article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
our sun is known to act as a gravitational lens .
the angular shift of the apparent position of a star located behind the sun was first observed during the solar eclipse of 1919 @xcite , in conformity with the predictions of general relativity @xcite .
many years later , einstein ( 1936 ) discussed the lens like action of the sun s gravitational field more generally .
the influence of that discussion has caused the ring seen by the observer during perfect alignment of the source , lens and the observer to be called the einstein ring around a point lens .
for an opaque lens , the result of exterior focusing of the parallel light rays incident with an impact parameter @xmath1 on the lens plane is a deflection @xmath2 .
the deflection @xmath2 is inversely proportional to the impact parameter @xmath1 according to the formula first derived by einstein ( 1911 ) @xmath3 where @xmath1 is the impact parameter of the passing light ray , @xmath4 is the deflecting mass , @xmath5 is the gravitational constant and @xmath6 is the speed of light in vacuum . assuming the deflection angle is small , the minimum focal length for the opaque sun is approximated by @xmath7 where @xmath8 and @xmath9 are the total mass and radius of the sun , respectively .
the uncertainty in @xmath10 is based on statistical errors on @xmath11 and @xmath12 @xcite .
slight offsets of the lens or source from the optic axis will break the einstein ring into two bright images as seen by the observer .
therefore the opaque sun produces multiple images only beyond @xmath10 .
the possibility of using the _ transparent _ sun as a gravitational lens was first discussed by cyranski and lubkin ( 1974 ) and more recently by nemiroff and ftaclas ( 1997 ) .
demkov and puchkov ( 2000 ) computed a minimum focal length of the transparent sun to be about 24 au .
starting with the standard solar model as it was known in 1989 @xcite , they computed the gravitational lens deflection angle as a function of impact parameter and then approximated this with a taylor s series near the sun s center . throughout the present analysis , the more recent 2005 standard solar model ( ssm ) data
is used @xcite . a more complete model of the gravitational lens characteristics of the transparent sun
is computed , including a more accurate minimum focal length ( 23.5 @xmath13 0.1 au ) , regions of multiple images , and the locations of caustics and critical curves .
the plan of the paper is as follows : @xmath14 2 outlines two alternative approaches to obtain the minimal focal length of the transparent sun .
the strength of a composite lens capable of producing multiple images is given a through treatment in @xmath15 3 .
the critical curves , caustics , magnification and multiple image zones are computed in @xmath15 4 , and conclusions summarized in @xmath15 5 .
it is well known that only a point lens can generate two images in the weak deflection limit ( schneider , ehlers & falco 1992 ) . however
, a composite lens can produce multiple images .
the location of images in the lens plane depends on the enclosed projected mass within the impact radius .
burke s odd number theorem @xcite states that a composite lens can produce only odd number of images of the source , and the images always appear ( disappear ) in pairs as the source moves inside ( outside ) of a caustic . in the standard geometrical optics approximation
, light rays start out straight from a source and then change directions discretely at the deflector plane , subsequently reaching the observer . to utilize geometric optics for a composite lens ,
one must project the three - dimensional mass of the deflector onto a plane perpendicular to the optical axis connecting the source to the observer . a plane wave approximation is assumed by considering source radiation that has a wavelength @xmath16 , where @xmath17 , is the schwarzschild radius of the deflecting lens mass .
it is also assumed that the sources are point like and there is no contribution due to internally converging ricci focusing by the mass within a pencil of rays . in other words the light bundle is assumed to be infinitesimally thin and the deflection is purely due to weyl focusing by the projected mass lying within the impact radius @xcite .
the mass outside of the impact radius does not contribute to the deflection , following newton s and birkhoff s theorem .
is at a distance @xmath18 from the observer @xmath19 and a distance @xmath20 from the source @xmath21 . @xmath22 and @xmath23 are the angles subtended by the unlensed source and its image visible to the observer . ] the geometry of fig .
[ schematic ] satisfies the equation @xmath24 if one considers sources that are very far away , then @xmath25 .
hence the lens equation takes a simplified form @xmath26 in terms on the two dimensional projected density @xmath27 of the lens equation ( [ def ] ) can be recast in the form @xmath28 where @xmath29is the value of the projected mass density at the point @xmath30 from the center of the lens , @xmath30 is the impact parameter , and @xmath21 the surface area in a polar ( @xmath31 ) coordinate system .
the center of the lens is made to coincide with the center of the preferred coordinate system for simplicity and owing to the cylindrical symmetry . defining a dimensionless parameter @xmath32 where @xmath33 is the critical density , a reference measure to determine the strength of the lens from an observer s stand point , see for example ( subramanian & cowling 1986 ; schneider et al .
1992 ) : @xmath34 where @xmath18 is the distance between the lens and the observer . using equation ( [ kappa ] ) and
equation ( [ critden ] ) in equation ( [ lensden ] ) and scaling the radial coordinate using the relation , @xmath35 , yields @xmath36 where @xmath37 is the angular position of the image , as seen by the observer located at a distance @xmath18 , corresponding to an impact radius @xmath30 . a scalar potential @xmath38 can be defined such that , @xmath39 where @xmath40 is the polar angular coordinate . comparing equation ( [ scalarpot ] ) with equation ( [ lensden2 ] ) and noting that @xmath41 , @xmath42 a generalized lens equation can be obtained by using equation ( [ scalardef ] ) in equation ( [ lens ] ) : @xmath43 essentially equation ( [ sol ] ) is a mapping from @xmath44 , as the observer traces the ray back to the source .
the numerical solution of equation ( [ sol ] ) yields the image locations .
the @xmath45 potential of the lens plane , then appears in the lens equation as elements of the inverse metric @xmath46 @xcite .
@xmath47 is a function of @xmath48 ( radius ) only , owing to the circular symmetry . in general
it is a function of two parameters of a polar coordinate system . in other words , the source and the image locations
do not align on a straight line . and the image locations @xmath23 for three different scenarios .
the continuous curve represents the solutions of the lens equation for a centrally condensed lens .
the disjoint set of curves delineates a point lens of the same mass and the straight line corresponds to the absence of a lens , where unlensed source positions and image positions coincide at all times . ]
the solutions to the lens equation have a maximum of three roots .
any vertical line in the @xmath49 plane corresponds to a fixed source location , the points of intersection of which are the corresponding image locations .
inspection of fig .
[ betatheta ] shows that sources far from the optic axis are seen as a single image .
the @xmath49 curves are conformal transformations for the mapping @xmath44 for every observer location @xmath18 .
note that the fig .
[ betatheta ] is based on an inverted map that is used to calculate the magnification . at @xmath50
the two images are equally separated from the optic axis .
the assumed spherical symmetry of the lens will generate a circular einstein ring on the lens plane .
the formation of a real einstein ring requires that a source be placed exactly along the line joining the observer through the center of the lens , which , for realistic cases , only occurs for sources of finite size .
nevertheless , an einstein ring still has theoretical significance since it separates sets of images @xcite .
for example , as the source moves behind the lens , no source image will ever be seen to cross an einstein ring . along the optic axis .
the curve ( @xmath51 ) corresponds to the sun and curve ( @xmath1 ) corresponds to a homogenous lens with a radius and mean density equal to the sun .
the angle subtended at the limb of the sun in both cases converge to @xmath52 it is important to make a distinction between what is meant by focal length for an optical lens as apposed to a gravitational lens .
the focal length of an optical lens is the distance between the center of the lens and the point where paraxial light rays converge or appear to diverge .
the definition of deflection angle is @xmath53 where @xmath1 , @xmath54 and @xmath2 are the impact parameter , focal length and deflection angle respectively .
the gravitational lens has a focal length for each impact parameter bounded by a minimum focal length .
this is because the deflection angle is not a linear function of impact parameter .
paraxial light rays converge at different points along the optic axis .
the existence of multiple images is a necessary condition for a consistent definition of focal length . past the minimum focal length , the observed merger of two source images results in the formation of an einstein ring
. numerical simulations ( discussed more in the appendix ) involving different values of @xmath18 and burke s odd number theorem will impose the following mathematical condition , for the existence of multiple images @xcite : @xmath55 the least value of @xmath18 that conforms to the above condition , the minimum focal length , is found numerically to be @xmath56 au .
this implies that , for values of @xmath18 less than @xmath56 au there are no multiple images and hence no focal point .
the perfect alignment of the source , lens and the observer at a distance less than @xmath56 au will not result in the formation of einstein ring .
not only does the sun s mass density decrease monotonically with radius , but its mass density projected onto a lens plane also decreases monotonically with radius .
surface density normalized to the critical surface density needed to generate multiple images is referred to as a dimensionless surface density @xmath57 and is a measure of the strength of the lens @xcite . if @xmath58 for _ some _ regions on the lens then the lens is termed strong . " for instructive comparisons to our sun , a constant density " sphere of radius @xmath9 with a volume density equal to the average density of the sun is used .
a weak " lens is characterized by @xmath59 and can not produce multiple images @xcite .
the constant density sphere does not produce multiple images at @xmath60 au .
this is illustrated in fig .
[ critden ] , where curve ( @xmath1 ) has a value of @xmath59 throughout the entire the lens at 23.5 au .
weak lenses are only weak for observers sufficiently close
observers further than some minimal focal length will see the same object as a strong lens that creates multiple images . in case of the constant density sphere of solar mass
, this value for @xmath18 is @xmath61 au .
the validity of equation ( [ multiplecond ] ) implies that , the transition of the lens from being weak to strong should have at least so many points , such that the @xmath62 curve is continuous at the minimum focal length .
therefore , more than a single method is required to establish and verify the existence of a minimum focal length more precisely , accounting for the less than one percent combined error of the data @xcite and simulations . plotted on the vertical axis
is a monotonically decreasing function of radius .
the curves ( a ) corresponds to the sun for observer
lens distance @xmath63 au , whereas curves ( b ) and ( c ) represents the homogeneous density sphere at @xmath18 of @xmath64 and @xmath65 au .
the dimensionless surface density parameter suggests that the transparent sun is a strong lens just above @xmath56 au and a homogeneous model lens is a weak lens @xmath66 at the same @xmath18 .
the very same model lens is capable of producing multiple images at observer lens distance ( @xmath18 ) greater than @xmath67 au . ] alternatively , the minimum focal length of the sun can be obtained by using the values of the mass enclosed and the impact parameter corresponding to the value of @xmath68 . using equation ( [ focal ] ) and the substituting the value @xmath69 and the radius of the enclosed mass @xmath70 one obtains a focal length of @xmath71 au , which lies between the orbits of uranus and neptune .
the lens plane is defined as the plane perpendicular to the line joining a point source and a fixed observer . according to the odd number theorem for transparent gravitational lenses @xcite
, the observer will see an odd number of images of the point source no matter what position the center of the lens occupies in the lens plane .
areas might exist in the lens plane where the lens center can be placed to create a specific odd number of images visible to the observer .
the boundaries between these areas are called critical curves . for a spherically symmetric lens ,
a critical curve is a circle .
were a lens center to lie on a critical curve , a formally infinite magnification of a point source would be seen by the observer .
the source plane is defined as a plane perpendicular to the line extending from a fixed observer through the lens center . in analogy with the lens plane
, areas might exist in the source plane where a point source can be placed so the ( fixed ) lens creates a specific odd number of images visible to the ( fixed ) observer .
the boundaries between these areas of the source plane are called caustics . for spherical lenses ,
the caustics are also circles . were a point source to lie on a caustic , a formally infinite magnification would be seen by the observer . for a given lens , the @xmath23
versus @xmath22 curve may show points of diverging slope .
these points correspond to infinite magnification , and hence yield the angular radii of the corresponding caustic ( @xmath22 ) and critical ( @xmath23 ) circles .
let s now consider specifically our sun .
the @xmath23 versus @xmath22 curve is shown in fig .
[ betatheta ] . at 30 au , just outside the minimum focal length of 23.5 au , the radius of the caustic
is found to be 10 @xmath72 where @xmath73 , the angular einstein ring radius . in terms of the dimensions of the sun ,
the radius of the einstein ring is .1 @xmath9 and the caustic is .01 @xmath9 .
similarly the radius of the critical curve is found to be 50 @xmath72 or @xmath74 .06 @xmath75 .
table [ radii ] summarizes the relative radii of einstein ring , caustic and critical curve for different observer locations .
@xmath76 is the angular measure of the sun s radius for the corresponding @xmath18 .
lrrrrr 23.0 & 0 & 0 & 0 & 0 & 0 + 25.0 & 5.62 @xmath77 & 3.68 @xmath77 & 0.481 & 2.07 @xmath78 & 2.70 @xmath77 + 27.0 & 8.24 @xmath77 & 6.14 @xmath77 & 0.518 & 5.06 @xmath78 & 4.28 @xmath77 + 30.0 & 0.113 & 0.100 & 0.519 & 1.13 @xmath77 & 5.87 @xmath77 + [ radii ] along the horizontal axis . ] the apparent relative motion of an angularly small source behind the solar lens will result in a light curve with sharp spikes when the source crosses a caustic .
the magnification at a given location can be obtained from the @xmath49 curve corresponding to that location @xmath79 , expressed as an angular measure of @xmath80 . as the distance separating the source and the lens increases , a pair of images disappears .
the vertical axis is the magnification in logarithmic scale and the horizontal axis represents the distance scale normalized to angular size of the einstein ring @xmath81 . ]
the straight line motion of the source behind a weakly acting lens will result in a light curve which peaks at the minimum distance the source appears from the lens .
this minimum angular distance is called the impact parameter , @xmath82 for the source . the distance separating the projected source on the lens plane and the center of the lens scales as @xmath22 and the corresponding distance for the single image scales as @xmath23 for small @xmath22 and @xmath23 . for large distances of the source from the center of the lens ,
the image nearly coincides with the source .
but , as this distance decreases , the image and the source locations on the lens plane diverge as seen by the observer . for a strong lens ,
unlensed source positions inside a circular caustic create three images of the source visible to the observer .
regardless of the number of observed source images , a finite size source will undergo a finite total magnification .
large magnifications allows us a more sensitive look for neutrino flux from nearby stars , as well as high energy gravitational radiation .
the gravitational radiation with the wavelength of @xmath83 will be lensed by the sun @xcite . in order to satisfy plane wave approximation and to account for diffraction effects , the characteristic length scale ,
@xmath84 schwarzschild radius of the sun puts an upper limit on the wavelengths of detectable gravitational wave spectrum .
high energy neutrinos , by definition , have energies greater than 100 gev@xcite .
the sun admits neutrinos , unimpeded by electron scattering , only for energies up to 300 gev@xcite .
so , unlike gravitational waves , the neutrino spectrum is primarily low energy .
the mass distribution of the sun within the impact radius that allows for a minimum focal length , puts an upper limit on the neutrino energy spectrum .
the effects of plasma and core rotation in the interior of the sun , thought to be small , were ignored during this analysis .
the minimum focal length , the critical curve and caustics were computed numerically .
the magnification of point sources for typical source separations was analyzed .
the code can be modified to simulate galaxies and dark matter halos with an assumed density profile as input parameters . to date , four spacecraft have traveled past the minimum transparent focal length of our sun , just beyond the orbit of uranus . launching a spacecraft to this distance with sufficiently sensitive gravitational wave or neutrino detectors remains a dream , however . in a different paper , the possibility of detecting strong gravitational radiation and a wide range of neutrino energy spectrum
will be examined in some detail . with the advancement of modern technology and the improvement in the resolution power of detectors ,
such an endeavor could shed new light on sources emitting neutrinos , high energy gravitational waves or hitherto hypothetical particles that are predictions of at least some theories .
the image locations for a given impact parameter are found by solving equation ( [ lens ] . by allowing for a range of both positive and negative values for the deflection angle @xmath2 for all positive values of source position @xmath22 , and interpolating within bounds ,
one obtains the image locations @xmath23 .
therefore , this can be termed as a controlled ray tracing algorithm @xmath85 where @xmath1 , and @xmath18 are the impact parameter and fixed observer distance from the lens center .
the resulting curve is shown in fig .
[ betatheta ] . for point sources
the magnification can be computed directly from the curves using equation ( [ mag ] ) .
magnifications for extended sources can be computed by taking the average magnifications of individual randomly generated points , lying within a circle or by choosing neighboring points and taking the ratio of their areas in the lens plane and source planes . for a circularly symmetric lens , the deflection depends only on the impact parameter . in that case
, a new angular measure @xmath82 can be defined as in fig .
[ angles ] , the perpendicular distance from the source on the lens plane . the observer can choose a coordinate system with her @xmath86 coordinate and origin made to coincide with @xmath87 through the center of the lens . now at different source positions , as seen by the observer , the source will subtend an angular measure @xmath88 from the @xmath89 axis , where @xmath90 . ) behind the lens , making an impact parameter @xmath82 .
the image ( @xmath91 ) and the source always lies along a straight line for a spherical symmetric lens .
therefore , the angular distance of the source from the center of the lens along the source trajectory , as seen by the observer at any given time is @xmath92 bahcall , j. n. 1989 , neutrino astrophysics , ( cambridge : cambridge university press ) bachall , j. n. et al .
2005 , , 621 , l85 burke , w. l. 1981 , , 244 , l1 cox , a. 1999 , astrophysical quantities ( 4th ed . ;
new york : springer ) cyranski , j. f. & lubkin , e. 1974 , annals of physics , 87 , 205 demkov , yu . n. & puchkov , a. m. 2000 , , 61 , 083001 - 1 dyer , c. c. 1977 , , 180 , 231 dyson , f. w. , eddington , a. s. & davidson c. 1920 , philos .
royal soc .
london , 220a , 291 einstein , a. ( 1911)1994 , in the collected papers of albert einstein , vol .
3 ( princeton : princeton university press ) , 379 einstein , a. ( 1916)1997 , in the collected papers of albert einstein , vol . 6 ( princeton : princeton university press ) , 147 einstein , a. 1936 , science , 84 , 506 escribano , r. , frere , j. m. , monderen , d. & van elewyck , v. 2001 , phys .
lett . , b512 , 8 gaisser , t.k . ,
halzen , f. & stanev , t. 1995 , phys .
, 258 , 173 ; erratum : 1996 , 271 , 355 isaacson , r. a. 1968 , phys .
166 , no . 5 , 1263 narayan , r. & bartelmann , m. 1996 , preprint ( astro - ph/9606001 ) nemiroff , r. j. 1993 , american journal of physics , 61 , issue 7 , 619 nemiroff , r. j. & ftaclas , c. 1997 , baas , 190 , 38.01 schneider , p. , ehlers , j. , & falco , e. e. 1992 , gravitational lenses ( berlin : springer - verlag ) subramanian , k. & cowling , s. a. 1986 , mnras , 219 , 333 thorne , k. s. 1989 , in 300 years of gravitation , ed . hawking , s. & israel , w .
( cambridge : cambridge university press ) , 330 | the transparent sun is modeled as a spherically symmetric and centrally condensed gravitational lens using recent standard solar model ( ssm ) data .
the sun s minimum focal length is computed to a refined accuracy of 23.5 @xmath0 au , just beyond the orbit of uranus .
the sun creates a single image of a distant point source visible to observers inside this minimum focal length and to observers sufficiently removed from the line connecting the source through the sun s center .
regions of space are mapped where three images of a distant point source are created , along with their associated magnifications .
solar caustics , critical curves , and einstein rings are computed and discussed .
extremely high gravitational lens magnifications exist for observers situated so that an angularly small , unlensed source appears near a three - image caustic .
types of radiations that might undergo significant solar lens magnifications as they can traverse the core of the sun , including neutrinos and gravitational radiation , are discussed . |
the traditional way for a firm to repurchase its own shares is through open - market repurchase programs ( omr ) .
however , as reported in @xcite , after they have announced they intend to buy back shares , a substantial number of firms do not commit to their initial plan .
unexpected shocks on prices or on the liquidity of the stock may indeed provide incentives to slow down , postpone , or even cancel repurchase programs . in order to make a credible commitment to buy back shares , an increasing number of firms enter accelerated share repurchase contracts with investment banks .
accelerated share repurchase contracts are of several kinds and we traditionally distinguish between asr with fixed number of shares and fixed notional asr . in the case of an asr with fixed number of shares ( @xmath1 ) ,
the contract is the following : 1 . at time @xmath2
, the bank borrows @xmath1 shares from shareholders ( ordinarily institutions ) and gives the shares to the firm in exchange of a fixed amount @xmath3 where @xmath4 is the mark - to - market ( mtm ) price of the stock at time @xmath2 .
the bank then has to progressively buy back @xmath1 shares on the market to give back @xmath1 shares to the initial shareholders , and go from a short position to a flat position on the stock ; 2 .
the bank is long an option with payoff @xmath5 ( the firm being short of this option ) , where @xmath6 is the average price between @xmath7 and @xmath8 ( in practice the average of closing prices or the average of daily vwaps ) and where @xmath9 is chosen by the bank , among a set of specified dates , during which the option can not be exercised . ]
@xmath10 .
hence , the firm eventually pays on average the price @xmath11 for each of the @xmath1 shares that have been repurchased . in the case of a fixed notional asr with notional @xmath0
, the contract is the following : 1 . at time @xmath2
, the bank borrows @xmath1 shares from shareholders usually @xmath12 , where @xmath13 $ ] is usually around @xmath14 , and gives the shares to the firm in exchange of the fixed amount @xmath0 ; 2 .
an option is embedded in the contract : when the bank exercises this option at date @xmath9 ( the date is chosen by the bank , among a set of specified dates @xmath10 ) , there is a transfer of @xmath15 shares between the bank and the firm is chosen small enough so that the transfer is almost always from the bank to the firm , _
i.e. _ @xmath16 . in
what follows , we implicitly assume that the settlement is from the bank to the firm . ] , so that the actual number of shares obtained by the firm is @xmath17 . hence , the firm pays @xmath0 and eventually gets @xmath17 shares .
as above , the bank will also progressively buy shares on the market to give back @xmath1 shares to the initial shareholders . pricing these contracts and hedging them are mathematical problems that can not be solved in a satisfactory way using the classical theory of derivatives pricing . at time @xmath2
, the bank has a short position on the stock and it will buy shares on the market over a period of a few weeks to a few months . as a consequence ,
the problem is at the same time a problem of optimal execution and a problem of option pricing : a cash - settled exotic option with bermudean exercise dates in the case of an asr with fixed number of shares and a physically - settled exotic option with bermudean exercise dates in the case of a fixed notional asr .
ignoring the interactions between the two problems would lead to a suboptimal strategy and to mispricing .
moreover , since the nominal / notional of these options is large , execution costs must be taken into account and we propose a framework coming from the literature on optimal execution to price and execute optimally these contracts .
the problem of pricing derivative products with large nominal has already been tackled in the literature using the tools of optimal execution .
rogers and singh @xcite considered execution costs that are not linear in ( proportional to ) the volume executed , but rather strictly convex to account for liquidity effects ( as opposed to the literature on pricing with transaction costs see for instance @xcite ) .
they considered an objective function that penalizes both execution costs and mean - squared hedging error at maturity .
they obtained , in this close - to - mean - variance framework , a closed - form approximation for the optimal hedging strategy of a vanilla option when illiquidity costs are small .
also , li and almgren @xcite , motivated by saw - tooth patterns recently observed on several us stocks ( see @xcite ) , considered a model with both permanent and temporary impact .
in their model , they assumed execution costs are quadratic and they considered a constant @xmath18 approximation . using another objective function , they obtained a closed form expression for the hedging strategy of a call option .
finally , guant and pu proposed in @xcite a method to price and hedge a vanilla option in a utility - based framework , under general assumptions on market impact .
in particular , they considered both the case of physical settlement and the case of cash settlement .
the case of asr contracts is however more complex than the case of vanilla options , because the payoff is not a classical european payoff , but rather an exotic one with asian and bermudean features , and because the problem to be solved is both a problem of option pricing / hedging and a problem of optimal execution .
economists have discussed the role of asr contracts among the different ways to buy back shares ( see for instance @xcite ) .
the mathematical literature on accelerated share repurchase contracts is however rather limited and very recent .
the first paper on the topic is an interesting paper by jaimungal , kinzebulatov and rubisov @xcite .
the authors focus on asr contracts with fixed number of shares . for finding the optimal buying strategy
, they propose a model in continuous time with the following characteristics : stock prices are modeled with a perturbed geometric brownian motion ( the perturbation accounts for permanent market impact ) , execution costs are quadratic as in the original almgren - chriss models @xcite , the range of possible exercise dates is @xmath19 $ ] ( hence the product is american rather than bermudean ) .
the optimal strategy is found in @xcite for an agent who is risk - neutral , although the authors add inventory penalties as in @xcite .
the main interest of @xcite is that the authors manage to reduce the problem to a 3-variable pde , whereas the initial problem is in dimension 5 . in particular , they show that the exercise boundary only depends on the time to maturity and the ratio between the stock price and its average value since inception .
the case of asr contracts with fixed number of shares has also been considered in a different model by guant , pu and royer ( see @xcite ) . in this discrete - time model , a risk - averse agent
is considered in an expected utility framework .
permanent market impact is considered as in @xcite , but stock prices are assumed to be normal rather than log - normal .
also , a general form of execution cost is allowed as in @xcite , and exercise dates are among a finite set of dates . as in @xcite , the model in @xcite
boils down to a set of equations with 3 variables : time , the number of shares to be bought , and the difference between the current stock price and the average price since time @xmath2 .
an original fast tree - based numerical method is proposed in @xcite .
the model we propose in this paper is inspired from @xcite .
in particular , we use the same expected utility framework .
however since we focus on the specific case of fixed notional asr contracts , the reduction of the problem to a 3-dimension one is not possible anymore . in particular , the original tree - based method proposed in @xcite to approximate
numerically the optimal strategy can not be used .
we present in section 2 the framework of the model without permanent market impact . as in @xcite ,
the model is in discrete time and we end up with a characterization of the optimal buying / hedging strategy with recursive bellman equations . in section 3 ,
we introduce these bellman equations , along with a change of variables to reduce the dimensionality of the problem ( from 5 to 4 ) .
we also introduce the indifference price of the asr contract . in section 4
, we present a numerical method involving trees and splines to approximate the solution of the problem , along with examples .
the introduction of permanent market impact is presented in appendix a.
we consider the pricing and hedging of a fixed notional asr from the point of view of an investment bank buying back shares for a client ( a firm ) . throughout the paper , the notional of the asr contract
will be denoted by @xmath0 and the maturity of the contract will be denoted by @xmath20 .
the model we consider is a discrete - time model where each period of time ( of length @xmath21 ) corresponds to one day . in other words ,
if the interval @xmath19 $ ] corresponds to @xmath22 days , we assume that decisions are made at times @xmath23 . in order to introduce random variables , we introduce a probability space @xmath24 , where the filtration satisfies the usual assumptions .
all random variables will be defined on this probability space .
as far as prices are concerned , we start with an initial price @xmath25 and we consider that the dynamics of the price is given by : @xmath26 where @xmath27 is @xmath28-measurable and where the random variables @xmath29 are assumed to be i.i.d . with mean @xmath7 , variance @xmath30 .
we also assume that these variables have a moment - generating function defined on @xmath31 .
depending on the exact payoff of the considered accelerated share repurchase contract , @xmath32 can be regarded as the daily vwap over the period @xmath33 $ ] ( @xmath34 ) , or the closing price of the @xmath35 day .
depends on what @xmath36 stands for . ]
the average price process entering the definition of the payoff in the asr contract is denoted by @xmath37 : @xmath38 in order to buy shares , we assume that the bank sends every day an order to be executed over the day . at time @xmath39 , the size of the order sent by the bank is denoted by @xmath40 where the process @xmath41 is assumed to be adapted .
hence , the number of shares bought on the market to give them back to the initial shareholders is given by : @xmath42 the price paid by the bank for the shares bought over @xmath43 $ ] is assumed to be @xmath44 plus execution costs ; models vwap , we assume implicitly that the bank has sent a vwap order . if @xmath36 models closing prices , we implicitly assume that a target close order has been sent by the bank . ] the execution costs being modeled by a function @xmath45 satisfying the following assumptions : * @xmath46 , * @xmath47 strictly convex , even and increasing on @xmath48 , * @xmath47 is asymptotically superlinear , _
i.e. _ : @xmath49 the cash spent by the bank is modeled by the process @xmath50 defined by : @xmath51 where @xmath52 is the market volume over the period @xmath43 $ ] .
we assume that the process @xmath53 is @xmath54-measurable . in applications
, @xmath47 is often a power function , i.e. @xmath55 with @xmath56 , or a function of the form @xmath57 with @xmath58 . in other words ,
the execution costs per share are of the form @xmath59 where @xmath60 is the participation rate .
the coefficients for this form of temporary market impact function has been estimated in @xcite by almgren et al . for the model to be realistic
, we also assume that the bank can not buy / sell too many shares compared to the available liquidity .
in other words , we impose that @xmath61 , where @xmath62 can be of either sign .
we then introduce a non - empty set @xmath63 corresponding to time indices at which the bank can choose to exercise the option ( note that @xmath64 because the bank has no choice at time @xmath20 : the settlement is done at time @xmath20 if it has not been done beforehand ) . in practice , this set is usually of the form @xmath65 where @xmath66 .
when the bank decides to exercise the option , or at maturity if the option has not been exercised , the settlement of the contract occurs .
we denote by @xmath67 the time index corresponding to the settlement . at time @xmath68
, the bank delivers @xmath69 shares to the firm and then the bank has to return to a flat position .
overall , the bank will have to buy @xmath70 shares after time @xmath68 . as this problem is a pure optimal execution problem that only depends on the past through @xmath71
, we can assume that the cost of buying these shares is of the form @xmath72 where the function @xmath73 represents a risk - liquidity premium to account for the execution costs and the risk in the execution process after time @xmath68 ( see remark 2 below and @xcite ) .
the optimization problem the bank faces is given by : @xmath74,\ ] ] where @xmath75 is the set of admissible strategies defined as : @xmath76 and where @xmath77 is the absolute risk aversion parameter of the bank .
[ l ] to be consistent with the expected utility framework , the penalty function @xmath73 should be linked to the indifference price of a block of @xmath78 shares ( which is indeed of the form @xmath79 ) .
expressions for @xmath73 have been given in a continuous time model in @xcite .
if we assume that execution after @xmath68 is at constant participation rate , we can also consider for @xmath73 the risk - liquidity premium associated with a liquidation at participation rate @xmath60 ( see @xcite ) : @xmath80 where @xmath81 is the average value of the market volume process .
this is the function we consider throughout this article see also @xcite .
solving this problem requires to determine both the optimal execution strategy of the bank and the optimal stopping time . for that purpose ,
we use dynamic programming .
we introduce the value functions @xmath82 defined by : @xmath83@xmath84,\ ] ] where @xmath85 is the set of admissible strategies at time @xmath39 defined as : @xmath86 and where the state variables are defined for @xmath87 and @xmath88 by : @xmath89 in fact , when @xmath90 , the function @xmath91 is independent of @xmath92 .
it is then straightforward to prove that the value functions are characterized by the recursive equations of proposition [ rec ] : [ rec ] the family of functions @xmath93 defined above is the unique solution of the bellman equation : @xmath94 where @xmath95.\end{aligned}\ ] ] our goal will now be to simplify the problem so that a solution can be computed numerically . by `` solution '' , we mean two different things : ( i ) an optimal strategy for both optimal execution and optimal stopping and ( ii ) a price for the asr contract . for the second problem
we use the concept of indifference price ( see below ) . following the classical economic fact that the current wealth of an agent has no impact on its decision - making process
when he has a constant absolute risk aversion , we introduce the change of variables : @xmath96 the functions @xmath97 are then characterized by the recursive equations of proposition [ theta ] : [ theta ] @xmath98 satisfies : @xmath99 where @xmath100 as above for @xmath101 , @xmath102 is independent of @xmath92 , and we write @xmath103 instead of @xmath104
. this change of variables along with the associated recursive equations show that the dimension of the problem can be reduced from 5 to 4 .
indeed , the variable @xmath105 does not appear anymore in the functions @xmath106 . in @xcite , and
similarly in @xcite , the authors manage , in the case of an asr contract with fixed number of shares , to reduce the dimensionality of the problem from 5 to 3 .
they consider indeed respectively the ratio @xmath107 and the spread @xmath108 instead of the couple @xmath109 . in the case of a fixed notional asr ,
this dimension reduction is not anymore possible due to the additional convexity in the payoff ( the term @xmath110 ) , and the problem to be solved is in dimension 4 . to find the optimal strategy ,
the method is first to solve recursively the equations for @xmath106 .
then , when @xmath111 , the option is exercised if @xmath112 . in the case where the option is not exercised ,
the optimal strategy consists in sending an order of size @xmath113 , where @xmath114 is a minimizer in the definition of @xmath115 .
in addition to a reduction in the dimension of the problem , the main interest of the change of variables introduced earlier lies in the link between the functions @xmath106 and the indifference price of the asr contract .
the indifference price of the asr contract is , by definition , the amount of cash ( positive or negative ) to give to the bank so that it is indifferent between accepting and not accepting the asr contract .
mathematically , it is defined by : @xmath116@xmath117\geq -1 \big\}.\ ] ] the link between the price and the functions @xmath106 is given by the following proposition : [ price ] @xmath118 when @xmath119 is positive , it means that the execution costs and the cost of the risk borne by the bank are larger than the potential gain for the bank associated with the option embedded in the contract . on the contrary , if @xmath119 is negative , the bank values the option sufficiently high to compensate the execution costs and the risk associated with the execution strategy . in practice , when a firm wants to buy back shares , competition occurs between banks willing to enter the deal ( banks for which @xmath120 ) .
however , instead of proposing rebates in cash that would correspond to at most @xmath121 , banks propose a discount @xmath122 on the average price .
the optimization problem is then : @xmath123.\ ] ] from a mathematical point of view , this problem is very similar to the original one .
however , the maximum rebate @xmath124 defined as @xmath125 @xmath126 \ge -1 \right\rbrace,\ ] ] can not be computed directly .
the functions @xmath106 also enable to understand the different effects at stake when it comes to the execution strategy and the choice of the stopping time .
for that purpose , let us expand the expression for @xmath127 ( we omit superscripts for the sake of simplicity ) .
as proved in appendix b , we have : @xmath128 to understand the different effects , one has to distinguish between what happens before the option is exercised and after the option is exercised . before time @xmath68
, the bank buys back shares on the market . as for all execution problems
, there is an execution cost component and a risk component : ( i ) the bank pays execution costs depending on its participation rate to the market ( this is `` liquidity term i '' ) , and ( ii ) the bank is exposed to price moves . however , due to the form of the payoff , the risk is partially hedged .
when prices go up , the bank buys shares at higher prices but the number of shares to buy ( @xmath110 ) decreases . similarly the other way round ,
when prices go down , the bank buys shares at lower prices but the number of shares to buy increases .
contrary to what happens with asr contracts with fixed number of shares , the hedge is only partial in the case of a fixed notional asr : `` risk term i '' can not be set to 0 using an adapted strategy .
after time @xmath68 , the execution costs are similar but the risk is not anymore partially hedged : this is the `` post - exercise risk - liquidity term '' ( it requires to choose @xmath73 properly , see remark 2 ) .
the above analysis is linked to the execution process used for buying back shares .
coming to the option , the bank has an incentive to exercise it when @xmath36 is below @xmath92 , since @xmath110 will mechanically go up , as @xmath92 goes towards @xmath36 .
the cost of not exercising now is somehow given by the `` spread term '' .
we see therefore that there is a trade - off between exercising when @xmath36 goes below @xmath92 in order to eventually deliver less shares , and not executing too soon in order to ( partially ) hedge the risk associated with the execution process .
this implies accelerating the buying process when @xmath36 decreases in order to have less to hedge , and decelerating the buying process ( or even selling shares ) when @xmath36 increases .
to solve the problem numerically , we introduce a tree - based method .. ] this method uses a tree to model the diffusion of prices . for each node in the tree , indexed by a time index @xmath129 and a price @xmath36
, we shall compute the values of the function @xmath130 on a grid .
we consider that innovations @xmath131 have the following distribution : & = & 0\\ \mathbb{e}\left[\epsilon_{n}^{2}\right ] & = & 1\\ \mathbb{e}\left[\epsilon_{n}^{3}\right ] & = & 0\\ \mathbb{e}\left[\epsilon_{n}^{4}\right ] & = & 3 \end{array}\right.\ ] ] ] @xmath132 we introduce , for @xmath133 , and for @xmath134 the function @xmath135 defined by : @xmath136 following proposition [ theta ] , the family of functions @xmath137 satisfies : @xmath138 where @xmath139}\logexpo q\sigma\sqrt{\delta t}\epsilon_{n+1}-l\left(\frac{v}{v_{n+1}}\right)v_{n+1}\delta t\\ & & -\theta_{n}^{\zeta+\left(\epsilon_{n+1}+2\right)}\biggl(q+v\delta t,\frac{n}{n+1}a\\ & & \quad+\frac{1}{n+1}\left(s_{0}+\sigma\sqrt{\delta t}\left(\zeta-2n\right)\right)+\frac{1}{n+1}\sigma\sqrt{\delta t}\epsilon_{n+1}\biggr)\logexpc.\end{aligned}\ ] ] the functions @xmath140 are computed recursively on a grid .
each index @xmath141 corresponds to a node of the pentanomial tree . at each node of the tree , we compute the values of @xmath143 for @xmath144 , where @xmath145 is a grid of the form @xmath146 and where @xmath147 is a grid of the form @xmath148 . for each grid , we need to choose the number of points in the grid ( @xmath149 and @xmath150 ) and the width of the grid ( linked to @xmath151 and @xmath152 ) .
the number of shares to deliver is linked to the ratio @xmath110 .
consequently , it is not bounded and we need to choose a minimal value for @xmath92 . since @xmath153\simeq\dfrac{1}{3}n\sigma^{2}\delta t$ ] , it is natural to consider values of @xmath92 in a range of the form @xmath154 $ ] where @xmath155 is the number of standard deviations we want to keep . in practice , we choose @xmath156 , that is around @xmath157 standard deviations . then , it is natural for @xmath151 , in coherence with the lower bound for @xmath92 , to consider a value @xmath158 . + in our numerical method , we consider the optimization of the execution strategy on the grid , that is : @xmath159\cap\mathcal{g}_{q}}\ ] ] @xmath160@xmath161 in practice , if @xmath162 is a constant process , it is better to have @xmath163 and @xmath164 on the @xmath78-grid . for computing the minimum in the previous equation , and hence for computing @xmath165 , we need the values of @xmath166 for @xmath167 as @xmath168 does not necessary lie on the grid @xmath147 , we use interpolation with natural cubic splines and , if necessary , we extrapolate linearly the functions outside of the domain ( for @xmath92 ) .
+ using the above recursive equations , we can approximate the values of the functions @xmath106 . moreover , at each step in time and for each combination of the tuple @xmath169 for which we computed the value
, we can easily compute the optimal strategy .
we obtain @xmath170 from the minimization procedure that permits to compute @xmath165 .
if @xmath171 and if we are allowed to exercise , then we exercise the option . otherwise , we send an order of size @xmath172 to the market .
we now turn to the practical use of our numerical method on examples .
we consider the following reference case , that corresponds to rounded values for a major stock of the cac 40 index .
this case will be used throughout the remainder of this text .
* @xmath173 . *
@xmath174 @xmath175 , which corresponds to an annual volatility approximately equal to @xmath176
. * @xmath177 stocks@xmath178 @xmath179 . * @xmath180 with @xmath181 @xmath182 and @xmath183 .
* @xmath184 days . *
the set of possible dates for early exercising the option is @xmath185\cap \mathbb n$ ] .
* @xmath186. * participation rates are bounded by @xmath187 and @xmath188 . * risk aversion is @xmath189 @xmath190 .
* @xmath73 is as described in remark 2 . in other words , after the option has been exercised , we assume that execution occurs at constant participation rate of @xmath191 . * @xmath192 stocks . *
* @xmath194 , corresponding to @xmath195 standard deviations .
* @xmath196 .
we first start with three examples corresponding to three trajectories for the price .
the first example , on figure [ ref_up ] , corresponds to an increasing trend and the stock price is therefore above its average .
as explained in the previous section , there is no reason in that case to early exercise the option as @xmath92 increases ( the dot corresponds to the date when the option is exercised ) .
the second example , on figure [ ref_mid ] , corresponds to a stock price oscillating around its average value .
we see , as expected , that the buying process accelerates when the stock price decreases and decelerates when the stock price increases .
we even see that the buying process can turn into a selling process as prices increase .
the rationale for that is hedging , and it can be seen on the term we called `` risk term i '' : @xmath197 . as prices increase , we expect the number of shares to be eventually bought ( that is @xmath198 ) to decrease and the exercise date ( _ i.e. _ @xmath68 ) to be later than initially thought
. therefore , the term @xmath199 decreases and @xmath78 has to go down in order to reach the same level of risk .
we also see on this second example that the option is exercised near maturity .
as the optimal strategy leads to buying and selling shares , it is interesting to understand what happens when we force the bank to use a buy - only strategy .
if we impose @xmath200 , we obtain the optimal strategy of figure [ constr_mid ] that is substantially different . )
.,scaledwidth=75.0% ] the third trajectory we consider corresponds to decreasing prices ( see figure [ ref_down ] ) . in that case
, the bank buys shares quite rapidly and exercises the option as soon as it can , although it has not yet bought the required number of shares . by exercising the option , the bank wants in fact to avoid the natural decrease in @xmath92 that would lead to an increase in the number of shares it would have to buy .
in addition to optimal strategies , we can compute in our reference case , the price @xmath119 of the fixed notional asr contract . here , this price is negative as we found @xmath201 .
this means , in utility terms , that the gain associated with the optionality component of the asr contract is important enough to compensate ( in utility terms ) the execution costs and the risk of the contract . in the case where we impose the use of a buy - only strategy ,
we obtain @xmath202 . in order to understand the role of the main parameters , we carry out comparative statics .
we focus on the liquidity of the stock , through the parameter @xmath203 , on the volatility of the stock price , through @xmath204 , and on the risk aversion parameter @xmath77 .
regarding execution costs , we consider our reference case with 3 values for the parameter @xmath203 : @xmath205 and @xmath206 .
we concentrate on the second price trajectory ( figure [ eta_mid ] ) as it shows very well the role of @xmath203 : the more liquid the stock , the larger the number of round trips on the stock for hedging purposes .
.,scaledwidth=75.0% ] the differences in terms of prices are the following : [ cols="^,^,^,^",options="header " , ]
this paper is a contribution to a new literature on pricing inspired from the literature on optimal execution .
specifically , we have presented in this article a model for characterizing and computing the optimal strategy of a bank entering a fixed notional asr contract with a firm .
our discrete - time model enables to model the interactions between the execution problem linked to an asr and the asian / bermudean option that is part of the contract , whereas classical approaches would separate the two questions .
in addition to providing an optimal strategy , we define an indifference price for the contract .
the numerical method we developed for approximating solutions works very well in practice and provides both optimal strategies and prices .
in the setup presented above , market impact was only temporary and boiled down to execution costs . here
, we briefly show how to generalize the previous model in order to introduce a linear form of permanent market impact ( see @xcite for the rationale underlying a linear form of market impact ) . in this appendix
, we assume that @xmath44 is the closing price of the @xmath207 day and that the average price in the payoff of the asr contract is the average of the closing prices .
the dynamics of the portfolio is the same : @xmath42 the dynamics of the price is however impacted by the strategy of the bank : @xmath208 where the additional term means that the price goes up when the bank buys shares and goes down when the bank sells shares , as in the classical almgren - chriss framework .
the average price is still @xmath38 if we assume that the order of size @xmath40 is executed evenly over the interval @xmath43 $ ] ( in a continuous time almgren - chriss model ) , this leads to the following dynamics for the cash account : @xmath209 where @xmath210 are i.i.d . random variables with moment generating function defined on @xmath31 , with @xmath211=0 $ ] and @xmath212=\left(\begin{array}{cc } 1 & \dfrac{\sqrt{3}}{2}\\ \dfrac{\sqrt{3}}{2 } & 1 \end{array}\right)$ ] .
the term @xmath213 comes from the fact that the expected value of the price paid for the shares is the average between @xmath214 and @xmath215 .
the noise term @xmath216 corresponds to the fact that we execute progressively and evenly over the day while the price @xmath44 is the closing price .
permanent market impact also has to be taken into account in the price paid for the shares after time @xmath68 . for that purpose ,
we replace @xmath217 by @xmath218 .
hence , the optimization problem the bank faces becomes : @xmath219.\ ] ] the problem can be solved recursively using the bellman equations : @xmath83@xmath220 where @xmath221.\label{eq : def - utilde2}\end{aligned}\ ] ] if we define @xmath222 the recursive equations for @xmath223 given in proposition [ theta ] become : @xmath224 where @xmath225 in that case , approximations with splines need to be used for @xmath36 and @xmath92 .
the problem is therefore slightly more complicated from a numerical point of view , but our framework can be used even in the case when one wants to model permanent market impact . | be it for taking advantage of stock undervaluation or in order to distribute part of their profits to shareholders , firms may buy back their own shares .
one of the way they proceed is by including accelerated share repurchases ( asr ) as part of their repurchase programs . in this article
, we study the pricing and optimal execution strategy of an asr contract with fixed notional .
in such a contract the firm pays a fixed notional @xmath0 to the bank and receives , in exchange , a number of shares corresponding to the ratio between @xmath0 and the average stock price over the purchase period , the duration of this period being decided upon by the bank . from a mathematical point of view , the problem is related to both optimal execution and exotic option pricing .
* key words : * optimal execution , asr contracts , optimal stopping , stochastic optimal control , utility indifference pricing . |
it is an old question to find the resistance between two adjacent grid points of an infinite square lattice in which all the edges represent identical resistances @xmath1 .
it is well known that the result is @xmath2 , and an elegant and elementary solution of the problem is given by aitchison @xcite .
the electric - circuit theory is discussed in detail in a classic text by van der pol and bremmer@xcite and they derive the resistance between nearby points on the square lattice . in doyle s and snells@xcite
book the connection between random walks and electric networks is presented , including many interesting results and useful references .
recently venezian @xcite and atkinson et .
@xcite also studied the problem and in these papers the reader can find additional references .
venezian s method for finding the resistance between two arbitrary grid points of an infinite square lattice is based on the principle of the superposition of current distributions .
this method was further utilized by atkinson et .
@xcite and applied to two dimensional infinite triangular and hexagonal lattices as well as infinite cubic and hypercubic lattices . in this paper
we present an alternative approach using lattice green s functions .
this green s function method may have several advantages : ( i ) it can be used straightforwardly for more complicated lattice structures such as body and face centered cubic lattices .
( ii ) the results derived by this method reflect the symmetry of the lattice structures although they may not be suitable for numerical purposes
. however , from these results one may derive other integral representations of the resistance between two nodes , which can be used for evaluating the integrals either algebraically or numerically with high precision . later in this paper
we shall give examples for this procedure .
( iii ) from the equation for the green s function one can , in principle , derive some so - called recurrence formulas for the resistances between arbitrary grid points of an infinite lattice . in this paper
such recurrence formulas are derived for an infinite square lattice @xcite ( for the first time to the best of the author s knowledge ) .
( iv ) in condensed matter physics the application of the lattice green s function has become a very efficient tool .
the analytical behavior of the lattice green s function has been extensively studied over the past three decades for several lattice structures .
we make use of the knowledge of this analytical behavior for an infinite cubic lattice and give the asymptotic value of the resistance as the separation of the two nodes tends to infinity . throughout this paper
we shall utilize the results known in the literature about the lattice green s function and also give some important references .
( v ) finally , our approach for networks of resistors may serve as a didactically good example for introducing the green s function method as well as many basic concepts such as the brillouin zone ( bz ) used in solid state physics .
we therefore feel that our green s function method is of some physical interest .
the application of the green s function proved to be a very effective method for studying the transport in inhomogeneous conductors and it has been used successfully by kirkpatrick @xcite for percolating networks of resistors .
our approach for obtaining the resistance of a lattice of resistors is closely related to that used by kirkpatrick .
economou s book@xcite gives an excellent introduction to the green s function .
a review of the lattice green s function is given by katsura et .
@xcite the lattice green s function is also utilized in the theory of the kosterlitz thouless
berezinskii@xcite phase transition of the screening of topological defects ( vortices ) .
a review of the latter problem is given by chaikin et .
al . in their book @xcite .
the phase transition in classical two - dimensional lattice coulomb gases has been studied by lee et .
@xcite also by using the lattice green s function in their monte carlo simulations .
the above examples of the applications of the lattice green s function are just a selection of many problems known in the literature .
a review of the green s function of the so - called tight - binding hamiltonian ( tbh ) used for describing the electronic band structures of crystal lattices is presented in economou s book @xcite .
the lattice green s function defined in this paper is related to the green s function of the tbh .
below we shall point out that the resistance in a given lattice of resistors is related to the green s function of the tbh at the energy at which the density of states is singular .
this singularity is one of the van hove singularities of the density of states @xcite .
consider a @xmath0 dimensional lattice which consists of all lattice points specified by position vectors @xmath3 given in the form @xmath4 where @xmath5 are independent primitive translation vectors , and @xmath6 range through all integer values ( i.e. positive and negative integers , as well as zero ) . in the case of a @xmath0 dimensional
hypercube all the primitive translation vectors have the same magnitude , say @xmath7 , @xmath8 @xmath9 . here
@xmath7 is the lattice constant of the @xmath0 dimensional hypercube . in the network of resistors
we assume that the resistance of all the edges of the hypercube is the same , say @xmath1 .
we wish to find the resistance between the origin and a given lattice point @xmath10 of the infinite hypercube .
we denote the current that can enter at lattice point @xmath3 by @xmath11 . however , for measuring the resistance between two sites the current has to be zero at all other sites .
similarly , the potential at site @xmath3 will be denoted by @xmath12 . then , at site @xmath3 , according to ohm s and kirchhoff s laws , we may write @xmath13 , \label{aram-1}\ ] ] where the @xmath14 are the vectors from site @xmath3 to its nearest neighbors ( @xmath15 ) . the right hand side of eq .
( [ aram-1 ] ) may be expressed by the so - called _ lattice laplacian _ defined on the hypercubic lattice : @xmath16 .
\label{laplace}\ ] ] the above defined lattice laplacian corresponds to the finite - difference representation of the laplace operator . the lattice laplacian @xmath17 yields the correct form of the laplacian in the continuum limit
_ i.e. _ @xmath18 .
the lattice laplacian is widely used to solve partial differential equations with the finite - difference method@xcite . to measure the resistance between the origin and an arbitrary lattice point @xmath10
we assume that a current @xmath19 enters at the origin and exits at lattice point @xmath10 .
therefore , the current is zero at all the lattice points except for @xmath20 and @xmath10 , where it is @xmath19 and @xmath21 , respectively .
thus , eq .
( [ aram-1 ] ) , with the lattice laplacian , can be rewritten as @xmath22 where current at lattice point @xmath3 is @xmath23 the resistance between the origin and @xmath10 is @xmath24 to find the resistance we need to solve eq .
( [ aram-2 ] ) .
this is a poisson - like equation and may be solved by using the lattice green s function : @xmath25 where the lattice green s function is defined by @xmath26 finally , the resistance between the origin and @xmath10 can be expressed by the lattice green s function . using eq .
( [ aram-3 ] ) and ( [ ellen-2 ] ) we obtain @xmath27 , \label{r - veg}\ ] ] where we have made use of the fact that the lattice green s function is even , @xmath28 @xmath29 .
equation ( [ r - veg ] ) is our central result for the resistance . to find the lattice green s function defined by eq .
( [ green-1 ] ) we take periodic boundary conditions at the edges of the hypercube .
consider a hypercube with @xmath30 lattice points along each side .
thus the total number of sites in the @xmath0 dimensional hypercube is @xmath31 . substituting the fourier
transform @xmath32 of the lattice green s function into eq .
( [ green-1 ] ) , we find @xmath33 for the @xmath0 dimensional hypercube where we have defined @xmath34 owing to the periodic boundary conditions , the wave vector @xmath35 in eq .
( [ g - k ] ) is limited to the first brillouin zone and is given by @xmath36 where @xmath37 are integers such that @xmath38 for @xmath39 , and @xmath40 are the reciprocal lattice vectors defined by @xmath41 , @xmath42 . here
we assumed that @xmath30 is an even integer , which will be irrelevant in the limit @xmath43 .
the mathematical description of the crystal lattice and the concept of the brillouin zone can be found in many books on solid state physics@xcite .
finally , the lattice green s function takes the form @xmath44 if we take the limit @xmath43 then the discrete summation over @xmath35 can be substituted by an integral @xcite : @xmath45 where @xmath46 is the volume of the unit cell of the @xmath0 dimensional hypercube .
thus the lattice green s function is @xmath47 using eqs .
( [ r - veg ] ) and ( [ g - veg ] ) in @xmath0 dimensions the resistance between the origin and lattice point @xmath10 is @xmath48 the above result can be simplified if we specify the lattice point as @xmath49 : @xmath50 from this final expression of the resistance one can see that the resistance does not depend on the angles between the unit vectors @xmath5 .
physically this means that the hypercube can be deformed without the change of the resistance between any two lattice points .
the resistance in topologically equivalent lattices is the same . for further references
we also give the lattice green s function for a @xmath0 dimensional hypercube : @xmath51 the resistance in an infinite conducting medium can be obtained by taking the limit as the lattice constant @xmath7 tends to zero in eq .
( [ r - hyper-1 ] ) . denoting the electrical conductivity of the medium by @xmath52 , the resistance of a @xmath0 dimensional hypercube with lattice constant @xmath7 , according to ohm s law , is given by @xmath53 using the approximation @xmath54 for @xmath55 , eq .
( [ r - hyper-1 ] ) can easily be reduced to @xmath56 the same result for the resistance of a conducting medium is given in chaikin s book@xcite .
consider a linear chain of identical resistors @xmath1 .
the resistance between the origin and site @xmath58 can be obtained by taking @xmath57 in the general result given in eq .
( [ r - hyper-2 ] ) : @xmath59 the integral can be evaluated by the method of residues@xcite and gives the following very simple result : @xmath60 this can be interpreted as the resistance of @xmath58 resistances @xmath1 in series .
the current flows only between the two sites separated by a finite distance .
the two semi - infinite segments of the chain do not affect the resistance .
using eq .
( [ r - hyper-2 ] ) , the resistance in two dimensions between the origin and @xmath62 is @xmath63 = r \int_{-\pi}^{\pi } \frac{d x_1}{2\pi } \int_{-\pi}^{\pi } \frac{d x_2}{2\pi}\ , \ , \ , \frac{1-\cos\left(l_1 x_1+l_2 x_2\right)}{2-\cos x_1 - \cos x_2}. \label{r - d2-b}\end{aligned}\ ] ] the resistance between two adjacent lattice sites
can easily be obtained from the above expression without evaluating the integrals in eq .
( [ r - d2-a ] ) or ( [ r - d2-b ] ) .
note that interchanging @xmath64 and @xmath65 in eq .
( [ r - d2-b ] ) does not change the resistance , @xmath8 @xmath66 .
this is consistent with the symmetry of the lattice .
then , from eq . ( [ r - d2-b ] ) , the sum of @xmath67 and @xmath68 yields @xmath69 thus @xmath70 .
the resistance between two adjacent lattice sites is @xmath2 , which is a well known result @xcite . in general
the integrals in eq .
( [ r - d2-a ] ) have to be evaluated numerically .
it is shown in appendix [ appendix-1 ] how one integral can be performed in eq .
( [ r - d2-a ] ) .
we found the same result as venezian@xcite and atkinson@xcite : @xmath71 where @xmath72 it turns out that this expression is more stable numerically than eq .
( [ r - d2-b ] ) . as an application of the above formula
one can calculate the resistance between second nearest neighbors exactly .
after some algebra one finds @xmath73 the energy dependent lattice green s function of the tight - binding hamiltonian for a square lattice is given by @xmath74 [ see eq .
( 5.31 ) in economou s book@xcite ] .
this is a generalization of our green s function by introducing a new variable @xmath75 instead of the value 2 in the denominator in eq .
( [ g - hypercube ] ) for @xmath61 .
note that a factor 2 appearing in the denominator of our green s function in eq .
( [ g - hypercube ] ) is missing in eq .
( [ g_e - d2 ] ) .
this is related to the fact that in the schrdinger equation the laplacian is multiplied by a factor of @xmath76 while in our case the laplace equation is solved . comparing eqs . (
[ g - hypercube ] ) ( for @xmath61 ) and ( [ g_e - d2 ] ) one can see that the resistance is @xmath77 .
\label{vanhove - sq}\ ] ] based on the equation for the green s function morita @xcite derived the recurrence formulas for the green s function @xmath78 for an infinite square lattice ( see eqs .
( 3.8 ) and ( 4.2)-(4.4 ) in morita s paper@xcite ) .
applying morita s results ( with @xmath79 ) to the resistance given in eq .
( [ vanhove - sq ] ) we obtained the following recurrence formulas for the resistance : @xmath80 we have seen that @xmath81 and @xmath82 since we know the exact values of @xmath68 and @xmath83 ( obviously @xmath84 ) one can calculate the resistance exactly for arbitrary nodes by using the above given recurrence formulas . this way we obtained the same results as atkinson et .
@xcite using mathematica .
the advantages of our recurrence relations are that they provide a new , very simple and effective tool to calculate the resistance between arbitrary nodes on a square lattice .
we note that van der pol and bremmer@xcite also gave the exact values of the resistance for nearby points in a square lattice using a different approach .
it is interesting to see the asymptotic form of the resistance for large values of @xmath64 or / and @xmath65 . in appendix [ appendix-2 ]
we derive the asymptotic form of the lattice green s function for a square lattice [ see eq .
( [ g - asym - veg ] ) ] . inserting eq . ( [ g - asym - veg ] ) into the general result of the resistance given in eq .
( [ r - veg ] ) the asymptotic form of the resistance is @xmath85 where @xmath86 is the euler - mascheroni constant@xcite .
the same result was obtained by venezian@xcite except that we got an exact value of the numerical constant in eq .
( [ r - d2-asym ] ) whereas it was numerically approximated in venezian s paper .
the resistance is logarithmically divergent for large values of @xmath64 and @xmath65 .
a similar behavior was found for conducting medium by chaikin et .
al . in their book@xcite . in the theory of the kosterlitz
berezinskii@xcite phase transition of the screening of topological defects ( vortices ) the same asymptotic form of the green s function ( as given in eq .
( [ g - asym - veg ] ) ) has been used for a square lattice @xcite .
finally , we note that doyle and snell showed in their book@xcite ( pp .
122123 ) that the resistance goes to infinity as the separation between nodes tends to infinity but the asymptotic form was not derived . in three dimensions
the resistance between the origin and a lattice point @xmath88 can be obtained from eq .
( [ r - hyper-2 ] ) : @xmath89 & = & r\int_{-\pi}^{\pi } \frac{d x_1}{2\pi } \int_{-\pi}^{\pi } \frac{d x_2}{2\pi } \int_{-\pi}^{\pi } \frac{d x_3}{2\pi}\ , \ , \ , \frac{1-\cos\left(l_1 x_1+l_2 x_2 + l_3 x_3\right ) } { 3-\cos x_1 - \cos x_2 -\cos x_3}. \label{r - d3-b}\end{aligned}\ ] ] similarly to the case of a square lattice , the exact value of the resistance between two adjacent lattice sites can be obtained from the above expression . clearly , from eq .
( [ r - d3-b ] ) ( and for symmetry reasons , too ) @xmath90 and @xmath91 therefore , the resistance between adjacent sites is @xmath92 .
it is easy to show much in the same way that in a @xmath0 dimensional hypercube the resistance between adjacent sites is @xmath93 .
unlike in a square lattice , in a simple cubic lattice the resistance does not diverge as the separation of the entering and exiting sites increases , but tends to a finite value . one can write for the resistance @xmath94 $ ] where @xmath95 is given in eq .
( [ g - hypercube ] ) for @xmath96 it is well known from the theory of fourier series ( riemann s lemma ) that @xmath97 for any integrable function @xmath98 hence , @xmath99 ( which indeed corresponds to the boundary condition of the green s function at infinity ) and thus , @xmath100 when any of the @xmath101 the value of @xmath102 was evaluated for the first time by watson @xcite and subsequently by joyce @xcite in a closed form in terms of elliptic integrals .
the following exact result was found @xmath103}^2 = 0.505462\dots , \label{exact - kocka}\ ] ] where @xmath104 and @xmath105 is the complete elliptic integral of the first kind .
it is worth mentioning that a simpler result was obtained by glasser et .
@xcite ( see also doyle s and snell s book@xcite ) who calculated the integrals in terms of gamma functions : @xmath106 thus , the resistance in units of @xmath1 for a simple cubic lattice tends to the finite value 0.505462@xmath107 when the separation between the entering and exiting sites tends to infinity . for finite separations the formula for the resistance given in eq .
( [ r - d3-b ] ) is not suitable for numerical purposes since the integrals converge slowly with increasing the number of mesh points .
similarly to the case of a square lattice , we can use the energy dependent lattice green s function of the tight - binding hamiltonian defined for a simple cubic lattice as @xmath108 [ see eq .
( 5.49 ) in economou s book @xcite ] .
this is a generalization of our green s function by introducing a new variable @xmath75 instead of the value 3 in the denominator in eq .
( [ g - hypercube ] ) for @xmath87 .
note that the reason for the missing factor of 2 in eq .
( [ g_e - d3 ] ) is the same as that explained in the previous section .
comparison of eqs .
( [ g - hypercube ] ) ( for @xmath87 ) and ( [ g_e - d3 ] ) yields @xmath109 . \label{vanhove}\ ] ] the analytic behavior of the lattice green s function @xmath110 has been extensively studied over the past three decades .
numerical values of @xmath110 were given by koster et .
@xcite and wolfram et .
al . , @xcite by using the following representation of the green s function for a simple cubic lattice : @xmath111 where @xmath112 is the modified bessel function of order @xmath58 . however , it turns out that another representation of the lattice green s functions is more suitable for numerical calculations . for @xmath113 the following form of the green s function along an axis
was given in terms of complete elliptic integrals of the first kind by horiguchi @xcite @xmath114 where @xmath115 and @xmath116 is defined in eq .
( [ elliptic ] ) . in fig .
[ r - gorbe - kocka ] we plotted the numerical values of the resistance using eq .
( [ hori ] ) for @xmath117 and @xmath118 .
= 8.5 cm it is seen from the figure that the resistance tends rapidly to its asymptotic value given in eq .
( [ exact - kocka ] ) .
we would mention that glasser et .
@xcite gave other useful integral representations of the lattice green s function for the hypercubic lattice for arbitrary dimension @xmath0 .
it was shown by joyce @xcite that the function @xmath119 can be expressed in the form of a product of two complete elliptic integrals of the first kind .
horiguchi@xcite and later on morita@xcite obtained recurrence relations for the function @xmath110 of the simple cubic lattice for an arbitrary site @xmath120 in terms only of @xmath119 , @xmath121 and @xmath122 .
the last two green s functions were expressed by horiguchi and morita @xcite in closed form in terms of complete elliptic integrals . from these results
one can calculate the green s function for an arbitrary lattice point . as an example , one of the recurrence formulas we obtained from the recurrence formula of the green s function ( at @xmath117 ) given by horiguchi@xcite is @xmath123 together with other recurrence formulas given in horiguchi s and morita s paper one can find @xmath124 for arbitrary values of @xmath120 .
the numerical values of the resistance for small @xmath120 are listed in table i. .numerical values of @xmath124 in units of @xmath1 for a simple cubic lattice . [
cols="^,^ " , ] our results are in agreement with those given by atkinson@xcite .
it is worth mentioning some useful references for the lattice green s function of other three dimensional lattices , such as body centered and face centered .
the exact values of @xmath125 as well as other exact results can be found in joyce s paper@xcite for a body centered cubic lattice .
morita @xcite gave the recurrence relations of the lattice green s function for a body centered cubic lattice .
inoue@xcite derived the recurrence relations for a face centered cubic lattice and obtained the exact values of the lattice green s function at some lattice points .
based on the results given in these papers one can calculate the resistance between arbitrary nodes for cubic lattices . finally we would like to point out that in eq .
( [ vanhove ] ) the resistance is related to the energy dependent green s function at energy @xmath117 . however , it is known that the density of states at this energy corresponds to one type of the van hove singularities @xcite .
therefore , the resistance for a simple cubic lattice is related to the green s function of the tight - binding hamiltonian at the value of the energy at which the density of states has a van hove type singularity . in the general case of a @xmath0 dimensional hypercube
the green s function of the tight - binding hamiltonian is infinite at @xmath126 for @xmath61 ( logarithmically divergent ) while it has a finite value for @xmath87 as it was shown above . in the latter case
the derivative of the green s function with respect to @xmath75 is singular .
in this section we shall calculate the resistance of a rectangular lattice in which the resistance of each edge is proportional to its length .
consider a rectangular lattice with unit vectors @xmath127 and @xmath128 and introduce parameter @xmath129 let @xmath1 be the resistance of the edge along the direction of @xmath128 . to find the resistance between the origin and site @xmath10 assume a current @xmath19 enters at the origin and exits at site @xmath10 .
we denote the potential at lattice point @xmath3 by @xmath12 .
then , according to ohm s and kirchhoff s laws , we may write @xmath130 where the ` rectangular ' laplacian is defined by @xmath131 then the equation for the lattice green s function corresponding to the ` rectangular ' laplace equation is @xmath132 the green s function can be calculated in a way similar to the hypercubic case in sec .
[ sec : hyper ] .
we have @xmath133 where @xmath134 is the area of the unit cell , and the brillouin zone is a rectangle with sides @xmath135 and @xmath136 along the directions of @xmath127 and @xmath128 , respectively , and @xmath137.\ ] ] using eq .
( [ r - veg ] ) , the resistance between the origin and lattice point @xmath138 for a given @xmath139 is @xmath140 & = & r \int_{-\pi}^{\pi } \frac{d x_1}{2\pi } \int_{-\pi}^{\pi } \frac{d x_2}{2\pi}\ , \ , \ , \frac{1-\cos\left(l_1 x_1+l_2 x_2\right ) } { \frac{1}{p}\left(1-\cos x_1\right ) + 1- \cos x_2}. \label{r - rect - b}\end{aligned}\ ] ] the integral over @xmath141 can be evaluated similarly to a square lattice ( see appendix [ appendix-1 ] ) .
we find @xmath142 where @xmath143 note that @xmath64 and @xmath65 are interchanged here compared to eq .
( [ r - d2-one_int ] ) .
now it is interesting to see how the resistance between adjacent sites varies with @xmath129 the resistance between nearest neighbors is different along the @xmath144 and the @xmath145 axis .
thus , unlike in the case of a square lattice , @xmath146 for a rectangular lattice except for the trivial case @xmath147 ( square lattice ) .
no sum rule exists for @xmath148 as in the case of a square lattice . in fig .
[ r - gorbe - tegla ] the resistances @xmath149 and @xmath150 are plotted as functions of @xmath139 .
= 8.5 cm from fig .
[ r - gorbe - tegla ] one can see that resistance @xmath149 increases with increasing @xmath139 .
it can be shown that @xmath151 as @xmath152 .
this is physically clear since the lattice constant along the @xmath144 axis increases resulting in an increasing resistance of each of the segments in this direction . on the other hand along the @xmath145 axis
a saturation of @xmath150 can be seen , which is not obvious at all .
expanding the integral in the expression of @xmath150 in powers of @xmath139 , for large @xmath139 we get @xmath153 .
thus , @xmath154 when @xmath152 .
some additional results are presented for the energy dependent lattice green s function for a rectangular lattice in the papers of morita et .
@xcite and katsura et .
al @xcite .
in this section we calculate the resistance in a triangular lattice in which the resistance of each edge is identical , say @xmath1 . first we consider a triangular lattice with unit vectors @xmath127 and @xmath128 , and with a lattice constant @xmath155 .
we choose @xmath127 and @xmath128 such that the angle between them is @xmath156 .
we introduce a third vector by @xmath157 .
the vectors drawn from each lattice point to its 6 nearest neighbors are @xmath158 .
assume that a current @xmath19 enters at the origin and exits at site @xmath10 , and that the potential at site @xmath3 is @xmath12 .
again , from ohm s and kirchhoff s laws , we find @xmath159 where the ` triangular ' laplacian is defined by @xmath160 .
\label{triang - laplacian}\ ] ] it is important to note that the triangular laplacian @xmath161 defined above is @xmath162 times that used for solving the laplace equation with the finite - difference method on a triangular lattice . the factor @xmath162 ensures that the lattice laplacian in the finite - difference method yields the correct form of the laplacian in the continuum limit ( @xmath18 ) .
however , in our case this limit does not exist since there are also connections ( resistors ) between adjacent lattice points along the direction of the vector @xmath163 .
the equation for the lattice green s function corresponding to the ` triangular ' laplace equation is given by @xmath164 the green s function can be calculated in the same way as in the hypercubic case .
we have @xmath165 where @xmath166 is the area of the unit cell , and the vector @xmath35 given in eq .
( [ k - vector ] ) runs over the brillouin zone of the triangular lattice , which is a regular hexagon@xcite .
the resistance between the origin and site @xmath10 can be obtained from eq .
( [ r - veg ] ) with the lattice green s function for a triangular lattice and it yields @xmath167 note that the factor @xmath168 dropped out from the denominator . using eq .
( [ r - triang - veg ] ) we can easily find the resistance between two adjacent lattice sites . for symmetry reasons it is clear that @xmath169 . on the other hand @xmath170 in the last step we have made use of the fact that the volume of the brillouin zone is @xmath171
hence , the resistance between two adjacent lattice sites is @xmath172 .
it is worth mentioning that without changing the resistance between two arbitrary lattice points one can transform the triangular lattice to a square lattice in which there are also resistors between the end points of one of the diagonals of each square .
this topologically equivalent lattice is more suitable for evaluating the necessary integrals over the brillouin zone since the brillouin zone becomes a square .
the same transformation was used by atkinson et .
al . in their paper@xcite
if we choose @xmath173 and @xmath174 , then the resistance between the origin and lattice point @xmath175 is given by @xmath176 in appendix [ appendix-3 ] we perform one integral in eq .
( [ r - triang - veg - trafo ] ) using the method of residues@xcite in a similar way to appendix [ appendix-1 ] .
after some algebra it is easy to see that the result given in eq .
( [ triang - r - egyint - veg - app ] ) is in agreement with atkinson s result@xcite : @xmath177 where @xmath178 for @xmath179 in eq .
( [ r - triang - one_int ] ) we have @xmath180 evaluating this expression by means of the program maple we obtained the same results for @xmath179 as atkinson@xcite .
like in the case of a square lattice we believe that similar recurrence formulas exist for a triangular lattice but further work is necessary along this line .
similarly , it would be interesting to find the asymptotic form of the resistance as the separation between the two nodes tends to infinity .
according to some preliminary work the resistance is again logarithmically divergent .
in this section we calculate the resistances in an infinite honeycomb lattice of resistors .
atkinson and steenwijk@xcite have studied this lattice structure exploiting the fact that the hexagonal lattice can be constructed from the triangular lattice by the application of the so - called @xmath181 transformation@xcite .
one advantage of our green s function method is that we do not use such a @xmath181 transformation , specific only for the honeycomb lattice , therefore our method can also be used in a straightforward manner for other lattice structures .
as we shall see later , in the honeycomb lattice each unit cell contains @xmath182 lattice points , which is the main structural difference from the triangular lattice . including more than one lattice point in the unit cell , the method outlined in this section can be viewed as a generalization of the green s function method discussed in the previous sections .
we assume that all the resistors have the same resistance @xmath1 .
the lattice structure and the unit cell are shown in fig .
[ hatszog - abra ] .
the angle between @xmath127 and @xmath128 is @xmath156 , and @xmath183 , where @xmath7 is the length of the edges of the hexagons .
there are two types of lattice points in each unit cell denoted by @xmath184 and @xmath185 in fig .
[ hatszog - abra ] .
= 8.5 cm from now on the position of the lattice points @xmath184 and @xmath185 will be given by the position of the unit cell in which they are located .
assume the origin is at one of the lattice points @xmath184 , and then the position of a unit cell can be specified by the position vector @xmath186 , where @xmath187 are arbitrary integers .
we denote the potential and the current in one of the unit cells by @xmath188 and @xmath189 , and @xmath190 and @xmath191 , respectively , where subscripts @xmath184 and @xmath185 refer to the corresponding lattice points . owing to ohm s and
kirchhoff s laws the currents @xmath190 and @xmath191 in the unit cell specified by @xmath3 satisfy the following equations : @xmath192
i_b({\bf r } ) & = & \frac{v_b({\bf r})-v_a({\bf r})}{r } + \frac{v_b({\bf r})-v_a({\bf r}+{\bf a}_1)}{r}+ \frac{v_b({\bf r})-v_a({\bf r}+{\bf a}_1+{\bf a}_2)}{r}. \label{aram - honey}\end{aligned}\ ] ] assuming periodic boundary conditions again the potential @xmath188 can be given by its fourier transform @xmath193 where @xmath194 is the number of unit cells , and analogous expressions are valid for @xmath195 and @xmath191 .
here @xmath35 is in the brillouin zone@xcite .
thus , we may rewrite eq .
( [ aram - honey ] ) as @xmath196 = -r\left [ \begin{array } { c } i_a({\bf k } ) \\ i_a({\bf k } ) \end{array } \right ] , \label{fourier}\ ] ] where @xmath197 and @xmath198 equation ( [ fourier ] ) is indeed the fourier transform of the poisson - like equation that determines the potentials for a given current distribution .
however , in this case the laplace operator is a 2x2 matrix . in @xmath3-represantion
the analogous equation for a hypercubic lattice was given in eq .
( [ aram-2 ] ) . the equation for the fourier transform of the green s function ( which is a 2x2 matrix for a honeycomb lattice , too ) can be defined analogously to a hypercubic lattice in eq .
( [ green-1 ] ) : @xmath199 the solution of eq .
( [ green - hatsz - def ] ) for the green s function is @xmath200 where @xmath201 and we have introduced a third vector , @xmath202 .
there are four types of resistance .
we denote the resistance between a lattice point @xmath184 as the origin and site @xmath175 ( which is an @xmath184-type site ) by @xmath203 , while the resistance between the origin and site @xmath204 ( which is a @xmath185-type site in the unit cell at @xmath10 ) is denoted by @xmath205 . for symmetry reasons
, it follows that for the other two types of resistance : @xmath206 and @xmath207 to measure @xmath203 the current at sites @xmath184 and @xmath185 in the unit cell at @xmath3 are @xmath208 while for measuring @xmath205 we have @xmath209 thus @xmath210 r_{ab}({\bf r}_0 ) & = & \frac{v_a({\bf r}={\bf 0})-v_b({\bf r}_0)}{i}. \label{r - b}\end{aligned}\ ] ] first , consider the resistance @xmath203 . from eqs .
( [ fourier ] ) , ( [ green - hatsz - def])([aram - a ] ) we obtain @xmath211 hence , using eq .
( [ hatsz - fourier - pot ] ) we can obtain the @xmath3-dependence of the potential @xmath188 and then eq . ( [ r - a ] ) yields @xmath212 using eq .
( [ sum - int ] ) , the summation over @xmath35 can be substituted by an integral and we obtain @xmath213 where @xmath214 is the area of the unit cell .
one can see that the same expression was found for a triangular lattice except for the factor @xmath215 here [ see eq .
( [ r - triang - veg ] ) ] .
note that although the area of the unit cell @xmath216 is 6 times bigger than that for the triangular lattice , the area of the brillouin zone is 6 times less and thus , the ratio of the resistances for honeycomb and triangular lattices is 3 .
therefore , the resistance between two @xmath184-type nodes is three times the corresponding resistance in the triangular lattice
. this again agrees with atkinson s result .
similarly , using eqs .
( [ fourier ] ) , ( [ green - hatsz - def ] ) , ( [ green - hatsz ] ) and ( [ aram - b ] ) we can obtain @xmath217 and @xmath218 . from eq .
( [ hatsz - fourier - pot ] ) the potential @xmath188 can be determined , and analogously @xmath189 .
finally , eq . ( [ r - b ] ) leads to @xmath219 & = & rv_0\int_{{\bf k } \in { \rm bz } } \ , \frac{d^2 { \bf k}}{{(2\pi)}^2}\ , \frac{3-\cos { \bf k}{\bf r}_0- \cos { \bf k}\left({\bf r}_0+{\bf a}_1\right ) -\cos { \bf k}\left({\bf r}_0+{\bf a}_1+{\bf a}_2\right ) } { 3-\cos { \bf k}{\bf a}_1-\cos { \bf k}{\bf a}_2- \cos { \bf k}{\bf a}_3 } \nonumber \\[1ex ] & = & \frac{1}{3 } \left[r_{aa}({\bf r}_0 ) + r_{aa}({\bf r}_0+{\bf a}_1 ) + r_{aa}({\bf r}_0+{\bf a}_1+{\bf a}_2 ) \right ] .
\label{r - b - veg}\end{aligned}\ ] ] the same result was found by atkinson and steenwijk .
the resistance between second nearest neighbor lattice sites may be found from eq .
( [ r - a - veg ] ) . for symmetry reasons @xmath220 and from eq .
( [ r - a - veg ] ) we have @xmath221 thus , the resistance between second nearest neighbor lattice sites is @xmath222 using eq .
( [ r - b - veg ] ) , the resistance for adjacent lattice sites is @xmath223=2r/3 $ ] , since obviously , @xmath224 .
we would just mention that the limiting value of the resistance is again infinite as the separation between nodes tends to infinity ( see p. 139 of doyle s and snell s book@xcite ) .
we note that one can transform the honeycomb lattice to a topologically equivalent brick - type lattice shown in fig .
[ tegla - abra ] without changing the resistance between two arbitrary lattice points .
= 8.5 cm this fact has been utilized in the case of the triangular lattice in the previous section . as it can be seen from the above results , the resistance between arbitrary lattice points in a honeycomb lattice can be related to the corresponding triangular lattice .
this kind of relation is the consequence of the so - called _ duality transformation _ in which the variables are transformed to fourier transform variables ( for more details see chaikin s book@xcite pp .
578584 ) .
the dual lattice of a triangular lattice is a honeycomb lattice .
however , if the unit cell contains more than two non - equivalent lattice points then such a connection might not be used . on the other hand , the green s function method outlined in the example of the honeycomb lattice can still be applied straightforwardly .
the author wishes to thank g. tichy , t. geszti , p. gndig , a. pirth , s. redner and l. glasser for helpful discussions .
this work was supported by the hungarian science foundation otka t025866 , the hungarian ministry of education ( fkfp 0159/1997 ) and omfb within the program `` dynamics of nanostructures '' .
starting from eq . ( [ r - d2-a ] ) for the resistance
@xmath225 we have @xmath226 where @xmath227 since @xmath228 , we take @xmath229 . to proceed further ,
we perform the integral in @xmath230 using the method of residues . introducing the complex variable @xmath231 the integral @xmath230 reads @xmath232 where the path of integration is the unit circle .
the denominator has roots at @xmath233 and @xmath234 , where @xmath235 and @xmath236 satisfy the equation @xmath237 and @xmath238 it is clear that for @xmath239 we have @xmath240 , therefore the solution for @xmath241 is purely imaginary
. thus we introduce @xmath242 with @xmath243 , where @xmath242 satisfies the equation @xmath244 for @xmath239 it is true that @xmath245 , so the two poles of the integrand in @xmath230 are real numbers and satisfy the following inequalities : @xmath246 and @xmath247 .
thus the pole @xmath248 is within the unit circle , while @xmath249 is outside .
according to the residue theorem@xcite @xmath250 . using eq .
( [ gyok ] ) the residue of the integrand of @xmath230 at @xmath248 is @xmath251 we obtain @xmath252 where @xmath242 satisfies eq .
( [ gyok ] ) . finally , the resistance @xmath225 for arbitrary integers @xmath187 becomes @xmath253 the same result was found by venezian@xcite .
in this appendix we derive the asymptotic form of the lattice green s function for a square lattice .
the lattice green s function at site @xmath20 is divergent since @xmath254 for @xmath255 .
therefore we calculate the asymptotic form of @xmath256 . starting from eq .
( [ g - hypercube ] ) the lattice green s function for site @xmath257 in a square lattice becomes @xmath258 the integral over @xmath144 is the same as @xmath230 in eq .
( [ i - y ] ) , so we can use the result obtained in eq . ( [ i - y - veg ] ) : @xmath259 we follow the same method for deriving the asymptotic form of the lattice green s function for large @xmath260 and @xmath261 as venezian@xcite .
a similar method was used in chaikin s book@xcite ( see pp .
295296 ) in the case of a continuous medium in two dimensions .
we break the integral in eq .
( [ g - asym-1 ] ) into three parts : @xmath262 where @xmath263 i_2 & = & \int_{0}^{\pi } \frac{d y}{2\pi } \,\ , \left(\frac{1}{\sinh s } -\frac{1}{y}\right ) , \nonumber \\[1ex ] i_3 & = & \int_{0}^{\pi } \frac{d y}{2\pi } \,\ , \left(\frac{e^{-\left|n\right|y}\cos my}{y } - \frac{e^{-\left|n\right|s}\cos my}{\sinh s}\right ) , \label{i-123}\end{aligned}\ ] ] and @xmath242 satisfies eq .
( [ gyok ] ) .
the first integral @xmath264 can be expressed by the integral exponential @xmath265 @xcite : @xmath266 \right \},\ ] ] where @xmath265 is defined by @xmath267 for large values of its argument , @xmath268 , where @xmath86 is the euler - mascheroni constant .
thus , for large @xmath261 and @xmath260 @xmath264 can be approximated by @xmath269 using eq .
( [ gyok ] ) the integral @xmath270 can be evaluated exactly : @xmath271 in the integral @xmath272 the integrand is close to zero for small values of @xmath145 and @xmath242 since @xmath273 , while for larger values of @xmath145 and @xmath242 the exponentials are negligible , therefore @xmath274 .
finally , we find that the lattice green s function for large arguments , @xmath8 @xmath275 becomes @xmath276 the same result is quoted on page 296 of chaikin s book@xcite . in the theory of the kosterlitz thouless
berezinkii phase transition @xcite kosterlitz used the same asymptotic form of the green s function for a square lattice @xcite .
introducing the coordinate transformations @xmath277 and @xmath278 , and using the complex variable @xmath279 we have @xmath280 from eq .
( [ r - triang - veg - trafo ] ) , where @xmath281 and the path of integration is the unit circle .
the factor 1/2 in front of @xmath225 is the jacobian corresponding to the transformations of variables . since @xmath282 , we take @xmath283 the denominator has roots in @xmath284 and @xmath285 and it is easy to see that they satisfy equations @xmath286 and @xmath287 .
for @xmath288 it is clear that @xmath235 is purely imaginary .
if we choose @xmath235 such that @xmath289 , then @xmath290 and @xmath291 thus , with @xmath292 , the residue of the integrand of @xmath293 at @xmath248 ( which is inside the unit circle , @xmath8 @xmath294 ) is @xmath295 where @xmath242 satisfies the equation @xmath296 finally , the resistance for arbitrary integers @xmath187 is @xmath297 after the bulk of this work was completed i received an e - mail from sidney redner who pointed out several references including atkinson s and steenwijk s work@xcite . reading these papers and
demonstrating the usefulness of the green s function method inspired me to include the recurrence formulas for an infinite square lattice and to show that our resistance formula for a triangular lattice agrees with that given by atkinson and steenwijk .
v. l. berezinskii , `` destruction of long - range order in one - dimensional and two - dimensional systems having a continuous symmetry group i. classical systems , ' ' sov .
jetp * 32 , * 493500 ( 1971 ) , [ zh . eksp . teor .
* 59 * , 907920 ( 1970 ) ] . | we calculate the resistance between two arbitrary grid points of several infinite lattice structures of resistors by using lattice green s functions . the resistance for @xmath0 dimensional hypercubic , rectangular , triangular and honeycomb lattices of resistors
is discussed in detail .
we give recurrence formulas for the resistance between arbitrary lattice points of the square lattice . for large separation between nodes
we calculate the asymptotic form of the resistance for a square lattice and the finite limiting value of the resistance for a simple cubic lattice .
we point out the relation between the resistance of the lattice and the van hove singularity of the tight - binding hamiltonian .
our green s function method can be applied in a straightforward manner to other types of lattice structures and can be useful didactically for introducing many concepts used in condensed matter physics . |
the international network of large - scale interferometric gravitational wave ( gw ) detectors is currently undergoing an extensive technological upgrade towards what is commonly referred to as the `` second generation '' of gw detectors . upon completion , these second generation observatories , namely geo - hf @xcite , advanced ligo @xcite and advanced virgo @xcite , complemented by the new observatory kagra @xcite , will reach unprecedented sensitivities to gw - induced strain .
this sensitivity improvement constitutes a major leap towards reaching the long - standing goal of the first direct measurement of gws . besides power recycling ( pr ) , which was already implemented in the first generation ligo and virgo interferometers , signal recycling
@xcite ( sr ) has been adopted for the optical layouts of all second generation detectors .
dual recycling , i.e. the combination of pr and sr , has already been successfully implemented and operated in the geo600 detector @xcite .
furthermore , the injection of squeezed vacuum a technique proposed by caves to improve the quantum noise limited sensitivity of gw detectors @xcite will be employed to enhance the performance of geo - hf as well as the advanced ligo hanford observatory .
the current geo600 detector has been the first large - scale interferometer to successfully demonstrate long - term stable squeezing injection and the associated broadband sensitivity improvement , already for the case of carrier - tuned sr . @xcite the compatibility of _ detuned _ sr interferometers with squeezing injection was investigated theoretically by harms et al .
this work led to the insight that the combination of these two techniques requires additional filter cavities in the squeezing path to take full advantage of the benefit from squeezing injection on the detector sensitivity .
the use of filter cavities was previously suggested by kimble et al . to compensate for radiation pressure noise in gw detectors @xcite . as an alternative to conventional detuned sr , thring et al . proposed the technique of twin signal recycling @xcite ( tsr ) which is based on introducing a second recycling mirror in the asymmetric port of the interferometer ( cf .
[ fig : drmi_tsr_topologies ] ) .
it was shown that , in contrast to detuned sr , an interferometer featuring tsr does not require additional filter cavities to maximize the benefit from squeezing injection in the shot noise - limited frequency regime .
besides this , thring points out further advantages of tsr over conventional detuned sr in his paper : a sensitivity improvement over a wide frequency band and a less challenging gw signal readout scheme .
the former can be attributed to the simultaneous resonant enhancement of both the upper and the lower gw signal sidebands , which can be arranged for in tsr configurations .
the latter stems from the fact that in tsr configurations the gw signal is entirely contained in the phase quadrature of the carrier field , similar to the case of a conventional michelson interferometer . in this paper
we describe the electronic stabilization and long - term operation of a power recycled michelson interferometer with tsr , thus laying the foundation for further investigations of this topology to evaluate its viability for the application in gw interferometry .
one such aspect , the broadband enhancement of the interferometer by squeezed light injection , was carried out in a subsequent experiment and was already reported on in @xcite .
the starting point of our investigations was an existing dual recycling michelson interferometer ( drmi ) tabletop experiment @xcite a scaled and largely simplified model of the geo600 interferometer which we were aiming to extend to a power recycled tsr interferometer with a minimum of invasive changes .
thus , we were able not only to experimentally test the power recycled tsr topology but also to evaluate the viability of tsr as an upgrade for existing interferometers with conventional sr . by introducing an additional recycling mirror into the optical setup of the drmi ,
the former signal recycling cavity ( src ) is converted into a system of two optically coupled cavities which we refer to as the _ twin signal recycling cavity _ ( tsrc ) throughout this paper .
this cavity is formed by the newly introduced mirror and the michelson mirrors acting as the end mirrors , and the former signal recycling mirror acting as the coupling mirror . for clarity we will keep referring to the former signal recycling mirror as srm and
denote the newly introduced mirror as _ twin signal recycling mirror _ ( tsrm ) .
we will refer to the length of the tsrm
srm cavity as @xmath0 and , correspondingly , the average length of the cavity formed by the srm and the michelson end mirrors ( i.e. the former src ) will be denoted as @xmath1 , in accordance with the nomenclature used in @xcite .
three free parameters of the tsrm were to be determined : ( i ) reflectivity , ( ii ) radius of curvature ( roc ) and ( iii ) distance to the srm .
these , either directly or indirectly , affect the frequency response as well as the structure of the sensing matrix of the `` optical plant '' which determines the complexity of the longitudinal control scheme .
the reflectivity of the tsrm influences the level of resonant enhancement of the signal sidebands , which is in close analogy to the role of the srm reflectivity in a signal recycled interferometer .
we chose a reflectivity of @xmath2% for our experiment .
the roc of the tsrm and its distance to the srm can , generally , not be chosen independently .
the boundary conditions for a stable and ( theoretically ) well mode matched cavity constrain the values these parameters can reasonably take on . from the control point of view
it was desirable to chose @xmath3 such that , for given phase modulation ( pm ) control sideband frequencies , an error signal could be extracted for feedback to the tsrm which preferably exhibits only small coupling to the other longitudinal degrees of freedom ( dof ) in the interferometer . for a known length @xmath3 the corresponding tsrm curvature , which results in a stable and mode matched cavity , can then be easily determined .
we will discuss our approach of determining the optimal length @xmath0 , driven by control considerations , in the following section .
for the design of the sensing scheme we performed a series of numerical simulations , based on a steady - state frequency domain model of the power recycled tsr interferometer realized with the software _ finesse _ @xcite .
besides the length @xmath3 the optimal locations of the length signal extration ports to obtain error signals for the four longitudinal dof of the interferometer were to be determined with the aid of the numerical model .
these dof are the differential mode of the michelson arms ( mich ) , the power recycling cavity ( prc ) and the microscopic positions of the tsrm relative to the srm ( @xmath4 ) and the srm relative to the average position of the michelson end mirrors ( @xmath5 ) , respectively . due to the coupled nature of the longitudinal dof in a power recycled tsr interferometer the extracted heterodyne length signals , likewise ,
are coupled .
this coupling can often be reduced by prudently designing the optical plant and the sensing system .
however , the rf modulation scheme was carried over from the precursor drmi experiment and modulation frequencies were thus not intended as free parameters for the optimization of the sensing scheme . as a consequence ,
the resulting sensing matrices were non - orthogonal but could be arranged for to have at least full rank by choosing appropriate values for the remaining free parameters . in this case
the technique of _ gain hierarchy _ @xcite can often be applied and was successfully used in earlier interferometry experiments , e.g. in @xcite . .modulation / demodulation parameters of the signal extraction system . in the first column
the longitudinal dof are listed along with , in brackets , the mirrors which were actuated upon to tune the respective dof .
[ cols="<,^,^,^",options="header " , ] [ tab : exeriment_params ]
in this section we discuss the aspect of bringing the aligned and mode matched , yet uncontrolled interferometer to a fully locked state of its four longitudinal dof .
the proportional gains of the feedback loops were optimized for long - term stable operation at @xmath6min.,scaledwidth=97.0% ] .
these signals were amplified and in turn applied to piezoelectric transducers actuating on the power recycling mirror ( blue trace ) , one of the michelson end mirrors ( orange trace ) , the signal recycling ( green trace ) and the twin signal recycling mirror ( magenta trace ) , respectively .
long term drifts eventually led to a lock - loss of the instrument due to the limited range of the actuators.,scaledwidth=97.0% ] the hierarchic stabilization of the lengths in the interferometer was carried out in the order which was derived from the numerical interferometer model , as described in sec .
[ subsec : development_of_a_sensing_scheme ] . due to different types of imperfections ,
which inevitably occur in any real interferometer , the extracted heterodyne signals were likely to deviate from those obtained from the strongly idealized numerical model . among these imperfections
are e.g. non - ideal mode matching and the presence of higher - order modes , non - optimal demodulation phases , deviations from nominal lengths and also the coupling of length fluctuations from yet unstabilized dof , which was not included in the model .
several iterations were necessary to optimize the relevant parameters and settings to fully lock the interferometer for the first time . in this stage of the experiment
the ccd camera in the detection port helped to determine whether the interferometer was accidentally locked on some residual higher order mode or on the desired fundamental mode .
as a figure of merit , and also to document the transition of the interferometer from the initial , unlocked state to a fully locked final state , light powers in different optical ports along with the correction signals generated by the servo controllers were monitored during the acquisition phase and after lock was acquired .
we used two digital storage oscilloscopes to record the data of two consecutive exemplary measurements which we present in figs .
[ fig : dc_levels_locking ] and [ fig : control_signals_locking ] .
the shaded regions in the plots represent different stages of the lock acquisition process : 1 .
all dof unlocked , mirror positions randomly fluctuating .
2 . mich manually tuned close to the dark fringe , prc length locked ( red traces ) .
mich locked on the dark fringe using p - polarized light ( green traces ) .
the tsrm position was scanned by applying a triangular drive to its pzt .
the srm was tuned manually to its operating point to obtain a decent error signal for feedback to the tsrm , to lock @xmath7 ( yellow traces ) .
the position of srm stabilized to its designated operating point , @xmath8 locked ( blue traces ) .
all longitudinal dof are electronically stabilized .
proportional gains of the servo loops adjusted to optimize loop suppression . in fig .
[ fig : control_signals_locking ] , when @xmath8 lock is acquired at 1.4 minutes , we can observe a correlated jump in the srm and tsrm correction signals .
this is due to the attempt of the tsrm servo to maintain the operating point for @xmath7 , i.e. the servo commands the tsrm to swiftly follow the srm when the srm position is locked .
the p - polarization - based probing of the mich degree of freedom gave rise to a small amount of s - polarized light leaking out of the asymmetric port .
this is due to a non - degeneracy of s- and p- polarized light with respect to covered optical path lengths in the interferometer and resulted in a slight offset of the michelson operating point from the dark fringe for the s - polarized field .
special care needed to be taken to prevent oscillations to occur in the servo controllers during acquisition . due to mutual dependencies of the control loops , a consequence of the coupling of the longitudinal dof , the proportional gains of the servo controllers needed to be set carefully , to thoroughly balance the suppression of disturbances that would cause immediate lock - loss versus loop instability .
we also found that slightly tuning the higher sideband frequency improved the robustness of our locking scheme .
this could be attributed to small deviations of the recycling cavity lengths from their nominal values .
our optimizations of the experiment led to a gradual increase of the time the interferometer remained in a fully locked state of up to tens of minutes .
the limited duration of the locks could be explained by long - term drifts in the longitudinal dof of the interferometer , e.g. due to thermal expansion of components in the setup but also due to drifts of the input laser frequency .
drifts exceeding the range of the pzt actuators eventually led to lock - loss of the interferometer . in the example shown in fig .
[ fig : dc_levels_locking ] and fig .
[ fig : control_signals_locking ] , lock of all four longitudinal dof was acquired 1.4 minutes after the measurement started and lasted for more than 20 minutes until the actuator range of the prm was exceeded .
a significant improvement of the duty cycle could be expected from operating the interferometer in vacuum .
this was , however , not within the scope of the experiment described here .
in this paper we described the design , the implementation and the successful operation of a laboratory - scale power recycled michelson interferometer with twin signal recycling . with the aid of numerical simulations we derived a length sensing and control scheme which we successfully applied to the experiment , to stably lock the interferometer in its four longitudinal degrees of freedom .
after a series of iterative optimizations we were able to operate the fully locked interferometer on time scales of tens of minutes , limited by long - term drifts in the optical setup exceeding the range of the actuators .
based on the setup discussed in this paper , the aspect of squeezing injection into a twin signal recycling interferometer was investigated in a subsequent experiment .
results of this work were already published in @xcite .
future work will include the transfer of the twin signal recycling technique to other detector topologies , e.g. michelson interferometers with fabry - perot arm cavities , which will form the basis of all second generation as well as the planned third generation observatories , e.g. the einstein telescope @xcite .
this work has been supported by the international max planck research school ( imprs ) and the cluster of excellence quest ( centre for quantum engineering and space - time research ) . | the techniques of power recycling and signal recycling have proven as key concepts to increase the sensitivity of large - scale gravitational wave detectors by independent resonant enhancement of light power and signal sidebands within the interferometer .
developing the latter concept further , _ twin signal recycling _ was proposed as an alternative to conventional detuned signal recycling .
twin signal recycling features the narrow - band sensitivity gain of conventional detuned signal recycling but furthermore facilitates the injection of squeezed states of light , increases the detector sensitivity over a wide frequency band and requires a less complex detection scheme for optimal signal readout .
these benefits come at the expense of an additional recycling mirror , thus increasing the number of degrees of freedom in the interferometer which need to be controlled . in this article
we describe the development of a length sensing and control scheme and its successful application to a tabletop - scale power recycled michelson interferometer with twin signal recycling .
we were able to lock the interferometer in all relevant longitudinal degrees of freedom , enabling the long - term stable operation of the experiment .
we thus laid the foundation for further investigations of this interferometer topology to evaluate its viability for the application in gravitational wave detectors .
99 b. willke for the geo collaboration , `` the geo - hf project , '' class .
quantum grav .
* 23 , * s207s214 ( 2006 ) .
g. m. harry for the ligo scientific collaborations , `` advanced ligo : the next generation of gravitational wave detectors , '' class
. quantum grav .
* 27 , * 084006 ( 2010 )
. the virgo collaboration , `` status of the virgo project , '' class .
quantum grav .
* 28 , * 114002 ( 2011 ) .
k. somiya for the kagra collaboration , `` detector configuration of kagra the japanese cryogenic gravitational - wave detector , '' class .
quantum grav .
* 29 , * 124007 ( 2012 ) .
b. j. meers , `` recycling in laser - interferometric gravitational - wave detectors , '' phys .
rev .
d * 38 , * 23172326 ( 1988 ) .
s. hild , h. grote , m. hewitson , h. lck , j. r. smith , k. a. strain , b. willke , and k. danzmann , `` demonstration and comparison of tuned and detuned signal recycling in a large - scale gravitational wave detector , '' class .
quantum grav .
* 24 , * 15131523 ( 2007 ) . c. m. caves , `` quantum - mechanical noise in an interferometer , '' phys .
rev .
d * 23 , * 16931708 ( 1981 ) .
r. schnabel for the ligo scientific collaboration , `` a gravitational wave observatory operating beyond the quantum shot - noise limit , '' nature physics * 7 , * 962965 ( 2011 ) .
j. harms , y. chen , s. chelkowski , a. franzen , h. vahlbruch , k. danzmann , and r. schnabel , `` squeezed - input , optical - spring , signal - recycled gravitational - wave detectors , '' phys .
rev .
d * 68 , * 042001 ( 2003 ) h. j. kimble , y. levin , a. b. matsko , k. s. thorne , and s. p. vyatchanin , `` conversion of conventional gravitational - wave interferometers into quantum nondemolition interferometers by modifying their input and/or output optics , '' phys .
rev .
d * 65 , * 022002 ( 2002 ) .
a. thring , r. schnabel , h. lck , and k. danzmann , `` detuned twin signal recycling for ultra - high precision interferometers , '' 985 ( 2007 ) .
a. thring , c. grf , h. vahlbruch , m. mehmet , k. danzmann , and r. schnabel `` broadband squeezing of quantum noise in a michelson interferometer with twin - signal - recycling , '' 824 ( 2009 ) .
h. vahlbruch , s. chelkowski , b. hage , a. franzen , k. danzmann , and r. schnabel , `` demonstration of a squeezed - light - enhanced power- and signal - recycled michelson interferometer , '' 211102 ( 2005 ) .
a. freise , g. heinzel , h. lck , r. schilling , b. willke , and k. danzmann , `` frequency - domain interferometer simulation with higher - order spatial modes , '' class . quantum grav . * 21 , * 1067 ( 2004 )
. k. a. strain , g. mller , t. delker , d. h. reitze , d. b. tanner , j. e. mason , p. a. willems , d. a. shaddock , m. b. gray , c. mow - lowry , and d. e. mcclelland `` sensing and control in dual - recycling laser interferometer gravitational - wave detectors , '' 1244 ( 2003 ) .
m. w. regehr , f. j. raab , and s. e. whitcomb , `` demonstration of a power - recycled michelson interferometer with fabry - perot arms by frontal modulation , '' 1507 ( 1995 ) . r. w. p. drever , j. l. hall , f. v. kowalski , j. hough , g. m. ford , a. j. munley , and h. ward `` laser phase and frequency stabilization using an optical resonator , '' appl .
phys .
b * 31 , * 97105 ( 1983 ) .
j. mizuno , `` comparison of optical configurations for laser - interferometric gravitational - wave detectors , '' phd thesis , ( universitt hannover , 1995 ) .
m. punturo et al . `` the einstein telescope : a third - generation gravitational wave observatory , '' class .
quantum grav .
* 27 , * 194002 ( 2010 ) . |
with the recent development of large format infrared array detectors , high quality photometric surveys are routinely conducted at wavelengths between 12.5 @xmath3 m . soon the completion of the 2 micron all sky survey ( 2mass ; skrutskie 1997 ) and denis ( epchtein et al .
1997 ) will provide comprehensive catalogues of near infrared sources with detection limits sensitive to a wide variety of stellar and non stellar objects .
infrared spectra will be required for appropriate identification of many of these sources , and for further study of their astrophysical properties .
the pioneering study of johnson and mendez ( 1970 ) was the first to explore the spectra of a large sample of normal stars in the near infrared .
however many years passed before improvements in instrumentation made possible similar observations of large numbers of targets of astrophysical interest .
the majority of the work done in near infrared spectroscopy to date has been focused on the k band , in large part because intrinsically cool or heavily obscured objects are typically brighter at k
band than in the j or h bands . in 1986 ,
kleinmann and hall ( 1986 ; kh86 ) provided the first comprehensive medium resolution atlas ( @xmath4 ) of stellar spectra in the k - band , covering all luminosity classes , but restricted to spectral types between f8-m7 .
more recently , wallace and hinkle ( 1997 ; wh97 ) have extended the kh96 k band atlas , using the same fts spectrograph on the kpno 4 m with @xmath4 , but including stellar spectra spanning spectral types o - m and luminosity classes i - v .
they also summarize the considerable body of work directed toward k band spectroscopy in the last decade . while in many situations
, the k band will be the wavelength selection of choice for spectroscopic studies of highly obscured or very cool objects , the presence of circumstellar dust ( @xmath5 ; pollack et al .
1994 ) often results in significant excess continuum emission longward of @xmath6 m .
this continuum excess is commonly found in two important classes of objects : young stars with circumstellar disks ( e.g. meyer , calvet , & hillenbrand , 1997 ) and evolved stars with extensive envelopes from mass - loss ( e.g. le bertre 1997 ) . near
infrared excess due to warm dust can also complicate spectroscopic studies of composite stellar systems aimed at discerning the stellar populations of other galaxies ( e.g. schinnerer et al .
continuum excess longward of @xmath6 m will weaken or even render invisible the photospheric features in the k band , while the photosphere will dominate at shorter wavelengths .
in such a situation , near infrared spectra shortward of 2 @xmath3 m will be required to see the stellar photosphere too obscured to be detected optically . to date
, there has been relatively little work in the h band ( 1.551.75 @xmath3 m ) .
recent publications include : i ) observations of 40 g , k , and m stars of luminosity class i and iii at @xmath7 by origlia , moorwood , and oliva ( 1993 ) ; ii ) the library of 56 spectra o m of luminosity class i , ii , and v at @xmath8 ( lancon & rocca volmerange , 1992 ) ; iii ) a library of 37 stars of luminosity classes i , iii , v at @xmath9 ( dallier , boisson , & joly 1996 ) over a limited portion of the h band ; and iv ) a study of 9 ob stars at @xmath10 ( blum et al . 1997 ) . here
we present an h band spectral atlas for 85 stars of nearly solar abundance with spectral types on the mk classification system ranging from 07-m5 and luminosity classes i - v .
these @xmath4 spectra were collected with the same fts at the kpno 4 m as the k band atlases of kh86 and wh97 . in section 2
, we describe the sample selection and in section 3 we describe the observations and calibration of the data . in section 4
we discuss the dependence of the spectral features on temperature and luminosity and suggest a two dimensional classification appropriate for late - type stars . in section 5
we discuss near ir spectral classification with regard to wavelength range / spectral resolution , and conclude with a summary of our results .
in our sample selection , we chose optically visible stars which had previously been identified on the temperature and luminosity scales of the revised mk system ( keenan 1987 ) .
the majority of the stars were drawn from the following fundamental lists : i ) morgan , abt , and tapscott ( 1978 ) for 29 stars o6g0 ; ii ) keenan and mcneil ( 1989 ) for 45 stars g0k5 ; and iii ) kirkpatrick , henry , and mccarthy ( 1991 ) for 5 late type dwarfs k5m3 .
we supplemented these primary standards with an additional 5 secondary standards from the compilation of jaschek , conde , and de sierra ( 1964 ) and one late
type dwarf classified by henry , kirkpatrick , and simons ( 1994 ) . in order to cover as complete a range of stellar temperature and luminosity as possible , we defined a two - dimensional grid with 26 bins of spectral type and three bins of luminosity class .
our temperature grid is binned @xmath11 more coarsely in spectral subclass than the revised mk system , so that we typically sample only every other mk subclass .
the three luminosity bins are divided into supergiants ( i ii ) , giants ( iii ) , and subgiants / dwarf stars ( iv v ) .
a full sampling of this grid would have resulted in 78 distinct temperature / luminosity pairs .
our atlas includes a total of 85 sources with 53 of the bins filled .
grid coverage was finer among the later spectral types , where for stars go and later , we filled 26 of the 27 bins ( 9 spectral types @xmath12 3 luminosity classes ) .
in contrast , for stars earlier than go , only 27 of the 51 bins were covered ( 17 spectral types @xmath12 3 luminosity classes ) .
the 85 individual stars in our h band survey are listed in table [ sample ] along with relevent stellar properties taken from the bright star catalogue ( hoffleit & jaschek 1982 ) .
additional restrictions on the sample selection included : i ) v@xmath13 @xmath14 250 km s@xmath15 with exception of hr3982 ( b7 v ) ; ii ) near - solar metallicity ( avoiding those mk standards which exhibit spectral peculiarities due to enhanced or deficient metal abundance ) ; and iii ) no visual companions within the beam ( separations @xmath16 ) .
our program was begun in advance of the k band fts atlas of wh97 , and their sources are drawn in large measure from our sample .
we note in table [ sample ] the stars for which k band spectra can be found in the wh97 digital atlas .
table [ temp ] and figure [ fig1 ] provide additional insight into the temperature and luminosity coverage of our sample . in table
[ temp ] , we list each of the spectral type and luminosity bins we have `` filled '' . for each bin
in which there is at least one spectrum , we give the corresponding effective temperature . for most stars
, we adopted the temperature scale of tokunaga ( 1996 ) , except for giants earlier than g0 where we adopted schmidt
kaler ( 1982 ) .
figure [ fig1 ] provides a schematic illustrating the temperature and luminosity coverage for the 85 stars in our sample . in this illustration
we have applied the same main sequence bolometric corrections to both dwarf ( 27 ) and subgiant ( 11 ) stars ; as such they are indistinguishable in this diagram .
observations of our 85 sample stars were obtained at the mayall 4 m telescope at kitt peak national observatory during four separate observing runs from 19931994 ( table [ ftslog ] ) .
we used the fourier transform spectrometer ( fts ) dual output interferometer ( hall et al .
the fts was ideal for this program for several reasons .
first , the wavelength coverage of the fts is limited only by the bandpass of the blocking filters , independent of the spectral resolution .
this gave us complete coverage in the j and h bands which would have been difficult to obtain with available grating spectrographs .
for example , our h band spectral range is a factor of two greater than the spectra of dallier et al .
secondly , the spectral resolution is fixed by the path difference scanned with the interferometer so we were able to chose the highest resolution possible and achieve s / n in excess of 75 for the majority of our sources . finally , because of the novel background subtraction algorithm of the 4 m fts described below , we were able to observe the brightest stars in our sample ( @xmath17 ) during good daytime conditions ( typically mornings ) .
combining daytime observations with targeted nighttime observations of key faint sources , the fts provided a uniform set of high quality spectra for a large sample of spectral standards .
our observing program included simultaneous spectral coverage in both the j band and the h band .
however , the j band data presented difficulties which made it expeditious to focus our initial effort on the h band . the primary problems with the j - band spectra were ; i ) the inherent difficulty in data reduction due to rapid temporal variations in telluric water vapor absorption ; and ii ) and the relative paucity of strong features which would allow spectral classification over the full range of stellar temperature and luminosity .
we defer discussion of the j band spectra to a future contribution .
spectra were collected simultaneously in the j and h bands with the use of a dichroic beamsplitter to separate the wavelengths longward and shortward of 1.5 @xmath3 m .
each star was centered within an input aperture of 3.8 arcsec while sky background was measured through an identical aperture 50.0 arcsec away .
the interferogram was scanned at a rate of 1 khz as the path difference was varied continuously from 0.00.75 cm providing an unapodized resolution of 0.8 cm@xmath15 .
data were obtained as separate scan pairs , with the path difference varied first in one direction and then the other .
a forward backward scan pair was treated as an `` observation '' and observations were repeated in beam switching mode ( a b b a ) . because the sky background from each aperture produces an interferogram shifted in phase by 180@xmath19 at each set of detectors , source spectra are background subtracted in fourier space as they are collected .
this permits observations of bright stars to be obtained during good daytime conditions .
these beam switched observations were repeated and scans were averaged until adequate signal to noise ratio ( snr ) was achieved .
the interferograms were transformed at kitt peak national observatory yielding spectra in units of relative flux versus wavenumber ( @xmath20 in cm@xmath15 ) .
the transformed spectra were converted into fits format images and all data reduction was performed using the iraf software package .
the spectra were then convolved with a gaussian filter of half
width @xmath21 cm@xmath15 .
this procedure , commonly referred to as apodization , eliminates `` ringing '' observed in the fts spectra due to the finite scan path of the interferometer . the resulting apodized resolution ( rayleigh criterion )
was @xmath22 cm@xmath15 giving a mean resolving power of @xmath4 in the h band . at this stage
the j and h band spectra were separated for ease of reduction .
the slope of the continuum was normalized to 1.0 using a four segment spline fitting function .
care was taken to keep the residuals from this fit to within 1 % .
next we corrected the spectra for telluric absorption features present in the spectra which varied with zenith angle .
we attempted to construct an opacity map for the earth s atmosphere by dividing normalized spectra obtained of the a0 star standards at different airmass . because of the simplicity of the a0 star spectra , showing primarily hydrogen lines in absorption , it was relatively easy to monitor the degree to which this procedure was successful .
in dividing two normalized spectra of the same star taken at different airmass , all stellar photospheric absorption features should directly cancel , leaving only those absorption features due to the earth s atmosphere .
if we assume that the opacity of the telluric absorption is directly proportional to airmass we derive : @xmath23\ ] ] where @xmath24 is the atmospheric opacity , @xmath25 is the low airmass value , and @xmath26 is the high airmass value .
a typical opacity map derived in this way for the h band is shown in figure [ fig2 ] .
several of the features in this map identified with known constituents of the earth s atmosphere such as water vapor , methane , and carbon dioxide , are denoted in figure [ fig2 ] . again
if the atmospheric opacity varies linearly with airmass we can simply scale the opacity for each star so that @xmath27 . using this technique
we corrected the spectra to zero airmass ; @xmath28 we used the highest signal to noise a0 standard star spectra ( @xmath29 ) with the largest @xmath30 to define the opacity .
we found some residual telluric absorptions , possibly due to water vapor which do not vary strictly with airmass .
such features severely complicate the reduction of the j band spectra .
finally , the forward and backward scans of each star were averaged and residuals of the differenced spectra were calculated in order to evaluate the average snr .
the observations were obtained with the goal of achieving snr of 75 or greater . in most cases
this was achieved with the highest quality spectra reaching values of several hundreds .
the average snr for each stellar spectrum is included in table [ ftshewsup ] table [ ftshewdw ] below .
representative h band spectra are shown in figures [ fig3] [ fig6 ] for luminosity classes i ii , iii , iv , and v , with prominent atomic and molecular features identified .
line identifications were made for the strongest lines from comparison with the solar photospheric and umbral near infrared atlases ( livingston & wallace 1991 ( lw91 ) ; wallace & livingston 1992 ( wl92 ) ) .
however , at our moderate spectral resolution many features are blended and we found the model atmosphere calculations of oliva , moorwood , and origlia ( 1993 ) to be useful in identifying the dominant contributors to a blend in late type stars .
visual inspection of the features in figures [ fig3] [ fig6 ] reveals that @xmath4 h band spectra contain sufficent temperature and luminosity sensitive features to enable spectral classes to be distinguished . beginning with the early type stars , the dominant spectral features are hei 5882 cm@xmath15 ( 1.700 @xmath3 m ) and the brackett series of hydrogen from lines 4 - 10 ( 1.736 @xmath3 m ) to 4 - 16 ( 1.556 @xmath3 m ) .
the he i line exceeds the strength of the brackett lines in the very earliest stars ( 06 to b0 ) , with a maximum equivalent width of @xmath31 cm@xmath15 ( hr1903 ; b0 ia ) , and recedes to undetectable levels ( @xmath32 cm@xmath15 ) by spectral type b8 . from the late b to early f stars , the brackett series dominates the spectrum , after which lines of neutral atomic metals begin to take prominence .
the strongest metallic lines include mgi , sii , cai , ali , and fei , which increase in strength toward the k stars .
finally molecular features of oh and co dominate the spectra of the latest type stars from k5m5 .
the most striking luminosity - sensitive feature is the second overtone co bandhead @xmath33 $ ] at 6177 cm@xmath15 ( 1.619 @xmath3 m ) , which is found in the spectra of the k and m stars .
this feature is signficantly stronger in stars of lower surface gravity at equivalent spectral type . to further enable spectral classification in the h band ,
we have identified a set of 9 features which are prominent in stars of spectral type a - m .
these include a relatively isolated brackett line ( h4 - 11 ) , 5 neutral metals , and 3 molecular bands . in table
[ ftshband ] , we define 9 narrow band indices with bandpasses ranging from 10 to 50 cm@xmath15 , which include each of these features . the variable widths of the bandpasses were selected to minimize line blending , contamination from residual telluric absorption , and sensitivity to radial velocity shifts .
table [ ftshband ] also identifies the wavenumber of the dominant contributor and the lower state energy level , the central wavenumber and passband of the index , and additional species which may contribute to the index strength .
the equivalent widths of these 9 indices were evaluated from the normalized spectra of our 85 survey stars , and are tabulated in tables [ ftshewsup ] to [ ftshewdw ] in units of cm@xmath15 .
\times 10 ^ 8 $ ] ] uncertainties in these equivalent widths depend on the snr of the spectrum in question and the bandpass / strength of the feature .
errors range from @xmath34 cm@xmath15 exceeding this upper limit in very few cases .
multiple observations of several sources are listed for comparison .
the temperature and luminosity dependence for four representative indices is illustrated in figure [ fig7 ] .
the 411 brackett line ( hi5950 ) behaves as expected , with a rapid rise to a maximum ( at a peak equivalent width of @xmath35 3 cm@xmath15 ) as @xmath36 approaches 10000 k , and a slower decline toward higher temperatures .
the behavior of the index is similar in both the dwarfs and the giants , although the luminosity class i / ii sources show a larger scatter , presumably due to intrinsic variability ( e.g. kaufer et al .
the general behavior of the neutral atomic features is illustrated by the mg6345 index . in luminosity classes
iv v , this index reaches a maximum strength between 5000 - 6000 k , with a peak equivalent width of @xmath35 2.5 cm@xmath15 .
in contrast , the maximum strength of this index in the lower surface gravity objects ( also @xmath35 2.5 cm@xmath15 ) is found in the coolest stars in our sample , monotonically decreasing toward higher temperatures ; as expected given the behavior of ionization state as a function of surface gravity .
the two sii indices exhibit similar behavior , but the ali index , ( not shown ) turns over at much lower temperatures in our dwarf stars because of its lower ionization potential .
note that we have chosen not to form an index based on the strongest sii line at 6292 cm@xmath15 ( 1.5892 @xmath3 m ) because it is coincident with the 414 brackett line of hi at 6297 cm@xmath15 ( 1.5881 @xmath3 m ) .
the behavior of the molecular features is illustrated for both the second
overtone @xmath37co ( 6,3 ) and the oh ( @xmath38 ) indices .
both indices exhibit a similar behavior with temperature and luminosity , becoming detectable around @xmath36 = 5000 , with a strength in the giants approximately twice that in the dwarfs .
similar behavior was noted by kh86 in the first
overtone co features in the k band . in dwarf stars
the second overtone co index reaches a maximum before m5 , and displays a turnover toward the coolest stars
. this may be due in part to features of cai and fei which contaminate the index for intermediate spectral types ( f5k3 ) .
ali et al . ( 1995 ) find that the relationship between t@xmath39 and equivalent widths of the first
overtone co bandheads flatten out between 35005000 k in dwarf stars . from high resolution ( @xmath40 )
fts spectra , wallace & hinkle ( 1996 ) observed that the 2 @xmath3 m continuum in late
type dwarf stars is suppressed by numerous water vapor features which are blended at intermediate to low resolution .
predicting the equivalent widths of features where the apparent continuum is subject to temperature and luminosity effects is not straight forward .
in contrast , both the co and oh indices continue to rise at the coolest temperatures for stars of higher luminosity . however , the magnitude ( and temperature ) of the maximum in the dwarf stars differs between the co and oh indices , which we use in the next section to define a two - dimensional classification scheme for late type stars .
we note that the oh index begins to include a contribution from the stark broadened 4 - 11 brackett line at 5949 cm@xmath15 creating a secondary maxiumum in the strength of this index around 10,000 k in the dwarf stars .
while the temperature and luminosity dependence of the atomic features is readily understood through application of the saha and boltzman equations governing the population of the ionization states and energy levels respectively , the explanation behind the behavior of the molecular features is more subtle .
two possibilities for the factor of two enhancement in the molecular bands in the giants over the dwarfs have been explored in the literature .
one attributes the luminosity dependence in the molecular features to differing microturbulence in the atmospheres of dwarfs and giants .
the expectation is that larger microturbulence in the lower surface gravity giants effectively broadens the opacity of the feature over a larger frequency interval in these saturated features , thereby enhancing the equivalent width ( mcwilliams & lambert 1984 ) .
another possible contributor is the differing depth of the line formation region in the dwarfs versus the giants , which is fixed by the h@xmath41 opacity . as described by gray ( 1992 ) higher surface gravity results in a higher electron pressure ( and thus h@xmath42 column density ) .
this brings the co line formation region closer to the stellar surface reducing @xmath43 according to the following proportionality : @xmath44 in any case , this luminosity dependence of the band strength gives an excellent empirical discriminant between giants and dwarfs , which we exploit below to develop a two dimensional spectral index . to discern surface gravity effects between the super giants and giants or between sub giants and dwarf stars requires more careful study . a detailed examination of line strengths as a function of surface gravity at a fixed temperature
reveal the expected trends .
however , this behavior does not reveal itself in the coarse analysis afforded by our narrow band indices .
while the temperature and luminosity sensitivities outlined above can provide good spectral classification in many instances , discriminants which do not rely on absolute line strength are required when a star is subject to near - infrared continuum veiling . in this
case line ratio diagnostics are to be preferred , since absorption features will appear shallower in the presence of continuum excess but line ratios will be preserved as long as the excess is not strongly wavelength dependent .
we have identified one diagnostic based on line ratios which can be used to evaluate both temperature and luminosity for stars from k3-m5 in the presence of continuum veiling .
this two dimensional spectral index is defined as : @xmath45}{ew[mg6345 ] } \
vs. \ \frac{ew[co6018+co6170]}{ew[mg6345]}\ ] ] where ew is the equivalent width in cm@xmath15 for the indices identified in table [ ftshband ] and listed in tables [ ftshewsup] [ ftshewdw ] .
the temperature and luminosity dependence of this diagnostic is illustrated in figure [ fig8 ] . in this diagnostic ,
the ratio of the oh5920 to mg6345 indices is temperature sensitive , with distinct temperature dependences for dwarfs and giants .
specifically we find @xmath46}{ew[mg6345]}\ ] ] and @xmath47}{ew[mg6345]}\ ] ] the comparison of this temperature sensitive ratio with the sum of the two @xmath37co indices , also normalized to mg6345 , then provides an excellent means of identifying both the temperature and luminosity class of late type stars .
formal errors in the equivalent width suggest that spectral types can be evaluated to within @xmath2 subclasses ( @xmath48 300 k ) from k3m5 using spectra with @xmath1 based on these indices alone .
spectral classification in the near - infrared will become increasingly important in the next decade , as the 2mass and denis near - infrared sky surveys reveal unprecented numbers of stars which are optically invisible . because the 12.5 @xmath3 m region is on the rayleigh
jeans tail of most stellar seds , it is not an ideal wavelength regime to pursue spectral classification . yet
, there are sufficient features in both the h and the k bands to allow most stellar photospheres to be classified . for heavily reddened sources , the k band will be the wavelength of choice .
however , continuum emission from circumstellar dust with temperatures less than 2000 k can heavily veil stellar photospheres at wavelengths greater than 2.0 @xmath3 m . in this case ,
shorter wavelength spectra are required in order to identify the underlying star .
the early - type stars are probably the most challenging for spectral classification in the near infrared .
we find that a rough classification from 07-b8 can be made in the h
band by the relative strengths of hei 5882 cm@xmath15 and the brackett series ( see also blum et al .
hanson , conti , and rieke ( 1995 ) have established a classification scheme in the k
band for o b stars relying on lines of helium as well as higher ionization species obtained at @xmath49 . for stars
a through early k , the h band may be superior to the k band in providing a large number of intermediate ionization potential species with strong features such as mgi , sii , and fei in addition to the numerous brackett series features .
stars k3m5 are probably best classified in the k band ( kh86 ; ali et al .
1995 ; wh97 ) using atomic features of mgi , cai , and nai as well as the first
overtone co bandheads observed at @xmath50 .
however we have found that these stars also have strong temperature and luminosity sensitive features in the h band such as mgi , ali , oh , and the second overtone co bandheads .
the very latest type stars ( @xmath51 ) have very strong , broad , molecular features which can be identified at resolutions as low as @xmath52 .
kirkpatrick et al .
1993 ( see also jones et al .
1996 ) have classified stars in the i or the j band employing features due to vo , tio , and feh . in addition , broad water vapor bands observed throughout the 12.5 @xmath3 m region ( jones et al .
1995 ) are an important opacity source in the atmospheres of the coolest stars as well as brown dwarfs ( allard & hauschildt 1995 ) . while the i or j bands are probably the best spectral regions to classify extremely cool stars ( as they lie on the wien side of the planck function for these objects ) more heavily obscured objects can still be profitably observed at low resolution in the j and h bands ( e.g. nicmos on hst ) or in the k band ( wilking , greene , & meyer 1998 ) in search of these water vapor absorptions . the h band spectral atlas we have presented is comprised of moderate resolution spectra with @xmath53 . in contrast , most spectral classification is typically carried out with @xmath54 .
the strongest and broadest features in the h band are the co(6 - 3 ) bands and the brackett lines .
these features could be identified with much lower spectral resolution than our survey , at @xmath55 .
the most crowded region in the h band spectra is that in the vicinity of the hi line at 5948.50 cm@xmath15 ( 1.68110 @xmath3 m ) , the ali triplet at 59645980 cm@xmath15 ( 1.6771.672 @xmath3 m ) , and the sii line at 5993.29 cm@xmath15 ( 1.66853 @xmath3 m ) . in order to properly separate these important features from each other ,
a resolving power of @xmath50 is required . at this resolution
, one can also obtain measurements of the hei line at 5882 cm@xmath15 ( 1.700 @xmath3 m ) . at @xmath56
one can resolve individual components of the ali triplet , the mg i doublet , and the co bandheads , as well as the stark broadened brackett lines in the early type dwarf stars ( table [ ftshband ] ) .
an additional issue in the near infrared is the significant contribution to shot - noise from air - glow lines . in the h band air
glow from oh is sufficently bright and variable that they compromise @xmath57 h band spectral classification for very faint sources .
spectral resolution as high as @xmath58 will be required to resolve the bulk of these air glow features and to obtain adequate snr spectra of faint objects . in summary
, we present an h band spectral atlas at a resolving power of @xmath56 that spans a wide range in stellar temperature ( o7-m5 ) and luminosity class ( i - v ) .
this spectral region contains a number of temperature and/or luminosity sensitive atomic and molecular features which will allow spectral classification to be carried out in the h - band . as an example of the efficacy of this spectral range for distinguishing stellar spectral types
, we define a set of narrow band indices which , with @xmath59 , permit classification of late type stars on the mk system within @xmath2 subclasses .
it appears however , that for most applications obtaining h band spectra at @xmath50 will be sufficient for classification .
the final reduced averaged spectra as well as the difference of the forward and backward scan pairs ( see section 3 for description of the reduction procedure ) are available through the astronomical data center ( adc ) for each observation listed in this paper .
the adc can be contacted directly : i ) by post at astronomical data center , nasa goddard space flight center , code 631 , greenbelt , md 20771 ; or ii ) by telephone at ( 301 ) 2868310 ; or iii ) by fax at ( 301 ) 2861771 ; or iv ) via the internet at http://adc.gsfc.nasa.gov .
the data are in fits format with pertinent header information included for each image .
these fits format files , useful plotting routines , and other relevent information are also available on the world wide web at http://donald.phast.umass.edu .
the raw fts data are also available directly from noao ( contact khh for details ) .
we would like to thank lori allen , ed chang , lynne hillenbrand , susan kleinmann , michael skrutskie , and lloyd wallace for helpful discussions .
special thanks to john carpenter for assisting in the initial compilation of the standard star lists , and to karen strom and stephen friedman for their assistance in making the data available electronically .
antonella romano provided assistance in preparing the tables and figures for publication .
support for mrm during the final stages of this work was provided by nasa through hubble fellowship grant # hf01098.0197a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa under contract nas 526555 .
se acknowledges support from the national science foundation s faculty award for women program . this work was supported in part through a grant from the national science foundation ( # ast9114863 ) to ses .
schmidt kaler , t.h .
1982 , physical parameters of stars , landolt - bornstein new series , vol . 2b , astronomy and astrophysics , stars and star clusters , ed . k. shaifers & h. h. voigt , ( springer verlag : new york ) lllllll & & & & & & + 1903 & * 46 @xmath60 ori & 2.4 & b0 & ia & 26sb & 87 + 1203 & 44 @xmath61 per & 3.4 & b1 & ib & 20sb & 59 + 2827 & 31 @xmath62 cma & 2.5 & b5 & ia & 41v & 45 + 1713 & * 19 @xmath63 ori & 0.1 & b8 & ia & 21sb & 33 + 3975 & * 30 @xmath62 leo & 3.3 & a0 & ib & 3v & 20 + 7924 & * 50 @xmath64 cyg & 1.0 & a2 & ia & -5sbo & 21 + 1865 & 11 @xmath64 lep & 2.0 & f0 & ib & 24 & 13 + 1017 & * 33 @xmath64 per & 0.9 & f5 & ib & -2v & 18 + 7796 & * 37 @xmath65 cyg & 1.1 & f8 & ib & -8v & 20 + 8232 & * 22 @xmath63 aqr & 1.5 & g0 & ib & 7v ? & 18 + 7479 & 5 @xmath64 sge & 2.4 & g1 & ii & 2v ? & 0 + 7063 & @xmath63 sct & 2.2 & g4 & iia & -22sb10 & 10 + 8752 & * - & 3.6 & g4v & @xmath66 & -58v ? & 35 + 7314 & * 21 @xmath67 lyr & 2.1 & k0 & + ii & -31v & @xmath1419 + 6713 & 93 her & 2.4 & k0.5 & iib & -24 & @xmath1417 + 8465 & * 21 @xmath61 cep & 1.1 & k1.5 & ib & -18sb & @xmath1417 + 6498 & 49 @xmath20 oph & 1.9 & k2 & ii & -27 & @xmath1419 + 603 & 57 @xmath68 and & -0.5 & k3 & -iib & -12sb & @xmath1417 + 8089 & * 63 cyg & 1.5 & k4 & ib - iia & -26v & + 8079 & * 62 @xmath69 cyg & 0.5 & k4.5 & ib - ii & -20sb & @xmath1417 + 2061 & 58 @xmath64 ori & -2 . ? &
m1 - 2 & ia - iab & 21sb & + 1155 & * & 0.5 & m2 & + iiab & -3v & + 921 & 25 @xmath70 per & -1.7 & m4 & ii & 28 & + 7009 & * & 0.6 & m4.5-m5 & + ii & -19 & + 6406 & * 64 @xmath71 her & -2.4 & m5 & ib - ii & -33v & 21 + 1899 & 44 @xmath72 ori & 3.5 & o9 & iii & 22sb2o & 130 + 1552 & 3 @xmath73 ori & 4.1 & b2+b2 & iii & 23sbo & 40 + 5291 & * 11 @xmath64 dra & 3.5 & a0 & iii & -13sbo & 18 + 403 & * 37 @xmath74 cas & 2.3 & a5 & iii - iv & 7sb & 113 + 1412 & * 78 @xmath75 tau & 2.9 & a7 & iii & 40sb1o & 78 + 4031 & * 36 @xmath61 leo & 2.8 & f0 & iii & -16sb & 84 + 21 & * 11 @xmath63 cas & 1.6 & f2 & iii iv & 12sb & 70 + 5017 & * 20 cvn & 3.9 & f3 & iii & 8v ? & 17 + 2706 & 48 gem & 5.0 & f5 & iii - iv & 13v & 74 + 8905 & * 68 @xmath76 peg & 3.3 & f8 & iii & -11 & 79 + 4883 & * 31 com & 3.0 & g0 & iii & -1v ? & 77 + 4716 & 5 cvn & 2.8 & g6 & iii & -12sb & @xmath1417 + 7328 & 1 @xmath77 cyg & 1.6 & g9 & iii & -29sb & @xmath1417 + 7949 & 53 @xmath60 cyg & 0.2 & k0 & -iii & -11sb ? & @xmath1417 + 8317 & * 11 cep & 2.3 & k1 & iii & -37 & @xmath1417 + 6299 & * 27 @xmath77 oph & 0.8 & k2 & iii & -56v & @xmath1417 + 165 & 31 @xmath74 and & 0.5 & k3 & iii & -7sb1o & @xmath1417 + 6705 & * 33 @xmath65 dra & -1.2 & k5 & iii & -28 & @xmath1417 + 152 & & 1.7 & k5-m0 & iii & -33v & @xmath1417 + 4517 & * 3 @xmath78 vir & 0.3 & m1 & iiiab & 51v ? & + 6242 & & 0.4 & m4 & + iii - iiia & -7v ? & + 7886 & * & -0.6 & m6 & iii & -66v ? &
+ lllllll & & & & & & + 6588 & 85 @xmath72 her & 4.3 & b3 & iv & -20sb1o & 11 + 4033 & * 33 @xmath79 uma & 3.3 & a2 & iv & 18v & 48 + 1351 & 57 tau & 4.9 & f0 & iv & 42sb1 ? & 109 + 5235 & 8 @xmath62 boo & 1.5 & g0 & iv & 0sb1o & 13 + 5409 & 105 @xmath80 vir & 2.8 & g2 & iv & -10sb & 0 + 6623 & 86 @xmath3 her & 1.4 & g5 & iv & -16v & 20 + 995 & 59 ari & 3.9 & g6 & iv & 0v & + 7602 & 60 @xmath63 aql & 1.7 & g8 & iv & -40v & @xmath14 16 + 7957 & * 3 @xmath62 cep & 1.2 & k0 & iv & -87 & @xmath1417 + 5901 & 11 @xmath77 crb & 2.5 & k1 & iva & -24 & @xmath1417 + 6014 & & 3.6 & k1.5 & iv & -4v & + 2456 & * 15 mon & 5.5 & o7 & v(e ) & 33sb & 63 + 5191 & * 85 @xmath62 uma & 2.4 & b3 & v & -11sb ? & 205 + 3982 & * 32 @xmath64 leo & 1.6 & b7 & v & 6sb & 329 + 7001 & * @xmath64 lyr & 0.0 & a0 & v & -14v & 15 + 2491 & * 9 @xmath64 cma & -1.5 & a1 & vm & -8sbo & 13 + 4534 & * 94 @xmath63 leo & 2.0 & a3 & v & 0v & 121 + 4357 & * 68 @xmath74 leo & 2.3 & a4 & v & -20v & 181 + 4931 & 78 uma & 4.1 & f2 & v & -10v ? & 92 + 1279 & & 5.1 & f3 & v & 36sb1 ? & 25 + 2943 & * 10 @xmath64 cmi & -0.6 & f5 & iv - v & -3sbo & 6 + 1538 & 59 eri & 2.3 & f6 & v & 35 & + 4375 & * 53 @xmath69 uma & 3.0 & g0 & v & -16sb1o & 1 + 4983 & 43 @xmath63 com & 3.1 & f9.5 & v & 6sb ? & 6 + 483 & * & 3.7 & g1.5 & v & 4v ? & 2 + 4374 & 53 @xmath69 uma & 3.5 & g0 & v & -16sb1o & 3 + 5072 & 70 vir & 3.6 & g4 & v & 5v & 1 + 4496 & * 61 uma & 3.8 & g8 & v & -5v & @xmath1417 + 7462 & * 61 @xmath20 dra & 3.0 & k0 & v & 27v & @xmath1417 + 1084 & * 18 @xmath60 eri & 1.6 & k2 & v & 15v ? & @xmath1417 + 8832 & * & 3.2 & k3 & v & -18v & + & gl570a & 3.0 & k4 & v & - & - + 8085 & * 61 cyg & 2.4 & k5 & v & -64v & @xmath1417 + 8086 & 61 cyg & 3.1 & k7 & v & -64v ? & = @xmath1425 + & gl338a & 4.5 & m0 & v & - & - + & gl526 & 4.5 & m1.5 & v & - & - + & * gl411 & 3.6 & m2 & v & - & - + & * gl725a & 4.7 & m3 & v & - & - + rlll & & & + o6-o8 & & 37000 & 38000 + o9 & 32500 & 32000 & 33200 + o9.5 & & & 31450 + b0 & 26000 & 29000 & 29700 + b1 & 20700 & 24000 & 25600 + b2 & 17800 & 20300 & 22300 + b3 & 15600 & 17100 & 19000 + b4 & 13900 & & 17200 + b5 & 13400 & 15000 & 15400 + b6 & 12700 & 14100 & 14100
+ b7 & 12000 & 13200 & 13000 + b8 & 11200 & 12400 & 11800 + b9 & 10500 & 11000 & 10700 + a0 & 9730 & 10100 & 9480 + a1 & 9230 & 9480 & + a2 & 9080 & 9000 & 8810 + a5 & 8510 & 8100 & 8160 + a7 & & 7650 & 7930 + f0 & 7700 & 7150 & 7020 + f2 & 7170 & 6870 & 6750 + f5 & 6640 & 6470 & 6530 + f7 & & & 6240 + f8 & 6100 & 6150 & + g0 & 5510 & 5910 & 5930 + g2 & & & 5830 + g3 & 4980 & & + g4 & & 5190 & 5740 + g6 & & 5050 & 5620 + g8 & 4590 & 4960 & + k0 & 4420 & 4810 & 5240
+ k1 & 4330 & 4610 & + k2 & 4260 & 4500 & 5010 + k3 & 4130 & 4320 & + k4 & & 4080 & 4560 + k5 & 3850 & 3980 & 4340 + k7 & & & 4040 + m0 & 3650 & 3820 & 3800 + m1 & 3550 & 3780 & 3680 + m2 & 3450 & 3710 & 3530 + m3 & 3200 & 3630 & 3380 + m4 & 2980 & 3560 & 3180 + m5 & & 3420 & 3030 + m6 & & 3250 & 2850 + lllllll & & & & & & + mgi(4s4p ) & 5.39 & 5843.41 & 1.71133 & 5844 & 10 & co , fe , ni , oh + oh(@xmath38 ) & 0.76 & 5920 : & 1.689 & 5920 & 20 & c , co , fe , ni + hi(4 - 11 ) & 12.75 & 5948.50 & 1.68110 & 5950 & 20 & co , fe , ni , si + ali(4p4d tr ) & 4.09 & 5963.76 & 1.67679 & 5972.5 & 25 & co , fe , ni , oh + & & 5968.31 & 1.67552 & & & + & & 5979.60 & 1.67235 & & & + sii(4p3d ) & 5.98 & 5993.29 & 1.66853 & 5993 & 10 & co , fe , ni , oh + @xmath37co(8,5)bh & 1.55 & 6018 & 1.662 & 6017.5 & 15 & fe , oh , s + @xmath37co(6,3)bh & 1.05 & 6177 & 1.619 & 6170 & 50 & ca , fe , ni , oh , si + sii(4p5s ) & 5.98 & 6263.92 & 1.59644 & 6264 & 10 & fe , mg , ni , oh + mgi(4s4p tr ) & 5.93 & 6341.10 & 1.57701 & 6345 & 20 & cn , co , fe , h@xmath81o , ni , oh + & & 6347.88 & 1.57533 & & & + & & 6351.22 & 1.57450 & & & + llllllllllll & & & & & & & & & & & + hr1903 & 26000 & 203 & -0.03 & 0.09 & 0.75 & 0.17 & -0.02 & 0.11 & 0.27 & 0.11 & 0.47 + hr1903 & 26000 & 167 & 0.01 & 0.07 & 0.73 & 0.11 & -0.00 & 0.09 & 0.19 & 0.12 & 0.32 + hr1203 & 20700 & 160 & -0.01 & 0.12 & 0.80 & 0.18 & 0.01 & 0.11 & 0.23 & 0.11 & 0.56 + hr2827 & 13400 & 045 & -0.05 & -0.02 & 1.06 & -0.04 & 0.20 & -0.04 & 0.01 & -0.05 & 0.08 + hr1713 & 11200 & 218 & 0.04 & 0.06 & 1.27 & 0.23 & 0.05 & 0.09 & 0.19 & 0.17 & 0.50 + hr3975 & 9730 & 068 & -0.08 & 0.11 & 2.12 & 0.23 & 0.12 & -0.09 & 0.33 & 0.20 & 0.58 + hr7924 & 9080 & 225 & -0.05 & 0.00 & 1.40 & 0.20 & -0.02 & 0.03 & -0.14 & 0.08 & 0.08 + hr1865 & 7700 & 196 & -0.01 & 0.22 & 2.21 & 0.32 & 0.14 & -0.13 & 0.30 & 0.25 & 0.52 + hr1017 & 6640 & 248 & 0.08 & 0.25 & 1.88 & 0.43 & 0.06 & -0.02 & 0.53 & 0.41 & 1.05 + hr7796 & 6100 & 324 & 0.10 & 0.29 & 1.58 & 0.69 & 0.14 & 0.03 & 0.79 & 0.60 & 0.94 + hr8232 & 5510 & 290 & 0.09 & 0.37 & 1.28 & 0.67 & 0.17 & 0.03 & 1.18 & 0.67 & 1.19 + hr8752 & 5510 & 084 & -0.13 & 0.34 & 1.16 & 0.14 & 0.07 & 0.02 & 0.16 & 0.16 & 0.41 + hr7479 & 5333 & 096 & 0.15 & 0.13 & 1.13 & 0.70 & 0.10 & 0.00 & 0.92 & 0.41 & 0.69 + hr7479 & 5333 & 109 & 0.10 & 0.22 & 1.07 & 0.49 & 0.11 & 0.02 & 0.76 & 0.54 & 1.04 + hr7479 & 5333 & 182 & 0.13 & 0.12 & 1.08 & 0.56 & 0.05 & 0.07 & 0.94 & 0.49 & 1.06 + hr7479 & 5333 & 093 & 0.06 & -0.02 & 1.20 & 0.53 & 0.03 & 0.14 & 1.05 & 0.54 & 0.93 + hr7063 & 4902 & 260 & 0.17 & 0.12 & 0.94 & 0.84 & 0.14 & 0.25 & 1.66 & 0.76 & 1.06 + hr7314 & 4420 & 193 & 0.23 & 0.15 & 0.84 & 0.88 & 0.15 & 0.46 & 2.25 & 1.00 & 1.60 + hr7314 & 4420 & 257 & 0.15 & 0.17 & 0.67 & 0.59 & 0.10 & 0.37 & 1.91 & 0.86 & 1.36 + hr6713 & 4375 & 265 & 0.22 & 0.19 & 0.81 & 0.80 & 0.16 & 0.40 & 1.86 & 0.79 & 1.25 + hr8465 & 4295 & 308 & 0.13 & 0.28 & 0.68 & 0.70 & 0.19 & 0.63 & 2.35 & 1.02 & 1.79 + hr6498 & 4260 & 247 & 0.24 & 0.30 & 0.84 & 1.00 & 0.21 & 0.83 & 2.75 & 1.11 & 1.84 + hr603 & 4130 & 453 & 0.24 & 0.34 & 0.69 & 0.61 & 0.19 & 0.49 & 2.09 & 0.87 & 1.46 + hr8089 & 3990 & 202 & 0.21 & 0.45 & 0.50 & 0.67 & 0.14 & 0.68 & 2.26 & 0.94 & 1.59 + hr8079 & 3920 & 445 & 0.24 & 0.67 & 0.62 & 0.80 & 0.24 & 1.05 & 2.70 & 1.05 & 1.72 + hr2061 & 3550 & 327 & 0.54 & 1.08 & 0.36 & 1.39 & 0.23 & 1.46 & 3.92 & 0.89 & 2.22 + hr1155 & 3450 & 237 & 0.46 & 1.31 & 0.58 & 1.12 & 0.38 & 1.37 & 3.35 & 1.12 & 2.03 + hr921 & 2980 & 225 & 0.77 & 1.41 & 0.29 & 1.27 & 0.37 & 1.10 & 4.08 & 0.67 & 1.52 + hr7009 & 2925 & 292 & 0.63 & 1.43 & 0.79 & 1.55 & 0.54 & 1.79 & 4.68 & 1.26 & 2.21 + hr6406 & 2800 & 319 & 0.59 & 1.41 & 0.80 & 1.66 & 0.54 & 1.85 & 4.98 & 1.31 & 2.31 + llllllllllll & & & & & & & & & & & + hr1899 & 32000 & 164 & -0.03 & 0.17 & 0.54 & 0.15 & -0.04 & 0.03 & 0.18 & 0.10 & 0.49 + hr1899 & 32000 & 182 & -0.04 & 0.20 & 0.62 & 0.11 & 0.05 & -0.03 & 0.07 & -0.00 & 0.07 + hr1552 & 20300 & 115 & -0.07 & 0.13 & 1.29 & 0.31 & 0.10 & -0.09 & 0.12 & -0.01 & 0.09 + hr5291 & 10100 & 159 & -0.07 & 0.19 & 2.65 & 0.78 & 0.02 & 0.05 & 0.28 & 0.11 & 0.41 + hr403 & 8100 & 202 & -0.02 & 0.39 & 2.79 & 0.77 & 0.09 & -0.13 & 0.25 & 0.08 & 0.54 + hr1412 & 7650 & 186 & -0.02 & 0.43 & 2.70 & 0.71 & 0.05 & -0.10 & 0.60 & 0.25 & 0.91 + hr4031 & 7150 & 193 & 0.02 & 0.29 & 2.30 & 0.54 & 0.12 & -0.09 & 0.25 & 0.13 & 0.48 + hr21 & 6870 & 245 & 0.03 & 0.32 & 1.99 & 0.55 & 0.04 & -0.11 & 0.20 & 0.22 & 0.82 + hr21 & 6870 & 179 & 0.01 & 0.25 & 2.21 & 0.63 & 0.13 & -0.14 & 0.46 & 0.24 & 0.69 + hr5017 & 6700 & 174 & 0.04 & 0.28 & 2.58 & 0.83 & 0.07 & -0.03 & 0.43 & 0.36 & 0.96 + hr2706 & 6470 & 086 & 0.01 & 0.39 & 2.08 & 0.47 & 0.16 & -0.10 & 0.13 & 0.20 & 0.64 + hr8905 & 6270 & 059 & 0.06 & 0.13 & 1.06 & 0.45 & -0.01 & -0.16 & 0.30 & 0.28 & 1.02 + hr4883 & 5910 & 280 & 0.15 & 0.20 & 1.07 & 0.52 & 0.09 & 0.04 & 0.90 & 0.43 & 1.02 + hr4716 & 5050 & 133 & 0.17 & 0.10 & 0.97 & 0.47 & 0.13 & 0.07 & 1.29 & 0.54 & 0.92 + hr7328 & 4885 & 281 & 0.19 & 0.20 & 0.78 & 0.57 & 0.11 & 0.11 & 1.33 & 0.67 & 1.35 + hr7949 & 4810 & 369 & 0.21 & 0.24 & 0.77 & 0.59 & 0.13 & 0.18 & 1.39 & 0.67 & 1.38 + hr8317 & 4710 & 191 & 0.29 & 0.14 & 0.63 & 0.58 & 0.08 & 0.26 & 1.79 & 0.83 & 1.64 + hr6299 & 4500 & 532 & 0.26 & 0.09 & 0.77 & 0.81 & 0.10 & 0.44 & 1.92 & 0.76 & 1.40 + hr165 & 4320 & 266 & 0.34 & 0.24 & 0.64 & 0.58 & 0.18 & 0.41 & 1.98 & 0.89 & 1.45 + hr6705 & 3990 & 298 & 0.39 & 0.60 & 0.77 & 1.12 & 0.26 & 0.96 & 3.05 & 1.12 & 1.75 + hr6705 & 3990 & 694 & 0.41 & 0.61 & 0.76 & 1.07 & 0.25 & 0.89 & 2.87 & 1.06 & 1.66 + hr152 & 3956 & 270 & 0.37 & 0.86 & 0.58 & 0.92 & 0.34 & 0.87 & 2.59 & 0.78 & 1.40 + hr4517 & 3780 & 673 & 0.66 & 1.06 & 0.33 & 1.07 & 0.21 & 0.89 & 3.14 & 0.70 & 1.93 + hr6242 & 3560 & 458 & 0.52 & 1.12 & 0.57 & 1.18 & 0.39 & 1.45 & 3.93 & 1.16 & 1.96 + hr7886 & 3250 & 574 & 0.61 & 1.54 & 0.89 & 1.71 & 0.64 & 2.25 & 5.69 & 1.41 & 2.42 + llllllllllll & & & & & & & & & & & + hr6588 & 19000 & 162 & -0.03 & 0.10 & 1.65 & 0.48 & -0.03 & 0.02 & 0.17 & 0.07 & 0.40 + hr4033 & 8810 & 146 & -0.00 & 0.33 & 2.62 & 0.73 & 0.06 & 0.02 & 0.27 & 0.13 & 0.35 + hr1351 & 7020 & 094 & -0.06 & 0.40 & 2.18 & 0.62 & 0.19 & -0.19 & -0.01 & 0.10 & 0.24 + hr5235 & 5930 & 341 & 0.16 & 0.12 & 1.21 & 0.83 & 0.11 & 0.03 & 1.30 & 0.64 & 1.14 + hr5235 & 5930 & 263 & 0.17 & 0.26 & 1.26 & 0.68 & 0.18 & -0.01 & 0.97 & 0.50 & 1.03 + hr5409 & 5830 & 158 & 0.16 & 0.16 & 1.02 & 0.51 & 0.12 & -0.03 & 0.91 & 0.44 & 0.95 + hr6623 & 5680 & 211 & 0.33 & 0.12 & 0.87 & 0.87 & 0.09 & 0.09 & 1.21 & 0.69 & 1.38 + hr995 & 5620 & 143 & 0.22 & 0.08 & 0.80 & 0.61 & 0.23 & 0.03 & 1.09 & 0.48 & 1.14 + hr7602 & 5430 & 056 & 0.46 & 0.15 & 0.71 & 0.86 & 0.01 & 0.17 & 1.40 & 0.57 & 1.49 + hr7957 & 5240 & 189 & 0.22 & 0.09 & 0.68 & 0.70 & 0.07 & 0.15 & 1.04 & 0.46 & 0.99 + hr7957 & 5240 & 192 & 0.09 & 0.18 & 0.56 & 0.52 & -0.01 & 0.09 & 0.75 & 0.40 & 1.17 + hr5901 & 5125 & 395 & 0.37 & 0.13 & 0.55 & 0.62 & 0.08 & 0.21 & 1.61 & 0.74 & 1.60 + hr5901 & 5125 & 153 & 0.37 & 0.09 & 0.75 & 0.93 & 0.15 & 0.30 & 1.54 & 0.71 & 1.13 + hr6014 & 5068 & 072 & 0.31 & -0.12 & 0.39 & 0.59 & 0.14 & 0.05 & 1.10 & 0.70 & 1.40 + hr6014 & 5068 & 133 & 0.42 & 0.11 & 0.58 & 0.55 & 0.18 & 0.09 & 1.48 & 0.65 & 1.24 + llllllllllll & & & & & & & & & & & + hr2456 & 38000 & 053 & -0.03 & 0.36 & 0.39 & 0.26 & 0.16 & 0.12 & -0.32 & 0.11 & 0.31 + hr2456 & 38000 & 073 & -0.12 & 0.18 & 0.48 & 0.10 & 0.06 & -0.15 & 0.15 & 0.03 & 0.10 + hr5191 & 19000 & 282 & -0.05 & 0.19 & 1.71 & 0.57 & 0.08 & -0.07 & 0.22 & 0.07 & 0.24 + hr3982 & 13000 & 241 & -0.08 & 0.13 & 2.22 & 0.47 & 0.04 & -0.09 & 0.28 & 0.03 & 0.15 + hr3982 & 13000 & 241 & -0.07 & 0.09 & 2.14 & 0.51 & 0.06 & -0.09 & 0.32 & 0.12 & 0.26 + hr7001 & 9480 & 678 & -0.05 & 0.36 & 2.83 & 1.06 & 0.03 & -0.03 & 0.41 & 0.15 & 0.52 + hr7001 & 9480 & 146 & -0.10 & 0.38 & 2.81 & 1.02 & 0.10 & -0.05 & 0.43 & 0.17 & 0.41 + hr2491 & 9145 & 111 & -0.07 & 0.36 & 2.71 & 0.83 & 0.16 & -0.14 & 0.37 & -0.08 & 0.10 + hr2491 & 9145 & 148 & -0.14 & 0.40 & 2.67 & 0.93 & 0.12 & -0.13 & 0.39 & 0.05 & 0.20 + hr4534 & 8593 & 205 & -0.04 & 0.45 & 2.67 & 1.01 & 0.11 & -0.12 & 0.30 & 0.07 & 0.34 + hr4357 & 8377 & 192 & -0.05 & 0.40 & 2.83 & 0.96 & 0.09 & -0.12 & 0.24 & 0.06 & 0.38 + hr4931 & 6750 & 119 & 0.04 & 0.24 & 1.47 & 0.62 & -0.01 & -0.06 & 0.09 & 0.09 & 0.86 + hr1279 & 6677 & 080 & 0.10 & 0.32 & 1.62 & 0.67 & 0.15 & -0.09 & 0.26 & 0.12 & 0.64 + hr2943 & 6530 & 327 & 0.04 & 0.33 & 1.60 & 0.57 & 0.15 & -0.14 & 0.54 & 0.27 & 0.74 + hr1538 & 6385 & 281 & 0.06 & 0.34 & 1.26 & 0.53 & 0.18 & -0.11 & 0.51 & 0.22 & 0.90 + hr4375 & 6085 & 254 & 0.26 & 0.15 & 0.63 & 0.56 & 0.15 & -0.06 & 0.68 & 0.37 & 1.09 + hr4983 & 5930 & 123 & 0.21 & 0.16 & 0.75 & 0.66 & 0.09 & 0.04 & 0.71 & 0.45 & 1.20 + hr4983 & 5930 & 173 & 0.19 & 0.21 & 0.90 & 0.52 & 0.19 & -0.06 & 0.87 & 0.47 & 0.96 + hr483 & 5855 & 059 & 0.24 & 0.02 & 0.68 & 0.29 & 0.04 & -0.17 & 0.18 & 0.37 & 1.09 + hr483 & 5855 & 079 & 0.24 & 0.19 & 0.90 & 0.46 & 0.19 & -0.11 & 0.67 & 0.48 & 1.30 + hr4374 & 5830 & 250 & 0.28 & 0.20 & 0.62 & 0.63 & 0.10 & -0.01 & 0.90 & 0.47 & 1.35 + hr5072 & 5740 & 197 & 0.26 & 0.15 & 0.74 & 0.53 & 0.16 & -0.04 & 0.99 & 0.53 & 1.28 + hr4496 & 5430 & 105 & 0.35 & 0.17 & 0.59 & 0.59 & 0.21 & -0.08 & 1.12 & 0.61 & 1.45 + hr4496 & 5430 & 090 & 0.31 & 0.16 & 0.39 & 0.50 & 0.13 & -0.04 & 0.69 & 0.43 & 1.08 + hr7462 & 5240 & 131 & 0.52 & 0.04 & 0.38 & 0.72 & 0.09 & 0.09 & 1.20 & 0.62 & 1.89 + hr7462 & 5240 & 095 & 0.51 & 0.07 & 0.37 & 0.69 & 0.13 & 0.06 & 1.16 & 0.72 & 2.05 + hr1084 & 5010 & 223 & 0.61 & 0.19 & 0.44 & 0.74 & 0.24 & -0.01 & 1.59 & 0.76 & 2.00 + hr8832 & 4785 & 111 & 0.88 & 0.14 & 0.29 & 1.06 & 0.17 & 0.18 & 1.88 & 0.90 & 2.50 + gl570a & 4560 & 120 & 0.90 & 0.22 & 0.37 & 1.13 & 0.28 & 0.04 & 2.02 & 0.90 & 2.29 + hr8085 & 4340 & 125 & 0.65 & 0.18 & 0.01 & 1.15 & 0.11 & 0.15 & 1.59 & 0.60 & 2.38 + hr8086 & 4040 & 170 & 0.72 & 0.39 & 0.06 & 1.46 & 0.14 & 0.11 & 1.78 & 0.50 & 2.07 + hr8086 & 4040 & 181 & 0.72 & 0.32 & 0.07 & 1.48 & 0.15 & 0.12 & 1.86 & 0.52 & 2.18 + gl338a & 3800 & 072 & 0.98 & 0.37 & 0.08 & 1.36 & 0.13 & 0.02 & 1.51 & 0.34 & 1.80 + gl526 & 3605 & 068 & 0.62 & 0.42 & -0.13 & 1.32 & 0.16 & 0.10 & 1.04 & 0.04 & 0.92 + gl411 & 3530 & 202 & 0.48 & 0.53 & -0.04 & 1.29 & 0.16 & 0.01 & 1.17 & 0.10 & 1.19 + gl411 & 3530 & 189 & 0.47 & 0.55 & -0.07 & 1.29 & 0.18 & 0.06 & 1.21 & 0.12 & 1.23 + gl725a & 3380 & 083 & 0.52 & 0.57 & -0.10 & 1.15 & 0.06 & 0.03 & 0.93 & -0.00 & 0.99 + gl725a & 3380 & 107 & 0.54 & 0.56 & -0.12 & 1.25 & 0.02 & 0.08 & 1.20 & -0.02 & 1.09 + | we present a catalogue of h band spectra for 85 stars of approximately solar abundance observed at a resolving power of 3000 with the kpno mayall 4 m fts .
the atlas covers spectral types o7m5 and luminosity classes i - v as defined on the mk system .
we identify both atomic and molecular indices and line ratios which are temperature and luminosity sensitive allowing spectral classification to be carried out in the h band .
the line ratios permit spectral classification in the presence of continuum excess emission , which is commonly found in pre main sequence and evolved stars .
we demonstrate that with spectra of @xmath0 obtained at @xmath1 it is possible to derive spectral types within @xmath2 subclasses for late type stars .
these data are available electronically through the astronomical data center in addition to being served on the world wide web .
# 1#2#3#4#5#6#7 to = -0.2 truein |
early _ hubble space telescope
_ ( hst ) images of circumstellar disks in the orion nebula cluster have provided the first direct evidence that protoplanetary disks in rich young stellar clusters may be photo - evaporated by uv radiation of nearby massive stars @xcite .
the typical signature of photo - evaporation is a bright ionized cusp pointing to the brightest stars of the cluster , @xmath3ori - c in particular .
the cusps always surround a central low - mass star , that may occasionally be encircled by a compact dark disk seen in silhouette against the nebular background , or against the cusp itself .
the most recent atlas of protoplanetary disks ( `` proplyds '' ) in the orion nebula @xcite counts 178 of these comet - shaped sources .
the same atlas also counts other 36 disks which do not show evidence of photoevaporation , being seen as pure dark silhouettes either in absorption against the nebular background ( 28 sources ) or as dark lanes at the midplane of some diffuse bipolar emission ( 8 sources ) .
disk photo - evaporation naturally raises the issue of disk survival in the harsh environment of young massive clusters ( see e.g. * ? ? ?
the frequency and spatial distribution of disks with and without evidence of photo - evaporation within the same cluster provides an excellent opportunity to investigate how environmental factors may affect the path to planet formation .
+ the basic physical process at the origin of disk photo - evaporation in massive clusters has been clarified by @xcite . the fuv ( 6 ev @xmath4 13.6 ev ; 2000@xmath5 912 ) radiation reaches nearly undisturbed the disk surface , creating a photodissociation region ( pdr ) with temperature @xmath6k .
depending on the radial distance from the ionizing stars , the heated disk surface may reach thermal velocity higher than the escape velocity .
the resulting neutral outflow , largely composed by molecular @xmath7 gas dragging a non negligible amount of small dust grains @xcite soon encounters the euv ( @xmath8 13.6 ev ; @xmath9 912 ) radiation coming directly from the ionizing stars , creating a ionization front at @xmath10 k , i.e. the bright ionized cusp .
this model can explain both the general morphology of photoevaporated disks as well as more peculiar features , like the presence of [ oi ] emission from photodissociation of oh radicals produced by @xmath7 heating and dissociation observed in one of the most prominent objects , 182 - 413 @xcite .
it is generally assumed that disk photo - evaporation in rich clusters , as revealed by the presence of photo - ionized cusps , can occur only if the disk is directly exposed to the euv flux from the massive stars , consistently with the fact that the majority of photo - evaporated disks in the trapezium cluster lies in the immediate surroundings of @xmath3ori - c .
dark silhouette disks , lacking any evidence of photo - ionization , are believed to lie within the foreground veil of the orion nebula , shielded by euv radiation @xcite . however , dark silhouette disks can also be photo - evaporated by fuv radiation @xcite although the products of photo - dissociation in these systems have never been seen @xcite .
this because the parameter controlling photo - evaporation is the intensity of the fuv flux , which can be expressed in units of _ habing flux _
( the estimate average flux in the local interstellar medium ) , defined as @xmath11erg @xmath12 s@xmath13 . in the inner core of the trapezium , the fuv flux is elevated , reaching @xmath14 at the orion bar @xcite , about 110or 0.22 pc south of @xmath3ori - c ( assuming a distance @xmath15 pc * ? ? ?
* ) pc to be more accurate .
our estimates do not depend on the exact value of the assumed distance . ] .
sources in the outskirt of the region , shielded from the euv photons emitted e.g. by @xmath3ori - c , may still receive scattered euv and , especially , fuv radiation from the nebular environment .
the most spectacular case of dark silhouette in orion is the well know 114 - 426 disk @xcite . with an apparent diameter of over @xmath16 , which makes it visible also from the ground
, this disk spans about @xmath0 au , 10 times the size of our solar system or of other typical photo - ionized disks in orion .
it is seen nearly edge - on , the presence of the central star being noticeable through scattered polar emission , and is considered to be in a relatively quiescent , non photo - evaporated phase @xcite .
@xcite estimate for the central star an initrinisic brighness @xmath17 , corresponding to a 1.5m@xmath18 star at the distance and age of the orion cluster , but the uncertainites are large . using hst / nicmos high - resolution near - infrared images of 114 - 426 , @xcite find that the major axis of the disk in the pa@xmath19 at 1.87 @xmath20 m line is @xmath2120@xmath22 smaller than at 0.6@xmath20 m .
from this , they deduce a minimum disk mass of 10m@xmath23 , and possibly @xmath24m@xmath25 . on the other hand ,
114 - 426 has not been detected by the submillimeter array survey @xcite at either 880@xmath20 m or 1330@xmath20 m .
this non - detection places an upper mass limit of @xmath26 ( 0.5 - 12 @xmath27 ) , confining the disk mass into a range of two orders of magnitude .
this object has attracted interest because the variation of the disk size with wavelength has allowed to probe the properties of the disk grains , exploiting the wavelength dependence of the extinction of the bright nebular background through the translucent disk edges @xcite .
if the grain sizes were much smaller than the observed wavelengths , as in the rayleigh scattering limit , shorter wavelengths would be scattered much more efficiently than longer ones : in this case the disk would appear smaller at long wavelengths .
viceversa , if the grain sizes are of the same order of the wavelength of observation , the disk size is expected to remain nearly constant with wavelength ( `` gray '' scatter ) . using observations of 114 - 426 in the h@xmath19 ( 0.656 @xmath20 m ; hst / wfpc2 ) ,
pa@xmath19 ( 1.87 @xmath20 m ; hst / nicmos1 ) and br@xmath19 ( 4.05 @xmath20 m ; keck / nirspec ) , @xcite found evidence for chromatic extinction at the northern translucent edge of the disk , deriving a typical grains diameter @xmath28 m , but not @xmath29 4 @xmath20 m .
this represents one of the first evidences for grain growth in protoplanetary disks , the initial step in the process of planet formation @xcite . in this paper
we use the most recent hst / acs images of the 114 - 426 , together with a hst / wfpc2 image taken in the u - band , to clarify the disk structure and its physical status .
contrary to what is commonly assumed , we show that the disk appears strongly affected by the action of the environment and is photoevaporated .
its unusually large size and morphology ( warped and asymmetric with respect to the central star ) may result from photoevaporation .
the translucent northern edge of the disk appears originated by the outflow of neutral material rather than the truncated edge of an undisturbed disk .
we map the variation of extinction in the translucent region , showing how the typical grain size variation is compatible with an evaporative flow .
this allows us to estimate the mass - loss rate from the disk surface and the disk lifetime .
we also show that the disk is enshrouded by a large scale ( @xmath30 pc ) cocoon of dust with a bright ridge facing the side opposite to the cluster center .
this suggests that the disk photoevaporation is driven by the diffuse radiation field rather than by direct irradiation from the massive cluster stars . in section [ sec :
observations ] we briefly present our data ; in section [ sec : morphology ] we illustrate the geometry of the source , moving from the inner structure to the most extended features and the nebular surroundings ; in section [ sec : dust ] we analyze the grain size composition and distribution in the photoevaporated flow ; in section [ sec : discussion ] we derive the main physical parameters of the mass loss and discuss the possible heating sources .
our conclusions are presented in section [ sec : conclusions ] .
hst / acs - wfc and hst / wfpc2 images of 114 - 426 have been collected as part of the hst / treasury program on the orion nebula ( go-10246 , pi m. robberto ) . in this paper
we use the data taken in six broad - band filters : f336w with wfpc2 , and f435w , f555w , f658n , f775w , f850lp with acs ( fig .
[ fig1 ] ) .
the passbands are approximately equivalent to the standard @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 and @xmath36 photometric system , respectively , and for simplicity we will equivalently use both naming conventions . the instrument configuration , observation and data reduction strategy are presented elsewhere @xcite . here
we remind that the acs images have been drizzled to a scale 50 mas / pixels , corresponding to the nominal scale of the instrument , whereas the wfpc2 images have been coadded preserving the original 10 mas / pixel scale of the pc chips .
the fwpc2 image has been resampled and rotated to align it to the acs images .
a column of bad pixels in the wfpc2 image survived image processing and is visible in the f336w image .
however , it s location and extent does not affect our analysis .
in fig . [ fig1 ]
we present our six images ordered in order of increasing wavelength .
the scattered light from the central star , leaking through the polar lobes of the disk , becomes more prominent with wavelength .
the western bright lobe is always brighter than the eastern one , indicating that the disk , seen nearly edge - on , is tilted with the western side facing us .
the morphology also shows clear signs of multiple asymmetries .
first , the lobes are not symmetric with respect to the disk s polar axis , being displaced from the center and closer to the southern edge .
second , the lobes are not on the same vertical axis : a line joining the centers of the two lobes would clearly appear tilted with respect to the disk axis , with an inclination of @xmath37 . , @xmath32 , @xmath33 , @xmath34 , @xmath35 and @xmath36 passbands .
, scaledwidth=50.0% ] to further investigate the structure of this system , we have subtracted from the f850lp image the f658n image , properly scaled , to enhance the image contrast by exploiting the cromatic variation of dust opacity ( fig .
[ fig2 ] ) .
the disk appears warped , with an `` equatorial '' dark lane tracing in silhouette the asymmetry of the bright lobes .
the southern edge of the disk is remarkably sharp and thick , at a projected distance of @xmath38 au from the mid - point between the two lobes , which may be assumed as the position of the central sta .
we will take @xmath39 au as the nominal radius of the disk . in what concerns the northern edge of the disk , by stretching the gray levels of our f555w image
we can resolve it in a spur - shaped structure ( fig .
[ fig3 ] ) protruding beyond the darkest part of the disk . if we regard the spur as an outer appendix to the main disk , thus limiting the disk extension at the northern side , the symmetry of the warped disk with respect to the mid - point between the two lobes ( i.e. the position of the central star ) is nicely recovered .
we believe that the spur most probably represents an outflow of photoevaporated material .
it corresponds to the `` translucent '' part of the disk where @xcite and @xcite have measured the dust grain size . in figure [ fig2.a ]
se shows a cartoon of the system summarizing the main features : tilt , asymmetry and warp . observing on a larger scale the h@xmath19 image of the source , one can detect further signspots of ouflow of neutral material .
[ fig4 ] shows both a color composite image and a contranst enhanced v - b image of the target , stretched to enhance the contrast against the nebular background .
both images reveals a remarkable dark `` tail '' on the eastern side , characterized in the innermost region by twisted filaments that can be traced up to distances of several disk diameters .
the darker filament departs from the southern part of the disk , just to the south of the eastern lobe , and can be directly revealed by a contour plot of the disk intensity levels ( fig .
[ fig8 ] ) .
the same figure also shows a minor protusion on the north - western side of the disk , just to the north of the bright lobe .
notice that these two dark protrusions appear simmetric with respect to the disk center and antisymmetric with respect to the bright lobes .
figure [ fig5 ] shows our source with its immediate surroundings in a contrast enhanced @xmath34-v image .
an extended filamentary shell emitting in @xmath34 line encircles the western side of the disk .
the shell radius has a projected distance of about 2.5 disk diameters , @xmath40 pc .
the filaments can be traced , at much lower brightness levels , to the eastern side of the disk , up to a projected distance of 0.03 pc .
the tenuous bright rim delineates a characteristic foot - like " structure which seems to represent a region of enhanced density externally illuminated by the diffuse radiation field of the nebula .
remarkably , the two closest sources immediately to the east of 114 - 426 , 117 - 421 and 121 - 434 @xcite , are resolved in our acs images in two compact photo - ionized proplyds , pointing in the direction of @xmath3ori - c , lying about 100 to the north - east ( @xmath410.2 pc projected distance ) .
the small size of the two dark disks , unresolved in our acs images and in any case less than @xmath42 au radius ( bounded by the ionized cups ) , is in striking contrast with the huge dimensions of our target , about 20 times larger in size and without any sign of direct influence from @xmath3ori - c .
+ in conclusion , there are various elements of circumstantial evidence all pointing in the directions of 114 - 426 being photoevaporated , and we uphold that this is indeed the case . before envisioning a possible scenario for this unique object , it is appropriate to reconsider , on the basis of our data , the physical properties of the dust grains .
this is the subject of the next section .
as shown by @xcite and @xcite , it is possible to investigate the properties of dust particles of 114 - 426 via the wavelength dependence of the extinction of background light in partially - opaque areas of the disk .
following the same type of analysis , we are now able to exploit the spatial resolution of our acs images to map for the first time the particle size in different positions of the partially - opaque part of the disk .
we use to this purpose the classic mie theory , a fully analytical model which derives , for the special case of spherical particles , an exact series expansion of the scattered field ( see for example * ? ? ?
three parameters are needed for computing the cross section of dust : the radius @xmath43 , the real part of the refractive index @xmath44 , and the imaginary part of the refractive index @xmath45 , the latter two quantities being dependent on the wavelength of the incoming light . both the absorption ( @xmath46 ) and the scattering ( @xmath47 ) cross sections are thus determined , and therefore the extinction cross section given by @xmath48 . generally speaking ,
the dominant role is played by the radius @xmath43 , while the refractive index components have a relatively minor effect .
this fundamental property allows performing a measurement of the grain size from the shape of the cross section curve as a function of wavelength .
the refractive index of the grains can be estimated given the dust grain composition , which is typically considered to be a compound of different chemical species ( e.g. * ? ? ?
