nuclear evolution @xcite .
instead massive onemg white dwarfs can reach high enough densities for electrons to capture onto @xmath7 mg and @xmath8ne ( before ne ignition can occur ( see @xcite ) , causing the white dwarf to lose support against gravity ; ultimately this process is expected to lead to core collapse , a sn , and a ns remnant .
such a collapse could occur right at the progenitor mass boundary between the formation of massive , onemg white dwarfs and the collapse of iron cores into ns : hydrogen - rich , original progenitors of @xmath9m@xmath3 .
this mass range is , however , highly uncertain .
more recently podsiadlowski et al .
@xcite proposed that ecs events are relevant for stars of initial mass in the range @xmath10m@xmath3 in binary systems where they can avoid the second dredge - up phase on the asymptotic giant branch ( agb ) ( due to loss of the hydrogen - rich envelope ) and form bigger helium cores ; instead single stars of the same mass would form onemg white dwarfs .
they also suggested that ecs are relatively fast explosions compared to delayed neutrino - driven explosions of iron cores and could naturally lead to natal kicks of low magnitudes .
kitaura et al . modeled ecs explosions of a @xmath111.3m@xmath3 onemg white dwarf ( formed from a 2.2m@xmath3 helium core , which in turn formed from an 8.8m@xmath3 hydrogen - rich progenitor )
; they find that the explosions are rather faint ( low @xmath12ni production ) , produce low - mass ns with baryonic mass of @xmath13m@xmath3 , and are not prompt but still delayed and powered by neutrino heating .
t. janka during his talk at this conference @xcite presented more recent results and commented that the shock is not delayed long enough for kicks in excess of @xmath14kms@xmath0 to be produced .
last , poelarends et al .
@xcite investigated the late evolution of solar - metallicity _ single _ stars on the agb and concluded that stars in the mass range @xmath15m@xmath3 could experience ecs , implying an ecs fraction of 420% among all core - collapse sn , if one accounts for some of the model uncertainties ( agb stellar wind strengths and dredge - up efficiency , semi - convection , convective overshooting , rotational mixing ; see ) . in a similar study ,
siess concluded that the ecs mass range is indeed very narrow for single stars ( < 11.5m@xmath3 ) ; at solar metallicity he finds a mass range of 9.510.5m@xmath3 which is shifted downwards by 2m@xmath3 if core overshooting is included and could also vanish altogether depending on the metallicity dependence of agb winds . in view of all this recent work , our current understanding of ecs in connection to ns progenitors and kicks
can be summarized as follows : they can occur in nature and may contribute a small but non - negligible fraction of ns forming with natal kicks @xmath16kms@xmath0 .
their progenitor mass range in single and binary stars is _ highly uncertain _ , but _ may be as wide as _
@xmath17m@xmath3 , and even reach down to 5m@xmath3 for low - metallicity stars , potentially including helium stars less massive than 2m@xmath3 @xcite . a significant part of our understanding of how ns form and evolve comes from analyses of binary pulsars with two neutron stars ( ns - ns ) . in the `` standard '' ns - ns formation channel @xcite the first - born ns experiences a common - envelope phase when its hydrogen - rich companion fills its roche lobe .
as a result the orbit shrinks to the small orbits observed , the helium - rich progenitor of the second ns is revealed , and the first ns has a chance of getting mildly recycled due to accretion in the common envelope and/or mass transfer from the helium star companion .
brown @xcite challenged the survival of the first ns through the common - envelope phase and argued that the ns will collapse instead into a black hole due to hyper - critical accretion ( exceeding the photon eddington limit by many orders of magnitude ) .
he suggested an alternative formation scenario that envisions evolution through a double - common - envelope phase revealing two helium - rich ns progenitors in a tight orbit , followed by two consequent sn explosions .
this channel requires that the two hydrogen - rich massive progenitors have almost equal masses ( within 4% ) and could also potentially explain the almost equal ns masses in ns - ns .
we note however , that hydrodynamical studies of hypercritical accretion ( e.g. , @xcite ) and population studies of ns - ns formation that have accounted for hypercritical accretion ( e.g. , @xcite ) have concluded that if the maximum ns mass is @xmath18m@xmath3 or higher , then a significant fraction of ns can still survive the common - envelope phase and avoid collapse into a black hole .
for almost two decades now , it has been widely accepted that there is a minimum mass a helium - rich star must have for the formation of an iron core and ns formation at core collapse .
this minimum value is somewhat dependent on the detailed treatment of convection and late phases of nuclear burning in stellar models , but has been known to be in excess of 2m@xmath3 ( e.g. , in the range 2.12.3m@xmath3 , see habets ) .
one pathway for helium stars to be less massive than the above limit at the time of core collapse is provided by mass transfer onto its binary companion , which however does not abort iron - core formation .
recently , several independent groups have analysed the late stages of evolution of ns - ns progenitors and all reached the same conclusion : at least two of the known close ns - ns binaries ( psr j0737 - 3039 and psr b1534 + 12 ) were experiencing roche - lobe overflow mass transfer from the helium - rich progenitor of the second ns onto the first ns at the time of the second sn explosion @xcite .
although the mass accretion onto the ns may be rather small ( and yet still adequate for mild recycling ) , the mass of the helium star can be reduced significantly ( @xmath19m@xmath3 from its original mass by the time of its core collapse @xcite .
alternatively ( or even concurrently ? ) ns formed in ecs are also expected to originate from low - mass helium cores .
this is primarily because of the connection to an onemg core that just barely exceeds the chandrasekhar mass when electron capture and loss of support ensue .
although the ecs progenitor mass range is highly uncertain ( see previous section ) , it tends to lie below 1112m@xmath3 and potentially includes helium - rich progenitors less massive than 2m@xmath3 ( see ecs discussion in @xcite .
such low - mass progenitors would also naturally produce low - mass ns with baryonic masses very close to the chandrasekhar limit and gravitational masses of @xmath20m@xmath3 .
the discovery of the double pulsar @xcite and the mass measurement of pulsar b ( the second - born ns ) at 1.25m@xmath3 led podsiadlowski et al .
@xcite to suggest an ecs origin .
a number of independent groups analyzed the evolutionary history of this system , given the observed characteristics , and derived constraints on the immediate ( right before the sn ) progenitor of pulsar b. piran & shaviv @xcite were the first to allow as part of the priors in the statistical analysis helium star progenitors less massive than 2m@xmath3 ; they found that progenitor masses in the range 1.251.75m@xmath3 were favored by their analysis at the 30% confidence level , while masses up to 2m@xmath3 could just be reached at 97% confidence level ( see their figure 2 ) .
willems et al .
@xcite presented a detailed analysis that explored the influence of all uncertainties associated with the systems ( most notably its age and radial velocity with respect to the sun ) ; when allowing for low - mass progenitors they too concluded that the most likely helium star masses lie in the range 1.351.95m@xmath3 , while the 90% confidence intervals extended up to 2.5m@xmath3 ( see their figure 6 and table iv ) .
more recently , stairs et al .
@xcite addressed some of these questions using updated proper motion measurements , and they too concluded that the immediate progenitor of pulsar b was likely less massive than 2m@xmath3 with most likely values at @xmath21m@xmath3 ( see their figure 1 ) .
these conclusions are very different from those of a very similar analysis of the ns - ns system psr b1534 + 12 for which stairs et al .
@xcite report a most likely value of 2.45m@xmath3 and a 95% confidence interval 1.63.9m@xmath3 .
based on these results , it appears most probable that the immediate progenitor of pulsar b was less massive than 2m@xmath3 ( most likely @xmath21m@xmath3 ) , even though this is clearly not universal among close ns - ns binaries .
the next question becomes : _ what does this low mass tell us about ns formation ?
_ we would argue that the answer is still unclear .
it is indeed very tempting to associate this mass with an ecs event and declare the case closed , and this could very well be the answer .
however , we should not eliminate from our thinking that psr j0737 - 3039 has also been found by several independent studies to be in a phase of roche - lobe at the time of pulsar b formation .
consequently , we can not exclude that the low progenitor mass is due to mass transfer ( albeit with small accretion onto pulsar a ) initiated before the sn explosion .
_ we conclude that both of these possibilities remain as potentially relevant to the double pulsar formation .
_ at present no obvious secure observational test to distinguish between these two possibilities appears to exist .
however we suggest that two different types of theoretical studies could shed some light and allow us to make further progress : _ ( i ) _ detailed mass transfer sequences that can explain not just the low progenitor mass but also the orbital characteristics right at the explosion time as well as the measured ns masses ; _ ( ii ) _ population studies simulating evolution forward in time focusing on the question of the relative frequency of ecs vs. mass transfer effects depending on assumptions about the highly uncertain ecs progenitor mass range .
for over more than three decades now , estimates of kick velocity magnitudes imparted to ns at birth have been obtained through the analysis of the kinematics of single pulsar samples of varying sizes and reliability of proper motion measurements ( starting with gunn & ostriker @xcite ) .
arzoumanian et al .
@xcite analyzed a sample of 79 single pulsar proper motions and concluded that the ns kick magnitude distribution is best described by a two - component maxwellian velocity distribution with characteristic velocities ( three - dimensional velocity dispersions ) of 90kms@xmath0 and 500kms@xmath0 with almost equal relative weight .
more recently , hobbs et al .
@xcite adopted an updated electron density model and used a sample of 73 proper motion measurements for pulsars with short characteristic ages ( @xmath22myr ) ; they concluded that the sample is best described by a single maxwellian with a one - dimensional velocity dispersion of 265kms@xmath0 , shifting the mean kick velocity to values @xmath23kms@xmath0 .
very similar results were also obtained by zou et al .
it is recognized that these results could be somewhat contaminated by pulsars originating from disrupted binaries and the contribution of orbital velocities , and by a potential bias against low velocities due to the associated small and hence not easily measurable proper motions .
nevertheless , it is clear that overall these analyses point to significant kick magnitudes of a few hundreds of kms@xmath0 .
additional constraints on ns kick magnitudes have been obtained by the analysis of the evolutionary history of specific binary systems with ns known in the galaxy , most notably binary pulsars and x - ray binaries .
although the majority of such studies in the past reached conclusions in support of significant ns kicks , more recently a number of arguments and analyses point towards a need for at least some ns receiving low - magnitude kicks . in what follows
we summarize such evidence and place it in the context of our developing current understanding . * neutron stars in globular clusters . *
perhaps one of the earliest arguments in favor of modest or negligible ns birth kicks has been anchored on the need for ns retention in globular clusters .
the observed low - mass x - ray binaries ( persistent and transient ) and binary pulsars in galactic globular clusters have led to estimates for the total number of ns currently present in clusters in the range @xmath24 @xcite , implying retention fractions of @xmath25% .
given that current estimates of cluster escape velocities lie in the range @xmath26kms@xmath0 , it is evident that clusters could indeed pose a requirement for low kick magnitudes .
pfahl et al .
@xcite concluded that retention fractions of 110% can be achieved if kick magnitude distributions are maxwellians with one - dimensional dispersions smaller than @xmath27kms@xmath0 .
they also found that the vast majority of ns are retained in massive binaries in the clusters .
however the increased characteristic kick magnitudes derived by hobbs et al .
@xcite re - emphasized the issue of ns retention in clusters .
most recently , ivanova et al . @xcite and kuranov &
postnov @xcite concluded that the required retention fractions can still be obtained , if a significant fraction of ns form through ecs events , and if these ns acquire either no kicks or kicks reduced by a factor of 10 relative to kicks imparted to ns formed through iron core collapse .
* high mass x - ray binaries . * pfahl et al .
@xcite highlighted the existence of a subclass of high - mass x - ray binaries characterized by relatively low orbital eccentricities ( @xmath28 ) and wide orbits ( periods in excess of 30d ) .
they argued that their formation appears highly unlikely , unless some ns acquire only small kicks of @xmath29kms@xmath0 ( see their figures 3 and 5 ) .
subsequently podsiadlowski et al .
@xcite associated these low kicks with ns formation in ecs events and accounted for the subclass of low - eccentricity , wide high - mass x - ray binaries assuming a rather wide ecs mass range of @xmath30m@xmath3 . *
double neutron stars .
* the pattern of observed systems requiring both low and high ns kick magnitudes appears to be even more perplexed for the case of ns - ns binaries . to
our knowledge van den heuvel @xcite was the first to point to ns - ns and argue in favor of low ns kicks based on the occurrence of small eccentricities ( @xmath31 ) for 5 of the 7 systems known in the galactic field .
following pfahl et al .
@xcite and podsiadlowski et al .
@xcite , he suggested that _ some _ second - born ns must have received very small kicks ( @xmath32kms@xmath0 ) because of ecs formation . on the other hand , chaurasia & bailes @xcite and ihm et al .
@xcite showed that orbital evolution through gravitational radiation alters the eccentricity distribution : many of the high - eccentricity systems are drained from the population because they evolve fast and either merge and/or shift to lower eccentricities through circularization .
mclaughlin et al .
@xcite highlighted a correlation between the ns - ns eccentricities and the spin periods of the first - born ns in the systems , while faulkner et al .
@xcite argued that such a correlation would naturally arise if all second - born ns were formed symmetrically and no kicks were imparted to them . following this initial suggestion , dewi et al .
performed population studies adopting a simple spin - up model for the first - born ns and concluded that such a correlation appears only if _ all _ second - born ns acquire kicks from a maxwellian with a velocity dispersion of @xmath33kms@xmath0 . willems et al .
@xcite explored the potential role of polar kicks in forming this correlation and similarly conclude that low kicks for the second - born ns seem to be needed in order to explain this correlation .
not surprisingly , the particular case of the _ double pulsar _
( psr j0737 - 3039 ) has attracted a lot of attention in this discussion of ns kick magnitudes .
the binary pulsar has a low eccentricity ( 0.09 ) and a low second - born ns mass ( 1.25m@xmath3 ) , and most recently a low proper motion ( 10kms@xmath0 ) was reported @xcite .
hence , it appears as the prime candidate for a system where the second - born ns was formed through an ecs event and acquired a small kick .
willems et al .
@xcite assumed the `` standard '' minimum progenitor mass of 2m@xmath3 and adopted a flat prior distribution for the unknown radial velocity and found most probably kick magnitudes in the range 100300kms@xmath0 .
podsiadlowski et al .
@xcite used the low mass of pulsar b as the `` tell - tale sign '' for ecs and hence a small kick imparted to it , anticipating a small proper motion measurement .
piran & shaviv @xcite used the measured orbital characteristics and an earlier estimate of the proper motion to derive a probability distribution for the birth kick to pulsar b. in their calculation they allowed for the possibility of very low - mass progenitors and also made a number of restrictive assumptions about the system s age and its galactic motion ( the latter allowed them to by - pass the question of the unknown radial velocity ) .
they primarily highlighted their result derived at 30% confidence level favoring kick magnitudes either @xmath34kms@xmath0 or in the range @xmath35kms@xmath0 . at a safer confidence level of 97% ,
their results set an _ upper limit of _ 125kms@xmath0 ( see their figure 2c ) .
following these studies , willems et al .
@xcite and stairs et al .
@xcite presented the two most comprehensive analyses of this system to date ( the latter of the two using the most recent and reliable measurement from pulsar timing of the proper motion by kramer et al .
both reached rather similar conclusions for the _ most likely _ kick magnitude imparted to pulsar b : 50170kms@xmath0 ( willems et al .
@xcite ) and 4080kms@xmath0 ( stairs et al .
as expected these magnitudes reach much higher values if wider confidence intervals are considered
. willems et al .
@xcite found up to 300kms@xmath0 at 90% confidence level , assuming a gaussian distribution for the unknown radial velocity with dispersion of 130kms@xmath0 ( motivated by population synthesis calculations ) .
stairs et al .
@xcite found kicks in the range of 451000kms@xmath0 for the same radial velocity assumptions , and of 20140kms@xmath0 , assuming velocity vectors are isotropically distributed around the transverse velocity .
the issue of the radial velocities has been identified by these two studies in particular as quite important for the best possible determination of the kick magnitude . to avoid a long digression
we discuss this issue in more detail in the appendix where we show the simple assumption of isotropicity for the system velocity to not be justified .
it is interesting to compare the willems et al .
@xcite and stairs et al .
@xcite results to those obtained by piran & shaviv @xcite .
although each study has used somewhat different observational constraints as they became available , one can say that at zeroth order all three studies agree that , once immediate progenitor masses lower than 2m@xmath3 are allowed as part of the priors in the statistical analysis ( motivated by ecs events or collapse of rather small iron cores ) , then kick magnitudes in the range @xmath36kms@xmath0 are clearly favored at low confidence intervals ( @xmath37% ) . for these magnitudes , willems et al .
@xcite and stairs et al .
@xcite predicted a spin misalignment angle for pulsar a smaller than 20@xmath38 in very good agreement with the upper limit of 1015@xmath38 at similar confidence intervals just recently reported by ferdman et al .
@xcite .
some subtleties however do arise at a more detailed level of examination .
although these do not alter the basic conclusions in any way , it is somewhat important to briefly identify some of them , to avoid propagation to misunderstandings or misconceptions .
for example : ( i ) a ns progenitor of mass lower than 2m@xmath3 does not necessarily imply that kicks must also be very small ( @xmath32kms@xmath0 ) ; ( ii ) the current position of the system close to the galactic plane does not necessarily imply that the system and kick velocity must be small ( see the point made by figure 9b in @xcite ) ; ( iii ) the derived probability distributions are inherently multi - dimensional ( 5-d : progenitor mass , orbital separation , kick magnitude , and two kick direction angles ) and statements about preferred kick magnitudes are _ quite sensitive _ to projection effects in different dimensions because of the complex structure of the multi - dimensional surface ( as shown in table iii in @xcite , the statement of `` _ most likely _ ''
kick can vary by a factor of @xmath39 from 10 to 75kms@xmath0 depending on the projection dimensionality ) ; ( iv ) the system age is one of the poorest constrained priors in the determination of the probability distributions ( as seen from table iii in willems et al .
@xcite , the `` _ most likely _ '' kick can vary by a factor of @xmath40 from 5 to 170kms@xmath0 depending on the assumed system age ) ;
( v ) the most likely birth location of the system can be several kpc away from its current position ( see figure 8 in @xcite ) , so that following the three dimensional motion of the binary is important for the dynamics .
the results obtained for the _ other ns - ns binaries _ are particularly interesting in comparison to the case of the double pulsar . just like with the second - born ns progenitor masses , psr b1534 + 12 paints a rather different picture for ns kicks .
independent studies @xcite have concluded that a minimum kick of @xmath41kms@xmath0 is needed to explain its properties along with its significant spin - orbit misalignment .
it is further interesting to point out that this system has a rather small eccentricity ( < 0.3 ) _ and _ exhibits rather well the correlation between pulsar spin period and orbital eccentricity .
both of these attributes have been associated with second - born ns getting no or minimal ( @xmath32kms@xmath0 ) kicks @xcite .
psr 1534 + 12 therefore clearly presents a contradictory example .
psr b1913 + 16 also clearly requires a significant kick magnitude of at least 260kms@xmath0 ( * ? ? ?
* ; * ? ? ?
* and references therein ) .
it has been long recognized that first - born ns in ns - ns must have been spun - up ( recycled ) at some stage prior to the second ns formation event . however , the measured spin periods are all rather long in the range 10100ms , in excess of the `` classical '' millisecond pulsars , and these ns are thought to be only _ mildly _ recycled .
nevertheless , even mild recycling _ requires _ some level of accretion onto the ns at some stage between the two sn events .
this requirement is sometimes ignored in discussions of ns - ns formation , but can provide us with some insights and constraints . in the `` standard '' ns - ns formation channel , the first - born ns has in principle two opportunities for recycling : either during the common - envelope inspiral involving the hydrogen - rich progenitor of the second ns or during a mass - transfer phase driven by roche - lobe overflow from the helium core exposed after the common - envelope phase . as mentioned already , the progenitors of at least two of the known ns - ns binaries were indeed in such a mass transfer phase from the helium - rich companion to the first - born ns at the time of the second sn explosion . therefore the need for ns spin - up during the common envelope is alleviated for these systems .
it is also interesting that all explanations of the correlation between spin period and eccentricity rely on a ( linear ) scaling between the accretion amount responsible for the spin - up and the mass of the second - born ns progenitor right before the explosion .
such a scaling arises naturally if spin - up happens during a phase of roche - lobe overflow mass transfer from a helium - rich companion and is supported by mass - transfer calculations of ns helium - star binaries @xcite ( see also fig . 2 in @xcite ) .
willems et al .
@xcite furthermore noted that the presence of high - eccentricity systems on the spin period - eccentricity relation could support the existence of a similar scaling for ns accretion during common envelope phases .
last , it is important to realize that in the case of the double common - envelope channel put forward by brown @xcite , the opportunities for ns spin - up are limited and potentially negligible . given how similar the masses of the two progenitors are , the time between the two sn explosions is very short ( @xmath42yr ) and it is difficult to imagine that in this short time both the orbital and stellar characteristics would happen to be appropriate for establishing roche - lobe overflow and spinning up the first - born ns . in their population simulations belczynski et al .
@xcite did indeed identify this ns - ns formation channel , but , for a maximum ns mass of 3m@xmath3 , it only contributed 8% of the total ns - ns population .
most importantly though belczynski et al .
did not find any opportunity for ns recycling .
they also considered a model with a maximum ns mass of 1.5m@xmath3 ( model d2 ) and naturally found a significant contribution to ns - ns formation from the double common - envelope channel , but the issue of recycling remained .
more recently dewi et al .
@xcite modeled ns - ns formation through this channel and obtained rates very similar to those by belczynski et al .
@xcite , but did not comment on the difficulty with recycling the first - born ns in the systems . at present
we would say that the question of when and under what circumstances first - born ns in ns - ns binaries get recycled is rather unclear .
if the double common - envelope channel is shown to be dominant in some way , then spin - up during mass transfer from the helium - rich companion would be necessary .
it is possible that accretion during common envelope inspiral or wind accretion alone without roche - lobe overflow would be sufficient to recycle the first born ns , although so far neither of these have been examined quantitatively . also , if the correlation between spin period and orbital eccentricity persists , the amount of mass accreted by the ns must scale with the donor mass regardless of the mode of mass transfer .
the picture of ns formation from 10 years ago has clearly changed and paradigms have shifted - ns formation now appears more diverse than we thought both in terms of progenitor masses and birth kicks , and thus also in terms of core - collapse physics , asymmetry developments , and maybe even kick mechanisms . in the last 3 years , the number of papers favoring small or even zero ns kicks has mushroomed .
a lot of this recent work is driven by the increasing observational information regarding binary pulsars as well as new detections .
we summarize the current picture as follows : ( 1 ) ns form from the iron core collapse of helium - rich progenitors with masses above @xmath43 , though lower masses may be possible if significant mass loss through roche lobe overflow occurs shortly before ns formation .
alternatively , ns form from lower - mass helium stars through ecs of massive onemg cores .
the ecs progenitor mass range is currently highly uncertain , but the channel may be particularly accessible for ns formation in binaries where the ns progenitors lose their hydrogen - rich envelope before the second dredge - up on the agb .
( 2 ) typical ns kicks of 200 - 300kms@xmath0 are still favored by analyses of pulsar kinematics and are required for many of the known binary ns systems .
however sub - populations of or some specific ns binaries clearly favor smaller kick magnitudes ( lower than @xmath14kms@xmath0 , although probably not as low as @xmath2kms@xmath0 ( t. janka 2007 , private communication ) ; these may be naturally provided during ecs events because of the short duration of the explosion limiting the development of asymmetries and build - up of larger kicks .
( 3 ) ns recycling back to @xmath44ms spin periods occurs through long - term , steady accretion through roche - lobe overflow . for some of the known double ns systems
such mass transfer from the helium - rich progenitor of the second ns is actually required ; however , depending on the dominant formation channel(s ) , mild recycling to @xmath5ms may be possible ( or even required ) during the common envelope phase or through wind accretion from the helium - rich ns companion .
the diversity is apparent and it is providing us with new clues for the formation and origin of ns ( see also @xcite for a related discussion ) . in the near future
we look forward to more analyses that can hopefully constrain some of the current uncertainties : what is the progenitor mass range for ns formation through both iron - core collapse and ecs , especially in binaries over a range of metallicities ? is there a minimum and/or maximum ns kick associated with ecs events ? do ns survive hypercritical accretion in common envelopes ? which progenitor properties determine the birth ns masses ? between ns - ns systemic velocities and our line - of - sight , for population synthesis models adopting maxwellian kick velocity distributions with dispersions of 50kms@xmath0 ( left ) and 260kms@xmath0 ( right ) .
solid lines represent the full population of ns - ns binaries , dashed lines the sub - population of systems within 50pc from the galactic plane ( cut0 ) , and dot - dashed lines the sub - population of systems within 50pc from the galactic plane with a proper motion of less than 10kms@xmath0 ( cut1 ) . for comparison ,
the isotropic @xmath45-distribution @xmath46 is also shown ( dotted line ) .
all pdfs are constructed using the ns - ns populations synthesized by willems et al .
( 2006).,title="fig : " ] between ns - ns systemic velocities and our line - of - sight , for population synthesis models adopting maxwellian kick velocity distributions with dispersions of 50kms@xmath0 ( left ) and 260kms@xmath0 ( right ) .
solid lines represent the full population of ns - ns binaries , dashed lines the sub - population of systems within 50pc from the galactic plane ( cut0 ) , and dot - dashed lines the sub - population of systems within 50pc from the galactic plane with a proper motion of less than 10kms@xmath0 ( cut1 ) . for comparison , the isotropic @xmath45-distribution @xmath46 is also shown ( dotted line ) .
all pdfs are constructed using the ns - ns populations synthesized by willems et al .
( 2006).,title="fig : " ] , but with different constraints imposed on the population .
solid lines represent the full population of ns - ns binaries ( as in figure [ f1 ] ) , and dashed lines the sub - population of systems located within @xmath47 from psr j0737 - 3039 s position in galactic longitude and latitude ( @xmath48 , @xmath49 ; cut3 ) .
the dotted line represents the isotropic @xmath45-distribution @xmath46 .
, title="fig : " ] , but with different constraints imposed on the population .
solid lines represent the full population of ns - ns binaries ( as in figure [ f1 ] ) , and dashed lines the sub - population of systems located within @xmath47 from psr j0737 - 3039 s position in galactic longitude and latitude ( @xmath48 , @xmath49 ; cut3 ) .
the dotted line represents the isotropic @xmath45-distribution @xmath46 .
, title="fig : " ] we are grateful to ingrid stairs for useful discussions and comments on this manuscript , and to jeff kaplan and chris belczynski for allowing us use of their codes from their earlier work in collaboration with vk and bw .
this work is partially supported by a packard foundation fellowship in science and engineering and a nsf career grant ( ast-0449558 ) to vk .
all of the analyses of evolutionary histories and kinematics of pulsar systems suffer from the lack of any constraints on their radial velocities @xcite . in the case of psr j0737 - 3039 ,
probably the most exhaustively studied binary pulsar , the question of the appropriate radial velocity distribution that should be adopted is one of the most important remaining uncertainties in the models ( along with the age ) .
piran & shaviv @xcite did not assume anything , but therefore had to restrict the galactic motion to one dimension ; willems et al .
@xcite assumed a broad flat radial velocity distribution ; willems et al .
@xcite used population synthesis for the formation and motion of ns - ns binaries to derive theoretical radial velocity distributions filling in the void left by the lack of empirical constraints ; stairs et al .
@xcite also adopted the latter in combination with updated proper motion constraints , but at the same time argued in favor of a radial velocity distribution obtained by assuming an isotropic distribution for the angle @xmath45 between the full 3-d space velocity and our line - of - sight .
the choice between these two priors alters their most likely kick velocity at 95% confidence level from 451005kms@xmath0 to 20140kms@xmath0 .
given this level of influence on results , we examine this question in more detail in what follows .
the isotropic assumption appears reasonable in the absence of other constraints and most importantly appears intuitive .
quantitatively , it leads to a much narrower radial velocity pdf than the one obtained from population synthesis models ( see figure 1 in @xcite ) .
however , it is possible that the motion of the ns - ns population introduces velocity anisotropicities , especially when proper motion and location constraints are incorporated .
in particular , if the motion is dominated by galactic rotation , as it is for the progenitors of ns - ns , the systemic velocity distribution will be strongly anisotropic and depend on the direction of the line - of - sight .
specifically let us consider the following _ thought experiment _ : if only small kicks are imparted to the second - born ns at birth , an observer located at the galactic center will see a @xmath45-distribution that is _ strongly peaked _ near angles of @xmath50 .
it is hard to envision that such a distribution would appear isotropic when the observer s viewpoint is shifted from the galactic center to the position of the sun . to illustrate this further
, we use the theoretical ns - ns populations synthesized by willems et al .
@xcite in the case of isotropic sn kick velocities with magnitudes drawn from maxwellian distributions with dispersions of 50kms@xmath0 ( as an example of pre - dominantly low kicks ) and 260kms@xmath0 ( as inferred by hobbs et al .
we follow the formation and motion of these systems in the galaxy and use the results to derive present - day radial and transverse velocities of ns - ns binaries as well as pdfs for the angle @xmath45 between the 3-d space velocities and the line - of - sight .
we note that with this definition for @xmath45 the radial velocity @xmath51 is related to the transverse velocity @xmath52 as @xmath53 , so @xmath54 points away from us and @xmath55 towards us .
stairs et al .
@xcite used this relation to derive an empirical radial velocity distribution for psr j0737 - 3039 using @xmath56 @xcite and assuming an isotropic distribution for @xmath45 . in figure [ f1 ] , we show the resulting present day pdfs of the angle @xmath45 for the entire ns - ns population in the two kick velocity models , for the sub - population of systems within 50pc from the galactic plane , and for the sub - population of systems within 50pc from the galactic plane with a proper motion of less than10kms@xmath0 .
the distributions are _ clearly anisotropic _ for both kick velocity models , with a strong preference for directions along the line - of - sight ( towards or away from us ) emerging when the population of ns - ns systems is restricted to binaries within 50pc from the galactic plane _ and _ a small proper motion of less than 10kms@xmath0 , characteristic of psr j0737 - 3039 . in figure
[ f2 ] , we show the @xmath45-pdfs for the sub - population of ns - ns binaries located within @xmath47 from psr j0737 - 3039 s position in galactic longitude and latitude . here too , the distribution of @xmath45-angles is clearly anisotropic , with velocity vectors pointing primarily towards us .
the anisotropy persists if the sample population is restricted to systems within 1kpc from the sun ( the current upper limit on the distance from the sun to psr j0737 - 3039 @xcite ) .
the same holds true when combining the constraints on the vertical distance to the galactic plane , the proper motion , and the position in terms of galactic coordinates , though the statistics of the simulated samples become considerably poorer .
we conclude that , although it appears intuitive that the direction of the space velocity of the double pulsar is distributed isotropically around the line - of - sight , the system s motion due to galactic rotation and the post - sn peculiar velocity combined with the non - central position of observers on earth induce significant anisotropies in the double pulsar s most likely space velocity . such anisotropies are implicitly taken into account in the theoretical radial velocity distributions derived and used by willems et al .
@xcite to constrain the formation of the second - born pulsar in psr j0737 - 3039 ; therefore the resulting broader distributions of radial velocities are physically motivated and self - consistent compared to the narrower distribution obtained with the ad hoc assumption of isotropic velocities with respect to our line - of - sight . 9 tauris , t. m. , & van den heuvel , e. p. j. 2006 , compact stellar x - ray sources , 623 bhattacharya , d. , & van den heuvel , e. p. j. 1991 , physics reports , 203 , 1 miyaji , s. , nomoto , k. , yokoi , k. , & sugimoto d. 1980 , pasj , 32 , 303 nomoto , k. 1984 , apj , 277 , 791 nomoto , k. 1987 , apj , 322 , 206 wheeler , j. c. , cowan , j. j. , & hillebrandt , w. 1998 , apj , 493 , l101 wanajo , s. , tamamura , m. , itoh , n. , nomoto , k. , ishimaru , y. , beers , t. c. , & nozawa , s. 2003 , apj , 593 , 968 podsiadlowski , p. , langer , n. , poelarends , a. j. t. , rappaport , s. , heger , a. , & pfahl , e. 2004 , apj , 612 , 1044 kitaura , f. s. , janka , h .-
t . , & hillebrandt , w. 2006 , a&a , 450 , 345 janka , t. 2008 , `` supernova explosions and neutron star birth '' , in this volume poelarends , a. j. t. , herwig , f. , langer , n. , & heger , a. 2007 , arxiv e - prints , 705 , arxiv:0705.4643 eldridge , j. j. , & tout , c. a. 2004 , mnras , 353 , 87 siess , l. 2007 , a&a , 476 , 893 gil - pons , p. , gutirrez , j. , & garca - berro , e. 2007 , a&a , 464 , 667 miyaji , s. , & nomoto , k. 1987 , apj , 318 , 307 hashimoto , m. , iwamoto , k. , & nomoto , k. 1993 , apj , 414 , l105 brown , g. e. 1995 , apj , 440 , 270 fryer , c. l. , benz , w. , & herant , m. 1996 , apj , 460 , 801 fryer , c. , & kalogera , v. 1997 , apj , 489 , 244 belczynski , k. , kalogera , v. , & bulik , t. 2002 , apj , 572 , 407 habets , g. m. h. j. 1986 , a&a , 167 , 61 dewi , j. d. m. , & van den heuvel , e. p.
j. 2004 , mnras , 349 , 169 willems , b. , & kalogera , v. 2004 , apj , 603 , l101 willems , b. , kalogera , v. , & henninger , m. 2004 , apj , 616 , 414 thorsett , s. e. , dewey , r. j. , & stairs , i. h. 2005 , apj , 619 , 1036 stairs , i. h. , thorsett , s. e. , dewey , r. j. , kramer , m. , & mcphee , c. a. 2006 , mnras , 373 , l50 dewi , j. d. m. , pols , o. r. , savonije , g. j. , & van den heuvel , e. p. j. 2002 ,
mnras , 331 , 1027 ivanova , n. , belczynski , k. , kalogera , v. , rasio , f. a. , & taam , r. e. 2003 , apj , 592 , 475 dewi , j. d. m. , & pols , o. r. 2003 , mnras , 344 , 629 belczynski , k. , kalogera , v. , rasio , f. a. , taam , r. e. , zezas , a. , bulik , t. , maccarone , t. j. , & ivanova , n. 2007 , apjs , in press ( astro - ph/0511811 ) ivanova , n. , heinke , c. , rasio , f. a. , belczynski , k. , & fregeau , j. 2007 , arxiv e - prints , 706 , arxiv:0706.4096 burgay , m. , et al .
2003 , nature , 426 , 531 podsiadlowski , p. , dewi , j. d. m. , lesaffre , p. , miller , j. c. , newton , w. g. , & stone , j. r. 2005 , mnras , 361 , 1243 piran , t. , & shaviv , n. j. 2005 , phys .
lett . , 94 , 051102 willems , b. , kaplan , j. , fragos , t. , kalogera , v. , & belczynski , k. 2006 , phys .
d , 74 , 043003 gunn , j. e. , & ostriker , j. p. 1970 ,
apj , 160 , 979 arzoumanian , z. , chernoff , d. f. , & cordes , j. m. 2002 , apj , 568 , 289 hobbs , g. , lorimer , d. r. , lyne , a. g. , & kramer , m. 2005 , mnras , 360 , 974 zou , w. z. , hobbs , g. , wang , n. , manchester , r. n. , wu , x. j. , & wang , h. x. 2005 , mnras , 362 , 1189 heinke , c. o. , grindlay , j. e. , edmonds , p. d. , cohn , h. n. , lugger , p. m. , camilo , f. , bogdanov , s. , & freire , p. c. 2005 , apj , 625 , 796 pfahl , e. , rappaport , s. , & podsiadlowski , p. 2002 , apj , 573 , 283 ivanova , n. , heinke , c. o. , rasio , f. a. , taam , r. e. , belczynski , k. , & fregeau , j. 2006 , mnras , 372 , 1043 kuranov , a. g. , & postnov , k. a. 2006 , astron .
, 32 , 393 pfahl , e. , rappaport , s. , podsiadlowski , p. , & spruit , h. 2002 , apj , 574 , 364 van den heuvel , e. p. j. 2004 , 5th integral workshop on the integral universe , 552 , 185 chaurasia , h. k. , & bailes , m. 2005 , apj , 632 , 1054 ihm , c. m. , kalogera , v. , & belczynski , k. 2006 , apj , 652 , 540 mclaughlin , m. a. , et al .
2005 , binary radio pulsars , 328 , 43 faulkner , a. j. , et al .
2005 , apj , 618 , l119 dewi , j. d. m. , podsiadlowski , p. , & pols , o. r. 2005 , mnras , 363 , l71 willems , b. , andrews , j. , kalogera , v. , & belczynski , k. 2008 , `` polar kicks and the spin period - eccentricity relation in double neutron stars '' , in this volume ( arxiv:0710.0345 ) kramer , m. , et al .
2006 , science , 314 , 97 ferdman , r. d. , et al .
2008 , `` the double pulsar : evolutionary constraints from the system geometry '' , in this volume ( arxiv:0711.4927 ) stairs , i.h .
2008 , `` double neutron star binaries '' , in this volume dewi , j. d. m. , podsiadlowski , p. , & sena , a. 2006 , mnras , 368 , 1742 wex , n. , kalogera , v. , & kramer , m. 2000 , apj , 528 , 401 piran , t. , & shaviv , n. j. 2005 , arxiv astrophysics e - prints , arxiv : astro - ph/0510584 | one of the primary goals when studying stellar systems with neutron stars has been to reveal the physical properties of progenitors and understand how neutron star spins and birth kicks are determined . over the years
a consensus understanding had been developed , but recently some of the basic elements of this understanding are being challenged by current observations of some binary systems and their theoretical interpretation . in what follows we review such recent developments and
highlight how they are interconnected ; we particularly emphasize some of the assumptions and caveats of theoretical interpretations and examine their validity ( e.g. , in connection to the unknown radial velocities of pulsars or the nuances of multi - dimensional statistical analysis ) .
the emerging picture does not erase our earlier understanding ; instead it broadens it as it reveals additional pathways for neutron star formation and evolution : neutron stars probably form at the end of both core collapse of fe cores of massive stars and electron - capture supernovae of onemg cores of lower - mass stars ; birth kicks are required to be high ( well in excess of 100kms@xmath0 ) for some neutron stars and low ( @xmath1kms@xmath0 ) for others depending on the formation process ; and spin up may occur not just through roche - lobe overflow but also through wind accretion or phases of hypercritical accretion during common envelope evolution .
address = northwestern university , department of physics and astronomy , 2131 tech drive , evanston , il 60208 , usa address = northwestern university , department of physics and astronomy , 2131 tech drive , evanston , il 60208 , usa address = northwestern university , department of physics and astronomy , 2131 tech drive , evanston , il 60208 , usa |
the ubiquity of supermassive black holes ( smbhs ) in the nuclei of many galaxies ( e.g. magorrian et al .
1998 ) suggests that binary and multiple smbh systems may also be widespread .
for example , smbh binaries should form in galaxy mergers ( begelman , blandford , & rees 1980 ) , which is a common process in hierarchical models for galaxy formation and evolution ( e.g. kauffmann & haehnelt 2000 ; menou , haiman , & narayanan 2001 ) .
smbh binaries have been invoked to explain periodic wiggles in extragalactic radio jets ( e.g. roos , kaastra , & hummel 1993 ) , periodic flares in the bl lac source oj 287 ( e.g. lehto & valtonen 1996 ) , variations in the apparent superluminal transverse velocities and position angles of the 3c 273 radio jet ( romero et al .
2000 ) , and the `` core - type '' nuclear surface brightness profiles of bright elliptical galaxies ( ebisuzaki , makino , & okumura 1991 ; quinlan & hernquist 1997 ; faber et al .
1997 ; milosavljevi & merritt 2001 ) .
smbh binary mergers would be powerful sources of gravitational waves ( thorne & braginsky 1976 ) which should be easily detectable from planned space - based gravitational wave observatories such as the _ laser interferometer space antenna _ ( bender et al .
1998 ) . however , it is still uncertain whether binary smbhs can evolve to small enough semimajor axes that gravitational radiation drives them to merge .
energy can be extracted from a wide binary through interactions with surrounding matter .
for example , encounters with passing stars will shrink the binary , but this process will eventually be limited by the depletion of stars on sufficiently radial orbits to encounter the ever tightening binary
. detailed work on this aspect of the problem has been done by numerous authors ( e.g. quinlan 1996 , quinlan & hernquist 1997 , milosavljevi & merritt 2001 ) .
a key uncertainty is the amount of stochastic wandering that the binary undergoes at the bottom of the potential well of the galaxy , allowing it to interact with many more stars .
another important process that has received relatively less attention is interaction with surrounding gas , which can also extract energy from the binary and drive it to merge ( e.g. armitage & natarajan 2002 and references therein ) .
large galaxies are typically the product of multiple mergers over the history of the universe .
if the characteristic merger time of binary smbhs is not much less than the characteristic time scale between mergers , then interactions with a third smbh or a second smbh binary will likely take place .
strong encounters within these multiple black hole systems may lead to slingshot ejections of smbhs from the nucleus of the galaxy ( valtonen et al .
1994 , valtonen 1996 ) .
on the other hand , the very existence of the additional galaxy merger may introduce more stellar encounters with the original central binary and drive it to merge ( roos 1988 ) . moreover , repeated encounters between a third black hole and the inner binary can increase the probability that the inner binary attains a large eccentricity , thereby accelerating the energy and angular momentum loss by gravitational radiation ( makino & ebisuzaki 1994 ) .
a modification of the last scenario is a case where the third black hole has evolved to the point that it has become bound to the smbh binary , but has not yet come close enough for an unstable three - body interaction to take place .
the system then forms a hierarchical triple and can be treated as consisting of an inner binary and an outer binary . in that case , if the mutual inclination angle between the inner and outer binaries is high enough , then the time - averaged tidal gravitational force on the inner binary can induce an oscillation in its eccentricity .
this effect is known as the kozai mechanism ( kozai 1962 ) .
provided all other dynamical influences are negligible ( see section 2 ) , an initially very small eccentricity in the inner binary will oscillate through a maximum value given by @xmath0 provided @xmath1 , i.e. the initial mutual inclination angle @xmath2 lies between roughly @xmath3 and @xmath4 ( kozai 1962 ) .
given that the gravitational wave merger time is a strong function of eccentricity ( approximately proportional to @xmath5^{7/2}$ ] ) , highly inclined orbits could in principle greatly reduce the merger time .
for example , a mutual inclination angle of @xmath6 results in eccentricity oscillations which , when at maximum amplitude , reduce the characteristic merger time by an order of magnitude .
although the true merger time will depend on how long the inner binary spends at high eccentricity , it appears promising that a random distribution of initial inclinations will result in dramatically accelerated mergers in many cases .
this is the subject we will explore in the present paper .
while this research was being completed , we learned of recent work by miller & hamilton ( 2002 ) , who investigate a very similar idea : the use of the kozai mechanism to accelerate binary mergers in triple black hole systems formed in globular clusters .
we begin in section 2 by going through the characteristic time scales which will affect the dynamics of triple supermassive black hole systems .
then in section 3 we summarize the evolution equations of an isolated triple used in our numerical calculations .
section 4 presents the results of those calculations , which help delineate the regions of initial condition parameter space where the kozai mechanism plays a substantial role in accelerating the merger of the inner binary .
we discuss our results and summarize our conclusions in section 5 .
consider a binary system consisting of two black holes with masses @xmath7 and @xmath8 , with initial semimajor axis @xmath9 and eccentricity @xmath10 .
the time it takes for the binary to merge due to gravitational wave emission ( peters 1964 ) can be written as @xmath11 where @xmath12 is a weak function of the initial eccentricity that is of order unity ( @xmath13 for all @xmath10 ) .
this gravitational wave merger time is a strong function of both the semimajor axis and the eccentricity .
hence gravitational radiation only becomes important late in the evolution of the binary .
prior to that time , the evolution is dominated by interactions between the binary and surrounding material , either stars or gas .
these interactions are very complex , due to the fact that the surroundings themselves evolve due to their interaction with the binary .
( see milosavljevi & merritt 2001 for a recent discussion of stellar interactions . ) as we noted above , it is not yet clear whether these interactions are sufficient to harden the binary to a point where gravitational radiation then causes it to merge .
the scenario we wish to explore in this paper is one where the binary s semimajor axis evolution has stalled because of insufficient interactions with surrounding material , and where a subsequent galaxy merger then introduces a third black hole in the system .
we envisage the third black hole eventually forming a bound , hierarchical triple with the binary .
the tidal gravitational torques exerted on the inner binary by the outer black hole can alter the eccentricity of the inner binary , thereby affecting the rate of gravitational wave emission .
general relativity causes periastron precession in the inner binary , with a period @xmath14 provided this general relativistic precession is unimportant , and the outer black hole is in a sufficiently inclined orbit around the inner binary , then the eccentricity of the inner binary will oscillate by the kozai mechanism .
the characteristic time scale for these oscillations is given by ( e.g. holman , touma , & tremaine 1997 ) @xmath15 where @xmath16 is the mass of the third body and @xmath17 and @xmath18 are the semimajor axis and eccentricity , respectively , of its orbit around the inner binary .
general relativistic precession can stop these eccentricity oscillations by destroying the kozai resonance ( e.g. holman et al .
1997 ) . a precise criterion for this not to happen
is easily derived ( see equation [ [ a2cubed ] ] in appendix ) : and factors of order unity , this is roughly equivalent to @xmath19 . equation ( [ a2a1 ] ) provides a more quantitatively accurate description of the behavior seen in our numerical results . ]
@xmath20 the outer black hole must therefore come quite close to the inner binary for eccentricity oscillations to take place , although the triple is still hierarchical in the sense that the semimajor axis ratio @xmath21 is large .
interactions with the surrounding stars and gas may cause the outer black hole s semimajor axis to evolve significantly over the merger time scale of the inner binary , a point to which we shall return in section 5 below . for now , however , we will neglect this fact and consider the evolution of an isolated , hierarchical black hole triple .
as we will show , once the outer black hole is sufficiently close that equation ( [ a2a1 ] ) is satisfied , the merger time of the inner binary drops substantially for sufficiently high mutual inclinations .
we assume that the triple is hierarchical , so that the orbit of the outer black hole is much larger in size than the orbit of the inner binary . in this case
the triple can be considered to consist of two binaries in approximately keplerian orbits : the inner binary consisting of black holes with masses @xmath7 and @xmath8 ; and the outer binary consisting of the center of mass of the inner binary , with mass @xmath22 , and the third black hole with mass @xmath16 .
let the semimajor axes of the inner and outer binaries be @xmath9 and @xmath17 , respectively , and the eccentricities be @xmath10 and @xmath18 .
we also define @xmath23 and @xmath24 to be the corresponding arguments of periastron .
the magnitude of the angular momenta @xmath25 and @xmath26 of the inner and outer binaries are given by @xmath27^{1/2}\ ] ] and @xmath28^{1/2}\ ] ] respectively , where @xmath29 is newton s gravitational constant .
let the total orbital angular momentum of the triple be @xmath30 . in the absence of gravitational radiation , tidal torques , or gravitational interactions with surrounding stars , this vector would be rigorously conserved .
let @xmath31 ( @xmath32 ) be the inclination of the inner ( outer ) binary , i.e. the angle between @xmath25 ( @xmath26 ) and @xmath33 .
then @xmath34 and @xmath35 the mutual inclination angle between the two binaries is @xmath36 .
marchal ( 1990 ) , krymolowski & mazeh ( 1999 ) and ford , kozinsky , & rasio ( 2000 ) have derived the orbit - averaged hamiltonian of an isolated , newtonian hierarchical triple of point masses , using secular perturbation theory to octupole order , i.e. to order @xmath37 .
we adopt the equations of motion for the orbital elements of ford et al . here , resulting from the canonical transformation of the von zeipel method .
we have corrected a sign error in the octupolar terms in the equations of motion of ford et al .
( 2000 ) . in equations ( 22 ) and ( 29)-(32 ) of their paper , all terms involving @xmath38 should have the opposite sign .
the equations of motion of krymolowski & mazeh ( 1999 ) also contain the same sign error . ] but modify them to incorporate two general relativistic effects on the inner binary : the precession of periastron and gravitational radiation .
we do this by simply adding orbit - averaged general relativistic correction terms to the ford et al .
( 2000 ) equations of motion for the orbital elements of the inner binary , the same correction terms which would exist if the inner binary were isolated .
we must also take into account the fact that @xmath33 is no longer conserved because of the radiative loss @xmath39 of the inner binary s orbital angular momentum .
we do this by noting that in the absence of interactions between the inner and outer binaries , gravitational radiation acts to change the magnitude of @xmath25 , but not its direction .
in addition , the vector @xmath26 remains unchanged .
the resulting equation is @xmath40 note that if the black holes are spinning , then the orbital plane of the inner binary can also evolve because of general relativistic spin - orbit coupling .
this is a higher order effect than general relativistic precession , and we neglect it entirely , assuming that we are essentially dealing with schwarzschild black holes .
thus our evolution equations for the orbital elements are @xmath41 @xmath42+{\theta\over g_2}[2+e_1 ^ 2(3 - 5\cos2g_1)]\right\ } \nonumber\\ & & + c_3e_2e_1\left({1\over g_2}+{\theta\over g_1}\right ) \left\{\sin g_1\sin g_2[a+10(3\theta^2 - 1)(1-e_1 ^ 2)]- 5\theta b\cos\phi\right\}\nonumber\\ & & -c_3e_2{1-e_1 ^ 2\over e_1g_1}\left[10\theta(1-\theta^2 ) ( 1 - 3e_1
^ 2)\sin g_1\sin g_2+\cos\phi(3a-10\theta^2 + 2)\right ] \nonumber\\ & & + { 3\over c^2a_1(1-e_1 ^ 2)}\left[{g(m_0+m_1)\over a_1 } \right]^{3/2 } , \label{dg1dt}\end{aligned}\ ] ] @xmath43\nonumber\\ & & -{304g^3m_0m_1(m_0+m_1)e_1\over15c^5a_1 ^ 4(1-e_1 ^ 2)^{5/2 } } \left(1+{121\over304}e_1 ^ 2\right ) , \label{de1dt}\end{aligned}\ ] ] @xmath44 @xmath45 + { 1\over g_2}[4 + 6e_1 ^ 2+(5\theta^2 - 3)(2 +
3e_1 ^ 2 - 5e_1 ^ 2 \cos2g_1)]\right\}\nonumber\\ & & -c_3e_1\sin g_1\sin g_2\bigg\{{4e_2 ^ 2 + 1\over e_2g_2}10\theta ( 1-\theta^2)(1-e_1 ^ 2 ) \nonumber\\ & & \qquad\qquad\qquad\qquad\quad -e_2\left({1\over g_1}+ { \theta\over g_2}\right)[a+10(3\theta^2 - 1)(1-e_1 ^ 2)]\bigg\ } \nonumber\\ & & -c_3e_1\cos\phi\left[5b\theta e_2\left({1\over g_1}+ { \theta\over g_2}\right)+{4e_2 ^ 2 + 1\over e_2g_2}a\right],\end{aligned}\ ] ] @xmath46 , \label{de2dt}\ ] ] and @xmath47^{1/2 } \left(1+{7\over 8}e_1 ^ 2\right){g_1+g_2\theta\over h}. \label{dhdt}\ ] ] the quantities @xmath48 and @xmath38 multiply the quadrupole and octupole perturbation terms , respectively .
they are defined by ( ford et al .
2000 ) @xmath49 and @xmath50 note that the octupole terms vanish if @xmath51 .
the quantities @xmath52 and @xmath53 in these terms are given by @xmath54 and @xmath55 finally , the quantity @xmath56 is the cosine of the mutual inclination of the binaries , @xmath57 and @xmath58 is the angle between the periastron directions , @xmath59 note that equation ( [ eqtheta ] ) determines the evolution of the mutual inclination through the time dependence of the eccentricities , @xmath9 , and @xmath60 .
the terms in the equations of motion which remain after setting @xmath61 are the general relativistic correction terms .
the last term in equation ( [ dg1dt ] ) is the general relativistic precession term , and the terms involving the speed of light @xmath62 in equations ( [ da1dt ] ) , ( [ de1dt ] ) , and ( [ dhdt ] ) are the gravitational radiation terms
. we may immediately deduce some important features of our evolution equations .
equations ( [ de1dt ] ) and ( [ de2dt ] ) imply that if both the inner and outer binaries are circular ( @xmath63 ) , then they will remain that way .
an eccentric outer binary will produce a nonzero eccentricity in an initially circular inner binary , and vice - versa , but only because of the octupole interaction terms .
note that if we switch off the quadrupole and octupole interaction terms by setting @xmath61 , then an immediate consequence is @xmath64 , as may be verified by differentiating equation ( [ eqtheta ] ) directly .
in other words , general relativistic effects alone do not change the mutual inclination of the inner and outer binaries if there are no interactions between them .
this is as one would expect , given our treatment of the gravitational wave angular momentum loss in equation ( [ eqdhrad ] ) .
our evolution equations also exhibit an important scaling .
if all masses and all initial semimajor axes are multiplied by some constant factor , then the merger time and all other time scales change by the same multiplicative factor .
this is a direct consequence of the nature of gravity , where mass , length , and time have the same dimensions in units where @xmath65 , and it reduces the parameter space of masses and semimajor axes which need to be explored numerically .
another similarly useful fact is that the equations are unchanged when @xmath7 and @xmath8 in the inner binary are interchanged and @xmath23 or @xmath24 is increased by @xmath66 .
the former corresponds to a relabeling of the inner binary black holes , while the latter represents a spatial inversion of the two orbits .
while our implementation of the general relativistic effects has been heuristic , the general relativistic precession terms for the inner binary can in fact be rigorously justified by orbit - averaging the post - newtonian hamiltonian . to octupole order ,
the resulting hamiltonian is @xmath67 \nonumber\\ & & + c_3e_1e_2\left[a\cos\phi+10\theta(1-\theta^2)(1-e_1 ^ 2 ) \sin g_1\sin g_2\right ] \nonumber\\ & & + { g^2m_0m_1\over c^2a_1
^ 2}\left[{15m_1 ^ 2 + 15m_0 ^ 2 + 29m_0m_1\over8(m_0+m_1 ) } -{3(m_0+m_1)\over(1-e_1 ^ 2)^{1/2}}\right ] .
\label{hamav}\end{aligned}\ ] ] a rigorous derivation of the radiation terms would be considerably more difficult , and is beyond the scope of this paper .
we believe that our equations of motion will capture the overall time evolution of the inner binary , because the early phases of this evolution will be dominated by interactions with the outer black hole , while the late phases will be dominated by gravitational radiation . in both those limits our evolution equations for the orbital elements of the inner binary are rigorously correct .
it is possible , however , that interesting effects may occur during the transition between these two phases of evolution , which may not be captured by our equations . in the absence of gravitational radiation ,
our equations conserve the orbit averaged hamiltonian in equation ( [ hamav ] ) .
we use this fact to check the accuracy of our numerical integrations .
the reader should bear in mind that our equations are based on orbit averaging and an expansion in the semimajor axis ratio @xmath68 , and may therefore be inaccurate when this ratio is not very small .
we will show in the next section that the octupole terms usually have fairly small effects on the merger time of the inner binary .
hence our neglect of even higher order terms , and our use of such an expansion in the first place , is probably justified for the primary purpose of this paper .
the major exception to this is the issue of stability .
we are investigating whether the presence of a third black hole can cause the inner binary to merge before the third black hole comes close enough to cause an unstable interaction .
there is a limit to how small @xmath21 can be for the triple to be stable and for the evolution we calculate to be valid .
completely general stability criteria for hierarchical triples in arbitrary initial configurations do not exist .
we adopt as a stability criterion the empirical formula used by mardling & aarseth ( 2001 ) , which may be written as @xmath69^{2/5}. \label{eqacrit}\ ] ] this criterion was derived for newtonian coplanar prograde orbits of the inner and outer binaries .
inclined orbits , which are the major focus of this paper , are expected to be more stable , so inequality ( [ eqacrit ] ) provides a conservative stability limit .
this inequality is also conservative in the sense that it is either close to or above the more complicated stability criterion proposed by eggleton & kiseleva ( 1995 ) , which was empirically verified over a wide range of mass ratios including nearly all those we numerically investigate in the next section .
all the calculations we present in this paper are done for systems which are stable according to inequality ( [ eqacrit ] ) .
the stability criterion ( [ eqacrit ] ) may be combined with the constraint of inequality ( [ a2a1 ] ) that general relativistic precession not destroy the kozai resonance to give a lower limit on the inner semimajor axis for which the kozai mechanism can reduce the inner binary merger time .
the result is @xmath70 a stable triple with initial conditions such that the inner semimajor axis @xmath9 violates this inequality will not have an accelerated inner binary merger by the kozai mechanism .
however , in that case @xmath9 is so small that the binary merger time by equation ( [ tmerge ] ) is @xmath71 at least for roughly equal mass triples and outer eccentricities that are not too high , this implies that the inner binary would already be rapidly merging and needs no help from the kozai mechanism . on the other hand , equations ( [ a1crit ] ) and ( [ tmcrit ] )
have a rather strong dependence on @xmath18 , so if the outer black hole s orbit is _ very _ eccentric , then the resulting triple may be too unstable for there to be time for the kozai mechanism to operate . however , numerical simulations of black hole binary evolution by milosavljevi & merritt ( 2001 ) suggest that this is rather unlikely .
they find rather modest initial eccentricities when the binary first forms , and the subsequent evolution of the binary does not go to very high eccentricity .
hence there is probably plenty of parameter space for stable hierarchical triples to form and evolve under the kozai mechanism .
we have numerically integrated equations ( [ da1dt])-([dhdt ] ) for a wide range of possible black hole masses and initial conditions . even for fixed black hole masses ,
the parameter space of possible initial conditions is huge , and it helps to understand which are the most important quantities affecting the merger time .
we therefore focus first on the detailed evolution of a particular triple consisting of nearly equal mass black holes .
specifically , we consider an inner binary with a @xmath72 m@xmath73 black hole and a @xmath74 m@xmath73 black hole , about which orbits another @xmath74 m@xmath73 black hole ten times further out .
we have purposely chosen the two inner black holes to have different masses so as to allow for octupole interaction effects in the evolution .
figure 1 depicts the evolution of inner semimajor axis @xmath9 , inner eccentricity @xmath10 , and outer eccentricity @xmath18 for a particular choice of initial conditions .
the inner binary starts out nearly circular ( @xmath75 ) with a semimajor axis which would give a gravitational wave merger time @xmath76 yr if the binary were isolated .
however , the presence of the outer black hole induces large amplitude eccentricity oscillations in the inner binary through the kozai mechanism .
the time spent at higher eccentricity reduces the gravitational wave merger time of the two inner black holes by roughly a factor thirty for the particular case shown .
in contrast to @xmath10 , the outer eccentricity @xmath18 stays close to its original value throughout the evolution .
equation ( [ de2dt ] ) shows that @xmath18 only changes as a result of octupole interactions , which are weaker than the quadrupolar interactions driving the inner binary eccentricity oscillations .
the lack of strong evolution in the outer eccentricity is a generic feature of all the numerical calculations we have done , implying that @xmath18 is unlikely to be driven to high enough values to destabilize an isolated triple .
as might be expected , the merger time turns out to be rather insensitive to the initial orientation of the roughly circular inner orbit , i.e. the initial value of the inner argument of periastron @xmath23 .
figure 2 shows the evolution of the same triple as depicted in figure 1 , except that the initial value of @xmath23 has been shifted by @xmath77 .
this results in a merger time which is about seven percent longer than that shown in figure 1 . the overall behavior of the eccentricity oscillations depicted in figures 1(b ) and 2(b )
can be understood by examining the evolution in the @xmath10 vs. @xmath23 phase space .
the initial phase space trajectories for triples with the same total angular momentum as the triples of figures 1 and 2 are depicted in figure 3 .
depending on the initial conditions , the evolution in @xmath23 is either one of libration about @xmath78 or @xmath79 , or one of circulation .
the quadrupolar fixed point of the kozai resonance lies inside the smallest libration contour at @xmath80 .
the only difference between the triples of figures 1 and 2 is that the former starts off circulating while the latter starts off librating .
figure 4 depicts snapshots of the phase space trajectories during the course of the evolution shown in figures 1 and 2 . as shown in figure 4(b ) , gravitational radiation drives an initially librating inner binary over into a circulating inner binary , thereby causing the minimum eccentricity to drop until it crosses the separatrix .
this separatrix crossing corresponds to the momentary zero eccentricity spike just before @xmath81 years in figure 2(b ) .
thereafter , the evolution is very similar to the case where the inner binary starts off circulating : the minimum eccentricity rises until gravitational radiation becomes so strong that it starts to circularize the orbit .
figure 5 shows the overall dependence of the merger time for varying initial inner argument of periastron @xmath23 , again for an inner orbit which is initially roughly circular : @xmath75 .
while the overall effect of @xmath23 on the merger time is not too great , the exact value of the merger time is extremely sensitive to the initial value of @xmath23 for triples close to the libration / circulation separatrix , or for those triples which start out librating ( @xmath23 initially around @xmath77 or @xmath79 ) and therefore subsequently cross the separatrix . as shown in figure 3 , the region near the separatrix is chaotic , and this chaos arises from the octupole interaction terms .
if these terms are neglected , the system then lacks the necessary degrees of freedom to exhibit chaos , and the dependence of the merger time on the initial value of @xmath23 is much smoother ( the dashed line in figure 5 ) .
note that the separatrix is associated with passing through very small values of @xmath10 [ cf .
figure 2(b ) ] .
one might therefore worry that our results suffer from numerical problems associated with the @xmath82 singularity in the octupole term of equation ( [ dg1dt ] ) .
however , following the suggestion of ford et al .
( 2000 ) , we have removed this singularity in our numerical integrations by first transforming the dependent variables from @xmath83 to @xmath84 .
in addition , the evolution of the hamiltonian , equation ( [ hamav ] ) , is completely smooth through the separatrix crossing , changing only as a result of gravitational radiation losses .
we therefore believe the `` noise '' exhibited in figure 5 is physical , and is caused by chaotic behavior during evolution through the separatrix .
the initial orientation of the inner orbit affects the merger time much more substantially when the inner orbit is more eccentric .
figure 6 depicts the merger time as a function of the initial inner eccentricity @xmath10 for two values of the initial inner argument of periastron : @xmath85 and @xmath77 .
the former case corresponds to a circulating inner binary which starts out with an eccentricity which is at the minimum in the kozai oscillation ( see figure 3 ) .
hence in this case the kozai mechanism _
always _ speeds up the merger compared to the time @xmath86 it would take an isolated binary with the same initial eccentricity to merge . in contrast ,
inner binaries with initial @xmath78 start at the minimum in the kozai eccentricity oscillation for values of @xmath10 which are below the kozai fixed point , and at the maximum for values of @xmath10 above it .
we would therefore expect the kozai mechanism to reduce the merger time in the former case and increase it in the latter .
this expectation is confirmed by the behavior shown in figure 6 , provided the binary does not start out too near the kozai fixed point which occurs at @xmath87 for this set of triples with fixed mutual inclination angles . near the kozai fixed point , the merger time is generally increased by the kozai mechanism , due perhaps to the fact that gravitational radiation drives the eccentricity at the kozai fixed point down to lower values with time .
we therefore conclude that the kozai mechanism always acts to reduce the merger time of an initially nearly circular binary , but for eccentric inner binaries it will either reduce or increase the merger time , depending on the orientation of the inner binary within the triple system .
the @xmath78 curve in figure 6 exhibits the chaos we typically find for inner binaries which start out librating , or circulating near the separatrix .
this chaos is also present at the lowest values of eccentricity @xmath10 in the @xmath85 curve , as the inner binary then starts out in the chaotic zone shown in figure 3 .
we turn now to the effects of the orientation and eccentricity of the outer binary on the merger time of the inner binary .
equations ( [ dg1dt])-([de1dt ] ) and ( [ eqc2])-([eqc3 ] ) show that increasing the eccentricity @xmath18 of the outer binary at fixed semimajor axis @xmath17 strengthens both the quadrupolar and octupolar interaction terms , with the latter being enhanced over the former .
this is of course physically reasonable as the distance of closest approach of the outer black hole with the inner binary is smaller . on the other hand , the outer argument of periastron
@xmath24 only affects the evolution of the inner binary through the octupolar interaction terms . we would therefore expect that increasing @xmath18 would generally decrease the merger time through the ( quadrupolar ) kozai mechanism , but that the initial value of @xmath24 could modify this significantly at high eccentricity due to the enhanced octupolar effects .
these expectations are confirmed by our numerical calculations shown in figure 7 .
note , however , that the effects of @xmath18 and @xmath24 are not that large , at least for this particular triple .
we have therefore chosen to fix their initial values to be @xmath88 and @xmath89 for all our other numerical calculations .
perhaps the most important initial condition parameters affecting the inner binary merger time are the semimajor axes @xmath9 and @xmath17 and the mutual inclination of the inner and outer orbits . figure 8 shows the dependence of the merger time on the semimajor axis ratio @xmath21 for triples with fixed initial inclination @xmath90 ( the same as in all previous figures ) and various values of the initial inner semimajor axis @xmath9 . for large initial semimajor axis ratios @xmath21 ( to the right of the dashed line in figure 8) , general relativistic precession destroys the kozai resonance and the eccentricity of the inner binary is largely unaffected by the outer black hole .
the merger time in this case is then the same as that of an isolated binary , and is given by equation ( [ tmerge ] ) .
on the other hand , if the outer black hole comes sufficiently close ( to the left of the dashed line in figure 8) , then the kozai resonance exists and substantial reduction in the merger time occurs as a result of eccentricity oscillations .
the equation for the dashed line itself which separates these two regimes comes from using the binary merger time in equation ( [ tmerge ] ) as a proxy for the inner semimajor axis @xmath9 , and employing equation ( [ a2a1 ] ) : @xmath91 figure 9 shows how the variation of merger time with initial semimajor axis ratio depends on the initial mutual inclination angle of the triple for initial @xmath92 pc .
as expected , significant reduction in the merger time can only occur when the initial inclination is high enough that the kozai resonance is present : @xmath93 . if this criterion is satisfied , then values of @xmath21 satisfying equation ( [ a2a1 ] ) , i.e. @xmath94 for this particular triple , will exhibit accelerated mergers , with higher inclinations showing the fastest merger times . as miller & hamilton ( 2002 ) point out for the stellar mass case , retrograde triples ( @xmath95 ) generally produce faster merger times , at least for nearly equal masses .
as we noted above in section 3 , our numerical results for the nearly equal mass triples of the previous subsection can be scaled to all triples with the same mass ratio , @xmath96 , provided the initial semimajor axes are scaled by the same factor as the masses .
the merger time then scales by exactly the same factor .
we have also investigated triples consisting of substantially unequal masses , and our results may also be scaled to other , similar mass ratio triples in the same manner .
figure 10 depicts the merger time for the same inner binary as shown in figure 8 , but with different outer black hole masses .
all cases shown start with an initial inner semimajor axis of @xmath92 pc , for which the merger time would be @xmath97 yr if the binary were isolated .
because we do not account for gravitational wave losses associated with the outer binary , all the calculations shown in the figure have @xmath21 large enough so that the nominal gravitational wave merger time of the outer binary from equation ( [ tmerge ] ) is at least ten times longer than that of the inner binary .
the @xmath98 m@xmath73 curve shows that an outer black hole with a much smaller mass than that of the inner binary does not have a substantial effect on the merger time of that binary . on the other hand ,
a larger outer black hole mass exerts a stronger tidal perturbation on the inner binary .
this reduces the merger time significantly compared to the nearly equal mass case , and also does it at larger semimajor axis ratios . in agreement with equation ( [ a2a1 ] ) ,
the value of @xmath21 required for the onset of the kozai resonance scales with the outer black hole mass as @xmath99 .
triples in which the outer black hole mass is much larger than that of the inner binary might arise from the merger of a smaller galaxy containing a stalled binary with a larger galaxy containing a bigger black hole .
figure 11 presents a more detailed look at the dependence of the merger time on the semimajor axes @xmath9 and @xmath17 for the case of an outer black hole mass of @xmath100 m@xmath73 . with the exception of the larger outer black hole mass , the initial triple parameters are identical to those of figure 8 .
note the change in scales on the axes : the larger outer black hole mass greatly reduces the merger time at much larger values of @xmath21 .
the dashed line comes once again from equation ( [ eqdash ] ) , and separates the region on the left where the kozai resonance exists from that on the right where relativistic periastron precession destroys the resonance .
figure 12 shows results for a substantially unequal mass inner binary consisting of @xmath101 and @xmath74 m@xmath73 black holes , in a triple with an outer @xmath101 m@xmath73 black hole .
we choose initial inner semimajor axes @xmath9 to be ten times larger than those of figures 8 and 11 so that @xmath86 is roughly the same .
very substantial reductions in the merger time occur once again , but for semimajor axis ratios @xmath21 that are even smaller than in the roughly equal mass case .
larger outer black hole masses can expand this range of @xmath21 ( see equation [ [ a2a1 ] ] ) , but would also accelerate the gravitational wave merger of the outer binary , increasing the likelihood of an unstable encounter .
we have only shown results here for prograde orbits in triples with substantially unequal masses .
in contrast to the nearly equal mass case , retrograde orbits do not produce merger times which are very different from their prograde counterparts .
this asymmetry between prograde and retrograde orbits arises from terms in the equations of motion which depend on odd powers of @xmath56 , which are neglected in the usual kozai approximation ( see appendix a ) .
these terms are also very small for the unequal mass cases we have calculated here .
we have shown that in cases where a bound hierarchical black hole triple forms , substantial reductions in the gravitational wave merger time of the inner binary can take place for reasonably large ranges of initial conditions of the triple .
for example , figure 9 shows that for a triple consisting of nearly equal mass black holes , the merger time can be reduced by more than a factor of ten in over fifty percent of all cases , provided the mutual inclination angle is randomly distributed in solid angle .
however , we have not addressed the issue of what the distribution of initial conditions is likely to be , as that would entail following the dynamical formation of the triple itself by interactions with the surrounding stars and gas . as we have shown in the previous section ,
all the initial parameters of the triple affect whether and by how much the merger of the inner binary is altered by the presence of the outer black hole . however , the most important are the mutual inclination angle @xmath102 and the semimajor axes @xmath9 and @xmath17 .
there is very little effect on the merger time of the inner binary unless the semimajor axis ratio @xmath21 is small enough that general relativistic precession does not destroy the kozai resonance ( equation [ [ a2a1 ] ] ) .
an issue related to the uncertainty in initial conditions is the fact that we have not considered the evolution of the semimajor axis of the outer black hole due either to interactions with surrounding material or to gravitational radiation .
the latter is not significant for the calculations shown here , as we have always chosen @xmath21 and @xmath18 to be such that the gravitational wave evolution time scale of the outer black hole is much longer than the merger time scale of the inner binary .
note that @xmath18 evolves purely due to octupolar interactions ( equation [ [ de2dt ] ] ) , and generally does not undergo the dramatic changes that @xmath10 exhibits .
hence we do not expect dramatic reductions in the _
binary merger time scale due to interactions with the inner binary .
however , that still leaves open the issue of whether the inner binary will merge before interactions with surrounding material cause the outer black hole to come sufficiently close that an unstable three body encounter occurs .
it would be interesting to follow the evolution of the triple by including the time dependence of @xmath17 starting at large enough values that the kozai resonance is destroyed by general relativistic periastron precession .
in addition to the kozai mechanism , the presence of a third black hole may affect the merger time scale of the inner binary in other ways .
in particular , being bound in a triple forces the inner binary to `` wander '' through space , albeit in a regular orbit rather than a stochastic trajectory .
this may enhance the binary s interactions with surrounding stars beyond what it would have had if the outer black hole were not present , and therefore the hardening rate might be increased .
eccentricity oscillations in the inner binary may be induced by other sources of tidal gravitational fields besides that of an orbiting outer black hole , e.g. nearby matter inhomogeneities or an aspherical distribution of surrounding stars and gas . in order to beat general relativistic periastron precession ,
equation ( [ a2cubed ] ) implies that the quadrupole moment @xmath103 of these exterior matter distributions must satisfy @xmath104 note that this critical quadrupole moment has a strong scaling with the binary semimajor axis @xmath9 , and it may be possible to achieve given the actual stellar mass densities observed in galactic nuclei ( e.g. faber et al . 1997 ) .
we thank e. ford , s. hughes , m. c. miller , n. murray , s. j. peale , s. tremaine , and y. wu for useful discussions , and the referee for suggestions that greatly improved this paper .
this work was supported in part by nasa grants nag5 7075 and nag5 3646 , and nsf grant ast 9970827 .
we also acknowledge the hospitality of the canadian institute for theoretical astrophysics , where part of this work was completed .
in this appendix we present analytic calculations which are useful in understanding how general relativistic precession modifies the kozai mechanism .
we neglect the octupole terms in the hierarchical triple evolution equations ( [ da1dt])-([dhdt ] ) throughout this section .
the only parameter of the outer binary which then changes is @xmath24 , and this evolution does not affect the inner binary .