* ; * ? ? ?
the bruggeman mixing rule @xcite provides a generic form that expresses the dielectric function of the heterogeneous particle , @xmath49 , as a combination of the dielectric functions of the single species . for a @xmath50-components mixture
it is : @xmath51 where @xmath52 and @xmath53 are the dielectric function and the volume density of the @xmath54 species ( @xmath55 ) , respectively . for the dust particles of circumstellar disk we consider porous composite grains made of astronomical silicates ( @xmath56 ) , carbonaceous materials ( @xmath57 ) , water ice ( @xmath58 ) and vacuum ( @xmath59 )
this is an approximated version of the @xcite model ( see * ? ? ?
the dielectric function is derived from the complex refraction index , @xmath60 .
we have assumed for the silicates the refraction index vs. the wavelength tabulated by @xcite for the silicates , by @xcite for the amorphous carbonacious material and by @xcite for the water ice . for vacuum , @xmath61 .
by solving the implicit equation [ eq : bruggeman ] for @xmath62 we derive for the real part of the refractive index @xmath63 and an imaginary part , @xmath64 .
having derived the effective refraction index ( or dielectric function ) , we derive the absorption and scattering cross sections , and therefore the extinction , from the mie theory .
the shape of the dust grains deserves some discussion .
real dust grains are not spherical but likely present complex internal structures due to the aggregation of smaller interstellar grains .
the most recent studies @xcite indicate that the scattering properties of particles of different shape and sub structure are not dramatically different from those derived using the mie theory for spheres . assuming that the dust particles are perfect spheres does not introduce uncertainties larger than a 30% , the size returned by the classic mie theory being typically underestimated . also the size distribution plays a role in determining the extinction . in this paper
we make the simplifying assumption of treating the dust grains as if they were monodisperse , just one radius , in each pixel . in a separate paper
the issue of the size distribution will be treated more in detail . to compare theoretical predictions with observations , we exploit the extinction efficiency factor , @xmath65 , namely the cross section divided by the geometrical cross section @xmath66 .
this is a parameter commonly used in scattering problems , and can easily be obtained on the basis of the mie expansion series as a function of the reduced size parameter @xmath67 , where @xmath68 .
the extinction curve can then be obtained at any @xmath69 for a given size @xmath43 , and then fitted to the data.the extinction data have been evaluated as the ratio of the average flux ( _ counts / s _ ) of the nebular background , @xmath70 , and the flux ( _ counts / s _ ) in a single pixel of the translucent zone index labels the broad - band filter .
] , @xmath71 .
in particular @xmath70 has been evaluated in an elliptic ring , all around the disk , close enough to be rapresentative of the effective nebular background of 114 - 426 .
more precisely we define @xmath65 , given by the observational data , as @xmath72 the surface density @xmath73 disappears when @xmath65 is normalized to some reference wavelength , for example the v - band ( f555w ) .
the plots in fig .
[ newmap ] show the data for three representative pixels and the corresponding best fit curves , providing radii equal to 0.60 @xmath20 m , 0.50 @xmath20 m and 0.36 @xmath20 m .
the attenuation in the v - band ( f555w ) varies from @xmath74 ( pixel a ) down to @xmath75 ( pixel c ) .
we estimated the error bars accounting for both the poisson statistics ( on @xmath70 and @xmath71 ) and non uniformity of the background ( for @xmath70 ) .
similar plots have been obtained for every pixel in the translucent zone , allowing us to map the distribution of dust grains size , also shown in fig .
[ newmap ] . from the outmost pixels ( blue ) to the innermost pixels ( light yellow ) , the radii increase from 0.2 @xmath20 m to 0.6 @xmath20 m .
the inner parts of the disk can not be studied with this approach since the optical thickness increases and the scattered light from the lobes dominates .
the absolute values of the radius @xmath43 we have derived depend on the assumed grain composition . for a generic grain of refraction index @xmath76 ,
the scattered field depends on the product @xmath77 : an increase of the refraction index @xmath76 compensated by a similar decrease of the grain radius @xmath43 provides the same result .
therefore , the uncertainties on the assumed refraction index affect only the absolute value of the grain size , not their ratios .
the evidence for a spatial variation of the grain sizes at the disk edge appears therefore strong . our analysis , based on eq .
( 2 ) , does not account for a scattering term .
heuristically , we not expect scattering to be dominant .
one the one hand , it is reasonable to assume that our dust particles are non - spherical , randomly oriented , heterogeneous and with variable refraction index . in this case , the oscillations of the individual cross sections predicted by the mie theory are smeared out and no large - scale chromatic patterns should be expected .
also , scattering by typical disk grains is highly non - isotropic , as mie theory predicts 90% of the scattered light is emitted in the forward hemisphere .
since our source is mostly back - illuminated , wave propagation toward us is dominated by the coherent superposition of the background light with the forward scattered wave , i.e. extinction .
third , even assuming isotropic scattering , it may be shown that the color effect induced by the scattered radiation produces an apparent modulation of the reddening law ( w. henney , private communication ) that seems unable to explain the observed range of attenuation , and therefore of grain size , shown by our data .
some evidence for the spatial variation of the size of dust grains has been obtained by recent interferometric observations of young disks in the sub - millimeter @xcite .
the general result found so far is that larger grains are found in the disk regions which are closer to the central star , and this is the same trend found for the 114 - 426 disk . in those cases , however , the dust probed by those observations is located inside the disk , and not in a photoevaporative wind as for the translucent north side of 114 - 426 .
the result presented in this paper is in general agreement with the expectations for the dynamics of dust grains entrained in a gaseous photoevaporative wind .
the drag force experienced by a dust grain is proportional to @xmath78 ( takeuchi et al .
2005 ) , and therefore the acceleration scales with @xmath79 .
this means that the wind selectively entrains grains with different sizes , with smaller grains being more easily dragged along the wind .
as shown by @xcite , the drag force on small grains can dominate over the stellar gravity allowing them to be carried to very large distances from the central star ( several hundreds of au ) .
this can explain , at least in a qualitative way , the radial gradient of the grain sizes obtained by our analysis over a spatial scale of @xmath80 au ( fig .
[ fig : mappa ] ) . however , the owen et al .
models treat the case of a photoevaporative wind driven by the ionizing radiation from the central star , whereas in the case of 114 - 426 the observed asymmetries indicate that the origin of the photoevaporative wind is external to the system , i.e. due to the diffuse fuv radiation field of the nebula ( section 3 ) .
this has an obvious effect on the geometry predicted for the photoevaporation of the disk . a model which calculates the photoevaporative wind induced by a diffuse non - ionizing radiation field
is needed to obtain a more quantitative comparison with the data . to build a possible scenario for 114 - 426 , we begin by observing that the disk surface appears shielded by the euv flux from @xmath3ori - c , unlike the nearby proplyds 117 - 421 and 121 - 434 ( fig .
[ fig5 ] ) .
this can be explained in two ways . on the one hand
, 114 - 426 may lie much further away from the trapezium than the other two proplyds , possibly within the foreground veil of the orion nebula , shielded by the euv photons from @xmath3ori - c by some extended ionization front , possibly at the inward - facing side of the veil . in this case
, it seems difficult to explain the existence of the bright filamentary structure encircling our source on the west side ( fig .
[ fig5 ] ) , which is basically the direction opposite to @xmath3ori - c .
this filament , mostly visible in h@xmath19 , most likely represents a ionization front produced by diffuse euv radiation coming from the outer regions of the nebula , absorbed by a density enhancement apparently centered on our target . at the same time
, there is no evidence of a comparably strong ionization front on the east side , facing @xmath3ori - c . in order to be viable , the `` within the veil '' scenario seems to require a rather peculiar combination of geometry , veil density and radiation field .
the second possibility is that the disk lies in the outer parts of the nebula in a location shielded by the euv radiation directly coming from @xmath3ori - c .
it is intriguing , in this respect , to notice that between 114 - 426 and the trapezium lies , projected on the plane of the sky , the high - density region hosting orion - s , a secondary center of star formation in the orion nebula ( fig .
[ fig : regionbig ] ) . according to most recent 3-d model of the orion nebula @xcite ,
the orion - s feature southwest of the trapezium is an optically thick molecular cloud within the body of ionized gas .
the orion - s cloud that may therefore prevent euv photons from the trapezium stars to reach 114 - 426 .
the two proplyds 121 - 434 and 117 - 421 would therefore lie in the foreground , far enough from the orion - s cloud to see directly the trapezium stars .
alternatively , they could be located deeper inside the nebula , exposed to the euv flux passing under the orion - s cloud .
we propose that the influence of the fuv photons mostly coming from the western regions of the orion nebula causes 114 - 426 to photoevaporate .
the contribution from other stellar sources to the east of 114 - 426 is neglible , in particular we estimate that the total flux from closest a0 star ks ori ( par 1685 ) at @xmath81 arcmin distance is about 1% of the diffuse fuv field ( see the next section ) .
the neutral outflow from the disk surface intercepts the weak diffuse field of euv photons at relatively large distances from the disk ; still , the ionized filamentary structure may well represent the analog of the bright ionized cusps that identify common proplyds .
the pressure gradient pushes the neutral material further to the east , originating``foot - like '' structure traced by a weak ionization front at its boundaries .
also the photoionized proplyds in the vicinity of @xmath82ori - c show dusty tails downstream of the ionizing radiation . in summary ,
114 - 426 can be regarded as an extreme case of proplyd photoevaporated by the diffuse radiation field .
the mass loss rate , @xmath83 , for large disks that evaporate in an external fuv field is given by @xcite : @xmath84 where @xmath85 is the fraction of solid angle subtended by the outflow , @xmath20 is the mean molecular mass of the gas , @xmath86 is the absorption cross section for dust grains to fuv radiation , @xmath87 is the sound speed of the flow , and @xmath88 is the disk radius . with @xcite , we assume @xmath89 . assuming a typical dust grain radius @xmath90 m , as derived in section [ sec : dust ] , and a gas - to - dust mass ratio of @xmath91 we derive @xmath92 @xmath93 , in line with the typical value for pdr regions @xcite . for the outer radius we use @xmath94 au , concentrating our analysis on the northern translucent region .
the sound speed , i.e. the temperature at the disk surface , is controlled by the amount of fuv illuminating the disk which can be estimated as follows .
both ioinizing ( euv ) and non - ionizing ( fuv ) fluxes emitted by the trapezium stars are scattered within the hii region .
the euv radiation , carrying about 2/3 of the original ionizing energy ( assuming menzell s case b recombination ) , is scattered by recombination processes , whereas fuv radiation is scattered only by dust particles ( see e.g. * ? ? ?
most of the scattering occurs at the edges of the hii region so that the diffuse field can be assumed to be constant within the cavity .
a disk shielded from the euv radiation and from direct illumination from the cluster center will only receive a flux of scattered fuv photons given by @xmath95 assuming that 1/2 of the radiation is redirected inward by the walls of the cavity , a scattering efficiency @xmath96 appropriate for large grains , a flux @xmath97 of non - ionizing photons emitted by the central o stars ( mostly @xmath3ori - c ) , and a typical radius of the cavity @xmath98 pc , the distance between @xmath3ori - c and the ionization front @xcite .
we assumeori - c is historically flagged with a spectral type o6.5pe .
however , recent observations indicate that this peculiar system is a binary with a o5.5 primary ( m=34.0@xmath99 ) and an o9.5 secondary ( m=15.5 m@xmath100 * ? ? ? * ) .
we therefore adopt the lyman continuum luminosity of a zams o5.5 star , according to @xcite .
our results to not depend on this assumption . ]
that @xmath3ori - c emits @xmath101 of its luminosity as fuv photons , @xmath102photons s@xmath13 and an effective wavelength @xmath103 . a shielded disk
will then receive a diffuse fuv flux @xmath104 . solving the equilibrium equation for the radiative heating and cooling of dust grains ( e.g. * ? ? ?
* ch24 ) , assuming the above fuv flux and the refraction index of our heterogeneous grain , we derive a dust temperature of 25 k , appropriate for a photo dissociation region .
for the temperature of the gas components , we can refer to fig.2(b ) of @xcite , which reports the temperature profiles calculated from the pdr code for an external fuv radiation field of @xmath105 as a function of visual extinction @xmath106 , for different values of the number density . at @xmath107
the gas temperature is in the range @xmath108k , corresponding to a sound speed @xmath109 km s@xmath13 from the disk .
using equation ( 3 ) we then derive for the mass loss rate of 114 - 426 @xmath110 where the spread is mainly dependent on the uncertainty on the sound speed .
the large disk size , combined with the modest upper limit for the disk mass , make 114 - 426 a rather unique object .
analyzing the thermal continuum ( 880 @xmath20 m ) emission from circumstellar disks in the @xmath111 myr old ophiucus star - forming region , @xcite have found a strong correlation between the disk mass and radius , larger disks being more massive .
a comparison of our values for the disk radius and the mass upper limit with the valuer reported by @xcite shows that our source does not follow the radius - mass relation traced by the other sources , being way too large for its mass .
if the relation radius - mass is an imprint of the initial conditions inherited by the disk at the formation epoch , as suggested by @xcite , then 114 - 426 is in a diverse evolutionary stage in which this dependence seems to be lost . assuming our range of estimated mass loss rates and the current upper limit for the disk mass , @xmath112m@xmath113 , the disk s life time ranges between @xmath114 yr to @xmath115 yr .
this is clearly a very short time scale , but it may be appropriate for a rather unique object like 114 - 426 . at the current evaporation rate ,
114 - 426 will be entirely dissipated in a few @xmath116 yr .
it may well have formed 1 myr ago ( about the age of the onc ) with an average disk mass of @xmath117 m@xmath18 and be reduced to the current mass maintaining level of photoevaporation nearly constant at the measured level .
overall , the picture seems consistent with 114 - 426 being a relatively low mass disk inflated and photoevaporated by the diffuse field of fuv radiation within the nebula , seen in the very last phases of its evolution . in this last section
we briefly discuss the disk morphology .
first , we note that the keplerian orbital period at the outer disk edge @xmath118 corresponds , for dust grains moving at 1km / s , to a traveled distance of about 1700 au , or 4 disk radii .
this is nicely compatible with the spatial scale of variation of the spiraling dust filaments on the eastern side of the disk , which may therefore be related to periodic enhancement of the disk mass loss rate syncronous with the passage , or appearance , of disk asymmetries . concerning the disk asymmetry ,
while the diffuse fuv radiation represents the main mechanism of destruction of 114 - 426 , it can not easily account for the extremely warped disk geometry unless one assumes it is highly non - isotropic .
a directional radiation flux impacting the surface of a flared disk can generate a non - uniform increase of the surface temperature and modulate the flaring angle @xcite , or just push the disk mid - plane out of its equilibrium position , depending on the disk angular velocity and surface density ( s. lubow , private communication ) .
since the disk is rotating , the development of a persistent warp will depend on factors like the cooling time scale , the turbulence regime , or any possible temporal variations of the radiation intensity , all basically unknown at the moment .
a second possibility is that the disk hosts a binary star , in which case the warping would result from a a misalignement between the disk plane and the plane of the binary orbit .
very little is known about the nature of the central star ( see * ? ? ? * for a discussion ) .
an intriguing aspect is that the colors of the two lobes are quite different : the south - eastern lobe is redder , to the point of becoming prominent in the near - ir , as revealed by the hst / nicmos images of @xcite .
it is not possible to explain such an inversion by assuming a standard flared disk model and common optical properties ( scattering / absorption efficiencies ) for the dust grains .
however , a binary system with two stars having comparable luminosity and different effective temperatures could represent a more viable solution .
direct spectroscopy of the two lobes can address this possibility .
stellar encounters may also perturb circumstellar disks leading to the fomation of warps , spiral waves , tidal tails and disk truncation @xcite .
those disturbances , observed e.g. in rw aur by @xcite , are expected to occur in the onc .
several studies have pointed out that in the dense inner regions of young clusters like the onc stellar enconters may be frequent enough to produce major effects on circumstellar disks ( see e.g. * ? ? ?
* and reference therein ) .
warps induced in the outer disk regions by the fly - by of another star propagate inward to a radius @xmath119 with a wavelength given by @xcite @xmath120 where @xmath121 is the local disk sound speed and @xmath33 and @xmath122 are the relative velocity and closest approach distance of the encounter , respectively .
the warp propagate inward to a radius @xmath123 at which @xmath124 , where the wavelength of the warp becomes larger than the disk size and gets reflected .
angular momentum , however , is still communicated to the inner disk so the disk becomes tilted , but not warped . the decay time - scale for the warp
is given approximately by @xmath125 where @xmath19 is the dimensionless viscosity parameter , typically in the range @xmath126@xcite .
for our disk with @xmath39au , assuming @xmath127 m@xmath113 , the warp may be expected to last for a time of around @xmath128yr . therefore , for the low values of the viscosity parameter @xmath19 typical of the outer disk regions , a warp imprinted to the disk by a past close encounter may still be propagating at these later times . an interplay between the warped outer disk , the flat and tilted inner disk , the energy input from the central star(s ) and the nebular environment supporting
a conspicuous photoevaporative flow may account for the complex phenomenology observed in this unique system . in the near future
, alma observations should greatly help clarifying the nature of 114 - 426 by mapping with high spatial resolution the spatial distribution of the disk and its kinematic status .
we have used a set of hst images taken with wfpc2 and acs to describe the complex morphology of the most prominent dark silhouette disk in the orion nebula , 114 - 426 .
the disk appears tilted , asymmetric and warped , with clear signs of photoevaporation of neutral material . mapping the distribution of dust grains at the northern translucent edge of the disk using the mie theory for standard circumstellar disk grains , we have found evidence for a spatial gradient of the grain radius , from 0.2 to 0.7 @xmath20 m , with grains becoming smaller as the distance from the disk increases .
these values are slightly below the minimal 1.9 @xmath20 m grain size estimated by shuping et al .
( 2003 ) , but still compatible considering the difficulty of observing this source at infrared wavelengths .
overall , the disk morphology and location in the nebula indicate that this system is photoevaporated by the diffuse radiation field of the orion nebula , while being shielded from the radiation coming directly from the central trapezium stars . using our derived grain properties ,
we have estimated the mass loss rate from the disk surface and the time - scale for total disk dissipation , which turns out to be of the order of @xmath2yr .
this short time scale , of the order of 1/100 of the cluster age , seems compatible with the exceptional nature of this system , namely its huge size and low mass .
finally , we briefly discuss the viability of possible mechanisms that may lead to the peculiar morphology of this system : external uv flux , binary star and past close encounter . 0.5 cm
_ acknowlegments _ : the authors would like to thank s. lubow , i. pascucci , j. pringle and k. stapelfeld for helpful discussions on this fascinating source , and the referee w. henney for prompt and useful feedback .
zolt levay provided support with image production .
support for go program 10246 was provided by nasa through a grant from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
kraus , s. , balega , y. y. , berger , j .-
p . , hofmann , k .- h .
, millan - gabet , r. , monnier , j. d. , ohnaka , k. , pedretti , e. , preibisch , th . ,
schertl , d. , schloerb , f. p. , traub , w. a. , & weigelt , g. 2007 , , 466 , 649 | deep hst broad - band images taken with acs and wfpc2 of the giant ( @xmath0 au diameter ) dark silhouette proplyd 114 - 426 in the orion nebula show that this system is tilted , asymmetric , warped and photoevaporated .
the exquisite angular resolution of acs allows us to map the distribution of dust grains at the northern translucent edge of the disk , dominated by the photoevaporative flow . using the mie theory for standard circumstellar disk grains , we find evidence for a spatial gradient in grain size . the typical dust radius , @xmath1 m ( less than what reported by previous studies ) becomes smaller as the distance from the disk center increases , consistent with the expectations for the dynamic of dust entrained in a gaseous photoevaporative wind .
our analysis of the disk morphology and location within the nebula indicates that this system is photoevaporated by the diffuse radiation field of the orion nebula , while being shielded from the radiation coming directly from the central trapezium stars .
we estimate the mass - loss rate from the disk surface and the time - scale for total disk dissipation , which turns out to be of the order of @xmath2yr . such a short time , of the order of 1/100 of the cluster age ,
indicates that this system is seen on the verge of destruction .
this is compatible with the exceptional nature of the disk , namely its combination of huge size and low mass .
finally , we briefly discuss the viability of possible mechanisms that may lead to the peculiar morphology of this system : external uv flux , binary star and past close encounter . |
graphene exhibits electronic properties that conventional metals do not show , e.g. , charge carriers with zero effective mass , anomalous quantum hall effect , and a minimum conductivity when carrier concentrations tend to zero @xcite .
the unconventional electronic properties of graphene occur due to its linear energy dispersion in the vicinity of the dirac point and two - dimensional structure @xcite .
recent theoretical studies show , however , that a linear energy dispersion relation is not unique to graphene @xcite .
for instance , a flat two - dimensional silicon ( si ) sheet with a honeycomb lattice ( silicene ) exhibits a linear energy dispersion too @xcite .
even buckling of the si honeycomb lattice ( si tends to form @xmath1 bonds rather than @xmath2 bonds ) does not remove the linearity in the energy dispersion relation nor does it open an energy gap @xcite .
more generally , it has been predicted that a group - iv nanosheet with a flat or buckled honeycomb lattice also shows a linear dispersion too @xcite .
it is the underlying symmetries of the honeycomb lattice that are responsible for the linear energy dispersion @xcite .
one - dimensional carbon nanostructures also exhibit unconventional electronic properties , e.g. , carbon nanotubes show different electronic behavior depending on their chirality . similarly to graphene , the unconventional electronic properties are not unique to carbon nanotubes @xcite .
si nanotubes also show electronic properties that are chirality dependent @xcite . the theoretical studies mentioned above motivated the experimental realization of two - dimensional sheets other than graphene .
silicon hexagonal sheets were chemically exfoliated from calcium disilicide ( casi@xmath3 ) @xcite and si nanoribbons have been recently synthesized @xcite .
there is a report of the observation of a linear electronic dispersion for the si nanoribbons using angle - resolved photoemission spectroscopy @xcite .
similarly to graphene , there is interest in modifying a silicene sheet in order to tailor its properties .
for example , studies have shown that si nanoribbons can display semiconducting and magnetic behavior @xcite .
hydrogenation of silicene can also open a gap ( e.g. , hydrogenated graphene ( graphane ) is an insulator of direct band gap @xcite ) and the resulting compound should be related to polysilane @xcite .
there are already a few ab initio calculations of related structures . indeed ,
density functional theory ( dft ) calculations within the local density approximation ( lda ) found that a si sheet without hydrogen is a semiconductor of gap zero @xcite . upon hydrogenation ,
the si sheet becomes an semiconductor of indirect band gap @xcite .
as for hydrogenated si nanotubes , a previous density functional tight - binding theory found that hydrogenated si nanotubes are semiconductors with a gap that shows little dependence on chirality @xcite .
it is the purpose of the present work to study the band structure of hydrogenated si nanosheets and nanotubes in more detail than had been done previously with the ab initio method . in particular , ( @xmath4 )
previous work had focused on structures , ( @xmath5 ) the calculations of the hydrogenated si sheet were all done using lda , which suffers from the band - gap problem , ( @xmath6 ) a discussion of the wave functions was missing from the earlier work , and ( @xmath7 ) a complete study of the band structures of si - h nanotubes has not yet been carried out . the most appropriate model is the empirical tight binding one . in this paper , we will first consider a hydrogenated si sheet ( also referred as silicane @xcite ) .
for the hydrogenated si sheet , we find that ( @xmath4 ) it is an semiconductor of indirect band gap of about @xmath0ev ; ( @xmath5 ) the lowest point of the conduction bands ( cb ) occurs at the m point and the highest point of the valence bands ( vb ) occurs at the @xmath8 point ; ( @xmath6 ) the indirect gap is closely related to the band structure of silicene and it is mainly due to band - filling effect ; ( @xmath7 ) the degeneracy at the dirac point is not lifted by hydrogenating the si sheet .
we then consider the electronic structure of hydrogenated single - walled si nanotubes ( si - h nts ) .
we find that ( @xmath4 ) si - h nts of certain chiralities are semiconductors with co - existing direct and indirect gaps of identical magnitudes ; ( @xmath5 ) the magnitude of the gaps is equal to that of the hydrogenated si sheet .
we now describe our tight - binding model .
we consider a @xmath9 hamiltonian with the following bloch states for si ( @xmath10 site ) , @xmath11 and similarly for the @xmath12 sites and for h ( @xmath13 and @xmath14 sites ) . here
@xmath15 are lwdin orbitals ; @xmath16 is a lattice vector of the hexagonal lattice ; @xmath17 is the position vector of the si and h atoms in the unit cell ; and @xmath18 is the number of unit cells . for si atoms ,
we consider first nearest neighbors located at @xmath19 and second nearest neighbors located at @xmath20 here , @xmath21 is the buckling height ; @xmath22 is the si - h bond length ; and @xmath23 is the lattice constant .
these structural parameters are taken from ab initio calculations @xcite .
the origin is placed at site a ( figure [ fig : sideview ] ) . for the hydrogen atoms ,
we only consider up to first - nearest - neighbor interactions .
we further assume that there is only a @xmath24 bond between the si and h atoms .
silicane has a hexagonal lattice with a basis composed of two si atoms and two hydrogen atoms ( figure [ fig : sideview ] ) . and @xmath12 .
hydrogen atoms are located at sites @xmath13 and @xmath14 .
the rectangle encloses the unit cell.,width=377,height=151 ] our tight - binding model is constructed as follows : ( @xmath4 ) the si - si tb parameters are taken from those of bulk si ( @xmath1 configuration ) ; ( @xmath5 ) the si - h tb parameters are obtained by using the universal harrison scaling rule @xcite ( @xmath6 ) the two - center approximation is used to account for the slight distortion of the hydrogenated sheet from pure @xmath1 hybridization .
a simple and fairly good parametrization of the bulk si structure is that of grosso _ et .
these parameters reproduce fairly well the band structure of bulk si , in particular in the k@xmath25 direction , which is important for the si nanosheets @xcite .
the full list of tb parameters is given in table [ t : tca - parameters ] .
si - si & + @xmath26 & @xmath27 + @xmath28 & @xmath29 + @xmath30 & @xmath31 + @xmath32 & @xmath33 + @xmath34 & @xmath35 + @xmath36 & @xmath37 + @xmath38 & @xmath39 + @xmath40 & @xmath39 + @xmath41 & @xmath42 + @xmath43 & @xmath44 + si - h & + @xmath45 & @xmath46 + @xmath47 & @xmath48 +
we now present the results obtained from our tight - binding model .
we first discuss the hydrogenated si sheet .
the hydrogenated si sheet is an semiconductor with indirect gap @xmath49ev ( figure [ fig : si_buckled_sheets ] ) .
our result is in between the @xmath50ev and @xmath51ev reported previously @xcite .
the lowest point of the cb is at m and it is one - fold degenerate ( figure [ fig : si_buckled_sheets ] and table [ t : buckled_sheet_w_h ] ) . the wave function at the m point of the lowest cb is given as follows ( table [ t : buckled_sheet_w_h ] ) : @xmath52 ( @xmath53% ) , @xmath54 ( @xmath55% ) , @xmath56 ( @xmath57% ) , and @xmath58 ( @xmath59% ) .
the highest point in the vb is at @xmath8 and it is two - fold degenerate ( figure [ fig : si_buckled_sheets ] and table [ t : buckled_sheet_w_h ] ) . the wave functions at the @xmath8 point of the highest vb is pure @xmath54 or pure @xmath60 ( table [ t : buckled_sheet_w_h ] ) .
this is the same as was found for graphane @xcite .
the fermi level is set to zero ( dashed line).,width=377,height=226 ] ccl @xmath61[ev ] & degeneracy & wave functions + @xmath62 & @xmath63 & @xmath64 + @xmath65 & @xmath63 & @xmath66 + @xmath67 & @xmath68 & @xmath69 + @xmath70 & @xmath63 & @xmath71 + @xmath72 & @xmath68 & @xmath73 + @xmath74 & @xmath63 & @xmath75 + @xmath76 & @xmath63 & @xmath77 + @xmath78 & @xmath63 & @xmath79 + @xmath80[ev ] & & + @xmath81 & @xmath63 & @xmath82 + @xmath83 & @xmath63 & @xmath84 + @xmath85 & @xmath63 & @xmath86 + @xmath87 & @xmath63 & @xmath88 + @xmath89 & @xmath63 & @xmath90 + @xmath91 & @xmath63 & @xmath92 + @xmath93 & @xmath63 & @xmath86 + @xmath94 & @xmath63 & @xmath95 + @xmath96 & @xmath63 & @xmath97 + @xmath98 & @xmath63 & @xmath99 + @xmath100[ev ] & & + @xmath101 & @xmath68 & @xmath82 + & & @xmath82 + @xmath102 & @xmath63 & @xmath103 + @xmath104 & @xmath68 & @xmath105 + & & @xmath106 + @xmath107 & @xmath68 & @xmath108 + & & @xmath109 + @xmath76 & @xmath63 & @xmath103 + @xmath110 & @xmath68 & @xmath111 + & & @xmath112 + the symmetries of the wave functions at @xmath8 , m , and k for the hydrogenated si sheet are given in table [ t : buckled_sheet_w_h ] .
note that the k point ( @xmath113 ) is two - fold degenerate ( table [ t : buckled_sheet_w_h ] ) .
this is expected since fully hydrogenating the si sheet does not break the mirror symmetry of the honeycomb lattice .
another interesting result is that there is little hydrogen @xmath52 orbital ( @xmath58 ) in the vb .
thus , the gap `` opening '' is , in reality , due to the hydrogen electrons filling up the originally empty lowest cb of silicene ; i.e. , the gap is mostly intrinsic to silicene .
next , we compare the band structures of graphane and silicane
. the band structures of graphane and silicane differ significantly .
first , graphane is an insulator of direct band gap @xcite .
the gap occurs at @xmath8 .
second , the band gap of graphane ( @xmath114ev ) is far larger than that of silicane @xcite .
this is expected since the c - c bond length in graphane is greater than the si - si bond length of silicane @xcite .
, @xmath59 ) , ( b ) armchair ( @xmath115,@xmath115 ) , ( c ) zig - zag ( @xmath59,@xmath72 ) , ( d ) zig - zag ( @xmath115,@xmath72 ) hydrogenated si nanotubes
. armchair ( @xmath116,@xmath116 ) nanotubes with @xmath116 even have simultaneously direct ( @xmath8 ) and indirect ( @xmath117x ) gap equal to that of the hydrogenated si sheet .
zig - zag ( @xmath116,@xmath72 ) nanotubes with @xmath116 even have simultaneously direct ( @xmath8 ) and indirect ( @xmath117x ) gap equal to that of the hydrogenated si sheet . for armchair nanotubes x@xmath118 and for zig - zag nanotubes x@xmath119 .
the fermi level is set to zero ( dashed line ) .
the dotted line in ( a ) and ( c ) is a guide for the eye to show that the lowest conduction states at @xmath8 and x are at the same energy .
the dotted line in ( b ) and ( d ) is a guide for the eye to show that the lowest conduction states at @xmath8 and x are not at the same energy.,width=377,height=264 ] , @xmath59 ) , ( b ) armchair ( @xmath115,@xmath115 ) , ( c ) zig - zag ( @xmath59,@xmath72 ) , ( d ) zig - zag ( @xmath115,@xmath72 ) hydrogenated si nanotubes .
armchair ( @xmath116,@xmath116 ) nanotubes with @xmath116 even have simultaneous direct and indirect gaps since its brillouin zone crosses m and m. zig - zag ( @xmath116,@xmath72 ) nanotubes with @xmath116 even have simultaneous and identical direct and indirect gaps since its brillouin zone crosses m , m and m.,width=377,height=453 ] we now consider the band structure of si - h nts .
si - h nts are fully characterized by the chiral vector @xmath120 , where @xmath121 are integer numbers @xcite .
the band structure of si - h nts is obtained from that of silicane by quantizing the wave vector @xmath122 along the chiral direction @xcite .
armchair ( @xmath116,@xmath116 ) and zig - zag ( @xmath116,@xmath72 ) si - h nts with @xmath116 even have concurrent and identical direct and indirect gaps equal to that of the hydrogenated si sheet ( @xmath123ev ) ( figure [ fig : si - h nts](a ) and ( c ) ) .
we note that this dual behavior is different from the so - called mixed character previously reported @xcite as the latter was for a nanosheet and was an approximate phenomenon .
armchair ( @xmath116,@xmath116 ) and zig - zag ( @xmath116,@xmath72 ) si - h nts with @xmath116 odd have direct gap equal to that of the hydrogenated si sheet ( figure [ fig : si - h nts](b ) and ( d ) ) .
therefore , si - h nts are semiconductors with either direct gaps , or with co - existing direct and indirect gaps of identical magnitude .
the electronic behavior of si - h nts is explained as follows .
we first consider armchair si nts
. for armchair si - h nts @xmath124 , quantization of the wave vector @xmath122 along the chiral direction gives their brillouin zone ( bz ) @xcite , @xmath125 where , @xmath126 and @xmath127 .
armchair ( @xmath116,@xmath116 ) si - h - nts with @xmath116 even have simultaneous and identical direct and indirect gaps since their bz , equation ( [ eq : armchair_bz ] ) , always crosses m and m ( figure [ fig : si - h - nt - bz ] ( a ) ) , @xmath128 armchair ( @xmath116,@xmath116 ) si - h nts with @xmath116 odd have only a direct gap since their bz always crosses m but not m ( figure [ fig : si - h - nt - bz ] ( b ) ) .
we now consider zig - zag si - h nts . for zig - zag
si - h nts @xmath129 , quantization of the wave vector @xmath122 along the chiral direction gives their bz @xcite , @xmath130 where , @xmath131 and @xmath132 .
zig - zag ( @xmath116,@xmath72 ) si - h nts with @xmath116 even have simultaneous and identical direct and indirect gaps since their bz , equation ( [ eq : zig - zag_bz ] ) , always crosses m , m and m ( figure [ fig : si - h - nt - bz ] ( c ) ) , @xmath133 zig - zag ( @xmath116,@xmath72 ) si - h nts with @xmath116 odd have only direct gap since their bz always crosses m but not m , m ( figure [ fig : si - h - nt - bz ] ( d ) ) .
the above results on the gap dependence on chirality are much more complex than the hamada rule @xcite obeyed by non - hydrogenated carbon and silicon nanotubes @xcite .
we now compare our si - h nts results to those of ref .
@xcite . ref .
@xcite presents density functional tight - binding ( dftb ) calculations of the si - h nt gap .
the dftb calculation finds that si - h nts are semiconductors whose gap has little dependence on chirality .
the gap of si - h nts rapidly converges to the gap of their hydrogenated si sheet ( @xmath134ev ) .
no reference is made as to whether the gap is direct or indirect . even though the electronic behavior of our tb model is in fairly good agreement with that of dftb
, one important difference is that our tb calculation did not find a dependence of the size of the gap at all .
we attribute this difference to curvature effects , which are neglected in our model .
in summary , we performed a tight - binding calculation of the band structure of fully hydrogenated si nanostructures . for the hydrogenated si sheet ( silicane )
, we find that it is a semiconductor with indirect band gap of about @xmath123ev .
the indirect band gap occurs between the m and @xmath8 points .
our study of the wave functions reveals that the gap is closely related to the band structure of silicene and is primarily due to a band - filling effect rather than to a gap opening . for si - h nts
, we find that they are semiconductors with either a direct gap or with co - existing direct and indirect band gaps depending on chirality . therefore , si - h nts do not follow hamada s rule .
this shows that hydrogenated si nanotubes can have properties distinct from carbon nanotubes . | the band structure of fully hydrogenated si nanosheets and nanotubes are elucidated by the use of an empirical tight - binding model .
the hydrogenated si sheet is a semiconductor with indirect band gap of about @xmath0ev .
the symmetries of the wave functions allow us to explain the origin of the gap .
we predict that , for certain chiralities , hydrogenated si nanotubes represent a new type of semiconductor , one with co - existing direct and indirect gaps of exactly the same magnitude .
this behavior is different from the hamada rule established for non - hydrogenated carbon and silicon nanotubes .
comparison to an ab initio calculation is made . |
ghost imaging is a transverse active - imaging technique that exploits the correlation between two light beams to image a target without spatially - resolving measurements of the light beam that has undergone target interaction .
the two beams are the signal , which interacts with the target and then is measured by a single - pixel bucket detector , and the reference , which is directly measured by a high spatial - resolution detector . an intensity cross - correlation between the signal beam that encounters the target and
the reference beam that impinges on the high spatial - resolution detector imparts target information the ghost image to the cross correlation between the photocurrents obtained from the two detectors , information that is unavailable from either photocurrent alone .
most ghost - imaging experiments and related theory deal with the transmissive case , in which the bucket detector is placed behind the target , and we image the target s intensity - transmission profile . in the first ghost imaging experiment , the signal and reference beams were the quantum - mechanically entangled outputs from a spontaneous parametric downconverter ( spdc ) @xcite , which afforded them a phase - sensitive cross correlation stronger than permitted by classical physics @xcite .
the system was run at low flux , in which the spdc s post - selected output state , within the photodetectors response time , could be taken to be a biphoton ; the ghost image was then formed by counting coincidences between the bucket and high spatial - resolution detections instead of a photocurrent cross correlation .
subsequently , ghost imaging was done with a high - flux pseudothermal source , i.e. , a classical - state source realized by sending a laser beam through a rotating ground - glass diffuser and a 50 - 50 beam splitter to create identical , spatially - incoherent signal and reference beams with a phase - insensitive cross correlation .
the ghost image was then obtained by continuous - time cross correlation of the bucket and high spatial - resolution photocurrents @xcite .
later , the theories of spdc and pseudothermal ghost imaging were united in a gaussian - state treatment that showed , in both cases , that ghost - image formation by photocurrent correlation arises from classical coherence propagation , but the nonclassical spdc source offers much higher image contrast and a modest spatial - resolution advantage in near - field operation @xcite .
recently , a computational ghost - imaging method was introduced , in which the need for a reference beam is removed @xcite .
conventional ghost imaging relies on the intensity cross - correlation between the signal beam on the target and the reference beam on the high spatial - resolution detector . in spdc and pseudothermal ghost imaging ,
these beam patterns are random .
if , however , a spatial light modulator is used to impart a sequence of known spatial patterns to a laser beam prior to propagation to the target , then diffraction theory can be used to calculate noiseless versions of the associated reference - beam intensity patterns .
these computed reference - beam results can be employed , for cross - correlation ghost - image formation , as if they came from photodetection of a physical reference beam .
more importantly , computational ghost imaging opens the door for more advanced computational image reconstruction techniques , notably compressive sensing @xcite .
all of the preceding discussion has been framed for ghost imaging of transmissive targets .
recent experiments have shown the feasibility of an alternate ghost - imaging configuration , in which the bucket detector views the target in reflection , rather than in transmission @xcite . in this reflective setup , the source and detector can , in principle , be moved arbitrarily far away from the target , suggesting that ghost imaging could be a viable remote - sensing technique .
atmospheric turbulence will surely be a relevant concern in any such standoff - sensing application , just as it is for astronomical imaging and laser radar . to date
, there has been theoretical study of the impact of turbulence on ghost imaging in transmission @xcite , but there has only been an initial turbulence - free theory development for ghost imaging in reflection @xcite , and that work was limited to pseudothermal sources . in this paper we will extend the analysis from @xcite to include spdc ghost imaging in reflection and , for both spdc and pseudothermal operation , the presence of atmospheric turbulence in the propagation paths .
the setup we shall consider is the lensless ghost - imaging configuration shown schematically in fig .
[ fig : setup ] @xcite .
pseudothermal ghost imaging can be described quantitatively using semiclassical photodetection theory in which the fields are described classically and photodetection incurs the shot noise associated with the discreteness of the electron charge whereas quantum photodetection theory is required for the spdc case , owing to the nonclassical nature of its output state @xcite . nevertheless , to provide a unified development
, we will employ quantum photodetection to characterize both pseudothermal and spdc ghost imaging .
we begin , in sec .
[ sec : setup ] , with the theoretical framework for our analysis . here
we will describe our models for the pseudothermal and spdc sources , reflection from the target , atmospheric propagation , quantum photodetection , and image formation . in sec .
[ sec : rescon ] we will derive expressions for the spatial resolution and image contrast for pseudothermal and spdc ghost images , and in sec . [
sec : snr ] we will derive their respective signal - to - noise ratios ( snrs ) . then , in sec .
[ sec : comp ] , we shall consider a computational ghost imager , in which the reference beam is calculated instead of measured @xcite . because the reference beam is not an optical field
, we find it convenient in sec .
[ sec : comp ] to employ the semiclassical theory .
this section will also include a quantitative performance comparison between pseudothermal , sdpc , and computational ghost images formed in reflection .
we conclude , in sec .
[ sec : discussion ] , with a discussion of our findings , and an assessment of the situations in which reflective ghost - imaging might have practical value .
m to the ccd camera , while the signal beam travels @xmath0 m to the target , and the reflected light travels @xmath0 m to the bucket detector . the image is then formed by cross correlation between the photocurrents obtained from the bucket detector and the high spatial - resolution ( ccd camera ) detector.,width=336 ]
we are interested in ghost imaging with classical pseudothermal light , and the nonclassical output of an spdc . in both cases our signal and reference beams
will be represented quantum - mechanically as @xmath1-units , @xmath2-going , positive - frequency , field operators@xmath3 and @xmath4 , as functions of transverse - coordinate vector @xmath5 and time at the output of the fig .
[ fig : setup ] beam splitter .
their baseband field operators have the canonical commutation relations @xmath6 = 0 , \\[.05 in ] \label{eq : comm2}&[{\hat{e}}_j({\bm\rho}_1,t_1),{\hat{e}}^\dagger_k({\bm\rho}_2,t_2 ) ] = \delta_{jk}\delta({\bm\rho}_1-{\bm\rho}_2)\delta(t_1-t_2),\end{aligned}\ ] ] for @xmath7 and @xmath8 , where @xmath9 is the kronecker delta , and @xmath10 is the unit impulse .
following @xcite , we will take both the pseudothermal and spdc sources to produce signal and reference beams that are in zero - mean , jointly gaussian states , which are then completely characterized by their nonzero autocorrelation and cross - correlation functions .
all the autocorrelations will be assumed to have the same gaussian - schell model form , @xmath11 & = \frac{2p}{\pi a_0 ^ 2}e^{-\frac{|{\bm\rho}_1|^2+|{\bm\rho}_2|^2}{a_0 ^ 2 } } e^{-\frac{|{\bm\rho}_1-{\bm\rho}_2|^2}{2\rho_0 ^ 2}}e^{-\frac{(t_1-t_2)^2}{2 t_0 ^ 2}},\end{aligned}\ ] ] for @xmath8 , where @xmath12 is the photon flux , @xmath13 is the source s @xmath14 intensity radius , @xmath15 is its coherence length , and @xmath16 is its coherence time .
the distinction between the gaussian - state models for the pseudothermal and spdc sources lies in their signal - reference cross correlations , as we will now explain .
the pseudothermal signal and reference beams start as continuous - wave ( cw ) laser light , that is passed through a rotating ground - glass diffuser to make it spatially incoherent .
signal and reference beams are then obtained by 50 - 50 beam splitting .
the signal and reference thus have no phase - sensitive cross correlation , and the maximum phase - insensitive cross correlation allowed by classical physics given their autocorrelation functions , i.e. , @xmath17 \begin{split}\label{eq : ccorrpis } { \langle { \hat{e}}^\dagger_s&({\bm\rho}_1,t_1){\hat{e}}_r({\bm\rho}_2,t_2 ) \rangle}_c \\[.05 in ] & = \frac{2p}{\pi a_0 ^ 2}e^{-\frac{|{\bm\rho}_1|^2+|{\bm\rho}_2|^2}{a_0 ^ 2 } } e^{-\frac{|{\bm\rho}_1-{\bm\rho}_2|^2}{2\rho_0 ^ 2}}e^{-\frac{(t_1-t_2)^2}{2 t_0 ^ 2 } } , \end{split}\end{aligned}\ ] ] where the subscript @xmath18 indicates that these are classical - state cross correlations .
the coherence time is directly related to how fast the ground - glass rotates , the coherence length is typically on the order of a few wavelengths , and the intensity radius is set by the radius of the initial laser beam .
a type - ii phase - matched , collinear , cw spdc source emits orthogonally - polarized , co - propagating signal and idler beams , which are separated by a polarizing beam splitter to become the signal and reference beams for ghost imaging .
they are in a maximally - entangled , jointly - gaussian state with no phase - insensitive cross correlation , and the maximum phase - sensitive cross correlation permitted by quantum theory given their autocorrelation functions , namely @xcite @xmath19 \begin{split}\label{eq : qcorrps } { \langle { \hat{e}}_s({\bm\rho}_1,t_1)&{\hat{e}}_r({\bm\rho}_2,t_2 ) \rangle}_q = \frac{2p}{\pi a_0 ^ 2 } e^{-\frac{|{\bm\rho}_1|^2+|{\bm\rho}_2|^2}{a_0 ^ 2 } } \!\ ! \left[e^{-\frac{|{\bm\rho}_1-{\bm\rho}_2|^2}{2\rho_0 ^
2}}e^{-\frac{(t_1-t_2)^2}{2t_0 ^ 2 } } + i \left(\frac{2}{\pi}\right)^\frac{1}{4 }
\!\sqrt{\frac{a_0 ^ 2}{p t_0 \rho_0 ^ 2}}e^{-\frac{|{\bm\rho}_1-{\bm\rho}_2|^2}{\rho_0 ^ 2}}e^{-\frac{(t_1-t_2)^2}{t_0 ^ 2 } } \right ] , \end{split}\end{aligned}\ ] ] where the subscript @xmath20 indicates that these are quantum ( nonclassical ) cross correlations . the signal and reference beams in fig .
[ fig : setup ] first propagate @xmath0 m over spatially - separated paths from their shared source location @xcite .
the reference beam is then measured by a high - spatial resolution ccd camera , while the signal field reflects off the target and travels @xmath0 m to the bucket detector .
we will assume statistically - independent turbulence on all three paths : the reference path ( @xmath21 ) from the source to the ccd ; the signal path ( @xmath22 ) from the source to the target ; and the target - return path ( @xmath23 ) from the target to the bucket detector . to account for the kolmogorov - spectrum turbulence that is uniformly distributed along the three propagation paths , we make use of the quantum version of the extended huygens - fresnel principle @xcite @xmath24 where @xmath25 , and
@xmath26 is a complex - valued random process that encapsulates the effects of turbulence on path @xmath27 from @xmath5 to @xmath28 .
the real and imaginary parts of @xmath29 , which we will denote @xmath30 and @xmath31 , are the logamplitude and phase fluctuations imposed on the field received at transverse coordinate @xmath28 over path @xmath27 from a point source at transverse coordinate @xmath5 . because all the path lengths in fig .
[ fig : setup ] are equal , we have suppressed the @xmath32 line - of - sight propagation delay in eq .
( [ eq : prop ] ) .
we have also suppressed any time - dependence of the turbulence .
physically , this amounts to assuming that the integration time , @xmath33 , employed in the fig .
[ fig : setup ] correlator is shorter than the @xmath34ms coherence time of the turbulence .
however , that assumption is _ not necessary for deriving the spatial resolution and image contrast achieved in the presence of turbulence , although it
_ will be relevant to our signal - to - noise ratio analysis .
_ _ the mutual coherence function of @xmath35 is taken to be @xcite @xmath36 & = e^{-(|{\bm\rho}'_1-{\bm\rho}'_2|^2+({\bm\rho}'_1-{\bm\rho}'_2)\cdot({\bm\rho}_1-{\bm\rho}_2 ) + |{\bm\rho}_1-{\bm\rho}_2|^2)/2\rho_m^2},\end{aligned}\ ] ] which is the square - law approximation to the rigorous @xmath37-law behavior for uniformly - distributed turbulence @xcite . in eq .
( [ eq : turbcorr ] ) , @xmath38 is the turbulence coherence length on path @xmath27 , given by @xmath39 in terms of the constant turbulence strength - parameter , @xmath40 , along path @xmath27 @xcite .
we will assume that the path length @xmath41 satisfies the far - field coherence - propagation conditions for pseudothermal and spdc light discussed in @xcite , viz .
, @xmath42 , and @xmath43 , respectively . for convenience ,
we also define @xmath44 and @xmath45 .
these are the far - field , turbulence - free , on - target coherence length and intensity radius produced by the gaussian - schell autocorrelation function from eq .
( [ eq : autocorr ] ) .
most real - world targets have surfaces that are sufficiently rough on the scale of an optical wavelength that reflections from them present fully - developed speckle patterns at appreciable distances from the surface @xcite .
we will model such a target as a planar 2-d object with random , microscopic height variations having depths on the order of several wavelengths , and transverse variations on the order a wavelength and an average intensity - reflection coefficient @xmath46 .
thus , following laser - radar theory @xcite , we shall assume that this target has a random field - reflection coefficient , @xmath47 , that is a zero - mean , complex - valued gaussian random process characterized by the autocorrelation function @xmath48 where the deterministic pattern @xmath46 is what we are tying to image .
target reflection is a passive process , so we require @xmath49 , which is in conflict with the gaussian statistics and the delta - function term in eq .
( [ eq : scatter ] ) . that delta function , however , leads to quasi - lambertian reflection , implying that at standoff distances a realistic bucket detector will only capture a very small fraction of the reflected light . hence using the reflection statistics from the preceding paragraph
is a reasonable approximation for ghost imaging of rough - surfaced targets .
however , because some light may be absorbed by or transmitted through the target , the quantum model for target reflection needs to be @xmath50 here @xmath51 is a vacuum - state field operator , needed to preserve the free - field commutator relations given , for the source fields , in eqs . and . in what follows we will assume that the target s intensity - reflection coefficient , @xmath46 , is space - limited to the region illuminated by the signal beam , i.e. , @xmath52 for @xmath53
this condition will give the ensemble - average pseudothermal and sdpc ghost images convolution - integral relationships with @xmath46 from which to determine their spatial resolutions .
see @xcite for information about ghost imaging field - of - view when the preceding restriction on @xmath46 is not satisfied .
loss also occurs in the photodetection process , because the detectors in fig .
[ fig : setup ] will , in general , have sub - unity quantum efficiencies .
therefore , the field operator @xmath54 driving the photocurrent , @xmath55 , from ccd pixel @xmath56 , and the field operator @xmath57 driving the photocurrent , @xmath58 , from the bucket detector satisfy @xcite @xmath59 for @xmath60 or @xmath61 . here , @xmath62 is the quantum efficiency , which is taken to be the same for both the bucket detector and the ccd camera , and @xmath63 is a vacuum - state field operator whose inclusion models the loss incurred by having @xmath64 , cf .
eq . ( [ eq : targetloss ] )
. we will assume ghost - image formation via continuous - time , pixel - wise correlation of the photocurrents from the ccd and the bucket detector , formalized as @xmath65 where @xmath66 is the center coordinate of the ccd s @xmath56th pixel , and @xmath33 is the correlation integration time . from quantum photodetection theory @xcite
, we have that @xmath67 where @xmath68 is the effective ( sub - unity quantum efficiency ) photon flux on the active region @xmath69 for detector @xmath27 , and @xmath70 is the impulse response of the detectors output circuit , which we have taken to be the same for both the bucket detector and the ccd camera .
to enable the performance evaluations in the sections to follow , we make the following assumptions .
first , each pixel is small enough that the the average photon flux impinging on it obeys @xmath71 , where @xmath72 is the area of the @xmath73 .
second , we assume that the detector has a gaussian impulse response , @xmath74 whose bandwidth , @xmath75 , greatly exceeds that of the incident light for the pseudothermal source ( @xmath76 ) , and is much smaller than that of the incident light for the spdc source ( @xmath77 ) .
finally , we assume that the correlation integration time is sufficiently long to capture a ghost image , so @xmath78 for the pseudothermal source , and @xmath79 for the spdc source .
the spatial resolution and contrast of the ghost image are properties of the average cross correlation , @xmath80 . to determine this ensemble - average cross correlation we : ( 1 ) write @xmath58 and @xmath55 in terms of the field operators that drive them , @xmath57 and @xmath54 ; ( 2 ) apply the canonical commutator relations to normally - order the resulting photocurrent - product expressions ; and ( 3 ) use eq . to write the detected fields in terms of @xmath81 and @xmath82 . from the linearity of expectation , we can bring the ensemble - average operation inside the time - domain integration to get @xmath83 & \times { \langle { \hat{e}}'^\dagger_r({\bm\rho}_p,\tau_1 ) { \hat{e}}'^\dagger_t({\bm\rho},\tau_2 ) { \hat{e}}'_r({\bm\rho}_p,\tau_1 ) { \hat{e}}'_t({\bm\rho},\tau_2 ) \rangle}.\end{aligned}\ ] ] we then use eqs . and to back - propagate the target - return field operator to the target , and find @xmath84 & = & \int\ ! d{\bm\rho}_2 \!\int\ !
d{\bm\rho}_3\ , { \langle t^*({\bm\rho}_2)t({\bm\rho}_3 ) \rangle}{\langle e^{\psi^*_t({\bm\rho},{\bm\rho}_2)}e^{\psi_t({\bm\rho},{\bm\rho}_3 ) } \rangle}\nonumber \\[.05 in ] & \times & { \langle { \hat{e}}'^\dagger_r({\bm\rho}_p,\tau_1 ) { \hat{e}}'^\dagger_s({\bm\rho}_2,\tau_2 ) { \hat{e}}'_r({\bm\rho}_p,\tau_1 ) { \hat{e}}'_s({\bm\rho}_3,\tau_2 ) \rangle } \nonumber \\[.05 in ] & \times & \frac{e^{-ik_0(|{\bm\rho}_2|^2-|{\bm\rho}_3|^2)/2l + ik_0{\bm\rho}\cdot({\bm\rho}_2-{\bm\rho}_3)/l}}{(\lambda_0l)^2},\end{aligned}\ ] ] where we have used the statistical independence of the target s reflection coefficient , the signal and reference fields at the source , and the turbulence on all three propagation paths . substituting for @xmath85 from eq . and performing the @xmath86 integral eliminates the turbulence term from the preceding @xmath80 expression , which shows that turbulence on the the target - to - bucket path has no impact on the ghost image s spatial resolution or image contrast .
now , by back - propagating the field operators to the source by means of eq . with @xmath87we
have that @xmath88 & = & \frac{1}{\lambda_0 ^ 4l^6}\int\ ! d{\bm\rho}_2 \ , { \mathcal{t}}({\bm\rho}_2 ) \int\ !
d{\bm\rho}'_1 \int\!d{\bm\rho}'_2 \int\!d{\bm\rho}''_1 \int \ ! d{\bm\rho}''_2 \nonumber \\[.05 in ] & \times & { \langle { \hat{e}}^\dagger_r({\bm\rho}'_1,\tau_1 ) { \hat{e}}^\dagger_s({\bm\rho}'_2,\tau_2 ) { \hat{e}}_r({\bm\rho}''_1,\tau_1 ) { \hat{e}}_s({\bm\rho}''_2,\tau_2 ) \rangle } \nonumber \\[.05 in ] & \times & { \langle e^{\psi^*_r({\bm\rho}_p,{\bm\rho}'_1)}e^{\psi_r({\bm\rho}_p,{\bm\rho}''_1 ) } \rangle}{\langle e^{\psi^*_s({\bm\rho}_2,{\bm\rho}'_2)}e^{\psi_s({\bm\rho}_2,{\bm\rho}''_2 ) } \rangle } \nonumber \\[.05 in ] & \times & e^{ik_0[{\bm\rho}_p\cdot({\bm\rho}'_1-{\bm\rho}''_1 ) + { \bm\rho}_2\cdot({\bm\rho}'_2 - { \bm\rho}''_2)]/l},\end{aligned}\ ] ] where we have again invoked the statistical independence of the signal and reference fields at the source and the turbulence on the source - to - ccd , and source - to - target paths
. we can now directly evaluate all remaining moments .
the turbulence moments are available from eq . , and the fourth - order field moment
can be expressed in terms of second - order moments with the gaussian - state moment - factoring theorem @xcite : @xmath89 until this point , our analysis applies equally well to the pseudothermal and spdc ghost imagers .
that equivalence disappears , however , when we evaluate the second - order moments in eq . using eqs .
for the pseudothermal ghost imager , and using eqs . , , and for the spdc ghost imager .
the results we obtain are @xmath90 , \label{eq : cavgmean}\end{aligned}\ ] ] for the pseudothermal ( classical ) imager , and @xmath91 & \times & \left [ 1 + \frac{\omega_b t_0}{4}\frac{e^{-|{\bm\rho}-{\bm\rho}_p|^2/\alpha\rho_l^2}}{\alpha}\left(1+\frac{1}{4\sqrt{\pi}\mathcal{i}}\right)\right ] .
\label{eq : qavgmean}\end{aligned}\ ] ] for the spdc ( quantum ) ghost imager , where @xmath92 is the area of @xmath93 , @xmath94 is the source brightness in photons per spatiotemporal mode , and @xmath95 is a turbulence - induced performance - degradation factor .
we see that both average correlation functions are the sum of the same featureless background term @xmath96 and different image - bearing terms : @xmath97 for the classical - state ( pseudothermal ) source , and @xmath98 & \times & \frac{\omega_b t_0}{4}\int\!d{\bm\rho}\ , { \mathcal{t}}(-{\bm\rho } ) e^{-\frac{|{\bm\rho}-{\bm\rho}_p|^2}{\alpha\rho_l^2}},\end{aligned}\ ] ] for the nonclassical - state ( quantum ) source .
the average pseudothermal ghost image is erect , and consists of a scaled version of the target s intensity - reflection coefficient @xmath46 convolved with a gaussian point - spread function whose @xmath99 radius is @xmath100 .
the average spdc ghost image is inverted , with a different scaling but the same point - spread function .
thus , both the pseudothermal and spdc ghost images have the same spatial resolution , @xmath101 } = \label{eq : res } } \\[.05 in ] & & \hspace*{-.1 in } \frac{\lambda_0l } { \pi a_0}\sqrt{1+\frac{a_0 ^ 2}{2 } ( 1.09 k_0 ^ 2 l)^\frac{6}{5}\left[(c_{n , s}^2)^\frac{6}{5 } + ( c_{n , r}^2)^\frac{6}{5}\right]},\end{aligned}\ ] ] where we have utilized eq . to write the turbulence coherence lengths on each path in terms of their respective strength parameters .
equation shows that turbulence - limited resolution prevails for the far - field lensless ghost image when either the source - to - target coherence length ( @xmath102 ) or the source - to - reference coherence length ( @xmath103 ) becomes smaller than the source size ( @xmath13 ) . when @xmath104 and @xmath105 , far - field lensless ghost imaging in reflection yields the same @xmath106 spatial resolution previously reported for far - field lensless ghost imaging in transmission @xcite
. however , even when turbulence does not degrade ghost - image spatial resolution , there is still a significant difference between the transmissive and reflective cases , namely the @xmath107 factor that is present in our results but is absent from those in @xcite .
this factor , which will obey @xmath108 in a standoff - sensing scenario , is the bucket detector s angular subtense at the target , and represents the fraction of the quasi - lambertian reflected light that this detector collects .
no such factor appeared in @xcite , because that work assumed the bucket detector collected all the light that was transmitted through the target .
the featureless background terms in our pseudothermal and spdc ghost images affect the images contrast . to assess these contrasts
, we will adopt the definition and approach presented in @xcite . in particular
, the image contrast @xmath109 is the ratio of the difference between the brightest and darkest pixels in the image to the featureless background : @xmath110-\min_{\mathcal{r}}[\hat{c}({\bm\rho}_p)]}{c_0},\end{aligned}\ ] ] where @xmath111 is the region within which @xmath46 is nonzero .
we will assume that the ghost imager completely resolves the target , so that we can use the approximation @xmath112 defining @xmath113 as the target s reflective cross - section , and taking @xmath114 = 1 $ ] , we get the following image contrast expressions for our two ghost imagers , @xmath115 and @xmath116 if @xmath117 where it is nonzero , then @xmath118 is the area of the target . moreover , @xmath119 is the spatial - resolution area in the absence of turbulence .
it follows that the pseudothermal ghost - image contrast is approximately @xmath120 , where @xmath121 is number of no - turbulence resolution cells in the image .
note that our analysis to this point has assumed a dc - coupled correlator in fig .
[ fig : setup ] .
the contrast of a pseudothermal ghost image can be improved by suppressing the background term through ac - coupling , i.e. , measuring the cross covariance , rather than the cross correlation , between the bucket and high spatial - resolution photodetector s outputs , as was done in the pseudothermal experiments of scarcelli _ et al .
_ the pseudothermal imager s contrast does not depend on its source brightness , @xmath122 , but that for the spdc does . at high source brightness ,
@xmath123 , the spdc image contrast reduces to the result we found for the pseudothermal case .
the normal operating regime for a cw spdc , however , is one of low brightness , @xmath124 , for which eq .
reduces to @xmath125 the number of photon pairs emitted by an spdc in one detector time - constant is low , so @xmath126 , and our assumption that the target is contained within the region illuminated by the signal beam implies that @xmath127 if @xmath117 where it is nonzero .
consequently , unlike the pseudothermal case , @xmath128 is the norm for sdpc ghost imaging in dc - coupled operation .
in keeping with the analysis done in @xcite , we will be evaluating the signal - to - noise ratio ( snr ) pixel - wise as the ratio of the square of the expected value to the variance for the photocurrent cross - correlation function , @xmath129 } = \frac{{\langle \hat{c}({\bm\rho}_p ) \rangle}^2}{{\langle \hat{c}^2({\bm\rho}_p ) \rangle}-{\langle
\hat{c}({\bm\rho}_p ) \rangle}^2}.\end{aligned}\ ] ] the main complication in evaluating eq . is the second moment , @xmath130 after using eq . to write the currents in terms of the fields , eqs .
and to normally order the fields , and eqs . and
to propagate the fields back to the source , we are left needing to evaluate an eighth - order moment of the fields , two sixth - order moments of the fields , a fourth - order moment of the fields , a fourth - order moment of the target , three fourth - order moments of the turbulence , and propagation of these terms over 12 unique paths . to make the analysis more tractable , we use several simplifying assumptions .
first , we ac - couple both of the detector outputs by inserting a narrowband notch centered at the origin in frequency space , resulting in the frequency response @xmath131 where @xmath132 .
we also assume that the ghost imager can resolve all structure on the target , allowing us to use the approximation in eq . .
to evaluate the fourth - order turbulence moments , we can no longer use the square - law approximation in eq .
that models the turbulence s mutual - coherence function as gaussian , because this approximation is not valid for fourth - order moments @xcite . to evaluate these terms
we assume that the turbulence coherence length on each path is much larger than the on - target coherence length @xmath133 .
we also assume the turbulence coherence lengths are all much larger than @xmath13 , implying that the turbulence functions are independent of coordinates at the source and detection planes .
although these assumptions are somewhat limiting , they still allow weak - to - medium strength turbulence for a typical situation .
for instance , if the target is standing off @xmath134 km , the source aperture has radius @xmath135 cm , and we are at a wavelength @xmath136 m , this assumption is satisfied for @xmath137 .
moreover , this situation has no spatial - averaging of the turbulence at the target , and thus represents a worst - case scenario for the snr . for @xmath138 and @xmath139
being coordinates at the ccd or bucket detectors , and @xmath140 being coordinates at the target , the preceding coherence - length assumptions imply that @xmath141 + 2{\rm re}[\psi_m(\bm 0 , { \bm\rho}_2 ) ] } \rangle}.\end{aligned}\ ] ] decomposing the turbulence function into its log - amplitude and phase components as @xmath142 , and taking the logamplitude to be gaussian distributed with mean equal to minus its variance @xcite , then gives @xmath143
\approx e^{4 k_{\chi_m } ( { \bm\rho}_1-{\bm\rho}_2)},\end{aligned}\ ] ] where @xmath144 for @xmath145 , is the log - amplitude covariance function . the logamplitude variance will be taken to be @xcite @xmath146 where we have used the rytov - approximation expression , whose validity is ensured by our assumption of weak - to - medium strength turbulence . even with these simplifications ,
the snr evaluation is quite tedious , if somewhat straightforward .
the higher - order field moments and the fourth - order target moment can be reduced to a sum of the products of second - order moments by application of the gaussian moment - factoring theorem .
the ac - coupling removes several of these terms , and we are left with integrals and fourier transforms of the product of gaussian terms , the turbulence correlation functions , and the target s intensity - reflection coefficient , @xmath46 . to simplify our final snr expressions ,
we define two new terms , @xmath147 where @xmath148 is the two - circle overlap function for circles of diameter @xmath149 , @xmath150 & & \hspace*{-.2in}\left\{\begin{array}{ll } \frac{d^2}{2}\left [ \cos^{-1 } \left ( \frac{|\bm\zeta|}{d}\right ) -\frac{|\bm\zeta|}{d}\sqrt{1-\frac{|\bm\zeta|^2}{d^2 } } \right ] , & \mbox{for $ |\bm\zeta|\le d$ } \\[.05 in ] 0 , & \text{elsewhere , } \end{array}\right.\end{aligned}\ ] ] @xmath151 , and we have utilized the dimensionless difference coordinates @xmath152 , where @xmath28 and @xmath153 are coordinates at the bucket detector .
both @xmath154 and @xmath155 have significant physical interpretations .
similar to @xmath118 , @xmath154 is another measure of the reflective area of the target , and thus directly related to the number of on - target resolution cells , and the subsequent amount of time it takes to form an image .
@xmath155 is a metric for the spatial averaging of the target - induced speckle over the bucket detector .
for very small detectors @xmath156 , while for large detectors it is inversely proportional to the area of the bucket detector , viz .
@xmath157 . to further simplify our snr expressions ,
we make two final assumptions : first that @xmath158 , which is equivalent to saying that the ghost image consists of at least @xmath159 resolution cells ; and second , that the bucket detector is at least as large as the area of the source beam , or @xmath160 . applying these conditions we arrive at the following snr expressions for pseudothermal and spdc reflective ghost imaging through atmospheric turbulence , @xmath161 ^ 2}{\displaystyle \frac{4}{\omega_bt_i}\delta^2\text{source } + \delta^2\text{path}\left[1+\frac{1}{4\sqrt{\pi}\mathcal{i}}\right]^2 + \frac{t_0 4\sqrt{2}}{t_i}\delta^2\text{detect}\left[1+\frac{1}{4\sqrt{\pi}\mathcal{i}}\right]+ \frac{t_0}{t_i}\delta^2\text{mix}\left[\frac{2}{\sqrt{3}}+\frac{1}{4\sqrt{\pi}\mathcal{i}}\right]}.\end{gathered}\ ] ] where , for ease of comparison with the pseudothermal imager
, we have inverted the image coordinates in the spdc case .
the terms that appear in the noise denominators of eqs .
( [ eq : csnr ] ) and ( [ eq : qsnr ] ) are as follows : @xmath162 , \\ & \delta^2\text{detect } = \frac{{\mathcal{t}}({\bm\rho}_p ) \rho_l^2 \sqrt{\pi}}{16 \sqrt{2 } a_p \eta^2 \mathcal{i}^2}\frac{l^2}{a_b } , \\ & \delta^2\text{mix } = \frac{{\mathcal{t}}({\bm\rho}_p)}{\eta\mathcal{i}}\frac { l^2}{a_b}e^{4\sigma_r^2}\nonumber\\ & \quad\quad+\frac{\pi\rho_l^2{\mathcal{t}}^2({\bm\rho}_p)}{a_p\eta\mathcal{i}}\left[\frac{4}{3}+\beta^{-1}\right]e^{4(\sigma_s^2+\sigma_t^2)},\end{aligned}\ ] ] where @xmath163 for @xmath164 denotes the turbulence - induced logamplitude variance on the reference ( @xmath21 ) , signal ( @xmath22 ) , and target ( @xmath23 ) paths .
each of these noise terms has an important physical interpretation , as we explain below .
the first noise term , @xmath165 , captures the fluctuations associated with decorrelation of transverse locations in the image arising from the source s time evolution .
thus , it is inversely proportional to the number of on - target resolution cells , @xmath166 , and the number of different source patterns measured .
for the pseudothermal imager , for which @xmath167 holds , this is the number of field - coherence times in the integration time , so we find @xmath165 scaled by @xmath168 . for the spdc imager we have @xmath169 , so it is the detectors response time that limits the measurements correlation behavior , hence we find @xmath165 scaled by @xmath170 in that imager s snr denominator .
the @xmath171 term is due to the optical - path interactions that degrade the measurement , i.e. , the turbulent atmosphere , and the scattering off the rough target .
the noise contribution from target scattering can be mitigated by aperture averaging , which , as noted above , is quantified by @xmath155 .
the effect of the turbulence is somewhat more complicated , as it contributes a scaling factor @xmath172 to all field - variation terms on path @xmath173 .
in addition , the logamplitude fluctuations also cause a deviation that results in the squared - mean term in the denominator of eq .
no longer being canceled .
these effects are particularly devastating because @xmath171 is the only noise term whose strength is independent of the detector integration time .
this is because we have assumed time - independent target behavior and @xmath33 less than the @xmath34ms turbulence coherence time .
consequently , regardless of the source brightness and integration time , @xmath171 sets an ultimate upper limit on the snr .
the noise term @xmath174 results from the statistics of photodetection at both the reference and bucket detectors , and as such is inversely proportional to the photon fluxes impinging on their active regions .
thus , @xmath174 is inversely proportional to the size of each pixel , @xmath72 , and the solid - angle subtense factor @xmath107 .
detection noise is mitigated by temporal averaging , but exacerbated by having a fast detector , so the pseudothermal @xmath174 is scaled by @xmath175 , while for the spdc imager it is scaled only by @xmath168 . the spdc case is also scaled by by @xmath176 , a result of that source s stronger - than classical correlation . the @xmath177 term contains noise contributed by mixing between source plus path noise on the reference arm with detection noise on the signal arm , and vice versa . for both imagers
this term has a temporal scaling of @xmath168 , and for the spdc imager the quantum correlation imparts an additional factor of @xmath178 . to get a better understanding of the snr behavior for pseudothermal and spdc ghost imagers , we will consider three limiting cases .
the first is the long integration - time limit in which @xmath171 dominates the noise denominators for both the pseudothermal and spdc ghost imagers , leading to the _
identical maximum ( saturation ) snr value @xmath179-e^{-4(\sigma_r^2+\sigma_s^2+\sigma_t^2)}},\end{aligned}\ ] ] regardless of the strength of the illumination . _ for our next two cases , we will assume the integration time is sufficiently short that snr saturation is not approached .
then , in the high - brightness ( @xmath180 ) case , for which detection noise can be neglected , performance is dominated by source fluctuations , @xmath165 , yielding @xmath181 for pseudothermal ( @xmath18 ) and spdc ( @xmath20 ) imagers , respectively . with all parameters other than source coherence - time being equal
, the spdc high - brightness snr exceeds that of the pseudothermal source by a factor @xmath167 , where @xmath16 is the pseudothermal source s coherence time .
this is because the much faster fluctuations in the spdc output fields lead to faster decorrelation , and hence higher snr .
that said , however , the reader is reminded that cw spdc operation is ordinarily in the low - brightness regime , hence we turn our attention now to that special case . in the low - brightness , @xmath182 , regime , snr behavior
is dominated by the beat between the detection noises , @xmath174 .
here we find that @xmath183 for the pseudothermal and spdc cases .
the spdc snr is linear in source brightness , whereas the pseudothermal snr is quadratic in that parameter .
this behavior is to be expected , because the spdc low - brightness output approaches the biphoton state , which is known to have lower detection noise than classical - state light .
furthermore , @xmath16 for the pseudothermal source will greatly exceed @xmath16 for the spdc , and , as noted in the preceding paragraph , @xmath167 holds for the pseudothermal source .
it follows that the low - brightness spdc snr is much higher than that of the pseudothermal imager .
the reader is cautioned to remember , however , that pseudothermal ghost imaging is typically performed in the high - brightness regime .
it is also instructive to compare our snr results which apply to reflective ghost imaging through atmospheric turbulence to the transmissive ghost imaging , no - turbulence results from @xcite .
the principal differences between the snr behaviors in these two cases is as follows .
our snr expressions have the average intensity reflection coefficient @xmath184 appearing in lieu of the magnitude - squared of the field reflection coefficient @xmath185 that is seen in the transmissive case . moreover , the transmissive - case snr from @xcite has neither target - induced speckle nor turbulence contributions , so it has neither @xmath171 term nor any turbulence - induced noise - magnification factors .
the reflective case s @xmath165 term has a factor of @xmath186 and its @xmath177 term has a factor of @xmath187 , but both @xmath188 contributions are absent from the transmissive case .
these additional reflective - case contributions are direct results of the rough - surface scattering . in both transmissive and reflective operation
there is averaging of the source randomness at the target , but in the reflective case the fields scatter incoherently off the target , and their fourier components are averaged at the bucket detector .
this second averaging depends on the ratio of the source size to the bucket size , i.e. , @xmath189 .
as @xmath189 increases , the bucket detector captures more light , this term becomes less significant , and we approach the transmissive behavior for these terms . finally , the reflective - snr terms associated with the bucket detector measurement in @xmath174 and @xmath177 are scaled by @xmath190 , relative to the corresponding terms in the transmissive case .
this scale factor is the inverse of the solid angle subtended at the target by the bucket detector .
its presence in the reflective case is a consequence of the quasi - lambertian scattering from the rough surface it follows from the snr differences highlighted above that the low - brightness snr asymptotes for transmissive and reflective operation only differ by the latter s including the solid - angle subtense factor @xmath107 .
their high - brightness snr asymptotes are also similar , with the only difference being the reflective case s factor of @xmath186 .
the most profound difference , however , is due to the @xmath171 term that is only present for reflective operation .
this term implies that the snr of reflective ghost imaging has a maximum value , snr@xmath191 , whereas no such saturation occurs in transmissive ghost imaging , i.e. , its snr can grow without bound as @xmath192 increases .
furthermore , in reflective imaging , this limiting effect can be quite severe : even with no turbulence , for @xmath193 we find @xmath194 , and for @xmath195 we have @xmath196 .
so , for realistic standoff sensing , the snr will be limited to single - digit values if no further measures are taken to suppress the fluctuations arising from target - induced speckle and atmospheric turbulence .
we have developed the performance of reflective ghost imaging from classical and nonclassical sources in a quantum gaussian - state framework .
next we consider a computational ghost imager , a variant of the pseudothermal case in which the `` chaotic '' signal field is created by deterministically modulating the wavefront of a laser with a spatial light modulator ( slm ) , and no reference - arm light beam is required . instead ,
knowledge of the slm phase pattern is used to calculate the signal - beam intensity at the target for use in lieu of a reference - beam measurement @xcite .
if an independent pseudorandom phase process is applied to each slm pixel , then far - field propagation can be argued to yield gaussian field statistics per the central limit theorem .
thus it is appropriate for us to model the computational ghost imager s far - field statistics with a gaussian - schell model , taking the source s coherence length to be approximately the width of an slm pixel , and source s intensity diameter to be the diameter of the slm .
the laser that illuminates the slm will be taken to emit coherent - state light , hence the slm s output beam is in classical state and semiclassical theory will give quantitatively identical results for its far - field photodetection statistics as the full quantum treatment @xcite . also , because the reference field at the target is computed , rather than measured , it is most naturally modeled as a complex - valued gaussian random process . consequently , and without loss of generality , we choose to use semiclassical theory for our analysis of computational ghost imaging because it lets us deal with only classical random processes , rather than a combination of quantum field operators and classical random processes .
we thus use @xmath197 to represent the complex envelope of the target - plane signal field , and @xmath198 to represent the computed complex - envelope of the field that would have been present on the ccd camera .
these two complex - valued , zero - mean , gaussian random processes are completely characterized by their nonzero autocorrelation and cross - correlation functions , which we take to be as given in eqs .
for their field - operator pseudothermal counterparts .
that these correlation functions are consistent with classical random - process theory follows from the pseudothermal source s producing classical - state light .
the principal distinction between computational and pseudothermal ghost imaging which is also its main advantage is that there is neither detection noise nor turbulence on its computed reference - arm intensity , which is found via @xmath199 with @xmath200 .
the bucket detector photocurrent , in semiclassical theory , is given by @xmath201 h(t-\tau ) , \end{aligned}\ ] ] where @xmath202 is the impinging photon flux , and conditioned on that photon flux the shot noise @xmath203 is a zero - mean random process with autocorrelation function @xmath204 .
the computational ghost image is then obtained from the correlation function @xmath205 because the shot - noise is zero mean , given @xmath206 , it does not contribute to @xmath207 , hence the computational ghost imager s spatial resolution and image contrast analysis is identical to the sec .
[ sec : rescon ] treatment of the pseudothermal ghost imager once we substitute the computational system s classical random processes for the pseudothermal imager s field operators , and then let @xmath208 , and set @xmath209 to account for the absence of turbulence on the computed reference . using this approach we immediately find the spatial resolution to be @xmath210 where @xmath211 and the image contrast is identical to that given for the pseudothermal imager in eq . . in the latter
regard we note that we could replace @xmath212 in eq .
( [ cgi ] ) with the zero - mean process , @xmath213 , thereby achieving the same high - contrast operation as is realized with ac - coupling in pseudothermal ghost imaging .
turning now to the computational ghost imager s signal - to - noise ratio , we find more significant differences between its analysis and that of the pseudothermal ghost imager . in our sec . [
sec : snr ] snr analysis , the second moment of the photocurrent cross - correlation function involved an eighth - order field moment , two sixth - order field moments , a fourth - order field - moment , three fourth - order moments of the turbulence , and a fourth - order moment of the target s field - reflection coefficient . for the computational ghost imager s snr analysis
we only have to evaluate an eighth - order field moment , a sixth - order field moment , two fourth - order moments of the turbulence , and the fourth - order moment of the target s field - reflection coefficient .
this simplification is because the other terms involved the path and detection noises on the reference arm , which are absent from the computational configuration .
the computational snr is then given by @xmath214-e^{-4(\sigma_s^2+\sigma_t^2 ) } ) + \frac{{\mathcal{t}}({\bm\rho}_p)}{\eta\mathcal{i}}\frac { l^2}{a_b}}.\end{aligned}\ ] ] it follows that computational ghost imaging shares the same saturation snr as pseudothermal and spdc ghost imaging , from eq .
( [ eq : snrsat ] ) , with @xmath215 set to zero because there is no turbulence in the computational imager s reference arm .
likewise , computational ghost imaging has the same high - brightness snr asymptote as pseudothermal ghost imaging , from eq .
( [ eq : snrc_h ] ) , with @xmath216 . on the other hand ,
computational ghost imaging has a significantly improved low - brightness snr asymptote , given by @xmath217 in comparison with _ both pseudothermal and spdc ghost imaging .
_ for quantitative comparison between the snrs of the pseudothermal , spdc , and computational systems we will first have to put them on equal footing .
the dimensionless brightness quantity @xmath122 that appears in our snr formulas is the photon flux per spatiotemporal mode of the source .
however , because of the dramatically different coherence times of the spdc ( @xmath169 ) and the pseudothermal and slm sources ( @xmath167 ) , comparisons based on equal @xmath122 values will be for sources with dramatically different photon fluxes , whereas we would prefer to compare snrs at equal photon fluxes .
to do so we introduce the dimensionless quantity @xmath218 , which is the photon flux per source spatial mode and detector temporal mode . because all three systems are taken to have detectors with identical characteristics , identical values for @xmath219 will yield identical photon fluxes .
making this substitution we can easily compare the low - brightness snrs from eqs .
, , and . in the extreme low - brightness regime , the computational imager has the best snr , followed by the spdc and pseudothermal imagers .
we can also compare the high - brightness limits eqs . and , where the shorter coherence time of the spdc source gives it a clear advantage .
because the different sources exit their low- and high - brightness snr regimes and enter into snr saturation for different photon fluxes , assessing that behavior requires us to explicitly choose some operating parameters .
first , we assume the operating wavelength is @xmath220 m , the source coherence length is @xmath221 mm , the source intensity radius is @xmath222 cm , and that the bucket detector matches the source size , i.e. , @xmath223 .
these give us the on - target parameters of @xmath224 m , and @xmath225 m .
the photodetectors are taken to have @xmath226 quantum efficiency , with the ccd pixel area being @xmath227 .
we assume the target is at @xmath228 km range , that it has an effective area of @xmath229 , and @xmath230 at the point for which we will perform our snr evaluations .
all paths are assumed to have moderate - strength ( @xmath231 ) turbulence . finally , we take the pseudothermal and computational sources to have temporal coherence times satisfying @xmath232hz , while for the spdc source @xmath233hz .
we consider two scenarios : first , when the source is weak enough ( @xmath234 ) that the spdc output is still a stream of biphotons , but we are not deep into the low - brightness regime ; and second , when the source is sufficiently strong ( @xmath235 ) to make the spdc output appear classical @xcite .
the resulting snr curves for these two cases are shown in figs .
[ fig : snr_l ] and [ fig : snr_h ] , respectively . from fig .
[ fig : snr_l ] we see that the spdc imager is the worst performer when it is in its biphoton regime , while the computational ghost imager is to the top performer there . from fig .
[ fig : snr_h ] , however , we see that when the spdc output appears to be classical , its much shorter coherence time allows it to reach the saturation - snr limit well before the other two systems , whose snr curves are nearly identical . .
the source brightness is @xmath234 , so that the spdc output is a stream of biphotons and all three systems are well below their respective high - brightness snr asymptotes .
other parameters used in computing these curves are given in the text.,width=336 ] .
the source brightness is @xmath235 , so that the spdc output appears to be classical and the pseudothermal and computational ghost imagers are in their high - brightness limits wherein they only differ by a factor of @xmath236 .
other parameters used in computing these curves are given in the text.,width=336 ]
we have derived the key performance characteristics of reflective ghost imaging in the context of realistic standoff sensing applications .
this was done within a gaussian - state framework that allowed for analysis of classical pseudothermal sources and the nonclassical output of an spdc .
our results for the spatial resolution , image contrast , and snr were analogous to those calculated for transmissive ghost imaging in the absence of turbulence @xcite , with a few significant differences that result from the roughness of the target surface , turbulence on the optical paths , and the standoff nature of the measurement
. first we shall discuss the impact of rough - surface scattering in the absence of turbulence , then we shall discuss the additional effects incurred when turbulence is present .
rough - surfaced targets scatter their impinging illumination in a random manner , casting random speckle patterns in their far fields that , on average , correspond to a quasi - lambertian distribution .
consequently , the ghost - imager s bucket detector will , on average , collect only a fraction , @xmath107 , of the target - scattered light .
this angular subtense factor which is absent in transmissive ghost imaging appears in the average photocurrent cross - correlation functions in eqs . and , and the noise terms associated with the bucket detector , i.e. , @xmath174 and @xmath177 .
the surface roughness has another major effect on the snr : its time - independent speckle pattern implies the existence of a finite saturation - snr value in the limit of long integration time .
the target - speckle contribution that leads to a saturation snr appears in @xmath171 , and is a function of @xmath189 , the ratio of the source size to the bucket detector size .
here we may incur a trade - off between spatial resolution and snr . specifically , by enlarging the bucket detector we can increase the averaging of the target speckle , and thus the saturation snr , without affecting the imager s spatial resolution .
ultimately , however , there will be a practical limit beyond which the receiving aperture size can no longer be increased .
beyond that point @xmath189 , and thus the saturation snr , can only be increased by decreasing the source size @xmath13 . however , as seen in eq .
, spatial resolution is inversely proportional to @xmath13 , so increasing the saturation snr in this manner will degrade spatial resolution .
the effects of turbulence on the image contrast and spatial resolution of reflective ghost imaging are similar those previously found for transmissive ghost imaging @xcite . in particular :
turbulence does not change the image contrast ; turbulence in the target - to - bucket path has no effect on spatial resolution ; and turbulence on the signal and reference paths degrades spatial resolution in the same manner , i.e. , degradation occurs when they become smaller than the source size . in both spdc and pseudothermal ghost
imaging it is therefore advantageous to propagate the reference field through a controlled ( turbulence - free ) environment .
turbulence also degrades ghost - image snr in two ways : first , its logamplitude fluctuations magnify several existing noise terms ; and second , these fluctuations also contribute to the time - independent noise term @xmath171 , decreasing the saturation snr from its speckle - only value . in computational ghost imaging , the reference field
is computed , removing the turbulence on that path , and thus improving both the spatial resolution and the snr .
the computational case also has significantly reduced noise for low - brightness illumination , as there is no longer detection noise on the reference arm , making its low - brightness snr behavior superior to that of both pseudothermal and spdc ghost imaging . more generally , figs .
[ fig : snr_l ] and [ fig : snr_h ] show that the computational ghost imager has a higher snr than the pseudothermal system in both low - brightness and high - brightness operation , and than the spdc system in its low - brightness ( biphoton ) regime .
however , in high - brightness operation , the spdc ghost imager s significantly shorter coherence time enables it to reach its saturation snr much quicker than either the computational or the pseudothermal ghost imagers .
although a cw spdc will not be able to reach that high - brightness regime , pulsed systems are capable of doing so @xcite . insofar as standoff sensing is concerned , the essential conclusion to be drawn from the preceding summary of our work is simple : the computational approach is the ghost - imaging configuration that shows the most promise for this application .
it has the best spatial resolution , the highest saturation - snr , and general snr performance second only to high - brightness spdc operation .
in addition , the computational approach obviates the need for a physical reference path , which makes it naturally amenable to obtaining 3d ghost images .
in particular , pseudothermal and spdc ghost imaging require that reference - arm measurements be made on an intensity pattern corresponding to the one that is projected onto the target .
as shown theoretically in @xcite and experimentally in @xcite , computational ghost imaging allows reference intensity patterns to be computed at a variety of target ranges so that ghost images can be formed for these target ranges from the same bucket - detector data .
this range sectioning is something that _ can not be done , without separate measurements for each possible target range , in pseudothermal or spdc ghost imaging . _
we have reserved our final comment for our choice of image reconstruction via photocurrent cross - correlation .
it has already been demonstrated @xcite that ghost imaging reconstruction can be performed in transmission through more advanced methods , such as compressive sensing .
a preliminary no - turbulence performance comparison between a simple cw - laser radar and ghost imaging in reflection via cross correlation has shown their similarity in spatial resolution and snr @xcite .
if advanced image - reconstruction techniques can be successfully applied to computational ghost imaging , the resulting system could have notable advantages over current laser radars .
this work was supported by the u.s .
army research office multidisciplinary university research initiative grant no .
w911nf-05 - 1 - 0197 , the w. m. keck foundation center for extreme quantum information theory , the darpa information in a photon program under u.s .
army research office grant no . w911nf-10 - 1 - 0404 , and the darpa quantum sensors program under afrl contract no .
fa8750 - 09-c-0194 , for the sake of generality , we have allowed there to be turbulence in the reference path , although a lens could be used to cast a minified reference - beam pattern onto a ccd camera within a protected environment near the source , hence avoiding turbulence on that path . that no additional quantum - noise term is needed to ensure that this expression for @xmath237 preserves free - field commutator brackets follows from the normal - mode decomposition for propagation through turbulence [ j. h. shapiro , appl
* 13 , * 2614 ( 1974 ) ] .
a similar square - law approximation is available for cases in which the turbulence strength is not uniformly distributed along each propagation path , but we have opted to limit our attention to uniform distributions . in quantum photodetection , the currents at the outputs of the bucket and ccd detectors are classical stochastic processes , @xmath238 and @xmath239 , whose statistics are equivalent to those of the quantum measurements @xmath58 and @xmath55 given in the next paragraph .
we say `` appear classical '' because the phase - sensitive cross - correlation function from eq .
( [ eq : qcorrps ] ) _ always exceeds the classical limit set by the autocorrelation functions from eq .
( [ eq : autocorr ] ) .
however , for @xmath123 , that cross - correlation function is so close in value to the classical limit , given in eq .
( [ eq : ccorrpis ] ) , to render this measure of nonclassicality exceedingly small . _ | recent work has indicated that ghost imaging may have applications in standoff sensing . however , most theoretical work has addressed transmission - based ghost imaging . to be a viable remote - sensing system
, the ghost imager needs to image rough - surfaced targets in reflection through long , turbulent optical paths .
we develop , within a gaussian - state framework , expressions for the spatial resolution , image contrast , and signal - to - noise ratio of such a system .
we consider rough - surfaced targets that create fully developed speckle in their returns , and kolmogorov - spectrum turbulence that is uniformly distributed along all propagation paths .
we address both classical and nonclassical optical sources , as well as a computational ghost imager . |
the most common method to drive a current through a conductor is by applying a voltage . in nanostructures , due to the ease with
which parameters can be tuned , another possibility opens up : _ pumping .
_ in an electron pump a dc current can be generated in a nanoscale conductor , even at zero - bias voltage , by periodically changing some of its properties in time .
if the time dependence of the parameters is slow compared to the internal time scales of the system , such as the dwell time of carriers , pumping is adiabatic . in this case
, the pumped charge does not depend on the detailed time evolution of the pumping cycle but only on its area .
most of the theoretical efforts on pumping have been addressed to systems where the electron - electron interaction is weak and can , therefore , be neglected or treated in a mean - field approach.@xcite for these weakly interacting nanoscale conductors , a well - established formalism for pumping exists based on the dynamical - scattering approach to mesoscopic transport.@xcite recently , several works have investigated pumping through interacting systems.@xcite most of these works have focused on few - electron quantum dots@xcite since these systems are the prototype for nanostructures with strong interaction effects .
an open question regards how the pumping characteristics of an interacting quantum dot depends on the dot s single - particle spectrum . in order to address this problem ,
we have decided to study a quantum dot with a dense spectrum , i.e. , with vanishing level spacing . due to its finite density of states ,
such a quantum dot is referred to as a metallic island .
indeed , pumping in systems of metallic islands has been investigated experimentally already in the early 1990s.@xcite pothier _
@xcite realized a single - electron pump in a setup with two tunnel - coupled metallic islands in series .
pumping was obtained by changing in time the gate voltages of the two islands .
the pumping cycle was chosen to enclose a degeneracy point , where three charge states of the double - island system are degenerate .
this resulted in pumping of exactly one electron per cycle . to describe this experiment ,
it was assumed that the system relaxes always to the ground state and that this transition occurs via first - order tunneling .
although this theoretical explanation is perfectly suited to describe the pumping cycle of ref .
, it would fail in describing weak pumping or pumping cycles nonencircling a degeneracy point .
more experiments on charge pumping through nanostructures were performed on metallic multijunction systems , @xcite double quantum dots in inas quantum wires , @xcite or making use of surface acoustic waves .
@xcite also spin pumping through a quantum dot was reported .
@xcite in the present paper , we develop a diagrammatic real - time approach to adiabatic pumping in systems consisting of metallic islands . to this aim
, we follow the lines of ref . , where adiabatic pumping through a single - level quantum dot was considered .
our approach relies on a systematic perturbative expansion in the tunnel - coupling strengths and therefore is restricted to weak tunnel coupling .
in particular , it is not suitable to study the very - low - temperature regime , where high - order tunneling processes become relevant .
however , it treats the coulomb interaction on the island exactly , allowing us to investigate effects due to strong interaction .
first , we apply our formalism to a single - island system consisting of one metallic island with coulomb interaction , tunnel coupled to two noninteracting leads .
we compute the pumped charge up to first order in the tunnel - coupling strength for any pair of pumping parameters chosen among the charging - energy gap of the system and the left and the right tunnel - coupling strengths . as a result
, we find that the contribution in first order is due solely to the renormalization of the charging - energy gap .
for the case of pumping with the two tunnel - coupling strengths , this term becomes the dominant one .
furthermore , we consider pumping with the charging - energy gaps in a more complex system consisting of two tunnel - coupled metallic islands called the double - island system .
we calculate the pumped charge through this system in lowest order in the tunnel - coupling strength and investigate the issue of pumped - charge quantization . with our technique
we are able to describe the transition from weak to strong pumping , finding charge quantization for strong pumping .
this paper is structured as follows . in sec .
ii we present the model .
the technique to compute the pumped current is introduced in sec .
the results for single - island and double - island systems are presented in secs .
iv and v , respectively .
the conclusions are given in sec .
details on the calculation of the diagrams can be found in appendixes a and b. to keep all formulas transparent we set @xmath0 throughout the paper .
we consider a chain of @xmath1 tunnel - coupled metallic islands .
the leftmost and rightmost islands are tunnel coupled to metallic leads .
the different metallic regions are labeled by the index @xmath2 , where @xmath3 and @xmath4 correspond to the noninteracting metallic leads and @xmath5 to the metallic islands .
the metallic islands , for typical experimental realizations , have a dense single - particle spectrum and can be treated as equilibrium reservoirs of fermionic degrees of freedom , described by a fermi function .
each island is capacitively coupled to a gate voltage , which can be used as a pumping parameter .
the state of the chain is described by the number of excess charges @xmath6 on each island , @xmath7 .
the hamiltonian of the system is @xmath8 .
the operator @xmath9 describes the system in the absence of tunnel coupling @xmath10 where @xmath11 are the single - particle energies , labeled by @xmath12 and by the transverse - channel index @xmath13 .
the spin degree of freedom is included in the sum over transverse channels . in the following
, we will consider the many - channel case , @xmath14 .
the coulomb interaction is taken into account by @xmath15 , which describes the electrostatic energy of a given charge state @xmath16.@xcite this charging energy depends on the gate voltages applied to each single island .
the tunnel coupling is described by the tunneling hamiltonian @xmath17 \ , , \end{aligned}\ ] ] with @xmath18 . for simplicity in eq .
( [ htunn ] ) we have assumed that the tunnel - matrix elements @xmath19 for tunneling through barrier @xmath2 do not depend on the quantum numbers @xmath12 and @xmath20 or on the transverse - channel index @xmath21 . notice that the tunneling hamiltonian eq .
( [ htunn ] ) conserves the transverse - channel index . in eq .
( [ htunn ] ) we have explicitly indicated that the tunnel - matrix elements can be time dependent . the operator @xmath22 is the canonical conjugate of the number operator @xmath23 for the excess charges on island @xmath2 , i.e. , @xmath24=i$ ] .
therefore , @xmath25 are simply ladder operators for the charge on island @xmath2 .
finally , we characterize the junction @xmath2 between the two consecutive islands @xmath2 and @xmath26 by the tunnel - coupling strength @xmath27 where @xmath28 is the quantum resistance , @xmath29 the tunnel resistance of the junction , @xmath30 and @xmath31 the densities of states at the fermi energy for channel @xmath21 of leads or islands @xmath2 and @xmath26 , respectively .
we define also the two - time coupling strength as @xmath32 .
we are interested in the dynamics of the excess charge on each island and not in the large number of noninteracting fermionic degrees of freedom which act effectively as baths.@xcite therefore , we trace over the states of the fermionic sector and we obtain the reduced density matrix @xmath33 , which just describes the charge degrees of freedom . in the basis defined by the states @xmath7 ( with excess charge @xmath6 on the island @xmath2 ) ,
the dynamics of the diagonal and off - diagonal matrix elements of the reduced density matrix are uncoupled .
furthermore , the off - diagonal matrix elements do not contribute to the current .
hence , we need only to consider the diagonal matrix elements , which are the occupation probabilities for the respective states and which we denote by @xmath34 .
the probabilities obey the generalized master equation @xmath35 where @xmath36 is the vector of the probabilities @xmath37 and the matrix element @xmath38 of the kernel @xmath39 includes all transitions from state @xmath40 at time @xmath41 to state @xmath42 at time @xmath43 .
we now proceed by performing an adiabatic expansion , i.e. , a perturbation expansion in the frequency @xmath44 of the time - dependent pumping fields @xmath45 . to describe the adiabatic pumping regime it is enough to retain terms linear in @xmath44 .
the first step consists in performing a taylor expansion of @xmath46 on the right - hand side ( r.h.s . ) of eq .
( [ eq_master ] ) around @xmath43 up to linear order , @xmath47\ , .\ ] ] this expansion is justified if the period of the pumping fields is much longer than the characteristic memory time of the system .
second , we make an adiabatic expansion of the kernels appearing in the generalized master equation as follows : @xmath48 the instantaneous term , @xmath49 , is simply obtained by replacing @xmath45 with @xmath50 . on the other hand , the first adiabatic correction , @xmath51
is obtained by performing the taylor expansion @xmath52 and retaining terms up to first order in the time derivatives . the index @xmath43 in eq .
( [ kernelexp ] ) indicates the time with respect to which the adiabatic expansion has been performed .
the adiabatic expansion for the kernels is valid if the period of the pumping fields is much longer than the characteristic response times of the system , i.e. , if @xmath53 , where @xmath54 are the charging - energy gaps of the islands .
finally , we expand also the probabilities , @xmath55 notice that by construction @xmath56 are the stationary probabilities for the time - independent problem where all parameters are frozen at their values at time @xmath43 . since all expressions now depend on the time difference @xmath57 it is useful to introduce the laplace transform @xmath58 . in the following we make use of the notation @xmath59 and @xmath60 .
even after having performed the adiabatic expansion , the problem can not be solved exactly due to the presence of the coulomb interaction and of the tunnel couplings .
therefore , in addition to the expansion in the pumping frequency @xmath44 , we perform a perturbative expansion in the tunnel - coupling strength @xmath61 .
this limits the range of validity of the method to the weak tunnel - coupling regime .
the order in the expansion in @xmath61 will be indicated by an integer superscript .
for example , @xmath62 indicates the first order in @xmath61 of the adiabatic correction of the kernel .
the perturbative expansion for the instantaneous probabilities is straightforward since it corresponds to the time - independent problem with all parameters frozen at time @xmath43 .
the instantaneous probabilities @xmath63 and @xmath64 are determined from @xmath65 complemented by the normalization conditions @xmath66 and @xmath67 , where @xmath68 is the vector with all components equal to one .
the first adiabatic correction to the probabilities obey the equations @xmath69 complemented by the normalization conditions @xmath70 and @xmath71 . the orders in @xmath61 on the r.h.s . of eqs .
( [ adiabatic-1 ] ) and ( [ adiabatic0 ] ) have been chosen to match those on the left - hand side .
we notice that the expansion in @xmath61 starts with the order @xmath72 .
although this result might seem surprising at first , it is in line with our expansion since @xmath73 .
the pumped current through the left lead can be written as @xmath74 with @xmath75 being the electron charge and @xmath76 .
the kernel @xmath77 describes all processes where the number of electrons entering reservoir @xmath2 minus the ones leaving it equals @xmath20 .
we proceed for the current in the same way as for the master equation ( [ eq_master ] ) performing an adiabatic expansion in the frequency @xmath44 and a perturbative expansion in the tunnel coupling @xmath61 .
since there is no applied transport voltage , the instantaneous current vanishes .
the adiabatic corrections to the current in zeroth and first order in @xmath61 read @xmath78.\end{aligned}\ ] ] the lowest - order contribution to the adiabatic current , given by eq .
( [ cur_ad_low ] ) , scales with @xmath44 and is independent of @xmath61 , in contrast to the dc current through a system with an applied transport voltage , which scales with @xmath61 . since @xmath53 ,
the pumped current goes to zero for vanishing tunnel coupling as it should .
the pumped charge @xmath79 per cycle is simply given by the integral of the pumped current over one period @xmath80 .
we start with summarizing the diagrammatic rules for the kernel @xmath81 , where @xmath21 indicates the order of the expansion in @xmath61 .
examples of first - order diagrams for the single - island system are drawn in fig .
[ fig1 ] .
* draw all topologically different diagrams with @xmath82 directed tunneling lines connecting pairs of vertices containing lead - electron operators .
assign a reservoir index @xmath83 and energy @xmath84 to each of these lines .
assign charge states @xmath85 and the corresponding energies @xmath86 to each element of the keldysh contour connecting two vertices .
furthermore , draw an external line from the upper leftmost beginning of an island propagator to the upper rightmost end of an island propagator that carries the ( imaginary ) energy @xmath87 .
* for each time segment between two adjacent vertices ( independent of whether they are on the same or on opposite branches of the keldysh contour ) write a resolvent @xmath88 , where @xmath89 is the difference of left - going minus right - going energies ( including energies of tunneling lines and the external line - the positive imaginary part of @xmath90 will keep all resolvents regularized ) .
* each vertex containing an island operator @xmath91 gives rise to a matrix element @xmath92 , where @xmath93 is the island state entering ( leaving ) the vertex with respect to the keldysh contour . * the contribution of a tunneling line of reservoir @xmath83 is @xmath94 if the line is going backward with respect to the closed time path and if it is going forward with @xmath95 and @xmath96 . *
the overall prefactor is given by @xmath97 , where @xmath98 is the total number of vertices on the backward propagator and @xmath99 the number of crossings of tunneling lines .
* integrate over all the energies of the tunneling lines and sum over the reservoir indices .
the possible time - dependent parameters are the tunnel - matrix elements @xmath100 and the charging - energy gaps @xmath101 .
the matrix elements @xmath100 enter in the contractions of tunnel vertices as @xmath102 .
the charging - energy gaps @xmath101 enter in the isolated - island propagator @xmath103 .
while for the instantaneous kernels all parameters were taken at time @xmath43 , we now perform an expansion around the same time @xmath43 and keep all contributions linear in a time derivative of the pumping parameters , @xmath104\ , .
\nonumber\end{aligned}\ ] ] the linear factors @xmath105 appearing in these expansions can be included in the diagrammatic rules by introducing an additional external frequency line with the imaginary energy @xmath106 from the vertex at @xmath107 to the rightmost vertex at @xmath43 , performing the first derivative with respect to @xmath108 , and then setting @xmath109 .
similarly , the quadratic terms @xmath110 can be included in the diagrammatic rules by introducing an additional external frequency line with the imaginary energy @xmath106 from the beginning of an island - propagator line at @xmath107 to the rightmost upper end of an island - propagator line at @xmath43 , performing the second derivative with respect to @xmath108 and then setting @xmath109 .
the external frequency lines are drawn as dotted lines in fig .
[ fig1 ] . the rules to compute the contribution to the adiabatic corrections @xmath111 due to the time dependence of @xmath112 read : *
add to all diagrams needed for @xmath81 additional external frequency lines between any vertex @xmath107 and the upper right corner of the diagram and assign to them an ( imaginary ) energy @xmath113 .
note that an external frequency line between two right corners of a diagram does not contribute and can always be omitted .
* follow the rules ( 1 ) to ( 6 ) taking into account the extra lines . *
perform a first derivative with respect to @xmath114 and multiply it by @xmath115 .
sum all the contributions obtained in this way . * set all the external frequencies @xmath108 and @xmath116 to @xmath117 .
the contribution to the adiabatic correction @xmath111 due to the time dependence of the charging energy can be computed in a similar way : * in addition to the external frequency lines added according to rule ( 7a ) , put one more external frequency line from the left corner of the diagram with no vertex to the right corner . * follow the rules ( 1 ) to ( 6 ) taking into account the extra lines . * perform a second derivative with respect to @xmath114 and multiply it by @xmath118 , where @xmath119 is the island state entering ( leaving ) the vertex of the external frequency line at @xmath120 with respect to the keldysh contour .
the term @xmath121 is omitted if the segment associated with @xmath122 does not belong to the diagram .
sum all the contributions obtained in this way . * set the external frequencies @xmath108 and @xmath116 to @xmath117 . finally to compute the current , we need the matrix elements of @xmath123 .
for each contribution @xmath124 all diagrams for which the number of tunneling lines with reservoir index @xmath83 running from the upper to the lower propagator minus the number of those with reservoir index @xmath83 running in the opposite direction equals @xmath125 need to be added .
in this section , we present the results for a single - island system consisting of one metallic island with coulomb interaction , tunnel coupled to two noninteracting leads kept at the same chemical potential . the coulomb interaction on the island
is described by the charging energy @xmath126 , where @xmath127 denotes the charges externally induced via the gate voltage @xmath128 and @xmath129 indicates the charge on the island .
the energy scale for the charging energy is @xmath130 with @xmath131 being the total island capacitance
. at low temperatures only two different charge states contribute to the transport .
we indicate these states as @xmath132 and @xmath133 and their energies as @xmath134 and @xmath135 , respectively .
the occupation probabilities for these states are labeled @xmath136 and @xmath137 .
the charging - energy gap of the island is defined as the energy difference @xmath138 .
an energy sketch of the single - island system is shown in fig .
[ fig2 ] . to calculate the pumped current in zeroth order in @xmath61 we need the instantaneous occupation probabilities in zeroth order in @xmath61 and their adiabatic correction in minus first order .
the instantaneous probabilities can be obtained from eq .
( [ instantan0 ] ) and are simply given by boltzmann factors , @xmath139 where @xmath140 is the inverse temperature , @xmath141 the energy of the island state @xmath42 , and @xmath142 the partition function .
the instantaneous average occupation number of the island in lowest order in @xmath61 can be calculated from the instantaneous probabilities and reads @xmath143 the adiabatic correction to the probabilities can be obtained from eqs .
( [ adiabatic-1 ] ) and ( [ eq_p_i,0 ] ) and is given by @xmath144 where we used the notation @xmath145 and @xmath146 .
furthermore , we define @xmath147 .
since this probability vector is proportional to the time derivative of the instantaneous probabilities @xmath63 , it is an eigenvector of the kernel @xmath148 . to compute the probabilities eqs .
( [ eq_p_i,0 ] ) and ( [ ad_prob_-1 ] ) , we need to evaluate only the instantaneous kernels in first order in @xmath61 .
the expressions for the matrix elements of @xmath148 are given in appendix [ appendix_a ] . by inserting the results for the probabilities in eq .
( [ cur_ad_low ] ) , it yields for the adiabatically pumped current through the single - island system in zeroth order @xmath61 , @xmath149 the average occupation number of the island in zeroth order @xmath61 depends only on the charging - energy gap @xmath150 and not on the tunnel - coupling strengths .
therefore , the pumped current in zeroth order @xmath61 is nonvanishing only if the charging - energy gap is one of the pumping parameters .
the result of eq .
( [ cur_a_0 ] ) has a simple interpretation : if the charging - energy gap is varied in time , the occupation of the island adjusts to the new situation and therefore electrons need to enter or leave the island . the fraction of the resulting electrical current which flows through barrier @xmath2 is given by the ratio @xmath151 .
therefore , this pumping mechanism can be considered as peristaltic .
we start by giving the expression for the instantaneous average occupation number of the island in first order in @xmath61 as follows : @xmath152 \;\dfrac{d}{d \delta } \left\langle n\right\rangle ^{(\text{i},0)}\biggr.\\ % & \;+\biggl.\bigl ( \left\langle n\right\rangle ^{(\text{i},0)}\;-\;\dfrac{1}{2}\bigr ) \;\ ; \partial_\delta\;\re\left [ \varrho(\delta)\right ] \biggr\rbrace \ , , \end{split}\ ] ] the function @xmath153 is defined via @xmath154 with @xmath155 . to keep the integrals convergent we introduce a high - energy cutoff , which in our case is provided by the charging - energy scale @xmath156
. then in the limit of @xmath157 we obtain for the real part of the @xmath158-function , @xmath159 = - 2\ ; \omega \ ; \biggl\lbrace \ln \left ( \dfrac{\beta e_\text{c}}{2\pi } \right ) - \re\psi\left ( i \dfrac{\beta \omega}{2\pi } \right)\biggr\rbrace \,,\ ] ] where @xmath160 denotes the digamma function .
we identify the first term in eq .
( [ eq_n_i,1 ] ) as the contribution due to the renormalization of the charging - energy gap @xmath150 , @xmath161\ ; \dfrac{d}{d \delta } \left\langle n\right\rangle ^{(\text{i},0 ) } \,,\ ] ] where the charging - energy gap is renormalized by @xmath162 $ ] .
the broadening of the spectral function of the island due to the tunnel coupling to the leads yields a contribution to the first - order correction of the island occupation number , which can be written as the sum of the contributions of the individual barriers , @xmath163 . the second term of eq .
( [ eq_n_i,1 ] ) can be identified as the contribution due to broadening , @xmath164 \,,\ ] ] where @xmath165 denotes the derivative with respect to the charging - energy gap @xmath150 .
the amplitude of the broadening depends on the occupation of the island : if the island is unoccupied , the prefactor is @xmath166 ; if the island is occupied by a single electron , the coefficient is @xmath167 . in order to compute the first - order correction to the pumped current , we need , besides @xmath64 , also @xmath168 , which can be computed by means of eq .
( [ adiabatic0 ] ) and reads @xmath169 \ ;
\partial_\delta\alpha(\delta ) \;\;\dfrac{d}{dt}\,\left [ f^+(\delta)\right]\\ & \quad \ ; -\;\dfrac{1}{\alpha(\delta)}\ ; \dfrac{d}{dt}\left ( \dfrac{\alpha_0}{2\pi}\;\ ; \re\left
[ \varrho(\delta)\right]\;\partial_\delta f^+(\delta)\ ; \right ) \;. \end{split}\ ] ] the expressions of the matrix elements of kernels @xmath62 and @xmath170 , which are needed to compute the probabilities eq . ( [ p_a,0 ] ) , can be found in appendix [ appendix_a ] .
finally , the pumped current in first order in @xmath61 can be written as @xmath171 the first term is related to the time variation of the correction of the average occupation number due to broadening of the resonance at @xmath150 induced by tunnel coupling the island to the left lead .
it is , therefore , associated with tunneling processes where electrons tunnel through the left barrier .
equation ( [ cur_a_1 ] ) shows that this first contribution to the current consists of the total time derivative of the correction of the average occupation number due to tunnel coupling to the left lead and thus the integral over a whole pumping period yields zero .
the second contribution to the pumped current in first order in @xmath61 is due to the renormalization of the charging - energy gap and has the same form as the zeroth - order contribution to the current , eq . ( [ cur_a_0 ] ) , with renormalized parameters . as the renormalized average occupation number of the island @xmath172 depends on the charging - energy gap as well as on the tunnel - coupling strengths @xmath173 , we obtain a nonvanishing pumped charge for any choice of pumping parameters . particularly intriguing is the case of pumping with the two barriers , since in this case the current @xmath174 vanishes and the first nonvanishing contribution to the current is due to the renormalization of the charging - energy gap , and therefore it is an effect purely due to quantum fluctuations and interaction .
now we calculate the pumped charge per cycle in the weak - pumping limit .
we write the pumping parameters as @xmath175 , where @xmath176 is the time average and @xmath177 the oscillating part .
the weak - pumping limit corresponds to computing the current in bilinear order in the time - varying parts of the parameters .
we start by considering one tunnel - coupling strength @xmath178 and the charging - energy gap @xmath150 as pumping parameters .
expanding the current @xmath179 up to bilinear order in the time variation of the pumping parameters and integrating over one cycle , we obtain for the pumped charge @xmath180\ ; \dfrac{d}{d\beta\bar{\delta } } \left\langle \bar{n}\right\rangle ^{(\text{i},0 ) } \right\rbrace \ , , \end{split}\ ] ] where @xmath181 is the area of the pumping cycle in parameter space , @xmath182 the instantaneous average occupation number , where the charging - energy gap has been replaced with its time average @xmath183 , and @xmath184 .
the first contribution to the pumped charge arises from the current in zeroth order in @xmath61 , the second from the first - order correction of the current due to the renormalization of the charging - energy gap as discussed in sec .
hence , the first term is the dominant one . in fig .
[ fig3 ] the pumped charge is plotted as a function of the time average of the charging - energy gap for different values of the high - energy cutoff .
the pumped charge is symmetric around zero and has its maximum there .
this peak is associated with transitions between charge states where the island is unoccupied or occupied by a single electron .
the pumped charge vanishes for charging energies much bigger than the temperature ( @xmath185 ) without changing its sign .
the overall sign of the pumped charge is determined by the traversal direction of the pumping cycle in parameter space . in fig .
[ fig3 ] we can clearly see that the maximum of the pumped charge decreases for increasing high - energy cutoff due to the correction introduced by the renormalization of the charging - energy gap . in units of @xmath186 as a function of the time average of the charging - energy gap for different values of @xmath156 .
the pumping parameters are @xmath187 and @xmath150 .
the tunnel - coupling strength is @xmath188 .
[ fig3 ] ] in units of @xmath189 as a function of the time average of the charging - energy gap for different values of @xmath156 .
the pumping parameters are @xmath187 and @xmath190 .
[ fig4 ] ] when we choose both tunnel - coupling strengths @xmath178 and @xmath191 as pumping parameters the contribution in zeroth order in @xmath61 vanishes , as already mentioned .
we therefore obtain for the pumped charge in the weak - pumping limit only the contribution due to the renormalization of the charging - energy gap , @xmath192\;\dfrac{d}{d\bar{\delta } } \left\langle \bar{n}\right\rangle ^{(\text{i},0)}\,,\ ] ] where @xmath193 is again the area of the pumping cycle in parameter space . the pumped charge as a function of the time - averaged charging - energy gap
is plotted in fig .
it is zero for @xmath194 and antisymmetric around zero . for charging energies much larger than the temperature ( @xmath185 ) the pumped charge vanishes .
increasing the high - energy cutoff @xmath156 , the amplitude of the pumped charge is increased .
finally , we emphasize that by adiabatic pumping with the two tunnel - coupling strengths we can explore the effects of renormalization of the charging - energy gap @xmath150 .
the pumping mechanism is peristaltic but it occurs via renormalization effects .
the sign change of the pumped charge around the resonance clearly distinguishes this mechanism from lowest - order pumping with one barrier and gate voltage .
we remark that such a sign change has been experimentally observed in charge pumping through carbon nanotubes with the help of surface acoustic waves.@xcite there , this feature has been understood as adiabatic pumping through an effective few - level double quantum dot with changing level positions .
in this section we consider pumping with the two charging - energy gaps in a system consisting of two tunnel - coupled metallic islands with no applied bias voltage .
this system corresponds to the single - electron pump of pothier _ et al .
_ @xcite the left and right islands are labeled by @xmath195 and @xmath196 , respectively .
notice that we will use the same labels for the the left and right leads . whenever confusion could arise
, we specify whether we refer to an island or a lead . the charging energy of the system reads @xmath197 where @xmath198 is the total charge on the left ( right ) island and @xmath199 the gate - induced charge on the left ( right ) island .
the prefactors @xmath200 , @xmath201 , and @xmath202 can be easily computed as a function of the junction and gate capacitances . at low temperature
, the hilbert space for the double - island system can be truncated to the three states @xmath203 , @xmath204 , and @xmath205 , corresponding to no excess charge in the system , one excess charge on the left island and none on the right one , and one excess charge on the right island and none on the left one , respectively .
the corresponding energies are @xmath206 , @xmath207 , and @xmath208 .
we define the two time - dependent charging - energy gaps @xmath209 and @xmath210 of the islands and the abbreviation @xmath211 .
the occupation probabilities for these charge states are @xmath136 , @xmath212 , and @xmath213 and are collected in the probability vector @xmath214 .
a sketch of the double - island system is shown in fig .
[ fig5 ] . in comparison to the single - island system ,
the tunneling events in the double - island system include additional tunneling between the two islands .
we define the direction of these tunneling lines to correspond to an electron tunneling from the right to the left island .
since no transport voltage is applied the pumped current vanishes in instantaneous order and we need to compute the first adiabatic correction . as in this system the current in zeroth order in the tunnel - coupling strength is the dominant term , we restrict ourselves to this contribution . to compute the pumped current in zeroth order in @xmath61 we need the probabilities @xmath215 , determined by eq .
( [ adiabatic-1 ] ) , together with eq .
( [ instantan0 ] ) . for this purpose ,
only the instantaneous kernels in first order in @xmath61 are necessary .
examples of diagrams contributing to the matrix elements of the kernel @xmath148 for the double - island system can be found in appendix [ appendix_b ] . following this procedure we find for the instantaneous occupation probabilities @xmath216 making use of the notation @xmath217 and @xmath218 with @xmath219 as well as @xmath220 for the denominator ,
the adiabatic probabilities can be written as @xmath221 by substituting these results in eq .
( [ cur_ad_low ] ) we obtain the adiabatically pumped current in the left lead in zeroth order in @xmath61 , @xmath222 where the functions @xmath223 and @xmath224 are defined by @xmath225 the current , eq . ( [ cur_exp ] ) , depends on the time derivative of the occupation probabilities of both islands .
the current in the right lead can be obtained by exchanging l and r in eq .
( [ cur_exp ] ) ; in doing so we have to replace also @xmath226 with @xmath227 . to compute the pumped charge per cycle we simply need to integrate the current @xmath174 , given in eq .
( [ cur_exp ] ) , over one period .
using simple mathematical manipulations , the pumped charge per period can be cast in the form @xmath228 where the emissivity for the left lead is given by @xmath229 following ref .
the pumped charge per period can be written as an integral over the area @xmath230 spanned by the cycle in parameter space and reads @xmath231 we now consider the weak - pumping regime and we write the pumping parameters , here the charging - energy gaps , as their mean values plus a small time - dependent variation around it , i.e. , @xmath232 .
equation ( [ brouwer ] ) in the weak - pumping limit yields for the zeroth - order contribution to the pumped charge @xmath233 \ , , \end{split}\ ] ] where @xmath234 is the area enclosed by the pumping cycle in parameter space . in the functions @xmath235 and the instantaneous occupation probabilities @xmath236 the charging - energy gaps have been replaced with their time averages @xmath237 . in units of @xmath238 as a function of the time averages of the pumping parameters @xmath239 and @xmath240 in units of the inverse temperature @xmath241 .
the white square marks the area in parameter space over which we integrate to obtain the pumped charge in the strong - pumping limit .
[ fig6 ] ] a density plot of the pumped charge as a function of the time averages of the pumping parameters @xmath239 and @xmath240 is shown in fig .
it consists in a peak with its maximum in the origin ( for symmetrically chosen tunnel couplings @xmath242 ) .
this can be seen more clearly in fig .
[ fig7 ] where we show a cross section of the peak along the line with @xmath243 ( solid line ) and along the line with @xmath244 ( dashed line ) .
notice that the peak is symmetric for the case of antisymmetrically chosen charging - energy gaps , while the symmetry is absent for symmetrically chosen charging - energy gaps . in units of @xmath238 for @xmath245 equal to @xmath246 ( dashed line ) and @xmath245 equal to @xmath247 ( solid line ) .
[ fig7 ] ] going around the origin of the plot in fig .
[ fig6 ] with constant radius @xmath248 , the pumped charge exhibits a maximum when the polar angle takes the values @xmath249 and @xmath250 , which correspond to @xmath244 . ]
so far , we discussed the pumped charge in the weak - pumping limit
. now , we want to describe the transition from weak to strong pumping .
therefore , we increase the variation of the pumping parameters @xmath251 and , hence , the area enclosed in the parameter space . the pumped charge over a large pumping cycle can be obtained by integrating the weak - pumped charge of eq .
( [ charge_exp ] ) over the area of the cycle . for the sake of definiteness
, we choose as a pumping cycle a square loop around the origin , as shown in fig .
the result for the charge as a function of the side of the loop , @xmath252 , obtained in this way is plotted in fig .
[ fig8 ] . in this figure
we see that we obtain charge @xmath75 if the size of the cycle is larger than @xmath253 .
this agrees with the results of the experiment of pothier _
et al._@xcite where charge quantization was measured for strong adiabatic pumping .
finally , we wish to stress that by means of our technique we are able to describe the transition from weak to strong pumping , recovering charge quantization for strong pumping cycles encircling the degeneracy point @xmath254 .
we developed a diagrammatic real - time approach to adiabatic pumping through a system of tunnel - coupled metallic islands , performing a systematic perturbative expansion in powers of the tunnel - coupling strengths .
this method allowed us to identify the different physical processes which contribute to the pumped charge .
we first applied our formalism to a single - island system consisting of one metallic island with coulomb interaction tunnel coupled to two noninteracting leads .
we computed the pumped charge up to first order in the tunnel - coupling strength , finding that the contribution in first order is due to the renormalization of the charging - energy gap . for the case of pumping with the two tunnel - coupling strengths ,
this term becomes the dominant one .
we emphasize that the single - island system works like a peristaltic pump if one tunnel - coupling strength and the charging - energy gap are the pumping parameters . for pumping with both tunnel - coupling strengths the peristalting mechanism occurs by means of charging - energy gap renormalization , therefore enabling experimental access to renormalization effects .
furthermore , we considered pumping with the charging - energy gaps in a system consisting of two tunnel - coupled metallic islands .
we calculated the pumped charge through this double - island system and studied the issue of pumped - charge quantization .
first , we investigated the pumped charge in the weak - pumping limit . for example , we found that in experiments a symmetric choice of the average of the charging - energy gaps should be preferred to maximize the pumped charge . by integrating the weak - pumped charge we obtained the pumped charge over a larger pumping cycle .
therefore , we were able to describe the transition from weak to strong pumping . in the strong - pumping limit
we found the charge to be quantized .
this is consistent with the results obtained by pothier _ et al._@xcite for strong pumping with the single - electron pump .
we acknowledge financial support from dfg via spp 1285 .
in this appendix , we give the results for the matrix elements @xmath255 for the single - island system .
these matrices are needed to compute the occupation probabilities of the metallic island .
we start with calculating the matrix elements @xmath256 of the instantaneous kernels in first order in @xmath61 .
the corresponding diagrams are shown in fig .
[ fig9 ] , where the additional dotted frequency lines must not be taken into account . as an example , we compute the two diagrams contributing to the matrix element @xmath257 , shown in fig .
[ fig9](b ) .
the sum of the two diagrams reads @xmath258 \nonumber \\ & = & 2 \pi \,\alpha^-(\delta)\;.\end{aligned}\ ] ] the instantaneous kernel in first order in @xmath61 reads @xmath259 for the derivative of the instantaneous kernel in first order in @xmath61 we get @xmath260 \left ( \begin{array}{cc } \quad1 & -1
\\ -1 & \quad1 \end{array } \right ) \;.\ ] ] to compute the adiabatic correction we need to introduce additional external frequency lines according to rules ( 7a ) and ( 8a ) , which are drawn as dotted lines in fig .
the evaluation of these diagrams leads us to the adiabatic correction to kernel in first order in @xmath61 , @xmath261 \right\rbrace \left ( \begin{array}{cc } \quad 1 & -1 \\ -1 & \quad 1 \end{array } \right ) \;.\ ] ] finally , we calculate the second - order contribution in @xmath61 to the matrix elements of the instantaneous kernel . as an example , we show all diagrams contributing to the matrix element @xmath262 in fig .
. we have to sum over all indices @xmath263 and to integrate over all frequencies @xmath264 . performing the limit @xmath265 and using cauchy s principal value ,
we get @xmath266 \\ & \quad & + 2 \pi \ ,
\alpha_0 \ ; \re\left [ \varrho(\delta)\right ] \;\
; \partial_\delta \;\alpha^-(\delta ) \,.\end{aligned}\ ] ] for the matrix elements of the instantaneous kernel in second order in @xmath61 we find @xmath267 \left ( \begin{array}{cc } \quad1 & -1 \\
-1 & \quad1 \end{array } \right ) \\ & \quad&+ 2\pi \ , \alpha_0 \ ;
\re\left [ \varrho(\delta)\right]\;\partial_\delta \left ( \begin{array}{cc } -\alpha^+(\delta ) & \quad\alpha^-(\delta ) \\
\quad\alpha^+(\delta ) & -\alpha^-(\delta ) \end{array } \right ) \;.\end{aligned}\ ] ]
in this appendix we give the results for the matrix elements @xmath268 of the instantaneous kernel in first order in @xmath61 for the double - island system .
examples of the corresponding diagrams are shown in fig .
[ fig11 ] .
their evaluation is similar to the one for the instantaneous matrix elements in first order in @xmath61 for the single - island system , see eq .
( [ w_i,1 ] ) , @xmath269\\ & = & 2\pi\alpha_\text{m}^-(\delta_\text{m } ) \;. \ ] ] j. m. martinis , m. nahum , and h. d. jensen , phys .
lett . * 72 * , 904 ( 1994 ) ; m. w. keller , j. m. martinis , n. m. zimmerman , and a. h. steinbach , appl .
. lett . * 69 * , 1804 ( 1996 ) ; r. l. kautz , m. w. keller , and j. m. martinis , phys .
b * 60 * , 8199 ( 1999 ) .
n. e. fletcher , j. ebbecke , t. j. b. m. janssen , f. j. ahlers , m. pepper , h. e. beere , and d. a. ritchie , phys .
b * 68 * , 245310 ( 2003 ) ; j. ebbecke , n. e. fletcher , t. j. b. m. janssen , f. j. ahlers , m. pepper , h. e. beere , and d. a. ritchie , appl .
. lett . * 84 * , 4319 ( 2004 ) . | we present a real - time diagrammatic formalism to study adiabatic pumping in chains of tunnel - coupled metallic islands .
this approach is based on an expansion to linear order in the frequency of the time - dependent parameters and on a systematic perturbation expansion in the tunnel - coupling strength .
we apply our formalism to single - island and double - island systems . in the single - island setup , we find that the first - order contribution in the tunnel - coupling strength is purely due to the renormalization of the charging - energy gap . in the double - island system
, we investigate the transition between weak and strong pumping . |
knowlege of the three dimensional distribution of matter in the universe at @xmath8 is limited .
we do not know if the matter distribution over scales @xmath9 is homogeneous or isotropic ( background radiation , self consistency of the standard model based on the assumption of a _
stable _ uniform density , etc .
do not provide direct evidence about the distribution of visible matter in the present epoch ) .
for @xmath10 the distribution of visible matter is clearly inhomogeneous , with large voids and clustering , and various analyses have produced results that are equivalent to claiming scale invariance for the correlation integral , that @xmath11 , with one school ( @xcite , @xcite ) reporting that @xmath12 for @xmath13 , whereas the other ( @xcite , @xcite ) reports that @xmath14 for @xmath15 .
the correlation integral and scaling exponent @xmath2 are defined in part [ sect : correlation_integral ] .
roughly speaking , one can think of the scaling exponent @xmath2 as a correlation dimension , but not as hausdorff or box counting dimension .
we will explain why the reported claims of scale invariance may be spurious unless the matter distribution actually is , to a very good approximation , monofractal .
the first aim of this paper is to explain the need for a more careful analysis of observational data than has heretofore been performed .
the second aim is to provide that analysis ( part [ sect : data ] ) .
the third is to explain and eliminate the confusion over the term multifractal .
finally we will explain why a `` nonanalytic '' density does not rule out the use of differential equations . in any attempt to extract scaling exponents from log log plots of correlation or generating functions
a conservative criterion in both critical phenomena and dynamical systems theory is that linearity should be exhibited over at least three decades , which would require data out to at least @xmath16 in astronomy .
the reason for this is that there are too many different functions @xmath17 that do nt scale with @xmath18 , @xmath19 , but @xmath20 vs. @xmath5 may nevertheless _ appear _ to have a constant slope over a short enough range of @xmath18 .
the function @xmath21 provides a relevant example : this function is not scale invariant because of _ two _ exponents @xmath22 and @xmath23 , but it is easy to exhibit the _ illusion _ of scale invariance by plotting @xmath20 vs. @xmath5 over only two decades ( see figure [ fig : logplot ] ) .
if one questions the controversial claim of scale invariance over two decades up to @xmath24 , then must also one question the _ widely accepted _ claim of scale invariance , _ also _ over only two decades , up to @xmath25 . to make our viewpoint clear to the reader we recapitulate briefly the controversy over scaling in this field .
the earliest attempts to extract @xmath2 were based upon the expectation of homogeneity at larger scales , with inhomogeneities confined to @xmath26 where @xmath27 is a correlation length . by confusing an amplitude with a correlation ( i.e. characteristic )
length @xcite @xmath28 was obtained , which is inconsistent with the observed clustering and voids out to @xmath29 , and beyond .
coleman and pietronero @xcite and martinez et al .
@xcite have argued instead that the known data are scale invariant , and correspondingly that there is no correlation length .
early attempts to dismiss this point of view as the result of `` large deviations '' due to `` unfair samples '' failed when it became clear that the apparent scale invariance is the rule and not the exception . as a @xmath30 order description
this seems to be correct , but we show in part [ sect : data ] that a refinement of the method of @xcite ( see part [ sect : correlation_integral ] ) leads to the conclusion that the data are inadequate to draw a conclusion for or against multifractal scaling , although it is clear that simple fractal scaling , with a single exponent @xmath2 as claimed by the pietronero school , is not indicated by the data .
fractal or not , it is clear that there is , as yet , no evidence from galaxy redshift surveys of any crossover to homogeneity .
this is a good question because , as one expert in statistical mechanics and nonlinear dynamics put it , if you ca nt calculate anything then you can still talk about scaling .
furthermore , most phenomena in nature do not scale , or at least have not been shown to scale ( the polls are not yet closed on the question of multifractal scaling in the inertial ( @xcite ; @xcite ) and dissipation ( @xcite & @xcite ) ranges of fluid turbulence , and the early returns are not entirely convincing ) .
in fact , there are only a few known reasons why anything should scale , aside from dimensional analysis ( reynolds number scaling , which works pretty well in fluid mechanics so long as one sticks to qualitative considerations and does not look hard at the numbers ) .
let us enumerate the ( known ) ways .
first , there is scaling of all sorts of correlation functions and other thermodynamic quantities if you are close enough ( within @xmath31 of @xmath32 , at least ) to a second order phase transition .
the problem with this is first that the universe is not in thermodynamic equilibrium ( tde ) : nonuniformities like spiral galaxies and dna are not generated systematically at small scales in a system in tde .
second , there is no reason why the universe should be tuned precisely to a critical point ( where @xmath33 ) .
critical phenomena are popular because the scaling indices are universal , depending only on symmetry and dimension , which allows a theorist to forget about worse details than those that plague experimentalists and calculate the scaling indices of real ferromagnets , for example , by using ising models or @xmath34 models ( @xcite )
. we can also consider dynamical critical phenomena , where universality classes can still be defined for scaling exponents based only on symmetry and dimension , but that does nt help : we are still restricted to systems that have only very small deviations from thermal equilibrium .
large excursions from tde are nt allowed at small scales in these systems .
galaxies have been modelled on the basis of critical phenomena far from tde by using a particular cellular automaton ( @xcite ) near the percolation threshhold .
nice patterns can be produced that look like spiral galaxies ( @xcite ) , but who tunes the galactic system to stay near the percolation threshold ?
this model does nt yet have enough physics in it to be falsifiable .
for those who believe that scaling is ubiquitous in nature , but do nt expect that mother nature tunes phenomena to a critical point , there is soc ( self organized criticality ) .
the idea of soc ( @xcite ) is based on driven dissipative dynamical systems far from equilibrium that quite naturally lie at a borderline of chaos , for a large range of parameter values , and therefore require no parameter tuning .
criticality ( a borderline of chaos ) means that all liapunov exponents must vanish ( some positive exponents are usually allowed in the literature because models where all of them vanish for a _
finite _ range of control parameter values ( @xcite ) are unknown ) . scaling exponents in soc
are argued to be universal because they are expected to be parameter independent .
the main problem with soc is that no one has yet found an example of a dynamical system where the idea has been realized , criticality without parameter tuning , criticality that persists while parameters are varied .
the soc idea is usually illustrated by a sandpile model that has no tunable parameters because the parameters in the model were _ implicitly _ tuned to criticality and then forgotten .
soc purports to provide a universal explanation of fractal scaling indices which , if we follow the scaling enthusiasts , should be ubiquitous in nature , but _ fractal and multifractal exponents are not universal and ca nt be used to define universality classes_. the different models used to try to describe the ( still inadequate ) data on the inertial and dissipation ranges of fluid turbulence provide examples of this ( @xcite & @xcite ) .
sandpile models of soc reproduce certain qualitative features of block spring models of earthquakes , but the block spring models do not produce the parameter independent criticality demanded by soc ( @xcite ) .
soc has not been defined unambiguously because universality classes for soc have not been defined . in the absence of universality classes
one can not claim that a simple automaton like the sandpile model represents a complicated or complex dynamical system that occurs in nature ( @xcite ) .
then , there are the fractals that are generated in the phase spaces of critical and _ chaotic driven dissipative _ dynamical systems far from thermal equilibrium . the scaling indices that describe the fractals that occur in chaos are _ not universal_. that s ok , because while the fractal dimensions are parameter - dependent the fractals persist ( in distorted form ) as the parameters are varied over relatively wide ranges ( this is usually what happens in soc models too ) . given the coarsegrained fractal support generated by a driven dissipative dynamical system ( `` support '' of a distribution
is defined in part [ sect : empirical_dist ] ) , nonuniform distributions on that support are typically multifractal , meaning that the coarsegrained density becomes more and more spiky ( and perhaps also intermittent with voids ) as the distribution is resolved at finer and finer scales of observation .
the corresponding densities would be nearly everywhere nondifferentiable if the mathematicians fiction of an infinite precision limit were not ruled out empirically .
conservative dynamical systems ( like gravity without dissipation / driving ) can not generate fractals in phase space : the support of any distribution generated by a conservative dynamical system is space filling ( liouville s theorem ) . space
filling means that the support has the dimension of the phase space , whereas a fractal support has a nonintegral dimension less than that of the phase space .
however , a conservative dynamical system far from thermal equilibrium can also generate multifractal coarsegrained distributions on the space filling support . _
a noninteger correlation integral scaling exponent @xmath2 does not suggest that the galaxy distribution has a fractal support : nonintegral @xmath2 is consistent with multifractal distributions on space filling supports . _
the problem with all of this is that it does not explain anything , as yet : the fractals and multifractals discussed above all occur in the very high dimensional phase space of all of the galaxies ( each galaxy is treated as a point particle here and below ) , and we do not know how those distributions would look when projected onto the three dimensional space of observation in astronomy . in other words , we do nt have a quantitative explanation for where fractal ( including multifractal ) galaxy clusters should come from . in practice , it makes more sense to consider hydrodynamic models of galaxy formation and clustering .
hydrodynamics demands a coarsegrained description of the density .
coarsegrained descriptions are precisely what are provided by the multifractal formalism ( see vergassola et al .
@xcite and references therein for a hydrodynamic approach to clustering and voids ) .
unable at this time to contribute to the dynamical theory of galaxy formation , let us forget temporarily that no theorist can yet convincingly explain the origin of fractal galaxy clustering , if it exists , and turn instead to the question how astronomical data should be analyzed in order to decide the much easier question _ whether _ fractal clustering is indicated by the observational data .
in what follows we consider only finitely many data points @xmath35 in some space , the observational data . to present the ideas in the clearest possible way we assume that the space is one dimensional ( excepting section [ sect : correlation_integral ] on the correlation integral , where the dimension of space is irrelevant ) .
the ideas of sections [ sect : clustering_voids][sect : optimal_partitioning ] are admittedly heuristic and can only be made rigorous by the use of generating partitions found in the phase space of certain nonintegrable dynamical systems ( @xcite ) .
in particular the heuristic description is limited to one dimension ( generating partitions are not so limited ) , but our one dimensional treatment is adequate to the purpose of resolving the prevailing confusion over `` multifractals '' and `` nonanalytic '' densities .
we assume that the @xmath35 data points in our one dimensional space are confined to the unit interval , because any finite interval can be converted into the unit interval by rescaling . with lengths denoted by @xmath36 ,
@xmath37 in all that follows ( planar and three dimensional cosmological data are also assumed to be rescaled so that @xmath38 in the discussion of part [ sect : correlation_integral ] ) .
coarsegraining of the data set requires only that we cover the @xmath35 points by @xmath39 nonoverlapping intervals of size @xmath40 .
clearly , we can take @xmath41 intervals of size @xmath42 , @xmath43 intervals of size @xmath44 , and so on . all that is required to avoid overlap is @xmath45 ; coarsegraining per se is not unique .
the only other requirement , so far , is that we must choose the intervals so that @xmath46 where @xmath47 is the smallest distance between two points in the sample ( a single point can not be coarsegrained , so that an interval containing only one point is meaningless ) .
we shall see that the desire of the theorist to approximate @xmath48 by @xmath49 , where @xmath49 goes to zero , has led to serious errors in data analysis . in our analysis
@xmath48 is always finite because it contains at least two _ real _ data points . in any efficient
coarsegraining the intervals that cover the points must be _ nonoverlapping _ , coarsegrainings that are space - filling ( meaning that @xmath50 ) satisfy this criterion but do not separate voids from clusters . for data with voids
we should construct a coarsegraining that is not space - filling , one where @xmath51 , in order to cover all of the clusters while excluding the largest voids .
such a coarsegraining is more efficient than an arbitrary one . with the desire for efficiency in mind
the idea is first to remove all of the largest voids .
then we choose the interval size @xmath52 , where the intervals @xmath53 , all of roughly the same characteristic size , represent the @xmath54 largest voids in the sample , and @xmath55 .
the @xmath56 first generation intervals required to cover the @xmath35 data points are not space - filling because @xmath57 , by construction . if , beneath the @xmath56 intervals now covering the data , there are still voids and clustering then we can continue systematically by removing all voids of next largest characteristic size : in the second generation of coarsegraining simply choose @xmath58 intervals of size @xmath59 where the intervals @xmath60 represent the sizes of @xmath61 largest voids covered by the @xmath56 intervals @xmath62 . this iterative procedure may be continued so long as we can distinguish clusters from voids .
there are two ways that it can terminate .
either we reach a scale @xmath63 where the points are relatively evenly spaced over those intervals ( so that there is no longer a distinction between clusters and voids ) , or else clustering continues all the way down to the finite limit @xmath64 , where @xmath48 roughly characterizes an interparticle spacing and will be defined more precisely in part [ sect : optimal_partitioning ] .
the procedure outlined above describes the idea of a more efficient partitioning than a space filling one , a more efficient coarsegraining of the data set because the largest voids are systematically excluded , generation by generation in n. we do not have only one partition but a hierarchy @xmath65 of partitions , each with interval sizes @xmath40 . a pencil and paper sketch of about sixteen points with two big voids of roughly the same size , but with three or more very nonuniform first generation clusters , e.g. figure [ fig : nonuniform ]
, shows that the method described above will produce an efficient partition only if the clustering is relatively uniform , only if all clusters in a each generation are all of about the same size .
when this is not the case , when the clustering is very nonuniform ( as in figure [ fig : nonuniform ] ) , then the procedure outlined above will not produce an efficient partition and may even fail to cover the set . in that case
we could try to repair the misfit by taking @xmath40 to be the largest of the intervals in the @xmath66th generation , but this may produce overlapping intervals , @xmath67 , which is intolerable .
we explain in the next section how``convergence problems '' arise in the analysis of empirical data whenever arbitrary ( rather than efficient ) partitions are used .
the motivation for the expectation that an _ optimal _ partition may exist , at least in _ some _ cases , is as follows : certain nonintegrable dynamical systems coarsegrain phase space uniquely ( @xcite , @xcite ) . in those systems the optimal partition is generated by the dynamics and is called the `` generating partition '' . as a simple example , the invariant set of the ternary tent map is the middle thirds cantor set ( @xcite ) .
the generating partition of the ternary tent map ( obtained by @xmath66 backward iterations of the unit interval using that map ) is given by @xmath41 intervals @xmath68 , for @xmath69 .
it is impossible to construct a more efficient coarsegraining of the middle thirds cantor set than this one .
the voids are the excluded open intervals @xmath70 , and so on ( initial conditions of the ternary tent map that lie in the voids iterate to minus infinity ) .
we restrict our considerations in this section to relatively uniform clustering ( the clusters in a generation @xmath66 of coarsegraining are all of about the same size @xmath40 , and the @xmath66th generation voids are also all of about the same size @xmath71 ) .
invariant quantities ( scaling exponents ) can only be constructed , if at all , from @xmath39 and @xmath40 as the generation @xmath66 of coarsegraining is increased , as we look at the data set with finer and finer , but _ never _ with _ pointwise _ resolution ( the smallest interval size @xmath48 always contains at least _ two _ data points ) . in what follows it
is conceptually useful to think of the hierarchy of intervals for @xmath65 as sitting on the branches of a tree of some order @xmath72 .
there are @xmath56 branches in generation 1 , @xmath58 in generation 2 , and so on , and the tree need not be complete .
scale invariance will be seen below to require that @xmath39 increases exponentially , @xmath73 , as @xmath40 decreases exponentially , @xmath74 .
complete trees have @xmath75 branches in generation @xmath66 with @xmath76 an integer , while incomplete ones have noninteger @xmath72 , in which case the order of the tree is the next integer larger than @xmath72 .
the middle thirds cantor set , an idealized model of uniform clustering exhibiting big voids on all scales , defines a complete binary tree ( @xmath77 ) .
we are only treating coarsegrained versions of @xmath35 points , so the simplest kind of scale invariance is geometric ( and so is the more complicated kind , as we shall see in parts [ sect : multi_optimal ] and [ sect : multi_empirical ] below ) . a fine - grained picture of the subset of any branch for @xmath78 looks , upon magnification , like the entire tree .
this kind of scale invariance ( @xcite ) is expressed by the exponent @xmath79 in @xmath80 in other words , @xmath81 is an exponent that reflects self similarity of a hierarchy of relatively _ uniform _ clusters ( clusters within clusters within ) . in order to check whether ( [ eq:1 ] ) holds approximately for a given ( very uniform )
data set it is very useful first to find as efficient a partition as possible .
the practical difference ( emphasizing the famous `` convergence '' problem ) between the use of efficient and inefficient partitions is best illustrated by an idealized example .
as an example of an optimal partition consider a very artificial data set constructed as follows : arbitrarily choose a finite number @xmath35 of points generated by ternary expansions of the form @xmath82 with @xmath83 or 2 ( @xmath84 is excluded ) .
these numbers belong to the middle
thirds cantor set ( all ternary numbers of this kind define the middle
thirds cantor set @xcite ) .
terminating strings @xmath85 define @xmath41 intervals @xmath68 given by @xmath86 $ ] and @xmath87 $ ] in generation one ( @xmath88 ) , @xmath89 $ ] , @xmath90 $ ] , @xmath91 $ ] , and @xmath92 $ ] for @xmath93 , and so on .
rational numbers of the form @xmath94 ( periodic expansions like @xmath95 ) and irrational ones @xmath96 ( where the digit string is nonperiodic ) are also covered by all of the @xmath39 intervals as well , so that the covering provided by those @xmath39 intervals @xmath68 is optimal , generation by generation .
the scaling law ( [ eq:1 ] ) yields @xmath97 .
this example illustrates a `` monofractal '' because the optimal covering is uniform ( all @xmath39 of the optimal intervals in one generation have the same size @xmath40 ) .
in contrast , we can try to estimate @xmath79 by using the uniform nonoptimal covering @xmath42 , a space - filling partition given by @xmath98 $ ] and @xmath99 $ ] for @xmath88 , @xmath100 , [ 1/4,1/2 ] , [ 1/2,3/4]$ ] , and @xmath101 $ ] for @xmath102 , etc .
, that ignores the voids alltogether . here
, with @xmath35 points in the data set @xmath103 must be very large before we can expect to observe scaling with an exponent close to @xmath104 : for @xmath105 points , e.g. , and using @xmath106 then from @xmath107 and 2 one gets @xmath108 , @xmath109 yields @xmath110 , and further attempts to extract @xmath79 are impossible unless the number @xmath35 of data points is increased .
this illustrates why , in practice only relatively efficient partitions are of interest .
we return next to the search for the optimal partition of a typically nonuniform empirical data set of @xmath35 points . for a typical empirical data set of @xmath35
points the most efficient partition that one can construct will rarely be uniform .
let @xmath40 denote the size of the largest cluster in generation @xmath66 after deleting the largest voids , as in part [ sect : clustering_voids ] .
this may yield an overlapping partition of @xmath39 uniform intervals @xmath40 where @xmath67 , but we can immediately improve upon that coarsegraining : in any generation @xmath66 the @xmath39 intervals so constructed will ( excepting the largest interval , which determines @xmath40 ) not end on data points but will extend beyond them . to make the covering efficient
simply shrink each interval until it ends on the nearest two points of the cluster that it was intended to cover in the first place .
the result is that the number of intervals is exactly the same as before , but we now have @xmath39 nonuniform intervals obeying both @xmath111 and @xmath112 . in other words , we have minimized the sum @xmath113 while holding @xmath39 fixed .
it is hard to see how a more efficient covering can be found , so for the purpose of this paper we call the hierarchy of intervals , so - constructed , the optimal partition ( see figure [ fig : nonuniform2 ] ) .
our definition of optimal partition is a finite precision realization of an `` optimal '' @xmath114covering ( a @xmath114covering approximates the `` infimum '' ) , as is used in the mathematicians definition of `` hausdorff measure '' ( @xcite ) .
the method in the cosmology literature that may come nearest to ours , in spirit , is the minimal spanning tree method ( @xcite ) . in dynamical systems
theory the optimal partition is called the generating partition and provides the geometric or finite precision , definition of a fractal . here s an idealized example of a dynamical system with a nonuniform optimal partition ( @xcite ) . the generating partition of the asymmetric tent map with slope magnitudes @xmath22 and @xmath23 is given by @xmath115 intervals @xmath116 where @xmath117 , and describes the two scale cantor set ( there are two first generation scales @xmath118 and @xmath119 ) . the idealized data set consists of interval end points and also of limits of infinite sequences of interval end points ( the latter corresponding to infinitely many backward iterations of the unit interval by the asymmetric tent map ) .
this description of the asymmetric tent map is correct if @xmath120 ( @xmath121 means space filling , while @xmath122 produces voids and clustering , representing an idealized nonuniform `` cantor dust '' ( @xcite ) .
one object of @xcite was to extract the generating partition of the henon map .
the generalization of the scaling law ( [ eq:1 ] ) to our hierarchy of nonuniform optimal partitions @xmath123 is @xmath124 where the scaling index @xmath125 is called the hausdorff dimension ( @xcite ) . whether a data set has a hausdorff dimension
can only be answered empirically , by constructing the optimal partition and checking to see whether ( [ eq:1b ] ) holds over many different generations n with the same exponent @xmath125 .
the hausdorff dimension of the two scale cantor set is given exactly in the first generation by @xmath126 .
in the empirical case , in contrast , @xmath103 is usually required in order that we have enough data points to see scaling via either ( [ eq:1 ] ) or ( [ eq:1b ] ) which , as mandelbrot @xcite pointed out early in his book , is usually confined to an intermediate range of interval sizes @xmath40 where @xmath127 .
here , @xmath128 is on the order of the size of the sample , @xmath40 is the largest of the @xmath39 @xmath66th generation intervals @xmath129 , and @xmath48 is in our case the smallest distance between two points in the sample . in general , we should not expect to observe scaling unless the largest intervals are much smaller than the size of the system , and unless all of the smallest ones contain more than a single data point .
however , the larger limit may not be applicable to present astronomical data because the systems of galactic clusters and voids are presumably much larger than any available sample size .
also , there is nothing to prevent our checking for scaling all the way down to @xmath48 . again ,
if an inefficient partition is chosen then scaling may not be observed even if the data set is fractal , because the number @xmath66 of generations needed for `` convergence '' to a scaling exponent @xmath79 or @xmath125 may exceed the number @xmath35 of data points in the sample . _ all definitions and approximations based upon the @xmath130 limit are systematically avoided because , as we explain below , they lead to formulae that generally do not apply to finite data sets . _ suppose that we have found the optimal nonuniform partition for a data set .
if we replace the uneven intervals in a nonuniform partition @xmath123 by the largest scale of each generation ( simply call it @xmath40 ) , then we obtain a less optimal uniform covering defined by @xmath131 because @xmath39 is the same as in ( [ eq:1b ] ) , and because @xmath132 , it follows that @xmath133 where @xmath79 is called the box counting dimension .
this procedure defines the most efficient _ uniform _ partitioning of the data set if the clustering is relatively uniform . in other words , whether or not @xmath79 provides a good estimate of @xmath125 for low values of @xmath134 depends on whether or not the uniform partition closely approximates the nonuniform optimal one .
this replacement amounts to the pointwise approximation of a nonuniform fractal data set by a monofractal .
the resulting idealized data set can be thought of as consisting of the end points of the uniform intervals @xmath40 .
for the two - scale cantor set with @xmath135 the box - counting dimension is @xmath136 . in the two - scale cantor
set @xmath137 and @xmath138 yield @xmath139 only if @xmath22 and @xmath23 are approximately equal .
how many generations are necessary in order to convince hardened sceptics that scaling has been observed ? in both critical phenomena and dynamical systems theory the rule of thumb is three decades on a log log plot , requiring astronomical data , e.g. , from .1 to 100 , or from 1 to 1000 depending on the smallest distance reported in a given catalog of galaxies . with @xmath140 , @xmath141
, we would need @xmath142 generations .
we call this criterion `` the geilo criterion '' because it was suggested at a geilo nato - asi .
the geilo criterion is not a matter of taste : it is advocated in order to deflect erroneous reports of scaling like that indicated in figure [ fig : logplot ] over only two decades , @xmath143 .
thinking of the optimal partition of a fractal as organized onto a tree of some order , if we write @xmath144 then the order of the tree is the next integer greater than or equal to @xmath72 .
if @xmath72 is nonintegral then the tree is incomplete . if @xmath145 then @xmath146 .
with @xmath141 the @xmath147model of fluid turbulence lies on an incomplete tree that is at least octal ( @xcite ) .
see @xcite for a discussion of the @xmath147model in cosmology .
a data set that obeys scaling ( [ eq:1b ] ) with a nonuniform optimal partition may be `` multifractal '' .
multifractal is always defined here to mean a spectrum of fractal dimensions ( @xcite ) .
each dimension in a multifractal spectrum describes the scaling of a subset of the optimal partition ( a nonuniform fractal is decomposed disjointly into a union of other fractals ) .
given the optimal partition a multifractal spectrum @xmath148 may be defined by parameterizing the hierarchy of coarsegrained intervals ( the parameter here is @xmath149 ) so that the partition is organized into ( nonoverlapping but ) interwoven sub partitions ( @xcite ) .
each fractal dimension in the multifractal spectrum is the hausdorff dimension of one subset of the partition ( a multifractal is always the union of a complete set of nonoverlapping but interwoven fractals labeled systematically by some index ) . to obtain scale invariance
the interval sizes @xmath150 must contract exponentially as @xmath66 increases .
both the @xmath66th generation intervals and their contraction rates are generally nonuniform : as an oversimplified example let @xmath151 denote the @xmath152th of @xmath39 intervals in generation @xmath66 .
this defines a simple nonuniform cantor set based on @xmath56 different first generation scales @xmath153 if @xmath154 ( in a chaotic dynamical system the contraction rate @xmath155 describes the intervals of the generating partition only asymptotically and approximately for @xmath156 , representing the inverse butterfly effect for integration backward in time along unstable manifolds ( @xcite ) ) .
suppose that there are @xmath157 intervals with the same contraction exponent @xmath149 ( in a dynamical system @xmath158 is the liapunov exponent for forward evolution in time , starting from a specific class of initial conditions , namely , all initial conditions that yield the same liapunov exponent @xmath158 )
. then @xmath159 ( where @xmath160 ) defines the hausdorff dimension @xmath148 of the subset of the partition labeled by @xmath149
. we can generalize ( [ eq:1b ] ) by writing down the generating function ( @xcite ) @xmath161 with @xmath162 for large enough @xmath66 , and where @xmath163 defines the hausdorff dimension of the entire fractal ( by ( [ eq:1b ] ) ) .
note that @xmath164 because @xmath165 .
this simple fractal , seen as multifractal , is the union of a complete set of interwoven , nonoverlapping monofractals ( neighboring branches on each generation of the tree generally are labeled by different indices @xmath149 ) . in the ovesimplified model above we have @xmath166 . as an example ( @xcite ) , consider the two scale cantor set ( with @xmath167 first generation intervals @xmath168 ) . in generation
@xmath66 the optimal covering is given by the @xmath169 intervals with sizes @xmath170 , of which have the same size @xmath171 . here ,
@xmath172 with @xmath173 . in each generation @xmath66
there are @xmath174 points in the multifractal spectrum , not more , and the number of points grows only linearly with @xmath66 ( still , this eliminates bi - fractals , tri - fractals , , and @xmath35fractals from our definition of `` multifractal '' ) . using stirling s approximation , so that , with @xmath175 , @xmath176 where @xmath177 is the boltzmann entropy ( divided by @xmath66 ) of all intervals with the same contraction exponent @xmath149 , we have @xmath178 , which shows the connection of fractal dimension to entropy ( @xmath179 and @xmath180 for the middle thirds cantor set ) . since @xmath181 the tree is generally binary but incomplete for any monofractal subset labeled by the contraction index @xmath149 in the two - scale cantor set .
note that `` multifractal '' is consistent with @xmath182 : the support may be space - filling , while subsets of the support are fractal ( @xmath183 ) . in the two - scale cantor
set @xmath182 corresponds to the space - filling condition @xmath184 .
ideas based on the generating function ( [ eq:2 ] ) have been used to analyze experiments on the transition to chaos in fluid dynamics ( @xcite ) .
the notion that multifractal scaling can be verified for small data sets ( @xcite ) is a misconception : fewer data points @xmath35 , with less precision , _ can not _ be required for the determination of a _ spectrum _ of fractal dimensions than are required for determining _ one _ dimension , say @xmath125 .
the error in the claim follows from confusion and errors made in defining `` multifractal '' ( see parts [ sect : correlation_integral ] , [ sect : generating_functions ] and [ sect : lognormal ] below ) .
the term multifractal has occasionally been incorrectly defined in the cosmology literature where at least three entirely different generating functions are confused together ( see parts [ sect : correlation_integral ] and [ sect : generating_functions ] ) as if their corresponding scaling exponents ( whenever scaling exists ) would define universality classes independently of probability distributions and partitionings used to define those exponents . in @xcite
a far to restrictive idea of multifractal is presented in part 6 .
multifractal spectra were introduced in analogy with critical exponents ( @xcite ) , but the expectation of universality ( @xcite ) was not realized ( a restricted and still unproven universality is merely postulated for vortex cascades in fluid turbulence ( @xcite ) ) . in the theory of chaotic dynamical systems two examples of topologic invariants are the tree order @xmath72 and its degree of incompleteness ( @xcite , @xcite ) .
we always restrict our formulation of the requirement for fractal and multifractal scaling to finite @xmath40 and to finitely many data points @xmath35 , completely avoiding mathematically idealized results that would require for their applicability the empirically and computationally unattainable limit where @xmath185 .
we shall see in part [ sect : multi_empirical ] that this rules out largest term approximations , whose validity would require values of @xmath40 that are too much small ( @xmath186 , with the range of @xmath40 extending over at least three decades ) to be consistent with the analysis of galaxy distributions . for an observational data set there is only one pointwise probability distribution , the empirical distribution @xmath187 defined by the @xmath35 data points : @xmath187 is simply the fraction of points lying to the left of ( and including ) @xmath188 , so that @xmath189 and @xmath190 , by construction .
the distribution is constant on the voids and increases discontinuously at each data point , so that the plot of @xmath187 is a staircase of @xmath191 steps of finite width .
the data staircase has the singular pointwise density @xmath192 given by @xmath193 of course , ( [ eq:3 ] ) is only a theorist s fiction : it neglects the error bars in the locations of positions . in reality
each position is specified empirically by a finite _ interval _ whose width is the uncertainty in location .
we assume here that these uncertainties are very small relative to the smallest separation @xmath48 between data points . otherwise coarsegraining and fractal / multifractal analysis are impossible .
the density ( [ eq:3 ] ) will be used to correct a more serious theoretical error in part [ sect : generating_functions ] below .
all that we need in what follows is the staircase @xmath187 along with an empirical technique for characterizing voids and clusters .
we emphasize that attempts to `` smooth '' the staircase ( via `` splines '' , e , g ) will discard important information about clustering . _ no pointwise probability distribution other than the staircase @xmath187
_ , @xmath194 _ is relevant for empirical data analysis . _
an arbitrary distribution @xmath187 of @xmath35 data points is generally not approximable in either the continuum ( infinite precision ) or hydrodynamic ( coarsegrained ) limit by a differentiable distribution .
we shall find next that the coarsegrained versions of the empirical distribution @xmath187 are typically too spiky ( too intermittent ) to be approximable by an everywhere differentiable distribution ( reminding us more of `` noise '' than of analyticity ) , even if @xmath125 is an integer ( even if the optimal partition is space filling ) .
the spikiness / intermittence represent clustering and voids in the sample .
hydrodynamics demands a coarsegrained description of a pointwise distribution , and we will discuss in part [ sect : homogeneity ] how the required densities can be defined . on an optimal or at least efficient partition a coarsegrained probability @xmath195 is defined by the difference @xmath196 over the @xmath152th closed interval of size @xmath40 , and is just the fraction @xmath197 of the total number of data points @xmath198 in the @xmath152th interval including the end points .
while each empirical distribution @xmath187 is a staircase of finitely - many steps , each coarsegrained distribution @xmath199 is a histogram on a finite support .
the optimal partition optimally defines the `` support '' of the hierarchy of coarsegrained empirical distributions @xmath199 : for the optimal partition , each interval end point coincides with a point where the staircase function @xmath187 increases discontinuously . whether an empirically - constructed optimal partition is the generating partition of a deterministic dynamical system is a separate question ( the main question , but very hard to answer ( @xcite , @xcite ) ) .
the coarsegrained probabilities @xmath199 can be used to perform averages that ignore the details of the dynamics at all scales smaller than @xmath40 ( coarsegraining the smaller scales is required in order to define hydrodynamics ) .
the only limitation , so far , is that @xmath200 .
bear in mind however , that before reaching the smallest scale @xmath48 , as @xmath66 is increased , we may not be able to distinguish clustering from voids without ambiguity .
even if clustering and voids are present at all scales they may not be scale invariant .
the construction of efficient and even optimal partitions does not presume scale invariance , rather , the converse is true , especially in practice . we have used the frequency definition of probabilities because it arises naturally in both empirical data analysis and computer simulations . using our simple example above ,
however , we can offer as an idealized staircase distribution the cantor function ( @xcite , @xcite ) @xmath201 where , because @xmath202 is a ternary number with @xmath83 or 2 , @xmath187 is a binary number ( because @xmath203 or 1 ) .
this staircase describes a mathematician s idealization of empirical data , namely , one ( of infinitely many ) distribution that can be constructed by using points in the middle
thirds cantor set : @xmath204
, @xmath205 , @xmath187 is constant on the open voids and increases discontinuously at the end point of any closed interval @xmath68 of the optimal covering , where the change in @xmath187 is @xmath206 . therefore , the cantor function defines a hierarchy of _ uniform _ coarsegrained distributions @xmath207 , generation by generation in @xmath66 , on the fractal support @xmath68 .
the plot of the cantor function , the mathematicians idealization of an empirical distribution , is called a _
devil s _ staircase because it has @xmath208 steps . a good way to look for voids is simply to plot the empirical staircase : @xmath187 is constant on the voids , where there are no data points .
this is illustrated by the idealized example described above ( see also figure [ fig : staircase ] ) .
unfortunately this method is not trivially generalizable to more than one dimensions .
the cantor distribution reflects statistical independence based upon the two first generation probabilities @xmath209 for occupying the two first generation intervals , each of length @xmath210 , and is generated by the ternary tent map for a special class of initial conditions .
for other initial conditions the map yields other distributions .
nonuniform distributions that lack statistical independence are trivially easy to construct via either the ternary bernoulli shift or the ternary tent map .
for example , iterate the ( ternary ) initial condition @xmath211 .
statistically independent distributions where @xmath212 can also be constructed with only a bit more effort ( see ch .
9 of @xcite for the method ) .
summarizing , in the beginning there is only a collection of @xmath35 points ( or a time series ) generated by some generally unknown dynamical system .
we can construct the empirical distribution @xmath187 immediately , but we can not construct the coarsegrained distributions @xmath199 without first extracting the optimal partition @xmath213 . in dynamical systems
theory the optimal partition is provided by the generating partition . the generating partition , if it exists , is the signature of the dynamical system because it shows how the dynamics coarsegrains phase space naturally .
in contrast , the histograms that appear on the optimal support can be produced by every system in the same topologic universality class ( symbolic dynamics is universal for all systems in the same universality class ( @xcite , @xcite ) ) , so that a particular statistical distribution @xmath199 can not be the signature of a particular dynamical system .
both the henon map and the logistic map @xmath214 , with @xmath215 , where @xmath216 is the period doubling critical point , belong to the same topologic universality class ( both the logistic map with @xmath217 and the binary tent map belong to a separate universality class ) . both systems ,
although of different spatial dimension , generate the same range of histograms ( for corresponding classes of initial conditions ) , but on different supports .
from the perspective of both dynamical systems theory and the search for scale invariance the central problem of data analysis is to extract the optimal partition of a particular set of data points .
@xcite and @xcite for examples of the extraction of optimal partitions from data in fluid mechanics .
a nonuniform distribution on a fractal support looks fractally - fragmented ( looks more and more spiky as the support is viewed with finer and finer resolution @xmath40 ) .
distributions on space - filling supports may also be fractally fragmented , as we shall see below .
a nonuniform distribution on a uniform or nonuniform support ( that is either fractal or space filling ) can be used to sort and label fractal subsets of the support .
each subset has its own fractal dimension @xmath218 , where @xmath219 is the labeling
index @xcite .
multifractal , in this paper , always means a spectrum of fractal dimensions , where each dimension describes the scaling via ( [ eq:1 ] ) or ( [ eq:1b ] ) of a subset of the support of @xmath187 .
we will show in parts [ sect : generating_functions ] and [ sect : lognormal ] why the attempt to use other definitions leads to confusion and failed expectations ( predictions of `` dimensions '' that are not the dimension of anything are discussed in parts [ sect : correlation_integral ] and [ sect : generating_functions ] ) .
first , note that for a uniform distribution on a uniform support we can write @xmath220 . to describe a nonuniform distribution on a uniform or nonuniform support in generation @xmath66 of coarsegraining , we try to define scaling exponents @xmath221 by writing @xmath222 , where the scaling index @xmath221 takes on the same value @xmath223 on @xmath224 different intervals which we denote ( using very sloppy but obvious notation ) by @xmath225 .
therefore , by ( [ eq:1b ] ) , we can write @xmath226 to define @xmath218 as the hausdorff dimension of the subset of the support where @xmath223 ( if , indeed , such scaling holds ) .
if we could accurately replace the optimal nonuniform partition by the largest term @xmath227 in the subset , then we would obtain @xmath228 , where @xmath229 .
in other words , @xmath230 is the box
counting dimension for @xmath224 nonoverlapping uniform intervals of size @xmath231 .
this replacement works only for relatively uniform sub clustering .
otherwise it is necessary to compute @xmath218 . given a nonuniform empirical distribution @xmath187 over an optimal uniform partition , whether or not the histograms @xmath199 scale over a reasonable range of different sizes of @xmath232 , @xmath233 , is the main question for empirical data analysis . in cosmology
@xmath234 is the fraction ( out of a total number @xmath35 of galaxies ) of galaxies in the @xmath152th interval @xmath232 . in all data analysis and computer simulations
there can be at most finitely many values of @xmath219 and @xmath230 ( and finitely many values of @xmath149 and @xmath148 as well ) because @xmath235 is finite ( @xmath236 for typical galaxy samples ) .
however , the number of points in a spectrum will grow generation by generation n for a multifractal spectrum ( again , this distinguishes multifractal from bi fractal , tri fractal , @xmath237 , @xmath35fractal ) within the cutoff limits @xmath238 .
we can illustrate this via a simple example of an @xmath230 spectrum , the one given by the two scale cantor set with first generation probabilities @xmath239 describing statistical independence ( @xcite ) in all higher generations , and optimal first generation intervals @xmath240 and @xmath241 . in this case , fixing the scaling index @xmath219 picks out a monofractal , so that @xmath242 because all intervals in the subset have the same size @xmath171 . by using stirling s approximation on @xmath243 ( requiring @xmath156 ) ,
we then obtain , with @xmath244 that @xmath245 where @xmath246 parameterizes both @xmath219 and @xmath230 .
there are @xmath174 points in the spectrum so that @xmath247 , where @xmath248 defines @xmath125 .
note also that @xmath249 , as expected .
we can summarize our present terminology by writing down either the generating function ( @xcite ) @xmath250 or the generation function ( @xcite ) @xmath251 where , in the empirical search for scaling laws , it is first necessary to find an approximately _ optimal partition _ in order correctly to extract a multifractal spectrum of dimensions @xmath230 , or even one fractal dimension @xmath125 .
otherwise the convergence requirements ( the number @xmath66 of generations in a hierarchy @xmath123 of interval sizes ) needed to see scaling with an approximately correct exponent almost always outruns the limitation placed by the number @xmath35 of points in an empirical sample . when multifractal scaling can be shown to hold over enough different generations @xmath66 of interval sizes @xmath40
, then we can also write @xmath252 and @xmath253 note that ( [ eq:6 ] ) generalizes ( [ eq:1b ] ) so that we can explicitly discuss nonuniform distributions of points on the support , as with ( [ eq:5 ] ) . by using @xmath254 in ( [ eq:6 ] )
we get a result that looks formally like the generating function @xmath255 in ( [ eq:2 ] ) above if we set @xmath256 and @xmath257 .
note , however , that no assumption about the distribution of points @xmath258 over the support @xmath123 was necessary in order to define the generating function ( [ eq:2 ] ) . in dynamical systems
theory the generating function @xmath255 can be rewritten via symbolic dynamics as the partition function for a one dimensional ising model ( @xcite ) with long range interactions in equilibrium statistical mechanics ( @xmath147 is then the inverse temperature ) .
most discussions of multifractals inevitably stress that the generating functions ( [ eq:5 ] ) and ( [ eq:6 ] ) should _ themselves _
scale approximately in the limit @xmath185 where , due to domination of the entire sum by @xmath224 largest terms ( @xcite , @xcite ) , all of the same size ( and parameterized by @xmath259 ) , @xmath260 or @xmath261 where ( because @xmath66 goes to infinity as @xmath185 ) we would hypothetically obtain an @xmath230 curve parameterized by @xmath259 . in this case , because there are enough points in the spectrum that @xmath230 may be differentiated accurately , the `` generalized dimensions '' @xmath262 can be defined by @xmath263 , where @xmath264 is the slope in the plot of @xmath265 vs. @xmath259 and @xmath266 is the slope of the @xmath230 curve .
_ this continuum limit is misleading because it is generally inapplicable to data analysis .
_ to see this , merely note that both generating functions are sums over all possible scales , @xmath267 ( or , in the case of a nonuniform distribution on a monofractal , over @xmath39 terms @xmath268 with different exponents @xmath221 ) . for finite @xmath40 the generating functions ( [ eq:5b ] ) and ( [ eq:6b ] ) _ can not _ scale approximately unless @xmath40 is small enough , @xmath269 ( we can not emphasize this requirement too strongly ) , that a largest term approximation is accurate , which is generally _ not _ the case .
formulations and expectations of scaling based on the @xmath185 limits ( [ eq:7 ] ) and ( [ eq:8 ] ) have been taken seriously enough to have been employed during data analysis within the cosmology community ( @xcite ) , as elsewhere . in data analysis
this approximation is usually a very bad one ( see theilor @xcite for a clear and comprehensive exposition of the usual assumptions made in discussions of multifractal generating functions ) . in typical data analyses found in the literature
the largest term approximation is implicit in any plot of the logarithm of a generating function vs. @xmath270 in the search for generalized dimensions @xmath262 .
we expect that most empirical data will not produce small enough values of @xmath40 for a largest term approximation to be applicable .