@xmath42+{\theta\over g_2}[2+e_1 ^ 2(3 - 5\cos2g_1)]\right\ } \nonumber\\ & & + { 3\over c^2a_1(1-e_1 ^ 2)}\left[{g(m_0+m_1)\over a_1 } \right]^{3/2 } , \label{dg1dtb}\end{aligned}\ ] ] kozai ( 1962 ) has provided an approximate analytic solution to these quadrupole evolution equations when all general relativistic effects are neglected .
the approximation neglects the term multiplied by @xmath106 in equation ( [ dg1dtb ] ) which is smaller than the first term by @xmath107 for comparable masses .
also , equation ( [ eqtheta ] ) implies that @xmath108 which is a constant when octupole terms and gravitational radiation are neglected . again taking @xmath109
, this gives an approximate integral of motion @xmath110 for @xmath111 , kozai s solutions exhibit two classes of dynamical behavior in the evolution of the argument of perihelion @xmath23 : libration about @xmath78 or @xmath79 , and circulation .
there then exists a critical inclination angle for which the librating solutions degenerate to a fixed point in the @xmath10 vs. @xmath23 phase space : @xmath112 . for @xmath113 ,
the resonance vanishes and only circulating solutions exist . in this case
the amplitude of eccentricity oscillations is quite small , so the existence of the resonance can be used as a necessary criterion for determining when large amplitude eccentricity oscillations become possible . because @xmath111 implies that @xmath114 , large amplitude oscillations are only possible for high inclinations : @xmath115 .
general relativistic precession is itself circulation in @xmath23 , so general relativity would be expected to reduce the parameter space where librations exist .
equation ( [ de1dtb ] ) implies that the fixed point , if it exists , still occurs at @xmath78 or @xmath79 .
if we continue to neglect the term multiplied by @xmath106 in equation ( [ dg1dtb ] ) , then setting @xmath116 gives as expected , general relativistic precession decreases @xmath118 , so that the fixed point can only exist if the inclination angle is pushed higher .
this can only be achieved so long as the right hand side of equation ( [ icrit ] ) remains positive , implying that @xmath119 which gives equation ( [ a2a1 ] ) .
equation ( [ icrit ] ) can be rearranged to give an equation for the eccentricity @xmath120 at the fixed point for triples with specified total angular momentum @xmath60 , and therefore specified kozai constant @xmath121 within the kozai approximation , @xmath122^{1/2}. \label{1me12}\ ] ] solving this equation for @xmath120 requires solving a quartic , but written in this way it is again obvious that general relativistic precession lowers the value of @xmath120 for triples of given angular momentum , thereby increasing the circulating portion of phase space at the expense of libration . | we consider the dynamical evolution of bound , hierarchical triples of supermassive black holes that might be formed in the nuclei of galaxies undergoing sequential mergers . the tidal force of the outer black hole on the inner binary produces eccentricity oscillations through the kozai mechanism , and
this can substantially reduce the gravitational wave merger time of the inner binary .
we numerically calculate the merger time for a wide range of initial conditions and black hole mass ratios , including the effects of octupole interactions in the triple as well as general relativistic periastron precession in the inner binary .
the semimajor axes and the mutual inclination of the inner and outer binaries are the most important factors affecting the merger time .
we find that for a random distribution of inclination angles and approximately equal mass black holes , it is possible to reduce the merger time of a near circular inner binary by more than a factor of ten in over fifty percent of all cases .
we estimate that a typical exterior quadrupole moment from surrounding matter in the galaxy may also be sufficient to excite eccentricity oscillations in supermassive black hole binaries , and also accelerate black hole mergers . |
recently , babar collaboration has discovered a narrow state with @xmath4 with a mass of 2317 mev , @xmath5 @xcite .
the existence of a second narrow resonance , @xmath6 with @xmath7 , was confirmed by cleo @xcite .
both states have been confirmed by belle @xcite . soon after the discovery , another set of charmed mesons , @xmath8 and @xmath9 which have the same quantum numbers @xmath10 as @xmath11 has been discovered by belle @xcite . before their discovery ,
quark model and lattice calculations predicted that the masses of these states , in particular @xmath5 and @xmath12 , would be significantly higher than observed @xcite,@xcite .
moreover , these states were predicted to be broad due to the fact that they can decay into @xmath13 and @xmath14 , respectively .
experimentally , the masses of @xmath5 and @xmath12 are below the @xmath13 and @xmath14 thresholds and hence they are very narrow . these facts inspired a lot of theorists to explain the puzzle @xcite . in this paper
we will focus our attention on the weak semileptonic transitions of @xmath15 mesons into lower lying @xmath0-wave charmed mesons ( @xmath16 ) .
these transitions were studied , within a quark model approach , for the first time in @xcite and , more recently , in @xcite where the authors take into account the symmetries of qcd for heavy quarks @xcite , already used in @xcite .
the light - front covariant model @xcite was adopted to study the same subject in @xcite .
the relevant form factors were also evaluated , in the framework of qcd sum rules @xcite , in @xcite . here
we employ a simple constituent quark model @xcite to evaluate semileptonic form factors of b mesons into @xmath0-wave charmed mesons .
the plan of the paper is the following . in the next section
we describe our quark model ; the third section is devoted to introduce and evaluate the @xmath17-wave to @xmath0-wave form factors .
our way to fix the free parameters of the model and the resulting form factors are discussed in section four , while in section five the heavy quark limit of the form factors are computed and compared with heavy quark effective theory predictions ; the @xmath2 , and @xmath3 are also evaluated . in the last section
we show and discuss our numerical results .
in our model @xcite any heavy meson @xmath18 , with @xmath19 and @xmath20 , is described by the matrix @xmath21 where @xmath22 ( @xmath23 ) stands for the heavy ( light ) quark mass ; @xmath24 are their @xmath25momenta ( cfr fig .
[ f : trbd ] ) .
@xmath26 indicates the meson s wave function which is fixed by using a phenomenological approach . the meson constituent quarks vertexes , @xmath27 in eq .
( [ e : hi ] ) , are fixed by using the correct transformation properties under c , p and to enforce the relation @xmath28 for the odd parity , @xmath17-wave heavy mesons @xmath29 , @xmath27 is given by @xmath30 { \sqrt { \frac{2\ , m_{q } m_{q}}{m_h^2 - ( m_{q}- m_{q})^2 } } } \;,\end{aligned}\ ] ] where @xmath31 is the polarization @xmath25vector of the ( vector ) meson @xmath32 .
the vertexes of the lower lying even parity heavy mesons , instead , are given by the matrices @xmath33 \gamma_5 \sqrt{\frac{3\ , m_q m_q}{m_h^2- ( m_q + m_q)^2}}\ ; , \\ \gamma_{^1\!p_1}(\varepsilon ,
q_1,q_2 ) & = & \varepsilon_\mu\left[\frac{(q_1^\mu - q_2^\mu)\ ; m_h}{m_h^2-(m_{q}-m_{q})^2}\right ] \gamma_5 \sqrt{\frac{6\ ; m_q m_q}{m_h^2-(m_q+m_q)^2}}\;.\end{aligned}\ ] ] as already discussed in @xcite , the @xmath25momentum conservation in the meson - constituent quarks vertexes can be obtained defining a heavy running quark mass . for details
we address the reader to the references @xcite . here , for the sake of utility , we recall all the remaining rules of our model for the evaluation of the hadronic matrix elements of weak currents : * for each quark loop with @xmath34-momentum @xmath35 we have @xmath36 a colour factor of 3 and a trace over dirac matrices ; * for the weak hadronic current , @xmath37 , one puts the factor @xmath38 where with @xmath39 we indicate a combination of dirac matrices .
in this section we evaluate the form factors parameterizing the @xmath40 and @xmath41 weak transitions .
the decomposition of these matrix elements of weak currents in terms of form factors are the following ( see also @xcite ) @xmath42 \right\}\ ; , \label{e : gfapam}\\ < h_{0^+}(p^\prime)|a_{\mu } |h_{0 ^ -}(p ) > & = & f_{+}(q^2)(p_\mu+p_\mu^\prime ) + f_{-}(q^2)(p_\mu - p_\mu^\prime)\ ; , \label{e : f1f00p}\\
< h_{^3\!p_1}(p^\prime,\varepsilon)|v_{\mu}-a_{\mu } |h_{0 ^ -}(p ) > & = & -\imath \left\{f^\prime(q^2)\ , \varepsilon^\ast_\mu\ + ( \varepsilon^\ast \cdot p)\ , \left [ \frac{}{}a^\prime_+(q^2)\,(p_\mu+p^\prime_\mu)+a^\prime_-(q^2)\,(p_\mu - p^\prime_\mu)\right ] \right\}\nonumber\\ & & + 2\ g^\prime(q^2)\ \epsilon^{\mu\nu\alpha\beta}\ \varepsilon^\ast_\nu\ , p_\alpha\ , p^\prime_\beta \ ; , \label{e : gfapam3p1}\\ < h_{^1\!p_1}(p^\prime,\varepsilon)|v_{\mu}-a_{\mu } |h_{0 ^ -}(p ) > & = & -\imath \left\{f(q^2)\ , \varepsilon^\ast_\mu\ + ( \varepsilon^\ast \cdot p)\ , \left [ \frac{}{}a_+(q^2)\,(p_\mu+p^\prime_\mu)+a_-(q^2)\,(p_\mu - p^\prime_\mu)\right ] \right\}\nonumber\\ & & + 2\ g(q^2)\ \epsilon^{\mu\nu\alpha\beta}\ \varepsilon^\ast_\nu\ , p_\alpha\ , p^\prime_\beta \;. \label{e : gfapam1p1}\end{aligned}\ ] ] the calculation of the form factors in eqs .
( [ e : f1f0 ] ) and ( [ e : gfapam ] ) for the case of @xmath43 transitions has been done in ref .
however , for the sake of utility , the analytical expressions are reported in appendix [ a:0m0m1 m ] .
+ one of the main results of this paper is the calculation of the form factors appearing in eqs .
( [ e : f1f00p ] ) , ( [ e : gfapam3p1 ] ) and ( [ e : gfapam1p1 ] ) . by way of an example , in the following we describe the calculation of the matrix element @xmath44 and give the expressions of the form factors @xmath45 .
in appendix [ a:0m1p ] we collect the expressions for @xmath46 , @xmath47 and @xmath48 .
note that all the calculations are done in the frame where @xmath49 , @xmath50 , @xmath51 .
let us start considering the @xmath52 transition , @xmath53 , \label{e:0m0p}\end{aligned}\ ] ] where @xmath54 is the integration domain ( see refs .
@xcite ) defined by @xmath55 with @xmath56 @xmath57 and @xmath58 are the azimuthal and the polar angles respectively for the tri - momentum @xmath59 . @xmath60 and @xmath61 ( @xmath62 ) is the mass of the initial ( final ) meson : in eq .
( [ e:0m0p ] ) @xmath63 .
we choose the @xmath64axis along the direction of @xmath65 , the ( tri-)momentum of the w boson ( cfr fig .
[ f : trbd ] ) .
the analytical expressions for the form factors can be obtained by comparing eq .
( [ e:0m0p ] ) with eq .
( [ e : f1f00p ] ) @xmath66\,,\\ \label{e : f1f00pe - b } f_-(q^2 ) & = & \int \frac{k^2 dk d cos\theta \psi_i(k)\psi_f^\ast(k)}{8\,{\pi } ^2\,{m_{{i}}}\ , { \sqrt{\left ( -{{d_{12}}}^2 + { { m_{{i}}}}^2 \right ) \ , \left ( { { m_f}}^2 - { { s_{23}}}^2 \right ) \,{{{e}}_1}\ , { { { e}}_3}}}}\times \nonumber\\ & & \left [ { m_2}{m_{{i}}}{s_{13 } } + \left ( { { m_{{i}}}}^2 + { d_{12}}{s_{23 } } \right ) { { { e}}_2 } - \frac{k}{{|\vec q| } } \cos ( \theta ) \left ( { { { { } { m}}_f}}^2{m_{{i } } } + { { { } { d}}_{12}}{s_{23}}\left({m_{{i}}}+ { { { } { { e}}}_f}\right ) + { { { { } { m}}_{{i}}}}^2 { { { } { { e}}}_f } \right)\right]\,,\end{aligned}\ ] ] where ( @xmath67 and @xmath68 ) . the expressions for the remaining form factors in eqs . (
[ e : f1f0])-([e : gfapam1p1 ] ) are collected in appendix [ a:0m0m1 m ] and [ a:0m1p ] .
the numerical evaluation of the form factors given in section [ s : dcfdf ] requires to specify the expression for the vertex functions and the values of the free parameters of the model . for the vertex functions we adopt two possible forms , the gaussian - type , extensively used in literature ( see for example @xcite ) @xmath69 and the exponential one @xmath70 which is able to fit the results of a relativistic quark model regarding the shape of the meson wave - functions @xcite . in our approach
@xmath71 is a free parameter which should be fixed by comparing a set of experimental data with the predictions of the model . in this paper
we choose to fix the free parameters by a fit to the experimental data on the @xmath72 @xcite and on the spectrum of @xmath73 process @xcite .
the quality of the agreement between fitted spectrum and the corresponding experimental data may be assessed by looking at the figure [ f : spettro ] .
it should be also observed the very small differences between the @xmath73 spectrum using the vertex functions in eqs .
( [ e : wfgauss])-([e : wfexp ] ) . regarding the @xmath74 branching ratio , we obtain @xmath75 for the exponential ( gaussian ) vertex function to be compared to the experimental value : @xmath76 for the charged ( neutral ) b meson . at this stage the two different form of the vertex functions agree equally well with the experimental data .
however , differences emerge when single form factors are considered ( cfr for example table [ t : fdfris1 ] ) .
.the two best fit sets of values of the free parameters are obtained for the exponential ( exp . ) and gaussian ( gauss . )
vertex function .
[ t : datafit1 ] [ cols="^,^,^",options="header " , ] regarding the @xmath1 dependance of the form factors , we find a very good agreement with numerical results assuming the following polar expression @xmath77 the fitted values of @xmath78 can be found in table [ t : fdfris2 ] .
it is interesting to observe that the effective pole mass is not far from the mass of the @xmath79 meson .
a different @xmath1 dependence exhibit the form factors @xmath80 and @xmath81 ; for them we use the form @xmath82 and the values of @xmath83 are collected in table [ t : fdfris2 ] . in conclusion we have obtained in a very simple constituent quark model all the semileptonic form factors relevant to the transition of b into the low - lying odd and even parity charmed mesons .
the free parameters of the model have been fixed by comparing model predictions with the @xmath73 spectrum and @xmath74 branching ratio .
our numerical results are generally larger than the results of other models
. however , form factors reproduce the scaling laws dictated by the hqet in the limit of infinitely heavy quark masses .
in this appendix we collect the analytical expressions for the @xmath88 and @xmath89 form factors defined in eqs .
( [ e : f1f0 ] ) and ( [ e : gfapam ] ) , respectively . @xmath90 @xmath91 where @xmath67 and @xmath68 ( with @xmath92 the mass of @xmath93quark ) ; @xmath61 and @xmath62 are the masses of the initial and final mesons , respectively .
@xmath94 represents the energy of the final meson .
the angle @xmath58 is defined in section [ s : dcfdf ] after eq .
( [ e : f ] ) .
in this appendix we give the expressions of the form factors appearing in eq .
( [ e : gfapam1p1 ] ) ( @xmath95 transitions ) .
we use the same notations of the previous appendix .
@xmath96 @xmath97 @xmath98 regarding the form factors in eq .
( [ e : gfapam3p1 ] ) we have @xmath99 @xmath100 @xmath101 99 b. aubert _ et al . _
[ babar collaboration ] , phys .
* 90 * ( 2003 ) 242001 [ arxiv : hep - ex/0304021 ] .
d. besson _ et al .
_ [ cleo collaboration ] , phys .
d * 68 * ( 2003 ) 032002 [ arxiv : hep - ex/0305100 ] .
p. krokovny _ et al .
_ [ belle collaboration ] , phys .
* 91 * ( 2003 )
262002 [ arxiv : hep - ex/0308019 ] . k. abe _ et al . _ [ belle collaboration ] , phys .
d * 69 * ( 2004 ) 112002 [ arxiv : hep - ex/0307021 ] .
s. godfrey and n. isgur , phys .
d * 32 * ( 1985 ) 189 ; s. godfrey and r. kokoski , phys .
d * 43 * , 1679 ( 1991 ) .
j. hein _ et al .
_ , phys .
d * 62 * , 074503 ( 2000 ) [ arxiv : hep - ph/0003130 ] ; p. boyle [ ukqcd collaboration ] , nucl .
* 63 * ( 1998 ) 314 [ arxiv : hep - lat/9710036 ] ; r. lewis and r. m. woloshyn , phys .
d * 62 * ( 2000 ) 114507 [ arxiv : hep - lat/0003011 ] . for a review
see for example e. s. swanson , phys .
* 429 * ( 2006 ) 243 [ arxiv : hep - ph/0601110 ] .
n. isgur , d. scora , b. grinstein and m. b. wise , phys .
d * 39 * ( 1989 ) 799 ; d. scora and n. isgur , phys .
d * 52 * ( 1995 ) 2783 [ arxiv : hep - ph/9503486 ] .
a. wambach , nucl .
b * 434 * ( 1995 ) 647 [ arxiv : hep - ph/9312227 ] ; s. veseli and i. dunietz , phys .
d * 54 * ( 1996 ) 6803 [ arxiv : hep - ph/9607293 ] ; s. veseli and m. g. olsson , phys . rev .
d * 54 * ( 1996 ) 886 [ arxiv : hep - ph/9601307 ] ; v. morenas , a. le yaouanc , l. oliver , o. pene and j. c. raynal , phys .
d * 56 * , 5668 ( 1997 ) [ arxiv : hep - ph/9706265 ] .
a. deandrea , n. di bartolomeo , r. gatto , g. nardulli and a. d. polosa , phys .
d * 58 * ( 1998 ) 034004 [ arxiv : hep - ph/9802308 ] ; y. b. dai , c. s. huang and h. y. jin , z. phys .
c * 60 * ( 1993 ) 527 ; d. ebert , r. n. faustov and v. o. galkin , phys .
d * 62 * ( 2000 ) 014032 [ arxiv : hep - ph/9912357 ] .
n. isgur and m. b. wise , phys .
b * 232 * ( 1989 ) 113 ; n. isgur and m. b. wise , phys .
b * 237 * ( 1990 ) 527 ; h. georgi , phys .
b * 240 * ( 1990 ) 447 ; a. f. falk , h. georgi , b. grinstein and m. b. wise , nucl .
b * 343 * ( 1990 ) 1 .
n. isgur and m.b .
wise , phys .
d * 43 * , 819 ( 1991 ) .
w. jaus , phys .
d * 60 * ( 1999 ) 054026 .
h. y. cheng , c. k. chua and c. w. hwang , phys .
d * 69 * ( 2004 ) 074025 [ arxiv : hep - ph/0310359 ] .
m. a. shifman , a. i. vainshtein and v. i. zakharov , nucl .
b147 * , 385 ( 1979 ) ; * b147 * , 448 ( 1979 ) ; v. a. novikov , m. a. shifman , a. i. vainshtein and v. i. zakharov , fortschr.phys . * 32 * , 11 ( 1984 ) .
p. colangelo , g. nardulli , a. a. ovchinnikov and n. paver , phys .
b * 269 * ( 1991 ) 201 ; p. colangelo , g. nardulli and n. paver , phys .
b * 293 * ( 1992 ) 207 ; y. b. dai and m. q. huang , phys .
d * 59 * ( 1999 ) 034018 [ arxiv : hep - ph/9807461 ] ; m. q. huang , c. z. li and y. b. dai , phys .
d * 61 * ( 2000 ) 054010 [ arxiv : hep - ph/9909307 ] ; m. q. huang and y. b. dai , phys .
d * 64 * ( 2001 ) 014034 [ arxiv : hep - ph/0102299 ] .
p. colangelo , f. de fazio , m. ladisa , g. nardulli , p. santorelli and a. tricarico , eur .
j. c * 8 * ( 1999 ) 81 [ arxiv : hep - ph/9809372 ] ; m. ladisa , g. nardulli and p. santorelli , phys .
b * 455 * ( 1999 ) 283 [ arxiv : hep - ph/9903206 ] .
m. de vito and p. santorelli , eur .
j. c * 40 * ( 2005 ) 193 [ arxiv : hep - ph/0412388 ] .
m. a. ivanov and p. santorelli , phys .
b * 456 * ( 1999 ) 248 [ arxiv : hep - ph/9903446 ] ; m. a. ivanov , j. g. korner and p. santorelli , phys .
d * 63 * ( 2001 ) 074010 [ arxiv : hep - ph/0007169 ] ; m. a. ivanov , j. g. korner and p. santorelli , phys .
d * 70 * ( 2004 ) 014005 [ arxiv : hep - ph/0311300 ] .
m. a. ivanov , j. g. korner and p. santorelli , phys .
d * 71 * , 094006 ( 2005 ) [ arxiv : hep - ph/0501051 ] .
m. a. ivanov , j. g. korner and p. santorelli , phys .
d * 73 * , 054024 ( 2006 ) [ arxiv : hep - ph/0602050 ] .
p. cea , p. colangelo , l. cosmai and g. nardulli , phys .
b * 206 * ( 1988 ) 691 ; p. colangelo , g. nardulli and m. pietroni , phys .
d * 43 * ( 1991 ) 3002 .
s. eidelman , et al .
, particle data group , phys .
* b 592 * , 1 ( 2004 ) .
r. a. briere _ et al . _
[ cleo collaboration ] , phys .
* 89 * ( 2002 ) 081803 [ arxiv : hep - ex/0203032 ] ; b. aubert [ babar collaboration ] , arxiv : hep - ex/0408027 .
v. morenas , a. le yaouanc , l. oliver , o. pene and j. c. raynal , phys .
d * 56 * ( 1997 ) 5668 [ arxiv : hep - ph/9706265 ] .
h. y. cheng , c. k. chua and c. w. hwang , phys .
d * 69 * ( 2004 ) 074025 [ arxiv : hep - ph/0310359 ] .
y. b. dai and m. q. huang , phys .
d * 59 * , 034018 ( 1999 ) [ arxiv : hep - ph/9807461 ] . j. d. bjorken , slac - pub-5278 _ invited talk given at les rencontre de la valle daoste , la thuile , italy , mar 18 - 24 , 1990 _ | by using a constituent quark model we compute the form factors relevant to semileptonic transitions of b mesons into low - lying @xmath0-wave charmed mesons .
we evaluate the @xmath1 dependence of these form factors and compare them with other model calculations . the isgur - wise functions @xmath2 and @xmath3 are also obtained in the heavy quark limit of our results . |
quantum interference of electrons in mesoscopic conductors manifests itself in a number of fundamentally important phenomena which can be directly observed in modern experiments .
one of them is the phenomenon of weak localization ( wl ) @xcite . in the absence of interactions
electron wave functions preserve their coherence and , hence , quantum interference remains efficient throughout a large part of the sample making wl a pronounced effect .
interactions between electrons and with other degrees of freedom may limit phase coherence thereby making quantum interference of electrons possible only within a finite length scale @xmath8 .
this so - called electron decoherence length as well as directly related to it decoherence time @xmath9 ( where @xmath7 is diffusion coefficient ) are crucial parameters indicating importance of quantum effects in the system under consideration . at sufficiently high temperatures quantum behavior of electrons in disordered conductors
is usually suppressed due to various types of interactions . however , as temperature gets lower , certain interaction mechanisms either `` freeze out '' or become less efficient in destroying quantum coherence . as a result , both @xmath8 and @xmath10 usually grow with decreasing temperature and quantum effects become progressively more important . should one expect @xmath8 and @xmath10 to diverge in the limit @xmath0 ? while some authors tend to give a positive answer to this question , numerous experiments performed on virtually all kinds of disordered conductors and in all dimensions demonstrate just the opposite , i.e. that at low enough @xmath11 both decoherence length and time _ saturate _ to a constant and do not anymore grow if temperature decreases further .
the list of corresponding structures and experiments , by far incomplete , includes quasi-1d metallic wires @xcite , quasi-1d semiconductors @xcite , carbon nanotubes @xcite , 2d metallic @xcite and semiconductor @xcite films , various 3d disordered metals @xcite and ( 0d ) quantum dots @xcite .
though dimensions and parameters of these systems are different , the low temperature saturation of @xmath10 remains the common feature of all these observations .
is this ubiquitous saturation of @xmath10 an _ intrinsic _ or _ extrinsic _ effect ? if intrinsic , decoherence of electrons at @xmath12 would be a fundamentally important conclusion which would shed a new light on the physical nature of the ground state of disordered conductors as well as on their low temperature transport properties .
while extrinsic saturation of @xmath10 could be caused by a variety of reasons , the choice of intrinsic dephasing mechanisms is , in fact , much more restricted
. there exists , however , at least one mechanism , electron - electron interactions , which remains important down to lowest temperatures and may destroy quantum interference of electrons even at @xmath12 @xcite . in a series of papers
@xcite we offered a theoretical approach that allows to describe electron interference effects in the presence of disorder and electron - electron interactions at any temperature including the most interesting limit @xmath0 .
this formalism extends chakravarty - schmid description @xcite of wl and generalizes feynman - vernon - caldeira - leggett path integral influence functional technique @xcite to fermionic systems with disorder and interactions . with the aid of our approach we have evaluated wl correction to conductance and electron decoherence time in the limit @xmath0 and demonstrated that low temperature saturation of @xmath10 can indeed be caused by electron - electron interactions .
our results allowed for a direct comparison with experiments and a good agreement between our theory and numerous experimental data for @xmath10 in the low temperature limit was found @xcite .
in particular , for quasi-1d wires with thicknesses exceeding the elastic electron mean free path @xmath13 at @xmath0 our theory predicts @xmath14 , where @xmath7 is the diffusion coefficient .
this scaling is indeed observed in experiments for not very strongly disordered wires typically with @xmath15 @xmath16/s ( see sec . 6 for more details ) . on the other hand , for strongly disordered structures with smaller values of @xmath7 this scaling is not anymore fulfilled and , moreover , an opposite trend is observed : @xmath10 was found _ to increase _ with decreasing @xmath7 @xcite .
this trend is _ not _ described by our expressions for @xmath10 @xcite .
another interesting scaling was observed in quantum dots : saturated values @xmath10 were argued @xcite to scale with the dot dwell times @xmath3 as @xmath17 .
our theory @xcite can not be directly used in order to explain the latter scaling either . in order to attempt to reconcile all these observations within one approach
it is necessary to develop a unified theoretical description which would cover essentially all types of disordered conductors .
it is worth pointing out that the technique @xcite is formally an exact procedure which should cover all situations .
however , for some structures , such as , e.g. , quantum dots and granular metals , it can be rather difficult to directly evaluate the wl correction within this technique for the following reasons .
first of all , our description in terms of quasiclassical electron trajectories may become insufficient in the above cases , and electron scattering on disorder should be treated on more general footing .
another purely technical point is averaging over disorder .
in our approach @xcite it is convenient to postpone disorder averaging until the last stage of the calculation . in some cases like ones studied below it might be , on the contrary , more appropriate to perform disorder averaging already in the beginning of the whole consideration .
in addition , it is desirable to deal with the model which would embrace various types of conductors with well defined properties both in the long and short wavelength limits . this feature will help to construct a fully self - contained theory free of any divergencies and additional cutoff parameters .
recently @xcite we made a first step towards this unified theory .
namely , we adopted a model for a disordered conductor consisting of an array of ( metallic ) quantum dots connected via junctions ( scatterers ) with arbitrary transmission distribution of their conducting channels .
this model allows to easily crossover between the limits of a granular metal and that with point - like impurities and to treat spatially restricted and spatially extended conductors within the same theoretical framework , as desired . within this model in ref .
@xcite we analyzed wl corrections to conductance merely for non - interacting electrons and included interaction effects by introducing the electron dephasing time @xmath10 just as a phenomenological parameter .
systematic analysis of the effect of electron - electron interactions on weak localization within this formalism will be developed in this paper .
this approach will allow to microscopically evaluate @xmath10 for all the structures under consideration .
the structure of our paper is as follows . in sec .
2 we will discuss qualitative arguments illustrating the role of scattering and interactions in electron dephasing . in sec .
3 we introduce our model of an array of quantum dots and outline a general theoretical framework which is then employed in sec . 4 and 5 for rigorous calculations of the wl correction to conductance and electron decoherence time in the presence of electron - electron interactions . a detailed comparison of our results with numerous experiments performed in various disordered conductors is carried out in sec .
. a brief summary of our main results and conclusions is presented in sec .
before turning to a detailed calculation it is instructive to discuss a simple qualitative picture demonstrating under which conditions electron dephasing by interaction is expected to occur . and @xmath18 are assumed to drop only across the barriers and not inside the dot.,width=200 ] consider first the simplest system of two scatterers separated by a cavity ( quantum dot , fig .
1 ) the wl correction to conductance of a disordered system @xmath19 is known to arise from interference of pairs of time - reversed electron paths @xcite . in the absence of interactions for a single quantum dot of fig .
1 this correction is evaluated in a general form @xcite .
the effect of electron - electron interactions can be described in terms of fluctuating voltages .
let us assume that the voltage can drop only across the barriers and consider two time - reversed electron paths which cross the left barrier ( with fluctuating voltage @xmath20 ) twice at times @xmath21 and @xmath22 as shown in fig .
it is easy to see that the voltage - dependent random phase factor @xmath23 acquired by the electron wave function @xmath24 along any path turns out to be exactly the same as that for its time - reversed counterpart .
hence , in the product @xmath25 these random phases cancel each other and quantum coherence of electrons remains fully preserved .
this implies that for the system of fig .
1a fluctuating voltages ( which can mediate electron - electron interactions ) _ do not cause any dephasing_. this qualitative conclusion can be verified by means of more rigorous considerations .
for instance , it was demonstrated @xcite that the scattering matrix of the system remains unitary in the presence of electron - electron interactions , which implies that the only effect of such interactions is transmission renormalization but not electron decoherence . in ref .
@xcite a similar conclusion was reached by directly evaluating the wl correction to the system conductance .
thus , for the system of two scatterers of fig .
1a electron - electron interactions can only yield energy dependent ( logarithmic at sufficiently low energies ) renormalization of the dot channel transmissions @xcite but not electron dephasing . ,
@xmath26 and @xmath18 are again assumed to drop only across the barriers.,width=283 ] let us now add one more scatterer and consider the system of two quantum dots depicted in fig .
we again assume that fluctuating voltages are concentrated at the barriers and not inside the cavities .
the phase factor accumulated along the path ( see fig .
2 ) which crosses the central barrier twice ( at times @xmath21 and @xmath27 ) and returns to the initial point ( at a time @xmath22 ) is @xmath28}$ ] , where @xmath29 is the fluctuating voltage across the central barrier . similarly , the phase factor picked up along the time - reversed path reads @xmath30}$ ] .
hence , the overall phase factor acquired by the product @xmath25 for a pair of time - reversed paths is @xmath31 , where @xmath32 averaging over phase fluctuations , which for simplicity are assumed gaussian , we obtain @xmath33 where we defined the phase correlation function @xmath34 should this function grow with time the electron phase coherence decays and , hence , @xmath19 has to be suppressed below its non - interacting value due to interaction - induced electron decoherence .
the above arguments are , of course , not specific to systems with three barriers only .
they can also be applied to any system with larger number of scatterers , i.e. virtually to any disordered conductor where
exactly for the same reasons
one also expects non - vanishing interaction - induced electron decoherence at any temperature including @xmath12 . in the next sections we will develop a quantitative theory which will confirm and extend our qualitative physical picture .
we are going to give a complete quantum mechanical analysis of the problem which fully accounts for fermi statistics of electrons and treats electron - electron interactions in terms quantum fields produced internally by fluctuating electrons . below we will non - perturbatively evaluate wl correction @xmath19 for arrays of metallic quantum dots in the presence of electron - electron interactions which will be shown to reduce phase coherence of electrons at any temperature down to @xmath12 .
quantum dots connected via @xmath35 barriers . , width=283 ] let us consider a 1d array of @xmath36 quantum dots connected in series chaotic quantum dots ( fig .
each quantum dot is characterized by its own mean level spacing @xmath37 .
adjacent quantum dots are connected to each other via barriers which can scatter electrons .
the first and the last dot are also connected to the leads ( with total resistance @xmath38 ) , i.e. altogether we have @xmath35 scatterers in our system .
each such scatterer is described by a set of transmissions of its conducting channels @xmath39 ( here @xmath40 labels the channels and @xmath41 labels the scatterers ) .
below we will focus our attention on the case of metallic quantum dots with the level spacing @xmath37 being the lowest energy parameter in the problem .
the system of fig .
3 will be described by the hamiltonian @xmath42 where @xmath43 is the capacitance matrix of the array , @xmath44 is the operator of electric potential on @xmath45th quantum dot , @xmath46 are the hamiltonians of the left and right leads , @xmath47 are the electric potentials of the leads fixed by the external voltage source , @xmath48 is the hamiltonian of @xmath41-th quantum dot and @xmath49 is a random matrix which belongs to the orthogonal ensemble .
finally , electron transfer between adjacent @xmath50-th and @xmath41-th quantum dots will be described by the hamiltonian @xmath51.\ ] ] here the integration runs over the junction area .
note that in ref .
@xcite we have already applied the model of fig . 3 in order to analyze wl effects in the absence of electron - electron interactions . in that paper
we have used the scattering matrix formalism combined with the non - linear @xmath52-model . in order to incorporate interaction effects into our consideration
it will be convenient for us to describe inter - dot electron transfer within the tunneling hamiltonian approach , as specified above .
we would like to emphasize that this choice is only a matter of technical convenience , since the approaches based on the tunneling hamiltonian and on the scattering matrix are fully equivalent to each other .
for the sake of completeness , let us briefly remind the reader the relation between these two approaches .
consider , e.g. , the @xmath41-th barrier between the @xmath50-th and @xmath41-th quantum dots and define the matrix elements @xmath53 between the @xmath54th wave function in the @xmath50-th dot and @xmath55th wave function in the @xmath41-th dot .
electron transfer across this barrier can then be described by the set of eigenvalues of this matrix @xmath56 .
these eigenvalues are related to the barrier channel transmissions @xmath39 as ( see , e.g. , @xcite ) @xmath57 this equation allows to keep track of the relation between two approaches at every stage of the calculation . to proceed
we will make use of the path integral keldysh technique .
a brief sketch of this approach is outlined below .
the time evolution of the density matrix of our system is described by the standard equation @xmath58 where @xmath59 is given by eq .
( [ h ] ) .
let us express the operators @xmath60 and @xmath61 via path integrals over the fluctuating electric potentials @xmath62 defined respectively on the forward and backward parts of the keldysh contour : @xmath63\right\ } , \nonumber\\ e^{i\hat ht}&=&\int dv_n^b\ ; \tilde{\rm t}\,\exp\left\{i\int_0^t dt'\hat h\left[v_n^b(t')\right]\right\}.\end{aligned}\ ] ] here @xmath64 ( @xmath65 ) stands for the time ordered ( anti - ordered ) exponent and the hamiltonians @xmath66 $ ] , @xmath67 $ ] are obtained from the original hamiltonian ( [ h ] ) if one replaces the operators @xmath68 respectively by the fluctuating voltages @xmath69 and @xmath70 .
the central object of our analysis is the effective action defined as @xmath71&=&\ln\left ( { \rm tr } \left [ { \rm t}\,\exp\left\{-i\int_0^t dt'\hat h\left[v_n^f(t')\right]\right\ } \right.\right . \nonumber\\ & & \times\ , \left.\left .
\hat\rho_0 \tilde{\rm t}\,\exp\left\{i\int_0^t dt'\hat h\left[v_n^b(t')\right]\right\ } \right]\right).\end{aligned}\ ] ] since the operators @xmath66 $ ] , @xmath67 $ ] are quadratic in the electron creation and annihilation operators , it is possible to rewrite the action in the form @xmath72 , \label{s}\ ] ] where @xmath73 is the inverse keldysh green function of the system .
the operator @xmath73 has the following matrix structure @xmath74 here each quantum dot as well as each of the two leads is represented by a @xmath75 diagonal block @xmath76 while barriers are described by off - diagonal blocks @xmath77 , which have the form @xmath78 below we will employ these general expressions in specific situations of two quantum dots and of an array with a large number of quantum dots
. the latter system will also serve as a model for spatially extended diffusive conductors .
let us first consider the system of three scatterers @xmath79 ( two quantum dots ) .
this system is important both in the context of electron dephasing in quantum dots @xcite and as a relatively simple example which illustrates all significant features of the effect of electron - electron interactions on wl in more complicated systems . here
we will treat interaction effects practically _ without any approximations_. for the sake of simplicity we will assume that the central ( i.e. second ) barrier is a tunnel junction with @xmath80 for all its transmission channels .
this assumption allows one to expand the action ( [ s ] ) in powers of the parameter @xmath81 .
we will also assume that dimensionless conductance of the central barrier @xmath82 is much lower than those of the first and third barriers which , in turn , strongly exceed unity , i.e , @xmath83 .
the effective expansion parameter in this case is @xmath84 .
then the wl correction to the conductance in the absence of interactions takes a particularly simple form @xcite .
e.g. , for fully open outer barriers we have @xmath85 the first term in eq .
( [ g ] ) defines the contribution of the first and third scatterers , while the term @xmath86 comes from the central junction .
experimentally one can access @xmath86 attaching additional voltage probes to the quantum dots .
the aim of our subsequent analysis is to demonstrate how the result ( [ g0 ] ) is modified in the presence of electron - electron interactions . setting @xmath79 in eq .
( [ h ] ) and defining the fluctuating phases across the barriers @xmath87 $ ] we perform the gauge transformation which yields the new expressions for the blocks @xmath88 and @xmath77 : @xmath89 @xmath90 since the central barrier transmission is small we can expand the action ( [ s ] ) in powers of the parameter @xmath91 .
proceeding with this expansion up to the fourth order we get @xmath92+is_2^{(2)}+is_2^{(4)}.\ ] ] here @xmath93=is_c+is_{\rm ext } + 2\,{\rm tr}\,\ln[\check g^{-1}]\big|_{t_2(\bm{r})=0 } \label{ss}\ ] ] is the effective action at zero transmission of the second barrier which also includes terms describing capacitances ( @xmath94 ) and the external circuit ( @xmath95 ) , the term @xmath96 is the standard ambegaokar - eckern - schn ( aes ) action @xcite and @xmath97 contains information which will allow us to evaluate the wl correction to the system conductance .
the corresponding expression reads @xmath98 here we use the convention @xmath99 , @xmath100 , @xmath101 , and @xmath102 are @xmath75 matrix green - keldysh functions in the first and the second quantum dots ( @xmath103 ) : @xmath104 here we will set the green functions @xmath105 and @xmath106 equal to their equilibrium values @xmath107 , where @xmath108 are retarded and advanced green functions , @xmath109 @xmath110 is the fermi function and @xmath111 .
this choice is sufficiently accurate for the problem in question .
we will return to this point towards the end of this section .
our next step amounts to averaging the products of retarded and advanced propagators in the action ( [ s2 ] ) over disorder in each of the two dots separately .
this averaging can be accomplished , e.g. , with the aid of the non - linear @xmath52-model or by other means .
for the first ( left ) dot we obtain ( cf .
, e.g. , @xcite ) @xmath112 where @xmath113 is the volume of the first ( left ) quantum dot , @xmath114 , @xmath115 and @xmath116 are respectively the density of states , the diffuson and the cooperon in the first ( left ) dot , @xmath117 , @xmath118 and @xmath13 are respectively the fermi wave vector and elastic mean free path .
the same averaging procedure applies to the second ( right ) dot . in the absence of magnetic field both the diffuson and the cooperon
satisfy the same diffusion equation @xmath119 with the appropriate boundary conditions at the contacts .
averaging the action ( [ s2 ] ) over disorder we will collect only the terms proportional to the product of the cooperons @xmath120 and @xmath121 and ignore all other contributions which are unimportant for the problem in question . in this way we arrive at the action
@xmath122 which describes weak localization effects in our system . ignoring for simplicity the coordinate dependence of the cooperons @xmath123 inside the dots , we obtain @xmath124 } \sin\frac{\varphi^-(t_1)}{2 } \nonumber\\ & & \times\ , \left [ h(\tau_1-t_2 ) e^{-i\frac{\varphi^-(t_2)}{2 } } + f(\tau_1-t_2 ) e^{i\frac{\varphi^-(t_2)}{2 } } \right ] \nonumber\\ & & \times\ , \left [ h(\tau_2-t_3)e^{i\frac{\varphi^-(t_3)}{2}}f(t_1+t_3-t_4-\tau_1 ) \right .
\nonumber\\ & & -\,\left .
f(\tau_2-t_3 ) e^{-i\frac{\varphi^-(t_3)}{2}}h(t_1+t_3-t_4-\tau_1 ) \right ] \nonumber\\ & & \times\ , \left [ e^{-i\frac{\varphi^-(t_4)}{2}}f(-t_1+t_2+t_4-\tau_2 ) + e^{i\frac{\varphi^-(t_4)}{2}}\right . \nonumber\\
& & \times\ , h(-t_1+t_2+t_4-\tau_2)\bigg ] + \,\big\ { 1\leftrightarrow 2 , \varphi^\pm\to -\varphi^\pm\big\}. \label{swl}\end{aligned}\ ] ] here we defined mean level spacing @xmath125 for both dots , introduced `` classical '' @xmath126 and `` quantum '' @xmath127 phases and made use of the fourier transforms of the fermi function @xmath128 .
the resulting action ( [ swl ] ) fully accounts for the effects of electron - electron interactions on wl via the fluctuating phases @xmath129 . in order to find the wl correction to the current across the central barrier @xmath130 we employ the following general formula @xmath131}{\delta \varphi^-}\ , e^{is_{s}+is^{(2)}_{2}}. \label{i}\ ] ]
the task at hand is to combine eqs .
( [ swl ] ) and ( [ i ] ) and to average over the phases @xmath129 by evaluating the path integral in ( [ i ] ) .
the contributions @xmath132 and @xmath133 in ( [ ss ] ) are quadratic in the fluctuating phases provided an external circuit consists of linear elements .
the remaining contribution to @xmath134 ( [ ss ] ) which describes transfer of electrons across the first and third barriers as well as aes action of the second barrier @xmath135 are in general non - gaussian . however , in the interesting limit @xmath136 phase fluctuations can be considered small down to exponentially low energies @xcite in which case it suffices to expand both contributions @xmath137\big|_{t_2(\bm{r})=0}$ ] and @xmath135 up to the second order @xmath129 .
furthermore , gaussian approximation for the first of these terms becomes essentially exact in the limit of fully open outer barriers with @xmath136 @xcite .
we conclude that the integral ( [ i ] ) remains gaussian at all relevant energies and can easily be performed .
after straightforward algebra we arrive at the final result @xmath138 \nonumber\\ & & -\,\big\ { v\to -v \big\}. \label{iwl1}\end{aligned}\ ] ] here @xmath26 is the average voltage across the central barrier , @xmath139 are the fourier transforms of the cooperons @xmath140 and the functions @xmath141 ( @xmath142 ) are defined as @xmath143 } p_{j}(t_i),\end{aligned}\ ] ] where @xmath144{q}_{j}(t_1,t_2,t_3 ) , \label{calp}\ ] ] @xmath145 and @xmath146 coincides with the phase correlation function ( [ f ] )
. the terms @xmath147 read @xmath148 } \nonumber\\ & & \times\ , \big\{2e^{i\left[k(|t_1+t_2+t_3|)+k(t_1+t_3)+k(t_1+t_2)\right ] } \nonumber\\ & & -\,e^{i\left[k(t_1+t_2+t_3)+k(|t_1+t_3|)+k(|t_1+t_2|)\right ] } \big\ } , \nonumber\\ & & { q}_2=e^{i\left[k(|t_1+t_2+t_3|)-k(t_2)-k(|t_3|)\right ] } \nonumber\\ & & \times\ , e^{i\left[k(t_1+t_3)-k(|t_1+t_2|)-k(t_3-t_2)\right ] } , \nonumber\\ & & { q}_3= e^{i\left[k(t_1+t_2+t_3)-k(|t_2|)-k(|t_3|)\right ] } \nonumber\\ & & \times\ , e^{-i\left[k(t_1+t_3)+k(t_1+t_2)-k(|t_3-t_2|)\right ] } , \label{pj}\end{aligned}\ ] ] where @xmath149 \right\rangle$ ] is the response function .
( [ iwl1])-([pj ] ) fully determine wl correction to the current in our system . the non - interacting result @xcite is reproduced by the first line of eq .
( [ iwl1 ] ) , while its second and third lines _ exactly _ account for the effect of interactions .
we observe that the whole effect of electron - electron interactions is encoded in two different correlators of fluctuating phases @xmath150 and @xmath151 defined as @xmath152\frac{1-\cos\omega t}{\omega^2 } , \label{fff}\\ k(t)&=&e^2\int\frac{d\omega}{2\pi}\,{\rm re}\big[z(\omega)\big ] \frac{\sin\omega t}{\omega } , \label{kkk}\end{aligned}\ ] ] where @xmath153 is an effective impedance `` seen '' by the central barrier .
let us recall that these correlation functions play a central in the so - called @xmath154-theory of coulomb blockade in tunnel barriers @xcite .
it turns out that exactly the same correlators also describe the effect of electron - electron interactions on weak localization .
below we will demonstrate that the phase correlation function ( [ fff ] ) is responsible for electron dephasing while the response function ( [ kkk ] ) describes the coulomb blockade correction to wl .
let us now turn to the analysis of the above general results . to begin with , we notice that both functions ( [ fff ] ) and ( [ kkk ] ) are purely real and , hence , @xmath155 .
furthermore , at sufficiently long times @xmath156 ( where @xmath157 is an effective @xmath158-time of our system to be defined later ) we obtain @xmath159 where @xmath160 , @xmath161 and @xmath162 is the euler constant .
the correlation function @xmath150 grows with time @xcite at any temperature including @xmath12 .
in contrast , the response function @xmath151 always remains small in the limit @xmath163 considered here .
hence , the combination ( [ calf ] ) should be fully kept in the exponent of ( [ calp ] ) while the correlator @xmath151 can be safely ignored in the leading order in @xmath164 .
then all @xmath165 , the fermi function @xmath110 drops out from the result and we get @xmath166 , where @xmath167 where @xmath168 identifying @xmath169 and @xmath170 in eq .
( [ t2t3 ] ) we observe that the exponent @xmath171 exactly coincides with the expression ( [ phase ] ) derived from simple considerations involving electrons propagating along time - reversed paths in an external fluctuating field .
thus , we arrive at an important conclusion : in the leading order in @xmath164 the wl correction @xmath19 is affected by electron - electron interactions via dephasing produced _ only _ by the `` classical '' component @xmath172 of the fluctuating field which mediates such interactions .
this effect is described only by the phase correlation function @xmath150 ( [ fff ] ) . at the same time
, fluctuations of the `` quantum '' field @xmath173 turn out to be irrelevant for dephasing and may only cause a ( weak ) coulomb blockade correction described by the response function @xmath151 ( [ kkk ] ) .
this latter effect will be analyzed in the next subsection . in order to simplify our consideration
let us assume that two quantum dots are identical , i.e. @xmath174 and @xmath175 , and set @xmath176 .
in this case the cooperons are @xmath177 , where @xmath178 is the dwell time for each of the two dots and @xmath179 is the dephasing time due to the external magnetic field @xmath180 which can be applied to the system .
an effective impedance @xmath181 takes the form @xmath182 , \label{imped}\ ] ] where @xmath183 , @xmath184 , @xmath185 and @xmath186 , @xmath187 and @xmath188 are the capacitances of respectively left(right ) barriers , the central junction and the gate electrode .
substituting the cooperons @xmath189 into eq .
( [ gwlf ] ) we arrive at the final expression for the wl correction @xmath190 in the presence of electron - electron interactions @xmath191 with a good accuracy the double time integral in eq .
( [ gwlf ] ) can be replaced by a single one , in which case the magnetoconductance @xmath19 can be expressed in a much simpler form @xmath192 the result ( [ gwlf ] ) for @xmath193 is plotted in fig .
we observe that electron - electron interactions always suppress @xmath190 below its non - interacting value ( [ g0 ] ) .
this is a direct consequence of interaction - induced electron dephasing described by the correlation function @xmath150 . in the presence of electron - electron interactions for @xmath194 .
, width=264 ] let us define @xmath195 and consider the limit of metallic dots @xmath196 . at @xmath197 and for @xmath198 we find @xmath199
whereas at @xmath200 the wl correction saturates to @xmath201 let us compare the magnitude of the wl correction ( [ t0 ] ) to that of the leading contribution to the system conductance evaluated in the presence of electron - electron interactions .
this contribution reads @xcite @xmath202 where @xmath203 is the gamma - function . setting @xmath204 in the above expression
, we observe that the wl correction ( [ t0 ] ) remains much smaller than eq .
( [ main ] ) by the parameter @xmath205 . in the presence of interactions for @xmath194 .
, width=245 ] let us phenomenologically define the electron decoherence time @xmath10 by taking the cooperons ( for @xmath206 ) in the form @xmath207 .
this definition obviously yields @xcite @xmath208 . resolving this equation with respect to @xmath10 , we obtain @xmath209 substituting the result ( [ tt ] ) into eq .
( [ tphi ] ) at sufficiently high temperatures @xmath197 and in the limit @xmath210 we obtain @xmath211 combining eqs .
( [ t0 ] ) and ( [ tphi ] ) at lower temperatures @xmath200 and for @xmath212 we find @xmath213 in the limit of large @xmath214 this expression yields @xmath215 the results for the electron decoherence time @xmath216 ( [ tphi ] ) are also plotted in fig . 5 for different values of @xmath2 .
we observe that at higher temperatures @xmath216 shows a @xmath2-dependent power law dependence on @xmath11 and eventually saturates to the value ( [ tauphi ] ) at lower @xmath11 .
this is exactly the behavior observed in a number of experiments with quantum dots @xcite .
we will postpone a detailed comparison between our theory and experiments to sec . 6 . let us now analyze the role of the response function @xmath151 which was disregarded in the previous subsection . for this purpose
let us expand the general expression for the current ( [ iwl1 ] ) to the first order in the interaction , i.e. to the first order in both @xmath150 and @xmath151 .
we get @xmath217 where @xmath218 , \label{wlf}\end{aligned}\ ] ] @xmath219 \nonumber\\ & & + \ , i\,{\rm im}\ , z(\omega)\ ; \tilde c_1(\omega)\tilde c_2(\omega ) \big\}. \label{wlk}\end{aligned}\ ] ] here we defined the function @xmath220 the above expressions allow to make several important observations . to begin with , we note that the term @xmath221 is linear in the bias voltage while @xmath222 is non - linear in @xmath26 .
the physical meaning of these two terms is entirely different .
while , as we already know , the correction @xmath221 describes electron dephasing , the term @xmath222 is nothing but the _ coulomb blockade correction _ to @xmath19 ( [ gwlf ] ) .
the latter conclusion is supported , for instance , by the observation that at large voltages @xmath222 tends to a constant offset value , which is a typical sign of coulomb blockade . just for the sake of illustration , let us for a moment assume that the environment remains gaussian at all energies . in this case
@xmath223 diverges at @xmath224 indicating the importance of higher order perturbative terms in the low energy limit . making use of analytical properties of the functions
@xmath225 one can exactly sum up diverging perturbative series to all orders and quite generally demonstrate that in the limit of zero voltage and temperature the wl correction tends to zero , @xmath226 , implying complete coulomb suppression of weak localization in this limit .
this strong coulomb blockade of wl can be recovered only if one fully accounts for the response function @xmath151 .
we also note that at large conductances weak coulomb blockade may turn into strong one only at exponentially small energies @xcite typically well below the inverse electron dwell time @xmath227 , otherwise coulomb suppression of wl remains weak . turning back to the first order terms ( [ wlf ] ) and ( [ wlk ] )
, we observe that in the linear response regime @xmath228 almost all contributions from @xmath221 and @xmath222 cancel each other exactly in the limit @xmath0 .
this ( partial ) cancellation of the so - called `` coth''([wlf ] ) and `` tanh '' ( [ wlk ] ) terms is a general feature of the first order perturbation theory .
for instance , in diffusive conductors it has been observed by various authors @xcite .
this cancellation is sometimes interpreted as a `` proof '' of zero dephasing of interacting electrons at @xmath12 _ assuming _ that the same `` coth - tanh '' combination should occur in every order of the perturbation theory @xcite .
our exact result ( [ iwl1 ] ) demonstrates , that this assumption is incorrect .
partial cancellation between dephasing and coulomb blockade terms occurs only in the first order and is of little importance for the issue of electron dephasing .
actually , the combination @xmath229 enters only in the first order , while the exact expression ( [ iwl1 ] ) depends on @xmath110 and @xmath230 and does not contain the combination @xmath231 at all .
this observation is fully consistent with our general result for the wl correction @xcite expressed in terms of the matrix elements of the operators @xmath232 and @xmath233 , where @xmath234 is the electron density matrix .
in fact , the analysis presented above is a way to explicitly evaluate these matrix elements for a particular case of two quantum dots .
finally , let us estimate the corrections @xmath221 and @xmath222 . for max@xmath235
we obtain @xmath236 , \label{dif}\\ & & \frac{di^k_{wl}}{dv}=\left\ { \begin{array}{cc } \frac{20}{3}\frac{e^2g_2 ^ 2}{g^3 t\tau_d } , & ev\ll t , \\ \frac{6}{\pi}\frac{e^2g_2 ^ 2}{g^3(ev\tau_d)^2 } , & ev \gg t ,
\end{array } \right . \label{dik}\end{aligned}\ ] ] i.e. at @xmath237 the ratio of the terms ( [ dik ] ) and ( [ dif ] ) is of order @xmath238 and it decreases further with increasing temperature as @xmath239 .
this estimate demonstrates that @xmath240 can be ignored as compared to the main contribution @xmath241 .
this conclusion remains valid also beyond the first order perturbation theory .
indeed , keeping the function @xmath150 in the exponent of eq .
( [ iwl1 ] ) and expanding this result only in powers of @xmath151 , at @xmath242 one finds @xmath243 thus supporting our conclusion about relative unimportance of the coulomb blockade correction to @xmath19 for the problem in question . for clarity ,
let us again summarize all the approximations used in the above analysis : @xmath244 throughout our calculation we have used the equilibrium form of the distribution function matrices @xmath245 ( [ distr ] ) which effectively implies neglecting the dependence of the green functions @xmath246 on the phases @xmath247 and @xmath248 .
this is accurate except at energies well below the inverse dwell time @xmath227 .
the wl correction ( [ gwlf1])-([t2t3 ] ) saturates at energies @xmath249 and , hence , is totally insensitive to this approximation .
the coulomb term @xmath250 should be treated somewhat more carefully in the limit @xmath251 . for @xmath176
this treatment also yields _ saturation _ of the coulomb correction to wl at energies of order @xmath227 , exactly as in the case of the interaction correction to the drude conductance @xcite .
in other words , at all @xmath252 the term @xmath250 remains constant of order @xmath253 .
thus , the above approximation is completely unimportant for any of our conclusions .
@xmath254 we have assumed @xmath136 and @xmath255 .
the first inequality just implies that our structure is metallic , while the second one is only a matter of technical convenience .
obviously , our results for @xmath10 remain qualitatively the same also for @xmath256 .
@xmath257 for completeness , let us also mention that in the final expressions for @xmath19 , e.g. , in eqs .
( [ tt ] ) and ( [ t0 ] ) , we have disregarded small renormalization of the conductance @xmath2 due to coulomb effects @xcite . this renormalization is strictly zero for fully open outer barriers , otherwise it can be trivially included into our final results .
thus , our treatment presented in this section does not contain any approximations which could influence our main results and conclusions . in the limit @xmath254 our general results for the wl correction to the @xmath258 curve in the presence of electron - electron interactions ( [ iwl1])-([pj ] ) are exact .
one of the main conclusions reached in the previous section is that the electron decoherence time is fully determined by fluctuations of the phase fields @xmath172 ( and the correlation function @xmath150 ) , whereas the phases @xmath173 ( and the response function @xmath151 ) are irrelevant for @xmath10 causing only a weak coulomb correction to @xmath19 .
this conclusion is general being independent of a number of scatterers in our system .
note that exactly the same conclusion was already reached in the case of diffusive metals by means of a different approach @xcite .
thus , in order to evaluate the decoherence time for interacting electrons in arrays of quantum dots it is sufficient to account for the fluctuating fields @xmath259 totally ignoring the fields @xmath260 .
the corresponding calculation is presented below .
we will specifically address the example of 1d arrays and argue that the final result for the zero temperature decoherence time @xmath261 actually does not depend on the dimensionality of the array .
we will also demonstrate how our present results for @xmath262 match with those obtained previously for weakly disordered metals @xcite .
let us consider a 1d array of @xmath36 quantum dots by @xmath35 identical barriers as shown in fig .
3 . for simplicity
, we will stick to the case of identical barriers ( with dimensionless conductance @xmath263 and fano factor @xmath264 ) and identical quantum dots ( with mean level spacing @xmath5 and dwell time @xmath265 ) .
the wl correction to the system conductance has the form @xcite
@xmath266 \nonumber\\ & & + \,(1-\beta)\big[c_{nn}(t)+c_{n-1,n-1}(t)\big ] \big\}. \label{gwl}\end{aligned}\ ] ] the cooperon @xmath267 is determined from a discrete version of the diffusion equation . for non - interacting electrons and in the absence of the magnetic field this equation reads @xmath268 the boundary conditions for this equation are @xmath269 as long as the index @xmath41 or @xmath270 belongs to one of the bulk electrode . the solution of eq .
( [ eqc ] ) with these boundary conditions can easily be obtained .
we have @xmath271 this solution can be represented in the form @xmath272 , where @xmath273 in the limit of large @xmath35 the term @xmath274 can be safely ignored and we obtain @xmath275 .
let us express the contribution @xmath276 as a sum over the integer valued paths @xmath277 , which start in the @xmath55th dot and end in the @xmath45th one ( i.e. @xmath278 , @xmath279 ) jumping from one dot to another at times @xmath280 .
this expression can be recovered if one expands eq .
( [ cbulk ] ) in powers of @xmath281 $ ] with subsequent summation over @xmath282 in every order of this expansion .
including additional phase factors acquired by electrons in the presence of the fluctuating fields @xmath283 , we obtain @xmath284\right\}.\end{aligned}\ ] ] averaging over gaussian fluctuations of voltages @xmath259 and utilizing the symmetry of the voltage correlator @xmath285 , we get @xmath286 \bigg\}. \label{c2}\end{aligned}\ ] ] the correlator of voltages can be obtained , e.g. , with the aid of the @xmath52-model approach employed in ref .
integrating over gaussian fluctuations of the @xmath287-fields one arrives at the quadratic action for the fluctuating fields @xmath259 which has the form @xmath288v^+_q(\omega)v^-_q(-\omega ) \nonumber\\ & & -\,\frac{2}{n}\sum_{q=1}^n\int\frac{d\omega}{2\pi } \frac{g\tau_d^2e^2}{\pi } \frac{\left(1-\cos\frac{\pi q}{n}\right)\omega\coth\frac{\omega}{2 t } } { \omega^2\tau_d^2+\left(1-\cos\frac{\pi q}{n}\right)^2 } \nonumber\\ & & \times\ , v^-_q(\omega)v^-_q(-\omega ) .
\label{s11}\end{aligned}\ ] ] here we defined @xmath289 the action ( [ s11 ] ) determines the expressions for both correlators @xmath290 ( @xmath291-function ) and @xmath292 ( @xmath293-function ) responsible respectively for decoherence and coulomb blockade correction to wl . since our aim is to describe electron decoherence , only the first out of these two correlation functions is of importance for us here .
it reads @xmath294 in the continuous limit @xmath295 and for sufficiently low frequencies @xmath296 both correlators @xmath297 and @xmath298 defined by eq .
( [ s11 ] ) reduce to those of a diffusive metal @xcite . to proceed
let us consider diffusive paths @xmath277 , in which case one has @xmath299 where @xmath300 is the diffuson .
for @xmath206 it exactly coincides with the cooperon for non - interacting electrons ( [ c0 ] ) , @xmath301 , i.e. @xmath302 substituting eq .
( [ vvd ] ) into ( [ c2 ] ) , we obtain @xmath303 where @xmath304 \label{f3}\end{aligned}\ ] ] is the function which controls the cooperon decay in time , i.e. describes electron decoherence for our 1d array of quantum dots .
the wl correction @xmath19 in the presence of electron - electron interactions is recovered by substituting the result ( [ c3 ] ) into eq .
( [ gwl ] ) . note that in the limit of very large @xmath35 and for @xmath305 eq .
( [ f3 ] ) ( combined with ( [ vv ] ) and ( [ dc0 ] ) ) reduces to eq .
( 22 ) of ref .
@xcite derived for metallic conductors by means of a different technique . replacing the sum in ( [ f3 ] ) by momentum integration , in the same limit one arrives at the result which matches with eq .
( 23 ) of @xcite provided one sets the charging energy terms equal to zero .
since the behavior of the latter formula was already analyzed in details earlier @xcite , there is no need to repeat this analysis here .
the dephasing time @xmath10 can be extracted from the equation @xmath306 . from eq .
( [ f3 ] ) with a good accuracy we obtain @xmath307 combining this formula with eqs .
( [ vv ] ) and ( [ dc0 ] ) , in the most interesting limit @xmath308 and for @xmath309 we find @xmath310 which yields @xmath311 where @xmath312 for @xmath313 and @xmath314 in the opposite case @xmath315 .
we observe that apart from an unimportant numerical factor of order one the result for @xmath261 ( [ t1 ] ) derived for 1d array of quantum dots coincides with the exact result ( [ tauphi ] ) derived in the previous section for the case of two quantum dots .
thus , we arrive at an important conclusion : in the low temperature limit the electron decoherence time @xmath316 is practically independent of the number of scatterers in the conductor ( provided the latter number exceeds two ) and is essentially determined by _ local _ properties of the system . in order to determine the dephasing length @xmath317
let us define the diffusion coefficient @xmath318 where @xmath319 is the average dot size .
combining eqs .
( [ t1 ] ) and ( [ ddd ] ) , at @xmath12 we obtain @xmath320 at non - zero @xmath11 thermal fluctuations provide an additional contribution to the dephasing rate @xmath321 . again substituting eqs .
( [ vv ] ) and ( [ dc0 ] ) into ( [ tau11 ] ) , we get @xmath322 where @xmath323 is the number of quantum dots within the length @xmath8 .
we observe that for sufficiently small @xmath324 ( but still @xmath325 ) the dephasing rate increases linearly both with temperature and with the number @xmath35 . at larger @xmath326}$ ] and/or at high enough temperatures @xmath327 becomes smaller than @xmath35 and eq .
( [ taut ] ) for @xmath10 should be resolved self - consistently . in this case
we obtain @xmath328 which matches with the aak result @xcite .
( [ taut ] ) also allows to estimate the temperature @xmath329 $ ] at which the crossover to the temperature - independent regime ( [ t1 ] ) occurs .
we find @xmath330}{2\pi n\tau_d},\;\;\ ; n\lesssim \sqrt{\frac{g}{\ln[{4\tilde e_c}/{\delta } ] } } , \nonumber\\ & & t^*\simeq \frac{3\ln^{3/2}[{4\tilde e_c}/{\delta}]}{2\pi \tau_d\sqrt{g}},\;\;\ ; n\gtrsim \sqrt{\frac{g}{\ln[{4\tilde e_c}/{\delta}]}}.\end{aligned}\ ] ] the above analysis and conclusions can be generalized further to the case 2d and 3d structures .
this generalization is absolutely straightforward ( see , e.g. @xcite ) and therefore is not presented here . at @xmath0
one again arrives at the same result for @xmath261 ( [ t1 ] ) .
now we discuss the relation between our present results and those derived earlier for weakly disordered metals by means of a different approach @xcite .
let us express the dot mean level spacing via the average dot size @xmath331 as @xmath332 ( where @xmath333 is the electron density of states at the fermi level ) .
then we obtain @xmath334 below we consider two different physical limits of @xmath335 good metals and @xmath336 strongly disordered ( granular ) conductors . for the model @xmath335
we assume that quantum dots are in a good contact with each other . in this case
@xmath2 scales linearly with the contact area @xmath337 , where @xmath338 is a numerical factor of order ( typically smaller than ) one which particular value depends on geometry . for weakly disordered metals most conducting channels in such contacts
can be considered open .
hence , @xmath339 and @xmath340 i.e. @xmath341 . comparing this estimate with the standard definition of @xmath7 for a bulk diffusive conductor , @xmath342
, we immediately observe that within our model the average dot size is comparable to the elastic mean free path , @xmath343 , as it should be for weakly disordered metals .
expressing @xmath261 ( [ t1 ] ) via @xmath7 , in this limit we get @xmath344 where @xmath270 is the electron mass and @xmath345 is constant which depends on @xmath346 .
estimating , e.g. , @xmath347 , one obtains @xmath348 .
note that apart from an unimportant numerical pre - factor and the logarithm in the denominator of eq .
( [ d3 ] ) the latter result for @xmath261 coincides with that derived in refs .
@xcite for a bulk diffusive metal within the framework of a completely different approach , cf .
, e.g. , eq .
( 81 ) in @xcite . within that approach local properties of the model could not be fully defined .
for this reason in the corresponding integrals in @xcite we could not avoid using an effective high frequency cutoff procedure which yields the correct leading dependence @xmath349 and it only does not allow to recover an additional logarithmic dependence on @xmath7 in ( [ d3 ] ) . our present approach is divergence - free and , hence , it does not require any cutoffs .
we can also add that eq .
( [ t1 ] ) also agrees with our earlier results @xcite derived for quasi-1d and quasi-2d metallic conductors .
provided the transversal size @xmath350 of our array is smaller than @xmath331 one should set @xmath351 for 2d and @xmath352 for 1d conductors
. then eq .
( [ t1 ] ) yields @xmath353 and @xmath354 respectively in 2d and 1d cases .
up to the factor @xmath355 these dependencies coincide with ones derived previously , cf .
, e.g. , eq .
( 32 ) in @xcite .
now let us turn to the model @xmath336 of strongly disordered or granular conductors .
in contrast to the situation @xmath335 , we will assume that the contact between dots ( grains ) is rather poor , and inter - grain electron transport may occur only via a limited number of conducting channels .
in this case the average dimensionless conductance @xmath2 can be approximated by some @xmath356-independent constant @xmath357 . substituting @xmath358 instead of @xmath2 into eq .
( [ dddd ] ) we observe that in the case of strongly disordered structures one can expect @xmath359 .
accordingly , for @xmath261 ( [ t1 ] ) one finds @xmath360 where @xmath361 again depends on @xmath346 .
for @xmath362 we have @xmath363 . hence , the dependence of @xmath261 on @xmath7 for strongly disordered or granular conductors ( [ d-3 ] ) is it qualitatively different from that for sufficiently clean metals ( [ d3 ] ) .
one can also roughly estimate the crossover between the regimes @xmath335 and @xmath336 by requiring the values of @xmath364 ( [ vfd ] ) and @xmath365 to be of the same order .
this condition yields @xmath366 , and we arrive at the estimate for @xmath7 at the crossover @xmath367 here we restored the planck constant @xmath368 set equal to unity elsewhere in our paper . in the next section
we will use the above results and carry out a detailed comparison between our theory and numerous available experimental data for @xmath261 in different types of disordered conductors .
turning to experiments , it is important to emphasize again that low temperature saturation of the electron decoherence time has been repeatedly observed in numerous experiments and is presently considered as firmly established and indisputably existing phenomenon . at the same time , the physical origin of this phenomenon still remains under debate . the key observations and remaining controversies are briefly summarized below . 1 .
the authors @xcite have analyzed the values of @xmath261 observed in their experiments with open quantum dots as well in earlier experiments by different groups @xcite .
for all 14 samples reported in @xcite the values @xmath261 were found to rather closely follow a simple dependence @xmath369 this _ approximate _ scaling was observed within the interval of dwell times @xmath3 of about 3 decades , see fig . 5 in ref .
@xcite . to the best of our knowledge , until now no physical interpretation of this observation has been suggested .
2 . in some of our earlier publications @xcite
we have demonstrated a good quantitative agreement between our theoretical predictions @xcite and experimental data for @xmath261 obtained for numerous metallic wires and quasi-1d semiconductors .
as our theory of dephasing by electron - electron interactions @xcite predicts a rather steep increase of @xmath261 with the system diffusion coefficient @xmath7 , e.g. for most metals as @xmath349 , we can conclude that for a large number of disordered conductors @xmath261 strongly _ increases _ with increasing @xmath7 .
3 . in a series of papers , see , e.g. , refs .
@xcite , lin and coworkers analyzed numerous experimental data for @xmath261 obtained by various groups in disordered conductors with @xmath370 @xmath16/s and observed systematic _
decrease _ of @xmath261 with increasing @xmath7 .
the data could be rather well fitted by the dependence @xmath371 with the power @xmath372 .
this trend is _ opposite _ to one observed in less disordered conductors with @xmath15 @xmath16/s and remained unexplained until now .
the authors @xcite pointed out a disagreement between our expressions for @xmath261 @xcite and the data obtained for a number of typically rather strongly disordered 2d and 3d structures . in some cases
this disagreement was argued to be as large as 4 to 5 orders of magnitude . in ref .
@xcite we countered this critique showing , on one hand , reasonable agreement for some of the samples in question and arguing , on the other hand , that our quasiclassical theory @xcite is applicable merely to weakly disordered conductors .
hence , it can not be used in order to quantitatively describe strongly disordered structures like , e.g. , granular metals , metallic glasses etc . quoted in ref .
@xcite . formally eliminating the controversy , this our argument
, however , did not yet allow to clarify the issue of low temperature saturation of @xmath10 in strongly disordered conductors which remained unclear until now .
pierre _ et al .
_ @xcite argued that low temperature saturation of @xmath262 in sufficiently clean samples can be caused by undetectably small number of magnetic impurities .
this idea , however , was not supported by the authors @xcite who demonstrated that at least in their experiments ( performed in high magnetic fields in order to fully polarize any magnetic moments should they exist ) the observed @xmath262-saturation can not be due to magnetic impurities .
later the issue was reanalyzed in refs .
@xcite on the basis of recently developed numerical renormalization group ( nrg ) theory of electron scattering by kondo impurities @xcite . in samples with implanted magnetic impurities a very good agreement between theory @xcite and experiments @xcite was demonstrated both above and below the kondo temperature @xmath373 .
however , at @xmath374 significant deviations from nrg predictions was observed and @xmath10 was found to saturate @xcite in all samples both with and without magnetic impurities .
interpretation of this saturation effect in terms of both underscreened and overscreened models is problematic @xcite .
let us now analyze the above problems and controversies point by point within the framework of our theory of electron - electron interactions .
we start from the case of quantum dots @xcite observe that our results for @xmath261 ( [ tauphi ] ) , ( [ t1 ] ) scale practically linearly with the dwell time @xmath3 which is essentially the scaling ( [ scdots ] ) suggested in ref .
@xcite . there is , however , one point which requires a comment . as it was argued in sec . 2
, one should expect no dephasing by electron - electron interactions in single quantum dots at any @xmath11 .
naively one could regard this statement as yet one more controversy with experiments @xcite where electron dephasing in _ single _ dots was clearly observed .
at this stage let us recall that qualitative arguments in sec .
2 as well as rigorous analysis in refs .
@xcite remain applicable to single dots provided fluctuating voltages drop strictly across the two barriers . in realistic quantum dots @xcite fluctuating voltages
most likely penetrate inside rather than drop only at the edges . within the framework of our model one
can easily mimic this situation by introducing additional scatterers inside the dot in which case fluctuating voltages already do dephase ( see sec .
2 ) . for illustration ,
let us consider a strongly asymmetric double dot system , i.e. we replace , say , the left barrier in fig . 1 by a small quantum dot with the electron flight time @xmath375
. the left dot will then model a barrier of a finite length in a single dot configuration of fig .
1 . applying eqs .
( [ gwlf1 ] ) , ( [ wlf ] ) and ( [ wlk ] ) , we evaluate the wl correction and again arrive at eqs .
( [ tt]-[t0 ] ) , where , however , @xmath376 . extracting the decoherence time @xmath261
, we obtain @xmath377 for @xmath378 and vanishing decoherence rate in the opposite limit @xmath379 .
provided the denominator in eq .
( [ tasym ] ) is not very large , we arrive at eq .
( [ scdots ] ) .
we also note that @xmath261 in a single dot ( [ tasym ] ) is always longer than that in a system of two dots ( [ tauphi0 ] ) .
alternatively , one can model a dot by a chain of several ( @xmath35 ) scatterers with @xmath380 defined in eqs .
( [ t1 ] ) , ( [ taut ] ) , ( [ aak ] ) .
then at higher temperatures we obtain @xmath381 with @xmath382 ranging from 2/3 to 1 , as observed in a number of dots @xcite . substituting @xmath383 by @xmath384 ( where @xmath385 are @xmath386 are respectively the dimensionless conductance and the dwell time of a composite dot ) , bearing in mind that @xmath387 @xcite and assuming @xmath35 to be not very large , at low @xmath11 from ( [ t1 ] ) one finds @xmath388 in agreement with experimental results presented in fig . 5 of ref . @xcite .
more complicated configurations of scatterers can also be considered with essentially the same results .
we conclude that our present theory is in a good agreement with experimental findings @xcite .
let us now turn to experiments with spatially extended conductors . in fig .