_ even if multifractal scaling should hold term by term in ( [ eq:5 ] ) , in the form of ( [ eq:5c ] ) , it can not be discovered by a plot of @xmath271 vs. @xmath270 . _ instead
, one must check for scaling term by term inside the sum ( [ eq:5 ] ) . _ in other words , forget the sums ( [ eq:5 ] ) and ( [ eq:6 ] ) and check each term separately for scaling _
, to verify whether @xmath272 with @xmath273 actually holds over a geilo - range of scale sizes .
the generating functions are not directly measurable anyway , so one need nt care whether or not they scale .
the coarsegrained density defined by @xmath274 is typically singular . even if the support is space filling (
even if @xmath182 ) the coarsegrained density will look more and more intermittent as the resolution is improved if the distribution @xmath187 has nonuniform clusters that are scale invariant ( see @xcite for a one dimensional example from fluid turbulence ) .
any attempt to replace a staircase @xmath187 by a distribution with a differentiable density may delete , mask , or , at best , unnecessarily complicate the description of clustering and intermittence .
why introduce the mathematical fiction of a continuous distribution when observation gives us tractable discreteness directly ?
in cosmology ( @xcite , @xcite , @xcite ) , as it was in empirical analyses of dynamical systems ten years ago ( before the partitioning ( @xcite ) and recycling ( @xcite ) of strange sets ) , it is usual to work with the correlation integral @xmath275 where `` the correlation integrand '' @xmath276 is the fraction of galaxies in a sphere of radius @xmath18 , centered on the @xmath152th of @xmath35 galaxies . here , we take @xmath277 .
this means that the original dimensional variable @xmath18 for each galaxy has been rescaled by dividing it by @xmath278 , where @xmath278 is the value of the unscaled variable @xmath18 for which the of radius @xmath18 , centered on galaxy @xmath152 just touches the boundary of the data set . in other words , spheres of
every radius @xmath18 lie _
completely _ within the boundaries of the data ( see figure [ fig : scal_method ] ) . we stress that data sets should not be `` extended '' by adding points beyond the boundaries of the observational data during box counting . to do so would change the data set from the one that we set out to analyze .
in other words , we agree with the pietronero school @xcite , but in part [ sect : data ] we will show how to refine the data analysis to eliminate a certain ( self ) inconsistency in that work .
whenever the distribution in ( [ eq:10 ] ) exhibits statistical independence then @xmath279 holds as well , as is implicit in standard treatments .
clearly , data that are statistically independent on an _ optimal _ partition ( like the cantor function differences @xmath280 over the closed intervals @xmath281 ) will not show statistical independence over an _ arbitrary _ partition . in the limit of small
length scales the generalized dimension @xmath282 coincides with the correlation integral dimension @xmath2 for the case of an empirical distribution @xmath187 that exhibits statistical independence on its optimal coarsegrained support ( @xcite ) , but we can not merely assume statistical independence of observational data , and the zero length limit is anyway physically unattainable .
therefore we do not expect to extract @xmath282 via data analysis .
loosely speaking , however , one can refer to @xmath2 as the `` correlation dimension '' .
note that @xmath283 requires a either a multifractal distribution on a fractal support , or , for a nearly uniform distribution , the support must be fractal .
the generalization of ( [ eq:9 ] ) is given by the ( not directly measurable ) generating function @xmath284 as in ( [ eq:9 ] ) , @xmath35 is not the number of intervals in a nonoverlapping efficient partition , but is the number of points in the data sample . _
the correlation integral was first emphasized in the literature because it appears to allow us to circumvent the need to find an optimal partitioning .
_ we will return to this point shortly .
our first main point is that whether one uses ( [ eq:5 ] ) , ( [ eq:6 ] ) or ( [ eq:11 ] ) to study galaxy counts is irrelevant : a generating function can not scale unless all terms in the sum scale separately , and only then if one term dominates .
if scaling holds locally but the data set is not monofractal , then each term in the correlation integral ( [ eq:9 ] ) must have the form @xmath285 where @xmath286 is the local correlation integral index , the scaling exponent ( formally somewhat analogous to @xmath221 in equation ( [ eq:5b ] ) , except that here there is no accurate way to define a spectrum of fractal dimensions @xmath287 because an efficient partition is not defined by ( [ eq:11 ] ) ) for galaxy counts for @xmath65 spheres with corresponding radii @xmath18 centered on galaxy @xmath152 .
clearly , in the absence of largest term dominance the correlation integral representing local scaling with @xmath35 _ different _ indices @xmath286 , @xmath288 is not scale invariant because each term in the sum scales differently : . if , in a plot of @xmath289 vs. @xmath290 , linearity is reported for a large enough range of values of @xmath18 that the result is not spurious ( see figure [ fig : logplot ] , then the likelihood is that all terms inside the sum have approximately the same scaling exponent @xmath291 , indicating that the data set is approximately monofractal . for example , with @xmath35 points distributed uniformly over the @xmath41 intervals of the optimal partition @xmath68 of the middle thirds cantor set , the correlation integral is dominated by a single term @xmath292 and scales when @xmath293 , @xmath294 where @xmath295 because @xmath296 . in other words , @xmath3 is approximately scale invariant for @xmath293 because every term @xmath297 in the sum ( [ eq:9 ] ) is the same and is also scale invariant .
here , @xmath298 holds because we have ( implicitly chosen to use ) a uniform distribution on a monofractal .
in particular the analysis of part [ sect : data ] shows , that one can not assume that @xmath299 , where @xmath300 is poisson noise .
in the analysis of empirical data , on the other hand , if scaling of ( [ eq:11 ] ) is reported but has only been observed over a non geilo range of values of @xmath18 then the resulting spectrum of generalized correlation dimensions @xmath6 in @xmath301 defined by @xmath302 may be spurious ( and appears only in the unphysical limits where @xmath303 and @xmath304 ) .
the variation of @xmath305 with @xmath259 obtained from a plot of @xmath306 vs. @xmath307 over an inadequate range of @xmath18values probably indicates that the generating function ( [ eq:11 ] ) does _ not _ scale . in @xcite
it is shown how one can even get a spurious `` generalized dimension '' spectrum from log log plots of a _ gaussian _ distribution .
_ we point out next that the hope that one could circumvent the need to extract the optimal partition from the empirical data was an illusion : _ the generating function ( [ eq:11 ] ) can not be used to compute either a hausdorff or box counting dimension .
setting @xmath308 in ( [ eq:14 ] ) , the standard approach would lead the expectation that @xmath309 in @xmath310 provides an estimate for the box counting dimension @xmath79 .
this is impossible , because neither the box counting nor information dimension is included in the @xmath6 spectrum ( appeals to the limit of vanishing @xmath18 ( @xcite ) do not help in the empirical case ) .
the reason is simple : the terms on the left hand side of ( [ eq:15 ] ) do nt define an efficient , nonoverlapping partition of @xmath311 intervals , each of size @xmath18 .
hence , the `` convergence '' difficulties reported by @xcite in the attempt to estimate the box counting dimension by computing @xmath309 . if we would try instead to define an interval @xmath312 by formally writing @xmath313 in ( [ eq:15 ] ) , then the result @xmath314 reminds us superficially of equation ( [ eq:1b ] ) above , but @xmath315 does not define a hausdorff dimension : the @xmath35 intervals @xmath312 overlap very badly with each other because the sum is over all @xmath35 galaxies instead of over an efficient nonoverlapping partition .
equation ( [ eq:15b ] ) was proposed by martinez ( @xcite ) as one that yields @xmath125 , as well as the @xmath262 spectrum for @xmath316 via the generalization ( see also @xcite ) @xmath317 but information about the spectrum @xmath262 , aside from an estimate of @xmath282 , is not included in these formulae . equation ( [ eq:15b ] ) is supposed to be based on the equation @xmath318 with @xmath319 discussed by @xcite , where @xmath312 is the nearest neighbor distance between two points in a sample consisting of @xmath35 points . however , ( [ eq:16 ] ) is not the same as ( [ eq:15b ] ) because the partitioning in ( [ eq:16 ] ) is _ nonoverlapping _ : each point is connected only to one other point . for any finite subset of the middle
thirds cantor set consisting only of end points , we find that @xmath320 because the number of end points @xmath321 is simply twice the number of intervals @xmath41 required to cover those points . in this case the required nearest neighbor intervals are simply the usual nonoverlapping intervals @xmath322 of the middle thirds cantor set .
notice also that one can not stray far from the optimal result @xmath68 by using nonterminating ternary expansions @xmath96 with @xmath323 or 2 to generate @xmath35 points of the middle
thirds cantor set rather than by using only terminating ternary strings .
while scaling holds _ exactly _ when we use nearest neighbor distances in ( [ eq:16 ] ) and ( [ eq:16b ] ) for the mathematical idealization of a cantor set , we should not expect an analogous `` convergence rate '' when we use empirical data .
equations ( [ eq:16 ] ) and ( [ eq:16b ] ) should only be expected to yield an accurate scaling index @xmath324 when the nearest neighbor distances are taken to be the end points of an optimal partition .
the expectation ( @xcite , @xcite ) that different generating functions can be used to compute `` the @xmath262 spectrum '' for different ranges of @xmath259 , even in the idealized limit where @xmath185 , is based on three unfullfilled expectations .
first , the box counting dimension does not belong to the @xmath6 spectrum ( neither does the information dimension ) .
second , the @xmath6 spectrum ( defined by ( [ eq:14 ] ) in the limit of infinitely small @xmath18 ) does not coincide with the @xmath262 spectrum of ( [ eq:5b ] ) and ( [ eq:6b ] ) ( which do include the box counting and information dimensions as @xmath185 ) .
third , one can not change probabilities and supports and expect scaling exponents to remain invariant : neither multifractal spectra @xmath230 , nor generalized dimensions @xmath262 derivable from multifractal spectra , can be used to define universality classes .
the misconception that an optimal partition is unnecessary has led to the expectation that partitions can be manipulated without changing @xmath230 and @xmath262 .
multifractal spectra and generalized dimensions are nonuniversal : they change whenever either the support or the histograms on that support is changed .
this is easily seen via the simplest possible examples . to emphasize this last assertion we demonstrate what happens when we try to get a complete @xmath262 spectrum by combining results from ( [ eq:5 ] ) and ( [ eq:6 ] ) for disjoint ranges of @xmath259 , but while using different distributions on different supports in each generating function .
the underlying point set is in each case taken to be a finite number of points in the two scale cantor set .
we consider first a uniform partition with uneven probabilities ( @xmath325 in generation @xmath88 ) and statistical independence ( for ease of calculation ) , whereas in the second case we take even probabilities @xmath209 also with statistical independence on the uneven but optimal partition with @xmath326 .
we can cover the finite ( or infinite ) point set in the first case by using an efficient uniform ( but not optimal ) partition based on @xmath327 , so that then we get @xmath328 and @xmath329 .
if in the second case , we use ( [ eq:6 ] ) with @xmath330 , then @xmath331 and @xmath332 .
the separate spectra do not lie on top of each other because the end points @xmath333 and @xmath334 do not coincide .
clearly , we can not combine these two different calculations in order to estimate disjoint parts of the @xmath262 spectrum of either case .
in this search for scaling we use only the correlation `` integrand '' @xmath297 , not the correlation integral @xmath3 for the reasons explained in parts [ sect : multi_empirical ] and [ sect : correlation_integral ] .
we calculate the number of galaxies within a sphere of radius @xmath18 centered on each galaxy as depicted in figure [ fig : scal_method ] .
_ we only use spheres that are completely within the sample geometry_. since we want to investigate the scaling properties we are not allowed to apply any boundary corrections that assume stationarity of the distribution of galaxies with respect to translations ( i.e. homogeneity ) or rotations ( i.e. isotropy ) .
such `` corrections '' would tend to introduce a spurious scaling with dimension three .
boundary `` corrections '' are inherent in all the usually used estimators for the two point correlation function , see e.g. @xcite .
similar corrections were used for estimators of the correlation integral see e.g. @xcite . for the same reasons as above we use volume limited samples . using flux ( or magnitude ) limited samples one is usually using a weighting scheme based on the selection function . in weighting with the selection function
one assumes homogeneity in giving a weight to galaxies proportional to the mean density , which is determined mainly from the nearby regions .
we only show galaxies in the plots where we have at least one neighbour in a radius range larger than @xmath335 , our `` scaling '' range ( with the limited data available it does not make sense to use decades , since we only have at maximum one decade available ) . to perform the scaling analysis more quantitatively we fit a straight line to the @xmath336@xmath5 plot and determine the slope , again only for galaxies having a `` scaling '' range larger than @xmath335 .
we are severely limited by the small number density ( or equivalently , the small sample size ) of the catalogues
. therefore we will not be able to extend @xmath335 over more than one decade which is obviously too small to draw any conclusions about scaling ( see figure [ fig : logplot ] ) . to estimate the influence of @xmath335 on the distribution of the slopes we look at three different @xmath335 . to get a geilo range of scales from pie
shaped samples ( as in figure [ fig : scal_method ] ) we would need observational data extending over several thousands .
= 6.5 cm = 6.5 cm we look , as an example of optically selected data , at the cfai catalogue a magnitude limited catalogue consisting of 1880 galaxies ( huchra @xcite ) . using this data set fractal and multifractal analysis
were performed e.g. by coleman & pietronero @xcite or by martinez et al .
@xcite .
first consider the volume limited sample with 40 depth with 360 galaxies in total , having a mean separation of 9.5 . in figure
[ fig : neigh_cfa_v4_n10 ] we show for 35 arbitrarily selected galaxies the number of neighbours @xmath297 against the radius of the sphere @xmath18 for galaxies in the volume limited sample with 40 depth . in this case
we demand a `` scaling '' range of @xmath337 , resulting in 157 galaxies having a ( more or less well ) defined slope . in figure
[ fig : neigh_cfa_v4_n20 ] we show the scaling properties for 35 arbitrarily selected galaxies from the 67 galaxies with `` scaling '' range @xmath338 , and in figure [ fig : neigh_cfa_v4_n30 ] we show the scaling properties of all the 22 galaxies which have a `` scaling '' range @xmath339 , spanning roughly one decade . as a first impression one recognizes that the scatter in the slope is large and does not decrease for larger @xmath335 , which should result in more reliable estimates for the slope . to make this impression more quantitative we plot the frequency of the slope for each of these samples in figure [ fig : hist_cfa_v4 ] .
one has to bear in mind that this frequency table is in the case of @xmath340 constructed from 22 galaxies only . in all three cases ,
the slope fluctuates over a range of 1.25 to 2.5 . in figure
[ fig : hist_cfa_v6 ] we plot the frequencies of the slopes for the volume limited sample with 60 , consisting of 215 galaxies with mean seperation 24 . imposing the constraints for the `` scaling '' range we are left with 67 galaxies for @xmath341 , 31 galaxies for @xmath342 , and only 9 galaxies for @xmath343 .
we see that no conclusions are possible with so few points .
the peak around 1.75 for @xmath343 ( determined from only 9 galaxies ! ) is a mainly due to sampling galaxies from the same region of the space in the center of the sample ( see figure [ fig : scal_method ] ) .
now we look at a sample of iras selected galaxies with limiting flux 1.2 jy fisher et al . @xcite consisting of 5313 galaxies .
the big advantages of this sample is the nearly complete covering of the sky , and the homogeneous flux calibration .
fractal and multifractal analysis of iras galaxies were performed e.g. by martinez & coles @xcite , xia et al .
@xcite and labini et al .
the last mentioned authors claim , that the iras samples are too sparse to estimate fractal dimensions reliably .
we proceed similar to the analysis of the cfa catalogue [ sect : cfa_analysis ] .
as discussed in kerscher et al .
@xcite we find differences between the northern and southern hemisphere , but since we do not want to focus on this topic we show the results for the combined data only . in figure [
fig : neigh_jy_v8_n10 ] we show for 35 randomly selected galaxies the number of neighbours @xmath297 against the radius of the sphere @xmath18 for galaxies in the volume limited sample with 80 depth ( mean seperation of the galaxies is 24 ) .
restricting ourselves to a `` scaling '' range of @xmath344 we are left with 359 galaxies of the total 788 galaxies in the volume limited sample . in figure
[ fig : neigh_jy_v8_n20 ] we show the scaling properties for 35 randomly selected galaxies from the 167 galaxies with a `` scaling '' range of @xmath345 , and in figure [ fig : neigh_jy_v8_n30 ] 35 randomly selected galaxies from the 72 galaxies with a `` scaling '' range of @xmath346 .
again we have a `` scaling '' range only spanning roughly one decade .
the sample with @xmath344 is clearly inapropriate for an analysis , since often only one neighbouring galaxy is within the scaling range , giving rise to a spurious slope of zero .
again we see a large scatter in the slopes for all @xmath335 , which does not decrease if we got to higher @xmath335 . to make this more quantitative we again plot histograms of the slopes , now for a sequence of volume limited samples .
the sample with 40 depth containing 646 galaxies is shown in figure [ fig : hist_jy_v4 ] . the sample with 60 depth containing 880 galaxies is shown in figure [ fig : hist_jy_v6 ] . the sample with 80 depth with 788 galaxies is shown in figure [ fig : hist_jy_v8 ] .
= 6.5 cm = 6.5 cm = 6.5 cm in both catalogues the local scaling exponents @xmath286 fluctuate over a broad range ( see figures [ fig : hist_jy_v4][fig : hist_jy_v8 ] and figures [ fig : hist_cfa_v4 ] , [ fig : hist_cfa_v6 ] ) indicating that there is _ no _ global monofractal scaling .
it mainly tells us that the distribution admits large fluctuations . from the limited data we are not able to judge whether these fluctuations are scale invariant ( over three decades ) or not .
therefore , fractal scaling , and certainly multifractal scaling , can not be deduced from that limited data as claimed in e.g. @xcite , @xcite , @xcite , @xcite , @xcite .
the fact that the correlation integral ( or the two point correlation function ) apparently scales with one exponent is in this case not related to the scaling of the galaxy distribution ( again , see figure [ fig : logplot ] ) .
we want to emphasize that the broad range of different slopes is _ not _ a sign of multifractality , it only shows that we have large fluctuations . in e.g.
@xcite the authors claim to see scale invariance over three decades ( from 1 to 1000 ) .
this statement is based on the scaling properties of only one @xmath297 . in this case
@xmath297 should be the number of galaxies in a sphere centered on our galaxy .
however , the analysis is carried out by using only intersections of a cone with a sphere as discussed in @xcite .
such an analysis is inconsistent with their own definition of fair sampling , very badly violating fig .
8 of @xcite . in fitting straight lines to a log log plot over roughly one decade we have performed a superficial data analysis , one that gives excessive weight to galaxies near the center of the sample .
unfortunately nothing more is possible without postulating `` boundary corrections '' that have no basis in observation , or without even more unfairly weighting points near the center of a conical sample more heavily than those near a boundary .
but even in doing so , we do not find any indication of global scaling .
the main message is that the current data are insufficient for a reliable scaling analysis .
we end this section with a quote from @xcite : `` if a sample contains too few points there may be no way to get any information from it .
in such a case one has to wait for better ( observational ) data . ''
the term multifractal is defined in jones @xcite by requiring only that the moments of an arbitrary distribution @xmath347 , @xmath348 where @xmath349 is a random variable defined in or on intervals of size @xmath36 , should scale , @xmath350 for small enough interval sizes @xmath36 .
however , without further requirements on @xmath349 and @xmath347 , there is no reason to expect that scaling exponents in ( [ eq:17b ] ) , if they exist _
at all _ for a given distribution @xmath347 , bear any relation to generalized dimensions @xmath262 derivable in the infinite precision limit from multifractal spectra @xmath148 or @xmath230 . to be specific , is the partition nonoverlapping ? efficient ? what is fractal about @xmath349 or @xmath347 ?
we will show via example in part [ sect : lognormal ] that equation ( [ eq:17 ] ) with a scaling law ( [ eq:17b ] ) generally does not describe intermittence due to voids , and that if one tries in those cases to force the definition @xmath351 ( or , as in @xcite , @xmath352 ) then the @xmath353 are not dimensions describing the support of the probability distribution , or of any other coarsegrained set or subset connected with an arbitrary distribution @xmath347 .
the generating function ( [ eq:17 ] ) combined with the scaling expectation ( [ eq:17b ] ) is used in definitions of _ multiaffine _ fractals ( @xcite ) , where a deterministic or random variable ( or field ) @xmath349 is continuous but has singular ( or no ) derivatives . if the distribution of singularities of the field @xmath349 can be described locally by writing @xmath354 and @xmath355 , then ( [ eq:17 ] ) may or may not yield scaling exponents @xmath356 that give rise to a spectrum of generalized dimensions , defined by @xmath357 in the limit @xmath358 , where @xmath359 and @xmath282 really are _ fractal dimensions _ of _ something _ in the model .
this happens _ only _ when @xmath360 describes a spectrum of fractal dimensions .
otherwise @xmath361 and @xmath360 are just a rewriting of a nonfractal probability distribution @xmath347 via a coordinate transformation , and _
nonfractal distributions can not be made fractal ( or the converse ) by a differentiable coordinate transformation_. stated another way , @xmath360 is not a spectrum of hausdorff dimensions unless `` @xmath36 '' represents an optimal or at least efficient partition of the support of an underlying pointwise distribution @xmath187 .
examples in the literature where @xmath354 with @xmath362 are used without any requirement of efficient partitioning of a support are height fluctuations in surface roughening ( @xcite ) , self organized criticality ( @xcite ) and velocity structure functions in the inertial range of fluid turbulence ( @xcite ) .
in contrast with multiaffine fractals ( where there is no idea of a generating partition ) , for a self similar fractal ( heretofore called `` fractal '' ) the point set is generally spatially - fragmented ( koch and peano curves are , however , continuous ) , like a cantor set , and @xmath195 scales like @xmath363 ( _ not _ like @xmath364 ) in order to describe a highly singular density @xmath365 on the optimal partition describing the support of @xmath187 . in a spirit similar to the attempt to define multifractal by using the moments ( [ eq:17 ] ) of an arbitrary distribution @xmath366 with a scaling law ( [ eq:17b ] )
, the definition @xmath367 where the probability distribution @xmath368 is undefined , is treated in various places ( @xcite ) as if it would be identical with the generating function @xmath369 although it is not . for an arbitrary probability distribution @xmath368 these two generating functions
are not even related ; their scaling exponents ( if scaling exponents exist in either case ) are not necessarily the same even if the generating functions _ are _ qualitatively related . _
multifractal spectra and generalized dimensions are not universal_. instead , they change with the histograms and their support .
furthermore , any deviation from the empirical distribution @xmath187 and its optimal coarsegrained descriptions @xmath199 is equivalent to changing the underlying data set . rather than imagining that the empirical distribution @xmath187 can be treated as a random field that fluctuates from one galaxy sample to the other ( as in jones @xcite ) , we view different @xmath187 s from different samples as disjoint pieces of a _ single _ global distribution of galaxies whose local properties can be discovered empirically , but whose entire global ( one hesitates to say `` universal '' ) aspect can never be known due to the inherent limitations on observation .
we do not want to try to replace what we do not know ( @xmath187 _ measured _ globally ) with speculations that can not be tested ( @xmath368 _ postulated _ globally ) .
the notion of statistical ensembles is useless here : there is only one universe , and it is not in equilibrium .
the source of confusing together entirely different generating functions can be traced to the use of the infinite precision limit in papers on dynamical systems theory written more than ten years ago .
it is suggested in @xcite ( see also @xcite ) that ( [ eq:5c_2 ] ) is analogous to the lebesgue integral @xmath370 and should yield the same generalized dimensions @xmath262 in the infinite precision limit , where @xmath187 is supposed to be `` the natural invariant measure '' of a chaotic dynamical system on a strange attractor ( see @xcite , e.g. , for one definition of `` natural '' ) and @xmath371 ) is the fraction of points lying within a ball of size @xmath36 covering ( but not necessarily centered on ) a data point @xmath188 .
there are two serious difficulties with the attempt to replace ( [ eq:5c_2 ] ) by ( [ eq:19 ] ) , and both revolve about lack of uniqueness .
first , there is no evidence from observation that a chaotic dynamical system generates `` a natural measure '' for the various initial conditions ( meaning also `` present conditions '' ) found in nature .
mathematically seen , a chaotic dynamical system can generate infinitely many different distributions ( `` measures '' ) @xmath187 for infinitely many different classes of initial conditions ( @xcite & @xcite ) .
empirically , the dependence on initial conditions is not a problem : the data are described by the empirical staircase @xmath372 without having made any theoretical assumptions that prejudice the data analysis we can say that the initial conditions , whatever they were , produced the empirical distribution @xmath187 via the time evolution of some dynamical system . given the empirical measure ( [ eq:3b_2 ] ) there is still ambiguity inherent in the attempt to use ( [ eq:19 ] ) as a replacement for ( [ eq:5c_2 ] ) . in finite precision
there are different possible definitions of the integral , depending on which subset of the data set we decide to measure ( before we can identify the function @xmath373 we must first define `` @xmath36 '' ) .
if we choose the balls / intervals @xmath374 to have arbitrary length @xmath36 , centered on a data point @xmath375 ( as in @xcite ) , then the fraction of points lying within each interval of size @xmath36 is given by @xmath376 where @xmath377 is the correlation integrand @xmath378 deleting the term @xmath379 in ( [ eq:10_2 ] ) is unimportant if the intervals are large enough to give `` good statistics '' ( pedantically , one can also replace the factor @xmath380 by @xmath381 in ( [ eq:10_2 ] ) ) .
insertion of the pointwise definitions @xmath382 and @xmath383 into the integral ( [ eq:19 ] ) yields the correlation integral generating function @xmath384 which differs significantly from ( [ eq:5c_2 ] ) in data analysis , as we have emphasized in section [ sect : correlation_integral ] . in dynamical systems
theory the @xmath35 intervals can in principle be chosen small enough not overlap with each other : on a mathematically defined strange attractor there are @xmath385 points in any neighborhood of any arbitrary point @xmath375 on the attractor ( @xmath385 is the cardinality of the attractor ) . here
, the @xmath35 intervals ( or balls ) @xmath386 of size @xmath36 can be chosen small enough not to overlap , but certainly do not partition the attractor efficiently , if at all .
equation ( [ eq:11c ] ) , which was not invented with partitioning in mind , is merely a time average over @xmath35 points on the attractor , and the uniform weight @xmath380 is correct because each point @xmath375 occurs exactly once ( so long as trajectories of the dynamical system are unique , which we assume here ) . in nonlinear dynamics calculations
the number @xmath35 of points may be increased by increasing the precision of the calculation . in cosmology , in contrast , @xmath35 is the total number of galaxies in a finite sample , so that the @xmath35 intervals of size @xmath36 are always overlapping .
there is a different way to define balls @xmath374 and a corresponding function @xmath373 .
instead of choosing @xmath35 uniform intervals where @xmath35 is the number of data points , requiring the pointwise definition @xmath376 as given by ( [ eq:10_2 ] ) ( which does _ not _ include a partition of the data set ) , we can instead choose our balls @xmath386 to be the @xmath39 intervals @xmath123 in the optimal partition of the data .
then , @xmath373 is given by the simple function @xmath387 where @xmath388 is the characteristic function for the partition @xmath123 of disjoint intervals @xcite and @xmath389 here , @xmath375 and @xmath390 are taken to be the end points of any of the @xmath39 _ optimal _ intervals @xmath123 . with this optimal choice of `` what to measure '' ( optimal choice of function @xmath373 to integrate with respect to the measure @xmath187 ) the integral ( [ eq:19 ] ) yields @xmath391 _ from the standpoint of both data analysis and measure theory the only significant difference between the distributions ( [ eq:10_2 ] ) and ( [ eq:21 ] ) is the lack of a partition in ( [ eq:10_2 ] ) , and the use of an optimal partition to define ( [ eq:21 ] ) . whether these two approaches do or do not , in the limit of @xmath185 for a mathematical fractal of cardinality @xmath385 , yield the same generalized dimensions ( whether @xmath392 as @xmath393 ) is of no importance whatsoever for the analysis of empirical data . _
what has lognormal to do with multifractal ?
the question arises because it has been asserted that the lognormal distribution is multifractal , that it defines a spectrum of generalized dimensions ( @xcite & @xcite , @xcite ) and an @xmath230 spectrum ( @xcite ) .
before answering this question we review how and where the lognormal distribution appears in discussions of multifractals , where a multifractal spectrum ( as defined in this paper , following halsey @xcite ) describes the spectrum of dimensions of nonoverlapping subsets of the support of a probability distribution . with the discussion of @xcite in mind , but following @xcite , let us make a largest term approximation on the generating function ( [ eq:5 ] ) . with @xmath259 fixed we first locate the largest term in ( [ eq:5 ] ) by minimizing the exponent @xmath394 , yielding @xmath395 . very near
( and only very near ) to the smallest exponent @xmath396 , where @xmath397 , we can write @xmath398 . according to a standard method ( @xcite ) we
next replace the sum over all these nearby terms by the integral @xmath399 where the range of integration @xmath400 is over the tiny region @xmath400 in @xmath219 containing all of the exponents @xmath401 that do not deviate from the minimum exponent @xmath402 more than quadratically in @xmath403 .
this quadratic approximation to deviations of the exponent @xmath404 from the minimum @xmath402 only works as @xmath185 , in which case the integrand in ( [ eq:21_2 ] ) is sharply enough peaked that , with small error , we may extend the integration limits to plus and minus infinity .
clearly , a locally ( not globally ) gaussian approximation to deviations from the minimum exponent @xmath402 is the same as saying that the deviations of @xmath405 from @xmath406 are locally ( not globally ) lognormal .
this local lognormality only contributes to the _ prefactor _ in ( [ eq:21_2 ] ) in the unphysical limit where @xmath185 , and not to the @xmath230 spectrum described by the exponent @xmath402 .
any time that an exponent @xmath361 has a gaussian distribution then the function @xmath407 is distributed lognormally ( see @xcite for an example from percolation theory , where permeabilities @xmath408 of sandstone and limestone deposits were long thought to be approximately lognormally distributed , with gaussian porosities @xmath409 , where @xmath410 ) .
_ this has nothing to do with the question whether the lognormal distribution is multifractal .
_ there are two ways , _ related mathematically to the above approximation _ , in which the lognormal distribution is called multifractal in @xcite and @xcite .
jones @xcite asserts that ( [ eq:17 ] ) with ( [ eq:17b ] ) defines multifractal , where @xmath411 and the @xmath412 are supposed to be generalized dimensions derivable from the halsey method as well .
_ this constitutes an entanglement of unrelated ideas_. following jones @xcite , we use @xmath413 in ( [ eq:17 ] ) along with a lognormal distribution of @xmath349 to compute @xmath414 ( the exponent @xmath361 is gaussian with mean @xmath415 and mean square fluctuation @xmath416 ) . using jone s @xcite second definition of @xmath417 ( instead of his first ) , where @xmath418 , we obtain @xmath419 . in other words a scaling law ( [ eq:17b ] )
generally does _ not _ follow : lognormal distributions per se , inserted into ( [ eq:17 ] ) do not yield scale invariance ( [ eq:17b ] ) , because the expected scaling exponent depends on @xmath420 .
a scaling law ( [ eq:17b ] ) follows _ only _ if we restrict to lognormal distributions where @xmath421 is proportional to @xmath422 . in this case
we obtain @xmath423 , where @xmath424 .
@xmath425 is not the dimension of the support of the lognormal distribution ( where @xmath426 ) , and the scaling exponents @xmath412 are not the dimensions of anything else in that distribution .
( we adhere to the assumption that fractals and multifractals are generated by deterministic dynamics , and do not consider the so called `` random fractals '' . )
jone s refinement of ( [ eq:17b ] ) is , in this case , equivalent to the imposition of the constraint @xmath427 in turbulence modelling ( see frisch @xcite ) . to try to model an eddy cascade in fluid turbulence
one must evaluate the average @xmath414 where @xmath36 represents the size of an @xmath66th generation eddy , and @xmath361 is supposed to be an exponent analogous to @xmath219 in multifractal spectra . there
, frisch @xcite makes a different identification than jones , namely , that @xmath428 for a lognormal distribution .
this yields @xmath429 ( space - filling support ) but the exponents @xmath353 for @xmath430 are not dimensions of anything in the model .
the origin of this apparent ( from our standpoint ) mislabeling of scaling exponents as multifractal is that frisch defines @xmath230 spectra ( and consequently @xmath353 spectra ) differently than we have .
his definition is not designed to agree with the fractal dimensions @xmath230 describing singular distributions ala halsey et al .
@xcite , but describes instead the _ cramer function _ in statistics ( @xcite ) .
cramer function may exist where nothing is fractal_. a cramer function , by construction , describes distributions of independent random variables @xmath431 or @xmath221 the limit where @xmath66 goes to infinity ( @xmath36 goes to zero ) .
in contrast , the indices @xmath219 in ( [ eq:5 ] ) and ( [ eq:21 ] ) are not random variables : they are scaling indices describing coarsegrained probabilities ( and occupation numbers @xmath432 ) @xmath433 . the cramer function is a systematic way of obtaining a description whereby @xmath354 and @xmath362 via a limit theorem in classical statistics , _ for the case where the @xmath431 are independent random variables _ , and has no necessary connection with the idea of spectra of hausdorff dimensions , or generalized dimensions derivable from spectra of hausdorff dimensions . the cramer function is based on the law of large numbers and appears in classical equilibrium statistical mechanics , for example .
this approach can be used to describe an alternative version of lognormality discussed in @xcite : in that case their spectrum @xmath434 describes the _ integrand _ of ( [ eq:17 ] ) , and not a scaling law ( [ eq:17b ] ) that might follow as a consequence of actually _ calculating _ the integral ( [ eq:17 ] ) . a scaling law ( [ eq:17b ] )
does not follow at all without assuming arbitrarily that @xmath435 varies as @xmath436 . with that restriction
we obtain @xmath437 with @xmath438 , analogous to frisch s result quoted above , but where we should now choose @xmath439 in order to describe the support of the lognormal distribution ( otherwise , @xmath79 is not the fractal dimension of anything in the model ) . even with this choice the remaining @xmath353 are not fractal dimensions , even for @xmath440 , of any aspect of the lognormal distribution . however , the _ same _ lognormal distribution can be understood from an entirely _ different _ standpoint : namely , as a combination of the gaussian integrand with the exponent @xmath441 in equation ( [ eq:21_2 ] ) of halsey et al above . in this case
@xmath79 is is _ not _ the hausdorff dimension of the support of the lognormal distribution , and the linear spectrum @xmath353 describes _ only _ the region near the peak of an unknown @xmath230 spectrum , a spectrum of box
counting dimensions where @xmath442 ( from the standpoint of eqn .
( [ eq:21_2 ] ) above one would interpret the @xmath356 spectrum derived from the lognormal distribution in turbulence modelling as the fragment of an @xmath230 spectrum that is valid only for very small values of @xmath443 , near @xmath444 , although this is certainly _ not _ the traditional interpretation ) . on the other hand , however , the _ generalization _ of the lognormal approximation represented by jone s @xcite equations ( 14 ) and ( 15 ) , and ( 30 ) , are not definitions of @xmath230 ala halsey et al .
@xcite , but represent instead the idea of a cramer function in classical statistics ( see frisch 95 ) : _ unless @xmath230 arises as the spectrum of hausdorff dimensions from an infimum condition on partitions , then @xmath230 is not ( by the halsey @xcite definition ) a multifractal spectrum .
_ this entanglement of different ideas did not originate in cosmology : one of the authors of halsey et al .
@xcite later used the cramer function , called it `` multifractal '' , and cited halsey et al .
@xcite as the reference ( @xcite )
. one can choose to follow halsey et al .
@xcite in defining @xmath230 and @xmath262 , or one can follow mandelbrot @xcite in defining `` @xmath230 and @xmath262 '' via cramer functions @xcite , but one should not mix these two different definitions together without comment as is done in @xcite .
we recommend the definition of multifractal given in this paper as the standard because , in that case , @xmath230 is always the hausdorff dimension of a subset of the support of the empirical distribution @xmath187 .
the necessity of an optimal partition in order to define @xmath230 is implicit in halsey et al .
@xcite ( see their `` infimum '' requirement ) but was not emphasized strongly enough at that time . the role played by the infimum requirement became clear only after the later work on generating partitions in nonlinear dynamics ( @xcite , @xcite and @xcite ) , which is little
known within the community of cosmologists .
when is a distribution @xmath187 multifractal ? to answer this question consider any distribution @xmath187 , empirical or theoretical , where @xmath189 , @xmath190 , and @xmath187 is nondecreasing . for idealized differentiable distributions
we have @xmath445 ( which is the same as @xmath446 , etc . ) and @xmath204 .
@xmath187 need not be differentiable , however , and generally is nt . in order to determine whether @xmath187 `` is multifractal '' ( admits a decomposition of its support into interwoven fractals with different dimensions @xmath230 )
one must determine the optimal partition and form the difference @xmath447 over each interval in that support to obtain the hierarchy of histograms @xmath199 ( for a differentiable distribution any space - filling partition will do the job ) .
having done that , one then investigates whether @xmath448 holds over @xmath449 intervals in the support as the interval sizes are systematically reduced .
all fractal distributions ( for points on a line ) have an density @xmath450 that is dense with singularities because @xmath451 , and if the distribution is multifractal then the indices @xmath221 will vary over the support according to @xmath449 .
a highly fragmented ( spiky ) coarsegrained density is typical of a multifractal distribution @xmath187 .
gaussian distributions are not multifractal . neither are lognormal distributions .
no smooth , differentiable distribution is multifractal because , by definition , such a distribution has a smooth density on a support with integer dimension @xmath79 .
smooth distributions can be differentiated everywhere , corresponding to the requirement that @xmath452 holds everywhere .
the cantor function @xmath187 in part [ sect : empirical_dist ] describes clustering and voids ) , the binomial distribution with @xmath453 on a space filling support describes intermittence without voids ( see @xcite for a physical example ) , but the lognormal distribution can not describe voids because it is differentiable . the distribution defined by the density @xmath454 is not differentiable at @xmath455 and 1 and is bifractal ( halsey et al .
@xcite ) : @xmath456 for @xmath457 but @xmath458 , @xmath459 for @xmath455 and 1 .
the bifractal staircase of she et al .
@xcite shows coalescence without voids only because a continuum of initial conditions was used rather than a finite number ( blocks of initial conditions with voids should also produce coalescence with voids in that model ) . summarizing , nontrivial @xmath230 spectra guarantee an intermittent probability density , one corresponding to a nondifferentiable probability distribution @xmath187 .
all fractal distributions have singular densities . having a fractal support guarantees a singular density @xmath460 with @xmath461 , but multifractal scaling can also hold for inhomogeneous distributions ( like those having statistical independence with uneven probabilities ) on a space - filling support .
the singular `` pointwise '' density of matter @xmath462 in any epoch is determined by the empirical staircase distribution @xmath463 , the generalization of ( [ eq:3b ] ) to three dimensions . a necessary but _
insufficient _ condition for a coarsegrained density that is smooth enough to be approximable by a differentiable function is that the support of the empirical distribution is space filling , meaning @xmath464 .
no information about @xmath125 is provided by the correlation integral exponent @xmath2 ( or by the correlation dimension @xmath282 ) unless , ( 1 ) the support is monofractal and the distribution is uniform ( @xmath465 ) , or ( 2 ) the support is space filling and the distribution is uniform or at least differentiable ( @xmath466 ) . in both cases @xmath467 . otherwise , we know only that @xmath468 and @xmath469 . an increase in @xmath2 with an increase in scale from intermediate toward cosmologic scales , @xmath470 , even if it would be found in the data , would not tell us anything about @xmath125 unless we would find that @xmath471 .
so long as @xmath283 no information about @xmath125 is provided by @xmath2 .
an increase of @xmath2 with increasing scale ( as is claimed in @xcite , @xcite ) would not imply that there exists a large scale coarsegraining where @xmath472 . if @xmath473 then the approximation of the empirical distribution by a differentiable one is impossible .
the same conclusion follows if @xmath474 .
fractal scaling of galaxies in an intermediate range @xmath475 , e.g. would still allow for the possibility of nonfractal matter distributions at the largest ( or `` cosmologic '' ) scales .
an empirical matter distribution @xmath463 that shows no clusters and voids `` at large enough scales '' @xmath40 would require a support with dimension @xmath476 . whether galaxies are distributed more or less uniformly over a nonuniform space - filling support @xmath477 is then a question of whether @xmath478 holds over enough generations of the hypothetical `` cosmologic - scale support '' so that one can define the derivatives of densities normally used in hydrodynamics . only for a uniform support would this condition reduce to the requirement of statistical independence with equal probabilities , @xmath479 .
this is the requirement for large - scale uniformity ( @xmath480 ) stated in the language of dynamical systems theory . a necessary condition for homogeneity in a given direction @xmath188 , at cosmological scales @xmath470 ,
can also be stated as follows : on what scale @xmath40 of cosmologic coarsegraining can a staircase @xmath187 of @xmath35 steps ( @xmath187 denotes the empirical distribution @xmath463 with @xmath481 and @xmath482 held constant ) be approximated by a differentiable distribution , @xmath483 , where @xmath484 is smooth , approximately analytic ? there are two requirements : the number @xmath198 of data points in each interval must be very large , and the spacing between points can not be very different from @xmath485 .
this is the same as saying that the steps in @xmath187 are nearly uniform and of very small width , and lie approximately on the straight line @xmath486 with slope @xmath487 ( uniform density ) .
if the pointwise spacing is not exactly @xmath485 , but @xmath464 and there are no voids and clustering , then the staircase may be a smooth deformation of the constant density distribution @xmath486 with variable but nearly smooth slope @xmath488 approximating a nonuniform smooth density .
the necessary and sufficient condition for large scale homogeneity , stated in the language of statisticians is given in stoyan et al .
contrary to the advice of part 9 in @xcite we point out that `` pencil beam surveys '' generally can not be treated as one dimensional cuts . in order to qualify as a one dimensional cut
, the maximum width of a pencil beam survey should be on the order of the size of a galaxy .
if we think in terms of hydrodynamic models of clustering , then space filling supports , by liouville s theorem , require conservative dynamical systems
. a smooth density at large scales can not be the result of dissipative hydrodynamics .
no empirical test can be performed globally on the scale of the universe .
the best that one can hope for is to gain information about the different local distributions of matter for various different samples of galaxies and clusters of galaxies at scales @xmath489 and test them for homogeneity and isotropy ( our euclidean language is applicable locally in a curved space - time ) .
if the scales required to exhibit evidence for homogeneity and isotropy should fall beyond the inherent limitations on all future observations , then the cosmological principle is not falsifiable and is only a matter of belief . a more useful question is how to combine fractal or multifractal distributions with hydrodynamics , now defining the density @xmath490 to include the masses of galaxies , as is done in part 6 of @xcite . in a related publication @xcite it
is asserted that `` phenomena in which _ intrinsic self similar irregularities develop at all scales _ and fluctuations can not be described in terms of analytic functions .
the theoretical methods used to describe this situation could not be based on ordinary differential equations because self
similarity implies the absence of analyticity and the familiar mathematical physics becomes inapplicable . ''
these claims are patently false : there is certainly no reason why one can not study the dynamics of eq .
( [ eq : mass_density ] ) as an n body problem using nonlinear differential equations , as was done by mathematicians from laplace to poincar and beyond !
why was such a sweeping statement made to begin with ?
clearly , by the extrapolation length scales to zero , to the empirically and physically meaningless mathematical limit where , essentially , two galaxies occupy the same position . only in this limit are fractal densities nonanalytic enough to be completely nondifferentiable .
in fact the coarsegrained picture of empirical distributions formulated in parts [ sect : empirical_dist ] and [ sect : multi_empirical ] above leads in principle to a hydrodynamic description in terms of the usual differential equations of mathematical physics . at any desired resolution @xmath491
, simply represent the density by @xmath492 where ( see eq .
( [ eq : mass_density ] ) above ) @xmath493 is the coarsegrained mass density for a partitioning @xmath123 .
one can certainly study the stability of this distribution ( taken as an initial condition ) via the usual differential equations of hydrodynamics , even if this may require solutions in the weak ( or distribution ) sense .
the results will not be correct at length scales smaller than the scales @xmath123 , but at smaller scales ( down to @xmath494 ) one can increase @xmath39 ( decreasing the size of intervals @xmath232 ) and again study stability questions via a finer grained density of the form of ( [ eq : mass_density_coarsegrained ] ) . uniform densities on space filling supports in newtonian cosmology are unstable if the universe is open , stable if the euclidian manifold is a flat 3torus @xcite .
in other words , homogeneity is globally unstable in an open newtonian universe .
the standard model of cosmology ( @xcite , @xcite ) is a paradigm of simplicity : the simplest possible solution of einstein s field equations ( global integrability based upon global symmetry ) is combined with what feynman has called `` the usual initial conditions of physics '' : random or thermal initial conditions .
feynman pointed out that biologists , geologists , and astronomers know that the usual initial conditions of physics ( and integrable dynamics , one must add ) can not be used to explain most of the phenomena that are observed in nature ( @xcite ) .
cosmologically seen , far from equilibrium phenomena occur at relatively small scales .
that these nonequilibrium nonuniformities at small scales should be consistent with perfect symmetry and random initial conditions at the largest scales is not at all clear .
if it would be true then , as plato @xcite believed , the heavens would be perfect while all disorder is confined to the `` sphere of the earth '' ( extended a bit , out to 150 , at least ) .
it is not necessary to assume that the galaxy distribution requires the nineteenth century notion of randomness except perhaps as a sometimes convenient approximation to deterministic chaotic dynamics .
we now understand how even the simplest chaotic dynamical systems can generate all possible historgams that can be constructed empirically ( @xcite ) . by randomness we mean a breakdown of the space time description of cause and effect ( as in quantum mechanics ) . statistical independence ( @xcite ) , in contrast ,
is a separate idea that occurs in deterministic dynamics .
physics in the last twenty years has begun to follow more the path laid out by poincar ( @xcite ) , deviating from the traditional path set down by boltzmann ( contrast the traditional emphasis on randomness found in @xcite and in @xcite , for example , with the perspective on randomness expressed in @xcite ) .
a deterministic dynamical system ( like the differential equations that generate the characteristic curves of newtonian or einsteinian cosmology ) can generate inhomogeneity that need not be fractal .
a dynamical system far from thermodynamic equilibrium does not generate a unique probability distribution , but instead generates infinitely many different classes of distributions depending on classes of initial conditions .
we would have no way to discover `` the initial conditions of the universe '' other than by accurate backward integrations in time starting from the present ( empirically unknown ) distribution of matter via the correct equations of motion .
the laws of physics alone do not tell us anything about initial conditions ( @xcite ) . in the far from equilibrium case , on small scales , we know that mother nature has not chosen `` the usual initial conditions of physics '' ( trees do nt grow from thermal equilibrium initial conditions , but arise instead from strong driving combined with dissipation ) . even if we knew the present cosmologic
scale distribution of matter the question whether the global matter distribution could have been generated from a thermally equilibrated initial state can not be answered by n body simulations , because no existing computer can reproduce , in backward integration in time , even the first digits of those initial conditions after integrations forward in time over billions of years .
accurate backward integrations were actually accomplished , for unknown reasons and for a very restricted range of energies , by building a special computer to try to simulate the evolution of the solar system via a chaotic symplectic map over millions of years ( @xcite ) .
one object was to try to understand the initial conditions that initially fascinated kepler , the so - called titius `` law '' ( see also @xcite ) .
the characteristic curves of the partial differential equations of newtonian cosmology are generated by a dynamical system with a phase space of at least six dimensions ( three degrees of freedom ) . the lagrangian method @xcite studies characteristics via backward integration in time , using the fact that initial conditions are trivially conserved along streamlines .
chaotic dynamics requires only a three dimensional phase space .
complex dynamics , dynamics equivalent to a universal computer ( no scaling laws , no generating partition , nothing to aid in forecasting the future statistically ) may occur in certain newtonian systems with only three degrees of freedom ( @xcite ) . there , scaling laws , attractors , and the cardinality of strange sets can at best be defined only locally , if at all .
it is unknown whether ( any form of ) fluid turbulence or the newtonian three body problem fall into the complexity category . expecting the universe to be describable by a completely integrable dynamical system , even at the largest scale of coarsegraining , seems unlikely in the light of what we now understand about deterministic dynamics .
we are reminded that the cosmological principle is not demanded by any known laws of physics , and is not itself a separate law of nature .
this article was written with the benefit of hindsight .
the author belongs to the subset of physicists who , in the past , has claimed evidence for @xmath402 on the basis of log log plots that did satisfy the geilo criterion .
i am very grateful to martin kerscher for generating the data analysis in section [ sect : data ] , to herbert wagner and thomas buchert for criticism and discussions and to herbert wagner and universitt mnchen for guestfriendship during sebt . feb . of my 199697 sabbatical year where i had the chance to learn some cosmology . the focus necessary to carry this project through
was generated by the oct .
1996 schloringberg conference on cosmology .
thomas buchert made it possible for me to attend that conference , while the encouragement to enter into the debate over fractals in galaxy statistics came from herbert wagner .
we are grateful to bernard jones for some comments , and to francesco sylos labini and luciano pietronero for an exchange of ideas via email . | unless there is evidence for fractal scaling with a single exponent over distances @xmath0 then the widely accepted notion of scale invariance of the correlation integral for @xmath1 must be questioned .
the attempt to extract a scaling exponent @xmath2 from the correlation integral @xmath3 by plotting @xmath4 vs. @xmath5 is unreliable unless the underlying point set is approximately monofractal .
the extraction of a spectrum of generalized dimensions @xmath6 from a plot of the correlation integral generating function @xmath7 by a similar procedure is probably an indication that @xmath7 does not scale at all .
we explain these assertions after defining the term multifractal , mutually
inconsistent definitions having been confused together in the cosmology literature .
part of this confusion is traced to a misleading speculation made earlier in the dynamical systems theory literature , while other errors follow from confusing together entirely different definitions of `` multifractal '' from two different schools of thought .
most important are serious errors in data analysis that follow from taking for granted a largest term approximation that is inevitably advertised in the literature on both fractals and dynamical systems theory . |
the formation of a black hole is , in principle , one of the most efficient mechanisms for generation of gravitational waves .
such sources tie together two major research initiatives .
laser interferometric gravity wave detectors@xcite hold out a promise of the detection of gravitational waves from astrophysical events . to interpret the results of the gravitational wave signals , and to help find signals in the detector noise
, a broad and detailed knowledge will be needed of astrophysical gravitational waveforms .
this is one of the underlying motivations for the `` grand challenge''@xcite in high performance computing , aimed at computing the coalescence of black hole binaries .
evolving numerical spacetimes and extracting outgoing radiation waveforms is indeed a challenge . in a straightforward numerical approach ,
a good estimate of the asymptotic waveform requires long numerical evolutions so that the emitted waves can be propagated far from the source .
the necessary long evolutions are difficult for a number of reasons .
general difficulties include throat stretching when black holes form , numerical instabilities associated with curvilinear coordinate systems , and the effects of outer boundary conditions which are approximate.@xcite we suggest here that at least part of the cure for this problem may lie in the use of the theory and techniques of nonspherical perturbations of the schwarzschild spacetime ( `` nps '' ) . by this
we mean the techniques for treating spacetimes as deviations , first order in some smallness parameter , from the schwarzschild spacetime .
these techniques differ from `` linearized theory '' which treats perturbations of the spacetime from minkowski spacetime and which can not describe black holes .
the basic ideas and methods were set down by many authors and lead to `` wave equations '' for the even parity@xcite and odd parity@xcite perturbations .
nps has been used to compute outgoing radiation waveforms from a wide variety of black hole processes , including the scattering of waves@xcite , particles falling into a hole@xcite , and stellar collapse to form a hole@xcite .
the general scheme of nps also underlies the techniques for extraction of radiation from numerically evolved spacetimes@xcite .
nps computations have recently been used in conjunction with fully numerical evolution , as a code test@xcite and as a strong - field radiation extraction procedure@xcite .
here we are interested in another sort of application of nps theory . to understand such applications we consider an example : two very relativistic neutron stars falling into each other , coalescing and forming a horizon , as depicted in fig
the curve hypersurface , " in fig . 1 , indicates a spacelike initial " surface .
the spacetime can be divided into three regions by this initial surface and the horizon .
the early evolution , in region i , below the initial hypersurface , is highly dynamical and nonspherical .
spherical perturbation theory is clearly inapplicable . above the initial surface
the spacetime remains highly nonspherical in region ii inside the event horizon , but outside the event horizon , in region iii , it may be justified to consider the spacetime to be a perturbation of a schwarzschild spacetime .
this is essentially guaranteed if the initial hypersurface is chosen late enough , in some sense , after the formation of the horizon .
the evolution in region iii , then , is determined by cauchy data on the initial hypersurface exterior to the horizon .
it is important to note that this is made possible by the fact that the horizon is a causal boundary which shields the outer region from the dynamics of the highly nonspherical central region .
the scheme inherent in this division of spacetime has the potential greatly to increase the efficiency of the computation of the radiation generated when strong field sources form black holes . if one starts from the cauchy data on the initial hypersurface , one can evolve forward in time with the linear equations of perturbation theory .
many of the long - time evolution problems of numerical relativity are avoided and the interpretation of the computed fields in terms of radiation is immediate .
the approach suggested would then seem to be : use numerical relativity up to the initial hypersurface ; use the techniques of nonspherical perturbations in the future of the initial hypersurface .
in fact , the efficiency that can be achieved may be even greater . in the early , highly nonspherical , pre - initial hypersurface phase of the development of the spacetime
, there may be relatively little generation of gravitational radiation . by using a computational technique which suppresses the radiative degrees of freedom
one may be able to compute the early stages of evolution relatively easily .
there are two very recent examples of just such applications of this viewpoint .
price and pullin@xcite used as initial data the misners@xcite solution to the initial value equations for two momentarily stationary black holes .
abrahams and cook@xcite considered two holes moving towards each other , and used numerical values of the initial value equations . in neither case was there _ any _ use of fully nonlinear numerical evolution .
the rather remarkable success of both computations suggests that there is something robust about the underlying idea of separating horizon - forming astrophysical scenarios into an early phase with no radiation and a late phase with small deviations from sphericity outside the horizon .
it is plausible that the bulk of the radiation in most processes is generated only in the very strong - field interactions around the time of horizon formation and that radiation generation in the early dynamics can be ignored .
one would , however , think that strong radiation would be emitted during the stages at which the early horizon is very nonspherical and at which time nonspherical perturbation theory would seem to be inapplicable .
there should be a tendency for this `` early '' radiation , produced very close to the horizon , to go inward into the developing black hole , so that the application of nonspherical perturbation theory to the exterior really requires that on the initial spacetime the perturbation are small only well outside the horizon .
it would seem that something of this sort would have to be happening to explain the accuracy of the price - pullin and abrahams - cook results .
whether or not many problems can be treated with _ no _ use of fully numerical evolution , it appears clear to us that these perturbation methods will be applied to a variety of problems in which data on the initial hypersurface is available numerically .
the primary purpose of this paper is to provide justification and background for earlier work on this subject and a clear recipe for future applications . in the next section
we discuss the meaning , and limitations , of extracting a `` perturbation '' from this numerical data and computing radiated energies .
the explicit process of extracting the perturbations from the numerical data is given in sec .
iii . in sec .
iv we demonstrate the use of this procedure via application to a specific example , the misner initial data .
we outline here the formalism for perturbation theory based on work by regge and wheeler@xcite and by zerilli@xcite , but we will draw heavily on the gauge invariant reformulation of those earlier works by moncrief@xcite .
our starting point is an initial hypersurface which can be taken as a surface of constant schwarzschild time .
we assume that the coordinates @xmath0 on that surface are almost schwarzschild coordinates @xmath1 and we assume that the values are known , on this hypersurface and in these coordinates , for the 3-metric @xmath2 and the extrinsic curvature @xmath3 .
the conditions for finding such a hypersurface and such coordinates will be made explicit in sec .
underlying perturbation theory is the idea of a family of metric functions @xmath4 , depending on the parameter @xmath5 , which satisfy the einstein equations for all @xmath5 , and which , in the limit @xmath6 , become the schwarzschild metric functions , such as @xmath7 .
( here @xmath8 and @xmath9 is the mass of the schwarzschild spacetime ; we use units throughout in which @xmath10 . )
nps theory amounts to the approximation @xmath11 it is of some practical importance to realize that the choice of the expansion parameter can have a considerable effect on the range over which perturbation theory gives a good approximation .
let us imagine that we introduce a new parameter @xmath12 which is a function of @xmath5 such that @xmath13 approaches unity as @xmath6 .
if we take @xmath12 to be the basis of our perturbation approach , the approximation becomes @xmath14 \left\{\frac{\epsilon'}{\epsilon}\right\}+{\cal o}(\epsilon'^2)\ .\label{prime}\end{aligned}\ ] ] at @xmath15 the derivative of @xmath16 with respect to @xmath5 and with respect to @xmath12 have the same values , so for a given spacetime that is , for a given value of @xmath5 the nonspherical perturbation in ( [ prime ] ) differs from that in ( [ pertbasic ] ) by the factor @xmath17 .
computed energies ( which are quadratic in the nonspherical perturbations ) will differ by the square of this ratio .
different choices of parameterization will change this factor and affect the accuracy of the linearized approximation . to show the effects of this parameterization dependence , we take as an example misner data @xcite@xcite for two holes .
the initial separation of the holes , in units of the mass of the spacetime , is described by misner s parameter @xmath18 .
the metric perturbations , however , are not analytic in @xmath18 as @xmath19 , so @xmath18 can not be used as the expansion parameter in ( [ pertbasic ] ) .
the actual expansion parameter used by price and pullin , was a function of @xmath18 denoted @xmath20 .
we consider here what would be the results of perturbation theory done with the expansion parameter @xmath21 figure 2 shows the results , along with the energies computed by numerical relativity applied to full nonlinear evolution @xcite . for all choices of @xmath22
the agreement between perturbation theory and numerical relativity is good at sufficiently small initial separation ( sufficiently small @xmath18 ) , but as @xmath18 grows larger , the agreement increasingly depends on the which parameterization is used .
the @xmath23 parameterization , the parameter of the price - pullin paper , is a reasonably good approximation even up to separations ( @xmath24 ) for which the initial apparent horizon consists of two disjoint parts . for positive values of @xmath22
the agreement is less impressive , while for @xmath25 , it appears that perturbation theory is giving excellent answers for initial data that are very nonspherical .
clearly the @xmath25 parameterization is `` better , '' at least for the purpose of computing radiated energy .
there exist yet better choices ; in principle a parameterization could be found for which the energy computed by linearized theory is perfect for any initial separation .
the crucial point is that we have no _ a priori _ way of choosing what is and what is not a good parameterization .
the choice of expansion parameter @xmath20 was made in the price - pullin analysis , because it occurred naturally in the mathematical expressions for the initial geometry .
there was no _ a priori _ reason for believing it to be a particularly good , or particularly bad parameterization .
this point will be discussed again , in connection with numerical results presented in sec .
the fact , demonstrated in fig . 2 , that the results of linear perturbation theory are arbitrary may seem to suggest that perturbation answers , from a formal expansion or numerical initial data , are of little value .
it should be realized , however , that the arbitrariness exhibited in fig . 2 is simply a demonstration of the fact that linearized perturbation results are uncertain to second order in the expansion parameter .
the fact that the results for different parameterizations start to differ from each other around @xmath26 simply signals that @xmath20 is around unity .
( in fact , @xmath27 for @xmath28 . )
higher order uncertainty is an unavoidable feature in the range where the expansion parameter is of order unity .
but there is a potential misunderstanding about the meaning of `` expansion parameter around unity . ''
to see this consider a change to a new expansion parameter @xmath29 .
the new expansion parameter @xmath5 is of order unity for @xmath30 , yet we know that perturbation fails dramatically for such a large value of @xmath18 .
the issue here is that we need some way of ascribing an appropriate `` normalization '' to the expansion parameter .
a sign that the normalization is good is that physically - based measures of distortion start getting large for @xmath5 around unity .
if we had reliable measures of this type then we could have some confidence about the range of the the expansion parameter for which we could neglect second order uncertainty , whether due to parameter arbitrariness or the omission of higher order terms in the calculation .
one can formulate interesting measures for the normalization of the expansion parameter , such as the extent to which the linearized initial conditions violates the exact hamiltonian constraint @xcite .
most such measures are useful only for finding a very rough normalization for @xmath20 ( equivalently , for roughly finding the range in which linearized perturbation theory is reliable ) .
the only reliable procedure for this is to carry out computations of radiated waveforms and energy to second order in the expansion parameter .
the ratio of second order corrections to first order results gives the only direct measure of the reliability of perturbation results .
if one computes an energy for which the second order correction to the first order result is 10% , then one knows that the third order correction ( due to a change in parameterization or an inclusion of third order terms in the computation ) will be on the order of 1% .
we turn now to the central question of this paper : how does one apply perturbation theory to numerically generated initial data ? to do this we consider our numerical initial data to be initial data for a solution in a parameterized family @xmath4 corresponding to @xmath31 .
the application of perturbation theory is equivalent to replacing @xmath32 by @xmath33 an added familiar complication is that we can introduce a family of coordinate transformations @xmath34 which reduces to @xmath35 for @xmath6 .
such a transformation takes the original family to a new family @xmath36 , which satisfies the same requirements as the original family .
we follow moncrief@xcite in constructing , from the 3-metric @xmath2 on constant-@xmath37 surfaces , quantities @xmath38 , which are invariant to first order in @xmath5 ( `` gauge invariant '' ) , for coordinate transformations .
the construction of these moncrief @xmath38 is done in two steps .
first , the multipole moments of the metric are extracted . in practice
this is done by multiplying the metric functions by certain angular factors and integrating over angles .
since we are only interested in quadrupole and higher order for radiation , this step also eliminates the spherically symmetric background parts of the metric function .
the second step is to form linear combinations of these multipoles and of their derivatives with respect to radius .
we symbolically represent the process of forming these quantities as @xmath39 here the symbol `` @xmath40 '' represents the process of multiplying by angular functions and integrating , then multiplying by certain functions of @xmath41 and taking linear combinations of the results .
( our notation here disagrees with that of moncrief@xcite in a potentially confusing way .
moncrief s perturbation quantities are independent of the size of @xmath5 . in order to have definitions that can be applied to numerical data we use quantities that to first order are proportional to @xmath5 . )
the moncrief gauge invariants play two different roles . for even parity one of the gauge invariants , @xmath42 ,
is a constraint ; it vanishes in linearized theory as a result of the initial value equations .
in linearized theory , the remaining moncrief quantities , denoted @xmath43 here , satisfy wave equations @xmath44 , the regge - wheeler equation in odd parity and zerilli equation in even parity . from our numerical data
we construct the quantities @xmath38 precisely according to ( [ monconstr ] ) .
our numerically constructed `` perturbation '' quantities will not be invariant under coordinate transformations , but rather will transform as @xmath45 . similarly , the linearized constraint , @xmath42 will not vanish , but
will be of order @xmath46 .
the numerically constructed wavefunctions @xmath43 will satisfy @xmath47 , where @xmath48 is the regge - wheeler or zerilli wave operators . the use of nps methods is equivalent to ignoring the second order terms in the wave equations .
the wavefunction @xmath38 can then be propagated from the initial hypersurface forward and the radiation waveforms extracted from it . to evolve @xmath43 off the initial hypersurface , however , requires the initial time derivative @xmath49 .
this can be computed from the initial extrinsic curvature , but some care is needed
. indeed , the possible ambiguities that arise here are the justification for the somewhat protracted discussion in this section . if @xmath50 is the future - directed unit normal to the initial hypersurface then the rate at which the 3-metric is changing is given by @xmath51 where @xmath3 is the extrinsic curvature and @xmath52 is the lie derivative along the unit normal .
the unit normal is related to the derivative with respect to schwarzschild time by @xmath53 .
the time derivative of the moncrief function then can be written @xmath54@xmath55 to evaluate the right hand side we need to know how @xmath56 changes when it is lie dragged by @xmath50 .
since @xmath56 depends only on @xmath2 it might appear that one need only lie drag @xmath2 to find the change in @xmath56 , and that @xmath57 .
from this it would follow that @xmath58 .
it is important to note that this is _ not _ the correct relationship between @xmath3 and the cauchy data for the wave equation .
the fallacy in this procedure lies in the fact that @xmath43 must be computed from the 3-metric on a slice for which schwarzschild time is constant ( to first order in @xmath59 ) .
lie dragging by @xmath50 moves the 3-metric to a surface that is not ( to first order ) a constant time surface .
the cure is clearly to compare quantities on surfaces of constant @xmath37 by using @xmath60 .
it is the schwarzschild time derivative that commutes with the schwarzschild radial derivative @xmath61 .
the correct prescription then follows from @xmath62@xmath63 @xmath64 we note that the perturbed schwarzschild metric does have a shift vector @xmath65 of order @xmath5 , and in principle the shift vector influences the time development of @xmath2 according to @xmath66 , where @xmath67 is a time coordinate in which the shift vector vanishes . but the shift vector can be considered to be `` pure gauge . ''
it is necessary if one wants a complete specification of the coordinates and the metric components , but its value is a matter of choice , and is not necessary for a complete specification of the physics .
the initial value , and evolution , of the gauge invariant quantity @xmath43 is invariant with respect to the choice of @xmath65 , and @xmath43 carries all the ( physically meaningful ) information about gravitational waves . the evaluation of @xmath43 from ( [ monconstr ] ) and @xmath49 from ( [ qdot ] ) completes the extraction , from the numerical data for @xmath68 of the cauchy data for the regge - wheeler or zerilli wave equation .
an alternative procedure arises if one uses the scalar wave - equations derived from the perturbative reduction of the nonlinear wave - equation for the extrinsic curvature which arises in a new explicitly hyperbolic form of the einstein equations@xcite . in this system ,
the scalar wave equations are one order lower in time derivative from the usual regge - wheeler and zerilli equations , so the cauchy data consists of the extrinsic curvature and its time - derivative ( which involves the 3-dimensional ricci curvature )
. from the above it is clear that linearized evolution should give good accuracy when applied to numerically generated initial data with sufficiently small deviations from sphericity . for initial data which are known in analytic form one can , of course , apply linearized theory even to cases in which initial deviations from sphericity are only marginally small . the results in fig . 2 , for example , show that the results of such application of perturbation theory give reasonable accuracy for values of @xmath18 at which an initial horizon is highly distorted .
it is worrisome to apply linearized evolution to marginally nonspherical initial data , which do not , for example , satisfy the constraint @xmath69 with reasonable accuracy .
such a procedure linear evolution of nonlinear initial data has , among other disadvantages , no clear theoretical framework .
we wish to point out here that nps methods can be used more broadly , and a procedure we call `` forced linearization '' can be applied to numerically generated initial data in a way that amounts to extracting the linearized part of the data and evolving linearly .
this procedure circumvents the difficulty of performing formal linearization to data which is only known numerically .
we imagine that we start with an initial value problem in which there is some adjustable parameter , call it @xmath70 , such that @xmath71 corresponds to the schwarzschild initial data .
there is no requirement that the family of solutions @xmath72 be analytic in @xmath70 at @xmath73 .
there may be additional parameters , call them @xmath74 , such as the parameters governing the initial momenta of holes .
to apply forced linearization we fix the values of the @xmath74 and make a choice of @xmath70 such that the computed initial data @xmath75 are `` very nearly spherical . ''
one criterion for this would be that @xmath42 is very small .
we then interpret this initial data as being essentially linearized data , to which the approximation in ( [ eq.linnum ] ) applies .
we extract multipoles , form a gauge invariant wave function @xmath43 , and evolve it with the zerilli or regge - wheeler equation , all as described above . the result of this
will be a late - time waveform @xmath76 and the energy @xmath77 that it carries .
the next step is to characterize the results with a well behaved gauge invariant parameter . to do this we choose some fiducial radius @xmath78 , and evaluate @xmath79 the gauge invariant wave function of the initial hypersurface at this radius .
next , we leave the @xmath74 unchanged , but choose a larger value of @xmath70 for which the numerically generated initial data set @xmath80 is `` marginal '' in that it corresponds to deviations from sphericity large enough so that it differs significantly form linearized initial conditions ; one sign of this would be that the condition @xmath69 is significantly violated . for this data set we go through the same procedure as above in characterizing the data set by a parameter @xmath81 . for this marginally spherical initial data we take the solution for the wavefunction and energy to be @xmath82 the idea underlying
this method is that the very nearly spherical data give us the solution for for @xmath83 . for the marginal initial data
set we then need only multiply this initial data by the appropriate factor telling us how much larger is the linear part of the nonsphericity than that of the very nearly spherical initial data .
the success of forced linearization requires then that @xmath5 evaluated at @xmath84 be a well behaved parameterization of the linearized part of the nonsphericity in the numerical data .
since our expansion parameter @xmath5 is the magnitude of the perturbation , it will be a good expansion parameter as long as it is evaluated in a region where the nonlinear deviations from sphericity are small , i.e. , where ( [ eq.linnum ] ) is a good approximation .
for this reason it is important that @xmath84 be chosen fairly large . for processes of the type pictured in fig . 1 , the deviations from sphericity fall off quickly in radius , so that at large enough @xmath41 one can be certain that the initial data are an excellent approximation to linearized data .
evidence for this is that the violations of the @xmath69 constraint are always confined to small radii .
one easily implemented check on the forced linearization procedure is to look at the factor @xmath85 and confirm that it is independent of @xmath41 for @xmath86 . in section iii we show that this test is easily passed by a numerical example , and that the results of forced linearization are essentially the same as those of formal linearized theory .
here we assume that the reader has numerical solutions for the 3-metric on an approximately t = const surface .
the first step in applying nps to numerical results is to transform to coordinates which are `` almost schwarzschild '' coordinates .
it is assumed that the numerical @xmath2 and @xmath3 are expressed in a coordinate system @xmath87 in which the approximate spherical symmetry is manifest .
this means that @xmath3 and ratios like @xmath88 must be small .
they all are , in fact , formally of order @xmath59 , so if they are not all reasonably small compared to unity there is little reason to think that nps will work .
a schwarzchild - like areal radial coordinate @xmath41 needs to be introduced .
this can be defined as a function of @xmath89 by @xmath90 where the integral is taken on a surface of constant @xmath89 .
the metric component @xmath91 , in terms of this quantity , gives us another test of how close the geometry is to that of a constant time schwarzschild slice .
the quantity @xmath92 should be nearly equal to the constant @xmath93 , where @xmath9 is the mass of the spacetime . the variability of this quantity in @xmath94 , and @xmath95 , is formally of order @xmath59 .
there are , of course , other ways of specifying the schwarzschild - like coordinates .
we could , for example , have defined @xmath96 all these coordinate choices , however , should agree to order @xmath59 and are therefore equivalent within a linearized gauge transformation . to compute the gauge invariant perturbation functions ,
we first assume that an @xmath97 multipole of the 3-metric may be expanded as @xmath98 where , for clarity , we have suppressed multipole indices and have replaced moncrief s @xmath99 and @xmath100 odd parity perturbation functions with @xmath101 , @xmath102 . the multipole moments @xmath103 and @xmath104 are computed by projection onto the relevant spherical harmonics which can be found in moncrief@xcite .
explicit formulas for the important special case of even parity , axisymmetric perturbations may be found in ref .
@xcite . for odd parity perturbations
, one function can be constructed from the amplitudes @xmath101 and @xmath102 which is gauge invariant and satisfies the regge - wheeler equation ( below ) , @xmath105 { s \over r}.\ ] ] the situation for even parity perturbations is more complicated .
two gauge invariant functions may be formed out of the multipole moments : @xmath106\end{aligned}\ ] ] from @xmath107 and @xmath108 it is possible to form two new functions , one which is radiative and one which is equivalent to the perturbed hamiltonian constraint @xmath109 + \ell(\ell+1 ) [ 2 s k_2 + ( 1-m/\{rs\ } ) k_1 ] .\end{aligned}\ ] ] the scaled function @xmath110 with @xmath111 satisfies the zerilli equation ( below ) .
the time derivatives of the radiative gauge invariant functions @xmath112 and @xmath113 are found by substituting @xmath114 for @xmath2 in the multipole moment computation and forming the same combinations of moments .
the wavefunctions @xmath112 and @xmath115 obey the regge - wheeler and zerilli wave equations respectively : @xmath116 where the wave operator appropriate to schwarzschild spacetime is @xmath117 in terms of the `` tortoise coordinate '' @xmath118 , and where the potentials are given by @xmath119\ ] ] and , @xmath120.\ ] ] once the zerilli and regge - wheeler equations are integrated for all the desired @xmath121 and @xmath122 modes , the total radiated energy can be calculated from the asymptotic timeseries for @xmath123 and @xmath124 : @xmath125
in this section we demonstrate the extraction of a perturbation from a numerical solution to the nonlinear constraint equations the misner data representing two black holes at a moment of time symmetry .
the misner 3-geometry may be written @xcite as @xmath126 for this exercise , we pretend that the initial geometry is known only numerically , so no explicit formal linearization can be done .
the odd parity perturbations vanish in the misner solution .
we compute the even parity gauge invariant wavefunction for @xmath131 using numerical evaluations of ( [ eq : miscon ] ) - ( [ eq : miskap ] ) .
specifically , we compute @xmath132 and @xmath133 of ( [ mncrform ] ) from @xmath134 all the other moments in ( [ mncrform ] ) vanish for the conformally schwarzschild metric of ( [ eq : mismet ] ) .
the function @xmath135 is evaluated at values of @xmath41 corresponding to the range @xmath136 to @xmath137 .
the initial value of @xmath135 ( along with its time - derivative which is zero for the misner time - symmetric initial data ) provides initial values for integration of ( [ eq : zer ] ) . at large radius , @xmath138
, the value of @xmath139 is used in ( [ eq : power ] ) to compute the radiated energy .
first , in fig . [ fig.misnl ] we show the result of directly computing the gauge invariant function @xmath135 from the nonlinear initial data , integrating the zerilli equation , and computing the radiated energy . for small values of @xmath18
the agreement with the explicitly linearized data of ref .
@xcite is excellent . at
about @xmath140 the agreement breaks down and the qualitative behavior becomes dramatically different .
it is interesting to note that the apparent horizon encompassing both black holes does not exist for @xmath141 , close to the dramatic reversal in the energy curve . in fig .
[ fig.miscon ] the violation of the linearized constraint by the nonlinear data is shown as a function of radius .
we plot the ratio of the constrained gauge invariant function , @xmath43 to the radiative function @xmath42 scaled in such a way as to compensate for large violation at @xmath142 .
the value of @xmath42 clearly grows much faster than the radiative variable @xmath43 as the separation is increased . as discussed in sec .
ii , it is possible to obtain the results of formal perturbation theory directly from the numerical data without ever making reference to the analytic solution . in fig .
[ fig.misfl ] we demonstrate the application of the forced linearization procedure to the nonlinear misner data for various values of the fiducial radius @xmath84 .
for very small values of @xmath18 , such as @xmath143 , the geometry outside the event horizon is everywhere well approximated by ( [ eq.linnum ] ) and forced linearization works even for small values of @xmath144 .
when @xmath18 is larger than around @xmath145 , on the other hand , the initial geometry near the horizon contains significant nonlinear effects , and large values of @xmath144 must be used to get results equivalent to those of formal linearized theory . as @xmath84 gets large ,
the results become indistinguishable from those of formal perturbation theory reported in ref .
@xcite . for @xmath146
the difference in radiated energy for @xmath147 is less than @xmath148 .
this high - accuracy equivalence deserves some explanation .
in particular , why is forced linearization equivalent to formal linearization with expansion parameter @xmath20 ?
why is that expansion parameter singled out ?
the equivalence is a result of two features of the way in which the linearizations were done : first , both the formal linearization of ref .
@xcite , and the forced linearization results in fig .
[ fig.misfl ] , use precisely the same coordinates .
( the forced linearization results , in fact , are not based on initial values that were generated by genuinely numerical means . rather , the closed form solutions for the misner metric functions were used .
the `` almost - schwarzschild '' coordinates of the forced linearization , were precisely the same as the `` almost - schwarzschild '' coordinates in ref .
@xcite ) . secondly ,
in the `` almost - schwarzschild '' coordinate system , the parameter @xmath20 is , to all perturbation orders , the coefficient of the dominant nonsphericity at large radius .
forced linearization ( in the limit of large @xmath84 ) results in a parameterization based on a gauge invariant measure of nonsphericity at large radius .
it therefore must be proportional to @xmath20 and produce results equivalent to those of the formal linearization of ref .
@xcite , in which @xmath20 was the expansion parameter .
it should be understood that this does not imply that the parameter @xmath20 is _ physically _ singled out .
a first order change in the `` almost - schwarzschild '' coordinates will change the coefficient of the dominant large - radius nonsphericity .
we might , for example , transform from the `` almost - schwarzschild '' radial coordinate @xmath41 of ( [ eq : mismet ] ) to a new coordinate @xmath149 $ ] . in this case
the coefficient of the leading large @xmath150 term in the metric will be @xmath151 , and the results of forced linearization with the resulting `` numerical '' data will differ , when perturbations are large , from the results in ref .
the forced linearization will have induced an expansion parameter different from @xmath20 .
a. a. abramovici _ et al .
_ , science * 256 * , 325 ( 1992 ) ; k. s. thorne , submitted to _
proceedings of snowmass 94 summer study on particle and nuclear astrophysics and cosmology _ , eds . w. w. kolb and r. peccei ( world scientific , singapore ) .
m. davis , r. ruffini , w. h. press , and r. h. price , phys .
* 27 * , 1466 ( 1971 ) ; l. petrich , s. shapiro , and i. wasserman , astrophys .
j. suppl .
* 58 * , 297(1985 ) .
s.l . detweiler and e. szedenits , astrophys.j . *
231 * , 211 ( 1979 ) | nonspherical perturbation theory has been necessary to understand the meaning of radiation in spacetimes generated through fully nonlinear numerical relativity .
recently , perturbation techniques have been found to be successful for the time evolution of initial data found by nonlinear methods .
anticipating that such an approach will prove useful in a variety of problems , we give here both the practical steps , and a discussion of the underlying theory , for taking numerically generated data on an initial hypersurface as initial value data and extracting data that can be considered to be nonspherical perturbations . |
long - range rapidity correlations between pairs of hadrons produced at small azimuthal angles with respect to each other were discovered recently in heavy ion ( @xmath3 ) @xcite , proton proton ( @xmath4 ) @xcite , and proton nucleus ( @xmath5 ) collisions @xcite . due to the particular shape of the corresponding correlation function , with a narrow correlation in the azimuthal angle @xmath2 and a wide correlation in pseudo - rapidity separation @xmath6 , these correlations are often referred to as the `` ridge '' .
there appears to be a consensus in the community that the origin of these long - range rapidity correlations is in the very early - time dynamics immediately following the collision .
a simple causality argument demonstrates that a correlation between two hadrons produced far apart in rapidity may arise only in their common causal past , that is , in the early stages of the collision @xcite .
however , the detailed dynamical origin of these `` ridge '' correlations is not completely clear .
it has been proposed in the literature @xcite that the `` ridge '' correlations may arise in the classical gluon field dynamics of the parton saturation physics / color glass condensate ( cgc ) .
( for reviews of saturation / cgc physics see @xcite . )
indeed classical gluon fields , which in the mclerran venugopalan ( mv ) model @xcite dominate gluon production in heavy ion collisions , do lead to a rapidity - independent distribution of the produced gluons @xcite over rapidity intervals of up to @xmath7 , which is the upper limit of their validity ( with @xmath8 the strong coupling constant ) . correlations between such classical fields , introduced in the process of averaging over their color sources , do have a long range in rapidity @xcite .
moreover , it was observed in @xcite that the diagrams giving rise to such rapidity correlations also lead to a narrow correlation in the azimuthal direction , in qualitative agreement with the shape of the `` ridge '' correlation .
one has to keep in mind that in heavy ion collisions the early - stage azimuthal correlation may be washed out by the final state interactions leading to thermalization of the produced medium and its hydrodynamic evolution .
( note though , that the rapidity correlation is not likely to be strongly affected by such late - time dynamics . )
it was argued , however , that the effect of the radial flow in the hydrodynamic evolution of the quark - gluon plasma ( qgp ) would be to ( re-)introduce the azimuthal correlations @xcite . another potential complication with the cgc explanation of the `` ridge '' is the fact that rapidity - dependent corrections to classical gluon fields do become important at rapidity of the order of @xmath9 .
these corrections come in through the non - linear balitsky kovchegov ( bk ) @xcite and jalilian - marian iancu mclerran weigert leonidov kovner ( jimwlk ) @xcite evolution equations .
rapidity - dependent corrections are very important for describing the hadron multiplicity distribution in rapidity , @xmath10 , in the cgc framework @xcite .
as @xmath10 does depend on rapidity rather strongly in rhic heavy ion data , and also strongly ( albeit less so ) in the lhc data , it is very hard to describe without the rapidity - dependent nonlinear evolution .
it is possible that similar rapidity - dependent corrections may significantly affect and potentially destroy the long - range structure of the rapidity correlations due to classical gluon fields of the mv model .
note that progress on this issue has been made in @xcite , indicating that inclusion of small-@xmath11 evolution still leaves the `` ridge '' reasonably flat in rapidity until @xmath12 when the correlation disappears . to elucidate the above questions and concerns , and to improve the precision of the cgc predictions for the `` ridge''-like correlations , it is important to be able to calculate the two - particle correlation in the cgc framework beyond the lowest order . while some works do consider the role of small-@xmath11 evolution and multiple rescatterings in the correlation function @xcite , most of the phenomenological approaches @xcite simply include the saturation effects into the lowest - order calculation by evolving the unintegrated gluon distributions with the running - coupling bk ( rcbk ) nonlinear evolution @xcite .
the aim of this paper is to begin to analytically include saturation effects into the two - gluon correlation function in nucleus nucleus collisions
. indeed full inclusion analytic of saturation effects originating in both nuclei would be a very hard problem : even the single gluon production in the quasi - classical mv limit of @xmath3 collisions can be dealt with only numerically at present @xcite . to make the problem more tractable we assume that one of the colliding nuclei is much larger than the other one , such that saturation effects are important only in interactions with the larger nucleus . in a more formal language @xcite for the quasi - classical mv model , we assume that @xmath13 , where @xmath14 and @xmath15 are the atomic numbers of the smaller and larger nuclei , while @xmath16 and @xmath17
. equivalently one can say that we are interested in production of gluons with transverse momentum @xmath18 , where @xmath19 is the saturation scale of the smaller nucleus .
we do not impose any constraints on @xmath20 compared to the saturation scale @xmath21 of the larger nucleus ( @xmath22 ) .
we stress here that this setup is not what is usually referred to as the @xmath5 collision in the saturation / cgc terminology : since @xmath23 , the two gluon production cross section is dominated by the gluons produced in interactions of _ different _ nucleons in the smaller nucleus , which may also be viewed as a lowest - order saturation correction .
below we calculate the two - gluon production cross section and the corresponding correlation function including the all - order saturation effects in the larger nucleus , while keeping them to the lowest non - trivial order in the smaller nucleus .
the paper is structured as follows . in sec .
[ sec : disc ] we set up the problem . in the process of properly defining the correlation function
we unexpectedly find a correlation originating in a simple fact that the nucleons in both nuclei are confined to within the nuclear radii .
the effect is quite generic , and is not specific to the saturation / cgc approach used in the calculations in much of this paper : this correlation should be present in any model of @xmath3 collisions which properly takes into account the geometry of the collision .
we refer to these correlations as the `` geometric correlations '' .
we argue that this effect gives a non - trivial contribution to the correlator even if the two - gluon production cross section is given by disconnected feynman diagrams , i.e. , the correlation is not generated dynamically in the interactions .
this correlation may be long - range in rapidity , though it does not have any non - trivial azimuthal structure .
the two - gluon production cross section in a heavy - light ion collision is calculated in sec .
[ sec : longrange ] in terms of the wilson line correlators .
we show that the two - gluon production cross section is related to various correlators of four adjoint wilson lines : the adjoint dipole , quadrupole , and a double - trace operator .
these correlators contain both the multiple rescatterings of the quasi - classical mv approximation , along with the small-@xmath11 bk / jimwlk evolution .
we study the long - range rapidity correlations in sec .
[ sec : lrrc ] .
we evaluate the wilson - line correlators describing the interaction with the target nucleus using the quasi - classical mv approximation and the large-@xmath24 limit . by analyzing the two - gluon correlation function at the lowest non - trivial order , and in agreement with the earlier calculations in the literature
@xcite , we find both the near - side ( @xmath1 ) and away - side ( @xmath25 ) long - range rapidity correlations .
( here @xmath2 is the azimuthal angle between the momenta of the produced gluons . )
the two correlations are identical in their azimuthal shapes ( as functions of @xmath2 ) , such that the correlator can be expanded into a fourier series in terms of only the even harmonics @xmath26 .
we discuss the possibility that such non - flow correlation may complicate experimental extraction of the contribution of the true qgp flow to the flow observables @xmath27 .
in addition to the geometric and long - range rapidity correlations , the obtained two - gluon production cross section contains hanbury - brown twiss ( hbt ) correlations @xcite .
a somewhat peculiar feature of this hbt correlation is that it is accompanied by an identical back - to - back peak as well .
we conclude in sec .
[ sec : conc ] by stressing the difference between near - side and away - side correlations calculated here and the mini - jet correlations : the near - side correlations in this work are long - range in rapidity , while the jet near - side correlations are local in rapidity .
following a standard approach used in experimental analyses of particle correlations @xcite the correlation function can be defined as @xmath28 where @xmath29 and @xmath30 are the single- and double - particle multiplicity distributions with @xmath31 the net inelastic nucleus
nucleus scattering cross section .
two - dimensional transverse vectors are denoted by @xmath32 with their length @xmath33 .
the normalization factor @xmath34 in is fixed by requiring that the number of particle pairs measured in the same ( `` real '' ) event @xmath35 is equal to the number of ( `` mixed '' ) pairs with particles coming from different events @xmath36 .
here we are interested in @xmath6-@xmath2 correlations with @xmath37 the difference between the azimuthal angles @xmath38 of the two momenta of the produced particles , and @xmath39 the difference between the pseudo - rapidities @xmath40 of the two particles .
for such a correlation , with the magnitudes of the transverse momenta @xmath41 and @xmath42 constrained to some chosen data bins , the normalization factor is fixed by @xmath43 ( we assume for simplicity that the number of produced particles is very large , @xmath44 , such that @xmath45 and @xmath31 is the same in both eqs . and
since the production cross section of producing exactly one particle is negligible . ) combining eqs .
, , , and we rewrite the correlation function in terms of cross sections as @xmath46 } { \left [ \int d \phi_1 \
, dy_1 \ , d \phi_2 \ , dy_2 \ , \frac{d \sigma}{d^2 k_1 dy_1 \ , d^2 k_2 dy_2 } \right ] } \ \frac{\frac{d \sigma}{d^2 k_1 dy_1 \ , d^2 k_2 dy_2 } } { \frac{d \sigma}{d^2 k_1 d y_1 } \ , \frac{d \sigma}{d^2 k_2 d y_2 } } - 1.\end{aligned}\ ] ] as outlined in the introduction , we take one of the nuclei ( the projectile ) to be much smaller than the other , @xmath47 , such that saturation effects in it are minimal and nucleons in this ( `` first '' ) nucleus interact with the other ( `` second '' ) nucleus independent of each other . at the same time , the projectile nucleus is still large enough , @xmath23 , such that the two - gluon production is dominated by the gluons generated by the interactions of two different nucleons from the first ( projectile ) nucleus with the second nucleus .
( this interaction with the two nucleons in the first nucleus can be considered a lowest - order saturation effect . )
denote the 1- and 2-nucleon wave functions of the projectile nucleus @xmath14 by @xmath48 and @xmath49 : they are normalized such that @xmath50 ) with a larger target nucleus ( @xmath15 ) .
the two smaller circles represent two nucleons in the nucleus @xmath14 ( see text for details).,width=188 ] with the help of these wave functions the single- and double - gluon production cross sections can be written as [ nna ] @xmath51 where @xmath52 is the impact parameter between the two nuclei and @xmath53 , @xmath54 , @xmath55 the transverse positions of the nucleons in the projectile nucleus , all measured with respect to the center of the second ( target ) nucleus , as shown in for the two - gluon production process .
( transverse vector @xmath56 labels the position of the incoming nucleon in the single gluon production case , while @xmath57 and @xmath55 label positions of projectile nucleons for two - gluon production . ) here @xmath58 is the cross section for the gluon production ( with fixed transverse momentum @xmath59 , rapidity @xmath60 , and transverse position @xmath56 ) in the collision of a nucleon ( @xmath61 ) with the target nucleus .
the angle brackets @xmath62 in denote averaging in the target nucleus wave function along with summation over all the nucleons in the target nucleus @xcite .
eqs . can be used in to give us the two - gluon correlation function in the heavy - light ion collision @xmath63 \ , \left [ \int d^2 b \
, d^2 b_2 \ , d \phi_2 \ , dy_2 \ , |\psi_i ( { \bm b } - { \bm b}_2)|^2 \ , \langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ ,
dy_2 \ , d^2 b_2 } \rangle \right]}{\left [ \int d^2 b \
, d^2 b_1 \ , d^2 b_2 \ , d \phi_1 \ , dy_1 \ , d \phi_2 \ , dy_2 \ , |\psi_{ii } ( \bm b - \bm b_1 , \bm b - \bm b_2 ) |^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , dy_1 \ , d^2 b_1 } \ , \frac{d \sigma^{p a_2}}{d^2 k_2 \ , dy_2 \ , d^2 b_2 } \right\rangle \right ] } \notag \\ & \times \ , \frac{\int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , |\psi_{ii } ( \bm b - \bm b_1 , \bm b - \bm b_2 ) |^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 dy_1 d^2 b_1 } \ , \frac{d \sigma^{p a_2}}{d^2 k_2 dy_2 d^2 b_2 } \right\rangle } { \left [ \int d^2 b \ , d^2 b_1
, |\psi_i ( { \bm b } - { \bm b}_1)|^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , d y_1 \ , d^2 b_1 } \right\rangle \right ] \ , \left [ \int d^2 b \ , d^2 b_2 \ , |\psi_i ( { \bm b } - { \bm b}_2)|^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , d y_2 \ , d^2 b_2 } \right\rangle \right ] } -1.\end{aligned}\ ] ] here we model the nucleus as a bag of independent nucleons , which is a correct description at the leading order in the atomic number @xmath64 @xcite . in such case
the single - nucleon light - cone wave function squared is simply equal to the nuclear profile function in the projectile nucleus , @xmath65 is defined by the integral over the longitudinal coordinate @xmath66 , @xmath67 such that for a spherical nucleus of radius @xmath68 and constant density @xmath69 it is @xmath70 . ) without any loss of generality we can perform the integral over @xmath71 in obtaining @xmath72 for a sufficiently large projectile nucleus , @xmath23 , one can assume that the two - nucleon wave function can be factorized , @xmath73 such that , with the help of we can write @xmath74 using this in we get @xmath75 substituting eqs . and into yields @xmath76 \ , \left[\int d^2 b_2 \ , d \phi_2 \ , d y_2 \ , \langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , dy_2 \ , d^2 b_2 } \rangle \right]}{\left [ \int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , d \phi_1 \ , d y_1 \ , d \phi_2 \ , d y_2 \ , t_1 ( \bm b - \bm b_1 ) \ , t_1 ( \bm b - \bm b_2 ) \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , dy_1 \ , d^2 b_1 } \ , \frac{d \sigma^{p a_2}}{d^2 k_2 \ ,
dy_2 \ , d^2 b_2 } \right\rangle \right ] } \notag \\
\times \ , \frac{\int d^2 b \ , d^2 b_1 \ , d^2 b_2
, t_1 ( \bm b - \bm b_1 ) \
, t_1 ( \bm b - \bm b_2 ) \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 dy_1 d^2 b_1 } \ , \frac{d \sigma^{p a_2}}{d^2 k_2 dy_2 d^2 b_2 } \right\rangle } { \int d^2 b_1 \ ,
\left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , d y_1 \ , d^2 b_1 } \right\rangle \ , \int d^2 b_2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , d y_2 \ , d^2 b_2 } \right\rangle } -1.\end{aligned}\ ] ] to complete the calculation one needs the single- and double - gluon production cross sections which , when used in , would give us the correlation function . before we proceed to construct them ,
let us study a simple example elucidating the nature of one of the correlation types contained in correlator .
let us consider the simplest possible example of particle ( gluon ) production mechanism where the interaction of the two nucleons in the first nucleus with the second nucleus in factorizes , @xmath77 this contribution comes from the disconnected feynman diagrams and is usually identified as the uncorrelated part of the two - gluon production cross section .
however , it is clear that substituting into does not reduce the correlation function to zero : instead one gets @xmath78 \ , \left[\int d^2 b_2 \ , d \phi_2 \ , d y_2 \ , \langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , dy_2 \ , d^2 b_2 } \rangle \right]}{\left [ \int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , d \phi_1 \ , d y_1 \ , d \phi_2 \ , d y_2 \ , t_1 ( \bm b - \bm b_1 ) \ , t_1 ( \bm b - \bm b_2 ) \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 dy_1 d^2 b_1 } \right\rangle \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_2 dy_2 d^2 b_2 } \right\rangle \right ] } \notag \\ \times \ , \frac{\int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , t_1 ( \bm b - \bm b_1 ) \ , t_1 ( \bm b - \bm b_2 ) \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 dy_1 d^2 b_1 } \right\rangle \ , \left\langle \frac{d \sigma^{p a_2}}{d^2
k_2 dy_2 d^2 b_2 } \right\rangle } { \int d^2 b_1 \
, \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , d y_1 \ , d^2 b_1 } \right\rangle \ , \int d^2 b_2 \
, \left\langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , d y_2 \ , d^2 b_2 } \right\rangle } -1,\end{aligned}\ ] ] which , in general , could be non - zero .
certainly if the @xmath56-dependence factorizes from the rapidity and azimuthal dependence in the cross section @xmath79 then the correlation function is zero : however , such factorization is not always the case . for gluon production in the saturation framework , the cross section is a complicated function of @xmath80 , which means it is not in a factorized form and thus the correlator is not zero .
note that in the mv model ( which does not contain the small-@xmath11 evolution ) , gluon production is rapidity - independent , and , if one neglects the dependence of the gluon production cross section on the angle between @xmath59 and @xmath56 , the correlator becomes zero .
( dependence of gluon production cross section on the collision geometry in the mv approximation is not very strong , peaking at non - perturbatively low momenta @xcite . ) for the general case in we observe a possible non - trivial correlation in the two - gluon production described by _ disconnected
_ feynman diagrams .
if the gluon production cross section is a slowly varying ( but not constant ) function of rapidity , as is the case in the saturation / cgc framework near mid - rapidity , this correlation would be long - range in rapidity .
the origin of this correlation is somewhat peculiar : even though the two - nucleon wave function in is factorized and , hence , represents uncorrelated nucleons , these two nucleons are correlated by the simple fact of being parts of the same bound state , the projectile nucleus . in other words ,
the probability of finding two nucleons at the impact parameters @xmath54 and @xmath81 is proportional to @xmath82 and is not a product of two independent probabilities after all impact parameters @xmath52 of the incoming nucleus are integrated over : this is a correlation .
note also that the presence of real wave - function correlations , that is , non - factorizable corrections to the right - hand - side of , would also lead to some nontrivial two - particle correlations in .
if we define the correlation function at the fixed nuclear impact parameter @xmath52 by not integrating over @xmath52 in eqs . and using the result in , we get @xmath83 \ , \left[\int d^2 b_2 \ , d \phi_2 \ , dy_2 \ , |\psi_i ( { \bm b } - { \bm b}_2)|^2 \ , \langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , dy_2 \ , d^2 b_2 } \rangle \right]}{\left [ \int d^2 b_1 \ , d^2 b_2 \ , d \phi_1 \ , dy_1 \ , d \phi_2 \ , dy_2 \ , |\psi_{ii } ( \bm b - \bm b_1 , \bm b - \bm b_2 ) |^2 \ , \langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , dy_1 \
, d^2 b_1 } \frac{d \sigma^{p a_2}}{d^2 k_2 \ , dy_2 \ , d^2 b_2 } \rangle \right ] } \notag \\
\times \ , \frac{\int d^2 b_1 \ ,
d^2 b_2 \ , |\psi_{ii } ( \bm b - \bm b_1 , \bm b - \bm b_2 ) |^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 dy_1 d^2 b_1 } \ , \frac{d \sigma^{p
a_2}}{d^2 k_2 dy_2 d^2 b_2 } \right\rangle } { \int d^2 b_1 \ , |\psi_i ( { \bm b } - { \bm b}_1)|^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_1 \ , d y_1 \ , d^2 b_1 } \right\rangle \ , \int d^2 b_2 \ , |\psi_i ( { \bm b } - { \bm b}_2)|^2 \ , \left\langle \frac{d \sigma^{p a_2}}{d^2 k_2 \ , d y_2 \ , d^2 b_2 } \right\rangle } -1,\end{aligned}\ ] ] one can see that this fixed - impact parameter correlation function in is zero , @xmath84 , for the factorized wave function from for disconnected - diagram interactions from .
thus di - gluon correlations due to our `` geometric '' correlation mechanism seem to also disappear when the impact parameter is fixed exactly . however , such precise determination of the impact parameter is impossible in an experimental analysis , where one is able to fix the collision centrality @xmath85 in a certain interval , but one can not fix the direction of @xmath52 .
the integration over any range of the impact parameter @xmath85 ( or the integration over the angles of @xmath52 keeping @xmath85 constant ) is likely to introduce these geometric correlations , as follows from .
note that the presence of non - trivial wave function correlations , i.e. , correlations beyond the factorization approximation in , may also lead to correlations which may survive in even for a fixed impact parameter @xmath52 and uncorrelated interactions . despite its simplicity
, the non - vanishing correlation in is one of the main results of this work . in the saturation / cgc framework
it may lead to long - range rapidity correlations similar to the observed `` ridge '' correlation .
indeed azimuthal correlations are missing in : such correlations may be formed in heavy ion collisions due to radial flow , as was argued in @xcite . indeed a lot more work is needed to compare this result with experiment .
in this section we are going to calculate the two - gluon production cross section for the heavy
light ion collisions working in the saturation / cgc framework .
as described above we assume that @xmath86 , such that the saturation effects are resummed to all orders only in the target nucleus with atomic number @xmath15 . while the saturation effects are not very important in the projectile nucleus with the atomic number @xmath14 , the fact that @xmath23 implies that the two gluons are predominantly produced in collisions of different nucleons in the projectile nucleus with the target nucleus .
according to the standard technique @xcite , the calculation will proceed using the light - cone perturbation theory ( lcpt ) @xcite to construct the gluon wave functions in the incoming projectile nucleus .
the production cross section will be obtained by convoluting the square of this light - cone wave function with the interaction with the target nucleus , which will be described by wilson lines along the light cone @xcite , which include the saturation effects in the form of both the glauber mueller multiple rescatterings @xcite and the bk / jimwlk evolution equations .
the diagrams contributing to the scattering amplitude of the two - gluon production in the collision of two nucleons from the projectile nucleus with the target nucleus are shown in . there
the two nucleons in the incoming nucleus are denoted by horizontal solid lines representing two quarks in the wave functions of those nucleons : throughout this work we will model the nucleons by a single valence quark each .
( generalization of our results to a more realistic nucleon wave function is straightforward . )
interaction with the target , which happens over a time - scale much shorter than the time scale of preparing gluons in the wave functions , is denoted by a vertical dashed line .
an essential part of the calculation is finding the light cone wave functions of the two interacting nucleons generating gluons in the incoming nucleus .
the diagrams in include vertical dotted lines denoting the intermediate states contributing light - cone energy denominators to the wave functions .
the energy denominators , and hence the wave functions of the two nucleons , factorize .
while this is obvious in diagrams c and d in , we will explain this factorization for other diagrams in a little more detail below .
we assume that the projectile nucleus is moving along the light cone `` + '' direction and the nucleons carry very large @xmath87 momenta . at the same time , the gluons carry light cone momenta @xmath88 and @xmath89 such that @xmath90 . the gluons will be produced with rapidities far away from the fragmentation regions of the colliding nuclei .
the small light cone momenta of the gluons make them dominate the energy denominators , generating a much larger contribution than that of the ( valence ) quarks @xcite .
the only difference between the diagrams a and b in is the ordering of the emitted gluons .
this will only affect the energy denominators that result from the light - cone perturbation theory rules @xcite .
when these two diagrams are added together these energy denominators factorize . denoting the light cone energies of the two gluons in diagrams a and b by @xmath91 and @xmath92 such that @xmath93 and @xmath94
we see that adding the contributions of the energy denominators in those graphs gives @xmath95 the result factorizes into the terms describing one or another gluon .
analysis of the diagrams e and f leads to the same factorization of energy denominators @xcite .
the two - gluon light cone wave function for diagrams a+b and e+f is @xmath96 where @xmath97 is the qcd coupling constant and the polarization vector is @xmath98 @xcite .
the factors of @xmath99 are the su@xmath100 generators in the fundamental representation with @xmath24 the number of quark colors , while the subscript @xmath101 denotes the associated nucleons .
the single nucleon s soft gluon wave function is @xcite @xmath102 the diagrams c+d have a wave function different from by a minus sign .
the minus sign difference will be absorbed in the interaction term .
our calculation for the two - gluon production cross section will be closely following that for the single gluon production in a collision of a single nucleon with a nucleus ( a @xmath5 collision ) .
for this reason , and also because we need the single inclusive gluon production cross section to construct the correlator , let us briefly review the results .
the single - gluon inclusive production cross section in @xmath5 collisions is illustrated in , which shows the square of the scattering amplitude .
the cross denotes the measured gluon .
the vertical solid straight line in is the final - state cut , while the vertical dashed lines denote interactions with the target just like in .
the gluon production cross section can be written as @xcite ( with the projectile proton represented by a single valence quark ) @xmath103 with the coordinate notation defined in .
the wave function squared summed over polarizations and colors is @xmath104 with @xmath105 the fundamental - representation casimir operator , while the interaction with the target is @xmath106 \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x } } u_{{\bm b}}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm b } } u_{{\bm y}}^\dagger ] \ ; + \ ; 1 \right\rangle.\end{aligned}\ ] ] the latter object is written in terms of expectation values ( in the target nucleus wave function ) of the adjoint wilson lines along the @xmath107 light cone : @xmath108 where @xmath109 is the gluon field of the target in the adjoint representation . defining the adjoint ( gluon ) dipole @xmath110-matrix by @xmath111 \right\rangle\end{aligned}\ ] ] we rewrite as @xmath112.\end{aligned}\ ] ] using in we obtain the gluon production cross section in the heavy light ion collision .
the dipole amplitude @xmath113 in contains the glauber - mueller multiple rescatterings @xcite ( recovered by putting @xmath114 in the argument of @xmath113 ) , along with the energy dependence included though the bk / jimwlk evolution equations @xcite : the cross section formula remains the same in both cases .
the former case would correspond to consistently treating the problem in the mv model @xcite .
the latter case would include small-@xmath11 evolution corrections in the rapidity interval between the produced gluon and the target nucleus : one may worry that such inclusion would be asymmetric , since the evolution in the rapidity range between the projectile and the produced target is not included .
a more symmetric treatment of the gluon production problem in @xmath5 collisions with evolution corrections included in all rapidity intervals is presented in @xcite ( see also @xcite for a more pedagogical presentation ) .
inclusion of evolution in the rapidity interval between the projectile and the produced gluon(s ) is beyond the scope of the present work : it should be possible along the lines of @xcite though
. the two - gluon production in heavy - light ion collisions is easily constructed by analogy to the single - gluon production calculation of sec .
[ sec : single ] .
the diagrams contributing to the square of the scattering amplitude for the double gluon production in heavy
light ion collisions are shown in , written as a direct product of the gluon production processes in the interactions of each of the nucleons from the projectile nucleus with the target . just as in the vertical dashed lines in represent interactions with the target , while the vertical solid line denotes the final state cut .
the evaluation of the diagrams in is straightforward .
the two - gluon wave function squared is obtained by using twice , which yields @xmath115 with all the coordinate labels explained in .
interactions with the target are treated similarly .
the traces of wilson lines in the top and bottom parts of are exactly the same as in and appear to factorize , suggesting absence of dynamically generated correlations .
this is not so .
note , that the averaging over the target is applied to both parts of simultaneously : we thus get an averaged product of traces of wilson lines , @xmath116 \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm
x}_1 } u_{{\bm b}_1}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm b}_1 } u_{{\bm y}_1}^\dagger ] \ ; + \ ; 1 \right ) \right .
\notag \\ & \times \left .
\left ( \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x}_2 } u_{{\bm y}_2}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x}_2 } u_{{\bm b}_2}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm b}_2 } u_{{\bm y}_2}^\dagger ] \ ; + \ ; 1 \right ) \right\rangle,\end{aligned}\ ] ] which does not , in general , factorize into the product of two separately target - averaged interactions from . using we write @xmath117 ^ 2 } \int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , t_1 ( { \bm b } - { \bm b}_1 ) \ , t_1 ( { \bm b } - { \bm b}_2 ) \ ,
, d^2 y_1 \ , d^2 x_2 \ , d^2 y_2 \
, e^{- i \ ; { \bm k}_1 \cdot ( { \bm x}_1-{\bm y}_1 ) - i \ ; { \bm k}_2 \cdot ( { \bm x}_2-{\bm y}_2 ) } \notag \\ & \times \ , \sum_{\lambda , \lambda ' , a , b } \psi^{ab } ( { \bm x}_1 , { \bm b_1 } , \lambda ; { \bm x}_2 , { \bm b_2 } , \lambda ' ) \ , \psi^{a b \ , * } ( { \bm y}_1 , { \bm b_1 } , \lambda ; { \bm y}_2 , { \bm b_2 } , \lambda ' ) \ , int_2 ( { \bm x}_1 , { \bm y}_1 , { \bm b}_1 ; { \bm x}_2 , { \bm y}_2 , { \bm b}_2),\end{aligned}\ ] ] which , with the help of eqs . and becomes ( cf .
@xcite ) @xmath118 \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x}_1 } u_{{\bm b}_1}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm b}_1 } u_{{\bm y}_1}^\dagger ] \ ; + \ ; 1 \right ) \right . \notag \\ & \times \left .
\left ( \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x}_2 } u_{{\bm y}_2}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm x}_2 } u_{{\bm b}_2}^\dagger ] \ ; - \ ; \frac{1}{n_c^2 - 1 } \ ; tr [ u_{{\bm b}_2 } u_{{\bm y}_2}^\dagger ] \ ; + \ ; 1 \right ) \right\rangle.\end{aligned}\ ] ] this is the expression for the two - gluon production cross section coming from the diagrams in .
the interaction with the target can be evaluated in the quasi - classical multiple rescattering approximation , as we will show later .
the rapidity evolution can be included using the jimwlk equation .
note , however , that when the rapidity difference between the two gluons is sufficiently large , @xmath119 , one has to include the evolution corrections in the rapidity interval between the produced gluons , such that the wilson lines in the two parenthesis in should be taken at different rapidities . a similar effect had to be included in the two - gluon production cross section in dis in @xcite . in such regime
one would also need to include evolution corrections in the rapidity window between the projectile and ( at least one of ) the produced gluons .
inclusion of the evolution corrections in terms of the weight functional @xmath120 of the jimwlk evolution equation into the two - gluon production cross section in nucleus
nucleus collisions was done in @xcite . in this work
we will limit ourselves to the quasi - classical regime where no evolution corrections are required and the cross sections are rapidity - independent .
note that for scattering on a large nuclear target , if one resums powers of @xmath121 , as is the case in the glauber - mueller ( gm ) rescatterings ( and in the bk / jimwlk evolution for which the gm rescatterings serve as the initial condition ) , and if one takes the large-@xmath24 limit , the expectation values of the traces in factorize , such that @xmath122 \ ; tr [ u_{{\bm x}_2 } u_{{\bm y}_2}^\dagger ] \right\rangle \bigg|_{\mbox{large}-n_c , \ , \mbox{large}-a_2 } \approx \left\langle tr [ u_{{\bm
x}_1 } u_{{\bm y}_1}^\dagger ] \right\rangle \ ; \left\langle tr [ u_{{\bm
x}_2 } u_{{\bm y}_2}^\dagger ] \right\rangle.\end{aligned}\ ] ] this leads to factorization of being valid in the large-@xmath24 and large - target - nucleus limit .
as shown above , even in this factorized regime one may obtain a non - trivial correlation function due to the geometric correlations .
before we proceed to evaluating the correlations contained in the cross section , let us point out another contribution to the two - gluon production cross section arising from squaring the sum of the diagrams in . when squaring the diagrams in it
is possible that the gluon emitted by one nucleon in the amplitude will be absorbed by another nucleon in the complex conjugate amplitude .
the corresponding contributions to the cross section are shown in figs .
[ crossed ] and [ crossed2 ] , where crossing gluon lines do not form a vertex .
we will refer to these diagrams as the `` crossed '' graphs .
the diagrams obtained from those in figs . [ crossed ] and [ crossed2 ] by a mirror reflection with respect to the cut correspond to the @xmath123 interchange , and will be automatically included in the cross section to be calculated below since it will contain integrals over all @xmath124 and @xmath125 .
the sum over different orderings of gluon emissions in the ( complex conjugate ) amplitude of is implied , but is not shown explicitly . in evaluating the graphs in figs .
[ crossed ] and [ crossed2 ] we note that there are non - trivial color factors due to the interaction terms , which we absorb into the wave function .
the expression for the wave function squared is @xmath126 defining the color - quadrupole operator @xcite @xmath127 \right\rangle\end{aligned}\ ] ] after some algebra the interactions with the target in both and can be shown to be equal to ( with the terms ordered in the same way as the diagrams in figs .
[ crossed ] and [ crossed2 ] and the transverse coordinates defined in the upper left diagram of each of these two figures ) @xmath128 the only difference between the contributions of the diagrams in figs .
[ crossed ] and [ crossed2 ] is in the exponential factors for the fourier transform into the transverse momentum space . the longitudinal momentum flow patterns in figs .
[ crossed ] and [ crossed2 ] are different from that in .
this is illustrated in , which shows the flow of the `` plus '' momentum component through the first diagram in .
note that the change in the `` plus '' momentum component is negligible in the eikonal interactions with the target considered here .
requiring that the incoming quark lines carry the same `` plus '' momentum both in the amplitude and in the complex conjugate amplitude , one would obtain @xmath129 : however such requirement is not correct .
the actual scattering happens between two nuclei , and it is the momenta of the whole incoming nuclei which have to be equal both in the amplitude and in the complex conjugate amplitude .
hence for @xmath130 the diagrams like that in would only correspond to different redistributions of the projectile nucleus momentum between the nucleons in it in the amplitude and in the complex conjugate amplitude without changing the same `` plus '' momentum of the whole nucleus on both sides of the cut .
hence the @xmath129 condition is not necessary for the `` crossed '' diagrams . combining the factors in eqs . and , and inserting fourier transform exponentials , we arrive at the expression for the two - gluon production cross section contribution resulting from the `` crossed '' diagrams from figs . [ crossed ] and [ crossed2 ]
@xmath131 ^ 2 } \
, \int d^2 b \ , d^2 b_1 \ , d^2 b_2
, t_1 ( { \bm b } - { \bm b}_1 ) \ , t_1 ( { \bm b } - { \bm b}_2 ) \ , d^2 x_1 \ , d^2 y_1 \ , d^2 x_2 \ , d^2 y_2 \notag \\ & \times \ , \left [ e^{- i \ ; { \bm k}_1 \cdot ( { \bm x}_1-{\bm y}_2 ) - i \ ; { \bm k}_2 \cdot ( { \bm x}_2-{\bm y}_1 ) } + e^{- i \ ; { \bm k}_1 \cdot ( { \bm x}_1-{\bm y}_2 ) + i \ ; { \bm k}_2 \cdot ( { \bm x}_2-{\bm y}_1 ) } \right ] \ , \frac{16 \ ; { \alpha}_s^2}{\pi^2 } \ , \frac{c_f}{2 n_c } \ ; \frac { { \bm x}_1 - { \bm b}_1}{|{\bm x}_1 - { \bm b}_1|^2 } \cdot \frac{{\bm y}_2 - { \bm b}_2 } {
|{\bm y}_2 - { \bm b}_2|^2 } \ , \frac{{\bm x}_2 - { \bm b}_2}{|{\bm
x}_2 - { \bm b}_2|^2 } \cdot \frac{{\bm y}_1 - { \bm b}_1 } { |{\bm y}_1 - { \bm b}_1|^2 } \notag
\\ & \times \ , \bigg [ q ( { \bm x}_1 , { \bm y}_1 , { \bm x}_2 , { \bm y}_2 ) - q ( { \bm x}_1 , { \bm y}_1 , { \bm x}_2 , { \bm b}_2 ) - q ( { \bm x}_1 , { \bm y}_1 , { \bm b}_2 , { \bm y}_2 ) + s_g ( { \bm x}_1 , { \bm y}_1 ) - q ( { \bm x}_1 , { \bm b}_1 , { \bm x}_2 , { \bm y}_2 ) \ ; + \ ; q ( { \bm x}_1 , { \bm b}_1 , { \bm x}_2 , { \bm b}_2 ) \notag \\ & + q ( { \bm x}_1 , { \bm b}_1 , { \bm b}_2 , { \bm y}_2 ) - s_g ( { \bm x}_1 , { \bm b}_1 ) - q ( { \bm b}_1 , { \bm y}_1 , { \bm x}_2 , { \bm y}_2 ) + q ( { \bm b}_1 , { \bm y}_1 , { \bm x}_2 , { \bm b}_2 ) + q ( { \bm b}_1 , { \bm y}_1 , { \bm b}_2 , { \bm y}_2 ) - s_g ( { \bm b}_1 , { \bm y}_1 ) + s_g ( { \bm x}_2 , { \bm y}_2 ) \notag \\ & - s_g ( { \bm x}_2 , { \bm b}_2 ) - s_g ( { \bm b}_2 , { \bm y}_2 ) + 1\bigg].\end{aligned}\ ] ] just like in , the dipole and quadrupole scattering amplitudes in can be evaluated either in the mv model or by using bk and jimwlk evolution equations .
the quadrupole amplitude evolution equation was derived in the large-@xmath24 limit in @xcite , and beyond the large-@xmath24 limit in @xcite .
again , is valid only as long as the rapidities @xmath132 and @xmath133 of the two produced gluons are close to each other , @xmath134 , such that no small-@xmath11 evolution corrections need to be included in the @xmath135 $ ] rapidity interval . equations ( [ eq:2glue_prod_main ] ) and ( [ crossed_xsect ] ) , when combined , give us the two - gluon production cross section in the heavy - light ion collisions : @xmath136 this production cross section is the main formal result of this work .
we analyze the properties of the cross section below .
our goal now is to evaluate the correlations resulting from the two - gluon production cross section , that is from the cross sections in eqs .
( [ eq:2glue_prod_main ] ) and ( [ crossed_xsect ] ) .
the first step is to evaluate the interaction with the target .
we will be working in the quasi - classical mv / gm limit , where the interaction and , hence , the production cross sections and , are rapidity - independent .
we begin with the cross section in . even in the quasi - classical limit ,
evaluation of the interaction in is calculationally intensive ( though conceptually rather straightforward ) . to simplify the calculation
we will also employ the large-@xmath24 expansion : we assume that the nucleons in the nuclei are made out of an order-@xmath137 valence quarks ( or gluons ) , such that the saturation scale , which in such case is proportional to @xmath138 , is constant in the t hooft s large-@xmath24 limit . in the saturation physics framework such approximation was used in @xcite for the quadrupole operator giving a reasonably good approximation to the exact answer @xcite .
similar approximations are frequently used ( albeit , often implicitly ) in applications of anti - de sitter space / conformal field theory ( ads / cft ) correspondence to collisions of heavy ions modeled by shock waves @xcite . at the leading order in @xmath139 expansion of the interaction with the target factorizes , as discussed around .
in addition , the cross section in is @xmath139-suppressed ( as compared to the leading term in ) and can be neglected at the leading order in @xmath24 . the correlation function is then given by with the single gluon production cross section from .
in the mv / gm approximation the gluon color dipole interaction with the target is @xcite @xmath140\end{aligned}\ ] ] with @xmath141 the rapidity - independent gluon saturation scale in the quasi - classical limit evaluated at the dipole center - of - mass @xmath142 and @xmath143 an infrared ( ir ) cutoff .
we see that the quasi - classical single gluon production cross section is rapidity - independent and , for unpolarized nuclei and for perturbatively large @xmath20 @xcite , is also independent of the azimuthal angle @xmath144 of the transverse momentum @xmath59 of the outgoing gluon .
we conclude that at the leading order in @xmath139 in the quasi - classical approximation the geometric correlations are almost absent and the correlation function is approximately zero .
non - trivial correlations can be obtained from by expanding the interaction term to the first non - trivial order in @xmath139 . to this end
we write a correlator of two wilson line traces as @xmath145 \ , tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle = \frac{1}{(n_c^2 - 1)^2 } \ , \langle tr [ u_{{\bm x}_1 } u_{{\bm x}_2}^\dagger ] \rangle \langle tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle + \delta ( { \bm x}_1 , { \bm x}_2 , { \bm x}_3 , { \bm x}_4 ) , \end{aligned}\ ] ] where the correction to the factorized expression , denoted by @xmath146 , is order-@xmath139 . (
note that each adjoint trace in is order-@xmath147 , such that its left - hand side , as well as the first term on its right - hand side , are order - one in @xmath24 counting . )
the leading order-@xmath139 contribution to @xmath146 is derived in the appendix , with the result being @xmath148 + o \left ( \frac{1}{n_c^4 } \right),\end{aligned}\ ] ] where we have defined [ ds ] @xmath149 \\ d_2 & = - \frac{q_{s0}^2}{4 } \left [ | \bm x_1 - \bm x_3 |^2 \ , \ln \left ( \frac{1}{| \bm x_1 - \bm x_3 | \lambda } \right ) + | \bm x_2 - \bm x_4 |^2 \ , \ln \left ( \frac{1}{| \bm x_2 - \bm x_4 | \lambda } \right ) \right ]
\\ d_3 & = - \frac{q_{s0}^2}{4 } \left [ | \bm x_1 - \bm x_4 |^2 \ , \ln \left ( \frac{1}{| \bm x_1 - \bm x_4 | \lambda } \right ) + | \bm x_2 - \bm x_3 |^2 \
, \ln \left ( \frac{1}{| \bm x_2 - \bm x_3 | \lambda } \right ) \right]\end{aligned}\ ] ] assuming , for simplicity , that all the saturation scales are evaluated at the same impact parameter . using in
we see that the correlated part of the two - gluon production cross section is @xmath150.\end{aligned}\ ] ] along with the expression for @xmath146 in are our most complete results for the contribution to the two - gluon production cross section coming from in the quasi - classical regime of the heavy
light ion collisions in the large-@xmath24 limit .
the evaluation of the full appears to be rather involved and is left for the future work .
instead we will expand and use the result in to obtain the correlated gluon production at the lowest non - trivial order .
this result can be used to elucidate the structure of the long - range rapidity correlations , along with the comparison to the existing expressions in the literature . at the lowest non - trivial order in @xmath151 s ( that is , since each @xmath151 represents a two - gluon exchange with the target , at the lowest order in the number of gluon exchanges corresponding to @xmath152 ) , becomes @xmath153 substituting this into we can further simplify the expression by assuming that for the connected diagrams that contribute to the @xmath146 s one has @xmath124 and @xmath125 perturbatively close to each other , with the typical separation between these two impact parameters much smaller than the nucleon radius . since the nuclear profile functions for a large nucleus do not vary much over perturbatively short distances we can put @xmath154 in the arguments of @xmath155 s and @xmath141 , where @xmath156 . defining @xmath157 we can write @xmath158 . the integral over @xmath159 can be then carried out with the help of a fourier transform @xmath160 used to replace @xmath161 in and , hence , all @xmath146 s in as a double fourier - integral : for instance @xmath162 \notag \\
\times \ , \left [ e^{- i \ , { \bm l } ' \cdot ( { \bm x}_1 - { \bm x}_2 ) } + e^{-i \ , { \bm l } ' \cdot ( { \bm y}_1 - { \bm y}_2 ) } - e^{- i \ , { \bm l } ' \cdot ( { \bm x}_1 - { \bm y}_2 ) } - e^{- i \ , { \bm l } ' \cdot ( { \bm y}_1 - { \bm x}_2 ) } \right ] \notag \\ = \frac{q_{s0}^4}{2 \ , n_c^2 } \int \frac{d^2 l \ , d^2 l'}{(2 \pi)^2 } \ , \ , \frac{1}{({\bm l}^2)^2 } \ , \frac{1}{({\bm l}'^2)^2 } \ , \left [ e^{i \ , { \bm l } \cdot ( { \tilde { \bm x}}_1 - { \tilde { \bm x}}_2 ) } + e^{i \ , { \bm l } \cdot ( { \tilde { \bm y}}_1 - { \tilde { \bm y}}_2 ) } - e^{i \ , { \bm l } \cdot ( { \tilde { \bm x}}_1 - { \tilde { \bm y}}_2 ) } - e^{i \ , { \bm l } \cdot ( { \tilde { \bm y}}_1 - { \tilde { \bm x}_2 ) } } \right ] \notag \\
\times \ , \left [ e^{- i \ ,
{ \bm l } ' \cdot ( { \tilde { \bm x}}_1 - { \tilde { \bm x}}_2 ) } + e^{-i \ , { \bm l } ' \cdot ( { \tilde { \bm y}}_1 - { \tilde { \bm y}}_2 ) } - e^{- i \ , { \bm l } ' \cdot ( { \tilde { \bm x}}_1 - { \tilde { \bm y}}_2 ) } - e^{- i \ , { \bm l } ' \cdot ( { \tilde { \bm y}}_1 - { \tilde { \bm x}}_2 ) } \right ] \ , e^{i \ , ( { \bm l } - { \bm l } ' ) \cdot \delta { \bm b}}\end{aligned}\ ] ] where we employed the substitution @xmath163 performing similar substitutions for all @xmath146 s in , we can integrate over @xmath159 , @xmath164 , @xmath165 , @xmath166 , @xmath167 , and @xmath168 .
after some algebra one arrives at ( cf .
@xcite ) @xmath169 ^ 2 \ , \frac{q_{s0}^4 ( { \bm b})}{{\bm k}_1 ^ 2 \ , { \bm k}_2 ^ 2 } \ , \int\limits_\lambda \frac{d^2 l}{({\bm l}^2)^2 } \ , \left [ \frac{1}{({\bm k}_1 - { \bm l})^2 \ , ( { \bm k}_2 + { \bm l})^2 } + \frac{1}{({\bm k}_1 - { \bm l})^2 \ , ( { \bm k}_2 - { \bm l})^2 } \right],\end{aligned}\ ] ] where the subscript @xmath170 denotes the lowest - order cross section .
equation is illustrated in by regular feynman diagrams that contribute to its right - hand - side ( cf .
@xcite ) : these diagrams are referred to as the `` glasma '' graphs in the literature .
the momenta of the gluon lines are labeled in , and the triple gluon vertices , marked by the dark circles , are the effective lipatov vertices .
these particular graphs contribute to the first and the second terms in the square brackets of correspondingly , and can be calculated by taking two lipatov vertices squared .
the obtained expression contains both the near - side and away - side azimuthal correlations @xcite : clearly the first term in the square brackets of contains poles at @xmath171 and @xmath172 , which , after integration over @xmath173 , lead to a contribution , which leads to a correlated contribution independent of the azimuthal angle between @xmath174 and @xmath175 . ]
@xmath176 characteristic of the away - side correlations .
the second term in the square brackets of has poles at @xmath177 and @xmath178 , yielding a contribution @xmath179 indicating near - side correlations .
note that all correlations are long - range in rapidity since the cross section is rapidity - independent .
now we turn our attention to . there
the cross section contribution itself is @xmath139-suppressed as compared to the leading ( uncorrelated ) part of : hence we need to evaluate the interaction with the target in using the large-@xmath24 limit .
we work in the same quasi - classical mv / gm approximation , along with the large-@xmath24 limit .
the fundamental ( quark ) quadrupole amplitude was evaluated in this approximation in @xcite ( see eq .
( 14 ) there ) , yielding @xmath180.\end{aligned}\ ] ] since the adjoint ( gluon ) quadrupole in the large-@xmath24 limit is simply @xmath181 ^ 2\end{aligned}\ ] ] we get @xmath182 ^ 2.\end{aligned}\ ] ] equations and , when used in , give us the remaining contribution to the two - gluon production cross section in the quasi - classical approximation , which has to be added to .
again the full expression is hard to evaluate : instead , similar to , we will consider the limit of large @xmath183 and @xmath184 : @xmath185
. expanding in the powers of @xmath151 s yields @xmath186 + o \left ( d_i^3 \right).\end{aligned}\ ] ] using this result along with a similar expansion for in , and employing again the fourier transform along with the substitution one can write @xmath187 \notag \\ + \left ( 1 - e^{i \ , { \bm l } \cdot { \bm { \tilde x}}_1 } \right ) \ , \left ( 1 - e^{- i \ , { \bm l } ' \cdot { \bm { \tilde x}}_2 } \right ) \ , \left ( 1 - e^{- i \ , { \bm l } \cdot { \bm { \tilde y}}_1 } \right ) \ , \left ( 1 - e^{i \ , { \bm l } ' \cdot { \bm { \tilde y}}_2 } \right ) \bigg\}\end{aligned}\ ] ] where the @xmath188 , @xmath189 , and @xmath190 symmetries of the integrand were utilized to cast the expression in its present form . using to replace the interaction with the target in and
integrating over @xmath191 , @xmath192 , @xmath193 , @xmath194 , and , in some terms , over @xmath168 , leads to the following result @xmath195 where @xmath196 ^ 2 \ , \frac{q_{s0}^4 ( { \bm b})}{{\bm k}_1 ^ 2 \ , { \bm k}_2 ^ 2 } \ , \int \frac{d^2 l}{({\bm l}^2)^2 \ , ( ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2)^2 \ , ( ( { \bm k}_1 - { \bm l})^2)^2 \ , ( ( { \bm k}_2 + { \bm l})^2)^2 } \notag \\ & \times \big\ { \left [ { \bm l}^2 \ , ( { \bm k}_2 + { \bm l})^2 + ( { \bm k}_1 - { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_1 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] \notag \\ & \times \ , \left [ { \bm l}^2 \ , ( { \bm k}_1 - { \bm l})^2 + ( { \bm k}_2 + { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_2 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] \notag \\ & + 4 \ , { \bm l}^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 \ , \left [ ( ( { \bm k}_1 - { \bm l})^2)^2 + ( ( { \bm k}_2 + { \bm l})^2)^2 \right ] \big\ } + ( { \bm k}_2 \rightarrow - { \bm k}_2)\end{aligned}\ ] ] and @xmath197 ^ 2 \ , \frac{q_{s0}^4 ( { \bm b})}{{\bm k}_1 ^ 2 \ , { \bm k}_2 ^ 2 } \ , \left [ \delta^2 ( { \bm k}_1 - { \bm k}_2 ) + \delta^2 ( { \bm k}_1 + { \bm k}_2 ) \right ] \ , \notag \\ & \times \ , \int\limits_\lambda \frac{d^2 l \ , d^2 l'}{({\bm l}^2)^2 \ , ( { \bm l ' } ^2)^2 \ , ( ( { \bm k}_1 - { \bm l})^2)^2 \ , ( ( { \bm k}_1 + { \bm l'})^2)^2 } \ , \left [ { \bm l}^2 \ , ( { \bm k}_1 + { \bm l'})^2 + { \bm l'}^2 \ , ( { \bm k}_1 - { \bm l})^2 - { \bm k}_1 ^ 2 \ , ( { \bm l } + { \bm l'})^2 \right]^2.\end{aligned}\ ] ] we will explain the origin of the `` hbt '' label on the cross section in in a little while below : we defer the analysis of until then .
first we concentrate on the expression .
we note that by shifting the integration momentum @xmath198 one could reduce the second term in the curly brackets ( together with the @xmath199 term ) to that in , thus doubling .
that term is still described by the diagrams of the type shown in and gives the near- and away - side correlations in eqs . and .
the first term in the curly brackets of corresponds to a different class of diagrams , one of which is shown in .
one can see that the diagram in contributes two non - forward `` squares '' of the effective lipatov vertices .
an analysis of the poles of the integral in the first term of shows that it contains similar near- and away - side correlations to that in eqs . and , though with an additional enhancement due to a prefactor , @xmath200 \ , \left[\frac{1}{({\bm k}_1 - { \bm k}_2)^2 } + \frac{1}{({\bm k}_1 + { \bm k}_2)^2 } \right].\end{aligned}\ ] ] the origin of this contribution to correlations is in the non - forward squares of the lipatov vertices , which is similar to the mechanism for generating long - range rapidity correlations proposed in @xcite using two bfkl ladders with non - zero momentum transfer .
note , however , that we do not obtain the correlation proportional to the first power of @xmath201 advocated in @xcite , perhaps due to the lowest - order nature of the result .
it appears that the correlations resulting from the lowest - order diagrams like the one depicted in have not been considered in the literature yet .
note that the away- and the near - side correlations enter eqs . and on equal footing :
in fact , one could be obtained from another by a simple @xmath202 substitution in either of those expressions . while evaluating the integrals over @xmath173 in eqs . and analytically appears to be rather algebra - intensive , we do not need to do this to observe that , once the @xmath173-integral is carried out for one of the terms in the square brackets of , the answer for the other term is obtained by substituting @xmath203 in the result .
equation simply contains an additive @xmath204 term .
since the correlated cross sections and are sums of two terms related by the @xmath199 substitution and are symmetric under the @xmath205 interchange , we conclude that they are functions only of even powers of @xmath174 and @xmath175 , that is functions of @xmath206 , @xmath207 , @xmath208 , and possibly @xmath209 .
clearly this implies that the fourier series representation of eqs . and would only contain even cosine harmonics of the azimuthal angle , that is @xmath210 where @xmath211 is the angle between momenta @xmath174 and @xmath175 while @xmath183 and latexmath:[$k_{2 t } = be determined by an exact calculation .
it is quite interesting that only even harmonics contribute to the correlated cross section in .
let us stress here that we have not made any assumptions about the centrality of the collision : we do not have an almond - shaped overlap of the two nuclei .
in fact we integrate over all impact parameters @xmath52 .
( also the impact parameter dependence factorizes from the rest of the expression . )
hence the correlation in is not caused by the geometry of the collision . to construct the correlation function
one has to use eqs . and in . in the latter
, the uncorrelated two - gluon production would dominate in the denominator of the normalization factor , that is , in the denominator of the first factor on its right - hand side .
the single - gluon production cross - section at the lowest order is equal to ( see @xcite and references therein ) @xmath213 using this along with the sum of eqs . and in and taking the large-@xmath24 limit in we get @xmath214 ^ 2 \ , q_{s0}^4 ( { \bm b})}{\int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , t_1 ( { \bm b } - { \bm b}_1 ) \ , t_1 ( { \bm b } - { \bm b}_2 ) \ , q_{s0}^2 ( { \bm b}_1 ) \ ,
q_{s0}^2 ( { \bm b}_2 ) } \notag \\ & \times \ , \frac{{\bm k}_1 ^ 2 \ , { \bm k}_2 ^ 2}{\ln \frac{k_1 ^ 2}{\lambda^2 } \ , \ln \frac{k_2 ^ 2}{\lambda^2 } } \ , \bigg\ { 2 \ , \int\limits_\lambda \frac{d^2 l}{({\bm l}^2)^2 } \ , \left [ \frac{1}{({\bm k}_1 - { \bm l})^2 \ , ( { \bm k}_2 + { \bm l})^2 } + \frac{1}{({\bm k}_1 - { \bm l})^2 \ , ( { \bm k}_2 - { \bm l})^2 } \right ] \notag \\ & + \frac{1}{8 } \ , \bigg [ \int \frac{d^2 l}{({\bm l}^2)^2 \ , ( ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2)^2 \ , ( ( { \bm k}_1 - { \bm l})^2)^2 \ , ( ( { \bm k}_2 + { \bm l})^2)^2 } \ , \left [ { \bm l}^2 \ , ( { \bm k}_2 + { \bm l})^2 + ( { \bm k}_1 - { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_1 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] \notag \\ & \times \ , \left [ { \bm l}^2 \ , ( { \bm k}_1 - { \bm l})^2 + ( { \bm k}_2 + { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_2 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] + ( { \bm k}_2 \rightarrow - { \bm k}_2 ) \bigg ] \bigg\}.\end{aligned}\ ] ] to evaluate this further let us assume for a moment that both colliding nuclei are simply cylinders oriented along the collision axis , such that the nuclear profile functions of the nuclei are @xmath215 where @xmath216 labels the nuclei , @xmath69 is the ( constant ) nucleon number density in the nucleus , @xmath217 and @xmath218 are the radii of the projectile and target nuclei , and the cylindrical nucleus is assumed to have length @xmath219 along its axis . assuming that both nuclei are large and neglecting the edge effects
, we can neglect the @xmath124 and @xmath220 dependence in these nuclear profile functions of both nuclei .
since the gluon saturation scale in the mv model is @xmath221 we obtain in the @xmath222 limit @xmath223 \notag \\ & + \frac{1}{8 } \ , \bigg [ \int \frac{d^2 l}{({\bm l}^2)^2 \ , ( ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2)^2 \ , ( ( { \bm k}_1 - { \bm l})^2)^2 \ , ( ( { \bm k}_2 + { \bm l})^2)^2 } \ , \left [ { \bm l}^2 \ , ( { \bm k}_2 + { \bm l})^2 + ( { \bm k}_1 - { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_1 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] \notag \\ & \times \ , \left [ { \bm l}^2 \ , ( { \bm k}_1 - { \bm l})^2 + ( { \bm k}_2 + { \bm l})^2 \ , ( { \bm l } - { \bm k}_1 + { \bm k}_2)^2 - { \bm k}_2 ^ 2 \ , ( { \bm k}_2 - { \bm k}_1 + 2 \ , { \bm l})^2 \right ] + ( { \bm k}_2 \rightarrow - { \bm k}_2 ) \bigg ] \bigg\}.\end{aligned}\ ] ] indeed the correlator is suppressed by a power of @xmath137 and a power of the cross sectional area of the projectile nucleus @xmath224 , as discussed in the literature @xcite . while our conclusion about the correlator and the cross section contributing only to even fourier harmonics has been verified above at the lowest order only , it is true for the full two - gluon production cross section in heavy - light ion collisions .
this can be seen by noticing that @xmath202 substitution in does not change the cross section , as it is equivalent to the @xmath225 interchange of the integration variables .
the integrand of ( or , equivalently , of ) is invariant under @xmath225 interchange since the gluon is its own anti - particle such that @xmath226 = tr \left [ u_{\bm y } \ , u^\dagger_{\bm x } \right]$ ] .
the expression in is explicitly invariant under @xmath204 substitution .
hence the net correlated cross section is an even function of @xmath227 .
note also that eqs . and
are @xmath228-symmetric : in and in the term arising from the first exponential in the symmetry is a consequence of the symmetry of the integrand under the simultaneous @xmath229 , @xmath230 , and @xmath231 interchanges .
the term multiplying the second exponential in is @xmath232-symmetric due to the @xmath233 , @xmath225 symmetry of the integrand , which follows from the following property of the quadrupole operator : @xmath234 = tr \left [ u_{{\bm y}_2 } \ , u^\dagger_{{\bm x}_2 } \
, u_{{\bm y}_1 } \ , u^\dagger_{{\bm x}_1 } \right]$ ] .
we see that the net cross section is decomposable into a fourier series with even harmonics only .
therefore , the correlation function in the heavy - light ion collision can be also written as an even - harmonics series @xmath235 with some coefficients @xmath236 .
this conclusion also seems to hold in the case of classical gluon fields produced in a collision of two heavy ions , as can be seen from the result of the full numerical simulation of two - gluon production in heavy ion collisions due to classical gluon field carried out in @xcite .
the correlators in figs . 9 and 10 of @xcite do appear to have similar - looking maxima at @xmath1 and @xmath0 ( within the accuracy of the numerical error bars ) , though a more careful analysis is needed to figure out if our conclusion is true for two colliding heavy ions . to better visualize the correlator
let us envision a toy model in which multiple rescatterings ( and other saturation effects ) regulate the singularities at @xmath237 by the saturation scale @xmath141 in such a way that the correlator can be modeled as proportional to @xmath238 this correlator is plotted in using arbitrary units along the vertical axis as a function of the azimuthal angle @xmath239 between @xmath174 and @xmath175 for @xmath240 .
the shape illustrates what the full correlation function may look like , having identical near- and away - side correlation peaks .
our conclusion in this section is that the saturation / cgc dynamics in nuclear collisions appears to generate the long - range rapidity correlations which have identical maxima at both @xmath1 and @xmath0 in the azimuthal angle .
such correlations are non - flow in nature , since they do not arise due to almond - shaped geometry of the collisions .
( in fact the correlations should persist for the most central heavy ion collisions , though they should be suppressed by an inverse power of the overlap area as discussed above and in @xcite . ) however , since the elliptic flow observable @xmath241 ( even in the reaction - plane method ) is determined from two - particle correlations , it is possible that the correlations discussed here may contribute to the elliptic flow ( and to higher - order even - harmonics flow observables @xmath242 ) measured experimentally .
it is , therefore , very important to experimentally separate our initial - state correlations from the late - time qgp effects .
naturally the cumulant analysis @xcite is likely to remove these two - particle non - flow correlations from the flow observables .
however , the effectiveness of the cumulant method may again depend on the collision geometry .
let us illustrate this by considering the fourth order cumulant for elliptic flow , defined by @xcite @xmath243 where the angle brackets denote event averages along with the averaging over all angles @xmath244 , and @xmath245 of the four particles employed in the definition . using the lowest - order correlations employed in arriving at one
can straightforwardly show that the cumulant due to these correlations only is proportional to @xmath246 ^ 2 \ , [ t_1 ( { \bm b } - { \bm b}_2)]^2 \ , q_{s0}^4 ( { \bm b}_1 ) \ , q_{s0}^4 ( { \bm b}_2)}{\int d^2 b \ ,
d^2 b_1 \ , d^2 b_2 \ , d^2 b_3 \ , d^2 b_4 \ , t_1 ( { \bm b } - { \bm b}_1 ) \ , t_1 ( { \bm b } - { \bm b}_2 ) \ , t_1 ( { \bm b } - { \bm b}_3 ) \ , t_1 ( { \bm b } - { \bm b}_4 ) \ , q_{s0}^2 ( { \bm b}_1 ) \ , q_{s0}^2 ( { \bm b}_2 ) \ , q_{s0}^2 ( { \bm b}_3 ) \ , q_{s0}^2 ( { \bm b}_4 ) } \notag \\ & - \left [ \frac{\int d^2 b \ , d^2 b \ , [ t_1 ( { \bm b } - { \bm b})]^2 \ , q_{s0}^4 ( { \bm b})}{\int d^2 b \ , d^2 b_1 \ , d^2 b_2 \ , t_1 ( { \bm b } - { \bm b}_1 ) \ , t_1 ( { \bm b } - { \bm b}_2 ) \ , q_{s0}^2 ( { \bm b}_1 ) \ , q_{s0}^2 ( { \bm b}_2 ) } \right]^2\end{aligned}\ ] ] if the 4-particle correlator in ( the correlated part of the first term on its right - hand - side ) is due to the pairwise correlations of particles @xmath247 and @xmath248 or @xmath249 and @xmath250 . clearly , in the general case , @xmath251 from is non - zero . ,
and for cylindrical nuclei , one gets @xmath252 .
] moreover , the fourth order cumulant would contain the azimuthal angles dependence resulting from but not shown explicitly in .
we see that the geometric correlations from sec .
[ sec : disc ] may prevent complete removal of these non - flow correlations in the cumulants . just like with the correlator of ,
the above non - flow correlations can be completely removed from the cumulants for the fixed impact parameter @xmath52 : if we fix @xmath52 in ( that is , remove all the @xmath253 integrations from it ) , we get @xmath252 , which is exactly what the cumulant is designed to do @xcite completely cancel for non - flow correlations . note
that since , in the actual experimental analyses one effectively integrates over @xmath52 in a given centrality bin , it is possible that some non - flow correlations would remain in the cumulant .
even fixing @xmath85 precisely and integrating over the angles of @xmath52 may generate a non - zero contribution of these correlations to the cumulant .
the question of the interplay of the true qgp flow and the non - flow correlations discussed here has to be resolved by a more detailed numerical study . to clarify the physical meaning of the cross section obtained in we again consider a collisions of cylindrical nuclei .
with this simplification we can consider one of the terms in the interaction with the target , say the @xmath254 in . among many terms which contribute to the quadrupole amplitude in , there is a term @xmath255 using in and changing the interaction variables to those defined in we see that the remaining @xmath124 and @xmath125 integrals , after shifting those variables by @xmath52 , become simple fourier transforms of the projectile nucleus profile function @xmath256 , @xmath257 in the limit of a sufficiently large projectile nucleus the fourier transforms give a delta - function , yielding @xmath258 in agreement with the result in .
our present estimate shows that a more careful evaluation of would give a smother peak in @xmath259 at @xmath260 with the width determined by the inverse radius of the projectile nucleus @xmath217 . the first term on the right - hand - side of ( or ) has the trademark form of the hanbury brown
twiss ( hbt ) correlations @xcite , which are widely studied in heavy ion physics @xcite .
these correlations are normally local both in rapidity ( @xmath261 ) and in the transverse momentum @xmath260 : however , in ( or , in ) we only have locality in transverse momenta .
this can be explained by the fact that the extent of the interaction region in the longitudinal direction , commonly labeled @xmath262 , is very small in our case due to the extreme lorentz - contraction of the two colliding nuclei leading to smearing of the longitudinal hbt correlations .
we conclude that the correlations resulting from include hbt .
diagrammatic representation of the hbt correlations was discussed earlier in @xcite .
an example of the diagram giving rise to the correlations in is shown in .
interestingly though , due to the @xmath263 symmetry of the two - gluon production cross section , the hbt peak at @xmath260 in is accompanied by an identical peak at @xmath264 resulting from @xmath265 .
we thus obtain a back - to - back hbt correlation resulting from multiple rescatterings in nuclei .
( the origin of our back - to - back hbt correlations is different from that of the back - to - back hbt - like correlations proposed in @xcite . )
note that it is possible that the process of hadronization would affect the phases of the produced gluons , possibly destroying the perturbative hbt correlations of and replacing them with the hbt correlations of the non - perturbative origin .
the same hadronization process may also destroy the back - to - back hbt correlations from .
above we have derived the cross section for two - gluon production in the heavy - light ion collisions . the cross section is given by , with the two contributions shown in eqs . and .
concentrating on gluon production with a long - range separation in rapidity of the two gluons we evaluated the interaction with the target in using the mv / gm approximation in .
the interaction with the target in in the mv - model framework can be obtained from eqs . and .
analyzing the corresponding correlation function we identified four main types of correlations : ( i ) geometric correlations , ( ii ) hbt correlations , ( iii ) away - side correlations , and ( iv ) the near - side correlations .
hbt correlations ( ii ) are local in rapidity and thus are outside the main scope of our work here : however , we find that hbt correlations arising in our calculation are accompanied by a back - to - back hbt peak , as follows from . geometric correlations ( i ) arise in due to nucleons being confined within the same nucleus : more work is needed to see how strong they are and what role they are playing in experimental data . concentrating on the long - range rapidity correlations ( iii ) and ( iv )
we have shown that the two correlations have identical azimuthal profiles centered around @xmath266 and @xmath267 correspondingly .
this result is true to all orders of the calculation in the heavy - light ion collision considered here .
let us stress here the difference between this result and the regular jet correlations : indeed a two - hadron production cross section calculated in perturbative qcd ( pqcd ) , illustrated by the examples in , contains both the back - to - back and near - side correlations .
however , while the jet back - to - back correlations are long - range in rapidity , the jet near - side correlations are local in rapidity , being due to collinear jet showers around the trigger hadron ( see @xcite for an analysis of mini - jet correlations in heavy ion collisions ) . hence in standard pqcd formalism the strengths along with rapidity and azimuthal shapes of the back - to - back and near - side correlations are different .
this is in stark contrast to our correlations ( iii ) and ( iv ) , which are both long - range in rapidity , and have identical amplitudes and rapidity / azimuthal shapes .
the correlations ( iii ) and ( iv ) arise due to multiple rescatterings in both nuclei and are , therefore , typical of the ion
ion collisions , weakening in the limit when one ( or both ) of the ions is ( are ) small ( a proton ) .
it is possible that this correlation would complicate the experimental extraction of the elliptic flow observable @xmath241 due to the qgp flow in heavy ion collisions , though it may also be that the correlation is too weak to present a significant background .
more detailed numerical analysis of our result is needed to clarify this point .
the authors are grateful to adrian dumitru , ulrich heinz , jamal jalilian - marian , mike lisa , and kirill tuchin for inspiring discussions of two - particle correlations .
this research is sponsored in part by the u.s .
department of energy under grant no .
de - sc0004286 .
+ after this paper had been posted online as a preprint , a new experimental result @xcite appeared , reporting on the discovery of identical near- and away - side correlations found in the @xmath268 collisions at the lhc , in qualitative agreement with one of the main results of this work .
our goal here is to calculate the following object @xmath269 \ , tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle\end{aligned}\ ] ] in the quasi - classical mv / gm approximation at the lowest non - trivial order in @xmath139 expansion in the t hooft s large-@xmath24 limit .
we assume that the saturation scale is @xmath24-independent , that is @xmath270 , due to an order-@xmath137 number of `` valence '' partons in each nucleon in the target nucleus .
a sample of the diagrams contributing to is shown in , where the @xmath271-channel gluons are denoted by dashed lines to simplify the picture .
the calculation is straightforward @xcite : one simply has to exponentiate the two - gluon exchange interaction with a single nucleon ( the nucleons are denoted by shaded ovals in ) .
the only complication is that , unlike the dipole amplitude calculated in @xcite , the interaction now , for the double - trace operator , is a matrix in the color space .
a double trace operator like with the wilson lines in the fundamental representation was calculated earlier in @xcite ( for similar calculations see also @xcite ) . to exponentiate the matrix we have to choose a basis in the color space of four @xmath272-channel gluons in :
clearly the net color of the four gluons is always zero .
the color states of the four gluons can be classified according to the color states of the top two @xmath272-channel gluons , since the color of the bottom pair of @xmath272-channel gluons is determined by requiring net color - neutrality of the four @xmath272-channel gluon system .
the colors of a gluon pair can be decomposed in the following irreducible representations @xmath273 here we are following the notations introduced in @xcite , see page 120 there . labeling the colors of the four @xmath272-channel gluons in an arbitrary color state by @xmath274 , and @xmath275
as shown in we define the color states corresponding to representations @xmath276 , @xmath277 , and @xmath278 by @xmath279 where we differ from @xmath280 s in @xcite by prefactors , since here we demanded that our color states are normalized to one , @xmath281 .
other color states can be constructed as well @xcite , but we will only need the states in for the calculation below .
we denote by @xmath282 the interaction with a single nucleon by a two - gluon exchange : it is a matrix in the color space of the @xmath272-channel gluons . since in the correlator the top ( bottom ) two @xmath272-channel gluons are in the color - singlet state both before and after the interaction , we write @xmath283 \ , tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle = \langle p_1 | e^{\hat m } | p_1 \rangle . \end{aligned}\ ] ] expanding the exponential in a power series and inserting unit operators @xmath284 between all the @xmath282 s yields @xmath285 \ , tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle = \left ( e^m \right)_{11}\end{aligned}\ ] ] where the @xmath286 matrix @xmath287 is defined by its elements , @xmath288 all one has to do now is to find the matrix @xmath287 from , exponentiate it , picking up the `` @xmath289 '' matrix element of the exponential .
such a calculation , while straightforward , is rather involved : here we will utilize the large-@xmath24 limit to construct an approximate result .
calculating some of the elements of the matrix @xmath287 and evaluating the @xmath24-order of the remaining matrix elements yields @xmath290 with @xmath151 defined in eqs .
( [ ds ] ) and ellipsis denoting the matrix elements we do not need to calculate as they are at most order-1 in @xmath24 counting . from we
see that if we start in the color - singlet state @xmath291 for the top ( bottom ) gluon pair , a single interaction can either leave the two gluons in the color - singlet state , or can flip them into a color - octet state @xmath292 .
the latter transition comes in with an order-@xmath293 suppression factor . in order to evaluate we have to start and finish with a color - singlet state : to order-@xmath139
we may have at most two such transitions : @xmath294 and the inverse , @xmath295 .
once the system is in the color - octet @xmath296 state , it can continue its random walk through color space : however , if we want to keep the calculation at the order-@xmath139 , the interactions between the two transitions should be leading - order in @xmath24 .
this is why only the leading-@xmath24-order matrix elements @xmath297 , @xmath298 , and @xmath299 contribute in the @xmath300 matrix @xmath301 with @xmath302 : we do not need to calculate the @xmath293-suppressed elements in or the elements denoted by ellipsis which can not contribute . exponentiating the matrix @xmath287 from , picking up the `` @xmath289 '' element of the obtained matrix and expanding the result to order-@xmath139 yields @xmath303 \ , tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle = e^{d_1 } + \frac{(d_3-d_2)^2}{n_c^2 } \ , & \left [ \frac{e^{d_1}}{d_1-d_2 } - \frac{2 \ , e^{d_1}}{(d_1-d_2)^2 } + \frac{e^{d_1}}{d_1-d_3 } - \frac{2 \ , e^{d_1}}{(d_1-d_3)^2 } \right .
\notag \\ & \left .
+ \frac{2 \ , e^{\frac{1}{2}(d_1+d_2)}}{(d_1-d_2)^2 } + \frac{2 \ , e^{\frac{1}{2}(d_1+d_3)}}{(d_1-d_3)^2 } \right ] + o \left ( \frac{1}{n_c^4 } \right).\end{aligned}\ ] ] finally , noting that ( see ) @xmath304 \rangle \ , \langle tr [ u_{{\bm x}_3 } u_{{\bm x}_4}^\dagger ] \rangle = e^{d_1}\end{aligned}\ ] ] and using we obtain .
collaboration , j. adams _ et .
_ , _ distributions of charged hadrons associated with high transverse momentum particles in p p and au + au collisions at s(nn)**(1/2 ) = 200-gev _ , _ phys .
* 95 * ( 2005 ) 152301 , [ http://xxx.lanl.gov/abs/nucl-ex/0501016 [ nucl - ex/0501016 ] ] .
collaboration , b. alver _ et .
al . _ , _ high transverse momentum triggered correlations over a large pseudorapidity acceptance in au+au collisions at @xmath306=200 gev _ , _ phys .
* 104 * ( 2010 ) 062301 , [ http://xxx.lanl.gov/abs/0903.2811 [ arxiv:0903.2811 ] ] .
collaboration , b. abelev _ et .
_ , _ long range rapidity correlations and jet production in high energy nuclear collisions _ , _ phys.rev .
_ * c80 * ( 2009 ) 064912 , [ http://xxx.lanl.gov/abs/0909.0191[arxiv:0909.0191 ] ] .
collaboration , v. khachatryan _ et .
_ , _ observation of long - range near - side angular correlations in proton - proton collisions at the lhc _ , _ jhep _ * 09 * ( 2010 ) 091 , [ http://xxx.lanl.gov/abs/1009.4122[arxiv:1009.4122 ] ] .
a. dumitru , f. gelis , l. mclerran , and r. venugopalan , _ glasma flux tubes and the near side ridge phenomenon at rhic _ , _ nucl .
phys . _ * a810 * ( 2008 ) 91108 , [ http://xxx.lanl.gov/abs/0804.3858 [ arxiv:0804.3858 ] ] .
s. gavin , l. mclerran , and g. moschelli , _ long range correlations and the soft ridge in relativistic nuclear collisions _ , _ phys .
* c79 * ( 2009 ) 051902 , [ http://xxx.lanl.gov/abs/0806.4718 [ arxiv:0806.4718 ] ] .
n. armesto , l. mclerran , and c. pajares , _ long range forward - backward correlations and the color glass condensate _ , _ nucl.phys . _ * a781 * ( 2007 ) 201208 , [ http://xxx.lanl.gov/abs/hep-ph/0607345 [ hep - ph/0607345 ] ] .
k. dusling , f. gelis , t. lappi , and r. venugopalan , _ long range two - particle rapidity correlations in a+a collisions from high energy qcd evolution _ , _ nucl .
_ * a836 * ( 2010 ) 159182 , [ http://xxx.lanl.gov/abs/0911.2720[arxiv:0911.2720 ] ] .
a. dumitru , k. dusling , f. gelis , j. jalilian - marian , t. lappi , _ et .
_ , _ the ridge in proton - proton collisions at the lhc _ , _ phys.lett . _ * b697 * ( 2011 ) 2125 , [ http://xxx.lanl.gov/abs/1009.5295 [ arxiv:1009.5295 ] ] .
a. dumitru and j. jalilian - marian , _ two - particle correlations in high energy collisions and the gluon four - point function _ , _ phys .
* d81 * ( 2010 ) 094015 , [ http://xxx.lanl.gov/abs/1001.4820 [ arxiv:1001.4820 ] ] .
f. gelis , t. lappi , and r. venugopalan , _ high energy factorization in nucleus - nucleus collisions .
3 . long range rapidity correlations _ , _ phys.rev .
_ * d79 * ( 2009 ) 094017 , [ http://xxx.lanl.gov/abs/0810.4829[arxiv:0810.4829 ] ] .
j. jalilian - marian and y. v. kovchegov , _ saturation physics and deuteron gold collisions at rhic _ , _ prog .
* 56 * ( 2006 ) 104231 , [ http://xxx.lanl.gov/abs/hep-ph/0505052 [ hep - ph/0505052 ] ] . l. d. mclerran and r. venugopalan , _ gluon distribution functions for very large nuclei at small transverse momentum _ , _ phys .
* d49 * ( 1994 ) 33523355 , [ http://xxx.lanl.gov/abs/hep-ph/9311205 [ hep - ph/9311205 ] ] .
l. d. mclerran and r. venugopalan , _ computing quark and gluon distribution functions for very large nuclei _
* d49 * ( 1994 ) 22332241 , [ http://xxx.lanl.gov/abs/hep-ph/9309289 [ hep - ph/9309289 ] ] .
a. kovner , l. d. mclerran , and h. weigert , _ gluon production from nonabelian weizsacker - williams fields in nucleus - nucleus collisions _ , _ phys .
* d52 * ( 1995 ) 62316237 , [ http://xxx.lanl.gov/abs/hep-ph/9502289[hep-ph/9502289 ] ] .
y. v. kovchegov and d. h. rischke , _ classical gluon radiation in ultrarelativistic nucleus nucleus collisions _ , _ phys .
* c56 * ( 1997 ) 10841094 , [ http://xxx.lanl.gov/abs/hep-ph/9704201 [ hep - ph/9704201 ] ] .
a. krasnitz , y. nara , and r. venugopalan , _ classical gluodynamics of high energy nuclear collisions : an erratum and an update _ , _ nucl .
phys . _ * a727 * ( 2003 ) 427436 , [ http://xxx.lanl.gov/abs/hep-ph/0305112 [ hep - ph/0305112 ] ] .
j. jalilian - marian , a. kovner , and h. weigert , _ the wilson renormalization group for low x physics : gluon evolution at finite parton density _ , _ phys .
_ * d59 * ( 1998 ) 014015 , [ http://xxx.lanl.gov/abs/hep-ph/9709432[hep-ph/9709432 ] ] .
j. jalilian - marian , a. kovner , a. leonidov , and h. weigert , _ the wilson renormalization group for low x physics : towards the high density regime _ , _ phys .
* d59 * ( 1998 ) 014014 , [ http://xxx.lanl.gov/abs/hep-ph/9706377[hep-ph/9706377 ] ] .
k. dusling and r. venugopalan , _ azimuthal collimation of long range rapidity correlations by strong color fields in high multiplicity hadron - hadron collisions _ , _ phys.rev.lett . _
* 108 * ( 2012 ) 262001 , [ http://xxx.lanl.gov/abs/1201.2658[arxiv:1201.2658 ] ] .
e. gardi , j. kuokkanen , k. rummukainen , and h. weigert , _ running coupling and power corrections in nonlinear evolution at the high - energy limit _ , _ nucl .
phys . _ * a784 * ( 2007 ) 282340 , [ http://xxx.lanl.gov/abs/hep-ph/0609087[hep-ph/0609087 ] ] .
a. krasnitz and r. venugopalan , _ the initial energy density of gluons produced in very high energy nuclear collisions _ , _ phys .
* 84 * ( 2000 ) 43094312 , [ http://xxx.lanl.gov/abs/hep-ph/9909203[hep-ph/9909203 ] ] .
y. v. kovchegov , _ non - abelian weizsaecker - williams field and a two- dimensional effective color charge density for a very large nucleus _ , _ phys .
* d54 * ( 1996 ) 54635469 , [ http://xxx.lanl.gov/abs/hep-ph/9605446[hep-ph/9605446 ] ] .
y. v. kovchegov , _ quantum structure of the non - abelian weizsaecker - williams field for a very large nucleus _ , _ phys .
rev . _ * d55 * ( 1997 ) 54455455 , [ http://xxx.lanl.gov/abs/hep-ph/9701229[hep-ph/9701229 ] ] .
k. eggert , h. frenzel , w. thome , b. betev , p. darriulat , _ et .
_ , _ angular correlations between the charged particles produced in @xmath310 collisions at isr energies _ , _ nucl.phys .
_ * b86 * ( 1975 ) 201 .
y. v. kovchegov and a. h. mueller , _ gluon production in current nucleus and nucleon nucleus collisions in a quasi - classical approximation _ , _ nucl .
phys . _ * b529 * ( 1998 ) 451479 , [ http://xxx.lanl.gov/abs/hep-ph/9802440[hep-ph/9802440 ] ] .
b. z. kopeliovich , a. v. tarasov , and a. schafer , _ bremsstrahlung of a quark propagating through a nucleus _ , _ phys .
_ * c59 * ( 1999 ) 16091619 , [ http://xxx.lanl.gov/abs/hep-ph/9808378 [ hep - ph/9808378 ] ] .
j. p. blaizot , f. gelis , and r. venugopalan , _ high energy p a collisions in the color glass condensate approach .
i : gluon production and the cronin effect _ , _ nucl .
phys . _ * a743 * ( 2004 ) 1356 , [ http://xxx.lanl.gov/abs/hep-ph/0402256[hep-ph/0402256 ] ] .
j. jalilian - marian and y. v. kovchegov , _ inclusive two - gluon and valence quark - gluon production in dis and p a _ , _ phys .
rev . _ * d70 * ( 2004 ) 114017 , [ http://xxx.lanl.gov/abs/hep-ph/0405266 [ hep - ph/0405266 ] ] .
f. dominguez , a. mueller , s. munier , and b .- w .
xiao , _ on the small - x evolution of the color quadrupole and the weizscker - williams gluon distribution _ , _ phys.lett .
_ * b705 * ( 2011 ) 106111 , [ http://xxx.lanl.gov/abs/1108.1752[arxiv:1108.1752 ] ] .
a. dumitru , j. jalilian - marian , t. lappi , b. schenke , and r. venugopalan , _ renormalization group evolution of multi - gluon correlators in high energy qcd _ , _ phys.lett . _ * b706 * ( 2011 ) 219224 , [ http://xxx.lanl.gov/abs/1108.4764[arxiv:1108.4764 ] ] . p. m. chesler and l. g. yaffe , _ holography and colliding gravitational shock waves in asymptotically ads@xmath311 spacetime _ , _ phys.rev.lett . _
* 106 * ( 2011 ) 021601 , [ http://xxx.lanl.gov/abs/1011.3562 [ arxiv:1011.3562 ] ] .
t. lappi , s. srednyak , and r. venugopalan , _ non - perturbative computation of double inclusive gluon production in the glasma _ , _ jhep _ * 1001 * ( 2010 ) 066 , [ http://xxx.lanl.gov/abs/0911.2068 [ arxiv:0911.2068 ] ] .
n. borghini , p. m. dinh , and j .- y .
ollitrault , _ a new method for measuring azimuthal distributions in nucleus - nucleus collisions _ , _ phys.rev . _
* c63 * ( 2001 ) 054906 , [ http://xxx.lanl.gov/abs/nucl-th/0007063 [ nucl - th/0007063 ] ] .
u. w. heinz and b. v. jacak , _ two particle correlations in relativistic heavy ion collisions _ , _
_ * 49 * ( 1999 ) 529579 , [ http://xxx.lanl.gov/abs/nucl-th/9902020 [ nucl - th/9902020 ] ] .
collaboration , j. adams _ et .
_ , _ azimuthally sensitive hbt in au + au collisions at s(nn)**(1/2 ) = 200-gev _ , _ phys.rev.lett .
_ * 93 * ( 2004 ) 012301 , [ http://xxx.lanl.gov/abs/nucl-ex/0312009[nucl-ex/0312009 ] ] .
t. csorgo and s. s. padula , _ disappearance of squeezed back - to - back correlations : a new signal of hadron freeze - out from a supercooled quark gluon plasma _ , _ braz.j.phys . _
* 37 * ( 2007 ) 949962 , [ http://xxx.lanl.gov/abs/0706.4325[arxiv:0706.4325 ] ] . c. marquet and h. weigert , _ new observables to test the color glass condensate beyond the large-@xmath24 limit _ , _ nucl .
_ * a843 * ( 2010 ) 6897 , [ http://xxx.lanl.gov/abs/1003.0813 [ arxiv:1003.0813 ] ] .
n. nikolaev , w. schafer , and b. zakharov , _ nonlinear k(perpendicular)-factorization for gluon - gluon dijets produced off nuclear targets _ , _ phys.rev . _ * d72 * ( 2005 ) 114018 , [ http://xxx.lanl.gov/abs/hep-ph/0508310[hep-ph/0508310 ] ] . | we study two - particle long - range rapidity correlations arising in the early stages of heavy ion collisions in the saturation / color glass condensate framework , assuming for simplicity that one colliding nucleus is much larger than the other .
we calculate the two - gluon production cross section while including all - order saturation effects in the heavy nucleus with the lowest - order rescattering in the lighter nucleus .
we find four types of correlations in the two - gluon production cross section : ( i ) geometric correlations , ( ii ) hbt correlations accompanied by a back - to - back maximum , ( iii ) away - side correlations , and ( iv ) near - side azimuthal correlations which are long - range in rapidity . the geometric correlations ( i ) are due to the fact that nucleons are correlated by simply being confined within the same nucleus and may lead to long - range rapidity correlations for the produced particles without strong azimuthal angle dependence . somewhat surprisingly , long - range rapidity correlations ( iii ) and ( iv ) have exactly the same amplitudes along with azimuthal and rapidity shapes : one centered around @xmath0 with the other one centered around @xmath1 ( here @xmath2 is the azimuthal angle between the two produced gluons ) .
we thus observe that the early - time cgc dynamics in nucleus - nucleus collisions generates azimuthal non - flow correlations which are qualitatively different from jet correlations by being long - range in rapidity .
if strong enough , they have the potential of mimicking the elliptic ( and higher - order even - harmonic ) flow in the di - hadron correlators : one may need to take them into account in the experimental determination of the flow observables . |
on a smooth metal - dielectric interface light does not couple to elementary electronic excitations of the metal surfaces because of energy and momentum conservation .
but in a metal film that is perforated with a two - dimensional ( 2d ) periodic array of holes forming a 2d metallo - dielectric photonic crystal , the periodicity promotes zone folding that results in the formation of band structure .
the grating makes it possible for light to couple directly to surface electromagnetic waves by transfering them the reciprocal lattice momentum @xmath1 , where @xmath2 is the grating period @xcite .
indeed , it was recently found@xcite that the optical transmission through hole arrays fabricated on optically thick metallic films is enhanced by orders of magnitude at resonances ; this phenomenon has been dubbed `` extraordinary transmission '' ( eot ) .
the maxima in the zero - order transmission spectrum is alternated by deep minima caused by wood s anomalies formed due to the diffraction grating of the corrugated film .
the optical properties of various subwavelength hole array films have been widely studied and different models have been proposed to explain the underlying mechanism of the anomalous transmission . while earlier models atributed the enhancement to excitation of film s surface plasmon - polaritons ( spp ) @xcite , more recent works adrresed the role resonant cavity modes @xcite , dynamical diffraction resonances @xcite , and interference of diffracted evanescent waves@xcite . while the precise mechanism of extraordinary transmission is still under discussion , the transient dynamics of surface electromagnetic modes might shed more light on the origins of this phenomenon . here
we investigate the ultrafast dynamics of surface electromagnetic modes measured by using the pump - probe correlation spectroscopy of the transmission bands in 2d subwavelength hole array on aluminum ( al ) thin film .
we found significant difference with the dynamics observed previously in smooth metal films ( mostly gold and silver ) in both the time - resolved and spectrally - resolved signals . following an instantaneous rise at the onset of the impinging pulse , the transient differential transmission exhibits a fast rise on subpicosecond timescale followed by a plateau for time delays of 1 - 3 ps , and subsequent slow decay with characteristic time of @xmath3 ps .
the short - time dynamics is accompanied by a blueshift of the main transmission band in the transient spectra that persists for time delays as long as 100 ps .
the observed dynamics can be explained by a fast energy transfer in the al film from the electron gas to the lattice , with subsequent cooling of the al film by heat transfer to the glass substrate .
the fast lattice temperature rise is caused by strong electron - phonon interaction in al , which makes the electron - lattice energy transfer rate comparable to the rate of nonequilibrium electrons thermalization via electron - electron interactions@xcite .
importantly , the blueshift is incosistent with the spp mechanism and points to the interference as the source of the extraordinary transmission .
the plasmonic lattice sample was a 70 nm thick al film , 5 @xmath4 5 mm@xmath5 in area deposited on a glass substrate .
the metal film was perforated with circular holes having diameter @xmath6 nm in a square hole array with lattice constant @xmath7 nm resulting in a fractional aperture area of 25% .
the ultrafast laser system used for measuring the transient pm spectrum was a ti : sapphire regenerative amplifier with pulses of 100 fs duration at photon energies of 1.55 ev , with 400 @xmath8j energy per pulse at a repetition rate of 1 khz .
the second harmonic of the fundamental pulses at 390 nm ( 3.1 ev ) were used as the pump beam .
the probe beam was a white light super - continuum generated in a 1-mm thick sapphire plate that covers the spectral range from 1.6 to 2.8 ev ; the white light super - continuum was also used for measuring the unperturbed transmission spectrum @xmath9 .
the pump and probe beams were directed onto the perforated al film from the air side to maximize coupling of light to the sew .
the transient pm spectrum was obtained from the photoinduced change ( @xmath10 ) in @xmath11 using a phase - sensitive technique with a resolution @xmath12 that corresponds to a photoexcited electron density in the al metal of @xmath13 @xmath14pulse . through an al plasmonic lattice comprised of a square array of nanoholes with lattice constant 300 nm and hole diameter 150 nm .
the inset shows the sample afm image .
numerical fit is based on the effective medium model for linear transmission in sec .
[ sec : effective ] . ]
figure [ fig:1 ] shows the eot spectrum of the al plasmonic lattice at normal incidence .
several eot bands and transmission anti - resonances ( ar ) are observed .
we note that @xmath9 at resonance frequencies substantially surpasses the fractional aperture area of 25% .
the lowest frequency eot band peaks at @xmath15 ev ; whereas the lowest frequency ar feature that is associated with wood s anomaly at the al / glass interface with wavevector @xmath16 ( where @xmath17 and @xmath1 ) is in the form of a dip at 2.6 ev .
the ar s at higher frequencies are higher - order wood s anomalies that correspond to wavevectors @xmath18 ( _ m , n _ are integers ) as follows : 3.7 ev ( @xmath19 ) that corresponds to the ( 1,1 ) branch at the al / glass interface ; and 4.0 ev ( @xmath20 ) that corresponds to the ( 0,1 ) branch at the al / air interface .
it is noteworthy that the lowest frequency eot resonance ( at @xmath15 ev ) occurs at considerably lower ( by @xmath21 ev ) frequency than that predicted by the lowest - order spp with wavevector satisfying the bragg condition @xcite @xmath22^{1/2}.\end{aligned}\ ] ] . , width=240 ] the transient pm spectra of the perforated al film are shown in figs .
[ fig:2 ] - [ fig:4 ] .
the time - evolution of pump - probe signal at 650 nm , corresponding to probe photon energy of 1.9 ev , reveals two component of the dynamics .
firstly , there is a fast rise of pump - probe signal on timescale @xmath23 ps , followed by a plateau at 1 - 3 ps ( see fig .
[ fig:2 ] ) .
secondly , after about 3 ps the saturated pm decays with characteristic time of @xmath3 ps ( see fig .
[ fig:3 ] ) due the heat transfer to the glass substrate .
figure [ fig:4](a ) shows the transient pm spectral change from 1.7 to 2.4 ev corresponding roughly to the main transmission band spectral width .
a transient blueshift is apparent that persists for times up to 100 ps , i e. much longer that electron gas thermalization time , pointing to a prominent role of _ e - ph _ interactions in the dynamics .
however this shift can not be well reproduced by a simple spectral derivative [ see fig .
[ fig:4](b ) ] indicating that a perturbation approach can not properly describe the observed changes in optical spectra for our excitation intensities . at the same time , although the data becomes more noisy at higher energies as one approaches to the wood anomaly , a clear nonlinearity in the frequency dependence emerges .
the energy dependence of @xmath24 is determined by transient change in the mode dispersion that reflects the periodicity of the structure . .