6 we have collected experimental data for @xmath316 obtained in over 120 metallic samples with diffusion coefficients varying by @xmath389 decades , from @xmath390 @xmath16/s to @xmath391 @xmath16/s .
the data were taken from about 30 different publications listed in figure caption .
we see that the the measured values of @xmath261 strongly depend on @xmath7 .
furthermore , this dependence turns out to be non - monotonous : for relatively weakly disordered structures with @xmath15 @xmath16/s @xmath261 clearly increases with increasing @xmath7 , while for strongly disordered conductors with @xmath392 @xmath16/s the opposite trend takes place .
in addition to the data points in fig .
6 we indicate the dependencies @xmath393 ( [ d3 ] ) and ( [ d-3 ] ) for two models @xmath335 and @xmath336 discussed in sec
. 5.2 . ) ; au @xcite ( @xmath394 ) ; au , ag @xcite ( @xmath395 ) ; au pd @xcite ( @xmath396 ) ; @xcite ( @xmath397 ) ; au @xcite ( @xmath398 ) ; @xcite ( @xmath399 , ten points within the box ) ; @xcite au@xmath400al ( @xmath401 ) , sb ( @xmath402 ) , sc@xmath403ag@xmath404 ( @xmath405 ) , v@xmath406al ( @xmath407 ) ; @xcite ( @xmath408 ) ; @xcite cugeau ( @xmath409 ) ; aupd-3 and al @xcite ( @xmath410 ) au @xcite(@xmath411);.,width=283 ] we observe that for @xmath15 @xmath16/s the data points clearly follow the scaling ( [ d3 ] ) . practically all data points remain within the strip between the two lines corresponding to eq .
( [ d3 ] ) with @xmath412 ( dashed line ) and @xmath413 ( solid line ) . on the other hand , for more disordered conductors with @xmath370 @xmath16/s the data are consistent with the scaling ( [ d-3 ] ) obtained within the model @xmath336 .
we would like to emphasize that theoretical curves ( [ d3 ] ) and ( [ d-3 ] ) are presented in fig . 6 without any additional fit parameters except for a geometry factor @xmath338 for the first dependence and the value @xmath414 for the second one .
this value of @xmath358 was estimated from the crossover condition ( [ estd ] ) with @xmath415 @xmath16/s and @xmath416 .
note that quite a few data points with @xmath15 @xmath16/sek correspond to the samples contaminated by magnetic impurities .
remarkably , these data points also demonstrate though with somewhat larger scatter _ systematic _ increase of @xmath261 with increasing @xmath7 . at the same time , for similar values of @xmath7 the samples with higher concentration of magnetic impurities have systematically lower @xmath261 than samples with few or no magnetic impurities .
these observations indicate that for samples with relatively high concentration of magnetic impurities both mechanisms of electron - electron interactions and spin - flip scattering provide substantial contributions to @xmath10 , being responsible respectively for the scaling @xmath349 and for additional non - universal shift of the data points downwards . in order to carry out more accurate comparison with our theory of electron - electron interactions
let us now leave out the data points taken from the samples with high concentration of magnetic impurities . in fig .
7 we selected the data for 37 different metallic wires with no or few magnetic impurities and with diffusion coefficients in the range 9 @xmath16/s @xmath417@xmath16/s from refs .
two minor adjustments of some data points are in order .
firstly , in order to eliminate the uncertainty related to different definitions of @xmath10 used by different groups @xcite , we have adjusted 6 data points @xcite and 2 data points @xcite according to the definition of @xmath10 used in other works @xcite . secondly
, 5 data points corresponding to samples with high diffusion coefficients @xcite have been adjusted in order to eliminate the temperature dependent contribution to @xmath10 which remains substantial for these samples down to the lowest @xmath418 mk @xcite .
both these adjustments can in no way influence our conclusions ( in fact , non - adjusted data points also remain in - between solid and dashed lines in fig .
/s @xmath417@xmath16/s .
only the data for the samples with low concentration or with no magnetic impurities have been selected : au-1 , au-3 , au-4 , au-6 @xcite , au-7 @xcite , au-9 @xcite and cf-1 ,
cf-2 @xcite ( @xmath419 ) ; a to f ( aupd ) @xcite @xmath420 ) ; ag(6n)a , b , c , d , au(6n ) , ag(5n)a , b , and cu(6n)a , b @xcite ( @xmath395 ) ; au @xcite , au1@xcite , ag1 @xcite and ag2 @xcite ( @xmath421 ) ; 2 samples from @xcite , still to be included , 2 au @xcite ( @xmath398 ) ; d ( aupd ) , f , h ( ag ) @xcite ( @xmath408 ) ; aupd-3 and al @xcite(@xmath410 ) au @xcite ( @xmath411).,width=283 ] the data in fig .
7 again clearly support the scaling of @xmath261 with @xmath7 ( [ d3 ] ) due to electron - electron interactions .
practically all data points are now located in - between the solid and dashed lines which indicate the dependence ( [ d3 ] ) with respectively @xmath422 and @xmath423 .
we can not exclude that the remaining scatter among the data points with similar values of @xmath7 to a certain extent may be due to relatively small amount of magnetic impurities possibly residing in some samples .
also minor differences in metal parameters ( e.g , fermi velocities ) may contribute to this effect .
it appears , however , that sample - to - sample fluctuations of the geometry parameter @xmath338 play the most important role .
this parameter is defined as a ratio between the square root of the inter - dot ( inter - grain ) contact area @xmath424 and the average dot ( grain ) size @xmath425 , i.e. @xmath338 accounts for _ local _ properties of the sample which can be highly non - universal .
electrons in metallic wires get scattered both in the bulk and on the surface .
hence , e.g. , the surface quality and/or details of the wire geometry along with other factors may significantly impact the value of @xmath338 .
since @xmath261 ( [ d3 ] ) depends quite strongly on @xmath338 , it is by no means surprising that different wires with similar values of @xmath7 may have decoherence times differing by few times .
as an illustration of this point let us consider 4 quasi-1d aupd samples c , d , e , f @xcite with nominally identical values of @xmath426 @xmath16/s .
although these samples were fabricated in the same way and measured in one experiment @xcite , their dephasing times were found to differ by up to @xmath427 times .
this difference can hardly be ascribed to magnetic impurities since the measured level of dephasing would require unrealistically high concentration of such impurities ( in the range of 100 ppm ) and , in addition , rather exotic values of the kondo temperature @xmath373 . at the same time , sample - to - sample fluctuations of the parameter @xmath338 of only @xmath428 per cent ( which is easy to assume given , e.g. , somewhat different geometry of the samples ) would fully account for the above difference of dephasing times .
we conclude that our theoretical expression ( [ d3 ] ) is in a good quantitative agreement with the available experimental data for quasi-1d metallic wires with diffusion coefficients in the range 9 @xmath16/s @xmath429 @xmath16/s .
now let us return to fig .
6 and consider the data for strongly disordered conductors with @xmath430 @xmath16/s . as we already pointed out , the agreement between the data and the dependence ( [ d-3 ] ) predicted within our simple model @xmath336 is reasonable , in particular for samples with @xmath431@xmath16/s . at higher diffusion coefficients
most of the data points indicate a weaker dependence of @xmath261 on @xmath7 which appears natural in the vicinity of the crossover to the dependence ( [ d3 ] ) .
the best fit for the whole range 0.3 @xmath16/s @xmath432 @xmath16/s is achieved with the function @xmath371 with the power @xmath433 .
further modifications of our clearly oversimplified
model @xmath336 can help to achieve even better agreement between theory and experiment .
although such modifications can certainly be worked out , it is not our aim to do it here .
more importantly , our analysis of sec .
5.2 allows to qualitatively understand and explain seemingly contradicting dependencies of @xmath261 on @xmath7 observed in weakly and strongly disordered conductors .
while the trend `` less disorder less decoherence '' ( [ d3 ] ) for sufficiently clean conductors is quite obvious , the opposite trend `` more disorder less decoherence '' in strongly disordered structures requires a comment .
effectively the latter dependence implies that with increasing disorder electrons spend more time in the areas with fluctuating in time but spatially uniform potentials which do not dephase , as we also discussed in sec .
2 . in other words , in this case an effective dwell time @xmath3 in eq .
( [ t1 ] ) becomes longer with increasing disorder and , hence , the electron decoherence time @xmath261 does so too .
is the physical picture of @xmath7 _ decreasing _ with increasing dot ( or grain ) size ( employed within the model @xmath336 ) realistic ?
although for cleaner conductors the tendency is usually just the opposite , for strongly disordered structures increasing resistivity with increasing grain size has been observed in a various experiments @xcite .
in addition , since local conductance fluctuations increase with increasing disorder , several grains can form a cluster with internal inter - grain conductances strongly exceeding those at its edges . in this case
fluctuating potentials remain almost uniform inside the whole cluster which will then play a role of an effective ( bigger ) grain / dot . accordingly , the average volume of such `` composite dots '' @xmath434 may grow with increasing disorder , electrons will spend more time in such bigger dots and , hence , the electron decoherence time ( [ t1 ] ) will increase .
the above comparison with experiments confirms that our previous quasiclassical results for @xmath261 @xcite are applicable to relatively weakly disordered structures with @xmath435 @xmath16/s , while for conductors with stronger disorder different expressions for @xmath261 ( e.g. , eq .
( [ d-3 ] ) ) should be used .
for instance , the claimed in ref .
@xcite `` disagreement '' between our theory and some experiments by 4 to 5 orders of magnitude is solely due to misuse of our results @xcite far beyond their applicability range . a glance at fig . 6 is sufficient to realize that any attempt to apply eq .
( [ d3 ] ) to structures with , e.g. , @xmath436 @xmath16/s can easily lead to a `` disagreement '' with the data , say , by 6 orders of magnitude or so .
this observation , however , does not imply any real disagreement , rather it indicates that more general results for @xmath261 , e.g. , eq .
( [ t1 ] ) , should be used .
the same argument invalidates the comparison between our quasiclassical results @xcite and the data for the sample au-5 @xcite , see fig .
@xcite , or , to a somewhat lesser extent , for the sample c of ref .
@xcite , see table ii of that paper @xcite .
finally , our analysis allows to rule out scattering on magnetic impurities as a cause of low temperature saturation of @xmath10 .
this mechanism can explain neither strong and non - trivial dependence of the electron decoherence time on @xmath7 ( see our figs .
6 and 7 as well as , e.g. , fig . 4 in @xcite , fig . 1 in @xcite and
fig . 5 in @xcite ) nor
even the level of dephasing observed in numerous experiments .
e.g. , in order to be able to attribute dephasing times as short as @xmath437 s to magnetic impurities one needs to assume huge concentration of such impurities ranging from few _ hundreds _ to few _ thousands _ ppm which appears highly unrealistic , in particular for systems like carbon nanotubes , 2degs or quantum dots .
similar arguments were independently emphasized by lin and coworkers @xcite . even in metallic wires with high values of @xmath7 and long dephasing times , at @xmath374 one observes clear saturation of @xmath10 @xcite which is in a _ quantitative _ agreement with our theory of electron - electron interactions ( see fig . 7 ) and is very hard to explain otherwise @xcite .
thus , although electron dephasing due to scattering on magnetic impurities is by itself an interesting issue , its role in low temperature saturation of @xmath10 in disordered conductors is sometimes strongly overemphasized .
since the latter phenomenon has been repeatedly observed in _ all _ types of disordered conductors , the physics behind it should most likely be universal and fundamental .
we believe and have demonstrated here that it is indeed the case : zero temperature electron decoherence in all types of conductors discussed above is caused by electron - electron interactions .
in this paper we have employed a model of an array of quantum dots / scatterers ( fig .
3 ) which embraces various types of disordered conductors and allows to study electron transport in the presence of interactions within a very general theoretical framework .
we have non - perturbatively analyzed the impact of electron - electron interactions on weak localization for such structures with the emphasis put on the interaction - induced decoherence of electrons at low temperatures .
we have formulated a fully self - contained theory free of any divergencies and cutoffs which allows to conveniently handle disorder averaging and treat electron scattering without residing to quasiclassics . in the case of two quantum dots ( or three scatterers )
it was possible to find an exact solution of the problem ( sec .
4 ) and to evaluate the wl correction to the system conductance practically without approximations . with the aid of our approach
we have formulated a unified description of electron dephasing by coulomb interaction in different structures including ( i ) weakly disordered conductors ( e.g. , metallic wires with @xmath15 @xmath16/s ) , ( ii ) strongly disordered conductors ( @xmath370 @xmath16/s ) and ( iii ) metallic quantum dots .
we have demonstrated that in all these cases at @xmath0 the electron decoherence time is determined by the same simple formula @xmath438 . in the case ( i )
this formula yields @xmath439 and matches with our previous quasiclassical results @xcite while in the cases ( ii ) and ( iii ) it illustrates new physics which was not yet explored before .
in particular , this formula emphasizes the dependence of @xmath261 on the electron dwell time @xmath3 in single quantum dots @xcite and helps to understand the ( at the first sight counterintuitive ) trend `` more disorder
less decoherence '' observed in strongly disordered conductors @xcite .
we have carried out a detailed comparison of our theoretical predictions with the results of numerous experiments for the whole scope of structures ( i ) , ( ii ) and ( iii ) ( sec .
6 ) . in all cases
we found a good agreement between theory and experiment which further supports our main conclusion that low temperature saturation of @xmath10 is universally caused by electron - electron interactions .
we are grateful to c. bauerle , j. bird , p. hakonen , p. mohanty , d. natelson , m. paalanen , j .- j .
lin , l. saminadayar and r.a .
webb for useful discussions on various experimental aspects of electron decoherence and/or for providing us with their experimental data .
this work is part of the european community s framework programme nmp4-ct-2003 - 505457 ultra-1d `` experimental and theoretical investigation of electron transport in ultra - narrow one - dimensional nanostructures '' .
10 g. bergmann , phys .
rep . 107 ( 1984 ) 1 .
altshuler , a.g .
aronov , in _ electron - electron interactions in disordered conductors _ , eds .
efros and m. polak , elsevier science publishers b.v .
, new york ( 1985 ) .
s. chakravarty , a. schmid , phys .
140 ( 1986 ) 193 .
p. mohanty , e.m.q .
jariwala , r.a .
webb , phys .
78 ( 1997 ) 3366 .
webb , p. mohanty , e.m.q .
jariwala , fortsch . phys . 46 ( 1998 ) 779 .
d. natelson , r.l .
willett , k.w .
west , l.n .
pfeiffer , phys .
86 ( 2001 ) 1821 .
f. pierre , a.b .
gougam , a. anthore , h. pothier , d. esteve , n.o .
birge , phys .
b 68 ( 2003 ) 085413 .
p. mohanty , r.a .
webb , phys .
91 ( 2003 ) 066604 .
f. schopfer , c. buerle , w. rabaud , l. saminadayar , phys .
( 2003 ) 056801 .
lin , j.p .
bird , l. rotkina , p.a .
bennett , appl .
82 ( 2003 ) 802 ; j.f .
lin , j.p .
bird , l. rotkina , physica e 19 ( 2003 ) 112 .
a. trionfi , s. lee , d. natelson , phys .
b 72 ( 2005 ) 035407 . c. buerle ,
f. mallet , f. shopfer , d. mailly , g. eska , l. saminadayar , phys .
95 ( 2005 ) 266805 .
p. wakaya , y. tsukatani , n. namasaki , k. murakami , s. abo , m. takai , j. phys .
series 38 ( 2006 ) 120 .
f. mallet , j. ericsson , d. mailly , s. unlubayir , d. reuter , a. melnikov , a.d .
wieck , t. micklitz , a. rosch , t.a .
costi , l. saminadayar , c. buerle , phys .
97 ( 2006 ) 226804 .
alzoubi , n.o .
birge , phys .
97 ( 2006 ) 226803 .
l. saminadayar , p. mohanty , r.a .
webb , p. degiovanni , c. buerle , physica e ( 2007 ) , this volume .
pooke , n. paquin , m. pepper , a. gundlach , j. phys . : condens .
matter 1 ( 1989 ) 3289 .
m. noguchi , t. ikoma , t. odagiri , h. sakakibara , s.n .
wang , j. appl . phys . 80 ( 1996 ) 5138 .
khavin , m.e .
gershenson , a.l .
bogdanov , phys .
81 ( 1998 ) 1066 .
l. langer , v. bayot , e. grivei , j .-
issi , j. p. heremans , c.h .
olk , l. stockman , c. van haesendonck , y. bruynseraede , phys .
76 ( 1996 ) 479 .
n. kang , j.s .
hu , w.j . kong , l. lu , d.l .
zhang , z.w .
pan , s.s .
xie , phys .
b 66 ( 2002 ) 241403 .
r. tarkiainen , m. ahlskog , a. zyuzin , p. hakonen , m. paalanen , phys .
b 69 ( 2004 ) 033402 . j.j .
lin , n. giordano , phys .
b 35 ( 1987 ) 585 .
a. sahnoune , j.o .
strom - olsen , h.e .
fisher , phys . rev .
b 46 ( 1992 ) 10035 .
p.m. echternach , m.e .
gershenson , h.m .
bozler , phys rev .
b 47 ( 1993 ) 13659 . j.j .
lin , j.p .
bird , j. phys . : condens .
matter 14 ( 2002 ) r501 , and further references therein .
lin , t.c .
lee , s.w .
wang , physica e ( 2007 ) this volume , see also further references therein . g. brunthaler , a. prinz , g. bauer , v.m .
pudalov , phys .
87 ( 2001 ) 096802 .
proskuryakov , a.k .
savchenko , s.s .
safonov , m. pepper , m.y .
simmons , d.a .
ritchie , phys .
86 ( 2001 ) 4895 .
y. yaish , o. prus , e. buchstab , g. ben yosef , u. sivan , i. ussishkin , a. stern , cond - mat/0109469 .
lin , l.y .
kao , j. phys . : condens .
matter 13 ( 2001 ) l119 .
bird , k. ishibashi , d.k .
ferry , y. ochiai , y. aoyagi , t. sugano , phys .
b 51 ( 1995 ) 18037 . r.m .
clarke , i.h .
chan , c.m .
marcus , c.i .
duruoz , j.s .
harris , k. campman , a.c .
gossard , phys .
b 52 ( 1995 ) 2656 . d.p .
pivin , a. andresen , j.p .
bird , d.k .
ferry , phys .
( 1999 ) 4687 .
huibers , j.a .
folk , s.r .
patel , c.m .
marcus , c.i .
duruoz , j.s .
harris , phys .
83 ( 1999 ) 5090 .
b. hackens , s. faniel , c. gustin , x. wallart , s. bollaert , a. cappy , v. bayot , phys .
94 ( 2005 ) 146802 .
golubev , a.d .
zaikin , phys .
81 ( 1998 ) 1074 .
golubev , a.d .
zaikin , phys .
b 59 ( 1999 ) 9195 .
golubev , a.d .
zaikin , phys .
b 62 ( 2000 ) 14061 . d.s .
golubev , a.d .
zaikin , j. low temp .
( 2003 ) 11 .
feynman , a.r .
hibbs , _ quantum mechanics and path integrals _ ( mcgraw hill , ny , 1965 ) , ch . 12 .
caldeira , a.j .
leggett , ann .
149 ( 1983 ) 347 .
g. schn , a.d .
zaikin , phys .
198 ( 1990 ) 237 .
u. weiss , _ quantum dissipative systems _ , world scientific , singapore , 2nd edition ( 1999 ) .
golubev , a.d .
zaikin , physica b 255 ( 1998 ) 164 .
golubev , a.d .
zaikin , g. schn , j. low temp .
( 2002 ) 1355 .
huang , h. akimoto , k. kono , j.j .
lin , unpublished . d.s .
golubev , a.d .
zaikin , phys .
b 74 ( 2006 ) 245329 . c.w.j .
beenakker , rev .
69 ( 1997 ) 731 .
golubev , a.d .
zaikin , phys .
b 69 ( 2004 ) 075318 . p.w .
brouwer , a. lamacraft , k. flensberg , phys .
b 72 ( 2005 ) 075316 . d.s .
golubev , a.d .
zaikin , phys .
86 ( 2001 ) 4887 .
bagrets , yu.v .
nazarov , phys .
94 ( 2005 ) 056801 .
n. taniguchi , b.d .
simons , b.l .
altshuler , phys .
b 53 ( 1996 ) 7618 . s.v .
panyukov , a.d .
zaikin , phys .
67 ( 1991 ) 3168 .
nazarov , phys .
82 ( 1999 ) 1245 .
golubev , a.v .
galaktionov , a.d .
zaikin , phys .
b 72 ( 2005 ) 205417 . g.l .
ingold , yu.v . nazarov . in : _
single charge tunneling _
by h. grabert and m.h .
devoret , nato asi series b 294 ( plenum press , new york 1992 ) p. 21 . strictly speaking , eq .
( [ fas ] ) is valid for @xmath440 .
panyukov , a.d .
zaikin , j. low temp .
73 ( 1998 ) 1 .
h. fukuyama , e. abrahams , phys .
b 27 ( 1983 ) 5976 .
aleiner , b.l .
altshuler , m.e .
gershenson , waves random media 9 ( 1999 ) 201 .
in addition one should _
assume _ that remaining ( `` non - cancelled '' ) terms , e.g. , @xmath441 , can be interpreted as an effective renormalization of the non - interacting result ( [ g0 ] ) .
this assumption turns out to be inconsistent with our exact result ( [ iwl1 ] ) either .
golubev , a.d .
zaikin , phys . rev .
b 70 ( 2004 ) 165423 .
altshuler , a.g .
aronov , d.e .
khmelnitskii , j. phys .
c 15 ( 1982 ) 7367 .
golubev , a.d .
zaikin , phys .
82 ( 1999 ) 3191 ; cond - mat/9811185 .
g. zarand , l. borda , j. von delft , n. andrei , phys .
( 2004 ) 107204 ; see also g. zarand , l. borda , physica e ( this volume ) and further references therein .
p.m. echternach , m.e .
gershenson , h.m .
bozler , a.l .
bogdanov , b. nilsson , phys .
b 48 ( 1993 ) 11516 .
p. mohanty , unpublished .
f. altomare , a.m. chang , m.r .
melloch , y. hong , c.w .
tu , appl .
86 ( 2005 ) 172501 .
different groups use different functions in order to fit their magnetoresistance data . as a result
the values of @xmath10 extracted from such fits may differ by a numerical prefactor . here
we stick to the standard definition of @xmath10 based on , e.g. , eq .
( 1 ) of ref . @xcite . with this choice
, the values of @xmath10 obtained from fits with the airy function @xcite should be ajusted ( moved upwards ) by a numerical factor @xmath389 .
for further discussion of this point we refer the reader to appendix b in @xcite .
we also note that the adjustment of the data points @xcite was carried out only in fig . 7 but not in fig .
according to our result ( [ d3 ] ) @xmath316 can be very long at large values of @xmath7 .
hence , saturation of @xmath10 in such samples can only be observed at much lower temperatures as compared to samples with smaller values of @xmath7 .
e.g. , the samples ag6na , b , c , d and au6n reported in ref .
@xcite down to temperature @xmath442 mk were still not in the true saturation regime .
later the samples with similar values of @xmath7 were cooled down to lower temperatures @xcite and saturation of @xmath10 was indeed observed at @xmath443 mk , see fig .
thus , in order to perform a fair comparison , the values @xmath261 for the samples ag6na , b , c , d and au6n @xcite should be determined by fitting the measured temperature dependencies for @xmath10 to our theory .
this procedure was described in detais in ref .
g. ramaswamy , k.a .
raychaudhuri , j. goswami , s.a .
shivashankar , j. appl .
82 ( 1997 ) 3797 .
lee , j.s .
min , s.k .
rha , s.s .
chun , c.o .
park , j. material sci .
: materials in electronics 7 ( 1996 ) 111 .
kang , s.h .
han , s.w .
rhee , thin solid films 350 ( 1999 ) 10
. the data for the samples d , g and h @xcite agree well with our theory @xcite , as is also seen in fig .
note that the ( saturated ) experimental values of @xmath8 for the samples g and h are respectively 1.9 @xmath444 m and 2.6 @xmath444 m @xcite .
their ratio @xmath445 matches perfectly with our theoretical prediction @xmath446 , cf .
table ii in @xcite .
d. natelson , private communication . | we develop a new unified theoretical approach enabling us to non - perturbatively study the effect of electron - electron interactions on weak localization in arbitrary arrays of quantum dots .
our model embraces ( i ) weakly disordered conductors ( ii ) strongly disordered conductors and ( iii ) metallic quantum dots . in all these cases at @xmath0 the electron decoherence time is determined by the universal formula @xmath1 , where @xmath2 , @xmath3 , @xmath4 and @xmath5 are respectively dimensionless conductance , dwell time , charging energy and level spacing of a single dot . in the case ( i )
this formula yields @xmath6 ( @xmath7 is the diffusion coefficient ) and matches with our previous quasiclassical results [ d.s .
golubev , a.d .
zaikin , phys .
rev .
lett .
81 ( 1998 ) 1074 ] , while in the cases ( ii ) and ( iii ) it illustrates new physics not explored earlier .
a detailed comparison between our theory and numerous experiments provides an overwhelming evidence that zero temperature electron decoherence in disordered conductors is universally caused by electron - electron interactions rather than by magnetic impurities . , weak localization , decoherence , electron - electron interactions , disorder , quantum dots 73.63.kv , 73.21.la , 73.20.fz , 73.23.-b |
collective behavior , which occurs commonly in physical , chemical and biological systems , has been a subject of continued interest in nonlinear science over the last decades @xcite . in neural and biological systems , a typical collective behavior observed is the coherent motion of oscillators .
this phenomenon of synchronization has been widely regarded as having important implications to the function and performance of those systems @xcite .
for instance , asynchronous contraction of the heart the may be triggered by electrical spiral waves , which eventually leads to heart dysfunction @xcite .
recently , much attention has been paid to a particular hybrid state in which an ensemble of identical oscillators with identical coupling spontaneously degenerate to one group with synchronization ( coherent ) and the other group with desynchronization ( incoherent ) @xcite .
this fascinating counterintuitive state was first discovered by kuramoto and co - workers @xcite , and named chimera state " by strogatz @xcite .
the experimental confirmation of the existence of chimera states has been reported independently almost at the same time in diverse systems such as coupled maps @xcite , chemical oscillators @xcite and mechanical pendulums @xcite . in two - dimensional systems , chimera states
take the form of spiral waves @xcite , called spiral wave chimeras with phase - locking oscillators in the spiral arm but a phase - randomized spiral core @xcite . an analytical solution for a spiral wave chimera
was further demonstrated by strogatz _
et al_. using a kuramoto - type phase equation with nonlocal coupling @xcite .
recently , the spiral wave chimera state has been reported experimentally in chemical oscillators @xcite and numerically in complex and even chaotic oscillators @xcite where nonlocal coupling is introduced .
it was initially believed that the chimera state arises from so - called nonlocal coupling , i.e. each oscillator in the system will be affected _ instantaneously _ by a group of oscillators within certain interacted range .
therefore , nonlocal coupling is intermediate between the cases of local and global coupling .
however , some recent works show that the non - locality conditions for occurrence of the chimera state can be further relaxed @xcite . for example , sethia _
et al_. have demonstrated a generalized chimera state called `` amplitude mediated chimera state '' in a population of oscillators even in the case of global coupling rather than the nonlocal coupling @xcite .
it is worth pointing out , on one hand , that in the seminal work of kuramoto _
et al_. , a key assumption to observe spiral wave chimeras is that the third component of reaction - diffusion ( rd ) changes so fast that it can be eliminated adiabatically @xcite . in this way
, the three - component rd system is essentially reduced to an effective two - component one with an extra term represented by a nonlocal coupling . on the other hand , to observe chimera state in experiments , the realization of nonlocal coupling strongly relies on a computer @xcite .
therefore , it remains an open question whether spiral wave chimeras exist in more natural systems .
in such systems , the time scale of each component in the system may be comparable and the coupling will be mediated by a natural law such as diffusion .
further more , spiral waves can rotate outwardly or inwardly @xcite , however all the spiral wave chimeras reported previously are outward .
there is very little information on the existence of inwardly rotating spiral wave chimera . in this work ,
we report the first numerical evidence of the existence of spiral wave chimeras in three - component rd system where each oscillatory element is locally coupled by diffusion , i.e. , by nearest neighbor interaction . differing from previously observed spiral wave chimeras in nonlocally coupled systems
, our study shows remarkable novel properties of spiral wave chimeras : first the observed spiral wave chimeras rotate inwardly , i.e. , a coherent wave propagates toward the phase - randomized center .
second , the non - diffusing components in the rd system exhibit chimera properties , while the diffusing components show a coherent spiral structure . a continuous transition from spiral wave with a smooth core to spiral wave chimeras
is also identified .
our findings in rd systems show that the occurrence of the chimera states does not require nonlocal or global coupling and therefore provide key hints to explore the chimera state in the natural world .
_ model description . _
our starting point is a three - component rd system in two spatial dimensions @xcite : @xmath0 these equations describe the evolution of concentrations of chemical reactants @xmath1 , @xmath2 and @xmath3 , where @xmath1 is an activator and @xmath2 and @xmath3 are inhibitors . @xmath4 and @xmath5 denote the diffusion coefficients of chemical species of @xmath1 and @xmath3 , respectively .
the dimensionless parameter @xmath6 represents the characteristic time scale of the system variable @xmath3 , and will be assigned a finite non - zero value . in principle , the parameter @xmath7 can be absorbed to @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 and @xmath13 which represent other parameters , but we still write it explicitly here to keep the same form as in ref . @xcite .
this three - component rd system actually is an extension of the two - component fitzhugh - nagumo model by coupling to a third variable @xmath3 , which was proposed to study pattern formation in the belousov - zhabotinsky ( bz ) reaction dispersed in water droplets of a water - in - oil aerosol ot ( aot ) microemulsion system ( bz - aot system ) @xcite and to model spot dynamics in gas discharges @xcite .
this study will differ from previous work on pattern formation in rd systems in two aspects .
the first modification is related to the ratio of the diffusion coefficients @xmath14 .
traditionally @xmath15 , but in our study we take @xmath16 or @xmath17 .
therefore , we consider eqs . from a dynamical point as a model describing a large number of oscillators that communicate with each other via a diffusive environment . such kind of models may be related to various systems such as bacteria , yeast cells and chemical oscillators .
the second important thing is on the time scale of @xmath6 .
our second modification is to omit the key assumption that the third variable @xmath3 changes fast , i.e. , @xmath18 @xcite .
hence , in our case the above three - component rd system can no longer be reduced a two - component system with traditional nonlocal coupling . _ numerical methods .
_ we numerically integrated eqs . in time
using the explicit forward euler method with spatial step @xmath19 and time step @xmath20 , in a grid of @xmath21 by @xmath22 oscillators .
we use a five - point stencil to evaluate the laplacian term in .
note that the diffusion term in is effectively implemented as the interaction of each oscillator with its four nearest neighbours on a rectangular lattice . in this paper ,
we focus on the dynamics of system as we vary the model parameters @xmath8 , @xmath13 , with fixed value of @xmath23 , @xmath24 , @xmath25 , @xmath26 , @xmath27 , @xmath28 . in the case of @xmath29 ,
changing the diffusion coefficient @xmath5 is equivalently to rescale the size of the system , and thus we keep the same value @xmath30 through the work . and @xmath31 .
( a ) denotes snapshot of the variable @xmath1 ; ( b ) is the magnification the square region in ( a ) ; ( c ) represents the variable of @xmath3 ; ( d ) is the snapshot of phase defined @xmath32 and ( e ) is a cross section along the line in ( d ) by defining @xmath33 .
it is noted that spiral core is incoherent for the variable @xmath1 , while it is coherent for the variable @xmath3 .
( f ) is the spatiotemporal pattern of @xmath1 along the middle horizontal line of ( a ) showing wave propagation towards the spiral chimera core . in panels ( a - d )
, the wave pattern rotates clockwise . ] . in the case of small @xmath34 , a coherent spiral wave
is observed ( a ) and it is gradually changed to a spiral wave chimera ( b ) as we increase the system parameter @xmath35 . further increasing @xmath8 led to the formation of completely disorder state ( c ) where @xmath36 .
we here also show the corresponding time averaged order parameter @xmath37(refer to ( d - f ) ) and the order parameter but along the center of medium ( g - i ) . except of @xmath8 , other parameters are taken as in fig . 1 . ] _ order parameter . _ to analyze the state of the coupled oscillators , we introduce the oscillator phase @xmath38 by @xmath32 . since several observed states will be incoherent at the level of the simulation grid
, we consider the domain as a discrete set of oscillators at positions @xmath39 with phase @xmath40 . to quantitatively study the size of the region with incoherent oscillations , we define the spatiotemporal order parameter @xmath41\right|\ ] ] where @xmath42 is the oscillator phase as defined above .
the notation @xmath43 denotes the set of nearest neighbors include itself and @xmath44 is a normalization factor where @xmath45 denotes the number of oscillators with the nearest coupling along a given spatial dimension . in the present case ,
we set @xmath46 , since we know the diffusion term in eqs .
was computed using a 5-point discrete laplacian .
finally , the time - averaged order parameter was computed as @xmath47 , where @xmath48 is a time interval , e.g , @xmath49 in our simulations .
this quantity was used to estimate the radius of the incoherent state @xmath50 where @xmath51 ( @xmath52 ) denotes maxim distance between the grid points along the @xmath53 ( @xmath54 ) line when @xmath55 .
_ existence of chimera state in a rd system with inhibitor diffusion only ( @xmath56 ) .
_ figure 1 demonstrates the existence of a spiral wave chimera state in the locally coupled rd system . the snapshot of variable @xmath1 and its magnification around the core are shown in fig .
1(a ) and ( b ) , respectively at time @xmath57 after the initiation of the spiral wave .
note that the @xmath1-variable around the spiral core is discontinuous while it is smooth far from the core , which is similar to previously reported spiral waves chimeras in nonlocal systems .
this property is also true for the other non - diffusing variable @xmath2 .
( figure not shown ) . however , in contrast to @xmath1 and @xmath2 , for the diffusing inhibitor @xmath3 , the whole spiral pattern is smooth even in the region closed to the spiral core as shown in fig .
this property is different from the spiral wave chimeras found in nonlocally coupled oscillator systems where all of variables describing the oscillator demonstrate a similar chimera characteristic .
the corresponding phase distribution of @xmath42 and the dependency of @xmath58 on the position along a horizontal cross section through the center of the medium ( @xmath59 ) is illustrated in fig .
1(d - e ) .
as one notes that close to spiral center , the quantity @xmath60 are nonzero , which means the phases of the oscillators around the spiral core are discontinuous or incoherent ; while far way from it , the phases becomes continuous or coherent as indicated by the zero quantity of @xmath60 . this behavior is the defining property of a chimera state for a group of identical oscillators : some oscillate in a coherent way while a localized subgroup oscillate incoherently @xcite .
it is worth pointing out that the spiral wave chimera shown in fig .
1 rotates inwardly , i.e. one sees the coherent waves propagate toward the phase - randomized core .
this can also be seen from the spatiotemporal pattern for @xmath1 variable along a horizontal cross section through the center of the medium ( @xmath59 ) shown in fig .
the inwardly rotating spiral waves , also called antispiral waves , have been reported 15 years ago @xcite . however , to our best knowledge , this is the first time to report such inwardly rotating wave chimeras in a rd system .
_ impact of local dynamics on the formation of spiral wave chimera .
_ to get more insights to the formation and robustness of the spiral wave chimera , we investigate its behavior under a sweep of the parameter @xmath8 . from @xcite , we recall that the system changes from a stationary state to an oscillatory state via a hopf bifurcation when the value of the parameter @xmath8 is increased . for @xmath61 , we observe spiral waves ( inward ) with a smooth core . from , @xmath62 onwards ,
spiral chimera states are found , whose core radius increases with increasing @xmath8 ; compare panels ( a ) and ( b ) from fig .
the incoherent region of the chimera state monotonically grows with increasing @xmath8 , until the incoherent state fully occupies the domain .
this completely incoherent state is already reached in fig .
2(c ) for @xmath63 .
we have found continuous transition from normal spiral waves to spiral wave chimeras by controlling the parameter @xmath8 . in fig .
2(d - f ) , we show the time - averaged order parameter for different values of @xmath8 . it is noted that in the incoherent chimera core , @xmath64 is smaller than one whereas in the outer region @xmath65 due to the continuity of phase .
therefore , in the case of spiral waves , the region with @xmath65 is extremely small , see fig .
2(d ) . on the contrary , for the turbulent state , @xmath66 everywhere .
these statements can also be seen clearly from fig .