, width=240 ] the short - time dynamics exhibts significant differences with that observed in gold and silver films @xcite . in these studies ,
the picosecond dynamics consisted of two well - separated in time stages : thermalization of nonequilibrium electrons due to electron - electron ( _ e - e _ ) scatering at 200 - 500 fs accompanied by the signal rise , followed by the decay on a 1 - 2 ps timescale due to the cooling of hot electron gas to the metal lattice .
the dominant factor that determines the transient dynamics in noble metals is the change in the interband succeptibilily of the highly polarizable _
d_-band electrons that is very sensitive to the smearing of the fermi edge due to the rise in electron temperature @xmath25 .
in contrast , in simple metals such as al , the transient optical spectra are governed by the changes in the intraband dielectric function @xmath26 , where @xmath27 is the bulk plasmon energy and @xmath28 is the optical relaxation rate @xcite .
the absorbance of the film and hence the transient dynamics is related to the deviation of @xmath28 from its intrinsic value @xmath29 that mainly originates from the change in the electron - phonon ( _ e - ph _ ) scattering , while the contributions from _ e - e _ , and impurity scatering are relatively weak @xcite .
note that the _ e - ph _ interaction constant @xmath30 in al is an order of magnitude larger than it is in noble metals ( for comparison , @xmath31 w / km@xmath32 vs. @xmath33 w / km@xmath32 ) , so that the energy exchange between the electron and phonon subsystems takes place on sub - picosecond times @xcite .
the rapid increase in the lattice temperature @xmath34 and hence in the _ e - ph _ relaxation rate accounts for the fast rise of the pump - probe signal until the equilibrium between electron and phonon subsystems is reached at @xmath35 ps .
this fast dynamics is superimposed with a slow decay due to the heat transfer from metal to glass substrate resulting in a plateau at 1 - 3 ps ( see figs . [ fig:2]-[fig:3 ] ) .
importantly , the merging of electron thermalization and electron to lattice cooling stages implies that the short time dynamics is intrinsically nonequilibrium ; it is therefore described below within the boltzmann equation approach . .
( b ) log derivative of measured zero - order transmission .
, width=240 ] the observed pm blueshift also indicates that the eot is sensitive to the imaginary part of permittivity , @xmath36 , for the following reason . within the drude model for al
, we have @xmath37 . with such @xmath38 , the spp bragg condition eq .
( [ bragg ] ) leads to the following expression for , e.g. , ( 0,1 ) mode frequency ( at @xmath39 ) : @xmath40 $ ] , where @xmath41 is the frequency associated with the reciprocal lattice period .
then an increase in @xmath28 upon pulsed excitation leads to a weak redshift of the resonance frequency , in disagreement with the data .
we thus conclude that the observed spectral blueshift can not be explained by the present version of the spp mechanism for the eot phenomenon @xcite . at the the same time
, an increase in @xmath28 causes an increase in the _ attenuation _ , @xmath42 , which for @xmath43 can be approximated by @xmath44 .
if @xmath42 participates in determining the position and lineshape of the eot bands , then the increase in @xmath28 implies _ higher _ resonant frequencies ( blueshift ) in agreement with the data .
the linear transmission is modeled below using effective medium approach .
the thansient optical spectra in metal films can be described in terms of time - dependent dielectric function @xmath45 , where @xmath46 is pm contribution due to a small ( compared to ground state ) fraction of photoexcited carriers .
the differential transmission @xmath47 can be then caculated from the linear transmission @xmath9 with @xmath46 @xcite . while the time - resolved signal is primarily governed by the electron dynamics in metal film encoded in @xmath48 , the transient _ spectra
_ are determined by the effect of the latter on the _ dispersion _ of surface electromagnetic modes .
modeling of the dynamics thus involves two distinct tasks : ( i ) description of the linear transmision through nanohole array for _ real _ metal with absorptive dielectric function , and ( ii ) simulation of non - equilibrium electron - phonon dynamics in al film .
we begin with _ analytical _ model for the main transmission band that incorporates the effect of the absorptive part of metal dielectric function .
we are interested in the transmission of _
p_-polarized light normally incident on a metal film perforated with square array of cylindrical nanoholes .
we assume that the scatering in the direction normal to the plane of incidence is weak and adopt one - dimentional grating approximation for transverse magenetic and electric fields , @xmath49 and @xmath50 .
magnetic fields in air ( media 1 ) , perforated metal ( media 2 ) , and glass ( media 3 ) , can be written as @xmath51,\end{aligned}\ ] ] where @xmath52 and @xmath53 ( here @xmath54 ) , and @xmath55 , and @xmath56 are reflection and transmission amplitudes of order @xmath57 in medium @xmath58 respectively ( @xmath59 ) . in air / glass , the fields
are expanded in the bloch functions basis . in the medium 2 ( perforated metal ) ,
the expansion goes over the normalized eigenstates of maxwell equations ( me ) in periodic medium , @xmath60 , where eigenvalue @xmath61 is the wave - vector along the normal to the surface .
the eigenfunctions @xmath62 are localized in the hole region and are periodic over the array .
the transverse electric fields in all three media are expressed through transverse magnetic fields via me @xmath63 . in the relevant spectral range , it is sufficient to restrict to the lowest modes in air / glass ( @xmath64 ) , and in perforated metal @xmath65 .
the transmission coefficient is obtained by matching transverse electric and magnetic fields at air / metal and metal / glass interfaces .
it is convenient to perform this matching in the bloch basis for electric field , and in the basis of eigenstates @xmath66 for magnetic field .
this choice takes into account the fact that , for low volume fraction of holes , @xmath67 , the electric field is mostly reflected back into the air at the air / metal interface and transmitted to the glass at the metal / glass interface , while the magnetic field , conversely , mostly penetrates into the metal at the air / metal interface , and gets reflected back into the metal from the metal / glass boundary .
we define magnetic and electric field moments averaged over the brilloin zone ( bz ) as @xmath68 where @xmath69 is periodic dielectric function , @xmath70 ( @xmath71 ) being dielectric function in metal ( hole ) region , and @xmath72 is the periodic step - function vanishing outside the holes . because we consider only the fundamental mode , we drop index @xmath73 hereafter . by performing matching procedure at the interfaces @xmath74 and
@xmath75 we obtain the following system of equations @xmath76 with @xmath77 , yielding the zero - order transmission @xmath78 where @xmath79 , @xmath80 are effective transmission amplitudes through the air / metal , and metal / glass interfaces respectively , and @xmath81 , @xmath82 are the corresponding effective reflection amplitudes .
the effective refraction indices @xmath83 are given by @xmath84 where @xmath85 are refraction indices in air / glass ( @xmath86 ) .
the transmission coefficient is thus expressed via the _ overlap integrals _ , @xmath87 and @xmath88 , between eigenstates in the periodic medium and in the air / glass .
such overlap causes hybridization of the states in perforated metal via the em waves at the interfaces ( both propagating and evanescent ) .
it is this hybridization that determines the position and the lineshape of the transmission peak . in order to proceed further
, we need to evaluate the overlap integrals , @xmath87 and @xmath88 , and the eigenvalue @xmath89 .
the former can be determined from the following self - consistency argument . while in the bulk of perforated metal the states @xmath66 are strongly localized within holes , near the interfaces ( where the overlap integrals are calculated ) these states become delocalized due to the hybridization via em in the dielectric . in other words , in the close vicinity of the interface , the profile of transverse magnetic field ( that is continuous together with its derivative accross the interfaces ) is similar at the both sides of the interface and can be approximated as @xmath90 .
the relation between @xmath91 and @xmath92 on the metal side can therefore be determined from me for @xmath93 , @xmath94 with periodically modulated dielectric function , @xmath95 , where @xmath96 is the average and @xmath97 , with @xmath98 , is the modulation amplitude @xcite . in the first order in modulation
, we obtain @xmath99 where @xmath91 is determined from the normalization condition .
the moments of electric field , @xmath88 , are straightforwardly calculated from eq .
( [ bz ] ) as @xmath100/k_0,\end{aligned}\ ] ] where @xmath101 .
let us now turn to evaluation of the wavevector @xmath89 that represents the lowest eigenvalue of me eq .
( [ me ] ) .
the eigenvalues are determined by the bulk states of the optically thick perforated metal film , and therefore are only weakly affected by the interfaces . in the bulk , the eigenstates are strongly localized within the holes and the weakly - modulated medium approximation does not apply . in this case , the eigenvalues can be evaluated by using the analogy between me and schrdinger equation ( se ) with complex potential .
namely , for each region , eq . (
[ me ] ) can be presented in the form @xmath102 , where @xmath103 , \nonumber\\ & & v(x ) = k_0 ^ 2\sum_n\bigl[\epsilon(\omega)-\epsilon_h\bigr]\theta\bigl(d/2 -|x+na|\bigr),\end{aligned}\ ] ] are the energy and potential , respectively .
the latter is kronig - penney type attractive potential of periodically spaced quantum wells .
note that @xmath104 corresponds to @xmath105 that is the value of @xmath89 for a solid metal film .
the fundametal mode is associated to the ground state of eq .
( [ se ] ) with the energy spectrum being a band with the width determined by the tunnel transparency between the wells . in the case of high metal
/ dielectric contrast of @xmath106 , so that @xmath107 , this transparency is exponentially small and the ground state energy can be approximated by that of the lowest level in a single well .
the latter can be found by replacing rectangular well by the delta - function potential , @xmath108 , with the strength @xmath109 , where we averaged over bz .
we then obtain @xmath110 , yielding @xmath111 .
we thus arrive at the following physical picture for the main transmission band .
diffraction on an individual hole can be viewed as scattering on the bound state in the associated se .
the interference of diffracted waves , caused by the hybridization of localized hole states via em waves in dielectrics , leads to the single - period modulation of magnetic field profile at the interfaces .
the latter is determined by the _ effective medium _ characterized by weakly - modulated dielectric function depending on the hole volume fraction @xmath112 .
note that eigenvalue @xmath89 as well as magnetic / electric field moments @xmath113 are complex due to absorptive part of metal dielectric function @xmath114 . in fig .
[ fig:1 ] we show the fit of linear trasmission obtained using the drude form of dielectric function of al , @xmath70 , with bulk plasmon frequency @xmath115 ev and damping rate @xmath116 ev @xcite .
it can be see that the main transmission band is described accurately , including the position of wood s anomaly at @xmath117 ev .
higher order maxima , including possible spp peak at @xmath118 ev are not captured within this model .
we now turn to the modeling of ultrafast dynamics .
the transient spectra in a perforated film can be obtained from the linear transmission @xmath9 by the replacement @xmath119 , where @xmath120 is the non - equilibrium contribution to the metal dielectric function . as noted above
, @xmath120 is mainly due to transient change in the _ e - ph _ relaxation rate @xmath121 in the drude dielectric function .
the distinctive feature of electron dynamics in al films is a very fast ( hundreds of fs ) energy exchange between electron and phonon subsystems caused by strong _ e - ph _ interaction in al .
this energy exchange takes place on the same timescale as the electron gas thermalization due to _ e - e _ scattering , rendering the transient dynamics inherently non - equilibrium .
note however that the high - energy fraction of of phoexcited electrons with the excess energy @xmath122 , scatter down on the timescale of tens of fs due to the quadratic energy dependence of _ e - e _ scattering rate , @xmath123 , whereas the thermalization of electron population is reached on the longer timescale of hundreds of fs @xcite . therefore , the higher energy end of electron distribution relaxes primarily via _ e - e _ scattering processes while the energy transfer to the phonons primarily goes through the lower energy electrons .
since the overall heating of the lattice is of the order of debye temperature ( @xmath124 in al ) , the phonons that gain energy from athermal electron gas can be considered as quasi - equilibrium . the energy exchange between phonon and electron subsystems
can be then described in terms of time - dependent latice temperature @xmath125 as @xmath126 where @xmath127 is the lattice heat capacity ( @xmath128 j / kkg for al ) , @xmath129 is the electron density of states ( dos ) , and @xmath130 is the electron distribution function that satisfies the boltzmann equation @xmath131 here the first two terms in r.h.s .
are due to _ e - e _ and _ e - ph _ scattering processes respectively , and @xmath132 stands for electron - gas excitation by the pump pulse .
naturally , only the _ e - ph _ scattering term contributes to the energy exchange rate eq .
( [ lattice ] ) . since the typical phonon energy is much smaller than the electron energy , the _ e - ph _ term can be more conveniently presented in the differential operator form @xcite . in the deformation potential approximation ,
the _ e - ph _ scattering term reads @xmath133 , \label{diffe - ph}\ ] ] where @xmath134 and @xmath30 is _ e - ph _ interaction constant . by substituting eq .
( [ diffe - ph ] ) into ( [ lattice ] ) we obtain the rate equation for the lattice temperature , @xmath135 where @xmath136 \label{teff}\ ] ] plays the role of _ effective _ temperature of _ athermal _ electrons .
note that for thermal electron distribution @xmath137 , @xmath138 coinsides with the `` regular '' electron temperature @xmath139 .
the last term in eq .
( [ 2tphonon ] ) is added to describe heat exchange between the metal film and the glass substrate . with boundary conditions that the temperature and heat flux are continuous accross the metal / glass interface
, it has the form @xcite : @xmath140 , where @xmath141 and @xmath142 are the glass temperature and thermal conductivity , respectively ( @xmath143 j / mk for sio@xmath144 ) .
the above system should be supplemented by equation for heat transport in the glass , @xmath145 where @xmath146 is the glass heat capacity ( @xmath147 j / m@xmath5k for sio@xmath144 ) .
note , however , that metal / glass dynamics is important only for long times ( see below ) .
in the case short - time dynamics , the evolution of lattice temperature @xmath125 is governed by the interaction with strongly athermal electrons gas , and thus requires solution of boltzmann equation for _ e - ph _ system ( [ boltzmann]-[2tphonon ] ) .
we use the integral form for the _ e - e _ scattering term for parabolic band @xcite , @xmath148_{x_{min}}^{x_{max } } , \label{dfdtee}\end{aligned}\ ] ] where @xmath149 , @xmath150 is static dielectric constant of metal , and @xmath151 is the thomas - fermi energy .
the phase space for scattering is determined by @xmath152 , where @xmath153 and @xmath154 , while the angular integrals of coulomb interaction yield the expression in the brackets in the integrand which is evaluated between the limints : @xmath155 /e_s$ ] . the last term in eq .
( [ boltzmann ] ) , describing the energy deposited by the pump pulse , is given by @xmath156 with gaussian pulse profile @xmath157 where @xmath158 is the absorbed pump pulse energy density and @xmath159 = 100 fs is the pulsewidth .
the factor @xmath160 , describing the phase space of phonon - assisted electronic transitions after absorption of a photon with energy @xmath161 , is available in the literature @xcite and is included in the numerical calculations . in fig .
[ fig:5 ] , we plot the time - dependent contribution to the electron distribution function , @xmath162 , versus energy ( measured from the fermi level ) at several time delays . for short times ,
electron distribution is strongly non - equilibrium , as indicated by the long high - energy tail of @xmath163 . at @xmath164
ps , electrons are almost thermalized and their distribution is close to the fermi function of a hot electron gas . note that , at 1 ps , the magnitude of @xmath163 has decreased significantly pointing to the energy transfer from electron gas to the lattice . . ,
the transient change in the lattice temperature , @xmath125 , provides the main contribution to the time - modulation of metal dielectric function , @xmath165,\end{aligned}\ ] ] where @xmath166 is the optical relaxation rate of an electron - hole pair excited by the probe light ( @xmath29 stands for the room - temperature relaxation rate ) . in the case
when the lattice temperature remains much smaller than @xmath161 , the relaxation rate is nearly @xmath161-independent @xcite , and is given by @xmath167 where @xmath168 , and @xmath169 is a material - dependent constant . in the relevant temperature range @xmath170 , the @xmath34dependence of @xmath28 is nearly linear , @xmath171 .
note that @xmath29 contains contributions also from the electron - impurity and _ e - e _ scattering processes ; however those are not significantly affected by the pump pulse due to a relatively small change in the scattering phase space by the fraction of photoexcited electrons .
the results of numerical simulations of short - time dynamics are shown in fig .
[ fig:2 ] .
the calculated differential transmission @xmath24 is in excellent agreement with the data , reproducing the signal rise at subpicosecond times . in fig .
[ fig:6 ] we show the evolution of the lattice temperature and the effective electron temperature .
the rapid drop of @xmath172 after reaching its peak value at @xmath173 fs is accompanied by increase in @xmath34 during the period @xmath174 ps , indicating energy intake by the phonon subsystem while the electron gas is still in the thermalization stage .
after electron and phonon subsystems equilibrate , the signal changes weakly ( plateau ) until the slower process of energy transfer to the glass substrate takes over , leading to the decay of pump - probe signal from the metal [ see fig . [ fig:3 ] ] . for long time delays , electron and phonon subsystems
are in thermal equilibrium with each other , and the long - time dynamics is well descrbed by the _ two - temperature model _ that incorporates the heat exchange between the metal and the glass @xcite . to compare to boltzmann equation approach , here we also included in calculations the rate equation for electron temperature @xmath175 that can be straightforwardly obtained from eqs .
( [ boltzmann]-[2tphonon ] ) by the replacement of @xmath130 with thermal distribution : @xmath176 where @xmath177 ( with @xmath178 j m@xmath179 k@xmath180 for al ) is the electronic heat capacity .
after @xmath3 ps , the signal reaches a plateau corresponding to thermal equilibrium between film and glass [ see fig . [ fig:3 ] ] .
here the evolution of @xmath24 also follows mainly the phonon dynamics [ see fig .
[ fig:6](b ) ] .
note , however , that the subpicosecond signal rise is noticably faster than the one calculated using boltzmann equation approach [ cf .
[ fig:2 ] ] .
the reason is that thermal distribution overestimates the fraction of lower energy electrons and , hence , the rate of their energy transfer to the lattice .
the measured signal also exhibits weak oscillations with a period @xmath181 ps ; those could be attributed to the vibrational mode accross the film that is excited by the rapid heating of the lattice , but are not captured by the present model .
the calculated transient spectra are shown in fig .
[ fig:4 ] for several time delays in the long - time regime when the electron and phonon subsystems are in thermal equilibrium with each other .
the blueshift is well reproduced as well as is the nonlinear @xmath161-dependence of @xmath24 in the main transmission band frequency range . the origin of the blueshift can be understood as follows .
the eigenmodes in a nanohole array are the result of interference of multiple waves scattered from periodic metal / dielectric regions as described by the above effective medium model . in the presence of _ attenuation
_ , the phase of scattered wave has a contribution @xmath182 .
therefore , a transient change in attenuation causes corresponding shifts in scattering phases .
this , in turn , changes the interference of waves making up an eigenmode . to compensate for the phase shifts ,
a transient increase in @xmath28 must be accompanied by a corresponding increase in @xmath161 ; and consequently the transmission resonances blueshift .
in summary , we investigated the ultrafast dynamics of al - based plasmonic lattice that reveals a non - trivial role played by the plasma attenuation in determining the eot resonant bands . while its effect on the spp dispersion is relatively weak , attenuation is more important for modes comprised of a multitude of scattered waves . by directly altering their interference ,
ultrafast time - resolved optical techniques unravel the unique properties of photoexcited sew in plasmonic metals . | we study the ultrafast dynamics of surface electromagnetic waves photogenerated on aluminum film perforated with subwavelength holes array by means of transient photomodulation with @xmath0 fs time resolution . we observed a pronounced _ blueshift _ of the resonant transmission band that reveals the important role of plasma attenuation in the dynamics and that is inconsistent with plasmon - polariton mechanism of extraordinary transmission .
the transient photomodulation spectra were successfully modeled within the boltzmann equation approach for the electron - phonon relaxation dynamics , involving non - equilibrium hot electrons and quasi - equilibrium phonons . |
of the more than 130 different molecules found in interstellar space , an important class is formed by the unsaturated carbonaceous species .
carbohydrides ranging from small molecules such as c@xmath2h @xcite to long chains like c@xmath6h @xcite and hc@xmath7n @xcite have been detected through their millimeter transitions in cold dark clouds like tmc-1 for several decades .
some of these molecules have also been seen in their cyclic form , with c - c@xmath0h@xmath2 as the best - known example @xcite .
even negative ions , in particular c@xmath8h@xmath9 , have now been detected @xcite .
bare carbon chains are likely present as well but do not have a permanent dipole moment and can therefore not be observed through their pure rotational transitions in emission . instead , the smallest members of this family , c@xmath2 and c@xmath0 , have been detected in diffuse interstellar clouds , i.e . , clouds which are not completely opaque to visible and ultraviolet radiation through their electronic absorptions against bright background stars @xcite .
@xmath4c@xmath0h@xmath2 has been seen in diffuse clouds as well , through absorption at radio wavelengths against background quasars @xcite .
the relatively large abundances of these non - saturated molecules , in spite of the fact that there is @xmath10 times more hydrogen than carbon in interstellar clouds , are a vivid demonstration that interstellar chemistry is not in thermodynamic equilibrium . instead
, the kinetics of the reactions that form and destroy these molecules need to be taken into account explicitly in order to explain their abundances . in cold dark clouds ,
these traditionally involve a series of ion - molecule and neutral - neutral reactions @xcite .
more recently , several of the carbohydrides c@xmath11h@xmath12 have also been observed in so - called photon - dominated regions ( pdrs ) @xcite , i.e. , clouds which are exposed to intense ultraviolet ( uv ) radiation .
species like c@xmath2h and c - c@xmath0h@xmath2 have been detected in the orion bar pdr @xcite and even more complex molecules like c@xmath0h@xmath2 and c@xmath1h have been seen in the horsehead nebula , m38 @xcite . together with the above mentioned observations of diffuse clouds , the data demonstrate that these molecules can exist in uv - exposed environments with abundances comparable to those in cold dark clouds .
traditional pdr chemical models can not explain the high abundances of these molecules , leading to speculations that they are perhaps produced by fragmentation of even larger carbonaceous molecules such as polycyclic aromatic hydrocarbons ( pahs ) @xcite .
however , a major uncertainty in these and other models are the photodissociation rates of the carbohydrides , which are basically unknown . other regions in which ultraviolet photons play a role in the chemistry include the envelopes of evolved stars and cometary atmospheres .
the same carbohydride molecules have been detected in the envelope of the carbon - rich late - type star irc+10216 , where their similar distributions are a puzzle for photochemical models @xcite . in cometary atmospheres ,
species as complex as c@xmath1h have been inferred , whose presence is not readily explained by the standard parent - daughter photochemical models @xcite .
many theoretical studies of the electronic structure of small carbohydride molecules exist in the literature ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* e.g. , ) .
however , most of them are limited to the ground and lowest few excited electronic states .
only few studies of higher states exist , most of them performed by peyerimhoff and her co - workers starting more than 30 years ago , e.g. , @xcite as will be shown here , those results are still highly relevant and provide an excellent starting point for further studies .
the aim of this paper is to provide insight into the photodissociation processes of carbohydride molecules through _ ab initio _ quantum chemical calculations of the vertical excitation energies and oscillator strengths for as large a number of excited electronic states as feasible with current programs . by assuming that most of the absorptions with energies above the dissociation limit lead to destruction , estimates of the ( upper limits of ) photodissociation rates under interstellar conditions can be obtained .
even though such calculations are necessarily limited to the smaller members of the carbo - hydride family , they do provide quantitative constraints to test the basic interstellar chemical networks .
specifically , c@xmath2h , c@xmath0 , c@xmath1 , @xmath3 and @xmath4c@xmath0h , @xmath3 and @xmath4 c@xmath0h@xmath2 , hc@xmath0h , @xmath3c@xmath1h and @xmath3c@xmath5h are studied here . except for hc@xmath0h
, all these species have been chosen to have been observed in interstellar space .
for small diatomic and triatomic molecules , quantum chemical calculations of the potential energy surfaces and transition dipole moments combined with dynamical calculations of the nuclear motions can provide photodissociation cross sections and oscillator strengths that agree with experiments to better than 2030% , e.g. , @xcite ( see @xcite for an early review ) .
for the polyatomic molecules considered here , calculations of the full potential surfaces including all degrees of freedom become very time consuming , as do the multi - dimensional dynamics .
moreover , such detail is not needed in order to compute photorates , since those are largely determined by the potentials and transition moments in the franck - condon region .
the simplest alternative is therefore to only compute the vertical excitation energies and transition dipole moment of the molecule at its equilibrium position and assume a certain probability that absorption into each excited state above the dissociation limit leads to dissociation . for the examples cited above
, such an approach leads to similar rates within the accuracy of the calculations .
the focus of our calculations is therefore on electric dipole - allowed transitions to states lying above the lowest dissociation limit but below 13.6 ev .
the interstellar radiation field has a broad spectrum from visible to extreme ultraviolet wavelengths , peaking in intensity around 7 ev and cutting off at the atomic h ionization potential at 13.6 ev @xcite .
all calculations presented here were performed with the wuppertal - bonn mrdci set of programs as implemented in the gamess - uk program package version 7.0 @xcite . for c ,
the tzvp atomic orbital basis set was used , and for h , the dzp basis set @xcite . to allow for a proper description of molecular rydberg states , two diffuse @xmath13 and two diffuse @xmath14 functions
were put on specific sites . for c@xmath2 and c@xmath2h , this was the middle of the cc bond ; for c@xmath2 and c@xmath1h the middle of the c1-c2 and c3-c4 bonds ; for c@xmath0 , @xmath3c@xmath0h , @xmath3c@xmath0h@xmath2 and hc@xmath0h the middle c ; for c@xmath5 on c2 and c4 ; for @xmath4c@xmath0h on the c connected to h ; and for @xmath4c@xmath0h@xmath2 on the lone c. these sites were chosen to avoid linear dependencies in the atomic orbital basis set due to large overlap of diffuse functions positioned on adjacent atoms .
cartesian @xmath14 functions were used , of which the spherical component served as the @xmath15 basis function .
molecular orbitals were generated using up to the maximum number of reference states , 255 .
the selection threshold was generally set at 0.5 @xmath16hartree . the total number of configurations included in the configuration interaction ( ci ) calculation ranged from 200000400000 per symmetry .
the sum over the coefficients - squared in the final ci wave function is typically 0.95 for the smaller species , dropping to 0.9 for the larger molecules .
the ci energies were extrapolated and corrected using the davidson extrapolation @xcite .
the aim was to compute as many excited electronic states as feasible , up to 9 per symmetry .
for the lower states , comparison with existing calculations and experiments indicates accuracies within 0.3 ev , generally better .
oscillator strengths to the lower states agree within 30% or better . for the higher states , typically the 5th root and higher per symmetry
, the accuracy decreases because many states and orbitals can mix .
nevertheless , such calculations should still provide insight into the location of those states , in particular whether they are above or below the ionization potential and below the 13.6 ev cutoff of the interstellar radiation field
. moreover , the magnitude of the oscillator strengths ( strong or weak ) should be reliable .
note that the precise values of the excitation energies are not so important for the purposes of calculating interstellar photorates because of the broad range of incident energies .
the only exception is possible overlap with lyman @xmath17 radiation at 10.2 ev , which is important for certain astrophysical environments @xcite . in all calculations ,
the largest abelian subgroup of the full @xmath18 group , @xmath19 , was used , with the molecule put along the @xmath20axis .
the doubly degenerate states are then split into the @xmath21 and @xmath22 irreducible representations for @xmath23 and @xmath24 states , and into @xmath25 and @xmath26 for the @xmath27 states . in general , the degeneracies for the @xmath27 states are recovered within a few hundreds of an ev in the calculations . the average of the two @xmath25 and @xmath26 values is tabulated here . the @xmath25 irreducible representation also contains the @xmath28 states , and the @xmath26 the @xmath29 states .
some excited states with electronic angular momentum higher than 2 ( in particular @xmath24 states ) are also found in our calculations but they are not tabulated here since the electric dipole transition moments from the ground state to these states are zero , implying that they do not contribute to photodissociation . the valence or rydberg character of the states is determined from calculations of the @xmath30 , @xmath31 and @xmath32 expectation values .
calculations have been limited to the equilibrium geometry corresponding to the lowest energy .
for c@xmath0 , c@xmath2h , c@xmath1 , c@xmath1h and c@xmath5h this is the linear geometry .
c@xmath0h and c@xmath0h@xmath2 have been detected in various isomeric forms so both cyclic and ( near-)linear forms have been studied .
the dissociation energies for c@xmath11h are calculated as the difference between the ground state energy of c@xmath11h at its equilibrium geometry and the energy of ground state c@xmath11 , also at its equilibrium geometry , calculated in the presence of an h atom positioned at a distance of 20 bohr from the center of mass . in this way
the errors associated with the lack of size consistency are significantly reduced .
note that the spatial symmetry of the molecule and the fragment generally differs .
oscillator strengths to all excited states have been calculated according to @xmath33 where all quantities are in atomic units ( a.u . ) , @xmath34 is the transition dipole moment from lower state @xmath35 to upper state @xmath36 , and @xmath37 is a degeneracy factor which is 2 for a @xmath38 transitions and 1 for any other transition . for linear molecules , the dipole moment operator contained in @xmath34
corresponds to @xmath39 for transitions between states with the same value of the electronic angular momentum projection quantum number @xmath40 and to @xmath41 for transitions with @xmath42 .
the computed excitation energies @xmath43 were used in this formula , not any experimental values .
hence , differences with other work can stem from differences in both @xmath34 and in @xmath44 .
the photodissociation rate of a molecule can be computed from @xmath45 where @xmath46 is the photodissociation cross section in @xmath47 and @xmath48 is the mean intensity of the radiation in photons @xmath49
s@xmath50 @xmath50 @xcite . under interstellar conditions ,
only single photon processes are important . for photodissociation initiated by line absorptions ( e.g. , predissociation ) , the rate becomes @xmath51 where @xmath52 the dissociation efficiency of state @xmath36 , which lies between 0 and 1 .
the numerical value of the factor @xmath53 is @xmath54 in the adopted units with @xmath55 in .
the total photodissociation rate of a molecule is obtained by summing over all channels . in this work ,
no dynamical calculations are performed to obtain continuous cross sections for dissociative states .
hence , all photorates are computed with eq . ( 3 ) .
it is furthermore assumed that the dissociation efficiency @xmath52=1 , either through direct dissociation in a repulsive state or by predissociation .
the motivation for this choice is that for larger molecules , internal conversion to a lower ( dissociative ) electronic state is usually much more rapid than any radiative decay rates owing to the high density of states @xcite .
specific experimental evidence of high dissociation efficiencies will be presented in the results section for individual molecules .
the c@xmath2h radical , detected in interstellar clouds since 1974 @xcite , is an important step in the formation of longer carbon chains . in comets
, it could be a photodissociation product of c@xmath2h@xmath2 and a precursor of the widely observed c@xmath2 molecule .
given also its importance in combustion processes , this radical has received ample theoretical attention , starting with the papers by shih , peyerimhoff and co - workers @xcite and culminating with the more recent two - dimensional potential surfaces by duflot et al .
@xcite . as a test of our computational procedure , we present in table [ tabc2h ] our computed vertical excitation energies , oscillator strengths and the main configurations of each state .
our energies generally agree well within 0.3 ev with those of koures and harding @xcite ( their ci + dv2 results ) and lie in between those of duflot et al .
@xcite and shih et al .
close inspection shows that the larger differences are usually for states with rydberg character ( which are not included in duflot et al .
@xcite ) or for states with a strong interaction with a neighboring state , leading sometimes even to a switch in the character of the state . for example
, the 4@xmath56 state in our calculation has rydberg @xmath57 character whereas it has ...
@xmath58 in duflot et al . , resulting also in a very different transition dipole moment ( see below ) . our computed @xmath59 transition dipole moment of 0.22 a.u
. is very close to that of 0.23 a.u . computed by duflot et al
. @xcite and peric et al . @xcite .
because of the slightly higher excitation energy in our work , the oscillator strengths show somewhat larger differences .
comparison for other states is difficult since duflot et al
. do not give any numerical values .
however , their figure 7 shows that neither the 3 and 4 @xmath56 , nor the 2 and 3 @xmath60 states have significant transition dipole moment .
this is generally consistent with our results . for the 5@xmath56 state
we find a huge transition dipole moment of 1.1 a.u . due to the rydberg character of our wavefunction .
thus , this state around 10 ev will dominate the interstellar c@xmath2h photodissociation but the higher @xmath60 states in the 8.510.5 ev range can also contribute significantly up to the ionization potential of @xmath6111.4 ev ( table [ tabdiss ] ) .
the c@xmath0 molecule was detected in cometary spectra in 1882 through its @xmath62 system at 4050 @xcite and in interstellar clouds more than a century later @xcite .
the molecule is also seen in the atmospheres of cool carbon stars through its mid - infrared @xcite and far - infrared @xcite transitions ( see @xcite for review ) .
c@xmath0 is a linear molecule with a ground x@xmath63 state .
its dissociation energy is computed to be about 4.6 ev , whereas its ionization energy is around 12 ev ( table [ tabdiss ] ) .
quantum chemical studies range from the early work by chabalowski et al.@xcite to the recent calculations by terentyev et al .
table [ tabc3 ] summarizes our computed excitation energies , together with the oscillator strengths and main configurations . for the low - lying valence states ,
our results agree to better than 0.3 ev with the mr - aqcc values of monninger et al.@xcite .
these results also show that the @xmath64 ( @xmath65 ) transition at 8.17 ev has by far the largest oscillator strength , as predicted first by pitzer and clementi @xcite .
this is confirmed by the combined experimental and theoretical study by monninger et al.@xcite whose 11005600 spectrum demonstrates that the @xmath66 transition around 1700 is indeed the strongest band .
their experiments in ne and ar matrices show a broad band but with some progressions superposed , which are evidence for interstate vibronic coupling with adjacent @xmath67 states .
indeed , the @xmath68 state is predicted to be unstable to bending , leading to avoided crossings with the lower - lying @xmath67 states @xcite .
these interactions can also lead to dissociative channels to produce c@xmath2 + c. both linear and cyclic c@xmath0h have been detected in the interstellar medium by their transitions at millimeter wavelengths @xcite .
the @xmath60 ground state of @xmath3c@xmath0h lies about 0.40.6 ev above the @xmath69 ground state of @xmath4c@xmath0h @xcite , with a small barrier toward the cyclic state .
hence , when c@xmath0h is produced in its linear form by some sequence of ion molecule or other chemical processes , it can be stable , and both isomers are therefore considered in this work .
the geometry of the @xmath70 state was taken from the experimental work of mccarthy and thaddeus @xcite .
table [ tabl - c3h ] summarizes our computed vertical excitation energies , oscillator strengths , and the corresponding configurations .
comparison with the casscf results of ding et al .
@xcite shows good agreement for the lowest states .
the dissociation energy of @xmath3c@xmath0h to c@xmath0 + h is computed at 3.3 ev . of the dipole - allowed states above this dissociation limit , the higher @xmath60 states around 7.8 ev have the largest oscillator strengths , in particular the 8@xmath60 state , which will dominate the photodissociation of the molecule .
the @xmath4c@xmath0h radical has @xmath19 geometry in its @xmath71 ground state , with the equilibrium structure for our calculations taken from yamamoto and saito @xcite .
table [ tabc - c3h ] summarizes our computed vertical excitation energies , oscillator strengths and configurations .
comparison with the casscf results of ding et al.@xcite shows again good agreement for the lowest states to 0.10.2 ev .
the dissociation energy of @xmath4c@xmath0h to c@xmath0 + h is around 4.3 ev whereas its ionization potential is computed to lie at 9.6 ev ( table [ tabdiss ] ) . of the dipole - allowed states above the dissociation limit , the @xmath72 state around 5 ev and the higher @xmath73 states around 7.5 ev have the largest oscillator strengths , but
the sum over the other states is comparable .
thus , there appear to be many potential routes to photodissociation for @xmath4c@xmath0h .
cyclopropenylidene , @xmath4c@xmath0h@xmath2 , was the first cyclic molecule to be detected in the interstellar medium @xcite and subsequently found to be ubiquitous throughout the galaxy , even in diffuse gas @xcite .
the near - linear form hccch ( propargylene ) , denoted here as hc@xmath0h , lies 0.40.8 ev higher in energy @xcite , depending whether the zero - point vibrational energy is included .
since it has a near - zero dipole moment , it has not yet been observed in interstellar clouds through radio transitions but is likely present as well .
there are various other stable isomers of c@xmath0h@xmath2 , of which another linear form , h@xmath2ccc ( vinylidenecarbene or propadienylidene ) , denoted here as @xmath3c@xmath0h@xmath2 , was discovered in interstellar space in 1991 @xcite .
it lies @xmath610.6 ev above the cyclic ground state including zero - point vibrational energy corrections @xcite .
hence , all three isomers are considered in this work .
although there are many theoretical studies of the ground states of the various c@xmath0h@xmath2 isomers dating back to 1976 @xcite , the excited states are largely unexplored .
the equilibrium structures of the electronic ground states are taken from seburg et al .
, their figure 5 @xcite .
our calculated relative ordering of the c@xmath0h@xmath2 isomers is consistent with previous findings but due to different amounts of recovered correlation energy , our energy splittings are larger than in other studies .
the adiabatic ionization potentials of the various c@xmath0h@xmath2 isomers lie at 8.96 ( hc@xmath0h ) , 9.2 ( @xmath4c@xmath0h@xmath2 ) and 10.4 ev ( @xmath3c@xmath0h@xmath2 ) , respectively ( table [ tabdiss ] ) . above these thresholds , the photoionization efficiency of all three c@xmath0h@xmath2 isomers increases rapidly so that photoionization will become the main pathway @xcite .
some of these ionizations are likely to be dissociative , but this option is not considered here .
the hc@xmath0h radical has @xmath74 geometry with a @xmath75 ground state . in this case
, the @xmath74 axis was put through the middle c atom and the middle of the h1h2 line .
table [ tabl - c3h2 ] summarizes our computed vertical excitation energies , oscillator strengths and configurations . for the lowest four excited states , our results are in good agreement with those of mebel et al .
@xcite ( their mrci+d(4,8)/ano(2 + ) results ) .
the dissociation energy of hc@xmath0h to @xmath3c@xmath0h + h is computed to be around 3.1 ev , with the dissociation energy to c@xmath0 + h@xmath2 perhaps even less @xcite . of the dipole - allowed states above the dissociation and below the ionization limit at 8.96 ev , the higher @xmath76 states at 6.2 and 7.5 ev , both of which have @xmath77 character , have the largest oscillator strengths .
the @xmath4c@xmath0h@xmath2 molecule has a @xmath78 ground state in its @xmath19 geometry , with a large dipole moment of 3.4 debye owing to the two unpaired electrons on one of the three carbon atoms .
table [ tabc - c3h2 ] summarizes our computed vertical excitation energies and oscillator strengths , which again agree well in terms of energies with those of mebel et al .
@xcite , although our oscillator strengths are somewhat lower . compared with @xmath4c@xmath0h
, @xmath4c@xmath0h@xmath2 has only few low - lying electronic states , and all of the dipole - allowed transitions lie above the dissociation energy of 4.4 ev @xcite .
those to the @xmath79 ( @xmath77 ) at 9.4 , @xmath80 at 10.6 and @xmath81 state at 11.2 ev have the largest oscillator strengths . however ,
none of these states lie below the ionization potential of @xmath4c@xmath0h@xmath2 at 9.2 ev .
thus , its photodissociation rate will differ substantially whether or not these states are included ( see 4 ) .
the vinylidenecarbene or propadienylidene isomer of c@xmath0h@xmath2 , denoted here as @xmath3c@xmath0h@xmath2 , also has @xmath78 symmetry in its @xmath19 ground state geometry .
our computed vertical excitation energies ( table [ tabv - c3h2 ] ) of 1.86 , 2.44 and 5.54 ev to the @xmath82 , @xmath83 and @xmath84 states are consistent with the experimental adiabatic values @xcite of 1.73 , 2.00 and 4.84 ev , respectively , and agree within 0.10.2 ev with the vertical values computed by mebel et al .
@xcite the dissociation energy of @xmath3c@xmath0h@xmath2 to c@xmath0h + h is computed to lie around 3.9 ev , so that only the @xmath84 and higher states can lead to photodissociation .
the @xmath85 state shows a well - resolved progression in its electronic spectrum @xcite , but the resolution of those data is not high enough to measure predissociation rates .
thus , we consider the photodissociation both with and without taking the @xmath85 state into account . of the dipole - allowed transitions below the ionization potential at 10.4 ev , those to the higher @xmath86 states around 9 ev have the strongest oscillator strengths .
these states all have @xmath87 character and likely belong to the rydberg series converging to the lowest ionization potential .
similarly , the higher @xmath88 states belong to the @xmath89 rydberg series .
the lowest energy @xmath90 ground state of c@xmath1 occurs for the linear geometry .
c@xmath1 has been searched for in diffuse interstellar clouds through its @xmath91 transition around 3789 but not yet detected @xcite .
however , a pattern of bands at 57 @xmath16 m observed with the _ infrared space observatory _ toward a handful of objects @xcite is consistent with transitions in the @xmath92 bending mode @xcite .
rhombic c@xmath1 , which is almost isoenergetic with @xmath3c@xmath1 , has not yet been detected in interstellar space .
our calculations use the @xmath3c@xmath1 equilibrium values computed by botschwina @xcite , which are consistent with experiments @xcite .
the results are presented in table [ tabc4 ] .
our excitation energies of the lower states are in good agreement with experiments and with the calculations of mass et al.@xcite , except for the 2@xmath93 state .
note that the experimental values refer to @xmath94 rather than @xmath95 , which can differ by a few tenths of ev .
our computed dissociation energy @xmath96 is 4.7 ev , in good agreement with the experimentally inferred @xmath97 value of 4.71 @xmath980.15 ev @xcite .
the lowest excited state above the dissociation limit to which dipole allowed transitions are possible is the @xmath99 state at 5.1 ev .
indeed , photofragment yield spectra in the 2.225.40 ev range by choi et al .
@xcite show significant single - photon dissociation into c@xmath0 + c at @xmath1005.2 ev , with a minor channel to c@xmath2 + c@xmath2 .
their spectra are well reproduced by phase space theory models in which the product state distributions are statistical .
this implies that absorption into the excited state is most likely followed by rapid internal conversion to the ground state potential energy surface with no barriers present along the dissociation coordinates . by far
the strongest absorption occurs into the @xmath101 state around 6.95 ev , which has an oscillator strength of 1.56 . even though higher @xmath102 and @xmath103 states will contribute
, the @xmath101 channel will dominate the interstellar photodissociation of c@xmath1 , if indeed every absorption is followed by dissociation .
the @xmath3c@xmath1h ( butadiynyl ) radical was detected in the envelopes of carbon - rich evolved stars nearly 30 years ago @xcite .
it was subsequently found to be very abundant in cold dark clouds like tmc-1 @xcite and even detected in comets @xcite .
it is one of the carbon - bearing molecules found at the edges of pdrs @xcite , where it can photodissociate into smaller species .
c@xmath1h has linear symmetry with a @xmath56 ground state .
the equilibrium coordinates in our calculations were taken from ref .
table [ tabc4h ] summarizes our computed vertical excitation energies , oscillator strengths and configurations .
comparison with the results of graf et al .
@xcite shows good agreement in both energies and transition dipole moments for the lowest states , but poor agreement for the higher @xmath60 states , where the graf et al .
energies are generally lower by up to 1 ev . there are two main resons for this .
first , graf et al . did not include diffuse ( rydberg ) functions in their basis set , which start to become important for the higher states .
second , their caspt2 perturbation method does not guarantee a lower bound to the energies .
our calculations agree in the fact that none of the higher lying @xmath60 states have large oscillator strengths .
the threshold for photodissociation , corresponding to dissociation of c@xmath1h to c@xmath1 + h , lies around 4.7 ev ( table [ tabdiss ] ) . of the dipole - allowed transitions below the ionization potential at 9.6 ev , the 4 and 5 @xmath56 states at 7.7 and 8.7 ev ,
respectively , have orders of magnitude larger oscillator strengths than other states and will thus dominate the photodissociation .
both states have @xmath104 rydberg character .
thus , our overall photodissociation rate will be significantly larger than that using the data from graf et al .
@xcite like c@xmath1h , @xmath3c@xmath5h is detected toward the carbon - rich evolved star irc+10216 @xcite and in cold dark clouds @xcite .
it also has linear symmetry but with a @xmath60 ground state .
the equilibrium coordinates in our calculations were taken from ref.@xcite .
table [ tabc5h ] summarizes our results .
comparison with haubrich et al .
@xcite shows excellent agreement in both energies and oscillator strengths , except for the higher - lying @xmath56 states .
this difference is likely due to the explicit inclusion of rydberg states in our work .
for the 3 and 4 @xmath60 states , the oscillator strengths differ by a factor of 2 but these states show considerable interaction between the @xmath105 and @xmath106 @xmath107 excitations .
differences in mixing ratios can lead to large changes in transition dipole moments to individual states , but not in energies .
the computed excitation energies for the lowest two excited states also agree well with those measured and computed by ding et al .
@xcite .
the threshold for photodissociation lies around 3.6 ev ( table [ tabdiss ] ) . of the dipole - allowed transitions below the ionization potential at 7.4 ev , both the 4 @xmath60 ( @xmath108 ) state at 4.2 ev and the 6 @xmath60 ( @xmath109 state around 6.1 ev have large oscillator strengths .
all other states have typical oscillator strengths of a few @xmath110 and thus contribute at a lower level .
in table [ tabisrates ] , the computed photodissociation rates in the unshielded interstellar radiation field cf .
draine @xcite are presented , using the oscillator strengths given in tables [ tabc2h ] to [ tabc5h ] and asuming @xmath52=1 for all states .
thus , these rates should be regarded as upper limits .
only states above the dissociation limit and below the ionization potential have been taken into account ( see table [ tabdiss ] for adopted values ) .
no corrections have been made for possible higher - lying states below the ionization limit not computed in this work . since the oscillator strengths for rydberg states decrease roughly as @xmath111 , it is assumed that any such corrections would be small since the lowest rydberg members are calculated explicitly . for reference ,
inclusion of a hypothetical state at 9 ev with an oscillator strength of 0.1 would increase the photodissociation rates by only @xmath112 s@xmath50 . for c@xmath2h ,
our new rate is a factor of 3 larger than that given in van dishoeck et al .
@xcite , which was based on the energies and oscillator strengths of shih et al . @xcite . the increase is mostly due to the higher @xmath56 and @xmath60 states which were not computed in that work . for c@xmath0 ,
the new rate is only 30% larger , mostly because the photodissociation rate of this molecule is dominated by the very strong absorption into the @xmath113 state around 8 ev , which was included in previous estimates .
it is seen that the rates for the various carbon - bearing molecules span a range of a factor of 8 , with the rates for the bare carbon chains ( c@xmath0 , c@xmath1 ) being largest and those for the odd - numbered c@xmath11h species lowest . of the different c@xmath0h@xmath2 isomers
, @xmath3c@xmath0h@xmath2 has the largest photodissociation rate and @xmath4c@xmath0h@xmath2 the smallest by a factor of 3 .
however , @xmath4c@xmath0h@xmath2 differs from the other isomers in that it has several states with large oscillator strengths above the ionization potential of 9.15 ev . if those states were included , the @xmath4c@xmath0h@xmath2 photodissociation rate would be increased by a factor of 2 . for @xmath3c@xmath0h@xmath2 ,
the rate would drop from @xmath114 to @xmath115 s@xmath50 if the @xmath85 state is not included . in spite of the range of values , all rates are above @xmath116 s@xmath50 , coresponding to a lifetime of less than 30 yr at the edge of an interstellar cloud . in regions such as the orion bar and the horsehead nebula ,
the radiation field is enhanced by factors of @xmath117 compared with the standard field adopted here , decreasing the lifetimes to less than 1 month .
thus , there must be rapid production routes of these molecules in order to explain their high abundances in uv - exposed regions .
this conclusion is not changed if only 10% , say , of the absorptions would lead to dissociation rather than the 100% assumed here .
as argued in 2 and 3 , it is plausible that a substantial fraction of the absorptions lead to dissociation for these larger molecules so that the upper limits should be close to the actual values .
we have presented vertical excitation energies , transition dipole moments and oscillator strengths for states up to the 9th root of each symmetry for several carbonaceous molecules of astrophysical interest . for lower - lying states ,
good agreement is generally found with previous studies .
several new , higher - lying states with large oscillator strengths , often of rydberg character , are revealed in this work .
the calculated photodissociation rates of the small carbon - bearing molecules studied here are substantial , leading to lifetimes at the edges of interstellar clouds of less than 30 yr .
these high rates assume that all absorptions above the dissociation limit indeed lead to dissociation , so that the rates should be viewed as upper limits .
further experimental work is needed to quantify the dissociation efficiencies for the strongest states found in this work .
our calculations of the photodissociation processes of astrophysical molecules , which started 30 years ago , were made possible through the wuppertal - bonn mrd - ci set of quantum - chemical programs kindly made available by prof .
peyerimhoff and prof .
r.j . buenker .
the authors enjoyed many stimulating conversations with prof .
peyerimhoff on excited electronic states of small molecules over the past decades .
astrochemistry in leiden is supported by a spinoza grant from the netherlands organization for scientific research ( nwo ) .
m. gulin , r. lucas , r. neri , m. bremer , d. broguire , in : y.c .
minh , e.f .
van dishoeck ( eds . ) astrochemistry : from molecular clouds to planetary systems , iau symposium 197 , astronomical society of the pacific , san francisco , 2000 , p.365 gamess - uk is a package of ab initio programs .
see : http://www.cfs.dl.ac.uk/gamess-uk/index.shtm , m.f .
guest , i. j. bush , h.j.j .
van dam , p. sherwood , j.m.h .
thomas , j.h .
van lenthe , r.w.a havenith , j. kendrick , the gamess - uk electronic structure package : algorithms , developments and applications , molecular physics , vol .
6 - 8 , 20 march-20 april 2005 , 719 - 747 . | astronomical observations have shown that small carbonaceous molecules can persist in interstellar clouds exposed to intense ultraviolet radiation .
current astrochemical models lack quantitative information on photodissociation rates in order to interpret these data .
we here present _
ab initio _ multi - reference configuration - interaction calculations of the vertical excitation energies , transition dipole moments and oscillator strengths for a number of astrophysically relevant molecules : c@xmath0 , c@xmath1 , c@xmath2h , @xmath3 and @xmath4c@xmath0h , @xmath3 and @xmath4c@xmath0h@xmath2 , hc@xmath0h , @xmath3c@xmath1h and @xmath3c@xmath5h .
highly excited states up to the 9th root of each symmetry are computed , and several new states with large oscillator strengths are found below the ionization potentials .
these data are used to calculate upper limits on photodissociation rates in the unattenuated interstellar radiation field by assuming that all absorptions above the dissociation limit lead to dissociation .
molecular data , excited electronic states , oscillator strengths , photodissociation , interstellar molecules 31.15.ar , 31.50.df , 33.15.fm , 33.20 , 33.70.ca , 33.80.gj , 95.30.ky , 98.38.bn , 98.38.-j |
galaxy mergers are a natural and crucial ingredient of the @xmath0cdm hierarchical galaxy formation paradigm .
although the fraction of galaxies undergoing a merger at any given time is relatively small , nearly all galaxies will experience a merger at some point in their histories ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
particularly significant are ` major ' mergers , which can be transformative . in these cases
, mergers violently alter the orbits of the stars in the galaxies and can transform rotationally supported discs into dispersion - supported spheroids ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
furthermore , tidal torques exerted by the galaxies upon one another drive gas inwards ( e.g. , * ? ? ? * ; * ? ? ?
* ) , thereby resulting in powerful starbursts ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , triggering active galactic nuclei ( agn ) ( e.g. , * ? ? ?
* ) , altering metallicity gradients @xcite , and leaving behind signatures of the starbursts and agn activity in the form of compact stellar cores ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and supermassive black holes ( bhs ; e.g. , * ? ? ?
also , mergers may drive the size evolution of quiescent galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
it has thus been proposed that various seemingly different observational classes of objects in the universe including blue star - forming disc galaxies , irregular galaxies of a variety of morphologies , heavily dust - obscured ( ultra-)luminous infrared galaxies ( ( u)lirgs ) , both obscured and unobscured agn , ` post - starburst ' ( aka ` k+a ' ) galaxies , and ` red and dead ' elliptical galaxies may be related in a merger - driven evolutionary sequence ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
galaxy mergers have been studied using numerical simulations for more than forty years ( @xcite ) , and for more than seventy years if one considers the pioneering laboratory method of @xcite .
although the early simulations included only gravity , they provided much insight into the effects of mergers on galaxy morphologies , the formation of tidal tails and shells , and the kinematics of merger remnants .
more sophisticated simulations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) included also gas dynamical processes , which can be important for many galaxies because a significant fraction of the baryonic mass is in gas , and , unlike the stars , the gas is dissipational .
hydrodynamic simulations of galaxy mergers have helped us to understand the driving mechanism of starbursts ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , supermassive bh fueling and feedback ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , feedback from supernovae ( e.g. , * ? ? ?
* ) , the kinematics of merger remnants ( e.g. , @xcite ; @xcite ) , the survivability of discs during mergers ( e.g. , * ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the sizes of merger remnants ( e.g. , @xcite ; @xcite ; @xcite ) , and the formation of local ( e.g. , * ? ? ?
* ) and high - redshift ulirgs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , among many other topics .
whereas cosmological simulations are routinely performed using both grid - based eulerian and particle - based pseudo - lagrangian methods , idealised isolated ( i.e. , non - cosmological ) galaxy merger simulations have almost always been performed using pseudo - lagrangian smoothed particle hydrodynamics ( sph ; @xcite ; see @xcite , @xcite , and @xcite for recent reviews ) .
sph is well - suited to simulating galaxy mergers because it naturally treats the large bulk velocities present before the discs coalesce and it concentrates resolution elements in regions where the mass is located ( the galaxy centre(s ) ; see section 9.4 of @xcite for further discussion ) .
furthermore , the other primary method used for galaxy formation studies , adaptive mesh refinement ( amr ) , does not treat self - gravity as accurately as particle - based approaches @xcite . for these reasons , relatively few galaxy merger simulations have been performed using amr ( to our knowledge , the only published examples of such simulations are @xcite ) .
recent studies that used simple , idealised test problems and cosmological simulations have highlighted potentially significant problems that are inherent in the standard formulation of the sph technique .
@xcite demonstrated that the standard implementation of sph artificially suppresses the kelvin
helmholtz ( kh ) and raleigh taylor ( rt ) instabilities because the method introduces spurious pressure forces near steep density gradients that allow a gap to be created between particles , thereby reducing their interactions .
furthermore , sph can artificially damp subsonic turbulence @xcite and restricts gas stripping from substructures falling on to haloes @xcite .
however , fixed grid - based methods also have drawbacks ; for example , these codes are not galilean invariant and can produce over - mixing @xcite .
partially motivated by the limitations inherent in the traditional sph technique , @xcite developed a novel moving - mesh hydrodynamics code known as arepo .
although similar approaches have been proposed earlier , such techniques have not yet seen wide - spread use in astrophysical applications .
arepo is quasi - lagrangian because the unstructured mesh is advected with the flow .
whereas the motion of the grid reduces mass exchange between cells compared with a static mesh ( e.g. , * ? ? ?
* ) , arepo is not strictly lagrangian because the mass in a given cell can change with time .
nevertheless , arepo offers a number of advantages over other methods that make it attractive for astrophysical applications .
for example , it is galilean invariant and naturally concentrates resolution elements in dense regions , similar to particle - based techniques . because arepo uses a finite - volume method to solve the euler equations ,
it is better than traditional sph at capturing shocks and contact discontinuities and does not artificially suppress fluid instabilities ( see also * ? ? ?
* ; * ? ? ?
its ability to better capture weak shocks than sph is potentially significant for cosmological problems because these features are ubiquitous in the cosmic web ( e.g. , * ? ? ?
because of its hybrid nature , arepo performs better than ( or at least as well as ) both traditional sph and grid - based approaches for idealised test problems . thus , it is useful to compare the results of arepo simulations to those of simulations performed using other techniques to investigate what effects , if any , the shortcomings of the traditional methods have on the results of cosmological and idealised galaxy merger simulations . some comparisons between cosmological simulations performed with arepoand gadget-3 in which all other ingredients , including the gravity solver and sub - resolution models , are identical have already been made . in some situations , the baryonic properties of the simulations performed with the two codes differ strikingly . for example
, @xcite demonstrated that gas cooling is more efficient in the arepo simulations , which results in more star formation at late times .
they attribute the difference to spurious heating in the outer regions of virialized haloes in the gadget-3 simulations and the inability of conventional sph to correctly develop a turbulent cascade to smaller scales .
the arepo simulations produce galaxies with extended , relatively smooth gas discs , whereas in the gadget-3 simulations , the discs are more compact and clumpy @xcite .
@xcite demonstrated that the relative fraction of gas supplied to galaxies in halos of moderate to large size through ` cold - mode accretion ' ( aka ` cold flows ' ) is dramatically less in the arepo simulations because with traditional sph , much of the cold gas that reaches the central galaxies does so because it is locked in ` blobs ' of purely numerical origin and because the rate at which gas cools from galaxies hot haloes is higher in arepobecause it allows for a proper cascade of turbulent energy to small scales .
because of the nature of the aforementioned comparisons , the differences can only originate from differences in the hydrodynamical solver .
it is sometimes argued ( e.g. , * ? ? ?
* ) that the differences caused by inaccuracies in the numerical technique employed are subdominant to the effects caused by varying the prescriptions for star formation and stellar and agn feedback and hence may be ignored .
however , the results of @xcite caution against this because they find differences in the accretion rate of hot gas on to galaxies between arepo and gadget-3 that in some cases approach two orders of magnitude , which is far greater than the discrepancies between simulations and observations that feedback effects are invoked to resolve ( see @xcite and @xcite for comparisons of arepocosmological hydrodynamical simulations with observations ) . with the exception of a single modest - resolution merger simulation presented in @xcite , the moving - mesh approach has not yet been used to simulate galaxy mergers . here , we present the first detailed study of idealised galaxy merger simulations performed using moving - mesh hydrodynamics .
we present a small suite of simulations of equal - mass mergers simulated with both gadget-3 and arepo . to isolate differences caused by the different hydrodynamical solvers
, we have kept all other components of the simulations
namely , the gravity solver and the sub - resolution models for star formation , the interstellar medium ( ism ) , bh accretion , and agn feedback as similar as possible .
other , more - comprehensive comparisons , such as the _ aquila _ @xcite and agora @xcite projects , allow multiple ingredients of the simulations to vary simultaneously , thereby yielding a general characterization of the systematic uncertainties due to different numerical methods and sub - resolution models .
our goal is much more specific : we aim to isolate effects of the hydrodynamical solver from those caused by differences in the gravity solver or sub - resolution models , which is not readily possible in these other comparisons . in our work
, we have chosen to use a ` standard ' implementation of sph , as incorporated in the gadget-3 code .
we have not explored recent variants of sph that are designed to improve its reliability in some circumstances ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) for several reasons .
the most important is that we wish to assess the reliability of previous simulations of gas dynamics in galaxy mergers that were performed using traditional formulations of sph .
we also note that many of the sph modifications that have recently been proposed have not yet been tested under a wide range of conditions .
it is presently thus still unclear which of the numerous modification should eventually be adopted in a new ` best sph ' variant .
also , a number of problems with sph are still unresolved even in the most recent proposed revisions of the method .
we discuss these issues further below .
the remainder of this paper is organised as follows . in section [
s : methods ] , we describe the two different hydrodynamical methods used and the sub - resolution models for star formation , bh accretion , and supernova and agn feedback . in section [ s : results_no_bhs ] ( [ s : results_bhs ] ) , we compare gadget-3 and arepo simulations that do not ( do ) include bh accretion and agn feedback .
section [ s : tests ] presents tests of different methods for treating bh accretion and agn feedback that we used to inform our choice of a fiducial treatment . in section [ s : discussion ] , we review some of the major issues with sph , discuss why sph works reasonably well for some applications but not others , and outline which previous work is likely to be robust to the hydrodynamical solver used and which may need to be reconsidered .
section [ s : conclusions ] presents our conclusions .
as noted above , we use two different codes , gadget-3 and arepo , because we wish to investigate differences in the outcome that are driven by variations in the method used to solve the hydrodynamics .
gadget-3 uses sph @xcite , a pseudo - lagrangian method . in sph ,
the gas is discretised into particles , which are typically of fixed mass .
the density field and other continuous quantities are calculated by taking the mean of the values of some number of nearest particles ( we use 32 ) weighted by the smoothing kernel . to derive the equations of motion
, the lagrangian can be discretised and then the variational principle used @xcite .
the formulation of sph used in gadget-3 is explicitly conservative even if the smoothing lengths vary @xcite .
the advantages of modern sph include explicit conservation of mass , energy , entropy , and linear and angular momentum ; galilean invariance ; resolution that naturally becomes finer in regions of high gas density ; and accurate treatment of self - gravity .
arepo adopts a novel version of the other primary technique used in astrophysical hydrodynamics , the finite - volume ( i.e. , eulerian ) grid - based approach . in traditional amr codes , cubic cells at fixed spatial locations are employed , and the cells are refined and de - refined according to some criteria ( e.g. , the mass of cells can be kept approximately constant ) . in arepo ,
the grid cells are not fixed in space ; rather , mesh - generating points are advected with the flow and a voronoi tesselation is used to generate an unstructured grid from the points .
the euler equations are solved using a finite - volume approach .
specifically , arepo uses a second - order unsplit godunov scheme with an exact riemann solver .
advantages of this moving - mesh approach compared with sph include the following : it is better at resolving shocks and contact discontinuities ; it does not suppress fluid instabilities ; and the density field across a resolution element ( a cell ) can be reconstructed to first order ( unlike in sph , for which the density can only be constructed to zeroth order ) .
both codes use the same tree - based gravity solver , which is a modified version of that used in gadget-2@xcite .
collisionless particles are used to represent dark matter and stars ; these particles are assigned fixed gravitational softening lengths . in gadget-3 ,
the gas particles also have fixed gravitational softening .
this treatment guarantees that , by using suitably small softening , one can have sufficient force resolution at all times .
in contrast , the gravity solvers traditionally used in amr codes have force resolution that depends on the cell size and thus varies in time and space . even in collisionless simulations ,
this treatment can lead to the suppression of small - scale structure ( i.e. , dwarf galaxies ) if cells around forming haloes are not refined sufficiently early @xcite . because arepo treats the collisionless component using a tree - based method , it does not suffer from this problem .
the quasi - lagrangian nature of the moving mesh cells also enables a superior treatment of gas self - gravity .
arepo treats each cell as if the mass were concentrated at the cell centre and calculates the softened gravitational force using a softening that is of order the cell radius . for all components ,
gravitational interactions are softened using a cubic spline that has compact support ( e.g. , * ? ? ?
full details of the treatment of self - gravity used in arepo can be found in section 5 of @xcite . in both gadget-3 and arepo , star formation and supernova feedback
are implemented via the effective equation of state ( eos ) method of @xcite . only gas particles with density greater than a low - density cutoff ( @xmath1 @xmath2 )
are assumed to have an eos governed by the sub - resolution model . in the @xcite model ,
the ism is considered to consist of two phases in pressure equilibrium : cold dense clouds and a hot diffuse medium in which the cold clouds are embedded .
the instantaneous sfr for each particle is calculated using a volume density - dependent kennicutt schmidt @xcite prescription , @xmath3 , with @xmath4 .
star particles are spawned from gas particles or cells probabilistically according to their sfrs .
feedback from supernovae is included as an effective pressurization of the ism such that the equation of state of the gas is stiffer than an isothermal eos .
see @xcite and @xcite for further details .
we stress that the sub - resolution models for star formation and feedback in gadget-3 and arepo are as similar as possible , which enables us to investigate differences caused solely by variations in the calculation of the hydrodynamics . as for the star formation and stellar feedback sub - resolution models , we have attempted to keep the sub - resolution models for bh accretion and agn feedback in the two codes as similar as possible .
however , for reasons we discuss in section [ s : tests ] , the default sub - resolution models for bh accretion and feedback used in gadget-3 and arepo differ slightly . because we shall explore different numerical implementations of the bh accretion and feedback model ,
we describe the model in some detail here , but see @xcite for a more thorough presentation .
each disc galaxy is initialised with a @xmath5 central sink particle that undergoes modified eddington - limited bondi
lyttleton accretion @xcite .
the accretion rate is the bondi
lyttleton accretion rate multiplied by a dimensionless factor @xmath6 : @xmath7 where @xmath8 is the gravitational constant , @xmath9 is the bh mass , @xmath10 is the local gas density , @xmath11 is the local sound speed , and @xmath12 is the velocity of the bh relative to the gas . because we typically do not resolve the bondi radius , the parameter @xmath6 , which we set equal to 100 , is used to account for the fact that we underestimate the density at the bondi radius . in practice , the choice of @xmath6 matters only at early times because most of the mass growth occurs during times of eddington - limited accretion , and the mass of the final bh is insensitive to the value of @xmath6 .
the eddington - limited accretion rate is @xmath13 where @xmath14 is the proton mass , @xmath15 is the thomson cross - section , and @xmath16 is the radiative efficiency , which is defined by @xmath17 where @xmath18 is the luminosity of the bh .
we assume @xmath19 @xcite . the bh particles accrete mass at a rate @xmath20 in which the @xmath21 factor accounts for the rest - mass of the energy radiated away by the agn ( this factor was not included in the original @xcite treatment ) . in gadget-3 ,
gas particles are swallowed stochastically . in arepo ,
the more natural treatment is to continuously drain gas from the cell in which the bh is located .
we have compared this treatment with one analogous to that in gadget-3 , in which cells are swallowed stochastically ; we found that for the mass and time resolutions of our simulations , the differences in the results are negligible ( but lower - resolution simulations can exhibit differences ) .
to calculate the accretion rate given by equation ( [ eq : bondi ] ) , we must determine @xmath10 and @xmath22 near the bh . in gadget-3 , the sph estimates for these quantities are used . in arepo , we can adopt analogous estimates , but we can instead also employ the quantities for the cell in which the bh is located . in principle , the latter should better represent the properties of the gas around the bh
. however , the individual cell values can also be more noisy , so it is not clear a priori which is preferred .
we compare the results obtained using the different treatments in section [ s : accretion_test ] .
based on those tests , we chose to use the cell density and sph - like estimate of the sound speed in our default treatment .
one potential issue when calculating the accretion rate using either the sph density estimate or the cell density is that as the bh consumes or expels the nearby gas , the region used for the density estimate can grow in size .
consequently , the bh can continue to accrete gas from larger and larger scales , whereas in reality , bh accretion should terminate once the gas near the bh is consumed or expelled . in sph , this problem is unavoidable unless one reduces the number of neighbours used to estimate the density ( which is problematic because the noise in the density estimate will be increased significantly ) or a scheme that is more complicated than the simple bondi
lyttleton approach is employed ; one consequence is that lower - resolution simulations can sometimes exhibit more bh growth ( e.g. , * ? ? ?
* ) . in arepo
, this problem may be avoided by preventing cells near the bh from becoming too large .
we do this by forcing the cells within some radius of the bh to be refined if they have a radius greater than some maximum value .
we explore the effects of this refinement in section [ s : refinement_test ] . based on our tests
, we decided to force cells within 500 pc of a bh to have a maximum size of 50 pc .
bh accreting from a gas with sound speed @xmath23 km @xmath24 .
thus , an even stricter refinement criterion could be justified , but because of the resolution of the simulations , there is no structure on smaller scales . ] finally , because the initial bh particles are similar in mass to the stellar and dark matter particles , two - body interactions can cause the bh to stray from the centre of the potential well .
however , in reality , dynamical friction would cause the bhs to rapidly sink to the potential minimum .
thus , we pin the bh to the halo potential minimum . for this reason
, we also neglect the @xmath12 term in the denominator of equation ( [ eq : bondi ] ) .
we also include a simple model for thermal feedback from the agn @xcite .
the bh particles deposit some fraction @xmath25 of their luminosity ( as thermal energy ) to the surrounding gas .
we use @xmath26 because in previous gadget-2 simulations , this value yielded an @xmath27 relation normalization consistent with that observed @xcite .
we scale the number of gas particles over which we distribute the feedback energy with resolution such that the total mass of the particles is constant .
as we demonstrate in section [ s : fb_test ] , this scaling minimizes the resolution dependence of the sub - resolution model .
it is desirable to have sub - resolution models that do not depend significantly on numerical resolution , at least as long as the physics that the sub - resolution treatment is meant to represent remains inaccessible .
otherwise , the problem is not well - posed numerically and the interpretation of the sub - resolution model becomes unclear .
the initial disc galaxies are created following the procedure described in @xcite .
the galaxies consist of dark matter haloes described by a @xcite profile with virial velocity @xmath28 and concentration @xmath29 , an exponential stellar disc with scalelength @xmath30 and scaleheight @xmath31 , an exponential gaseous disc with scalelength @xmath32 and scaleheight determined by requiring the disc to be rotationally supported , and a bulge described by a @xcite profile with scalelength @xmath33 .
note that , unlike in @xcite , the gaseous and stellar discs do not have the same scalelength ; rather , the gaseous discs can be significantly more extended than the stellar discs . unlike sph , grid - based methods can not treat empty space .
thus , for the arepo simulations , we must add a background grid of low - density cells to the initial conditions used for the gadget-3 simulations such that the entire simulation volume has positive gas density .
see section 9.4 of @xcite for details of how the background mesh is added .
our intention is not to present a comprehensive suite of merger simulations that addresses the full parameter space but rather to directly compare the results obtained using arepo and gadget-3 for a set of simulations based on galaxy models that are representative of a variety of actual galaxies . for definiteness
, we use galaxy models that are similar to those of @xcite and the same merger parameters as @xcite .
two of the isolated disc galaxies are intended to be milky way ( ) and small magellanic cloud ( ) analogues .
the other two represent a dwarf starburst ( ) and a redshift @xmath34 disc galaxy ( ) .
thus , the initial discs capture much of the diversity of real disc galaxies , but the sampling is by no means complete .
the properties of the disc galaxies are given in table [ tab : galaxy_models ] .
we simulate each disc galaxy with two different resolutions , which are specified in table [ tab : resolutions ] .
we evolve each disc in isolation for 3 gyr . for the mergers ,
two identical disc galaxies are placed on parabolic orbits with initial separation and pericentric passage distance as specified in table [ tab : orb_params ] . as in @xcite ,
two orbits , the and orbits of @xcite , are used . for the orbit ,
the directions of the spin axes of the discs given in spherical coordinates are @xmath35 . for , @xmath36 .
note that neither orbit is coplanar ; thus , our general conclusions should be relatively insensitive to orbit ( mergers with perfectly coplanar orbits , which are of course highly unlikely in nature , can exhibit pathological behavior that is not characteristic of the behaviour for other orbits ) .
because our focus is to compare the results of otherwise - identical gadget-3and areposimulations , two orbits is sufficient ; a comprehensive suite of simulations would ideally include a much wider variety of orbits ( see , e.g. , * ? ? ?
* ; * ? ? ?
we run each merger simulation for @xmath37 gyr , depending on the simulation , which is sufficient time for the galaxies to coalesce .
we typically simulate each merger at two different resolutions ( see table [ tab : resolutions ] for details ) .
furthermore , to strengthen our conclusions regarding whether any resolution dependence is systematic , we performed the simulation with bh accretion and agn feedback included at a third , higher resolution . except where otherwise noted ,
we always plot the results of the highest - resolution run .
we first present a comparison of gadget-3 and arepo simulations for which we disabled the bh accretion and agn feedback treatments in the codes .
gyr , the absolute values of the sfrs for , , , and are @xmath38 , @xmath39 , @xmath40 , and @xmath41 , respectively .
the sfhs are almost identical regardless of the resolution or code used.,title="fig : " ] + gyr , the absolute values of the sfrs for , , , and are @xmath38 , @xmath39 , @xmath40 , and @xmath41 , respectively .
the sfhs are almost identical regardless of the resolution or code used.,title="fig : " ] + a comparison of the star formation histories ( sfhs ) and cumulative stellar mass formed versus time for the isolated disc simulations is shown in fig .
[ fig : iso_nb_sfhs ] .
the solid ( dashed ) lines indicate the higher ( lower ) resolution runs , and blueish ( reddish ) colours indicate arepo ( gadget-3 ) simulations . the isolated discs ( fig . [ fig : iso_nb_sfhs ] )
evolve in the expected manner : after some initial settling into equilibrium , as the gas is consumed ( in these idealised simulations , no additional gas is supplied during the simulations ) , the sfr decreases and the stellar mass formed increases . for the simulations ,
there are some minor resolution- and code - dependent differences in the sfhs at @xmath42 gyr .
the different runs are almost identical .
the simulations exhibit minor resolution - dependent differences in the sfhs throughout the simulation .
for the model , there are minor differences in the sfrs of the gadget-3 and arepo simulations at @xmath43 gyr . in all cases ,
the curves of cumulative stellar mass formed versus time are almost indistinguishable .
these results demonstrate that for the isolated disc simulations , the two codes agree very well and the simulations are converged with respect to particle number ( at least in terms of their sfhs ) .
[ fig : merger_nb_sfhs ] shows the sfhs for the merger simulations without bh accretion and agn feedback . as expected from much previous work , the mergers exhibit the following generic evolution .
the sfr initially oscillates for a short time as the discs settle into equilibrium .
then , there is a slight elevation when the discs are at first pericentric passage , but the bulges prevent the discs from becoming very unstable .
as the discs approach final coalescence , in many cases , strong tidal torques drive gas into the nucleus and fuel a starburst .
however , the strength and shape of the starburst depend on the progenitor properties and orbit . after the strong starbursts ( , , , and )
, the sfr decreases to the pre - merger level or significantly below it .
note that the decrease is driven solely by gas consumption and shock - heating of the gas because these simulations do not include agn feedback . in the other merger simulations , in which a strong starburst is not induced
, the sfr can remain elevated for the duration of the simulation ( see especially and ) . as can be inferred from fig .
[ fig : merger_nb_sfhs ] , the agreement for the merger simulations is also excellent , although there are more noticeable differences than for the isolated disc cases .
the most prominent difference is that for , , , and , the time at which the starburst occurs depends on resolution .
this is likely caused by resolution - dependent variations in the hydrodynamical drag experienced by the galaxies as they collide , which can slightly alter the merging time - scales .
however , for a given resolution , the sfhs during the starburst are similar for the two codes .
the most significant , albeit still relatively minor , differences between the arepo and gadget-3 simulations occur after the starbursts , e.g. , at @xmath44 , 2.9 , and 2.6 gyr for , , and , respectively .
the sfhs do not vary systematically depending on the code used , e.g. , for the post - starburst phase of , the arepo sfrs tend to be lower , whereas for , the opposite is true . as for the isolated discs , the differences in the cumulative stellar mass formed versus time are negligible .
+ the agreement of the sfhs of the gadget-3 and arepo simulations for both the isolated disc and merger simulations indicates that the differences in the numerical schemes do not dramatically affect the global evolution of the simulated systems .
however , it is possible that the detailed properties of the simulated mergers differ despite the good agreement for the integrated quantities , and this is indeed the case .
[ fig : nb_gas_comparison ] presents gas surface density plots for four different simulation snapshots .
( the interested reader is invited to visit the aforementioned url to view animations that compare the time - evolution of the gas surface density for the highest - resolution runs of all simulations presented in this work . )
the top row shows the arepo result , and the bottom row shows the gadget-3 result for the same snapshot and resolution .
the first column shows the isolated disc at @xmath45 gyr .
the arepo and gadget-3 results are qualitatively similar , but the gas distribution appears slightly smoother in the arepo simulation and the orientation of the bar differs .
the second column shows the merger at @xmath46 gyr , which is near final coalescence .
again , the morphologies are very similar , but the spatial extent of the tidal tails differs slightly .
the third column shows the merger near the peak of the starburst ( @xmath47 gyr ) . here , the gas distribution is again smoother in the arepo run , and the gas morphology of the nuclear region differs quite dramatically .
finally , the fourth column shows the merger at @xmath48 gyr ( @xmath49 gyr after the peak of the starburst ) . here , the differences are the most dramatic : the nuclear disc that has re - formed is oriented edge - on in the arepo simulation but face - on in the gadget-3 simulation
this is likely driven by the stochastic nature of the torques acting on the gas that accumulates in the centre of the remnant ( e.g. , * ? ? ?
furthermore , the gadget-3 simulation features a clumpy , extended hot halo that is not present in the arepo simulation . in general ,
the arepo morphologies tend to be smoother than those yielded by gadget-3 , and the clumps that are often observed in the gadget-3 simulations ( and are spurious ; e.g. , @xcite ) are not present in the arepo simulations .
the differences between the two codes are most pronounced during the starburst and post - starburst phases .
we discuss the physical reasons for these differences in detail in section [ s : reasons_for_agreement ] .
we shall now compare the gas phase structure in the gadget-3 and arepo simulations .
animations showing the evolution of the gas phase structure for all simulations are available at the aforementioned url . for brevity
, we will only discuss the general trends and present an illustrative example . throughout the simulations ,
the evolution of the gas phase structure is very similar in the arepo and gadget-3 simulations when bh accretion and agn feedback are not included . given the results presented above , this result is not surprising : if the phase structure differed significantly , then the sfhs would not agree so well . as for the gas morphology
, the differences are more pronounced in the starburst and post - starburst periods of the simulations .
specifically , the arepo simulations tend to exhibit less low - density , hot halo gas . in both the arepo and gadget-3 simulations , a hot halo forms when gas is shock - heated during final coalescence of the discs . in the arepo simulations
, however , the hot halo gas cools more effectively for the reasons discussed in section [ s : reasons_for_agreement ] .
[ fig : nb_pd_comparison ] demonstrates this difference .
this figure shows example phase diagrams for the post - starburst phase ( @xmath48 gyr , @xmath50 myr after the starburst ) of the merger simulation performed using arepo ( left ) and gadget-3 ( right ) .
note that these correspond to the same simulation and output time as the gas surface density plots shown in the fourth column of fig .
[ fig : nb_gas_comparison ] . in the arepo simulation , there is less hot halo gas .
the enhanced cooling of hot halo gas in the arepo simulation explains why the sfr is somewhat higher in the arepo simulation at this time ( see fig .
[ fig : merger_nb_sfhs ] ) , which is also the case in some of the other merger simulations ( e.g. , ) .
we now compare simulations that include bh accretion and agn feedback .
here , we use the default accretion and feedback schemes discussed above , but we explore the implications of different choices in section [ s : tests ] . + + + + + + + + + + + fig .
[ fig : iso_sfhs ] shows the sfhs and cumulative stellar mass formed versus time for the isolated disc simulations .
blue ( red ) indicates arepo ( gadget-3 ) simulations with bh accretion and agn feedback ; the corresponding simulations without bh accretion and agn feedback are indicated in cyan ( magenta ) for comparison .
solid ( dashed ) lines indicate higher ( lower ) resolution simulations . in general , for the isolated disc simulations , the agreement among the sfhs for the two codes and different resolutions is still relatively good , but the differences are clearly more significant than when bh accretion and agn feedback are disabled . for the case
( first row of fig .
[ fig : iso_sfhs ] ) , the arepo simulations tend to have slightly higher sfrs , but the differences between the simulations sfrs are less than @xmath51 dex at all times .
the cumulative stellar mass formed is indistinguishable .
the sfhs for the simulations ( second row ) agree similarly well , and the cumulative stellar mass formed is nearly the same .
for the simulations ( third row ) , the sfrs differ by as much as @xmath52 dex , but the resolution - dependent variations are as significant as those between the codes and the differences for simulations that vary only in whether they include bh accretion and agn feedback . in this case , the total stellar mass formed over the course of the simulations ( 3 gyr ) is @xmath51 dex less in the gadget-3 simulations with agn feedback , but all other simulations agree very well . finally , the sfrs of the isolated disc simulations ( fourth row ) can vary by as much as @xmath53 dex , and the primary cause of the difference is whether bh accretion and agn feedback are included ( in such simulations , the sfrs are systematically lower , as expected ) .
however , the stellar mass formed is the same to within @xmath54 dex . figs .
[ fig : e_sfhs ] and [ fig : f_sfhs ] show the sfhs and cumulative stellar mass formed versus time for the -orbit and -orbit merger simulations , respectively . for brevity , we will not discuss each panel individually but rather highlight general trends . for a given progenitor combination and orbit , the shapes of the sfhs are qualitatively similar for both codes and all resolutions ( except in the cases for which agn feedback significantly impacts the post - starburst sfr ) .
however , there are significant quantitative code- and resolution - dependent differences . for many of the simulations ( i.e. , , , , , and ) , the sfhs are almost identical up to final coalescence .
if a strong starburst is triggered , the precise time and amplitude of the starburst can vary depending on the code and resolution . in some but not all cases ( and , in particular ) , inclusion of agn feedback causes the post - starburst sfr to decrease more rapidly compared with the corresponding simulations without agn feedback . although the sfhs differ significantly at some times , the cumulative stellar mass formed versus time is often very similar for the different codes and resolutions , as we saw above for the simulations without agn feedback and the isolated disc simulations with agn feedback . in most cases , the cumulative stellar mass formed differs by less than @xmath51 dex at all times .
the merger simulations exhibit the most significant differences in the cumulative stellar mass formed : when agn feedback is included , the cumulative stellar mass formed in the arepo simulations is @xmath55 dex greater than in the gadget-3 simulations .
+ + + + another quantity of interest is the bh mass versus time because differences in the bh masses would alter the strength of the agn feedback and thereby influence whether or not the merger remnants lie on the @xmath27 relation @xcite .
it is natural to expect that the bh mass evolution is more code- and resolution - dependent than the sfh because the bh growth depends sensitively on the gas conditions in the nuclear region(s ) and could thus be affected by small - scale variations that would not significantly alter the integrated sfh .
[ fig : iso_bh_mass ] shows the bh mass versus time for the isolated disc simulations . in the and simulations ,
the bhs grow by only a modest amount ( @xmath56 and @xmath57 dex , respectively ) over the 3.0 gyr of the simulations . in the and simulations ,
the bhs rapidly increase in mass by 2 - 3 orders of magnitude .
this strong bh growth in the absence of a merger is driven by bar instabilities , which is clear from examination of the gas morphologies .
the bh growth terminates once the gas near the bh is consumed or expelled . for a given code ,
the final bh masses in the isolated disc simulations differ with resolution by @xmath58 dex . however , the code - dependent variations can be more significant . in particular , note that the final bh masses in the arepo simulations are a factor of @xmath59 greater than those in the gadget-3 simulations . the bh mass evolution for the merger simulations is shown in fig . [
fig : merger_bh_mass ] . in these plots ,
the total mass of all bhs recall that each progenitor disc is seeded with a bh is plotted . for a given progenitor ,
the bhs grow significantly more in the merger simulations than in the isolated disc simulations .
the merger exhibits the weakest growth ( @xmath60 dex ) , and the merger simulations exhibit the strongest bh growth ( almost four orders of magnitude ) . as for the sfhs ,
the code- and resolution - dependent differences amongst the bh masses in the merger simulations are more significant than for the isolated disc simulations . in many but not all examples , the bh masses are greater in the arepo simulations than in the gadget-3 runs . the resolution - dependent variations ( which are at most @xmath40 dex and usually significantly less ) are typically less than the code - dependent differences ( for a given resolution , these can be as great as an order of magnitude ) , and the resolution dependence is not systematic .
consequently , for a given code , the final bh masses should be considered uncertain by as much as a factor of a few .
this reflects the high degree of non - linearity in the feedback - regulated bh growth .
any small variation in the local gas conditions at the bh s position can influence its exponential growth rate and hence become strongly amplified with time .
note also that the bh accretion histories can differ significantly depending on the resolution and code ; thus , during the merger , the bh masses at a given time can differ more significantly than the final bh masses ( i.e. , the bh masses after the bh growth has terminated , which can occur at different times for different resolutions and codes ) .
interestingly , for the merger , which was simulated at three resolutions , the bh mass evolution in the two higher - resolution ( resolutions r3 and r4 ) simulations performed with a given code is almost identical , but the @xmath49 dex difference in the final bh masses yielded by the two different codes persists .
thus , it is possible that the bh masses yielded by a given code would converge if all simulations were performed at even higher resolution , but we have not performed such simulations because of the computational expense and because the code - dependent differences , which are the focus of this work , remain even for the highest - resolution simulations .
+ fig .
[ fig : gas_comparison ] shows example gas surface density plots for the arepo ( top row ) and gadget-3 ( bottom row ) simulations with bh accretion and agn feedback . as for the simulations without bh accretion and agn feedback , the morphologies are qualitatively similar , but
the details differ . furthermore , as for the sfhs ,
the code - dependent differences in the gas morphologies are more significant when bh accretion and agn feedback are included .
the first column shows the isolated disc at @xmath61 gyr . at this time , a bar is evident in the arepo simulation and much of the gas in the central region has been consumed because of the bar instability . in the gadget-3 simulation ,
a bar is not evident ; rather , a large , irregularly shaped cavity , which contains some small clumps of dense gas , has formed .
part of the reason for the significant differences between the arepo and gadget-3 results is that in the arepo simulations , the bh can only consume gas that is within 50 pc , whereas in gadget-3 , the region from which gas is accreted grows as the gas near the bh is depleted because a fixed number of neighbours is used to calculate the sph density estimate .
note , however , that generically , holes tends to form around the bhs for the following physical reason : in the quiescent state that is eventually reached in isolation or at the end of a merger ( when the bh growth has effectively shut off ) , a small bubble of hot , low - density gas around the bh is created by the pressure that is sustained in our feedback model by the residual accretion .
the second column shows the merger simulation at @xmath62 gyr ( after first pericentric passage ) .
the results of both codes exhibit extended , smooth tidal features , and the morphologies are almost indistinguishable .
this column exemplifies the good agreement that is characteristic of the pre - starburst phase of the merger simulations . the third column of fig .
[ fig : gas_comparison ] shows the merger near the peak of the starburst ( @xmath63 gyr ) .
some of the filamentary structure is similar in the arepo and gadget-3 simulations , but the detailed morphologies differ significantly . as noted above , the gadget-3 result exhibits many spurious clumps of gas that are not present in the arepo simulation . finally , the fourth column shows the merger at @xmath48 gyr , @xmath50 myr after the starburst .
note that for comparison , this is the same simulation and time as shown in the fourth column of fig .
[ fig : nb_gas_comparison ] , except that bh accretion and agn feedback are included here . for a given code ,
the gas morphologies are similar to those of the simulations for which bh accretion and agn feedback were not included ( fig . [ fig : nb_gas_comparison ] ) .
one notable difference is that in the gadget-3 simulation , the gas disc is less pronounced and the hot halo is more prominent .
once again , the gadget-3 simulation features an extended halo of hot gas and spurious clumps , both of which are not present in the arepo simulation .
note that in this example , the code - dependent differences in the gas morphologies are more significant than those caused by the inclusion of agn feedback .
this result is a counterexample to the conclusion of @xcite regarding the effects of various star formation and stellar feedback models compared with differences between codes and implies that it is highly desirable to use the most accurate hydrodynamical solver possible . as for the simulations without bh accretion and agn feedback ,
the gas phase structure in the gadget-3 and arepo simulations is similar in the pre - starburst phase but can differ significantly during and after the starburst .
again , we only present one example to illustrate the characteristic differences here , but the interested reader can visit the aforementioned url to examine the evolution of the gas phase structure for all simulations .
[ fig : pd_comparison ] shows gas phase diagrams for the simulations with bh accretion and agn feedback at the same time as in fig .
[ fig : nb_pd_comparison ] ( @xmath48 gyr , @xmath64 myr after the peak of the starburst and agn activity ) . for a given code ,
the inclusion of agn feedback causes there to be more gas in the hot halo and correspondingly less gas on the eos ( the thin line in the lower - right corner ) . as for the simulations that did not include agn feedback , the gadget-3 simulation features more hot halo gas than the arepo simulation , in which the hot halo gas cools more efficiently . here , we present various tests that demonstrate the effects of the different treatments of bh accretion and agn feedback discussed above .
we use the analogue as one test case . whereas for this simulation , the differences in the results are small in an absolute sense , they are systematic and can have more significant effects in other simulations .
we chose to use the analogue as a test case because the sfh and bh growth are comparatively simple ; thus , differences can be more easily understood .
furthermore , the bh grows very little ( @xmath65 , the seed mass , throughout the simulation ) and should have a negligible effect on the sfh of the galaxy .
thus , the ` no bh ' case can be used as the baseline with which to compare the other runs ; ideally , the sfhs should be the same for the bh and ` no bh ' cases .
we also investigate the effects of different bh accretion and agn feedback for the merger simulation . in this significantly more complicated case ,
the interpretation of the comparisons among the different treatments of bh accretion and agn feedback is less straightforward , but many of the effects observed for the case are also observed here .
the tests presented here justify our fiducial choices for the bh accretion and feedback implementations and demonstrate some important numerical effects that are not always appreciated in the literature .
+ , but for the merger .
the results are almost independent of the @xmath66 value for the higher resolution because the resolution around the bh is sufficiently high without the forced refinement .
the final bh masses for the two resolutions agree better when @xmath67 pc , but this agreement does not indicate better convergence .
see the text for details.,title="fig : " ] + , but for the merger .
the results are almost independent of the @xmath66 value for the higher resolution because the resolution around the bh is sufficiently high without the forced refinement .
the final bh masses for the two resolutions agree better when @xmath67 pc , but this agreement does not indicate better convergence .
see the text for details.,title="fig : " ] in sph , as the gas density in the immediate vicinity of a bh decreases because of gas consumption and expulsion , the radius over which the density is calculated increases because the number of neighbours used for the density estimate and the particle masses are fixed .
thus , in some situations , the bh accretion rate can be overestimated and gas fuels the bh from unphysically large scales ; this is especially problematic in cosmological simulations , for which the resolution is often not better than a kiloparsec . in arepo , the standard cell refinement scheme attempts to keep cell masses comparable ; thus , a similar effect , in which cells near the bh grow large , can occur .
however , unlike in sph , we can overcome this potential problem by preventing cells near the bh from becoming too large and using the gas density of the cell that contains the bh to calculate the accretion rate .
consequently , if the gas density in the vicinity of the bh decreases , the accretion rate decreases concomitantly .
[ fig : refinement_test ] demonstrates the effects of limiting the maximum cell size near the bh particle for the isolated disc case .
we force cells located within 500 pc of the bh to be refined if their size is greater than some value @xmath66 .
we show the results for @xmath68 and 200 pc in fig .
[ fig : refinement_test ] .
the top panel indicates that the choice of @xmath66 has no effect on the sfh , and the slight difference between the different resolutions at late times is independent of @xmath66 .
however , the growth of the bh differs systematically . for both resolutions
, the bh grows more when @xmath67 pc because of the effect described above , and the consequences are more severe for the lower - resolution simulation .
when @xmath68 pc , the final bh mass is slightly less and the two different resolutions agree perfectly .
[ fig : refinement_test_merger ] shows the results of changing @xmath66 for the merger simulation . in this case , the sfh is better converged when a maximum cell when @xmath68 pc . for the higher - resolution simulations ,
the bh growth history is unaffected by the choice of @xmath66 .
thus , for this resolution ( @xmath69 pc ) , the refinement near the bh is sufficiently high even without forcing refinement .
interestingly , the final bh mass of the lower - resolution simulation with @xmath67 pc agrees better with that of the higher - resolution simulations than when @xmath68 pc . in this case , allowing larger cells near the bh by setting @xmath67 pc serves to mitigate some of the resolution dependence : in the lower - resolution simulation with @xmath68 pc , during the starburst at final coalescence , the central gas density is greater than in the higher - resolution runs .
consequently , the bh grows more rapidly during that time . using @xmath67 pc results in a decreased central gas density and thus less - massive bh .
however , this result should not be taken as an indication that the bh mass is better - converged when @xmath67 pc : if the bh mass were truly converged , the @xmath68 pc simulation should agree at least as well because in this simulation , the number of resolution elements is greater than or equal to that of the lower - resolution simulation .
+ + + + to calculate the bh accretion rate using equation ( [ eq : bondi ] ) , we require the gas density and sound speed near the bh . in arepo
, both quantities can be estimated in a manner analogous to that in sph , i.e. , by averaging over some number of nearest neighbour cells ( typically , 32 ) .
however , it is also possible to use the gas density and sound speed for the cell in which the bh is located . in principle
, the cell values should be more representative of the conditions near the bh and thus ensure a more physical calculation of the accretion rate , but they may be too noisy to be useful .
we will explore the effects of these choices now .
[ fig : accretion_test ] demonstrates the variations that arise from using different methods to calculate the bh accretion rate in arepofor the isolated disc case .
the gadget-3 results are shown for comparison . in all cases ,
the mass over which the feedback energy is distributed is kept constant , i.e. , the number of cells or particles over which the energy is distributed is increased as the cell mass is decreased .
furthermore , cells within 500 pc of the bh are forced to have size less than @xmath68 pc .
the differences in the sfh ( upper - left panel ) for the various treatments are small and comparable with the resolution - dependent differences .
the bh growth history , in contrast , can be affected significantly ; the upper - right panel indicates that the final bh mass can differ by as much as 0.4 dex , and the effect in merger simulations can be even larger .
the bh grows most in arepo when the sph density and sound speed are used ( the blue curve ) .
the reason is that , as shown in the lower - left panel , in arepo , the sph density estimate is systematically greater than the cell density ( and similar to the gadget-3 density ) .
the gadget-3 simulations ( red curve ) exhibit the next highest bh growth ; the reason that they feature less growth compared with the analogous arepo simulations is that the sph sound speed estimate ( lower - right panel ) is systematically higher in gadget-3than arepo .
when the cell density is used , the bh grows less because the cell density is systematically lower than the sph density .
the growth is most suppressed when the cell sound speed is used ( cyan curve ) because the cell sound speed is systematically higher than the sph sound speed .
furthermore , the cell sound speed is very noisy , with variations of @xmath70 km@xmath24 from the mean value , and the cell density is significantly more noisy when the cell sound speed is used to calculate the bh accretion rate ( compare the cyan and green curves in the lower - left panel of fig .
[ fig : accretion_test ] ) . fig .
[ fig : accretion_test_merger ] compares the different bh accretion rate treatments for the merger simulation . as for the isolated disc case ,
the bh grows most when the sph estimates for the gas density and sound speed are used to calculate the accretion rate .
using the cell density rather than the sph estimate results in a very similar final bh mass .
furthermore , using both the cell density and sound speed results in the smallest final bh mass of the three lower - resolution simulations because the cell sound speed can be more than a factor of three greater than the sph estimate .
although the differences in the bh masses caused by varying the sub - resolution model accretion rate calculation are significant ( @xmath52 dex ) , they are less than the differences between the two resolutions for a fixed code and those between the two codes for a fixed number of particles / cells . based on these results , we chose to use the cell density for our production runs because this is more representative of the density near the bh than is the sph density , yet the two have comparable amounts of noise .
however , we chose to use the sph estimate for the sound speed because the cell value is too noisy .
+ + + + the manner in which the agn feedback energy is distributed can also affect the bh growth .
as explained in section [ s : bh_fb_method ] , our preferred method is to keep the mass over which the feedback energy is distributed ( the ` feedback mass ' ) constant .
thus , for higher - resolution runs , in which the particle / cell mass is lower , the number of particles / cells over which the feedback energy is distributed should be increased .
however , this is not the only possible approach ; an alternative approach is to keep the number of particles / cells over which the feedback energy is distributed fixed .
but , as we will show now , this causes a stronger resolution dependence , which is undesirable for a sub - resolution model .
[ fig : fb_test ] demonstrates the differences between these two choices for the isolated disc case . as in the previous test , there are minor variations in the sfhs at late times , but the magnitude of the difference is similar to that caused by varying the resolution .
however , varying the method for distributing the feedback energy causes significant and systematic changes in the bh growth history . first , note that the dashed and solid blue lines , which correspond to the lower - resolution arepo simulation and the higher - resolution simulation in which the feedback mass is kept constant , agree perfectly .
the corresponding curves for the density and sound speed are ( necessarily ) in good agreement . in contrast
, the arepo run in which the number of cells over which the feedback is distributed is kept constant ( the cyan line ) has systematically less bh growth .
the reason for this discrepancy is that the latter simulation has systematically lower gas density ( bottom left ) and higher sound speed ( bottom right ) because the agn feedback heats a smaller mass of gas compared with the lower - resolution run .
consequently , the steady state that is reached is at a higher temperature and lower density .
the analogous effect can be observed for gadget-3 , but in this case , there is still some resolution dependence even when the feedback mass is kept constant .
[ fig : fb_test_merger ] shows a comparison of the different methods for distributing the feedback energy for the merger . for this model ,
the sfh differs most significantly in the areporun in which an equal cell number is used ; in this case , star formation is quenched more effectively by the agn after the starburst than in any of the other simulations , and the sfh in this regime differs most significantly from that of the lower - resolution run .
for the gadget-3simulations , the final bh mass of the lower - resolution run agrees slightly better with that of the higher - resolution run in which the mass over which the feedback energy is distributed is kept constant , as was observed for the case .
however , for the areposimulations , the final bh mass in the higher - resolution run in which the number of cells over which the feedback energy is distributed is kept constant agrees better with the final bh mass of the lower - resolution simulation .
the reason for this behaviour is that in the equal - mass run , the initial stage of rapid bh growth ( @xmath71 gyr ) is terminated slightly earlier and at a slightly lower bh mass than in the equal - cell - number run , primarily because the rapid decline in the gas density occurs earlier in the equal - mass run .
consequently , during the final - coalescence phase , in which the bh undergoes eddington - limited accretion , the less - massive bh grows less .
this unexpected behaviour demonstrates the difficulty of predicting the effects of the sub - resolution model in the significantly more complex context of galaxy mergers and highlights the aforementioned conclusion that the final bh masses should only be considered robust to within a factor of a few .
furthermore , the difference in the final bh masses of the two higher - resolution simulations is less significant than the difference between either of these masses and that of the lower - resolution run . the purpose of the sub - resolution model is to encapsulate in a simple manner physics that is not included in the simulations .
the implicit assumption in our model is that the feedback energy deposited by agn is thermalised over some physical scale , and this physical scale does not and should not depend on the resolution of our simulations .
thus , in our production runs , we kept the feedback mass constant .
in earlier work @xcite , we have compared results from gadget-3and arepoand , in some cases , identified significant differences that we attributed to numerical issues with the conventional formulation of sph . in the present paper and in studies of the ly-@xmath6 forest ( e.g. , * ? ? ? * ; * ? ? ? * ) , it has been found that sph can produce results in certain regimes that agree well with grid - based codes .
to understand this situation , we first briefly review the primary limitations of the sph approach that have become clear in recent work .
this then allows us to discuss why sph can be expected to work reasonably well in some applications but not in others . finally , we comment on what this implies for the numerical robustness of different types of previous work .
the traditional formulation of sph used in this work has been and still is
being used for many astrophysical simulations .
it is thus important to understand which of these results may be influenced by numerical artefacts of the type discussed below . whereas we acknowledge that there have been impressive efforts to address at least some of these issues ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , we feel that there remain many lingering misconceptions about sph that muddle the interpretation of the reliability of simulations performed even with updated versions of this algorithm .
the transition from the continuum equations of fluid dynamics to the discrete form used by sph involves a two - step procedure ( e.g. , * ? ? ?
first , the exact fluid quantities are replaced by smoothed versions via a convolution with the smoothing kernel .
second , the integral forms of these convolutions are replaced by discrete sums over the sph particles such that they can be evaluated numerically .
the error made in the discretization step depends not on the number of sph particles , @xmath72 , but instead on the number of neighbours in the discrete sums , @xmath73 .
if , as is typically done , @xmath73 is held fixed as @xmath72 is increased , there will be a _
constant _ source of error in the local estimates even as the resolution of the simulation is nominally increased @xcite .
consequently , local fluid estimates are often noisy and are not guaranteed to approach their continuum values as @xmath72 is increased .
this source of noise , although small , may have a significant impact on flows in which the energy content is dominated by internal energy rather than kinetic or gravitational energy .
the noise is particularly strong in gradients of interpolated quantities , most notably in the pressure force @xcite .
furthermore , if @xmath72 is increased without simultaneously increasing @xmath73 , the solution may asymptote to a fixed result that is different from the true solution because of this constant source of error @xcite .
tests by @xcite demonstrated that conventional implementations of sph do not accurately describe jumps in physical quantities across contact discontinuities because the pressure effectively becomes multi - valued at the interface , thereby resulting in artificial repulsive forces that act as a macroscopic surface tension .
if two fluid phases in pressure equilibrium shear relative to one another , the spurious surface tension inhibits the proper growth of kelvin - helmholtz instabilities and the two phases will not mix together correctly . instead , the colder
, more dense phase can fragment into clumps that retain their identity because of the presence of the spurious surface tension ( e.g. , * ? ? ?
* ; * ? ? ?
various new formulations of sph have shown promise in alleviating this problem ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) ; thus , we will not dwell on it here .
however , we mention it for completeness and because nearly all previous studies of galaxy mergers using sph have employed traditional formulations of sph that are susceptible to this surface tension issue .
in most implementations of sph , the mass continuity equation is not integrated in detail ; instead , estimates of the fluid density at any given simulation time are made using kernel interpolation ( e.g. , * ? ? ?
* ) applied to the current particle positions .
partly for this reason , sph is often referred to as a ` lagrangian ' method . however , sph is only ` pseudo - lagrangian ' because the particle mass is fixed in time and particle shapes are not allowed to become distorted arbitrarily by the flow ( see @xcite for a detailed discussion ) .
consequently , sph can not properly describe fluid mixing on small scales , whereas in arepo , mixing is not suppressed because the implied mass exchange between cells is computed correctly according to the continuity equation . in
essentially all widely used sph codes , shocks are captured using some form of artificial viscosity . ideally , this viscosity should operate only in and near shocks because it can have unintended consequences for the properties of the flow in other regions if it is not accurately controlled ; for example , it can cause spurious cooling in cosmological applications ( e.g. , * ? ? ?
moreover , the action of the viscosity within shocks is to locally broaden them over several smoothing lengths , thereby degrading the effective spatial resolution in shocked regions . many grid - based codes , including arepo , instead treat shocks by solving the riemann problem across all cell - cell interfaces .
this treatment has several advantages because it implies that such codes minimize the additional source of unphysical diffusion that would arise from artificial viscosity and because shocks can be spatially resolved more precisely than in sph . from the discussion in section [ s : limitations_of_sph ]
, we can now provide arguments as to why sph yields results that are reliable in some situations and why it fails in others .
for definiteness , we consider four applications : ( 1 ) idealized tests of driven supersonic and subsonic turbulence , ( 2 ) the intergalactic medium ( igm ) , ( 3 ) gas accretion on to galaxies , and ( 4 ) starbursts and agn activity in galaxy mergers , as described in this paper .
@xcite compared gadget-3and arepofor idealized simulations of isothermal turbulence in periodic boxes subject to large - scale forcing .
their tests demonstrate that the traditional formulation of sph , as incorporated in gadget-3 , does not yield a proper cascade of energy to small scales when the turbulence is subsonic , as illustrated in their fig .
however , otherwise identical simulations performed with arepoand the navier - stokes version of arepo@xcite produced a well - developed turbulence spectrum , not only in the inertial range but also through the dissipational range .
in contrast , sph was found to perform significantly more reliably for supersonic turbulence @xcite , yielding results that are in reasonable agreement with arepoindependent of the motion of the mesh .
these apparently contradictory conclusions can be readily explained by the limitations discussed above in section [ s : local_noise ] and section [ s : shock_capturing ] . in the supersonic limit , the energy density of the fluid is dominated by kinetic energy . although noise is still present in the local fluid quantities ( see section [ s : local_noise ] )
, its influence is subdominant in this regime .
similarly , spurious entropy generation from the artificial viscosity ( see section [ s : shock_capturing ] ) is also a minor source of error .
in contrast , for subsonic turbulence , the internal energy is comparable in magnitude to the kinetic energy ; thus , the force errors from gradient noise and excessive spurious dissipation corrupt the solution on small scales such that the correct cascade of turbulent energy is not reproduced .
early simulation results for the ly-@xmath6 forest obtained using both sph @xcite and grid codes @xcite agreed well .
more refined comparisons ( e.g. , * ? ? ?
* ; * ? ? ?
* ) have demonstrated that when applied to the same initial conditions , sph and grid codes yield statistical measures for the ly-@xmath6 forest , such as flux probability distribution functions and power spectra , that agree at the @xmath74 level .
the reasons for this agreement can be understood based on the discussion in section [ s : limitations_of_sph ] .
the energy density of the igm gas responsible for the ly-@xmath6 forest is dominated by kinetic and gravitational energy ; thus , errors in the local fluid quantities due to noise are subdominant .
furthermore , the physical state of the gas is simple , in the sense that different phases of gas are not in close proximity .
thus , errors such as those discussed in sections [ s : inaccurate_instabilities ] and [ s : inaccurate_continuity ] will not greatly affect the igm , and hence the sph results are fairly reliable .
whether this conclusion also extends to metal lines , for which issues of mixing of galactic outflows with pristine igm gas become important , has not been investigated thus far . within galaxy haloes ,
however , the physical state of the gas can differ significantly between , e.g. , gadget-3and arepo@xcite .
why should sph predict the physical state of the gas responsible for the ly-@xmath6 forest so reliably yet fail so spectacularly within the haloes of galaxies ?
this can be understood by realizing that the internal energy in approximately hydrostatic gas is no longer negligible compared with the kinetic and gravitational energy .
noise in the local sph estimates can then significantly affect the physical state of the halo gas by , for example , producing spurious viscous heating effects that reduce cooling flows ( e.g. , * ? ? ?
moreover , the gas within haloes can exhibit complex phase structure , with cold , dense gas in close proximity to and shearing relative to shock - heated diffuse gas .
when simulated with traditional sph , the different gas phases will not mix correctly , as demonstrated by @xcite , because fluid instabilities ( section [ s : inaccurate_instabilities ] ) and mixing due to fluid motions ( section [ s : inaccurate_continuity ] ) are not treated properly . instead , the cold gas can fragment into clumps that remain intact ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , thereby delivering an artificial supply of cold , low - angular - momentum gas to the central galaxy , which in turn can inhibit the formation of a rotationally supported disc . in this paper , we have demonstrated that the results of idealised ( i.e. , non - cosmological ) numerical experiments involving isolated disc galaxies and galaxy mergers are relatively similar between sph and the moving - mesh approach , in contrast with cosmological simulations of forming galaxies .
this finding especially holds for simulations in which agn feedback is not included or , more generally , during early stages of mergers when the gas structure is relatively simple .
later , once the gas becomes virialized and feedback from bh growth drives large - scale outflows , some detailed differences do however appear .
the discussion in section [ s : limitations_of_sph ] can again be used as a guide to understand this behaviour . in the simulations presented here ,
we construct models of disc galaxies that evolve in isolation or merge . in these models ,
the gas is initially rotationally supported in the galaxy potential ; thus , its internal energy is small compared with its kinetic and gravitational energy and the flows are effectively supersonic . in this regime ,
similar to the gas that produces the ly-@xmath6 forest , we expect that noise in the sph estimates ( see section [ s : local_noise ] ) should not be a significant source of error .
moreover , in the approach adopted here , as formulated originally in @xcite , there is no effort made to resolve the multiphase structure of the star - forming gas . instead
, the ism is described using a sub - resolution model .
because of this , as for the gas in the igm , the gas locally has a simple structure and different phases of gas do not exist in close proximity .
as long as the gas can be characterised well in this manner , we do not expect errors associated with an inaccurate treatment of fluid instabilities ( section [ s : inaccurate_instabilities ] ) or mixing ( section [ s : inaccurate_continuity ] ) to affect our model galaxies .
these conditions can start to be violated in a galaxy merger as gas is shock heated and virialized and as different gas phases start to shear relative to one another because of , e.g. , the action of agn feedback .
we note that these conditions would also be violated in simulations that have enough resolution to truly resolve the multi - phase structure of the ism . in this case
, extremely strong local density constrasts would be prevalent .
furthermore , in contrast with cosmological simulations , in which gas cooling is a crucial determinant of how and when gas is supplied to galaxies , the gas flows in galaxy merger simulations are driven primarily by gravitational torques . in both the gadget-3 and arepo simulations presented here , we use the same accurate tree - based gravity solver .
consequently , the gravitational forces are treated equally accurately in both codes .
this is another reason that the results agree relatively well .
for the same reason , the dark matter halo mass functions of arepo and gadget-3 cosmological simulations agree very well despite the significant differences in the distributions of baryons @xcite .
as stressed above , the agreement is less good during the starburst and post - starburst phases of the merger simulations . during the starburst ,
gas is shock - heated and a hot halo forms .
if agn feedback is included , the amount of gas in the hot halo is increased .
once this hot halo forms , cooling becomes important for the post - starburst sfr and gas morphology and thus the relevant differences between sph and grid - based approaches become manifest . as in cosmological simulations @xcite and idealised simulations of ` inside - out ' disc formation @xcite , gas cools more effectively from the hot haloes formed in the mergers in arepo than in gadget-3 .
furthermore , some gas that is ejected during the starburst and agn activity falls back on to the remnant during the post - starburst phase . in the arepo simulations , infalling clumps and filaments are effectively disrupted , whereas in the gadget-3 simulations they survive ( see section [ s : inaccurate_instabilities ] ) ; the differences in this regard are the same that explain why ` cold flows ' are less prominent in cosmological simulations performed with arepo @xcite , as discussed above .
it is worthwhile considering the implications of the differences we have demonstrated , regarding both the robustness ( or lack thereof ) of previous work and comparisons of simulations with observations .
we restrict ourselves to comments on the rich literature of simulations of isolated galaxies and galaxy mergers , which forms the central topic of this paper . for the reasons discussed above
, we in general expect many of these past works to be at most weakly affected by the inaccuracies of traditional sph provided they use cold gas from the start and do not have sufficient resolution to resolve a truly multi - phase ism . however , the differences that we found are much more significant once hot haloes are present in the simulations .
hence , we expect that the results of simulations that initially include hot haloes in the progenitor galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) could differ significantly if a moving - mesh technique rather than traditional sph were used .
furthermore , conclusions regarding hot haloes produced by gas shocking in mergers ( e.g. , * ? ? ?
* ) , the x - ray emission from hot haloes ( e.g. , * ? ? ?
* ) , and the properties of starburst- and agn - driven winds ( e.g. , * ? ? ?
* ; * ? ? ?
* ) may depend on the numerical method employed .
also , because the re - formation of discs in mergers depends on gas cooling from the hot halo of the remnant , studies of disc re - formation ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) might also be affected by the spurious suppression of cooling that is inherent in sph .
indeed , we have already noted that the discs that re - form in the mergers presented here are very different in the gadget-3 and arepo simulations ( see the fourth columns of figs .
[ fig : nb_gas_comparison ] and [ fig : gas_comparison ] ) .
the differences between traditional sph and moving - mesh simulations may also have important implications for studying agn feedback with merger simulations .
in particular , the differences in the bh masses yielded by arepo and gadget-3can affect the strength of the feedback , the agn luminosity and duty cycle ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , and the @xmath75 relation of the merger remnants ( e.g. , * ? ? ? * ) .
however , we caution that for the thermal agn feedback model presented here , the final bh masses depend primarily on the feedback coupling efficiency @xmath25 @xcite . because this number is very uncertain , it may be possible and reasonable to simply use a slightly reduced value of @xmath25 in arepo to make the final bh masses more consistent with those yielded by gadget-3 .
furthermore , the scatter in the bh masses for different resolutions suggests that the values of the final bh masses are only robust to within a factor of a few ( see also * ? ? ?
recently , @xcite have presented idealised isolated disc and galaxy merger simulations performed with gadget-3that include models for star formation and stellar feedback that are significantly more sophisticated than the eos approach used here .
this work attempts to directly resolve the ism multi - phase structure .
hence , one possible concern is the use of traditional sph .
however , @xcite also demonstrate that the results of their simulations agree rather well with the results of gadget-3simulations that employ the much simpler eos approach .
also , @xcite compared simulations performed using the traditional density - entropy formulation of sph used in gadget-3with simulations performed using an alternative pressure - entropy formulation of sph @xcite that is designed to overcome the inaccurate treatment of fluid instabilities ; they found that the spurious cold clumps present in the outflows in the standard gadget-3simulations were not formed when the new pressure - entropy flavour of sph was used .
thus , many of the conclusions of @xcite are likely robust to the hydrodynamical solver employed because the rotational support is still dominant over pressure forces , although some details , such as the post - starburst gas morphologies , are clearly affected .
we have compared a suite of idealised isolated disc and galaxy merger simulations performed using the sph code gadget-3 and the moving - mesh hydrodynamics code arepo , both with and without bh accretion and agn feedback . to isolate the effects of the hydrodynamical solver used
, we have kept all other aspects of the simulations ( i.e. , the gravity solver , the treatment of cooling , and the sub - resolution models for star formation , supernova feedback , bh accretion , and agn feedback ) as similar as possible .
our principal conclusions are the following : 1 . unlike for cosmological hydrodynamical simulations ,
the results of idealised ( non - cosmological ) isolated disc and merger simulations performed with arepo and gadget-3 are similar because ( 1 ) in these simulations , the gas is already initialised by hand in a rotationally supported disc , and thus the flow is supersonic in character and noise in the sph estimates is not a significant source of error ; ( 2 ) the gas phase structure is relatively simple ; and ( 3 ) gravitational torques ( rather than cooling ) are the primary determinant of the gas inflows .
2 . when bh accretion and agn feedback are not included , the results are quite insensitive to the code used .
the sfhs and cumulative stellar mass formed versus time are remarkably similar .
3 . when bh accretion and agn feedback are included , the results of the two codes are qualitatively similar , but the quantitative differences are more significant . in particular , the bh masses can differ by as much as an order of magnitude .
the arepo simulations typically yield larger bh masses .
4 . the gas morphologies and phase structures are also similar but differ in detail .
primarily , arepoyields smoother , less clumpy morphologies and less prominent hot gaseous haloes in the post - starburst phase . 5 . as for cosmological simulations ,
the differences between the sph and moving - mesh results are primarily caused by more efficient cooling of hot halo gas and stripping of gas clumps and filaments in arepo .
much of the previously published results of idealised isolated disc and galaxy merger simulations are likely robust to the inaccuracies that are inherent in traditional sph . however , for some studies , such as simulations in which the progenitors are initialised with hot haloes and studies of disc re - formation in mergers , the arepo results may differ qualitatively from those yielded by traditional sph .
consequently , it would be interesting to revisit such studies using arepo .
it is certainly reassuring that the bulk of our results for merger simulations exhibit good quantitative agreement between sph and the very different moving - mesh technique . interpreting this as a vindication of using sph for merger simulations
would nevertheless be incorrect .
as we have discussed , unsolved conceptual problems with the accuracy of sph and its convergence rate remain even in the most recent incarnations of the proposed improved versions of sph .
we therefore think that more accurate numerical techniques , such as our moving - mesh approach , should clearly be preferred over traditional sph to perform such simulations , especially in future calculations for which a higher numerical precision and the resolution of multi - phase media is desired .
we thank shy genel , phil hopkins , federico marinacci , rdiger pakmor , ewald puchwein , and debora sijacki for useful discussion ; jorge moreno and franois schweizer for detailed comments on the manuscript ; and the anonymous referee for a critical report that helped to improve the paper .
cch is grateful to the klaus tschira foundation for financial support , acknowledges the hospitality of the aspen center for physics , which is supported by the national science foundation grant no .
phy-1066293 , and heartily thanks andreas bauer for many python tips , which were very helpful for ( almost completely ) curing a long - standing idl dependency .
vs acknowledges support from the european research council under erc - stg grant exagal-308037 . this research has made use of nasa s astrophysics data system bibliographic services . | galaxy mergers have been investigated for decades using smoothed particle hydrodynamics ( sph ) , but recent work highlighting inaccuracies inherent in the traditional sph technique calls into question the reliability of previous studies .
we explore this issue by comparing a suite of gadget-3 sph simulations of idealised ( i.e. , non - cosmological ) isolated discs and galaxy mergers with otherwise identical calculations performed using the moving - mesh code arepo .
when black hole ( bh ) accretion and active galactic nucleus ( agn ) feedback are not included , the star formation histories ( sfhs ) obtained from the two codes agree well . when bhs are included , the code- and resolution - dependent variations in the sfhs are more significant , but the agreement is still good , and the stellar mass formed over the course of a simulation is robust to variations in the numerical method . during a merger ,
the gas morphology and phase structure are initially similar prior to the starburst phase .
however , once a hot gaseous halo has formed from shock heating and agn feedback ( when included ) , the agreement is less good . in particular , during the post - starburst phase , the sph simulations feature more prominent hot gaseous haloes and spurious clumps , whereas with arepo , gas clumps and filaments are less apparent and the hot halo gas can cool more efficiently .
we discuss the origin of these differences and explain why the sph technique yields trustworthy results for some applications ( such as the idealised isolated disc and galaxy merger simulations presented here ) but not others ( e.g. , gas flows onto galaxies in cosmological hydrodynamical simulations ) .
[ firstpage ] hydrodynamics methods : numerical galaxies : interactions galaxies : starburst galaxies : active galaxies : formation . |
while some long - duration gamma - ray bursts ( grbs ) are clearly associated with supernovae ( sne ) , a deeper understanding of the grb / sn connection remains elusive .
the grb / sn link was first confirmed observationally with the detection of the low - reshift grb 980425/sn 1998bw ( @xmath0 ) @xcite .
grb 980425 , however , was not a typical grb ; it was under - luminous in gamma - rays and had no detected optical afterglow ( oag ) .
sne were later associated with typical grbs at cosmological redshifts ( e.g. * ? ? ?
* ; * ? ? ?
however , another low gamma - ray luminosity event similar to grb 980425/sn 1998bw was not observed until grb 031203 .
this burst was also orders of magnitude under - energetic and , despite its low redshift ( @xmath1 ) @xcite , was followed by only a dim oag @xcite .
follow - up observations of this burst detected a sn - like brightening @xcite and sn 2003lw was confirmed spectroscopically by @xcite .
the spectra of sn 2003lw were reminiscent of those of sn 1998bw @xcite .
the the two sn also had similiar peak magnitudes , although sn 2003lw was somewhat brighter and evolved more slowly .
their light curve shapes were also qualitatively different , with sn 1998bw climbing smoothly to peak while sn 2003lw experienced a broad plateau .
the low gamma - ray luminosity of grb 980425 and 031203 suggested that they might represent a new grb category .
alternatively , they could be normal grbs that appear under - luminous because their jetted emission is observed off - axis ( e.g. * ? ? ?
* ; * ? ? ?
a comparison of the event rate of low- to typical - luminosity grbs was warranted but only a small and inhomogeneous sample of well - localized grbs existed before _
swift _ , as pre-_swift _ grbs were detected using multiple instruments , each with unique sensitivity and sky coverage .
_ swift _ provides a large and homogeneous grb sample that is well - suited for rate calculations . on 2006
february 18 at 03:34:30 ut _ swift _ detected a new low - luminosity event : grb 060218 @xcite .
this was an unusual grb with weak gamma - ray emission lasting over 2000 seconds @xcite .
this burst was followed by an unusual oag that brightened for 10 hours before decaying like a typical oag ( e.g. * ? ? ?
this was the first _ swift _ grb to be associated with a sn : sn 2006aj .
the sn was initially noted in spectral observations @xcite and then detected as an optical re - brightening ( e.g. * ? ? ?
* ; * ? ? ?
* etc . ) . at @xmath2 , grb 060218
is now the second closest grb with a measured redshift @xcite .
this is the third example of a grb - related sn in which the gamma - rays are highly under - luminous and the sn light curve is clearly distinct from the grb s oag .
hereafter , we will refer to these long - duration , low - luminosity events as l@xmath3grbs .
smarts observations of sn 2006aj began on 2006 feb 22 at 00:35 ut @xcite .
we present optical / ir data obtained between 5 and 30 days following grb 060218 .
our homogeneous data demonstrate that the light curve of sn 2006aj is qualitatively similar to that of the pre-_swift _
l@xmath3-grb sne 1998bw/2003lw .
we argue in 4 that the detection of this single event in the _ swift _ era already places strong constraints on the nature of l@xmath3-grbs .
our data was obtained using the andicam instrument mounted on the 1.3 m telescope at cerro tololo inter - american observatory .
this telescope is operated as part of the small and moderate aperture research telescope system ( smarts ) consortium .
nightly imaging was obtained over 26 days with occasional interruptions for weather and equipment problems .
the grb / sn was only observable for a limited period of time immediately after twilight ( @xmath4hr ) .
consequently , all observations were obtained at high airmass ( @xmath5 ) .
each nightly data set consisted of 6 individual 360-second i - band observations obtained simultaneously with 30 dithered 60-second j - band images .
the data were reduced in the same way as in @xcite .
a few additional steps were added , including cosmic ray removal in the i - band images using the l.a . cosmic program @xcite and
i - band fringe correction using an iterative masking technique .
some images were not included in the final frames because of excessive background due to twilight or because of telescope drift . typically , the final frames were equivalent to 30 minutes of i / j - band exposure time .
the relative magnitude of the sn + host galaxy was determined by comparison with 11 ( 3 ) on - chip , non - variable objects in i ( j ) using seeing matched aperture photometry .
differential magnitudes were converted to apparent magnitudes by comparison , on photometric nights , with landolt standard stars in the fields of ru149 and pg1047 @xcite for the i - band images , and with 3 on - chip 2mass stars @xcite for the j - band images .
the difference in airmass value between the landolt standard frames and science frames was corrected for using an extinction coefficient of 0.066 magnitudes per airmass .
the light curves are shown in figure 1 and the photometric data are summarized in table 1 .
the error bars represent the photometric measurement error , which accurately reflects nightly variations in image quality but does not account for systematic measurement errors .
in addition to the relative night - to - night uncertainty , there is a systematic error of 0.05 magnitudes in i and j resulting from uncertainties in the photometric calibration .
figure 1 shows that grb 060218 s optical and ir counterpart brightened for the first two weeks and then proceeded to gradually decay .
this behavior is not consistent with that of a standard grb oag ( e.g. * ? ? ?
* ) but is reminiscent of the low redshift events sne 1998bw/2003lw .
identification of this optical emission as a sn is possible from our data alone due to our dense observations and the object s particular transient behavior ; spectral evidence obtained by other groups clearly supports our claim @xcite . the well - sampled nature of our observations allows us to determine an unambiguous time of peak brightness in i and j. this parameter will be important for determining the amount of mass ejected in the supernova explosion , thought this modeling is beyond the scope of this paper .
the position of peak brightness was determined by fitting second order cubic splines to the data points , with errors derived from the formal chi - squared error on the fits in combination with the error on the measured magnitudes .
the combination of the host galaxy and sn reaches a peak apparent magnitude in i of @xmath6 mag after @xmath7 days and in j of @xmath8 mag after @xmath9 days .
the rest - frame time to peak is , therefore , approximately 12.7 days in i and 17.0 days in j. the galactic extinction correction along the line of sight to the host galaxy is taken to be @xmath10 mag and @xmath11 mag , assuming the galactic extinction curves of @xcite and a measured reddening value of e(b - v)=0.127 mag @xcite .
the pre - burst sdss model magnitude of the host galaxy is @xmath12 mag , not corrected for galactic extinction @xcite . using the transformation equations derived by @xcite ,
this corresponds to @xmath13 mag .
the peak absolute magnitude of the supernova is , therefore , @xmath14 mag .
no k - correction has been applied , but this should result in minimal error due to the low redshift of the burst . a correction of -0.04 magnitudes was applied to account for spectral stretching .
the exact pre - burst j magnitude of the host is unknown as the host galaxy is too dim to appear in the 2mass catalog .
our observations indicate the host galaxy must have a j magnitude @xmath15 . assuming a range of host magnitudes from 18 mag to 20 mag , the peak absolute magnitude of the supernova in j is approximately @xmath16 mag .
note that 2006aj clearly peaks later in j than in i. this later peak at redder wavelengths follows the trend seen in sn 1998bw , which , in the rest - frame , peaked 1.6 days earlier in v than in i. the rest - frame v - band peak of sn 2006aj occurred at approximately 9.7 days @xcite , which is 2.9 days prior to the i - band peak .
likewise , sn 2003lw peaked in v at @xmath17 days @xcite and in i at @xmath18 days @xcite .
the combined light of the galaxy and the sn reddens from i - j = 0.0 mag during the first week to i - j = 0.6 mag for the last few observations .
this is a stronger evolution in i - j color than experienced by either sn 1998bw or sn 2003lw , whose i - j colors in the first month only change by about 0.3 magnitudes @xcite .
this comparison is complicated , however , by the unknown intrinsic i - j color of the host galaxy of sn 2006aj .
our data , together with those of @xcite , @xcite and @xcite , show that grb 060218 is the third detected grb to be followed by a dominant sn .
with such a small sample being used to extrapolate the characteristics for an entire category , it is important to collect homogeneous and detailed observations over a wide range of wavelengths .
our data provides dense ir coverage , which extends the total sn 2006aj dataset out to redder bands than reported thus far .
it is instructive to compare all three cases in which l@xmath3grbs have been detected , each of which was followed by a type ic sn : 1998bw , 2003lw and 2006aj ( see figure 2 ) .
we note that the limit on observing l@xmath3grb events are more stringent than those on observing the associated sne , so it is unlikely that grbs for which no optical counterpart are observed are of this character .
the properties of these three events are shown in table 2 .
all three sne are very similar in peak brightness , though 2003lw may be half a magnitude brighter than the others .
the biggest difference between the bursts is their rise times , with 2006aj peaking the fastest and 2003lw taking the longest time to peak .
the rest - frame photon energy at which the grb spectrum peaks ( @xmath19 ) appears to increase with increasing sn rise time . as discussed in the introduction
, it would be instructive to compare the frequency of these l@xmath3 bursts with that of high - luminosity bursts using the homogeneous _ swift _ dataset .
such a comparison is now possible since grb 060218 is the first _
swift_-detected burst that falls in the category of low - luminosity grbs associated with type ic sne .
the low observed fluxes and redshifts of l@xmath3grbs suggest that the underlying rate of l@xmath3grb events may be quite high ( see also * ? ? ?
* ; * ? ? ?
* ) , since the volume in which these sources can be observed is much smaller than that of typical grbs .
the ratio of the event rate of l@xmath3grbs to ordinary long - duration grbs is expected to be @xmath20 where @xmath21 denotes the true rate per comoving volume of the two kinds of events , @xmath22 is the observed event rate seen by a given experiment , and @xmath23 is the comoving radial distance out to which the events could be observed by that experiment . in the case of
_ swift _ , the limiting volume in which l@xmath3grbs are observable is so small that the observation of even one of these events implies an extremely high ratio of true rates .
specifically , grb 060218 would not have triggered the bat event monitor if it had been @xmath24 times fainter , which corresponds to a maximum redshift of @xmath25 .
by contrast , the 31 high - luminosity , long - duration _ swift _ grbs that have measured redshifts have an average redshift of @xmath26 . many of these bursts would have been detected even if they had occurred at @xmath27 .
we will take a conservative approach , however , and assume that most of these bursts would not have been detected had they been at extreme redshifts and adopt @xmath26 as the redshift below which the high - luminosity grb samples are complete .
assuming the concordance cosmology of @xmath28 , @xmath29 and a hubble constant of 71 km s@xmath30 mpc@xmath30 , the ratio of the comoving radial distance cubed for these two events is @xmath31 . assuming that the ratio of rates is equal to the number of long - duration grb events observed to date ( 1 l@xmath3grb per about 100 grbs ) , we find a true event ratio of @xmath32 . in the near future , thanks to _ swift _ , the cosmological grb sample may be complete out to at least a redshift of @xmath33 . for this volume , the true event ratio increases by a factor of @xmath24 .
if the current highest grb redshift measurement , @xmath34 , is representative of the redshift out to which we have a complete sample , then the true event ratio is @xmath35 .
we also note that we are assuming a constant grb rate per unit time .
if , instead , the rate scales with the star formation rate ( e.g. * ? ? ?
* ) , then the relevant ratio would be higher still .
these event ratios provide a constraint on the origins of l@xmath3grbs .
one explanation of these events is that they are standard grbs observed off - axis ( e.g. * ? ? ?
* ; * ? ? ?
this is an attractive option as it accounts for all long - duration grbs using a `` unified model '' , with observed differences attributed only to viewing angle .
however , this scenario implies a maximum true rate ratio , which would be generated if the l@xmath3grbs could be observed from any angle .
this upper limit is @xmath36 , where @xmath37 is the jet opening solid angle of a typical grb ( in steradians ) .
jet breaks observed in x - ray / optical ag light curves constrain the jet opening angle to @xmath38 ( e.g. * ? ? ?
* ) , so we infer a maximum true rate ratio of 65 in this model .
this ratio is a factor of five lower than the ratio of @xmath32 , which we inferred for @xmath26 , and lower by @xmath39 than for @xmath34 .
since _ swift _ has observed only one l@xmath3grb , the underlying event rate remains uncertain ; grb 060128/sn 2006aj could have been a serendipitous event . from poisson statistics ,
the 90% lower confidence limit of the true event rate is an order of magnitude lower than assumed above , which implies that the off - axis scenario is still acceptable , provided one assumes a limit of @xmath26 and a constant grb rate . if _ swift _ detected a second l@xmath3grb , the off - axis scenario would become significantly less probable . such a poissonian analysis addresses the question `` given an event rate , what is the probability of seeing an event '' , whereas in this case one might more appropriately ask the bayesian question `` having seen an event , what is the probability of a given event rate '' . assuming a uniform prior on the distribution of event rates ( an assumption for which there is no real basis ) , we find that the off - axis scenario is implausible at the 98% level .
of course , the undetermined luminosity function of grbs could complicate this calculation , as could cosmic evolution of the grb source population . at face value
, however , the above calculation of the event rate suggests that there is a category of grb events that is intrinsically different from that of typical grbs .
several suggestions have been made for how these differences can be accounted for including the possibility that the gamma - rays are produced in supernova shock breakout @xcite or `` failed collapsars '' in which highly relativistic jets fail to develop due to baryon loading @xcite .
the orientation - corrected energies of grbs have been claimed to be constant at @xmath40 ergs @xcite .
however , if intrinsically low - energy grbs exist as a separate population , efforts to use grbs as standard candles ( e.g. * ? ? ?
* ; * ? ? ?
* ) may be compromised .
we thank smarts observers d. gonzalez and j. espinoza for their dedication and s. tourtellotte for assistance with optical data reduction .
we greatly appreciate discussions with a. cantrell and j. emerson .
this work is supported by nsf graduate fellowship dge0202738 to bec and nsf / ast grant 0407063 and _ swift _ grant nng05gm63 g to cdb .
adelman - mccarthy , j. k. et al . 2006 , , in press amati , l. 2006 , ( astro - ph/0601553 ) amati , l. et al .
2006 , grb circular network , 4846 barthelmy , s. et al .
2006 , grb circular network , 4806 bloom , j. s. , et al . 1999 , , 401 , 453 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 cobb , b. e. et al .
2004 , , 608 , l93 cobb , b. e. , bailyn , c. d. 2006 , grb circular network , 4837 cusumano , g. et al .
2006 , grb circular network , 4775 davanzo , p. et al .
2006 , grb circular network , 4810 della valle , m. , et al .
2003 , , 406 , l33 frail , d. a. , et al .
2001 , , 562 , l55 galama , t. j. , et al .
1998 , , 395 , 670 gal - yam , a. , et al .
2004 , , 609 , l59 ghirlanda , g. , et al .
2004 , , 613 , l13 guenther , e. w. et al .
2006 , grb circular network , 4863 landolt , a. u. 1992 , , 104 , 340 lupton , r. 2005 , http://www.sdss.org/dr4/ lazzati , d. et al .
2006 , ( astro - ph/0602216 ) malesani , d. , et al .
2004 , , 609 , l5 marshall , f. , et al .
2006 , grb circular network , 4800 masetti , n. et al .
2006 , grb circular network , 4803 matzner , c. d. , & mckee , c. f. 1999 , , 510 , 379 mirabal , n. et al . 2006 , ( astro - ph/0603686 ) mirabal , n. , & halpern , j. p. 2006 , grb circular network , 4792 modjaz , m. et al .
2006 , ( astro - ph/0603377 ) natarajan , p. et al .
2005 , , 364 , l8 ovaldsen , j. et al . 2006 , grb circular network , 4816 patat , f. , et al .
2001 , , 555 , 900 pian , e. , et al .
2000 , , 536 , 778 pian , e. , et al . 2006 , ( astro - ph/0603530 ) prochaska , j. x. , et al .
2004 , , 611 , 200 ramirez - ruiz , e. et al .
2005 , 625 , l91 sakamoto , t. et al .
2006 , grb circular network , 4822 sazonov , s. y. , lutovinov , a. a. , & sunyaev , r. a. 2004 , , 430 , 646 skrutskie , m. f. , et al . 2006 , , 131 , 1163 sollerman , j. et al .
2006 , ( astro - ph/0603495 ) soderberg , a. m. , et al . 2004 , , 430 , 648 stanek , k. z. , et al .
2003 , , 591 , l17 tagliaferri , g. , et al .
2004 , grb circular network , 2545 tagliaferri , g. , et al .
2005 , , 443 , l1 tan , j. c. , matzner , c. d. , & mckee , c. f. 2001 , , 551 , 946 thomsen , b. , et al .
2004 , , 419 , l21 van dokkum , p. g. 2001 , , 113 , 1420 woosley , s. e. , & macfadyen , a. i. 1999 , , 138 , 499 yamazaki , r. , yonetoku , d. , & nakamura , t. 2003 , , 594 , l79 lcccc redshift & 0.0085 & 0.1055 & 0.033 + fluence ( @xmath47 erg cm@xmath30 ) & @xmath49 & @xmath50 & @xmath51 + total duration ( s ) & @xmath52 & @xmath52 & @xmath53 + @xmath19 ( kev ) & @xmath54 & @xmath55 & @xmath56 + @xmath57 ( @xmath58 erg ) & @xmath59 & @xmath60 & @xmath61 + i - band t@xmath62 ( days ) & @xmath63 & @xmath64 & @xmath65 + peak @xmath66 ( mag ) & @xmath67 & @xmath68 to @xmath69 & @xmath70 + i - j sn color , @xmath71t@xmath72 & 0.5 & @xmath73 & @xmath74 + = 0.39 mag . for clarity ,
the i - band points have been shifted by + 0.1 magnitudes .
error bars are photometric measurement errors and do not include possible systematic effects .
the curves are fit with second order cubic splines . _
bottom panel : _ i - j color evolution , the combined light is observed to redden with time.,scaledwidth=100.0% ] | we present smarts consortium optical / ir light curves of sn 2006aj associated with grb 060218 .
we find that this event is broadly similar to two previously observed events sn 1998bw / grb 980425 and sn 2003lw / grb 031203 .
in particular , all of these events are greatly under - luminous in gamma - rays compared to typical long - duration grbs . we find that the observation by _
swift _ of even one such event implies a large enough true event rate to create difficulties in interpreting these events as typical grbs observed off - axis .
thus these events appear to be intrinsically different from and much more common than high - luminosity grbs , which have been observed in large numbers out to a redshift of at least 6.3 .
the existence of a range of intrinsic energies of grbs may present challenges to using grbs as standard candles . |
though the fundamental components of all physical systems follow the laws of quantum mechanics , the objects in our everyday reality obey newton s laws .
clearly , in composite objects , the quantum interactions of numerous particles somehow conspire to produce a `` whole '' that obeys newtonian physics .
unfortunately , quantum theory does not provide a natural or obvious framework for the derivation of that classical reality , or for indicating the border between quantum and classical behavior . in the early days of quantum theory
the demarcation between quantum and classical behavior was often placed , at least approximately , at the boundary between the microscopic and macroscopic . but this could nt be the ultimate answer , for no object is truly macroscopic in the sense that it is a single particle of macroscopic dimension and weight . instead , to call something macroscopic is to approximate it by ignoring the internal states of its microscopic components , thereby essentially reducing the problem to center - of - mass variables .
in the past few decades , the failures of the macro / micro distinction have also been highlighted by new experimental results .
for example , a nanometer - scale superconducting electrode connected to a reservoir via a josephson junction constitutes a macroscopic two - level electronic system in which the two charge states can be coherently superposed @xcite .
more recently , nanomechanical resonators coupled to electrical circuits systems of billions of atoms have been measured to be oscillating in their quantum ground state @xcite .
but the grandest demand for a seamless approach to the quantum / classical divide comes from cosmology : since the universe has no place for an external observer , a theory of quantum cosmology must include classical physics as a property that emerges from purely quantum laws @xcite .
the mechanism by which the classical world emerges from a set of quantum particles is called decoherence .
one way systems decohere is through interactions with an external system , the environment , which typically consists of a large number of degrees of freedom , and is not subject to measurement .
the coupling means that the state of the environment will be sensitive to the state of the system , and will in a sense , `` observe '' the system , leading to a destruction of coherence .
mathematically , this occurs because our lack of knowledge about the environment requires that we trace over the environmental degrees of freedom , leading to an evolution of the observed system in which the density matrix is reduced to a classical probability distribution .
there are many environmental mechanisms that can cause this .
for example , the molecules in a beam can collide with lighter particles @xcite , or interact with thermal radiation @xcite . while much work has been done to examine the decoherence of systems due to external factors ( see , for example , @xcite ) as cosmologists have pointed out
, we need not turn to external systems as the cause of the emergence of classical behavior . such behavior can arise from the make - up of the object itself that is , from the object s own internal states , which can act essentially as an `` external environment '' that travels with the object , allowing the body s center - of - mass motion to be approximated by the classical equations of motion @xcite . in this paper
we investigate the mechanism for this transition to classical behavior through the influence - functional methods of feynman and vernon @xcite .
that method is tailor - made for this analysis because it allows us to integrate out the internal degrees of freedom , leaving an expression for the center - of - mass coordinates alone .
we will then illustrate the effect by considering a specific scattering problem .
we find that when the scattered particles possess internal states , as the entanglement between internal state and environment grows , the interference diminishes .
for simplicity of analysis , we will consider a composite system with only a single internal degree of freedom . obviously , more complicated systems with many internal degrees of freedom will likely exhibit much stronger decoherence effects , but analyzing even a single internal degree of freedom is both revealing and fairly challenging .
consider a two - particle system in an external potential @xmath0 , and subject to an internal potential @xmath1 .
assume both particles have mass @xmath2 .
we can change variables to separate the center - of - mass and internal degrees of freedom : @xmath3 or inverting the relationship , @xmath4 in terms of these new variables , the kinetic energy is @xmath5 where @xmath6 is the mass of the composite system . assuming that @xmath1 has the form @xmath7
, the potential energy is @xmath8 the lagrangian in terms of @xmath9 and @xmath10 is thus @xmath11 the propagator for this system is given by the path integral @xmath12 where @xmath13 is @xmath14 if @xmath15 the composite - particle propagator in ( [ eq : propagator ] ) would represent the propagator for a single particle of mass @xmath6 and `` charge '' 2 , i.e. , moving in a potential @xmath16 .
this is a good approximation when the contributions from the path integral in ( [ eq : internalpropagator ] ) are dominated by the path @xmath17 . in general , though this may not hold , the paths that are important in ( [ eq : internalpropagator ] ) are nevertheless of order @xmath18 , the `` size '' of the composite system .
if the external potential , @xmath19 , varies slowly over such distances , this suggests approximating ( [ eq : internalpropagator ] ) employing a taylor expansion around @xmath20 : @xmath21 the @xmath16 term is just a potential acting on the center - of - mass degree of freedom , and can be absorbed into the action for the center - of - mass .
the second term has the form of parametric driving on the internal degree of freedom .
this gives the following approximation for the propagator in ( [ eq : propagator ] ) : @xmath22 we will make this approximation in the following section , in which we consider the effective transition probabilities for center - of - mass states , assuming that the final state of the internal degree of freedom goes unobserved . note that , since the harmonic force driving the internal state variable depends on the path in the center - of - mass path integral , there is no equivalent approximation in the hamiltonian formalism . from the propagator we can derive the probability for the degrees of freedom to make a transition from an initial product wave function @xmath23 to a final product wave function @xmath24 : @xmath25 we fix the initial state of the internal degree of freedom and obtain an effective transition probability for the center - of - mass state alone by summing over ( i.e. , tracing out ) the final state of the internal degree of freedom : @xmath26 , \nonumber \label{eq : transitionprob}\end{aligned}\ ] ] where @xmath27 $ ] is the _ influence functional _ : @xmath28 & = & \int\int\int dy_f dy_i dy'_i \psi^*_i(y'_i ) \psi_i(y_i ) \nonumber\\ & & \times \tilde{k}^*[y'(t ) ; y_f , t_t ; y'_i , t_i ) \\ & & \times \tilde{k}[y(t ) ; y_f , t_t ; y_i , t_i ) .
\nonumber \label{eq : influencefunctional}\end{aligned}\ ] ] this integral has an interesting interpretation .
the center - of - mass trajectories in the path integral for the propagator of the full system ( [ eq : propagator ] ) appear here as fixed functions of time , @xmath20 and @xmath29 .
if we consider these to be classical driving functions , then we can define two wave functions @xmath30 these are the wave functions of the internal degree of freedom that would result if that degree of freedom began in the state @xmath31 , and was then `` classically driven '' by a potential that depends on the center - of - mass trajectories @xmath20 and @xmath29 , respectively , namely : @xmath32 and @xmath33 the influence functional is the scalar product of those two wave functions at time @xmath34 : @xmath35 = \bracket{\tilde{\psi}_y}{\tilde{\psi}_{y ' } } .\ ] ] though we have a simple physical interpretation , evaluating the path integrals in ( [ eq : transitionprob ] ) for general potentials is very difficult .
but we can make more progress , and see a simpler physical interpretation , if we use the approximation we alluded to above : @xmath36 we can now more cleanly separate the terms that depend only on @xmath20 and @xmath29 from those that depend on on @xmath37 and @xmath38 , giving us the following approximate expression for the influence functional : @xmath39 \nonumber\\ & = & e^{-\frac{2i}{\hbar } \int_{t_i}^{t_f } ( v(y(t))-v(y'(t ) ) ) dt } \nonumber\\ & & \times \int dy_f \left [ \int dy'_i k^*[y'(t ) ; y_f , t_f;y'_i , t_i ) \psi^*_i(y'_i ) \right ] \nonumber\\ & & \times \left [ \int dy_i k[y(t ) ; y_f , t_f;y_i , t_i ) \psi_i(y_i ) \right ] \nonumber\\ & = & e^{-\frac{2i}{\hbar } \int_{t_i}^{t_f } ( v(y(t))-v(y'(t ) ) ) dt}\bracket{{\psi}_y}{{\psi}_{y ' } } , \end{aligned}\ ] ] here , @xmath40 is the propagator corresponding to the time - dependent ( and center - of - mass - path - dependent ) hamiltonian @xmath41 and @xmath42 is the wave function of the internal degree of freedom at time @xmath34 after evolving via that hamiltonian from the initial wave function @xmath43 . employing ( [ eq : propagatortwo ] ) we obtain the following approximation for ( [ eq : transitionprob ] ) : @xmath44 this expression shows clearly that the effect of the internal states is to modulate the integrand for the center - of - mass motion through the factor @xmath45 .
when @xmath46 , the compound particle acts as if it is not composite , but rather a single particle with mass equal to the sum of the component masses .
the degree to which @xmath45 deviates from unity determines the `` degree of compositeness '' of the compound particle .
we expect decoherence effects such as the suppression of interference when @xmath47 for @xmath48 .
this will depend on both the internal potential @xmath49 and the coupling term @xmath50 . if @xmath19 is a quadratic potential for instance , a harmonic oscillator
then this term will be constant and there will be no coupling between the internal and external degrees of freedom . therefore decoherence should only occur if the external potential @xmath19 is not harmonic .
however , even in that case , the effects will also depend on the _ internal _ potential , as well as the state of the internal degree of freedom .
we now specialize to the case where the internal potential is harmonic .
as mentioned after eq .
( [ eq : transitionprobapprox ] ) , a natural question arises as to what degree a compound system can be treated as a single particle , rather than a composite system
. it would be useful to have a measure of `` compositeness , '' which would indicate to what degree we can ignore the internal degrees of freedom of the system . to put this another way , we would like to ask to what extent we can describe the behavior of the center - of - mass degree of freedom while neglecting the internal degree ( or degrees ) of freedom . consider a composite system in an initial product state @xmath51 the hamiltonian of the joint system is given by @xmath52 the term @xmath53 couples the internal and center - of - mass degrees of freedom .
how much does this influence the evolution of the center - of - mass ?
this will in general depend both on the particular system ( the choice of potentials @xmath19 and @xmath49 ) and the states of the internal and external degrees of freedom .
we can define a `` decoupled '' hamiltonian @xmath54 which omits the coupling term .
we can then compare the alternative evolutions with and without the coupling , by looking at how the overlap between the states changes for the two evolutions : @xmath55 the rate of change of the overlap is given by @xmath56 specializing to @xmath57 , we get @xmath58 the magnitude of this rate has some properties that one would want in a measure of compositeness , but it is not perfect . in particular , the rate of change of the overlap does not distinguish between physically significant changes in the state due to coupling , and physically meaningless changes in the global phase .
it also treats the internal and center - of - mass degrees of freedom symmetrically , while we are more interested in the question of how well the center - of - mass degree of freedom is approximated as a single indivisible particle .
this overlap does suggest a related approach , however , which can overcome these difficulties .
let us define the state in the _ interaction picture _ : @xmath59 we can then define the reduced density matrix for the center - of - mass in the interaction picture by taking a partial trace over the internal degrees of freedom : @xmath60 to solve for the evolution of @xmath61 or @xmath62 in general requires the ability to solve the schrdinger equation .
but at @xmath57 with an initial product state , the rate of change of @xmath63 simplifies : @xmath64 , \ ] ] where @xmath65 is an effective hamiltonian : @xmath66 this evolution will produce a nontrivial change in @xmath67 if and only if @xmath68 is not an eigenstate of @xmath65 . we can measure this by looking at the variance : @xmath69 this quantity @xmath70 is real and nonnegative , and is nonzero if and only if there is nontrivial coupling between the internal and center - of - mass degrees of freedom .
the value of @xmath70 depends on the nature of the potential @xmath19 , and on the states @xmath68 and @xmath71 .
the factor of @xmath72 means that this quantity does nt have a simple maximum ( and the approximate expansion of @xmath73 to second order will not be valid if @xmath74 is too large ) , but it should have a minimum , which may or may not be nonzero , depending on the system .
it also suggests that the states which will show large effects of compositeness have both @xmath74 and @xmath75 large , and are not eigenstates of @xmath75 . choosing @xmath70 to be the square root of the variance has a suggestive physical interpretation .
since the operator @xmath65 is essentially an effective interaction hamiltonian , we can define an energy - time uncertainty principle relating the variance of @xmath65 and the timescale @xmath76 of a state transition : @xmath77 so we can interpret @xmath70 as proportional to the rate at which the interaction between the center - of - mass and internal degrees of freedom will cause the center - of - mass state to become orthogonal to the center - of - mass state _ without _ interaction .
we expect this measure to be nonzero whenever interaction with the internal degrees of freedom causes decoherence ; but it may be nonzero even in cases with no decoherence , if the composite structure produces a change in the center - of - mass evolution . the dependence of @xmath65 on the second derivative of the potential is at first a bit surprising .
( the second derivative is simply related to a quantity in mechanics known as the _ jerk _ or _
jolt_. ) but a bit of reflection reveals that the leading order contribution to the coupling between the internal and center - of - mass degrees of freedom actually depends on the _ third _ derivative . to see that ,
suppose we expand the potential @xmath78 about some point @xmath79 .
we can see how different terms in this expansion will affect the internal and external degrees of freedom : * there is a set of terms involving only @xmath9 : @xmath80 .
these produce no coupling between the internal and center - of - mass degrees of freedom . *
all terms that are linear in @xmath10 cancel out in the expansion , as do all terms of odd power in y. * there is a term @xmath81 that will produce a harmonic potential for the internal degree of freedom , but which is not coupled to the center of mass .
( if the internal potential @xmath49 is also harmonic , this is just a constant frequency shift . ) * the lowest order term including both the internal and center - of - mass variables is @xmath82 .
this term will produce a parametric driving of the internal degree of freedom dependent on the center - of - mass position , and a force on the center - of - mass that depends on the internal degree of freedom .
we see from this that to produce nontrivial coupling between the internal and center - of - mass degrees of freedom requires a nonvanishing third derivative . without that
, the coupling produces at most a constant frequency shift , and can affect only the global phase of the center - of - mass state .
the proposed measure @xmath70 is nonzero precisely when such nontrivial couplings are present .
a number of straightforward variants of this measure are possible .
for instance , we could relax the restriction to product states , or to pure states .
we could also look at the change over a finite time interval , rather than the instantaneous rate of change .
these differences would entail more computational complexity , but are worth exploring in future work .
suppose the internal potential is harmonic , which is a good approximation for many simple molecules . @xmath83
the propagator becomes @xmath84 for the purpose of evaluating the path integral over the internal degree of freedom , let us _
assume that the center - of - mass trajectory _
is fixed_. we can then define the time - dependent internal potential @xmath85 where @xmath86 first let us set @xmath87 and @xmath88 .
let us make a change of variables : @xmath89 in terms of these new variables the effective hamiltonian corresponding to the internal degree of freedom in ( [ eq : propagatorthree ] ) : @xmath90 where @xmath91 employing this hamiltonian , for each trajectory in the center - of - mass path integral in ( [ eq : propagatorthree ] ) , we can reduce the problem of determining the dynamics of the internal degree of freedom , @xmath10 , to the solution of a single ordinary ( nonlinear ) differential equation . to solve for the dynamics ,
we begin by noting that operators can be put in normal - ordered form @xcite : @xmath92 where the coefficients @xmath93 are complex numbers .
every such operator is therefore in one - to - one correspondence with a complex function @xmath94 , where @xmath95 we can now use the fact that the function @xmath94 determines the operator @xmath96 to solve for the unitary propagator of the parametrically - driven oscillator .
let @xmath97 be the unitary propagator from the initial time to time @xmath98 .
it satisfies the operator schrdinger equation : @xmath99 u(t ) .
\nonumber\end{aligned}\ ] ] we can write @xmath100 in normal - ordered form , @xmath101 this function obeys a differential equation : @xmath102 u(t)\ket{\alpha } .
\nonumber\end{aligned}\ ] ] employing the simple results : @xmath103 @xmath104 the differential equation for @xmath105 becomes @xmath106 \bar{u}(\alpha,\alpha^*,t ) .
\nonumber \label{eq : ubardiffeq}\end{aligned}\ ] ] we make an ansatz for the solution : @xmath107 where @xmath108 if we can solve for the functions of time in ( [ eq : ansatz ] ) , we can easily recover the effective propagator for the internal states : @xmath109 note that one might expect also terms proportional to @xmath110 and @xmath111 but these terms can be chosen to be identically zero . at @xmath57
we want @xmath112 , which gives us initial conditions : @xmath113 plugging the above expression into the differential equation ( [ eq : ubardiffeq ] ) yields the following differential equations for the functions @xmath114 , @xmath115 , @xmath116 and @xmath117 : @xmath118 ( d(t ) + 1 ) , \\
i \frac{de}{dt } & = & \beta(t ) + 2 \omega(t ) e(t ) + 4\beta(t ) e^2(t ) , \nonumber\\ i \frac{df}{dt } & = & \beta(t ) ( d(t ) + 1)^2 .
\nonumber\end{aligned}\ ] ] looking at the structure of these equations , we see that one can solve for @xmath114 , @xmath115 and @xmath117 in terms of @xmath116 : @xmath119 @xmath120 @xmath121 the equation for @xmath116 is self - contained , and must be solved before plugging it into these expressions : @xmath122 for certain choices of @xmath123 this equation may be solvable in closed form ; but generically , this equation may only be solvable numerically
. this would be sufficient to allow the solution of the influence functional ( [ eq : propagatortwo ] ) for particular choices of @xmath20 and @xmath29 .
it is not practical to evaluate the path integrals in ( [ eq : transitionprobapprox ] ) in this way , however it is often the case that a path integral is dominated by the classical path @xcite . in that case , by replacing @xmath20 and @xmath29 in @xmath45 by @xmath124 and @xmath125 , the leading order to ( [ eq : transitionprobapprox ] ) could be calculated .
we can see from the above results that coupling to internal degrees of freedom should generically lead to suppression of interference effects , just like other kinds of decoherence .
it is hard to estimate the magnitude of this effect analytically , however . in this section ,
we give a numerical analysis of interference in a simple scattering process , and show that even a single internal degree of freedom can significantly suppress interference .
suppose two wave packets , initially separated from each other , approach a potential of compact support , such as a square well , from opposite directions .
each wave packet , taken alone , would produce both a reflected and transmitted wave . by linearity , when we start with the superposition of the two , each will produce such waves , and they will interfere in some manner , depending on the initial conditions .
this interference is a strictly quantum effect with no classical counterpart .
we show below that , when the scattered particles possess internal states , as the entanglement between internal state and environment grows , the interference diminishes .
consider a particle moving in one dimension with a potential @xmath126 which is zero everywhere outside of a compact region of size
l : @xmath127 . for simplicity
, we will assume a simple square well potential : @xmath128 where @xmath129 .
a narrow wave packet incident on this potential will generically scatter into two transmitted and reflected wave packets .
classically , incident particles would not be reflected from such a well ; such reflection is already a quantum - mechanical effect .
suppose we have a particle of mass @xmath130 incident from the left in a plane wave state with momentum @xmath131 ; then it is straightforward to calculate the norms of the reflection and transmission coefficients @xmath132 and @xmath133 : @xmath134 by choosing an appropriate set of values @xmath131 , @xmath18 and @xmath135 it is possible to have this square well act like a 50/50 beam splitter for wave - packets centered around momentum @xmath131 .
if we choose @xmath136 and @xmath137 in dimensionless units , this 50/50 reflection / transmission occurs for @xmath138 .
we can then model a situation like the second half of a michelson interferometer : a superposition of two identical wave packets incident from the right and left with the same magnitude of @xmath139 can interfere almost perfectly , so that a single wavefunction emerges either to the right or to the left ( depending on the relative phase of the two incident wave packets ) .
now suppose that instead of a single particle of mass @xmath130 we have a composite particle , comprising two particles of mass @xmath140 bound by a harmonic potential . because the potential is piecewise constant , @xmath141 everywhere except at @xmath142 .
so from the results of the previous section , the internal and center - of - mass degrees of freedom are decoupled everywhere except at those two points , where there is an impulsive coupling between them .
if we prepare the center - of - mass degree of freedom in a narrow wave packet and scatter it off of this potential well , entanglement can be produced between the internal and center - of - mass degrees of freedom .
whether entanglement is in fact produced depends on the initial state of the internal degree of freedom . in a classical harmonic oscillator , the strength of parametric driving depends on the initial amplitude of oscillation .
an oscillator in equilibrium experiences no driving .
similarly , if the internal degree of freedom begins in the ground state , the parametric driving from the center - of - mass degree of freedom will be ineffective in coupling the two degrees of freedom .
if the internal degree of freedom begins in an excited state , however , it will be more strongly coupled to the center - of - mass degree of freedom .
we have numerically simulated the interference experiment described in the previous subsection for two coupled particles with a variety of initial states
. the center - of - mass degree of freedom begins in the superposition state @xmath143 and the internal degree of freedom begins in the state @xmath144 where @xmath145 , @xmath146 , @xmath136 , @xmath137 , and @xmath147 .
the parameter @xmath148 is a relative phase between the two wave packets ; we choose this phase @xmath148 to maximize the interference between them .
we choose the internal coupling strength @xmath149 , and the width of the initial wave packet for the internal degree of freedom to be @xmath150 ; with this choice , the initial state @xmath151 is the ground state for @xmath152 , and an excited state for @xmath153 .
we have simulated this system from its initial state until after the scattering ( into two wave packets ) from the potential well for a range of values of @xmath154 .
decoherence generally has its effect by suppressing interference . in this case
, the interference is between the reflected and scattered components of the two incoming wave packets .
we choose the relative phase @xmath148 to maximize the probability after scattering that the scattered particle will be moving _
leftward_. ( of course , the situation would be completely symmetric if we chose to maximize the outgoing rightward component . )
since the state of this system is pure at all times , the proper measure of entanglement between the internal and center - of - mass degrees of freedom would be the _ entropy of entanglement _ between them .
however , this is a somewhat messy quantity to calculate for high - dimensional systems like those we are treating here .
instead , we use a simpler proxy quantity : the _ impurity _ of the reduced density matrix of the internal degree of freedom .
this is simply @xmath155 results of the simulations are plotted in figs .
[ fig : entanglement ] and [ fig : interference ] . fig .
[ fig : entanglement ] plots the impurity of the reduced density matrix for the internal degree of freedom after scattering as a function of its initial displacement . fig .
[ fig : interference ] plots the maximum leftward probability of the scattered particle .
we see that if the internal degree of freedom begins in the ground state , the parametric driving is ineffective in coupling the internal and center - of - mass degrees of freedom , so the two degrees of freedom remain unentangled and the interference is maximal .
this is the case for simple molecules like water , in which @xmath156 is about 18 times @xmath157 at room temperature @xcite .
as we increase the displacement of the initial state of the internal degree of freedom , the state of the two degrees of freedom becomes entangled , and the interference is reduced . increasing the displacement increases the entanglement and decreases the interference , up to a point where the two curves turn over and the entanglement diminishes .
this turn over is not too surprising .
coupling to a single degree of freedom can only modestly suppress interference , in general ; such small systems commonly exhibit recurrences and `` re - coherences . ''
models of environmental decoherence that produce monotonic decay of interference typically assume large heat baths with many degrees of freedom .
such an effect could probably be caused by a large composite system with many internal degrees of freedom , provided that the center - of - mass coupled strongly enough to these internal modes .
but the difficulties of solving such a system analytically are great , and they quickly become intractable numerically as well . however , even in this very simple system , we see a very close relationship between the degree of entanglement and the suppression of interference .
this relationship makes intuitive sense , based on earlier experience with external environments the state of the internal degree of freedom records `` which path '' information about the center - of - mass , specifically , information about whether it was transmitted or reflected by the potential well . what about the measure of compositeness defined in section [ sec : compositeness ] ?
we can calculate this numerically for the scattering system described in this section .
we use the same potential as in the simulations , and calculate @xmath158 where @xmath159 we choose states @xmath160 for the center - of - mass and @xmath161 for the internal degree of freedom of the same form as used in the numerical simulations , @xmath162 @xmath144 with the two wavepackets centered at @xmath79 and @xmath154 , respectively .
we can then calculate @xmath70 for different choices of @xmath79 and @xmath154 . for wavepacket centered at @xmath79 ( center - of - mass ) and @xmath163 ( internal ) .
the parameters in this plot are the same as in the numerical simulations.,width=288 ] for wavepacket centered at @xmath79 ( center - of - mass ) and @xmath163 ( internal ) .
the parameters in this plot are the same as in the numerical simulations . because the potential well is narrow compared to the width of the center - of - mass wavepacket , the composite system scatters from the potential as a whole.,width=288 ] figures [ fig : expectq ] and [ fig : delq2a ] show the expectation @xmath164 and the compositeness measure @xmath165 for the same parameters used in the numerical simulations above .
while in the expectation @xmath164 we can see the effects of the two potential well edges , the width of the wavepacket means that the composite system essentially scatters off of the potential well as a whole . from fig .
[ fig : delq2a ] we see that the evolution is significantly affected by the coupling between the internal and center - of - mass degrees of freedom throughout the scattering region . for wavepacket centered at @xmath79 ( center - of - mass ) and @xmath163 ( internal ) .
here we have used a narrower wavepacket @xmath166 and a wider potential well @xmath167 so that we can see coupling due to separate scatterings from the edges of the potential well , as well as from the potential as a whole.,width=288 ] for a narrow wave packet and/or a broader potential well , the coupling should reflect separate scatterings off of the two edges of the potential .
indeed , plotting @xmath70 for a case like that demonstrates the features of these two separate scatterings . in both these cases ,
we see that the coupling between internal and center - of - mass degrees of freedom leads to significant differences from the evolution of a single , indivisible particle .
as we have shown in this paper , it is possible for coupling between the center - of - mass and internal degrees of freedom in a composite object to give rise to decoherence .
we have calculated expressions for coupling to a single internal vibrational mode , and numerically demonstrated that this can lead to suppression of interference in a simple scattering experiment . just as in coupling to an external environment , a signature of this decoherence is the growth of entanglement between the center - of - mass and internal degrees of freedom . for this work ,
the overall state of the system was pure , so the impurity of the reduced density matrix for the internal degree of freedom served as an easy - to - calculate proxy for the entropy of entanglement
. however , there are certain important differences between decoherence due to internal degrees of freedom and due to an external environment .
the coupling between the internal degrees of freedom and the center - of - mass takes the form of a parametric driving term , and is dependent on the second derivative of the background potential .
therefore we would only expect to see such an effect when the background potential is suitably nontrivial , and only when the internal degrees of freedom are in an appropriate initial state . for internal degrees of freedom in an initial ground state ,
the effect of parametric driving is weak or nonexistent .
this raises the intriguing question of whether it would be possible to see this effect experimentally in a controllable fashion .
matter interferometry experiments have been done with molecules , including large molecules with many internal degrees of freedom
. it might be possible to vary the coupling between the internal and center - of - mass degrees of freedom by exciting one or more vibrational modes of the molecules before the interference experiment , and observing the effect on the visibility of the interference . for three - dimensional systems
, there can also be potential couplings with rotational degrees of freedom @xcite .
a number of interesting questions remain unanswered .
a single internal mode can not really act as a decohering bath .
it might be possible to derive effects for a composite particle with many internal degrees of freedom . in this paper
we have suggested one possible measure of compositeness that is , the degree to which we can approximate a composite system as a single particle without internal structure .
this quantity is based on the coupling between the internal and center - of - mass degrees of freedom , and how large an effect this has on the reduced state of the center - of - mass degrees of freedom . but a broader exploration of how well this measure correlates with ( for example ) suppression of interference by internal degrees of freedom has yet to be made .
this measure could also be varied in a number of ways .
these questions should be grounds for interesting future work .
lm and tab would like to acknowledges mark hillery , jos gonzalez , kung - chuang hsu , jan florjanczyk , and christopher cantwell for interesting and useful conversations .
they also acknowledge the support of caltech s institute for quantum information and matter ( iqim ) .
p. joyez , p. lafarge , a. filipe , d. esteve and m. h. devoret , _ observation of parity - induced suppression of josephson tunneling in the superconducting single electron transistor _ , phys .
lett . * 72 * , 24582461 ( 1994 ) .
j. chan , t. p. mayer alegre , a. h. safavi - naeini , j. t. hill , a. krause , s. grblacher , m. aspelmeyer and o. painter , _ laser cooling of a nanomechanical oscillator into its quantum ground state _ ,
nature * 478 * , 89 - 92 ( 2011 ) .
w. h. miller , _ the semiclassical initial value representation : a potentially practical way for adding quantum effects to classical molecular dynamics simulations _ , j. phys .
a * 105 * , 29422955 ( 2001 ) . | we consider a composite system consisting of coupled particles , and investigate decoherence due to coupling of the center - of - mass degree of freedom with the internal degrees of freedom . for a simple model of two bound particles ,
we show that in general such a decoherence effect exists , and leads to suppression of interference between different paths of the center - of - mass . for the special case of two harmonically - bound particles moving in an external potential in one dimension
, we show that the coupling between the center - of - mass and internal degrees of freedom can be approximated as parametric driving , and that nontrivial coupling depends on the second derivative of the external potential .
we find a partial solution to this parametric driving problem .
for a simple interference experiment , consisting of two wave packets scattering off of a square well , we perform numerical simulations and show a close connection between suppression of interference and entanglement between the center - of - mass and internal degrees of freedom .
we also propose a _ measure of compositeness _ which quantifies the extent to which a composite system can not be approximated as a single , indivisible particle .
we numerically calculate this quantity for our square well example system . |
polymer networks are unique soft solids which can be significantly deformed without irreversible damage to their structure . a network is formed by cross - linking a melt or a semidilute solution of polymer chains .
once a homogeneous ( on length scales large compared to its `` mesh '' size ) network is formed , one can generate large - scale patterns in it by further cross - linking , followed by swelling ( and possibly stretching ) of the network , resulting in a gel inhomogeously swollen by solvent .
this can be done , for example , by adding light - sensitive cross - links to a transparent network . focusing a laser beam in regions inside the gel one can write information into gel structure in the form of @xmath1 or @xmath2 patterns of cross - linking density . in this paper
we show that although such information is hidden at preparation conditions , it can be recovered by swelling the gel since unobservable variations of cross - link density in the melt are transformed into observable variations of monomer density in the swollen gel .
regions of a gel with increased cross - link concentration can be considered as inclusions with enhanced elastic modulus .
if such inclusions deform differently from polymer matrix , as in case of any normal elastic solids , they would induce elastic stresses in the gel and initial pattern would be significantly distorted due to long range character of elastic interactions .
this scenario determines , for example , the elastic properties of amorphous polycrystalline solids but it does not apply to polymer gels , because of the unusual character of gel elasticity .
we show that in swollen gels that are isotropically stretched by absorption of solvent , the observed _ monomer density pattern _ is not distorted and is simply an affinely stretched variant of the _ initial cross - linking pattern_. such gels can serve as a magnifying glass that enlarges the initially written pattern without distorting its shape .
the corresponding magnification factor can be very large in case of super - elasic networks .
in this paper we use the simplest mean field model of a gel with free energy @xcite:@xmath3
\right ) d\mathbf{x}_{0 } \label{ai}\ ] ] here @xmath4 is monomer density as function of coordinates @xmath5 in preparation state .
we assume that the gel was initially cross - linked in a polymer melt and then a pre - programmed pattern in cross - link concentration ( i.e. , a well - defined region of higher cross - link density compared to that of the surrounding network ) is created in the network using , say , a light - sensitive cross - linking technique ( the case of cross - linking in semi - dilute solution in good solvent is analyzed in si ) . here
@xmath6 is the polymer contribution to the elastic modulus of the cross - linked melt ( which is proportional to the local cross - link density)@xmath7 and @xmath8 represents the variations of cross - link density introduced by the second cross - linking step ( fig .
[ pattern]a ) .
pattern3d @xmath9 is the osmotic ( interaction ) part of the free energy of the gel , with monomer density @xmath10 .
@xmath11 is the deformation gradient tensor@xmath12 and @xmath13 are coordinates of deformed gel .
it is convenient to assume that the gel is deformed with respect to preparation state in two stages:@xmath14 thus , the gel is first stretched with respect to preparation state by factors @xmath15 along axes @xmath16 .
for such a deformation @xmath17 with components @xmath18 and we get @xmath19 where @xmath20 is the uniform monomer density in the undeformed state of preparation ( fig .
[ pattern]b ) .
notice that the coordinates @xmath13 describe a stretched network with inhomogeneous cross - link density but a _ homogeneous monomer density _
[ pattern]c and d ) .
even though such a homogeneous ( in monomer density ) state does not minimize the free energy and therefore is not an equilibrium state of the deformed gel , we use it as a reference state .
the true equilibrium state of the deformed network has an inhomogeneous monomer density profile and is defined by introducing a displacement field @xmath21 defined with respect to the above reference state : @xmath22 and we get gradient tensor and monomer density as function of coordinates @xmath13@xmath23 minimizing the free energy in eq .
( [ ai ] ) with respect to displacements @xmath24 at the preparation state ( all @xmath25 ) we conclude that in a melt the cross - links and the monomers will remain at their previous possitions and the elastic reference state will not change after relaxation .
we conclude that information about the pattern written on network structure is hidden in preparation state and can only be revealed after swelling .
in a swollen state the monomer density is small and the interaction energy can be expanded as @xmath26 , where @xmath27 is boltzmann constant , @xmath28 is temperature and @xmath29 is second virial coefficient . expanding the free energy in powers of @xmath24 and integrating over the volume of the undeformed network with measure @xmath30 we get@xmath31 \frac{d\mathbf{x}}{\prod_{i}\lambda _ { i}},\quad k_{\text{os}}=k_{b}tb\bar{\rho}^{2}. \label{a}\ ] ] the equilibrium deformation of the gel is found by minimizing this free energy .
its variation is@xmath32 and therefore , the minimum condition is@xmath33 we are interested only in variations of monomer density@xmath34 where @xmath35 is average density . taking the gradient of both sides of eq .
( [ gu ] ) we obtain an equation for the variations of monomer density@xmath36 where@xmath37 the solution of this equation@xmath38 \tilde{g}\left ( \lambda^{-1}\mathbf{\cdot x}\right ) \frac{d\mathbf{y}}{\prod_{j}\gamma_{j } } \label{rg}\ ] ] is expressed through green s functions of the laplace equation in 2 and 3 dimensions , respectively:@xmath39 in case of isotropically stretched / swollen gel with all @xmath40 the equilibrium monomer density depends on local cross - link concentration,@xmath41 we conclude that under isotropic deformation such as swelling , the monomer density produces an undistorted , uniformly stretched image of the pattern of cross - link density originally written on the homogeneous network ( compare figs .