2(g - i ) , which show the local order parameter @xmath64 along the line @xmath59 . the dependency of the chimera core radius on the parameter @xmath8 is shown in fig .
3 . when @xmath67 , the radius @xmath68 and thus in such a case , a classical spiral wave emerges . however , when @xmath8 lies between @xmath69 and @xmath70 , the radius of incoherent spiral core is clearly finite , corresponding to the spiral wave chimera state .
further , we observe a fast transition from spiral waves chimera to completely turbulent state beyond @xmath71 . , for @xmath72 . ] to give the global picture about how the local dynamics parameter affects the behaviors of the spiral wave chimeras , we further identified the occurrence of the chimera state in a wide range of the @xmath8-@xmath73 parameter space .
the results are summarized in fig .
4 . in this figure
, full circles represent the stable spiral wave chimeras , the full triangles and squares denote the spiral waves and incoherent state .
the cross means that oscillations are not sustained . in the parameter space ,
the separation line between spiral and spiral wave chimera is almost vertical , which means that @xmath8 plays the key role in determining such state , while @xmath13 has a more important role in determining whether spiral chimeras or turbulent states appear .
parameter space showing occurrence of spiral waves(@xmath74 ) , fully incoherent state ( @xmath75 ) , spiral wave chimera states ( @xmath76 ) and a stable resting state ( @xmath77 ) .
all the pattern were observed at @xmath57 after the initiation of the spiral wave . ] _ spiral wave chimeras with finite activator diffusion ( @xmath78 ) . _ till now , we only consider the rd systems where only the diffusion of inhibitor , @xmath3 , is presented , i.e. , @xmath17 .
a natural question is : how such kind of spiral wave chimeras would change when we increase @xmath78 from zero to a finite value ?
typical results are presented in fig .
5 . for @xmath79 , spiral wave chimeras survive as we increase to @xmath80 , see fig . 5(a ) and magnification of the core region 5(d ) . the core region becomes smooth but somehow spilt as we increase to @xmath81 , see fig .
5(b ) and 5(e ) .
further increasing @xmath82 finally leads to the appearance of the spiral wave with smooth core as illustrated in fig .
5(c ) and 5(f ) .
these results are within our expectation as the diffusion coupling of the activator has a clear smoothing effect . on spiral wave chimera .
( a ) @xmath80 spiral wave chimeras ; ( b ) @xmath81 spiral wave with split core ; ( c ) @xmath82 spiral wave with smooth core .
( d - e ) are the corresponding core region as indicated by the white square region in ( a - c ) .
@xmath83 , @xmath36 . ]
in this study , we have found states in locally coupled rd systems that are highly similar to the chimera states found in non - locally coupled oscillator systems @xcite .
we have identified this spiral wave chimera state for a broad parameter range in our model that is based on a chemical system .
our results are not limited to the particular rd system , when it is are linearly coupled to the a third variable that diffuses . since
no parameter tuning was involved in our numerical experiments , we believe that the robust nature of spiral wave chimeras found in locally coupled systems indicates that spiral wave chimeras may be a frequent self - organized spatiotemporal pattern in complex systems , which may arise from weak coupling of a set of oscillators by a diffusing variable . therefore , we expect that natural chimera states will be found in oscillatory biological or chemical systems .
previous works on chimera states were focused on nonlocal or global coupling . in both cases ,
each element in the system was treated as a discrete oscillator . whether chimera states exist in a spatially extended continuous media
remains an open question . in a recent note @xcite ,
one - dimensional chimera states in networks of pure local coupling were recently reported . however
, this author did not consider spiral chimera states . in our work
, we started from the spatial discretization of a laplacian operator in eq . on a cartesian grid , and found that the resulting wave became incoherent at the level of the grid scale .
furthermore , we have verified that using a finer discretization , the chimera state still persists .
thus , in locally coupled systems , the elusive chimera state breaks the continuum representation of the medium and forces one to consider the domain as a set of individual oscillators .
therefore , future analytical approaches will need to match the continuum dynamics in the outer region with discrete non - linear dynamics in the spiral core region .
it is noted that although we take the form of local coupling between the @xmath3 variable , in the limit case of @xmath17 , such kind of local coupling of the inhibitor @xmath3 may give rise to the nonlocal effect due to the in absence of diffusion of activator @xmath1 .
however , these nonlocal effects , which largely differs from the traditional nonlocal coupling ( where each oscillator in the system will be affected _ instantaneously _ by a group of oscillators within certain interacted range ) , are not only spatial but also temporal , because of the finite @xmath6 .
the systems with the local coupling that nevertheless demonstrate spatiotemporally nonlocal effects have been largely overlooked in the last decades .
however , such systems may be very common in biological systems and thus deserve further investigation in the future
. there may be various reasons for not observing chimeras in natural experimental settings before . from our study
we note two necessary conditions : first , the local dynamics of each element needs to be oscillatory .
secondly , one requires close - to - vanishing spatial coupling of the observed variable .
finally , we note the essential property of the system ( 1 ) used in our study is that each oscillator ( described by @xmath1 and @xmath2 ) is coupled indirectly via a nonuniform dynamical environment which described by the variable @xmath3 .
the systems with such property may denote a broad class of the systems such as chemical oscillators e.g. , bz - aot @xcite and bz particles imersed in the solutions @xcite , engineered gene network @xcite , yeast cells @xcite . from the results in this study , we expect that natural chimera states can be found in many oscillatory biological or chemical systems .
in summary , we have shown that ( inward ) spiral wave chimeras do exist in spatially extended oscillatory media where only nearest - nearest interaction between the element is present . in such system ,
the nondiffusing components appear as a spiral wave chimera , while the diffusing variables show a coherent spiral wave structure . a continuous transition from coherent spiral waves to spiral waves chimeras
is observed as we increase the model parameter @xmath8 that controls the hopf bifurcation . a phase diagram for spiral wave chimeras
is identified in the wide parameter space .
we further discussed the smooth effects on the spiral core by the diffusion of the activator @xmath1 .
our results on the spiral wave chimera in locally coupled oscillator systems largely improves our understanding of the chimera state and provides indications that the chimera state can be found in natural systems such as biological and chemical systems where each oscillating element communicates via a diffusive dynamical environment . | the recently discovered chimera state involves the coexistence of synchronized and desynchronized states for a group of identical oscillators .
this fascinating chimera state has until now been found only in non - local or globally coupled oscillator systems . in this work ,
we for the first time show numerical evidence of the existence of spiral wave chimeras in reaction - diffusion systems where each element is _ locally _ coupled by diffusion .
this spiral wave chimera rotates inwardly , i.e. , coherent waves propagate toward the phase randomized core .
a continuous transition from spiral waves with smooth core to spiral wave chimeras is found as we change the local dynamics of the system .
our findings on the spiral wave chimera in locally coupled oscillator systems largely improve our understanding of the chimera state and suggest that spiral chimera states may be found in natural systems which can be modeled by a set of oscillators indirectly coupled by a diffusive environment . |
pascal s triangle is constructed from binomial coefficients @xmath2 , where @xmath3 is the @xmath4 entry of the @xmath5 row ( see figure [ pastri ] ) .
it is well - known that pascal s triangle exhibits a fractal nature mod @xmath0 where @xmath0 is a prime . for example , in the mod 2 case , the triangle follows the pattern of sierpinski s fractal , the fractal which begins with a triangle and removes the center quarter , doing the same for each remaining upright triangle at each iteration , as in figure [ sieriter ] .
@xmath6 one can correlate each iteration of sierpinski s fractal to the first @xmath7 rows of pascal s triangle mod 2 , where the `` removed '' portion is all the zeros mod 2 , that is , all the even numbers .
the first 8 rows of this triangle , corresponding to the second iteration of sierpinski s fractal , are shown in figure [ pascal2 ] .
@xmath8 beyond @xmath9 , the exact fractal produced mod @xmath0 is different than sierpinski s , but exhibits the same repetitive structure .
this structure can be described mathematically by expressing @xmath10 and @xmath11 in base @xmath0 .
we will use @xmath12 and @xmath13 to notate this .
let @xmath14 where @xmath15 for all @xmath16 .
then we say @xmath17 .
when the context is unclear , we may use the subscript @xmath0 on either @xmath10 or its expansion .
for example , when @xmath18 , we write @xmath19 . the pertinent result is a theorem of lucas : [ lucas ] let @xmath0 be a prime and @xmath20 .
then @xmath21 we have reproduced theorem [ lucas ] as it appears in @xcite . a proof of the theorem can be found in @xcite .
what the theorem essentially means is that the exact residue class of any binomial coefficient mod @xmath0 can be determined by considering only binomials in the first @xmath0 rows of pascal s triangle , since the @xmath22 are between 0 and @xmath23 .
binomial coefficients can be generalized to produce other triangles , which may also be studied mod @xmath0 . in this paper
we will specifically study the fibonomial triangle , which is constructed from the fibonacci sequence .
let @xmath24 and @xmath25 for @xmath26 ,
so that @xmath27 is the fibonacci sequence @xmath28 there are many combinatorial interpretations of these numbers ; for example , @xmath29 counts the number of ways to tile a row of @xmath30 tiles with dominos and squares ( see @xcite ) .
we will use this result in a later proof .
the sequence also appears in various places throughout nature , such as in pinecone spirals and the unencumbered growth of rabbit populations .
we define a generalization of factorials in terms of @xmath27 , appropriately named fibotorials .
we follow @xcite in our notation by denoting this number as @xmath31 . for @xmath32 ,
let @xmath33 .
we define @xmath34 to be 1 . for example ,
having defined a generalization of factorials , we can now give the definition of fibonomial coefficients .
these are defined identically to binomial coefficients , except with fibotorials instead of factorials : the fibonomial coefficient @xmath36 for @xmath37 .
otherwise , @xmath38 . thus @xmath39 and @xmath40
. it would be reasonable to suspect that @xmath41 has a combinatorial interpretation , and in fact it does . given a @xmath42 board with a lattice path @xmath43 running from the lower left corner to the upper right corner , @xmath41 counts the number of ways over all possible paths @xmath43 to tile the rows above @xmath43 with dominos and squares while tiling the columns below @xmath43 with tilings of dominos and squares , with the added proviso that each column must have a domino in its first two ( lowest ) tiles . a proof of this fact is given in @xcite .
thus @xmath41 is an integer for all @xmath44 .
fibonomial coefficients satisfy an analogue of the binomial relation @xmath45 , in that @xmath46 .
benjamin and reiland @xcite offer a combinatorial proof of this fact by breaking the tilings into two disjoint sets , the first being tilings of lattice paths ending with a vertical step , of which there are @xmath47 , and the second being tilings of lattice paths ending with a horizontal step , of which there are @xmath48 . using this recurrence or the definition
, one can generate the fibonomial triangle .
we have reproduced the first 8 rows in figure [ fibtri ] . given this triangle ,
we would like to derive conditions similar to theorem [ lucas ] for when a prime @xmath0 divides @xmath41 .
such a coefficient exists , since there is a first fibonacci number which @xmath0 divides .
we follow @xcite in our notation of this number .
@xmath49 given a prime number @xmath0 , we define @xmath50 to be the least positive integer @xmath10 such that @xmath51 . for example ,
@xmath52 , since @xmath53 is the first fibonacci number which 11 divides .
there are several different conventions for naming @xmath50 . in @xcite
it is referred to as @xmath54 , _ the rank of apparition of @xmath0_. in @xcite it is referred to as @xmath55 , _ the fibonacci entry point_. throughout the paper we will simply use @xmath50 .
the existence of @xmath50 gives a new base to consider when using the fibonacci numbers .
let @xmath56 .
@xmath57 has all the usual properties of a base , since @xmath58 and @xmath59 for all @xmath60 .
thus every number @xmath10 can be written uniquely as @xmath61 where @xmath62 and @xmath15 for all other @xmath22 .
this is a natural base to use for the fibonacci numbers , since @xmath63 , just as in base @xmath0 we have @xmath64 .
one might wonder naively whether this base could be used to show a similar congruence to theorem [ lucas ] , namely whether the congruence @xmath65 holds for all primes @xmath0 .
but this is not the case .
as shown in @xcite , the congruence holds when @xmath9 , but for @xmath66 the relationship only maintains divisibility by 3 .
in fact , we will show that whenever @xmath67 , the relationship does not even preserve divisibility by @xmath0 .
in contrast to these cases , if @xmath68 , it appears that divisibility by @xmath0 is maintained .
we thus introduce the following conjecture : [ openprob ] let @xmath68 and let @xmath57 be as above .
express @xmath69 and @xmath70 .
then @xmath71 .
in fact , since @xmath72 and @xmath73 , we have that @xmath74 and @xmath75 ( so @xmath76 and @xmath77 ) , meaning conjecture [ openprob ] has already been shown for the first two cases in @xcite .
the rest of the paper will be devoted to studying the validity of conjecture [ openprob ] .
section 2 will detail the important results to consider in the literature , as well as the techniques that will aid us in our endeavors . in section 3
we will show that conjecture [ openprob ] is true when @xmath78 , and will prove an immediate corrollary . section 4 will prove the necessary conditions for conjecture [ openprob ] , namely that it fails when @xmath67 .
section 5 concludes with a discussion of possible directions for future research .
chen and sagan @xcite have studied the fractal nature of the fibonomial triangle mod 2 and 3 , deriving the exact residue class of @xmath41 in each case for all @xmath10 and @xmath11 .
their main theorem in the mod 2 case is as follows : [ chesagmain ] given @xmath79 and @xmath80 we have @xmath81 what this essentially means is that the fibonomial triangle mod 2 is quite similar to pascal s triangle mod 2 , in that it correlates to sierpinski s fractal , except with an initial triangle of height three , rather than height 2 .
see figure [ fibtri2 ] .
chen and sagan s primary method of proof is combinatorial .
using the tiling interpretation of @xmath41 , they pair tilings together by means of an involution , and determine the residue class based on whether or not there is an unpaired tiling left over . as secondary methods , they also show theorem [ chesagmain ] from a number theoretic perspective , and then with an inductive argument .
@xmath82 we will make use of methods similar to chen and sagan s number theoretic proof of theorem [ chesagmain ] .
while making this argument , they use the base @xmath83 .
since @xmath74 , this is identical to the base @xmath84 we defined in the introduction . using this base , they offer the following theorem : let @xmath85 and @xmath86
. then @xmath87 .
as noted in the introduction , since @xmath88 , this shows conjecture [ openprob ] for when @xmath9 .
chen and sagan prove their theorem with a case - by - case analysis of the congruence classes of @xmath10 and @xmath11 , relying on a theorem of knuth and wilf involving the @xmath0-adic valuation of @xmath41 .
we will use this theorem in section 3 to show conjecture [ openprob ] holds when @xmath78 .
knuth and wilf @xcite study generalized binomial coefficients and derive conditions for when the prime @xmath0 divides @xmath41 .
their main theorem regarding fibonomials is concerned with the @xmath0-adic valuation of @xmath41 , that is , the highest power of @xmath0 dividing @xmath41 .
let @xmath89 , and let @xmath0 be a prime .
then @xmath90 is the @xmath0-valuation of @xmath91 , i.e. , the highest power of @xmath0 dividing @xmath91 .
the simplest way to calculate @xmath92 is to calculate the @xmath0-valuation of the numerator and subtract from this the @xmath0-valuation of the denominator .
for example , @xmath93 = 1 - 0 - 0 = 1 $ ] .
these definitions give us enough information to state the pertinent theorem of knuth and wilf : [ knuwil ] the highest power of the odd prime @xmath0 that divides the fibonomial coefficient @xmath94 is the number of carries that occur to the left of the radix point when @xmath95 is added to @xmath96 in @xmath0-ary notation , plus @xmath97 if a carry occurs across the radix point .
for example , when @xmath18 , we have @xmath98 since @xmath99 is the first fibonacci number which 7 divides .
additionallly , we have @xmath100 . then given @xmath101
, we have @xmath102 and @xmath103 . to divide by 8 in base 7
, we must express 8 as @xmath104 , which gives @xmath105 and @xmath106 .
then adding gives @xmath107 where one carry occurs across the radix point and one carry occurs from the 1 s place to the 7 s place .
thus by theorem [ knuwil ] , @xmath108 .
we wish to use theorem [ knuwil ] to show that conjecture [ openprob ] holds when @xmath78 , as these methods illustrate how the conjecture might be proven in the case of a more general prime @xmath0 . for our purposes in the @xmath109 case
, we need the fact that @xmath110 , since the first fibonacci number divisible by 5 is @xmath111
. then @xmath112 . using this fact
, we will prove the following corollary of knuth and wilf s theorem : [ fiveval ] the highest power of 5 that divides the fibonomial coefficient @xmath94 is the number of carries that occur when @xmath113 is added to @xmath114 in 5-ary notation .
break the carries down into three types : those occurring to the left of the radix point , a carry occurring across it , and those occurring to the right of the radix point .
theorem [ knuwil ] simply counts the number of the first type of carry , as does this corollary . for a carry occurring across the radix point , theorem [ knuwil ]
counts @xmath97 .
but we have @xmath115 , so we can simply count whether or not this carry occurs .
lastly , theorem [ knuwil ] does nt count carries to the right of the radix point .
but no such carries occur for our corollary : since division by 5 in base 5 simply moves the radix point one space to the left ( identical to division by 10 in base 10 ) , the numbers @xmath113 and @xmath114 each terminate one place to the right of the radix point , thus eliminating the possibility of any carries further to the right of the radix point . so counting the total number of carries
is identical to counting the number of carries described in theorem [ knuwil ] .
@xmath116 to give necessary conditions for conjecture [ openprob ] , we will need several well - known facts about the divisibility of fibonacci numbers . among these
are @xmath117 and its corollary @xmath118 , as well as binet s formula , which denotes the distinct roots of @xmath119 as @xmath120 and @xmath121 , with the fact that @xmath122 .
we must also introduce the lucas numbers @xmath123 , which have the similar relation @xmath124 . in @xcite ,
hoggatt and bergum make use of these facts to prove a theorem comparing the divisibility by @xmath0 of @xmath29 with that of @xmath125 : [ berhog ] if @xmath0 is an odd prime and @xmath51 then @xmath126 for all @xmath127 we desire a slightly stronger result , namely , that the @xmath0-adic valuation increases by exactly 1 when the subscript of a fibonacci number divisible by @xmath0 is multipled by @xmath0 .
this is in fact true , and we can generalize the proof of theorem [ berhog ] to prove the desired result : [ valinc ] let @xmath0 be an odd prime .
then if @xmath128 , we have @xmath129 .
assume that @xmath130 for some positive @xmath11 .
we wish to show that @xmath129 .
using binet s formula , we have @xmath131 .
then by factoring @xmath132 , we arrive at @xmath133 @xmath134 . in the summation @xmath135 , the middle term is @xmath136 since @xmath137 . as in @xcite
we can show that the sum of the @xmath138 and @xmath139 terms , where @xmath140 , is @xmath141 @xmath142 .
we follow @xcite in writing the subscript this way because of the relation @xmath143 @xmath144 .
thus we have that the sum of the two terms is @xmath145 @xmath146 .
grouping these terms together in the summation , we arrive at @xmath147 . now using our hypothesis that @xmath130 , let us inspect the term in parentheses .
each fibonacci number in the summation has a subscript that @xmath10 divides , giving that @xmath29 divides each summand and hence @xmath148 divides each summand .
but then the whole term takes the form @xmath149 , so only @xmath150 divides it , and no higher powers . hence @xmath151 .
@xmath152 since we are concerned with the fibonomial triangle mod 5 , we have reproduced the first 10 rows of that triangle in figure [ fibtri5 ] . as noted in @xcite , divisibility by 5 of @xmath41
is entirely understood , since the 5-adic valuation of a fibonacci number ( @xmath153 ) is the same as the 5-adic valuation of its subscript ( @xmath154 ) , and hence @xmath155 , since @xmath3 is obtained by replacing the fibonacci numbers in @xmath41 with their subscripts .
thus by theorem [ lucas ] , conjecture [ openprob ] holds for @xmath78 .
however , we will show this result using corollary [ fiveval ] , to illustrate how the result could be extended to more general primes .
as observed previously , we have @xmath110 .
thus the base @xmath156 is simply base 5 .
hence we wish to show the following lemma : [ weightlift ] let @xmath157 and @xmath158
. then @xmath159 first note that by our conditions on @xmath22 and @xmath160 , for each choice of @xmath22 and @xmath160 we have @xmath161 appearing in one of the following 25 positions on the fibonomial triangle : @xmath162 the inverted triangle of zeroes contains the coefficients @xmath161 with @xmath163 .
thus @xmath164 , since no other multiples of five appear in the 25 positions in question . with that in mind
, we proceed with the proof .
@xmath165 by corollary [ fiveval ] , if @xmath166 then there is a carry occurring in the addition of @xmath167 and @xmath168 .
but as noted above , division by 5 in base 5 simply moves the radix point one point to the left , so this means there is a carry in the addition of @xmath169 and @xmath170 .
let the first carry occur at the @xmath171 position , that is , let the addition @xmath172 have a carry such that for @xmath173 has no carry .
there are ten possibilities for @xmath174 , and @xmath175 , as shown in the table below : @xmath176 each of these cases corresponds to a fibonomial @xmath177 that has @xmath178 .
but then as noted , we have @xmath179 and consequently @xmath180 .
@xmath181 conversely , let @xmath180 .
then @xmath182 such that @xmath179 .
without loss of generality let @xmath183 be the least such @xmath183 .
we will show that @xmath174 , and @xmath175 are as in one of the cases in the above table .
first , note that since @xmath179 , we have @xmath178 . thus @xmath177 is one of the 10 cases above , since those are the only cases where @xmath178 .
then we can show that @xmath175 follows by minimality of @xmath183 .
we are working with the following subtraction problem : @xmath184 for each @xmath185 , since @xmath186 .
then by definition , each @xmath161 has @xmath187 .
so there is no carrying involved in calculating @xmath188 , which is exactly @xmath189 for each @xmath190 : @xmath191 then @xmath175 is simply @xmath192 , where the 5 is carried from @xmath193 .
this verifies that @xmath175 is as in the table above .
but then a carry occurs in the addition of @xmath169 and @xmath170 , meaning a carry occurs in the addition of @xmath167 and @xmath168 , giving @xmath194 thus conjecture [ openprob ] holds when @xmath78 .
lemma [ weightlift ] gives the following corollary , which we include because of its similarity to theorem [ chesagmain ] .
[ maincorollary ] let @xmath195 .
then for @xmath196 , we have @xmath197 .
let @xmath198 , and let @xmath199 be the base 5 expansions of @xmath10 and @xmath11 , as in lemma [ weightlift ] . by lemma [ weightlift ] , @xmath200 .
now since @xmath201 , we have @xmath202 and hence @xmath203
. so @xmath204 .
but then by another direct application of lemma [ weightlift ] , this is true if and only if @xmath194
since conjecture [ openprob ] holds for the first three primes , it would be reasonable to attempt to show it holds for all primes . for @xmath18
, data suggests that this pattern continues .
however , the very next case is a counterexample to this trend . when @xmath205 , we have @xmath52 , as noted in the introduction , giving @xmath206 .
the fibonomial coefficent @xmath207 is not divisible by 11 .
however , @xmath208 and @xmath209 , so @xmath210 and @xmath211 , giving @xmath212 .
this problem arises because @xmath67 . we will show that the same problem arises in general when this is the case . here
we will make use of our result from section 2 .
[ valincrestate ] let @xmath0 be an odd prime .
then if @xmath128 , we have @xmath129 .
since @xmath213 by definition of @xmath50 , we have the following corollary : [ neccor ] given an odd prime @xmath0 , @xmath214 .
what we have essentially shown is that multiplication by @xmath0 in the subscript of a fibonacci number already divisible by @xmath0 raises the @xmath0-valuation by one .
now we must show that if @xmath215 , then multiplying by @xmath50 in the subscript does nt change the @xmath0-valuation , i.e. , that @xmath216 . it is well - known that for all primes @xmath0 except 5 , @xmath217 , and hence @xmath215 .
[ neclem ] given an odd prime @xmath0 not equal to 5 , let @xmath218 .
then @xmath219 . @xmath220 since @xmath221 , so @xmath222 .
but @xmath223 .
since @xmath224 , we have @xmath217 and hence @xmath225 .
so @xmath226 .
but then since @xmath227 ( by corollary [ neccor ] ) , this implies that @xmath228 , which gives @xmath219 .
@xmath116 using lemma [ neclem ] , we can show that conjecture [ openprob ] will not hold if @xmath67 . the pertinent counterexample , as we saw in the @xmath205 case with @xmath207 , is @xmath229 . in the base @xmath230 , using @xmath67 , we have @xmath231 and @xmath63 . given this scenario ,
we have the following theorem . here
we will express @xmath41 as the fraction @xmath232 : [ nectheo ] let @xmath0 be a prime such that @xmath67 . then @xmath233 but @xmath234 . certainly by definition
since @xmath235 , we have @xmath233 . now @xmath236 .
but for @xmath237 , we have that @xmath238 , so @xmath239 , since @xmath240 where @xmath241 .
similarly , by definition of @xmath242 . thus @xmath243 and @xmath244 . but by lemma [ neclem ] , we have @xmath245 .
so @xmath246 , giving @xmath234 .
@xmath116 thus for conjecture [ openprob ] to hold , we must in general have @xmath68 .
it is well - known that the largest value @xmath50 can take is @xmath247 , so our conjecture is only interested in 5 and primes @xmath0 whose @xmath50 attains this maximum .
the list of such @xmath0 is @xmath248 .
it is not known whether this list is finite or infinite ( see @xcite ) .
we can thus assert the necessary conditions placed on @xmath0 in the conjecture .
[ openprobrestate ] let @xmath68 and let @xmath57 be as above .
express @xmath69 and @xmath70 .
then @xmath71 . chen and sagan
have shown that conjecture [ openprob ] holds for @xmath249 , and our lemma [ weightlift ] handles @xmath109 .
preliminary computations show that conjecture [ openprob ] holds for the first 500 rows of the fibonomial triangle for @xmath250 , and 83 .
we now discuss further directions of research related to the fibonomial triangle .
extending the divisibility results of corollary [ maincorollary ] to exact residue classes for all @xmath41 will not be possible using only theorem [ knuwil ] , since that theorem relates to divisibility of @xmath41 by a prime @xmath0 , not the exact residue class .
however , one can derive simple relations using combinatorial and other number theoretic methods .
take for example the following lemma , which makes use of the fact mentioned in the introduction , that @xmath29 counts the number of tilings of a row of @xmath30 squares by dominos and monominos : @xmath252 counts the number of ways to tile @xmath253 squares with dominos and monominos .
these tilings can be broken up into two cases .
number the lines between squares as @xmath254 .
mark line @xmath10 .
let the first type of tiling have a domino crossing this line and the second type not have a domino crossing the line .
certainly these cases are disjoint and each tiling falls under one of the cases .
the first type is the number of ways to tile the first @xmath30 tiles for each tiling of the last 3 tiles ( @xmath255 ) .
the second type is the number of ways to tile the first @xmath10 tiles for each tiling of the last 4 tiles ( @xmath256 ) .
this gives the relation @xmath257 .
replacing the known values , we arrive at the desired result : @xmath258 .
@xmath116 as shown in section 4 , one can not use the base @xmath57 to derive a conjecture similar to conjecture [ openprob ] for primes where @xmath67 .
another direction to research is to determine whether there is a more appropriate base to choose when this is the case .
this seems unlikely , since @xmath57 is constructed so that for each term in the base , the fibonacci number with that subscript has a greater @xmath0-valuation than the fibonacci number with the previous term as its subscript , and hence seems like the correct " base to use when studying the fibonacci numbers mod @xmath0 .
however , perhaps there is a different principle to use in constructing a base when @xmath67 which will avoid the problems arising from @xmath57 . | the fibonomial triangle has been shown by chen and sagan to have a fractal nature mod 2 and 3 .
both these primes have the property that the fibonacci entry point of @xmath0 is @xmath1 .
we study the fibonomial triangle mod 5 , showing with a theorem of knuth and wilf that the triangle has a recurring structure under divisibility by five .
while this result is not new , our method of proof is new and suggests a conjecture for the divisibility of a fibonomial coefficient by a general prime @xmath0 .
we give necessary conditions for such primes , namely that the fibonacci entry point must be greater than or equal to @xmath0 , and offer numerical evidence for the validity of the conjecture .
lastly , we conclude with a discussion concerning further directions of research . |
we consider simple undirected graphs , and are interested in the classical packings of graphs as introduced in @xcite , and their generalization , called limited packings , as presented in @xcite . in the literature , the classical packings can be referred to under different names , e.g. , as ( distance ) @xmath9-packings @xcite , closed neighborhood packings @xcite , or strong stable sets @xcite .
they can also be considered as generalizations of independent ( stable ) sets , which would be ( distance ) @xmath10-packings , following the terminology of @xcite .
a vertex set @xmath0 in a graph @xmath1 is a
_ @xmath2-limited packing _
if for any vertex @xmath3 , @xmath4 \cap x\right| \le k$ ] , where @xmath5 $ ] is the closed neighbourhood of @xmath6 .
@xmath2-limited packing number _ @xmath7 of a graph @xmath1 is the maximum size of a @xmath2-limited packing in @xmath1 . in these terms , the classical ( distance ) @xmath9-packings are @xmath10-limited packings , and hence @xmath11 , where @xmath12 is the 2-packing number .
@xmath9-packings ( @xmath10-limited packings ) are well - studied in the literature ( e.g. , @xcite ) .
also , several papers discuss connections between packings and dominating sets in graphs ( e.g. , @xcite ) .
notice that the corresponding problems have a very different nature : one of the problems ( packings ) is a maximization problem not to break some ( security ) constraints , and the other is a minimization problem ( dominating sets ) to satisfy some reliability requirements .
for example , given a graph @xmath1 , the definitions imply a simple inequality @xmath13 , where @xmath14 is the domination number of @xmath1 ( e.g. , see @xcite )
. however , the difference between @xmath12 and @xmath14 can be arbitrarily large .
in general , the problems of packings in graphs can be viewed as facility location problems subject to some ( security ) constraints .
the problem of finding a @xmath9-packing ( @xmath10-limited packing ) of maximum size is shown to be @xmath8-complete in @xcite . in @xcite
, it is shown that the problem of finding a maximum size @xmath2-limited packing is @xmath8-complete even in split or bipartite graphs .
we develop the probabilistic method for @xmath2-limited packings and , in particular , for @xmath9-packings ( @xmath10-limited packings ) , with a comparison to a greedy approach .
we present two new lower bounds for the @xmath2-limited packing number @xmath7 , and a randomized algorithm to find @xmath2-limited packings satisfying the lower bounds .
also , using a greedy algorithm approach , we provide an improved lower bound for @xmath11 .
it can be shown that the main lower bound is asymptotically sharp , and the improvements for @xmath10-limited packings from the greedy algorithm approach are asymptotically irrelevant for almost all graphs .
we also provide some new upper bounds for @xmath7 .
details for some of these results can be found in @xcite .
denote by @xmath15 = @xmath16 the maximum vertex degree of @xmath1 , and put @xmath17 . notice that @xmath18 for @xmath19 .
we put @xmath20 .
the following theorem gives a new lower bound for the @xmath2-limited packing number .
[ th1 ] for any graph @xmath1 of order @xmath21 with @xmath22 , @xmath23 the probabilistic construction used to prove theorem [ th1 ] implies a randomized algorithm to find a @xmath2-limited packing set , whose size satisfies bound ( [ main_bound ] ) with a positive probability .
a pseudocode presented in algorithm [ alg : k - limited ] below explicitly describes the randomized algorithm .
algorithm [ alg : k - limited ] can be implemented to run in @xmath24 time .
also , it can be implemented in parallel or as a local distributed algorithm , which is important in some applications .
[ alg : k - limited ] the lower bound of theorem [ th1 ] can be written in a simple weaker form as follows : for any graph @xmath1 of order @xmath21 , @xmath25 in the case @xmath26 , theorem [ th1 ] gives the following lower bound : [ cor22 ] for any graph @xmath1 of order @xmath21 with @xmath27 , @xmath28 let @xmath29 = @xmath30 denote the minimum vertex degree of @xmath1 .
the lower bound ( [ bcor22 ] ) can be improved as follows : [ th22 ] for any graph @xmath1 of order @xmath21 , @xmath31 the proof of theorem [ th22 ] provides a greedy algorithm to find a distance @xmath9-packing ( @xmath10-limited packing ) satisfying bound ( [ greedy_bound ] ) .
however , as explained later , the lower bound of theorem [ th22 ] is as good as lower bound ( [ bcor22 ] ) of corollary [ cor22 ] for almost all graphs .
refining and developing the ideas and techniques used in the proofs of the above results , we have been able to show the following more subtle results .
[ tha ] for any graph @xmath1 of order @xmath21 , with vertex degrees @xmath32 , @xmath33 , @xmath34 [ thb ] for any @xmath15-regular graph @xmath1 of order @xmath21 , @xmath35 , @xmath36
the lower bound of theorem [ th1 ] is asymptotically best possible for some ` large ' values of @xmath2 . bound ( [ main_bound ] )
can be written as @xmath37{{\delta \choose k } ( \delta + 1 ) } } $ ] for @xmath38 . combining this with the upper bound of lemma @xmath39 from @xcite , for any connected graph @xmath1 with @xmath40
, we have : @xmath41{{\delta \choose k } ( \delta + 1 ) } } \times { k \over k+1 } n \;\le\ ; l_k(g)\ ;
\le \;{k \over k+1 } n.\ ] ] notice that the upper bound in the inequality ( [ ineq5 ] ) is sharp ( see @xcite ) , and bounds ( [ ineq5 ] ) provide an interval of possible values for @xmath7 in terms of @xmath2 and @xmath15 ( when @xmath42 ) . for regular graphs , @xmath43 , and , when @xmath44 , we have @xmath45{{\delta \choose k } ( \delta + 1 ) } } = { 1 \over \left(k+1\right)^{1/k } } \longrightarrow 1\ \quad\mbox{as } \quad \ k \to \infty$ ] .
therefore , theorem [ th1 ] is asymptotically sharp for regular connected graphs in the case @xmath44 .
a similar statement can be proved when @xmath46 .
thus , the following result is true : as stated in theorem [ th22 ] , in contrast to the situation for relatively ` large ' values of @xmath2 , bound ( [ main_bound ] ) of theorem [ th1 ] ( corollary [ cor22 ] ) can be improved for distance @xmath9-packings ( @xmath10-limited packings ) , i.e. when @xmath26 . however , this improvement is irrelevant for almost all graphs .
a @xmath10-limited packing set @xmath0 in @xmath1 has a very strong property that any two vertices of @xmath0 are at distance at least @xmath48 in @xmath1 .
it is well known that almost every graph has diameter equal to @xmath9 .
therefore , @xmath49 for almost all graphs .
thus , in the case @xmath26 , theorem [ th1 ] yields a lower bound of 1 for almost all graphs and is as good as theorem [ th22 ] .
notice that the bound of theorem [ th22 ] is sharp , for example for any number of disjoint copies of the petersen graph .
when @xmath1 has a diameter larger than @xmath9 , it should be better to use the greedy algorithm and lower bound ( [ greedy_bound ] ) provided by theorem [ th22 ] : it improves bound ( [ bcor22 ] ) of corollary [ cor22 ] by a factor of @xmath50 . as mentioned earlier , @xmath51 . in @xcite ,
the authors provide several upper bounds for @xmath7 , e.g. , given any graph @xmath1 , @xmath52 .
considering the @xmath2-tuple domination number @xmath53 and results of @xcite , it is possible to prove the following upper bounds for @xmath7 . a set @xmath0 is a _
@xmath2-tuple dominating set _ of @xmath1 if for every vertex @xmath3 , @xmath54\cap x|\ge k$ ] .
the minimum cardinality of a @xmath2-tuple dominating set of @xmath1 is the _ @xmath2-tuple domination number _ @xmath53 .
the @xmath2-tuple domination number is only defined for graphs with @xmath55 . in @xcite , we define @xmath56 and , for @xmath57 , @xmath58 .
a. gagarin , v. zverovich , the probabilistic approach to limited packings in graphs , corr abs/1311.1707 ( 2013 ) .
r. gallant , g. gunther , b. hartnell , d. rall , limited packings in graphs , _ discrete appl . math . _ * 158*(12 ) ( 2010 ) 13571364 . | we consider ( closed neighbourhood ) packings and their generalization in graphs called limited packings .
a vertex set @xmath0 in a graph @xmath1 is a _ @xmath2-limited packing _
if for any vertex @xmath3 , @xmath4 \cap x\right| \le k$ ] , where @xmath5 $ ] is the closed neighbourhood of @xmath6 .
the _ @xmath2-limited packing number _
@xmath7 is the largest size of a @xmath2-limited packing in @xmath1 .
limited packing problems can be considered as secure facility location problems in networks .
we develop probabilistic and greedy approaches to limited packings in graphs , providing lower bounds for the @xmath2-limited packing number , and randomized and greedy algorithms to find @xmath2-limited packings satisfying the bounds .
some upper bounds for @xmath7 are given as well .
the problem of finding a maximum size @xmath2-limited packing is known to be @xmath8-complete even in split or bipartite graphs . |
the potential of ant - based models @xcite in solving difficult optimization problems has been well emphasized by successful results obtained in many and varied fields including transportation optimization , quadratic assignment , scheduling , vehicle routing and protein folding .
inspired by the real - world collective behaviour of social insects , _ ant colony system ( acs ) _
algorithms @xcite rely on the stigmergic interactions between many identical artificial ants to find solutions to a given problem .
each ant generates a complete tour ( associated to a problem solution ) by probabilistically choosing the next node at each path intersection based on the cost and the amount of pheromone on the connecting edge .
stronger pheromone trails are preferred and the most promising tours build up higher amounts of pheromone in time . inducing heterogeneity in the population by enabling each artificial ant to react in a different way to the same environment @xcite represents a promising approach to the application of ant - based models for solving complex real - world problems possibly with a dynamic character .
each individual ant can be endowed with a certain level of sensitivity to the pheromone trail triggering various types of reactions to a changing environment .
the variable pheromone sensitivity within the same population of ants can potentially intensify the search ( normally through high sensitivity levels ) while in the same time inducing diversity for the exploration of the environment .
the decision of a low - level sensitive ant regarding the action to be performed crucially contributes to the quality of the search process and solutions .
the generalized vehicle routing problem ( gvrp ) is an extension of the vehicle routing problem ( vrp ) and was introduced by ghiani and improta @xcite .
the gvrp is the problem of designing optimal delivery or collection routes , subject to capacity restrictions , from a given depot to a number of predefined , mutually exclusive and exhaustive node - sets ( clusters ) .
the gvrp belongs to the class of generalized combinatorial optimization problems , which are natural extensions of combinatorial optimization problems by considering a related problem relative to a given partition of the nodes of the graph into node sets , while the feasibility constraints are expressed in terms of the clusters . in the literature we can find several generalized problems such as the generalized minimum spanning tree problem ( see @xcite ) , the generalized traveling salesman problem , the generalized vehicle routing problem , the generalized ( subset ) assignment problem , etc .
these generalized problems belong to the class of np - complete problems , are harder than the classical ones and nowadays are intensively studied due to the interesting properties and applications in the real world , even though many practitioners are reluctant to use them for practical modeling problems because of the complexity of finding optimal or near - optimal solutions .
ghiani and improta @xcite showed that the problem can be transformed into a capacitated arc routing problem ( carp ) and baldacci et al .