[ pattern]a and e ) .
equilibrium displacement is expressed through the variation of monomer density , eq .
( [ dr ] ) , as@xmath42 we conclude that although density variations in isotropically deformed gels are strictly local , there is long - range strain field decaying as power law of the distance @xmath43 .
this strain induces a stress distribution in the gel , which can be observed by measuring the birefringence of transmitted light ( stress - optical law @xcite ) . in
anisotropically deformed networks the pattern is strongly distorted ( compare figs .
[ pattern]a and f ) and @xmath44 decays as power law of a distance @xmath45 from the localized cross - link density inhomogeneity @xmath46 . observe that variations of monomer density are largest along the direction of stretching .
this effect is closely related to the well known butterfly picture in contour plots of neutron scattering from random inhomogeneities of network structure in anisotropely deformed swollen gels @xcite .
in order to understand the difference between gels and normal solids we recall that the free energy of any solid is a functional of the nonlinear strain tensor @xmath47 @xcite,@xmath48 while the last term is usually neglected in the linear theory elasticity of solids because solids behave elastically only under small deformations , it can be shown that only this nonlinear term contributes to the elasticity of gels and that the elastic part of the free energy of gels ( eq .
( [ ai ] ) ) is linear in this nonlinear strain@xcite .
physically , the difference between elastic energy _ of a solid , _ which is a quadratic form in the linear strain and _ of a gel , _ which is linear in the nonlinear strain tensor , stems from the fact that while in solids there is a stress - free state of equilibrium ( crystal lattice ) that minimizes the energy of interaction between the atoms , the equilibrium state of gels is not stress - free .
polymer networks are made of entropic springs and , in the absence of osmotic pressure due to permeation by good solvent or due to excluded volume interactions in the melt state , such networks would collapse to the size of a single spring .
the finite length of entropic springs in the swollen gel is the result of osmotic pressure which can be replaced by equivalent isotropic stretching forces that act on the outer boundaries of the gel @xcite .
the difference between gels and solids becomes apparent when considering two simple toy models of heterogeneous gel and solid as two hookean springs with moduli @xmath49 and @xmath50 , connected in series as in fig .
[ toy ] : toy _ a ) gel model : _ osmotic pressure is represented by a force @xmath51 applied to free ends of the connected springs . in the presence of this force
the equilibrium lengths of the gaussian springs become @xmath52 and @xmath53 , and if we apply additional force @xmath54 , each of the springs will deform affinally with distance @xmath55 between the ends to which the force is applied ( boundaries of the system ) : @xmath56 _ b ) solid model : _ the springs of a solid have equilibrium lengths @xmath57 and @xmath58 in the stress - free state . during stretching due to force @xmath54
applied to the ends of the two - spring system , such a solid deforms nonaffinelly:@xmath59{c}r_{1}=r_{1}^{\text{eq}}+\left ( \lambda-1\right ) \left ( r_{1}^{\text{eq}}+r_{2}^{\text{eq}}\right ) \dfrac{k_{2}}{k_{1}+k_{2}},\\ r_{2}=r_{2}^{\text{eq}}+\left ( \lambda-1\right ) \left ( r_{1}^{\text{eq}}+r_{2}^{\text{eq}}\right ) \dfrac{k_{1}}{k_{1}+k_{2}}\end{array}\ ] ] with the soft spring ( @xmath60 ) stretched more than the rigid one .
these two simple toy models illustrate why under isotropic deformations , cross - linking density patterns in gels are stretched affinally , whereas soft regions in solids would undergo larger deformation compared to more rigid regions , thus distorting the original pattern .
we studied the combined effect of swelling and deformation on inhomogeneous networks , prepared by cross - linking a melt of polymer chains .
it is well - known that cross - link density heterogeneities that have no effect on the monomer density in the state of preparation ( a melt or a concentrated polymer solution ) , can be revealed by swelling the gel and observing the enhancement of light , x - ray and neutron scattering from the resulting monomer density inhomogeneities@xcite . in this paper we focused on a related phenomenon , namely that when large - scale cross - link density patterns are written into the network structure , the hidden image can be revealed by swelling and stretching the gel and observing the corresponding patterns of monomer density . using the mean field theory of elasticity of polymer gels we showed that stretching / swelling in good solvent acts as a magnifying glass : while isotropic stretching reproduces an enlarged but otherwise undistorted version of the original pattern , anisotropic stretching distorts this pattern , see figure [ pattern ] .
we compared these results with those obtained for ordinary elastic solids with inhomogeneous elastic moduli and found that in this case even isotropic deformations lead to distorted patterns .
we showed that the fundamental difference between response of inhomogeneous gels and solids to isotropic stretching can be traced back to the fact that unlike regular springs that have an equilibrium length even in the absence of stress , the equilibrium length of entropic springs is entirely determined by the osmotic forces that isotropically stretch the polymer gel .
finally , we would like to comment on the possibility of experimental verification and on possible applications of our results . in most application involving gels such as biomimetic sensors , actuators and artificial muscles @xcite
, macroscopically inhomogeneous ( layered ) gels undergo shape transitions when the thermodynamic conditions are changed or in response to application of external fields @xcite . in our case , the cross - link density pattern imprinted into the gel structure by , say , activation of light - sensitive cross - links , can be microscopic ( micron size ) and therefore would have little effect on the shape of the gel . upon swelling and/or isotropic stretching in good solvent , the magnified density pattern can be imaged on a light - sensitive screen .
the contrast can be significantly enhanced by stretching the entire gel in poor solvent or by focusing a laser beam on the localized pattern and heating it , resulting in local change of the quality of solvent .
finally , the sensitivity of the image to quality of solvent ( the distortion under anisotropic deformation disappears in @xmath0 solvent - see si ) can be useful for sensor devices .
* acknowledgments * yr s research was supported by the i - core program of the planning and budgeting committee and the israel science foundation , and by the us - israel binational science foundation . * supporting information . * in the si
we show that patterns obtained by cross - linking a semi - dilute polymer solution , deform affinely ( non - affinely ) under isotropic ( anisotropic ) deformation , just like in the case of cross - linking in the melt .
we then analyze how the pattern deforms under several different solvent conditions .
we show that the pattern always stretches affinely in a @xmath0-solvent , even under anisotropic deformations .
since the contrast between the high and the low monomer density regions can be significantly enhanced in a poor solvent we proceed to analyze the density profiles in gels that are isotropically stretched in mildly poor solvents ( at lower solubility , such stretched gels will undergo a transition into a strongly inhomogeneous state characterized by the appearance of dense filamentous structures @xcite ) .
we find that when the amplitude of cross - link density variations is sufficiently low , the image stretches affinely with the isotropic deformation but that for larger density contrasts the image becomes distorted , especially near the edges and corners of the pattern .
99 onuki a. , _ adv .
_ * 1993 * , _ 109 _ , 63121 . panyukov s. and rabin y. , _ macromolecules _ * 1996 * , _ 29 _ , 79607975 .
doi m. and edwards s.f . , _ the theory of polymer dynamics _ , clarendon : oxford , 1986 . bastide j. , leibler l. and prost j. , _ macromolecules _ * 1990 * , _ 23 _ , 18311837 .
landau l.d . and
lifshitz e.m . ,
_ theory of elasticity _ , pergamon : oxford , 1970 . panyukov s. and rabin y. , _ physics reports _ * 1996 * , _ 269 _ , 1131 . alexander s. , _ physics reports _ * 1998 * , _ 296 _ , 65236 .
flory p.j . and rehner j. , _ j. chem
. phys . _ * 1943 * , _ 11 _ , 521526 .
candau s. , bastide j. and delsanti m. , _ adv .
* 1982 * , _ 44 _ , 2771 .
horkay f. , hecht a .-
m . , mallam s. , geissler e. and rennie a.r . ,
_ macromolecules _ * 1991 * , _ 24 _ , 28962902 .
shibayama m. , tanaka t. and han c.c . ,
. phys _ * 1992 * , _ 97 _ , 68296841 .
bastide j. and candau s. , in _ the physical properties of polymeric gels _
cohen addad j.p . , wiley , new york , 1996 , p. 143 .
shahinpoor m. , bar - cohen y. , simpson j.o . and
smith j , _ smart mater .
struct . _ * 1998 * , _ 7 _ , r15r30 . klein y. , efrati e. and sharon e. , _ science _ * 2007 * , _ 315 _ , 11161120 .
peleg o. , kroger m. , hecht i. and rabin y. , _ europhys .
* 2007 * , _ 77 _ , 5800758012 .
* supplementary information to : *
consider a gel prepared in a good solvent at the monomer density @xmath20 that is swollen to density @xmath10 .
its free energy is the sum of elastic and osmotic contributions .
the osmotic pressure @xmath61 of the gel in a good solvent increases proportionally to the @xmath62 power of monomer density @xmath10 @xcite@xmath63 where @xmath27 is boltzmann constant , @xmath28 is temperature and @xmath64 is monomer size .
the osmotic part @xmath65 of the free energy per polymer chain between network junctions is proportional to the free energy density ( @xmath66 ) divided by the number of chains per unit volume ( @xmath67 ) , where @xmath68 is the chain degree of polymerization:@xmath69 the dimension of the chain along the main axis @xmath70 of deformation is @xmath71 , _ _
_ _ where @xmath72 is deformation factor along this axis ( defined as eigenvalue of local deformation gradient tensor @xmath11 ) and @xmath73 is the chain size in the state at which the gel was formed . the elastic free energy per chain is@xmath74 where @xmath75 is the amplitude of fluctuations of the chain in the deformed state . in a heterogeneous network
the direction of the triad of deformation axes @xmath70 depends on its position , and the sum of squares of local deformation factors in eq .
( [ si_el ] ) can be rewritten through the deformation gradient tensor @xmath11 as @xmath76 since the mean - square amplitude of chain fluctuations is proportional to the mean - square polymer size at semi - dilute good solvent conditions and scales with monomer density as@xcite @xmath77 , while the mean - square chain size in the preparation conditions scales as @xmath78 , the elastic free energy per chain is@xmath79 at the equilibrium swelling ( @xmath80 ) in the absence of additional deformations the total free energy of the gel per chain is:@xmath81 \label{si_aeq}\ ] ] and it is minimized at the density@xcite@xmath82 corresponding to maximum swelling ratio@xmath83 note that similar expression for @xmath84 is obtained in mean field model of a gel with second virial coefficient @xmath85 , see main text .
this conclusion can also be extended to our solution of the image storing problem . * * * * since both elastic ( eq . ( [ si_el1 ] ) ) and osmotic ( eq . ( [ si_osm ] ) )
terms in the gel free energy are multiplied by the same scaling factor @xmath86 such scaling renormalization does not change the results obtained for the mean field model .
in a @xmath0-solvent the second virial coefficient vanishes ( @xmath87 ) and equation @xmath38 \tilde{g}\left ( \lambda^{-1}\mathbf{\cdot x}\right ) \frac{d\mathbf{y}}{\prod_{j}\gamma_{j } } \label{si_rg}\ ] ] reproduces without distortion affinely stretched initial pattern @xmath88 even for anisotropically stretched gels ( small deviations from affinity are expected because of the non - vanishing third virial coefficient ) strong enhancement of monomer density contrast can be obtained by placing the gel ( with fixed boundaries otherwise it would collapse ) in a poor solvent with negative second virial coefficient @xmath89 . in case of very poor solvent with @xmath90 the gel becomes unstable with respect to formation of domains with different monomer density@xcite .
below we consider the case of poor solvent close to @xmath0-conditions with small @xmath89 and positive @xmath91 . at small @xmath92 the amplitude of density variations
@xmath93 can be significantly increased because of the small denominator in equation@xmath94 and we have to take into account corrections due to second order in @xmath24 term @xmath95 in expression for monomer density,@xmath96 where @xmath97 to first order in @xmath98 we find@xmath99 \label{si_corr}\ ] ] where the equilibrium displacement in @xmath100 is determined as@xmath101 we conclude that the correction term in eq .
( [ si_corr ] ) enhances the contrast between the high and the low monomer density regions of the profile ( fig .
[ si_poor ] ) .
poor3d this effect is the mostly pronounced near the corners of the pattern where several edges converge and it leads to distortion of the otherwise affinely stretched profile at these points .
frozen - in random heterogeneities of network structure can change the image beyond recognition @xcite .
the free energy of a gel with frozen - in heterogeities was derived in ref .
. the only source of heterogeneities in the melt with fixed monomer density is statistical distribution of cross - links in the state of preparation that arises as the consequence of the random process of cross - linking .
this frozen - in distribution is described by an additional contribution to the free energy : @xmath102 where @xmath103 is random gaussian function of coordinate @xmath13 , characterized by correlation function@xmath104 comparing the amplitude of frozen - in fluctuations on a scale @xmath105 with variations of elastic modulus @xmath106 on this scale we conclude that the contribution of frozen - in heterogeneities can be neglected if@xmath107 where @xmath108 is average cross - link concentration and thus , frozen - in heterogeities have no influence on large - scale patterns .
the suppression of frozen - in heterogeneities of monomer density is due to strong overlap of network chains @xcite . | using the theory of elasticity of polymer gels we show that large - scale _ cross - link _ density patterns written into the structure of the network in the melt state , can be revealed upon swelling by monitoring the _ monomer _ density patterns .
we find that while isotropic deformations in good solvent yield magnified images of the original pattern , anisotropic deformations distort the image ( both types of deformation yield affinely stretched images in @xmath0 solvents ) . we show that in ordinary solids with spatially inhomogeneous profile of the shear modulus , isotropic stretching leads to distorted density image of this profile under isotropic deformation .
using simple physical arguments we demonstrate that the different response to isotropic stretching stems from fundamental differences between the theory of elasticity of solids and that of gels .
possible tests of our predictions and some potential applications are discussed . |
gauge symmetries provide mathematical basis for known fundamental interactions . within the generalized hamiltonian framework @xcite
, gauge theories correspond to first - class constraints systems . upon gauge fixing , these systems convert to second - class constraint systems .
the operator quantization schemes for constraints systems have been developed by dirac @xcite . the path integral quantization has also been developed and found to be especially effective for gauge theories ( for reviews see @xcite ) .
besides conventional operator formulation of quantum mechanics and the path integral method , the popular approach to quantization of classical systems is based on the groenewold star - product formalism @xcite .
it takes the origin from the weyl s association rule @xcite between operators in the hilbert space and functions in phase space and the wigner function @xcite .
the star - product formalism is known also under the names of the deformation quantization and the moyal quantization @xcite .
the skew - symmetric part of the star - product , named the moyal bracket , governs the evolution of quantum systems in phase space , just like the poisson bracket governs the evolution of classical unconstrained systems and the dirac bracket governs the evolution of classical constraint systems .
the moyal bracket represents the quantum deformation of the poisson bracket .
the quantum deformation of the dirac bracket has been constructed recently @xcite .
the outline of the paper is as follows : in the next sect . , we give a pedagogical introduction to the weyl s association rule using the elegant method developed by stratonovich @xcite and give an introduction to the star - product formalism .
more details on this subject can be found in articles @xcite .
the phase - space functions and the dirac bracket do not make any physical sense outside of constraint submanifolds . in ref .
@xcite we constructed the quantum deformation of the dirac bracket on the constraint submanifold , sufficient for the purpose of generating time evolution of quantum constraint systems .
it would , however , be interesting from the abstract point of view to have a quantum - mechanical extension of the dirac bracket which matches smoothly at @xmath0 with the classical dirac bracket outside of the constraint submanifold also
. this problem is addressed and solved in sects .
iii and iv . in sect .
iii , we reformulate the classical constraint dynamics using projection formalism and present the classical dirac bracket of functions in terms of the poisson bracket of functions projected onto constraint submanifold . sect .
iv gives the quantum - mechanical generalization of the method proposed .
iii - d and iv - b , c contain new results , the others is a pedagogical exposition of earlier works ( mainly @xcite ) . in conclusion ,
we summarize results .
systems with @xmath1 degrees of freedom are described by @xmath2 canonical coordinates and momenta @xmath3 . these variables parameterize phase space @xmath4 defined as the cotangent bundle of @xmath1-dimensional configuration space @xmath5 .
canonical variables satisfy the poisson bracket relations @xmath6 the skew - symmetric matrix @xmath7 has the form @xmath8 where @xmath9 is the @xmath10 identity matrix and imparts to @xmath4 a skew - symmetric bilinear form .
the phase space acquires thereby structure of symplectic space .
the distance between two points in phase space is not defined .
one can measure , however , areas stretched on any two vectors @xmath11 and @xmath12 as @xmath13 where @xmath14 so that @xmath15 .
principal similarities and distinctions between euclidean and symplectic spaces are cataloguized in table 1 . for skew - gradients of functions ,
short notation @xmath16 is used .
@xmath17 in quantum mechanics , canonical variables @xmath18 are associated to operators of canonical coordinates and momenta @xmath19 acting in the hilbert space , which obey the commutation relations @xmath20=-i\hbar
i^{kl}. \label{comm}\ ] ] the weyl s association rule extends the correspondence @xmath21 to phase - space functions @xmath22 and operators @xmath23 .
it can be illustrated as follows : @xmath24 \nonumber \\
f(\xi ) \in c^{\infty } ( t_{*}\mathbb{r}^{n } ) & \longleftrightarrow & \mathfrak{f}\in op(l^{2}(\mathbb{r}^{n } ) ) \nonumber\end{aligned}\ ] ] the set of operators @xmath25 acting in the hilbert space is closed under multiplication of operators by @xmath26-numbers and summation of operators .
such a set constitutes vector space : @xmath27 elements of basis of such a vector space can be labelled by canonical variables @xmath28 . the commonly used weyl s basis looks like @xmath29 the objects @xmath30 satisfy relations @xcite @xmath31 & = & 1 , \nonumber \\
\int \frac{d^{2n}\xi } { ( 2\pi \hbar ) ^{n}}\frak{b}(\xi ) & = & \frak{1 } , \nonumber \\ \int \frac{d^{2n}\xi } { ( 2\pi \hbar ) ^{n}}\frak{b}(\xi ) tr[\frak{b}(\xi ) \frak{f } ] & = & \frak{f } , \nonumber \\
tr[\frak{b}(\xi ) \frak{b}(\xi ^{\prime } ) ] = ( 2\pi \hbar ) ^{n}\delta ^{2n}(\xi & -&\xi ^{\prime } ) , \nonumber \\
\frak{b}(\xi ) \exp ( -\frac{i\hbar } { 2}\mathcal{p}_{\xi \xi ^{\prime } } ) \frak{b}(\xi ^{\prime } ) & = & \nonumber \\ = ( 2\pi \hbar ) ^{n}\delta ^{2n}(\xi - \xi ^{\prime } ) \frak{b}(\xi ^{\prime } ) .&\ ; & \nonumber\end{aligned}\ ] ] here , @xmath32 is the so - called poisson operator .
the weyl s association rule for a function @xmath33 and an operator @xmath25 has the form @xcite @xmath34 , \label{s } \\
\mathfrak{f } & = & \int \frac{d^{2n}\xi } { ( 2\pi \hbar ) ^{n}}f(\xi ) \mathfrak{b}(\xi ) .
\label{inv}\end{aligned}\ ] ] in particular , @xmath35 \label{coordi1 } \\
\mathfrak{x}^{i } & = & \int \frac{d^{2n}\xi } { ( 2\pi \hbar ) ^{n}}\xi ^{i}\mathfrak{b}(\xi ) .
\label{coordi2}\end{aligned}\ ] ] the function @xmath33 can be treated as the coordinate of @xmath25 in the basis @xmath30 , while the right side of eq.([s ] ) can be interpreted as the scalar product of @xmath30 and @xmath25 .
alternative operator bases and their relations are discussed in refs .
@xcite . one can make , in particular , operator transforms on @xmath30 and @xmath26-number transforms on @xmath18 .
ambiguities in the choice of operator basis are connected to ambiguities in quantization of classical systems , better known as `` operator ordering problem '' .
the set of operators is closed under multiplication of operators .
the vector space of operators is endowed thereby with an associative algebra structure . given two functions @xmath36 $ ] and @xmath37 $ ] , one can construct a third function @xmath38 .
\label{gr}\ ] ] this operation is called star - product
. it has been introduced by groenewold @xcite .
the explicit form of the star - product is as follows : @xmath39 where @xmath40 .
the star - product splits into symmetric and skew - symmetric parts @xmath41 the skew - symmetric part @xmath42 is known under the name of moyal bracket .
it is essentially unique @xcite .
it governs quantum evolution in phase space and endows the set of functions with the poisson algebra structure : @xmath43 the average values of a physical observable described by function @xmath44 are calculated in terms of the wigner function @xmath45 .
\label{wign}\ ] ] it is normalized to unity @xmath46 if @xmath47 and @xmath48 where @xmath49 is the density matrix , then @xmath50 & = & \int \frac{d^{2n}\xi } { ( 2\pi \hbar)^{n}}f(\xi ) \star w(\xi ) \nonumber \\ & = & \int \frac{d^{2n}\xi } { ( 2\pi \hbar)^{n}}f(\xi ) w(\xi ) .
\label{tr}\end{aligned}\ ] ] under the sign of integral , the star - product can be replaced with the pointwise product @xcite .
real functions in phase space stand for physical observables , which constitute in turn the poisson algebra .
if the associative product @xmath51 does not commute , its skew - symmetric part gives automatically the skew - symmetric product which satisfies the leibniz law @xmath52 this equation is valid separately for symmetric and skew - symmetric parts of the star - product . in the last case , eq.([leib ] ) provides the jacobi identity .
the validity of the leibniz law allows to link the moyal bracket with time derivative of functions and build up thereby an evolution equation for functions in phase space . in classical limit
, the moyal bracket turns to the poisson bracket : @xmath53
second - class constraints @xmath54 with @xmath55 and @xmath56 have the poisson bracket relations which form a non - degenerate @xmath57 matrix @xmath58 if this condition is not fulfilled , it would mean that gauge degrees of freedom appear in the system . after imposing gauge - fixing conditions
, we could arrive at inequality ( [ nongen ] ) .
alternatively , breaking condition ( [ nongen ] ) would mean that constraint functions are dependent .
after removing redundant constraints , we arrive at inequality ( [ nongen ] ) .
constraint functions are equivalent if they describe the same constraint submanifold . within this class
one can make transformations without changing dynamics . for arbitrary point @xmath59 of constraint submanifold @xmath60
, there is a neighborhood where one may find equivalent constraint functions in terms of which the poisson bracket relations look like @xmath61 where @xmath62 here , @xmath63 is the identity @xmath64 matrix , @xmath65 .
the global existence of symplectic basis ( [ sb ] ) is an opened question in general case .
the basis ( [ sb ] ) always exists locally , i.e. , in a finite neighborhood of any point of the constraint submanifold .
this is sufficient for needs of perturbation theory .
the formalism presented in this section can therefore to be used to formulate evolution problem of any second - class constraints system in phase space in the sense of the perturbation theory .
the existence of the local symplectic basis ( [ sb ] ) is on the line with the darboux s theorem ( see , e.g. , @xcite ) which states that in symplectic space around any point @xmath59 there exists coordinate system in @xmath66 such that @xmath67 where symplectic structure takes the standard canonical form .
symplectic spaces can be covered by such coordinate systems .
this is in contrast to riemannian geometry where metric tensor at any given point @xmath68 can always be made minkowskian , but in any neighborhood of @xmath68 the variance of the riemannian metric with the minkowskian metric is , in general , @xmath69 .
physically , by passing to inertial coordinate frame one can remove gravitation fields at any given point , but not in an entire neighborhood of that point .
the darboux s theorem states , reversely , that the symplectic structure can be made to take the standard canonical form in an entire neighborhood @xmath66 of any point @xmath59
. in riemannian spaces , locally means at some given point . in symplectic spaces ,
locally means at some given point and in an entire neighborhood of that point .
locally , all symplectic spaces are indistinguishable .
conditionally , one can say that any surface in symplectic space , including any constraint surface , is a plane
. in the view of this marked dissimilarity , the validity of eqs.([pois ] ) in a finite domain looks indispensable .
the concept of skew - gradient projection @xmath70 of canonical variables @xmath59 onto constraint submanifold plays very important role in the moyal quantization of constraint systems .
geometrically , skew - gradient projection acts along phase flows @xmath71 generated by constraint functions .
these flows are commutative in virtue of eqs.([sb ] ) : using eqs.([sb ] ) and the jacobi identity , one gets @xmath72 for any function @xmath73 , so the point of intersection with @xmath74 is unique .
skew - gradient projections are investigated in refs . @xcite and independently in refs .
@xcite . to construct skew - gradient projections , we start from equations @xmath75 which say that point @xmath76 is left invariant by phase flows generated by @xmath77 . using symplectic basis ( [ sb ] ) for the constraints and expanding @xmath78 in the power series of @xmath79 ,
one gets @xmath80 similar projection can be made for function @xmath33 : @xmath81 it satisfies @xmath82 constraint functions are in involution with projected function : @xmath83 consequently , @xmath84 does not vary along @xmath85 , since @xmath86 applying eqs.([fsg ] ) and ( [ fssf ] ) to constraint functions @xmath87 , one concludes that the point @xmath70 belongs to the constraint submanifold @xmath88 the constraint submanifold can therefore be described equivalently as @xmath89 .
an average of function @xmath44 is calculated using the probability density distribution @xmath90 and the liouville measure restricted to the constraint submanifold @xcite : @xmath91 on the constraint submanifold @xmath92 , so @xmath33 and @xmath90 can be replaced with @xmath84 and @xmath93 . there
exist therefore equivalence classes of functions in phase space : @xmath94 the symbol @xmath95 means that functions are equal in the weak sense , @xmath96 , i.e. , on the constraint submanifold .
we shall see that symbols @xmath97 and @xmath98 acquire distinct meaning upon quantization .
note that @xmath99 eqs.([fssf ] ) and ( [ on ] ) imply @xmath100 .
constraint functions belong to an equivalence class containing zero .
given hamiltonian function @xmath101 , the evolution of function @xmath73 is described using the dirac bracket @xcite @xmath102 in the symplectic basis ( [ sb ] ) , the dirac bracket looks like @xmath103 on the constraint submanifold , one has @xmath104 calculation of the dirac bracket can be replaced therefore with calculation of the poisson bracket for functions projected onto the constraint submanifold .
two functions are equivalent provided they coincide on the constraint submanifold .
the hamiltonian functions determine the evolution of systems and play thereby special role .
two hamiltonian functions are equivalent if they generate within @xmath74 phase flows whose projections onto the tangent plane of the constraint submanifold are identical .
one may suppose that the equivalence relation for functions , defined above , does not apply to hamiltonian functions , since skew - gradients of hamiltonian functions enter the problem either .
this is not the case , however .
the components of the hamiltonian phase flow , which belong to a subspace spanned at @xmath74 by phase flows of the constraint functions , do not affect dynamics and could be different , whereas the skew - gradient projection ( [ fsg ] ) _ does not modify components of skew - gradients of functions , tangent to constraint submanifold_. we illustrate it schematically on fig .
2 . the geometrical sense of the dirac bracket reduces to dropping the component of the hamiltonian phase flow which does not belong to tangent plane of the constraint submanifold .
equivalently , those components can be made to vanish with the help of the skew - gradient projection . @xmath101 and @xmath105 are thereby _ dynamically equivalent _ , so eq.([eq ] ) characterizes an equivalence class for the hamiltonian functions either . among functions of this class
, @xmath105 is the one whose phase flow is skew - orthogonal to phase flows of the constraint functions , i.e. , @xmath106 . and
@xmath107 generated by hamiltonian function @xmath108 and projected hamiltonian function @xmath109 at point @xmath59 of constraint submanifold @xmath74 .
the phase flow @xmath107 belongs to the tangent plane of @xmath74 .
the hamiltonian phase flow @xmath110 admits decomposition @xmath111 . within the constraint submanifold ( i.e. @xmath112 and @xmath113 ) one
has @xmath114 and therefore @xmath115 .
the first term @xmath116 is therefore skew - orthogonal to any vector @xmath117 of the tangent plane .
, width=5 ] replacing @xmath101 with @xmath105 , one can rewrite the evolution equation in terms of the poisson bracket ( cf .
eq.([ev ] ) ) : @xmath118 the evolution does not mix up the equivalence classes . _ the physical observables in second - class constraints systems are associated with the equivalence classes of real functions in the unconstrained phase space .
the equivalence classes constitute a vector space _
@xmath119 _ _ equipped with two multiplication operations , the associative pointwise product and the skew - symmetric dirac bracket @xmath120 which confer _ _ @xmath119 _ _ a poisson algebra structure . _ _ instead of working with equivalence classes of functions @xmath121 , one can work with their representatives @xmath122 defined uniquely by the skew - gradient projection .
the one - to - one mapping @xmath123 induces a poisson algebra structure on the set of projected functions .
the sum @xmath124 converts to @xmath125 , the associative product @xmath126 converts to the pointwise product @xmath127 , while the dirac bracket becomes the poisson bracket : @xmath128 these operations satisfy the leibniz law and the jacobi identity and , since @xmath129 , @xmath130 , and @xmath131 , keep the set of projected functions closed
. outside of the constraint submanifold functions do not make any physical sense .
it is sufficient thus to work with the dirac bracket on the constraint submanifold .
the evolution problem in such a case can consistently be formulated in terms of the poisson bracket for functions projected onto the constraint submanifold .
the dirac bracket is , however , well defined in the whole phase space .
redefinition of constraint functions by shifts @xmath132 leaves the dirac bracket unchanged , because it depends on derivatives of constraint functions only .
it is not the case for the poisson bracket applied to projected functions .
this is why eq.([db1 ] ) is valid on constraint submanifold only .
one can modify projection formalism to fit the above - mentioned property of the dirac bracket .
suppose we wish to find the dirac bracket of functions @xmath133 and @xmath134 at a point @xmath135 outside of the constraint submanifold .
the intersection of level sets @xmath136 can be considered as new constraint submanifold defined by constraint functions @xmath137 projected functions depend thereby on both @xmath138 and @xmath139 : @xmath140 and similarly for @xmath141 the poisson brackets are calculated with respect to @xmath138 while @xmath59 is a parameter .
the appropriate extension looks like @xmath142 in eq.([db1 ] ) all four terms are pairwise distinct functions in the whole phase space .
these functions coincide on the constraint submanifold only . in eq.([outside ] ) all four terms coincide in the whole phase space . if @xmath112 , we reproduce the result ( [ db1 ] ) derived earlier .
scheme presented in the previous sect .
3 is suitable to approach description of quantum constraint systems in phase space .
we give final results and refer to @xcite for intermediate steps .
we remind that classical hamiltonian function @xmath108 and constraint functions @xmath87 are distinct in general from their quantum analogues @xmath143 and @xmath144 .
these dissimilarities are connected to the usual ambiguities in quantization of classical systems , being not specific for the problem we are interested in .
it is required only @xmath145 in what follows @xmath146 .
the quantum constraint functions @xmath144 satisfy @xmath147 in classical limit , @xmath144 turn to @xmath87 . the quantum - mechanical version of the skew - gradient projections is defined with the use of the moyal bracket @xmath148 the projected canonical variables have the form @xmath149 the quantum analogue of eq.([fsg ] ) is @xmath150 the function @xmath151 obeys equation @xmath152 the evolution equation which is the analogue of eq.([pev ] ) takes the form @xmath153 where @xmath154 is the hamiltonian function projected onto the constraint submanifold as prescribed by eq.([sgrad4 ] ) . taking projection of eq.([pev2 ] ) we get evolution equation in the closed form for projected functions : @xmath155 _ the quantum deformation of the dirac bracket represents the moyal bracket for two functions projected quantum - mechanically onto the constraint submanifold .
_ the formal structure of the dynamical quantum system is described by the scheme ( [ scheme ] ) with the word `` functions '' replaced by the phrase `` projected functions '' and @xmath73 and @xmath156 replaced by @xmath157 and @xmath158 , respectively .
the star - product is an associative operation , whereas the moyal bracket for projected functions satisfies the leibniz law and , respectively , the jacobi identity .
projected functions in phase space are objects associated to quantum observables .
functions which have the same projections are physically equivalent .
we can unify such functions into equivalence classes .
the star - product and the moyal bracket for projected functions generate for equivalence classes a poisson algebra structure accordingly . .
brackets which govern evolution in phase space of functions ( second column ) and projected functions ( third column ) of classical systems ( first row ) and quantum systems ( second row ) .
the right upper corner shows the dirac bracket expressed in terms of the poisson bracket of functions projected onto the constraint submanifold .
the left upper corner is the poisson bracket .
the left lower corner is the moyal bracket , which represents the quantum deformation of the poisson bracket .
the operation @xmath159 is the quantum deformation of the dirac bracket .
[ cols="<,^,^",options="header " , ] the bracket @xmath160 constructed in @xcite gives the deformation of the dirac bracket on @xmath74 .
what about the whole phase space ?
one can generalize the operation @xmath161 to match in classical limit the dirac bracket outside of the constraint submanifold .
we can proceed like in the classical case by writing projected functions in the form @xmath162 where @xmath163 the moyal brackets and the @xmath164-products entering this equation are calculated with respect to @xmath138 .
the desired extension looks like @xmath165 it is assumed that the constraint functions @xmath144 satisfy the bracket relations ( [ sbas ] ) at @xmath166 .
expression ( [ outsidequa ] ) is valid on and outside of the constraint submanifold . if @xmath112 , we reproduce operation @xmath167 announced earlier .
the set of operators @xmath168 is known to be complete , so that any operator @xmath25 can be represented as a symmetrized ( probably infinite ) weighted sum of products of operators @xmath168 . in the sense of the taylor expansion ,
one can write @xmath169 . the one - to - one correspondence between operators @xmath170 and functions in phase space @xmath44 , based on the taylor expansion , is equivalent to the weyl s association rule .
the similar completeness condition holds for projected operators of canonical variables @xmath171 which are inverse weyl s transforms of @xmath172 .
apparently , any operator @xmath25 acting in the hilbert space can be represented as an operator function @xmath173 . applying projection to the symmetrized product of @xmath174 constraint operators @xmath175 , which are inverse weyl s transforms of @xmath176 ,
one gets a series like @xmath177 , and so @xmath178 the taylor series of @xmath173 generates thereby vanishing terms involving @xmath175 .
we thus obtain @xmath179 respectively , any function projected quantum - mechanically onto the constraint submanifold can be represented in the form @xmath180 one can pass to classical limit to get eq.([fssf ] ) .
constructing @xmath181 from @xmath44 is a non - trivial task equivalent to solving constraints .
the operator counterpart of eq.([great ] ) , @xmath182 demonstrates the completeness of projected set of operators of canonical coordinates and momenta .
accordingly , eq.([great ] ) shows completeness of the set of @xmath172 in description of projected functions .
it is worthwhile to notice that eq.([ggggt ] ) does not extend to antisymmetric products of @xmath175 as one sees from @xmath183_{t } = ( - \mathcal{i}^{ab})_{t } = - \mathcal{i}^{ab } \neq [ \mathfrak{g}^{a}_{t},\mathfrak{g}^{b}_{t } ] = 0 $ ] where condition @xmath184 is taken into account .
we made short introduction to the weyl s association rule and the groenewold star - product technique for unconstrained and constraint systems .
the attention was focused to the evolution problem .
the use of skew - gradient projection formalism allows to treat unconstrained and constraint systems essentially on the same footing .
projections of solutions of quantum evolution equations onto the constraint submanifold comprise the entire information on quantum dynamics of constraint systems . m. nakamura and n. mishima , nuovo cim .
* 79b * , 287 ( 1984 ) ; prog .
theor . phys . * 81 * , 451 ( 1989 ) ; m. nakamura and h. minowa , j. math .
phys . * 34 * , 50 ( 1993 ) ; m. nakamura and k. kojima , nuovo cim . * 116b * , 287 ( 2001 ) . | quantization of constraint systems within the weyl - wigner - groenewold - moyal framework is discussed .
constraint dynamics of classical and quantum systems is reformulated using the skew - gradient projection formalism .
the quantum deformation of the dirac bracket is generalized to match smoothly the classical dirac bracket in and outside of the constraint submanifold in the limit @xmath0 . |
calculation of the form factors for semileptonic transitions of @xmath9 mesons has been being a subject discussed intensely .
recently , it has been shown that the @xmath10 transition form factor can be consistently analyzed by using the different approaches in the different @xmath11 regions @xcite . the perturbative qcd ( pqcd )
can be applied to the @xmath10 form factor in the large recoil ( small @xmath11 ) region and it is reliable when the involved energy scale is large enough @xcite .
the qcd light - cone sum rules ( lcsr ) can involve both the hard and soft contributions to the @xmath10 form factor below @xmath12 @xcite .
the lattice qcd simulations of the @xmath10 transition form factor @xcite are available only for the soft region @xmath13 , because of the restriction to the @xmath14 energy smaller than the inverse lattice spacing .
thus the results from these three approaches might be complementary to each other . in ref.@xcite
we recalculate the @xmath10 form factor in the pqcd approach , with the transverse momentum dependence included for both the hard scattering part and the nonperturbative wave functions(of @xmath14 and @xmath9 ) to get a more reliable pqcd result . by combining the results from these three methods we obtain a full understanding of the @xmath10 transition form factor in its physical region @xmath15 .
up to now , in comparison with heavy - to - light cases the calculations on heavy - to - heavy transitions can be done only for a certain specific kinematical range , although there have been a lot of discussions in the literature . in the bsw model @xcite ,
the relevant form factors at zero momentum transfer are expressed as an overlap of initial and final meson wave functions for which they take the solutions of the bethe - salpeter ( bs ) equation in a relativistic harmonic oscillator potential .
then one extrapolates the result at @xmath16 to the whole kinematical region assuming the nearest pole dominance . with the discovering of the heavy quark symmetry , the @xmath17 form factor
have been known better at zero recoil .
this is because of the fact that in the heavy quark limit the resulting form factors isgur - wise functions @xcite at zero recoil are rigorously normalized . including the leading symmetry breaking corrections , the deviation from this limit
can be estimated at an order of a few percent due to luke s theorem and therefore the value of the form factor at this point can be determined within a higher accuracy @xcite
. however , the dependence of the form factor on the velocity transfer @xmath18 ( with @xmath19 and @xmath20 being the velocities of the @xmath9 and @xmath1 mesons respectively ) is difficult to get even in the leading order , in view of the arbitrary function @xmath21 @xcite which is introduced to simulate higher - resonances in the heavy quark effective theory ( hqet ) .
the lattice qcd , despite a rigorous nonperturbative approach , is just adequate to estimate the behavior of the form factors near the zero recoil @xcite . among the other approaches are the qcd sum rules and pqcd .
ref.@xcite applies the traditional @xmath22-point sum rule to calculate the form factor at zero momentum transfer .
it is concluded in ref.@xcite that pqcd approach is applicable in the large recoil region and can give a consistent result with the experiment .
it is necessary that there is a reliable estimate of @xmath23 transition in the whole kinematically accessible range @xmath24 , in order to account for the data on @xmath0 . for this purpose , it is practical , as shown in @xmath25 case , to combine the result of qcd lcsr with those from the lattice qcd , heavy quark symmetry and pqcd .
the lcsr approach @xcite , where the non - perturbative dynamics are effectively parameterized in so - called light - cone wave functions , is regarded as an effective tool to deal with heavy - to - light exclusive processes .
although the @xmath17 transition in question is a heavy - to - heavy one , the @xmath26-quark is much lighter compared to @xmath27-quark and so discussing it with lcsr is plausible for the kinematical range where the ope near light - cone @xmath3 is valid .
the other problem with our practical calculation is that the higher twist da s of d meson , which are important but less studied , would enter into the sum rule result .
however , an effective approach @xcite has been presented to avoid the pollution by some higher - twist da s .
this improved lscr method uses a certain chiral current correlator as the starting point so that the relevant twist-@xmath22 wave functions make no contributions and the reliability of calculation can be enhanced to a large extend .
its applicability has been examined by a great deal of studies @xcite . in this paper
we would like to employ the improved lcsr to discuss the form factor for the @xmath17 transition and try to give a full understanding of qcd dynamics involved in the @xmath28 .
this paper is organized as follows . in the following section we derive the lcsr for the form factor for @xmath23 .
a discussion of the da models for the @xmath1-meson is given in section iii .
section iv is devoted to the numerical analysis and comparison with other approaches .
the last section is reserved for summary .
the @xmath17 weak form factors @xmath29 and @xmath30 are usually defined as : @xmath31 with @xmath32 being the momentum transfer . on the other hand ,
when applying the heavy quark symmetry to do discussion the following definition is advisable , @xmath33.\ ] ] if we neglect the masses of leptons in the decay final state of @xmath34 , only @xmath29 is relevant and thus we can confine us to the discussion on @xmath29 .
obviously , the following relation is observed between @xmath29 and @xmath35 , @xmath36 where @xmath37 , @xmath38 . using the heavy quark symmetry ,
the value of form factor @xmath39 at zero recoil could be fixed better . since in heavy quark limit @xmath40 and @xmath41 , the form factor @xmath39 should be close to @xmath42 .
a systematic investigation gives @xmath43@xcite , with a less model dependence .
this result is also confirmed with lattice calculations @xcite .
pqcd analyses are also made in the large recoil region @xmath44 , yielding a result consistent with the data .
the lcsr calculation can help to understand @xmath45 in the whole kinematical region in complementary to the lattice qcd with the heavy quark symmetry and pqcd approaches . to achieve a lcsr estimate of @xmath46 ,
we follow @xcite and use the following chiral current correlator @xmath47 : @xmath48 in the first place , we discuss the hadronic representation for the correlator .
this can be done by inserting the complete intermediate states with the same quantum numbers as the current operator @xmath49 . isolating the pole contribution due to the lowest pseudoscalar @xmath9 meson
, we have the hadronic representation in the following : @xmath50 note that the intermediate states @xmath51 contain not only the pseudoscalar resonance of masses greater than @xmath52 , but also the scalar resonances with @xmath53 , corresponding to the operator @xmath54 . with eq.([eq : def ] ) and the definition of the decay constant @xmath55 of the @xmath9-meson @xmath56 and expressing the contributions of higher resonances and continuum states in a form of dispersion integration , the invariant amplitudes @xmath57 and @xmath58 read , @xmath59=\frac{2f(q^2)m_b^2f_b}{m_b(m_b^2-(p+q)^2)}+ \int^\infty_{s_0}{\frac{\rho^h(s)}{s-(p+q)^2}ds}+\mbox{subtractions},\ ] ] and @xmath60=\frac{\tilde{f}(q^2)m_b^2f_b}{m_b(m_b^2-(p+q)^2)}+ \int^\infty_{s_0}{\frac{\tilde{\rho}^h(s)}{s-(p+q)^2}ds}+\mbox{subtractions},\ ] ] where the threshold parameter @xmath61 should be set near the squared mass of the lowest scalar @xmath9 meson , the spectral densities @xmath62 and @xmath63 can be approximated by invoking the quark - hadron duality ansatz @xmath64 on the other hand , we need to calculate the corrector in qcd theory to obtain the desired sum rule result .
in fact , there is an effective kinematical region which makes ope applicable : @xmath65 for the @xmath66 channel and @xmath67 for the momentum transfer . for the present purpose , it is sufficient to consider the invariant amplitude @xmath68 which contains the desired form factor .
the leading contribution is derived easily by contracting the @xmath69quark operators to a free propagator : @xmath70substituting eq.([eq : pro ] ) into eq.([eq : cc ] ) , we have the two - particle contribution to the correlator , @xmath71 an important observation , as in ref.@xcite , is that only the leading non - local matrix element @xmath72 contributions to the correlator , while the nonlocal matrix elements @xmath73 and @xmath74 whose leading terms are of twist @xmath22 , disappear from the sum rule . proceeding to eq.([eq : qq ] )
, we can expand the nonlocal matrix element @xmath75 as @xmath76 where @xmath77 is the twist-2 da of d meson with @xmath78 being the longitudinal momentum fraction carried by the @xmath26-quark , those da s entering the higher - twist terms are of at least twist @xmath79 .
the use of eq.([eq : da ] ) yields @xmath80=2f_dm_b\int_0 ^ 1{du\frac{\varphi_d(u)}{m_b^2-(up+q)^2}}+\mbox{higher twist terms}. \label{eq : qq1}\ ] ] invoking a correction term due to the interaction of the b quark with a background field gluon into ( 10 ) , the three - particle contribution @xmath81 is achievable .
however , the practical calculation shows that the corresponding matrix element whose leading term is of twist @xmath22 also vanishes .
thus , if we work to the twist-3 accuracy , only the leading twist da @xmath82 is needed to yield a lcsr prediction .
furthermore , we carry out the subtraction procedure of the continuum spectrum , make the borel transformations with respect to @xmath83 in the hadronic and the qcd expressions , and then equate them . finally , from eq.([eq : f12 ] ) follows the lcsr for @xmath84 , which is applicable to the velocity transfer region @xmath85 , @xmath86 } \varphi_d(u)},\label{eq : ff}\end{aligned}\ ] ] where @xmath87 and @xmath88 has been used .
now let s do a brief discussion on an important nonperturbative parameter appearing in the lcsr formula([eq : ff ] ) , the leading twist da of @xmath1-meson , @xmath89 .
@xmath1-meson is composed of the heavy quark @xmath26 and the light anti - quark @xmath90
. the longitudinal momentum distribution should be asymmetry and the peak of the distribution should be approximately at @xmath91 . according to the definition in eq.([eq : da ] ) , @xmath89 satisfies the normalization condition @xmath92 which is derived by the leptonic decay @xmath93 . in the pqcd calculations @xcite ,
a simple model ( we call model i ) is adopted as @xmath94 which is based on the expansion of the gegenbauer polynomials .
eq.([eq : da1 ] ) has a free parameter @xmath95 which ranges from @xmath96 to @xmath42
. we will take @xmath97 as input .
on the other hand , it was suggested in @xcite that the light - cone wave function of the @xmath1-meson be taken as : @xmath98}\label{eq : wf}\ ] ] which is derived from the brosky - huang - lepage(bhl ) prescription @xcite .
@xmath99 can be related to the da by the definition @xmath100 substituting eq.([eq : wf ] ) into eq.([eq : wf - da ] ) , we have a model of the da(model ii ) @xmath101},\label{eq : da2}\ ] ] where the parameters @xmath102 and @xmath103 can be fixed by the normalization([eq : constraint1 ] ) and the probability of finding the @xmath104 fock state in the @xmath1- meson , @xmath105 @xmath106 as discussed in ref.@xcite , @xmath107 is a good approximation for the @xmath1-meson ( as we have checked , change of @xmath105 makes a numerical effect less than @xmath108 ) .
then , taking @xmath109 , @xmath110 , @xmath111 and @xmath112 , we have @xmath113 , @xmath114 .
furthermore , as argued in ref.@xcite , a more complete form of the light - cone wave function should include the melosh rotation effect in spin space : @xmath115}\ ] ] with the melosh factor , @xmath116 it can be seen from eq.([eq : melosh ] ) that @xmath117 as @xmath118 , since there is no spin interaction between the two quarks in the heavy - flavor meson , ie .
, the spin of the heavy constituent decouples from the gluon field , in the heavy quark limit @xcite .
however the @xmath26-quark is not heavy enough to neglect the melosh factor .
after integration over @xmath119 the full form of @xmath1 meson da can be achieved ( model iii ) : @xmath120 \exp{\left[-b_d^2\frac{(xm_d^2+(1-x)m_c^2-y^2)}{x(1-x)}\right]}\label{eq : da3},\ ] ] where @xmath121 and the error function @xmath122 is defined as @xmath123 . using the same constraints as in eq.([eq : constraint1 ] ) and
( [ eq : constraint2 ] ) , the parameters @xmath102 and @xmath103 are fixed as @xmath124 and @xmath125 . in this paper we will employ the above three kinds of models to do numerical calculation .
all these da s of the @xmath1-meson are plotted in fig.([fig : wf ] ) for a comparison .
it is shown that they are of similar shape and all of them exhibit a maximum at @xmath126 as expected .
apart from the da of @xmath1-meson , the decay constant of @xmath9-meson @xmath55 is among the important nonperturbative inputs . for consistency , we use the following corrector @xmath127 to recalculate it in the two - point sum rules .
the calculation should be limited to leading order in qcd , since the qcd radiative corrections to the sum rule for @xmath46 are not taken into account .
the value of the threshold parameter @xmath128 is determined by a best fit requirement in the region @xmath129 , where @xmath130 is the corresponding borel parameter .
the same procedure is performed for @xmath46 , resulting in different values of the threshold parameter @xmath61 .
the result is listed in tab.[tab : s0 ] .
choosing @xmath61 and @xmath128 in this way , the dependence of @xmath55 and @xmath46 on the borel parameter is very weak and thus we can simply evaluate them at @xmath131 .
the other input parameters are taken as @xmath132 .
as we have ignored all the radiation corrections , we do nt expect our values of @xmath55 to be good predictions of that quantity .
.parameter sets for @xmath55 and @xmath46 , @xmath128 and @xmath61 for @xmath55 and @xmath46 respectively ; @xmath133 and @xmath55 are given in gev , @xmath61 and @xmath128 in @xmath134 . [ cols="<,<,^,^,^",options="header " , ] with the parameters chosen , it is straightforward to calculate the form factor @xmath46 in the region @xmath135 .
the results with different sets of parameters are plotted in fig.([fig : f(mb ) ] ) , where only model ii has been used for simplicity .
it is shown that the change of parameters can induce a uncertainty of about @xmath136 if we let @xmath133 vary between @xmath137 . by fitting the data
, the behavior of @xmath45 has been known using the parametrization @xmath138,\ ] ] with @xmath139 corresponding to the linear and quadratic fits @xcite , respectively . with the three da models ,
the resulting dependence of @xmath46 on the velocity transfer @xmath140 , along with that extracted experimentally is illustrated in fig.([fig : f(y ) ] ) . in what follows
, we denote the lcsr results for the form factor by @xmath141 and those extracted experimentally by @xmath142 . for comparison , a figure - copy which expresses the pqcd results in @xcite
is given in fig.([fig : pqcd ] ) .
in the region to which the lcsr method is applicable , the central values of @xmath141 turn out to be a bit smaller than the corresponding those of @xmath143 , using the da models ii and iii as inputs ; however , both of them are in accordance with each other within the error . the situation with model i da is about the same .
the central value of the form factor at the largest recoil is @xmath144 versus @xmath145 , depending on the da models .
we note that the behavior of @xmath141 is essentially unchanged when the three different da s are used . from the present calculations , therefore it is too early to draw a conclusion which da model is more suitable to reflect the characteristics of qcd dynamics inside the @xmath1 meson .
when a comparison is made between the pqcd and lcsr predictions , the consistent results can also be observed at the larger recoil .
of course , the two approaches describe the different dynamics in @xmath146 transitions . whereas the use of lcsr approach is to assume that the soft exchanges dominate in the weak decay in question , applying pqcd method to do calculation corresponds to the viewpoint that the hard exchanges do . in fact , the kinematical region we give , which makes lcsr results valid , is a conservative estimate .
it is possible to extrapolate the present lcsr calculation to the small recoil region .
if it is true , we find that in the whole kinematically accessible range @xmath147 , the yielding lcsr estimates are compatible with the data .
for instance , at zero recoil it follows that @xmath148 ( using model iii ) , which is in a good agreement with the evaluation obtained using the heavy quark symmetry : @xmath43 .
nevertheless , we have to emphasize that a full understanding of the dynamics involved in @xmath149 transition should be obtained by combining the three different approaches the lattice qcd calculations with the heavy quark symmetry considered , lcsr results and pqcd predictions , which are complementary to each other .
the lcsr results with chiral current correlator may act as a bridge connecting those of other approaches .
@xmath150 @xmath151 @xmath152 @xmath153
we have discussed the form factor for @xmath17 transitions @xmath46 , using the improved qcd lcsr approach where with the chiral current correlator chosen only the leading twist da of the @xmath1-meson is relevant at twist-3 accuracy .
the resulting lcsr s for @xmath46 are available in the velocity transfer region @xmath2 .
calculation is done using three different twist-2 da models for d meson .
it is shown the numerical results are less sensitive to the choice of da , and are of a central value slight smaller than but within the error in a agreement with those obtained by fitting the data on @xmath4 . in the larger recoil region @xmath5 where pqcd is applicable , the results presented here are consistent with ones of pqcd . from the practical calculations , we find that the present results might be extrapolated to the smaller recoil region so that the @xmath17 transitions are calculable in the whole kinematically accessible range , using the improved lcsr approach . also , we argue that for understanding the form factor for @xmath154 in the whole kinematical range a combined use is necessary of three different methods : the lattice qcd ( with the heavy quark symmetry considered ) , improved lcsr and pqcd approaches , which are adequate to do calculation in different kinematical regions and so could be complementary to each other .
the lcsr approach plays a bridge role in doing such calculation .
the present findings can be improved once the qcd radiative correction to the lcsr is taken into account and a more reliable
twist-2 da of @xmath1 meson becomes available . from the previous discussion in @xcite , however , it is expected that the qcd radiative correction can not change the present results too much .
s. j. brodsky , t. huang and g. p. lepage , in _ particles and fields-2 _ , proceedings of the banff summer institute , banff , alberta , 1981 , edited by a. z. capri and a. n. kamal ( plenum , new york , 1983 ) , p143 ; g. p. lepage , s. j. brodsky , t. huang , and p. b. mackenize , _ ibid .
_ , p83 ; t. huang , _ in proceedings of xxth international conference on high energy physics _ , madison , wisconsin , 1980 , edited by l. durand and l. g. pondrom , aip conf .
69 ( aip , new york , 1981 ) , p1000 . | within the framework of qcd light - cone sum rules ( lcsr ) , we calculate the form factor for @xmath0 transitions with chiral current correlator .
the resulting form factor depends on the distribution amplitude ( da ) of the @xmath1-meson .
we try to use three kinds of da models of the @xmath1-meson . in the velocity transfer region @xmath2 , which renders the operator product expansion ( ope ) near
light - cone @xmath3 go effectively , the yielding behavior of form factor is in agreement with that extracted from the data on @xmath4 , within the error . in the large recoil region @xmath5 ,
the results are observed consistent with those of perturbative qcd ( pqcd ) .
the presented calculation can play a bridge role connecting those from the lattice qcd , heavy quark symmetry and pqcd to have an all - around understanding of @xmath0 transitions .
* form factor for @xmath0 in the light - cone sum rules with chiral current correlator * * .3 cm fen zuo,@xmath6 zuo - hong li @xmath7 and tao huang @xmath8 * |
recently , there has been considerable progress in the thermodynamic description of small ( bio- ) systems @xcite which are prone to thermal fluctuations and typically driven far out of equilibrium . it has been shown that thermodynamic quantities obey various exact work and fluctuation theorems @xcite . the jarzynski relation connects the equilibrium free energy difference with an average of work values obtained from a series of nonequilibrium transitions @xcite .
while biomolecules are typically described by a langevin equation where effects of the thermal bath are included in friction and fluctuation terms , the jarzynski relation has originally been derived in the framework of purely hamiltonian dynamics with initial canonical equilibrium distribution @xcite .
this situation corresponds to an experiment where the heat bath of temperature @xmath0 is decoupled from a small hamiltonian system at time @xmath1 . during a time intevall
@xmath2 $ ] a control parameter @xmath3 is varied time - dependently .
such dynamics has been used in various generalizations of the jarzynski relation @xcite and for the optimization of the free energy reconstruction from nonequilibrium work data @xcite . while this approach is elegant from a theoretical point of view , it is important to note that such a dynamics has quite different properties than a system subject to a permanent heat bath modelled _
e. g. _ by a langevin equation .
the most important difference concerns the quasistatic work which is achieved for infinitesimally slow transitions @xmath4 , independent of the detailed shape of the protocol for the control parameter @xmath3 . for a system coupled permanently to a heat bath conserving the canonical distribution for fixed control parameter , this quasistatic work is equivalent to the free energy difference .
this property gives rise to the free energy calculation method termed thermodynamic integration where the free energy is approximated by the work for a very slow transition .
for purely hamiltonian dynamics , however , the quasistatic work ( which then is called adiabatic work ) can exceed the free energy difference @xcite .
still , the jarzynski relation is valid and can be used to estimate free energy differences .
it can be shown that under reasonable assumptions ( mainly ergodicity ) , both the work is a monotonically decreasing function of the transition time @xmath5 for a given protocol shape and the adiabatic work is independent of the shape of the protocol @xcite .
there has been considerable interest in deriving fluctuation theorems and analogues of the jarzynski relation also for systems in the quantum domain @xcite . here
, it is more subtle than in the classical case to define thermodynamic quantities such as work on a single trajectory @xcite .
apart from recent progresses for open quantum systems @xcite , most studies @xcite consider the case where the energy is measured at beginning and end , defining the work as the difference between final and initial energy .
these studies rely on schrdinger dynamics which has similar properties as hamiltonian dynamics .
in particular , the quasistatic work then is also different from the free energy difference . in this paper
, we study optimal protocols of the control parameter @xmath6 leading to a minimal mean work for a finite given transition time @xmath5 .
this question has been tackled for langevin dynamics in harmonic potentials @xcite with the result of unexpected singularities such as jumps in the overdamped case and even delta - type singularities in the underdamped case . for purely hamiltonian or schrdinger dynamics in harmonic potentials
, we will show that the optimal protocol is degenerate and that the optimal work is equal to the adiabatic work .
this result implies that the adiabatic work , which is achieved for an infinitely long transition time for any protocol , can be reached in an arbitrarily short time by using an appropriate protocol for the control parameter .
this surprising result shows that purely hamiltonian dynamics can beat langevin evolution since for the latter short transition times , @xmath7 , yield the work for an instantaneous jump . in the totally different context of collective escape over a barrier
, it has been shown that hamiltonian dynamics also beats langevin dynamics in the sense that it leads to a faster escape @xcite .
we will also calculate the optimal work numerically for an anharmonic potential .
however , it is difficult to decide from the numerical data , whether the adiabatic work can be achieved in a finite transition time also for such an anharmonic potential .
in this section , we will study minimum work protocols for hamiltonian dynamics in three different types of time - dependent potentials : ( i ) a moving harmonic potential , ( ii ) a harmonic potential with time - dependent stiffness and ( iii ) an anharmonic quartic potential .
we consider a particle of mass @xmath8 subject to a moving harmonic potential @xmath9 where @xmath10 is the ( constant ) stiffness of the potential .
the minimum of the potential @xmath6 is changed time - dependently from an initial position @xmath11 to a final position @xmath12 .
the hamiltonian dynamics of position and momentum of the particle at time @xmath13 then is governed by the equations of motion @xmath14 where the prime denotes the derivative with respect to @xmath15 and the dot the derivative with respect to the time @xmath13 .
the particle initially is in thermal equilibrium with probability density @xmath16 \exp[-\beta p_0 ^ 2 / ( 2 m ) ] \label{eq : ham_rho}\ ] ] for the initial position @xmath17 and the initial momentum @xmath18 with normalization constant @xmath19 and inverse temperature @xmath20 with boltzmann s constant @xmath21 .
since there is no heat transfer during the ( purely hamiltonian ) transition , the work is given by the change of internal energy .
averaging over all initial conditions yields the mean work @xmath22_0^t \nonumber \\ & = & \left [ \frac { { \left \langle p^2 \right \rangle } } { 2 m } + \frac{k}{2 } \left ( { \left \langle x^2 \right \rangle } - 2 \lambda { \left \langle x \right \rangle } + \lambda^2 \right ) \right]_0^t \label{eq : ham_w_mov}\end{aligned}\ ] ] where @xmath23 denotes the average over the initial distribution of position and momentum ( [ eq : ham_rho ] ) .
it can easily be shown that the variances @xmath24 and @xmath25 are time independent for any protocol @xmath6 .
thus , the work ( [ eq : ham_w_mov ] ) can be written as @xmath26_0^t \nonumber \\ & = & \left [ \frac { { \left \langle p \right \rangle}^2 } { 2 m } + \frac{k}{2 } \left ( { \left \langle x \right
\rangle}-\lambda \right ) ^2 \right]_0^t\end{aligned}\ ] ] which becomes zero , _ i.e. _ , minimal for @xmath27 the evolution equations for these mean values are given by @xmath28 with initial values @xmath29 and @xmath30 .
with the freedom to choose the continuous function @xmath6 which corresponds to an infinite number of free parameters , it is quite obvious that the two final conditions can easily be met .
in fact , the optimal protocol is highly degenerate . for a specific choice , taking , _
e. g. _ , a third order polynomial @xmath31 with two free parameters @xmath32 suffices to meet the optimality conditions .
the solution of the evolution equations ( [ eq : ham : mean ] ) can be solved analytically and , inserted into equation ( [ eq : ham_mov_cond ] ) , leads to @xmath33 with @xmath34 . for short transitions times @xmath35 ,
the coefficients diverge as @xmath36 and @xmath37 . as a quite different choice
, both conditions can also be met by a linear protocol with delta singularities at the boundaries , in analogy to the underdamped stochastic dynamics case @xcite .
indeed , it can easily be verified that the protocol @xmath38\ ] ] fulfills the conditions ( [ eq : ham_mov_cond ] ) .
this expression could also have been obtained as the limit of the underdamped optimal protocol for a vanishing friction coefficient @xmath39 .
thus , for any given total transition time @xmath5 the minimal mean work is given by @xmath40 for infinitely long transition times , quite generally , the work is given by the adiabatic work .
for the studied model system , we have @xmath41 .
as shown above , this lower bound on the work can even be achieved for arbitrarily short transition time .
this surprising result shows that purely hamiltonian dynamics can beat langevin evolution where short transition times @xmath7 yield the work for an instantaneous jump @xmath42 .
the optimal protocols show a distinct positive peak in the first part of the transition and a distinct negative peak in the second part of the transition . this peak structure can be understood intuitively in complete analogy to the underdamped dynamics case @xcite .
the first peak accelerates the particle and leads to a finite mean velocity which is necessary to reach a final mean particle position at the potential minimum .
the second negative peak decelerates the particle , thus recovering the invested work from the kinetic energy of the particle .
we consider the time dependent potential @xmath43 where the stiffness @xmath6 is changed from an initial value @xmath44 to a final value @xmath45 in a finite time @xmath5 .
the work exerted on the system during the finite time transition is given by the change in internal energy @xmath46_0^t = \left [ \frac { { \left \langle p^2 \right \rangle } } { 2 m } + \frac \lambda 2 { \left \langle x^2 \right \rangle } \right]_0^t.\ ] ] the dynamics ( [ eq : ham_dyn ] ) leads to coupled evolution equations for the mean squared position @xmath47 and the mean squared momentum @xmath48 @xmath49 multiplication of equation ( [ eq : ham_ddotw ] ) with a factor @xmath50 and insertion of equation ( [ eq : ham_dotz ] ) leads to the relation @xmath51 which , using the equilibrium initial conditions @xmath52 , @xmath53 , and @xmath54 can be integrated to yield @xmath55 at the final time @xmath5 , minimization requires @xmath56 which is equivalent to the final condition @xmath57 the ( mean ) total work thus becomes @xmath58.\ ] ] optimizing this expression with respect to @xmath59 leads to @xmath60 which yields the optimal work @xmath61 similar to the first case study , the optimal protocol is highly degenerate . for a third order polynomial ,
the equations of motion can not be solved analytically . however , as expected , the two parameters can be chosen such that the boundary conditions ( [ eq : ham_z_opt ] ) and ( [ eq : ham_w_opt ] ) are fulfilled . optimal third order polynomial protocols @xmath62 are shown in fig .
[ fig : ham_stiff_prot ] for different transition times @xmath5 .
a short transition time requires pronounced peaks with increasing height for decreasing transition times .
these peaks serve to accelerate the brownian particle in the first part of the transition and to deccelerate the particle in the second part of the transition .
another convenient choice is a linear protocol with superimposed delta peak singularities at beginning and end .
the initial singularity allows for setting the initial mean squared momentum instantaneously to a value @xmath63 which can be tuned to meet the boundary condition on @xmath64 .
the second delta singularity at the final time @xmath5 allows for setting @xmath56 .
note that , similar to the first case study , the minimal work ( [ eq : meanw_ham ] ) does not depend on the transition time @xmath5 .
naively , in the limit @xmath7 , one would expect the result obtained for an instantaneous jump of the protocol @xmath65 which is substantially larger than the optimal work .
however , the possibility to use a protocol with increasing absolute values of @xmath3 ( and thus increasing forces ) for decreasing transition times @xmath5 leads to the singular limit at @xmath7 . for any fixed shape of the protocol @xmath66 , the work approaches @xmath67 with @xmath7 . in both harmonic case studies , the optimal work @xmath68 is given by the adiabatic work @xmath69 , independent of the total transition time @xmath5
the optimal protocols @xmath70 are highly degenerate .
it is interesting to see whether these ( unexpected ) features persist for anharmonic potentials where the optimality conditions can not be cast into only two conditions for the moments of the probability distribution .
however , it is computationally quite difficult to determine optimal protocols in the anharmonic case . for a case study , we use the quartic potential @xmath71 with @xmath72 , @xmath73 , and @xmath74 together with a fourier ansatz for the optimal protocol @xmath75 where we omit higher odd frequencies because their symmetry is not suitable to produce the peak structure found in the two harmonic cases .
we next optimize the parameters @xmath76 for a minimal mean work which we calculate as the average of work values from @xmath77 randomly sampled initial values for @xmath17 and @xmath18 .
for the minimization , we use a standard mathematica algorithm . note that the optimal work values obtained in this numerical procedure are upper bounds for the true optimal work since we have used a finite number of parameters .
the adiabatic work can be calculated using the microcanonical distribution for given energy @xmath78 and given @xmath3 .
the change of internal energy then is given by @xmath79 with the microcanonical partition function @xmath80 the evolution equation ( [ eq : ham_dgl_qs ] ) is an ordinary differential equation for @xmath81 .
for the potential @xmath82 , we obtain @xmath83 which has the solution @xmath84 with initial energy @xmath85 . averaging over initial conditions @xmath86 according to the appropriate initial boltzmann distribution ( [ eq : ham_rho ] )
then yields @xmath87 { 2 } - 1 ) \gamma ( 5/4 ) } { \gamma ( 1/4 ) } \simeq 0.1949 .\ ] ] in fig .
[ fig : ham_x4_w]a , we compare optimal work values with the adiabatic work .
even though it looks like convergence may be reached already by the small number of fourier modes , we can not decide this question conclusively . particularly for very small transition times ,
optimal protocols feature extreme protocol values which may require a large number of fourier modes .
moreover , we can not exclude the possibility that the minimum search is trapped in a local minimum preventing the convergence towards the global minimum .
therefore , we can not exclude the possibility that the adiabatic work can be reached in a finite time also in this case study .
more likely , however , is the following scenario : the optimal work is discontinuous at @xmath88 .
for any transition time , work values well below the work for an instantaneous jump @xmath89 but above the adiabatic work can be obtained .
clearly , even for very fast transitions compared to the characteristic oscillation time @xmath90 , the optimal work is well below the work for a linear protocol which has the limit @xmath89 for @xmath7 .
the optimal protocol again shows pronounced peaks for short transition times @xmath5 , see fig .
[ fig : ham_x4_w]b .
since we use an ansatz with a finite number of parameters , it is impossible to decide whether the optimal protocol is degenerate . even though we could not decide whether the adiabatic work can be reached in a finite time , we can show that free energy differences as determined via the jarzynski relation indeed improve for an optimal protocol compared to a linear protocol , see fig .
[ fig_df ] and its caption for details .
in this section , we investigate optimal protocols for quantum system . fortunately , for harmonic systems which are comparable to the first two case studies from the previous sections , work distributions @xmath91
have recently been calculated for schrdinger dynamics @xcite . by using their results
, it is easy to derive optimal protocols since in both cases , the work distributions ( and therefore also the mean work ) depend on a single `` rapidity '' parameter which vanishes for quasistatic transitions . in the last subsection
, we study the two level system of a single spin in a time dependent magnetic field .
if we impose that the absolute value of the magnetic field stays fixed , the adiabatic work can only be reached in a finite ( non - zero ) time .
we consider the quantum analogue of a moving harmonic potential @xmath92 with hamiltonian @xmath93 + \frac { m \omega^2 } { 2 } \lambda(\tau)^2\ ] ] with creation and annihilation operators @xmath94 and @xmath95 , planck constant @xmath96 , frequency @xmath97 and a time dependent minimum of the harmonic potential @xmath6 which is changed from an initial value @xmath98 to a final value @xmath45 during a given finite time @xmath5 .
it has recently been noted that the mean work is given exactly by the classical expression for this case study @xcite .
thus , the optimal protocols as discussed in sect . [ sect_ham_mov_pot ] also apply to the quantum case .
however , the work distributions are different for classical and quantum systems
. it has been shown @xcite that the work distribution @xmath91 for an initially thermalized ensemble depends only on the absolute value @xmath99 of a `` rapidity '' parameter @xmath100 which vanishes for adiabatic transitions with @xmath4 . however , even for transitions in a finite time @xmath5 , there exist appropriate protocols @xmath6 which lead to a vanishing `` rapidity parameter '' @xmath101 and thus to a vanishing mean work @xmath102 .
in addition to the optimal protocols from sect . [ sect_ham_mov_pot ] , it also suffices to consider the simple trigonometric protocol @xmath103\ ] ] which can be easily shown to yield @xmath104 for any integer @xmath105 .
again , the optimal protocol is highly degenerate . this degeneracy arises because the continuous protocol @xmath6 with an infinite number of parameters has to satisfy only two boundary conditions @xmath98 , @xmath45 and the adiabaticity condition @xmath106 .
a similar argument can be applied to the quantum analogue of the harmonic potential with time - dependent stiffness @xmath107 with hamiltonian @xmath108 the control parameter @xmath3 is changed time - dependently from an initial value @xmath44 to a final value @xmath45 during a finite transition time @xmath5 , which corresponds to a time - dependent oscillator frequency @xmath109 .
it has been shown @xcite that the work distribution and therefore also the mean work only depends on a single parameter @xmath110 - 1\end{aligned}\ ] ] where @xmath111 and @xmath112 are solutions for the classical harmonic oscillator equation @xmath113 with initial conditions @xmath114 the dynamics of @xmath115 and @xmath116 is constrained by their time evolution with common protocol @xmath6 which yields the relation @xmath117 addition of the term @xmath118 on both sides and subsequent integration leads to @xmath119 where the integration constant is determined by the initial conditions ( [ eq : ham_qm_stiff_cond ] ) . using standard techniques
, we perform a minimization of @xmath120 under the constraint ( [ eq : ham_constr ] ) with respect to @xmath111 , @xmath121 , @xmath112 , and @xmath122 .
the optimality conditions then are given by @xmath123 where @xmath124 can be chosen arbitrarily .
the minimal value of the parameter @xmath120 then is @xmath125 and the optimal work is given by the adiabatic work which can be calculated @xcite as @xmath126 where @xmath127 and @xmath128 .
again , only a finite number of optimality conditions have to be fulfilled while the continuous protocol @xmath6 corresponds to an infinite number of parameters .
the optimal protocol @xmath70 thus is highly degenerate and the adiabatic work can be reached in any given ( arbitrarily short ) transition time .
we next study a simple two level system consisting of a single spin in a time dependently rotating magnetic field @xmath129 where @xmath130 is a unit vector which is to be changed from an initial value @xmath131 to a final value @xmath132 in a given finite transition time @xmath5 , keeping the absolute value of the magnetic field @xmath133 fixed . here @xmath134 denotes the unit vector in @xmath135-direction , @xmath136 .
the hamiltonian of the system is given by @xmath137 with pauli matrices @xmath138 and gyromagnetic ratio @xmath139 .
in the basis of eigenvectors of @xmath140 , the initial canonical density matrix is given by @xmath141 where @xmath142 is eigenvector of @xmath140 with eigenvalue @xmath143 , @xmath144 is eigenvector of @xmath140 with eigenvalue @xmath145 and @xmath146 is the canonical partition function .
the work in such a finite time transition then is given by @xmath147 where the time evolution of the density matrix @xmath148 is given by the liouville - von - neumann equation @xmath149 .\ ] ] quite generally , due to the adiabatic theorem in quantum mechanics , the occupation probabilities of the states do not change for quasistatic driving .
the density matrix at the end of an adiabatic transition is therefore given by @xmath150 thus , initial and final energy are equal @xmath151 and the adiabatic work is zero @xmath41 .
we next try to identify optimal protocols for such a transition .
generally , any pure spin state corresponds to a bloch vector of unit length .
applying a magnetic field @xmath152 leads to a precession of the bloch vector around the axis of the magnetic field with angular frequency @xmath153 . therefore ,
after an appropriate time @xmath154 , the direction of a bloch vector perpendicular to the magnetic field is reversed . during the adiabatic evolution from initial state @xmath155 to the final state @xmath156 , the initial bloch vectors in @xmath157 and @xmath158 directions are also reversed . in order to reach the adiabatic work in a total transition time @xmath159
, it therefore suffices to choose the protocol @xmath160 which corresponds to an initial jump to a ( constant ) magnetic field perpendicular to the @xmath101-direction , _ i.e. _ @xmath161 , and a final jump to the magnetic field @xmath162 . at the end of the transition ,
the density matrix is given by @xmath163 and the mean work thus is equal to the adiabatic work @xmath102 .
the adiabatic work thus can be reached in the finite time @xmath164 . in order to reach the adiabatic work for any larger transition time @xmath165
, it suffices to choose the protocol @xmath166 for times @xmath167 , keeping the magnetic field @xmath168 unchanged for all times @xmath169 . for transition times @xmath170 ,
the adiabatic work can not be reached since there is no possibility to obtain the density matrix @xmath156 from the given dynamics in a time @xmath170 .
however , if we had not constrained the magnetic field to have a given absolute value @xmath133 but allowed for any value of the magnetic field during the transition , we would have found that the adiabatic work can be reached in any transition time @xmath5 .
this is obvious because the spin precession frequency is proportional to @xmath133 and thus any precession frequency can be achieved in such a setup . for a given transition time @xmath5 , the magnetic field
then can be tuned to yield a precession frequency @xmath171 which leads to @xmath172 . more generally , if one allows any form of the hamiltonian , it is obvious that a work value arbitrarily close to the adiabatic work can be reached in any given ( short ) time @xmath5 for the following reason .
assume that the work @xmath173 can be reached in a slow transition with dynamics @xmath174 for the time dependent hamiltonian @xmath175 with @xmath176 .
the hamiltonian @xmath177 then leads to an equivalent dynamics @xmath178 and therefore to an equivalent final mean internal energy at time @xmath5 .
thus , the work @xmath173 ( which is arbitrarily close to @xmath69 for long times @xmath179 ) can be reached in the ( arbitrarily short ) time @xmath5 .
note , however , that this argument is no longer true if one allows only for any form of the potential @xmath180 since the kinetic energy ( and thus the dependence on the momentum @xmath181 ) in the hamiltonian @xmath182 is given and thus can not be rescaled in this manner .
we have investigated optimal protocols for a minimal mean work for both hamiltonian and schrdinger dynamics with canonical initial conditions .
this extends our previous studies of optimal protocols for overdamped and underdamped langevin dynamics @xcite where we found unexpected jumps in the overdamped case and even delta peaks in the underdamped case . for hamiltonian dynamics and harmonic potentials ,
the present optimization shows two unexpected features which do not occur for langevin dynamics : ( i ) independently of the transition time @xmath5 , the optimal work @xmath68 is given by the adiabatic work obtained for an infinitely slow transition , and ( ii ) the optimal protocol @xmath70 is highly degenerate since only two conditions on the moments of the probability distribution have to be fulfilled . for hamiltonian dynamics in an anharmonic potential
, we performed numeric calculations of the optimal protocol in a restricted state space .
we found that the optimal work is well below the work for an instantaneous jump even for very fast transitions and we thus conjecture that the optimal work is quite generally discontinuous at @xmath183 .
however , we could not decide from our numerical data whether the adiabatic work can be obtained in an arbitrarily short time @xmath5 also for anharmonic potentials . in order to decide this question conclusively , extensive numerics is necessary which will be left for future research . for both harmonic and anharmonic potentials ,
values very close to the adiabatic work can be reached with transition times well below the typical oscillation times .
these results suggest that the efficiency of free energy calculations for hamiltonian dynamics via the jarzynski relation can be significantly improved by using the optimal protocol .
this insight is even more relevant since thermodynamic integration leads to a biased estimate because of the inequality of adiabatic work and free energy difference . quite generally
, the question whether the adiabatic work can be achieved in an arbitrarily short time will depend on the dependence of the potential @xmath184 on the protocol @xmath6 . in the harmonic case studies ,
any constraint on the maximal possible absolute value of @xmath6 for hamiltonian dynamics would lead to optimal work values well above the adiabatic work for short transition times .
such a constraint could , _
e. g. _ , be implemented by replacing @xmath185 , which yields values between @xmath186 and @xmath187 for any value of @xmath188 .
consequently , the question arises whether the adiabatic work can be reached in an arbitrary short time using any type of intermediate potentials which does include not only those compatible with a given functional form @xmath184 .
it is obvious that , if one allowed also for a time - dependent mass @xmath189 , the adiabatic work could be reached in any given time , in complete analogy to the discussion in the quantum case . for quantum systems in an initially thermalized state ,
minimal work protocols show the same features as in the classical case for harmonic potentials . for a two level system which could serve as a paradigm for a finite quantum system , it depends on the allowed parameter space for the `` potential '' , _ i. e. _ the magnetic field , whether the adiabatic work can be reached in an arbitrarily short time .
generally , the adiabatic work can be reached in arbitrarily short time if any functional form of the time - dependent hamiltonian @xmath182 is allowed . for most practical purposes
, the range of the control parameter will be limited by the experimental conditions . for quantum systems , refined strategies for the optimal control of the time evolution of wave functions under such constraints
have been developped @xcite .
these techniques may be generalized in order to specify minimum work protocols for quantum nonequilibrium transitions in complex systems with additional constraints on the possible values of the control parameter @xmath6 .
10 c. bustamante , j. liphardt , and f. ritort , `` the nonequilibrium thermodynamics of small systems '' , physics today * 58(7 ) * , 43 ( 2005 ) .
u. seifert , `` stochastic thermodynamics : principles and perspectives '' , eur .
j. b * 64 * , 423 ( 2008 ) .
d. j. evans , e. g. d. cohen , and g. p. morriss , `` probability of second law violations in shearing steady states '' , phys . rev
. lett . * 71 * , 2401 ( 1993 ) .
g. gallavotti and e. g. d. cohen , `` dynamical ensembles in nonequilibrium statistical mechanics '' , phys .
lett . * 74 * , 2694 ( 1995 ) . c. jarzynski , `` nonequilibrium equality for free energy differences '' , phys .
lett . * 78 * , 2690 ( 1997 ) .
g. e. crooks , `` path - ensemble averages in systems driven far from equilibrium '' , phys .
e * 61 * , 2361 ( 2000 ) .
u. seifert , `` entropy production along a stochastic trajectory and an integral fluctuation theorem . '' , phys .
lett . * 95 * , 040602 ( 2005 ) . c. jarzynski ,
`` equilibrium free - energy differences from nonequilibrium measurements : a master - equation approach '' , phys . rev .
e * 56 * , 5018 ( 1997 ) . c. jarzynski ,
`` hamiltonian derivation of a detailed fluctuation theorem '' , j. stat .
phys . * 98 * , 77 ( 2000 ) .
r. kawai , j. m. r. parrondo , and c. v. den broeck , `` dissipation : the phase - space perspective '' , phys .
lett . * 98 * , 080602 ( 2007 ) .
s. x. sun , `` generating generalized distributions from dynamical simulation '' , j. chem .
phys . * 118 * , 5769 ( 2003 ) .
h. oberhofer , c. dellago , and p. l. geissler , `` biased sampling of nonequilibrium trajectories : can fast switching simulations outperform conventional free energy calculation methods ? '' , j. phys .
b * 109 * , 6902 ( 2005 ) .
s. vaikuntanathan and c. jarzynski , `` escorted free energy simulations : improving convergence by reducing dissipation '' , phys .
lett . * 100 * , 190601 ( 2008 ) .
a. e. allahverdyan and t. m. nieuwenhuizen , `` minimal - work principle and its limits for classical systems '' , phys .
e * 75 * , 051124 ( 2007 ) .
s. mukamel , `` quantum extension of the jarzynski relation : analogy with stochastic dephasing '' , phys .
lett . * 90 * , 170604 ( 2003 ) . w. de roeck and c. maes , `` quantum version of free - energy - irreversible - work relations '' , phys .
e * 69 * , 026115 ( 2004 ) .
a. engel and r. nolte , `` failure of jarzynski identity for a simple quantum system '' , epl * 79 * , 10003 ( 2007 ) .
p. talkner , e. lutz , and p. hnggi , `` fluctuation theorems : work is not an observable '' , phys .
e * 75 * , 050102 ( 2007 ) .
g. e. crooks , `` on the jarzynski relation for dissipative quantum dynamics '' , j. stat .
: theor . exp . , p10023 ( 2008 ) .
m. esposito , u. harbola , and s. mukamel , `` nonequilibrium fluctuations , fluctuation theorems , and counting statistics in quantum systems '' , arxiv:0811.3717 , ( 2008 ) .
j. teifel and g. mahler , `` model studies on the quantum jarzynski relation '' , phys .
e * 76 * , 051126 ( 2007 ) .
s. deffner and e. lutz , `` nonequilibrium work distribution of a quantum harmonic oscillator '' , phys . rev .
e * 77 * , 021128 ( 2008 ) .
p. talkner , b. p. s , and h. p , `` statistics of work performed on a forced quantum oscillator '' , phys . rev .
e * 78 * , 011115 ( 2008 ) .
t. schmiedl and u. seifert , `` optimal finite - time processes in stochastic thermodynamics '' , phys .
* 98 * , 108301 ( 2007 ) .
a. gomez - marin , t. schmiedl , and u. seifert , `` optimal protocols for minimal work processes in underdamped stochastic thermodynamics '' , j. chem . phys . * 129 * , 024114 ( 2008 ) .
d. hennig , l. schimansky - geier , and p. hnggi , `` self - organized , noise - free escape of a coupled nonlinear oscillator chain '' , epl * 78 * , 20002 ( 2007 ) .
m. campisi , `` increase of boltzmann entropy in a quantum forced harmonic oscillator '' , phys .
e * 78 * , 011115 ( 2008 ) . in ref .
@xcite , talkner et al .
discuss the case of a time - dependent , spatially constant force acting on a quantum harmonic oscillator .
this additional force tilts the potential and leads to both a moving potential minimum and a lowering of the potential .
this situation is slightly different from the moving harmonic potential where the minimum of the potential is only moved horizontally . however , both hamiltonians can be transferred into each other by adding a term proportional to the square of the force .
s. h. shi , a. woody , and h. rabitz , `` optimal - control of selective vibrational - excitation in harmonic linear - chain molecules '' , j. chem .
phys . * 88 * , 6870 ( 1988 ) .
r. kosloff , s. a. rice , p. gaspard , s. tersigni , and d. j. tannor , `` wavepacket dancing : achieving chemical selectivity by shaping light pulses '' , chem . phys . * 139 * , 201 ( 1989 ) .
r. s. judson and h. rabitz , `` teaching lasers to control molecules '' , phys .
lett . * 68 * , 1500 ( 1992 ) .
y. j. yan , r. e. gillilan , r. m. whitnell , k. r. wilson , and s. mukamel , `` optical control of molecular - dynamics - liouville - space theory '' , j. phys .
chem . * 97 * , 2320 ( 1993 ) .
r. j. gordon and s. a. rice , `` active control of the dynamics of atoms and molecules '' , annu .
phys . chem . *
48 * , 601 ( 1997 ) . | for systems in an externally controllable time - dependent potential , the optimal protocol minimizes the mean work spent in a finite - time transition between given initial and final values of a control parameter . for an initially thermalized ensemble ,
we consider both hamiltonian evolution for classical systems and schrdinger evolution for quantum systems . in both cases
, we show that for harmonic potentials , the optimal work is given by the adiabatic work even in the limit of short transition times .
this result is counter - intuitive because the adiabatic work is substantially smaller than the work for an instantaneous jump .
we also perform numerical calculations of the optimal protocol for hamiltonian dynamics in an anharmonic quartic potential . for a two - level spin system ,
we give examples where the adiabatic work can be reached in either a finite or an arbitrarily short transition time depending on the allowed parameter space . |
recently , there has been significant experimental improvements in the measurements of flavor - changing neutral current ( fcnc ) processes of @xmath4 mesons at b factories and tevatron . for the @xmath5 transition , experimental data for @xmath2 and @xmath6 ( @xmath7 ) decays , @xmath8 oscillation , and @xmath3 decay have already started to constrain possible contributions from new physics beyond the standard model . here
we focus our attention to one of the @xmath5 processes , the decay into neutrino pairs @xcite , bs| .
[ eq - bsnunu ] it is known that the decays of the b mesons induced by the partonic process ( [ eq - bsnunu ] ) , especially the inclusive branching ratio @xmath9 , have small theoretical uncertainty due to the absence of photonic penguin and strong suppression of light quark contributions . on the other hand , experimental search of the decay ( [ eq - bsnunu ] ) is a hard task . at present , only the upper bounds are known for both inclusive @xcite and exclusive @xcite branching ratios , at 90% c.l . , & & _
br(|bx_s|)<6.410 ^ -4 , + & & _
br(b^+k^+ |)<1.410 ^ -5 , + & & _
br(|b^0k^0_s |)<1.610 ^ -4 , + & & _ br(|b^0k^*0|)<3.410 ^ -4 , + & & _
br(b^+k^*+ |)<1.410 ^ -4 , which are still one order of magnitude larger than the standard model predictions for the inclusive @xcite and exclusive @xcite modes , & & _ br(|bx_s|)_sm= ( 3.70.2)10 ^ -5 , + & & _ br(|bk|)_sm= ( 3.8^+1.2_-0.6 ) 10 ^ -6 , + & & _ br(|bk^*|)_sm= ( 1.3^+0.4_-0.3 ) 10 ^ -5 .
[ eq - bsnunusm ] future upgrades of the b factories @xcite will extend the search region for the exclusive decays .
for example , @xmath10 around the level of the standard model prediction ( [ eq - bsnunusm ] ) is expected to be measured at the precision of 20% with integrated luminosity 50100 ab@xmath11 . on the other hand , a future @xmath12 collider running on the @xmath13-boson resonance ( giga - z ) has a potential @xcite to produce very large number of @xmath14 events , and possibility to greatly improve previous studies of the inclusive modes @xcite at the lep i , to measure the inclusive branching ratio . in this paper
, we consider the decay ( [ eq - bsnunu ] ) in the framework of the minimal supersymmetric standard model ( mssm ) @xcite with general flavor mixing of squarks , and study the contributions of new particles , namely the supersymmetric ( susy ) particles and higgs bosons . in cases where the value of @xmath1 , the ratio of the vacuum expectation values of two higgs boson doublets in the mssm , is not much larger than unity , it is shown @xcite that the chargino - squark loops give main part of the new physics contributions to the decay , and may become sizable when large flavor mixing is present in the left - right mixing part of the up - type squark mass matrix . note that this is also the case for the susy contributions to the related decays @xmath15 @xcite . at large @xmath1 , say similar to or larger than @xmath16 , the mssm loop contributions other than charginos might become also important , at least in principle .
for example , gluino - squark loop contributions are generated by @xmath1-enhanced large left - right mixing of down - type squarks .
when , in addition , sizable mixing between down - type squarks in the second and third generations are present , gluino contribution might become sizable .
it is also possible that , as explained later , charged higgs boson might give sizable loop contributions due to the flavor - changing effective higgs - quark couplings , generated by @xmath17 susy loop corrections , as pointed out in ref .
@xcite for the @xmath15 decays .
however , parameters in the susy and higgs sectors should receive stringent constraints from existing measurements of the fcnc processes , which might suppress possible magnitudes of their contributions to @xmath0 . in this paper
, we will present a rough estimate of the possible constraints from the decay @xmath2 , by showing correlations between the new physics contributions to the wilson coefficients for @xmath0 and those for @xmath2 , for each susy / higgs sector separately .
we will also comment on the implication of the @xmath3 decay to the higgs boson contributions .
the paper is organized as follows . in sec .
[ sect - bsnunu ] , we present basic formulas for the analysis of the @xmath0 decay in the mssm . in sec .
[ sect - mssmcontr ] , numerical results for the new physics contributions in the mssm to @xmath0 are presented as correlations with those to @xmath2 for each new physics sector . an additional constraint from the @xmath3 decay on the higgs boson contributions is briefly commented in sec .
[ sect - bsmumu ] .
finally , conclusion is given in sec .
[ sect - conclu ] .
the @xmath0 decay is described by the effective hamiltonian , in the notation of ref .
@xcite , h_eff = - k^*_tsk_tb [ c_o_l + c_o_r ] , [ heff ] where @xmath18 is the cabibbo - kobayashi - maskawa ( ckm ) matrix . here
the relevant operators are _ l & = & ( |s_l^b_l)(|_l__l ) , [ eq - onunul ] + o_r & = & ( |s_r^b_r)(|_l__l ) .
[ eq - onunur ] the inclusive branching ratio is then expressed in terms of the wilson coefficients ( @xmath19 , @xmath20 ) in eq .
( [ heff ] ) as @xcite _ br(|bx_s|)~ ( |bx_c e|_e ) ( |c_|^2 + |c_|^2 ) , [ eq - inclwidth ] up to the qcd corrections and @xmath21 corrections to the semileptonic decays @xmath22 .
interference between @xmath19 and @xmath20 appears in the branching ratios of the exclusive modes @xmath23 @xcite . note that , in the massless quark limit , @xmath24 are independent of the renormalization scale in qcd . in the mssm , the interaction ( [ heff ] )
is generated by the @xmath13-boson penguin and box diagrams .
the standard model particles only contribute to @xmath19 , giving at the leading order in qcd @xcite , c_,sm = - , where @xmath25 .
numerically , @xmath26 is about @xmath27 for @xmath28 gev .
new particles in the mssm , namely the susy particles and higgs bosons , may contribute to both @xmath19 and @xmath20 , & & c_=c_,sm+c_(new ) , c_=c_(new ) , + & & c^()_(new ) = c^( ) _ , + c^()_,^+c^()_,^0+c^()_,h^. @xmath29 consists of the contributions of the gluino @xmath30 - down type squark loops , chargino @xmath31 - up - type squark loops , neutralino @xmath32 - down - type squark loops , and charged higgs boson @xmath33 - top quark loops .
below we list the analytic forms of these one - loop contributions for each sector : c _ , & = & - ( ^_dl)_2i(_dr)_ik ( ^_dr)_kj(_dl)_j3 c_24(_i,_j , ) , [ clnugl ] + c _ , & = & ( ^_dr)_2i(_dl)_ik ( ^_dl)_kj(_dr)_j3 c_24(_i,_j , ) , [ crnugl ] + c_,^ & = & + & & + u_k1u^*_l1m_^_km_^_l d_0(_i,^_k,^_l,^- ) , [ clnuch ] + c_,^ & = & + & & - u_k1u^*_l1 d_27(_i,^_k,^_l,^- ) , [ crnuch ] + c_,^0 & = & + & & + , + [ clnune ] + c_,^0 & = & + & & - , + [ crnune ] + c_,h^ & = & ( 1-x_th+x_th ) , [ clnuhp ] + c_,h^ & = & - ( 1-x_th+x_th ) , [ crnuhp ] where @xmath34 , @xmath35 .
we assume flavor degeneracy in the lepton and slepton sectors .
the formulas ( [ clnugl][crnuhp ] ) are derived from previous studies of the @xmath0 decays in the mssm @xcite , as well as related works on the @xmath15 decays @xcite . @xmath36 and @xmath37 are the three - point functions for the @xmath13-penguin diagrams and four - point functions for the box diagrams , respectively @xcite , in the convention of ref . @xcite
. ultraviolet divergence of @xmath38 cancels out in the formulas ( [ clnugl][crnune ] ) .
we ignore the masses of @xmath39 quarks , and include those of @xmath40 only when they are multiplied by @xmath1 . in this approximation ,
the neutral higgs boson contributions to @xmath41 vanish . the couplings and mixing matrices in eqs .
( [ clnugl][crnuhp ] ) are given as follows : the squark mixing matrices @xmath42 @xmath43 give relations between the mass eigenstates @xmath44 to the gauge eigenstates in the `` super - ckm '' basis @xmath45 , which are related to the mass eigenbasis of the quarks @xmath46 $ ] by susy transformation , as @xmath47 these matrices are determined to diagonalize the @xmath48 mass matrices of squarks in the super - ckm basis , m_^2 & = & ( cc m^2 _ ll & ( m^2 _ rl)^ + m^2 _ rl & m^2 _ rr ) , + ( m^2 _ ll ) _ & = & ( m^2 _ ll ) _ + ( m_q^(0))^(m_q^(0))+ _ ( i_3q_l - e_q^2_w)m_z^22 , + ( m^2 _ rr ) _ & = & ( m^2 _ rr ) _
+ ( m_q^(0))(m_q^(0))^+ _ e_q^2_w m_z^22 , + ( m^2 _ rl ) _ & = & ( m^2 _ rl ) _ - m_u^(0)^ * , + ( m^2 _ rl ) _ & = & ( m^2 _ rl ) _ - m_d^(0 ) ^*. [ msq2 ] in eq .
( [ msq2 ] ) , off - diagonal elements of the soft susy breaking mass matrices @xmath49 induce flavor mixings which are not constrained by the ckm matrix in general , and may cause potentially large fcnc .
@xmath50 are the `` bare '' mass matrices of the quarks . for the up - type squarks
, it is just the running mass matrix @xmath51 in the standard model . for the down - type squarks , in contrast
, @xmath52 may substantially deviate from the standard model mass matrix @xmath53 , as explained later .
the quark - squark - chargino and quark - squark - neutralino couplings @xmath54 are then given in terms of the mixing matrices for squarks ( [ eq - gammasq ] ) , for charginos @xmath55 , and for neutralinos @xmath56 @xcite , as a^c_ik & = & g_2 ( _ ul)_i v_k1^ * k _ - h_t ( _ ur)_i3 v_k2^ * k_t , + b^c_ik & = & -(_ul)_i u_k2 k _ ( _ d)^ * _ , + a^n_ik & = & ( - n_k2^ * + n_k1^ * ) ( _ dl)_i + ( _ d ) _ n_k3^ * ( _ dr)_i , + b^n_ik & = & n_k1 ( _ dr)_i + ( _ d)^ * _ n_k3 ( _ dl)_i , [ eq - abino ] finally , @xmath57 in eqs . ( [ clnune ] , [ crnune ] ) denote the neutrino - sneutrino - neutralino couplings .
we need some explanation for @xmath58 , the bare yukawa coupling matrix for down - type quarks .
we start from the effective lagrangian for the couplings of @xmath59 to the higgs boson doublets ( @xmath60 , @xmath61 ) in the mssm , after integrating out the susy particles , _ eff = -(_d)_ij |d_ir ( d_jl h_d^0 - k^*_kju_kl h_d^- ) -(y_d)_ij |d_ir ( d_jl h_u^0 * + k^*_kju_kl h_u^- ) + ( h.c ) .
[ eq - leff ] the couplings @xmath62 are forbidden at the tree - level by supersymmetry , but induced by susy particle loops with soft susy breaking .
the running mass matrix in the standard model @xmath53 is then given by ( m_d ) _ & = & _ , + & & [ m_d^(0 ) + m_d ] _ .
[ eq - md ] although the loop - generated @xmath63 is suppressed relative to the tree - level coupling @xmath64 , its contribution to @xmath65 , @xmath66 , is enhanced by @xmath1 , as seen in eq .
( [ eq - md ] ) , and may become numerically comparable to the tree - level part @xmath67 at large @xmath1 @xcite . on the other hand ,
the couplings of ( @xmath59 , @xmath68 ) to heavier higgs bosons ( @xmath69 , @xmath70 , @xmath33 ) and higgsinos @xmath71 are determined by @xmath64 , as shown in eqs .
( [ crnuhp ] , [ eq - abino ] ) , without @xmath1-enhanced contributions from @xmath63 . as a consequence , at large @xmath1
, these couplings may significantly deviate from the tree - level values @xcite given in terms of @xmath72 and , since @xmath63 is not flavor diagonal in general , include flavor - mixing parts not determined by the ckm matrix , even in the super - ckm basis .
the bare quark mass matrix @xmath73 should be also used in the mass matrix ( [ msq2 ] ) of the down - type squarks , which also receives no contributions from @xmath63 .
the correction ( [ eq - md ] ) therefore affects the masses and mixing matrices @xmath74 of the down - type squarks , generating additional flavor mixing for squarks .
these @xmath1-enhanced corrections to the down - type quarks and squarks are often comparable to the tree - level contributions in the mssm at large @xmath1 , and should be included in realistic analysis of processes involving these particles @xcite .
now we turn to the behavior of the susy and higgs contributions ( [ clnugl][crnuhp ] ) to @xmath75 .
the main part of these contributions comes from the @xmath13 penguin diagrams through effective @xmath76 and @xmath77 vertices .
appearance of these vertices needs both the mixing between the second and third generations of quarks / squarks , and the su(2)@xmath78u(1 ) gauge symmetry breaking in the loops . for small or moderate value of @xmath1 , the largest su(2 ) breaking in the loops
are provided by the top quark and squarks . as a consequence , @xmath79
( [ clnuhp ] ) and @xmath80 ( [ clnuch ] ) are relevant .
the former , however , is suppressed by @xmath81 and only relevant for @xmath82 , which is disfavored by experimental lower limit on the mass of the lightest higgs boson .
therefore , only the latter , @xmath80 , is left as a potentially important susy contribution to @xmath0 .
previous studies have shown @xcite that @xmath83 is enhanced by large @xmath84 , especially by its flavor - mixing parts .
similar behavior is observed for the chargino contributions to the @xmath15 decays @xcite . at large @xmath1 ,
however , other contributions to @xmath0 have the possibility to become sizable , by the following reasons : first , the su(2)-breaking left - right mixing of the down - type squarks @xmath85 increases as @xmath1 and may enhance the gluino contribution .
second , off - diagonal parts of the effective yukawa coupling @xmath64 in eq .
( [ eq - md ] ) induce the flavor - changing couplings of the down - type quarks , which are enhanced by @xmath1 and not necessarily suppressed by the corresponding ckm matrix elements or quark masses . especially , the element @xmath86 , induced by flavor mixing in @xmath87 , might give large yukawa couplings of @xmath88 and enhance @xmath89 .
this is similar to the case of @xmath15 at large @xmath1 @xcite , where loop - induced couplings ( @xmath90 , @xmath86 ) give large effective @xmath91 coupling .
therefore , the gluino ( [ clnugl ] , [ crnugl ] ) and charged higgs boson ( [ crnuhp ] ) contributions must be considered in the analysis of @xmath0 at large @xmath1 .
we present numerical results for the new physics contributions ( [ clnugl][crnuhp ] ) to the @xmath0 decay in the mssm .
we concentrate on the cases with large @xmath1 , which were not considered in previous studies . in the estimation of possible magnitudes of the new physics contributions ( [ clnugl][crnuhp ] ) to @xmath0
, we need to take into account the constraints on susy and higgs parameters from other fcnc processes . in this section
, we consider the implication of the constraints from the radiative decay @xmath2 .
this constraint is expected to be crucial since the su(2)@xmath78u(1 ) breaking and flavor mixing between quarks / squarks in the second and third generations , which are necessary to enhance the contributions to @xmath0 , may also give large contributions to @xmath2 .
another reason to focus on @xmath2 is the rather good agreement between experimental data @xcite and the standard model prediction @xcite of the inclusive branching ratio @xmath92 .
indeed , the decay @xmath2 in the mssm have been shown to give strong constraints on the higgs and susy parameters .
it should also be noted that the susy contributions to @xmath2 are enhanced by @xmath1 @xcite .
here we do not attempt precise calculation of the experimental constraints from @xmath2 .
instead , we present a very rough estimation of the expected constraints in terms of the wilson coefficients @xmath93 for @xmath2 , defined as & & h_eff = -k^*_tsk_tb ( c_7 ( ) o_7 ( ) + c_7 ( ) o_7 ( ) ) , + & & o_7 = m_b ( ) ( |s_l^b_r)f _ , + & & o_7 = m_b ( ) ( |s_r^b_l)f _ .
[ eq - bsg ] below we show the correlations between @xmath29 , eqs .
( [ clnugl][crnuhp ] ) , and new physics contributions to @xmath94 , @xmath95 , for each sector of new physics : namely , the gluino - squark , chargino - squark , and charged higgs boson - top quark loop contributions , varying squarks mixing parameters which are relevant to @xmath0 . for simplicity
, we assume the flavor structures of the soft susy breaking terms in the squark mass matrices ( [ msq2 ] ) as m^2 _ xx & = & m_^2 ( ccc 1 & 0 & 0 + 0 & 1 & ( _ xx^q)_23 + 0 & ( _ xx^q)_23 & 1 ) ( xx = ll , rr ) , [ msq2ll ] + m^2 _ rl & = & m_t ( ccc 0 & 0 & 0 + 0 & 0 & 0 + 0 & ( a_u)_32 & ( a_u)_33 ) , [ msu2rl ] since cp violation is not essential for the analysis in this paper , all susy and higgs parameters , including those in eqs .
( [ msq2ll ] , [ msu2rl ] ) are set to be real .
we also set @xmath96 in eq .
( [ msq2 ] ) since its contribution to @xmath97 is , when the vacuum stability bounds @xcite is applied , @xmath98 and subdominant compared to the second term @xmath99 .
note that the condition ( [ msq2ll ] ) for @xmath100 may be imposed only either @xmath101 or @xmath102 , due to the su(2 ) symmetry @xmath103 .
we calculate the new physics contributions to @xmath104 and @xmath94 at the leading one - loop order ( see refs .
@xcite for the formulas of @xmath94 ) , but improved by including the @xmath1-enhanced corrections to the quark / squark yukawa couplings from eq .
( [ eq - md ] ) and , for @xmath94 , also from the proper vertex corrections .
however , we ignored the corrections in eq .
( [ clnuhp ] ) , since @xmath79 itself is strongly suppressed by @xmath81 and numerically negligible . ] to the @xmath105 couplings to @xmath106 @xcite , in the effective lagrangian formalism @xcite . in these formulas
, we use the running quark masses and @xmath107 at the renormalization scale @xmath108 , calculated from @xmath109 gev , @xmath110 gev , @xmath111 mev , @xmath112 and @xmath113 , which give @xmath114 and @xmath115 at the renormalization scale @xmath108 . for susy and higgs parameters ,
we fix the following parameters : @xmath116 , @xmath117 gev , @xmath118 gev , @xmath119 gev , @xmath120 gev , while varying other parameters .
we also impose the bounds @xmath121 gev and @xmath122 gev , suggested by experimental search limits for susy particles . for each sector of the new physics , rough estimates of the bounds on the contributions to @xmath24
are obtained by requiring that the magnitudes of @xmath123 should be smaller than the standard model contribution @xmath124 .
the gluino - squark contributions @xmath125 are induced by the flavor and left - right mixing of the down - type squarks . in fig .
[ fig - gluino ] , the gluino contribution @xmath126 is shown as a correlation with @xmath127 , for parameter scan over @xmath128 $ ] , @xmath129 $ ] , and @xmath130 $ ] gev . @xmath131 and @xmath132 are set to 0 .
correlation between @xmath133 and @xmath134 for the same parameters is obtained from fig .
[ fig - gluino ] by changing the sign of the horizontal axis .
large @xmath135 is obtained for large negative @xmath136 and large @xmath137 , which cause large @xmath138 mixing .
it is seen that @xmath135 can be larger than @xmath139 , which gives about @xmath140 % correction to the standard model prediction of the decay width ( [ eq - inclwidth ] ) .
however , by requiring @xmath141 , magnitudes of @xmath142 are constrained to be much smaller than @xmath143 .
therefore , without very precise cancellation between new physics contributions to @xmath2 , gluino contributions to @xmath0 should be completely negligible , even for @xmath144 , to satisfy the bound from @xmath2 . here
we briefly comment on the neutralino contributions @xmath145 , eqs .
( [ clnune ] , [ crnune ] ) .
similar to the gluino contributions , @xmath145 are induced by the @xmath146 mixing in the loops . however , due to small couplings , these contributions are much smaller than the gluino contributions @xmath125 for most parameter regions and therefore not discussed here .
the chargino - squark loop contributions @xmath83 , eq .
( [ clnuch ] ) , have been studied in previous works @xcite at small or moderate value of @xmath1 . in these works
, it has been shown that they might give sizable contributions , larger than the uncertainty of the standard model predictions ( [ eq - bsnunusm ] ) , for large flavor - mixing element of @xmath147 in eq .
( [ msu2rl ] ) , especially its ( @xmath148 , @xmath149)-mixing element @xmath150 .
figure [ fig - chargino ] shows the correlation between @xmath80 and @xmath151 , for varying parameters over @xmath152 $ ] gev , @xmath153 $ ] gev , and @xmath154 $ ] .
other parameters are fixed at @xmath155 gev , @xmath156 gev , and @xmath157 .
for these parameters , @xmath158 is negligibly small ( @xmath159 ) and not shown here . as is the case of the gluino contributions , susy parameters which give large @xmath83
tend to also give large @xmath160 .
the resulting constraint on @xmath83 gets tighter as @xmath1 increases , since @xmath160 are enhanced by @xmath1 while @xmath83 is not . nevertheless , the correlation is not so strong as in the gluino sector , as seen in fig .
[ fig - chargino ] .
this is due to the different dependences of @xmath83 and @xmath160 on two a - term elements , @xmath131 and @xmath132 in eq .
( [ msu2rl ] ) .
in fact , as seen in fig . [ fig - chargino ] , we may have @xmath161 while keeping @xmath162 .
even larger value of @xmath83 might be possible by careful choice of the susy parameters .
the resulting deviations of the decay widths from the standard model predictions ( [ eq - bsnunusm ] ) could be proved at future b factories , if the theoretical uncertainties of the exclusive widths in eq .
( [ eq - bsnunusm ] ) , mainly coming from the meson form factors , are reduced .
however , one must note that the large chargino contribution is realized by the fine tuning between susy parameters , especially @xmath131 and @xmath132 , to realize small @xmath160 . as discussed in the previous section , only @xmath89 , eq .
( [ crnuhp ] ) , is relevant at large @xmath1 .
this contribution comes from the @xmath163 coupling @xmath164 , which is generated by the flavor mixing involving @xmath165 through the @xmath1-enhanced loop corrections ( [ eq - md ] ) . in fig .
[ fig - hp ] , we show the correlations between @xmath89 and @xmath166 at @xmath167 gev , @xmath168 gev , @xmath169 gev , @xmath170 $ ] , @xmath171 $ ] , and @xmath172 $ ] gev .
in contrast to the gluino and chargino contributions to @xmath2 , the main parts of @xmath166 are not enhanced by @xmath1 .
moreover , the correlations between @xmath89 and @xmath173 is severely affected by different parameter dependences of two generation - mixing @xmath33 couplings : the effective @xmath174 coupling @xmath175 in @xmath89 and @xmath176 , and @xmath17 proper vertex corrections to the @xmath177 coupling @xcite in @xmath173 . as a result , similar to the case of chargino contributions , there is possiblity to have sizable @xmath89 while keeping @xmath166 small .
in addition to @xmath2 , several other @xmath5 fcnc processes have been measured in recent experiments .
since most of these measurements show rather good consistency with the standard model predictions , they should give additional constraints on the susy and higgs parameters , and their contributions to @xmath0 .
for example , measurements of the @xmath8 oscillation @xcite impose constraints on the @xmath146 mixing , especially on @xmath178 and @xmath179 @xcite .
here we just show a case of these constraints : implication of the upper limit of the branching ratio for @xmath3 on the @xmath33 contributions @xmath89 , at large @xmath1 . as seen in eq .
( [ crnuhp ] ) , large value of @xmath89 is obtained for parameters which give large effective @xmath180 yukawa coupling @xmath181 . as discussed in sect
. [ sect - bsnunu ] , the parameter @xmath86 also gives the flavor - changing @xmath182 couplings of the heavier neutral higgs bosons @xmath183 . on the other hand , at large @xmath1
, this coupling gives `` tree - level '' contributions to the @xmath3 decay by the higgs penguin diagrams @xcite , which are often much larger than the standard model contributions by orders of magnitude .
requiring that these higgs penguin contributions do not saturate the experimental upper bound @xmath184 at 95% c.l .
@xcite , and neglecting mass difference between ( @xmath69 , @xmath70 ) , the condition |(_d)_32|^2+|(_d)_23|^2 < 0.2 ^ 2 ( m_a/500gev)^4 , [ eq - bsmumubound ] is imposed on the @xmath185 mixing yukawa couplings ( @xmath186 , @xmath86 ) at the renormalization scale @xmath187 . in the approximation of neglecting the qcd running between @xmath188 and @xmath189 , and also the @xmath17-enhanced correction to the @xmath190 coupling @xmath191 , eq .
( [ eq - bsmumubound ] ) implies the bound @xmath192 for @xmath116 and @xmath193 gev , which is completely negligible compared to @xmath26 .
we expect that this strong constraint still holds when more rigorous estimation of @xmath3 is adopted .
we have studied the flavor - changing decay @xmath0 in the mssm , at large @xmath1 and with general flavor mixing of squarks .
this case is interesting since the gluino and @xmath33 loops , which are negligible at moderate value of @xmath1 and with minimal flavor violation for squarks , are enhanced and might give contributions to this decay , comparable to the standard model and chargino loop contributions .
this is due to the @xmath1-enhanced su(2)@xmath78u(1 ) gauge symmetry breaking and flavor mixing in the down - type squark sector , and loop - generated effective flavor - changing couplings of the charged higgs boson to quarks and squarks .
however , the contributions to @xmath0 by new physics should be constrained by experimental data for other @xmath5 processes . in this paper , we have focused our attention to the constraints from the radiative decay @xmath2 , since both of the @xmath0 and @xmath2 decays are enhanced by the su(2)@xmath78u(1 ) symmetry breaking and flavor mixing between the second and third generations of the quarks / squarks in the loops . as a very rough estimation of the constraints by @xmath2
, we have calculated the correlations between new physics contributions to the wilson coefficients @xmath41 and @xmath94 for the @xmath0 and @xmath2 decays , respectively , for each new physics sector : gluino - squark , chargino - squark , and charged higgs - quark loops .
calculation has been done at the leading order , but including @xmath1-enhanced corrections to the quark yukawa couplings in the loops .
it has been demonstrated that the requirement that the new physics contributions @xmath95 for each sector are smaller than @xmath194 strongly constrains the new physics contributions @xmath29 .
especially , the gluino contributions @xmath125 are suppressed much below @xmath26 due to their strong correlation with @xmath195 . in contrast , although the constraints by @xmath94 are also tight for chargino and charged higgs boson contributions , there still remains a possibility that their contributions to @xmath41 become sizable , @xmath196% , while keeping contributions to @xmath94 below @xmath194 . as an example of the constraints by other @xmath5 processes , we have also considered the higgs penguin contributions to the decay @xmath3 , which might become much larger than the standard model contribution at large @xmath1 .
it has been shown that the present experimental upper bound of the decay ratio may impose strong constraints on @xmath89 , suppressing it much below @xmath26 . for more realistic analysis of @xmath0 in the mssm and estimation of the new physics contributions ,
we need to scan over wider parameter space , including correlations between different contributions to @xmath41 , main parts of the qcd corrections and hadronic effects , and constraints from other flavor - changing processes using more precise formulas of the new physics contributions .
we leave such studies for future works .
the author thanks francesca borzumati for earlier collaboration .
the work was supported in part by the grant - in - aid for scientific research on priority areas from the ministry of education , culture , sports , science and technology of japan , no .
16081202 and 17340062 .
g. buchalla and a. j. buras , nucl .
b * 398 * , 285 ( 1993 ) ; g. buchalla and a. j. buras , nucl .
b * 400 * , 225 ( 1993 ) ; m. misiak and j. urban , phys .
b * 451 * , 161 ( 1999 ) [ arxiv : hep - ph/9901278 ] ; g. buchalla and a. j. buras , nucl .
b * 548 * , 309 ( 1999 ) [ arxiv : hep - ph/9901288 ] .
p. colangelo , f. de fazio , p. santorelli and e. scrimieri , phys .
b * 395 * , 339 ( 1997 ) [ arxiv : hep - ph/9610297 ] ; d. melikhov , n. nikitin and s. simula , phys . lett .
b * 410 * , 290 ( 1997 ) [ arxiv : hep - ph/9704268 ] ; d. melikhov , n. nikitin and s. simula , phys .
d * 57 * , 6814 ( 1998 ) [ arxiv : hep - ph/9711362 ] ; d. melikhov , n. nikitin and s. simula , phys .
b * 428 * , 171 ( 1998 ) [ arxiv : hep - ph/9803269 ] ; t. m. aliev , a. zpineci and m. savci , phys .
b * 506 * , 77 ( 2001 ) [ arxiv : hep - ph/0101066 ] .
r. barate _
[ aleph collaboration ] , eur .
j. c * 19 * , 213 ( 2001 ) [ arxiv : hep - ex/0010022 ] .
k. f. chen _ et al .
_ [ belle collaboration ] , phys . rev .
* 99 * , 221802 ( 2007 ) [ arxiv:0707.0138 [ hep - ex ] ] . c. bobeth _
et al . _ ,
b * 726 * , 252 ( 2005 ) [ arxiv : hep - ph/0505110 ] .
g. buchalla , g. hiller and g. isidori , phys .
d * 63 * , 014015 ( 2001 ) [ arxiv : hep - ph/0006136 ] .
a. g. akeroyd _ et al .
_ [ superkekb physics working group ] , arxiv : hep - ex/0406071 ; j. l. hewett _ et al .
_ , arxiv : hep - ph/0503261 ; m. bona _ et al .
_ , arxiv:0709.0451 [ hep - ex ] .
a. ali , d. benson , i. i. y. bigi , r. hawkings and t. mannel , arxiv : hep - ph/0012218 .
h. p. nilles , phys . rep .
* 110 * , 1 ( 1984 ) ; h. e. haber and g. l. kane , phys . rep . * 117 * , 75 ( 1985 ) .
s. bertolini , f. borzumati , a. masiero and g. ridolfi , nucl . phys .
b * 353 * , 591 ( 1991 ) .
y. grossman , z. ligeti and e. nardi , nucl .
b * 465 * , 369 ( 1996 ) [ erratum - ibid .
b * 480 * , 753 ( 1996 ) ] [ arxiv : hep - ph/9510378 ] ; t. goto , y. okada , y. shimizu and m. tanaka , phys .
d * 55 * , 4273 ( 1997 ) [ erratum - ibid .
d * 66 * , 019901 ( 2002 ) ] [ arxiv : hep - ph/9609512 ] ; a. j. buras , p. gambino , m. gorbahn , s. jger and l. silvestrini , nucl .
b * 592 * , 55 ( 2001 ) [ arxiv : hep - ph/0007313 ] ; c. bobeth , a. j. buras , f. krger and j. urban , nucl .
b * 630 * , 87 ( 2002 ) [ arxiv : hep - ph/0112305 ] .
s. bertolini and a. masiero , phys .
b * 174 * , 343 ( 1986 ) ; g. f. giudice , z. phys .
c * 34 * , 57 ( 1987 ) ; i. i. y. bigi and f. gabbiani , nucl .
b * 367 * , 3 ( 1991 ) ; g. couture and h. knig , z. phys .
c * 69 * , 167 ( 1995 ) [ arxiv : hep - ph/9503299 ] .
y. nir and m. p. worah , phys .
b * 423 * , 319 ( 1998 ) [ arxiv : hep - ph/9711215 ] ; a. j. buras , a. romanino and l. silvestrini , nucl .
b * 520 * , 3 ( 1998 ) [ arxiv : hep - ph/9712398 ] ; g. colangelo and g. isidori , j. high energy phys . * 9809 * , 009 ( 1998 ) [ arxiv : hep - ph/9808487 ] .
a. j. buras , g. colangelo , g. isidori , a. romanino and l. silvestrini , nucl .
b * 566 * , 3 ( 2000 ) [ arxiv : hep - ph/9908371 ] ; a. j. buras , t. ewerth , s. jger and j. rosiek , nucl .
b * 714 * , 103 ( 2005 ) [ arxiv : hep - ph/0408142 ] ; g. isidori , f. mescia , p. paradisi , c. smith and s. trine , j. high energy phys . * 0608 * , 064 ( 2006 ) [ arxiv : hep - ph/0604074 ] .
g. isidori and p. paradisi , phys .
d * 73 * , 055017 ( 2006 ) [ arxiv : hep - ph/0601094 ] .
t. inami and c. s. lim , prog .
phys . * 65 * , 297 ( 1981 ) [ erratum - ibid . *
65 * , 1772 ( 1981 ) ] . g. passarino and m. j. g. veltman , nucl .
b * 160 * , 151 ( 1979 ) .
k. hagiwara , s. matsumoto , d. haidt and c. s. kim , z. phys .
c * 64 * , 559 ( 1994 ) [ erratum - ibid .
c * 68 * , 352 ( 1995 ) ] [ arxiv : hep - ph/9409380 ] .
j. f. gunion and h. e. haber , nucl .
b * 272 * , 1 ( 1986 ) [ erratum - ibid .
b * 402 * , 567 ( 1993 ) ] .
t. banks , nucl .
* b303 * , 172 ( 1988 ) ; r. hempfling , phys . rev .
d * 49 * , 6168 ( 1994 ) ; l. j. hall , r. rattazzi and u. sarid , phys . rev .
d * 50 * , 7048 ( 1994 ) [ arxiv : hep - ph/9306309 ] ; m. carena , m. olechowski , s. pokorski and c. e. m. wagner , nucl .
phys . * b426 * , 269 ( 1994 ) [ arxiv : hep - ph/9402253 ] ; t. blaek , s. raby and s. pokorski , phys . rev .
d * 52 * , 4151 ( 1995 ) [ arxiv : hep - ph/9504364 ] .
m. carena , s. mrenna and c. e. m. wagner , phys .
d * 60 * , 075010 ( 1999 ) [ arxiv : hep - ph/9808312 ] ; k. s. babu and c. f. kolda , phys . lett .
b * 451 * , 77 ( 1999 ) [ arxiv : hep - ph/9811308 ] ; f. borzumati , g. r. farrar , n. polonsky and s. thomas , nucl .
* b555 * , 53 ( 1999 ) [ arxiv : hep - ph/9902443 ] ; h. eberl , k. hidaka , s. kraml , w. majerotto and y. yamada , phys .
d * 62 * , 055006 ( 2000 ) [ arxiv : hep - ph/9912463 ] ; m. carena , d. garcia , u. nierste and c. e. m. wagner , nucl .
b577 * , 88 ( 2000 ) [ arxiv : hep - ph/9912516 ] ; h. e. haber _ et al .
_ , phys .
d * 63 * , 055004 ( 2001 ) [ arxiv : hep - ph/0007006 ] ; m. j. herrero , s. pearanda and d. temes , phys .
d * 64 * , 115003 ( 2001 ) [ arxiv : hep - ph/0105097 ] ; d. a. demir , phys .
b * 571 * , 193 ( 2003 ) [ arxiv : hep - ph/0303249 ] ; a. j. buras , p. h. chankowski , j. rosiek and .
sawianowska , phys .
b * 546 * , 96 ( 2002 ) [ arxiv : hep - ph/0207241 ] ; a. j. buras , p. h. chankowski , j. rosiek and .
sawianowska , nucl .
b * 659 * , 3 ( 2003 ) [ arxiv : hep - ph/0210145 ] ; j. foster , k. i. okumura and l. roszkowski , phys .
b * 609 * , 102 ( 2005 ) [ arxiv : hep - ph/0410323 ] ; j. foster , k. i. okumura and l. roszkowski , j. high energy phys . * 0508 * , 094 ( 2005 ) [ arxiv : hep - ph/0506146 ] ; g. isidori and p. paradisi , phys .
b * 639 * , 499 ( 2006 ) [ arxiv : hep - ph/0605012 ] ; e. lunghi , w. porod and o. vives , phys .
d * 74 * , 075003 ( 2006 ) [ arxiv : hep - ph/0605177 ] .
e. barberio _ et al .
_ [ heavy flavor averaging group ( hfag ) collaboration ] , arxiv:0704.3575 [ hep - ex ] .
m. misiak _ et al .
_ , phys .
lett . * 98 * , 022002 ( 2007 ) [ arxiv : hep - ph/0609232 ] . s. bertolini , f. borzumati and a. masiero , phys .
b * 192 * , 437 ( 1987 ) .
n. oshimo , nucl .
* b404 * , 20 ( 1993 ) ; m. a. daz , phys .
b * 304 * , 278 ( 1993 ) [ arxiv : hep - ph/9303280 ] ; r. barbieri and g. f. giudice , phys .
b * 309 * , 86 ( 1993 ) [ arxiv : hep - ph/9303270 ] ; y. okada , phys .
b * 315 * , 119 ( 1993 ) [ arxiv : hep - ph/9307249 ] ; r. garisto and j. n. ng , phys . lett .
b * 315 * , 372 ( 1993 ) [ arxiv : hep - ph/9307301 ] ; f. m. borzumati , z. phys .
c * 63 * , 291 ( 1994 ) [ arxiv : hep - ph/9310212 ] ; s. bertolini and f. vissani , z. phys .
c * 67 * , 513 ( 1995 ) [ arxiv : hep - ph/9403397 ] .
f. m. borzumati , m. olechowski and s. pokorski , phys .
b * 349 * , 311 ( 1995 ) [ arxiv : hep - ph/9412379 ] ; r. rattazzi and u. sarid , nucl .
* b501 * , 297 ( 1997 ) [ arxiv : hep - ph/9612464 ] ; h. baer , m. brhlik , d. castao and x. tata , phys . rev .
d * 58 * , 015007 ( 1998 ) [ arxiv : hep - ph/9712305 ] ; t. blaek and s. raby , phys .
d * 59 * , 095002 ( 1999 ) [ arxiv : hep - ph/9712257 ] .
f. borzumati , c. greub , t. hurth and d. wyler , phys .
d * 62 * , 075005 ( 2000 ) [ arxiv : hep - ph/9911245 ]
. g. degrassi , p. gambino and g. f. giudice , j. high energy phys . * 0012 * , 009 ( 2000 ) [ arxiv : hep - ph/0009337 ] ; m. carena , d. garcia , u. nierste and c. e. m. wagner , phys .
b * 499 * , 141 ( 2001 ) [ arxiv : hep - ph/0010003 ] ; k. i. okumura and l. roszkowski , phys .
lett . * 92 * , 161801 ( 2004 ) [ arxiv : hep - ph/0208101 ] ; k. i. okumura and l. roszkowski , j. high energy phys . * 0310 * , 024 ( 2003 ) [ arxiv : hep - ph/0308102 ] .
f. borzumati , c. greub and y. yamada , phys .
d * 69 * , 055005 ( 2004 ) [ arxiv : hep - ph/0311151 ] ; g. degrassi , p. gambino and p. slavich , phys .
b * 635 * , 335 ( 2006 ) [ arxiv : hep - ph/0601135 ] .
j. a. casas and s. dimopoulos , phys .
b * 387 * , 107 ( 1996 ) [ arxiv : hep - ph/9606237 ] . v. m. abazov _ et al . _
[ d0 collaboration ] , phys .
lett . * 97 * , 021802 ( 2006 ) [ arxiv : hep - ex/0603029 ] ; a. abulencia _ et al . _ [ cdf - run ii collaboration ] , phys .
* 97 * , 062003 ( 2006 ) [ arxiv : hep - ex/0606027 ] ; a. abulencia _ et al . _
[ cdf collaboration ] , phys .
* 97 * , 242003 ( 2006 ) [ arxiv : hep - ex/0609040 ] .
m. ciuchini and l. silvestrini , phys .
lett . * 97 * , 021803 ( 2006 ) [ arxiv : hep - ph/0603114 ] ; m. endo and s. mishima , phys .
b * 640 * , 205 ( 2006 ) [ arxiv : hep - ph/0603251 ] ; j. foster , k. i. okumura and l. roszkowski , phys .
b * 641 * , 452 ( 2006 ) [ arxiv : hep - ph/0604121 ] ; s. baek , j. high energy phys . * 0609 * , 077 ( 2006 ) [ arxiv : hep - ph/0605182 ] ; r. arnowitt , b. dutta , b. hu and s. oh , phys .
b * 641 * , 305 ( 2006 ) [ arxiv : hep - ph/0606130 ] .
s. r. choudhury and n. gaur , phys .
b * 451 * , 86 ( 1999 ) [ arxiv : hep - ph/9810307 ] ; k. s. babu and c. f. kolda , phys .
lett . * 84 * , 228 ( 2000 ) [ arxiv : hep - ph/9909476 ] ; c. s. huang , w. liao , q. s. yan and s. h. zhu , phys .
d * 63 * , 114021 ( 2001 ) [ erratum - ibid .
d * 64 * , 059902 ( 2001 ) ] [ arxiv : hep - ph/0006250 ] ; p. h. chankowski and .
sawianowska , phys .
d * 63 * , 054012 ( 2001 ) [ arxiv : hep - ph/0008046 ] . c. bobeth , t. ewerth , f. krger and j. urban , phys .
d * 64 * , 074014 ( 2001 ) [ arxiv : hep - ph/0104284 ] .
f. scuri , arxiv:0705.3004 [ hep - ex ] . | the decay @xmath0 is discussed in the minimal supersymmetric standard model with general flavor mixing for squarks , at large @xmath1 . in this case , in addition to the chargino loop contributions which were analyzed in previous studies , @xmath1-enhanced contributions from the gluino and charged higgs boson loops might become sizable compared with the standard model contribution , at least in principle .
however , it is demonstrated that the experimental bounds on the new physics contributions to the radiative decay @xmath2 should strongly constrain these contributions to @xmath0 , especially on the gluino contribution .
we also briefly comment on a possible constraint from the @xmath3 decay .
tu-796 + youichi
yamada [ cols= " < " , ] pacs : 13.20.he , 12.60.jv |
the complex cluster morphology seen in x - ray images and in galaxy distributions gave support to the hypothesis that large structures form hierarchically ; that is , that small groups of galaxies merge to form low - mass subclusters , which then merge to form a massive rich cluster with the infalling groups aligned along large filaments .
galaxies and groups are the building blocks of the observable universe and contain the bulk of the observable baryons .
groups are estimated to contain a significant fraction , 20 - 30% , of the total matter in the universe .
thus groups are important cosmological indicators of the distribution and properties of the dark matter . however , because they are not as luminous as clusters , they have received less study than their more massive cousins .
observations of rich clusters show many examples of pending or ongoing mergers of subclusters .
in particular in x - ray cluster catalogs , about 40% of rich clusters show substructure @xcite .
virtually all stages of cluster mergers have been thoroughly investigated with both the _ chandra _ and xmm - newton observatories @xcite . however
, less attention has been paid to the merging of groups and the formation of low - mass clusters due to their lower x - ray luminosity and the paucity of examples in the local universe . to our knowledge , the only nearby example of the early stages of the merger of two roughly equal mass groups is the ngc 499/ngc 507 pair @xcite . in the hierarchical scenario ,
group mergers represent a critical transitional phase in the formation of larger scale structure .
an understanding of the group merger process is thus fundamental to understanding the growth of structure . in this paper , we report results from analysis of a short _ chandra_/acis - s observation of the nearby elliptical galaxy ngc 7618 ( @xmath0=0.017309 or d@xmath7=74.1 mpc for wmap cosmology @xcite - 1@xmath8=350 pc ) .
the x - ray luminosity of ngc 7618 is @xmath27@xmath910@xmath10 ergs s@xmath11 in the 0.1 - 10 kev band , typical of groups , not isolated elliptical galaxies , although it appears to be optically isolated @xcite .
we find a sharp surface brightness discontinuity in the x - ray emission north of of the nucleus of ngc 7618 , and an extended tail to the south .
we conclude that these features are either the result of a major group - group merger with ugc 12491 , a group that lies 14.1@xmath12 on the sky from ngc 7618 at virtually identical redshift ( @xmath0=0.017365 ) @xcite , or ram - pressure stripping due to infall of ngc 7618 into a larger gravitational potential ( which may include ugc 12491 ) .
this pair has been poorly studied in large part because of its relatively low galactic latitude ( @xmath13 , @xmath14 , @xmath15=0.97 , @xmath16=1.19@xmath910@xmath17 @xmath18 ) ) .
ngc 7618 was observed for 18.4 ks with chandra / acis - s on december 10 , 1999 ( obsid 802 ) .
the lightcurve of events in the 5.0 - 10.0 kev bandpass on the entire s3 chip , excluding the ngc 7618 nucleus and any point sources visible by eye , was created using 259 s bins and examined for periods of flaring background .
intervals where the rate was more than 3@xmath19 above the mean rate were removed .
there was considerable background flaring during this observations , and almost 10 ks of data were excluded .
only 8438 s of good time remained .
bad pixels , hot columns , and columns along node boundaries were also removed .
we present data from the s2 and s3 chips in this paper .
absorption by foreground gas in our galaxy ( @xmath16=1.19@xmath910@xmath17 @xmath18 ) was included in all spectral fits .
ngc 7618 was also observed by asca for @xmath254 ks on july 7 , 1998 .
ugc 12491 is contained within the fov of the asca / gis , and we use these data to measure the temperature of the gas around this galaxy and the diffuse emission between ngc 7618 and ugc 12941 .
data from the sis was not used because of its smaller field of view .
an adaptively smoothed asca / gis image of the ngc 7618/ugc 12491 pair ( usin the ciao program ` csmooth ' ) is shown in figure [ gisimg ] .
the optical and x - ray properties of both groups are summarized in table [ galtab ] .
their x - ray luminosities are each @xmath26 - 7@xmath910@xmath10 ergs s@xmath11 , and their x - ray emission extends to radii of 150 - 200 kpc .
the luminosities and spatial extents far exceed those of typical isolated elliptical galaxies and are more representative of poor clusters or fossil groups @xcite .
the central elliptical galaxies represent only a small fraction of the gravitating mass that resides in a much larger dark matter halo .
no optical census of the galaxy populations around either ngc 7618 or ugc 12491 has been undertaken , but their recessional velocities differ by only 17 km s@xmath11 .
the temperature of the gas around these galaxies is @xmath20.8 kev , typical of groups .
based on their x - ray luminosities , x - ray extents , and spatial proximity we conclude that these two objects are groups likely within @xmath2300 kpc of each other and gravitationally interacting . an adaptively smoothed , exposure corrected , background subtracted _ chandra_/acis - s image in the 0.5 - 2.0 kev band of ngc 7618 is shown in figure [ acisimgb ] .
the x - ray bright active nucleus of the host galaxy is labeled at the center .
there are three unusual features to note in this image .
first , the peak of the diffuse x - ray emission ( shown orange in figure [ acisimgb ] ) , and therefore presumably the peak in gas density , as well as the center of the host galaxy lie @xmath21@xmath12 north of the center of the larger scale diffuse x - ray emission ( shown green ) .
an x - ray ` tail ' extends @xmath23.3@xmath12 ( 69.3 kpc ) south of the nucleus .
the statistical significance of this ` tail ' can be seen in the x - ray surface brightness profiles shown in figure [ tailwedge ] .
these profiles were made in two 90@xmath1 sectors to the north ( green ) and south ( red ) with the nucleus of ngc 7618 at the vertex .
the surface brightness in the s sector at distances larger than 70@xmath8 from the nucleus is several times higher than in the north sector .
either the dark matter has a very unusual distribution , or the gas is not in hydrostatic equilibrium in the gravitational potential .
second , there is a sharp surface brightness discontinuity @xmath241@xmath8 ( 14.4 kpc ) north of the nucleus that spans @xmath2130@xmath1 .
this discontinuity is delineated by the white arrows in figure [ acisimgb ] . and is probably a contact discontinuity between two moving fluids .
third , on smaller scales , the x - ray emission from the gas peaks @xmath29@xmath8 ( 3.2 kpc ) to the e of the nucleus of the host galaxy as seen in figure [ xoptovl ] . in the central 10 kpc of the host galaxy , the stars will dominate the gravitating mass , so there clearly is an offset between the hot gas and the gravitating mass .
all of these strongly suggest that the gas has partially separated from the gravitating matter and indicate non - hydrostatic motions .
there is an x - ray point source coincident with the optical nucleus , presumably a low - luminosity agn .
the point source contains 48 counts in the 0.5 - 5.0 kev band . assuming a power - law spectrum with photon index 1.7 and galactic absorption , the x - ray luminosity of the active nucleus is 4.2@xmath910@xmath20 ergs s@xmath11 in the 0.1 - 10.0 kev band ( unabsorbed ) .
the x - ray luminosity of the nucleus is roughly an order of magnitude larger than that expected based on the correlation of x - ray and radio cores @xcite .
this may be a considerable underestimate if the nucleus is heavily absorbed , but is typical of that found in other `` normal '' elliptical galaxies @xcite .
the surface brightness profile in a 60@xmath1 sector to the north of ngc 7618 , shown in figure [ sbprof ] , drops by approximately a factor of two across the discontinuity . at the sharpest region of the edge to the ne of the nucleus ,
the surface brightness drops by a factor of 4 over a distance of @xmath24@xmath8 ( 1.4 kpc ) . the surface brightness profile in a 30@xmath1 sector to the ne
is shown in figure [ newedge ] .
the morphology of this discontinuity is similar to that seen in _
_ observations of cluster cold - fronts , although there is no evidence for a temperature discontinuity between the two moving fluids as commonly seen in clusters of galaxies @xcite .
we fitted absorbed apec models to the spectra in 90@xmath1 sectors of two annular regions , one inside the discontinuity and one outside .
the thickness of the inner and outer annuli were 33.0@xmath8 ( 11.6 kpc ) and 66.4@xmath8 ( 23.2 kpc ) , respectively .
the gas temperature does not change significantly across the discontinuity ( @[email protected] kev , @[email protected] kev ) .
there is marginal evidence for a jump in the elemental abundance ( @[email protected] , @[email protected] - all uncertainties at 90% confidence ) , however .
this suggests that the discontinuity could be an ` abundance ' front due to a sharp discontinuity in the elemental abundance , and therefore the emissivity of the gas as observed in ngc 507 @xcite .
the temperature and elemental abundance structure in the gas is probably more complex than we have assumed here , but the short exposure time and limited quality of the data prevent a more detailed analysis .
fitting power laws to the surface brightness profiles interior to and exterior to the discontinuity and assuming hydrostatic equilibrium , we find a large change in the power law index across the discontinuity , @xmath23=0.30 and @xmath24=0.64 .
the change in @xmath25 is almost certainly not related to a change in the gravitational potential and suggestive of non - hydrostatic gas motions .
we estimate the gas density on both sides of the discontinuity by deprojecting the surface brightness profile ( assuming spherical symmetry ) and find proton densities of 6.0@xmath26 and 5.0@xmath27 @xmath910@xmath28 inside and outside the discontinuity , respectively .
the upper limit of the velocity of the gas interior to the discontinuity estimated from the maximum pressure difference is mach 0.9 or @xmath2 420 km s@xmath11 @xcite .
the x - ray morphology of the asca / gis image ( figure [ gisimg ] ) suggests that the two groups reside in a larger - scale dark matter potential .
a third x - ray peak is seen 8.5@xmath12 to the east of ngc 7618 , and diffuse emission extends at least 15@xmath12 to the north and 10@xmath12 to the south .
any gas that resides in this larger scale dark matter halo should be hotter than the gas in the ngc 7618 group .
we fitted the asca / gis spectrum in three regions , two 7@xmath12 radius circles centered on each galaxy and a third region 15@xmath12 in radius centered between ngc 7618 and ugc 12491 excluding the two 7@xmath12 radius circles centered on each galaxy .
the radius of the two circles centered on ngc 7618 and ugc 12491 corresponds to the 90% encircled energy radius for the asca / gis in the 1 - 2 kev band .
background was determined from asca / gis high - latitude , blank sky observations taken from the heasarc and generated using the ftool ` mkgisbgd ' .
we fit absorbed , single temperature apec models to each spectrum with @xmath16 frozen at the galactic value and the elemental abundance frozen at 0.6@xmath29 .
the results of these fits are summarized in the top half of table [ spectab ] .
single temperature models are poor fits for the regions centered on ngc 7618 and ugc 12491 , but provide an adequate description of the diffuse emission between the galaxies .
the temperature of the diffuse gas between ngc 7618 and ugc 12491 is 2.32@xmath30 .
we also fit two temperature models to the two spectra extracted from the regions centered on ngc 7618 and ugc 12491 . as before ,
both the @xmath16 and the elemental abundance were frozen . for ngc 7618
, we also froze the temperature of one component at 0.8 kev , the value determined from the acis - s spectral fitting , to reduce the number of free parameters .
if this parameter is allowed to freely vary , the fit value is consistent with 0.8 kev . for ugc 12491
, the temperatures of both components were allowed to freely vary .
the results of these fits are summarized in the bottom half of table [ spectab ] .
the two temperature model provides acceptable fits in both cases , and the temperature of the hotter component is @xmath22.3 kev .
there are at least four possible explanations for the observed x - ray structures .
first , the complex x - ray morphology could be the result of a radio lobe / igm interaction .
ngc 7618 is a radio source .
it is detected in the nvss with a flux density of 20 mjy , with some evidence of extension nw / se . at higher frequencies ( 5 and 8 ghz )
two 15-min archival vla observations only detect a point - like core coincident with the center of the galaxy , with a flux density of 4.5 mjy at 4.9 ghz and 3.0 mjy at 8.4 ghz .
however , at lower frequencies , the flux density is much higher : 1.2 jy in the 151-mhz 6c catalog @xcite , 0.26 jy in the 408-mhz b3 catalog @xcite , and 1.0 jy in the 74-mhz vla low - frequency sky survey ( vlss : http://lwa.nrl.navy.mil/vlss/ ) .
the high flux density at low frequencies suggests that there may be a relic radio source , the aged synchrotron plasma from a more energetic phase of the active nucleus .
the x - ray morphology is very different from that seen in other examples of radio lobe / ism interactions , however .
there are no obvious x - ray cavities as are commonly seen in such radio lobe / ism interactions ( e.g. @xcite ) , nor is there evidence of hot , shock - heated shell that would be present if the radio lobes were expanding supersonically @xcite .
it is not clear whether any of these features would be expected in the case of a relic radio source , however
. a sensitive low - frequency radio map would give us a better understanding of the possible interaction between radio - emitting plasma and the hot gas .
second , it is possible that the ngc 7618 gas is oscillating , or sloshing , in the gravitational potential because of a recent merger / interaction with a lower mass sub - group .
a dust lane has been detected in the optical host galaxy , perhaps indicative of a recent merger and adding support to this hypothesis @xcite .
similar structures ( on larger scales ) to those presented here have been seen in x - ray observations of clusters of galaxies @xcite . in this scenario ,
the gas is oscillating , or ` sloshing ' , in the dark matter potential due to a recent merger with a sub - cluster .
the infall of the sub - cluster drives a shock into the gas that displaces it from the gravitating matter .
the gas oscillates around the center of mass , eventually returning to hydrostatic equilibrium via viscous dissipation .
such processes can contribute a significant amount of energy to the gas and may disrupt or prevent the formation of cooling flows in clusters of galaxies .
if we naively assume the gas on both sides of the discontinuity is in hydrostatic equilibrium , we find an unphysical discontinuity in the gravitating mass .
the apparent mass discontinuity is the result of the non - zero acceleration of the gas inside the discontinuity .
the distribution of the gas will reflect the reduced gravity .
that is , the acceleration term in euler s equation is non - zero for gas inside the discontinuity .
assuming that the core is at maximum displacement from the center , the gravitational potential energy of the gas is @xmath317@xmath910@xmath32 ergs gm@xmath11 , or roughly one third of the thermal energy of the gas .
there is one important difficulty with this interpretation , however . on tens of kpc scales , the emission peak of the gas and the optical galaxy ( ngc 7618 )
are both offset ( to the n ) relative to the larger scale x - ray isophotes ( shown green in figure [ acisimgb ] ) .
if the sloshing scenario is correct only the gas and not the galaxy / gravitating mass , should be moving relative to the large scale dark matter halo .
the host galaxy should be resting at the center of the gas distribution .
in addition , the peak of the x - ray emission lies to the east of the nucleus of the host galaxy , but the larger scale ` sloshing ' is north / south .
this suggests that there are significant gas motions along perpendicular axes , and that the gas is coupled to the gravitating mass ( the stars ) along the north / south axis , but partially decoupled along the east / west axis .
it is difficult to see how there could be the case unless the merger were off - axis .
if this is the case , the merging galaxy should be detectable by an optical census .
the third possibility is that the observed structures are the result of a near - miss flyby of the nearby ugc 12491 group . in this scenario ,
the ugc 12491 group passed by ngc 7618 from the se toward its current position to the wnw , and the complex x - ray morphology of ngc 7618 is the result of fluid motions in the gas induced by the gravitational and hydrodynamical interaction .
the observed x - ray morphology is the result of the ngc 7618 core being displaced relative to the larger scale gravitational potential , and the surface brightness discontinuity represents a discontinuity in the elemental abundance .
x - ray observations of groups and clusters show that the elemental abundance is strongly peaked toward the center .
the displacement of the core of a group relative to its halo could create the observed structures .
hydrodynamic simulations of head - on collisions of equal mass clusters show that the close approach of the cores can create strong non - hydrostatic motions in the gas @xcite .
elongation of the dark matter potential , and thus the gas distribution , is a natural consequence of these collisions . in these simulations , the gas becomes partially separated from the gravitating matter and is not a good tracer of the dark matter distribution .
the separation between ugc 12491 and ngc 7618 was probably ( much ) smaller in the past in order to create such a large disturbance in the ngc 7618 gas .
the simulations of merging clusters show little separation of the gas from the dark matter as the two clusters initially approach each other .
it is only as the cores collide / merge and subsequently separate that significant gas velocities develop .
if the current positions of the two groups are , in fact , their closest approach so far , non - hydrostatic effects should just be starting to manifest themselves .
therefore , the two groups must have already passed each other at least once .
there are several difficulties with this scenario , however . if the hotter gas seen in the asca / gis spectrum is shock - heated group gas , the measured temperature ratio ( 0.8 to 2.3 kev ) implies an infall / approach velocity of the two groups of @xmath21170 km s@xmath11 @xcite , much larger than typical peculiar velocities
. it would also be difficult to explain the large amount of hot gas ( @xmath210@xmath33 m@xmath34 ) seen in the asca / gis observation in this scenario under the assumption that this component is entirely shock - heated group gas .
the total ( thermal ) energy of the hotter component is only @xmath21.3@xmath910@xmath35 ergs , a reasonable value for the collison of two groups ( we note that an agn outburst could easily supply this much energy to the gas as well @xcite ) .
if the hot component is , in fact , shock - heated group gas , it is not bound to the group potential and a transient phenomenon , escaping as a wind .
it is likely that these two groups are gravitationally bound .
ignoring angular momentum ( which is not believed to be important on large scales @xcite ) and dissipative forces , and assuming that the mass of each group is 10@xmath36 m@xmath34 , the escape velocity of this pair is @xmath2550 km s@xmath11 .
if either of these groups has a velocity relative to the center of mass larger than this value , the pair will be unbound .
this is larger than typical peculiar velocities of galaxies / groups , and it is reasonable to conclude that the system is bound . significant angular momentum will decrease this value , whereas dissipative forces will increase it .
the dynamic parameters of this pair are too poorly known to make a definitive statement , however .
the fourth possibility is that ngc 7618 is falling into a larger gravitational potential . in this scenario ,
either ugc 12491 is at the center of the potential and ngc 7618 is falling into it , or both ngc 7618 and ugc 12491 are falling into a larger scale potential .
the existense of the hotter ( @xmath22.3 kev ) gas component in the asca / gis observation ( figure [ gisimg ] ) supports this scenario .
the x - ray structures seen in the gas around ngc 7618 are then the result of ram - pressure stripping .
this is the classic ` cold - front ' scenario discussed by @xcite .
the effects of ram - pressure stripping on the hot gas atmospheres of early - type galaxies falling into clusters have been studied in _
observations of ngc 4472 , ngc 4552 ( falling into the virgo cluster ) and ngc 1404 ( falling into / toward ngc 1399 , the dominant member of the fornax cluster ) @xcite . in these cases ,
a sharp surface brightness discontinuity in the direction of infall and a diffuse tail in the opposite direction have been observed .
the gas around ngc 7618 is being stripped by hotter , lower density gas that presumably resides in the larger / deeper dark matter halo .
if this scenario is correct , either ngc 7618 is falling into ugc 12491 or both groups are merging with / falling into a larger dark matter potential in which there is no central dominant elliptical galaxy .
the high temperature of the second spectral component in the asca analysis and the lack of an azimuthally symmetric x - ray halo around ugc 12491 ( figure [ gisimg ] ) support the latter hypothesis .
this phenomenon has in fact already been observed on a smaller scale in the pegasus i group @xcite .
neither of the massive ellipticals in this group , ngc 7619 or ngc 7626 , lie at the center of the extended x - ray halo .
_ chandra _ observations of ngc 7619 show a sharp surface brightness discontinuity to the ne ( presumably the direction of infall ) , and an extended , ram - pressure stripped tail in the opposite direction .
if we assume the hotter ( 2.3 kev ) component is in hydrostatic equilibrium with the dark matter potential and that the gas density follows a beta - model profile with @xmath25=0.67 and @xmath37=50 kpc ( typical for clusters of galaxies ) , the gravitating mass , @xmath38 , within a radius of 325 kpc of the midpoint between ngc 7618 and ugc 12491 is @xmath25.6@xmath910@xmath32 m@xmath34 .
the gas mass , @xmath39 , required to account for the observed asca / gis flux of the hotter component within this radius is @xmath27@xmath910@xmath40 m@xmath34 for this density profile .
ignoring the stellar component , which is insignificant on these spatial scales , the baryon fraction , @xmath41 is @xmath21.5% .
such a low value of baryon fraction is not improbable for a 2.3 kev cluster .
we speculate that this larger dark matter halo may be a ` failed ' cluster in which much of the gas was blown off during a cataclysmic event early in its formation ( either a merger or a powerful agn remnant ) .
we caution , however , that the spatial resolution of the asca / gis is poor and our knowledge about the morphology of the gas limited .
in addition , this gas may not be in hydrostatic equilibrium , so the uncertainties on both ( gas and gravitating ) mass estimates are large .
a moderate xmm - newton observation of this pair could measure the temperature and morphology of the hotter component and resolve this issue .
we note that in this scenario the surface brightness discontinuity north of the ngc 7618 nucleus can not be the stagnation point between the hot gas in ngc 7618 and the gas of this putative larger scale halo .
if the observed x - ray surface brightness discontinuity is in fact the stagnation point between these two gases , the temperature of the gas exterior to the discontinuity must be on the order or higher than that of the halo gas ( from bernoulli s equation ) .
the stagnation point therefore must lie beyond the observable emission , and the surface brightness discontinuity represents a contact discontinuity between two fluids .
unless the infall is highly supersonic , the gas interior to the stagnation point should not be highly disturbed and should remain in rough hydrostatic equilibrium in the gravitational potential of ngc 7618 . a deeper _ chandra _ observation of the central regions of ngc 7618 is required to elucidate the hydrodynamics of the gas .
we have observed a sharp surface brightness discontinuity in the x - ray emission from the hot gas in the ngc 7618 group , and an x - ray ` tail ' extending 70 kpc in the opposite direction in an 8 ks _ chandra_/acis - s observation .
archival asca / gis observations indicate the presence of a hotter ( 2.3 kev ) component , although the morphology of this gas is poorly constrained .
we conclude that there are three possible explanations for these features .
first , the ngc 7618/ugc 12491 pair underwent a recent ` near - miss ' flyby . if this is the case , this pair is the nearest early - stage merger of two roughly equal mass groups .
second , ugc 12491 may be at the center of a cluster , and ngc 7618 is falling into it .
third , ngc 7618 and ugc 12491 are both falling into a gas poor cluster with no dominant central elliptical galaxy .
whether the observed features are the result of a group - group merger , or the infall of two groups into a larger dark matter potential , the ngc 7618/ugc 12491 pair is one of the best examples of an ongoing merger in the local universe .
deeper x - ray observations are required to better constrain the thermodynamic parameters of the gas in the central regions and the larger scale halo .
radio observations will be critical in assessing the role of radio plasma / igm interactions .
if the observed x - ray features are the result of ram - pressure stripping or a merger interaction between the groups , the effects on the relic radio halo are likely to have been dramatic
. a detailed optical census including velocities of the other galaxies in the both groups will be useful to constrain their dynamics and their relationship to the larger dark matter potential .
this work was supported by nasa contracts nas8 - 38248 , nas8 - 39073 , the chandra x - ray center , and the smithsonian institution .
we would like to thank the anonymous referee for comments that improved this paper .
.summary of the x - ray and optical properties of the ngc 7618 and ugc 12491 galaxies .
the x - ray luminosity is in the 0.1 - 10 kev bandpass ( unabsorbed ) within 7@xmath12 ( 146 kpc ) of the nucleus .
absolute magnitudes have been corrected for extinction . [ cols="<,^,^",options="header " , ] + @xmath42 ( kev ) & 1.43@xmath43 & 1.32@xmath44 & 2.32@xmath30 + flux & 3.06@xmath910@xmath45 & 5.37@xmath910@xmath45 & 2.53@xmath910@xmath45 + @xmath46 & 1.95 & 1.60 & 0.73 + + @xmath47 ( kev ) & 0.80 & 0.86@xmath48 & + flux & 2.26@xmath910@xmath45 & 3.69@xmath910@xmath45 & + @xmath49 ( kev ) & 2.21@xmath50 & 2.26@xmath51 & + flux & 1.48@xmath910@xmath45 & 2.15@xmath910@xmath45 & + @xmath46 & 1.03 & 0.78 & + | we present results from a short _ chandra_/acis - s observation of ngc 7618 , the dominant central galaxy of a nearby ( @xmath0=0.017309 , d=74.1 mpc ) group .
we detect a sharp surface brightness discontinuity 14.4 kpc n of the nucleus subtending an angle of 130@xmath1 with an x - ray tail extending @xmath2 70 kpc in the opposite direction . the temperature of the gas inside and outside the discontinuity is [email protected] and [email protected] kev , respectively .
there is marginal evidence for a discontinuous change in the elemental abundance ( @[email protected],@[email protected] at 90% confidence ) , suggesting that this may be an ` abundance ' front .
fitting a two - temperature model to the asca / gis spectrum of the ngc 7618/ugc 12491 pair shows the presence of a second , much hotter ( @[email protected] kev ) component .
we consider several scenarios for the origin of the edge and the tail including a radio lobe / igm interaction , non - hydrostatic ` sloshing ' , equal - mass merger and collision , and ram - pressure stripping . in the last case
, we consider the possibility that ngc 7618 is falling into ugc 12491 , or that both groups are falling into a gas poor cluster potential .
there are significant problems with the first two models , however , and we conclude that the discontinuity and tail are most likely the result of ram pressure stripping of the ngc 7618 group as it falls into a larger dark matter potential . |
the experimental realization of degenerate atomic gases @xcite have opened up the possibility of engineering strongly correlated many - body quantum states .
such states are fundamental for the development of quantum technologies such as those implementing quantum information protocols and quantum computation .
a paramount example is the realization of a mott insulator ( mi ) @xcite in the lowest bloch band of an optical lattice which can serve as a quantum register @xcite .
bose - einstein condensates of alkali atoms have also been proposed as systems to realize macroscopic superposition states @xcite or number correlated states ( also called twin fock states ) which are useful for quantum metrology @xcite .
in @xcite we proposed a method for engineering twin fock states via a two - component mi to superfluid ( sf ) transition .
the proposed scheme manipulates hyperfine states of atom pairs pinned to single lattice sites and decoupled from one another in the mi regime @xcite .
the mi limit has already proven to be useful starting point for quantum state engineering since the motional dynamics are frozen out and the system decouples to a set of single site hamiltonians to good approximation . in the case of a two - component system with doubly occupied sites spin - changing collisions will occur and can be accurately controlled with external magnetic or microwave fields @xcite . as a result
the spin - dynamics of atom pairs behaves like an effective two - level system undergoing rabi - like oscillations .
in contrast to the usual rabi - model for the coupling of a pair of _ single - particle _ states this represents a coherent coupling between a pair of _ two - particle states _ induced by collisional interactions .
these local _ two - particle _ correlations are then transformed into _ multi - particle _ correlations extending over the whole system by melting the mi into a sf .
this quantum melting can be experimentally implemented by adiabatically ramping down the depth of the lattice potential .
the build up of long - range correlations when a periodic potential is dynamically lowered is a highly non trivial process that occurs rather quickly in a one - component system @xcite .
the quantum melting of a multi - component system is still an open problem that is even more intriguing due to the different phases that appear in the @xmath0 phase diagram depending on the ratios of the interparticle interactions and of the tunneling rates of each component @xcite .
an analysis of the two - component symmetric case with all interparticle interactions equal can be found in @xcite . in @xcite
we studied the adiabatic melting of an optical lattice in a two - component system with different values for the inter - species and intra - species interactions .
we showed that one can create twin fock states with high fidelity via melting of an initial mi state with two atoms per site in different internal states .
we inspected possible sources of experimental imperfections such as atom loss during melting or depletion caused by final finite interactions .
the possibility of creating macroscopic superposition states via the adiabatic melting of on - site entangled mi states in an infinitely connected lattice was also briefly considered in @xcite .
an examination of the adiabatic melting in an infinitely connected lattice , including the effect of the dynamical relative phase acquired by the state can be found in @xcite . in this paper
we extend the results in @xcite and analyze the melting of on - site superposition states in a lattice with only nearest neighbor hopping .
we show that the final sf states obtained from such on - site superposition states are highly depleted even if the initial state is a ground mi state . for the melted states we calculate the momentum distribution and the momentum space particle number correlations .
these quantities are experimentally accessible either by time of flight expansion of the cloud or via shot noise measurements @xcite .
this paper is organized as follows . in section [ sec : bhm ] we introduce the two - mode bose - hubbard model ( bhm ) and the initial mott insulating states that we consider . in section [ sec : adiabatic ] we describe the adiabatic approximation and the no - crossing rule that we use to infer the final superfluid states . in section [ sec : melt_ab ] we illustrate how one can create macroscopic twin fock states if the inter - species interaction is greater than the intra - species interaction during melting . in [ sec : messy ] we show using the symmetry properties of the initial states and a series of approximations , that an on - site entangled mi ground state melts into a highly depleted sf state in the limit of intra - species interaction greater than the inter - species interaction .
in this section we introduce the two - component bhm that describes the dynamics of ultra - cold atoms in an optical lattice .
we discuss its experimental feasibility and analyze the symmetries of the hamiltonian .
finally we introduce the initial mi states under consideration .
our focus is on a state - dependent optical lattice which is sufficiently deep that two hyperfine states @xmath1 and @xmath2 of the atoms are be trapped in the lowest bloch band of the lattice .
the resulting system is then accurately described @xcite by the two - component bhm with the corresponding hamiltonian ( @xmath3 ) @xmath4 where @xmath5 is the bosonic destruction operator for an @xmath6-atom localized in lattice site @xmath7 , @xmath8 and @xmath9 , while @xmath10 denotes summation over nearest - neighbors .
the parameter @xmath11 is the tunnelling matrix element for atoms in state @xmath1(@xmath2 ) ; @xmath12 and @xmath13 are the on - site intra- and inter - species interaction matrix elements , respectively .
a crucial feature of the optical lattice setup is that the ratio between the tunnelling and interaction matrix elements in eq .
( [ eq : ham ] ) is experimentally controllable via the lattice laser intensities .
further independent control of the interaction matrix elements @xmath13 , @xmath14 can be achieved with feshbach resonances @xcite , or by shifting the @xmath1 and @xmath2 atoms away from each other using state - dependent lattices @xcite .
all of these methods of control can be exploited dynamically to permit different regimes of the hamiltonian in eq .
( [ eq : ham ] ) to be unitarily connected .
together with the long decoherence times of ultra - cold atoms this dynamical manipulation can often be implemented on near - adiabatic timescales .
based on the considerable flexibility of the optical lattice setup we make a number of simplifications in our analysis .
first , we restrict the couplings to the symmetric case where @xmath15 and @xmath16 .
second , we consider states of the system @xmath17 in which the lattice is commensurately filled with two atoms per site so @xmath18 has @xmath19 , where @xmath20 is the total particle number and @xmath21 is the total number of sites .
finally , we confine our attention to a 1d lattice with periodic boundary conditions ( i.e. @xmath22 ) in addition to the spatial homogeneity of the matrix elements in eq .
( [ eq : ham ] ) .
this is not only a good approximation for the center of a large system , where trapping and boundary effects have a negligible influence , but is also directly applicable to recently demonstrated ring - shaped optical lattices @xcite .
the most basic conserved quantity of the two - component bhm is the total number of atoms @xmath23 , @xmath24 in the either of the internal states @xmath1 or @xmath2 , respectively . with the additional assumptions outlined a number of other symmetries
arise which we exploit in this work .
the homogeneity and periodic boundary conditions assumed results in translational invariance . to exploit this
we introduce quasi - momentum operators for each component as @xmath25 where @xmath26 is the quasi - momentum in the first brillouin zone and @xmath27 is the position of the site @xmath7 .
the quasi - momentum states spanning the first brillouin zone have @xmath28 with @xmath29 and @xmath30 being the distance between lattice sites . for @xmath21
even @xmath31 and @xmath32 , while for @xmath21 odd @xmath33 .
the total momentum of the system is then @xmath34 , where @xmath35 and @xmath36 , and is a conserved quantity .
the total momentum generates _ cyclic _ shifts of the lattice sites , e.g. @xmath37 for the site indices of all operators , via the unitary @xmath38 .
its action on the bosonic creation operators is @xmath39 simultaneously for all sites @xmath7 .
the homogeneity of the system also gives mirror reflection symmetry in which site indices for operators interchange as @xmath40 .
in this case the _ parity _ operator @xmath41 is the conserved quantity and implements the transformation @xmath42 simultaneously for all sites @xmath7 .
finally , the homogeneity and symmetry of the matrix elements result in _ color _ inversion symmetry @xmath43 . like parity
there is a hermitian - unitary operator @xmath44 which is the conserved quantity and implements the transformation @xmath45 simultaneously for all sites @xmath7 .
these three symmetries are schematically represented in fig .
[ fig : sym](a)-(c ) . and
@xmath2.,width=302 ] symmetries are by definition related to the presence of degeneracies @xcite or more precisely to a degeneracy which remains for all values of the couplings in the hamiltonian - a so - called permanent degeneracy . by using these symmetries of @xmath46
it follows that any of its eigenstates which are connected by a combination of these symmetry operators are necessarily degenerate
. moreover , any eigenstates which can not be mapped on to each other by symmetry operators will be non - degenerate if the symmetries of the system are exhaustive @xcite .
other types of degeneracies can occur when the couplings attain a particular value which indicates that at this point the system acquires a higher symmetry as compared to other points .
this enhancement of symmetry frequently occurs in the limits of zero or infinite coupling .
an important example in this work is when @xmath47 where @xmath46 decouples into a set of single - site hamiltonians and acquires full permutational symmetry .
the breaking of the full permutational symmetry for finite @xmath48 splits the degenerate eigenstates of the @xmath47 hamiltonian as we will see in proceeding sections .
particle number conservation for each component breaks up the system hilbert space into sectors where @xmath49 and @xmath50 .
the evolution of each component of an initial state in each of these sectors can therefore be treated independently .
we shall use @xmath51 to denote the sectors which possess the same total particle number @xmath52 where @xmath53 and @xmath54 using @xmath55 .
we denote as @xmath56 the completely symmetric subspace of the direct sum @xmath57 , defined by @xmath58 , @xmath59 and @xmath60 for @xmath61 where @xmath62 . as we shall now outline for the specific initial states considered here we fix @xmath19 throughout and the only relevant sectors are those where @xmath63 is restricted to be even or @xmath64 .
our approach to exploit the two - component bhm to engineer quantum states is based on using two experimentally accessible initial states .
both of these states are generated in the mi limit where interactions are dominant , i.e. @xmath65 and @xmath66 , and precisely with two atoms localized on every site . by controlling the spin - changing collisions a doubly occupied single - component mi formed in some additional hyperfine state can be transformed with near unit efficiency to the state @xcite @xmath67 with exactly the same number of @xmath1 and @xmath2 atoms in each lattice site .
note that @xmath68 is exclusively contained in the symmetric subspace @xmath69 . by applying a further @xmath70-raman pulse to the state @xmath68 an on - site entangled mi state @xmath71 can be created .
this state is an equal superposition of all @xmath72 possible combinations of the atom- pair states @xmath73 and @xmath74 over @xmath21 sites .
the states in the superposition are contained in the symmetric subspaces @xmath75 with @xmath76 , where @xmath77 denotes the integer part .
specifically , in each sector @xmath78 we form the state @xmath79 where @xmath80 stands for the normalized superposition of all permutations of the lattice sites .
since the state @xmath81 possesses full permutational symmetry it is manifestly cyclic- and mirror - symmetric . if we then symmetrize the color as where @xmath82 .
] @xmath83 we obtain states which are each contained in @xmath84 , respectively .
the state @xmath85 can then be seen to be a binomial superposition of the states @xmath86 for all even-@xmath63 symmetric subspaces as @xmath87 the two initial states in eqs .
( [ eq : sab ] ) and ( [ eq : sup ] ) are exact symmetric ground states of the two - component bhm in the limit @xmath47 for different regimes of @xmath88 .
this is the mi regime where the particle - hole spectrum is gapped . in the case @xmath89 where intra - species interaction is dominant , and in the subspace @xmath90 , the on - site state @xmath91
is energetically favored making @xmath68 the nondegenerate ground state .
this is easily confirmed by comparing it to all other configurations and moreover establishes that the ground state remains nondegenerate for non - zero @xmath48 .
for the opposite case @xmath92 where the inter - species interaction is dominant phase "- separated configurations composed of on - site states @xmath93 and @xmath94 are energetically favored .
thus at @xmath47 the full permutational and color symmetry results in @xmath95 degenerate ground states even and @xmath96 where the degeneracy his given by @xmath97 spanned by all configurations of the @xmath63 states @xmath98 and @xmath99 states @xmath100 distributed amongst the @xmath101 sites .
some of these states can be mapped by the cyclic shift symmetry @xmath102 , the mirror @xmath41 or the color @xmath44 transformation . we denote as @xmath103 the symmetrized ( with respect to @xmath44 , @xmath102 and @xmath41 ) ground state manifold of the bhm in eq .
( [ eq : ham ] ) for @xmath47 and @xmath92 with dimension denoted by @xmath104 . on this basis
we see that each of the states @xmath105
composing @xmath85 is one of many possible ground states in the symmetric subspace @xmath106 .
however , as will be described in detail in sec .
[ sec : small_messy ] , once @xmath48 is non - zero this degeneracy is partially split resulting in @xmath85 mapping to a superposition of numerous low - lying excited states .
this is schematically shown in fig .
[ fig : nocrossingrule ] .
the reason for this can be readily seen . for @xmath107
nearest - neighbor hopping will reduce the energy of configurations like @xmath108 where @xmath1 and @xmath2 atoms form contiguous regions , compared to configurations like @xmath109 .
since @xmath85 contains contributions from all configurations it necessarily maps to a superposition of numerous low - lying excited states of the two - component bhm in eq .
( [ eq : ham ] ) for each subspace @xmath75 when @xmath48 is non - zero .
this feature significantly complicates the analysis of this state even under adiabatic conditions .
here we explore how adiabatic quantum melting which is an accessible coherent process that can be applied to the system and performs a highly non - trivial many - body unitary transformation can be exploited as a means of quantum state engineering . for both initial states @xmath68 and
@xmath85 we take the quantum melting as ending in the non - interacting sf limit where @xmath110 . in this limit
the two - component bhm possesses a nondegenerate ground state @xmath111 for any sector @xmath112 in which all the @xmath1 and @xmath2 atoms bose - condense into their corresponding zero quasi - momentum state . and fixed @xmath88 .
the no - crossing rule assures that the system follows the instantaneous eigenstates and performs no level crossings if the states have all the same symmetry , there is only one varying parameter and the evolution is performed sufficiently slow.,width=226 ] a feasible way to analyze the properties of the final melted states is to use the adiabatic approximation @xcite in which the system follows the instantaneous eigenstates of the two - component bhm in eq .
( [ eq : ham ] ) for each value @xmath113 and @xmath114 during the melting .
the first requirement for the analysis to be tractable is the lack of level crossings .
this condition is greatly aided by the no - crossing rule @xcite which states that if the energy levels of a quantum system are plotted as a function of only one parameter , the curves for levels of the same symmetry do not cross .
the no - crossing rule can be followed by using the flexibility of the optical lattice setup to ensure that the dynamical melting only varies the ratio @xmath113 whilst keeping @xmath88 constant .
thus the analysis can proceed by following the evolution of eigenstates in the symmetric subspace only . also , the bhm is a non - integrable system that shows a chaotic spectrum and thus presents energy level repulsion @xcite .
the work in @xcite provides numerical evidence that for small systems the symmetries we consider are exhaustive for the bhm ( i.e. no crossings are seen ) .
we extend this assumption to arbitrary sized systems .
the energy levels may approach each other forming so - called avoided crossings , but for infinitely slow dynamics there will be no transitions between eigenstates . for finite times
the landau - zener model gives the diabatic transition probabilities @xcite between two levels , and this generalizes to brundobler - elser rule @xcite for a multistate problem . in principle , adiabatic evolution requires that the dynamical parameter @xmath113 changes in an infinitely long time .
a crude estimate of the time scale @xmath115 needed for adiabatic melting of an optical lattice is given by @xmath116 @xcite , where @xmath117 stands for the initial value in the mi regime .
importantly the adiabatic condition does not scale with the system size but with the density @xmath118 .
numerical computations in @xcite indicate that the adiabatic time scale is on the order of @xmath119 for a finite realistically sized one - dimensional system .
these values are attainable in experimental setups and make the adiabatic approximation a reasonable assumption .
these estimates are valid for the case when @xmath120 when eq .
( [ eq : ham ] ) decouples as two one - component bhm s . finding analogous time scales for the two - component system with @xmath121 is still an open problem but it is reasonable to expect that the corresponding time scales are also attainable in experimental setups .
we consider first the melting of the product mi state @xmath68 .
our analysis begins by examining the outcome of quantum melting in a small system by using an exact numerical calculation .
the results found then generalize to an arbitrarily large system and give rise to a final state whose usefulness for interferometry is analyzed . here
we provide some numerical evidence for the claims made in the previous section by performing an exact numerical diagonalization of the two - component bhm hamiltonian @xmath46 for a system with @xmath122 sites . for the state @xmath68 we need only consider the projection of @xmath46 in the subspace @xmath123 .
we then compute the ground state and low - lying excitations for a sequence of linearly decreasing interaction strengths @xmath113 with a fixed ratio @xmath124 throughout . in fig .
[ fig : small_ab_results](a ) the eigenenergies are shown confirming that the ground state in @xmath123 remains nondegenerate during the melting , and that within the symmetric subspace level repulsion ensures that no level crossings occur consistent with the no - crossing rule .
the final sf ground state is then @xmath125 in which the zero quasi - momentum mode of each component is equally occupied .
thus the quantum melting transforms local initial correlations between pairs of atoms into correlations between delocalized and macroscopically occupied modes . for this reason the final state @xmath126
is called a twin - fock state @xcite . in fig .
[ fig : small_ab_results](b ) the overlap @xmath127 between the instantaneous ground state @xmath128 during the ramp and the sf twin - fock @xmath125 is seen to converge to unity as a @xmath129 . for @xmath122 sites in the symmetric subspace @xmath123 ,
i.e. with @xmath130 , as a function of @xmath113 with @xmath124 throughout .
to the far left ( not plotted ) where @xmath131 the state @xmath68 is the non - degenerate ground state . to the right
where @xmath132 the non - interacting sf state @xmath133 is the non - degenerate ground state . as expected these two ground states
are adiabatically connected ( solid line ) .
the first and second excited states ( dotted and dashed line , respectively ) are also shown and become degenerate in the @xmath134 limit .