@xcite proved that the reverse transformation is valid .
recently , pop @xcite provided a new efficient transformation of the gvrp into the classical vehicle routing problem ( vrp ) . as far as we know ,
the only specific algorithm for solving the gvrp was developed by pop et al .
@xcite and was based on ant colony optimization .
the aim of this paper is to investigate the performance of the _ sensitive ant model ( sam ) _
@xcite in solving the _ generalized vehicle routing problem ( gvrp ) _ @xcite .
we report numerical results of the _ sam _ model for several _ gvrp _ benchmark problems and discuss the performance of _ sam _ compared to the standard _
acs _ technique .
let @xmath0 be a directed graph with @xmath1 as the set of vertices and the set of arcs @xmath2 .
a nonnegative cost @xmath3 associated with each arc @xmath4 .
the set of vertices ( nodes ) is partitioned into @xmath5 mutually exclusive nonempty subsets , called clusters , @xmath6 ( i.e. @xmath7 and @xmath8 for all @xmath9 and @xmath10 ) . the cluster @xmath11 has only one vertex @xmath12 , which represents the depot , and remaining n nodes belonging to the remaining @xmath13 clusters represent geographically dispersed customers .
each customer has a certain amount of demand and the total demand of each cluster can be satisfied via any of its nodes .
there exist @xmath14 identical vehicles , each with a capacity @xmath15 .
the generalized vehicle routing problem ( gvrp ) consists in finding the minimum total cost tours of starting and ending at the depot , such that each cluster should be visited exactly once , the entering and leaving nodes of each cluster is the same and the sum of all the demands of any tour ( route ) does not exceed the capacity of the vehicle @xmath15 .
an illustrative scheme of the gvrp and a feasible tour is shown in the figure [ gvrp ] .
the gvrp reduces to the classical vehicle routing problem ( vrp ) when all the clusters are singletons and to the generalized traveling salesman problem ( gtsp ) when @xmath16 and @xmath17 .
the gvrp is @xmath18-hard because it includes the generalized traveling salesman problem as a special case when @xmath19 and @xmath20 .
several real - world situations can be modelled as a gvrp .
the post - box collection problem described in laporte et al .
@xcite becomes an asymmetric gvrp if more than one vehicle is required .
furthermore , the gvrp is able to model the distribution of goods by sea to a number of customers situated in an archipelago as in philippines , new zeeland , indonesia , italy , greece and croatia . in this application ,
a number of potential harbours is selected for every island and a fleet of ships is required to visit exactly one harbour for every island .
several applications of the gtsp ( laporte et al .
@xcite ) may be extended naturally to gvrp .
in addition , several other situations can be modeled as a gvrp , these include : * the traveling salesman problem ( tsp ) with profits ( feillet et al .
@xcite ) ; * a number of vehicle routing problem ( vrp ) extensions : the vrp with selective back hauls , the covering vrp , the periodic vrp , the capacitated general windy routing problem , etc . ; * the design of tandem configurations for automated guided vehicles ( baldacci et al .
the acs - based algorithm for gvrp @xcite uses artificial ants in order to construct vehicle routes by successively choosing exactly one node from each cluster .
this task continues until each cluster has been visited . whenever the choice of another node from a cluster would lead to an infeasible solution because of vehicles capacity
, the depot is chosen and a new route is started . initially ,
the nearest neighbor ( nn ) algorithm - with the rule always go to the nearest as - yet - unvisited location - is considered .
the best solution of nearest neighbor ( nn ) algorithm ( @xmath21 ) is used for acs - based algorithm start .
the number of ants corresponds to the number of gvrp customers _
m_. at the beginning of an iteration , an ant is placed at each node ( customer ) .
after initializing the basic ant system algorithm , the two steps : ( i ) construction of vehicle routes and ( ii ) trail update are repeated for a given number of iterations . to favor the selection of an edge with a high pheromone level and high visibility , a probability function @xmath22
is defined as follows : @xmath23^\beta}{\sum_{o\in j^k_{i}}\tau_{io}^k(t)[\eta_{io}^k(t)]^\beta } \label{eq_acs}\ ] ] where @xmath24 is the set of unvisited neighbors of node @xmath25 by ant @xmath13 , @xmath26 and @xmath27 is a parameter used for tuning the relative importance of visibility .
after an artificial ant has constructed a feasible solution , the pheromone trails are laid depending on the objective value @xmath28 . for each edge that was used by ant @xmath13 ,
the pheromone trail is updated according to the following rule : @xmath29 where @xmath30 is an evaporation rate parameter .
a tabu list prevents ants visiting clusters they have previously visited .
the ant tabu list is cleared after each completed tour .
the global update rule , applied by the elitist ants , as in acs @xcite is : @xmath31 where @xmath21 is the so far best solution .
the _ sensitive ant model ( sam ) _ technique proposed in @xcite is engaged for solving the _ gvrp_.
the general approach to solving gvrp using sam is the same with the acs approach presented in the previous section except that the transition probabilities defined by sam are used .
the initialization of the algorithm , the update rules and the maintenance of a tabu list are kept the same in sam for gvrp .
the _ sam _ algorithm involves several ants able to communicate in a stigmergic manner ( influenced by pheromone trails ) for solving complex search problems . within the _ sam _ model , each ant
is characterized by a pheromone sensitivity level ( _ psl _ ) .
the _ psl _
value is expressed by a real number in the unit interval [ 0 , 1 ] .
psl _ is null the ant completely ignores stigmergic information and when _ psl _ is one the agent has maximum pheromone sensitivity .
the ants with a low _ psl _ value are more independent and are considered environment explorers .
they have the potential to autonomously discover new promising regions of the solution space .
the ants with high _ psl _ values are very sensitive to pheromone traces .
they are influenced by stigmergic information and therefore intensively exploit the promising search regions already identified .
sam introduces a measure of randomness proportional to the level of individual _ psl _ in the decisions of ants regarding the path to follow .
this is achieved by modifying the transition probabilities using the _ psl _ values in a renormalization process @xcite .
the sam renormalized transition probability for ant @xmath13 ( influenced by _ psl _ )
is denoted by @xmath32 and is given by the following equation : @xmath33 where @xmath34 is the probability for ant _ k _ to choose the next node _
j _ from current node _
i _ ( as given in _
acs _ - see equation [ eq_acs ] ) and @xmath35 represents the _ psl _ value of ant _ k _ at time _
it can be noticed that the sam probability of selecting the next node is the same with the acs one when psl value is one . in order to associate a standard probability distribution to the system ,
the sam virtual state corresponding to the lost probability of @xmath36 has to be defined .
the associated virtual state decision rule specifies the action to be taken when the virtual state is selected using the renormalized transition mechanism .
the following rule is used in the current paper : the ant randomly chooses an available node with uniform probability if the virtual state is selected .
this approach favors the increasing of randomness in the selection process with the decreasing of sensitivity level to pheromone .
the performance of the sam and acs for solving _
gvrp _ is investigated .
numerical experiments focus on seven benchmark problems from the _ tsplib _ library @xcite .
these problems contain between 51 and 101 customers ( nodes ) , which are partitioned into a given number of clusters , and in addition the depot .
originally the set of nodes in these problems is not divided into clusters . the clustering procedure proposed by fischetti _
@xcite is used to divide data into node - sets .
this procedure sets the number of clusters @xmath37 $ ] , identifies the @xmath14 farthest nodes from each other and assigns each remaining node to its nearest center .
table [ tab : problems ] contains the description of the gvrp instances addressed in this paper .
[ tab : problems ] .problem characteristics for the ant - based algorithms for _ gvrp _ [ cols="^,^,^,^,^,^",options="header " , ] the computational values are the result of the average of @xmath38 successively runs of both algorithms .
termination criterion is either the maximum number of iterations , @xmath39 or the maximum running time ( five minutes ) on a amd 2600 , 1.9ghz and 1024 mb .
when comparing the best solution reported in @xmath38 runs , the performance of the acs and sam algorithms are similar .
sam reports better values for three out seven problems while acs does better for other three problems .
it should be noticed that whenever a method is able to obtain a better solution , it also reports a longer running time compared to the other one .
the average solutions obtained by sam are clearly improved compared to acs although the running time has slightly increased for some of the gvrp instances .
sam detects a better average solution ( calculated based on the 25 runs ) for six out of the seven benchmark problems .
overall better average sam results are facilitated by a better exploration of the search space and exploitation of new solutions .
this is due to the variable sensitivity induced in sam via random individual psl values .
sensitive heterogeneous ant - based models facilitate a balanced search process by endowing ants with different pheromone sensitivity levels translated into different search strategies .
an effective exploration of the search space is performed particularly by ants having low pheromone sensitivity while the exploitation of intermediary solutions is facilitated by highly - sensitive ants .
the performance of hybrid ant - based models is investigated with successful results for solving the np - hard generalized vehicle routing problem .
variable pheromone sensitivity in ant - based models proves to be benefic to the search process leading to better results compared to the ant colony system algorithm .
numerical results encourage the exploration of new ways to induce heterogeneity in ant - based models . +
this work was supported by cncsis uefiscsu , project number pnii idei 508/2007 _ new computational paradigms for dynamic complex problems_. p.c.pop , c.m.pintea , i.zelina and d.dumitrescu . _ solving the generalized vehicle routing problem with an acs - based algorithm_. american institute of physics ( aip ) springer , conference proceedings : bics 2008 , 1117(1):157162 , 2009 .
efficient transformations of the generalized combinatorial optimization problems into classical variants .
_ proceedings of the 9-th balkan conference on operational research , constanta , romania , 2 - 6 | the idea of sensitivity in ant colony systems has been exploited in hybrid ant - based models with promising results for many combinatorial optimization problems .
heterogeneity is induced in the ant population by endowing individual ants with a certain level of sensitivity to the pheromone trail .
the variable pheromone sensitivity within the same population of ants can potentially intensify the search while in the same time inducing diversity for the exploration of the environment .
the performance of sensitive ant models is investigated for solving the generalized vehicle routing problem .
numerical results and comparisons are discussed and analysed with a focus on emphasizing any particular aspects and potential benefits related to hybrid ant - based models .
ant - based models , optimization , sensitivity , complex problems |
the jordan - wigner transformation maps a pauli spin 1/2 system onto a ` hard - core ' boson model @xcite .
this leads to a bosonic description of the collective excitons whilst maintaining pauli s exclusion principle , which forbids double excitation of a single site . in our case
each ` spin ' represents one of @xmath0 identical optical emitters / absorbers and the spin s ` up ' / ` down ' projection along the @xmath86-axis denotes the presence / absence of an exciton on the respective site .
we assume a ring - like geometrical arrangement as shown in fig .
1a of the main paper .
considering a ring rather than a spin chain with free ends introduces the additional complexity of an alternating periodic or anti - periodic boundary condition depending on the number of excitations @xcite .
the jordan - wigner transformation is achieved by introducing bosonic creation and annihilation operators , which are defined in terms of the pauli spin raising and lowering operators as follows : @xmath87 here , the @xmath88 factors are traditionally referred to as ` strings ' , and their inclusion is necessary for producing the correct anti - commutation relation for fermions , which also apply to hard - core bosons : @xmath89 performing the transform ( [ eq : jw1 ] , [ eq : jw2 ] ) to the hamiltonian of the main paper given by eq .
( 1 ) yields : @xmath90 where @xmath91 counts the number of excitons on the ring .
the final term in eq .
( [ eq : ham1 ] ) has been introduced to implement the required alternating boundary condition @xcite .
note that this elegant and compact form of the hamiltonian double - counts the interaction term for @xmath92 , but this is of no concern here since we are only interested in @xmath93 .
clearly , the operator @xmath94 commutes with the hamiltonian ( [ eq : ham1 ] ) , so that the eigenvalues of @xmath94 are good quantum numbers .
the system can thus be partitioned into two parity subspaces of even and odd exciton number @xmath95 , each of which can be solved separately .
we complete the diagonalisation with the help of the fourier transform @xmath96 where the @xmath97 are phase factors defined by @xmath98 @xmath99{\includegraphics[width=2in]{figs / excitoncounts.pdf } } & \subfigure[]{\includegraphics[width=2in]{figs / excitoncountsfd.pdf } } & \subfigure[]{\includegraphics[width=2in]{figs / excitoncountsnp.pdf } } \end{array}$ ] each of the above @xmath0 distinct @xmath100-values corresponds to a single exciton state of the system .
the eigenstates of the full multi - exciton system are given by combinations of these states .
the eigenstates are thus defined by sets @xmath11 whose elements are the @xmath100 s , and where the number of elements in @xmath11 represents the number of excitons in the state ( @xmath101 ) .
the pauli exclusion principle manifests itself by allowing only different single exciton states to form a multiple exciton state .
the eigenvalues are then given by @xcite @xmath102 with corresponding eigenstates @xmath103 where @xmath104 is the ground state .
obviously , eq . ( [ eq : gen ] ) yields an eigenstate with @xmath95 excitations and each band comprises a total of @xmath105 possible states .
in its diagonalised form the hamiltonian eq .
( [ eq : ham1 ] ) is then simply given by @xmath106
the optical transition operator @xmath107 connects eigenstates differing by one exciton .
the transition rates are proportional to @xmath108 and can be explicitly evaluated with the help of slater determinants , albeit involving some tedious calculations @xcite , yielding : @xmath109 where @xmath110 is 1 if @xmath111 for integer @xmath112 and zero otherwise .
the indices @xmath113 inside the sums and products of the above expression refer to the elements of unique pairs of @xmath100-values composing @xmath11 or @xmath114 ( see @xcite for full details ) .
in fig . 2 of the main text we display the steady - state exciton population in the ratchets model as a function of phonon and photon temperatures , and also the relative population of the ratchets model over the forced dark ( fd ) scenario .
to complete the picture , in fig .
[ fig : steadyex ] we show the absolute values of the steady - state exciton count for all three scenarios . in the no phonons ( np ) case ,
as expected there is no dependence on the phonon temperature . indeed , since there is no dark - state protection in the np case , we find a lower exciton population than in the other two cases ; the difference is more pronounced at lower phonon bath temperature , when dark states are most likely to be populated . naively , one might expect the surface for the fd scenario to plateau at one , however , the ladder of bright states remains available in this case and via it population can access and get trapped in higher lying dark states above the single exciton band .
we expect this to happen for high optical and low but finite phonon temperature , and indeed in this region the surface rises above one .
note that unlike panels ( a ) and ( c ) in fig .
[ fig : steadyex ] , ( b ) starts at a phonon bath temperature of @xmath115 k due to numerical instabilities in our steady - state solver . at much lower phonon temperature , where phonon - assisted relaxation becomes effectively uni - directional , we expect the fd surface to drop towards one , as it gets increasingly unlikely for population to bypass or escape the single excitation subspace . using alternative methods ,
we have verified this is indeed the case ( not shown ) . )
plane of the figure . ] for the collective extraction model , enhancement of power output for ratchet states over np ( left , displayed as a multiplicative factor ) and fd ( right , displayed as a percentage ) , as a function of both hopping strength @xmath4 and extraction rate @xmath76 .
the simulations use the same model and parameters as fig .
[ fig : collabs ] . ]
our photocell cycle relies on the extraction of the excitons from the ring to the trap site .
below we give an explicit description of two different mechanisms and models , one of which features in the main text , the other whose results are displayed for comparison in this supplement . for the incoherent hopping process that we use in the main text
we assume that excitons are extracted incoherently from a single site of the ring to the trap .
we then simply include a dissipator : @xmath116 = \gamma_{x } { \cal d}[\sigma^{-}_1 \sigma^{+}_{t } , \rho ] \,.\ ] ] for optimal extraction the trap s energy is resonant with the desired extraction level , i.e. for the ratcheting and fd cases , the lowest state in the first exciton band , @xmath59 ; for the np case , the trap is instead made resonant with the bright state , @xmath117 . for consistency and a fair comparison across the different mechanisms ,
the energy of the trap site is also adjusted in the same way for the extraction mechanism described next .
we use the single site extraction process in the paper since it is relatively straightforward to implement .
however , it is not necessarily the most efficient so in this supplement we will also consider the possibility of extraction directly from particular eigenstates of the ring .
this can sometimes be achieved by suitable positioning of the trap : e.g. placing the trap at the ring s centre with equal coupling to all other sites allows extraction from symmetric states @xcite .
unfortunately , this does not straightforwardly generalise to the more complex anti - symmetric dark states .
for the special cases of @xmath118 alternating phases on the interaction terms achieves the desired coupling @xcite ; however , scaling up to more sites would require increasingly exotic molecular orbitals .
we therefore implement a collective state extraction method using a set of operationally motivated lindblad operators @xmath119 . in the case of the ratchets and the fd model , the @xmath119 move population from the lowest state of a particular band of @xmath120 into the lowest state in the exciton band below , exciting the trap in the process . in the np case , the @xmath119
go from the highest state of one band to the highest of the band below ( since these are the only states accessed in this model ) .
we incorporate this process into our model through the dissipator @xmath121 = \gamma_{x } \sum_{i } { \cal d}[\hat{c}_{i } , \rho ] \,,\ ] ] where , once again , @xmath122 is the effective extraction rate .
once on the trap , excitons are assumed to undergo charge separation at rate @xmath68 , which we model as an irreversible decay process .
power ( solid ) and its limit ( dashed ) as a function of voltage based on single site incoherent lindblad extraction .
three scenarios are shown , ratcheting ( blue ) , forced - dark ( fd , green ) and np ( red ) .
the power limit is given by the maximum possible voltage multiplied by the current @xmath123 @xcite .
model parameters are from table i of the main text .
each plot is found by varying @xmath68 independently for the different models .
the upper @xmath124-axis shows how @xmath68 varies in the ratchets model over many orders of magnitude for relatively small changes on the voltage scale . ]
[ fig : collabs ] shows the absolute power for the ratchets and np model for the collective trapping model .
this plot is equivalent to fig . 3 of the main paper which displayed the same quantities for the single site incoherent trapping model .
similarly , fig .
[ fig : collrel ] shows the _ relative _ power enhancement of ratcheting over both artificial models for the collective trapping model again this is the equivalent plot to fig . 4 in the main paper but here for the collective extraction model .
unsurprisingly , the absolute power extracted in the collective model is larger than the incoherent model .
this is because we are here extracting directly from the ratchet or dark states and thus have access to the ` entire ' exciton rather than just its projection onto a single site of the ring .
the relative advantage of ratcheting over fd has a similar form for both extraction mechanisms , though it is smaller for the collective model .
in figs . 3 and 4 of the main paper we are displaying maximum output power as a function of @xmath4 and @xmath76 . each point in those plots
is a result of generating a power - voltage curve for each choice of parameter , and finding the maximum power .
each such curve is generated by varying the trap decay rate @xmath62 .
we show examples of these power - voltage curves in fig .
[ fig : fig3old ] , considering the usual three scenarios : ratchets , fd and np . on the upper @xmath124-axis
we also show the values of @xmath62 corresponding to each voltage taking the example of the ratchets case ; as shown @xmath62 varies over orders of magnitude for only modest voltage shifts .
these curves make use of the parameters specified in table i of the main text , and a comparison with figs .
3 and 4 of the main paper shows that @xmath125 ev and @xmath126 ev are in fact not optimal for showing the full ratcheting advantage .
further examples of @xmath127 curves are shown in fig .
[ fig : bottleneck ] , each for different single site incoherent extraction rates , fixing @xmath125 ev .
we can see here the advantage of ratcheting in the intermediate and severe bottleneck regions .
for very fast extraction , any absorbed photon gets immediately transferred to the trap , so that neither dark - state protection nor ratcheting offer any advantage ( the system is already operating as efficiently as possible ) . in this case
np achieves the same current as ratchets and fd , but ends up producing a higher resultant power as its trap transition frequency and thus effective voltage is higher , as explained in sec .
[ sec : ext ] .
in fig . [ fig:2dbottle ]
we show several 2d sections of the surfaces in fig . 3 of the main paper and fig . [
fig : collabs ]
i.e. we show absolute output power as a function of @xmath4 ( for fixed @xmath128ev ) in the left panel and as a function of @xmath76 ( for fixed @xmath129ev ) in the right panel .
these are displayed for all three scenarios ( ratchets , fd , np ) and for both collective and incoherent extraction mechanisms .
( for fixed @xmath128ev ) .
right : absolute output power as a function of @xmath76 ( for fixed @xmath129ev ) .
curves for all three scenarios ratchets ( r ) , np , fd are displayed .
the two extraction mechanisms of incoherent single site ( inc ) and incoherent collective ( col ) are also shown .
all other parameters are as specified in table i of the main text . ]
to model disorder , we generate random site energies that are normally distributed around the transition frequency @xmath130 ev with standard deviation @xmath131 . the @xmath132 value pairs for each trap decay frequency @xmath62
are averaged over 100 different implementations , and are based on the bloch - redfield dissipator @xcite for the optical and phonon interactions .
results for different degrees of disorder are shown in the left of fig .
[ fig : disorder ] . for small levels of disorder @xmath133
, there is little deviation from the ideal case , and 100 implementations are indeed enough to give fairly smooth curves .
once @xmath131 approaches @xmath4 ( 0.02 ev corresponds to a @xmath131 of around 1.14% ) , however , not only is the achievable power with ratchets visibly reduced but our statistical ensemble is also not large enough to smooth out those larger performance variations , resulting in jagged - looking curves .
interestingly , the performance of the np case improves with increasing disorder .
this may be due to the fact that even in asymmetric circumstances eigenstates with a varied range of interesting optical properties may arise @xcite in the fully accessible hilbert space , and the absorption dynamics is no longer restricted to only the single lowest ladder transition .
we do not show the fd scenario here , as the presence of disorder makes it impossible to properly distinguish between allowed ` ladder ' and fd ` forbidden ' optical transitions .
however , we note that random sampling for small levels of disorder ( well below relative site energy fluctuations on the order of a percent ) where the symmetries and properties of system eigenstates remain largely unperturbed indicates that fd always performs worse than the full ratchets model , as expected . the right panel of fig .
[ fig : disorder ] shows the averaging from a single up to 1000 implementations of randomness for @xmath134 .
interestingly , the single run has some datapoints which can even exceed the ideal , fully ordered , ratcheting performance though the ensemble average power is clearly below that in the fully ordered case .
we suspect these outliers correspond to highly efficient asymmetric configurations : for the case of a dimer and dark - state protection it has recently been established that certain asymmetric arrangements can potentially outperform symmetric ones @xcite . however , the unravelling of the precise interplay of mechanisms at play will be more complicated in this larger system , but would be an interesting line of investigation for a future study .
note that we keep our extraction and trap model unaltered for our disorder calculations , i.e. we extract with a trap energy of @xmath135 for the np and ratchets case , respectively .
this simplifying choice may in general entail slightly over- or underestimating the effective voltage of our trap .
however , the relative voltage discrepancy arising from this assumption is expected to be limited by energy shifts due to disorder given by @xmath131 . as seen in both panels of fig .
[ fig : disorder ] , disorder in the model affects the power output for both ratchets and np much more significantly than that , meaning our results are clearly dominated by other processes as opposed to just being an artefact of our simplified trapping model .
we model non - radiative decay by including a full set of uni - directional lindblad lowering processes acting in the site basis ( as opposed to the collective optical decay events which are present due to the interaction with the shared electromagnetic environment ) .
increasing non - radiative decay rates significantly decreases the achievable current and power output of the photocell as shown in fig .
[ fig : non - rad ] .
interestingly , non - radiative decay does not affect all three scenarios equally , and the ratcheting advantage diminishes with larger non - radiative recombination rate .
the ratchets states enjoy no special protection against non - radiative decay acting on individual sites , and when they carry more than one excitation the impact of non - radiative decay is felt more acutely
. however , the combined ratcheting and dark state protection advantage persists as long as non - radiative decay does not become the dominant loss channel .
we now discuss the impact of exciton - exciton annihilation ( eea ) on the ratcheting efficiency enhancement .
eea is the result of a two step dynamical process @xcite .
first , excitons in adjacent sites on the ring can fuse together , creating an excitation on a single chromophore that has twice the energy of the first excited exciton state so far considered .
the mechanism for this is , like exciton transfer , frster coupling
i.e. a dynamic dipole - dipole interaction between the adjacent excitations .
the second step is fast non - radiative decay of the double energy state on the excited chromophore back down to the regular single exciton state .
the hamiltonian underlying the first step might be expressed by using an extension of the notation we have employed in the main paper .
we now allow three levels per chromophore : the already - considered ground state @xmath136 and low energy exciton state @xmath137 , as well as the newly introduced double energy excitation @xmath138 .
with these state labels the pauli operator we use in the main text is @xmath139 , and now we define a new spin raising operator @xmath140 .
the hamiltonian then becomes : @xmath141 where we have made the simplifying assumption that the two excitation energy transfer has the same strength ( @xmath4 ) as the single excitation one .
this is reasonable , implying a similar transition dipole from @xmath136 to @xmath137 as from @xmath137 to @xmath138 . by diagonalizing eq .
( [ eq : ham2 ] ) for an @xmath16 system , in a basis of states @xmath142 on each site , we are able to straightforwardly obtain the new eigenstates of the system . again , since the hamiltonian preserves total excitation number ( assuming that @xmath138 states count as two excitations ) , then the eigenstates fall into a series of bands , each corresponding to a different excitation number .
the zero excitation ground state and the four single excitation eigenstates are obviously unaffected by the @xmath138 states , but the second band now consists of ten , not six , possible states , which are now combinations of both @xmath138 states on single sites and @xmath137 states on two sites .
the lowest energy state in this second band ( let us label it @xmath143 ) would be reached following rapid relaxation following ratchet action from the first band .
we have calculated that @xmath143 contains only a 16% @xmath138 state character ( by probability ) .
considering that population of the @xmath138 state is crucial for eea to occur , we can conclude that eea will be significantly suppressed from @xmath143 .
moreover , we have calculated the electric transition dipole moment between @xmath143 and all states in the first band .
the dipole is zero for transitions into all but one of the first band states , and only 18% of the dipole of a single uncoupled chromophore for the remaining state .
by contrast , if the @xmath138 states are not included , the equivalent transition dipole is 41% of the single chromophore value .
thus , we can conclude that the lowest state of the second band is more protected from optical re - emission if the @xmath138 states are included in the model .
this means that including the possibility of doubly - excited sites surprisingly lends a certain degree of extra ` dark - state protection ' character to the second excitation band , allowing more time for exciton harvesting , and thus improving the efficiency of the device .
there are materials that do not quickly decay through non - radiative transition to the lowest energy state of a given band - these materials then do not obey _
kasha s rule
_ @xcite , which states that optical emission is always from the lowest energy state of a band . for these materials ,
the reasoning above regarding the composition of the lowest second band state is less relevant .
however , in that case , the second step involved in eea - the non - radiative decay of the @xmath138 state - is also inhibited , and this too leads to an inhibition of eea and so to more efficient energy harvesting devices . indeed
, finding materials that violate kasha s rule for novel optoelectronic applications such as this is a very active area of research @xcite . as discussed above ,
we expect the @xmath143 to have suppressed eea , but we do not expect it to disappear completely .
we therefore look at its impact by introducing eea it into our model by means of one - way lindblad operators taking population from all doubly or higher excited states to the highest energy ( ladder ) state in the single excitation subspace . in fig .
[ fig : annihilation ] we show power - voltage characteristics with such eea taken into account .
each plot compares the ratchets , fd and np models , for the value of the annihilation rate indicated above it .
as expected , neither the np case nor the fd scenario are visibly affected . as annihilation increases
, the maximal current achieved by the ratchets case eventually reduces to that of fd but , importantly , the ratcheting advantage only fully vanishes once the exciton - exciton annihilation rate exceeds the spontaneous emission rate by an order of magnitude or more . on close numerical inspection ,
fd and np also suffer very slightly under eea , because in both those cases the ladder of bright states remains available and there is thus slight chance of multiple excitation .
however , unlike in the case of non - radiative recombination , annihilation never kills the dark - state protection advantage .
in this section we mention some possible systems in which the action of ratchet states might be initially detected . ideally , we require a system composed of at least four coupled , radiation - absorbing elements , and where the spacing between each element is either much less than the wavelength of light , or else where the elements can be placed such that the radiation with which they interact is in phase with all of them .
this latter condition allows for the possibility of using systems placed an integer number of wavelengths apart , and would include , for example , superconducting qubits interacting with microwave cavity fields .
for example , two such qubits have been used to demonstrate super - radiance @xcite . including two more would enable ratchets states to be reached , initially by driving the bright excitation and then by altering single qubit phases through local manipulation .
a system closer in spirit to what we have described here would involve coupled optical absorbers . for example
, nv@xmath144 centers in diamond are established as systems on which to test quantum information protocols @xcite and fundamental features of quantum theory @xcite .
nvs can be individually addressed optically , and have spin selective coupling to light that allows optical measurement of their spin state .
moreover , they have been produced in patterns with a precision of around @xmath145 , where @xmath146 is the wavelength at which they emit @xcite .
the prospect of collective photon emission ( superrradiance ) with multiple nv centres has recently been studied @xcite .
nvs have strain dependent splitting so would have some disorder , though as we have seen in section [ sec : disorder ] above , this does not mean that ratchets states lose all function .
however , it may be that silicon vacancies ( sivs ) in diamond @xcite , or alternatively ( di)vacancies in sic , make better test systems .
for example , sivs in diamond can be placed with @xmath147 precision and are very stable , with inhomogeneous broadening heavily suppressed by their inversion symmetry .
for all these kinds of emitter , the frster coupling could be enhanced using a waveguide or cavity to promote the interaction of several centers with a common optical mode @xcite .
transient luminescence and absorption techniques could be used to track the light excitation in time as it moves through the system , and demonstrate behavior consistent with ratcheting .
quantum dots are another system that can exhibit strong optical coupling , and could for example be arranged in sub - wavelength patterns using dna origami @xcite .
evidence of collective photon emission of lateral quantum dot ensembles has also already been reported @xcite .
of course , the final device likely to be used in a real photocell would be a molecular system .
such systems can be made with high reproducibility , and such that they are very closely spaced with respect to their emission wavelength .
h - aggregates of chromophores have a positive frster coupling that shifts the bright ( absorbing ) state up in energy , as is the case in our model @xcite .
alternatively , light - harvesting @xmath148-conjugated supramolecular rings can be made with twelve or more repeating units @xcite , specifically to mimic the ring structures found in biology .
indeed , the symmetry of this structure generates a dark @xmath149 transition in the absence of vibrational distortion @xcite .
further design optimization of both the individual systems and rigidity of the ring could result in a system ready to meet the requirements for ratcheting . | natural and artificial light harvesting systems often operate in a regime where the flux of photons is relatively low . besides absorbing as many photons as possible
it is therefore paramount to prevent excitons from annihilation via photon re - emission until they have undergone an irreversible energy conversion process .
taking inspiration from photosynthetic antenna structures , we here consider ring - like systems and introduce a class of states we call ratchets : excited states capable of absorbing but not emitting light .
this allows our antennae to absorb further photons whilst retaining the excitations from those that have already been captured .
simulations for a ring of four sites reveal a peak power enhancement by up to a factor of 35 under ambient conditions owing to a combination of ratcheting and the prevention of emission through dark - state population . in the slow extraction limit the achievable power enhancement due to ratcheting alone exceeds 20% . _ introduction _ the absorption of light and prevention of its reemission is essential for the efficient operation of solar energy harvesting devices @xcite . from the many causes of device inefficiency ,
few are as fundamental and seemingly insurmountable as energy loss via radiative recombination : any absorption process must have a companion emission process .
this inherent absorption inefficiency is a result of the principle of ` detailed balance ' and constitutes a key contribution to the famous shockley - queisser limit @xcite .
however , pioneering work by scully showed that it is possible to break detailed balance , given an external source of coherence @xcite ; later work showed that this can be achieved by clever internal design alone , by using an optically dark state to prevent exciton recombination @xcite .
such dark states are populated passively if the energy separation between dark and bright states falls into the vibrational spectrum of the absorbing nanostructure : dissipation then preferentially mediates transfer into states from which optical decay can not occur .
this is thought to play a role in photosynthetic light harvesters @xcite , e.g. by means of dynamic localisation reducing the effective optical dipole strength @xcite .
however , time spent in dark states is ` dead time ' with respect to absorbing further photons . in organic light harvesting systems the time needed to extract an exciton from a dark state and turn it into useful energy is often orders of magnitude slower than typical light absorption rates @xcite , and this results in the loss of any subsequently arriving photons . in this letter
, we show that certain exciton states , which we will call _ ratchets _ , can enhance photocell efficiency by enabling absorption of these subsequent photons , while preventing emission . _ model _ we consider a ring of @xmath0 identical two - level optical emitters ( see fig . [ fig1 ] ) , with nearest pairs coupled by a transition dipole - dipole interaction , governed by the hamiltonian ( @xmath1 ) : @xmath2 where @xmath3 is the bare transition energy of the sites , @xmath4 is the hopping strength , the @xmath5 denote the usual raising and lowering operators which create and destroy an excitation on site @xmath6 , and @xmath7 . for a single exciton ,
the eigenstates are equal superposition of excitons localized at the different ring positions , with each being characterized by a relative phase @xmath8 with @xmath9 between exciton states in adjacent positions .
full diagonalization can be achieved by the successive application of the jordan - wigner and fourier transforms @xcite , whence the hamiltonian becomes @xmath10 .