( b ) the overlap @xmath135 between the instantaneous ground state @xmath128 and the sf state @xmath126 as a function of @xmath113.,width=321 ] we expand here the results that were investigated in @xcite . by using the fact that @xmath68 is a nondegenerate mi ground state and the no - crossing rule the outcome of adiabatic melting found for a small system can be readily generalized to any size system as @xmath136 for arbitrary @xmath52 .
quantum melting of the mi state @xmath68 therefore provides a direct means of creating a @xmath52-particle twin - fock state @xmath137 , as depicted in fig .
[ fig : mzi](a ) .
twin - fock states @xcite have been proposed as useful input states in interferometric experiments which could potentially allow measurement sensitivities which scale as @xmath138 .
this is the heisenberg limit for interferometry and improves on the standard limit obtained from an uncorrelated source where the sensitivity scales as @xmath139 @xcite .
a typical experimental scheme is a mach - zenhder interferometer ( mzi ) that consists of an initial beam splitter and two interferometric paths that acquire a phase difference @xmath140 and are recombined in a final beam splitter , as shown in fig .
[ fig : mzi](b ) . the beam splitter operations could be implemented by rapid resonant @xmath70-raman pulses .
finally one extracts the phase difference measuring a phase dependent quantity . in to the twin - fock sf state @xmath137 .
( b ) the two internal states @xmath1 and @xmath2 represent the arms of an interferometer with the rotations r1 , r2 and the phase - shift @xmath140 induced by appropriate laser pulses . finally a @xmath140 dependent observable is measured at @xmath141 from which @xmath140 can be deduced.,width=321 ] to analyze this two - component system within the context of interferometry we will use the schwinger boson formalism for constructing su(2 ) operators @xcite @xmath142 and we denote @xmath143 . in many cases @xmath144 is the phase - dependent measurement made in the interferometer .
however , for the twin - fock state @xmath145 so a different quantity is required .
one possibility is the next order angular momentum @xmath146 . note that inefficient detection of such quantities can reduce the sensitivities back to the standard limit @xcite and
indirect measurement schemes of @xmath146 are needed @xcite .
if we use @xmath146 as the phase - dependent observable the resulting sensitivity can be calculated using error propagation theory .
this gives @xmath147 where @xmath148 and @xmath149 are the average and the spread respectively .
to calculate @xmath150 one can either rotate the state or the operators @xcite . since the action of the mzi is equivalent to the unitary operation @xmath151 , applying it to the operators gives @xmath152 and @xmath153 where @xmath154 is the anticommutator . for states which are zero eigenvectors of @xmath144 , such as twin - fock states , this yields @xmath155
the minimum value of this quantity occurs at @xmath156 and the sensitivity reduces to @xmath157 after using the equality @xmath158 which holds for such states . the sensitivity can be expressed entirely in terms of the one - particle density matrices @xmath159 and @xmath160 using @xmath161 .
\label{eq : jx2}\ ] ] for the initial ( uncorrelated ) mi state @xmath68 , which possesses no off - diagonal correlations @xmath162 , this yields a sensitivity at the standard limit @xmath163 . for the final sf @xmath164 , which possesses long - range correlations @xmath165 , we see that asymptotically the heisenberg limit @xmath166 is recovered
. thus the scaling @xmath167 changes from @xmath168 to @xmath169 during the melting . an inspection of eq .
( [ eq : jx2 ] ) shows that one needs non - zero values of the long - range correlations @xmath170 in order to have better scaling than @xmath52 given by the first term in eq .
( [ eq : jx2 ] ) .
we showed in @xcite that one can obtain sensitivities approaching the hl scaling even in the presence of depletion if @xmath120 .
recent results in @xcite indicate that our conclusions about the effect of depletion are also valid for @xmath171 .
moreover , we also showed in @xcite that one can obtain hl sensitivities ( @xmath172 ) if particles are lost during the melting process .
in this section we analyze the melting of the entangled mi state @xmath85 . in previous work @xcite we demonstrated that this state may melt into a maximally
entangled state if the lattice is infinitely connected . in a typical experimental set - up with only nearest neighbor hopping
, the initial permutational symmetry for @xmath47 is broken for finite @xmath48 and the final sf state is highly excited .
we begin by again considering the outcome of quantum melting in a small system . finally after applying a number of approximations we compute the momentum correlations of the final melted state for a 1d lattice . for @xmath122
the entangled mi state @xmath85 is an equal superposition of all 64 possible combinations of the atom - pair states @xmath98 and @xmath100 over @xmath21 sites .
following eq .
( [ eq : aabb_state ] ) it can be written as @xmath173 where the phase - separated states in each subspace @xmath84 are @xmath174 , \nonumber \\
{ \left | \ , \psi_{\rm ps}^6 \right \rangle } & = & \frac{1}{\sqrt{20}}\left[\sqrt{2}\,\mathbbm{s}\left({\left | \ , \bullet\circ\bullet\circ\bullet\,\circ \right \rangle}\right ) + \sqrt{12}\,\mathbbm{s}\left({\left | \ , \bullet\bullet\circ\bullet\circ\,\circ \right \rangle}\right)\right . \nonumber\\ & & + \,\left.\sqrt{6}\,\mathbbm{s}\left({\left | \ ,
\bullet\bullet\bullet\circ\circ\,\circ \right \rangle}\right)\right].\label{state_small}\end{aligned}\ ] ] here we have used @xmath175 to denote the normalized equal superposition of all equivalent configurations under the symmetry operations @xmath176 and @xmath44 . each phase separated state @xmath177 is a member of @xmath178 and is a superposition of all the @xmath179 inequivalent symmetric states within it ( note that @xmath180 , while @xmath181 ) .
the weights of this superposition eq .
( [ state_small ] ) are given by the degeneracies due to symmetrization of these inequivalent states , which we denote by @xmath182 where @xmath183 .
for @xmath122 sites in the symmetric subspace @xmath184 as a function of @xmath114 with @xmath185 throughout .
the initial state @xmath186 projected in this subspace is @xmath187 .
the adiabatic evolution of this state involves the ground state and two lowest - lying excited states . in the sf limit with @xmath120
the first - excited level is triply degenerate .
consistent with the no - crossing rule we see that energy levels within @xmath184 repel each other and only avoided crossings are seen .
( b ) the momentum distribution @xmath188 for one component ( it does not matter which due to color symmetry ) of the final adiabatically melted state @xmath189 derived from @xmath186 with @xmath122.,width=321 ] we now proceed to melt each of these phase - separated contributions separately . for the subspaces @xmath190 and @xmath191 the corresponding states @xmath192 and @xmath193 are non - degenerate mi ground states .
analogous to the melting of @xmath68 in the subspace @xmath123 exact numerical diagonalization of the two - component bhm with @xmath194 confirms that there are no level crossings and so these two states are adiabatically connected to their corresponding symmetric sf ground states @xmath195 in this notation @xmath196 we omit the dynamical phase @xmath197 acquired by the state during adiabatic melting @xcite .
in contrast subspaces @xmath198 and @xmath199 are three - fold degenerate and the states @xmath200 and @xmath201 are both composed of 3 inequivalent mi configurations . this situation becomes more typical with increasing system size . in the @xmath47 limit
these configurations are degenerate , however once @xmath48 is non - zero they split in energy . in fig .
[ fig : small_messy](a ) the energies for the ground state and three lowest lying excitations in @xmath184 are shown for a linear ramping of @xmath114 with @xmath202 .
this , together with a similar result for @xmath123 ( not shown ) , confirm that no level crossings occur .
for this reason we can map the mi configurations via their ordering in energy with their corresponding ground state and excitations in the sf limit .
the lowest energy configurations adiabatically connected to the symmetric sf ground state are @xmath203 respectively , while the two remaining configurations map to the lowest and second - lowest excitations in the sf limit .
quite generally there are many degenerate sf excitations in the @xmath110 limit . for the subspace @xmath123
the first two lowest - lying excitations are degenerate , whereas the first three are for @xmath184 . to determine the appropriate state to which each configuration maps
to we apply first - order perturbation theory ( see appendix [ app : sfpert ] ) in the inter- and intra - species interaction which splits these states and identifies the correct excitations .
this then gives for @xmath123 @xmath204 and for @xmath184 we have @xmath205 where @xmath206 and @xmath207 denote the lowest and second - lowest sf excitations in @xmath56 ( see appendix [ app : sfpert ] ) .
note that this calculation of the lowest excitations works for arbitrary @xmath21 , although it becomes cumbersome for larger system size as we would need a very large number of sf excitations . combining these results
we can write the overall final sf state which is adiabatically connected to @xmath208 with @xmath122 as @xmath209 . \nonumber\end{aligned}\ ] ] where we have now written explicitly the dynamical phases acquired by each of the states in the superposition .
an observable of experimental significance is the momentum distribution for either component .
in fact the momentum distribution @xmath210 is not dependent on the precise form of the excited states @xmath211 since all states in the same degenerate manifold have the same expectation value ( @xmath212 in this case ) . the resulting momentum distribution is shown in fig .
[ fig : small_messy](b ) , and gives the fraction of the population outside the zero - momentum mode ( i.e. depletion of the bose - condensate ) for either component as @xmath213 .
both the intra - species , @xmath214 , and the inter - species , @xmath215 , momentum space particle number correlations do depend on the precise form of the excited states @xmath211 .
neither the momentum distributions nor the momentum space particle number correlations depend on the relative dynamical phases . in fig . [
fig : mom_aa_ab](a)(b ) these correlations are plotted using the excited states determined from first - order perturbation theory with @xmath216 and @xmath124 . of the final melted state @xmath189 for one component ( it does not matter which due to color symmetry ) for @xmath122 .
( b ) the inter - species momentum space particle number correlations @xmath217 for @xmath189.,width=302 ] the final sf state obtained via adiabatic melting the entangled mi state @xmath85 is given by @xmath218 where @xmath219 if we assume that @xmath88 is fixed during melting and that the symmetries considered are exhaustive and thus no level crossings occur .
the numerical calculation for @xmath122 in section [ sec : small_messy ] hints that it is very complicated to obtain the final state for larger @xmath21 which are inaccessible with exact numerics . resolving the exact splitting of the states in each subspace
@xmath84 in the mi limit and in the sf limit is both complex and impractical for an arbitrarily large @xmath21 .
instead , we introduce here some approximations that allow us to calculate the momentum distributions and momentum correlations of the final states . in the mi regime for @xmath47 and
@xmath194 the color - symmetric ground state manifold is @xmath220 degenerate . for finite @xmath48 ,
the degeneracy is broken and @xmath104 states ( obtained after symmetrization in @xmath176 ) are not coupled by the evolution . as shown below for large @xmath21 we can approximate the initial mi state @xmath221 as an equal superposition of the @xmath104 states . in this way , using the adiabatic approximation we can map these @xmath179 states into the lowest @xmath179 states in the sf regime .
we calculate the lowest eigenstates of the bhm in the sf regime @xmath110 and their degeneracies @xmath222 such that @xmath223 . even without calculating the exact coefficients of the final @xmath179 states in the sf eigenbasis
, we can compute the momentum distribution and the momentum space particle number correlations of the final melted state to good approximation . if @xmath224 and @xmath47 the phase - separated state @xmath225 belongs to the ground state manifold @xmath103 .
it is constructed as an equal superposition of the @xmath226 atom - pair states @xmath98 and @xmath100 distributed over @xmath101 sites .
these atom - pair states belong to the ground state manifold . however , only @xmath227 of them define the symmetrized ground - state manifold @xmath228 and are not coupled by evolution for finite @xmath48 .
we showed in section [ sec : small_messy ] that e.g. @xmath229 and that for @xmath122 , the state @xmath230 is a superposition of @xmath231 states that evolve into @xmath231 different final sf states .
in general , we can calculate the dimension of the subspace @xmath178 , @xmath104 , by noticing that it is isomorphic to the number of color symmetric configurations in a bracelet ( c.f .
fig.[fig : sym ] ) with @xmath63 beads of one color and @xmath232 beads of the other color .
such mathematical structures have been extensively studied @xcite and there are analytical formulas available for their properties . for even @xmath21 ,
the number of complementable bracelets with @xmath21 beads and half of them of each color is given by @xcite @xmath233 \hspace{2 mm } \frac{m}{2 } \hspace{1 mm } \textrm{odd } \nonumber \\ & g(m/2,m)=\frac{1}{2 } [ f(\frac{m}{2})+2^{\frac{m}{2}-2}+\binom{\frac{m}{2}}{\frac{m}{4 } } ] \hspace{2 mm } \frac{m}{2 } \hspace{1 mm } \textrm{even},\nonumber\end{aligned}\ ] ] where @xmath234 and @xmath235 is euler s totient function . for bracelets with odd total number of beads
@xmath21 or even @xmath21 and @xmath236 , the number of complementable bracelets is given by @xcite @xmath237 .
\nonumber\end{aligned}\ ] ] here , @xmath238 gives the last term of the list @xmath239 generated by the cumulative application of the function @xmath240 to the elements of the input list @xmath241 and @xmath242 note that we recover @xmath181 obtained for the states @xmath243 and @xmath244 given in sec.[sec : small_messy ] .
we now need to write our initial state @xmath245 as a superposition of the @xmath104 states that evolve independently for finite @xmath48 . using the no - crossing rule we can map the initial superposition @xmath225 at @xmath47 into a final superposition @xmath246 .
the coefficients of the superposition are given by the the number of bracelet configurations @xmath182 that can be obtained by mirror inversion @xmath41 or cyclic permutation @xmath102 of any of the independent states @xmath247 in the subspace @xmath228 .
we show @xmath182 for @xmath248 in fig .
[ fig : deg ] ( a ) .
note that @xmath249 .
numerical results in fig .
[ fig : deg ] ( b ) show that most ( @xmath250 ) of the @xmath179 inequivalent configurations have the same degeneracy @xmath251 and that this behavior increases with size @xmath21 ( for @xmath252 already @xmath253 of the @xmath179 bracelets present the same degeneracy ) .
we can thus assume that the number of states isomorphic to each independent configuration is constant @xmath254 .
within this approximation , @xmath246 is thus an equal superposition of the first @xmath179 eigenstates of the sf hamiltonian . )
that belong to each of the @xmath255 independent states for finite @xmath48 for a lattice with @xmath256 and @xmath194 .
( b ) number of independent states with a given degeneracy @xmath257 .
the total number of independent states is equal to @xmath255 .
( c ) first energy levels ( @xmath258 ) in the sf regime ( @xmath110 ) and their degeneracy @xmath259 ( bar plot ) for @xmath248 .
the points ( @xmath260 ) show the quadratic approximation to the dispersion relation .
we plot only the first states such that @xmath223 for @xmath261.,width=321 ] in appendix b we show how to calculate the eigenstates @xmath262 of the hamiltonian ( [ eq : ham ] ) in the sf regime where @xmath110 .
the lowest eigenstates in ascending order in energy for large @xmath21 can be calculated efficiently assuming a quadratic approximation @xmath263 to the non - interacting single - particle energy @xmath264 . due to the color symmetry ,
the sf eigenstates @xmath265 are degenerate .
the degeneracies of the states @xmath222 can be calculated numerically .
we show in fig .
[ fig : deg ] ( c ) the eigenenergies of the first sf states and their degeneracies for @xmath248 . to obtain the momentum distribution of @xmath266 we do not need to compute the exact distribution of the final states @xmath267 in the @xmath268 basis
( see the appendix [ app : sfpert ] ) because all states in the same degenerate manifold have the same expectation value @xmath269 .
we just need to calculate the momentum distribution of the final eigenstates @xmath268 and their degeneracies @xmath222 and sum their contributions up to the highest excited state such that @xmath270 . on the other hand , we saw in sec .
[ sec : small_messy ] that we need the exact form of the final states @xmath267 in order to obtain the momentum - number correlations .
note , however , that @xmath271 where we sum over a complete degenerate manifold .
thus , if we make the further approximation that the state @xmath246 fully populates the highest excited manifold we can calculate its momentum space particle number correlations without knowing the exact states @xmath267 .
this is justified for large @xmath179 because the exact form of the highest excited states is concealed by the lower manifolds that are fully populated .
obtained via adiabatically melting the mi state @xmath272 with @xmath273 and @xmath224 .
( b ) the depletion @xmath274 of the melted state @xmath275 for @xmath224 as a function of the number of particles @xmath52 .
the solid line shows the best logarithmic fit @xmath276 to the calculated values ( shown as @xmath260).,width=321 ] we show the momentum distribution of the melted state @xmath277 in fig .
[ fig : mom ] ( a ) .
the final state is highly depleted .
a plot of the depletion @xmath278 of the term @xmath275 as a function of @xmath52 is shown in fig . [
fig : mom ] ( b ) .
it indicates the behavior of the depletion of the full state @xmath277 because this term has the highest binomial coefficient in the sum ( [ eq : psiaabbm ] ) and a very large @xmath179 .
one can see that the depletion increases very rapidly with the number of particles .
the momentum - space particle number correlations of the melted state @xmath277 are shown in figure [ fig : momc ] .
we observe that the population outside the zero momentum mode is higher for the inter - species momentum correlations .
our calculations clearly indicate that melting two component mott insulators leads to quantum states with a complicated structure and long range correlations .
a number of approximations were necessary to arrive at these results .
we believe that an exact treatment is unlikely to reveal a simple structure which allows an intuitive interpretation of the melting process . for one mode for the state @xmath279 obtained via the adiabatic melting of @xmath85 with @xmath280 and @xmath224 .
( b ) the inter - species momentum correlations @xmath281 of @xmath282 with @xmath280 .
note that a twin - fock state ( [ eq : tf ] ) would only show a central peak in both distributions.,width=302 ]
we have analyzed the quantum melting of correlated two - component mi states into the sf regime using the adiabatic approximation .
based on the experimental fact that the one component mi - sf transition can be performed near adiabatically , we have applied it to the two - mode case . using the no - crossing rule we have shown that quantum melting of a two - component mi in an optical lattice provides a viable route for engineering twin fock states useful in interferometric experiments .
the twin fock states present only macroscopic occupation of the final sf ground state although in general this is not the case .
we have shown that due to the breaking of the translational symmetry when there is nearest neighbor hopping the final sf connected to a ground initial state can be highly depleted even in the limit of adiabatic behavior . using a set of well justified approximations
, we have calculated the momentum correlations and momentum space particle number correlations for the final states obtained via melting on - site entangled states created in the mi regime .
our results clearly indicate that the melted sf states present a complicated structure and long range correlations .
further work is needed to ascertain if these correlations can be exploited for interferometry or other tasks .
m. r. thanks k. surmacz , j.j . garca - ripoll and r. a. molina for fruitful discussions and the spanish mcyt under programme juan de la cierva for support .
this research was supported by the epsrc projects ep / e041612/1 and ep / c51933/1 , and by the national science foundation under grant no .
nsf phy05 - 51164 .
m. r. and d. j acknowledge support from the eu through the strep project olaqui .
in the non - interacting sf limit where @xmath110 the excited states are made up by expelling pairs of atoms of opposite momentum from the ground state eq .
( [ eq : sf ] ) . of particular relevance to the exact calculation of the melting of @xmath85 in a small system are excitations in the subspaces @xmath69 , where @xmath283 , and @xmath284 , where @xmath285 and @xmath101 . here
we shall demonstrate that the lowest - lying excitations in @xmath69 and @xmath284 are split by the inter- and intra - species interaction to first - order in @xmath113 and @xmath114 . for the subspace @xmath286
the two lowest - lying excitations are degenerate .
these form an excitation subspace which is spanned by states where a single pair of atoms , of either component , is expelled from the condensate to equal and opposite momenta with the lowest possible magnitude as @xmath287 where @xmath288 is used to denote the smallest quasi - momenta @xmath289 .
the inter- and intra - species interaction matrix elements in this excitation subspace are @xmath290.\ ] ] upon diagonalizing this @xmath291 matrix the resulting gap and non - degenerate excitations @xmath292 and @xmath293 ( which will in general be some superposition of the states @xmath294 and @xmath295 ) are obtained in first - order perturbation theory . in a similar way for subspace
@xmath284 the three lowest - lying excitations are degenerate .
this excitation subspace is again spanned by states where a single pair of atoms is expelled from the condensate to equal and opposite momenta with the lowest possible magnitude as @xmath296 to first - order in perturbation theory these states are split according to the diagonalization of the @xmath297 matrix @xmath298.\ ] ] resulting in the non - degenerate excitations @xmath299 , @xmath300 and @xmath301 .
in the sf regime ( @xmath110 ) the ground state in the subspace @xmath84 is given by @xmath302 the excited states @xmath303 are constructed by expelling pairs of atoms of opposite momentum @xmath304 .
we can denote them by @xmath305 where @xmath306 and we omit the negative values because @xmath307 .
an efficient way of generating the lowest eigenstates in ascending order in energy for large @xmath21 is to assume a quadratic approximation @xmath308 to the dispersion relation @xmath309 for the single particle energy in a tight - binding lattice .
we can construct the eigenstates in ascending order of energy as @xmath310 where @xmath311 is an integer that runs from @xmath312 to some maximum value @xmath313 . due to the color symmetry , the excited states @xmath265 show a degeneracy @xmath222
( see for example the first excited degenerate states eq .
( [ eq : exc ] ) ) which can be calculated numerically .
o. mandel _
et al . _ ,
nature * 425 * , 937 ( 2003 ) ; d. jaksch , h .- j .
briegel , j.i .
cirac , c.w .
gardiner and p. zoller , phys .
* 82 * , 1975 ( 1999 ) .
j. dunningham , k. burnett , r. roth , w. d. phillips , quant - ph/0608242 ; k. mlmer and a. srensen , phys .
lett . * 82 * , 1835 ( 1999 ) ; a. micheli , d. jaksch , j. i. cirac and p. zoller , phys .
a * 67 * , 013607 ( 2003 ) .
a. widera _
et al . _ ,
* 95 * , 190405 ( 2005 ) ; f. gerbier , a. widera , s. flling , o. mandel , i. bloch , phys . rev .
a * 73 * , 041602(r ) ( 2006 ) ; g. thalhammer _
_ , phys .
lett . * 96 * , 050402(2006 ) .
e. altman , w. hofstetter , e. demler and m. d. lukin , new j phys * 5 * , 113 ( 2003 ) .
t. keilmann and j. j. garca - ripoll , cond - mat/0711.1483 m. rodriguez , s. r. clark and d. jaksch , j. physics , conf .
series xxx 2007 .
s. flling , f. gerbier , a. widera , o. mandel , t. gericke and i. bloch , nature * 434 * , 481 ( 2005 ) ; e. altman , e. demler and m. d. lukin , phys . rev . a * 70
* , 013603 ( 2004 ) .
e. a. donley , n. r. claussen , s. t. thompson , and c. e. wieman , nature * 417 * , 529 ( 2002 ) ; a. marte _ et al .
_ , phys.rev .
lett . * 89 * , 283202 ( 2002 ) ; m. theis _
et al . _ ,
lett . * 93 * , 123001 ( 2004 ) .
l. amico , a. osterloh and f. cataliotti , phys .
lett.*95 * , 063201 ( 2005 ) .
z. nussinov and g. ortiz , cond - mat/0702377 l. d. landau and e. m. lifshitz , _ quantum mechanics : non - relativistic theory _ ( butterworth - heinemann , 2005 ) vol 3 , 79 and 96 .
b. oles and k. sacha , cond - mat/0701.212 the sequence is 1 , 2 , 3 , 7 , 13 , 35 , 85 , 257 , 765 , 2518 , 8359 , 28968 , 101340 , 361270 , 1297879 , 4707969 , 17179435 , 63068876 , 232615771 , 861725794 .. and can be found in http://www.research.att.com/@xmath314njas/sequences/a006840 .
http://www.research.att.com/@xmath314njas/sequences/a073020 , http://www.research.att.com/@xmath314njas/sequences/a078925 .
the full series and the generating algorithm is given in http://www.research.att.com/@xmath314njas/sequences/a052307 . | we analyze the outcome of a mott insulator to superfluid transition for a two - component bose gas with two atoms per site in an optical lattice in the limit of slow ramping down the lattice potential .
this manipulation of the initial mott insulating state transforms local correlations between hyperfine states of atom pairs into multiparticle correlations extending over the whole system .
we show how to create macroscopic twin fock states in this way an that , in general , the obtained superfluid states are highly depleted even for initial ground mott insulator states . |
the fundamental quantity in galaxy dynamics is the distribution function ( df ) , which specifies the distribution of the stars in a galaxy over position and velocity . by jeans theorem ,
the df is a function of the isolating integrals that are conserved by the corresponding potential ( e.g. , lynden - bell 1962 ; binney 1982 ) .
axisymmetric galaxies , which we will study here , conserve at least the two classical integrals of motion , the energy @xmath1 and the vertical component of the angular momentum @xmath2 .
the df can generally not be measured directly , but since observationally accessible quantities , such as the projected density and the line - of - sight velocity , are simple moments of the df , photometric and kinematic observations can provide information on its properties . in some cases , the part of a two - integral distribution function that is even in the velocities , @xmath3 , can be obtained analytically by using integral transforms to solve the relation between the df and the intrinsic density @xmath4 , where @xmath5 are the usual cylindrical coordinates . the odd part can be found similarly from @xmath6 , where @xmath7 is the mean streaming velocity .
these laplace ( lynden - bell 1962 ; lake 1981 ) , stieltjes ( hunter 1975b ) and laplace - mellin ( dejonghe 1986 ) transforms have the drawback that numerical implementation is difficult and that they require @xmath8 and @xmath9 to be explicit functions of the potential @xmath10 and the cylindrical radius @xmath11 . a more general formalism ( hunter & qian 1993 ; qian et al . 1995 , hereafter q95 ) ,
the hq - method , uses contour integration instead of integral transforms .
this means it is simpler to implement , does not explicitly require @xmath12 and @xmath13 , but a suitable contour for the integration has to be chosen , which may not always be at hand . because of the drawbacks of these analytical formalisms , numerical methods that can be applied to arbitrary potential - density pairs are more attractive .
various methods have been developed , many of which circumvent knowledge of the df by solving the jeans equations directly ( van der marel et al . 1994 ; magorrian 1995 ) , while others assume that the df can be represented by a superposition of basis functions ( dehnen & gerhard 1994 ; kuijken 1995 ) .
accurate estimates of the mass - to - light ratio @xmath14 can be obtained in both ways ( van der marel 1991 ; shaw et al .
1993 ; van den bosch 1996 ) .
the predicted central black hole masses that are obtained with two - integral models ( magorrian et al .
1998 ) seem to overestimate the true values , but are still very useful to narrow down the parameter range for more sophisticated modeling ( van der marel et al .
1998 , hereafter vdm98 ; gebhardt et al .
2000 ; bower et al .
2001 ) .
numerical orbit integrations and observations of , e.g. , the anisotropy of the stellar dispersions in the solar neighbourhood show that , in addition to @xmath1 and @xmath2 , a third integral is conserved for most orbits . in separable potentials ,
this third integral is exact and has a closed form ( e.g. , kuzmin 1956 ; de zeeuw 1985 ) , allowing a direct calculation of the df ( e.g. , dejonghe & de zeeuw 1988 ) .
the most general family of these potentials corresponds to flattened mass models with constant - density cores ( de zeeuw , franx & peletier 1986 ) , which provide a poor description of the inner regions of most galaxies , since these contain a central density cusp ( lauer et al.1995 ) .
we conclude that many of the existing two - integral , as well as the analytical three - integral methods , are applicable to a limited range of galaxy models .
a flexible method for calculating numerical galaxy models was designed by schwarzschild ( 1979 , 1982 ) , who represented the df numerically by the occupation numbers in a superposition of building blocks .
there are no restrictions on the density or the potential , and no a priori assumptions have to be made about the shape or the degree of anisotropy of the galaxy .
schwarzschild s original implementation was aimed at reproducing a given triaxial density distribution and was subsequently applied to a large variety of galaxy models , from spherically and axially symmetric ( richstone & tremaine 1984 ; levison & richstone 1985 ) to triaxial density distributions ( e.g. , vietri 1986 ; statler 1987 ; schwarzschild 1993 , merritt & fridman 1996 ; siopis & kandrup 2000 ) .
more general versions of the method can also reproduce kinematical observations of spherically and axially symmetric galaxies that obey up to three integrals of motion ( zhao 1996 ; rix et al .
1997 , hereafter r97 ; cretton et al . 1999 , hereafter c99 ; cretton et al .
2000 ; hfner et al .
2000).=-2 in most implementations of schwarzschild s method , orbits are used as individual building blocks . in non - separable potentials the third integral of motion is not known explicitly , so that the orbital properties can only be obtained by solving the equations of motion numerically .
this orbit integration has to be carried out until the relevant quantities have averaged out and do not change upon a new time - step ( pfenniger 1984 ; copin et al .
2000 ) . especially for orbits that are near - stochastic , or that have nearly commensurate fundamental frequencies ,
this condition is difficult to achieve and the orbit integration can be very time - consuming . furthermore , an orbit fills a region in space that can have sharp edges , depending on the combination of integrals of motion that it obeys .
a superposition of a finite ( and relatively small ) number of orbits can therefore show artefacts that are caused by these edges .
one way to solve this is to ` blur ' orbits in phase - space randomly , another is to use building blocks that are smoother ( e.g. , zhao 1996 ; merritt & fridman 1996 ; r97 ; c99 ) .
of particular interest are the so - called isotropic and two - integral components ( ics and tics ) .
isotropic components , which are completely specified by their energy @xmath1 , fill the volume inside the equipotential at @xmath1 , while tics , which are fixed once @xmath1 and @xmath2 are chosen , completely fill the corresponding zero - velocity curve ( zvc ) .
the advantages of these building blocks is apparent : both the equipotential and the zvc are smooth surfaces , and ics and tics can be considered as weighted combinations of all orbits with the same energy @xmath1 ( or @xmath1 and @xmath2 ) , including the irregular orbits .
these building blocks can be used in schwarzschild s method in two ways .
they can be added to every energy or every combination @xmath15 of a ( three - integral ) orbit library , which will ` automatically ' take care of the stochastic orbits ( zhao 1996 ; r97 ; c99 ) .
alternatively , a fully isotropic or two - integral model can be constructed by using only ics or tics as building blocks .
c99 suggested that this might be a practical way of constructing such models , and derived some analytic properties , but did not pursue these models further .
we do so in this paper and concentrate on the two - integral case , since the properties of the isotropic components follow from those of the tics by taking @xmath16 . in [ sec
: method ] , we collect the properties of the tics , and summarize how we use them in schwarzschild s method .
the numerical aspects are discussed in detail in [ sec : implementation ] .
we show in [ sec : examples ] that our method can reproduce a model with a known analytical distribution function with high accuracy .
we also study a set of mass models with a central black hole , and show that our method is able to detect when a self - consistent solution does not exist .
we summarize our conclusions in [ sec : summary ] .
applications of our software to model the observed two - dimensional surface brightness and kinematics of nearby elliptical galaxies observed with sauron and stis are described in two follow - up papers ( verolme et al .
2002 , in preparation ; mcdermid et al .
2002 , in preparation ) .
we introduce the two - integral components ( tics ) and calculate their observable properties , and then describe how we use them to construct two - integral models for galaxies .
the properties of ics follow by taking @xmath16 .
a tic is completely specified by its energy @xmath17 and angular momentum @xmath18 ( where @xmath19 is a label for the tic ) and is defined by its distribution function @xmath20}=\!\left\{\begin{array}{l l } \ ! { \cal c}\,\delta(e\!-\!e_j)\,\delta(l_z\!-\!l_{z , j } ) & { \rm inside\,\,\,zvc,}\\ \ ! 0 & { \rm elsewhere,}\end{array}\right .
\label{eq : tic - df}\end{aligned}\ ] ] with @xmath21 a constant that is used to normalize the mass of the tic .
the abbreviation zvc denotes the zero - velocity curve in the @xmath22-plane , defined as the locus of points for which @xmath23 where @xmath24 is the effective potential .
the zvc defines a toroidal volume around the @xmath25-axis ( fig .
[ fig : zvc ] ) with inner and outer radii @xmath26 and @xmath27 .
the size of the central hole is determined by the value of the angular momentum : when @xmath16 , there is no hole at all ( @xmath28 ) and the torus fills the equipotential at the given energy completely , so that the tic reduces to the isotropic component ( ic ) at that energy . for the maximum value @xmath29 ( with @xmath30 the circular velocity at radius @xmath31 ) the torus reduces to the circular orbit at @xmath17 , and @xmath32 .
the density @xmath33 of a tic is defined as the integral of the df over all velocities , and is given by @xmath34 the velocity moments are defined as the average of ( powers of ) the velocity components over all velocities . from symmetry arguments and from the fact that @xmath35 , the only nonzero first and second moments are @xmath36 , \label{eq : moments - def}\end{aligned}\ ] ] where these expressions are valid inside the zvc .
all moments vanish outside the zvc .
we assume that the galaxy is observed at an inclination @xmath37 , so that the projected coordinates are given by @xmath38 where @xmath39 and @xmath40 are chosen in the plane of the sky , with @xmath40 along the projected minor axis of the galaxy , and @xmath41 is measured along the line of sight .
the projected density @xmath42 is the integral of the intrinsic density along the line of sight @xmath43 which can be evaluated analytically .
the intrinsic density of a tic is only defined for points that lie inside the toroidal volume defined by the zvc , so that the integral ( [ eq : surf - den - def ] ) is defined on the @xmath41-intervals delimited by the two or four points where the line - of - sight crosses the zvc ( fig .
[ fig : zvcproj ] ) .
we give all expressions for four intersections , labeled as @xmath44 .
the case of two intersections follows by taking @xmath45 .
we obtain @xmath46 with @xmath47 , \qquad k=1,\ldots,4 , \label{eq : surfcart}\ ] ] where we have defined @xmath48 and @xmath49 the limiting case of edge - on viewing ( @xmath50 ) is given by @xmath51 the velocity profile @xmath52 , defined as the distribution of stars over velocity @xmath53 at the position @xmath54 , is the integral of @xmath0 over the line of sight and the two velocity components @xmath55 and @xmath56 in the plane of the sky . since the distribution function of a tic is the product of two @xmath57-functions , the triple integral in the expression for @xmath58 reduces to a single integral , @xmath59 where the remaining integration in @xmath41 is over the same intervals as in eq .
( [ eq : surf - den - def ] ) , and @xmath60 is the jacobian for the change of variables from @xmath61 to @xmath62 . it can be written as ( c99 ) @xmath63 with @xmath64(x'^2 \cos^2 i\!+\!d^2_{z'})\!- \!l^2_{z , j } , \nonumber\\ b\!\!&=&\!\ ! -2 x'\,l_{z , j}\,\sin i ,
\\ c\!\!&=&\!\ ! -r'^2_{z ' } , \nonumber\end{aligned}\ ] ] with @xmath65 and @xmath66 from eqs ( [ eq : r - def ] ) and ( [ eq : d - def ] ) , respectively .
the integral of the velocity profile over all allowed values of @xmath53 is equal to @xmath42 .
the projected velocity moments @xmath67 ( with @xmath68 ) can be evaluated similarly , resulting in @xmath69^{v_+}_{v_- } \nonumber\\ \langle v^2_{z'}\rangle \!\!&=&\!\!-\frac{\cal c}{2}\,\int\!\!\frac{dz ' } { r'^2_{z'}}\left[(v-\frac{3b}{2c})j_{e_jl_{z , j}}+\right.\nonumber\\ & \,&\left.\frac{3\,b^2 - 4\,a\,c}{4\,c\,r'_{z'}}\,{\cal u}_{v_{z'}}\right]_{v_-}^{v_+},\end{aligned}\ ] ] with @xmath70 , \label{eq : u}\end{aligned}\ ] ] and @xmath60 is given in eq .
( [ eq : jacobian ] ) .
the values @xmath71 and @xmath72 are the limits of the integration in @xmath53 .
the individual tics can be used to model galaxies with schwarzschild s orbit superposition method .
the procedure follows the usual steps ( e.g. , c99 ) .
first determine the density distribution , either from a theoretical model or by deprojecting the observed projected surface density of a galaxy and choosing a mass - to - light ratio . then calculate the corresponding gravitational potential from poisson s equation .
in this potential , define a collection of tics by specifying the integrals @xmath73 .
jeans theorem guarantees that a sampling of the full range in energies and angular momenta results in a tic library that is representative for this potential .
this can be achieved by sampling the radius on a grid of @xmath74 circular radii @xmath75 , chosen e.g. logarithmically on a range that includes @xmath76% of the mass . from this
, the energy grid @xmath77 follows by calculating the energy of the circular orbit at the @xmath75 .
the @xmath78 angular momenta can then be chosen linearly between the minimum ( @xmath16 , orbits confined to the meridional plane ) and maximum ( @xmath29 , the circular orbit ) at every @xmath79 .
this choice of grids defines a library of @xmath80 tics .
schwarzschild s method then determines the weighted sum of the tics in this library that is closest to a data set @xmath81 ( generally consisting of both photometric and kinematic data ) .
if @xmath82 is the contribution of the @xmath19-th tic to the @xmath37-th aperture , the problem reduces to solving for the weights @xmath83 in @xmath84 where @xmath85 is the number of apertures for which observations are available .
the @xmath83 , which determine the weight of each individual tic in this superposition , are found by using a least - squares method , with the additional constraint that @xmath86 ( e.g. , r97 ) .
furthermore , @xmath87 , since every tic is labeled by a unique combination of @xmath1 and @xmath2 .
the df that corresponds to this set of weights can then be found from ( vandervoort 1984 ) @xmath88 where @xmath89 is the mass of the tic and @xmath90 is the area of the cell around @xmath1 and @xmath2 in integral space . since the matrix problem of eq .
( [ eq : matrix ] ) is often ill - conditioned , the weights and therefore the reconstructed df may vary rapidly as a function of @xmath1 and @xmath2 .
this can be avoided by adding regularisation constraints to the problem ( e.g. , zhao 1996 ; r97 ) .
these constraints force the weights @xmath91 towards a smooth function of @xmath1 and @xmath2 by minimizing the @xmath92-th order derivatives @xmath93 .
the degree of smoothening is determined by the order @xmath92 and by the maximum value that the derivatives are allowed to have .
the predicted observables that are obtained by calculating the weighted sum of the tics usually differ from the observed values @xmath81 .
there are various reasons for this : the df of the galaxy may not be of the form @xmath0 , model parameters such as inclination , stellar mass - to - light ratio or central black hole mass may not be chosen properly , the observations can contain systematical errors , and the calculations are done on finite grids , which introduces discretization effects . to measure the discrepancy
, we define a quality - of - fit parameter , @xmath94 , as @xmath95 in which @xmath96 is the model prediction at the @xmath37-th aperture and @xmath97 is the ( observational ) uncertainty that is associated with @xmath81 .
the numerical implementation of schwarzschild s method for three - integral axisymmetric models including kinematic constraints is described in detail in c99 . here
we discuss the simplifications for models containing tics , which can be used to speed up the calculations .
readers who are more interested in applications , can skip to [ sec : examples ] .
all observables are calculated on grids that are adapted to the size of the galaxy and to the constraints that have to be reproduced .
the meridional plane density is stored on a polar grid in the meridional plane , with a radial grid that is sampled logarithmically on the same range as the @xmath75 and a linear grid in polar angle between @xmath98 and @xmath99 .
the projected density is calculated on a cartesian grid in the sky plane , and the velocity profile is stored on a three - dimensional cartesian grid in @xmath100 and @xmath53 .
the bin sizes of both grids are adapted to the resolution of the kinematic data to be reproduced .
the observables are averaged over these grids , so that a continuous observable @xmath101 is represented by a discrete set of averaged grid entries @xmath102 all observables are related to the df through triple ( intrinsic density and velocity profile ) or quadruple ( projected density ) integrals , two of which are trivial for tics , because of their @xmath57-function behaviour .
this means that the averages @xmath103 can be written as no worse than quadruple integrals over the region enclosed by the cross - section of the zvc and the grid cell .
depending on the relative position of the grid cell and the zvc , the four integrals can be interchanged and the expressions can be simplified . in the following ,
we show how this is achieved for all observables .
the simple form of the meridional plane density ( [ eq : tic - density ] ) leads to considerable simplifications in the evaluation of the averaged meridional plane density @xmath104 .
the contribution of a general grid cell bordered by @xmath105 and @xmath106 is @xmath107 for cells that are completely surrounded by the zvc , which are relatively numerous for tics with small energies ( corresponding to large zvcs ) , this integral can be evaluated and is given by @xmath108 when , on the other hand , @xmath17 is large , so that the zvc is less extended , the cell is ( much ) larger than the zvc .
this means that the integration region in eq .
( [ eq : numerical ] ) can be adjusted to match the zvc size better , which speeds up the calculations considerably .
table [ tab : table1 ] and fig .
[ fig : zvcgrid ] show for which cells one can apply these two simplifications . calculating all
meridional plane densities for a library of 70 x 20 tics takes a few minutes on a machine with a 1 ghz processor .
.[tab : table1 ] for each cell in fig .
[ fig : zvcgrid ] ( and for one cell that is not shown , labeled with @xmath109 ) , the simplifications in the calculation of the averaged meridional plane density are given .
the equation that has to be used is listed , or a value of @xmath98 when the cell is located outside the zvc . when eq .
( [ eq : numerical ] ) applies , the new integration limits are given . [ cols="<,^,^,^,^,^",options="header " , ] in the previous subsections , we described how to obtain the average of the tic observables over the cells of the storage grids .
these quantities can only be compared with observational data when point spread function ( psf ) effects are taken into account correctly .
this means that observationally accessible quantities such as the projected density and the velocity profile have to be convolved with the appropriate psf , which is usually described in terms of a ( sum of ) gaussian functions .
the extra two - dimensional integration that this convolution introduces can be carried out analytically when it is combined with the average over the storage cell ( see appendix d of q95 ) .
this means that an extra term , consisting of error functions , appears in the expressions for the observables that were given previously .
to test our method , we apply the numerical implementation to three theoretical potential - density pairs .
we investigate whether our formalism rules out the same models as previous methods and determine the accuracy with which a theoretical distribution function can be reproduced .
this also allows us to study the behaviour of the @xmath94 ( [ eq : chisq ] ) and the regularisation algorithm , which is similar to the one that is used in the three - integral schwarzschild methods ( c99 ) .
applications to two - dimensional photometry and kinematics of galaxies are described in verolme et al . and mcdermid et al .
we first test the ability of our method to reproduce a known @xmath0 , given @xmath8 and @xmath10 , and investigate the effect of regularisation on the results .
we use the density and potential of the kuzmin kutuzov ( 1962 ) model @xmath110 with @xmath111 .
the density distribution is nearly spheroidal , and many of the properties of these models , including the projected surface density , can be evaluated explicitly . the even part of the @xmath112 distribution function for the kuzmin
kutuzov models is given by ( dejonghe & de zeeuw 1988 ; batsleer & dejonghe 1993 ) @xmath113}{(1 - 2\,a\,e\,t\,\sqrt{1-t^2 } + \epsilon\,l_z\,t\sqrt{2\,h\,e})^5},\nonumber\end{aligned}\ ] ] with @xmath114 this integral can be expressed in terms of elementary functions , but this leads to an exceedingly lengthy expression , and it is best evaluated numerically .
a grey - scale plot of the kuzmin - kutuzov df as a function of energy and fractional angular momentum is shown in the upper panel of fig .
[ fig : dfs ] . we attempted to reproduce this df by calculating a two - integral schwarzschild model that fits the meridional plane mass ( [ eq : rhokk ] ) .
we used @xmath115 and @xmath116 , including @xmath16 , and adopted @xmath117 radial and @xmath118 angular bins , like c99 . with these parameters ,
the model density was fitted with very high accuracy ( fractional difference smaller than @xmath119 ) , but the tic weights vary dramatically with energy and angular momentum , resulting in a very jagged df ( the middle panel in fig . [ fig : dfs ] ) .
including regularisation constraints enforces the weights to vary smoothly with tic index , resulting in a df that reproduces the theoretical df to within 0.1% .
this is illustrated in the lower panel of fig .
[ fig : dfs ] , which was obtained by requiring the same fitting accuracy ( @xmath120 in eq .
[ eq : chisq ] ) for both the meridional plane masses and the regularisation constraints .
this means that both constraint types have equal influence on the final model , so that the model df is smooth , but still reproduces the density . in order to measure
how fitting to the mass determines the kinematics of the model , we have calculated a theoretical velocity profile on the minor axis directly from the distribution function eq .
( [ eq : gamdf ] ) , and compared this with the predicted profile , given by the superposition of all the individual tic velocity profiles . fig .
[ fig : vpcomparison ] demonstrates that the agreement between the two curves is significantly better than the typical measurement error in observed velocity profiles ( van der marel & franx 1993 ; joseph et al .
2001 ) .
we now consider @xmath0 models with a central black hole , represented by a point mass with potential @xmath121 . for spherically symmetric galaxies , there is a direct relation between the luminous density distribution and the isotropic distribution function @xmath122 ( eddington 1916 ) .
when the df is required to be non - negative , this relation implies that no spherical , isotropic galaxies with a central black hole exist for cusp slopes shallower than @xmath123 ( e.g. , binney & tremaine 1987 ) . an indication that this restriction disappears when a higher degree of anisotropy is allowed is illustrated by the @xmath124 ( with @xmath125 the angular momentum ) that is obtained by placing all stars on circular orbits ( hunter 1975a ) . to investigate this effect for general anisotropic distribution functions @xmath0
, q95 applied the hq - method to density distributions that are stratified on similar concentric spheroids .
they considered models with a density profile that follows a double power law , given by @xmath126^{\beta},\end{aligned}\ ] ] where @xmath127 and @xmath128 are the central density and cusp slope , @xmath129 with @xmath130 the intrinsic axis ratio , and @xmath131 is the break radius .
q95 found that all oblate ( @xmath132 ) double power law profiles correspond to a non - negative distribution function .
however , adding a central black to these models limits the central cusp slope to @xmath133 , as in the spherical case .
this restriction is independent of black hole mass , as long as @xmath134 .
de bruijne et al . ( 1996 ) showed that this result is valid for a much larger family of cusped spheroidal density profiles with a central black hole : when the potential is spherically symmetric , the distribution function @xmath135 corresponds to the density @xmath136 in which the @xmath137 are given in terms of @xmath138-functions by eq .
( 17 ) of de bruijne et al .
the argument of a @xmath138-function must be positive , which implies that , near a black hole , where the potential falls off as @xmath139 ( @xmath140 in their notation ) , eqs ( [ dfjos ] ) and ( [ densjos ] ) exist only for @xmath141 we see from eq .
( [ dfjos ] ) that the limiting case of a two - integral df @xmath0 is given by @xmath142 , so that eq .
( [ condition1 ] ) reduces to @xmath143 this means that density distributions that can be expanded in terms of @xmath144 must obey condition ( [ condition2 ] ) on each of the terms of the expansion in order to have a physical df near the black hole .
the lowest - order term of this expansion , usually with @xmath145 , places the most stringent condition on @xmath128 , i.e. @xmath133 .
these results show that analytical methods are very powerful for deriving criteria for the existence of models .
in contrast , schwarzschild s method always finds an orbit superposition that fits the constraints to some degree .
the @xmath94 ( [ eq : chisq ] ) , determined from residual errors , is the only measure for the existence of the underlying distribution function .
a priori , it is not clear whether this parameter is able to reproduce the sharp limits on the model parameters that are found with analytical methods .
for this reason , we constructed two - integral models for the density profile ( [ eq : double_power_law ] ) .
we added a central point mass @xmath146% of the total galaxy mass , mimicking a central black hole , and varied the cusp slope at constant flattening . as in
[ sec : kk ] , we took @xmath115 and @xmath116 ( again including @xmath16 ) , and adopted 16 radial and 7 angular bins.=-2 fig .
[ fig : twoint ] shows the resulting @xmath94 as function of @xmath128 for self - consistent two - integral models . while @xmath147 for values of @xmath128 below @xmath123 , it increases rapidly as soon as @xmath128 increases above @xmath123 .
this shows that our method recovers the models of q95 , and that our goodness - of - fit parameter @xmath94 is a good discriminator of the ( non)-existence of a given model .
we also investigated the axisymmetric version of the triaxial models with central cusps that were introduced by de zeeuw & carollo ( 1996 ) .
the potential and density are given in appendix a. the behaviour of these models is completely determined by their axis ratios at small and large radii , @xmath148 and @xmath149 , and the central cusp slope @xmath128 .
the density distribution is dimpled on the symmetry axis , and , for some choices of @xmath150 , even becomes negative . in advance
, we can therefore rule out a ( small ) range of parameters for which the df does not exist .
the physical range of @xmath148 decreases when @xmath151 is decreased . at @xmath152 , which is considered here ,
models with @xmath153 show this behaviour .
we study the range @xmath154 .
we added a central black hole with a mass @xmath155% of the total mass , and calculated self - consistent two - integral models using the same grid settings as in [ sec : kk ] .
[ fig : alpha_q ] shows the contours of constant @xmath94 as a function of @xmath128 and @xmath148 .
as expected , the @xmath94 is zero for values @xmath156 and rises rapidly above @xmath157 , indicating that self - consistent two - integral distribution functions do not exist at these values of the central cusp slope .
however , for @xmath158 , the transition to @xmath159 occurs at values of @xmath160 , which means that these flatter models have an even larger freedom in possible central cusp slopes .
the points that contribute most to this nonzero @xmath94 are located in the center of the model , i.e. , near the black hole , and experiments with a denser sampling in the integrals of motion give similar results .
the limiting behavior of the density is ( cf.eq ( [ rhopl ] ) @xmath161,\end{aligned}\ ] ] so that the importance of the constant term in the expansion ( [ densjos ] ) decreases when @xmath148 decreases .
this may gradually relax the restriction on @xmath128 in the numerical model .
a similar effect has been found recently in oblate sridhar - touma models ( jalali & de zeeuw 2001 ) , which resemble the power - law models in many respects . however , the values of @xmath148 for which this occurs are much smaller than the observed flattenings of early - type galaxies .
we have presented an alternative to the existing analytical and numerical methods for calculating two - integral distribution functions .
it is based upon schwarzschild s orbit superposition method and able to deal with arbitrary density profiles and galaxy potentials . instead of orbital building blocks
, we use the so - called two - integral components , which are smoother and implicitly include stochastic orbits . due to their delta - function behaviour ,
the observables can be calculated ( semi-)analytically , which speeds up the calculations considerably .
we checked that the method is able to reproduce a known combination of potential , density and distribution function and tested the regularisation algorithm in the process .
this test shows that fitting the meridional plane masses alone is not enough to constrain the df , while including a ( modest ) amount of regularisation is enough to smoothen the df toward the theoretical curve .
we have also tested the @xmath94 ( [ eq : chisq ] ) , which measures the quality - of - fit of the resulting model , against analytical investigations .
this demonstrates that a high value of @xmath94 can be taken as a sure indication that the model does not exist .
these results show that our method is flexible and reproduces analytical results very closely .
additionally , many of the characteristics of the implementation that we used , such as the regularisation method and the @xmath94-parameter , are used in more general ( three - integral ) models .
regularisation is therefore also a very useful and necessary tool to smoothen the distribution function that is found by these methods , and the @xmath94-parameter is a useful diagnostic to test the existence of the resulting models .
applications in which we calculate fully three - integral models ( including tics ) of observed kinematics will follow in a subsequent paper .
* acknowledgements * + it is a pleasure to thank glenn van de ven , michele cappellari and roeland van der marel for useful discussions and a critical reading of the manuscript .
furthermore , we are very thankful to roeland van der marel for kindly providing his three - integral schwarzschild software . 99 batsleer p. , dejonghe h. , 1993 , a&a , 271 , 104 binney j.j .
, 1982 , mnras , 201 , 15 binney j.j .
, tremaine s.d . , 1987 .
galactic dynamics , princeton university press , princeton bower g.a .
et al . , 2001 , apj , 550 , 75 van den bosch f.c . ,
de zeeuw p.t . , 1996 , mnras , 283 , 381 de bruijne j.h.j . , van der marel r.p .
, de zeeuw p.t . , 1996 , mnras , 282 , 909 copin y. , zhao h.s . , de zeeuw p.t
. , 2000 , mnras , 318 , 781 cretton n. et al . , 1999 ,
apjs , 124 , 383 ( c99 ) cretton n. , rix .
, de zeeuw p.t . , 2000 , apj , 536 , 319 dehnen w. , gerhard o.e . , 1994 , mnras , 268 , 1019 dejonghe h. , 1986 , phys .
133 , 218 dejonghe h. , de zeeuw p.t . , 1988 , apj , 333 , 90 eddington a.s . , 1916 , mnras , 76 , 572 gebhardt k. , et al . , 2000 ,
aj , 119 , 115 hfner r. , et al . , 2000 , mnras , 314 , 433 hunter c. , 1975a , iaus 69 , 195h hunter c. , 1975b , aj , 80 , 783 hunter c. , qian e.e . , 1993 , mnras , 262 , 401 jalali m.a .
& de zeeuw , p.t . ,
mnras , submitted joseph c.l .
et al . , 2001 ,
apj , 550 , 668 kuijken k. , 1995 , apj , 446 , 194 kuzmin g.g . , 1956 , astr .
, 33 , 27 kuzmin g.g . ,
kutuzov , s.a . , 1962 , bull .
, 27 , 82 lake g. , 1981 , apj , 243 , 111 lauer t. , et al . , 1995 ,
aj , 110 , 2622 levison h.f . , richstone d. , 1985 , apj , 295 , 340 lynden - bell d. , 1962 , mnras , 123 , 447 van der marel r.p . , 1991 , mnras , 253 , 710 van der marel r.p . & franx m. , 1993 , apj , 407 , 525 van der marel r.p .
et al . , 1994 ,
mnras , 271 , 99 van der marel r.p .
et al . , 1998 ,
apj , 493 , 613 ( vdm98 ) magorrian j. , 1995 , mnras , 277 , 1185 magorrian j. et al . , 1998 ,
aj , 115 , 2285 merritt d.r .
, fridman t. , 1996 , apj , 460 , 136 pfenniger d. , 1984 , a&a , 141 , 171 qian e.e .
. , 1995 , mnras , 274 , 602 ( q95 ) richstone d. , tremaine s. , 1984 , apj , 286 , 27 rix h .- w . ,
et al . , 1997 , apj , 232 , 236 ( r97 ) schwarzschild m. , 1979 , apj , 232 , 239 schwarzschild m. , 1982 , apj , 263 , 000 schwarzschild m. , 1993 , apj , 409 , 563 shaw m. , et al . , a&a , 268 , 511 siopis c. , kandrup h. e. , 2000 , mnras , 319 , 43 statler , t.s . , 1987 , apj , 321 , 113 vietri m. , 1986 , apj , 306 , 48 de zeeuw p.t . , 1985 , mnras , 216 , 273 de zeeuw p.t . , carollo c.m . , 1996 , mnras , 281 , 1333 zhao h.s . , 1996 , mnras , 283 , 149
the axisymmetric limit of the triaxial models introduced by de zeeuw & carollo ( 1996 ) is defined by the potential @xmath162 with @xmath163 } & { \rm for}\,\,\alpha\neq -2\end{array}\right.\nonumber\\ v(r)&=&-\frac{a\,r^{2+\alpha}}{(r+r_2)^{4+\alpha}},\end{aligned}\ ] ] where @xmath164 and @xmath27 are constants , we have chosen @xmath165 , and consider @xmath166 .
the corresponding density distribution is @xmath167\end{aligned}\ ] ] with @xmath168 } { 4\pi(r+r_2)^{6+\alpha}}.\end{aligned}\ ] ] the constants @xmath164 and @xmath169 are related to the axis ratios of the density distribution at small and large radii , @xmath148 and @xmath151 , respectively , by @xmath170^{1/{(4+\alpha)}},\end{aligned}\ ] ] [ lastpage ] | we describe a practical method for constructing axisymmetric two - integral galaxy models ( with distribution functions of the form @xmath0 , in which @xmath1 is the orbital energy , and @xmath2 is the vertical component of the angular momentum ) , based on schwarzschild s orbit superposition method .
other @xmath0-methods are mostly based on solving the jeans equations or on finding the distribution function directly from the density , which often places restrictions on the shape of the galaxy . here , no assumptions are made and any axisymmetric density distribution is possible .
the observables are calculated ( semi-)analytically , so that our method is faster than most previous , fully numerical implementations .
various aspects are tested extensively , the results of which apply directly to three - integral schwarzschild methods .
we show that a given distribution function can be reproduced with high accuracy and investigate the behaviour of the parameter that is used to measure the goodness - of - fit .
furthermore , we show that the method correctly identifies the range of cusp clopes for which axisymmetric two - integral models with a central black hole do not exist .
[ firstpage ] galaxies : elliptical and lenticular , cd - galaxies : kinematics and dynamics - galaxies : structure |
motivation for our work @xcite comes from various speculations that came up mainly from : * the analogue spacetime programme @xcite ; * various different approaches to `` quantum gravity '' , + ( loop quantum gravity @xcite , string models @xcite , causal dynamical triangulations @xcite ) ; * various different approaches to `` quantum gravity phenomenology '' , + ( dsr @xcite , vsr @xcite , noncommutative geometry @xcite ) ; * various different approaches to `` exotic '' cosmologies @xcite .
many of these approaches strongly suggest the possibility of high energy lorentz symmetry breaking @xcite and/or birefringence ( or multi - refringence ) @xcite .
if we wish to create a geometric theory encoding causal relations more complicated than those of special relativity , then we must unfortunately depart from the standard concept of pseudo - riemannian geometry , and adopt something more general .
this more general geometry is almost inevitably a non - positive definite analogue of finsler geometry @xcite , that is , a geometry which can most descriptively be called a pseudo - finsler geometry @xcite .
( one might also try to generalize the spacetime in a slightly different direction by adopting the notion of an `` area metric '' @xcite , but we shall not take that particular route for now . )
one particularly attractive route to constructing lorentz symmetry breaking geometries is to consider analogue spacetimes .
these are classical systems providing us , at the macroscopic level , with emergent phenomena similar to gravity , while being radically different at the microscopic level and at the same time are physically well understood @xcite .
they provide us with analogues of many aspects of quantum gravity phenomenology , and are also possible sources of experimental tests for quantum gravity phenomenology .
( we can even hope for the detection of hawking radiation @xcite . ) in our specific case @xcite we chose a particular analogue spacetime model that of a birefringent crystal @xcite providing us with explicit lorentz symmetry breaking relations , which at the same time has a simple enough structure ( the dispersion relations factorize , in general , to give a two - cone structure ) so as to give us a chance to create in an intuitive way a full spacetime pseudo - finsler geometry ( with both metric and co - metric structures ) .
birefringent crystals have in the past traditionally been analysed in terms of 3-dimensional _ spatial _ finsler geometry .
( see , for instance , antonelli , ingarden , and matsumoto @xcite , and duval @xcite . ) but these analyses are purely _ spatial _ they effectively pick a fixed inertial reference frame , the rest frame of the crystal , and hence pick a preferred @xmath0 spacelike slices , so these approaches do not cross the boundaries of classical nonrelativistic mechanics . in contrast , we wish to construct a full @xmath1 _ spacetime _ pseudo - geometry based on birefringent crystals a fully @xmath2-dimensional construction .
how should the theory of crystal optics provide us with a lorentz symmetry breaking geometry ?
the theory gives us a clear ideas how the waves propagate , specified by the possible four - wavevectors @xmath3 with their corresponding phase velocities @xmath4 , and a clear idea about how the energy is transferred , specified by the possible group three - velocities @xmath5 .
the group three - velocity can most naively be extended to a group four - velocity by the prescription @xmath6 .
more generally , if we choose to parameterize the integral curves of the group velocity by some arbitrary parameter @xmath7 , then the possible group four - velocities become @xmath8 = ( v^0 ; \,\mathbf{v } ) \propto ( 1 ; \,\mathbf{v}_g)$ ] . these objects can be explicitly interpreted as describing how quanta of electromagnetic energy ( hence photons ) propagate .
note that any multiple of a possible four - wavevector is again a possible four - wavevector , and similarly by construction any multiple of a possible group four - velocity is again a possible group four - velocity .
so the spacetime geometry we are seeking should provide us with a pair of pseudo - norms @xmath9 and @xmath10 specifying as null four - vectors exactly the possible group four - velocities and four - wavevectors respectively .
that is , the pseudo - norm should give us the relation @xmath11 or @xmath12 for the possible group four - velocities / four - wavevectors , and _ only _ for the possible group four - velocities / four - wavevectors .
let us now precisely state the conditions we should impose on a pseudo - norm in order for it to be physically interesting : a pseudo - norm @xmath9 on four - vectors is a map from @xmath13 , which is @xcite : * 1-homogeneous , * the square @xmath14 is smooth on @xmath15 , and * the gradient @xmath16 $ ] is a bijection from @xmath17 , the vector - space dual to @xmath18 .
this concept of pseudo - norm now specifies what we mean by pseudo - finsler geometry @xcite . to see that we can build up a full spacetime geometry out of this concept ,
we shall define a few other geometric concepts in complete analogy to the traditional positive definite finsler case .
the metric tensor is defined , as usual , by @xmath19}{\partial v^{i}\ ; \partial v^{j}}.\ ] ] we see that this definition reduces to standard pseudo - riemann case for @xmath20 where ( in this case ) @xmath21 does not depend on @xmath22 . by using euler s theorem , which states that for @xmath23-homogeneous functions @xmath24 we can easily define a mapping from vectors ( four - velocities ) to co - vectors ( four - wavevectors ) as @xmath25}{\partial v^{i } } = g_{i j}(v)\;v^{j}.\ ] ] then by using a legendre transform , we can easily define the co - finsler pseudo - norm @xmath10 as @xmath26 and co - finsler metric tensor as @xmath27}{\partial k^{i}(v ) \ ; \partial k^{j}(v)}.\ ] ] then it can be proven that @xmath28}{\partial k_{i } } = g^{ij}(k ) \ ; k_j,\ ] ] where @xmath29 and @xmath30 are inverse mappings : @xmath31 .
this shows that we have a meaningful generalization of pseudo - riemann geometry .
note that we can easily define a pseudo - finsler connection , and hence curvature , in this formalism simply by extremalizing the finsler distance ( @xmath32 ) : @xmath33
what are the specific physical constraints given by birefringent crystals on our geometry ? we consider it meaningful to impose the following constraints : * the pseudo - norm should , for vectors approaching the purely timelike @xmath34 vector ( in the fixed coordinates `` at rest '' with respect to the crystal ) , approach the pseudo - riemann flat - space pseudo - norm .
* the pseudo - norm should break lorentz invariance as in a birefringent crystal . * each `` constant proper time '' hypersurface , specified by @xmath35 , should be connected , and should contain the point @xmath34 .
let us discuss these constraints a little bit : the first is just requirement that we want to recover low - energy physics as we know it ( in a pseudo - riemann form ) .
the second is trivial : it is the specific contribution of using birefringent crystal as an analogue model ( unlike the first and the third condition , which can be considered as generic ) .
the third condition states that if we have a massive particle , it should not be able to have two distinct sets of four - velocities between which it is not able to undergo a smooth transition .
it also means that any massive particle should be able to move arbitrarily slowly in the rest frame of the crystal .
now focus on the classical theory of birefringent crystals ( see @xcite for a detailed exposition ) .
it is typically formulated in a preferred coordinate frame , which is the inertial frame of the crystal , with spatial coordinate axes aligned with the principal axes of the crystal .
in fact in the rest fame the crystal is described by a @xmath36 dielectric tensor @xmath37 , where the principal axes are defined as axes where the dielectric tensor is diagonalized .
the relation between the components of the possible four - wavevectors can be extracted from the fresnel equation for wave normals @xcite : @xmath38 this factorizes into @xmath39 where @xmath40 and @xmath41 as we see , in the general case this relation encodes a two - cone structure , one for each polarization , each taking in general a distinct phase velocity @xmath42 .
so we see that here the space - time causal structure is much more complicated than in special relativity . since the four - wavevector is in fact a covector ,
these relations give rise to a co - finsler structure rather then to a finsler structure .
hence they give us a generic candidate for the co - pseudo - finsler norm , which can be written as @xmath43 or more precisely @xmath44 , where @xmath45 is an otherwise arbitrary function fulfilling following constraints : * @xmath46}{\partial k^{i}}$ ] should be a bijection from @xmath47 ; * @xmath45 is a @xmath48 homogeneous map to @xmath49 ; * smooth on @xmath50 ; * inside both signal cones negative , outside both positive , between them nonzero ; * @xmath51 should approximate the flat - space pseudo - riemann norm ( squared ) in the given coordinates for @xmath52 close to @xmath34 and such vectors should lie on some hypersurface ( mass - shell ) given by : @xmath53 , which should be always connected .
these conditions follow trivially from the conditions we imposed on any physically meaningful finslerian geometry .
we should also consider the finsler pseudo - norm ; this should be extracted from the equations describing how energy propagates , which is the equation determining how photons propagate ( their group velocities ) .
this should be put into the form @xmath54 , so that it encodes the pseudo - finsler metric tensor as opposed to co - finsler metric tensor .
the information regarding energy propagation is given by the fresnel ray equation , which is : @xmath55 here @xmath56 using the four vector @xmath57 , the fresnel ray equation can be transformed into @xmath58 where @xmath59 and @xmath60 obviously the the ray equation also encodes a two - cone structure .
the generic finsler pseudo - norm is in this case expressed exactly in the same way as was the co - finsler pseudo - norm in the previous case , by the function @xmath61 , where @xmath62 is again an arbitrary function fulfilling exactly the same conditions as the @xmath45 function , we just have to exchange @xmath63 for @xmath18 . now the last step in putting constraints on geometric construction is to relate the finsler and co - finsler structures . the items which are as yet undetermined are the functions @xmath62 , and @xmath45 .
so they have to fulfill the last condition , which is the condition of forming the full united geometry .
this condition is given by the legendre transform relation : @xmath64}{\partial v^{i}}\right)\right]^2=n(v)^2.\ ] ] in our language , the function @xmath62 must be connected to @xmath45 by : @xmath65}{\partial v^{i}}\right ) \ ; f'\left(\frac{\partial [ m(v)f(v)]}{\partial v^{i}}\right)=m(v ) \ ; f(v ) .
\label{e : compatibility}\ ] ] thus we see that if we have somehow found an appropriate @xmath62 , then @xmath45 will be uniquely determined .
to find solutions of such an abstract ( although precisely formulated ) mathematical exercise is not an easy task . at this early stage
it is quite doubtful whether general solutions exist .
we shall look at particular cases , and try to check some intuitive solutions , since our original objective was to build up geometry in an intuitive way .
the simplest case of crystal optics is an isotropic medium @xmath66 .
the obvious interpretation in this case is as a pseudo - riemann metric , but with @xmath67 playing the role of the speed of light @xcite .
( strictly speaking , even this obvious solution does not fulfill our condition ( e ) , unless @xmath68 , which implies there is no crystal at all . ) in a uniaxial crystal ( @xmath69 ) , the @xmath70 and @xmath71 functions decompose into the product of two riemannian structures @xcite , but simple and intuitive ansatz for the @xmath72 and @xmath10 functions lead to serious difficulties . for instance a suggestion where either @xmath73 ( with the two relations being related by legendre transformation ) encounters problems with condition ( e ) , the putative finsler norms are nonreal between the two signal cones , and not smooth at the signal cones .
( even second derivatives do not exist there , hence the pseudo-(co)-finsler metric tensor is undefined there @xcite ) . in the case of biaxial crystal ( @xmath74 )
there are even bigger problems with this solution " even the individual @xmath75 , @xmath76 and @xmath77 , @xmath78 are not themselves pseudo - riemannian structures ; they are nonsmooth at the optical axes @xcite .
obviously , the previous problems with the `` unified '' @xmath79 or legendre - equivalent @xmath80 structures remain
. we could try to work out more sophisticated solutions to our `` compatibility problem '' ( [ e : compatibility ] ) , and at least prove existence .
we have found a partial solution for this problem , except for proving the bijection condition , which is quite tricky .
but the fact that this construction lacks any intuitive basis is quite disturbing . in these circumstances we have to accept the unwelcome conclusion
that although birefringent crystal have given us many ideas about the construction of pseudo - finsler geometries , they have not provided us with a clear and intuitive example of a `` unified '' geometry capable of handling both polarization modes simultaneously .
so the original aim we posed at the beginning of this article has not been accomplished . in conclusion :
our results are much less encouraging than we had originally hoped , and suggest that while pseudo - finsler spacetimes are certainly useful constructs , in birefringent situations it does not appear possible to naturally and intuitively construct a `` unified '' pseudo - finsler spacetime such that the pseudo - finsler metric is null on both `` signal cones '' , but has no other zeros or singularities it seems physically more appropriate to think of physics as taking place in a single topological manifold that carries distinct pseudo - finsler metrics , one for each polarization mode .
69 j. skakala and m. visser , `` the pseudo - finslerian spacetimes of birefringent optics , '' arxiv:0806.0950 [ gr - qc ] . m. visser , `` acoustic propagation in fluids : an unexpected example of lorentzian geometry , '' arxiv : gr - qc/9311028
. + m. visser , `` acoustic black holes : horizons , ergospheres , and hawking radiation , '' class .
grav . * 15 * ( 1998 ) 1767 [ arxiv : gr - qc/9712010 ] .
+ c. barcelo , s. liberati and m. visser , `` analogue gravity , '' living rev .
* 8 * ( 2005 ) 12 [ arxiv : gr - qc/0505065 ] .
r. gambini and j. pullin , `` nonstandard optics from quantum spacetime , '' phys .
d * 59 * ( 1999 ) 124021 [ arxiv : gr - qc/9809038 ] .
+ m. bojowald , h. a. morales - tecotl and h. sahlmann , `` on loop quantum gravity phenomenology and the issue of lorentz invariance , '' phys .
d * 71 * ( 2005 ) 084012 [ arxiv : gr - qc/0411101 ] .
n. e. mavromatos , `` lorentz invariance violation from string theory , '' arxiv:0708.2250 [ hep - th ] .
+ n. e. mavromatos , `` quantum gravity , cosmology , ( liouville ) strings and lorentz invariance , '' arxiv : hep - th/0210079 .
+ v. a. kostelecky and r. potting , phys .
b * 381 * , 89 ( 1996 ) [ arxiv : hep - th/9605088 ] .
+ s. i. vacaru , `` finsler and lagrange geometries in einstein and string gravity '' , arxiv:0801.4958 [ gr - qc ] . j. ambjorn , j. jurkiewicz and r. loll , `` emergence of a 4d world from causal quantum gravity , '' phys . rev .
* 93 * ( 2004 ) 131301 [ arxiv : hep - th/0404156 ] .
+ j. ambjorn , r. loll , y. watabiki , w. westra and s. zohren , `` a string field theory based on causal dynamical triangulations , '' jhep * 0805 * ( 2008 ) 032 [ arxiv:0802.0719 [ hep - th ] ] .
s. judes and m. visser , `` conservation laws in doubly special relativity , '' phys .
d * 68 * ( 2003 ) 045001 [ arxiv : gr - qc/0205067 ] .
+ s. liberati , s. sonego and m. visser , `` interpreting doubly special relativity as a modified theory of measurement , '' phys .
d * 71 * ( 2005 ) 045001 [ arxiv : gr - qc/0410113 ] .
a. g. cohen and s. l. glashow , `` very special relativity , '' phys .
* 97 * ( 2006 ) 021601 [ arxiv : hep - ph/0601236 ] .
+ g. w. gibbons , j. gomis and c. n. pope , `` general very special relativity is finsler geometry , '' phys .
d * 76 * ( 2007 ) 081701 [ arxiv:0707.2174 [ hep - th ] ]
. + a. p. kouretsis , m. stathakopoulos and p. c. stavrinos , `` the general very special relativity in finsler cosmology , '' arxiv:0810.3267 [ gr - qc ] .
f. girelli and e. r. livine , `` special relativity as a non commutative geometry : lessons for deformed special relativity , '' arxiv : gr - qc/0407098 .
+ a. connes , m. r. douglas and a. s. schwarz , `` noncommutative geometry and matrix theory : compactification on tori , '' jhep * 9802 * ( 1998 ) 003 [ arxiv : hep - th/9711162 ] .
g. f. r. ellis and j. p. r. uzan , `` c is the speed of light , is nt it ?
, '' am .
j. phys .
* 73 * ( 2005 ) 240 [ arxiv : gr - qc/0305099 ] .
+ g. f. r. ellis , `` note on varying speed of light cosmologies , '' gen .
* 39 * ( 2007 ) 511 [ arxiv : astro - ph/0703751 ] .
+ b. a. bassett , s. liberati , c. molina - paris and m. visser , `` geometrodynamics of variable speed of light cosmologies , '' phys .
d * 62 * ( 2000 ) 103518 [ arxiv : astro - ph/0001441 ] .
+ s. liberati , b. a. bassett , c. molina - paris and m. visser , `` chi variable - speed - of - light cosmologies , '' nucl . phys .
* 88 * ( 2000 ) 259 [ arxiv : astro - ph/0001481 ] .
t. jacobson , s. liberati and d. mattingly , `` lorentz violation at high energy : concepts , phenomena and astrophysical constraints , '' annals phys . * 321 * ( 2006 ) 150 [ arxiv : astro - ph/0505267 ] .
+ t. a. jacobson , s. liberati , d. mattingly and f. w. stecker , `` new limits on planck scale lorentz violation in qed , '' phys .
* 93 * ( 2004 ) 021101 [ arxiv : astro - ph/0309681 ] .
+ t. jacobson , s. liberati and d. mattingly , `` lorentz violation and crab synchrotron emission : a new constraint far beyond the planck scale , '' nature * 424 * ( 2003 ) 1019 [ arxiv : astro - ph/0212190 ]
. + t. jacobson , s. liberati and d. mattingly , `` threshold effects and planck scale lorentz violation : combined constraints from high energy astrophysics , '' phys . rev .
d * 67 * ( 2003 ) 124011 [ arxiv : hep - ph/0209264 ] .
+ t. jacobson , s. liberati and d. mattingly , `` tev astrophysics constraints on planck scale lorentz violation , '' phys
. rev .
d * 66 * ( 2002 ) 081302 [ arxiv : hep - ph/0112207 ] .
m. visser , c. barcelo and s. liberati , `` acoustics in bose - einstein condensates as an example of broken lorentz symmetry , '' arxiv : hep - th/0109033 .
+ s. weinfurtner , s. liberati and m. visser , `` modelling planck - scale lorentz violation via analogue models , '' j. phys .
ser . * 33 * ( 2006 ) 373 [ arxiv : gr - qc/0512127 ] .
+ s. liberati , m. visser and s. weinfurtner , `` naturalness in emergent spacetime , '' phys .
* 96 * ( 2006 ) 151301 [ arxiv : gr - qc/0512139 ] .
+ s. weinfurtner , s. liberati and m. visser , `` analogue spacetime based on 2-component bose - einstein condensates , '' lect .
notes phys .
* 718 * ( 2007 ) 115 [ arxiv : gr - qc/0605121 ] .
+ s. weinfurtner , s. liberati and m. visser , `` analogue model for quantum gravity phenomenology , '' j. phys .
a * 39 * ( 2006 ) 6807 [ arxiv : gr - qc/0511105 ]
. + p. jain , s. weinfurtner , m. visser and c. w. gardiner , `` analogue model of a frw universe in bose - einstein condensates : application of the classical field method , '' phys .
rev . * a 76 * ( 2007 ) 033616 [ arxiv:0705.2077 [ cond-mat.other ] ] .
+ s. weinfurtner , p. jain , m. visser and c. w. gardiner , `` cosmological particle production in emergent rainbow spacetimes , '' arxiv:0801.2673 [ gr - qc ] .
+ m. visser and s. weinfurtner , `` analogue spacetimes : toy models for quantum gravity , '' pos * qg - ph * ( 2008 ) 042 [ arxiv:0712.0427 [ gr - qc ] ] .
+ s. weinfurtner , m. visser , p. jain and c. w. gardiner , `` on the phenomenon of emergent spacetimes : an instruction guide for experimental cosmology , '' pos * qg - ph * ( 2008 ) 044 [ arxiv:0804.1346 [ gr - qc ] ] .
+ m. visser , `` emergent rainbow spacetimes : two pedagogical examples , '' arxiv:0712.0810 [ gr - qc ] . c. barcel , s. liberati and m. visser , `` analog gravity from field theory normal modes ? '' , class .
* 18 * ( 2001 ) 3595 [ arxiv : gr - qc/0104001 ] .
+ c. barcel , s. liberati and m. visser , `` refringence , field theory , and normal modes '' , class .
* 19 * ( 2002 ) 2961 [ arxiv : gr - qc/0111059 ] .
+ m. visser , c. barcel and s. liberati , `` bi - refringence versus bi - metricity '' , arxiv : gr - qc/0204017 .
p. finsler , `` ber kurven und flachen in allgemeinen raumen '' , dissertation , gottingen , 1918 , published by verlag birkhauser basel , 1951 . + h. buseman , `` the geometry of finsler space '' , bull .
soc . * 56 * , number 1 , part 1 ( 1950 ) , 5 - 16 .
+ h. rund , _ the differential geometry of finsler spaces _ , ( springer , berlin , 1959 ) .
+ d. bao , s.s .
chern , and z. shen , _ an introduction to riemann finsler geometry _ , ( springer , new york , 2000 ) .
+ d. bao , s.s .
chern , and z. shen ( editors ) , _ finsler geometry _ , proceedings of the joint summer research conference on finsler geometry ,
july 1995 , seattle , washington , ( american mathematical society , providence , rhode island , 1996 ) .
chern , `` finsler geometry is just riemannian geometry without the quadratic restriction '' , not .
* 43 * ( 1996 ) 959963 .
+ z. shen , _ differential geometry of spray and finsler spaces _ , ( kluwer , dordrecht , 2001 ) .
+ z. shen , _ lectures on finsler geometry _ , ( world scientific , singapore , 2001 ) .
antonelli and b.c .
lackey ( editors ) , _ the theory of finslerian laplacians and applications _ , ( kluwer , dordrecht , 1998 ) .
+ a. bejancu , _ finsler geometry and applications _ , ( ellis horwood , chichester , england , 1990 ) .
antonelli , r.s .
ingarden , m. matsumoto , _ the theory of sprays and finsler spaces with applications in physics and biology _ , ( springer [ kluwer ] , berlin , 1993 ) .
s. liberati , `` quantum gravity phenomenology via lorentz violations '' , pos * p2gc * ( 2007 ) 018 [ arxiv:0706.0142 [ gr - qc ] ] .
+ f. girelli , s. liberati and l. sindoni , `` phenomenology of quantum gravity and finsler geometry '' , phys .
d * 75 * ( 2007 ) 064015 [ arxiv : gr - qc/0611024 ] .
+ l. sindoni , `` the higgs mechanism in finsler spacetimes '' , arxiv:0712.3518 [ gr - qc ] .
+ v. perlick , `` fermat principle in finsler spacetimes '' , gen .
* 38 * ( 2006 ) 365 [ arxiv : gr - qc/0508029 ] .
+ s. siparov , introduction to the problem of anisotropy in geometrodynamics , arxiv 0809.1817 [ gr - qc ] .
r. punzi , m. n. r. wohlfarth and f. p. schuller , `` propagation of light and fluids in string backgrounds , '' arxiv:0711.3771 [ hep - th ] .
+ f. p.
schuller and m. n. r. wohlfarth , `` radiation - dominated area metric cosmology , '' jcap * 0712 * ( 2007 ) 013 [ arxiv:0705.4656 [ hep - th ] ] .
+ r. punzi , f. p. schuller and m. n. r. wohlfarth , `` area metric gravity and accelerating cosmology , '' jhep * 0702 * ( 2007 ) 030 [ arxiv : hep - th/0612141 ] .
+ f. p. schuller and m. n. r. wohlfarth , `` geometry of manifolds with area metric , '' nucl .
b * 747 * ( 2006 ) 398 [ arxiv : hep - th/0508170 ] .
m. born and e. wolf , _ principles of optics _ , 7th edition , ( cambridge university press , england , 1999 ) . | based on the analogue spacetime programme , and many other ideas currently mooted in `` quantum gravity '' , there is considerable ongoing speculation that the usual pseudo - riemannian ( lorentzian ) manifolds of general relativity might eventually be modified at short distances .
two specific modifications that are often advocated are the adoption of finsler geometries ( or more specifically , pseudo - finsler spacetimes ) and the possibility of birefringence ( or more generally , multi - refringence ) .
we have investigated the possibility of whether it is possible to usefully and cleanly deal with these two possibilities _ simultaneously_. that is , given two ( or more ) `` signal cones '' : is it possible to naturally and intuitively construct a `` unified '' pseudo - finsler spacetime such that the pseudo - finsler metric is null on these `` signal cones '' , _ but has no other zeros or singularities _ ?
our results are much less encouraging than we had originally hoped , and suggest that while pseudo - finsler spacetimes are certainly useful constructs , it is physically more appropriate to think of physics as taking place in a single topological manifold that carries several distinct pseudo - finsler metrics , one for each polarization mode
. [ section ] [ thm]corollary [ thm]lemma [ thm]proposition [ section ] [ section ] |
_ continuous time tasep . _
the totally asymmetric simple exclusion process ( tasep ) is the simplest non - reversible interacting stochastic particle system . in tasep , particles are on the lattice of integers , @xmath0 , with at most one particle at each site ( exclusion principle ) .
the dynamics is defined as follows .
particles jump to the neighboring right site with rate @xmath5 provided that the site is empty .
jumps are independent of each other and take place after an exponential waiting time with mean @xmath5 , which is counted from the time instant when the right neighbor site is empty .
it is known that the only translation invariant stationary measures are bernoulli product measures with a given density @xmath6 $ ] ( see @xcite ) . in the sequel we fix a @xmath7 to avoid the trivial cases @xmath8 ( no particles ) and @xmath9 ( all sites occupied ) .
one quantity of interest is the fluctuations of the currents of particles during a large time @xmath10 .
consider the so - called _ characteristic line _ given by @xmath11 .
then , the current fluctuation seen from the characteristic are @xmath12 ( in agreement with the scaling for the two - point function established in @xcite ) , and the limit distribution function has been determined ( conjectured in @xcite and proved in @xcite , see also @xcite ) .
the limiting distribution was first discovered in the context of a directed last passage percolation model @xcite . on the other hand ,
if one looks at the current along a line different from the characteristic line , say @xmath13 with @xmath14 , then for large time @xmath10 one just sees the gaussian fluctuations from the initial condition on a @xmath15 scale , since the dynamics generates fluctuations only @xmath12 , thus irrelevant @xcite .
a non - trivial interplay between the dynamically generated fluctuations and the one in the initial condition occurs in a region of order @xmath4 around the characteristic line .
one of the main results of this paper is the determination of multi - point limits of the current fluctuations around the characteristics @xmath16 ( see theorem [ theoremtasep ] and theorem [ theoremtasepb ] ) .
there has been a great deal of work pertaining to the limiting fluctuation of tasep over the last ten years since the well - known work of johansson @xcite .
see for example , the review paper @xcite for the limit processes which arise from deterministic initial conditions .
a recent paper @xcite , building on the earlier work @xcite , is concerned on the situation where random and deterministic initial condition are both present , but not stationary . _
directed percolation model .
_ it is well - known that the currents of tasep ( of arbitrary initial condition ) can be expressible in terms of the last passage time of an associated directed last passage percolation model ( see for example , @xcite ) .
we first discuss the asymptotic result of the following directed percolation model associated to stationary tasep @xcite : see the paragraphs preceding theorem [ theoremtasep ] below for the exact relation to stationary tasep .
let @xmath17 , @xmath18 , @xmath19 , be independent random variables with the following distributions @xmath20 here the notation @xmath21 means that @xmath22 is a random variable exponentially distributed with expectation @xmath23 . an _ up - right _
path @xmath24 from @xmath25 to @xmath26 is a sequence of points , starting from the origin , @xmath27 , ending at @xmath28 , @xmath29 , and satisfying @xmath30 .
denote by @xmath31 .
then , the last passage time is defined by @xmath32 we are interested in the limit distribution of the properly rescaled last passage time of @xmath33 , [ yn])$ ] in the @xmath34 limit . when there are no ` borders ' i.e. @xmath35 when @xmath36 or @xmath37 , the limiting distribution of @xmath38 was first obtained in @xcite .
the case of ` single - border ' when @xmath39 but @xmath40 was studied in @xcite .
the above ` double - border ' case was considered in @xcite .
( the poisson and geometric variations were studied earlier in @xcite . )
one can also consider a more general model when @xmath41 for @xmath42 and @xmath43 for @xmath44 , where @xmath45 .
such model was considered in @xcite and a conjecture on the limiting distribution left in that paper was recently confirmed in @xcite . for the poisson and geometric variations ,
this generalization was considered in @xcite earlier when @xmath46 .
an interesting characteristic behavior in such bordered models is the transition phenomenon .
one can imagine that the last passage time comes from the competition between the contributions from the ` bulk ' @xmath47 , and the edges @xmath36 or @xmath37 .
see for example , section 6 of @xcite for an illustration of such heuristic ideas . for the model , a crucial role
is played by the critical direction ( which corresponds to the characteristic line of tasep ) , @xmath48 it is easy to check that along a direction other than the critical direction , the fluctuations of @xmath33 , [ yn])$ ] is given by gaussian distribution on the @xmath49 scale , see appendix [ appgaussian ] .
this situation corresponds to the gaussian fluctuations along a non - characteristic line in a stationary tasep @xcite mentioned earlier . along the critical direction and also in a @xmath50 neighborhood of the critical direction ,
the fluctuations of @xmath33 , [ yn])$ ] are on the @xmath51 scale .
precisely , the following , among other things , is proven in @xcite .
let us set @xmath52 then set the parameter @xmath53 , where @xmath54 is considered to be large .
consider the scaling @xmath55 the parameter @xmath56 measures the displacement of the focus with respect to the critical line ( on a @xmath57 scale ) .
in particular , for @xmath58 one looks exactly along the critical direction .
@xmath59 is the macroscopic value of the last passage time for large @xmath54 , while the parameter @xmath60 in @xmath61 measures the amount of the fluctuations ( on a @xmath62 scale ) of the last passage time .
then for any fixed @xmath63 , @xmath64 for an explicit distribution function @xmath65 , which satisfies @xmath66 .
an analogous result was first obtained in @xcite for a poissonized version of the problem in which @xmath65 was obtained in terms of a solution to the painlev ii equation ( see ( 3.22 ) of @xcite : @xmath67 ) . based on this result , prhofer and spohn conjectured in @xcite that holds .
this conjecture was proven in @xcite whose analysis was somewhat different from @xcite .
this results in a different formula of @xmath65 , expressed in terms of the airy function ( see ( 1.20 ) in @xcite ) .
it can be checked that these two formulas do agree .
one of the main objects of study in this paper is the limit of multi - point distribution @xmath68 , @xmath69 .
the limit of the process for the no - border case was first obtained in @xcite following the earlier work of @xcite on the poissonized version of the model .
the limit of the process for the double - border case was considered in @xcite for the model when @xmath70 and @xmath41 for @xmath42 , @xmath43 for @xmath44 when @xmath71 and @xmath72 .
note that when @xmath73 , the random variable @xmath74 becomes singular in this model : the restriction @xmath75 is significant in connection to stationary tasep .
the geometric counterpart of the model was studied earlier by imamura and sasamoto @xcite .
denoting by @xmath76 a random variable with the probability mass function @xmath77 , @xmath78 , the authors of @xcite considered the model where @xmath75 , @xmath79 , @xmath80 and @xmath81 for @xmath82 . since the exponential model can be obtained as a limiting case of @xmath83 , the analysis of @xcite can in principle be used to yield the corresponding result for the model .
nevertheless , this paper differs from @xcite on the following aspects .
( a ) the authors of @xcite obtained explicit limiting distribution functions for the case that the situation @xmath72 ( and @xmath84 ) in .
however , they left the case corresponding to @xmath85 as the limit of the @xmath72 case and did not compute the limiting distribution explicitly ( see remarks after theorem 5.1 in @xcite ) .
this critical case is the most interesting ( and the most difficult ) case for our situation and we give an explicit formula in theorem [ mainthm ] below .
( b ) the justification of the limits of fredholm determinant and other quantities appearing in the analysis requires proper conjugations of corresponding operators in order to make sense of the fredholm determinant and trace class limit .
these issues were not discussed in @xcite ( the main issue was to determine the possible limit regimes and not specifically the @xmath85 case ) .
( c ) in addition to the multi - point distribution on the line @xmath86 considered in @xcite , we also obtain limit of the process result for points @xmath87 not necessarily on the same line ( see theorem [ theoremdppext ] ) .
( d ) the limit of the process for points at more general positions than on a line mentioned in ( c ) is used to prove the limit of the process ( in the sense of finite distribution ) of stationary tasep ( see theorem [ theoremtasep ] and theorem [ theoremtasepb ] below ) .
we first state the result extending to the joint distributions at points on the line @xmath88 at and near the critical direction .
we first need some definitions .
[ defin1 ] fix @xmath89 . for real numbers
@xmath90 and @xmath91 , set @xmath92}\frac{e^{-\frac23\tau_i^3-\tau_i x}}{\sqrt{4\pi(\tau_i-\tau_1)}}\int_{-\infty}^{s_1-x}dy\ , e^{-\frac{y^2}{4(\tau_i-\tau_1 ) } } -\int_{0}^\infty dy\ , \operatorname*{ai}(y+x+\tau_i^2)e^{\tau_i y}. \end{aligned}\ ] ] for @xmath93 , where @xmath94 denotes the airy function .
the first result is the following .
[ mainthm ] fix @xmath89 . for real numbers
@xmath90 and @xmath91 , with the scaling given in , @xmath95 here @xmath96 is equipped with the standard measure @xmath97 where @xmath98 is the counting measure on @xmath99 .
@xmath100 denotes the projection operator @xmath101}$ ] , and @xmath102 is the so - called extended airy kernel @xcite with shifted entries defined by the kernel @xmath103_{i , j}(x , y)\\ & = \begin{cases}\displaystyle { \int_0^\infty d\lambda \operatorname*{ai}(x+\lambda+\tau_i^2)\operatorname*{ai}(y+\lambda+\tau_j^2 ) e^{-\lambda(\tau_j-\tau_i)}},&\textrm { if } \tau_i\leq \tau_j,\\ \displaystyle { -\int_{-\infty}^0 d\lambda \operatorname*{ai}(x+\lambda+\tau_i^2)\operatorname*{ai}(y+\lambda+\tau_j^2 ) e^{-\lambda(\tau_j-\tau_i)}},&\textrm { if } \tau_i > \tau_j .
\end{cases } \end{aligned}\ ] ] the function @xmath104 is defined by @xmath105 where @xmath106 and @xmath107 , @xmath108 .
finally the functions @xmath109 , @xmath110 , and @xmath111 are defined in definition [ defin1 ] .
observe that @xmath112 .
when @xmath113 , agrees with the limiting function in ( 1.20 ) of @xcite , as one may expect .
[ rem : inv ] the fact that @xmath114 is invertible follows from the fact that @xmath115 is trace class ( see @xcite ) and that @xmath116 for all given @xmath117 .
see lemma [ leminvertibility ] in appendix [ appinvertibility ] .
the shift in the integrand by @xmath118 is due to the fact that the last passage time is ( macroscopically ) a linear function along the line @xmath119 , in contrast with the non - border case ( i.e. , @xmath120 ) where the last passage time has a non - zero curvature .
therefore , when @xmath121 , the contribution from @xmath122 will very close to one and the main contribution comes from @xmath104 .
the second theorem is a generalization of theorem [ mainthm ] to the case when the points @xmath87 are not necessarily on the same line @xmath123 , @xmath124 .
we show that the fluctuation is unchanged even if some of the points are away from the line to the order smaller than @xmath125 .
this is due to the slow de - correlation phenomena obtained in @xcite : along the critical direction the fluctuations decorrelate to order @xmath126 over a time scale @xmath125 ( instead of @xmath127 ) .
more precisely , if we compare the last passage time @xmath128 at two points @xmath28 and @xmath129 with @xmath130 , their fluctuation will be @xmath131 ( see lemma [ lemslowdecorrelation ] below for details ) .
consider a @xmath132 and any fixed number @xmath133 and consider the scaling ( a generalization of ( [ eq6 ] ) ) given by @xmath134 see figure [ figuredppext ] for an illustration .
[ cb]@xmath135 [ cb]@xmath136 [ c]@xmath137 [ c]@xmath138 [ l]critical line for some @xmath139 away from the line @xmath138 .
then , the fluctuations of the passage time at the locations of the black dots are , on the @xmath62 scale , the same as the one of their projection along the critical direction to the line @xmath138 , the white dots.,title="fig:",height=188 ] [ theoremdppext ] fix @xmath140 and @xmath132 . for real numbers @xmath90 , @xmath141 , and @xmath91 , with the scaling given in , @xmath142
this generalization of theorem [ mainthm ] is proven in section [ sectproofmainthm ] .
_ stationary tasep and directed percolation .
_ we now discuss the result in terms of stationary tasep .
the mapping between tasep and last passage percolation model is as follows .
we assign label @xmath143 to the particle sitting at the smallest positive integer site initially . for the rest we use the right - to - left ordering so that @xmath144
. then @xmath145 for all @xmath146 , since the tasep preserves the ordering of the particles . for @xmath147 such that @xmath148 , let @xmath149 be the waiting time of particle with label @xmath150 to jump from site @xmath151 to site @xmath152 ( the waiting time
is counted from the instant where particle can jump , i.e. , particle is at @xmath151 and site @xmath152 is empty ) .
the @xmath149 are iid @xmath153 random variables .
let @xmath154 be the last passage time to @xmath26 along a directed path in the domain @xmath155 , starting any point in the domain .
this is a curve - to - point optimization problem .
an example of the non - trivial part of the boundary of this domain is illustrated in figure [ figuretasep ] below as the curve at @xmath156 .
then a straightforward generalization of the step - initial case of @xcite shows that @xmath157 when the initial condition is random , the domain of directed percolation is also random . note that ( see figure [ figuretasep ] ) , the part of the domain @xmath158 in the first quadrant is the rectangle @xmath159 , but the parts in the second quadrant @xmath160 and the forth quadrant @xmath161 are of random shape .
observe that @xmath158 does not intersect with the third quadrant @xmath162 .
define @xmath163 to be the right - most empty site in @xmath164 in the initial particles configuration and @xmath165 to be the position of the left - most particle in @xmath166 .
first let us focus on the case @xmath167 , @xmath168 ( as in figure [ figuretasep ] ) . then , since the initial condition is stationary bernoulli , an interpretation of burke s theorem @xcite , see also @xcite , shows that the @xmath169 is distributed as @xmath170 where @xmath171 s are iid @xmath172 distributed .
similarly , by considering holes instead of particles one finds that @xmath173 is distributed as @xmath174 where @xmath175 s are iid @xmath176 random variables .
hence @xmath154 has the same distribution as @xmath38 defined from .
this argument was sketched in section 2 of @xcite .
consequently , we have for @xmath177 , @xmath178 .
@xmath179 for stationary tasep when initially the position @xmath143 is empty and the position @xmath5 is occupied . for generic @xmath180 and @xmath181 ,
@xmath154 is distributed as @xmath38 where now we assume in that @xmath182 where @xmath183 and @xmath184 .
but as shown in proposition 2.2 in @xcite , this change does not affect the asymptotics .
geometrically , the tasep and directed percolation can be thought of as two different cuts of the three - dimensional object @xmath185 ( a ) the directed percolation problem we analyze is the cut at @xmath186 , + ( b ) the particles configuration of the tasep at time @xmath10 is the cut at @xmath187 .
+ we have shown in theorem [ theoremdppext ] that the limit process does not depend on the cut chosen for the analysis as long as we avoid the cut along the characteristic , @xmath188 .
thus , to get the fluctuations around the characteristic line it is enough to project on @xmath189 ( see figure [ figuretasep ] for an illustration ) , for which the limit theorem was proven in theorem [ mainthm ] .
[ c]@xmath150 [ l]@xmath190 [ cb]@xmath136 [ c]@xmath137 [ bc]@xmath156 [ bl]@xmath10 [ c]@xmath138 [ l]characteristics and some later time @xmath10 . the position of particle with label @xmath191 is its projection on the @xmath192-axis .
interpolating between particles as in this example , one gets a line configuration , which is interpreted as height function @xmath193 above position @xmath150.,title="fig:",height=226 ] the following two theorems are proven in section [ sectproofmainthm ] .
define the functions @xmath194 [ theoremtasep ] fix @xmath140 . for real numbers
@xmath90 and @xmath91 , @xmath195 an equivalent but geometrically slightly different way of representing the tasep is via a height function ( see e.g. @xcite ) .
let us define the occupation variable , @xmath196 if there is a particle at site @xmath197 at time @xmath10 and @xmath198 otherwise .
then , define the height function @xmath199 where @xmath200 is the number of particles which jumped from site @xmath143 to site @xmath5 during the time - span @xmath201 $ ] .
then the link between the height functions and the locations of particles is @xmath202 let us consider the scaling @xmath203 [ theoremtasepb ] fix @xmath140
. for real numbers @xmath90 and @xmath91 .
then , @xmath204 for simplicity , in theorems [ theoremtasep ] and [ theoremtasepb ] we stated the result only for fixed time .
however , the statements can be extended to different times in a similar manner of the extension from theorem [ mainthm ] to theorem [ theoremdppext ] . _ queues in tandem .
_ there is a direct relation between queues in tandem and tasep .
suppose that there are infinitely many servers , and we assume that the service time of a customer at each server is independent and distributed as @xmath153 .
once a customer is served at the server @xmath197 , then the customer joins at the @xmath205th queue , and so on . in other words ,
the departure process from the @xmath197th queue is the arrival process at the @xmath205th queue .
suppose that the system is in equilibrium with parameter @xmath1 : the arrival process at each queue is independent poisson process of rate @xmath1 .
then the departure process at each queue is also independent poisson process of rate @xmath1 due to burke s theorem ( see for example , @xcite ; see also @xcite ) .
this also implies that at each time , the number of customers in each queue is distributed as @xmath206 .
( here @xmath207 means that @xmath208 , @xmath78 . )
now consider a fixed time @xmath156 and arbitrary select one customer and assign label @xmath143 to that customer . for convenience
, we call the queue in which that customer is in at time @xmath143 as the @xmath143th queue .
we assign labels to the other customers so that the labels decreases for the customers ahead in the queues ( see figure [ figurequeues ] ) .
[ b]@xmath209 [ b]@xmath210 [ b]@xmath211 [ b]@xmath212 [ b]@xmath213 [ b]@xmath143 [ b]@xmath5 [ b]@xmath214 [ b]@xmath215 [ b]@xmath216 [ cb]c@xmath217 [ cb]c@xmath218 [ cb]c@xmath219 [ cb]c@xmath220 [ cb]c@xmath221 [ cb]c@xmath222 [ cb]server @xmath212 [ cb]server @xmath213 [ cb]server @xmath143 [ cb]server @xmath5 [ cb]server @xmath214 let @xmath223 denote the label of the queue in which the @xmath150th customer is in at time @xmath10 .
the mapping from the queueing model to the tasep is obtained by setting @xmath224 ( see figure [ figurequeues ] ) .
the equilibrium condition implies that the initial condition for the corresponding tasep is stationary .
hence if we define @xmath225 be the time the @xmath150th customer exits the queue @xmath197 , then we find that @xmath226 hence using and theorem [ theoremdppext ] , we immediately obtain the following result .
[ theoremqueue ] fix @xmath140 and @xmath132 . for real numbers @xmath90 , @xmath141 , and @xmath91 , with the scaling given in , @xmath227 to prove theorem [ mainthm ] , which is the basis for theorems [ theoremtasep ] and [ theoremtasepb ]
, we first consider a slightly different directed percolation model which is known to be determinantal .
this model is then related to ( [ eqdp ] ) by a shift argument ( section [ sectshift ] ) followed by an analytic continuation ( section [ sectanalytic ] ) , resulting in an explicit formula of the joint distribution of the last passage times of a more general version of when @xmath41 for @xmath42 and @xmath43 for @xmath44 , for @xmath228 ( see theorem [ thm : pab ] ) .
the formula for the model is obtained by setting @xmath229 .
these arguments are a generalization of the arguments given in @xcite and @xcite for one - point distribution . in section [ sectasympt ]
we carry out the asymptotic analysis of the formula obtained in theorem [ thm : pab ] .
the proofs of theorems [ mainthm ] , [ theoremtasep ] and [ theoremtasepb ] are given in section [ sectproofmainthm ] .
some technical computations are given in appendices : various expressions of integrals involving airy functions are given in appendix [ sectairyfunctions ] , the invertibility of an operator appearing in theorem [ mainthm ] is discussed in appendix [ appinvertibility ] , certain operator are shown to be trace - class in appendix [ secttraceclass ] , and finally we explain the gaussian fluctuation along non - critical directions in appendix [ appgaussian ] .
as mentioned above , theorem [ thm : pab ] below contains a formula for the model with @xmath41 for @xmath42 and @xmath43 for @xmath44 , for @xmath228 , which is more general than .
this would correspond to other random initial data which possibly allows a shock .
essentially all the ingredients for the asymptotic analysis for this more general case are in this paper , and one can obtain the limit laws for such models .
however , for the sake of simplicity , we do not pursue this direction here .
the work of jinho baik was supported in part by nsf grant dms075709 .
patrik l. ferrari started the work while being at wias - berlin .
finally , we would like to thank the referee for reading the manuscript carefully and suggesting many small changes which helped improving the presentation .
we first consider a slightly different directed percolation model .
let @xmath230 where the parameters @xmath231 and @xmath232 satisfy @xmath233 denote by @xmath234 to be the last passage time from @xmath25 to @xmath28 for this modified model .
this model is well - studied and has nice mathematical structure . in particular , the correlation functions and joint distribution functions on @xmath138
are determinantal , which is well - suited for an asymptotic analysis .
note that the original model given in corresponds to the case when @xmath85 and @xmath235 .
we will show in sections [ sectshift ] and [ sectanalytic ] how to obtain a joint distribution formula for the original model from this modified model .
let @xmath236 be the last passage time for the model with @xmath237 replaced by @xmath143 .
we proceed as follows : + * ( 1 ) shift argument * : for @xmath238 with @xmath72 , we relate the distribution of @xmath239 with the one of @xmath240 . + * ( 2 ) analytic continuation * : we determine an expression for @xmath239 which can be analytically continued in all @xmath238
. + * ( 3 ) choice of parameter * : finally we set @xmath241 and @xmath242 .
+ the step ( 1 ) is done in section [ sectshift ] and step ( 2 ) is presented in section [ sectanalytic ] .
[ propshift ] let @xmath238 with @xmath72 .
let @xmath243 be the a set distinct of points in @xmath244 and define @xmath245 then , @xmath246 proposition [ propshift ] let us only consider @xmath247 : the proof for general @xmath248 is a straightforward generalization and we leave it to the reader .
set @xmath249 .
then @xmath250 consider the laplace transform @xmath251 since @xmath252 when either @xmath253 or @xmath254 , we can restrict the integral from @xmath255 to @xmath256 .
then the above equals @xmath257 multiplying by @xmath258 and integrating by parts leads to @xmath259 since this holds for all @xmath260 , by inverting the laplace transform , we obtain @xmath261 as mentioned earlier , the probability @xmath262 has an explicit determinantal expression .
in fact , the joint distribution function at points on @xmath138 for the directed percolation model with , , has a determinantal structure ( it is a limit of geometric random variables studied in @xcite or of the schur measure @xcite ; see @xcite for a direct approach to exponential random variables ) .
specifically , this follows , for example , from theorem 3.14 and ( 1.3 ) of @xcite after substituting @xmath263 and @xmath264 in ( 1.3 ) and taking the limit @xmath265 .
the kernel in this limit is explicitly derived in theorem 3 of @xcite . for good survey papers on the topic ,
see @xcite . by specializing to the case with weights as in
, we have the following result .
consider a set of distinct points @xmath243 on the line , which can be ordered and parametrized by @xmath266 $ ] , @xmath267 set @xmath268 and , the notation @xmath269 means that the integration path goes anticlockwise around the points in @xmath270 but does not include any other poles . ]
@xmath271 define the kernels @xmath272 where @xmath273}}{2\pi{{\rm i}}}\int_{{{\rm i}}{\mathbb{r } } } d\widetilde z\ , e^{\widetilde z ( y - x)}\frac{\phi_i(\widetilde z)}{\phi_j(\widetilde z)},\\ \widetilde k_{i , j}(x , y ) & = \frac{-1}{(2\pi{{\rm i}})^2}\oint_{\gamma_{-1/2}}dz \oint_{\gamma_{1/2}}dw\ , e^{zy - wx}\frac{\phi_i(w)}{\phi_j(z)}\frac{1}{w - z}. \end{aligned}\ ] ] then , @xmath274 where @xmath275}$ ] and the operator @xmath276 is defined by the kernel @xmath277 in , we have used a slight abuse of notation as @xmath278 is not a trace - class operator : one can indeed observe that for @xmath279 , @xmath280 as @xmath281 . nevertheless with a suitable multiplication operator @xmath282 ,
the conjugate operator @xmath283 becomes trace - class so that the fredholm determinant is well - defined , and the identity becomes valid analytically .
more concretely , if we take @xmath282 as in below , then it is shown in appendix [ secttraceclass ] below that @xmath283 is trace - class operator for @xmath284 .
thus , after conjugation , the fredholm determinant is well - defined , and the identity becomes valid analytically , as proved in the following .
[ propstartingkernel ] let @xmath285 .
fix constants @xmath286 satisfying @xmath287 define the conjugated operator @xmath288 by the kernel @xmath289 then , @xmath290 is a trace - class operator on @xmath291 and @xmath292 proposition [ propstartingkernel ] and give the standard extension from one - point kernel to the multi - point ( or extended ) kernel .
all the ingredients are included in appendix b and section 3 of @xcite , and also in @xcite .
the extended kernel can be found in the more recent work @xcite . there
the setting is more general allowing several lines / columns to have waiting times different from @xmath5 .
observe @xmath293 is independent of @xmath231 and @xmath232 , and the only dependence on @xmath231 and @xmath232 in @xmath276 is through the rank - one term @xmath294 . using this , we can find an expression of @xmath262 in which the condition can be relaxed to .
this relaxation is important when we take the limit @xmath295 in the next section .
[ propmodifiedkernel ] let @xmath285 .
fix constants @xmath286 satisfying @xmath296 define the conjugated operator @xmath297 by the kernel @xmath298 then @xmath299 where the notation @xmath300 denotes the real inner product in @xmath301 . one can check that the expression can be analytically extended to @xmath302 with @xmath72 . nevertheless since we will discuss the issue of extending the domain analyticity of a formula of @xmath303 to @xmath302 ( with no restriction of @xmath72 ) in the next section
, we do not discuss the detail here .
proposition [ propmodifiedkernel ] from , @xmath304 where @xmath305 and @xmath306 are multiplication operators by the functions @xmath307 , @xmath308 .
the functions @xmath309 and @xmath310 are @xmath311 : see the end of the proof of proposition [ proptraceclass ] . for the convenience of notations , set @xmath312=p_u k^{\rm conj } p_u$ ] , @xmath313=p_u \overline k^{\rm conj } p_u$ ] , @xmath314=p_uf^{\rm conj}$ ] and @xmath315=g^{\rm conj}p_u$ ] .
thus @xmath316\right ) & = \det\left({\mathbbm{1}}-(a+b)({\mathbbm{1}}-[\overline k])^{-1}[f]\otimes [ g]\right)\cdot \det\left({\mathbbm{1}}-[\overline k]\right),\\ & = \left(1-(a+b)\langle({\mathbbm{1}}-[\overline k])^{-1}[f],[g]\,\rangle\right ) \cdot \det\left({\mathbbm{1}}-[\overline k]\right ) .
\end{aligned}\ ] ] the above step holds assuming that @xmath317={\mathbbm{1}}-p_u\overline k^{\rm conj } p_u$ ] is invertible .
the invertibility can be verified as follows .
consider the modification of the directed percolation model where @xmath318 if @xmath197 and/or @xmath150 are equal to @xmath143 , and @xmath319 for @xmath82 ( i.e. , model without sources ) .
denote by @xmath320 the new measure .
this is the model with @xmath321 in , and hence @xmath322 for any given @xmath323 .
it is easy to obtain a lower bound @xmath324 , with @xmath325 . indeed
, it is enough to take the configurations with @xmath326 for @xmath327 such that @xmath328 .
this implies that @xmath329 , and hence @xmath330 is invertible .
finally , since @xmath331 is a bounded operator in @xmath311 , and , @xmath332)^{-1}[f],[g]\,\rangle$ ] equals @xmath333 by conjugating back @xmath282 .
now the condition for the @xmath334 s can be relaxed to the condition since @xmath297 is trace - class under this assumption : see proposition [ proptraceclass ] .
we now find a formula extending to the case when @xmath228 without the restriction that @xmath72 .
for this purpose , we show that both sides of are analytic in the parameter @xmath335 .
analyticity of the left - hand - side of i.e. , of @xmath336 is a straightforward generalization of proposition 5.1 in @xcite .
this section is devoted to finding the analytic continuation of the right - hand - side of . by using ( [ eq25 ] )
, the right - hand - side of becomes @xmath337 the last determinant is independent of @xmath338 .
it is enough show that @xmath339 is analytically continued in @xmath238 .
note that by changing the contour , @xmath340 where @xmath341 [ propanalyticcont ] for @xmath284 , @xmath342 where @xmath343 and @xmath344 with @xmath345 } \int_{-\infty}^{u_1 } dy\ , v_{i,1}(x , y ) f_1^{(a)}(y).\ ] ] the term @xmath346 is analytic in @xmath302 , and @xmath347 is convergent and is analytic in @xmath302 .
hence the right - hand - side of is an analytic continuation of to @xmath348 . combining , and
, we finally obtain the following representation of @xmath349 defined in .
[ thm : pab ] recall the conditions and definitions from through above . for @xmath228 , @xmath350 , \end{aligned}\ ] ] for @xmath351 , where @xmath346 and @xmath352 are defined in proposition [ propanalyticcont ] .
proposition [ propanalyticcont ] _
part i : decomposition_. first we decompose the contribution coming from the pole at @xmath353 and @xmath231 of @xmath354 , namely @xmath355 from lemma [ lemma1piuv ] below , @xmath356 , where @xmath357 is defined in ( [ eq3.18 ] ) . here , @xmath358 is well - defined pointwise and is in @xmath311 as we can check as in the proof of lemma [ lemma1piuv ] . hence using the identity ( recall that @xmath359 in ) @xmath360
the last term in ( [ eq34 ] ) becomes @xmath361 observe that the function @xmath362 is precisely the last two terms in ( [ eq31 ] ) . hence from , we obtain @xmath363 now a direct computation shows that @xmath364 here in the third equality , we used the fact that the contour @xmath365 can be made to be on the left of the point @xmath366 since @xmath367 . hence and imply . _
part ii : analyticity_. we now show that the functions on the right - hand - side of are analytic in @xmath348 . clearly , @xmath346 is analytic in since both the poles @xmath366 and @xmath368 lie inside the integration contour .
we need to show that @xmath369 is analytic .
note that @xmath293 , @xmath370 and @xmath371 are independent of @xmath372 . as @xmath373 is analytic in @xmath374
, @xmath375 is analytic in @xmath231 in the same domain .
also it is clear from the integral representation that @xmath376 is analytic in @xmath377 .
hence it it enough to show that @xmath369 is well - defined for @xmath378 .
fix @xmath379 .
let @xmath380 $ ] .
using the identity @xmath381 and the estimates and in lemma [ lemmaanalytic ] below , we see that @xmath382 for some constant @xmath383 . on the other hand , from , @xmath384 for some constant @xmath383 . therefore , the inner product is convergent , and the proposition is obtained .
[ lemma1piuv ] let @xmath385 .
define the function @xmath357 in @xmath386 by @xmath387 then @xmath388 defined by @xmath389 is well - defined for each @xmath390 and is in @xmath311 .
moreover , @xmath391 lemma [ lemma1piuv ] recall from that @xmath392 .
the estimate shows that the integral in is well - defined .
@xmath393 is well - defined pointwise .
note that @xmath394 for some function @xmath395 ( see ) .
hence the integral in equals @xmath396 .
hence the @xmath311 norm of the integral is bounded by @xmath397 . but
as @xmath398 by , we find that the integral in is in @xmath311 . now let @xmath399 and @xmath400 be fixed . in the integral formula of @xmath401 in , we can change the contour @xmath402 to either @xmath403 or @xmath404 .
we will use the contour @xmath403 when @xmath405 and the contour @xmath404 when @xmath406 . then for @xmath407 , @xmath408 where the integral
is evaluated using cauchy s formula . taking into account the case when @xmath409 , we obtain .
[ lemmaanalytic ] for any @xmath410 $ ] , there exists a constant @xmath383 such that @xmath411},\qquad x , y\in { \mathbb{r}},\ ] ] and @xmath412 for any @xmath377 , there is a constant @xmath383 such that @xmath413 finally , suppose that @xmath374 is given . for any @xmath414 $ ]
, there is a constant @xmath383 such that @xmath415 here the constants are uniform if the parameters @xmath416 are in compact sets .
lemma [ lemmaanalytic ] when @xmath417 , by deforming the contour to in , we obtain for @xmath418 , @xmath419 since the integrand is absolutely convergent ( recall that @xmath420 for @xmath279 ) .
we obtain the bound similarly by using the contour @xmath421 when @xmath422 .
the estimate is easily obtained by taking the contours @xmath423 and @xmath424 as the circles of radii of @xmath425 centered at @xmath426 and @xmath353 , respectively , in . recalling @xmath427 ( from ) we obtain the estimate by evaluating the residue at @xmath428 and making the remaining contour @xmath423 small enough . finally , in order to estimate @xmath375 , first note that @xmath429 by using the integral representation with the contour given by the circle of radius @xmath425 centered at @xmath353 . also from and , . using these estimates and @xmath430 in the definition , @xmath431 } \int_{-\infty}^{u_1 } dy\ , v_{i,1}(x , y ) f_1^{(a)}(y),\ ] ]
we obtain for @xmath136 bounded below .
in particular , to have the integrals over @xmath137 bounded we need @xmath432 for the integral with @xmath433 and @xmath434 for the integral with @xmath435 .
both conditions are satisfied for @xmath436 .
by setting @xmath229 in theorem [ thm : pab ] , we obtain @xmath437 . \end{aligned}\ ] ] we now begin asymptotic analysis of this formula . to obtain our main theorem ( theorem [ mainthm ] ) we need to consider the following scaling limit .
fix @xmath438 .
set @xmath439 , @xmath242 , and @xmath241 . for a large parameter @xmath54 , according to ( [ eq6 ] ) and ( [ eq : xycond ] ) , we consider @xmath440 where we order @xmath90 with @xmath441 $ ] for all @xmath197 . the convergence of the fredholm determinants is ensured only after ( yet another ) proper conjugation . for this purpose , set @xmath442 [ propasymptkbar ] consider the scaling ( [ eqscaling ] ) .
then @xmath443_{i , j}(s_i , s_j),\ ] ] uniformly for @xmath444 in a bounded set .
the operator @xmath102 is defined in ( [ eqkhat ] ) .
proposition [ propasymptkbar ] recall that @xmath445 , with @xmath446 for @xmath447 .
the same structure holds for @xmath102 . indeed , using the identity ( [ eqairyf ] ) of lemma [ lemairyexpressions ] ,
we can rewrite ( [ eqkhat ] ) as follows , @xmath448_{i , j}(s_i , s_j ) & = \int_0^\infty d\lambda \operatorname*{ai}(s_i+\lambda+\tau_i^2)\operatorname*{ai}(s_j+\lambda+\tau_j^2 ) e^{-\lambda(\tau_j-\tau_i)}\\ & -\frac{\exp\left(-\frac{(s_i - s_j)^2}{4(\tau_i-\tau_j)}+\frac23(\tau_j^3-\tau_i^3)+ ( \tau_js_j-\tau_is_i)\right)}{\sqrt{4\pi ( \tau_i-\tau_j ) } } { \mathbbm{1}}(\tau_i>\tau_j ) \end{aligned}\ ] ] the proof is divided into the convergence for @xmath449 in lemma [ lemasymptv ] and of @xmath450 in lemma [ lemasymptk ] below .
[ lemasymptv ] consider the scaling ( [ eqscaling ] ) and @xmath279 . then @xmath451 uniform for @xmath452 in a bounded set .
lemma [ lemasymptv ] recall that @xmath453 .
we derive the asymptotics by saddle point analysis . set @xmath454,\\ g_1(\widetilde z)&:=-\widetilde z ( s_i - s_j)\chi^{-1/3 } , \\
g(z ) & : = g_0(z)+t^{-1/3}g_1(z ) . \end{aligned}\ ] ] then by plugging into , @xmath455 there is a unique critical point for @xmath456 in the interval @xmath457 , namely @xmath458 a straightforward computation gives @xmath459 from which @xmath460 .
also observe that @xmath461 for the saddle point analysis we use the contour @xmath462 .
first let us show that the contribution coming from @xmath463 is negligible in the @xmath464 limit .
denote by @xmath465 .
then we have @xmath466\right ) \\ & = \left(\frac{t}{\chi}\right)^{1/3}\frac{1}{\pi } \int_{\delta}^\infty dt \frac{1}{(|1-{{\rm i}}t/(1/2-\widetilde z_c)| |1+{{\rm i}}t/(1/2+\widetilde z_c)|)^\eta } \\ & = \left(\frac{t}{\chi}\right)^{1/3}\frac{1}{\pi } \int_{\delta}^\infty dt \frac{1}{(1+t^2/\rho^2)^{\eta/2}(1+t^2/(1-\rho)^2)^{\eta/2}}\\ \end{aligned}\ ] ] with @xmath467 when @xmath54 is large enough . for @xmath468
, we have the bound @xmath469 hence using the linear lower bound @xmath470 for all @xmath471 , we have @xmath472 therefore there exist a constant @xmath473 and a constant @xmath474 such that @xmath475 when @xmath476 , we obtain the same estimate by just replacing in some of the bounds @xmath1 by @xmath477 .
next we determine the contribution from a @xmath425-neighborhood of the critical point .
noting the taylor series @xmath478 we find that @xmath479 thus by choosing @xmath425 small enough , we find that for @xmath480 , @xmath481 and also @xmath482 for some constant @xmath383 . using , and the general identity @xmath483 we deduce that @xmath484 finally , since @xmath485 for some constant @xmath486 , which is uniform for @xmath452 in a bounded set , we obtain from and . before discussing the asymptotic of @xmath370
we obtain also an exponential bound of @xmath487 when @xmath488 , which will be used later .
[ lemasymptvbound ] consider the scaling ( [ eqscaling ] ) and @xmath279 ( i.e. , @xmath453 ) .
then , for any given @xmath489 , there exists a @xmath490 large enough such that @xmath491 for all @xmath492 and @xmath493 .
the constant @xmath494 is uniform in @xmath444 and in @xmath495 .
lemma [ lemasymptvbound ] set @xmath496 a+\frac{{\varepsilon}\chi^{1/3}}{2(\tau_i-\tau_j)},&\textrm { if } s_i - s_j > { \varepsilon}t^{1/3},\\[0.5em ] a-\frac{{\varepsilon}\chi^{1/3}}{2(\tau_i-\tau_j)},&\textrm { if } s_i - s_j < { \varepsilon}t^{1/3 } , \end{array } \right.\ ] ] where @xmath497 is a fixed constant chosen small enough so that @xmath498 lies in a compact subset of @xmath457 .
taking the integration path as , a computation as in yields ( the function @xmath499 is defined in ) @xmath500 where @xmath501 when @xmath54 is large enough .
hence @xmath502 the last integral satisfies @xmath503 as @xmath504 .
therefore we find that there is a constant @xmath383 such that for @xmath54 large enough , @xmath505 now from and ( with @xmath506 and @xmath507 ) , if we have taken @xmath508 small enough , then @xmath509 now we plugging the value of @xmath498 in for three difference cases . when @xmath510 , @xmath511 when @xmath512 , then @xmath513 when @xmath514 , then @xmath515 therefore , we obtain @xmath516 therefore , for any given @xmath517 , by taking @xmath497 small enough but fixed and then taking @xmath518 large enough , there is a constant @xmath383 such that for all @xmath493 , @xmath519 and hence @xmath520 finally , given any @xmath489 , by take @xmath521
. then when @xmath522 , then @xmath523 since @xmath524 .
similarly , when @xmath525 , then @xmath526 this implies .
we now prove an asymptotic result for @xmath370 .
[ lemasymptk ] consider the scaling ( [ eqscaling ] ) . then , @xmath527 uniformly for @xmath444 in a bounded set .
lemma [ lemasymptk ] by the definition , @xmath528 the steepest - descent analysis of integrals very similar to this one with the same scaling has been performed repeatedly in various places ( see for example , @xcite ) .
the only difference here is that we have a double integral and we have to make sure that the two path do not touch .
however this can be easily handled by locally modifying the steep(est)-decent contours near the critical point . except for this modification ,
the analysis of our case is similar to those in the literature .
nevertheless we provide the proof for the completeness of the paper , and also since the analysis of this lemma and lemma [ lemasymptkbound ] is a prototype of all the other lemmas in this section below ( except for lemma [ lem : vlow ] ) .
the first step is to determine a steep descent path for @xmath529 and @xmath530 .
let us write @xmath531 where @xmath532,\\ h_{2,i}(w)&=-s_i w\chi^{-1/3}. \end{aligned}\ ] ] note that @xmath533 is independent of @xmath197 .
for the steep descent analysis we need to determine a steep descent path for @xmath534 .
the steep descent path will always be taken symmetric with respect to complex conjugation .
thus we can restrict the discussion below to the part lying in the upper half plane .
we have @xmath535 which both vanishes at the critical point @xmath536 , and @xmath537 .
let @xmath538 with @xmath539 .
then , for , @xmath540 is strictly decreasing in @xmath133 .
indeed , @xmath541 since the last parenthesis is strictly positive : for @xmath542 , @xmath543 and @xmath544 , while @xmath545 . in the neighborhood of the critical point
we consider a second path , @xmath546\}$ ] . along that path
, we have @xmath547 the second degree term , @xmath548 , is strictly positive for all @xmath7 and @xmath549 $ ] .
therefore also @xmath550 is a steep descent path , and close to the critical point will be steepest descent .
now we can define the steep descent path used in the analysis .
let @xmath551 .
in a similar way , one obtains a steep descent path for @xmath552 , namely , @xmath553 is the path obtained by rotation around the origin of the steep descent path @xmath554 but with @xmath477 instead of @xmath1 . finally ,
as we shall discuss below , any local modification of the contours in a region of order @xmath555 around the critical point is allowed , see figure [ figsteepk ] , so to have @xmath556 for a fixed @xmath497 .
[ c]@xmath554 [ c]@xmath553 [ c]@xmath231 [ c]@xmath353 [ c]@xmath426 and @xmath553 are both anticlockwise oriented .
the modification of @xmath553 close to the critical point @xmath231 is of order @xmath555.,title="fig:",height=188 ] the path @xmath554 and @xmath553 are steep descent path for @xmath533 and @xmath552 .
therefore , for any given small @xmath557 , the contribution to the double integral ( [ eq84 ] ) coming from @xmath558 is only of order @xmath559 for some @xmath473 ( with @xmath560 for small @xmath425 ) , which is smaller than @xmath561 .
note that @xmath562 is the value of the integrand at the critical point .
next we analyze the contribution coming from a @xmath425-neighborhood of the critical point @xmath563 .
there we can use taylor series expansion of @xmath564 , which are given by @xmath565 therefore , the main contribution to ( [ eq84 ] ) is given by @xmath566 where @xmath567 and @xmath568 are the pieces of @xmath554 and @xmath553 which lies in a @xmath425-neighborhood of the critical point @xmath563 .
more precisely , setting @xmath569 one has that @xmath570 assume that @xmath571 , then @xmath572 and @xmath573 thus , it is an easy computation to check that one can find @xmath574 small enough so that @xmath575 assume that @xmath576 .
then it is easy to show as in ( [ borneot13 ] ) , lemma [ lemasymptv ] that @xmath577 it thus remains us only to determine the asymptotic of ( [ eq90 ] ) .
we make the change of variables @xmath578 and @xmath579 and obtain @xmath580 where the integration paths do not touch .
we can now extend the integration path to infinity , since @xmath581 the conjugation by @xmath582 and the identity ( [ eqairya ] ) end the proof .
now we give a bound which holds uniformly for @xmath444 bounded from below .
let @xmath583 be given .
[ lemasymptkbound ] consider the scaling ( [ eqscaling ] ) .
then , for any given @xmath489 , there exists a @xmath490 large enough such that for all @xmath584 and @xmath585 @xmath586 the constant @xmath494 depends on @xmath587 only . lemma [ lemasymptkbound ] the upper bound of the integrals similar to has also been obtained in various places ( see for example , @xcite ) .
the analysis in our case is similar to those in the literature . however , again we provide the proof for the completeness of the paper ; also the analysis in this proof is going to be used and adapted in all the lemmas below in this section ( except for lemma [ lem : vlow ] ) .
first of all , we can rewrite @xmath588 set @xmath589 ( which corresponds to @xmath590 ) .
then @xmath591 where @xmath592 we now show the exponential decay of @xmath593 for large positive @xmath594 .
lemma [ lemasymptk ] indeed implies the result when @xmath595 .
it is thus enough to consider the case where @xmath596 .
the analysis of @xmath597 is made in exactly the same way , up to a rotation around the origin and exchange of @xmath1 with @xmath477 . to this aim
we modify the contour given on figure [ figsteepk ] as follows , defining a new contour @xmath598 .
call @xmath599 the part of the contour lying in the upper half - plane @xmath600 .
let @xmath489 be a fixed constant and @xmath601 then we set @xmath602 the contour @xmath598 is then completed by adjoining the conjugate of @xmath599 .
this modification of the contour @xmath554 has no impact on the saddle point analysis made in the proof of lemma [ lemasymptk ] .
indeed , for @xmath54 large enough @xmath603 where @xmath604 is a uniformly bounded constant as @xmath605 is chosen in a compact interval of @xmath606 along @xmath598 it holds , @xmath607 thus , along @xmath598 @xmath608 using the saddle point argument used in the proof of lemma [ lemasymptk ] and ( [ star ] ) , we easily deduce that there exists @xmath609 independent of @xmath594 such that for @xmath54 large enough @xmath610 thus , @xmath611 for some other constant @xmath494 .
[ c]@xmath612 [ c]@xmath612 [ c]@xmath231 [ c](a ) [ c](b ) [ c](c ) [ c]@xmath353 [ c]@xmath426 that are only of order @xmath555.,title="fig:",height=188 ] the saddle point analysis made for @xmath370 in the preceding lemmas will now be used , up to minor modifications , to consider the asymptotics of the remaining terms .
[ lemr ] uniformly for @xmath613 in a bounded interval one has that @xmath614 moreover , for any given @xmath489 , there exists a @xmath490 large enough , such that for all @xmath495 it holds @xmath615 for any @xmath616 .
the constant @xmath494 does not depend on @xmath617 or @xmath613 .
lemma [ lemr ] we have @xmath618 and also @xmath619\ ] ] with @xmath620 s defined in ( [ eq89 ] ) .
as steep descent path we use almost the same contour @xmath553 as in figure [ figsteepk ] , up to @xmath555-local deformation so that it passes to the right of @xmath231 , see figure [ figsteeptwo ] ( a ) .
then , using the same asymptotic argument as in the proof of lemma [ lemasymptk ] , we deduce that @xmath621 where with @xmath622 we mean the path from @xmath623 to @xmath624 and passing on the right of @xmath143
. we can take it to its left up to adding a residue term , which is just @xmath613 .
then the identity ( [ eqairyc ] ) gives the first formula of the lemma .
+ for the exponential decay , one first computes the residue at @xmath625 , yielding that @xmath626 the contour now lies to the left of @xmath231 , which ensures using the same argument as in lemma [ lemasymptkbound ] , the exponential decay for large positive @xmath627 now we we turn to the asymptotics of the function @xmath628 defined in ( [ eq22 ] ) and the different terms of the function @xmath352 defined in ( [ eq31 ] ) .
[ lemf12 ] consider the scaling ( [ eqscaling ] ) .
then @xmath629 uniformly for @xmath594 in a bounded set .
moreover , for any given @xmath489 , there exists a @xmath490 large enough , such that for all @xmath495 it holds @xmath630 for some constant @xmath494 independent of @xmath54 and uniform for @xmath594 bounded from below .
lemma [ lemf12 ] we have @xmath631 with the @xmath620 s defined in ( [ eq89 ] ) . as a steep descent path we use @xmath554 in figure [ figsteepk ] with a local modification on the @xmath555 scale so that it passes on the right of @xmath231 , see figure [ figsteeptwo ] ( b ) .
then , the same asymptotic argument of the proofs of lemma [ lemasymptk ] and lemma [ lemasymptkbound ] still holds and gives us the claimed result
. we also use the identity ( [ eqairyd ] ) to express the result in terms of airy functions .
[ lemkfa ] consider the scaling ( [ eqscaling ] ) .
then @xmath632 uniformly for @xmath633 in a bounded set .
moreover , for any given @xmath489 , there exists a @xmath490 large enough , such that for all @xmath495 it holds @xmath634 for some constant @xmath494 independent of @xmath54 and uniform for @xmath635 bounded from below .
lemma [ lemkfa ] notice that @xmath636 .
then , choosing the integration path for @xmath637 in ( [ eq24 ] ) so that @xmath638 , we can explicitly integrate the variable @xmath137 and get @xmath639 where the integration path for @xmath637 does not include @xmath231 and does not cross the integration path for @xmath640 .
the rest follows along the same lines of lemma [ lemasymptk ] and lemma [ lemasymptkbound ] .
finally one uses ( [ eqairye ] ) to rewrite the double integral expression in terms of airy functions .
[ lemvfa][lem : vlow ] consider the scaling ( [ eqscaling ] ) and let @xmath641 .
then @xmath642 uniformly for @xmath643 in a bounded set .
moreover , for any given @xmath489 , there exists a @xmath490 large enough , such that for all @xmath495 it holds @xmath644}e^{-2\tau_i^3/3-\tau_is_i } \bigg)\ ] ] for some constant @xmath494 independent of @xmath54 and uniform in @xmath635 .
lemma [ lemvfa ] using @xmath645 and modifying the integration path in ( [ eq24 ] ) to @xmath646 for any @xmath647 , we can integrate out the @xmath137 variable with the result @xmath648 for the saddle point analysis we then deform back our integration path to @xmath649 with only a local deviation order @xmath555 around @xmath625 to make it passing on the right of @xmath231 . specifically , for a given small @xmath497 , we use the integration path @xmath650\},\ ] ] see figure [ figsteeptwo ] ( c ) . the asymptotic analysis for @xmath643 in a bounded set
is the same as in lemma [ lemasymptv ] except that the integral has an extra @xmath651 in the denominator .
explicitly , we get @xmath652 for any @xmath497 .
the function @xmath653 satisfies @xmath654 and the boundary condition @xmath655 therefore , @xmath656 is a gaussian distribution function , so that our limiting function is @xmath657 in order to prove , we adapt the proof of lemma [ lemasymptvbound ] .
first , suppose that @xmath658 .
for this case , we take the contour @xmath659 as in the proof of lemma [ lemasymptvbound ] .
note that @xmath660 .
hence and the term @xmath661 in is bounded by @xmath126 and an analysis as in the proof of lemma [ lemasymptvbound ] as in the proof ( noting the presence of @xmath62 in ) implies .
secondly , suppose that @xmath662 .
we take the same contour . in this case , the residue at the simple pole @xmath231 should be taken into account . except for this term ,
the analysis on the contour @xmath554 is the same and we obtain . finally , when @xmath663 , we take the contour @xmath664 where @xmath550 is the part of the contour @xmath665 which lies outside the circle of radius @xmath666 centered at @xmath231 , and @xmath667 is the part of the circle of radius @xmath23 centered at @xmath231 whose real part is at least @xmath498 .
this contour is same as the straight line @xmath665 , except that it goes around @xmath231 on the right in a @xmath668-neighborgood of @xmath231 . on @xmath550 , as @xmath669 , and hence we obtain , as in the proof of lemma [ lemasymptvbound ] , @xmath670 on the other hand , on @xmath667 , observe that @xmath671 and the arc length of @xmath667 is also @xmath668 . by using the estimate and , which holds on @xmath667 if @xmath54 is large enough , we obtain that @xmath672 is bounded for @xmath673 ( where @xmath499 is in ) .
hence ( recall the first equation of ) @xmath674 this , together with implies .
the last term to be computed is the asymptotic of @xmath628 .
[ lemg ] consider the scaling ( [ eqscaling ] ) .
then @xmath675 uniformly for @xmath676 in a bounded set . moreover , for any given @xmath489 , there exists a @xmath490 large enough , such that for all @xmath495 it holds @xmath677 for some constant @xmath494 independent of @xmath54 and uniform for @xmath676 bounded from below .
lemma [ lemg ] we have @xmath678 with the @xmath620 s defined in ( [ eq89 ] ) . as steep descent path we use @xmath553 in figure [ figsteepk ] with a local modification on the @xmath555 scale so that it passes on the right of @xmath231 , see figure [ figsteeptwo ] ( a ) .
then , the asymptotic argument of the proofs of lemma [ lemasymptk ] and lemma [ lemasymptkbound ] still holds and gives us the claimed result .
the only small difference is that for the bound the integration path for large @xmath676 s has to be chosen to pass on the left of @xmath231 , which is fine up to adding the residue at @xmath231 .
the factor @xmath679 comes from the multiplication by @xmath680 . to rewrite the result in terms of airy functions we used ( [ eqairyb ] ) .
in this section we prove theorem [ mainthm ] . recall @xmath681 . \end{split}\ ] ] where @xmath682 and @xmath241 .
we take the limit @xmath464 with the scaling .
note that the scaling of is related to the scaling by the identities @xmath683 and @xmath684 .
the difference ( at most ) @xmath685 does not contribute to the asymptotics and only makes the notations complicated . for this reason
, we will ignore this difference @xmath685 in the following presentation .
[ propconvkai ] fix @xmath686 , and real numbers @xmath90 , @xmath687 .
let @xmath688 , @xmath689 and @xmath690 be defined as in
. then it holds @xmath691 where @xmath692}$ ] and @xmath101}$ ] .
proposition [ propconvkai ] proposition [ propasymptkbar ] implies that for @xmath444 in a bounded interval @xmath693 @xmath694_{i , j}(s_i , s_j),\ ] ] and that the convergence is uniform on @xmath693 .
now lemmas [ lemasymptvbound ] and [ lemasymptkbound ] then imply the convergence in trace class norm of the rescaled kernels @xmath695 and @xmath696 .
this completes the proof .
[ propconvscalarproduct ] fix @xmath686 real numbers @xmath90 , .
let @xmath688 , @xmath689 and @xmath690 be defined as in .
then @xmath697 where @xmath698 is defined in .
proposition [ propconvscalarproduct ] consider the term @xmath699 .
we first scale and also conjugate by some operators .
define the function @xmath700 set @xmath701 , and define the multiplication operator @xmath702 set @xmath703 and @xmath704 where one recalls the definitions of @xmath705 in ( [ eq31 ] ) and @xmath706 in ( [ eq22 ] ) . here
the superscript @xmath54 indicates the dependence on @xmath54 . by lemmas [ lemf12 ] , [ lemkfa ] , [ lemvfa ] and [ lemg ] , for any given @xmath489
, there exists a @xmath490 large enough , such that for all @xmath495 , @xmath707 and @xmath708 are @xmath709 for any given @xmath594 .
then for @xmath54 large enough , @xmath710 let @xmath711 be the multiplication operator @xmath712 and set @xmath713 where the functions @xmath714 are defined in .
observe that the functions @xmath715 .
since @xmath716 it is enough to prove @xmath717 where .
clearly , @xmath718 first consider @xmath719 . as the operator - norm is bounded by the hilbert - schmidt norm @xmath720 , @xmath721 by taking @xmath722 in lemmas [ lemasymptkbound ] and [ lemasymptvbound ] , and using the dominated convergence theorem and proposition [ propasymptkbar ] , the limit as @xmath464 of becomes @xmath143 . by lemma [ leminvertibility ] ,
the operator @xmath723 is invertible and it follows that @xmath724 as @xmath464 .
now consider the term @xmath725 . from of lemma [ lemg ]
, we have @xmath726 for @xmath727 for a fixed @xmath676 .
the conjugation @xmath728 is introduced to make this function decay exponentially as @xmath729 .
hence @xmath730 is bounded uniformly for large enough @xmath54 . also noting that the right - hand - side of is precisely @xmath731 , it follows from dominated convergence theorem that @xmath732 as @xmath464 .
it follows from lemmas [ lemf12 ] , [ lemkfa ] and [ lemvfa ] that @xmath733 for @xmath734 for a fixed @xmath676 .
hence @xmath735 is uniformly bounded for large enough @xmath54 if we set for example @xmath736 .
also it follows that @xmath737 as @xmath464 .
hence we have shown that converges to .
the remaining term in @xmath738 is @xmath739 .
but lemma [ lemr ] shows that this converges to @xmath740 ( after the changes of variables @xmath741 ) .
this completes the proof .
the above two propositions proves that @xmath742 converges to @xmath743 as @xmath464 .
it remains to show that the derivatives with respect to @xmath744 s also converge .
let us first consider @xmath745 of @xmath738 .
as in , this equals @xmath746 .
note that the dependence on @xmath744 is only through the projection operator @xmath100 .
hence by simple translations , @xmath747 where @xmath748 , @xmath749 and @xmath750 , and all the ( real ) inner product , the operator and the functions are all defined on @xmath751 .
then @xmath752 hence we need to control the asymptotics of the derivatives of the kernels .
but since all the asymptotic bounds for large @xmath136 in the lemmas in section [ sectasympt ] are all exponential , while the derivatives of the kernels yields only polynomial terms , we can check that the dominated convergence theorem still applies and obtain the convergence of to the corresponding derivatives of the limit .
we omit the detail .
the other terms are also similar .
this completes the proof of theorem [ mainthm ] . to prove theorems [ theoremdppext]-[theoremtasepb ] we use the following slow - decorrelation result , which is an extension of proposition 8 of @xcite .
[ lemslowdecorrelation ] let @xmath753 for some @xmath754 .
let then with @xmath755 with @xmath756 .
then , for any @xmath757 , it holds @xmath758 lemma [ lemslowdecorrelation ] first of all , it is easy to check by proposition 2.2 of @xcite that the difference between the model and the last passage percolation model corresponding exactly to the stationary tasep becomes irrelevant in the order as @xmath464 .
denote @xmath759 the measure for the last passage percolation corresponding to the stationary tasep . due to the stationarity in space and time of tasep
, we observe that @xmath760 and @xmath761 have the same distribution .
therefore , @xmath762 but since the distribution of @xmath763 converges to @xmath764 @xcite , and @xmath765 as @xmath464 , we find that converges to @xmath5 as @xmath464 .
this completes the proof .
theorem [ theoremdppext ] the projection of @xmath766 on the line along the critical direction @xmath767 is @xmath768 with @xmath769 defined in ( [ eq6 ] ) .
we have @xmath770 with @xmath771 . in particular
then , @xmath773 with @xmath774 by lemma [ lemslowdecorrelation ] we have @xmath775 so that @xmath776 see the proof of theorem 1 of @xcite for detailed steps .
this is by theorem [ mainthm ] the desired result .
theorem [ theoremtasep ] the goal is the express our setting into the one of theorem [ theoremdppext ] . denote @xmath777 , which is given by @xmath778 and recall @xmath779 ( since @xmath780 and @xmath781 are integers , the above formulas are are exact up to the error of size @xmath214 at most : since this difference does not affect the asymptotics but only complicates the formulas , we drop this difference in the following presentation . ) by ( [ eqdptasep ] ) we have @xmath782 denote by @xmath783 the projection of @xmath784 on the line along the critical direction @xmath767 .
we get @xmath785 where @xmath786 let us further denote @xmath787 then , @xmath788 where @xmath789 is defined in .
then , replacing ( [ eq5.22 ] ) into ( [ eq5.21 ] ) we get @xmath790 where @xmath791 are defined in ( [ eq6 ] ) . with these notations
we can rewrite ( [ eq5.20 ] ) as follows , @xmath792 where @xmath793 . then
, by theorem [ theoremdppext ] the result follows . theorem [ theoremtasepb ] the proof of theorem [ theoremtasepb ] is very similar to the one of theorem [ theoremtasep ] .
the scaling ( [ eqscalingheight ] ) corresponds , in terms of @xmath769 , to the scaling @xmath794 the projections @xmath783 of @xmath28 on the line @xmath795 along the critical direction @xmath767 are given by @xmath796 , @xmath797 .
explicitly we find @xmath798 setting @xmath799 we get @xmath800 then , setting @xmath801 we get ( [ eq5.25 ] ) and then by theorem [ theoremdppext ] the result follows .
in the asymptotic analysis we get two basic integral expressions which can be rewritten in terms of airy functions and exponential , namely @xmath802 starting from these two formulas we state some identities .
[ lemairyexpressions ] @xmath803 @xmath804 @xmath805 @xmath806 @xmath807 @xmath808 lemma [ lemairyexpressions ] ( a ) since we can choose the paths for @xmath809 and @xmath810 such that @xmath811 , using @xmath812 and ( [ eqairy ] ) we get ( [ eqairya ] ) .
\(b ) let @xmath813 . then differentiating and using ( [ eqairy ] ) we get @xmath814 .
together with the boundary condition @xmath815 , we get ( [ eqairyb ] ) .
\(c ) let @xmath816 . differentiating twice and applying ( [ eqairy ] ) we obtain @xmath817 .
the boundary conditions are @xmath818 . integrating twice and shifting the integration bounds to zero , we get ( [ eqairyc ] ) .
\(d ) this can be derived by ( [ eqairyb ] ) by change of variable @xmath819 .
\(e ) this identity can be found by first using ( [ eq141 ] ) and then the two complex integrals are decoupled . for the integral over
@xmath809 one uses ( [ eqairy ] ) , while for the integral over @xmath810 one uses ( [ eqairyb ] ) .
\(f ) one applies the identity ( 2.20 ) of @xcite , which holds for @xmath820 , @xmath821 and then we set @xmath822 , @xmath823 .
[ leminvertibility ] for any fixed real numbers @xmath91 , @xmath824 exists .
lemma [ leminvertibility ] first of all , notice that @xmath825 now , let @xmath826 . then , @xmath827 the equality with the goe tracy - widom distribution , @xmath828 , is proven in corollary 1.3 of @xcite , while the strict inequality follows from the monotonicity of the distribution function and the large @xmath829 asymptotics ( see for example , @xcite ) . moreover , as shown in section 2.2 of @xcite , @xmath115 is trace - class ( the shift by @xmath830 is irrelevant for that property , since it holds for any @xmath744 s ) .
the proof ends by applying a known result on fredholm determinant , see e.g. theorem xiii.105 ( b ) in @xcite : let @xmath831 be a trace - class operator , then @xmath832
in this section , we discuss the basic properties of the operators @xmath293 and @xmath276 in section [ sectshift ] .
[ proptraceclass]@xmath833 * the operators @xmath834 and @xmath278 are bounded in @xmath291 .
* let @xmath835 .
fix constants @xmath286 such that @xmath836 then @xmath837 defined in is a trace - class operator on @xmath291 . on the other hand
, @xmath838 defined in is a trace - class in the same space if the constants satisfy more restricted condition @xmath839 proposition [ proptraceclass ] we only prove ( ii ) .
the proof of ( i ) follows easily by suitably modifying the analysis for part ( ii ) .
we start with @xmath837 .
since the set of trace - class operators is a linear space , it is enough to prove that @xmath840 is a trace - class operator in @xmath311 for each @xmath841 . as @xmath842 ( see ) , we prove that each of the two operators is trace - class . _
( a ) operator @xmath450 : _ it is easy to check that the operator can be re - expressed as ( see ( [ eq96 ] ) ) @xmath843 where @xmath844 by choosing the integration paths close enough to @xmath353 , resp .
@xmath426 , a simple estimate shows that there exists @xmath845 so that @xmath846 from this it follows immediately that @xmath847 and @xmath848 are hilbert - schmidt operators on @xmath311 . so is the conjugated kernel of @xmath450 trace - class .
_ ( b ) operator @xmath449 : _ as @xmath446 for @xmath849 , assume that @xmath279 . in this case
, we have @xmath850 from ( [ eq24 ] ) , by plugging in @xmath851 and @xmath852 and by changing the contour from @xmath402 to @xmath853 , we obtain @xmath854 where @xmath855 hence @xmath856 the fourier transform of @xmath354 .
note that @xmath857 where @xmath858 as @xmath859 so then is @xmath860 .
moreover , as @xmath861 is well - defined pointwise except at @xmath862 ( since it is conditionally convergent ) , @xmath863 equals almost everywhere .
furthermore , we can compute explicitly and obtain @xmath864}.\ ] ] similarly , @xmath865}.\ ] ] we have @xmath866 hence @xmath867 here all the steps make sense since @xmath868 .
note @xmath869 for @xmath279 , and @xmath870 . by using , we see that @xmath871 where @xmath872 and @xmath873 are the operators with kernel @xmath874},\\ \mathcal{h}(x , y)&= \frac{(x - y)^{t_i - t_j-1}}{(t_i - t_j-1 ) ! } e^{-\frac{(x - y)}2 } { \mathbbm{1}}_{[x - y>0 ] } , \end{aligned}\ ] ] and @xmath875 is the multiplication operator @xmath876 then @xmath877 changing the variable @xmath137 by @xmath878 ,
the above equals @xmath879 which is finite .
hence @xmath880 is a hilbert - schmidt operator .
similarly , @xmath881 is finite , and hence @xmath882 is a hilbert - schmidt operator .
therefore , @xmath883 is a trace - class operator .
now we consider @xmath288 . from ,
@xmath884 but by changing the contour in , @xmath885 hence there is a constant @xmath383 such that @xmath886 for some constant @xmath887 for @xmath888 .
therefore , @xmath889 if @xmath890 .
similarly , @xmath891 if @xmath892
. being a product of two hilbert - schmidt operator , @xmath893 is trace - class if holds .
this completes the proof .
consider the directed percolation model defined in ( [ eqdp ] ) , but look away from the characteristic line .
we set @xmath894 denote @xmath895 the passage time from @xmath896 to @xmath28 .
then @xmath897 in section 6 of @xcite ( see also @xcite ) a last passage percolation with one source only is considered .
nevertheless , the arguments given therein can readily be used to prove the following : if @xmath898 @xmath899 where @xmath900 and @xmath901 if @xmath902 then there exists a constant @xmath903 ( see @xcite ) such that @xmath904 where @xmath905 and @xmath906 is the gue tracy - widom distribution . by symmetry , if @xmath902 , setting @xmath907 and @xmath908 , it holds @xmath909 and for @xmath910 @xmath911 consider now the case @xmath912 .
then since @xmath913 we have @xmath914 with the notation @xmath915 and ( [ eqgauss0 ] ) we obtain @xmath916 we take the limit @xmath34 on both sides and by ( [ eqgauss1 ] ) and ( [ eqgauss3 ] ) we get @xmath917 p.l .
ferrari . .
in j. baik , t. kriecherbauer , l - c .
li , k. mclaughlin , and c. tomei , editors , _ integrable systems and random matrices : in honor of percy deift _ , contemporary math . ,
pages 321332 .
soc . , 2008 .
k. johansson .
random matrices and determinantal processes . in a.
bovier , f. dunlop , a. van enter , f. den hollander , and j. dalibard , editors , _ mathematical statistical physics , session lxxxiii : lecture notes of the les houches summer school 2005 _ , pages 156 .
elsevier science , 2006 . | the totally asymmetric simple exclusion process ( tasep ) on @xmath0 with the bernoulli-@xmath1 measure as initial conditions , @xmath2 , is stationary .
it is known that along the characteristic line , the current fluctuates as of order @xmath3 .
the limiting distribution has also been obtained explicitly . in this paper
we determine the limiting multi - point distribution of the current fluctuations moving away from the characteristics by the order @xmath4 .
the main tool is the analysis of a related directed last percolation model .
we also discuss the process limit in tandem queues in equilibrium . |
the hyperbolic distance has turned out to be a useful tool in geometric function theory .
the basic models for the hyperbolic distance are the unit ball model and the upper half space model . using these models in the plane case @xmath0
, we can find the hyperbolic distance in any domain with at least 2 boundary points via the riemann mapping theorem . in higher dimensions @xmath1 , there are no such results we could use to consider the hyperbolic distance in general domains .
a solution to this is to use other distance functions , which approximate the hyperbolic distance and are easier to evaluate .
we call this kind of distance functions hyperbolic type distances .
the study of hyperbolic distances was initiated four decades ago by gehring , palka , martin and osgood @xcite .
thereafter many researchers have studied hyperbolic type metrics or used them as a tool in their work , see for example @xcite . in this article
we are interested in the growth of hyperbolic type distances in proper subdomains @xmath2 .
we consider the growth along a euclidean line segment from @xmath3 to @xmath4 . by a linear transformation
we may assume that @xmath5 . to ensure that the line segment @xmath6 is in @xmath7 we restrict our study to starlike domains , which means that for every @xmath8 the euclidean line segment @xmath9 $ ] is contained in @xmath7 .
let @xmath10 be a starlike domain and @xmath4 .
let @xmath11 and denote @xmath12 .
we study the behaviour of the function @xmath13 and the function @xmath14 . ] where @xmath15 is a hyperbolic type distance , see figure [ fig : function f ] .
note that now @xmath16 is a continuous mapping from @xmath17 to @xmath18 . to study this function
we start of with an easier one @xmath19 all the results of our study are true in a more general setting as long as the line segment @xmath20 is contained in the domain . however , we consider starlike domains as then this condition is clearly fulfilled . our main result is the following schwarz lemma type theorem
. for definition of the distances see the section named after the distance .
[ thm : main theorem ] let @xmath2 be a domain with @xmath21 and @xmath22 be the function defined in for any @xmath4 . for @xmath23 @xmath24 and for @xmath25 @xmath26 where the upper and lower bounds are best possible , and @xmath27
in this section we introduce notation and consider examples of the hyperbolic distance in the upper half space and a ball . for @xmath29 we denote the closed euclidean line segment between the points by @xmath30 = \ { c \in { { \mathbb{r}^n}}\colon c = x+t(y - x ) , \ , 0 \le t \le 1 \}$ ] .
we also use notation @xmath31 , @xmath32 and @xmath33 $ ] for open and half - closed line segments in @xmath34 .
we denote the smallest angle between line segments @xmath30 $ ] and @xmath35 $ ] by @xmath36 .
we denote euclidean balls and spheres with centre @xmath37 and radius @xmath38 , respectively , by @xmath39 and @xmath40 .
a domain @xmath2 , @xmath21 , is said to be starlike , if it is strictly starlike with respect to @xmath41 : for every @xmath4 the line segment @xmath20 is contained in @xmath7 .
we say that distance function @xmath42 in @xmath43 is hyperbolic type , if @xmath44 , where @xmath45 is the hyperbolic distance in the unit ball defined in .
let @xmath46 .
the hyperbolic distance for all @xmath47 is defined as @xmath48 note that we use shifted version of the upper half space , as the usual upper half space @xmath49 does not contain the origin .
[ exa : rho in hnb ] let @xmath50 and @xmath11 with @xmath51 .
we show that in this case @xmath52 .
now for @xmath53 we have @xmath54 and by differentiation we obtain @xmath55 we consider @xmath56 by taking the limit @xmath57 as @xmath58 .
@xmath59 let @xmath60 be a ball with @xmath37 and @xmath61 .
for @xmath62 the hyperbolic distance is defined by @xmath63 [ exa : rho in b ] let @xmath64 and @xmath65 .
we show that in this case @xmath66 .
now for @xmath67 @xmath68 and @xmath69 . taking the limit we obtain @xmath70 as @xmath58 .
@xmath59 examples [ exa : rho in hnb ] and [ exa : rho in b ] suggest that for hyperbolic type distances @xmath71 could hold for @xmath72 or perhaps even for @xmath73 .
it turns out that this conjecture is not true in general .
however , based on our study in this article , it seems that the following is true for a hyperbolic type distance @xmath15 there exists constants @xmath72 and @xmath74 such that @xmath75
in this section we study the problem for the distance to the boundary function @xmath77 formulated in .
let us begin our study with few simple planar examples , which are easy to reconstruct in higher dimensions .
[ exm : simple domain ] we consider starlike domain @xmath78 where @xmath79 with @xmath80 . for a given @xmath81
we derive a formula for @xmath77 and show that @xmath82 and thus linear for large values of @xmath83 .
we choose @xmath84 for @xmath85 and note that @xmath4 . of examples [ exm : simple domain ] ( on left ) and
[ exm : circular domain ] ( on right).,title="fig : " ] of examples [ exm : simple domain ] ( on left ) and [ exm : circular domain ] ( on right).,title="fig : " ] when @xmath83 is small , then @xmath86 is close to @xmath41 and @xmath87 . to be more explicit
we can write @xmath88 where @xmath89 is the point in @xmath90 $ ] , with @xmath91 , see figure [ fig : simple example ] . if @xmath83 is faraway from @xmath41 , the we have @xmath92 where @xmath93 and @xmath94 .
next we want to express our function @xmath77 in terms of @xmath83 and points @xmath95 and @xmath96 .
we easily obtain @xmath97 , @xmath98 , @xmath99 , @xmath100 and @xmath101 putting all together gives @xmath102 and here @xmath98 .
@xmath59 [ exm : circular domain ]
let us consider starlike domain @xmath103 and @xmath104 , see figure [ fig : simple example ] .
for @xmath11 and @xmath105 we have @xmath106 now @xmath107 implying @xmath108 as @xmath109
. @xmath59 [ exm : polynomial domain ] in example [ exm : circular domain ] we could place the circular arc with any decreasing function : first define a decreasing function @xmath110 \to [ 0,1]$ ] with @xmath111 and @xmath112 . then reflect the function across the real axis and then reflect both function across the imaginary axis . for @xmath113 we can define polynomial function @xmath114 to obtain a domain @xmath7 .
now for @xmath104 we have @xmath115 and since @xmath116 we obtain @xmath117 as @xmath109 .
@xmath59 let us then consider general case @xmath2 .
how quickly and how slowly @xmath77 can decrease for large values of @xmath83 ? how quickly and how slowly @xmath77 can increase for @xmath83 close to @xmath41 ? before considering the bounds for @xmath77 , we introduce angular domain . for @xmath118 and @xmath119
we define @xmath120 if @xmath83 is close to @xmath41 , then the slowest growth for @xmath77 occurs in the following case : @xmath121 for some @xmath122 and @xmath123 , see figure [ fig : domain g2 ] .
now for @xmath124 $ ] , we have @xmath125 defined in for the extremal case , when @xmath83 is close to @xmath41 . ]
if @xmath83 is faraway from the boundary , then the slowest growth occurs in the case @xmath126 for any @xmath122 and @xmath127 .
now @xmath128 for @xmath129 .
as we will later see , it turns out that the domains @xmath130 and @xmath131 defined respectively in and can be used as extremal domains for many hyperbolic type distances . note that if @xmath7 was not starlike , we could use @xmath132 and @xmath133 instead of @xmath130 and @xmath131 .
next we consider how quickly @xmath77 can decrease .
close to the origin the fastest decrement occurs in the domain @xmath134 and it is @xmath135 for @xmath136 .
faraway from the origin the decrement can made arbitrarily slow as can be observed from examples [ exm : simple domain ] , [ exm : circular domain ] and [ exm : polynomial domain ] .
we have arrived to the following theorem : [ thm : distance to the boundary estimates ] for any starlike domain @xmath7 and point @xmath4 we have @xmath137 moreover , @xmath138
in this section we estimate @xmath140 for the distance ratio distance @xmath139 , which is defined in any open subset @xmath2 for points @xmath141 by @xmath142 the distance ratio distance was introduced by vuorinen in the 1980 s @xcite and in a slightly different form by gehring and osgood @xcite . for @xmath140 we can use the same domains as for @xmath77 to consider the extremal cases .
note that the following result is true also in non starlike domains .
[ thm : distance ratio estimates ] let @xmath2 be a starlike domain and @xmath4 . for the distance ratio distance @xmath139 we have @xmath143 , @xmath144 and @xmath145 let us first consider lower bound for @xmath146 .
we denote a closest boundary @xmath147 with @xmath148 and @xmath4 .
now @xmath149 with @xmath51 we have @xmath150 this is obtained for example as in in the domain @xmath151 for some @xmath122 and @xmath123 .
let us then consider upper bound for @xmath146 . now as @xmath152 is a constant we may assume @xmath153 and obtain by @xmath154 this situation is obtained in the domain @xmath155 , because for any @xmath7 we have @xmath156 .
the upper and lower bounds of @xmath140 give us @xmath157 which yields @xmath143 .
note that the lower bound for @xmath140 in theorem [ thm : distance ratio estimates ] works for all @xmath83 whereas the upper bound holds only close to the origin . as a matter of fact , by theorem [ thm :
distance to the boundary estimates ] , the function @xmath158 can be arbitrarily close to zero near the boundary and therefore we can not find an upper bound for @xmath140 in this case .
next we consider the quasihyperbolic distance .
let @xmath2 be a domain .
we define the quasihyperbolic length of a curve @xmath160 by @xmath161 for @xmath141 the quasihyperbolic distance between @xmath162 and @xmath163 is define by @xmath164 where the infimum is taken over all rectifiable curves @xmath165 joining @xmath162 and @xmath163 in @xmath7 .
the quasihyperbolic distance was introduced in the 1970 s by gehring and palka @xcite .
[ thm : quasihyperbolic estimates ] let @xmath2 be a starlike domain and @xmath4 .
for the quasihyperbolic distance @xmath159 we have @xmath166 , @xmath167 and @xmath168 we start by finding a lower bound for @xmath169 .
let @xmath4 and @xmath11 .
for @xmath170 we estimate @xmath171 } \frac{|du|}{d_g(0)+u } = \int_0^t \frac{du}{d_g(0)+u}\\ & = & \log(d_g(0)+t)-\log d_g(0 ) = \log \left ( 1+\frac{t}{d_g(0 ) } \right ) .
\end{aligned}\ ] ] for @xmath172 and @xmath173 clearly @xmath174 and we obtain @xmath175 the lower and the upper bounds of @xmath169 are equal to the corresponding bounds for @xmath140 and the expression for @xmath176 follows as in the proof of theorem [ thm : distance ratio estimates ] .
exactly same observation for @xmath169 near the boundary can be made that we made for the distance ratio distance .
theorem [ thm : quasihyperbolic estimates ] gives a lower bound and an upper bound can not be obtained .
for a domain @xmath2 and points @xmath141 we define the triangular ratio distance by @xmath179 geometrically the supremum is attained at a point @xmath180 such that it is either on line segment @xmath181 $ ] or if this is not possible , then @xmath180 is on the largest ellipsoid with focii @xmath162 and @xmath163 contained in @xmath7 .
the triangular ratio distance was introduced by hst in the 2000 s @xcite .
since @xmath182 $ ] we observe that @xmath177 is not hyperbolic type .
for @xmath141 we define @xmath183 to obtain a hyperbolic type distance . as in the case of the distance to the boundary function and the distance ratio distance
we can use the same extremal domains .
the domain @xmath151 for some @xmath122 and @xmath123 gives lower bound @xmath184 for @xmath129 .
the domain @xmath185 for any @xmath122 and @xmath127 gives upper bound @xmath186 for @xmath187 .
combining these estimates we obtain : [ thm : sigma distance ] let @xmath2 be a starlike domain and @xmath4 . then @xmath188 for @xmath129 and @xmath189 for @xmath187 .
moreover , @xmath190 . the bounds for @xmath191 follow from and .
by differentiation we obtain @xmath192 and thus @xmath193 .
[ rem : sigma remark ] theorem [ thm : sigma distance ] suggests that an alternative way to define a hyperbolic type distance by using the triangular ratio distance , could be @xmath194 . for this distance function @xmath195 .
for a domain @xmath2 and points @xmath141 we define the cassinian distance by @xmath197 the cassinian distance was introduced by ibragimov in the 2000 s @xcite . in the plane case the supremum is attained at a point @xmath180 that is on the largest cassinian oval with focii @xmath162 and @xmath163 contained in @xmath7 .
[ thm : cassinian distance ] let @xmath2 be a starlike domain and @xmath4 . then @xmath198 for @xmath129 and @xmath199 for @xmath187 .
moreover , @xmath200 .
the domain @xmath151 for some @xmath122 and @xmath123 gives lower bound @xmath198 for @xmath129 .
the domain @xmath185 for any @xmath122 and @xmath127 gives upper bound @xmath201 for @xmath187 .
differentiation gives @xmath202 and clearly @xmath200 .
let @xmath204 be a domain such that @xmath205 is not contained in a sphere in @xmath206 .
then for @xmath141 we define the apollonian distance ( * ? ? ?
* theorem 1.1 ) by @xmath207 the apollonian distance was introduced by barbilian in the 1930 s @xcite and reintroduced by beardon in the 1990 s in connection with the hyperbolic metric @xcite .
it is worth pointing out that we can write @xmath208 and in the case @xmath0 each quotient defines an apollonian circle .
taking the supremum means that we take the largest possible apollonian circles in @xmath209 .
[ lem : apollonian estimate ] let @xmath210 , @xmath38 and @xmath211 .
denote @xmath212 .
then @xmath213 and for @xmath214 and @xmath215 @xmath216 to simplify notation we may assume that @xmath217 and @xmath218 for @xmath219 .
now @xmath220 } \left\ { \frac{|w - re^{i \zeta}|}{r } \right\ } = \frac{|w - re^{i \pi}|}{r } = \frac{r+|w|}{r},\ ] ] because @xmath221 and the function @xmath222 is increasing for @xmath223 $ ]
. similarly we obtain @xmath224 } \left\ { \frac{r}{|w - re^{i \zeta}| } \right\ } = \frac{r}{|w - re^{i 0}| } = \frac{r}{r-|w|}.\ ] ] next we note that the function @xmath225 , @xmath226 , is decreasing on @xmath227 , because @xmath228 .
the function @xmath229 , @xmath226 , is decreasing on @xmath230 , because @xmath231 .
combining monotonicity of @xmath232 with gives @xmath233 and similarly monotonicity of @xmath234 with gives @xmath235 [ thm : f for alpha ] let @xmath236 be a starlike domain and @xmath4 .
then @xmath237 for @xmath129 and @xmath238 for @xmath136 .
moreover , we have @xmath239 and the upper bound for @xmath240 and the lower bound for @xmath241 are best possible . by we
can estimate @xmath242 by estimating @xmath243 and @xmath244 separately . by lemma [ lem : apollonian estimate ]
we obtain upper bounds @xmath245 and @xmath246 these inequalities give @xmath247 we can also use lemma [ lem : apollonian estimate ] for lower bound @xmath248 and we can trivially estimate @xmath249 together these two inequalities give us @xmath250 differentiation of gives @xmath251 and differentiation of gives @xmath252 . thus @xmath239 . finally , we give two example domains , which show that the upper bound for @xmath240 and the lower bound for @xmath241 are best possible .
the domain @xmath253 for some @xmath122 and @xmath127 shows that the upper bound is best possible .
now the suprema in the definition of the apollonian distance are obtained at @xmath254 and @xmath95 , and thus @xmath255 the domain @xmath131 defined in shows that the lower bound for @xmath241 is best possible .
we recall that @xmath256 for any @xmath122 and @xmath127 . for @xmath257 with @xmath258
we have @xmath259 and differentiation together with taking the limit gives @xmath260 as @xmath58 .
note that in the theorem [ thm : f for alpha ] also the upper bound for @xmath241 is best possible , because the upper bound for @xmath240 is best possible and both upper and lower bounds of @xmath240 tend to @xmath41 as @xmath58 .
let @xmath2 be a domain . for @xmath141
we define the seittenranta distance by @xmath262 the seittenranta distance was introduced by seittenranta in the 1990 s ( * ? ? ?
* theorem 3.3 ) . before estimating @xmath263 we find general upper and lower bound for @xmath264 .
[ lem : upper bound for delta ] let @xmath2 be a domain .
then for all @xmath141 we have @xmath265 by the euclidean triangle inequality we obtain @xmath266 and the assertion follows .
the following lower bound for @xmath264 is from ( * ? ? ?
* theorem 3.11 ) .
[ pro : lower bound for delta ] let @xmath10 be a domain and @xmath205 is not contained in a sphere in @xmath206 .
then for all @xmath267 we have @xmath268 we also need exact formulas for the seittenranta distance in two starlike domains . [
lem : formula for delta in special domains ] let @xmath131 be the domain defined in for some @xmath122 and define a starlike domain @xmath269 \(1 ) for @xmath270 we have @xmath271 \(2 ) for @xmath270 we have @xmath272 in both cases we have @xmath273 where @xmath274 .
the idea of our proof is to first show that for any @xmath275 the supremum over @xmath276 is attained at @xmath277 . using this property we can find supremum over @xmath278 .
\(1 ) we denote @xmath279 and @xmath280 . for @xmath281 the angle @xmath282 and by the law of cosines @xmath283
we fix @xmath278 and find the point @xmath89 , @xmath284 , which gives the minimum value for @xmath285 since @xmath286 we have @xmath287 and @xmath288 the denominator of @xmath289 is positive , because @xmath290 , @xmath291 and @xmath292 is equivalent to @xmath293 .
the numerator of @xmath289 equals zero , when @xmath294 and the function @xmath295 obtains its minimum either as @xmath296 or as @xmath297 .
we have @xmath298 and @xmath299 let us now consider @xmath300 .
if @xmath301 , then @xmath302 and the suprema in the definition of the seittenranta metric are obtained at @xmath303 , @xmath304 implying @xmath305 if @xmath306 , then @xmath307 and the suprema in the definition of the seittenranta metric are obtained at @xmath308 , @xmath309 implying @xmath305 \(2 ) it is easy to see that @xmath310 consists of @xmath311 and a line @xmath312 . if both @xmath313 , then the assertion follows from ( 1 ) .
we assume @xmath314 and denote @xmath315 . for any @xmath316 with @xmath284
we may assume that @xmath317 is maximal .
this immediately implies that @xmath318 .
we denote @xmath319 .
now for @xmath320 we have by the pythagorean theorem @xmath321 and we want to find maximum of @xmath322 we show that @xmath323 is a decreasing function .
differentiation gives @xmath324 and the numerator of @xmath325 equals zero whenever @xmath326 for @xmath327 if we choose minus in @xmath328 then clearly @xmath329 . if we choose plus in @xmath328 , then the equation @xmath330 has solution @xmath331 and @xmath332 .
now @xmath333 is either positive or negative for @xmath334 .
we estimate @xmath335 we have obtained @xmath329 and since @xmath336 we have @xmath337 for all @xmath338 .
now we are ready to consider @xmath339 . for any fixed @xmath340
, the largest value for @xmath341 is obtained for @xmath303 .
thus the suprema in the definition of the seittenranta metric are obtained at @xmath303 , @xmath342 implying @xmath343 where the second equality is obtained by the pythagorean theorem as @xmath344 implying @xmath345 and @xmath346 .
now we are ready to find bounds for @xmath263 and @xmath347 .
[ thm : seittenranta distance ] let @xmath2 be a starlike domain .
then @xmath348 for @xmath129 and @xmath349 for @xmath136
. moreover , @xmath350 and the bounds for @xmath347 are best possible .
lower bound for @xmath263 and @xmath347 follow from proposition [ pro : lower bound for delta ] and theorem [ thm : f for alpha ] . for the upper bound we use lemma [ lem : upper bound for delta ] to obtain @xmath351 for @xmath136 .
differentiation and taking the limit gives @xmath352 as @xmath58 , which proves the upper bound for @xmath347 .
finally , we show that the bounds for @xmath347 are best possible .
sharpness of the lower bound occurs in the domain @xmath131 defined in . by lemma [ lem : formula for delta in special domains ]
( 1 ) @xmath353 and by differentiation we obtain @xmath252 .
now as @xmath58 we get the @xmath354 .
sharpness of the upper bound occurs in the domain @xmath355 defined in lemma [ lem : formula for delta in special domains ] .
now by lemma [ lem : formula for delta in special domains ] ( 2 ) @xmath356 by differentiation and taking the limit we obtain @xmath357 as @xmath58 .
let @xmath2 be a domain . for @xmath141
we define the visual angle distance by @xmath360 the visual angle distance was introduced in the 2010 s in @xcite .
clearly @xmath361 $ ] and therefore we define for @xmath141 a hyperbolic type distance @xmath362 [ pro : increasing visual angle distance ] let @xmath363 be a line and @xmath364 .
then the function @xmath365 is strictly increasing for @xmath366 . now
@xmath367 and the point @xmath368 moves along the line @xmath369 .
it is easy to see that @xmath370 , @xmath371 and @xmath372 is strictly increasing as a function of @xmath373 . by the law of sines @xmath374 and
since @xmath375 is constant we obtain for @xmath376 @xmath377 now @xmath378 is a constant and @xmath372 is strictly increasing .
thus we need to show that for @xmath379 the function @xmath380 is strictly increasing in @xmath381 . by differentiation
we obtain @xmath382 and the assertion follows . in proposition
[ pro : increasing visual angle distance ] the line @xmath369 did not contain 0 . if @xmath383 with @xmath384 , then for any @xmath385 we have @xmath386,\\ \pi , & \textrm{if } 0 \in [ u , w].\\ \end{array } \right.\ ] ] [ thm : tau distance ] let @xmath2 be a starlike domain
. then @xmath387 for all @xmath129 and @xmath388 for all @xmath136 . moreover , the bounds for @xmath389 are best possible and @xmath390 . by
we note that @xmath391 for all @xmath129 and by proposition [ pro : increasing visual angle distance ] we obtain @xmath392 for all @xmath129 .
the lower bound @xmath41 is obtained in the domain @xmath131 introduced in .
the upper bound for @xmath389 is attained in @xmath393 . by
* ( 3.3 ) ) , @xmath394 and thus @xmath395 by differentiation and taking the limit we obtain @xmath396 as @xmath58 . [ rem : tau remark ] as in remark [ rem : sigma remark ] , theorem [ thm : tau distance ] suggests that we could define @xmath397 . for this distance function @xmath398 .
in this final section we prove our main result and consider the estimates in domains other than starlike .
until now we have considered starlike domains and found estimates for the function @xmath22 . moreover , the upper bound for @xmath22 is obtained only when @xmath83 is close to the origin ( @xmath136 ) , whereas the lower bound is valid also for large values ( @xmath129 ) .
let now @xmath2 be any domain with @xmath21 and @xmath4 .
our results hold also in @xmath7 , when @xmath399 . in other words ,
the lower and upper bounds for @xmath22 are true for @xmath136 and the results for @xmath400 are also true .
we initially choose @xmath41 to simplify notation .
it could be any point in @xmath7 .
the result for @xmath139 follows from theorem [ thm : distance ratio estimates ] , for @xmath159 from theorem [ thm : quasihyperbolic estimates ] , for @xmath401 form theorem [ thm : sigma distance ] and remark [ rem : sigma remark ] , and for @xmath196 from [ thm : cassinian distance ] .
in theorem [ thm : main theorem ] we assumed @xmath4 .
this is not needed , if we only consider @xmath295 close to the origin .
we can define for any @xmath403 @xmath404 as an application of theorem [ thm : main theorem ] we obtain if we restrict to starlike john domains , then we can easily get lower bound for @xmath22 .
a domain @xmath7 is @xmath408-john domain , @xmath409 , if there is a distinguished point @xmath410 such that any @xmath411 can be connected to @xmath412 by a rectifiable curve @xmath413 \to g$ ] , which is parametrised by arclength and with @xmath414 , @xmath415 and @xmath416 for every @xmath417 $ ] .
we can choose @xmath418 and it is easy to see that for example @xmath419 .
we consider next monotonicity of the function @xmath22 . by
* theorem 4.8 ) the metric balls @xmath420 are starlike whenever @xmath7 is starlike .
this implies that in starlike domains @xmath140 is an increasing function .
same is also true for the quasihyperbolic distance ( * ? ? ?
* theorem 2.10 ) , the apollonian distance ( * ? ? ?
* theorem 3.5 ) , the triangular ratio distance ( * ? ? ? * theorem 1.2 ) and the visual angle distance ( by definition ) . for the seittenranta distance
this is not know ( * ? ? ?
* open problem 4.10 ( 1 ) ) .
monotonicity is not true for the function @xmath77 .
however , in convex domains @xmath77 behaves well : there exists a point @xmath421 such that @xmath77 is increasing on @xmath422 $ ] and decreasing on @xmath423 .
the next example shows that @xmath77 has not this property in starlike domains .
[ exm : comb ] we show that @xmath77 misbehaves in the domain @xmath424 \right ) , \quad a_l = ( 1 - 2^{-l})e_1 + 2^{-(l+1 ) } e_2,\ ] ] for @xmath104 . @xmath7 is not strictly starlike , but by replacing line segments @xmath425 $ ] with angular domains @xmath426 , where @xmath427 is small enough , we could construct a strictly starlike domain with the same effect .
we stick to the line segment version to make the computation easier to follow .
we demonstrate that on each interval @xmath428 $ ] the function @xmath77 obtains its maximum in @xmath429 and @xmath430 .
this means that @xmath77 has infinitely many local maxima and minima . of example
[ exm : comb ] .
right : function @xmath77 in example [ exm : comb].,title="fig:",scaledwidth=30.0% ] of example [ exm : comb ] .
right : function @xmath77 in example [ exm : comb].,title="fig:",scaledwidth=50.0% ] let us fix @xmath431 . now @xmath432 and thus @xmath433 .
we show that for @xmath434 , @xmath435 by construction of @xmath7 it is clear that @xmath436),|b_le_1-a_{l+1}| \}.\ ] ] by geometry we have @xmath437 ) \ge 2^{-(l+1 ) } = g(1 - 2^{-l}).\ ] ] we calculate @xmath438 and @xmath439 implying @xmath440 inequalities and imply and thus @xmath77 misbehaves ( it has infinitely many local maxima and minima ) . @xmath59 | we study the growth of hyperbolic type distances in starlike domains .
we derive estimates for various hyperbolic type distances and consider the asymptotic sharpness of the estimates .
@equation=@theorem keywords : hyperbolic type distance , starlike domain msc2010 : 30f45 , 51m10 , 30c65 |
one of the many ways to determine @xmath4 is through the measurement of jet production in dis since the production cross section in leading order of @xmath7 jets ( not counting the proton remnant ) is proportional to @xmath8 . at hera
this measurement is particularly interesting since the hadronic final state is opened up in pseudo - rapidity .
the measurements of jet production cross - section presented here have many points in common which facilitates comparison of the extracted values of @xmath4 .
in particular , jets are measured throughout using the longitudinally invariant @xmath9 algorithm in the breit - frame of reference , where in the naive quark parton model the struck quark and the virtual boson collide head - on .
hence , substantial transverse energy @xmath3 in the breit frame stems necessarily from qcd radiation and provides a natural energy scale for the determination of @xmath4 .
in the h1 ( zeus ) inclusive - jet analysis , all jets in dis events with @xmath10 @xmath11(@xmath12 @xmath11 ) are counted if their transverse energy @xmath3 exceeds 7 @xmath1(8 @xmath1 ) .
furthermore , the pseudo - rapidity of each jet in the laboratory frame ( breit frame ) is required to lie between -1.0 and 2.5 ( -2.0 and 1.5 ) in the h1 ( zeus ) analysis .
the inclusive jet cross - section is measured single- and double - differentially in @xmath2 and in @xmath3 , and also differentially in the pseudo - rapidity @xmath13 of the jet in the breit frame ( zeus analysis ) @xcite . for the measurement of dijet and trijet cross - sections
a lower minimal transverse energy of @xmath14 @xmath1 is required in both analyses , and jets in the same pseudo - rapidity interval as for the h1 inclusive jet analysis are considered . in order to ensure infrared - safeness of the qcd predictions for multi - jet cross - sections , the invariant mass of the 2 or 3 jets is required to be above 25 @xmath1 .
the h1 and zeus analyses differ in the kinematic range of the dis events : @xmath15 @xmath11 for h1 and @xmath16 @xmath11 for zeus .
the multi - jet cross - sections are measured differentially in @xmath2 ( h1 ) and @xmath2 , @xmath3 , and @xmath13(zeus ) , and the ratio @xmath5 of the trijet and the dijet cross - sections is measured as a function of @xmath2 ( fig .
[ fig : hmuljetr32 ] ) @xcite . in all analyses ,
the data are corrected bin - by - bin for detector effects and qed radiation using fully simulated monte carlo events , using leading - order ( lo ) event generators .
the dependence of the detector correction on the different models for parton showering leading - log parton showers ( as implemented in rapgap and lepto ) and the colour - dipole model ( as implemented in ariadne ) results in a systematic uncertainty of e.g. @xmath17 for the inclusive jet cross - section ( zeus ) .
a systematic error of comparable size stems from the uncertainty of the absolute energy scale of the hadronic calorimeters .
it is evaluated by varying the calorimeter energy scale by @xmath18 for all events and fully resconstructing these ( h1 ) or simply by varying the transverse energy of the jets @xmath3 by @xmath1913% depending on @xmath3 ( zeus ) . typically , this results in an uncertainty of @xmath20 on the inclusive - jet cross - section as measured by zeus and of 1% ( 4% ) on the dijet ( trijet ) cross - section as measured by h1 . other experimental uncertainties , such as systematic errors on the measurement of luminosity or on the reconstruction of the scattered electron , typically contribute to the overall experimental error on a level of 12 per - cent .
the measured jet cross - sections are compared to nlo pqcd predictions calculated using the programs disent ( @xmath21 ) and nlojet++ ( @xmath22 ) .
however , these codes do not include parton - showering and fragmentation , and deliver cross - section values for parton final states only . in order to compare these calculations to measured data
, a hadronization correction factor is applied to the pqcd predictions as the ratio of the cross - sections for hadron and parton final states as calculated by the above mentioned event generators .
this procedure relies on the assumption that the hadronisation correction is not significantly affected by the difference between lo and nlo parton final states .
the theoretical error on this hadronization correction is estimated by comparing the correction coefficients obtained with the two different models for parton showering .
perturbative qcd calculations at fixed order unavoidably exhibit an explicit dependence on the renormalisation scale @xmath23reflecting the absence of higher orders in the perturbative expansion .
it has been shown in these analyses that the qcd predictions for the discussed jet observables are not very sensitive to the particular choice for @xmath23 and the factorization scale @xmath24 among the possible scales of each event:@xmath2 , @xmath3 , or an arithmetic combination of both , but rather to the absolute value of @xmath23(cf .
[ fig : zincxsctq2 ] ) . in order to assess the theoretical uncertainty on the jet cross - sections and @xmath4 due to missing higher orders
, the renormalization scale @xmath23 is varied .
it has to be pointed out that the particular choice of varying @xmath23 up and down factor of 2 is a mere convention - that is followed in the presented analyses - but has no theoretical motivation .
for the inclusive jet cross - section ( e.g. zeus ) , the uncertainty resulting from the renormalization scale dependence ( @xmath25 ) dominates over the other contributions , e.g parametrization of the proton pdf ( @xmath26 ) , to the overall error of the pqcd prediction .
this is also true for the multi - jet cross - sections ( e.g. h1 ) : predicted cross - section for dijets ( trijets ) vary by as much as @xmath27 ( @xmath28 ) with @xmath23 whereas the uncertainty from the hadronization correction is @xmath29 ( @xmath30 ) .
the lower part of figure [ fig : zincxsctq2 ] demonstrates the good agreement of the nlo pqcd prediction with the measured inclusive jet cross - section ( zeus ) as it drops over five orders of magnitude on a wide range in @xmath2 .
theory and experiment are compatible within experimental error which are generally smaller than the estimated theoretical uncertainty .
the good agreement holds for the multi - jet rates and their ratio as well , as can be seen on fig .
[ fig : hmuljetr32 ] except at very high @xmath2 where electro - weak effects are not accounted for in the nlo pqcd programs .
the strong coupling constant is determined by fitting the pqcd predictions as a function of @xmath4to the measured values of the jet observables , i.e. the cross - section value in a particular bin in @xmath2 or @xmath3 or the ratio @xmath5 . in practice ,
the pqcd prediction for each observable is parametrised as a second order polynomial in @xmath4 ( with vanishing constant term ) .
figure [ fig : hincjetas ] shows the running @xmath31 as extracted from the differential inclusive jet cross - section in @xmath3 ( h1 ) , and the corresponding value evolved to the z@xmath32-boson mass . by simultaneously fitting the observables over the range where they are well described by pqcd , and also taking partial correllations of systematic experimental and theory errors into account ,
an @xmath33-measurement is extracted for each of the four analyses : for the same observable , both experiments obtain almost identical values of @xmath33 with slightly lower values and higher experimental error for a measurement using @xmath5 .
the @xmath33 values obtained from different observables , agree well with each other , with an average of previous hera measurements from inclusive dis and jet data @xcite ( in which @xcite is used ) , and also with a world average of @xmath34 as determined by @xcite at a level of experimental precision competitive with the most precise measurements for jets .
it should be noted that the error from the pqcd calculation is roughly three times higher than the experimental uncertainty for the most precise of these four measurements @xcite . although the latter could probably be reduced by increased statistics and better understanding of systematical errors , future progress on the overall precision for @xmath4 hinges on an improved estimation of contributions beyond nlo to the pqcd predictions .
99 h1 collaboration , contributed paper no .
629 at this conference + zeus collaboration , contributed paper no .
375 at this conference + h1 collaboration , contributed paper no .
625 at this conference + s. chekanov et al . , _
european physical journal _ * c44 * ( 2005 ) 183 - 193 [ hep - ex/0502007 ] + c. glasman , _ xiii int .
workshop on deep inelastic scattering , madison _
( 2005 ) [ hep - ex/0506035 ] + s. bethke ,
_ nucl.phys.proc.suppl .
( 2004 ) 345 - 352 [ hep - ex/0407021 ] + | the inclusive jet , dijet and trijet production cross - sections in deep - inelastic scattering ( dis ) at @xmath0 @xmath1 have been measured at the electron - proton collider hera by the h1 and zeus collaborations using data taken in 19982000 .
the jet cross - sections have been measured differentially in four - momentum transfer squared @xmath2 and jet transverse energy in the breit - frame @xmath3 with a typical precision of 510% limited by systematic uncertainties such as hadronization corrections and hadronic energy scale .
all jet observables are well described by perturbative qcd ( pqcd ) predictions at next - to - leading order ( nlo ) within the estimated accuracy of these calculations which is limited by the absence higher orders , and which is in general inferior to the experimental precision .
the values for the coupling constant @xmath4 of the strong interaction as determined by fits of pqcd predictions to the inclusive jet cross - section and the trijet - to - dijet production ratio @xmath5 are consistent for both experiments and both observables , and also are in excellent agreement with the world average . combining the individual measurements , a common value of @xmath6 is obtained . |
the one - dimensional ( 1d ) tight - binding model of non - interacting electrons with random nearest neighbor hopping has been extensively studied since dyson s exact solution to density of states of one - dimensional random harmonic oscillators @xcite , which can be mapped to the random nearest neighbor hopping model .
mertsching gave a node counting scheme to study the density of states in a similar model@xcite .
since the early 80s , supersymmetry methods have been used to study such problems with randomness@xcite .
all these studies show that there is a singularity in the one - electron density of states @xmath1 at the band center of the form @xmath2 .
concurrently , as a consequence of the thouless theorem @xcite , there is a logarithmic divergence of the localization length @xmath3 @xcite .
the behavior found there suggests that the strong disorder fixed point , which determines the functional form of low temperature magnetic susceptibility in the spin problem , ( for nearest neighbor hopping this is related to the density of states singularity in the fermionic model ) survives for power law exponents @xmath4 above a critical value , which they determined to be 2 ( within their numerical accuracy of 5 - 10 per cent ) . using the jordan wigner transformation on the 1d xy spin chain with long range couplings , while yielding a particle - hole symmetric fermion hamiltonian , does not give a pure noninteracting fermion problem with power law hopping [ terms such as @xmath5 appear as well ] .
while the _ exact _ connection between the fermionic and spin models is lost when further neighbor hoppings are added to the hamiltonian , we can nevertheless expect some similarities between the two models - xy random spin chain and free fermions with 1d random hopping - when the two have exchanges / hoppings with a power - law fall off , at least for large power - law exponents .
the result for the long range spin model also motivates paying close attention to our model s behavior when @xmath4 is in the vicinity of 2 .
the outline of our paper is as follows . in the next section ,
we define the model hamiltonian .
the following section discusses the methods and relevant formalisms used in computation .
we then present the results for the density of states , localization length of eigenstates , and the spin - spin correlation function for the equivalent magnetic model , obtained from various numerical calculations .
we find that the density of states retains the singularity at the band center , but this singularity is gradually reduced by the long range hopping .
the eigenstates are not well described as exponentially localized states , but have power - law - like tails determined by @xmath4 .
finally , the spin - spin correlation function also shows power - law decay where @xmath4 is not too small .
the model we have studied is built on a 1d bipartite lattice , consisting of odd and even sites along the chain .
the hamiltonian contains fermionic hopping terms between the two sublattices consisting of the odd and even sites , respectively , while hopping within the same sublattice is forbidden .
thus we have , @xmath6 where @xmath7 and @xmath8 are summed over members of the odd and even sublattices respectively .
this bipartite property preserves the particle - antiparticle symmetry and leads to interesting properties in the density of states and localization length .
the hopping parameters @xmath9 are random with zero average , defined in terms of the distance @xmath10 and a gaussian random variable , as described below .
two models with power law fall - off of @xmath11 have been studied .
model a allows hopping between every pair formed by picking one site from each of the two sublattices , while @xmath12 decreases with distance as a power law for each pair @xmath13 .
specifically , we take @xmath14 where @xmath15 is a gaussian random number with distribution @xmath16 in model b , @xmath12 is constant , but the probability for a hopping term to be present decreases with the hopping distance as a power - law .
thus @xmath17 where @xmath15 is a gaussian random variable as in model a , and @xmath18 is chosen randomly to be either 1 ( with probability @xmath19 ) or 0 [ with probability @xmath20 ) ] .
we expect these two models to exhibit some similarities .
we have used a = 0.2 throughout this paper . in both models , therefore , we have only one parameter @xmath4 which characterizes the model .
a small value of @xmath4 produces longer range hopping .
the hamiltonian of this model is an @xmath21 matrix for a system of n sites , with elements @xmath22 when @xmath10 is a multiple of 2 . because of this bipartite property
, the schrdinger equation can be written as two coupled equations when number of sites @xmath23 is even .
let @xmath24 be the wavefunction on the sublattice consisting of the odd numbered sites , and @xmath25 that on the sublattice consisting of even sites .
if the matrix @xmath26 represents the hopping terms from even sites to odd sites , then @xmath27 will consist of the hopping terms from odd sites to even sites .
the schrdinger equation can then be written as two equations : @xmath28 these equations lead to @xmath29 the above equations reduce the size of the matrix of the @xmath30 hamiltonian to @xmath31 .
further , from the above equations , it is clear that the eigenvalues come in pairs @xmath32 , and their eigenvectors are related by @xmath33 a direct consequence of the above is that the density of states is symmetric for each realization of the disorder .
by exploiting the particle - antiparticle symmetry , we gain efficiency in our numerical calculation . +
the density of states can be obtained by directly diagonalizing large matrices . in this approach ,
the difficulty is that we have to use large matrices which do not have significant finite size effects .
since we re interested in the property of the singularity at the band center , which by particle - hole symmetry , lies at @xmath34 , we need to get enough statistics near the band center .
we identify finite size effects by comparing the results of different sizes .
+ the other approach to calculating the density of states is a recursive method .
first , the dense matrix is cut off at a range large enough , so that the remaining part is an acceptable approximation for the original matrix of arbitary size .
[ for finite range hopping , this method gives no approximation at this stage , but for long - range ( power law hopping ) this is a different cutoff scheme , with a somewhat different finite size effect ] .
then the matrix @xmath35 is transformed into diagonal form by a similarity transformation , which rotates certain columns and rows to eliminate off - diagonal terms .
the remaining diagonal elements are not eigenvalues , but they retain the signature of the matrix , _
i.e. , _ a positive element implies an eigenvalue satisfying @xmath36 , a negative element corresponds to an eigenvalue satisfying @xmath37 . by counting the number of positive or negative diagonal elements
we can obtain the integrated density of states .
this process can be continued for arbitary length with no finite - size effect as in direct diagonalization , until the statistical error of the integrated density of states is smaller than our requirement . for nearest neighbor model ,
the recursion equation for diagonal element is : @xmath38 where @xmath39 is the only off - diagonal term in the hamiltonian , @xmath40 is the remaining diagonal element after the transformation .
suppose @xmath40 has a distribution @xmath41 as @xmath42 , then it is given by the integral equation @xmath43 where @xmath44 is the distribution function of @xmath45 .
although this equation is similar to dyson s approach @xcite , it has not been solved analytically due to the peculiar argument of the @xmath46 function .
the numerical approach is just to generate a sequence of @xmath40 and calculate its distribution . + to take care of the finite size effect introduced by the cut - off , we compare results using different cut offs .
such a procedure allows us to ascertain the range of energies for which the integrated density of states has converged : lower the energy , longer the cut - off required ( the cutoff required varies roughly logarithmically with energy ) . for a fixed cutoff , the energy variation of the integrated density of states obtained by this method mimics that of the nearest neighbor _ i.e. , _ finite range model - this sets the lower bound of energies close to the band center for which this method is applicable for that cutoff .
+ once we have obtained the eigenfunctions by direct diagonalization , we can calculate the correlation function of the corresponding spin model . the jordan - wigner transformation
@xcite transforms the xy - spin chain in zero external field to a half - filled band of fermions , because the following identity @xmath47 the ground state fills the @xmath48 states with negative energy which we can get by diagonalizing the hamiltonian .
we can further calculate the spin correlation function on the ground state .
we compute the zz part of the spin - spin correlation function defined by @xmath49 @xmath50 denotes the expectation value in the ground state , while the bar on top denotes ensemble average over the disorder . with @xmath51 , and @xmath52 ,
the correlation function can be written as @xmath53 where we have used the fact @xmath54 due to half filling .
the jordan - wigner transformation contains a phase factor due to the anticommutation relation of fermion operators on different sites , but the phase factor coming from different fermion operators cancels each other in our expression of @xmath55 . in the random singlet phase@xcite appropriate for describing the nearest neighbor only model in the spin operator language ,
all these correlation functions have the same dependence on distance .
the four fermion term can be expanded into a hartree term and a fock term .
@xmath56 here the sum is over all occupied states , i.e. all n/2 states with negative energy .
the hartree term can be evaluated directly to be @xmath57 because the wavefunctions are orthonormal using equations ( [ e5 ] ) and ( [ e6 ] ) .
therefore the correlation function is given by the fock term only .
further , if @xmath7 and @xmath8 are on the same sublattice , we see from equations ( [ e2 ] ) and ( [ e3 ] ) that @xmath58 and @xmath59 ( n is among @xmath48 occupied states ) are both eigenvectors of either @xmath60 or @xmath61 , and consequently orthogonal to each other . thus , when @xmath7 and @xmath8 are on the same sublattice , the correlation function is exactly zero . in the numerical results presented below
, we only display the spin correlation functions between 2 sublattices , and @xmath55 is replaced by @xmath62 where @xmath63 .
@xmath64 this expression can be evaluated directly , but since the computing time for evaluating c(x ) is even more than that required for diagonalization , the system sizes we use are smaller than for diagonalization .
the data presented for c(x ) are all obtained from systems of 256 sites .
however , we are able to discern the behavior reasonably well from the data for @xmath62 subject to this limitation . +
our results on model a show that the density of states remains singular at the band center when the long range hoppings are present .
figure [ fig:1 ] shows the density of states @xmath1 obtained by diagonalization as a function of energy @xmath65 on a double logarithmic plot for both the nearest neighbor model , and for the long range model for different values of the power law exponent @xmath4 .
as can be seen , the singularity at the band center ( @xmath66 ) persists at least for large @xmath4 , though its magnitude clearly decreases .
the inset to fig [ fig:1 ] compares the data for the nearest neighbor model along with the data for the lowest power law ( @xmath67 ) on a linear scale , which gives a clearer idea of the extent of this decrease . for quantitative purposes , it is better to plot the same data as @xmath68 vs.@xmath69 .
on such a plot , shown in fig [ fig:2 ] , the nearest neighbor model is supposed to lie on a straight line , which it clearly does .
further , the curves show little deviation from the nearest neighbor model as long as @xmath70 . the deviation for smaller @xmath4
are consistent with the singularity being gradually weakened as @xmath4 decreases ; if we fit the data with @xmath71 then when @xmath72 , the best fit @xmath73 is about 5 . as @xmath4 becomes lower than 1 , direct diagonalization is almost incapable of revealing any detail of the singularity : we only see a large value at the band center , and the density of states approaches the wigner semicircle@xcite .
nevertheless , the thermodynamic limit remains well defined upto @xmath74 , below which the bandwidth starts to increase with system size , i.e. we need to scale the hopping magnitude with system size to have a sensible thermodynamic limit .
this critical value of @xmath4 can be predicted by writing the model in path - integral form , using either replica technique or supersymmetry , and averaging over the random variables .
such an approach gives a four fermion term proportional to @xmath75 , where @xmath7 and @xmath8 are site indices , therefore when @xmath4 is less than 0.5 , the sum will diverge .
+ we have diagonalized several different sizes of samples to ensure that the data shown are not corrupted by finite size effects .
[ fig:3 ] shows an example of a finite size effect , which appears as a size - dependent rounding of density of states at low energy .
all the data plotted correspond to n = 1024 sites unless stated otherwise , and do not suffer significant finite size effects .
the results of model b are very similar to model a , except that the smallest @xmath4 for a proper thermodynamic limit to exist is 1 .
+ fig [ fig:4 ] shows our results for the density of states obtained using the recursive method . here
, one obtains the integrated density of states @xmath76 ( from @xmath77 to @xmath65 ) . for the nearest neighbor model , the exact asymptotic form is given by : @xmath78 fig [ fig:4 ] plots the inverse square root of @xmath79 versus @xmath80 ; for the nearest neighbor model , the expected straight line behavior is seen . for power law
hopping , the data shows measurable curvature certainly for @xmath81 . for larger @xmath4
it is difficult to see whether the data suggest curvature , or simply a changing slope with decreasing @xmath4 . while it is tempting to fit these curves with a form like @xmath82 , which will lead to an singularity in the density of states like @xmath83 , the data are better fit with several @xmath84 .
this suggests that corrections to the asymptotic form may be important for power law hopping .
+ in figures [ fig:5],[fig:6 ] , we show the accuracy of this recursive method .
figure [ fig:5 ] shows a direct comparison of the two methods.the recursive method agrees with the diagonalization results before the finite size effect sets in .
figure [ fig:6 ] shows the convergence of the recursive method as the range of cut - off is increased .
( below a certain energy , which corresponds to distances beyond the cutoff length , the recursive method will behave like nearest model , since the wavefunction spreads out of the range of hopping ) .
the density of states allows us calculate the specific heat and spin susceptibility for this this model of non - interacting fermions . using the standard formulae@xcite
, we can see that the specific heat prefactor ( @xmath85 , where @xmath86 ) and susceptibility ( @xmath87 ) at low temperature are singular at the center of the band .
thus , in the vicinity of the band center the zero temperature susceptibility is the form : @xmath88 if the singularity in density of states is latexmath:[$ 1 \over \epsilon @xmath85 .
the nearest neighbor model is known to have a sigularity in the localization length ( @xmath90 ) at center of the band .
it s asymptotic form is @xmath91 this singularity can be deduced from the singularity of density of states using the thouless theorem@xcite . however , in the longe range hopping model , the theorem does not apply , and our numerical calculation suggests that the behavior of the two models is rather different . + for finite energy , all states are found to decay from a central maximum in both models , and the decay becomes weaker as the band center is approached , as in the nearest neighbor model . however , the behavior with distance is rather different for the two models . * in the case of model a which has genuine long range ( power law ) behavior of the hopping parameter @xmath9 , the tail of the wavefunctions actually decay in a power - law manner , @xmath92 .
this form is obviously determined by the power - law long range hopping term .
if we apply the usual method of looking at the asymptotic behavior to determine localization length , the localization length is infinite for any power law exponent ! * in the case of model b , the wavefunction is found to be decaying exponentially at long distances , like the nearest neighbor model , and the localization length can be obtained by several methods , which agree with the theoretical prediction .
fig [ fig:7 ] shows a log - log plot of the averaged probability density ( wavefunction amplitude squared @xmath93 ) as a function of the distance from the center of the wavefunction , averaged over typically 512 states , as a function of energy away from the band center for model a. at long distances , the behavior is clearly linear on this double logarithmic plot , implying a power law decay at long distances .
fitting the data shows clearly that the decay is related to the power law behavior of @xmath9 - the tail of @xmath93 decays as @xmath94 .
this tells us that we can not use the localization length , as defined by thouless for an exponentially decaying state here .
however , we may usefully define moments of the wavefunction to compute , for example , inverse participation ratios@xcite . + the problem of defining the localization length in model a is not shared by model b , where we find that the wavefunctions always decay exponentially . to show this difference , in figures [ fig:8 ] to [ fig:11 ]
we have plotted typical wavefunctions for model a and model b ( _ on a log - linear plot of probability density versus distance _ ) near the center and in the tail of the band .
the difference between the two models near the center of the band is not obvious due to the large fluctuations , but in the tail of band , they look clearly different - model b shows straight exponential decay ( see fig .
[ fig:10 ] ) down to 50 orders of magnitude for @xmath95 , while model a ( see fig .
[ fig:9 ] ) shows clear upwards concave curvature in the tail , characteristic of a slower decaying function ( like a power law ) at long distances .
we now present results for the spin - spin correlation function for the associated model in terms of the spin operators obtained by a jordan wigner transformation .
the motivation is to compare the results with those obtained for the long range random antiferromagnetic xy spin chain@xcite .
we reiterate that because of long range hopping , our model contains terms in addition to those in the pure power law xy model studied by houck and bhatt ; however , because of the existance of the same long range two - spin interactions in both studies , there may be several points in common .
figure [ fig:12 ] shows the average correlation function @xmath62 obtained by averaging 256 samples of model a each having 256 sites plotted on a double logarithmic plot .
all the values of @xmath62 are negative ( corresponding to antiferromagnetic correlations ) .
our results suggest that there are three regions of distinct behavior , which can be summarized as given below : * for fast power law decays ( _ i.e. , _
exponents @xmath96 ) , the long distance behavior of the correlation function remains unchanged from the nearest neighbor model .
thus in figure [ fig:12 ] , the curves are parallel to each other at large @xmath97 values , consistent with the result for the nearest neighbor model , for which a slope of 2 is predicted@xcite .
( for example , a best fit of @xmath62 for x within the interval @xmath98 $ ] @xmath99 yields a slope of 1.97 ) . * when @xmath4 gets below 2 , the slope begins to change . for @xmath4 close to but less than 2 , the slope appears to be given by @xmath4 itself , implying @xmath100 .
( a least squares fit to @xmath62 of the form @xmath101 , gives @xmath102 for @xmath103 , @xmath104 for @xmath105 and @xmath106 for @xmath107 ) .
our numerically exact results thus show that the model s magnetic counterpart changes its low energy behavior different behavior for power law exponents @xmath4 below 2 .
this is precisely the value around which houck and bhatt@xcite found that the perturbative real - space rg procedure appears to break down , perhaps signalling a change of phase .
* below @xmath108 , the deviations of @xmath62 from the nearest neighbor model become rather significant . to exhibit this more clearly , we show a linear plot in figure [ fig:13 ] .
c(x ) seems to decay very slowly , and our data are consistent with it approaching a finite ( negative ) value , implying antiferromagnetic order .
( for example , the curve for @xmath109 is well fitted by @xmath110 , where @xmath111 is 1/2 within numerical accuracy ) .
we caution , however , that for such low power law decays , finite size effects can be large , and more detailed calculations with larger system sizes and more samples is necessary before this result can be stated with certainty from numerical studies . in summary ,
the numerical results on the spin correlation function from the long range fermion model ( which we can solve numerically exactly ) supports the earlier observations on the xy chain with random long range couplings using perturbative numerical rg methods that the random singlet phase is unstable for power law couplings with exponents less than 2 .
however , unlike the numerical rg study , which sees this as a breakdown of the rg scheme , we are able to go into the new phase , which appears to be characterized by continuously varying exponents , like a critical phase .
we also find evidence for a possible transition to long range order at still smaller @xmath4 .
it should , however , be borne in mind that these observations , are from numerical calculations in _ finite _ systems , and subject to statistical errors due to finite sampling of the quenched random variable .
in this paper , we have presented results of a numerical study of a one dimensional lattice model of non - interacting fermions with random long range ( power - law ) hopping , which maintains particle - hole symmetry of the nearest neighbor model by allowing hopping only between even and odd sites ( _ i.e. , _ no hopping allowed between odd sites , or between even sites ) .
we have studied two models - one with genuine power - law hopping , and another with long range hopping with a power law fall - off of the probability of such a hopping to be present .
the results on density of states , localization and spin correlation function of the two models have been presented and analysed . for the density of states , we observe that the singularity at the center of the band , present for the nearest neighbor model , is weakened by long range hopping ( model a ) .
the change is gradual , and at least for power - law exponents @xmath4 greater than 3 , is consistent with a change in the prefactor of the singularity . for @xmath4 less than 3 ,
though , the numerical data for the desity of states in the range available appear to fit better with a somewhat different power of the logarithm of the energy . beyond @xmath112 ,
the data are consistent with no singularity being present , until @xmath113 , when the thermodynamic limit becomes ill - defined .
similar results are seen in model b , except that the thermodynamic limit becomes ill - defined at @xmath112 . + the two models exhibit rather different behavior of the electronic wavefunctions .
model b is conventional , in that its wavefunctions are exponentially localized , just as the eigenstates of nearest neighbor model . in model a , however , the wavefunctions are actually localized in a power - law manner rather than exponential . consequently
, the usual definition of localization length in terms of the logarithm of the long distance behavior of the wavefunction is invalid ; however , several inverse participation ratios can still be defined . by transforming the fermion model back to a spin model using jordan - wigner transformation , we calculated the spin correlation function in the ground state .
based on the data , three different phases may be possible - ( 1 ) the random singlet phase , which seems to be stable for power law exponents down to @xmath114 ; ( 2 ) a critical type phase with a continuously varying exponent of the power - law characterizing the spin - spin correlation function between @xmath114 and @xmath115 ; and ( 3 ) a possibly long - range ordered phase for @xmath116 .
this research was supported by nsf dmr-9809483 . | the one - dimensional ( 1d ) tight binding model with random nearest neighbor hopping is known to have a singularity of the density of states and of the localization length at the band center .
we study numerically the effects of random long range ( power - law ) hopping with an ensemble averaged magnitude @xmath0 becomes smaller than 2 . |
the intriguing phenomenon , the strong collinearity of cores in emulsion experiments @xcite , closely related to coplanar scattering of secondary particles in the interaction , has been observed long time ago .
so far there is no simple satisfactory explanation of these cosmic ray observations in spite of numerous attempts to find it ( see , for instance , @xcite and references therein ) . among them ,
the jet - like mechanism @xcite looks very attractive and gives the natural explanation of alignment of three spots along the straight line which results from momentum conservation in a simple parton picture of scattering .
besides , the strong momentum correlation of particles inside a jet and correlation between jet axes due to singularity of qcd matrix elements allow us to suggest the high degree of alignment for more than 3 spots .
this has been already demonstrated for the four cores in @xcite but using a simplified picture of hadronization .
with increasing energy of colliding hadrons ( nuclei ) hard and semi - hard jets begin to play an important role due to growth of their production cross sections .
thus the jet activity is likely to be a feature of all events above certain threshold collision energy .
one of the manifestations of this activity and the strict momentum ordering inside a hard enough jet can be the observed strong collinearity of spots in emulsion experiments .
the main purpose of the present paper is just to trace this relation in detail . in sect .
2 we formulate the problem on the whole .
section 3 describes the results of numerical simulation made under conditions close to emulsion experiments in the framework of pythia @xcite , and some discussion .
a summary can be found in sect .
in the pamir experiment the observed events ( @xmath4-hadron families with the alignment ) are produced , mostly , by a proton with energy @xmath5 tev interacting at a height of several hundred metres to several kilometres in the atmosphere above the chamber @xcite .
the collision products are observed within a radial distance up to several centimetres in the emulsion where the spot separation is of the order of 1 mm .
one can estimate the typical transverse momentum in the events under consideration using the ratio ( see also ( [ position ] ) ) : @xmath6 where @xmath7 is the energy deposition in the spot , @xmath8 is its spacing in the @xmath9-ray film , @xmath10 is the height of interaction .
so @xmath11 is of the order of 10 gev for @xmath12 mm , @xmath13 km , @xmath14 tev .
the particles with such transverse momenta can be typically initiated by a jet with @xmath15 gev or larger , since the most probable value of a fraction of jet energy carried by leading particles is @xmath16 @xcite .
such energetic jets are already enough collimated : their effective angular cone size @xmath17 due to the strict ordering of transverse and longitudinal particle momenta in the leading logarithm approximation of perturbative qcd @xcite .
qcd teaches " us also that this effective size decreases with the growth of the jet hardness " ( transverse momentum ) as @xmath18 where @xmath19 is the dimensional qcd parameter .
the main conjecture is that the particle distribution from the hard enough jets in the @xmath9-ray film plane can lead to the alignment of the emulsion spots due to the strong collimation of such particles and dynamical correlation between jet axis directions .
let us consider the kinematics in detail .
for our analysis it is convenient to parametrize 4-momentum of each produced particle @xmath20 under consideration with its transverse momentum @xmath21 ( relative to the collision axis @xmath22 ) , azimuthal angle @xmath23 and rapidity @xmath24 in the center - of - mass system : @xmath25.\ ] ] in this case the transformation from the center - of - mass system to the laboratory one reduces to a simple rapidity shift : @xmath26 , where @xmath27 , @xmath28 are the rapidities of particle @xmath20 and the center - of - mass system correspondingly in the laboratory reference frame . if we neglect the further interactions of particles propagating through the atmosphere ( this gives the maximum estimation of the alignment effect ) , then their position in the transverse @xmath29-plane is easily calculated @xmath30 where @xmath31 and @xmath32 are the radial and longitudinal components of particle velocity respectively ( @xmath33 is the particle energy in the laboratory frame ) .
since the size of the observation region is of the order of several centimetres , these radial distances must obey the following restriction : @xmath34 @xmath35 we set @xmath36 mm , @xmath37 mm .
the restriction ( [ mini ] ) simply means that spots are not mixed with the central one formed by the particles which fly close to the collision axis ( mainly from the fragmentation region of an incident proton ) .
the separation of spots in the @xmath9-ray film gives another restriction on the distance between particles @xmath38 it must be larger than 1 mm : @xmath39 in the opposite case the particles must be combined in one particle - cluster until there remain only particles and/or particle - clusters with the mutual distances larger than @xmath40 , each such particle - cluster being considered as a single particle with coordinates defined in the same way as center - of - mass coordinates of two bodies : @xmath41 then we select @xmath42 clusters / particles which are most energetic and obey the restrictions ( [ mini ] , [ max ] , [ dijres ] ) and calculate the alignment @xmath43 using the conventional definition @xcite : @xmath44 and taking into account the central cluster , i.e. @xmath45 . here @xmath46 is the angle between two vectors ( @xmath47 ) and ( @xmath48 ) ( for the central spot @xmath49 ) .
this parameter characterizes the location of @xmath50 points just along the straight line and varies from @xmath51 to @xmath52 .
for instance , in the case of the symmetrical and close to most probable random configuration of three points in a plane ( the equilateral triangle ) @xmath53 .
the ultimate case of perfect alignment is @xmath54 when all points lie exactly along the straight line , while for an isotropic distribution @xmath55 .
the alignment degree @xmath56 is defined as a fraction of events with @xmath57 @xcite with the number of cores not less than @xmath50 .
if the hypothesis about the relation of alignment to the prevailing jet character of events at super high energies is valid , then this must manifest itself first of all in nucleon - nucleon collisions .
therefore , to be specific we consider a collision of two protons and fix a primary energy in the laboratory system @xmath58 tev , that is equivalent to @xmath59 tev just the energy attainable at lhc ( the rapidity shift being @xmath60 after the transformation from the center - of - mass system to the laboratory one ) . to simulate a collision of two protons with such energies we use the monte carlo generator pythia @xcite , which basically well describes jet events in hadron - hadron interactions and is tuned using the available experimental accelerator data .
the results of numerical simulation which follows the consideration in the previous section are presented in fig .
1 ( solid curve ) with the parameters @xmath61 mm , @xmath62 mm , @xmath63 m , which are close to the conditions of emulsion experiments , with the additional restriction on the energy threshold of particle registration in the emulsion : @xmath64 tev .
the estimated alignment degree @xmath56 for @xmath50 cores is considerably larger than that for randomly selected chaoticly located spots in the @xmath9-ray film , but is still too small ( by a factor of 34 ) to describe the experimental data @xcite even taking into account their large errors .
this can mean that the jet activity is not sufficient at such energies or the jet mechanism can not , in principle , give the large experimentally observable alignment . in order to try to answer this question
let us consider the influence of the applied restrictions ( [ mini ] , [ max ] ) ( the laboratory acceptance criterion ) on the spectrum of particles selected to calculate the alignment . for particles with high enough transverse momenta @xmath21 relative to their masses
@xmath65 these conditions ( [ mini ] , [ max ] ) reduce , mainly , to the restriction on the available particle rapidities in the center - of - mass system : @xmath66 @xmath67 since in this case @xmath68 for @xmath69 , where @xmath70 due to the kinematical restriction @xcite , @xmath71 a production of harder jets with larger rapidities becomes possible with the growth of @xmath72 .
the rapidity region ( [ mini1 ] , [ max1 ] ) just corresponds to the transition from soft to hard qcd physics , where the jet activity _ could _ manifest itself . here
one should note that ultrarelativistic particles ( @xmath73 ) are detected in the @xmath9-ray film from the restricted rapidity region ( [ mini1 ] , [ max1 ] ) which excludes such configurations as back - to - back hard jets with rapidities close to zero in the center - of - mass system .
but just such configurations with scattering of hard partons at angles close to @xmath74 in the considered hadronic center - of - mass system ( which in this case practically coincides with the partonic center - of - mass system ) can be expected to be responsible for the alignment phenomenon .
the point is that leading particles from both these hard jets have quantitatively comparable energies in the laboratory frame together with the `` strong memory '' of scattering plane .
meanwhile leading particles from any other back - to - back hard jets with the relatively large modulo rapidities @xmath75 in the partonic center - of - mass system have essentially different laboratory energies due to lorentz boost . and the energy distinction is mainly determined by the value of @xmath76 . in the latter case particles from a forward
hard jet produce as a rule the most energetic clusters ( apart from the central one ) in the laboratory frame , even if the particles from a backward ( in the partonic center - of - mass system ) jet hit the detection region .
however such most energetic clusters from one jet are less correlated with the primary scattering plane and therefore will not be much aligned as clusters from both hard jets .
this argumentation is confirmed by our simulation . in this connection
it is necessary to comment on the work @xcite in which the jet hypothesis has been suggested for the first time for the explanation of alignment phenomenon .
there the high degree of alignment has been demonstrated for the four cores only , using a simplified picture of fragmentation process .
in fact , an axis distribution has been calculated in the partonic center - of - mass system in the first order of perturbative qcd theory at the partonic level , considering three partons in the final state only .
then the lorentz transformation has been done in order to find their directions and localizations with respect to the central spot in the laboratory frame , considering each parton - jet as one long - living system ( the fragmentation time is of the order of flight time ) with some effective mass and aggregate group velocity .
the velocity has been fixed ( @xmath77 in @xcite ) so as to be able to include the events giving the high degree of alignment and corresponding to the two final - state parton - jets in the backward hemisphere and one in the forward hemisphere in the partonic center - of - mass system . in our variables
this means that a mass factor @xmath78 must be very small for a such massive system ( @xmath79 ) so that it hits the detection region even with negative rapidities ( i.e. corresponds to the backward hemisphere in the partonic center - of - mass system ) , if one uses the same boost parameters as for the hadronic center - of - mass system without taking into account the possible additional boost due to the distinction between these partonic and hadronic frames .
note that for real particles , e.g. @xmath80-mesons which mainly contribute to the multiplicity , this mass factor becomes significant for very small transverse momenta , @xmath81 gev , only .
however , as our investigation shows , falling of appropriately correlated particles into the observation region is still not sufficient to obtain the high degree of alignment because of the energy selection procedure , if the total number of particles is large and they generate many distinctly separated spots . for completeness one should also mention that high transverse momentum jet production has a connection to the double - core configuration of cosmic - ray events as it has been pointed out in @xcite . under certain conditions a hard forward ( in the partonic center - of - mass system )
jet together with a central bunch gives two relatively far separated clusters with large energies .
the detailed studies of double - core ( or binocular ) phenomena with estimations of event rates and average lateral spread of the @xmath4-family using a pqcd based monte carlo can be found , for instance , in @xcite .
ultrarelativistic particles from the central rapidity region in the hadronic center - of - mass system ( as possible sources of appropriately correlated spots ) can hit the observation region owing to the decrease of @xmath82 _ only _ , i.e. the decrease of the height @xmath10 of primary interaction or the increase of the rapidity @xmath28 of the center - of - mass system due to the growth of energy @xmath72 , as it follows from ( [ r0 ] ) .
the energy growth seems preferable , if we intend to be closer to emulsion experiments and increase the jet activity .
however this demands the extrapolation of pythia parameters and their special tuning to the experimentally untested energy domain .
updating can be done appropriately after the lhc operation starts .
moreover at present this generator already uses the extrapolation of experimentally tested cross sections and structure functions to the lhc energy region @xmath59 tev in order to estimate the effects expected at such energies . for illustration
we utilize the first `` less dangerous '' alternative decrease the interaction height by a factor of @xmath83 rather than increase the energy @xmath72 by the same factor of @xmath83 at the initial height so that particles from both hard jets ( with back - to - back structure ) , hitting the registration region , come from some rapidity range near @xmath84 including adjoint positive and negative values . in this case
the alignment degree becomes strongly dependent on the minimum transverse momentum of hard process , @xmath85 , which is a parameter of pythia . at the height @xmath13 km
such dependence was not visible , although we might catch some marginal tendency of the alignment degree to grow with the increase of @xmath85 at that height .
however without the restriction on @xmath85 from below ( minimum bias ) the result coincides practically with one obtained earlier ( solid curve in fig .
1 ) that shows some general characteristics of jet structure of events . if @xmath86 tev , particles from these hard jets together with particles flying close to @xmath22-axis ( within the transverse radius @xmath87 mm ) result in the alignment degree ( dashed curve ) comparable with the experimentally observed one @xcite .
thus the jet - like mechanism can , in principle , attempt to explain the results of emulsion experiments .
for such an explanation it is necessary ( but not sufficient ) that particles from both hard jets ( with rapidities near @xmath84 in the center - of - mass system ) hit the observation region .
this is possible at the relatively small height @xmath88 m and @xmath59 tev ; or at the height @xmath63 m , but the considerably higher energy @xmath89 tev ; or at some reasonable and acceptable intermediate combination of @xmath10 , @xmath72 and @xmath90 which meets the following condition : @xmath91 where @xmath92 is the proton mass .
@xmath93 is needed in order to have particles with @xmath94 that hit the detection region ( see ( [ max1 ] ) ) .
we verified the decisive significance of condition ( [ r00 ] ) to allow the observation of large degree of alignment and its dependence on the process hardness for the smaller energy @xmath95 tev ( where the prediction of pythia is quite adequate ) and the height @xmath96 m ( in accordance with ( [ r00 ] ) ) thereby confirming this peculiar kinematic `` scaling '' . at @xmath97 tev jets carry away about half of the energy of colliding protons in the center - of - mass system due to the relationship in a parton picture @xmath98 , where @xmath99 is a fraction of proton energy carried by each interacting parton ( quark or gluon ) .
the striking feature of such configurations in the @xmath9-ray film is approximate equality of energy deposition in the central and the rest most energetic clusters , that can be one of the physical guideline to select the events with very hard jets not only at the generator level ( simulation ) .
if we simply apply the additional threshold on the minimum energy of detected clusters needed in the alignment analysis , then we still obtain neither the desirable selection of jet hardness nor the increase of the alignment degree .
the small variation of resolution parameter @xmath40 does not provide the desirable effect also .
however , introduction of another threshold on the total energy of all @xmath100 selected clusters @xmath101 ( without taking into account the energy deposition in the central cluster around @xmath102 ) , @xmath103 allows us to select the events with hard jets only in a `` natural '' physical way and to reduce the hypothesis to the really active mechanism .
figure 2 shows that the alignment degree increases with the growth of @xmath104 ( the restriction on @xmath85 is absent at all ! ) , and it becomes large enough ( dashed curve ) and comparable with the experimentally observed one @xcite above the threshold @xmath105 pev . though one should note that our estimations give still too steep dependence on @xmath50 as one can see in figs .
1b , 2b from comparison of slopes of straight lines with the experimental behaviour . to give the reader a feeling for the various measures of alignment we present in figs .
3 and 4 the spatial distributions of most energetic clusters in the @xmath29-plane for a few generated events along with the corresponding values of @xmath43 .
some spots are hardly visible because of their small sizes which are proportional to the cluster energies ( especially in the case @xmath106 ) or because they are outside a square @xmath107 mm@xmath108 mm ( but inside a circle @xmath12 mm ) as it sometimes happens in the case @xmath109 . besides for @xmath106 we can distinctly see three relatively large spots resulted from two hard jets and a central bunch . here
one should note that there was slightly other criterion for the selection of families for the analysis in the works of pamir collaboration : the families with the total energy of @xmath4-quanta larger than a certain threshold and at least one hadron present were selected and analyzed .
the alignment becomes apparent considerably at @xmath110 pev ( the families being produced , mostly , by a proton with energy @xmath111 pev ) . since the adequate comparison of our estimations with experimental data is impossible without a full simulation of particle propagation through the atmosphere , taking into account the energy distribution of primary cosmic particles , etc .
, then in order to demonstrate the possibility of appearance of high alignment degree due to the jet mechanism we restrict ourself to the simpler ( but as concerns physics essentially close to experiment ) criterion of selection over the total energy of all particles .
these particles are mostly @xmath80-mesons , the neutrals among them being the main source of the detected @xmath4-quanta .
it is natural that the threshold on the total energy of all particles must be larger than the similar threshold on the total energy of @xmath4-quanta at the same collision energy . for comparison
we estimate also the alignment degree selecting only the most energetic @xmath4-quanta with their total energy larger than certain threshold @xmath112 ( fig .
5 ) : @xmath113 the result is close to that obtained previously with the threshold imposed on the total energy of all particles . besides for jet events @xmath114 with a high accuracy ( see figs .
1b , 2b , 5b , which present the dependence of alignment degree on the number of considered cores at the different values of hardness parameter ( 1b ) and threshold total energy ( 2b , 5b ) in the logarithmic scale ) .
this constant depends on @xmath85 , @xmath104 , @xmath112 , decreasing with their growth , and could in principle be determined by the kernels of the gribov
altarelli parisi
dokshitzer equations @xcite which describe the process of radiation of quarks and gluons in the initial and final states . and
, in fact , this process is implemented in the pythia generator together with the subsequent hadronization of quarks and gluons .
if nevertheless particles from the central rapidity region @xmath84 and the jet - like mechanism are insufficient to describe the observed alignment and there is another mechanism of its appearance at the energy @xmath115 tev and the height @xmath116 m ( mostly used in emulsion experiment estimations ) , then in any case some sort of alignment should arise at lhc too in the rapidity region ( [ mini1 ] , [ max1 ] ) .
this region must be investigated more carefully on the purpose to study the azimuthal anisotropy of energy flux in accordance with the procedure applied in the emulsion and other experiments , i.e. one should analyze the energy deposition in the cells of @xmath117-space in the rapidity interval ( [ mini1 ] , [ max1 ] ) ( the equivalent threshold minimum particle energy being @xmath118 gev in the center - of - mass system ) .
note that the absolute rapidity interval can be shifted : it is necessary only that the difference ( @xmath119 is equal to @xmath120 in accordance with the variation of radial distance by a factor of 15 ( @xmath121 ) due to the relationship @xmath122 ( independently of @xmath82 ) . in other words , since we use particle momenta in the center - of - mass system , then future data should be treated in accordance with the algorithm described earlier in sects . 2 , 3 introducing the corresponding laboratory observables .
our analysis shows that for @xmath123-collision at a fixed height of primary interaction above the energy @xmath72 , when the condition ( [ r00 ] ) is fulfilled that is ultrarelativistic particles from the rapidity interval near @xmath84 in the center - of - mass system fall into the observation region inside the radius @xmath90 in the laboratory frame due to the large lorentz factor the alignment of spots arises ( this , in principle , explains the existence of the experimental energy threshold of this effect ) and the alignment degree becomes strongly dependent on the process hardness .
if the process hardness is close to maximum for the given energy @xmath72 , the estimated degree of alignment is already comparable with the experimentally observed one . introducing another additional threshold (
the scale of which is determined by the energy of an incident proton ) on the total energy of all @xmath100 selected most energetic clusters ( without taking into account the energy deposition in the central cluster ) allows us to select the events with high hardness in a
natural physical way and thereby support the jet - like hypothesis , which later on may be accepted ( or refuted ) in further investigations of , for instance , the energy cluster distribution and their particle composition with regard for interactions in the atmosphere , etc . meanwhile we suggest the more careful investigation of the rapidity region ( [ mini1 ] , [ max1 ] ) at lhc in order to reveal the new still unknown mechanisms of alignment if they exist . for this purpose one should perform the analysis of energy deposition in calorimeters of cms and atlas experiments in accordance with the procedure described in sects . 2 , 3 ( i.e. calculating the appropriate observables in the laboratory frame ) .
such investigation can clarify the origin of the alignment , test the alternative hypotheses and give the new restrictions on the values of height and energy .
it is pleasure to thank a.i .
demianov , s.v .
molodtsov , s.a .
slavatinsky , l.g .
sveshnikova , k.yu .
teplov and g.t .
zatsepin for discussions .
this work is supported by grant n 04 - 02 - 16333 of russian foundation for basic research .
pamir collaboration , in proceedings of the 21st international cosmic ray conference , adelaide , australia ( 1989 ) , edited by r.j.protheroe ( university of adelaide , australia ) , 227 ( 1990 ) ; + s.a .
slavatinsky , in proceedings of the 5th international symposium on very high energy cosmic ray interactions , lodz , poland ( 1988 ) , edited by m. giler ( university of lodz , lodz , poland ) , 90 ( 1989 ) .
kopenkin , a.k .
managadze , i.v .
rakobolskaya , t.m .
roganova , phys .
d * 52 * , 2766 ( 1995 ) f. halzen , d.a .
morris , phys .
d * 42 * , 1435 ( 1990 ) t. sjostrand , comp .
phys . com .
* 135 * , 238 ( 2001 ) i.v .
rakobolskaya et al .
the peculiarity of hadron interactions of cosmic rays at super high energies ( msu , moscow , 2000 ) ( in russian ) .
dokshitzer , d.i .
dyakonov , s.i .
troyan , phys .
* 58 * , 269 ( 1980 ) v.v .
kopenkin , a.k .
managadze , i.v .
rakobolskaya , t.m .
roganova , izv .
. fiz . * 58 * , 13 ( 1994 ) d. cline , f. halzen , j. luthe , phys .
lett . * 31 * , 491 ( 1973 ) z. cao , l.k .
ding , q.q .
zhu , y.d .
he , phys .
lett . * 72 * , 1794 ( 1994 ) z. cao , l.k .
ding , q.q .
zhu , y.d .
he , phys .
d * 56 * , 7361 ( 1997 ) v.n .
gribov , l.n .
lipatov , sov .
. phys . * 15 * , 438 ( 1972 ) ; + v.n .
gribov , l.n .
lipatov , sov .
15 * , 675 ( 1972 ) l.n .
lipatov , sov .
. phys . * 20 * , 94 ( 1974 ) yu.l .
dokshitzer , sov .
j. jetp * 46 * , 641 ( 1977 ) g. altarelli , g. parisi , nucl .
b * 126 * , 298 ( 1977 ) | the hypothesis about the relation between the observed alignment of spots in the x - ray film in cosmic ray emulsion experiments and the features of events in which jets prevail at super high energies is tested .
due to strong dynamical correlation between jet axis directions and that between momenta of jet particles ( almost collinearity ) , the evaluated degree of alignment is considerably larger than that for randomly selected chaoticly located spots in the x - ray film .
it appears comparable with experimental data provided that the height of primary interaction , the collision energy and the total energy of selected clusters meet certain conditions .
the monte carlo generator pythia , which basically well describes jet events in hadron - hadron interactions , was used for the analysis .
= -15 mm # 1#23.6pt * jet activity versus alignment * i.p .
lokhtin@xmath0 , a.k .
managadze@xmath1 , l.i .
sarycheva@xmath2 , a.m. snigirev@xmath3 + m.v.lomonosov moscow state university , d.v.skobeltsyn institute of nuclear physics , + 119992 , vorobievy gory , moscow , russia + + @xmath0e - mail : [email protected] + @xmath1e - mail : [email protected] + @xmath2e - mail : [email protected] + @xmath3e - mail : [email protected] + |
for last few years , there was large progress in supersymmetric field theories in three - dimension .
many physical quantities in strongly coupled field theories have been computed by using localization method .
for example , partition functions of @xmath3 supersymmetric field theories in @xmath4 @xcite and @xmath5 @xcite are computed exactly , and they have been used to study non - perturbative aspects of three - dimensional field theories , such as dualities among three - dimensional field theories @xcite and relation to m - theory via ads / cft correspondence @xcite in this paper , we investigate the partition function of three - dimensional @xmath3 supersymmetric field theories in the orbifold @xmath6 @xcite . due to the non - trivial homotopy of the orbifold , @xmath7 , a gauge theory defined in it has degenerate vacua specified by the holonomy associated with the gauge symmetry .
their contributions are summed up to obtain the total partition function . in general , the partition function of a euclidean theory is complex .
we usually focus only on its absolute value and the phase is disregarded .
this is , however , not allowed when we compute the partition functions of different sectors which are summed up .
even when we are interested only in the absolute value of the total partition function , we need to care about the relative phase of each contribution .
the purpose of this paper is to determine appropriate phase factors in the holonomy sum in some gauge theories and look for a general rule for these phases .
we consider two gauge theories which are known to have dual field theories without vector multiplet .
on one side of the dualities , in the non - gauge theories , we can compute the absolute value of the partition function up to overall constant factor independent of parameters . by comparing the partition functions of gauge and non - gauge theories in each dual pair ,
we infer the relative phases in the holonomy sum in the gauge theories .
this paper is organized as follows . in [ pf.sec ] , we summarize a general formula of the orbifold partition function .
it is written by using orbifold extension of the double sign function , which we denote by @xmath8 . in [ dual.sec ] , we consider two dual pairs and determine the phase factors in the holonomy sum in the gauge theories so that the sum agrees with the partition function of the dual non - gauge theories .
we find that when the order @xmath2 of the orbifold group is odd the phase factor is absorbed in the definition of the function @xmath8 . in [ der.sec ] we consider more dual pairs which are derived from one of the dualities studied in [ dual.sec ] .
the last section is devoted to the conclusions and discussions .
let us first summarize the partition function on @xmath4 without orbifolding @xcite .
we consider the squashed @xmath4 with the metric @xmath9,\ ] ] where @xmath10 ( @xmath11 ) are the left - invariant differentials on @xmath12 .
we set @xmath13 in the following .
@xmath14 is the squashing parameter .
we also define parameters @xmath15 and @xmath16 by @xmath17 for later use .
the isometry of the round sphere @xmath18 is broken by the squashing to @xmath19 we consider an @xmath3 supersymmetric gauge theory on this manifold .
the supercharges belong to the representation @xmath20 of ( [ su2u1 ] ) .
a general formula for the partition function for the squashed sphere is given in @xcite .
the same partition function is first obtained in @xcite for a different deformation of @xmath4 .
see also @xcite for its relation to 4d superconformal index .
the formula is @xmath21e^{-s_0(\lambda)}z^{\rm 1-loop } , \label{s3formula}\ ] ] where @xmath22 is the one - loop determinant @xmath23 the index @xmath24 labels chiral multiplets , and @xmath25 and @xmath26 are the weight vectors and the weyl weights of the chiral multiplets .
the integration variable @xmath27 is an element of the cartan subalgebra of the gauge group @xmath28 parameterizing the coulomb branch . if @xmath28 is @xmath29 or product of @xmath29 , the integration measure @xmath30 $ ] is defined by @xmath31=\frac{1}{|w|}\prod_{a=1}^{{\rm rank } g}d\lambda_a , \label{s3measure}\ ] ] where @xmath32 is the order of the weyl group of @xmath28 , and @xmath33 are the diagonal components in the fundamental representation of @xmath27 .
different normalizations are also used in the literature .
the normalization ( [ s3measure ] ) is chosen so that theories in the dual pairs we will consider in [ dual.sec ] have the same partition functions .
we use in this paper is different from that in @xcite .
@xmath27 here is related to the constant mode @xmath34 of scalar field defined in @xcite by @xmath35 . ]
@xmath36 is the classical action .
if the theory has the chern - simons term @xmath37 and the fayet - iliopoulos term @xmath38 then the supersymmetric completion of these actions contributes to the classical action by @xmath39 @xmath40 is the double sine function defined by is directly related to the non - compact quantum dilogarithm @xmath41 @xcite by @xmath42 \varphi_b(z)$ ] .
see also @xcite for more details on @xmath40 . ]
@xmath43 the formula ( [ zs3a ] ) is derived by using localization , which reduces the path integral to gaussian integral for infinite number of non - zero modes and finite dimensional integral for zero modes .
the gaussian integral is performed by using spherical harmonics expansion . for a vector multiplet with weight vector @xmath44 , the gaussian integral of modes with specific @xmath45 quantum numbers @xmath46
gives @xmath47 notice that @xmath48 dependence appears only when @xmath4 is squashed . for round sphere with @xmath49 , there is no @xmath48 dependence due to the unbroken @xmath45 symmetry . if we set @xmath50 ( [ jmvector ] ) becomes the factor in the definition ( [ doublesine ] ) of the double sine function @xmath40 with argument @xmath51 , and by taking the product over @xmath52 and @xmath53 , we obtain the double sine function appearing in the numerator in ( [ zs3a ] ) .
the denominator in ( [ zs3a ] ) also arises from the gaussian integral of non - zero modes of chiral multiplets .
note that variables @xmath52 and @xmath53 appear differently in the numerator and the denominator in ( [ doublesine ] ) . in the numerator @xmath52 and @xmath53 appear in the coefficients of @xmath54 and @xmath16 , respectively , and in the denominator , the relation is reversed . because they are dummy variables we can exchange them in the numerator or in the denominator so that they appear in the same manner .
however , the relation to the @xmath45 quantum numbers ( [ mnpq ] ) holds only when we write the infinite product as in ( [ doublesine ] ) .
this becomes important when we consider the orbifolding by @xmath55 in the next subsection . from the definition ( [ doublesine ] )
we can easily show the following relations for @xmath40 . * self - duality and reflection property @xmath56 * functional equations @xmath57[2\sinh(\pi b^{-1 } z)]}. \label{funceq}\end{aligned}\ ] ] we consider the left - invariant orbifold @xmath6 with @xmath59 .
the partition function on the orbifold is obtained in @xcite for theories without matter fields in a general lens space @xmath60 without squashing .
it is extended to theories with chiral multiplets in background with nontrivial squashing parameter in @xcite .
our orbifold corresponds to @xmath61 .
because supercharges are @xmath62 neutral , the orbifolding by @xmath63 does not break any supersymmetry , and we can define @xmath3 supersymmetric theories on the orbifold . a gauge theory in this orbifold has degenerate vacua specified by the holonomy @xmath64 where @xmath65 is the generator of the fundamental group @xmath7 .
the consistency to @xmath66 requires @xmath67 .
( note that we define @xmath48 with the factor @xmath2 in ( [ mdef ] ) . )
the holonomy can be turned on for both global and gauge symmetries .
the holonomy for gauge symmetries should be summed up in the path integral .
the partition function is given by @xmath68e^{-s_0(\lambda , m ) } z^{\rm 1-loop}(\lambda , m ) , \label{znpartitionfunc}\ ] ] where @xmath69 and @xmath70 are the classical action and the one - loop determinant .
the summation is taken over the holonomy associated with gauge symmetry , which is denoted by @xmath71 in ( [ znpartitionfunc ] ) .
the holonomy for global symmetry @xmath72 is not summed , and the partition function @xmath73 depends on @xmath72 .
the integration measure @xmath30 $ ] is defined by @xmath31=\frac{1}{|w|}\prod_{a=1}^{{\rm rank } g}\frac{d\lambda_a}{n}. \label{orbmeasure}\ ] ] we introduce the factor @xmath74 for each integration variable for later convenience .
one may think that the classical action for @xmath6 is obtained by dividing that for @xmath4 by @xmath2 .
this naive expectation is not correct .
the classical action @xmath69 consists of two parts ; @xmath75 one is @xmath74 of the classical action for @xmath4 , and has the same origin as the @xmath4 case .
the other part comes from the chern - simons term . due to the non - trivial topology of @xmath6 ,
the chern - simons term gives non - vanishing contribution even for a flat gauge connection @xcite ; @xmath76 this phase plays an important role in dualities in @xmath6 .
the factor @xmath77 may be ill - defined depending on the coefficient .
if @xmath78 is odd , the holonomies @xmath79 and @xmath80 , which are identified in @xmath81 , give different phases .
we will meet such an ambiguity in the example in
[ jy.sec ] , and there we will give an additional rule to fix the ambiguity .
the one - loop partition function for the orbifold can be obtained by projecting out the factors in ( [ doublesine ] ) which originate from @xmath81-variant modes .
let @xmath82 be a field with a weight vector @xmath83 .
on @xmath4 it is fourier expanded as @xmath84 where @xmath85 is the coordinate along the hopf fiber of @xmath4 , and @xmath48 is the @xmath45 magnetic quantum number .
after @xmath81 orbifolding , the field must satisfy the boundary condition @xmath86 and only modes @xmath87 with the index @xmath48 satisfying @xmath88 survive after the orbifold projection .
we define @xmath8 as the function obtained from ( [ doublesine ] ) by restricting the product over @xmath89 by ( [ pqcond ] ) .
this restricted product is realized by substituting @xmath90_n,\quad q = nq'+k , \label{pqbypqk}\ ] ] to ( [ doublesine ] ) , and perform the product with respect to non - negative integers @xmath91 and @xmath92 , and @xmath93 .
@xmath94_n$ ] represents the remainder when @xmath48 is divided by @xmath2 .
it is convenient to introduce notation @xmath95 defined by @xmath96_n+\frac{1}{2}\right)-\frac{1}{2}.\ ] ] this satisfies the relations @xmath97 we rewrite the numerator in ( [ doublesine ] ) as @xmath98.\end{aligned}\ ] ] the denominator in ( [ doublesine ] ) is also rewritten in a similar way , and we obtain @xmath99 in the second line we replaced @xmath100 in the denominator by @xmath101 and used the second relation in ( [ lrangrel ] ) . in the final expression the product with respect to @xmath91 and @xmath92 has the same form as that in the definition ( [ doublesine ] ) of the double sine function @xmath40 , and we obtain @xmath102 by definition , the product of @xmath8 over all @xmath103 reproduces the original double sine function ; @xmath104 the function @xmath8 satisfies the following formulae , which are analogs of ( [ self ] ) and ( [ funceq ] ) .
* self - duality and reflection property @xmath105 * functional equations @xmath106 [ 2\sinh\left(\frac{\pi b^{-1 } z-\pi ih}{n}\right)]}.\end{aligned}\ ] ] the one - loop determinant for @xmath107 is obtained simply by replacing @xmath40 in ( [ zs3a ] ) by @xmath8 . @xmath108
the @xmath109-loop determinant of vector multiplets can be rewritten in terms of elementary functions .
@xmath110 \left[2\sinh\frac{\pi}{n}\left(b^{-1}\alpha(\lambda)-i\alpha(m)\right)\right].\end{aligned}\ ] ] when @xmath111 , this agree with the partition function in the lens space @xmath61 given in @xcite .
a gauge theory in @xmath6 has degenerate vacua labeled by holonomies associated with the gauge symmetry .
the contributions of these vacua should be summed up to obtain the total partition function . in this section
we consider two dual pairs and confirm that the partition functions of theories dual to each other agree if appropriate phase factors are inserted in the holonomy sum .
we first consider the mirror symmetry between an @xmath3 sqed and the xyz model @xcite .
on one side of the duality , we consider @xmath3 sqed with two chiral multiplets @xmath53 and @xmath112 with @xmath113 charge @xmath114 and @xmath115 , respectively .
we assume that @xmath53 and @xmath112 have the weyl weight @xmath116 .
the mirror theory , the xyz model , consists of three chiral multiplets @xmath117 , @xmath118 and @xmath119 interacting through the superpotential @xmath120 although three fields @xmath119 , @xmath117 , and @xmath118 in this model are symmetric , we treat @xmath117 and @xmath118 as a quark and an antiquark , and @xmath119 as a neutral field , because the mirror pair considered here is a special case of a series of mirror pairs , which are studied in [ nf.sec ] , and in a general mirror pair @xmath117 and @xmath118 are replaced by charged multiplets and @xmath119 by neutral ones . by the operator relation @xmath121 and the marginality of the superpotential ( [ qsq ] )
, we can determine the weyl weights of the fields in this theory as @xmath122 although the correct value of @xmath116 at the infra - red fixed point is @xmath123 , the equality of partition functions holds regardless of @xmath116 @xcite , and we leave @xmath116 unfixed . the global symmetry which is the same for these two theories is @xmath124 .
the charge assignments are shown in table [ table:1 ] .
.the charge assignment of the global symmetry @xmath124 of sqed and xyz model which are mirror to each other . @xmath48 and @xmath125 are the monopole and anti - monopole operators . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] + @xmath126 in the mirror theory is a topological @xmath113 symmetry coupling to monopole operators , and the corresponding mass parameter is the fayet - iliopoulos parameter .
although this mirror symmetry holds for an arbitrary rank @xmath127 of the gauge groups , we only consider abelian ( @xmath128 ) case for simplicity .
let us first consider the orbifold partition function in the mirror theory . when the gauge group is @xmath113 , the system consists of two sectors decoupled from each other : the neutral hypermultiplet and the @xmath129 sqed with one flavor . in the sqed
we need to carry out the holonomy sum to obtain the partition function , and it is non - trivial how we should choose the relative phases .
we here take the sign function ( [ eq : sign ] ) used in [ hyp.subsec ] .
we will see shortly that the partition function obtained with this phase factor is indeed reproduced in the abjm model , too , with an appropriate choice of the phases in the holonomy sum in the abjm model .
the partition function of the mirror theory is @xmath130 with the contribution of each holonomy sector @xmath131 & = z^{\rm hyper}(m_a ,-
h_t)|_{\delta=\frac{1}{2 } } z^{{\cal n}=4}(m_b , m_t;h , h_b , h_t)|_{\delta=\frac{1}{2}},\end{aligned}\ ] ] where we introduce mass parameters @xmath132 and holonomies @xmath133 corresponding to the flavor symmetry @xmath134 .
@xmath103 is the holonomy for the gauge symmetry . with the relation ( [ hypern4 ] ) , we can rewrite the partition function ( [ zmirrortot ] ) as the product of two @xmath135 ; @xmath136 on the abjm side , we need to sum up @xmath137 contributions parameterized by a pair of holonomies @xmath138 for the gauge group @xmath139 .
the partition function of the sector specified by @xmath138 is @xmath140 } { \begin{array}{l } s_{b , h_a+h_t+h_1-h_2}(m_a+m_t+\lambda-{\widetilde}\lambda-\frac{i}{2v } ) \\\times s_{b ,- h_a+h_t+h_1-h_2}(-m_a+m_t+\lambda-{\widetilde}\lambda-\frac{i}{2v } ) \\\times s_{b , h_b - h_t - h_1+h_2}(m_b - m_t+{\widetilde}\lambda-\lambda-\frac{i}{2v } ) \\\times s_{b ,- h_b - h_t - h_1+h_2}(-m_b - m_t+{\widetilde}\lambda-\lambda-\frac{i}{2v } ) \end{array } } .
\label{abjmzholo}\end{aligned}\ ] ] a question is if it is possible to choose an appropriate phases in the holonomy sum .
the answer is rather simple .
we do not need any non - trivial phases in this sum .
let us confirm this by summing up ( [ abjmzholo ] ) over holonomies @xmath141 and @xmath142 .
if we define @xmath143 and replace @xmath141 by @xmath144 , @xmath142 appears only in the phase factor @xmath145 the summation with respect to @xmath142 gives non - vanishing result only when @xmath146 , and we obtain @xmath147 } { \begin{array}{l } s_{b , h_a+h_t}(m_a+m_t+\lambda-{\widetilde}\lambda-\frac{i}{2v } ) \\\times s_{b ,- h_a+h_t}(-m_a+m_t+\lambda-{\widetilde}\lambda-\frac{i}{2v } ) \\\times s_{b , h_b - h_t}(m_b - m_t+{\widetilde}\lambda-\lambda-\frac{i}{2v } ) \\\times s_{b ,- h_b - h_t}(-m_b - m_t+{\widetilde}\lambda-\lambda-\frac{i}{2v } ) \end{array } } .\end{aligned}\ ] ] we can easily perform the integral and have @xmath148 & = z^{\rm hyper}(m_a ,-
h_t)|_{\delta=\frac{1}{2 } } z^{\rm hyper}(m_b , m_t;h_b , h_t)|_{\delta=\frac{1}{2 } } .
\label{zabjm}\end{aligned}\ ] ] this result precisely agrees with the partition function of the mirror theory ( [ zmirror ] ) .
we investigated relative phases in the holonomy sum , which is necessary to obtain the partition functions of gauge theories in @xmath149 .
we used dualities between gauge theories and non - gauge theories to determine the phases .
we first consider mirror symmetry between @xmath3 sqed with one flavor and the xyz model containing three chiral multiplets .
we showed that with the appropriate choice of the phases in the holonomy sum the partition functions of these theories coincides .
furthermore , we found that when @xmath2 is odd , the phase factor is absorbed by the redefinition of the single function @xmath8 , the orbifold extension of the double sine function .
we also consider the duality between @xmath150 gauge theory and a chiral multiplet proposed by jafferis and yin .
we could again find phase factors which makes the duality relation hold .
when @xmath2 is odd the phases are absorbed by redefining the function @xmath8 in the same way as in the first example .
we also confirmed that it is possible to find phase factors in three more examples of dual pairs , which are derived from the mirror symmetry we studied first .
when @xmath2 is odd , in all these examples , the phase factors can be absorbed by the definition of the function @xmath8 .
this fact strongly suggests that the modified function @xmath151 in ( [ shat ] ) always gives a `` correct '' partition function in some sense
. it would be interesting to check whether the modified function gives the same partition functions for theories in dual pairs we did not studied in this paper . for the purpose of obtaining non - trivial evidences for dualities
, it is desirable that we first fix the relative phases in the holonomy sum in each theory without relying on dualities .
this should be possible at least for saddle points belonging to the same component of the configuration space . in this case
we should obtain the relative phases by analyzing carefully the behavior of the integration measure in the path integral under adiabatic deformations connecting saddle points .
it is also interesting to search for guiding principles to fix the ambiguity for the relative phases for the contribution of topologically disconnected sectors .
the large @xmath127 limit of the partition function of the abjm model on the oribfold is considered in @xcite , and the coincidence of the free energy calculated from the abjm model and that from the gravity dual is confirmed . though many sectors with different holonomies contribute to the partition function the authors found that in the large @xmath127 limit it is dominated by the specific contribution with a certain holonomy configuration and the relative phases are not important . however , when one consider the next leading order of @xmath152 the other contributions become significant , therefore , the phase plays an important role there .
we are grateful to m. honda for valuable discussions .
the work of y.i . was partially supported by grant - in - aid for scientific research ( c ) ( no.24540260 ) , ministry of education , science and culture , japan . d.y .
acknowledges the financial support from the global center of excellence program by mext , japan through the `` nanoscience and quantum physics '' project of the tokyo institute of technology .
99 a. kapustin , b. willett and i. yaakov , `` exact results for wilson loops in superconformal chern - simons theories with matter , '' jhep * 1003 * , 089 ( 2010 ) [ arxiv:0909.4559 [ hep - th ] ] .
d. l. jafferis , `` the exact superconformal r - symmetry extremizes z , '' arxiv:1012.3210 [ hep - th ] . n. hama , k. hosomichi and s. lee , `` notes on susy gauge theories on three - sphere , '' jhep * 1103 * , 127 ( 2011 ) [ arxiv:1012.3512 [ hep - th ] ] .
s. kim , `` the complete superconformal index for n=6 chern - simons theory , '' nucl .
b * 821 * , 241 ( 2009 ) [ arxiv:0903.4172 [ hep - th ] ] .
y. imamura and s. yokoyama , `` index for three dimensional superconformal field theories with general r - charge assignments , '' jhep * 1104 * , 007 ( 2011 ) [ arxiv:1101.0557 [ hep - th ] ] .
a. kapustin , b. willett and i. yaakov , `` nonperturbative tests of three - dimensional dualities , '' jhep * 1010 * , 013 ( 2010 ) [ arxiv:1003.5694 [ hep - th ] ] .
a. kapustin , b. willett and i. yaakov , `` tests of seiberg - like duality in three dimensions , '' arxiv:1012.4021 [ hep - th ] .
a. kapustin , `` seiberg - like duality in three dimensions for orthogonal gauge groups , '' arxiv:1104.0466 [ hep - th ] .
b. willett and i. yaakov , `` n=2 dualities and z extremization in three dimensions , '' arxiv:1104.0487 [ hep - th ] .
n. drukker , m. marino , p. putrov , `` from weak to strong coupling in abjm theory , '' commun .
phys . * 306 * , 511 - 563 ( 2011 ) .
[ arxiv:1007.3837 [ hep - th ] ] .
c. p. herzog , i. r. klebanov , s. s. pufu and t. tesileanu , `` multi - matrix models and tri - sasaki einstein spaces , '' phys .
d * 83 * , 046001 ( 2011 ) [ arxiv:1011.5487 [ hep - th ] ] .
d. gang , `` chern - simons theory on l(p , q ) lens spaces and localization , '' arxiv:0912.4664 [ hep - th ] .
f. benini , t. nishioka and m. yamazaki , `` 4d index to 3d index and 2d tqft , '' arxiv:1109.0283 [ hep - th ] .
n. hama , k. hosomichi and s. lee , `` susy gauge theories on squashed three - spheres , '' jhep * 1105 * , 014 ( 2011 ) [ arxiv:1102.4716 [ hep - th ] ] .
y. imamura and d. yokoyama , `` n=2 supersymmetric theories on squashed three - sphere , '' phys . rev .
d * 85 * , 025015 ( 2012 ) [ arxiv:1109.4734 [ hep - th ] ] .
f. a. h. dolan , v. p. spiridonov and g. s. vartanov , `` from 4d superconformal indices to 3d partition functions , '' phys .
b * 704 * , 234 ( 2011 ) [ arxiv:1104.1787 [ hep - th ] ] .
a. gadde and w. yan , `` reducing the 4d index to the @xmath153 partition function , '' arxiv:1104.2592 [ hep - th ] .
y. imamura , `` relation between the 4d superconformal index and the s@xmath154 partition function , '' [ arxiv:1104.4482 [ hep - th ] ] .
s. kharchev , d. lebedev , m. semenov - tian - shansky , `` unitary representations of u(q ) ( sl(2 , r ) ) , the modular double , and the multiparticle q deformed toda chains , '' commun .
phys . *
225 * , 573 - 609 ( 2002 ) .
[ hep - th/0102180 ] .
s. k. hansen , t. takata `` reshetikhin - turaev invariants of seifert 3-manifolds for classical simple lie algebras , and their asymptotic expansions , '' arxiv : math/0209403 [ math.gt ] l. griguolo , d. seminara , r. j. szabo and a. tanzini , `` black holes , instanton counting on toric singularities and q - deformed two - dimensional yang - mills theory , '' nucl .
b * 772 * , 1 ( 2007 ) [ hep - th/0610155 ] .
v. v. bazhanov , v. v. mangazeev and s. m. sergeev , `` faddeev - volkov solution of the yang - baxter equation and discrete conformal symmetry , '' nucl .
b * 784 * , 234 ( 2007 ) [ hep - th/0703041 ] .
b. ponsot and j. teschner , `` clebsch - gordan and racah - wigner coefficients for a continuous series of representations of u(q)(sl(2,r ) ) , '' commun .
* 224 * , 613 ( 2001 ) [ math/0007097 [ math - qa ] ] .
r. m. kashaev , `` the non - compact quantum dilogarithm and the baxter equations , '' j. statist .
phys . * 102 * , 923 - 936 ( 2001 ) .
k. a. intriligator and n. seiberg , `` mirror symmetry in three - dimensional gauge theories , '' phys .
b * 387 * , 513 ( 1996 ) [ hep - th/9607207 ] .
j. de boer , k. hori , y. oz and z. yin , `` branes and mirror symmetry in n = 2 supersymmetric gauge theories in three dimensions , '' nucl .
b * 502 * , 107 ( 1997 ) [ arxiv : hep - th/9702154 ] .
a. kapustin and b. willett , `` generalized superconformal index for three dimensional field theories , '' arxiv:1106.2484 [ hep - th ] .
o. aharony , o. bergman , d. l. jafferis and j. maldacena , `` @xmath155 superconformal chern - simons - matter theories , m2-branes and their gravity duals , '' jhep * 0810 * ( 2008 ) 091 [ arxiv:0806.1218 [ hep - th ] ] .
f. benini , c. closset and s. cremonesi , jhep * 1002 * , 036 ( 2010 ) [ arxiv:0911.4127 [ hep - th ] ] .
l. f. alday , m. fluder and j. sparks , `` the large n limit of m2-branes on lens spaces , '' arxiv:1204.1280 [ hep - th ] . | we investigate @xmath0 partition function of @xmath1 supersymmetric gauge theories . a gauge theory on the orbifold
has degenerate vacua specified by the holonomy .
the partition function is obtained by summing up the contributions of saddle points with different holonomies .
an appropriate choice of the phase of each contribution is essential to obtain the partition function .
we determine the relative phases in the holonomy sum in a few examples by using duality to non - gauge theories . in the case of odd @xmath2
the phase factors can be absorbed by modifying a single function appearing in the partition function . |
the dynamics of recurrent neural network models with synchronous updating has been of interest ever since the work of little @xcite .
it is expected that little s model with binary units and symmetric hebbian couplings should exhibit either cycles of period two or fixed - point attractors and it has been suggested that the cycles could arise in statistical mechanics from a duplication of phase space by means of two state variables for every unit .
an equilibrium analysis , based on replica symmetry , predicted that period - two cycles associated with a full spin flip ( all units changing sign at each time step ) should occur in a zero temperature paramagnetic phase .
this was recognized as an unphysical phase apparently disconnected from the ordinary high temperature paramagnetic phase . a somewhat large negative self - interaction @xmath0 ( the diagonal elements of the synaptic matrix ) between units turned out to be necessary for this cyclic solution with full spin flip to appear @xcite .
early numerical simulations failed to show the presence of those cycles and lead to the conjecture that the problem with the statistical mechanics approach could be due to the assumption of replica symmetry .
indeed , the entropy of this phase was found to be negative , going to @xmath1 as @xmath2 @xcite , a feature that usually goes together with replica symmetry @xcite . in a further work
, a zero - temperature calculation of the average number of cycles as a function of the fraction of flipping spins demonstrated that period - two cycles with full spin flip are by far the most common type of cycles below a storage ratio of patterns @xmath3 @xcite .
however , the usual calculation of the average number of metastable states does not say anything about the macroscopic properties , in particular about the overlap with a chosen pattern .
fixed - point solutions for the macroscopic parameters that emerge from a dynamics with a finite fraction of spins changing sign at each time step are common , but not exclusive , to the phase diagram of little s model .
indeed , they also appear in recent studies of three - state ising and blume - emery - griffiths neural network models with synchronous updating .
these models can be thought as extensions of little s model with generalized synaptic interactions and multi - state units @xcite .
only a small fraction of neurons that change sign appear to be involved in the stationary states @xcite , and the work in those studies was mainly devoted to fixed - point solutions for the macroscopic parameters . on the other hand , stationary period
- two solutions for the macroscopic parameters , that arise from a fraction of units changing sign at each time step , have been found in recent work on the synchronous dynamics of symmetric sequence processing without a self - interaction @xcite .
the relation between the dynamics of the macroscopic parameters and the fraction of units that change sign at each time step is crucial to understand the behavior of synchronous networks .
a dynamical study may be helpful to get further insight into the asymptotic behavior of little s model .
so far , only the results of an approximate synchronous dynamics are available , which become exact for an asymmetrically random diluted network in the limit of extreme dilution @xcite .
this dynamics predicts that , at @xmath4 and @xmath5 , either frozen - in fixed points or frozen - in cycles of period two may appear when @xmath6 , for a positive or a negative self - interaction , respectively , where @xmath7 is the initial overlap with a chosen pattern . on the other hand
, there is a flow to a fixed point @xmath8 reached in time @xmath9 , when @xmath10 .
thus , both the size and the sign of the self - interaction play a crucial role and the same dynamics , for small non - zero @xmath11 , predicts a flow either to a paramagnetic or to a retrieval state .
frozen - in states do not evolve in time and they are an undesirable feature for associative memory . in the present context , they are states in which the overlap either remains fixed at @xmath7 ( frozen - in fixed - point ) or oscillates between @xmath7 and @xmath12 ( frozen - in cycle of period two ) .
the correlation function between two consecutive time steps , related to the fraction of flipping spins , is @xmath13 for a frozen - in fixed - point and @xmath14 for a frozen - in cycle of period two .
the question that arises is if the frozen - in states actually are stable in little s fully connected model subject to synaptic and/or stochastic noise , and the main purpose of this paper is to deal with that issue following the transient behavior of the dynamics that yields the stationary states of the network .
the full dynamics of the hebbian synchronous recurrent network of binary units , either at zero or at finite temperature with stochastic noise , including eventually the region where period - two cycles dominate has , apparently , not been carried out before . we make
use of a generating functional approach ( gfa ) @xcite , exact in the mean - field limit of an infinitely large system , combined with the numerical simulation procedure of eissfeller and opper ( eo ) , based on the gfa for the dynamics of disordered spin systems .
this is a procedure free of finite size effects @xcite .
the problem with spin - glass - like models ( neural networks among them ) is that the dynamics may be very slow .
due to that , in particular for not too small @xmath15 in both cyclic and in retrieval states , the numerical simulation may require a large number of time steps in order to reach a stationary behavior , making the computation prohibitive with the eo procedure .
a further purpose of the paper is to overcome this problem by means of an alternative procedure that is introduced in this work and which consists mainly of two features .
one is the neglect of memory effects that involve the response function in the self - energy term of the single - site effective field , which amounts to a signal - to - noise approximation ( sna ) @xcite .
the second feature consists in doing analytically the summation over states in the gfa .
thereby one accounts for the microscopic variables in the calculation of the dynamics of the macroscopic properties in a way that involves a much smaller number of variables .
the procedure is exact for @xmath5 , and yields an approximation for small @xmath11 , that can be done in a short computing time up to a large number of time steps producing results in excellent agreement with the procedure of eo for finite @xmath16 and sufficiently large values of @xmath17 .
we make use of both our and the eo procedure in what follows .
the outline of the paper is the following . in section 2
we introduce the model and present a brief summary of the now well known gfa supplemented with an adaptation of the eo procedure to our system . in section 3
we discuss our alternative procedure and in section 4 we show the results of the transient dynamics up to the asymptotic states for the overlap and the correlation function between two consecutive times .
we conclude with a summary and a further discussion in section 5 .
we consider a network of @xmath18 ising neurons in a microscopic state @xmath19 , at the time step @xmath9 in which each @xmath20 represents an active or inactive neuron , respectively .
the states of all neurons are updated simultaneously at each discrete time step according to the alignment of each spin with its local field @xmath21 following a microscopic stochastic single spin - flip dynamics with transition probability @xmath22 = \frac{1}{2 } \big{\ { } 1+\sigma_i(t+1)\tanh[\beta h_i(t ) ] \big{\ } } \,\,\ , , \label{2}\ ] ] ruled by the noise - control parameter @xmath23 , where @xmath16 is the synaptic noise .
the dynamics is a deterministic one when @xmath4 and fully random when @xmath24 . in the former case ,
@xmath25 $ ] . here
, @xmath26 is an external stimulus and @xmath27 is the synaptic coupling @xmath28 between units @xmath29 and @xmath30 , which is assumed to have a hebbian form when @xmath31 , with a macroscopic set @xmath32 , @xmath33 of @xmath34 independent and identically distributed quenched random patterns .
each @xmath35 with probability @xmath36 and @xmath0 is a variable self - interaction .
the latter plays a crucial role leading either to fixed - point or cyclic behavior .
the dynamical evolution of the system is described by the moment generating functional @xcite @xmath37 \big{\rangle } \nonumber \\ \fl & = & \sum_{\bsigma(0),\dots,\bsigma(t ) } \mathrm{prob}[\bsigma(0),\dots,\bsigma(t ) ] \exp\big[-\rmi \sum_{i } \sum_{s=0}^{t } \psi_{i}(s ) \ , \sigma_{i}(s)\big]\,\ , , \label{4}\end{aligned}\ ] ] for a finite number of time steps @xmath9 , where @xmath38 is a set of auxiliary variables that generates averages of moments of the states and the brackets denote an average over all possible paths of states with probability @xmath39 = p[\bsigma(0 ) ] \prod_{s=0}^{t-1 } \prod_{i } \frac{\exp[\,\beta\,\sigma_{i}(s+1)\ , h_{i}(s)]}{2\,\cosh{[\beta\,h_{i}(s ) ] } } \label{5}\ ] ] that follows from ( 2 ) . assuming that for @xmath40 only the statistical properties of the stored patterns will influence the macroscopic behavior , one obtains the relevant quantities which are the overlap @xmath41 of @xmath42 with the condensed pattern @xmath43 , say , the two - time correlation function @xmath44 and the response function @xmath45 , given by @xmath46 @xmath47 and @xmath48 where the bar denotes the configurational average with the non - condensed patters @xmath49 , @xmath50 , and the restriction @xmath51 is due to causality . calling @xmath52 ,
the fraction of flipping spins between two consecutive times becomes @xmath53/2 $ ] .
following the now standard procedure in which the disorder average is done before the sum over the neuron states one obtains exactly , in the large @xmath18 limit , the generating functional @xcite @xmath54 \exp\big{[}-\rmi\sum_i \sum_{s=0}^{t } \psi_i(s ) \ , \sigma_i(s ) \big { ] } \nonumber \\
\prod_{i } \prod_{s < t } \big{\ { } \int \rmd h_i(s ) \delta[h_i(s)-h_{i}^{\mathrm{eff}}(s)]\ , \mathrm{w}[\sigma_i(s+1)|h_i(s ) ] \big{\ } } \big{\rangle}_{\{\phi_i(s)\ } } \label{9}\end{aligned}\ ] ] in which @xmath55= \prod_{i}p[\sigma_{i}(0)]$ ] is the probability of the initial microscopic configuration while @xmath56 denotes an average over the correlated gaussian random variables @xmath57 , with zero - average and a correlation matrix given below .
these random variables turn out to be uncorrelated on different sites and one is left with a single - site effective theory in which a neuron evolves in time according to the probability @xmath58= \frac{1}{2 } \big\ { 1 + \sigma(t+1 ) \tanh{\big [ \beta h^{\mathrm{eff}}(t ) \big ] } \big\}\,\ , , \label{10}\ ] ] with an effective local field given by @xmath59 where we dropped the label of the condensed pattern and assumed that @xmath60 .
the two non - trivial contributions to the effective local field for @xmath61 come from a retarded self - interaction involving the matrix elements @xmath62_{t , t^{\prime } } \label{12}\ ] ] and the zero - average temporarily correlated gaussian noise @xmath63 with variance @xmath64_{t , t^{\prime}}\,\,.\\ \label{13}\ ] ] both these terms account for memory effects in the network .
the generating functional ( [ 9 ] ) is the functional of eissfeller and opper that applies to little s model and the specific algorithm that implements the numerical simulations is described in the literature @xcite .
the dynamics of each of the macroscopic quantities , given by ( [ 6])-([8 ] ) , is obtained from the statistics of the effective single neuron process through the average @xmath65 where @xmath66 and @xmath67 are now single - site vectors that follow a path in discrete times , and @xmath68 \prod_{s
< t } \mathrm{w}[\sigma(s+1)|h^{\mathrm{eff}}(s ) ] \,\ , \label{15 } \\\ ] ] is the single - spin path probability given the gaussian noise @xmath69 in the effective field , with a distribution @xmath70 the macroscopic parameters ( overlap , correlation and response function ) can , in principle , be calculated in closed form for any time step @xmath9 but that becomes non - practical since it requires an increasingly large number of macroscopic quantities .
thus , analytic calculations are only feasible for the first few time steps and , as it turns out , also for the asymptotic stationary state , albeit under some conditions which are not always fulfilled . in order to obtain the full dynamic description of the transients , we make use of the procedure of eissfeller and opper in which the effective single - site dynamics given by ( [ 10])-([13 ] ) is simulated by a monte - carlo method
. there are no finite - size effects , but a large number @xmath71 of stochastic trajectories has to be generated for the single - site process and the macroscopic parameters can be obtained from the average @xmath72 where @xmath73 denotes the set of spins along the path @xmath74 . the number of stochastic trajectories @xmath71 should not be confused with the number of neurons @xmath18 , which goes to infinity . to keep the numerical errors small , a sufficiently large @xmath71 must be used .
the alternative procedure consists in doing analytically the sum over the microscopic paths ( states in discrete times ) @xmath75 , to start with , in order to have a much smaller set of parameters to iterate in the numerical calculations . due to the presence of memory effects in the terms involving the retarded self - interaction @xmath76 , given by (
[ 12 ] ) , which relate the state of the system at time @xmath9 to that at all previous times , the full elimination of the microscopic parameters becomes a formidable task . in the following we consider the simplest approximation that consists in assuming that @xmath77 , which is the so - called signal - to - noise approximation @xcite .
this implies , in turn , that the variance of the gaussian correlated noise becomes @xmath78 according to ( [ 13 ] ) .
with that approximation , the effective local field with the time step denoted as a subindex , @xmath79 has still a stochastic gaussian noise for non - zero @xmath11 and a dependence on the spin variable @xmath80 at the same time .
we assume an initial distribution @xmath81 and proceed with the summation over states in ( [ 14 ] ) , which we denote as @xmath82 the summation can be done by inserting the following integral representation @xmath83 into ( [ 15 ] ) , with the effective local field given by ( [ campoefetivo1 ] ) , in order to separate the dependence on the states at two consecutive times .
the result , substituted in ( [ defg ] ) , yields @xmath84 } \sum_{\sigma_0 \dots \sigma_t } p(\sigma_{0 } ) f(\bsigma ) \exp\big{(}{\rmi \sum_{s=0}^{t-1 } x_s \sigma_{s+1 } } \big { ) } \nonumber \\
\prod_{s=0}^{t-1 } \big [ \cos{x_s } - \rmi \tanh{(\beta h_s ) } \sin{x_s } \big ] \,\,\ , .
\label{eqaux}\end{aligned}\ ] ] since the argument of the exponential is linear with respect to the set @xmath85 , we can now do the sum over the states separately for each time in order to obtain the form of the function @xmath86 in the case of the overlap and the two - time correlation function , with @xmath87 given by @xmath88 and @xmath89 ( @xmath90 ) , respectively .
the drawback is the appearance of @xmath9 multiple integrals over the full set @xmath91 , which can ultimately be reduced to a pair of integrals over @xmath92 and @xmath93 involving elements of either of the matrices @xmath94 in the representation @xmath95 where @xmath96 \\ 0 & \frac{1}{2 } [ \tanh{\beta(\xi m_{s } + j_{0 } + \sqrt{\alpha } \phi_{s } ) } - \tanh{\beta(\xi m_{s } - j_{0 } + \sqrt{\alpha } \phi_{s } ) } ] \end{array } \right ) \,\,,\ ] ] @xmath97 \\ i & \frac{1}{2 } [ \tanh{\beta(\xi m_{s } + j_{0 } + \sqrt{\alpha } \phi_{s } ) } + \tanh{\beta(\xi m_{s } - j_{0 } + \sqrt{\alpha } \phi_{s } ) } ] \end{array } \right ) \,\,.\ ] ] performing the discrete average over the pattern @xmath98 , we obtain the explicit expressions for the macroscopic parameters , that is , for the overlap @xmath99 u_{2\,2 } + \rmi \ ,
u_{1\,2 } \big{>}_{\bphi}\,\ , , \label{overlapeq}\end{aligned}\ ] ] for the correlation with the initial state @xmath100 u_{2\,2 } + \rmi \ , \,m_{0}\,u_{1\,2 } \big{>}_{\bphi}\,\ , , \label{correq}\end{aligned}\ ] ] and for the correlation with the states at other times @xmath101 , @xmath102 v_{2\,2 } + \rmi \ , v_{1\,2 } \big{>}_{\bphi}\,\ , , \label{correqa}\end{aligned}\ ] ] where @xmath103 and @xmath104 are the elements of the matrices @xmath105 and @xmath106 .
finally , for the binary spins in this work , the diagonal elements are @xmath107 . here , @xmath7 and @xmath108 are the initial overlap and gaussian noise , respectively , and the full correlation function is needed to calculate the average @xmath109 note that ( [ overlapeq])-([correqa ] ) are a set of coupled equations when @xmath110 involving the matrix @xmath111 and the overlap at previous time steps through the matrices @xmath105 and @xmath106 .
the remaining gaussian averages in the above equations have to be performed numerically replacing the integral over @xmath69 by a discrete average over a large number @xmath112 of randomly generated functions of the set of correlated gaussian variables @xmath113 , @xmath114 , at each time step @xmath9 , with mean zero and variance given by the correlation matrix @xmath111 . before presenting our numerical results
we discuss briefly the case where @xmath5 .
thus , in the absence of stochastic noise , the alternative procedure becomes exact and the average over @xmath69 drops out in ( [ overlapeq])-([correqa ] ) leading to the following two recursion relations , @xmath115 and @xmath116 for @xmath117 . setting @xmath118 and noting that @xmath107 , we get the two - consecutive time correlation function that yields the spin - flip order parameter @xmath119 satisfying the equation @xmath120 thus , ( [ overlaprecurr ] ) and ( [ q_0recurr ] ) are extensions for all @xmath16 of the equations obtained before for @xmath4 @xcite .
the asymptotic stationary order parameters are given by the fixed - point solutions of these equations , which become @xcite @xmath121 and @xmath122 in the @xmath4 limit , @xmath123 thus , at @xmath4 and @xmath5 , the system appears in a frozen - in state either in the initial overlap when @xmath124 or in a cycle when @xmath125 , in the first case , or else it flows to the retrieval state in the second case .
it is in the first case that we are interested in what follows . clearly , in that case the system is no longer useful as an associative memory but it could again become useful in the presence of synaptic and/or stochastic noise and it would be interesting to see how this occurs and we discuss that next .
when the dynamics is not too slow one may employ the procedure of eo , but otherwise we have to resort to our alternative approximate dynamics .
we start with the results for @xmath5 and any @xmath16 , for which the alternative procedure is exact . in figure
[ diagram ] we show the @xmath126 phase diagram of stationary states for various initial overlaps , as indicated , obtained by means of the iteration of ( [ overlaprecurr ] ) .
we also analyzed the stationary states of the correlation function between two consecutive times , by means of ( [ q_0recurr ] ) .
there is a paramagnetic phase ( p ) to the left of the curves with @xmath127 and @xmath128 for all @xmath16 , except at @xmath4 , where there is a state of frozen - in cycles with overlap @xmath129 and @xmath130 .
the network evolves to a retrieval fixed - point within the phase r for all @xmath131 , with an asymptotic overlap @xmath132 for low @xmath16 and @xmath133 . for @xmath4 ,
there is either a state of frozen - in fixed - points in phase r with @xmath134 and @xmath135 , if @xmath136 , or a retrieval state with @xmath137 and @xmath135 , if @xmath138 , in accordance with ( [ 33 ] ) .
the phase boundary obtained in the case of the initial overlap @xmath139 is precisely the same as that obtained from the equality of the free energies in the equilibrium analysis , where the appearance of a tricritical point and other features have been discussed @xcite . with initial overlaps @xmath140 and @xmath141 , from left to
right . ] in order to illustrate the instability to synaptic noise of the frozen - in fixed point in phase r , when @xmath5 , and the interesting transient crossover to retrieval behavior , we show in figure [ dinamalpha0retr ] the dynamical evolution of the overlap and of the two - time correlation function for @xmath142 , @xmath143 and initial overlap @xmath144 .
the results were obtained by means of the alternative procedure and , for comparison , we also show the results with the eo procedure . for these values of the parameters ,
well within the domain where @xmath136 already for a moderate value of @xmath0 , a large number of time steps is needed in order to reach the asymptotic behavior and even longer times are necessary for larger values of @xmath0 and/or smaller values of @xmath16 .
in contrast , the eo procedure yields inconclusive results within reasonable computing time , that could lead to wrong conclusions . with the exception of the dip around @xmath145 , which is an indication of crossover behavior in the overlap , the consecutive - time correlation function @xmath146 is very close to but not exactly equal to one through all the dynamics , which indicates that there is almost always a very small fraction of flipping spins , except at the crossover where that fraction is about five percent over a few number of steps . , @xmath147 , @xmath148 and @xmath5 .
superimposed are the results with the eo procedure ( dark circles).,title="fig : " ] , @xmath147 , @xmath148 and @xmath5 .
superimposed are the results with the eo procedure ( dark circles).,title="fig : " ] we consider now the instability to synaptic noise of the frozen - in cycles of period two in phase p , when @xmath5 .
the results obtained with the alternative approach and , for comparison with the eo procedure , are shown in figure [ dinamalpha0ciclo ] for @xmath149 , @xmath143 and initial overlap @xmath144 .
the oscillating overlap decreases continuously to zero with increasing @xmath9 , now with the parameters in the domain where @xmath150 and @xmath125 and , again , a large number of time steps is needed in order to reach a vanishing overlap characteristic of a paramagnetic phase .
the two - time correlation function @xmath146 is very slightly larger than @xmath151 through all the dynamics , which indicates the permanent presence of a small fraction of units that remains frozen between two consecutive times , causing a continuous decrease in the amplitude of the cycles as time evolves .
it turns out that the decrease is slower for larger values of @xmath15 and/or smaller values of @xmath16 .
we remark , again , that the convergence to the asymptotic behavior is rather slow and that results based on a much shorter time scale could lead to the wrong conclusion that there are cycles for finite @xmath16 . , @xmath152 , @xmath148 and @xmath5 .
superimposed are the results with the eo procedure ( dark circles ) . ]
consider next the situation for @xmath6 in the absence of synaptic noise ( @xmath4 ) and @xmath61 . in this case
we run into difficulties with the eo procedure due to the vanishing denominator in ( [ 16 ] ) after a relatively small number of time - steps , and the alternative dynamics turned out to be not a good approximation in this situation .
nevertheless , one can still draw conclusions from the calculations for a small number of time steps using the eo procedure , as we show next . in figure [ dinamalphafinretra ]
we illustrate the evolution of the overlap and of the correlation function @xmath146 for @xmath153 , @xmath154 and @xmath148 , within the domain where @xmath136 .
there is a transient crossover from a frozen - in to a retrieval state which takes place by means of an increase in the fraction of flipping spins , shown by the dip in @xmath146 , similar to the behavior discussed above .
the correlation function between consecutive time steps is very close to but smaller than one through all the dynamics , indicating a very small but finite fraction of flipping spins outside the crossover region .
it takes a longer time interval to reach the crossover the bigger the value of @xmath0 and/or the smaller the value of @xmath11 . for values of @xmath0 closer to @xmath7
the transient crossover takes place for even smaller values of @xmath11 , which suggests that this behavior should occur for any @xmath11 .
this , and the result obtained above , illustrates the interesting feature of emergence of retrieval behavior in the presence of a small amount of either synaptic or stochastic noise , just enough to draw the network from the frozen - in state . for a larger @xmath155 and the other parameters remaining the same
, we still find that @xmath156 through all the dynamics , but after an increase over the first few time steps the overlap follows a very slow decrease as time evolves until a remanent value is attained , which is characteristic of spin - glass behavior @xcite .
in figure [ dinamalphafinretrb ] we exhibit the dynamical behavior for @xmath149 , @xmath157 and @xmath148 , in which the amplitude of the oscillating overlap decreases rapidly and goes to zero . at the same time
, the correlation function @xmath146 decays to a value very close to @xmath151 indicating an almost full spin flip already after seven time steps . for smaller values of @xmath15 and @xmath11 , within the domain where @xmath158 ,
a similar behavior is reached within a shorter number of time steps .
this behavior is reminiscent of the paramagnetic phase in the presence of synaptic noise discussed above when @xmath5 . a stationary state with vanishing overlap for @xmath4 , @xmath159 and finite @xmath11 is in qualitative agreement with a paramagnetic phase found in the equilibrium replica symmetric approach @xcite .
it is worth noting that a paramagnetic state has also been found in an asymmetric extremely diluted synchronous network in the presence of stochastic noise at @xmath4 @xcite and @xmath160 .
we remind the reader , however , that we are dealing here with a fully connected network with symmetric interactions .
in contrast , for a slightly larger value of @xmath11 , and the other parameters being the same , @xmath146 no longer becomes @xmath151 and , instead , evolves towards a positive stationary value , indicating a partial spin flip , together with an oscillating overlap that decreases slowly until a finite remanent value is reached , characteristic of spin - glass behavior .
since our results suggest that either synaptic or stochastic noise , with finite @xmath16 or @xmath11 respectively , play a similar role in the dynamics when @xmath161 in drawing the network from the frozen - in states , one can expect that this will continue to be the case in the presence of both kinds of noise .
we checked this explicitly for @xmath144 , @xmath143 , @xmath162 and two values of @xmath0 .
as expected , for @xmath142 we get the transient crossover from a frozen - in state to retrieval behavior and , for @xmath149 , we find a continuous decreasing amplitude of the oscillating overlap as the time evolves , until a vanishingly overlap is reached . in both cases
we obtain @xmath163 through the dynamics , indicating either a small but finite fraction of flipping units or an almost full spin flip , for @xmath164 or @xmath165 , respectively .
due to the small values of @xmath16 and @xmath11 , it is necessary to go to larger times than those where the eo procedure is practically applicable , and we resorted for this purpose to the alternative approach which should be a good approximation for the value of @xmath11 used .
in fact , for the smaller times , there is very good agreement between the eo procedure with @xmath166 and the alternative approach with @xmath167 .
we used the numerical simulation procedure of eissfeller and opper and an alternative procedure developed in this work , both based on a generating functional approach , in order to study the effects of synaptic and stochastic noise on the dynamical evolution of little s model of a synchronous neural network with binary neurons and a finite self - interaction @xmath0 .
we analyzed the stability to both kinds of noise of frozen - in fixed points and frozen - in cycles that occupy a large part of the space of parameters and that were found in a previous work @xcite .
this is a crucial issue in the range of parameters when there is a sizeable self - interaction such that @xmath6 , for which there is no guaranty that there will be a flow to a retrieval fixed - point , even if @xmath168 .
that flow may be expected only if @xmath10 .
we found that the frozen - in states that appear at @xmath4 and @xmath5 are unstable to synaptic noise , with or without stochastic noise , leading either to retrieval or to paramagnetic states , for @xmath124 or @xmath125 , respectively .
our work also suggests the instability of the frozen - in states to stochastic noise in the absence of synaptic noise , with the same kind of final states .
this implies the absence of period - two cycles in any finite region of the phase diagram with @xmath16 and/or @xmath11 different of zero .
one has to be certain that the true asymptotic state is reached in the course of the dynamics .
it is in order to deal with very long transients that already appear for moderate values of @xmath15 and low values of synaptic and/or stochastic noise , which are heavily time consuming in the computations with the eo procedure , that we developed our alternative dynamical procedure within the generating functional approach .
the procedure consists in neglecting the retarded self - interaction term in the local field and performing explicitly the summation over states that enters into the calculation of the macroscopic quantities of interest , and it amounts to the numerical solution of a set of equations for the relevant macroscopic parameters that involves a gaussian correlated average , and it becomes exact in the absence of stochastic noise .
thus , we can perform the computations over a much longer number of time steps in a shorter computing time than the procedure of eissfeller and opper .
the procedure also yields results in very good agreement with the eo procedure in the case of small stochastic noise , finite @xmath16 and large values of @xmath17 , within the time scale where both procedures are applicable . finally , some concern about our alternative procedure . although it is exact only in the absence of stochastic noise , that is for @xmath5 , our results for @xmath131 and a rather small @xmath61 should be reliable and it does not seem to be justified at this stage , where we are not concerned with full phase diagrams , to include memory effects in the retarded self - interaction due to the states of the units at other than just the previous time step . on the other hand ,
reasonable calculations can be done with the alternative procedure over longer time scales which would be needed to obtain results for larger ratios of @xmath169 . despite the fact that the use of our alternative dynamics is justified for sufficiently small stochastic noise ( small values of @xmath11 )
, it may be interesting to consider its extension for a larger amount of noise .
even an approximate estimate of the retarded self - interaction @xmath76 , based on the dominating features of the response function @xmath45 , would be useful for that purpose .
a hint in that direction could come from simulations of eissfeller and opper for @xmath45 .
preliminary calculations in that direction , for small non - zero values of @xmath16 and @xmath11 , show that the response function is close to zero for most times , except for the presence of two peaks .
it is worth pointing out that the alternative procedure could be useful in the dynamics of other synchronous neural network models with a finite self - interaction @xmath0 .
the procedure should also be useful for the study of the dynamics of other disordered systems .
furthermore , the procedure can also be applied to neural network models with @xmath170 , including in the memory term of the effective local field ( [ 11 ] ) only the term which relates the state of the system at time @xmath9 to that at time @xmath171 . in this case
, it would be necessary to obtain an equation for the response function .
this , and related issues will be explored in a separate work .
we thank j. f. fontanari for an illuminating discussion .
the work of one of the authors ( wkt ) was financially supported , in part , by cnpq ( conselho nacional de desenvolvimento cientfico e tecnolgico ) , brazil . a grant from fapergs ( fundao de amparo pesquisa
do estado de rio grande do sul ) , brazil , to the same author is gratefully acknowledged .
f. l. metz acknowledges a fellowship from cnpq . | the full dynamics of a synchronous recurrent neural network model with ising binary units and a hebbian learning rule with a finite self - interaction is studied in order to determine the stability to synaptic and stochastic noise of frozen - in states that appear in the absence of both kinds of noise .
both , the numerical simulation procedure of eissfeller and opper and a new alternative procedure that allows to follow the dynamics over larger time scales have been used in this work .
it is shown that synaptic noise destabilizes the frozen - in states and yields either retrieval or paramagnetic states for not too large stochastic noise .
the indications are that the same results may follow in the absence of synaptic noise , for low stochastic noise . |
topological data analysis assigns homological invariants to data presented as a finite metric space ( a `` point cloud '' ) .
if we imagine this data as measurements sampled from some abstract universal space @xmath0 , the structure of that space is a metric measure space , having a notion both of distance between points and a notion of probability for the sampling . a standard homological approach to studying the samples is to assign a simplicial complex and compute its homology .
the construction of the associated simplicial complex for a point cloud depends on a choice of scale parameter .
the insight of `` persistence '' is that one should study homological invariants that encode change across scales ; the correct scale parameter is a priori unknown . as such , a first approach to studying the homology of @xmath0 from the samples is to simply compute the persistent homology @xmath1 of simplicial complexes associated to the sampled point cloud @xmath2 .
we can gain some perspective from imagining that we could make measurements on @xmath0 directly and interpret these measurements in terms of random sample points . with this in mind
, we immediately notice some defects with homology and persistent homology as invariants of @xmath0 . while the homology of @xmath0 captures information about the global topology of the metric space , the probability space structure plays no role .
this has bearing even if we assume @xmath0 is a compact riemannian manifold and the probability measure is the volume measure for the metric : handles which are small represent subsets of low probability but contribute to the homology in the same way as large handles . in this particular kind of example
, persistent homology can identify this type of phenomenon ( by encoding the scales at which homological features exist ) ; however , in a practical context , the metric on the sample may be ad hoc ( e.g. , @xcite ) and less closely related to the probability measure . in this case
, we could have handles that are medium size with respect to the metric but still low probability with respect to the measure .
homology and persistent homology have no mechanism for distinguishing low probability features from high probability features .
a closely related issue is the effect of small amounts of noise ( e.g. , a situation in which a fraction of the samples are corrupted ) .
a small proportion of bad samples can arbitrarily change the persistent homology .
these two kinds of phenomena are linked , insofar as decisions about whether low probability features are noise or not is part of data analysis .
the disconnect with the underlying probability measure presents a significant problem when trying to adapt persistent homology to the setting of _ hypothesis testing _ and
_ confidence intervals_. hypothesis testing involves making quantitative statements about the probability that the persistent homology computed from a sampling from a metric measure space is consistent with ( or refutes ) a hypothesis about the actual persistent homology .
confidence intervals provide a language to understand the variability in estimates introduced by the process of sampling .
because low probability features and a small proportion of bad samples can have a large effect on persistent homology computations , the persistent homology groups make poor test statistics for hypothesis testing and confidence intervals . to obtain useable test statistics ,
we need to develop invariants that better reflect the underlying measure and are less sensitive to large perturbation . to be precise about this
, we use the statistical notion of robustness .
a statistical estimator is _ robust _ when its value can not be arbitrarily perturbed by a constant proportion of bad samples .
for instance , the sample mean is not robust , as a single extremely large sample value can dominate the result . on the other hand ,
the sample median is robust . as we discuss in section [ sec : notrobust ] , persistent homology is not robust
. a small number of bad samples can cause large changes in the persistent homology , essentially as a reflection of the phenomenon of large metric low probability handles ( including spurious ones ) . in order to handle this ,
we adopt a standard statistical perspective , namely that the distribution of an estimator on some fixed finite number of samples is an appropriate way to grapple with such behavior .
to make this precise , we need to be able to talk about probability distributions on homological invariants . using the idea of an underlying metric measure space @xmath0 , formally the process of sampling amounts to considering random variables on the probability space @xmath3 equipped with the product probability measure .
the @xmath4-th persistent homology of a size @xmath5 sample is a random variable on @xmath6 taking values in the set @xmath7 of finite _ barcodes _
@xcite , where a barcode is essentially a multiset of intervals of the form @xmath8 .
the set @xmath7 of barcodes is equipped with a metric @xmath9 , the bottleneck metric @xcite , and we show in section [ sec : polish ] that it is separable and that its completion @xmath10 is also a space of barcodes .
then @xmath10 is polish , i.e. , complete and separable , which makes it amenable to probability theory ( see also @xcite for similar results ) . in particular , various metrics on the set of distributions on @xmath10
metrize weak convergence , including the prohorov metric @xmath11 and the wasserstein metric @xmath12 .
we consider the following probability distribution on barcode space @xmath10 ( restated in section [ sec : distinv ] as definition [ defn : phd ] ) .
[ defn : ph ] for a metric measure space @xmath13 and fixed @xmath14 , define @xmath15 to be the empirical measure induced by the @xmath4th @xmath5-sample persistent homology , i.e. , @xmath16 the probability distribution on the set of barcodes @xmath10 induced by pushforward along @xmath17 from the product measure @xmath18 on @xmath19 . in other words
, @xmath20 is the probability measure on the space of barcodes where the probability of a subset @xmath21 is the probability that a size @xmath5 sample from @xmath0 has @xmath4-th persistent homology landing in @xmath21 .
note that the pushforward makes sense since @xmath17 is a continuous and hence borel measurable function ; see section [ sec : distinv ] for a discussion . although complicated , @xmath22 is a continuous invariant of @xmath0 in the following sense .
the moduli space of metric measure spaces admits a metric ( in fact several ) that combine the ideas of the gromov - hausdorff distance on compact metric spaces and weak convergence of probability measures @xcite .
we follow @xcite , and use the _ gromov - prohorov metric _ , @xmath23 .
we prove the following theorem in section [ sec : distinv ] ( where it is restated as theorem [ thm : main ] ) .
[ thm : stability ] let @xmath13 and @xmath24 be compact metric measure spaces .
then we have the following inequality relating the prohorov and gromov - prohorov metrics : @xmath25 this inequality becomes increasingly tight as the right - hand side approaches @xmath26 ; we discuss precise estimates in section [ sec : distinv ] .
as we explain there , the fact that the bound increases with @xmath5 is expected behavior : @xmath5 should be thought of as a scale parameter , and increasing @xmath5 yields a more sensitive invariant .
the main import of theorem [ thm : stability ] is that for fixed @xmath5 , empirical approximations to @xmath27 computed from subsets @xmath28 are asymptotically convergent as the number of samples increase .
theorem [ thm : stability ] therefore validates computing @xmath20 in practice using empirical approximations , where we are given a large finite sample @xmath29 which we regard as drawn from @xmath0 . making @xmath29 a metric measure space via the subspace metric from @xmath0 and the empirical measure
, we can compute @xmath30 as an approximation to @xmath22 .
this procedure is justified by the fact that as the sample size increases , the empirical metric converges ( in @xmath23 ) to @xmath0 ; see corollary [ cor : largenum ] . in particular , this justifies a resampling procedure to approximate @xmath15 by subsampling from a large sample of size @xmath31 .
( we can also approximate using more sophisticated resampling methodology , a topic we study in future work . ) moreover , as a consequence of the continuity implied by the previous theorem , we can use @xmath20 to develop robust statistics : if we change @xmath0 by adjusting the metric arbitrarily on @xmath32 probability mass to produce @xmath33 , then the gromov - prohorov distance satisfies @xmath34 .
a difficulty with applying @xmath20 is that it can be hard to interpret or summarize the information contained in a distribution of barcodes , unlike distributions of numbers for which there are various moments ( e.g. , the mean and the variance ) which provide concise summaries of the distribution .
one approach is to develop `` topological summarizations '' of distributions of barcodes ; a version of this using frechet means is explored in @xcite .
another possibility is to embed the space of barcodes in a more tractable function space @xcite . in this paper , we instead consider cruder invariants which take values in @xmath35 .
one such invariant is the distance with respect to a reference distribution on barcodes @xmath36 , chosen to represent a hypothesis about the persistent homology of @xmath0 .
[ defn : hd ]
let @xmath13 be a compact metric measure space and let @xmath36 be a fixed reference distribution on @xmath10 .
fix @xmath37 .
define the homological distance on @xmath0 relative to @xmath36 to be @xmath38 we also consider a robust statistic @xmath39 related to @xmath40 without first computing the distribution @xmath20 . to construct @xmath39 ,
we start with a reference barcode and compute the median distance to the barcodes of subsamples .
[ defn : mhd ] let @xmath13 be a compact metric measure space and fix a reference barcode @xmath41 .
fix @xmath42 .
let @xmath43 denote the distribution on @xmath35 induced by applying @xmath44 to the barcode distribution @xmath45 .
define the median homological distance relative to @xmath46 to be @xmath47 the appearance of reference barcodes and distributions in the invariants above raises the question of where one obtains these quantities . as we illustrate below in section [ sec : examples ] , a common source of reference point is simply an a priori hypothesis about the data that we wish to test .
the frechet mean @xcite of a collection of samples provides a more principled approach to producing such reference points .
the use of the median rather than the mean in the preceding definition ensures that we compute a robust statistic , in the following sense .
[ defn : robust ] let @xmath48 be a function from finite metric spaces to a metric space @xmath49 .
we say that @xmath48 is robust with robustness coefficient @xmath50 if for any non - empty finite metric space @xmath51 , there exists a bound @xmath52 such that for any isometric embedding of @xmath0 into a finite metric space @xmath53 , @xmath54 implies @xmath55 , where @xmath56 denotes the number of elements of @xmath0 .
for example , under the analogous definition on finite multi - subsets of @xmath35 ( in place of finite metric spaces ) , median defines a function to @xmath35 that is robust with robustness coefficient @xmath57 for any @xmath32 since expanding a multi - subset @xmath0 to a larger one @xmath33 with fewer than twice as many elements will not change the median by more than the diameter of @xmath0 .
similarly , for a finite metric space @xmath0 , expanding @xmath0 to @xmath33 , the proportion of @xmath5-element samples of @xmath33 which are samples of @xmath0 is @xmath58 ; when this number is more than @xmath59 , the median value of any function @xmath48 on the set of @xmath5-element samples of @xmath33 is then bounded by the values of @xmath48 on @xmath5-element samples of @xmath0 . since @xmath60 for @xmath61 ,
any such function @xmath48 will be robust with robustness coefficient @xmath62 satisfying this bound , and in particular for @xmath63 .
[ thm : ahdrobust ] for any @xmath64 , the function @xmath65 from finite metric spaces ( with the uniform probability measure ) to @xmath35 is robust with robustness coefficient @xmath66 .
the function @xmath20 from finite metric spaces to distributions on @xmath10 and the function @xmath67 from finite metric spaces to @xmath35 are robust for any robustness coefficient for trivial reasons since the gromov - prohorov metric is bounded . however , for these functions we can give explicit uniform estimates for how much these functions change when expanding @xmath0 to @xmath33 just based on @xmath68 .
we introduce the following notion of uniform robustness which is strictly stronger than the notion of robustness .
[ defn : uniformrobust ] let @xmath48 be a function from finite metric spaces to a metric space @xmath49 .
we say that @xmath48 is uniformly robust with robustness coefficient @xmath50 and estimate bound @xmath52 if for any non - empty finite metric space @xmath51 and any isometric embedding of @xmath51 into a finite metric space @xmath53 , @xmath54 implies @xmath55 .
uniform robustness gives a uniform estimate on the change in the function from expanding the finite metric space .
for example , the median function does not satisfy the analogous notion of uniform robustness for functions on finite multi - subsets of @xmath35 .
we show in section [ sec : distinv ] that @xmath20 and @xmath67 satisfy this stronger notion of uniform robustness .
[ thm : hdunifrob ] for fixed @xmath69 , @xmath20 is uniformly robust with robustness coefficient @xmath62 and estimate bound @xmath70 for any @xmath62 . for fixed @xmath64
, @xmath71 is uniformly robust with robustness coefficient @xmath62 and estimate bound @xmath70 for any @xmath62 . as with @xmath20 itself
, the law of large numbers and the convergence implied by theorem [ thm : stability ] tells us that given a sufficiently large finite sample @xmath72 , we can approximate @xmath73 and @xmath74 of the metric measure space @xmath75 in a robust fashion from the persistent homology computations on @xmath29 .
( see corollaries [ cor : largenum ] , [ lem : largenumhd ] , and [ lem : largenumahd ] below . ) in light of the results on robustness and asymptotic convergence , @xmath73 , @xmath74 , and @xmath15 ( as well as various distributional invariants associated to @xmath15 ) provide good test statistics for hypothesis testing . furthermore , one of the benefits of the numerical statistics @xmath73 and @xmath74 is that we can use standard techniques to obtain confidence intervals , which provide a means for understanding the reliability of analyses of data sets .
we discuss hypothesis testing and the construction of confidence intervals in section [ sec : intervals ] , and explore examples in sections [ sec : examples ] and [ sec : notklein ] . in this paper
we primarily focus on analytic methods and monte carlo simulation for obtaining confidence intervals ; however , these statistics are well - suited for the construction of resampling confidence intervals . in a follow - up paper
@xcite we establish the asymptotic consistency of the bootstrap for @xmath73 and @xmath74 .
we regard this paper as a step towards providing a foundation for the integration of standard statistical methodology into computational algebraic topology .
our goal is to provide tools for practical use in topological data analysis .
we have developed an approach to using statistical tools to study persistent homological invariants for metric measure spaces accessed through finite samples .
there are a number of related approaches to studying the statistical properties of persistent homological estimators ; we quickly survey this work .
bubenik @xcite develops statistical inference via an embedding into function spaces called `` persistence landscapes '' , and with various co - authors in @xcite studies an approach using morse theory ( and hence taking advantage of the ambient metric space for smoothing ) .
the work of harer , mileyko , and mukerjee in @xcite parallels the development in section [ sec : polish ] and introduces probability measures on barcode space , and these ideas are developed further ( with turner ) in the context of frechet means as ways of summarizing barcode distributions in @xcite . in another direction ,
there has been a fair amount of work on the topological features of random simplicial complexes and noise due to kahle @xcite as well as adler , bobrowski , borman , subag , and weinberger @xcite .
this work is essential for understanding what persistent homological `` null hypotheses '' look like , and adapted to our setting should inform our statistical inference procedures . finally , there has also been a lot of excellent work arising on studying robustness in the context of understanding distances to measures for point clouds .
this approach was introduced by chazal , cohen - steiner , and merigot in @xcite , and was further developed by caillerie , chazal , dedecker , and michel in @xcite .
the basic idea is that the distribution of distances to a point cloud is a robust invariant of the point cloud ; indeed , this is closely related to the @xmath76 case of our central invariant .
since preservation of explicit distances is a goal of this approach , it is more closely related to rigid geometric inference ( and manifold learning ) than purely topological inference , as in our homological approach .
the authors would like to thank gunnar carlsson and michael lesnick for useful comments , rachel ward for comments on a prior draft , and olena blumberg for help with background and for assistance with the analysis of the tightness of the main theorem .
we would also like to thank the institute for mathematics and its applications for hospitality while revising this paper .
the paper is organized as follows . in section [ sec : background ] , we provide a rapid review of the necessary background on simplicial complexes , persistent homology , and metric measure spaces . in section [ sec :
polish ] , we study the space of barcodes , establishing foundations needed to work with distributions of barcodes . in section [ sec : notrobust ] , we discuss the robustness of persistent homology . in section [ sec : distinv ] , we study the properties of @xmath20 , @xmath74 , and @xmath73 and prove theorem [ thm : stability ] .
we discuss hypothesis testing and confidence intervals in section [ sec : intervals ] , which we illustrate with synthetic examples in section [ sec : examples ] .
section [ sec : notklein ] applies these ideas to the analysis of the natural images data in @xcite .
in this section we provide background for the framework for topological data analysis we study in this paper .
we focus on an approach which accesses the ambient metric measure space @xmath13 only through finite samples , i.e. , point clouds .
( we do not review other interesting approaches , for example , that use the ambient metric space for smoothing the data or for producing a variant simplicial complex @xcite . ) a standard approach in computational algebraic topology proceeds by assigning a simplicial complex ( which usually also depends on a scale parameter @xmath32 ) to a finite metric space @xmath77 .
recall that a simplicial complex is a combinatorial model of a topological space , defined as a collection of nonempty finite sets @xmath78 such that for any set @xmath79 , every nonempty subset of @xmath80 is also in @xmath78 .
associated to such a simplicial complex is the `` geometric realization '' , which is formed by gluing standard simplices of dimension @xmath81 via the subset relations . ( the standard @xmath5-simplex has @xmath82 vertexes . )
the most basic and widely used construction of a simplicial complex associated to a point cloud is the vietoris - rips complex : for @xmath83 , @xmath84 , the vietoris - rips complex @xmath85 is the simplicial complex with vertex set @xmath0 such that @xmath86 $ ] is an @xmath5-simplex when for each pair @xmath87 , the distance @xmath88 .
the vietoris - rips complex is determined by its @xmath89-skeleton .
the construction is functorial in the sense that for a continuous map @xmath90 with lipshitz constant @xmath91 and for @xmath92 , there is a commutative diagram @xmath93 \ar[d ] & \vr_{\kappa \epsilon}(y ) \ar[d ] \\
\vr_{\epsilon'}(x ) \ar[r ] & \vr_{\kappa \epsilon'}(y ) .
\\ } \end{gathered}\ ] ] the vietoris - rips complex is easy to compute , in the sense that is straightforward to determine when a simplex is in the complex .
more closely related to classical constructions in algebraic topology is the cech complex . for @xmath83 , @xmath84
, the cech complex @xmath94 is the simplicial complex with vertex set @xmath0 such that @xmath86 $ ] is an @xmath5-simplex when the intersection @xmath95 is non - empty , where here @xmath96 denotes the @xmath62-ball around @xmath97 .
the cech complex has analogous functoriality properties to the vietoris - rips complex .
the cech complex associated to a cover of a paracompact topological space satisfies the nerve lemma : if the cover consists of contractible spaces such all finite intersections are contractible or empty , the resulting simplicial complex is homotopy equivalent to the original space .
[ rem : witness ] both the vietoris - rips complex and the cech complex can be unmanageably large ; e.g. , for a set of points @xmath98 such that @xmath99 , every subset of @xmath100 specifies a simplex of the vietoris - rips complex . as a consequence ,
it is often very useful to define complexes with the vertices restricted to a small set of landmark points ; the weak witness complex is perhaps the best example of such a simplicial complex @xcite .
we discuss this construction further in section [ sec : notklein ] , as it is important in the applications .
the theory we develop in this paper is relatively insensitive to the specific details of the construction of a simplicial complex associated to a finite metric space ( and scale parameter ) . for reasons that will become evident when we discuss persistence in subsection [ sec : pers ] below
, the main thing we require is a procedure for assigning a complex to @xmath101 that is functorial in the vertical maps of diagram for @xmath102 . in light of the previous subsection ,
given a metric space @xmath51 , one defines the homology at the feature scale @xmath32 to be the homology of a simplicial complex associated to @xmath51 ; e.g. , @xmath103 or @xmath104 .
this latter definition is supported by the following essential consistency result , which is in line with the general philosophy that we are studying an underlying continuous geometric object via finite sets of samples .
let @xmath105 be a compact riemannian manifold equipped with an isometric embedding @xmath106 , and let @xmath107 be a finite independent identically - distributed sample drawn according to the volume measure on @xmath75 .
then for any @xmath108 , there are constants @xmath52 ( which depends on the curvature of @xmath75 and the embedding @xmath109 ) and @xmath110 such that if @xmath111 and @xmath112 then the probability that @xmath113 is an isomorphism is @xmath114 .
in fact , niyogi , smale , and weinberger prove an effective version of the previous result , in the sense that there are explicit numerical bounds dependent on @xmath115 and a `` condition number '' which incorporates data about the curvature of @xmath75 and the twisting of the embedding @xmath109 .
work by latschev provides an equivalent result for @xmath116 , with somewhat worse bounds , defined in terms of the injectivity radius of @xmath75 @xcite .
alternatively , one can show that in the limit @xmath116 captures the homotopy type of the underlying manifold using the fact that there are inclusions @xmath117 while reassuring , an unsatisfactory aspect of the preceding results is the dependence on a priori knowledge of the feature scale @xmath32 and the details of the intrinsic curvature of @xmath75 and the nature of the embedding . a convenient way to handle the fact that it is often hard to know a good choice of @xmath32 at the outset is to consider multi - scale homological invariants that encode the way homology changes as @xmath32 varies .
this leads us to the notion of persistent homology .
persistent homology arose more or less simultaneously and independently in work of robins @xcite , frosini and ferri and collaborators @xcite , and edelsbrunner and collaborators @xcite .
see the excellent survey of edelsbrunner and harer @xcite for a more expansive discussion of the history and development of these ideas .
the efficient algorithms and the algebraic presentation we apply herein is due to @xcite and @xcite . given a diagram of simplicial complexes indexed on @xmath35 , i.e. , a complex @xmath118 for each @xmath119 and maps @xmath120 for @xmath121 , there are natural maps @xmath122 induced by functoriality .
we say that a class @xmath123 is _ born _ at time @xmath124 if it is not in the image of @xmath125 for @xmath126 , and we say a class @xmath127 _ dies _ at time @xmath124 if the image of @xmath128 is 0 in @xmath125 for @xmath129 .
this information about the homology can be packaged up into an algebraic object : let @xmath130 be a diagram of simplicial complexes indexed on @xmath35 .
the @xmath115th persistent @xmath4th homology group of @xmath131 is defined to be @xmath132 where @xmath80 and @xmath46 denote the cycle and boundary groups respectively .
alternatively , @xmath133 is the image of the natural map @xmath134 barcodes provide a convenient reformulation of information from persistent homology .
although we will work over a field and in the presence of suitable finiteness hypotheses which are satisfied in our motivating examples , recent work makes it clear that this restriction could be weakened @xcite .
we assume that the values @xmath135 change only at a countable discrete subset of @xmath35 , so that by reindexing we have a direct system @xmath136 the direct system of simplicial complexes stabilizes at a finite stage and all homology groups are finitely - generated .
then a basic classification result of zomorodian - carlsson @xcite describes the persistent homology in terms of a barcode , a multiset of non - empty intervals of the form @xmath137 .
an interval in the barcode indicates the birth and death of a specific homological feature . for reasons we explain below
, the barcodes appearing in our context will always have finite length intervals .
the vietoris - rips ( or cech ) complexes associated to a point cloud @xmath138 fit into this context by looking at a sequence of varying values of @xmath32 : @xmath139 we can do this in several ways , for example , using the fact that the vietoris - rips complex changes only at discrete points @xmath140 and stabilizes for sufficiently large @xmath32 , or just choosing and fixing a finite sequence @xmath141 independently of @xmath0 .
the theory we present below makes sense for either of these choices , and we use the following notation .
let @xmath142 be a finite metric space
. for @xmath143 , denote the persistent homology of @xmath0 as @xmath144 for some chosen sequence @xmath145 and @xmath146 .
more generally , we can make analogous definitions for any functor @xmath147 where @xmath148 is the category of finite metric spaces and metric maps and @xmath149 denotes the category of simplicial complexes .
we will call such a @xmath150 `` good '' when the homology changes for only finitely values in @xmath35 . in this case
, we can choose the directed system of values of @xmath141 to contain these transition values .
we note that for large values of the parameter @xmath32 , @xmath151 will be contractible .
therefore , if we use the reduced homology group in dimension 0 , we get @xmath152 for all @xmath4 for large @xmath32 .
the barcodes associated to these persistent homologies therefore have only finite length bars . for convenience in computation
, we typically cut off @xmath32 at a moderately high value before this breakdown occurs .
the result is a truncation of the barcode to the cut - off point . by work of gromov ,
the set of isometry classes of compact metric spaces admits a useful metric structure , the gromov - hausdorff metric . for a pair of finite metric spaces
@xmath153 and @xmath154 , the gromov - hausdorff distance is defined as follows : for a compact metric space @xmath155 and closed subsets @xmath156 , the hausdorff distance is defined to be @xmath157 one then defines the gromov - hausdorff distance between @xmath158 and @xmath159 to be @xmath160 where here @xmath161 and @xmath162 are isometric embeddings .
since the topological invariants we are studying ultimately arise from finite metric spaces , a natural question to consider is the degree to which point clouds that are close in the gromov - hausdorff metric have similar homological invariants .
this question does not in general have a good answer in the setting of the homology of the point cloud , but in the context of persistent homology , chazal , et al .
@xcite provide a seminal theorem in this direction that we review as theorem [ thm : ghcompare ] below .
the statement of theorem [ thm : ghcompare ] involves a metric on the set of barcodes called the bottleneck distance and defined as follows .
recall that a barcode @xmath163 is a multiset of non - empty intervals . given two non - empty intervals @xmath164 and @xmath165 , define the distance between them to be @xmath166 we also make the convention @xmath167 for @xmath168 and @xmath169 . for the purposes of the following definition , we define a matching between two barcodes @xmath170 and @xmath171 to be a multi - subset @xmath172 of the underlying set of @xmath173 such that @xmath172 does not contain @xmath174 and each element @xmath175 of @xmath176 occurs as the first coordinate of an element of @xmath172 exactly the number of times ( counted with multiplicity ) of its multiplicity in @xmath176 , and likewise for every element of @xmath177 .
we get a more intuitive but less convenient description of a matching using the decomposition of @xmath178 into its evident four pieces : the basic data of @xmath172 consists of multi - subsets @xmath179 and @xmath180 together with a bijection ( properly accounting for multiplicities ) @xmath181 ; @xmath172 is then the ( disjoint ) union of the graph of @xmath109 viewed as a multi - subset of @xmath182 , the multi - subset @xmath183 of @xmath184 , and the multi - subset @xmath185 of @xmath186 . with this terminology
, we can define the bottleneck distance .
the bottleneck distance between barcodes @xmath187 and @xmath171 is @xmath188 where @xmath172 varies over all matchings between @xmath176 and @xmath177 .
although expressed slightly differently , this agrees with the bottleneck metric as defined in @xcite and @xcite . on the set of barcodes @xmath7 with finitely many finite length intervals , @xmath9 is obviously a metric .
more generally , for any @xmath189 , one can consider the @xmath190 version of this metric , @xmath191 for simplicity , we focus on @xmath9 in this paper , but analogues of our main theorems apply to these variant metrics as well .
we have the following essential stability theorem : [ thm : ghcompare ] for each @xmath4 , we have the bound @xmath192 note that truncating barcodes ( i.e. , truncating each persistent interval ) is a lipshitz map @xmath193 with lipshitz constant @xmath89 , so the bound above still holds when we use a large parameter cut - off in defining @xmath194 . the space of barcodes admits other metrics that are finer than the bottleneck metric for which versions of the stability theorem also hold ; these can be useful in practical situations .
notably , the papers @xcite study and apply a family of wasserstein ( mass transportation ) metrics on barcode space .
we believe that our results can be extended to this setting .
to establish more robust convergence results , we work with suitable metrics on the set of compact metric measure spaces .
specifically , following @xcite we use the idea of the gromov - hausdorff metric to extend certain standard metrics on distributions ( on a fixed metric measure space ) to a metric on the set of all compact metric measure spaces .
a basic metric of this kind is the gromov - prohorov metric @xcite .
( for the following formulas , see section 5 of @xcite and its references . )
this metric is defined in terms of the standard prohorov metric @xmath11 ( metrizing weak convergence of probability distributions on separable metric spaces ) .
first , recall that for measures @xmath195 and @xmath196 on a metric space @xmath80 , the prohorov metric is defined as @xmath197 where @xmath198 varies over all closed sets and @xmath199 is the set of points @xmath200 such that @xmath201 for some @xmath202 .
then the gromov - prohorov metric is defined as @xmath203 where the @xmath204 is computed over all isometric embeddings @xmath205 and @xmath206 into a target metric space @xmath207 .
it is very convenient to reformulate both the gromov - hausdorff and gromov - prohorov distances in terms of relations . for sets @xmath0 and @xmath100 ,
a relation @xmath208 is a correspondence if for each @xmath209 there exists at least one @xmath210 such that @xmath211 and for each @xmath212 there exists at least one @xmath213 such that @xmath214 . for a relation @xmath215 on metric spaces
@xmath138 and @xmath216 , we define the distortion as @xmath217 the gromov - hausdorff distance can be expressed as @xmath218 where we are taking the infimum over all correspondences @xmath219 .
similarly , we can reformulate the prohorov metric as follows .
given two measures @xmath195 and @xmath196 on a metric space @xmath0 , let a coupling of @xmath195 and @xmath196 be a measure @xmath220 on @xmath221 ( with the product metric ) such that @xmath222 and @xmath223 .
then we have @xmath224 then there exists a metric space @xmath80 and embeddings @xmath225 and @xmath226 such that @xmath227 .
this section introduces the spaces of barcodes @xmath228 and @xmath10 used in the distributional invariants @xmath20 of definition [ defn : ph ] . these spaces are complete and separable under the bottleneck metric .
this implies in particular that the prohorov metric on the set of probability measures in @xmath228 or @xmath10 metrizes convergence in probability , which justifies the perspective in the stability theorem [ thm : stability ] and the definition of the invariants @xmath71 in definition [ defn : hd ] .
a barcode is by definition a multi - set of intervals , in our case of the form @xmath8 for @xmath229 .
the set @xmath230 of all intervals of this form is of course in bijective correspondence with a subset of @xmath231 .
a multi - set @xmath21 of intervals is a multi - subset of @xmath230 , which concretely is a function from @xmath230 to the natural numbers @xmath232 which counts the number of multiples of each interval in @xmath21 .
we denote by @xmath233 the cardinality of @xmath21 , which we define as the sum of the values of the function @xmath234 specified by @xmath21 ( if finite , or countably or uncountably infinite , if not ) .
the space @xmath7 of barcodes of the introduction is the set of multi - sets of intervals @xmath21 such that @xmath235 .
we have the following important subsets of @xmath7 .
[ defn : abn ] for @xmath236 , let @xmath228 denote the set of multi - sets of intervals ( in @xmath230 ) @xmath21 with @xmath237 . the main result on @xmath228 is the following theorem , proved below .
( similar results can also be found in @xcite . )
[ thm : abn ] for each @xmath236 , @xmath228 is complete and separable under the bottleneck metric .
since the homology @xmath238 ( with any coefficients ) of any complex with @xmath5 vertices can have rank at most @xmath239 , our persistent homology barcodes will always land in one of the @xmath228 , with @xmath31 depending just on the size of the samples . as we let the size of the samples increase , @xmath31 may increase , and
so it is convenient to have a target independent of the number of samples .
the space @xmath240 is clearly not complete under the bottleneck metric ( consider a sequence of barcodes @xmath241 such that @xmath242 is produced from @xmath243 by adding a bar @xmath244 ) , so we introduce the following space of barcodes @xmath10 . [ defn : abc ] let @xmath10 be the space of multi - sets @xmath21 of intervals ( in @xmath230 ) with the property that for every @xmath245 , the multi - subset of @xmath21 of those intervals of length more than @xmath32 has finite cardinality .
clearly barcodes in @xmath10 have at most countable cardinality , and the bottleneck metric extends to a pseudo - metric @xmath246 .
the following lemma shows it is a metric .
[ lem : zeromet ] for @xmath247 , @xmath248 only if @xmath249 .
let @xmath247 with @xmath248 and assume without loss of generality that @xmath0 is not in @xmath228 for any @xmath31 .
then the possible distinct lengths of intervals in @xmath0 or @xmath100 form a countable set @xmath250 .
let @xmath251 and @xmath252 denote the multisubsets of @xmath0 and @xmath100 consisting of the intervals of length exactly @xmath253 .
let @xmath254 and in general let @xmath255 ( with each @xmath256 ) . for any @xmath5 and any @xmath257 , any matching @xmath172 of @xmath0 and @xmath100 with @xmath258 must induce a bijection between @xmath251 and @xmath252 for all @xmath259 ; moreover , if @xmath260 denotes the restriction of @xmath172 to a matching of @xmath251 and @xmath252 , @xmath261 . letting @xmath32 go to zero , we see that @xmath262 for all @xmath124 and that @xmath249 .
lemma [ lem : zeromet ] implies that @xmath9 extends to a metric on @xmath10 .
we prove the following theorem .
[ thm : abc ] @xmath10 is the completion of @xmath263 in the bottleneck metric . in particular
@xmath10 is complete and separable in the bottleneck metric .
the multi - sets of intervals with rational endpoints provides a countable dense subset for @xmath228 . to see that @xmath7 is dense in @xmath10 , given @xmath21 in @xmath10 and @xmath264 , let @xmath265 be the multi - subset of @xmath21 of those intervals of length @xmath266 . then by definition of @xmath10
, @xmath265 is in @xmath7 , and by definition of the bottleneck metric , using the matching coming from the inclusion of @xmath265 in @xmath21 , we have that @xmath267 it just remains to prove completeness of @xmath228 and @xmath10 . for this , given a cauchy sequence @xmath268 in @xmath7 it suffices to show that @xmath269 converges to an element @xmath0 in @xmath10 and that @xmath0 is in @xmath228 if all the @xmath269 are in @xmath228 .
let @xmath268 be a cauchy sequence in @xmath7 . by passing to a subsequence if necessary ,
we can assume without loss of generality that for @xmath270 , @xmath271 .
for each @xmath5 , we have @xmath272 ; choose a matching @xmath273 such that @xmath274 for all @xmath275 .
for each @xmath5 , define a finite sequence of intervals @xmath276 ,
, @xmath277 inductively as follows .
let @xmath278 .
let @xmath279 be the cardinality of the multi - subset of @xmath158 consisting of those intervals of length @xmath280 , and let @xmath281 ,
, @xmath282 be an enumeration of those intervals . by induction , @xmath276 ,
, @xmath277 is an enumeration of the intervals in @xmath269 of length @xmath283 such that for @xmath284 , the intervals @xmath285 and @xmath286 correspond under the matching @xmath287 .
for the inductive step , we note that if @xmath286 corresponds to @xmath288 under @xmath273 , then @xmath289 , so the length @xmath290 of @xmath288 is bigger than @xmath291 , and @xmath292 thus , we can choose @xmath293 to be the corresponding interval @xmath288 for @xmath294 , and we can choose the remaining intervals of length @xmath295 in an arbitrary order .
write @xmath296 and let @xmath297 since @xmath298 and @xmath299 , we have @xmath300 let @xmath0 be the multi - subset of @xmath230 consisting of the intervals @xmath301 for all @xmath124 ( or for all @xmath302 if @xmath303 is bounded ) .
first , we claim that @xmath0 is in @xmath10 .
given @xmath264 , choose @xmath31 large enough that @xmath304 . then for @xmath305 , the interval @xmath306 first appears in @xmath307 for some @xmath308 . looking at the matchings @xmath309 , ,
@xmath310 , we get a composite matching @xmath311 between @xmath312 and @xmath307 .
since each @xmath273 satisfied the bound @xmath313 , the matching @xmath311 must satisfy the bound @xmath314 since all intervals of length @xmath315 in @xmath312 appear as an @xmath316 , we must have that the length of @xmath317 in @xmath307 must be less than @xmath318 since each endpoint in @xmath306 differs from the endpoint of @xmath317 by at most @xmath319 , the length of @xmath306 can be at most @xmath320 thus , the cardinality of the multi - subset of @xmath0 of those intervals of length @xmath266 is at most @xmath321 .
next we claim that @xmath268 converges to @xmath0 .
we have a matching of @xmath269 with @xmath0 given by matching the intervals @xmath276 , , @xmath277 in @xmath269 with the corresponding intervals @xmath322 , , @xmath323 in @xmath0 .
our estimates above for @xmath324 and @xmath325 show that @xmath326 . by construction ,
each leftover interval in @xmath269 has length @xmath327 and the previous paragraph shows that each leftover interval in @xmath0 has length @xmath328 .
thus , @xmath329
. finally we note that if each @xmath269 is in @xmath228 for fixed @xmath31 , then each @xmath330 and so @xmath0 is in @xmath228 .
inevitably physical measurements will result in bad samples . as a consequence , we are interested in invariants which have limited sensitivity to a small proportion of arbitrarily bad samples .
many standard invariants not only have high sensitivity to a small proportion of bad samples , but in fact have high sensitivity to a small number of bad samples .
we do not claim particular novelty for the general nature of the results of this section , as these issues have been folklore for some time .
however , we do not know any place in the literature where precise statements are written down .
we use the following terminology to describe the instability of these invariants .
[ defn : unrobust ] a function @xmath48 from the set of finite metric spaces to @xmath35 is _ fragile _ if it not robust ( in the sense of definition [ defn : robust ] ) for any robustness coefficient @xmath331 . in some cases , an even stronger kind of sensitivity holds .
[ defn : notrobust ] a function @xmath48 from the set of finite metric spaces to @xmath35 is _ extremely fragile _ if there exists a constant @xmath4 such that for every non - empty finite metric space @xmath0 and constant @xmath31 there exists a metric space @xmath33 and an isometry @xmath332 such that @xmath333 and @xmath334 .
informally , extremely fragile in definition [ defn : notrobust ] means that adding a small constant number of points to any metric space can arbitrarily change the value of the invariant .
in particular , an extremely fragile function is fragile , but extremely fragile is much more unstable than just failing to be robust ( note the quantifier on the space @xmath0 ) . as we indicated in the introduction , gromov - hausdorff distance is fragile ; here we show it is extremely fragile .
let @xmath335 be a non - empty finite metric space .
the function @xmath336 is extremely fragile .
given @xmath337 , consider the space @xmath33 which is defined as a set to be the disjoint union of @xmath0 with a new point @xmath338 , and made a metric space by setting @xmath339 where @xmath340 .
we claim @xmath341 given any metric space @xmath342 and isometries @xmath343 , @xmath344 , we need to show that @xmath345 .
we have two cases .
first , if no point @xmath200 of @xmath80 has @xmath346 , then we have @xmath345 . on the other hand ,
if some point @xmath200 of @xmath80 has @xmath347 , then every point @xmath200 in @xmath80 satisfies @xmath348 . choosing some @xmath97 in @xmath0 , we see that for every @xmath200 in @xmath80 , @xmath349 .
it follows that @xmath350 the homology and persistent homology of a point cloud turns out to be a somewhat less sensitive invariant .
nonetheless , a similar kind of problem can occur .
it is instructive to consider the case of @xmath351 or @xmath352 . by adding @xmath353 points far from the original metric space @xmath0 ,
one can change either @xmath351 or @xmath352 by rank @xmath353 .
the further the distance of the points , the longer the additional bars in the barcode and we see for example that the distance @xmath44 in the bottleneck metric from any fixed barcode @xmath46 is a extremely fragile function .
( if we are truncating the barcodes , @xmath9 is bounded by the length of the interval we are considering , so technically is robust , but not in a meaningful way . )
we can also consider the rank of @xmath354 or of @xmath355 in a range ; here the distortion of the function depends on the number of points , but we see that the function is fragile . for @xmath356 and @xmath17 , @xmath357 ,
the same basic idea obtains : we add small spheres sufficiently far from the core of the points in order to adjust the required homology .
we work this out explicitly for @xmath358 . for each integer @xmath359 and real @xmath360 ,
let the metric circle @xmath361 denote the metric space with @xmath4 points @xmath362 such that @xmath363 for @xmath364 , the vietoris - rips complex associated to @xmath361 is just a collection of disconnected points .
it is clear that as long as @xmath365 , when @xmath366 , @xmath367 has the homotopy type of ( and is in fact homeomorphic to ) a circle .
in fact , we can say something more precise : for @xmath368 the rank of @xmath369 is at least @xmath89 .
consider the map @xmath48 from @xmath370 to the unit disk @xmath371 in @xmath231 that sends @xmath372 to @xmath373 and is linear on each simplex .
the condition @xmath374 precisely ensures that whenever @xmath375 forms a simplex @xmath376 in the vietoris - rips complex , the image vertices @xmath377 lie on an arc of angle @xmath378 on the unit circle , and so @xmath379 in particular lies in an open half plane through the origin .
it follows that the origin @xmath380 is not in the image of any simplex , and @xmath48 defines a map from @xmath370 to the punctured disk @xmath381 .
since @xmath382 , we have the @xmath89-cycle @xmath383+\dotsb + [ x_{k-1},x_{k}]+[x_{k},x_{1}]\ ] ] of @xmath370 which maps to a @xmath89-cycle in @xmath384 representing the generator of @xmath385 .
the length @xmath353 and number @xmath386 is arbitrary , so again , we conclude that functions like @xmath387 are extremely fragile .
results for higher dimensions ( using similar standard equidistributed models of @xmath5-spheres ) are completely analogous .
let @xmath46 be a barcode .
the functions @xmath388 from finite metric spaces to @xmath35 are extremely fragile .
in terms of rank , the lemma shows that we can increase the rank of first persistent homology group of a metric space @xmath0 on an interval @xmath389 $ ] by @xmath390 simply by adding extra points .
one can also typically reduce persistent homology intervals by adding points `` in the center '' of the representing cycle .
it is somewhat more complicated to precisely analyze the situation , so we give a representative example : suppose the cycle is represented by a collection of points @xmath362 such that the maximum distance @xmath391 .
then adding a point which is a distance @xmath52 from each of the other points reduces the lifetime of that cycle to @xmath52 .
in any case , the results of the lemma are sufficient to prove the following proposition .
the function that takes a finite metric space to the rank of @xmath194 on a fixed interval @xmath389 $ ] is fragile .
these computations suggest a problem with the stability of the usual invariants of computational topology .
a small number of bad samples can lead to arbitrary changes in these invariants .
fix a good functorial assignment of a simplicial complex to a finite metric space and a scale parameter @xmath32 . recall that we write @xmath17 of a finite metric space to denote the persistent homology of the associated direct system of complexes .
motivated by the concerns of the preceding section , we define @xmath15 as the distribution of barcodes induced by samples of size @xmath5 .
the basic idea motivating @xmath15 is that in order to obtain robust invariants , given a sample budget of @xmath31 samples from @xmath13 , instead of computing a single estimator from the @xmath31 samples it is preferable to look at _ the distribution _ of estimators produced by blocks of samples of size @xmath392 .
note that this is closely related to the idea behind bootstrap resampling .
it is also a more sophisticated version of computing a trimmed mean ( i.e. , a mean in which extremal samples are thrown out ) rather than removing extremal samples , we simply subsample at a rate such that the likelihood of seeing a bad sample is low .
ideally , this approach retains the information contained in those samples while also estimating the `` true '' value .
[ defn : phd ] for a metric measure space @xmath13 and fixed @xmath14 , define the @xmath4th @xmath5-sample persistent homology as @xmath16 the probability distribution on @xmath10 induced by pushforward along @xmath17 from the product measure @xmath18 on @xmath19 .
this definition makes sense because @xmath17 is a continuous function and the measures on the domain and codomain are both borel .
indeed , the stability theorem of chazal , et .
@xcite ( theorem [ thm : ghcompare ] above ) and the fact that the gromov hausdorff metric is less than or equal to the product metric in @xmath6 implies that @xmath17 is lipschitz with lipschitz constant at most 1 . in order to apply @xmath15
, we need to know two things .
first , that for fixed @xmath5 and @xmath4 the approximation to @xmath15 computed by choosing samples from the empirical measure on a large sample space @xmath29 drawn from @xmath13 converges in probability to the actual value ( as @xmath393 goes to infinity ) .
second , that for fixed @xmath5 and @xmath4 the approximation to @xmath15 obtained by computing the empirical measure from @xmath353 blocks of @xmath5 samples converges in probability to the actual value ( as @xmath353 goes to infinity ) . the latter follows from the weak law of large numbers for the empirical process .
the goal of this section is to prove the following theorem , which establishes the former asymptotic consistency .
for this ( and in the remainder of the section ) , we assume that we are computing @xmath394 using the vietoris - rips complex .
[ thm : main ] let @xmath13 and @xmath395 be compact metric measure spaces
. then we have the following inequality : @xmath396 assume that @xmath397 .
then we know that there exist embeddings @xmath398 and @xmath399 into a metric space @xmath80 and a coupling @xmath400 between @xmath401 and @xmath402 such that the probability mass of the set of pairs @xmath403 under @xmath400 such that @xmath404 is less than @xmath32 .
we can regard the restriction of @xmath405 to the full measure subspace @xmath406 of @xmath407 as a probability measure on @xmath408 .
this then induces a coupling between @xmath409 and @xmath410 on @xmath10 , which we now study .
consider @xmath5 samples @xmath411 from @xmath412 drawn according to the product distribution @xmath405 . now consider the probability that @xmath413 the triangle inequality implies that @xmath414 therefore , the union bound implies that the probability that @xmath415 is bounded by @xmath416 next , define a relation @xmath215 that matches @xmath417 and @xmath418 . by definition ,
the distortion of this relation is @xmath419 , and so @xmath420 by the stability theorem of chazal , et .
@xcite ( theorem [ thm : ghcompare ] above ) , this implies that the probability that @xmath421 is bounded by @xmath422 .
this further implies that the probability that @xmath423 is also bounded by @xmath424 .
therefore , we can conclude that @xmath425 we note the dependence on @xmath5 in the statement of the bound in theorem [ thm : main ] .
as @xmath5 increases , the quantity @xmath15 becomes a finer approximation of the persistent homology of the support of @xmath0 .
specifically , more points per sample means that @xmath15 is increasingly sensitive to small features of @xmath0 . in this light , it is not surprising that the bound becomes weaker for larger @xmath5 .
next we discuss the tightness of the bound in theorem [ thm : main ] .
clearly , this bound is vacuous when @xmath426 since the prohorov metric is bounded by @xmath89 , but we show that it becomes tight as @xmath427 approaches zero . reviewing the argument , starting from the hypothesis that@xmath428 , we used the union bound to obtain a bound of @xmath424 .
the exact bound in question is @xmath429 .
the leading term in the expansion of this quantity is @xmath424 , and so as @xmath430 the bound in the theorem becomes increasingly tight .
when @xmath32 is close to @xmath431 , using more terms in the expansion yields better bounds ( for example , when @xmath432 , @xmath433 and tends to @xmath434 for large @xmath5 ) .
the exact bound @xmath435 yields a tight estimate on @xmath436 ( using the @xmath437 product metric ) , as we can see by the following example .
consider the case of two finite metric spaces @xmath438 and @xmath439 , where @xmath440 and @xmath441 .
define @xmath442 via @xmath443 for @xmath444 , @xmath445 for @xmath446 , and @xmath447 for @xmath448 and @xmath449 . here
@xmath109 should be substantially larger than @xmath128 and @xmath450 .
we define @xmath451 analogously , using the same @xmath128 and @xmath450 but with @xmath452 distinct from @xmath109 ( and without loss of generality assume that @xmath453 ) .
consider the metric space @xmath80 formed from the disjoint union of @xmath454 , @xmath455 , and @xmath456 , and with the metric induced from @xmath442 and @xmath451 except that @xmath457 .
there are evident isometries @xmath458 and @xmath459 ; it is easy to see that @xmath460 for @xmath461 and moreover that this pair of embeddings minimizes the prohorov distance , so @xmath462 .
the induced embeddings @xmath463 , @xmath464 satisfy @xmath465 and a straight - forward combinatorics argument shows that this embedding also minimizes the prohorov distance , so @xmath466 .
( we thank olena blumberg for help with this example . ) by varying the parameters in the previous example , it is now clear that the bound on @xmath467 is tight and we can achieve the upper bound with a variety of barcode lengths . tightness for @xmath468 for @xmath469 is harder to analyze .
theorem [ thm : ghcompare ] is expected to be tight for all @xmath4 but nothing has yet appeared in the literature for @xmath469 .
if the bound in theorem [ thm : ghcompare ] is tight , it is reasonable to expect the bound in theorem [ thm : main ] also to be tight ; however , we do not know a rigorous argument . for a related discussion involving finite distance matrices , see ( * ? ? ?
* , 7 ) .
there the constant @xmath31 ( size of the matrix ) is analogous to the parameter @xmath5 above and enters into their formulas through the distance @xmath470 , which depends on @xmath31 .
we regard the bound as most useful for fixed @xmath5
. then a basic consequence of theorem [ thm : main ] is that consideration of large finite samples will suffice for computing @xmath15 . for a finite metric space @xmath138 , let @xmath471 and @xmath472 denote two measures on @xmath0 .
then we have the following inequality @xcite @xmath473 which follows by choosing a coupling that has measure at least the minimum of the two measures on each point .
[ cor : largenum ] let @xmath474 be a sequence of randomly drawn samples from @xmath13 .
we regard @xmath475 as a metric measure space using the subspace metric and the empirical measure .
then @xmath476 converges in probability to @xmath477 .
this result is a consequence of the fact that @xmath478 converges in probability to @xmath13 in the gromov - prohorov metric ( which can be checked directly using equation , as in @xcite , or can be deduced from the analogous convergence result for the gromov - wasserstein distance ( * ? ? ? * ( iii ) ) and the comparison between the gromov - prohorov distance and the gromov - wasserstein distance @xcite ) .
another consequence of theorem [ thm : main ] is that @xmath20 provide robust descriptors for metric measure spaces @xmath13 .
specifically , observe that if we have finite metric spaces @xmath479 and a probability measure @xmath480 on @xmath33 that restricts to @xmath481 on @xmath0 ( i.e. , for @xmath482 , @xmath483 ) , then equation implies that @xmath484 thus , when @xmath485 has probability @xmath486 , @xmath487 in particular , when @xmath0 and @xmath33 are finite metric spaces with the uniform measure , we get @xmath488 as an immediate consequence we obtain the following result .
[ thm : mainunifrob ] for fixed @xmath69 , @xmath20 is uniformly robust with robustness coefficient @xmath62 and estimate bound @xmath70 for any @xmath62
. it would be useful to prove analogues of the main theorem for other methods of assigning complexes ; e.g. , the witness complex ( see remark [ rem : witness ] and section [ sec : notklein ] ) .
we expect that the recent stability results of @xcite will be useful in this connection .
in this section , we describe various ways to use @xmath15 to perform statistical inference about the homological invariants of a point cloud .
the basic goal is to provide quantitative ways of saying what observed barcodes or empirical barcode distributions `` mean '' .
we are predominantly interested in addressing two kinds of questions : 1 .
are two given empirical barcode distributions coming from the same underlying distribution ?
2 . is a particular empirical barcode distribution consistent with the hypothesis that the underlying distribution has @xmath4 `` long bars '' ?
we approach both of these questions from the perspective of classical hypothesis testing , likelihood scores , and confidence intervals ; for example , see @xcite for a review .
we discuss a variety of test statistics derived from @xmath15 ; thus , the use of these procedures are justified in practice by theorem [ thm : main ] ( and specifically corollary [ cor : largenum ] ) .
moreover , we are able to use theorem [ thm : main ] to show that many of the test statistics we describe are robust .
we begin by explaining the basic procedure for computing approximations to @xmath15 .
corollary [ cor : largenum ] justifies the treatment of @xmath15 of the empirical measure on a sufficiently large sample @xmath489 of size @xmath31 as a good approximation for @xmath490 .
( note that the dependence on @xmath5 in the bound in theorem [ thm : main ] implies that we will have to choose @xmath392 in order to expect reasonable results ; see the discussion in the next section for some examples of how to choose @xmath5 . )
next , we can estimate @xmath15 on @xmath29 empirically via monte carlo simulation , i.e. , simply sampling blocks of @xmath5 samples from @xmath29 over and over again .
the weak law of large numbers for empirical distributions guarantees that this estimate converges in probability as the number of such samples increases .
therefore , we have an asymptotically convergent scheme for numerically approximating @xmath15 ( and hence quantities derived from it ) .
we now turn to questions of statistical inference .
the most basic question we can pose is whether a given observed barcode @xmath46 is more consistent with an underlying metric measure space @xmath491 or an alternate metric measure space @xmath492 .
a likelihood ratio provides a good test statistic to determine an answer to this question .
specifically , we can evaluate the likelihood of the hypotheses @xmath493 and @xmath494 that @xmath46 is within @xmath32 of a barcode drawn from @xmath491 and @xmath24 respectively . given an observed barcode @xmath46 ( e.g. , obtained by sampling @xmath5 points from an unknown metric measure space @xmath495 )
, we can compute the likelihood score @xmath496 for each hypothesis space @xmath13 and @xmath24 . the ratio @xmath497 then provides a test statistic for comparing the two hypotheses . to determine how to interpret the test statistics ( e.g. , to compute @xmath115-values ) ,
we require knowledge of the distribution of the test statistic induced by assuming that @xmath46 was drawn from @xmath477 and @xmath498 respectively . these distributions can be approximated by monte carlo simulation , i.e. , repeated sampling from the two distributions and computation of histograms . more generally , for a metric measure space @xmath499 and a particular subset @xmath29 of @xmath10 , we can test the hypothesis that the distribution @xmath22 has mass @xmath500 on @xmath29 as follows . for any hypothetical distribution on @xmath10 with mass @xmath500 on @xmath29 , the probability of an empirical sample of size @xmath31
having @xmath501 or fewer elements in @xmath29 is bounded above by the binomial cumulative distribution function @xmath502 then given an empirical approximation @xmath503 to @xmath20 obtained from @xmath31 samples , we can test the hypothesis that @xmath20 has mass @xmath500 in @xmath29 , by taking @xmath501 to be the number of such elements in @xmath503 .
when @xmath504 , we can reject this hypothesis at the @xmath505 level .
another kind of question we might ask is to determine whether to reject the hypothesis that two empirical distributions on barcode space ( i.e. , @xmath15 computed based on two different large samples @xmath29 and @xmath506 ) came from the same underlying distribution . in our setting
we can not assume very much about the class of possible distributions and so we are forced to rely on non - parametric methods .
this imposes significant constraints most asymptotic results on non - parametric tests for distribution comparison work only for distributions on @xmath35 .
thus , the first step is to project the data from barcode space into @xmath35 .
the following definition is the first of several kinds of projections we discuss .
[ defn : distdistr ]
let @xmath13 be a compact metric measure space .
fix @xmath37 . 1 .
define the distance distribution @xmath507 on @xmath35 to be the distribution on @xmath35 induced by applying @xmath508 to pairs @xmath509 drawn from @xmath510 .
2 . let @xmath46 be a fixed barcode in @xmath10 , and define @xmath511 to be the distribution induced by applying @xmath44 .
since both @xmath512 and @xmath513 are continuous with respect to the gromov - prohorov metric @xcite , corollary [ cor : largenum ] justifies working with empirical approximations to @xmath507 and @xmath511 .
one application of these projections is simply a direct use of the two - sample kolmogorov - smirnov statistic @xcite .
this test statistic gives a way to determine whether two observed empirical distributions were obtained from the same underlying distribution ; the salient feature about this statistic is that for distributions on @xmath35 the @xmath115-values of the test statistic are asymptotically independent of the underlying distribution as long as the samples are identically independently drawn . to compute the kolmogorov - smirnov test statistic for two sets of samples @xmath514 and @xmath515 , we first compute the empirical approximations @xmath516 and @xmath517 to the cumulative density functions , @xmath518 and use the test statistic @xmath519 . in practice , since @xmath520 is large we approximate @xmath521 using monte carlo simulation .
the distribution - independence of the statistic now implies that standard tables ( e.g. , in the appendix to @xcite ) or the built - in matlab functions can then be used to compute @xmath115-values for deciding if the statistic allows us to reject the hypothesis that the distributions are the same .
one might similary consider the mann - whitney test or various other nonparametric techniques for testing the same hypotheses @xcite .
for example , another way to handle this problem is to use a @xmath522 test for discrete distributions .
there are many ways to construct suitable distributions for this test ; we present two natural choices here . 1 .
take histograms from @xmath523 and @xmath524 with identical fixed numbers of bins and bin widths .
2 . fix a finite set @xmath525 of reference barcodes , where @xmath526 .
these reference barcodes should be chosen without reference to the observed data .
next , for each barcode with nonzero probability measure in ( the given empirical approximation to ) @xmath20 , assign the count to the nearest reference barcode .
the second method makes sense if we have a priori information about the expected shape of the barcode distributions .
let @xmath527 denote the count either for bin @xmath528 or for reference barcode @xmath529 in sample @xmath124 ( for @xmath530 ) .
the test statistic in the @xmath531 test for comparing @xmath532 and @xmath533 is then defined to be @xmath534 as the notation suggest , asymptotically this has a @xmath522 distribution with @xmath535 degrees of freedom ( where @xmath536 is the number of reference barcodes with nonzero counts ) @xcite . as such , we can again look up the @xmath115-values for this distribution in standard tables when performing hypothesis testing .
natural test statistics for studying hypotheses about empirical barcode distributions come from numerical summaries associated to @xmath15 .
for instance , a natural test statistic measures the distance to a fixed hypothesis distribution .
[ defn : hdd ] let @xmath13 be a compact metric measure space and let @xmath36 be a fixed reference distribution on @xmath10 . fix @xmath37 .
define the homological distance on @xmath0 relative to @xmath36 to be @xmath38 corollary [ cor : largenum ] again applies to show that large finite samples @xmath489 suffice to approximate @xmath73 .
in fact , the convergence is better since we are working over @xmath35 and the glivenko - cantelli theorem applies .
[ lem : largenumhd ] let @xmath474 be a sequence of randomly drawn samples from @xmath13 .
we regard @xmath475 as a metric measure space using the subspace metric and the empirical measure .
then for @xmath36 a fixed reference distribution on @xmath10 , @xmath537 converges almost surely to @xmath538 .
an immediate consequence of theorem [ thm : main ] is the following robustness result ( paralleling theorem [ thm : mainunifrob ] ) .
[ thm : mainunifrob2 ] for fixed @xmath64 , @xmath71 is uniformly robust with robustness coefficient @xmath62 and estimate bound @xmath70 for any @xmath62 .
another source of tractable test statistics is the moments of the distributions introduced in definition [ defn : distdistr ] .
a virtue of distributions on @xmath35 is that they can be naturally summarized by moments ; in contrast , moments for distributions on barcode space are hard to compute ( for instance , see @xcite ) .
even simply constructing meaningful centroids for a set of points in barcode space is challenging ; for example , geodesics between close points are not unique , although the barcode metric space is a length space ( it is straightforward to construct midpoints between any pair of barcodes ) . because we have emphasized robust statistics , we work with the median or a trimmed mean and introduce the following test statistics : [ defn : ahdbody ]
let @xmath13 be a compact metric measure space and fix a reference barcode @xmath539 .
fix @xmath37 .
define the median homological distance relative to @xmath46 to be @xmath540 for @xmath541 , define the @xmath128-trimmed mean homological distance to be @xmath542 where @xmath501 denotes the quantile function .