@xmath11 describes the eigenstate with energy @xmath12 and each @xmath11 now corresponds to a set of the phase factors @xmath13 , one for each excitation [ see supplementary information ( si ) @xcite ] .
the eigenstates fall into a series of bands , where each state within a band contains the same number of excitons @xmath14 , with @xmath15 the number of elements in set @xmath11 .
left : @xmath16 ring structure with attached trap .
right : energy level diagram of this ring showing dipole - allowed optical transitions ( red ) .
solid lines denote primary absorption transitions for the ratcheting cycle , others are indicated with thinner dashed lines .
only relevant phonon transitions are shown ( blue ) .
rarely populated levels are shown in grey . ]
optical transitions connect eigenstates differing by one exciton with rates proportional to @xmath17 where @xmath18 . an explicit analytical expression for @xmath19
can be derived @xcite and is given in @xcite .
only the fully symmetric @xmath20 single exciton eigenstate has a dipole matrix element with the ground state , since here the transition dipoles interfere constructively .
the other @xmath21 single exciton eigenstates do not couple to the ground state via light with a wavelength that is much longer than the separation between sites .
however , these @xmath21 states do have dipole - allowed transitions to the second exciton band .
they therefore have the potential for preventing re - emission of light , while still allowing further absorption of photons , creating more excitons that in a photovoltaic circuit can be converted into useful photocurrent .
such states are examples of ratchet states @xmath22 , which have the simultaneous properties : @xmath23 where @xmath24 is the sum of all transition rates connecting the ratchet eigenstate @xmath22 to the adjacent band below ( above ) and @xmath25 is the kronecker symbol . for @xmath16 and @xmath26
the bright state in the single exciton band has the highest energy @xmath27 , and the three ratchet states have energies @xmath28 and @xmath29 , see fig .
[ fig1]b and @xcite . in a molecular system
, the ratchet states can be reached following excitation into the bright state when the exciton loses a small amount of energy to molecular phonons .
importantly , such vibrational modes have much shorter wavelengths than optical modes and can often be assumed to be local to each site . unlike the photon field , the phonon interaction then breaks the symmetry and enables intraband transitions . _
dynamics _ in order to model the dynamics we use the qutip package @xcite to construct full open system dynamics in bloch - redfield theory , for each kind of environmental interaction .
we always retain the full form of the resulting dissipator tensor since the secular approximation can not be applied when a system has bands of closely spaced levels which may be more closely spaced than the corresponding dissipation rates @xcite .
first , consider the dynamics due to the light - matter interaction hamiltonian @xmath30 where @xmath31 and @xmath32 are creation and annihilation operators for photons of wavevector @xmath33 .
the photon spectral density is constant across all transition energies , and enters into the master equation dissipator only via the single emitter decay rate @xmath34 .
absorption and stimulated emission terms are weighted by the appropriate bose - einstein factor @xmath35 -1)$ ] , representing the thermal occupancy of the optical mode with interband transition frequency @xmath36 and at temperature @xmath37 ( @xmath38 k for solar radiation ) .
second , the exciton - phonon interaction is accounted for through @xcite @xmath39 where for each site @xmath6 @xmath40 and @xmath41 are the creation and annihilation operators for phonons of wavevector @xmath42 with energy @xmath43 and exciton - phonon coupling @xmath44 .
this interaction commutes with @xmath45 and so only mediates transitions within each excited band . for simplicity
, we again assume a spectral density @xmath46 that is approximately constant across all intraband transition energies @xmath47 , such that @xmath48 .
we calculate appropriate phonon matrix elements for each transition , letting absorption and stimulated emission terms carry the appropriate bose - einstein factors @xmath49 , with @xmath50 the ambient phonon temperature .
reflecting the fact that phonon relaxation typically proceeds orders of magnitude faster than optical transition rates @xcite , we fix @xmath51 .
this separation of timescales allows population to leave the bright states before reemission occurs . to complete the photovoltaic circuit
, we include an additional ` trap site ' @xmath52 , with excited state @xmath53 and ground state @xmath54 , where excitons are irreversibly converted into work .
the trap hamiltonian is @xmath55 , where @xmath56 represents its transition frequency , and @xmath57 . henceforth , we shall consider the joint density matrix of ring system and trap , @xmath58 . throughout the paper we use
an incoherent hopping from a single site in the ring to the trap ( see si @xcite for an alternative model ) , and we fix the trap energy to be resonant with the state in which population is most likely to accumulate unless otherwise stated , this is at the bottom of the first exciton band , @xmath59 .
.default model parameters ; italic rows are plot parameters in certain figures , as stated in the relevant captions . [ cols="^,^,^ " , ] ( a ) steady - state exciton population and ( b ) relative ratchet enhancement compared to the fd scenario ( see text ) , both as functions of the optical and phononic bath temperatures , and without any trap .
the former is given in solar temperature units , @xmath60 k , other parameters are as in table [ tab1 ] except that @xmath61 . ] the theory of quantum heat engines @xcite provides a way for us to assess the current and voltage output of the ring device : the action of a load across the device is mimicked by the trap decay rate @xmath62 , which varies depending on the load resistance .
the current is simply @xmath63 , with @xmath64 the steady state population of the excited trap state .
the potential difference seen by the load is given by the deviation of the trap s population from its thermal distribution @xcite : @xmath65 where @xmath66 is boltzmann s constant and @xmath67 the ( ambient ) phonon bath temperature .
to study the relationship between current , voltage , and therefore power , we alter the trapping rate @xmath68 . letting @xmath69 or @xmath70 leads to the well - known open and short circuit limit , respectively , however , our main interest in the following is in the region near the optimal power output point of the device . _
results _ we proceed by finding the steady state of the following master equation which incorporates the optical and phononic dissipator as well as the trap decay process : @xmath71 + d_{o}[\rho ] + d_{p}[\rho ] + d_{t}[\rho ] + d_x[\rho]\ , , \end{aligned}\ ] ] where @xmath72 $ ] , @xmath73 $ ] are the bloch redfield dissipators for the photon and phonon fields discussed above , and @xmath74 = \gamma_t(\sigma^-_t \rho \sigma^+_t -\frac{1}{2}\{\sigma^+_t\sigma^-_t , \rho\})$ ] is a standard lindblad dissipator representing trap decay .
@xmath75 $ ] represents exciton extraction via incoherent hopping at rate @xmath76 from a single site on the ring to the trap ( see fig .
[ fig1 ] a ) . to expose the effect of the optical ratchets we now contrast the full model of eq . with two other artificial cases that exclude the possibility of ratchets : in the first
, we remove all phonon - assisted relaxation ( @xmath77 ) , meaning the system only undergoes optical transitions along the dicke ladder of bright states @xcite we term this ` no phonons ' ( np ) .
in the second construct , the ratchet states are rendered fully dark by setting all upwards transition matrix elements from them to zero , a scenario we refer to as ` forced dark ' ( fd ) .
all results in this section are based on the steady state of the system for @xmath16 , obtained by an iterative numerical method performed with qutip @xcite , and using the default parameters from table [ tab1 ] unless explicitly stated otherwise .
our choice of @xmath78 ev@xmath79 differs from that in previous works @xcite : our subtraction of @xmath80 ensures that the highest ( bright state ) level in the first band always remains at 1.8 ev and absorbs at a fixed frequency . in fig .
[ fig2 ] we display the steady - state exciton population of the system without trapping as a function of optical and phonon bath temperatures .
we compare the ratchet model to fd , finding that the ratchet enhancement is greatest in the hot photons , cold phonons regime .
this is to be expected : hot photons quickly promote the system up the excited bands via ratchets , with cold phonons allowing a one - way protection of gained excitation energy .
for fd states at low phonon temperature the exciton population plateaus near one , as most population ends up trapped in the state at the bottom of the first band ( see @xcite for the fd data and an extended discussion ) .
by contrast , the ratchet states keep absorbing , allowing the steady state population to rise toward the infinite temperature limit of @xmath81 ( for @xmath16 ) .
absolute power output in the ratcheting ( blue surface ) and np ( red surface ) cases as a function of ( single site ) extraction rate @xmath76 and hopping strength @xmath4 .
other parameters are as in table i. for each point , the optimal trap decay rate is found by numerical search .
the sets of curves projected onto the side walls of the 3d plot are cuts through the data for fixed values of the appropriate parameter .
the two surfaces cross in the region where the bottleneck is lifted , with ratcheting generating more power for a severe and intermediate bottleneck . the line at which the curves cross is projected onto the bottom ( @xmath82 ) plane of the figure . ] using the heat engine model , we can now explore the performance of the four site ring as an energy harvester . we will focus on the dependence of the power output on two parameters : the coupling @xmath4 , and extraction rate @xmath76 . for each parameter set , the trap decay rate @xmath62 is varied for maximum power output ; example power traces as a function of @xmath62 can be found the si @xcite . in fig .
[ fig3 ] we display the _ absolute _ value of the power generated for the ratchet and np scenarios , whereas in fig . [ fig4 ] we display the _ relative _ power output of ratchet states over each of the two artificial scenarios . to ensure fair comparison , for np we set the trap energy to be resonant with the bright state at @xmath83 , since the ratchet states at lower energy are inaccessible here .
all three scenarios have lower output power when the trapping ( charge separation ) rates are low , since this creates a bottleneck in the cycle and limits the size of the photocurrent . however , the ratchet states hold a great advantage over the other scenarios in this bottlenecked region , since this is precisely the situation in which excitations need to be held for some time by the ring before extraction is possible .
the np scenario only performs poorly in this case , since excitations in all likelihood decay before being extracted .
indeed , the relative power output of ratcheting over np , shown in the left panel of fig
. [ fig4 ] , therefore rises to as high as a factor of 35 .
the right panel of fig .
[ fig4 ] demonstrates the importance of ratcheting over dark state protection alone , with the ratchet model delivering up to 20 % better performance than fd . in fig .
[ fig4 ] we can see that advantage of ratcheting persists into a region where there is only a moderate bottleneck , i.e. up to @xmath84 . referring to fig .
[ fig3 ] we see that the ratchet power output in this region is already close to its maximum .
even so , if the extraction rate was arbitrarily tunable , then even a moderate bottleneck could be avoided completely and the main advantage of ratcheting removed .
however , in photosynthesis , creating a fast extraction rate carries with it a clear resource cost : whereas antenna systems can be comparatively ` cheap ' , reaction centres carry a much larger spatial footprint , being typically embedded in membranes and requiring significant surrounding infrastructure ( such as concentration gradients produced by proton pumps ) .
the severity of the bottleneck is likely to be inversely proportional to the number of reaction centres
. a photocell design exploiting ratcheting in the moderate bottleneck regime would be likely then to generate optimal power per unit volume of material and similar design principles will apply to artificially designed molecular light harvesting systems .
the choice of interaction strength @xmath4 is also important .
ratcheting achieves optimal results for @xmath85 ev in the bottlenecked region .
this dependence arises because the size of @xmath4 determines the gap between bright and dark or ratchet states . in turn , this controls the effective rate for ` upwards ' phonon - assisted transitions within excitation bands , as those rely on the absorption of a phonon and are proportional to @xmath49 .
consequently , larger values of @xmath4 entail more directed dissipation into the lower states of each band , boosting the occupation of ratchet states . however at the same time , increasing @xmath4 leads to a lower ratchet state energy and so a lower trap energy and hence to a voltage drop .
the trade off between these two competing influences leads to a maximum in ratchet performance .
as we discuss in detail in the si , ratcheting continues to convey an advantage in the presence of moderate levels of various real - world imperfections , such as site energy disorder , non - radiative recombination , and exciton - exciton annihilation @xcite .
enhancement of power output for ratchet states over np ( left , displayed as a multiplicative factor ) and fd ( right , displayed as a percentage ) , as a function of both hopping strength @xmath4 and extraction rate @xmath76 .
parameters are as in fig .
[ fig3 ] . ] _ conclusion _ we have investigated the light harvesting properties of coupled ring structures , inspired by the molecular rings that serve as antennae in photosynthesis . considering a vibrational as well as an electromagnetic environment allows the system to explore the full hilbert space rather than just the restricted subset of dicke ladder states .
we have shown that the off - ladder states possess interesting and desirable properties , which can be harnessed for enhancing both the current and power of a ring - based photocell device .
several possible systems @xcite could be used to observe the effect , from superconducting qubits @xcite to macrocyclic molecules @xcite .
our approach generalises existing concepts for dark state protection @xcite to arbitrary numbers of sites and importantly includes multi - exciton states , which introduce the ratcheting effect as a distinct additional mechanism for enhancing the overall light - harvesting performance .
the optical ratchet enhancement is particularly well suited to situations where exciton extraction and conversion represent the bottleneck of a photocell cycle . in future work
it would be interesting to explore combining optical ratcheting with enhancements of the primary absorption process , for example by exploiting the phenomena of stimulated absorption @xcite or superabsorption @xcite .
the authors thank william b. brown for insightful discussions .
this work was supported by the epsrc and the leverhulme trust .
bwl thanks the royal society for a university research fellowship .
emg acknowledges support from the royal society of edinburgh and the scottish government .
30 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , * * , ( ) . ,
* * , ( ) . , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.104.207701 . , , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.111.253601
. , , , , , * * , ( ) , http://dx.doi.org/10.1039/c4cp05310a . , p. ( ) .
, , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.117.203603 . , , , * * , ( ) , , http://dx.doi.org/10.1021/jp984469g . , in _
_ , edited by ( , ) , vol . of _ _ , pp . , isbn , http://dx.doi.org/10.1007/978-3-642-82688-7_7
.
, , , , , * * ( ) , http://dx.doi.org/10.1038/ncomms4286 . ,
* * , ( ) , , http://www.pnas.org/content/109/34/13498.abstract . , _ _ ( , ) , ed . , * * , ( ) . , , , * * , ( ) , http://link.aps.org/doi/10.1103/physrevb.47.2019 . _
_ , * * , ( ) , http://link.aps.org/doi/10.1103/physrevlett.67.3424 . ,
_ _ ( , ) . , , , , , , * * ( ) , http://dx.doi.org/10.1038/ncomms5705 . , , , ,
, * * ( ) , http://dx.doi.org/10.1038/ncomms5236 . , , , , , ,
, * * , ( ) , http://dx.doi.org/10.1039/c3cs60269a . , , , , * * , ( ) , , http://www.pnas.org/content/110/8/2746.abstract . , , , ,
, * * , ( ) , , http://www.pnas.org/content/108/37/15097.abstract . , , , * * , ( ) ,
issn , http://www.sciencedirect.com/science/article/pii/s0010465512003955 .
link:\doibase { 10.1103/physreva.94.012110 } [ * * , ( ) ] http://amazon.com/o/asin/0306463385/[__ ] , ed .
( , ) , , , , * * , ( ) , http://dx.doi.org/10.1038/ncomms6186 . , , , , , , , , , , , * * , ( ) , http://dx.doi.org/10.1038/nature09683 . , * * , ( ) . * supplemental information : quantum - enhanced capture of photons using optical ratchet states * |
recent observations show that the most massive ( @xmath8@xmath1 ) spheroid - like galaxies at @xmath9 , irrespective of their star formation activity ( prez - gonzlez et al .
2008 ) , were much smaller ( a factor of @xmath7 ) than their local counterparts ( daddi et al . 2005 ; trujillo et al . 2006 , 2007 ; longhetti et al . 2007
; zirm et al . 2007 ; toft et al . 2007 ;
giavalisco et al .
2007 ; ravindranath et al .
2008 , cimatti et al .
2008 , hereafter c08 ; van dokkum et al . 2008 ; buitrago et al . 2008 , hereafter b08 ;
saracco et al .
2009 , damjanov et al .
2009 , ferreras et al .
the near absence of such systems ( @xmath10kpc ; @xmath8@xmath1 ) in the nearby universe ( @xmath11% ; trujillo et al .
2009 ) implies a strong evolution in the structural properties of these massive galaxies as cosmic time evolves .
different scenarios have been proposed to explain the dramatic size evolution of these galaxies since @xmath12 ( khochfar & silk 2006 ; naab et al .
2007 ; fan et al . 2008
; hopkins et al . 2009 ) .
the main difference among them is the role of mergers to increase the size of galaxies .
fan et al .
( 2008 ) support an evolutionary scheme where galaxies grow by the effect of quasar feedback , which removes huge amounts of cold gas from the central regions hence quenching the star formation .
the removal of gas makes galaxies to puff up in an scenario similar to the one proposed to explain the growth of globular clusters ( hills 1980 ) . in the merging scenario ,
however , merger remnants get larger sizes than those of their progenitors by transforming the kinetic energy of the colliding systems into potential energy . whereas both scenarios predict a strong size evolution for the most massive galaxies , they disagree on the expected evolution of the velocity dispersion of the massive galaxies at a given stellar mass .
the merging scenario basically predicts no evolution ( at most a 30% since @xmath12 ; hopkins et al .
2009 ) , whereas the puffing up scenario predicts central velocity dispersions to be @xmath13 times larger than in present - day massive galaxies . constraining the evolution of the velocity dispersion of spheroid - like massive galaxies over the last 10gyr turns therefore crucial to test the above models of galaxy evolution , as well as to determine the importance of dark matter halos in setting the velocity dispersion of galaxies as cosmic time evolves . for these reasons , in this letter we measure for the first time the velocity dispersion of such massive galaxies at @xmath2 , and compare it with a compilation of velocity dispersions for similar galaxies at @xmath14 .
in what follows , we adopt a cosmology of @xmath15 , @xmath16 and h@xmath17kms@xmath4mpc@xmath4 .
to determine the typical velocity dispersion of spheroid - like massive ( @xmath0@xmath1 ) galaxies at @xmath2 , we have used the publicly available stacked spectrum of galaxies at @xmath18 ( @xmath19 r.m.s .
standard deviation ) presented in c08 as part of the gmass ( galaxy mass assembly ultra - deep spectroscopic survey ) project .
the spectrum consists of an averaged spectrum of 13 massive galaxies with a total integration time of 480h . only 2 of them are classified as disks , whereas the remaining 11 galaxies have either a pure elliptical morphology or show a very concentrated and regular shape ( see c08 ) .
the stacked spectrum is consequently representative of massive spheroid - like objects at that redshift .
individual galaxy spectra were taken at vlt with fors2 , using the grism g300i and a 1arcsec width slit , providing a spectral resolution of fwhm @xmath20 in the range @xmath21(@xmath22 ) . before stacking ,
each individual galaxy spectrum was previously de - redshifted and assigned to have the same weight in the @xmath23 rest - frame range .
the resulting rest - frame stacked spectrum was set to 1 / pix over the range @xmath24 .
aimed at computing the velocity dispersion of the above stacked spectrum , it is required the use of reference template spectra of well known spectral resolution .
section [ template ] provides a full description of these data - sets , as they are basic input ingredients of the analysis carried out in this work .
we have compiled from previous work velocity dispersions estimates of spheroid - like massive galaxies at @xmath26 .
this includes data from van der wel et al .
( 2005 ) updated , when possible , with van der wel et al .
( 2008) for galaxies in the ranges @xmath27 and @xmath28 , and from di serego alighieri et al .
( 2005 ) for galaxies at @xmath29 .
these authors also provide stellar masses and effective radii for their galaxies . in both samples
the galaxies were classified visually .
the vast majority of these objects present a prominent spheroidal component an are identified as either e or s0 .
when necessary , stellar masses have been transformed assuming a chabrier ( 2003 ) imf . to allow a meaningful comparison along the entire redshift interval explored in this paper , only galaxies with stellar masses in the range @xmath30 have been considered . to have a local reference , we have retrieved velocity dispersions , effective radii and stellar masses ( chabrier imf ) from the sdss nyu value - added galaxy catalog ( dr6 ; blanton et al .
2005 ; blanton & roweis 2007 ) for those galaxies within the above stellar mass range having srsic ( 1968 ) indices ( in the r - band ) @xmath31 . selecting objects with @xmath31
assures that the majority of the sources have a hubble morphological type t @xmath32 ( i.e. ranging from e to s0 ; ravindranath et al .
we have estimated the average velocity dispersion of the above galaxies in two redshift bins : @xmath33 and @xmath34 .
we have not tried higher redshifts to assure individual signal - to - noise ratios ( s / n ) larger than 10 .
ccccccc coolcat stars & @xmath35 & @xmath36 & @xmath37 & @xmath38 & @xmath39 & @xmath40 + bc03+ngsl ssp models & @xmath35 & @xmath36 & @xmath41 & @xmath38 & @xmath42 & @xmath43 + keck / lris stars & @xmath44 & @xmath45 & @xmath46 & @xmath47 & @xmath48 & @xmath49 + [ table ] to estimate the velocity dispersion of the c08 stacked spectrum we use the movel code , available at the reduceme distribution ( cardiel 1999 ) .
movel is designed to characterize the line - of - sight velocity distribution of a galaxy spectrum by determining its first two moments : the mean radial velocity and the velocity dispersion .
based on fourier analysis , the code relies on the movel and optema algorithms of gonzlez ( 1993 ) .
a full explanation of the fourier techniques and their particular implementation in both routines is provided in the above work .
the movel algorithm , an improvement of the classic fourier quotient method by sargent et al .
( 1977 ) , is an iterative procedure in which a galaxy model is processed in parallel to the galaxy spectrum . in this way
, a comparison between the input and recovered broadening functions for the model allows to correct the galaxy power spectrum from any imperfections of the data handling in fourier space .
the optema algorithm is designed to overcome the typical template mismatch problem by computing an optimal template for the galaxy spectrum .
it is fed with a set of representative template spectra which are scaled , shifted and broadened according to initial values for the mean line strength , the radial velocity , and the velocity dispersion .
then , the algorithm finds the linear combination of the template spectra that best matches the observed galaxy spectrum .
this constitutes a first composite template which is fed into the movel algorithm .
the output kinematic parameters are then used to create an improved composite template and the process is iterated until it converges .
this iterative approach then provides an optimal template while simultaneously computing the radial velocity and velocity dispersion of the galaxy spectrum .
ideally , the reference template spectra are chosen to be representative stellar spectra observed with an identical technical setup to that of the galaxy spectra .
this way , the instrumental broadening is the same in both data sets , and the broadening difference between galaxy and template spectra , @xmath50 , is a direct measurement of the galaxy starlight velocity dispersion , @xmath51 .
since this is not the case for the c08 stacked spectrum , the different instrumental resolutions of the distinct data sets employed in this work must be taken into account .
this is explained in detail in sections [ template ] and [ results ] .
we have restricted the kinematical analysis of the c08 spectrum to those regions containing strong spectral features from the galaxy starlight . because of the lack of available template spectra of high - enough spectral resolution covering the full spectral range of interest , we employ three different data sets .
two of them allow us to analyze the near ultraviolet ( nuv ) region at @xmath52 , which comprises e.g. several feii lines at @xmath53 , the @xmath54 mgii doublet and the @xmath55 mgi line .
the third template set is chosen to provide a complementary analysis from the region @xmath56 , with features such as the @xmath57 nh and @xmath58 cno bands .
the three sets of template spectra , whose properties are summarized in table [ table ] , are described below : \i ) first , from the coolcat database , an hst / stis echelle catalog of normal late - type stars with r @xmath59 ( ayres 2005 ) , we have selected eight spectra ( from f5v to g0iii spectral types ) covering the range @xmath60 with no gaps in between .
they have been re - binned at 1 / pix to match the linear dispersion of the c08 spectrum .
using movel and their corresponding high - resolution coolcat spectra as reference templates , we have derived that the spectral resolution of the re - binned templates is @xmath61kms@xmath4 at @xmath62 , which corresponds to r @xmath63 .
\ii ) the second set of template spectra consists of a subsample of nineteen ssp spectral energy distributions by g. bruzual ( private communication ) .
they are based on the bruzual & charlot ( 2003 ) code and the next generation spectral library , ngsl ( gregg et al . 2004 ; heap & lindler 2007 ) , with an average resolution of r @xmath64 .
the employed ssp models span ages from 0.3 to @xmath65gyr with metallicities [ z / h ] @xmath66 , @xmath67 , and @xmath68 , hence enclosing the 1gyr and solar metallicity estimate of c08 for their stacked spectrum . again , using movel and the above high - resolution coolcat spectra , we have accurately derived the spectral resolution of bruzual s models in the range @xmath60 , which is @xmath69kms@xmath4 , or equivalently r @xmath70 .
\iii ) finally , for the red side of the c08 stacked spectrum , we employ a set of eighteen stellar spectra ( dwarfs and giants ranging from g0 to k7 spectral types ) observed with lris at keck as radial velocity templates for a different program ( beasley et al .
having used the grism 600/4000 and the 1arcsec longslit , the stellar spectra have a fwhm of @xmath71 ( @xmath72kms@xmath4 at 3565 ; r @xmath73 ) as determined from the width of the wavelength calibration arc lines .
the spectra span the range @xmath74 , although only the blue side is employed here .
we have analyzed the stacked spectrum of c08 with movel using the three sets of template spectra described in section [ template ] . in each case , to constrain the reliability of the final results , the procedure has been repeated several times using different initial movel parameters , like e.g. different starting velocity dispersions ( from 50 up to 300kms@xmath4 , in steps of 50kms@xmath4 ) . for each single movel solution ,
the stellar velocity dispersion of the c08 stacked spectrum , @xmath75 , is derived as @xmath76 where @xmath50 is the broadening difference between the optimal template spectra and the c08 spectrum ( as derived from movel ) , @xmath77 and @xmath78 are their intrinsic instrumental resolutions , and @xmath79 is the redshift ( @xmath80 ) .
note that , since the individual galaxy spectra of c08 were previously blue - shifted before stacking , the effective instrumental resolution at rest - frame decreases by ( @xmath81 ) . for each set of template spectra , table [ table ] presents @xmath77 , @xmath78 , and the mean values of @xmath50 and @xmath75 , with their errors accounting for the r.m.s .
standard deviation of the solutions obtained from the different movel trials .
an error - weighted mean of the three independent @xmath75 solutions , which are indeed consistent within the uncertainties , gives @xmath82kms@xmath4 .
it is worth noting that the above value could be slightly overestimated , as redshift uncertainties for the 13 stacked galaxy spectra may have enlarged artificially the width of the lines .
we have also checked that the strong @xmath54 mgii doublet in the stacked spectrum ( @xmath83 , table 2 of c08 ) is not significantly broadened by potential non - stellar mgii absorptions in surrounding galactic gas clouds .
such absorptions are typically much narrower ( @xmath84kms@xmath4 , as they follow much colder kinematics ) and @xmath85 times weaker ( @xmath80 , for typically @xmath86 clouds ; see e.g. steidel et al .
1994 ; churchill et al .
therefore , even if they were present in all galaxies , @xmath87 could be overestimated by only @xmath88kms@xmath4 ( up to @xmath86kms@xmath4 if all galaxies were damped - ly@xmath89 systems , which is very unlikely as no lyman absorptions are detected ) . on the other hand ,
if the two disk galaxies in the stacked spectrum had lower typical velocity dispersions of e.g.@xmath90kms@xmath4 , the derived @xmath91 would underestimate the true velocity dispersion of early - type galaxies by @xmath20kms@xmath4 .
overall , since the above systematics may cancel each other , @xmath92kms@xmath4is probably a reasonable estimation of the velocity dispersion of @xmath0@xmath1 spheroid - like galaxies at @xmath2 .
figure [ fig1 ] illustrates representative movel solutions as derived from each template set . at the nuv , coolcat stars ( panel @xmath93 ) provide a slightly more broadened solution than the ngsl based ssp models ( panel @xmath94 ) , probably due to the intrinsic differences between the two template sets and their effects on the template mismatch problem . at longer wavelengths ( panel @xmath95 ) ,
the galaxy spectral features look somewhat blurred , probably by sky line residuals commonplace of the near - infrared region at observed frame .
still , in all cases the solutions reproduce nicely the c08 spectrum . to further check the reliability of the movel results
, we carried out monte carlo simulations in which we added gaussian noise ( to match s / n = 5 to 30@xmath4 ) to artificially broadened reference template spectra ( @xmath96kms@xmath4 , in steps of 50kms@xmath4 ) from our three template sets .
for such _ fake _ galaxy spectra , we run the movel procedure in the same way as we did for the c08 stacked spectrum . as expected
, the accuracy in the final solution decreases with the decreasing s / n of the problem spectrum .
the typical scatter of the derived @xmath50 solutions averaged over all the simulations and initial movel conditions are 55 , 24 , 12 , and 8kms@xmath4 for s / n = 5 , 10 , 20 , and 30@xmath4 respectively .
more importantly , for all s / n values , the recovered mean @xmath50 is statistically consistent with the assumed input value , hence supporting the reliability of both the method and @xmath75 .
the fact that the velocity dispersion of @xmath0@xmath1spheroid - like galaxies at @xmath2 is @xmath3kms@xmath4has important consequences to understand their evolution : * it confirms that high - z spheroid - like massive galaxies are truly massive objects .
the unexpected strong size evolution of these objects has cast some doubts about the reliability of their stellar mass estimates , which relies on the assumption that the imf is the same at all redshifts .
in fact , an appropriate change of the imf with redshift would mitigate the problem of the strong size evolution .
however , the derived velocity dispersion at @xmath2 is similar to that of present - day galaxies with @xmath97@xmath1 , consequently constraining any potential change of the imf with redshift . *
it alleviates the problem of understanding how massive and compact galaxies at high - z can evolve through merging since that epoch .
an extraordinarily high velocity dispersion at high - z ( e.g. @xmath98kms@xmath4 ) would have implied that the gravitational potential depth of the system would be so intense that it would not easily evolve in size . to put in context the above result
, figure [ fig2 ] shows the sizes ( top panel ) and velocity dispersions ( bottom panel ) for the c08 galaxies and the compilation of spheroid - like galaxies of similar mass described in section [ thedata ] .
these galaxies follow nicely the observed strong size evolution found in other independent larger samples where completeness effects are accounted ( trujillo et al . 2007 ; b08 ;
dashed line ) , hence probing that the sizes of the objects explored in this paper are typical of average spheroidal - like massive galaxies at those redshifts .
it is also clear that , although massive spheroid - like galaxies have experienced a strong size evolution ( a factor of @xmath7 since @xmath99 ; trujillo et al .
2007 ) , the velocity dispersion has evolved mildly by a factor of @xmath100 in the same redshift interval ( see bernardi 2009 for a similar trend at @xmath101 ) .
this result is crucial to constrain the different scenarios proposed so far to explain the dramatic size evolution since @xmath102 , as they disagree in the amount of evolution expected in their velocity dispersions . in the puffing up scenario of fan et al .
( 2008 ) , velocity dispersions change as @xmath103 . using the observed size evolution of b08 , @xmath104 , the solid line in fig .
[ fig2 ] illustrates the expected @xmath51 evolution under this scenario .
it increases with redshift in a way that galaxies are expected to double their @xmath51 at @xmath99 .
the observed velocity dispersions are compatible within the error bars with this scenario up to @xmath105 .
however , at @xmath106 the discrepancy between theory and data is apparent . in the merging scenario of hopkins et al .
( 2009 ) , at a fixed stellar mass , @xmath51 evolves with redshift as @xmath107 where @xmath108 ) is the dark matter contribution to the central potential relative to that of the baryonic matter at @xmath109 , and @xmath110 is the effective radius of the halo .
again , we have used the b08 size evolution as an input into the merging model prediction . the grey area in fig .
[ fig2 ] encloses the expected @xmath51 evolution when @xmath111 varies between 1 and 2 .
the agreement between the merging scheme and the observed evolution looks reasonably good at all redshifts .
following hopkins et al .
( 2009 ) , to explain why @xmath51 has only changed weakly since @xmath102 it is necessary to consider that the observed velocity dispersion is driven by two components : the baryonic matter and the dark matter halo .
both components contribute linearly to the central gravitational potential of the galaxy , hence @xmath112 ) . assuming that @xmath113 evolves weakly with time ( most simulations show that halos build inside - out , so the central potential is set first and just the outer halo grows with time ) , the dark matter effect on the central potential of the galaxy ( i.e. on @xmath51 ) basically remains unchanged at a fixed @xmath114 .
however , the influence of the baryonic matter on the gravitational potential has changed strikingly since @xmath99 , at present - day being @xmath7 times smaller due to the expansion of the object . the relative influence of the dark matter on setting the inner potential increases with the decreasing effect of the baryonic matter . in fact , the data look in agreement with an almost symmetric influence of dark and baryonic matter ( i.e. @xmath115 ) in setting the central gravitational potential of present - day objects .
we acknowledge the referee for constructive comments .
this work has been possible thanks to the helpful contribution of several people .
g. bruzual provided us with his ssp model predictions based on the ngsl .
fruitful discussions with p. snchez - blzquez , a. vazdekis and c. conselice helped us to find the right strategy to achieve our goals .
n. cardiel gave us very valuable inputs on the movel capabilities .
finally , we benefited from discussion with r. guzmn , m. balcells , m. lpez - corredoira , and a. van der wel .
a.j.c . and i.t.c
are juan de la cierva and ramn y cajal fellows of the spanish ministry of science and innovation .
this work has been funded by the spanish ministry of science and innovation through grant aya2007 - 67752-c03 - 01 .
ayres , t. r. 2005 , 13th cambridge workshop on cool stars , stellar systems and the sun , 560 , 419 beasley , m. a. , cenarro , a. j. , strader , j. , & brodie , j. p. 2009
, , accepted , astro - ph .
bernardi , m. 2009 , arxiv:0901.1318 blanton , m. r. , et al .
2005 , , 129 , 2562 blanton , m. r. , & roweis , s. 2007 , , 133 , 734 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 buitrago , f. , trujillo , i. , conselice , c. j. , bouwens , r. j. , dickinson , m. , & yan , h. 2008 , , 687 , l61 ( b08 ) cardiel , n. 1999 , ph.d .
thesis , universidad complutense de madrid chabrier , g. 2003 , , 115 , 763 churchill , c. w. , mellon , r. r. , charlton , j. c. , jannuzi , b. t. , kirhakos , s. , steidel , c. c. , & schneider , d. p. 2000 , , 543 , 577 cimatti et al .
2008 , 482 , 21 ( c08 ) daddi , e. , et al .
2005 , , 626 , 680 damjanov , i. , et al .
2008 , arxiv:0807.1744 di serego alighieri , s. , et al .
2005 , , 442 , 125 fan , l. , lapi , a. , de zotti , g. , & danese , l. 2008 , , 689 , l101 ferreras , i. , lisker , t. , pasquali , a. , khochfar , s. , kaviraj , s. , 2009 , arxiv:0901.4555 gregg , m. d. , et al .
2004 , bulletin of the american astronomical society , 36 , 1496 giavalisco , m. , ravindranath , s. , daddi , e. , 2007 , ncimb , 122 , 1209 gonzlez , j. j. , 1993 , phdt , university of california heap , s. r. , & lindler , d. j. 2007 , from stars to galaxies : building the pieces to build up the universe , 374 , 409 hills , j. g. 1980 , , 235 , 986 hopkins , p. f. , hernquist , l. , cox , t. j. , keres , d. , & wuyts , s. 2009 , , 691 , 1424 khochfar , s. , & silk , j. 2006 , , 648 , l21 longhetti , m. , et al .
2007 , , 374 , 614 naab , t. , johansson , p. h. , ostriker , j. p. , & efstathiou , g. 2007 , , 658 , 710 prez - gonzlez , p. g. , trujillo , i. , barro , g. , gallego , j. , zamorano , j. , conselice , c. j. , 2008 , , 687 , 50 ravindranath , s. , et al .
2004 , , 604 , l9 ravindranath , s. , daddi , e. , giavalisco , m. , ferguson , h. c. , dickinson , m. e. , 2008 , iaus , 245 , 407 saracco , p. , longhetti , m. , & andreon , s. 2009 , , 392 , 718 sargent , w. l. w. , schechter , p. l. , boksenberg , a. , & shortridge , k. 1977 , , 212 , 326 sersic , j. l. 1968 , cordoba , argentina : observatorio astronomico , 1968 , steidel , c. c. , dickinson , m. , & persson , s. e. 1994 , , 437 , l75 toft , s. , et al .
2007 , , 671 , 285 trujillo , i. , et al .
2006 , , 373 , l36 trujillo , i. , conselice , c. j. , bundy , k. , cooper , m. c. , eisenhardt , p. , & ellis , r. s. 2007 , , 382 , 109 trujillo , i. , cenarro , a. j. , de lorenzo - caceres , a. , vazdekis , a. , de la rosa , i. g. , & cava , a. 2009 , , 692 , l118 van der wel , a. , franx , m. , van dokkum , p. g. , rix , h .- w . , illingworth , g. d. , & rosati , p. 2005
, , 631 , 145 van der wel , a. , holden , b. p. , zirm , a. w. , franx , m. , rettura , a. , illingworth , g. d. , & ford , h. c. 2008 , , 688 , 48 van dokkum , p. g. , et al .
2008 , , 677 , l5 zirm , a. w. , et al .
2007 , , 656 , 66 | making use of public spectra from cimatti et al .
( 2008 ) , we measure for the first time the velocity dispersion of spheroid - like massive ( @xmath0@xmath1 ) galaxies at @xmath2 . by comparing with galaxies of similar stellar mass at lower redshifts
, we find evidence for a mild evolution in velocity dispersion , decreasing from @xmath3kms@xmath4 at @xmath2 down to @xmath5kms@xmath4 at @xmath6 .
such mild evolution contrasts with the strong change in size ( a factor of @xmath7 ) found for these type of objects in the same cosmic time , and it is consistent with a progressive larger role , at lower redshift , of the dark matter halo in setting the velocity dispersion of these galaxies .
we discuss the implications of our results within the context of different scenarios proposed for the evolution of these massive objects . |
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.