( roughly speaking , we discard the fraction @xmath128 of the highest and lowest values and take the mean of the remainder . ) again , corollary [ cor : largenum ] implies that consideration of large finite samples @xmath489 suffices to approximate these test statistics .
[ lem : largenumahd ] let @xmath474 be a sequence of randomly drawn samples from @xmath13 .
we regard @xmath475 as a metric measure space using the subspace metric and the empirical measure .
let @xmath539 be a fixed reference barcode . 1 .
assume that @xmath543 has a distribution function with a positive derivative at the median .
then @xmath544 almost surely converges to @xmath545 .
@xmath546 almost surely converges to @xmath547 . as in the proof of corollary [ cor : largenum ]
, the fact that @xmath478 converges to @xmath13 in the gromov - prohorov metric implies that @xmath548 weakly converges to @xmath43 .
now the central limit theorem for the sample median ( see for instance ( * ? ? ? * iii.4.24 ) ) and the hypothesis about the derivative at the median implies the convergence of medians .
analogously , the central limit theorem for the trimmed mean @xcite ( which holds without further assumption provided that @xmath549 ) gives the second part of the result . the hypothesis on the median is the standard hypothesis for consistency of the central limit theorem ( and the bootstrap estimator for ) the sample median ; it is known that this hypothesis is required @xcite .
although it is our experience that this hypothesis holds in practice , it can be difficult to rigorously verify for an unknown underlying distribution . for this reason , the use of the trimmed mean may be preferable in cases where constraint on the possible hypotheses is unavailable .
as @xmath128 approaches @xmath550 , the trimmed mean converges to the median , and so choosing @xmath551 for small @xmath32 yields a reasonable alternative to the median . as discussed in the introduction , a counting argument yields the following robustness result .
[ thm : ahdrobustbody ] for any @xmath552 , the function @xmath553 from finite metric spaces ( with the uniform probability measure ) to @xmath35 is robust with robustness coefficient @xmath66 . for a finite metric space @xmath0 , expanding @xmath0 to @xmath33 , the proportion of @xmath5-element samples of @xmath33 which are samples of @xmath0 is @xmath58 ; when this number is more than @xmath59 , the median value of any function @xmath48 on the set of @xmath5-element samples of @xmath33 is then bounded by the values of @xmath48 on @xmath5-element samples of @xmath0 . since @xmath60 for @xmath61 ,
any such function @xmath48 will be robust with robustness coefficient @xmath62 satisfying this bound , and in particular for @xmath63 . in order to obtain the @xmath115-value cutoffs for performing hypothesis testing
, we can again use monte carlo simulation to estimate the distribution of these estimators under different hypotheses .
another possibility is to use asymptotic estimates , which we discuss below in the context of confidence intervals .
we close the section by remarking that there are many other possible numerical invariants one might associate to @xmath15 ( and apply as test statistics ) .
for instance , we define for a barcode @xmath46 the quantity @xmath554 where @xmath555 denotes the @xmath124th largest interval in @xmath46 . then the quantity @xmath556 and the related quantity @xmath557 are useful test statistics for determining if there is a group of `` long bars '' in the underlying distribution by checking for which ( if any @xmath390 ) has a large value of @xmath558 . for instance ,
when @xmath559 is small this suggests that the underlying metric measure space is generated by a topological space with no homology in dimension @xmath4 .
and large values of @xmath560 suggest that the underlying space has rank @xmath390 homology in dimension @xmath4 .
of course , in order to make precise statistical statements to replace `` suggests '' , we have to use monte carlo simulation in order to compute @xmath115-values and confidence intervals . for @xmath73 ,
the only way to produce @xmath115-values and confidence intervals is to use monte carlo simulation ( to estimate the distribution of @xmath73 on finite approximations to @xmath561 ) .
a particular advantage of @xmath74 is that we can define confidence intervals using the standard non - parametric techniques for determining confidence intervals for the median and trimmed mean @xcite . for the median ,
we use appropriate sample quantiles ( order statistics ) to determine the bounds for an interval which contains the actual median with confidence @xmath562 .
these confidence intervals then immediately yield cutoffs for @xmath115-values for hypothesis testing .
for example , a simple approximation can be obtained from the fact that order statistics asymptotically obey binomial distributions , which lead to the following definition using the normal approximation to the binomial distribution .
[ defn : confint ] let @xmath491 be a metric measure space and @xmath46 a fixed barcode .
fix @xmath563 and @xmath69 .
given @xmath390 samples from @xmath511 , let @xmath564 denote the samples sorted from smallest to largest .
let @xmath565 denote the @xmath566 significance threshold for a standard normal distribution .
the @xmath567 confidence interval for the sample median ( i.e. , @xmath74 ) is given by the interval @xmath568.\ ] ] for the trimmed mean , the situation is similar : asymptotic confidence intervals can be obtained from the sample standard deviation and an explicit formula @xcite . since we find that the median converges in practice , we do not write out the formula here ( as it involves a number of complicated auxiliary quantities ) and refer the interested reader to the cited reference . in the preceding discussion , the @xmath115-values and confidence intervals for our tests are always computed either via monte carlo simulation ( i.e. , sampling to estimate the distribution of the test statistic ) or using formulas derived from asymptotic results .
the latter are substantially easier and less computationally intensive to apply .
however , we may be concerned about whether sample sizes are large enough for the asymptotic @xmath115-value to be good approximations of the exact @xmath115-value this issue is a pervasive problem when applying such non - parametric tests based on asymptotic results ( e.g. , see @xcite ) .
a standard approach to mitigating such concerns is to perform monte carlo simulation of the distribution of these test statistics computed from representative models for @xmath15 ( e.g. , synthetic distributions generated by various standard manifolds ) ; this provides heuristic guidance about suitable sample sizes .
we provide some example calculations of this form in section [ sec : examples ] below .
however , the careful analyst with access to adequate samples and computer resources may simply choose to rely on monte carlo simulation methods .
when adequate samples are lacking , resampling methods also often provide a more reliable means to compute cutoffs than asymptotic results .
for instance , standard results about the consistency of the bootstrap for the sample median and sample trimmed mean @xcite allow us to compute @xmath115-value thresholds and cofidence intervals for @xmath74 via bootstrap resampling .
in this section , we provide numerical experiments on synthetic data sets to demonstrate the statistical inference procedures and robustness results described in the previous section .
we study a pair of examples embedded in @xmath569 ( an annulus and a pair of nested circles ) and three families of examples in @xmath570 ( spheres , tori , and uniform noise in a box ) .
although the examples embedded in @xmath569 are essentially trivial , the simplicity of the expected results allows us to focus on the methodology .
the examples in @xmath570 are more realistic but correspondingly are more complicated to interpret .
all of our experiments rely on the following procedures for producing empirical approximations to @xmath15 .
we fix a monte carlo parameter @xmath571 which is large ( we discuss estimates of how large @xmath571 needs to be below ) .
we then have the following basic algorithm : [ alg : absolute ] for a fixed metric measure space @xmath13 . 1 .
uniformly select @xmath571 subsamples of size @xmath5 from @xmath572 .
2 . compute the empirical approximation to @xmath15 from the @xmath571 subsamples .
to better represent the use of these procedures in practice , we have the following variant algorithm .
fix a subsample size @xmath31 .
[ alg : relative ] for a fixed metric measure space @xmath13 . 1 . uniformly sample @xmath31 points from @xmath572 .
2 . uniformly select @xmath571 subsamples of size @xmath5 from the empirical measure on the @xmath31 samples .
compute the empirical approximation to @xmath15 from the @xmath571 subsamples . to actually carry out these algorithms
, we used the perseus codebase @xcite to compute persistent homology from a finite metric space , executed from within a series of python and cython scripts that ran our various experimental setups . in order to avoid combinatorial explosion in the number of simplices when the scale parameter results in complete graphs , we typically capped the maximum scale parameter ( i.e. , truncated each of the bars in the barcodes )
the experiments were run on various stock linux machines ; no individual experiment took more than a few minutes to complete .
our random number generation was done using the gsl library @xcite to produce uniform and gaussian samples , and rejection sampling to simulate all other distributions ( as described below ) .
we first consider a simple example which illustrates the robustness of the distributional invariants .
the underlying metric measure space @xmath21 is an annulus of inner radius @xmath573 and outer radius of @xmath574 in @xmath575 ( see figure [ fig : ann ] ) , equipped with the subspace metric and the area measure .
the underlying manifold of @xmath21 is clearly homotopy equivalent to a circle .
we sample from the annulus via rejection sampling ; we sample uniformly from the bounding box @xmath576 \times [ -1.2 , 1.2]$ ] and only keep points @xmath577 such that @xmath578
. we began by examining the rate of convergence in corollary [ cor : largenum ] .
specifically , for @xmath579 and various @xmath5 , we consider subsamples @xmath475 in the annulus of increasing cardinality and study the convergence of various distributions derived from @xmath580 , using algorithms [ alg : absolute ] and [ alg : relative ] as a base .
we compute the distance distribution @xmath512 from @xmath580 and @xmath581 as the cardinality @xmath582 of @xmath475 increases and @xmath5 varies , using a barcode cutoff of @xmath583 .
we then used both the kolmogorov - smirnov test and the @xmath531 test on histograms to repeatedly compare the estimates computed from samples of cardinality @xmath582 to each other and to the reference distribution from @xmath21 .
fixing @xmath584 , our results indicated that @xmath585 sufficed to approximate the distribution for @xmath586 ; with these parameters , we were essentially never able to reject the null hypothesis that the two distributions were drawn from the same underlying distribution . [ cols="^,^,^,^,^,^,^,^",options="header " , ]
one of the most prominent applications of persistent homology in topological data analysis is the study of the natural images dataset described in @xcite .
this data consists of @xmath587 patches sampled from still photographs of `` natural '' scenes ( i.e. , pictures of rural areas without human artifacts ) .
the results of carlsson , ishkhanov , de silva , and zomorodian @xcite extract topological signals from this data set which can be interpreted in terms of collections of patches which are known to be meaningful based on the neurophysiology of the eye .
the goal of this section is to apply our statistical methodology to validate the conclusions of their work .
we compute the confidence intervals based on @xmath74 for a subset of patches from the natural images dataset as described in @xcite .
we briefly review the setup .
the dataset consists of 15000 points in @xmath588 , generated as follows .
from the natural images , @xmath587 patches ( dimensions given in pixels ) were sampled and the top @xmath589 with the highest contrast were retained .
these patches were then normalized twice , first by subtracting the mean intensity and then scaling so that the euclidean norm is @xmath89 .
the resulting dataset can be regarded as living on the surface of an @xmath590 embedded in @xmath588 . after performing density filtering ( with a parameter value of @xmath591 ; refer to @xcite for details ) and randomly selecting @xmath592 points , we are left with the dataset @xmath593 . at this density , one tends to see a barcode corresponding to 5 cycles in the @xmath594 . in the klein bottle model ,
these cycles are generated by three circles , intersecting pairwise at two points ( which can be visualized as unit circles lying on the @xmath595-plane , the @xmath596-plane , and the @xmath597-plane ) .
we computed empirical approximations to @xmath598 using algorithm [ alg : absolute ] , with @xmath584 and using a barcode cutoff of @xmath599 ( we used the value reported in @xcite as the maximal filtration value ) .
we found that ( after applying @xmath4-means clustering , as above ) the weight was distributed as 0.1% with one long bar , 1.1% with two long bars , 7.4% with three long bars , 34.2% with four long bars , and 57.2% with five long bars .
( here the threshold for a long bar was @xmath89 . ) analyzing likelihood test statistics as in section [ sec : like ] , we find that the underlying distribution has at least 95% of its mass on two , three , or four bars at the the 99.7% confidence level .
we also analyze the results using @xmath600 .
we use as the hypothesis barcode the multi - set @xmath601 .
we find using the nonparametric estimate from definition [ defn : confint ] that the 95% confidence interval for @xmath602 is @xmath603 $ ] .
the 99% confidence interval for @xmath604 is @xmath605 $ ] .
these results represents high confidence for the data to be further than @xmath606 but closer than @xmath607 to the reference barcode . on the other hand ,
when we compute the confidence intervals using the reference barcode the empty set , we find that both endpoints for the 95% and 99% confidence intervals are the cutoff value of @xmath599 .
we find the same results for hypothesis barcodes with @xmath353 bars @xmath608 for @xmath609 .
in particular , this means that the differences between the distance to the @xmath610 bar hypothesis and any other is statistically significant at the 99% level .
we interpret these results to suggest that the hypothesis barcode is consistent with the underlying distribution amongst barcode distributions that put all of their mass on a single barcode .
of course , these results also suggest that when sampling at 500 points , we simply do not expect to see a distribution that is heavily concentrated around a single barcode . in the next subsection
, we discuss the use of the witness complex , which does result in such a narrow distribution .
to validate the non - parametric estimate of the confidence interval , we also used bootstrap resampling to compute bootstrap confidence intervals .
although we do not justify or discuss further this procedure herein , we note that we observed the reassuring phenomenon that the bootstrap confidence intervals agreed closely with the non - parametric estimates for both the 95% confidence intervals and the 99% confidence intervals in each instance . because of the size of the datasets involved , the analysis performed in @xcite used the weak witness complex @xmath611 rather than the vietoris - rips complex @xmath612 .
the weak witness complex for a metric space @xmath77 depends on a subset @xmath613 of witnesses ; the size of the complexes is controlled by @xmath614 and not @xmath56 . for @xmath83 , @xmath84 and witness set @xmath615
, the weak witness complex @xmath616 is the simplicial complex with vertex set @xmath617 such that @xmath86 $ ] is an @xmath5-simplex when for each pair @xmath87 , there exists a point @xmath618 ( a witness ) such that the distances @xmath619 .
when working with the witness complex , we adapt our basic approach to study the induced distribution on barcodes which comes from fixing the point cloud and repeatedly sampling a fixed number of witnesses .
the theoretical guarantees we obtained for the vietoris - rips complex in this paper do not apply directly ; we intend to study the robustness and asymptotic behavior of this process in future work . here , we report preliminary numerical results .
specifically , we again computed empirical approximations to @xmath620 using algorithm [ alg : absolute ] , with @xmath621 and using a barcode cutoff of @xmath599 .
however , to produce the underlying complex , we use the @xmath5 points for each subsample as the landmark points @xmath617 in the construction of the witness complex rather than as the vertices for the vietoris - rips complex .
we use as the hypothesis barcode the multi - set @xmath622 as above .
we found using the non - parametric estimate of definition [ defn : confint ] that the 95% confidence interval for @xmath623 , is @xmath624 $ ] . the 99% confidence interval for @xmath623 was also
@xmath625 $ ] . when we computed the 95% confidence interval for @xmath626 we obtained @xmath627 $ ] .
the 99% confidence interval for @xmath628 was @xmath629 $ ] .
this represents high confidence for the data to be further than @xmath630 ( for @xmath631 ) and @xmath632 ( for @xmath633 ) but closer than @xmath632 ( for @xmath631 ) and @xmath634 ( for @xmath633 ) to the reference barcode .
we obtained essentially the same results @xmath635 as for @xmath636 . on the other hand , when using hypothesis barcodes with @xmath353 bars @xmath608 for @xmath609 , the confidence intervals start and end at @xmath599 .
again , this means that the difference between the distances to the 5-bar hypothesis and the other hypotheses is statistically significant at the 99% level .
we interpret these results to mean that the underlying distribution is essentially concentrated around the hypothesis barcode ; the distance of @xmath637 is essentially a consequence of noise .
in contrast , when we compute @xmath638 ( using the same experimental procedure as above ) , we find the confidence interval is @xmath639 $ ] .
when we compute @xmath640 , we find that the confidence interval is @xmath641 $ ] .
this represents high confidence that @xmath642 and @xmath643 are far from this reference barcode , which in light of the confidence intervals above for @xmath636 and @xmath635 appear to indicate that samples sizes 25 and 75 are too small .
p. bubenik , g. carlsson , p. t. kim , and z .- m .
statistical topology via morse theory persistence and nonparametric estimation . in _ algebraic methods in statistics and probability
math . , 516:7592 .
soc . , providence , ri , 2010 .
m. k. chung , p. bubenik , and p. t. kim .
persistence diagrams in cortical surface data . in _ information processing in medical imaging ( ipmi ) 2009
_ , 5636:386397 , lecture notes in computer science , 2009 . | we study distributions of persistent homology barcodes associated to taking subsamples of a fixed size from metric measure spaces .
we show that such distributions provide robust invariants of metric measure spaces , and illustrate their use in hypothesis testing and providing confidence intervals for topological data analysis . |
neutron rich matter is at the heart of many fundamental questions in nuclear physics and astrophysics .
* what are the high density phases of qcd ? * where did the chemical elements come from ?
* what is the structure of many compact and energetic objects in the heavens , and what determines their electromagnetic , neutrino , and gravitational - wave radiations ?
furthermore , neutron rich matter is being studied with an extraordinary variety of new tools such as the facility for rare isotope beams ( frib ) , a heavy ion accelerator to be built at michigan state university , and the laser interferometer gravitational wave observatory ( ligo ) .
indeed there are many , qualitatively different , probes of neutron rich matter including precision laboratory experiments on stable nuclei and experiments with neutron rich radioactive beams . while astrophysical observations probe neutron rich matter with electromagnetic radiation , neutrinos , and gravitational waves . finally , there are new theoretical probes of neutron rich matter that have been enabled by tremendous advances in computational science . in this paper
we give brief examples of how neutron rich matter is being studied with some of these different probes .
we start with the lead radius experiment ( prex ) @xcite .
this is a precision experiment to measure the neutron radius in @xmath0pb with parity violating electron scattering @xcite .
this has many implications for nuclear structure , astrophysics , atomic parity violation , and low energy tests of the standard model .
parity violation provides a model independent probe of neutrons , because the @xmath1 boson couples to the weak charge , and the weak charge of a proton @xmath2 , @xmath3 is much smaller than the weak charge of a neutron @xmath4 , @xmath5 in eq .
[ qwp ] , @xmath6 is the weak mixing angle .
therefore , the weak neutral current interaction , at low momentum transfers , couples almost completely to neutrons .
prex will measure the parity violating asymmetry @xmath7 .
this is the fractional cross section difference for scattering positive ( + ) , or negative ( - ) , helicity electrons , @xmath8 in born approximation , @xmath7 arrises from the interference of a weak amplitude of order @xmath9 , with @xmath10 the fermi constant and @xmath11 the momentum transfer , and an electromagnetic amplitude of order the fine structure constant @xmath12 @xcite , @xmath13 here the weak form factor @xmath14 is the fourier transform of the weak charge density @xmath15 , that is essentially the neutron density , @xmath16 likewise , the electromagnetic form factor @xmath17 is the fourier transform of the ( electromagnetic ) charge density @xmath18 .
this is known from elastic electron scattering @xcite .
therefore , in principle , measuring @xmath7 as a function of @xmath11 allows one to map out the neutron density @xmath19 .
note that , for a heavy nucleus , there are important corrections to eq .
[ apvborn ] from coulomb distortions .
however , these have been calculated exactly by solving the dirac equation for an electron moving in both a coulomb potnetial of order 10 mev and a weak axial vector potential of order electron volts @xcite .
therefore , even with coulomb distortions , one can accurately determine neutron densities .
note that this purely electroweak reaction is free from most strong interaction uncertainties .
the prex experiment in hall a at jefferson laboratory measures @xmath7 for 1.05 gev electrons elastically scattered from @xmath0pb at laboratory angles near five degrees . the experiment uses a 0.5 mm thick pb foil target that is enriched in @xmath0pb and sandwiched between two thin diamond foils .
these diamond foils help remove the beam heating and keep the target from melting . scattered electrons are deflected by septum magnets into the two high resolution spectrometers that separate elastically from inelastically scattered electrons .
the goal of this challenging experiment is to measure @xmath7 , that is of order 0.5 ppm , to 3% .
this allows one to determine the neutron root mean square radius to 1% ( @xmath20 fm ) .
the experiment ran during april to june 2010 .
krishna kumar , one of the prex spokespersons , provided the following preliminary results @xcite .
the experiment was successfully commissioned in march/ april .
high quality data were accumulated in transverse spin mode so as to make the systematic error from a potnetial 2 photon exchange amplitude negligible . however , radiation issues downstream from the main apparatus resulted in a significant reduction in experiment `` up time '' .
sufficient statistics were accumulated to provide a significant first constraint on the neutron radius .
the collaboration purposes to design appropriate engineering modifications to the beamline to mitigate the radiation problem and request additional beam time to achieve the original goal of a 1% constraint on the neutron radius .
if one can accurately measure the neutron radius with prex , then there will be important implications for nuclear structure , atomic parity violation @xcite , and astrophysics .
let us discuss some of the implications for astrophysics .
heavy nuclei are expected to have a neutron rich skin .
the pressure of neutron matter in this skin forces neutrons out against surface tension .
therefore the larger the pressure of neutron matter , the larger the neutron radius in @xmath0pb @xcite .
measuring the neutron radius of @xmath0pb allows us to determine the equation of state of neutron matter ( pressure as a function of density ) for slightly subnuclear densities near 0.1 @xmath21 .
this density represents an average of the surface and interior densities of @xmath0pb .
there is a very interesting relationship between the neutron radius of @xmath0pb , of order 6 fm , and the radius of a neutron star , of order 10 km @xcite .
this involves a breathtaking extrapolation of 18 orders of magnitude in size , or 55 orders of magnitude in mass .
nevertheless , both in the laboratory and in astrophysics , it is the same neutrons , the same strong interactions , the same neutron rich matter , and the same equation of state .
a measurement in one domain can have important implications in the other domain .
the radius of a neutron star depends on the pressure of neutron matter at normal nuclear density and above , because the central density of a neutron star can be a few or more times that of normal nuclear density .
it is important to have both low density information on the equation of state from prex , and high density information from measurements of neutron star radii .
this can constrain any possible density dependence of the equation of state from an interesting phase transition to a possible high density exotic phase such as quark matter , strange matter , or a color superconductor .
for example , if the @xmath0pb radius is relatively large , this shows the eos is stiff at low density ( has a high pressure ) . if at the same time , neutron stars have relatively small radii , than the high density eos is soft with a low pressure .
this softening of the eos with density could strongly suggest a phase transition to a soft high density exotic phase .
the radius of a neutron star @xmath22 can be deduced from x - ray measurements of luminosity @xmath23 and surface temperature @xmath24 , @xmath25 with @xmath26 the stefan boltzmann constant .
this allows one to deduce the surface area of a neutron star . however , there are important complications .
first one needs an accurate distance to the neutron star , for example from optical parallax measurements .
second , eq . [ nsr ] assumes a black - body and there are important non - blackbody corrections that must be determined from models of neutron star atmospheres . finally , gravity is so strong that space near a neutron star is strongly curved .
if one looks at the front face of a neutron star , one will also see about 30% of the far face , because of the curvature of space .
therefore the surface area that one observes in eq .
[ nsr ] depends on the curvature of space and the mass of the star .
thus what started out as a measurement of just the radius is , in fact , a measurement of a combination of mass and radius .
recently steiner , lattimer , and brown have deduced masses and radii @xcite from combined observations of seven neutron stars in three classes ( 1 ) x - ray bursts , ( 2 ) neutron stars in globular clusters , and ( 3 ) the nearby isolated star rxj 1856 . they conclude that observations favor a stiff high density equation of state that can support neutron stars with a maximum mass near 2 @xmath27 and that the equation of state is soft at low densities so that a 1.4 @xmath27 neutron star has a radius near 12 km .
they go on to predict that their equation of state , fit to observations of neutron stars , implies the neutron minus proton root mean square radii in @xmath0pb will be @xmath28 fm .
the steiner et al .
paper @xcite is potentially controversial because their results depend on , among other things , the model assumed for x - ray bursts .
ozel et al .
use a different model for x - ray bursts , with more optimistic assumptions about the errors , and get very small neutron star radii near 10 km or below @xcite . clearly
these determinations of neutron star masses and radii , if correct , have very important implications for the properties of cold dense qcd .
the neutron radius of @xmath0pb has implications for neutron star structure , in addition to the star s radius .
neutron stars have solid crusts over liquid cores , while a heavy nucleus is expected to have a neutron rich skin .
both the solid crust of the star , and the skin of the nucleus , are made of neutron rich matter at similar , slightly subnuclear , densities .
the common unknown is the equation of state of neutron matter .
a thick neutron skin in @xmath0pb , means a high pressure where the energy rises rapidly with density .
this quickly favors the transition to a uniform liquid phase .
therefore , we find a strong correlation between the neutron skin thickness , measured by prex , and the transition density in neutron stars from solid crust to liquid interior @xcite .
perhaps the presence of a solid crust deserves comment . we expect solids to form in compact stars , both white dwarfs and neutron stars
this may sound surprising since stars are often composed of not liquids or gasses , but plasmas
. nevertheless , the plasmas can be so dense that the ions actually freeze .
recently , there are significant new observations of how white dwarfs freeze @xcite and of how the solid crust of accreting neutron stars cools @xcite .
we hope to extract important new information , on the structure of solids in stars , by comparing these observations to our large scale molecular dynamics simulations of freezing in both white dwarfs @xcite and neutron stars @xcite .
we start with white dwarf crystallization .
winget et al .
deduce the luminosity function for white dwarfs in the globular star cluster ngc 6397 @xcite , see also @xcite .
this is simply the number of white dwarfs in different luminosity bins .
as white dwarfs cool , their luminosities decrease . the probability to find a white dwarf with a given luminosity depends on the cooling rate .
one is more likely to find stars at luminosities where they cool slowly , and hence spend a long time .
therefore , the luminosity function is expected to be proportional to one over the cooling rate .
eventually as a star cools , it freezes .
when this happens , the latent heat will slow the cooling rate until finally all of the latent heat is radiated away .
this is exactly like ice cubes slowing the rate of warming of your drink .
therefore there should be a peak in the luminosity function when the stars freeze .
indeed such a peak is observed by winget et al .
furthermore , the luminosity where this peak occurs , depends on the interior temperature .
this allows one to deduce the melting temperature of white dwarf cores .
we expect the cores of the white dwarfs in ngc 6397 to be a mixture of carbon and oxygen . from the observed melting temperature one
can infer the ratio of carbon to oxygen , if one knows the phase diagram for carbon / oxygen mixtures .
indeed the ratio of carbon to oxygen in white dwarf cores is important for stellar evolution and nucleosynthesis .
it depends on the rate for the , very interesting but incompletely known , nuclear reaction @xmath29c(@xmath30)@xmath31o .
note that there have been previous efforts to determine the composition of white dwarf cores using astrosysmology , where the frequencies of different oscillation modes are observed @xcite .
however , the interpretation of these frequencies may be ambiguous . to determine the carbon/ oxygen phase diagram
, we perform large scale molecular dynamics simulations of the freezing of carbon and oxygen plasma mixtures @xcite .
the temperature of a simulation volume containing a 27648 ion carbon / oxygen mixture is carefully adjusted so that half of the system remains solid while the other half is liquid , see fig .
carbon and oxygen ions are free to diffuse between the liquid and solid phases and , in general the solid becomes enriched in oxygen .
we measure the composition of each phase at the end of the simulation . because diffusion is slow , these simulations took about six months of computer time on special purpose mdgrape-2 boards .
our phase diagram agrees well with the recent model of mendin and cummings @xcite , but gives lower melting temperatures than earlier work by segratain et al .
@xcite . molecular dynamics simulation of a 27648 ion carbon and oxygen mixture showing a solid phase ( center ) in equilibrium with a liquid phase ( top and bottom ) @xcite.,width=499 ] winget et al.s observations of the white dwarf luminosity function of ngc 6397 , together with our carbon/ oxygen phase diagram
imply that the oxygen concentration ( by number ) @xmath32 in the core of these white dwarfs is @xmath33 @xcite .
the oxygen concentration in the core of a white dwarf depends on the rate of the nuclear reaction @xmath29c(@xmath30)@xmath31o and on the treatment of diffusion in stellar evolution models . for salaris et al.s models of 0.54 @xmath27 white dwarfs
, the constraint @xmath33 corresponds to a constraint on the effective astrophysical @xmath34 factor at 300 kev , describing the @xmath29c(@xmath30 ) reaction , of @xmath35 170 kev barns .
this is consistent with the recent experimental determination of buchmann and barnes of @xmath36 kev barns @xcite
. observations of crystallization in many more white dwarf systems is feasible and this may provide important benchmarks for both the @xmath29c(@xmath30 ) reaction , and our understanding of how nuclear reactions in stars make the carbon in our bones and the oxygen that we breathe .
we next consider freezing in accreting neutron stars .
material falling on a neutron star can undergo rapid proton capture ( or rp process ) nucleosynthesis to produce a range of nuclei with mass numbers @xmath37 that could be as high as @xmath38 @xcite . as this rp process ash is buried by further accretion ,
the rising electron fermi energy induces electron capture to produce a range of neutron rich nuclei from o to approximately se @xcite .
this material freezes , when the density reaches near @xmath39 g/@xmath40 . we have performed large scale md simulations of how this complex rp process ash freezes @xcite .
we find that chemical separation takes place and the liquid ocean is greatly enriched in low atomic number @xmath41 elements , while the newly formed solid crust is enriched in high @xmath41 elements .
furthermore , we find that a regular crystal lattice forms even though large numbers of impurities are present .
this regular crystal should have a high thermal conductivity .
we do not find an amorphous solid that would have a low thermal conductivity .
recent x - ray observations of neutron stars , find that the crust cools quickly , when heating from extended periods of accretion stops @xcite .
this is consistent with our md simulations , and strongly favors a crystalline crust over an amorphous solid that would cool more slowly @xcite .
see also the paper by ed .
brown in these proceedings @xcite .
we turn now to gravitational waves ( gw ) from crust `` mountains '' on rapidly rotating neutron stars .
albert einstein , almost 100 years ago , predicted the oscillation of space and time known as gravitational waves . within a few years , with the operation of advanced ligo and other sensitive interferometers , we anticipate their historic detection .
this will be a big deal , after decades of effort , and open a new window on the universe and on neutron rich matter .
strong gravitational wave sources often involve large accelerations of large amounts of neutron rich matter .
indeed the requirements for a strong source of continuous gw , at ligo frequencies , places extraordinary demands on neutron rich matter . generating gw sounds easy .
place a mass on a stick and shake vigorously .
however to have a detectable source , one may need not only a large mass , but also a very strong stick .
the stick is needed to help produce large accelerations .
since others have discussed large masses , let us focus here on the strong stick .
recently we performed large scale md simulations of the strength of neutron star crust @xcite .
a strong crust can support large deformations or `` mountains '' on neutron stars , see also @xcite .
these bumps , on rapidly rotating stars , can generate strong gravitational waves .
indeed bildstein and others @xcite have speculated that some stars in binary systems may spin so fast that they radiate angular momentum in gw at the same rate that angular momentum is gained from accretion .
this would explain why the fastest observed neutron stars are only spinning at about half of the breakup rate .
there are several very active ongoing and near future searches for continuous gravitational waves at ligo , virgo and other detectors , see for example @xcite .
often one searches at twice the frequency of known radio signals from pulsars because of the quadrupole nature of gw .
although the signals may be weak , one can gain sensitivity by integrating coherently for long observation times , although this can require very large computational resources .
no signal has yet been detected .
however , sensitive upper limits have been set .
these limits constrain the shape of neutron stars . in some cases the star s
elipticity @xmath42 , which is that fractional difference in moments of inertia @xmath43 is observed to be less than a part per million or even smaller .
here @xmath44 , @xmath45 , and @xmath46 are the principle moments of inertia
. how large can a neutron star mountain be before it collapses under the extreme gravity ?
this depends on the strength of the crust .
we performed large scale md simulations of crust breaking , where a sample was strained by moving top and bottom layers of frozen ions in opposite directions @xcite .
these simulations involve up to 12 million ions and explore the effects of defects , impurities , and grain boundaries on the breaking stress .
for example , in fig .
[ fig2 ] we show a polycrystalline sample involving 12 million ions . in the upper right panel
the initial system is shown , with the different colors indicating the eight original microcrystals that make up the sample .
the other panels are labeled with the strain , i.e. fractional deformation , of the system .
the red color indicates distortion of the body centered cubic crystal lattice .
the system starts to break along grain boundaries .
however the large pressure holds the microcrystals together and the system does not fail until large regions are deformed .
molecular dynamics simulation of a 12 million ion polycrystaline sample of neutron star crust .
the four panels are labeled by the strain , which is the fractional deformation .
the colors in the upper left panel , at zero strain , show the original eight microcrystals of different orientations .
the red color in the other three panels indicates distortion of the body centered cubic lattice @xcite.,width=537 ] we find that neutron star crust is very strong because the high pressure prevents the formation of voids or fractures and because the long range coulomb interactions insure many redundant `` bounds '' between planes of ions .
neutron star crust is the strongest material known , according to our simulations .
the breaking stress is 10 billion times larger than that for steel .
this is very promising for gw searches because it shows that large mountains are possible , and these could produce detectable signals .
neutrinos provide yet another , qualitatively different , probe of neutron rich matter . core collapse supernovae ( sn ) are gigantic stellar explosions that convert as much as @xmath47 of mass into @xmath48 neutrinos .
there are many exciting new underground experiments to search for dark matter , double beta decay , solar neutrinos , proton decay , oscillation of accelerator neutrinos , etc
. some of these experiments can be very sensitive to neutrinos from the next galactic supernova ( sn ) .
for example , the coming ton - scale dark matter experiments can be sensitive to sn neutrinos via neutrino - nucleus elastic scattering @xcite .
here , the spectrum of recoiling nuclei provides information of the spectrum of mu and tau neutrinos from a sn that is not available from the existing super - kamiokande detector .
super - k , a large water detector , is primarily sensitive to sn @xmath49 via @xmath50 .
note that most of the mass of super - k is oxygen and this is less sensitive to sn neutrinos than hydrogen .
in contrast , a neutrino - nucleus elastic detector involves very large coherent cross sections , is sensitive to all six flavors of sn neutrinos ( @xmath51 , and @xmath52 ) and all of the mass , in the fiducial volume , is active .
this gives very large yields of tens to hundreds of events per ton , instead of per kiloton , for a sn at 10 kiloparsecs @xcite .
neutrinos are emitted from a low density , @xmath53 g/@xmath40 , neutron rich neutrino - sphere region .
the pressure , composition , and long wave - length neutrino response @xcite of this region can be calculated from a model independent virial expansion @xcite that is based on nucleon - nucleon , nucleon - alpha , and alpha - alpha elastic scattering phase shifts .
the abundance of light nuclei near the neutrino - sphere can be important for the radiated @xmath54 spectrum @xcite .
the chance to combine electromagnetic , gravitational wave , and neutrino signals from the next galactic supernova should lead to tremendous results on the dynamics of supernova explosions , neutrinos and neutrino oscillations , the formation of neutron stars , and on nucleosynthesis .
we end with a new equation of state ( eos ) , pressure as a function of density , temperature , and proton fraction , for neutron rich matter .
this can be used in simulations of sn and neutron star mergers . at low densities
we use a virial expansion with nucleons , alphas , and thousands of species of heavy nuclei @xcite . at high densities
we use relativistic mean field calculations in a spherical wigner - seitz approximation @xcite . generating the eos table for a large range of temperatures @xmath55 to 80 mev ,
densities @xmath56 to 1.6 @xmath21 , and proton fractions took over 100,000 cpu hours .
this eos is exact at low density and contains detailed composition information for calculating neutrino interactions .
neutron rich matter is at the heart of many fundamental questions in nuclear physics and astrophysics .
what are the high density phases of qcd ? where did the chemical elements come from ?
what is the structure of many compact and energetic objects in the heavens , and what determines their electromagnetic , neutrino , and gravitational - wave radiations ?
moreover , neutron rich matter is being studied with an extraordinary variety of new tools such as facility for rare isotope beams ( frib ) and the laser interferometer gravitational wave observatory ( ligo ) .
we describe the lead radius experiment ( prex ) that is using parity violation to measure the neutron radius in 208pb .
this has important implications for neutron stars and their crusts . using large scale molecular dynamics , we model neutron rich matter and simulate how material in white dwarfs and neutron stars freeze . also ,
using large scale md simulations , we find neutron star crust to be the strongest material known , some 10 billion times stronger than steel .
it can support mountains on rotating neutron stars large enough to generate detectable gravitational waves .
finally , neutrinos from core collapse supernovae provide another , qualitatively different probe of neutron rich matter .
we describe a new equation of state for supernova and neutron star merger simulations based on the virial expansion at low densities , and on large scale relativistic mean field calculations at high densities .
this work was done in collaboration with many people including d. k. berry , e. f. brown , k. kadau , j. piekarewicz , and graduate students liliana caballero , helber dusan , joe hughto , andre schneider and gang shen .
this work was supported in part by doe grant de - fg02 - 87er40365 , by shared university research grants from ibm , inc . to indiana university , and by the national science foundation , teragrid grant tg - ast100014 .
r. wijnands et al .
, astro - ph/0405089 .
e. m. cackett et al . , mon . not .
. soc . * 372 * , 479 ( 2006 ) .
r. e. rutledge et al . , apj .
580 , 413 ( 2002 ) . c. j. horowitz , a. s. schneider , d. k. berry , phys .
* 104 * , 231101 ( 2010 ) . | neutron rich matter is at the heart of many fundamental questions in nuclear physics and astrophysics .
what are the high density phases of qcd ? where did the chemical elements come from ?
what is the structure of many compact and energetic objects in the heavens , and what determines their electromagnetic , neutrino , and gravitational - wave radiations ?
moreover , neutron rich matter is being studied with an extraordinary variety of new tools such as facility for rare isotope beams ( frib ) and the laser interferometer gravitational wave observatory ( ligo ) .
we describe the lead radius experiment ( prex ) that is using parity violation to measure the neutron radius in 208pb .
this has important implications for neutron stars and their crusts . using large scale molecular dynamics
, we model the formation of solids in both white dwarfs and neutron stars .
we find neutron star crust to be the strongest material known , some 10 billion times stronger than steel .
it can support mountains on rotating neutron stars large enough to generate detectable gravitational waves .
finally , we describe a new equation of state for supernova and neutron star merger simulations based on the virial expansion at low densities , and large scale relativistic mean field calculations . |
in 1950 , h. seifert @xcite asked whether every dynamical system on the 3-sphere with no singular points has a periodic trajectory .
the conjecture that the answer is yes became known as the seifert conjecture .
seifert proved the conjecture for perturbations of the flow parallel to the hopf fibration . in 1974 , p. a. schweitzer @xcite found a @xmath1 counterexample to the seifert conjecture , which was modified to a @xmath2 counterexample by j. harrison @xcite . in 1993 , h. hofer @xcite proved the conjecture for contact flows .
the main idea of this paper comes from the construction of a smooth counterexample to the seifert conjecture due to the second author @xcite . here
, we outline more general consequences of that idea .
in particular , we establish the following theorem : a foliation of any codimension of any manifold can be modified in an analytic or piecewise - linear fashion so that all minimal sets have codimension 1 .
[ thcd1 ] this theorem strengthens an earlier result of f. w. wilson @xcite which for oriented 1-foliations , establishes minimal sets of codimension 2 .
theorem [ thcd1 ] is trivial for foliations which are themselves codimension 1 .
the problem of opening all compact leaves in codimension 1 , which is a more natural question , has been partly settled by s. p. novikov@xcite and p. a. schweitzer @xcite .
novikov proved that every @xmath2 codimension 1 foliation of @xmath0 has a closed leaf ( which was later extended to continuous foliations by v. v. solodov @xcite and g. hector and u. hirsch @xcite ) , while schweitzer has shown that it is possible to modify any codimension 1 foliation in dimension 4 or higher in a @xmath1 fashion so that it has no compact leaf .
here and throughout the paper , smooth means @xmath3 and analytic means real analytic or @xmath4 .
all manifolds are assumed to be paracompact and hausdorff , and they are assumed to have no boundary unless explicitly stated otherwise .
also , dimension means covering dimension , cohomological dimension , small inductive dimension , or large inductive dimension ; they are all equal for topological spaces considered in this paper . in our context , we can freely convert between an ( oriented ) 1-dimensional foliation and a dynamical system with no singular points .
note that theorem [ thcd1 ] implies the following theorem : the 3-sphere @xmath0 has an analytic dynamical system such that all limit sets are 2-dimensional .
in particular , it has no circular trajectories .
[ th3sph ] table [ texamples ] shows the best known continuity of various kinds of foliations of 3-manifolds .
entries with no citation are covered by theorem [ th3sph ] or its analogue for arbitrary 3-manifolds .
.known foliations of 3-manifolds [ cols="^,^,^",options="header " , ] note that if @xmath5 is a flow bordism with entry region @xmath6 and exit region @xmath7 , then since @xmath6 and @xmath7 are the transverse boundaries together with some corner separation points , they possess trivially foliated neighborhoods . specifically , if @xmath8 is foliated by vertical fibers @xmath9 , there exists a leaf - preserving homeomorphism @xmath10 from an open neighborhood of @xmath11 to @xmath8 ; moreover , we can take @xmath12 for all @xmath13
. if @xmath5 is a flow bordism with connected support @xmath14 and @xmath15 has at least one circular or infinite leaf , then the entry stopped set @xmath16 ( resp .
the exit stopped set @xmath17 ) is non - empty if and only if @xmath6 ( resp .
@xmath7 ) is non - empty . in particular ,
if @xmath5 has matched ends ( and @xmath18 is non - empty ) , then @xmath5 is a plug .
[ lisplug ] let @xmath19 be a flow parallel to @xmath5 . if @xmath20 , then @xmath21 , the _ future longevity _ of @xmath22 , is the supremum of @xmath23 such that @xmath24 is defined .
similarly , the _ past longevity _
@xmath25 is the supremum of @xmath23 such that @xmath26 is defined .
thus we have two functions @xmath27 $ ] .
if @xmath28 or @xmath29 , then the leaf containing @xmath22 has an endpoint @xmath30 on @xmath7 or @xmath6 , respectively .
investigating the points in a neighborhood of such a @xmath30 , we see that two sets @xmath31 are both closed . if @xmath32 , then @xmath33 , and the set @xmath34 consisting of leaves that intersect @xmath6 is closed .
then @xmath14 is the union of two disjoint non - empty closed sets , @xmath34 and @xmath35 , which is a contradiction .
hence @xmath36 if @xmath37 .
similarly , @xmath38 if @xmath39 .
if @xmath5 is a semi - un - plug with support @xmath14 and entry region @xmath6 , then there is a foliation isomorphism @xmath40 , where @xmath41 is foliated by fibers @xmath42 . in particular ,
if @xmath5 is an un - plug with base @xmath43 , then there is a foliation isomorphism @xmath44 which extends the maps @xmath45 .
[ listrivial ] let @xmath19 be a flow parallel to @xmath5 .
let @xmath46 be the future longevity as in the proof of lemma [ lisplug ] . in each smoothness category other than the pl category ,
define @xmath47 .
the map @xmath48 has all of the desired properties .
unfortunately , the pl category is not closed under the arithmetic operation of division , so a more complicated argument is necessary .
let @xmath49 be the region in @xmath50 bounded by @xmath51 and the graph of @xmath46 restricted to @xmath6 .
the map @xmath52 is a foliation isomorphism from @xmath49 to @xmath14 ; it remains to find a foliation isomorphism @xmath53 from @xmath41 to @xmath49 . to construct @xmath53 ,
let @xmath54 be a triangulation with vertex set @xmath55 of @xmath6 such that @xmath46 is linear on each simplex of @xmath54 and such that @xmath54 is the barycentric subdivision of another triangulation .
since @xmath54 is a barycentric subdivision , there is a map @xmath56 , where @xmath57 is the dimension of @xmath6 , which is bijective when restricted to the vertices of any single @xmath57-simplex of @xmath54 .
if @xmath58 is such a simplex , consider a triangulation @xmath59 of @xmath60 , such that the simplex @xmath61 has vertices @xmath62 and @xmath63 .
then the map on vertices of @xmath60 given by @xmath64 and @xmath65 extends to a pl foliation isomorphism from @xmath60 to @xmath66 .
these maps fit together for different simplices @xmath58 of @xmath54 , yielding the desired map @xmath53 .
the following construction due to wilson @xcite turns a semi - plug into a plug .
if @xmath67 and @xmath68 are two flow bordisms such that the exit region of @xmath67 is the same as the entry region of @xmath68 , their _ concatenation _ is a flow bordism obtained by identifying trivially foliated neighborhoods of this shared region .
the _ mirror image _ @xmath69 of @xmath5 is given by reversing the orientation of the leaves of @xmath70 , which has the effect of switching the entry and exit regions .
the _ mirror - image construction _ is the concatenation of @xmath5 with @xmath71 ; it is easy to see that the result of this concatenation has matched ends .
( figure [ fmirror ] shows a schematic picture of a mirror - image construction . )
the primary purpose of plugs is the operation of _ insertion_. the geometric idea of insertion is illustrated in figure [ finsert ] . given a foliation @xmath72 of an @xmath57-manifold @xmath73 and an @xmath57-dimensional plug @xmath5 with base @xmath43 , we wish to find a leaf - preserving embedding of the trivially foliated @xmath74 in @xmath73 and replace it with @xmath5 . in full generality
, this procedure raises three technical issues : insertibility , attachability , and twistedness .
an _ insertion map _ for a plug @xmath5 into a foliation @xmath75 is an embedding @xmath76 of the base of @xmath5 which is transverse to @xmath75 .
such an insertion map can be extended to an embedding @xmath77 which takes the fiber foliation of @xmath74 to @xmath72 .
( it is convenient to denote the maps @xmath78 and @xmath79 by the same @xmath80 and this should not lead to a missunderstanding . )
an @xmath57-dimensional plug @xmath5 is _ insertible _ if @xmath43 admits an embedding in @xmath81 which is transverse to vertical lines .
such an embedding is equivalent to a _ bridge immersion _ of @xmath43 in @xmath82 , _
i.e. _ , an immersion which lifts to an embedding of @xmath74 in @xmath81 .
for example , if @xmath43 is 2-dimensional , orientable , and has non - empty boundary , then @xmath43 necessarily admits a bridge immersion .
figure [ fbridge ] shows a bridge immersion of a punctured torus @xmath83 and a bridge immersion of a surface of genus three with three punctures .
the corresponding embedding of @xmath84 is one of the main steps in schweitzer s counterexample to the seifert conjecture @xcite .
a plug is _ attachable _ if every leaf in the parallel boundary is finite . in general ,
if @xmath85 is a manifold with boundary , let @xmath86 denote an open neighborhood of @xmath87 in @xmath85 .
the next step in plug insertion is to remove @xmath88 from @xmath73 and glue the open lip @xmath89 to the support @xmath14 of an attachable plug @xmath5 by a leaf - preserving homeomorphism @xmath90 , where @xmath91 is a neighborhood of @xmath92 .
moreover , the identification @xmath93 should satisfy @xmath94 and @xmath95 .
a map @xmath93 with these properties is an attaching map for @xmath5 .
let @xmath96 be the parallel boundary of @xmath14 , so that @xmath97 .
recall that all of the leaves in @xmath96 have two endpoints ; this must also be true in a neighborhood @xmath98 of @xmath96 .
it follows by lemma [ listrivial ] that there exists a leaf - preserving homeomorphism @xmath99 .
this equivalence , together with the matched ends condition , ensures the existence of an attaching map for any attachable plug or un - plug .
let @xmath80 be an insertion map of a plug @xmath5 into a foliation @xmath75 on a manifold @xmath73 , and let @xmath100 be the foliation on the manifold @xmath101 resulting from inserting @xmath5 into @xmath75 .
the plug @xmath70 is _ untwisted _ if the attaching map @xmath93 extends to a homeomorphism @xmath102 .
the manifolds @xmath73 and @xmath101 need not be homeomorphic , but they are if @xmath5 is untwisted . if @xmath70 is a smooth or pl plug , then it is assumed that the extension of @xmath93 is smooth or pl just as @xmath93 is .
however , if @xmath5 is analytic , then an analytic extension of @xmath93 is usually not possible . in this case
, @xmath5 is untwisted if @xmath93 admits a smooth extension .
the resulting manifolds @xmath73 and @xmath101 have analytic structures which are a priori only smoothly diffeomorphic ; however , the morrey - grauert theorem ensures that they are in fact analytically equivalent .
all plugs in this paper are assumed to be untwisted unless explicitly stated otherwise .
a plug or a semi - plug is _ aperiodic _ if it has no circular leaves .
one of the main reasons to insert an aperiodic plug into another foliation is to break a circular leaf .
let @xmath5 be a plug with stopped set @xmath103 and base @xmath43 , and let @xmath80 be an insertion map into a foliation @xmath72 .
if @xmath104 intersects a circular leaf @xmath105 of @xmath72 , then the remnant of @xmath105 in @xmath100 is an infinite leaf . on the other hand ,
if @xmath105 is disjoint from @xmath104 but intersects @xmath106 , then there is a leaf @xmath107 in @xmath100 corresponding to @xmath105 with the same topology as @xmath105 , although the geometry of @xmath107 can differ from that of @xmath105 .
the simplest interesting target for an insertion is an irrational foliation of the 3-sphere @xmath0 .
given an irrational number @xmath108 , the irrational foliation of slope @xmath109 is parallel to the vector field @xmath110 on @xmath111 given by @xmath112 the foliation has two circles , and the closure of any other leaf is a torus lying between these two circles .
two copies of any aperiodic plug can be inserted in small regions meeting these circles and breaking them @xcite .
thus , to find a counterexample to the seifert conjecture , it suffices to construct an aperiodic plug .
if the plug is smooth or analytic , the corresponding foliation of @xmath0 is respectively smooth or analytic .
in 1935 , k. borsuk @xcite gave an example of a fixed point free homeomorphism of an acyclic compact subset of @xmath113 ; a solid cylinder with two narrowing and spiraling tunnels drilled out .
his example can be easily modified to obtain a semi - plug on @xmath114 ( @xmath115 is an @xmath57-cell ) with two circular leaves whose entry and exit stopped sets @xmath16 and @xmath17 are disks .
a similar construction on a solid torus ( borsuk s example can be obtained by cutting or unwrapping the torus ) was given in 1952 by f. b. fuller @xcite .
the consequence of fuller s result that @xmath116 , @xmath117 , admits a dynamical system whose only minimal set is an @xmath118-torus @xmath119 contained in a @xmath120 section motivated wilson s introduction of plugs , which he used to arrest flows globally .
every smooth @xmath57-manifold of euler characteristic zero or non - compact has a smooth dynamical system with a discrete set of minimal sets .
each of the minimal sets is an @xmath118-torus , and every trajectory originates ( resp . limits ) on one of these tori .
[ thwilson ] wilson s theorem implies that an analogue to the seifert conjecture for higher dimensional spheres of odd dimension is false . although wilson claimed theorem [ thwilson ] in the smooth category , the construction is actually analytic . also , with slight modifications , wilson s construction is valid in a pl setting .
wilson constructed a content - stopping , mirror - image plug and inserted many small copies of the plug into the foliation to break the leaves into small pieces .
more specifically , he proved that if @xmath121 is an oriented 1-foliation of an @xmath57-manifold @xmath85 , @xmath117 , and @xmath122 is open cover of @xmath85 , then @xmath75 can be modified to an oriented 1-foliation with every leaf contained in a star of an element of @xmath122 .
the following result of @xcite is therefore a special case of theorem [ thwilson ] . for every @xmath123
, there exists an oriented 1-foliation of @xmath124 with all leaves of diameter less than @xmath125 .
[ threed ] to establish an aperiodic version of theorem [ thwilson ] and [ threed ] in dimension 3 , it suffices to use an aperiodic plug that stops content , for instance , a variation of schweitzer s plug @xcite or a plug obtained by breaking the circles of wilson s periodic plug with any aperiodic plug . sections [ scounter ] , [ shigher ] , and [ spl ] give constructions of analytic and pl aperiodic plugs such that each has a unique minimal set and it has codimension 1 .
these plugs yield stronger versions of theorems [ thwilson ] and [ threed ] .
if @xmath85 is a continuous , @xmath126 , @xmath3 , @xmath4 , or @xmath127 manifold of dimension @xmath128 admitting an oriented 1-foliation in the same smoothness category , and @xmath122 is an open cover of @xmath85 , then there exists an aperiodic oriented 1-foliation of @xmath85 in the same smoothness category , such that each leaf is contained in an element of @xmath122 , and whose minimal sets have codimension one .
[ thgreg ] let @xmath85 be a @xmath127 manifold of dimension @xmath117 , @xmath129 , and let @xmath122 be an open cover of @xmath85 .
a @xmath127 1-foliation of @xmath85 can be modified in a @xmath127 fashion so that each leaf is contained in an element of @xmath122 , there are no circular leaves , and all minimal sets are @xmath130-dimensional .
[ thgregs ] the stopped set of each of the plugs constructed in this paper contains an arc . if the manifold @xmath85 is closed , then the foliation specified in wilson s theorem has finitely many minimal sets .
a plug can be `` weaved '' through the foliation so the arc in the stopped set meets a dense leaf in each of the minimal sets .
hence for closed manifolds , we get the following variations of the above theorems : if @xmath85 is a continuous , @xmath126 , @xmath3 , @xmath4 , or @xmath127 closed manifold of dimension @xmath128 admitting an oriented 1-foliation in the same smoothness category , then there exists an aperiodic oriented 1-foliation of @xmath85 in the same smoothness category , with exactly one minimal set , and the minimal set has codimension one .
[ thcompact ] let @xmath85 be a closed @xmath127 manifold of dimension @xmath117 and @xmath129 .
a @xmath127 1-foliation of @xmath85 can be modified in a @xmath127 fashion so that there are no circular leaves , and there is exactly one minimal set which is @xmath130-dimensional .
[ thcompactpl ]
as a warm - up to the important technique of self - insertion , we first define and analyze partial insertion of plugs .
let @xmath5 be an @xmath57-dimensional plug with base @xmath43 and support @xmath14 .
let @xmath131 be an attaching map for @xmath5 .
let @xmath73 be an @xmath57-manifold with an oriented @xmath132-foliation @xmath72 .
let @xmath133 be a closed domain in @xmath43 whose boundary in @xmath43 ( in the point - set topology sense of boundary ) is a properly embedded @xmath118-manifold @xmath34 transverse to @xmath134 .
it may happen that either @xmath34 or @xmath133 is not connected .
the domain @xmath133 is , in particular , a manifold with corners .
a _ partial insertion map _ of @xmath15 is an embedding @xmath135 such that @xmath136 , but @xmath137 is disjoint from @xmath138 .
the map @xmath80 can be extended to an embedding of @xmath139 which takes vertical fibers to leaves .
define @xmath101 to be @xmath140 glued to @xmath14 by the composition @xmath141 where it is defined .
define @xmath142 by gluing @xmath72 to @xmath70 in the same way .
figure [ fpartial ] shows an example of a partial insertion of a plug with base @xmath143 .
the partial insertion creates transverse boundary ( resembling a ledge and an overhang ) ; unlike with a complete insertion , some leaves in the result have endpoints . by the matched ends condition , all leaves of the partial insertion either have both endpoints or neither endpoint at the transverse boundary .
a leaf @xmath100 from the depths of @xmath73 might enter @xmath14 and never return to @xmath73 , but only by approaching some minimal set of @xmath5 and never by terminating .
let @xmath5 , @xmath34 , and @xmath133 be as before .
a _ self - insertion map _ for @xmath15 is an embedding @xmath144 such that the image of @xmath80 is transverse to @xmath5 and disjoint from @xmath11 , @xmath145 , @xmath146 is disjoint from @xmath92 , @xmath147 avoids leaves with endpoints in @xmath148 , and @xmath147 does not intersect any leaf twice .
choose an attaching map @xmath131 such that the image of @xmath80 is disjoint from the closure of @xmath149 . the map @xmath80 can be extended to an embedding of @xmath139 which is also disjoint from the closure of @xmath150 and from @xmath11 .
let @xmath151 be the space @xmath152 identified to itself by @xmath141 where it is defined .
the self - insertion of @xmath5 at @xmath80 is the foliation @xmath153 on @xmath151 obtained by the same gluing .
figure [ fentryexit ] shows an example of self - insertion of an un - plug with base @xmath154 . keeping the above notation , the surfaces @xmath155 and @xmath156 , which are on the boundary of @xmath14 , are in the interior of @xmath151 .
these are the _ internal entry _ and _ exit regions _ , respectively , of @xmath157 . similarly , @xmath158 and @xmath159 are the _ external entry _ and _ exit regions _ of @xmath157 .
many leaves of @xmath157 cross the internal entry and exit regions once or many times .
the points where a leaf @xmath105 does so are the _ transition _ points of @xmath105 , or the entry and exit points . in addition , @xmath105 may begin with an external entry and/or end with an external exit .
progressing along a leaf @xmath105 of @xmath157 in the positive direction , we encounter a history of transition points . in general @xmath105
consists of segments of leaves of @xmath5 separated by transition points . if @xmath160 is an entry ( exit ) , the leaf of @xmath5 preceding @xmath22 is _ interrupted _ ( _ interrupting _ ) leaf @xmath22 , while the leaf following @xmath22 is the interrupting ( interrupted ) leaf . in the self - insertion in figure [ fentryexit]a , a leaf has the following history of transition points : ( external ) entry , entry , entry , exit , exit , ( external ) exit . in the self - insertion in figure [ fentryexit]b ,
a leaf entering the bottom has the history ( external ) entry , entry , entry , entry , @xmath161 in a given history , an entry and an exit are _ matched _ if the exit follows the entry and they satisfy the following inductive rule : they are matched if they are adjacent , and otherwise they are matched if all entries and exits between them are matched to each other .
for example , if a leaf in a hypothetical self - insertion has the history given in figure [ frainbow ] , then the entries and exit are matched as indicated .
the proof of the following lemma is an instructive exercise for the reader ; it can also be found in @xcite .
note that the lemma would not hold for a hypothetical definition of self - insertion of a semi - plug .
if two transition points are matched , they have the same interrupted leaf and they are the endpoints of the same interrupting leaf .
[ linterrupt ] the definition of matched transition points and lemma [ linterrupt ] suggest an interpretation of the geometry of a self - insertion in terms of a recursive algorithm .
the following procedure follows a leaf of @xmath157 by following segments of leaves of @xmath5 .
the input to the procedure is a base point @xmath162 ; the procedure follows the corresponding leaf @xmath107 of @xmath157 .
: : procedure followleaf(@xmath22 ) 1 .
: : let @xmath105 be the leaf of @xmath5 that begins at @xmath163 .
2 . : : follow @xmath105 until some @xmath164 or @xmath165 is reached .
: if @xmath165 is reached , quit .
4 . : : if @xmath164 is reached , then do followleaf(@xmath30 ) and go to step 2 . in computer programming terms , a recursive procedure such as followleaf is implemented by means of a pushdown _ stack _ : when followleaf is called , its argument is pushed onto the top of the stack .
it is removed when that particular instantiation of followleaf terminates .
given a point @xmath166 , the stack of @xmath30 is the sequence of points @xmath167 , @xmath168 , @xmath169 where @xmath170 is a point in @xmath43 such that at least one of @xmath171 and @xmath172 is an endpoint of the leaf of @xmath5 containing @xmath22 .
the function @xmath173 is not defined on all of @xmath14 ; the stack at @xmath30 extends as long as @xmath173 is defined and may therefore be either a finite or infinite sequence .
it is easy to check that as @xmath30 moves along its leaf , its stack changes in the same way as the stack used to execute followleaf .
inspection of followleaf demonstrates that in a finite leaf , all transition points are matched .
in particular , the two endpoints are matched to each other .
this can be construed as saying that @xmath157 has matched ends .
unfortunately , @xmath157 is not even a flow bordism , because the top and bottom of @xmath151 have stair - steps and are not transverse to @xmath157 .
however , without disturbing the transition point scheme , it is possible to approximate the top and bottom of @xmath151 by surfaces @xmath174 and @xmath175 that are transverse to @xmath157 .
let @xmath176 be the manifold that results from cutting along these surfaces , and let @xmath177 , the self - inserted plug , be the restriction of @xmath157 to @xmath176 .
as figure [ fentryexit]b demonstrates , a leaf can have unmatched transition points .
indeed , by lemma [ linterrupt ] , if a transition point ( whether internal or external ) is the endpoint of an infinite leaf of @xmath5 , then it can not be matched in @xmath157 .
its leaf in @xmath153 ( and hence in @xmath178 ) is therefore either infinite or circular , and by lemma [ lisplug ] , @xmath178 is a plug . in particular ,
if @xmath162 is in the stopped set of @xmath5 and @xmath179 , then it lies in an infinite leaf of @xmath157 .
thus , a self - insertion can not close infinite leaves of this type . on the other hand ,
figure [ fentryexit ] also demonstrates that a self - insertion can create new circular leaves ; these necessarily have unmatched transition points .
finally , since a self - insertion changes the base of a plug , it may happen that the self - insertion of an insertible plug is not insertible .
it may also happen that the self - insertion of an untwisted plug is twisted .
the following example shows that a self - insertion may cause the boundary of a plug to disappear completely .
let @xmath180 be a plug with support @xmath114 such that the transverse boundary consists of the top and bottom of the cylinder , and the side of the cylinder is the parallel boundary .
a self - insertion of @xmath180 using an immersion of the entire disk @xmath143 results in a plug with a boundaryless support ( _ e.g. _ , @xmath181 ) .
this section contains a construction , similar to that of @xcite , of a two - component self - insertion of a plug @xmath182 which is a concatenation of two semi - plugs @xmath183 and @xmath184 , each a mirror image of the other .
parametrize @xmath185 by @xmath186 with @xmath187 .
let be @xmath188\times s^1 $ ] ( @xmath43 is an annulus ) , and the support of @xmath183 be the cylinder @xmath189 $ ] .
the coordinates of @xmath190 are @xmath109 , @xmath186 , and @xmath191 , similar to the cylindrical coordinates in @xmath192 .
the support of @xmath193 is also @xmath194 $ ] , but it will be denoted @xmath195 to avoid confusion ; let @xmath196 .
the semi - plugs @xmath183 and @xmath184 are generated by the vector fields @xmath197 and @xmath198 the concatenation identifies the top @xmath199 of @xmath190 with the bottom @xmath200 of @xmath195 ; the map @xmath201 sends @xmath43 onto the bottom of @xmath202 by @xmath203 .
the vector fields parallel to @xmath183 and @xmath184 are oriented in the positive @xmath191 direction .
since @xmath190 and @xmath195 are trivially foliated in the respective neighborhoods of @xmath204 and @xmath205 and they are analytic , the plug @xmath206 is analytic .
let @xmath207 and @xmath208 be the circles of @xmath183 and @xmath193 .
we define a self - insertion whose domain @xmath209 consists of two components @xmath210 and @xmath211 . the self - insertion map @xmath212 divides into a map @xmath213 given by the formula @xmath214 where @xmath210 is the set of all points @xmath215 such that @xmath216 lies in @xmath190 , and a map @xmath217 similarly defined by @xmath218 where @xmath211 is the set of all points @xmath215 such that @xmath219 lies in @xmath195 . to check that the equations for @xmath80 form a valid self - insertion
, we first note that @xmath210 and @xmath211 are disjoint , because if @xmath220 , then @xmath221 , while if @xmath222 , then @xmath223 .
each leaf of @xmath224 intersects @xmath225 at most once , because the slope @xmath226 of any leaf in @xmath224 is bigger than @xmath132 and is positive in @xmath190 .
the variable @xmath186 would have to increase by 8 for a leaf to connected @xmath227 with @xmath228 and by 10 for it to connect either component with itself , but this is impossible .
finally , the condition that @xmath80 avoid leaves in @xmath224 that begin in its domain is automatically satisfied , because such leaves have @xmath229 while @xmath225 only intersects @xmath224 at @xmath230 . the self - inserted plug @xmath231 given by @xmath80
is pictured in figure [ fdad ] .
observe that @xmath232 and that @xmath233 and that @xmath234 and @xmath235 are both in the stopped set of @xmath182 .
in other words , the self - insertion breaks both circles of @xmath182 .
moreover , if @xmath236 , then @xmath237 , with equality occurring only for two points of @xmath43 ; one point is sent to the circle @xmath54 by @xmath80 and the other is sent to @xmath238 .
this is the important _ radius inequality _ for self - insertions .
a point @xmath239 may be considered as a point in @xmath202 . to make a distinction between the leaves of @xmath231 and @xmath240 containing @xmath22 ,
the leaves are denoted by @xmath241 and @xmath105 .
the self - inserted plug @xmath231 has no circular leaves .
[ thnocirc ] let @xmath242 be the stack of some point in @xmath243 . by the radius inequality
, the points have strictly decreasing radii @xmath244 if some @xmath245 is either @xmath234 or @xmath235 , then the stack ends at @xmath245 , because @xmath246 lies on the circle @xmath54 or @xmath238 and @xmath247 is undefined . at all other points of @xmath248 ,
the radius inequality is a strict inequality .
suppose that @xmath241 is a circular leaf of @xmath231 and let @xmath22 be a point that varies along @xmath241 .
as @xmath22 goes all the way around @xmath241 , the stack of @xmath22 either grows , shrinks , or stays at the same level .
if it grows or shrinks , then it must be infinite , and moreover it must be eventually periodic , which is impossible by the radius inequality . if it stays at the same level , then for some @xmath249 the stack reaches a minimum height ; the segment of @xmath241 containing @xmath30 must therefore correspond to some leaf @xmath105 of @xmath182 which is interrupted but does not interrupt other leaves .
the leaf @xmath105 must be a circle , and must therefore be one of the two circles @xmath250 and @xmath251 .
however , the transition points on these circles lead to the entry points @xmath234 and @xmath235 which lie in the stopped set of @xmath182 and can not be matched . in conclusion ,
all avenues for a circular leaf in @xmath231 lead to contradiction .
if @xmath241 is a leaf of @xmath231 whose radii avoid the interval @xmath252 for some @xmath253 , then @xmath241 is a finite leaf .
[ lrlimit ] let @xmath254 be the radii of the stack of some point on such a leaf @xmath241 .
by hypothesis the points of the stack are at least @xmath125 away from the critical points @xmath234 and @xmath235 .
therefore there exists a @xmath255 such that @xmath256 , which in turn implies that the height of the stack is bounded by @xmath257 .
moreover , all leaves of @xmath182 that intersect the image of @xmath80 infinitely many times have @xmath258 , so there exists an @xmath259 such that all component leaves of @xmath241 intersect @xmath80 at most @xmath259 times . modelling the behavior of @xmath241 by followleaf ,
each instantiation of followleaf can only call followleaf at most n times , and the recursion can only extend to a depth of @xmath257 .
therefore there are at most @xmath260 internal calls , which means that @xmath241 has at most @xmath260 internal entry points , or @xmath261 transition points in total .
this implies that @xmath241 is finite , since none of the component leaves is infinite .
if @xmath241 is a leaf of @xmath231 with entry point at @xmath258 , then all subsequent transition points are matched and the positive limit set of @xmath241 contains the ( formerly circular ) leaf @xmath250 .
[ lrbig ] at the bottom level , the leaf @xmath241 follows a leaf @xmath105 of @xmath182 which approaches @xmath54 .
this leaf @xmath241 meets an infinite sequence of transition points , all of which have @xmath262 .
since the radii of higher - level transition points are only larger , lemma [ lrlimit ] implies that all of these transition points are matched .
a similar argument shows that the negative limit set of a leaf of @xmath263 with exit point at @xmath258 contains the leaf @xmath251 .
the plug @xmath231 has a unique non - trivial minimal set and it is 2-dimensional .
[ thdim ] by lemmas [ lrlimit ] and [ lrbig ] , the closure of any infinite leaf contains the leaf @xmath250 .
therefore there is only one minimal set and it contains @xmath250 . since @xmath232 , to understand the minimal set it suffices to follow the leaf @xmath241 of @xmath231 starting at @xmath264 .
let @xmath265 be the antiderivative of @xmath266 with @xmath267 ; in particular , @xmath268 by abuse of notation . solving the differential equation defined by @xmath269 , the leaves of @xmath183
are given by the equation @xmath270 when @xmath271 , and by @xmath272 when @xmath273 .
if @xmath105 is the leaf of @xmath183 starting at @xmath274 , then @xmath105 is given by the former equation with @xmath275 .
for each integer @xmath276 , @xmath105 intersects the surface @xmath277 with @xmath278 and for @xmath57 large enough , these points also lie in @xmath227 . for such @xmath57 , choose @xmath279 so that @xmath280 then the numbers @xmath281 as @xmath282 , while @xmath283 as @xmath282 by inspection of the relation @xmath284 let @xmath285 be the leaf of @xmath183 beginning at @xmath286 . by lemma [ lrbig ] , in the leaf @xmath241 , each leaf @xmath285 concatenated with its mirror image interrupts the leaf @xmath105 .
let @xmath287 be the other endpoint of @xmath285 in @xmath190 ; @xmath288 is given by @xmath289 combining several identities and approximations , @xmath290 for @xmath57 large .
thus , @xmath291 , but @xmath292 as @xmath282 .
it follows that @xmath293 is a dense subset of the circle @xmath294 , and that the circle @xmath58 of all points @xmath295 is in the closure of @xmath241 . following the leaves of @xmath296 and @xmath184 , the circle @xmath58 sweeps out the entire annulus @xmath259 between @xmath54 and @xmath238 .
the surface @xmath259 , except where it is cut by @xmath225 , is therefore in the closure of @xmath241 in @xmath231 , which demonstrates that the minimal set of @xmath297 is 2-dimensional .
finally , the following proposition was pointed out to the authors by .
ghys , who credits s. matsumoto @xcite .
it slightly simplifies section [ sstoppage ] , since it means that it is not necessary to insert @xmath231 into another content - stopping plug . the plug @xmath231 stops content . [ pcontent ]
the set of points @xmath298 with @xmath299 is the region in the @xmath277 plane with @xmath300 . in this region ,
the slope of a leaf of @xmath182 satisfies @xmath301 if @xmath105 is a leaf of @xmath206 at radius @xmath109 with @xmath302 , then in the region @xmath300 , the successive intersections of @xmath105 with @xmath225 are less than @xmath303 apart .
but the region @xmath300 has width @xmath304 in the @xmath191 direction , so it follows that the leaf in @xmath241 in @xmath231 containing segments of @xmath105 eventually meets an internal entry point with interrupting radius @xmath305 such that @xmath306 .
all entry points that @xmath105 meets at radius @xmath307 are matched by lemma [ lrbig ] , but an entry at radius @xmath308 can not be matched , and an entry at radius @xmath309 begins a new leaf @xmath310 of @xmath182 with the same properties as @xmath105 .
therefore the stack of @xmath241 grows indefinitely and @xmath241 is an infinite leaf . in conclusion ,
all points of @xmath311 with radius @xmath302 are in the stopped set of @xmath231 .
proposition [ pcontent ] applies equally well to any conceivable variation of the formulas defining the plug @xmath231 , with the conclusion that no such analytic ( or even @xmath1 ) plug can preserve volume given by a volume form .
on the other hand , proposition [ pcontent ] is valid for some but not all pl self - insertions .
nevertheless , it seems impossible to satisfy the radius inequality and preserve volume also .
reference @xcite gives a version of the schweitzer plug which preserves volume . like schweitzer s example , it is @xmath1 .
we first consider oriented 1-foliations in @xmath57 dimensions with @xmath312 .
let @xmath313 be an @xmath118-dimensional torus , parametrized by coordinates @xmath314 with period @xmath315 .
let @xmath316 be a vector field , where the coefficients @xmath317 are between @xmath308 and @xmath132 and are linearly independent over @xmath318 , the rational numbers .
all leaves of the foliation parallel to @xmath319 are dense .
parametrize the manifold @xmath320
\times t^{n-2 } \times [ -1,1]$ ] by the coordinates @xmath321 , and define the vector field @xmath322 on @xmath323 by the formula @xmath324 the foliation @xmath325 parallel to @xmath322 is a semi - plug , and the mirror - image construction yields a plug @xmath326 with base @xmath327 .
the plug @xmath326 is aperiodic and is similar to a construction of wilson which settled the higher - dimensional seifert conjecture for all manifolds with euler characteristic 0 .
however , its minimal sets are at most @xmath118-dimensional ; to achieve @xmath328-dimensional minimal sets we will perform a self - insertion .
let @xmath329 in the parametrization by the @xmath330 s , and let @xmath331 .
consider a self - insertion with two components , one component going into @xmath325 and the other going into @xmath332 .
the formula for the part of @xmath80 that maps into @xmath323 is @xmath333 while the formula that maps into @xmath334 is @xmath335 let @xmath336 be the self - inserted plug .
the self - insertion clearly satisfies the radius inequality with respect to the variable @xmath109 .
the base after self - insertion is @xmath327 with the boundary components connected by two boundary handles , which admits a bridge immersion in @xmath82 as illustrated in figure [ ftoungue ] in the case @xmath337 . following the argument for three dimensions ,
if @xmath105 is the leaf of @xmath183 beginning at @xmath338 , then @xmath241 is in the minimal set of @xmath336 . in @xmath241 ,
the leaf @xmath105 is interrupted by a sequence of leaves @xmath285 with radius @xmath339 . in the @xmath330 directions , these leaves progress in the @xmath330 directions by an amount that goes to infinity as @xmath282 , but at a rate that goes to 0 .
it follows that the entire surface with @xmath271 and @xmath340 in @xmath323 is in the closure of the leaves @xmath285 , which in turn forces the minimal set of @xmath336 to have codimension 1 . in the general case ,
the goal is to open the leaves of a @xmath130-dimensional foliation of an @xmath57-manifold so that the closure of any such leaf is at least @xmath328-dimensional .
if @xmath341 , there is almost nothing to prove . to eliminate the possibility of a minimal set of codimension 0 ( the whole manifold ) ,
choose a circle or properly embedded line transverse to the foliation and put in a reeb structure along this curve .
the interesting case is @xmath342 . following schweitzer @xcite ,
consider the manifold @xmath343 foliated by leaves @xmath344 , where @xmath105 is a leaf of @xmath345 and @xmath346 is the @xmath130-sphere .
the techniques of insertion and global stoppage readily generalize to such a foliation , and it is easy to check that the only minimal set has codimension 1 .
the pl construction mostly uses the same geometric ideas as the analytic ( and therefore smooth ) case , but they are realized somewhat differently because pl foliations do not have well - behaved parallel vector fields .
the construction has one new detail , a vertical annulus of circles , which gives the minimal set high dimension in a different way .
let @xmath347 be a compact manifold , usually with boundary or corners .
let @xmath348 \to h \times [ a , b]$ ] be a pl homeomorphism and let @xmath349 be a real number .
let @xmath350 be the foliation of @xmath351 \times [ 0,1]$ ] such that , for fixed @xmath352 and @xmath353 , the set @xmath354,z+lx \in [ a , b]\}$ ] is a leaf .
orient all such leaves from @xmath351 \times \{0\}$ ] to @xmath351 \times \{1\}$ ] .
the _ slanted suspension _ of @xmath355 with _ slant _ @xmath105 is defined as @xmath356 , the manifold @xmath351\times [ 0,1]$ ] with @xmath357 identified with @xmath358 , together with the foliation @xmath359 induced from @xmath350 . the slanted suspension @xmath360 is a pl foliation by construction , and moreover is a flow bordism with entry and exit regions @xmath361 .
let @xmath362 $ ] and @xmath363=[-2,2]$ ] .
let @xmath364\times [ -2,2 ] \to [ -1,1]\times [ -2,2]$ ] be a pl homeomorphism which is the identity at @xmath365\times [ -2,2])$ ] , such that @xmath366 , @xmath367 , and such that @xmath355 is linear on the two triangles and two trapezoids illustrated in figure [ fcollar ] .
it is easy to check that the slanted suspension of @xmath355 with slant @xmath368 is a semi - plug with an annulus of circular leaves corresponding to the line segment from @xmath369 to @xmath370 .
it is convenient to take a 20-fold covering of the slanted suspension , to obtain a semi - plug @xmath371 whose support @xmath372\times [ 0,20]\times [ -2,2]$ ] ( where 0=20 ) is parametrized by @xmath109 , @xmath186 , and @xmath191 with the suspension direction @xmath186 .
the semi - plug @xmath373 is similar to the analytic semi - plug @xmath183 , except that it has an annulus of circular leaves instead of one circular leaf .
let @xmath374 ( with support @xmath375 ) be a the mirror image of @xmath373 obtained by changing @xmath191 to @xmath376 .
let @xmath377 ( with support @xmath378 ) be the plug obtained from @xmath373 by the mirror - image construction , the concatenation of @xmath373 and @xmath379 .
although @xmath377 has two annuli of circles , @xmath54 and @xmath238 , it is still possible to break all of them with a self - insertion , since the stopped set is also a circle .
the radius inequality preserving self - insertion @xmath380 is defined on two parts of @xmath133 : @xmath210 containing the segment with endpoints @xmath381 , @xmath382 which is mapped linearly onto the whole segment of @xmath54 at @xmath383 , and @xmath211 contaning the segment with endpoints @xmath384 , @xmath385 which is mapped linearly onto the whole segment of @xmath238 at @xmath386 .
( figure [ fvertical ] shows the @xmath210 part of @xmath80 . )
its geometry is more or less the same as that in the analytic case , since the leaves of @xmath377 are at constant @xmath109 and the self - insertion satisfies the radius inequality .
lemma [ lrlimit ] , slightly modified lemma [ lrbig ] , and theorem [ thdim ] hold for @xmath387 .
the internal entries of a point at @xmath271 have radii that converge to @xmath271 , and their leaves converge to the annuli of circles of @xmath377 .
therefore the remnants of these annuli after self - insertion are contained in the minimal set , and the minimal set is still 2-dimensional .
the construction for higher - dimensional foliations also apply to the pl case without modification .
although most aperiodic self - inserted wilson - type plugs have 2-dimensional minimal sets , a carefully chosen self - insertion may result in a 1-dimensional minimal set .
such a self - insertion has interesting symbolic dynamics which can be described explicitly .
the self - insertion is easiest to define in the continuous category , but with yet more care it is also possible in the pl category .
first , we consider a self - insertion that only breaks one circle .
the other circle is the unique minimal set , but the leaf containing the broken circle is in a 1-dimensional invariant set which resembles the minimal set in the final construction .
let @xmath388 \times [ -3,\frac32]$ ] be a rectangle parametrized by @xmath109 and @xmath191 , and let @xmath389 be the pl homeomorphism illustrated in figure [ fmapf ] .
the map @xmath355 is linear in each of the regions delineated by the four rays @xmath390 , @xmath391 , @xmath392 , and @xmath393 . in the upper region , @xmath394
; in the lower region , @xmath395 , and in the two side regions , @xmath355 matches the unique linear transformation which makes it continuous along the four rays .
note that @xmath396 , and for any other point @xmath397 , the difference between @xmath191 and the @xmath191-coordinate of @xmath398 is greater than @xmath399 .
let @xmath400 be the slanted suspension of @xmath355 with slant @xmath401 , and let @xmath402 be its support .
more precisely , @xmath402 is obtained from @xmath403\times [ 0,1 ] \times [ -3,\frac32]$ ] by identifying @xmath404 with @xmath405 , where @xmath406 .
we use the coordinates @xmath407 , and the suspension is in the direction of @xmath186 .
each leaf of @xmath400 is the union of segments @xmath408 , z \in { { \bbb r } } , \theta \in [ 0,1 ] , z+{\frac32}\theta \in [ -2,2]\}$ ] .
the semi - plug @xmath400 is another pl analogue of the analytic semi - plug @xmath183 . like @xmath409
, it has one circular leaf @xmath54 .
note that @xmath54 passes through the point @xmath410 .
let @xmath411 be the mirror image of @xmath412 , and let @xmath238 be the circular leaf of @xmath411 .
let @xmath413 be the concatenation of the two mirror - image semi - plugs @xmath414 and @xmath415 .
the support @xmath55 of @xmath416 is obtained by idetifying the top of @xmath417 with the bottom of @xmath418 .
let @xmath43 be the base of @xmath413 .
let @xmath419 .
we define a self - insertion map @xmath420 on an appropriate @xmath421 containing @xmath422 .
the image of @xmath80 lies in the @xmath423 section of @xmath402 , and @xmath424 is described explicitly on @xmath422 by : @xmath425 @xmath426 and @xmath427 . on @xmath428 , @xmath80 tapers in the @xmath191 direction to have @xmath429 small enough so that leaves of @xmath430 intersect @xmath225 at most once , and none of the leaves that do intersect have an endpoint in @xmath133 .
figure [ fsigma ] shows the shape of @xmath225 in the section of @xmath402 with @xmath423 .
the formulas demonstrate that @xmath80 satisfies the radius inequality on @xmath422 .
outside @xmath422 the radius inequality is easy to achieve .
let @xmath431 be @xmath432 self - inserted by @xmath80 .
let @xmath250 be the leaf of @xmath431 containing segments of @xmath54 .
by lemma [ lrbig ] , @xmath433 determines the geometry of the leaf @xmath250 .
let @xmath434 be a sequence of closed disks in @xmath422 such that @xmath435 let @xmath436 be the tube of leaves of @xmath413 with endpoints in @xmath437 .
the intersection of each @xmath436 with the @xmath423 section is also shown in figure [ f1dim ] , along with the regions @xmath438 .
the figure suggests , and a computation shows , that @xmath439 for every @xmath57 .
let @xmath440 and let @xmath441 .
the set @xmath442 is also a union of disjoint disks , which we denote @xmath443 .
let @xmath444 be the tube of leaves of @xmath413 which begins at @xmath445 .
in general , for each @xmath130 , let @xmath446 be a sequence of disjoint closed disks whose union is @xmath447 , and let @xmath448 be the tube of leaves beginning at @xmath449 .
another computation shows that the diameter of @xmath450 goes to zero as @xmath130 goes to infinity , irrespective of the behavior of @xmath57 as a function of @xmath130 .
moreover , figure [ f1dim ] shows that there are infinitely many ( and in particular more than one ) disks @xmath451 in a given @xmath450 .
therefore the closure of the intersection @xmath452 is a cantor set @xmath255 . in the leaf @xmath250 , the leaf @xmath54 is interrupted by the leaf @xmath105 beginning at @xmath453 . since the leaf @xmath105 is infinite , the leaf @xmath250 never returns to @xmath54 . the leaf @xmath105 , in turn ,
is interrupted by an infinite sequence of internal entry points , and the @xmath57th such point lies in @xmath454 .
it follows that the closure of all internal entry points , which is a cross - section of the closure of @xmath250 , is the cantor set @xmath255 .
in particular , the closure of @xmath250 is 1-dimensional .
the main extra step in the full construction is to take the double - cover @xmath455 of @xmath413 .
similarly , let @xmath456 , @xmath457 , and @xmath458 be double - covers of @xmath402 , @xmath459 , and @xmath43 .
let @xmath460 be the covering map , let @xmath461 be the non - trivial deck translation ( also use @xmath462 and @xmath53 for the covering @xmath458 of @xmath43 ) , and let @xmath463 be the mirror - image involution which switches @xmath456 and @xmath457 .
let @xmath464 be a lift of @xmath465 , where the disk @xmath466 is a lift of the disk @xmath133 , let @xmath467 , and let @xmath468 be given by @xmath469 let @xmath470 , and define the self - insertion map @xmath471 to be both @xmath472 and @xmath473 .
the entire system of disks @xmath474 and tubes @xmath475 has a lift @xmath476 and @xmath477 and another lift @xmath478 and @xmath479 .
moreover , each @xmath480 intersects @xmath481 in some collection of disks @xmath482 , where @xmath483 and @xmath484 may both vary .
so the closure of @xmath250 still has a cantor set cross - section and is still 1-dimensional , but since the self - insertion at @xmath485 breaks both circles @xmath54 and @xmath238 , the closure of @xmath250 is in this case the minimal set .
the construction of theorem [ th1dplug ] gives rise to interesting symbolic dynamics .
consider the sequence of pairs of integers @xmath486 such that the @xmath487th internal entry of the leaf @xmath250 after the interruption of @xmath54 is in the disk @xmath488 .
the sequence runs : @xmath489 @xmath490 the sequence is generated by calling followdisks(1,@xmath491 ) , where the recursive procedure followdisks is defined as : * procedure followdisks(@xmath483,@xmath57 ) * print @xmath486 if @xmath57 is finite .
* do the following three steps with @xmath492 : * for each @xmath130 from @xmath132 to @xmath493 , do followdisks(@xmath487,@xmath130 ) if @xmath494 is odd .
* do followdisks(@xmath487,@xmath495 ) twice . * for each @xmath130 from @xmath493 to @xmath132 , do followdisks(@xmath487,@xmath130 ) if @xmath494 is even .
* quit .
finally , note that there are many different constructions that resemble the construction of theorem [ th1dplug ] , and in general they have similar but different symbolic dynamics .
for example , where followdisks says `` do followdisks(@xmath487,@xmath495 ) twice , '' its analogue for some other self - inserted plug may call itself once or three times or any other integer .
the above pl construction yielding a 1-dimensional minimal set easily generalizes to @xmath57-dimensional manifolds , @xmath117 : for any @xmath496 , there is an aperiodic pl plug with one minimal set which is of dimension @xmath130 . | using the theory of plugs and the self - insertion construction due to the second author , we prove that a foliation of any codimension of any manifold can be modified in a real analytic or piecewise - linear fashion so that all minimal sets have codimension 1 . in particular , the 3-sphere @xmath0 has a real analytic dynamical system such that all limit sets are 2-dimensional .
we also prove that a 1-dimensional foliation of a manifold of dimension at least 3 can be modified in a piecewise - linear fashion so that so that there are no closed leaves but all minimal sets are 1-dimensional .
these theorems provide new counterexamples to the seifert conjecture , which asserts that every dynamical system on @xmath0 with no singular points has a periodic trajectory .
[ theorem]proposition [ theorem]lemma [ theorem]definition [ theorem]corollary 1dim # 1 by # 2 ( # 3 ) to # 2 ' '' '' width # 1 height 0pt depth 0pt |
in this article , we consider a class of homogeneous partial differential operators on a finite dimensional vector space and study their associated heat kernels .
these operators , which we call nondegenerate - homogeneous operators , are seen to generalize the well - studied classes of semi - elliptic operators introduced by f. browder @xcite , also known as quasi - elliptic operators @xcite , and a special positive " subclass of semi - elliptic operators which appear as the spatial part of s. d. eidelman s @xmath0-parabolic operators @xcite . in particular , this class of operators contains all integer powers of the laplacian . to motivate the definition of nondegenerate - homogeneous operators , given in the next section
, we first introduce the class of semi - elliptic operators .
semi - elliptic operators are seen to be prototypical examples of nondegenerate - homogeneous operators ; in fact , the definition of nondegenerate - homogeneous operators is given to formulate the following construction in a basis - independent way .
given @xmath1-tuple of positive integers @xmath2 and a multi - index @xmath3 , set @xmath4 .
consider the constant coefficient partial differential operator @xmath5 with principal part ( relative to @xmath6 ) @xmath7 where @xmath8 and @xmath9 for each multi - index @xmath10 .
such an operator @xmath11 is said to be _ semi - elliptic _ if the symbol of @xmath12 , defined by @xmath13 for @xmath14 , is non - vanishing away from the origin .
if @xmath11 satisfies the stronger condition that @xmath15 is strictly positive away from the origin , we say that it is _ positive - semi - elliptic_. what seems to be the most important property of semi - elliptic operators is that their principal part @xmath12 is homogeneous in the following sense : if given any smooth function @xmath16 we put @xmath17 for all @xmath18 and @xmath19 , then @xmath20 for all @xmath18 .
this homogeneous structure was used explicitly in the work of f. browder and l. hrmander and , in this article , we generalize this notion .
we note that our definition for the differential operators @xmath21 is given to ensure a straightforward relationship between operators and symbols under our convention for the fourier transform ( defined in subsection [ subsec : preliminaries ] ) ; this definition differs only slightly from the standard references @xcite in which @xmath22 is replaced by @xmath23 . in both conventions , the symbol of the operator @xmath24 is the positive polynomial @xmath25 . in fact , the principal symbols of all positive - semi - elliptic operators agree in both conventions . +
as mentioned above , the class of semi - elliptic operators was introduced by f. browder in @xcite who studied spectral asymptotics for a related class of variable - coefficient operators ( operators of constant strength ) .
elliptic operators appeared later in l. hrmander s text @xcite as model examples of hypoelliptic operators on @xmath26 beyond the class of elliptic operators . around the same time ,
l. r. volevich @xcite independently introduced the same class of operators but instead called them `` quasi - elliptic '' .
since then , the theory of semi - elliptic operators , and hence quasi - elliptic operators , has reached a high level of sophistication and we refer the reader to the articles @xcite , which use the term semi - elliptic , and the articles @xcite , which use the term quasi - elliptic , for an account of this theory .
we would also like to point to the 1971 paper of m. troisi @xcite which gives a more complete list of references ( pertaining to quasi - elliptic operators ) .
+ shortly after f. browder s paper @xcite appeared , s. d. eidelman considered a subclass of semi - elliptic operators on @xmath27 ( and systems thereof ) of the form @xmath28 where @xmath29 and the coefficients @xmath30 are functions of @xmath31 and @xmath32 .
such an operator is said to be _
@xmath33-parabolic _ if its spatial part , @xmath34 , is ( uniformly ) positive - semi - elliptic .
we note however that eidelman s work and the existing literature refer exclusively to @xmath0-parabolic operators , i.e. , where @xmath35 , and for consistency we write @xmath0-parabolic henceforth @xcite . the relationship between positive - semi - elliptic operators and @xmath0-parabolic operators is analogous to the relationship between the laplacian and the heat operator and , in the context of this article , the relationship between nondegenerate - homogeneous and positive - homogeneous operators described by proposition [ prop : dichotomy ] .
the theory of @xmath0-parabolic operators , which generalizes the theory of parabolic partial differential equations ( and systems ) , has seen significant advancement by a number of mathematicians since eidelman s original work .
we encourage the reader to see the recent text @xcite which provides an account of this theory and an exhaustive list of references .
it should be noted however that the literature encompassing semi - elliptic operators and quasi - elliptic operators , as far as we can tell , has very few cross - references to the literature on @xmath0-parabolic operators beyond the 1960s .
we suspect that the absence of cross - references is due to the distinctness of vocabulary .
+ we motivate the study of homogeneous operators by first demonstrating the natural appearance of their heat kernels in the study of convolution powers of complex - valued functions . to this end , consider a finitely supported function @xmath37 and define its convolution powers iteratively by @xmath38 for @xmath39 where @xmath40 . in the special case that @xmath41 is a probability distribution , i.e. , @xmath41 is non - negative and has unit mass , @xmath41 drives a random walk on @xmath36 whose nth - step transition kernels are given by @xmath42 . under certain mild conditions on the random walk , @xmath43 is well - approximated by a single gaussian density ; this is the classical local limit theorem .
specifically , for a symmetric , aperiodic and irreducible random walk , the theorem states that @xmath44 uniformly for @xmath39 , where @xmath45 is the generalized gaussian density @xmath46 here , @xmath47 is the positive definite covariance matrix associated to @xmath41 and @xmath48 denotes the dot product @xcite .
the canonical example is that in which @xmath49 ( e.g. simple random walk ) and in this case @xmath43 is approximated by the so - called heat kernel @xmath50 defined by @xmath51 for @xmath18 and @xmath52 .
indeed , we observe that @xmath53 for each positive integer @xmath54 and @xmath39 and so the local limit theorem is written equivalently as @xmath55 uniformly for @xmath39 .
in addition to its natural appearance as the _ attractor _ in the local limit theorem above , @xmath56 is a fundamental solution to the heat equation @xmath57 on @xmath58 .
in fact , this connection to random walk underlies the heat equation s probabilistic / diffusive interpretation . beyond the probabilistic
setting , this link between convolution powers and fundamental solutions to partial differential equations persists as can be seen in the examples below . in what follows , the heat kernels @xmath59 are fundamental solutions to the corresponding heat - type equations of the form @xmath60 the appearance of @xmath61 in local limit theorems ( for @xmath43 ) is then found by evaluating @xmath62 at integer time @xmath63 and lattice point @xmath39 .
[ ex : intro1 ] consider @xmath64 defined by @xmath65 analogous to the probabilistic setting , the large @xmath54 behavior of @xmath43 is described by a generalized local limit theorem in which the attractor is a fundamental solution to a heat - type equation . specifically , the following local limit theorem holds ( see @xcite for details ) : @xmath66 uniformly for @xmath67 where @xmath59 is the `` heat '' kernel for the heat - type equation @xmath68 where @xmath69 this local limit theorem is illustrated in figure [ fig : ex1_intro ] which shows @xmath70 and the approximation @xmath71 when @xmath72 .
[ ex : intro2 ] consider @xmath73 defined by @xmath74 , where @xmath75 in this example , the following local limit theorem , which is illustrated by figure [ fig : ex2_intro ] , describes the limiting behavior of @xmath43 .
we have @xmath76 uniformly for @xmath67 where @xmath61 is again a fundamental solution to @xmath68 where , in this case , @xmath77 [ ex : intro3 ] consider @xmath78 defined by @xmath79 here , the following local limit theorem is valid : @xmath80 uniformly for @xmath67 . here again , the attractor @xmath61 is the fundamental solution to @xmath68 where @xmath81 looking back at preceding examples , we note that the operators appearing in examples [ ex : intro1 ] and [ ex : intro2 ] are both positive - semi - elliptic and consist only of their principal parts .
this is easily verified , for @xmath82 in example [ ex : intro1 ] and @xmath83 in example [ ex : intro2 ] .
in contrast to examples [ ex : intro1 ] and [ ex : intro2 ] , the operator @xmath11 which appears in example [ ex : intro3 ] is not semi - elliptic in the given coordinate system . after careful study ,
the @xmath11 appearing in example [ ex : intro3 ] can be written equivalently as @xmath84 where @xmath85 is the directional derivative in the @xmath86 direction and @xmath87 is the directional derivative in the @xmath88 direction . in this way
, @xmath11 is seen to be semi - elliptic with respect to some basis @xmath89 of @xmath90 and , with respect to this basis , we have @xmath91 .
for this reason , our formulation of nondegenerate - homogeneous operators ( and positive - homogeneous operators ) , given in the next section , is made in a basis - independent way .
+ all of the operators appearing in examples [ ex : intro1 ] , [ ex : intro2 ] and [ ex : intro3 ] share two important properties : homogeneity and positivity ( in the sense of symbols ) .
while we make these notions precise in the next section , loosely speaking , homogeneity is the property that @xmath11 `` plays well '' with some dilation structure on @xmath26 , though this structure is different in each example .
further , homogeneity for @xmath11 is reflected by an analogous one for the corresponding heat kernel @xmath61 ; in fact , the specific dilation structure is , in some sense , selected by @xmath43 as @xmath92 and leads to the corresponding local limit theorem . in further discussion of these examples ,
a very natural question arises : given @xmath37 , how does one compute the operator @xmath11 whose heat kernel @xmath61 appears as the attractor in the local limit theorem for @xmath43 ? in the examples we have looked at ,
one studies the taylor expansion of the fourier transform @xmath93 of @xmath41 near its local extrema and , here , the symbol of the relevant operator @xmath11 appears as certain scaled limit of this taylor expansion . in general , however , this is a very delicate business and , at present , there is no known algorithm to determine these operators .
in fact , it is possible that multiple ( distinct ) operators can appear by looking at the taylor expansions about distinct local extrema of @xmath93 ( when they exist ) and , in such cases , the corresponding local limit theorems involve sums of of heat kernels each corresponding to a distinct @xmath11 .
this study is carried out in the article @xcite wherein local limit theorems involve the heat kernels of the positive - homoegeneous operators studied in the present article .
we note that the theory presented in @xcite is not complete , for there are cases in which the associated taylor approximations yield symbols corresponding to operators @xmath11 which fail to be positive - homogeneous ( and hence fail to be positive - semi - elliptic ) and further , the heat kernels of these ( degenerate ) operators appear as limits of oscillatory integrals which correspond to the presence of odd " terms in @xmath11 , e.g. , the airy function . in one dimension , a complete theory of local limit theorems is known for the class of finitely supported functions @xmath94 . beyond one dimension ,
a theory for local limit theorems of complex - valued functions , in which the results of @xcite will fit , remains open .
+ the subject of this paper is an account of positive - homogeneous operators and their corresponding heat equations . in section [ sec : homogeneousoperators ] , we introduce positive - homogeneous operators and study their basic properties ; therein , we show that each positive - homogeneous operator is semi - elliptic in some coordinate system .
section [ sec : holder ] develops the necessary background to introduce the class of variable - coefficient operators studied in this article ; this is the class of @xmath95-positive - semi - elliptic operators introduced in section [ sec : uniformlysemielliptic]each of which is comparable to a constant - coefficient positive - homogeneous operator . in section
[ sec : fundamentalsolution ] , we study the heat equations corresponding to uniformly @xmath95-positive - semi - elliptic operators with hlder continuous coefficients .
specifically , we use the famous method of e. e. levi , adapted to parabolic systems by a. friedman and s. d. eidelman , to construct a fundamental solution to the corresponding heat equation .
our results in this direction are captured by those of s. d. eidelman @xcite and the works of his collaborators , notably s. d. ivashyshen and a. n. kochubei @xcite , concerning @xmath0-parabolic systems .
our focus in this presentation is to highlight the essential role played by the legendre - fenchel transform in heat kernel estimates which , to our knowledge , has not been pointed out in the context of semi - elliptic operators . in a forthcoming work ,
we study an analogous class of operators , written in divergence form , with measurable - coefficients and their corresponding heat kernels .
this class of measurable - coefficient operators does not appear to have been previously studied .
the results presented here , using the legendre - fenchel transform , provides the background and context for our work there .
* fourier analysis : * our setting is a real @xmath1-dimensional vector space @xmath96 equipped with haar ( lebesgue ) measure @xmath97 and the standard smooth structure ; we do not affix @xmath96 with a norm or basis . the dual space of @xmath96 is denoted by @xmath98 and the dual pairing is denoted by @xmath99 for @xmath100 and @xmath101 .
let @xmath102 be the haar measure on @xmath98 which we take to be normalized so that our convention for the fourier transform and inverse fourier transform , given below , makes each unitary . throughout this article ,
all functions on @xmath96 and @xmath98 are understood to be complex - valued .
the usual lebesgue spaces are denoted by @xmath103 and equipped with their usual norms @xmath104 for @xmath105 . in the case that @xmath106 , the corresponding inner product on @xmath107 is denoted by @xmath108 . of course
, we will also work with @xmath109 ; here the @xmath110-norm and inner product will be denoted by @xmath111 and @xmath112 respectively .
the fourier transform @xmath113 and inverse fourier transform @xmath114 are initially defined for schwartz functions @xmath115 and @xmath116 by @xmath117 for @xmath101 and @xmath100 respectively . for the remainder of this article (
mainly when duality is nt of interest ) , @xmath118 stands for any real @xmath1-dimensional vector space ( and so is interchangeable with @xmath96 or @xmath98 ) . for a non - empty open set @xmath119 ,
we denote by @xmath120 and @xmath121 the set of continuous functions on @xmath122 and bounded continuous functions on @xmath122 , respectively .
the set of smooth functions on @xmath122 is denoted by @xmath123 and the set of compactly supported smooth functions on @xmath122 is denoted by @xmath124 .
we denote by @xmath125 the space of distributions on @xmath122 ; this is dual to the space @xmath124 equipped with its usual topology given by seminorms . a partial differential operator @xmath126 on @xmath118 is said to be _
hypoelliptic _ if it satisfies the following property : given any open set @xmath119 and any distribution @xmath127 which satisfies @xmath128 in @xmath122 , then necessarily @xmath129 . +
* dilation structure : * denote by @xmath130 and @xmath131 the set of endomorphisms and isomorphisms of @xmath118 respectively . given @xmath132
, we consider the one - parameter group @xmath133 defined by @xmath134 for @xmath18 .
these one - parameter subgroups of @xmath131 allow us to define continuous one - parameter groups of operators on the space of distributions as follows : given @xmath132 and @xmath18 , first define @xmath135 for @xmath136 by @xmath137 for @xmath138 . extending this to the space of distribution on @xmath118 in the usual way ,
the collection @xmath139 is a continuous one - parameter group of operators on @xmath140 ; it will allow us to define homogeneity for partial differential operators in the next section .
+ * linear algebra , polynomials and the rest : * given a basis @xmath141 of @xmath118 , we define the map @xmath142 by setting @xmath143 whenever @xmath144 .
this map defines a global coordinate system on @xmath118 ; any such coordinate system is said to be a linear coordinate system on @xmath118 . by definition ,
a polynomial on @xmath118 is a function @xmath145 that is a polynomial function in every ( and hence any ) linear coordinate system on @xmath118 . a polynomial @xmath146 on @xmath118
is called a nondegenerate polynomial if @xmath147 for all @xmath148 .
further , @xmath146 is called a positive - definite polynomial if its real part , @xmath149 , is non - negative and has @xmath150 only when @xmath151 .
the symbols @xmath152 mean what they usually do , @xmath153 denotes the set of non - negative integers and @xmath154\subseteq\mathbb{r}$ ] .
the symbols @xmath155 , @xmath156 and @xmath157 denote the set of strictly positive elements of @xmath158 , @xmath153 and @xmath159 respectively .
likewise , @xmath160 , @xmath161 and @xmath162 respectively denote the set of @xmath1-tuples of these aforementioned sets . given @xmath163 and a basis @xmath141 of @xmath118 , we denote by @xmath164 the isomorphism of @xmath118 defined by @xmath165 for @xmath166 .
we say that two real - valued functions @xmath16 and @xmath167 on a set @xmath168 are comparable if , for some positive constant @xmath169 , @xmath170 for all @xmath171 ; in this case we write @xmath172 . adopting the summation notation for semi - elliptic operators of l. hrmander s treatise @xcite , for a fixed @xmath2 , we write @xmath173 for all multi - indices @xmath3 . finally , throughout the estimates made in this article , constants denoted by @xmath169 will change from line to line without explicit mention .
in this section we introduce two important classes of homogeneous constant - coefficient on @xmath96 .
these operators will serve as `` model '' operators in our theory in the way that integer powers of the laplacian serves a model operators in the elliptic theory of partial differential equations . to this end , let @xmath11 be a constant - coefficient partial differential operator on @xmath96 and let @xmath174 be its symbol . specifically , @xmath146 is the polynomial on @xmath98 defined by @xmath175 for @xmath101 ( this is independent of @xmath100 precisely because @xmath11 is a constant - coefficient operator ) .
we first introduce the following notion of homogeneity of operators ; it is mirrored by an analogous notion for symbols which we define shortly .
given @xmath176 , we say that a constant - coefficient partial differential operator @xmath11 is homogeneous with respect to the one - parameter group @xmath177 if @xmath178 for all @xmath18 ; in this case we say that @xmath179 is a member of the exponent set of @xmath11 and write @xmath180 .
a constant - coefficient partial differential operator @xmath11 need not be homogeneous with respect to a unique one - parameter group @xmath177 , i.e. , @xmath181 is not necessarily a singleton .
for instance , it is easily verified that , for the laplacian @xmath182 on @xmath26 , @xmath183 where @xmath184 is the identity and @xmath185 is the lie algebra of the orthogonal group , i.e. , is given by the set of skew - symmetric matrices . despite this lack of uniqueness , when @xmath11 is equipped with a nondegenerateness condition ( see definition [ def : homogeneousoperators ] ) , we will find that trace is the same for each member of @xmath181 and this allows us to uniquely define an order " for @xmath11 ; this is lemma [ lem : trace ] .
+ given a constant coefficient operator @xmath11 with symbol @xmath146 , one can quickly verify that @xmath180 if and only if @xmath186 for all @xmath18 and @xmath101 where @xmath187 is the adjoint of @xmath179 .
more generally , if @xmath146 is any continuous function on @xmath118 and is satisfied for some @xmath188 , we say that @xmath146 _ is homogeneous with respect to _ @xmath189 and write @xmath190 .
this admitted slight abuse of notation should not cause confusion . in this language , we see that @xmath191 if and only if @xmath192 . + we remark that the notion of homogeneity defined above is similar to that put forth for homogeneous operators on homogeneous ( lie ) groups , e.g. , rockland operators @xcite .
the difference is mostly a matter of perspective : a homogeneous group @xmath193 is equipped with a fixed dilation structure , i.e. , it comes with a one - parameter group @xmath194 , and homogeneity of operators is defined with respect to this fixed dilation structure . by contrast , we fix no dilation structure on @xmath96 and formulate homogeneity in terms of an operator @xmath11 and the existence of a one - parameter group @xmath177 that `` plays '' well with @xmath11 in sense defined above . as seen in the study of convolution powers on the square lattice ( see @xcite ) , it useful to have this freedom .
[ def : homogeneousoperators ] let @xmath11 be constant - coefficient partial differential operator on @xmath96 with symbol @xmath146 .
we say that @xmath11 is a nondegenerate - homogeneous operator if @xmath146 is a nondegenerate polynomial and @xmath181 contains a diagonalizable endomorphism .
we say that @xmath11 is a positive - homogeneous operator if @xmath146 is a positive - definite polynomial and @xmath181 contains a diagonalizable endomorphism . for any polynomial @xmath146 on a finite - dimensional vector space @xmath118
, @xmath146 is said to be _ nondegenerate - homogeneous _ if @xmath146 is nondegenerate and @xmath195 , defined as the set of @xmath196 for which holds , contains a diagonalizable endomorphism .
we say that @xmath146 is _ positive - homogeneous _ if it is a positive - definite polynomial and @xmath195 contains a diagonalizable endomorphism . in this language
, we have the following proposition .
[ prop : operatorsymbolequivalence ] let @xmath11 be a positive homogeneous operator on @xmath96 with symbol @xmath146 . then @xmath11 is a nondegenerate - homogeneous operator if and only if @xmath146 is a nondegenerate - homogeneous polynomial .
further , @xmath11 is a positive - homogeneous operator if and only if @xmath146 is a positive - homogeneous polynomial . since the adjectives `` nondegenerate '' and `` positive '' , in the sense of both operators and polynomials ,
are defined in terms of the symbol @xmath146 , all that needs to be verified is that @xmath181 contains a diagonalizable endomorphism if and only if @xmath195 contains a diagonalizable endomorphism . upon recalling that @xmath180 if and only if @xmath192 , this equivalence is verified by simply noting that diagonalizability is preserved under taking adjoints . to capture the class of nondegenerate - homogeneous operators ( or positive - homogeneous operators ) , in addition to requiring that that the symbol @xmath146 of an operator @xmath11 be nondegenerate ( or positive - definite ) , one can instead demand only that @xmath181 contains an endomorphism whose characteristic polynomial factors over @xmath158 or , equivalently , whose spectrum is real .
this a priori weaker condition is seen to be sufficient by an argument which makes use of the jordan - chevalley decomposition . in the positive - homogeneous case
, this argument is carried out in @xcite ( specifically proposition 2.2 ) wherein positive - homogeneous operators are first defined by this ( a priori weaker ) condition . for the nondegenerate case , the same argument pushes through with very little modification .
we observe easily that all positive - homogeneous operators are nondegenerate - homogeneous .
it is the `` heat '' kernels corresponding to positive - homogeneous operators that naturally appear in @xcite as the attractors of convolution powers of complex - valued functions .
the following proposition highlights the interplay between positive - homogeneity and nondegenerate - homogeneity for an operator @xmath11 on @xmath96 and its corresponding `` heat '' operator @xmath197 on @xmath198 . [
prop : dichotomy ] let @xmath11 be a constant - coefficient partial differential operator on @xmath96 whose exponent set @xmath181 contains a diagonalizable endomorphism .
let @xmath146 be the symbol of @xmath11 , set @xmath149 , and assume that there exists @xmath101 for which @xmath199 .
we have the following dichotomy : @xmath11 is a positive - homogeneous operator on @xmath96 if and only if @xmath197 is a nondegenerate - homogeneous operator on @xmath198 .
given a diagonalizable endomorphism @xmath180 , set @xmath200 where @xmath184 is the identity on @xmath158 .
obviously , @xmath201 is diagonalizable .
further , for any @xmath202 , @xmath203 for all @xmath204 and @xmath205 .
hence @xmath206 for all @xmath204 and therefore @xmath207 .
it remains to show that @xmath146 is positive - definite if and only if the symbol of @xmath197 is nondegenerate . to this end
, we first compute the symbol of @xmath197 which we denote by @xmath208 .
since the dual space of @xmath198 is isomorphic to @xmath209 , the characters of @xmath198 are represented by the collection of maps @xmath210 where @xmath211 .
consequently , @xmath212 for @xmath211 .
we note that @xmath213 because @xmath192 ; in fact , this happens whenever @xmath195 is non - empty . now if @xmath146 is a positive - definite polynomial , @xmath214 whenever @xmath215 .
thus to verify that @xmath208 is a nondegenerate polynomial , we simply must verify that @xmath216 for all non - zero @xmath217 .
this is easy to see because , in light of the above fact , @xmath218 whenever @xmath219 and hence @xmath208 is nondegenerate . for the other direction , we demonstrate the validity of the contrapositive statement . assuming that @xmath146 is not positive - definite , an application of the intermediate value theorem , using the condition that @xmath199 for some @xmath101 , guarantees that @xmath220 for some non - zero @xmath221 . here
, we observe that @xmath222 when @xmath223 and hence @xmath208 is not nondegenerate .
we will soon return to the discussion surrounding a positive - homogeneous operator @xmath11 and its heat operator @xmath197 .
it is useful to first provide representation formulas for nondegenerate - homogeneous and positive - homogeneous operators .
such representations connect our homogeneous operators to the class of semi - elliptic operators discussed in the introduction . to this end
, we define the `` base '' operators on @xmath96 .
first , for any element @xmath224 , we consider the differential operator @xmath225 defined originally for @xmath226 by @xmath227 for @xmath100 . fixing a basis @xmath228 of @xmath96
, we introduce , for each multi - index @xmath10 , @xmath229 .
[ prop : operatorrepresentation ] let @xmath11 be a nondegenerate - homogeneous operator on @xmath96 . then there exist a basis @xmath228 of @xmath96 and @xmath2 for which @xmath230 where @xmath231 . the isomorphism @xmath232 , defined by , is a member of @xmath181 .
further , if @xmath11 is positive - homogeneous , then @xmath233 for @xmath234 and hence @xmath235 we will sometimes refer to the @xmath6 and @xmath236 of the proposition as _
weights_. before addressing the proposition , we first prove the following mirrored result for symbols . [
lem : polynomialrepresentation ] let @xmath146 be a nondegenerate - homogeneous polynomial on a @xmath1-dimensional real vector space @xmath237 then there exists a basis @xmath141 of @xmath118 and @xmath2 for which @xmath238 for all @xmath239 where @xmath240 and @xmath241 .
the isomorphism @xmath242 , defined by , is a member of @xmath195 .
further , if @xmath146 is a positive - definite polynomial , i.e. , it is positive - homogeneous , then @xmath233 for @xmath234 and hence @xmath243 for @xmath244 .
let @xmath245 be diagonalizable and select a basis @xmath141 which diagonalizes @xmath179 , i.e. , @xmath246 where @xmath247 for @xmath248 . because @xmath146 is a polynomial , there exists a finite collection @xmath231 for which @xmath249 for @xmath244 . by invoking the homogeneity of @xmath146 with respect to @xmath179 and using the fact that @xmath250 for @xmath166
, we have @xmath251 for all @xmath244 and @xmath18 where @xmath252 . in view of the nondegenerateness of @xmath146 , the linear independence of distinct powers of @xmath32 and the polynomial functions @xmath253 , for distinct multi - indices @xmath254 ,
as @xmath255 functions ensures that @xmath256 unless @xmath257 .
we can therefore write @xmath258 for @xmath244 .
we now determine @xmath259 by evaluating this polynomial along the coordinate axes . to this end , by fixing @xmath166 and setting @xmath260 for @xmath261 , it is easy to see that the summation above collapses into a single term @xmath262 where @xmath263 ( here @xmath264 denotes the usual @xmath265th - euclidean basis vector in @xmath26 )
. consequently , @xmath266 for @xmath248 and thus , upon setting @xmath267 , yields @xmath238 for all @xmath244 as was asserted . in this notation
, it is also evident that @xmath268 . under the additional assumption that @xmath146 is positive - definite
, we again evaluate @xmath146 at the coordinate axes to see that @xmath269 for @xmath261 . in this case
, the positive - definiteness of @xmath146 requires @xmath270 and @xmath271 for each @xmath248 .
consequently , @xmath233 for @xmath234 as desired .
given a nondegenerate - homogeneous @xmath11 on @xmath96 with symbol @xmath146 , @xmath146 is necessarily a nondegenerate - homogeneous polynomial on @xmath98 in view of proposition [ prop : operatorsymbolequivalence ] .
we can therefore apply lemma [ lem : polynomialrepresentation ] to select a basis @xmath272 of @xmath98 and @xmath2 for which @xmath238 for all @xmath273 where @xmath231 .
we will denote by @xmath274 , the dual basis to @xmath275 , i.e. , @xmath228 is the unique basis of @xmath96 for which @xmath276 when @xmath277 and @xmath278 otherwise . in view of the duality of the bases @xmath274 and @xmath275 , it is straightforward to verify that , for each multi - index @xmath254 , the symbol of @xmath279 is @xmath280 in the notation of lemma [ lem : polynomialrepresentation ] .
consequently , the constant - coefficient partial differential operator defined by the right hand side of also has symbol @xmath146 and so it must be equal to @xmath11 because operators and symbols are in one - to - one correspondence . using , it is now straightforward to verify that @xmath281 .
the assertion that @xmath233 when @xmath11 is positive - homogeneous follows from the analogous conclusion of lemma [ lem : polynomialrepresentation ] by the same line of reasoning .
in view of proposition [ prop : operatorrepresentation ] , we see that all nondegenerate - homogeneous operators are semi - elliptic in some linear coordinate system ( that which is defined by @xmath274 ) .
an appeal to theorem 11.1.11 of @xcite immediately yields the following corollary . every nondegenerate - homogeneous operator @xmath11 on @xmath96 is hypoelliptic .
our next goal is to associate an `` order '' to each nondegenerate - homogeneous operator . for a positive - homogeneous operator @xmath11
, this order will be seen to govern the on - diagonal decay of its heat kernel @xmath61 and so , equivalently , the ultracontractivity of the semigroup @xmath282 . with the help of lemma [ lem : polynomialrepresentation ] ,
the few lemmas in this direction come easily .
[ lem : ptoinfty ] let @xmath146 be a nondegenerate - homogeneous polynomial on a @xmath1-dimensional real vector space @xmath118 . then @xmath283 ; here @xmath284 means that @xmath285 in any ( and hence every ) norm on @xmath118 .
the idea of the proof is to construct a function which bounds @xmath286 from below and obviously blows up at infinity . to this end , let @xmath287 be a basis for @xmath118 and take @xmath288 as guaranteed by lemma [ lem : polynomialrepresentation ] ; we have @xmath289 where @xmath290 for @xmath166 .
define @xmath291 by @xmath292 where @xmath293 .
we observe immediately @xmath294 because @xmath295 for @xmath166 .
an application of proposition [ prop : comparepoly ] ( a basic result appearing in our background section , section [ sec : holder ] ) , which uses the nondegenerateness of @xmath146 , gives a positive constant @xmath169 for which @xmath296 for all @xmath244 .
the lemma now follows by simply noting that @xmath297 as @xmath298 .
let @xmath146 be a polynomial on @xmath118 and denote by @xmath299 the set of @xmath300 for which @xmath301 for all @xmath244 .
if @xmath146 is a nondegenerate - homogeneous polynomial , then @xmath299 , called the symmetry group of @xmath146 , is a compact subgroup of @xmath131 .
our supposition that @xmath146 is a nondegenerate polynomial ensures that , for each @xmath302 , @xmath303 is empty and hence @xmath304 .
consequently , given @xmath305 and @xmath306 , we observe that @xmath307 and @xmath308 for all @xmath244 ; therefore @xmath299 is a subgroup of @xmath131 . to see that @xmath299 is compact , in view of the finite - dimensionality of @xmath131 and the heine - borel theorem
, it suffices to show that @xmath299 is closed and bounded .
first , for any sequence @xmath309 for which @xmath310 as @xmath92 , the continuity of @xmath146 ensures that @xmath311 for each @xmath244 and therefore @xmath299 is closed .
it remains to show that @xmath299 is bounded ; this is the only piece of the proof that makes use of the fact that @xmath146 is nondegenerate - homogeneous and not simply homogeneous .
assume that , to reach a contradiction , that there exists an unbounded sequence @xmath309 .
choosing a norm @xmath312 on @xmath118 , let @xmath313 be the corresponding unit sphere in @xmath118 .
then there exists a sequence @xmath314 for which @xmath315 for all @xmath316 but @xmath317 . in view of lemma [ lem :
ptoinfty ] , @xmath318 which can not be true for @xmath146 is necessarily bounded on @xmath313 because it is continuous .
[ lem : trace ] let @xmath11 be a nondegenerate - homogeneous operator . for any @xmath319 , @xmath320 let @xmath146 be the symbol of @xmath11 and take @xmath319 . since @xmath321 , @xmath322 for all @xmath18 . as @xmath299 is a compact group in view of the previous lemma , the determinant map @xmath323 , a lie group homomorphism , necessarily maps @xmath299 into the unit circle .
consequently , @xmath324 for all @xmath18 .
therefore , @xmath325 as desired . by the above lemma
, to each nondegenerate - homogenerous operator @xmath11 , we define the _ homogeneous order _ of @xmath11 to be the number @xmath326 for any @xmath180 . by an appeal to proposition [ prop : operatorrepresentation ] , @xmath281 for some @xmath327 and
so we observe that @xmath328 in particular , @xmath329 is a positive rational number .
we note that the term `` homogeneous - order '' does not coincide with the usual `` order '' for a partial differential operator . for instance
, the laplacian @xmath182 on @xmath26 is a second order operator ; however , because @xmath330 , its homogeneous order is @xmath331 .
we now restrict our attention to the study of positive - homogeneous operators and their associated heat kernels . to this end , let @xmath11 be a positive - homogeneous operator on @xmath96 with symbol @xmath146 and homogeneous order @xmath329 . the heat kernel for @xmath11 arises naturally from the study of the following cauchy problem for the corresponding heat equation @xmath68 : given initial data @xmath332
which is , say , bounded and continuous , find @xmath333 satisfying @xmath334 the initial value problem is solved by putting @xmath335 where @xmath336 is defined by @xmath337 for @xmath18 and @xmath100 ; we call @xmath61 _ the heat kernel _ associated to @xmath11 .
equivalently , @xmath61 is the integral ( convolution ) kernel of the continuous semigroup @xmath338 of bounded operators on @xmath107 with infinitesimal generator @xmath339 .
that is , for each @xmath340 , @xmath341 for @xmath18 and @xmath100 .
let us make some simple observations about @xmath61 .
first , by virtue of lemma [ lem : ptoinfty ] , it follows that @xmath342 for each @xmath18 .
further , for any @xmath180 , @xmath343 for @xmath18 and @xmath100 .
this computation immediately yields the so - called on - diagonal estimate for @xmath61 , @xmath344 for @xmath18 ; this is equivalently a statement of ultracontractivity for the semigroup @xmath282 .
as it turns out , we can say something much stronger .
[ prop : ccestimates ] let @xmath11 be a positive - homogeneous operator with symbol @xmath146 and homogeneous order @xmath329 .
let @xmath345 be the legendre - fenchel transform of @xmath149 defined by @xmath346 for @xmath100 .
also , let @xmath274 and @xmath29 be as guaranteed by proposition [ prop : operatorrepresentation ] .
then , there exit positive constants @xmath347 and @xmath348 and , for each multi - index @xmath254 , a positive constant @xmath349 such that , for all @xmath350 , @xmath351 for all @xmath352 and @xmath18 . in particular , @xmath353 for all @xmath352 and @xmath18 . in view of
, the exponent on the prefactor in can be equivalently written , for any multi - index @xmath254 and @xmath350 , as @xmath354 where @xmath355 .
we note that the estimates of proposition [ prop : ccestimates ] are written in terms of the difference @xmath356 and can ( trivially ) be expressed in terms of a single spatial variable @xmath31 .
the estimates are written in this way to emphasize the role that @xmath357 plays as an integral kernel .
we will later replace @xmath11 in by a comparable variable - coefficient operator @xmath126 and , in that setting , the associated heat kernel is not a convolution kernel and so we seek estimates involving two spatial variables @xmath31 and @xmath358 .
to that end , the estimates here form a template for estimates in the variable - coefficient setting .
we prove the proposition above in the section [ sec : fundamentalsolution ] ; the remainder of this section is dedicated to discussing the result and connecting it to the existing theory .
let us first note that the estimate is mirrored by an analogous space - time estimate , theorem 5.3 of @xcite , for the convolution powers of complex - valued functions on @xmath36 satisfying certain conditions ( see section 5 of @xcite ) .
the relationship between these two results , theorem 5.3 of @xcite and proposition [ prop : ccestimates ] , parallels the relationship between gaussian off - diagonal estimates for random walks and the analogous off - diagonal estimates enjoyed by the classical heat kernel @xcite .
+ let us first show that the estimates and recapture the well - known estimates of the theory of parabolic equations and systems in @xmath26 a theory in which the laplacian operator @xmath359 and its integer powers play a central role . to place things into the context of this article , let us observe that , for each positive integer @xmath360 , the partial differential operator @xmath361 is a positive - homogeneous operator on @xmath26 with symbol @xmath362 ; here , we identify @xmath26 as its own dual equipped with the dot product and euclidean norm @xmath312 .
indeed , one easily observes that @xmath363 is a positive - definite polynomial and @xmath364 where @xmath365 is the the identity .
consequently , the homogeneous order of @xmath361 is @xmath366 and the legendre - fenchel transform of @xmath367 is easily computed to be @xmath368 where @xmath369 .
hence , is the well - known estimate @xmath370 for @xmath371 and @xmath18 ; this so - called off - diagonal estimate is ubiquitous to the theory of higher - order " elliptic and parabolic equations @xcite . to write the derivative estimate in this context
, we first observe that the basis given by proposition [ prop : operatorrepresentation ] can be taken to be the standard euclidean basis , @xmath372 and further , @xmath373 is the ( isotropic ) weight given by the proposition .
writing @xmath374 and @xmath375 for each multi - index @xmath254 , takes the form @xmath376 for @xmath371 and @xmath18 , c.f . ,
* property 4 , p. 93 ) . + the appearance of the @xmath377-dimensional legendre - fenchel transform in heat kernel estimates
was previously recognized and exploited in @xcite and @xcite in the context of elliptic operators . due to the isotropic nature of elliptic operators ,
the @xmath377-dimensional transform is sufficient to capture the inherent isotropic decay of corresponding heat kernels . beyond the elliptic theory ,
the appearance of the full @xmath1-dimensional legendre - fenchel transform is remarkable because it sharply captures the general anisotropic decay of @xmath61 .
consider , for instance , the particularly simple positive - homogeneous operator @xmath378 on @xmath90 with symbol @xmath379 .
it is easily checked that the operator @xmath179 with matrix representation @xmath380 , in the standard euclidean basis , is a member of the @xmath181 and so the homogeneous order of @xmath11 is @xmath381 . here
we can compute the legendre - fenchel transform of @xmath382 directly to obtain @xmath383 for @xmath384 where @xmath385 and @xmath386 are positive constants . in this case ,
proposition [ prop : ccestimates ] gives positive constants @xmath347 and @xmath348 for which @xmath387 and @xmath18 where @xmath388 and @xmath389 .
we note however that @xmath11 is `` separable '' and so we can write @xmath390 where @xmath391 is the @xmath377-dimensional laplacian operator . in view of theorem 8 of @xcite and its subsequent remark ,
the estimate is seen to be sharp ( modulo the values of @xmath392 and @xmath169 ) . to further illustrate the proposition for a less simple positive - homogeneous operator , we consider the operator @xmath11 appearing in example [ ex : intro3 ] . in this case
, @xmath393 and one can verify directly that the @xmath394 , with matrix representation @xmath395 in the standard euclidean basis , is a member of @xmath181 . from this
, we immediately obtain @xmath396 and one can directly compute @xmath397 for @xmath384 where @xmath385 and @xmath386 are positive constants . an appeal to proposition [ prop : ccestimates ]
gives positive constants @xmath347 and @xmath348 for which @xmath398 for @xmath399 and @xmath18 where @xmath388 and @xmath389 .
furthermore , @xmath400 and the basis @xmath401 of @xmath90 given in discussion surrounding are precisely those guaranteed by proposition [ prop : operatorrepresentation ] .
appealing to the full strength of proposition [ prop : ccestimates ] , we obtain positive constants @xmath402 , @xmath348 and , for each multi - index @xmath254 , a positive constant @xmath403 such that , for each @xmath350 , @xmath404 for @xmath399 and @xmath18 where @xmath388 and @xmath389 . + in the context of homogeneous groups , the off - diagonal behavior for the heat kernel of a positive rockland operator ( a positive self - adjoint operator which is homogeneous with respect to the fixed dilation structure ) has been studied in @xcite
( see also @xcite ) .
given a positive rockland operator @xmath11 on homogeneous group @xmath193 , the best known estimate for the heat kernel @xmath61 , due to auscher , ter elst and robinson , is of the form @xmath405 where @xmath406 is a homogeneous norm on @xmath193 ( consistent with @xmath407 and @xmath408 is the highest order derivative appearing in @xmath11 . in the context of @xmath26 , given a symmetric and positive - homogeneous operator @xmath11 with symbol @xmath146 , the structure @xmath409 for @xmath410 where @xmath180 is a homogeneous group on which @xmath11 becomes a positive rockland operator .
on @xmath411 , it is quickly verified that @xmath412 is a homogeneous norm ( consistent with @xmath11 ) and so the above estimate is given in terms of @xmath413 which is , in general , dominated by the legendre - fenchel transform of @xmath414 . to see this ,
we need not look further than our previous and simple example in which @xmath415 . here
@xmath416 and so @xmath417 .
in view of , the estimate gives the correct decay along the @xmath418-coordinate axis ; however , the bounds decay at markedly different rates along the @xmath419-coordinate axis .
this illustrates that the estimate is suboptimal , at least in the context of @xmath26 , and thus leads to the natural question : for positive - homogeneous operators on a general homogeneous group @xmath193 , what is to replace the legendre - fenchel transform in heat kernel estimates ?
+ returning to the general picture , let @xmath11 be a positive - homogeneous operator on @xmath96 with symbol @xmath146 and homogeneous order @xmath329 . to highlight some remarkable properties about the estimates and in this general setting , the following proposition concerning @xmath420 is useful ; for a proof , see section 8.3 of @xcite .
[ prop : legendretransformproperties ] let @xmath11 be a positive - homogeneous operator with symbol @xmath146 and let @xmath420 be the legendre - fenchel transform of @xmath149 .
then , for any @xmath180 , @xmath421
. moreover @xmath420 is continuous , positive - definite in the sense that @xmath422 and @xmath423 only when @xmath424 .
further , @xmath420 grows superlinearly in the sense that , for any norm @xmath425 on @xmath96 , @xmath426 in particular , @xmath427 as @xmath428 .
let us first note that , in view of the proposition , we can easily rewrite , for any @xmath191 , as @xmath429 for @xmath352 and @xmath18 ; the analogous rewriting is true for .
the fact that @xmath420 is positive - definite and grows superlinearly ensures that the convolution operator @xmath282 defined by for @xmath18 is a bounded operator from @xmath430 to @xmath431 for any @xmath432 .
of course , we already knew this because @xmath433 is a schwartz function ; however , when replacing @xmath11 with a variable - coefficient operator @xmath126 , as we will do in the sections to follow , the validity of the estimate for the kernel of the semigroup @xmath434 initially defined on @xmath110 , guarantees that the semigroup extends to a strongly continuous semigroup @xmath435 on @xmath436 for all @xmath105 and , what s more , the respective infinitesimal generators @xmath437 have spectra independent of @xmath438 @xcite .
further , the estimate is key to establishing the boundedness of the riesz transform , it is connected to the resolution of kato s square root problem and it provides the appropriate starting point for uniqueness classes of solutions to @xmath439 @xcite . with this motivation in mind ,
following some background in section [ sec : holder ] , we introduce a class of variable - coefficient operators in section [ sec : uniformlysemielliptic ] called @xmath95-positive - semi - elliptic operators , each such operator @xmath126 comparable to a fixed positive - homogeneous operator . in section [ sec : fundamentalsolution ] , under the assumption that @xmath126 has hlder continuous coefficients and this notion of comparability is uniform , we construct a fundamental solution to the heat equation @xmath439 and show the essential role played by the legendre - fenchel transform in this construction . as mentioned previously , in a forthcoming work we will study the semigroup @xmath434 where @xmath126 is a divergence - form operator , which is comparable to a fixed positive - homogeneous operator , whose coefficients are at worst measurable . as the legendre - fenchel transform appears here by a complex change of variables followed by a minimization argument , in the measurable coefficient setting it appears quite naturally by an application of the so - called davies method , suitably adapted to the positive - homogeneous setting .
in this section , we provide the necessary background on one - parameter contracting groups , anisotropic hlder continuity , and the legendre - fenchel transform and its interplay with the two previous notions . in what follows , @xmath118 is a @xmath1-dimensional real vector space with a norm @xmath312 ; the corresponding operator norm on @xmath131 is denoted by @xmath406 . of course
, since everything is finite - dimensional , the usual topologies on @xmath118 and @xmath131 are insensitive to the specific choice of norms .
let @xmath440 be a continuous one - parameter group .
@xmath441 is said to be contracting if @xmath442 we easily observe that , for any diagonalizable @xmath132 with strictly positive spectrum , the corresponding one - parameter group @xmath443 is contracting . indeed ,
if there exists a basis @xmath141 of @xmath118 and a collection of positive numbers @xmath444 for which @xmath445 for @xmath248 , then the one parameter group @xmath443 has @xmath446 for @xmath248 and @xmath18 .
it then follows immediately that @xmath447 is contracting .
[ prop : comparepoly ] let @xmath208 and @xmath414 be continuous real - valued functions on @xmath118 . if @xmath448 for all @xmath148 and there exists @xmath449 for which @xmath447 is contracting , then , for some positive constant @xmath169 , @xmath450 for all @xmath451 .
if additionally @xmath452 for all @xmath148 , then @xmath453 .
let @xmath313 denote the unit sphere in @xmath118 and observe that @xmath454 because @xmath208 and @xmath414 are continuous and @xmath414 is non - zero on @xmath313 .
now , for any non - zero @xmath451 , the fact that @xmath455 is contracting implies that @xmath456 for some @xmath18 by virtue of the intermediate value theorem .
therefore , @xmath457 . in view of the continuity of @xmath208 and @xmath414 , this inequality must hold for all @xmath451 . when additionally @xmath452 for all non - zero @xmath458 , the conclusion that @xmath453 is obtained by reversing the roles of @xmath208 and @xmath414 in the preceding argument .
[ cor : movingconstants ] let @xmath11 be a positive - homogeneous operator on @xmath96 with symbol @xmath146 and let @xmath420 be the legendre - fenchel transform of @xmath149 .
then , for any positive constant @xmath348 , @xmath459 . by virtue of proposition [ prop : operatorrepresentation ] , let @xmath29 and @xmath274 be a basis for @xmath96 and for which @xmath460 . in view of proposition
[ prop : legendretransformproperties ] , @xmath420 and @xmath461 are both continuous , positive - definite and have @xmath462 . in view of
, it is easily verified that @xmath463 where @xmath464 and so it follows that @xmath465 is contracting .
the corollary now follows directly from proposition [ prop : comparepoly ] .
[ lem : scaling ] let @xmath146 be a positive - homogeneous polynomial on @xmath118 and let @xmath466 and @xmath287 be a basis for @xmath118 for which the conclusion of lemma [ lem : polynomialrepresentation ] holds .
let @xmath149 and let @xmath254 and @xmath467 be multi - indices such that @xmath468 ( in the standard partial ordering of multi - indices ) ; we shall assume the notation of lemma [ lem : polynomialrepresentation ] . 1 .
[ item : scaling1 ] for any @xmath316 such that @xmath469 , there exist positive constants @xmath348 and @xmath470 for which @xmath471 for all @xmath472 .
[ item : scaling2 ] if @xmath473 , there exist positive constants @xmath348 and @xmath470 for which @xmath474 for all @xmath475 .
[ item : scaling3 ] if @xmath473 and @xmath476 , then for every @xmath477 there exists a positive constant @xmath348 for which @xmath478 for all @xmath475 . assuming the notation of lemma [ lem : polynomialrepresentation ] ,
let @xmath479 and consider the contracting group @xmath480 on @xmath481 . because @xmath414 is a positive - definite polynomial , it immediately follows that @xmath482 is positive - definite .
let @xmath312 be a norm on @xmath481 and respectively denote by @xmath483 and @xmath313 the corresponding unit ball and unit sphere in this norm . to see item [ item : scaling1 ] , first observe that @xmath484 now , for any @xmath485 , because @xmath486 is contracting , it follows from the intermediate value theorem that , for some @xmath487 , @xmath488 .
correspondingly , @xmath489 because @xmath490 .
one obtains the constant @xmath470 and hence the desired inequality by simply noting that @xmath491 is bounded for all @xmath492 . for item [ item : scaling2 ]
, we use analogous reasoning to obtain a positive constant @xmath348 for which @xmath493 for all @xmath494 .
now , for any non - zero @xmath495 , the intermediate value theorem gives @xmath18 for which @xmath496 and hence @xmath497 where we have used the fact that @xmath498 and that @xmath499 .
as this inequality must also trivially hold at the origin , we can conclude that it holds for all @xmath472 , as desired .
finally , we prove item [ item : scaling3 ] . by virtue of item [ item : scaling2 ] , for any @xmath472 and @xmath18 , @xmath500 noting that @xmath501 because @xmath502 , we can make the coefficient of @xmath503 arbitrarily small by choosing @xmath32 sufficiently large and thereby obtaining the desired result . throughout the remainder of this article , @xmath274 will denote a fixed basis for @xmath96 and correspondingly we henceforth assume the notational conventions appearing in proposition [ prop : operatorrepresentation ] and @xmath233 is fixed . for @xmath504 , consider the homogeneous norm @xmath505 defined by @xmath506 for @xmath100 where @xmath507 .
as one can easily check , @xmath508 for all @xmath18 and @xmath100 where @xmath509 is defined by .
[ def : consistent ] let @xmath29 .
we say that @xmath504 is consistent with @xmath236 if @xmath510 for some @xmath511 .
as one can check , @xmath512 is consistent with @xmath236 if and only if @xmath513 where @xmath514 is defined by .
let @xmath515 and let @xmath516 .
we say that @xmath16 is @xmath274-hlder continuous on @xmath122 if for some @xmath517 and positive constant @xmath348 , @xmath518 .
in this case we will say that @xmath512 is the @xmath274-hlder exponent of @xmath16 .
if @xmath519 we will simply say that @xmath16 is @xmath274-hlder continuous with exponent @xmath512 .
the following proposition essentially states that , for bounded functions , hlder continuity is a local property ; its proof is straightforward and is omitted .
let @xmath520 be open and non - empty .
if @xmath16 is bounded and @xmath274-hlder continuous of order @xmath517 , then , for any @xmath521 , @xmath16 is also @xmath274-hlder continuous of order @xmath254 . in view of the proposition
, we immediately obtain the following corollary .
[ cor : holderconsistent ] let @xmath522 be open and non - empty and @xmath29 . if @xmath16 is bounded and @xmath274-hlder continuous on @xmath122 of order @xmath523 , there exists @xmath517 which is consistent with @xmath236 for which @xmath16 is also @xmath274-hlder continuous of order @xmath512 .
the statement follows from the proposition by choosing any @xmath512 , consistent with @xmath236 , such that @xmath524 .
the following definition captures the minimal regularity we will require of fundamental solutions to the heat equation .
let @xmath288 , @xmath274 be a basis of @xmath96 and let @xmath525 be a non - empty open subset of @xmath526\times\mathbb{v}$ ] .
a function @xmath333 is said to be @xmath527-regular on @xmath525 if on @xmath525 it is continuously differentiable in @xmath32 and has continuous ( spatial ) partial derivatives @xmath528 for all multi - indices @xmath254 for which @xmath529 . throughout this section ,
@xmath414 is the real part of the symbol @xmath146 of a positive - homogeneous operator @xmath11 on @xmath96 .
we assume the notation of proposition [ prop : legendretransformproperties ] ( and hence proposition [ prop : operatorrepresentation ] ) and write @xmath530 .
let us first record two important results which follow essentially from proposition [ prop : legendretransformproperties ] .
[ cor : legendrecomparehomogeneousnorm ] @xmath531 where @xmath514 was defined in . in view of propositions
[ prop : operatorrepresentation ] and [ prop : legendretransformproperties ] , @xmath532 . after recalling that @xmath465 is contracting , proposition [ prop : comparepoly ] yields the desired result immediately . by virtue of proposition [ prop : legendretransformproperties ]
, standard arguments immediately yield the following corollary .
[ cor : exponentialoflegendretransformintegrable ] for any @xmath477 and polynomial @xmath533 , i.e. , @xmath208 is a polynomial in any coordinate system , then @xmath534 [ lem : prelegendresubscale ] let @xmath535 .
then for any @xmath536 , there exists @xmath537 such that @xmath538 for all @xmath100 and @xmath539 . in view of corollary [ cor : legendrecomparehomogeneousnorm ] , it suffices to prove the statement @xmath540 for all @xmath100 and @xmath539 where @xmath537 and @xmath514 is given by .
but for any @xmath539 and @xmath100 , @xmath541 from which the result follows .
[ lem : legendresubscale ] let @xmath517 be consistent with @xmath236 . then there exists positive constants @xmath542 and @xmath543 such that @xmath544 and for any @xmath536 there exists @xmath537 such that @xmath545 for all @xmath100 and @xmath539 . by an appeal to corollary [ cor : legendrecomparehomogeneousnorm ] and lemma [ lem : prelegendresubscale ] , @xmath546 for all @xmath100 and @xmath539 . since @xmath512 is consistent with @xmath236 , @xmath513 where @xmath547 is that of definition [ def : consistent ] , the desired inequality follows by setting @xmath548 and @xmath549 . because @xmath550 , it is necessary that @xmath551 whence @xmath552 the following corollary is an immediate application of lemma [ lem : legendresubscale ] .
[ cor : holderlegendreestimate ] let @xmath332 be @xmath274-hlder continuous with exponent @xmath517 and suppose that @xmath512 is consistent with @xmath236 . then there exist positive constants @xmath542 and @xmath543 such that @xmath544 and , for any @xmath536 , there exists @xmath537 such that @xmath553 for all @xmath352 and @xmath539 .
in this section , we introduce a class of variable - coefficient operators on @xmath96 whose heat equations are studied in the next section .
these operators , in view of proposition [ prop : operatorrepresentation ] , generalize the class of positive - homogeneous operators . fix a basis @xmath274 of @xmath96 , @xmath29 and , in the notation of the previous section , consider a differential operator @xmath126 of the form @xmath554 where the coefficients @xmath555 are bounded functions .
the symbol of @xmath126 , @xmath556 , is defined by @xmath557 for @xmath558 and @xmath101 .
we shall call @xmath559 the principal part of @xmath126 and correspondingly , @xmath560 is its principal symbol .
let s also define @xmath561 by @xmath562 for @xmath101 . at times
, we will freeze the coefficients of @xmath126 and @xmath559 at a point @xmath558 and consider the constant - coefficient operators they define , namely @xmath563 and @xmath564 ( defined in the obvious way ) .
we note that , for each @xmath558 , @xmath564 is homogeneous with respect to the one - parameter group @xmath139 where @xmath565 is defined by .
that is , @xmath559 is homogeneous with respect to the same one - parameter group of dilations at each point in space .
this also allows us to uniquely define the _ homogeneous order of _
@xmath126 by @xmath566 we remark that this is consistent with our definition of homogeneous - order for constant - coefficient operators and we remind the reader that this notion differs from the usual order a partial differential operator ( see the discussion surrounding ) . as in the constant - coefficient setting
, @xmath564 is not necessarily homogeneous with respect to a unique group of dilations , i.e. , it is possible that @xmath567 contains members of @xmath568 distinct from @xmath179 .
however , we shall henceforth only work with the endomorphism @xmath179 , defined above , for worrying about this non - uniqueness of dilations does not aid our understanding nor will it sharpen our results .
let us further observe that , for each @xmath558 , @xmath569 and @xmath414 are homogeneous with respect to @xmath570 where @xmath571 .
the operator @xmath126 is called @xmath95-positive - semi - elliptic if for all @xmath558 , @xmath572 is a positive - definite polynomial .
@xmath126 is called uniformly @xmath95-positive - semi - elliptic if it is @xmath95-positive - semi - elliptic and there exists @xmath573 for which @xmath574 for all @xmath558 and @xmath101 . when the context is clear , we will simply say that @xmath126 is positive - semi - elliptic and uniformly positive - semi - elliptic respectively . in light of the above definition , a semi - elliptic operator @xmath126 is one that , at every point @xmath558 , its frozen - coefficient principal part @xmath564 , is a constant - coefficient positive - homogeneous operator which is homogeneous with respect to the same one - parameter group of dilations on @xmath96 .
a uniformly positive - semi - elliptic operator is one that is semi - elliptic and is uniformly comparable to a constant - coefficient positive - homogeneous operator , namely @xmath575 . in this way
, positive - homogeneous operators take a central role in this theory . in view of proposition
[ prop : operatorrepresentation ] , the definition of @xmath414 via agrees with that we have given for constant - coefficient positive - homogeneous operators . for an @xmath95-positive - semi - elliptic operator @xmath126 , uniform semi - ellipticity can be formulated in terms of @xmath576 for any @xmath577 ; such a notion is equivalent in view of proposition [ prop : comparepoly ] .
for a uniformly positive - semi - elliptic operator @xmath126 , we are interested in constructing a fundamental solution to the heat equation , @xmath578 on the cylinder @xmath526\times \mathbb{v}$ ] ; here and throughout @xmath536 is arbitrary but fixed . by definition ,
a fundamental solution to on @xmath526\times\mathbb{v}$ ] is a function @xmath579\times\mathbb{v}\times\mathbb{v}\rightarrow \mathbb{c}$ ] satisfying the following two properties : given a fundamental solution @xmath584 to , one can easily solve the cauchy problem : given @xmath582 , find @xmath333 satisfying @xmath585 this is , of course , solved by putting @xmath586 for @xmath100 and @xmath539 and interpreting @xmath587 as that defined by the limit of @xmath333 as @xmath588 .
the remainder of this paper is essentially dedicated to establishing the following result : [ thm : fundamentalsolution ] let @xmath126 be uniformly @xmath95-positive - semi - elliptic with bounded @xmath274-hlder continuous coefficients .
let @xmath414 and @xmath589 be defined by and respectively and denote by @xmath420 the legendre - fenchel transform of @xmath414 .
then , for any @xmath536 , there exists a fundamental solution @xmath579\times\mathbb{v}\times\mathbb{v}\rightarrow \mathbb{c}$ ] to on @xmath526\times\mathbb{v}$ ] such that , for some positive constants @xmath169 and @xmath348 , latexmath:[\[\label{eq : fundamentalsolution } we remark that , by definition , the fundamental solution @xmath584 given by theorem [ thm : fundamentalsolution ] is @xmath95-regular . thus @xmath584 is necessarily continuously differentiable in @xmath32 and has continuous spatial derivatives of all orders @xmath254 such that @xmath591 .
+ as we previously mentioned , the result above is implied by the work of s. d. eidelman for @xmath0-parabolic systems on @xmath26 ( where @xmath592 ) @xcite .
eidelman s systems , of the form , are slightly more general than we have considered here , for their coefficients are also allowed to depend on @xmath32 ( but in a uniformly hlder continuous way ) . admitting this @xmath32-dependence is a relatively straightforward matter and , for simplicity of presentation
, we have not included it ( see remark [ rmk : gp ] ) . in this slightly more general situation ,
stated in @xmath26 and in which @xmath593 is the standard euclidean basis , theorem 2.2 ( p.79 ) @xcite guarantees the existence of a fundamental solution @xmath594 to , which has the same regularity appearing in theorem [ thm : fundamentalsolution ] and satisfies latexmath:[\[\label{eq : eidelmansestimate } for @xmath371 and @xmath539 where @xmath169 and @xmath348 are positive constants . by an appeal to corollary [ cor : legendrecomparehomogeneousnorm ]
, we have @xmath596 and from this we see that the estimates and are comparable .
+ in view of corollary [ cor : holderconsistent ] , the hypothesis of theorem [ thm : fundamentalsolution ] concerning the coefficients of @xmath126 immediately imply the following a priori stronger condition : [ hypoth : holdercoefficients ] there exists @xmath517 which is consistent with @xmath236 and for which the coefficients of @xmath126 are bounded and @xmath274-hlder continuous on @xmath96 of order @xmath512 . in this subsection
, we construct a fundamental solution to under only the assumption that @xmath126 , a uniformly @xmath95-positive - semi - elliptic operator , satisfies hypothesis [ hypoth : holdercoefficients ] . henceforth , all statements include hypothesis [ hypoth : holdercoefficients ] without explicit mention .
we follow the famous method of e. e. levi , c.f . , @xcite as it was adopted for parabolic systems in @xcite and @xcite .
although well - known , levi s method is lengthy and tedious and we will break it into three steps . let s motivate these steps by first discussing the heuristics of the method
. + we start by considering the auxiliary equation where @xmath558 is treated as a parameter .
this is the so - called frozen - coefficient heat equation .
as one easily checks , for each @xmath558 , @xmath598 solves . by the uniform semi - ellipticity of @xmath126 , it is clear that @xmath599 for @xmath18 and @xmath558 .
as we shall see , more is true : @xmath600 is an approximate identity in the sense that @xmath601 for all @xmath582 .
thus , it is reasonable to seek a fundamental solution to of the form @xmath602 where @xmath41 is to be chosen to ensure that the correction term @xmath118 is @xmath95-regular , accounts for the fact that @xmath600 solves but not , and is `` small enough '' as @xmath603 so that the approximate identity aspect of @xmath584 is inherited directly from @xmath600 .
+ assuming for the moment that @xmath118 is sufficiently regular , let s apply the heat operator to with the goal of finding an appropriate @xmath41 to ensure that @xmath584 is a solution to . putting @xmath604 we have formally , @xmath605 where we have made use of leibniz rule and our assertion that @xmath600 is an approximate identity .
thus , for @xmath584 to satisfy , @xmath41 must satisfy the integral equation @xmath606 viewing @xmath607 as a linear integral operator , is the equation @xmath608 which has the solution @xmath609 provided the series converges in an appropriate sense .
+ taking the above as purely formal , our construction will proceed as follows : we first establish estimates for @xmath600 and show that @xmath600 is an approximate identity ; this is step 1 . in step 2
, we will define @xmath41 by and , after deducing some subtle estimates , show that @xmath41 s defining series converges whence is satisfied . finally in step @xmath610
, we will make use of the estimates from steps 1 and 2 to validate the formal calculation made in .
everything will be then pieced together to show that @xmath584 , defined by , is a fundamental solution to .
our entire construction depends on obtaining precise estimates for @xmath600 and for this we will rely heavily on the homogeneity of @xmath560 and the legendre - fenchel transform of @xmath414 .
+ [ rmk : gp ] one can allow the coefficients of @xmath126 to also depend on @xmath32 in a uniformly continuous way , and levi s method pushes though by instead taking @xmath600 as the solution to a frozen - coefficient initial value problem @xcite .
+ the lemma below is a basic building block used in our construction of a fundamental solution to via levi s method and it makes essential use of the uniform semi - ellipticity of @xmath126 .
we note however that the precise form of the constants obtained , as they depend on @xmath265 and @xmath254 , are more detailed than needed for the method to work .
also , the partial differential operators @xmath279 of the lemma are understood to act of the @xmath31 variable of @xmath611 .
[ lem : legendreestimate ] there exist positive constants @xmath348 and @xmath347 and , for each multi - index @xmath254 , a positive constant @xmath349 such that , for any @xmath350 , latexmath:[\[\label{eq : legendreestimate1 } before proving the lemma , let us note that @xmath613 for all @xmath18 and @xmath100 in view of proposition [ prop : legendretransformproperties ] .
thus the estimate can be written equivalently as latexmath:[\[\label{eq : legendreestimate11 } let us first observe that , for each @xmath352 and @xmath18 , @xmath615 where we have used the homogeneity of @xmath560 with respect to @xmath616 and the fact that @xmath617 .
therefore @xmath618 for all @xmath352 and @xmath18 .
thus , to establish ( equivalently ) it suffices to estimate the right hand side of which is independent of @xmath32 .
the proof of the desired estimate requires making a complex change of variables and for this reason we will work with the complexification of @xmath98 , whose members are denoted by @xmath619 for @xmath620 ; this space is isomorphic to @xmath621 .
we claim that there are positive constants @xmath622 and , for each multi - index @xmath254 , a positive constant @xmath349 such that , for each @xmath350 , latexmath:[\[\label{eq : legendreestimate3 } for all @xmath620 and @xmath558 .
let us first observe that @xmath624 for all @xmath625 and @xmath558 , where @xmath626 are bounded functions of @xmath358 arising from the coefficients of @xmath126 and the coefficients of the multinomial expansion . by virtue of the uniform semi - ellipticity of @xmath126 and the boundedness of the coefficients , we have @xmath627 for all @xmath620 and @xmath558 where @xmath169 is a positive constant . by applying lemma [ lem : scaling ] to each term @xmath491 in the summation , we can find a positive constant @xmath348 for which the entire summation is bounded above by @xmath628 for all @xmath620 . by setting @xmath629
, we have @xmath630 for all @xmath620 and @xmath558 . by analogous reasoning ( making use of item 1 of lemma [ lem : scaling ] )
, there exists a positive constant @xmath169 for which @xmath631 for all @xmath620 and @xmath558 .
thus , for any @xmath350 , latexmath:[\[\label{eq : legendreestimate5 } for all @xmath620 and @xmath558 where @xmath633 .
finally , for each multi - index @xmath254 , another application of lemma [ lem : scaling ] gives @xmath634 for which @xmath635 for all @xmath620 where @xmath636 has been chosen to satisfy @xmath637 .
consequently , there is a positive constant @xmath349 for which latexmath:[\[\label{eq : legendreestimate6 } @xmath620 . upon combining , and , we obtain the inequality @xmath639 which holds for all @xmath620 and @xmath558 . upon paying careful attention to the way in which our constants were chosen
, we observe the claim is established by setting @xmath640 .
from the claim above , it follows that , for any @xmath641 and @xmath558 , the following change of coordinates by means of a @xmath621 contour integral is justified : @xmath642 thus , by virtue of the estimate , @xmath643 for all @xmath352 and @xmath641 where we have absorbed the integral of @xmath644 into @xmath349 . upon minimizing with respect to @xmath641
, we have @xmath645 for all @xmath31 and @xmath558 because @xmath646 in this we see the natural appearance of the legendre - fenchel transform .
the replacement of @xmath647 by @xmath648 is done using corollary [ cor : movingconstants ] and , as required , the constant @xmath348 is independent of @xmath265 and @xmath254 . upon combining and ,
we obtain the desired estimate . given a positive - homogeneous operator @xmath11 , we invoke proposition [ prop : operatorrepresentation ] to obtain @xmath274 and @xmath236 for which @xmath649 .
in other words , @xmath11 is an @xmath95-positive - semi - elliptic operator which consists only of its principal part .
consequently , the heat kernel @xmath61 satisfies @xmath650 for all @xmath100 and @xmath18 and so we immediately obtain the estimate from the lemma .
[ lem : legendreholderestimate ] there is a positive constant @xmath348 and , to each multi - index @xmath254 , a positive constant @xmath349 such that @xmath651|\leq c_{\beta }
t^{-(\mu_h+|\beta:2\mathbf{m}|)}|h|_{\mathbf{v}}^{\alpha}\exp(-tmr^{\#}(x / t))\ ] ] for all @xmath18 , @xmath652 . here , in view of hypothesis [ hypoth : holdercoefficients ] , @xmath512 is the @xmath274-hlder continuity exponent for the coefficients of @xmath126 .
[ lem : approximateidentity ] suppose that @xmath653\times\mathbb{v})$ ] where @xmath654 .
then , on any compact set @xmath655\times\mathbb{v}$ ] , @xmath656 uniformly on @xmath208 as @xmath603 . in particular , for any @xmath582 , @xmath657 uniformly on all compact subsets of @xmath96 as @xmath603 . let @xmath208 be a compact subset of @xmath658\times\mathbb{v}$ ] and write @xmath659g(s - t , y)\,dy\\ & : = & i_t^{(1)}(s , x)+i_t^{(2)}(s , x).\end{aligned}\ ] ] let @xmath477 and , in view of corollary [ cor : exponentialoflegendretransformintegrable ] , let @xmath357 be a compact subset of @xmath96 for which @xmath660 where the constant @xmath348 is that given in of lemma [ lem : legendreestimate ] . using the continuity of @xmath167
, we have for sufficiently small @xmath18 , @xmath661 we note that , for any @xmath18 and @xmath100 , @xmath662 appealing to lemma [ lem : legendreestimate ] we have , for any @xmath663 , @xmath664 here we have made the change of variables : @xmath665 and used the homogeneity of @xmath560 to see that @xmath666 .
therefore @xmath667 uniformly on @xmath208 as @xmath603 .
let us now consider @xmath668
. with the help of lemmas [ lem : legendresubscale ] and [ lem : legendreholderestimate ] and by making similar arguments to those above we have @xmath669 for all @xmath670 $ ] , @xmath671 and @xmath100 ; here @xmath544 .
@xmath672 uniformly on @xmath208 as @xmath603 and the lemma is proved . * step 2 .
construction of @xmath41 and the integral equation * + for @xmath18 and @xmath352 , put @xmath674 and iteratively define @xmath675 where @xmath676 . in the sense of ,
note that @xmath677 .
+ we claim that for some @xmath678 and positive constants @xmath169 and @xmath348 , latexmath:[\[\label{eq : k1bound } all @xmath352 and @xmath539 .
indeed , observe that @xmath680 for all @xmath352 and @xmath18 where we have used the fact that the coefficients of @xmath126 are bounded . in view of lemma [ lem : legendreestimate ]
, we have @xmath681 for all @xmath352 and @xmath539 where @xmath682 using hypothesis [ hypoth : holdercoefficients ] , an appeal to corollary [ cor : holderlegendreestimate ] gives @xmath544 and @xmath683 for which @xmath684 for all @xmath352 and @xmath539 . our claim
is then justified by setting @xmath685 and adjusting the constants @xmath169 and @xmath348 appropriately to absorb the prefactor @xmath686 into the exponent .
it should be noted that the constant @xmath687 is inherently dependent on @xmath126 .
for it is clear that @xmath688 depends on @xmath126 .
the constants @xmath542 and @xmath543 are specified in lemma [ lem : legendresubscale ] and are defined in terms of the hlder exponent of the coefficients of @xmath126 and the weight @xmath236 .
+ taking cues from our heuristic discussion , we will soon form a series whose summands are the functions @xmath689 for @xmath690 in order to talk about the convergence of this series , our next task is to estimate these functions and in doing this we will observe two separate behaviors : a finite number of terms will exhibit singularities in @xmath32 at the origin ; the remainder of the terms will be absent of such singularities and will be estimated with the help of the gamma function .
we first address the terms with the singularities .
[ lem : kfiniteterms ] let @xmath678 be given by and @xmath537 be any constant for which is satisfied . for any positive natural number @xmath54 such that @xmath691 and @xmath477 for which @xmath692 , there is a constant @xmath693 such that @xmath694 for all @xmath352 and @xmath539 . in view of
, it is clear that the estimate holds when @xmath695 .
let us assume the estimate holds for @xmath696 such that @xmath697 and @xmath477 for which @xmath698 .
then @xmath699 for @xmath352 and @xmath539 where we have set @xmath700 .
observe that @xmath701 for all @xmath702 and @xmath703 .
using the fact that @xmath704 , guarantees that @xmath705 or equivalently @xmath706 for all @xmath702 and @xmath707 . combining and yields @xmath708\end{aligned}\ ] ] for all @xmath352 and @xmath539 , where we have used the fact that @xmath420 is non - negative .
let us focus our attention on the first term above .
for @xmath709 , @xmath710 because @xmath711 .
consequently , @xmath712 for all @xmath558 and @xmath18 .
we note that the second inequality is justified by making the change of variables @xmath713 ( thus canceling the term @xmath714 in the integral over @xmath715 ) and the final inequality is valid because @xmath716 . by similar reasoning ,
we obtain @xmath717 for all @xmath100 and @xmath18 . upon combining the estimates , and
, we have @xmath718 for all @xmath352 and @xmath539 where we have put @xmath719 and made use of corollary [ cor : exponentialoflegendretransformintegrable ] .
the estimate is an important one and will be used again . in the context of elliptic operators ,
i.e. , where @xmath368 , the analogous result is captured in lemma 5.1 of @xcite .
it is interesting to note that s. d. eidelman worked somewhat harder to prove it .
perhaps this is because the appearance of the legendre - fenchel transform was nt noticed .
it is clear from the previous lemma that for sufficiently large @xmath54 , @xmath689 is bounded by a positive power of @xmath32 .
the first such @xmath54 is @xmath720 . in view of the previous lemma , @xmath721 for all @xmath352 and @xmath539 where we have adjusted @xmath722 to account for this positive power of @xmath32 .
let @xmath723 and set @xmath724 upon combining preceding estimate with the estimates and , we have @xmath725 for all @xmath352 and @xmath539 where @xmath726 let us take this a little further . the euler - beta function @xmath730 satisfies the well - known identity @xmath731 . using this identity ,
one quickly obtains the estimate @xmath732 it therefore suffices to prove that latexmath:[\[\label{eq : kseriesboundeulerbeta } we first note that @xmath734 and so , for @xmath735 , follows directly from the calculation proceeding the lemma .
we shall induct on @xmath265 . by another application of and , we have @xmath736^{-1}|k_{\bar n+k+1}(t , x , y)|\\ & \leq & \int_{0}^t\int_{\mathbb{v}}(t - s)^{-(\mu_h+(1-\rho))}s^{-k\rho}\exp(-mr^{\#}((t - s)^{-e}(x - z)))\\ & & \hspace{6cm}\times\exp(-m_1r^{\#}(s^{-e}(z - y)))\,dz\,ds\\ & \leq & \exp(-m_1r^{\#}(t^{-e}(x - y)))\\ & & \hspace{1cm}\times\int_{0}^t\int_{\mathbb{v}}(t - s)^{-(\mu_h+(1-\rho))}s^{-k\rho}\exp(-m\delta r^{\#}((t - s)^{-e}(x - z)))\,dz\,ds\end{aligned}\ ] ] for all @xmath352 and @xmath539 . upon making the changes of variables
@xmath737 followed by @xmath738 , we have @xmath739 for all @xmath352 and @xmath539 .
therefore holds for @xmath740 as required .
[ prop : integralidentity ] let @xmath741\times\mathbb{v}\times\mathbb{v}\rightarrow\mathbb{c}$ ] be defined by @xmath742 this series converges uniformly for @xmath352 and @xmath743 where @xmath744 is any positive constant .
there exists @xmath745 for which latexmath:[\[\label{eq : phiestimate } using lemmas [ lem : kfiniteterms ] and [ lem : kseries ] we see that @xmath749\exp(-m_1r^{\#}(t^{-e}(x - y)))\ ] ] for all @xmath352 and @xmath539 from which and our assertion concerning uniform convergence follow . a similar calculation and an application of tonelli s theorem
justify the following use of fubini s theorem : for @xmath352 and @xmath539 , @xmath750 as desired .
the following hlder continuity estimate for @xmath41 is obtained by first showing the analogous estimate for @xmath357 and then deducing the desired result from the integral formula . as the proof is similar in character to those of the preceding two lemmas ,
we omit it . a full proof can be found in @xcite .
we also note here that the result is stronger than is required for our purposes ( see its use in the proof of lemma [ lem : wproperties ] ) .
all that is really required is that @xmath751 satisfies the hypotheses ( for @xmath16 ) in lemma [ lem : integraldifferentiation ] for each @xmath558 .
[ lem : integraldifferentiation ] let @xmath517 be consistent with @xmath236 and , for @xmath754 , let @xmath755\times\mathbb{v}\rightarrow\mathbb{c}$ ] be bounded and continuous . moreover , suppose that @xmath16 is uniformly @xmath274-hlder continuous in @xmath31 on @xmath756\times\mathbb{v}$ ] of order @xmath512 , by which we mean that there is a constant @xmath757 such that @xmath758}|f(t , x)-f(t , y)|\leq c|x - y|_{\mathbf{v}}^{\alpha}\ ] ] for all @xmath352 .
then @xmath759\times\mathbb{v}\rightarrow \mathbb{c}$ ] defined by @xmath760 is @xmath95-regular on @xmath761 .
moreover , @xmath762 and for any @xmath254 such that @xmath591 , we have @xmath763 for @xmath100 and @xmath764 . before starting the proof ,
let us observe that , for each multi - index @xmath254 , @xmath765 using the assumption that @xmath16 is bounded , we observe that @xmath766 for all @xmath743 and @xmath100 .
when @xmath767 , @xmath768 converges and consequently @xmath769 for all @xmath743 and @xmath100 . from this
it follows that @xmath528 is continuous on @xmath761 and moreover holds for such an @xmath254 in view of lebesgue s dominated convergence theorem . when @xmath473 , does not converge and hence the above argument fails .
the main issue in the proof below centers around using @xmath274-hlder continuity to remove this obstacle .
our argument proceeds in two steps .
the fist step deals with the spatial derivatives of @xmath770 . therein , we prove the asserted @xmath31-regularity and show that the formula holds .
in fact , we only need to consider the case where @xmath473 ; the case where @xmath767 was already treated in the paragraph proceeding the proof . in the second step ,
we address the time derivative of @xmath770 . as we will see , and the asserted @xmath32-regularity are partial consequences of the results proved in step 1 ; this is , in part , due to the fact that the time derivative of @xmath600 can be exchanged for spatial derivatives .
the regularity shown in the two steps together will automatically ensure that @xmath770 is @xmath95-regular on @xmath761 .
+ _ step 1 .
_ let @xmath254 be such that @xmath473 . for @xmath771 write @xmath772 and
observe that @xmath773 for all @xmath774 and @xmath100 ; it is clear that @xmath775 is continuous in @xmath32 and @xmath31 .
the fact that we can differentiate under the integral sign is justified because @xmath32 has been replaced by @xmath776 and hence the singularity in is avoided in the upper limit .
we will show that @xmath775 converges uniformly on all compact subsets of @xmath761 as @xmath777 .
this , of course , guarantees that @xmath778 exists , satisfies and is continuous on @xmath761 . to this end , let us write @xmath779 using our hypotheses , corollary [ cor : holderconsistent ] and lemma [ lem : legendresubscale ] , for some @xmath544 and @xmath683 , there is @xmath537 such that @xmath780 for all @xmath781 , @xmath782 $ ] and @xmath783 $ ]
. in view of the preceding estimate and lemma [ lem : legendreestimate ] , we have @xmath784 for all @xmath781 , @xmath782 $ ] and @xmath783 $ ] , where @xmath169 and @xmath348 are positive constants .
we then observe that @xmath785 for all @xmath782 $ ] and @xmath786 , where the validity of the second inequality is seen by making the change of variables @xmath787 and canceling the term @xmath788 .
consequently , @xmath789 exists for each @xmath782 $ ] and @xmath100 .
moreover , for all @xmath774 and @xmath100 , @xmath790 from this we see that @xmath791 converges uniformly on all compact subsets of @xmath761 .
we claim that for some @xmath678 , there exists @xmath757 such that @xmath792 for all @xmath100 and @xmath783 $ ] . indeed , @xmath793\big\vert_{y
= x}\,dz+\big [ d_{\mathbf{v}}^{\beta}\int_{\mathbb{v}}g_p(t - s , x - z;y)\,dz\big]\big\vert_{y = x}.\end{aligned}\ ] ] the first term above is estimated with the help of lemma [ lem : legendreholderestimate ] and by making arguments analogous to those in the previous paragraph ; the appearance of @xmath687 follows from an obvious application of lemma [ lem : legendresubscale ] .
this term is bounded by @xmath794 .
the second term is clearly zero and so our claim is justified . by essentially repeating the arguments made for @xmath795 and making use of
, we see that @xmath796 where this limit converges uniformly on all compact subsets of @xmath761 .
it follows from leibnitz rule that @xmath797 for all @xmath798 and @xmath100 .
now , in view of lemma [ lem : approximateidentity ] and our hypotheses concerning @xmath16 , @xmath799 where this limit converges uniformly on all compact subsets of @xmath761 . using the fact that @xmath800 , we see that @xmath801 for all @xmath802 and @xmath100 .
because the coefficients of @xmath126 are @xmath274-hlder continuous and bounded , for each @xmath254 , @xmath803 satisfies the same condition we have required for @xmath16 and so , in view of step 1 , it follows that @xmath804 converges uniformly on all compact subsets of @xmath761 as @xmath777 .
we thus conclude that @xmath805 exists , is continuous on @xmath761 and satisfies .
[ lem : wproperties ] let @xmath806\times\mathbb{v}\times\mathbb{v}\rightarrow \mathbb{c}$ ] be defined by @xmath807 for @xmath352 and @xmath539 . then , for each @xmath558 , @xmath808 is @xmath95-regular on @xmath581 and satisfies @xmath809 for all @xmath352 and @xmath810 .
moreover , there are positive constants @xmath169 and @xmath348 for which latexmath:[\[\label{eq : wproperties2 } the estimate follows from and by an analogous computation to that done in the proof of lemma [ lem : kfiniteterms ] . it remains to show that , for each @xmath558 , @xmath808 is @xmath95-regular and satisfies on @xmath581 .
these are both local properties and , as such , it suffices to examine them on @xmath761 for an arbitrary but fixed @xmath812 .
let us write @xmath813 for @xmath352 and @xmath814 .
in view of lemmas [ holderphilemma ] and [ lem : integraldifferentiation ] , for each @xmath558 , @xmath815 is @xmath95-regular on @xmath761 and @xmath816 for all @xmath100 and @xmath814 ; here we have used the fact that @xmath817 treating @xmath818 is easier because @xmath819 and , for each multi - index @xmath254 , @xmath820 are , as functions of @xmath715 and @xmath821 , absolutely integrable on @xmath822\times \mathbb{v}$ ] for every @xmath823 $ ] and @xmath100 by virtue of lemma [ lem : legendreestimate ] .
consequently , derivatives may be taken under the integral sign and
so it follows that , for each @xmath558 , @xmath824 is @xmath95-regular on @xmath761 and @xmath825 for @xmath100 and @xmath826 .
we can thus conclude that , for each @xmath558 , @xmath808 is @xmath95-regular on @xmath761 and , by combining and , @xmath827 for @xmath100 and @xmath826 . by , proposition
[ prop : integralidentity ] and the dominated convergence theorem , @xmath828 and therefore @xmath829 for all @xmath352 and @xmath814 .
the theorem below is our main result .
it is a more refined version of theorem [ thm : fundamentalsolution ] because it gives an explicit formula for the fundamental solution @xmath584 ; in particular theorem [ thm : fundamentalsolution ] is an immediate consequence of the result below .
let @xmath126 be a uniformly @xmath95-positive - semi - elliptic operator .
if @xmath126 satisfies hypothesis [ hypoth : holdercoefficients ] then @xmath579\times\mathbb{v}\times\mathbb{v}\rightarrow \mathbb{c}$ ] , defined by @xmath830 for @xmath352 and @xmath539 , is a fundamental solution to .
moreover , there are positive constants @xmath169 and
@xmath348 for which latexmath:[\[\label{eq : main2 } as @xmath678 , and lemma [ lem : legendreestimate ] imply the estimate . in view of lemma [ lem : wproperties ] and
corollary [ frozenfundsolcor ] , for each @xmath558 , @xmath580 is @xmath95-regular on @xmath581 and @xmath832 for all @xmath100 and @xmath833 .
it remains to show that for any @xmath582 , @xmath834 for all @xmath100 .
indeed , let @xmath582 and , in view of , observe that @xmath835 for all @xmath100 and @xmath539 .
an appeal to lemma [ lem : approximateidentity ] gives , for each @xmath100 , @xmath836 as required .
in fact , the above argument guarantees that this convergence happens uniformly on all compact subsets of @xmath96 .
we remind the reader that implicit in the definition of fundamental solution to is the condition that @xmath584 is @xmath95-regular .
in fact , one can further deduce estimates for the spatial derivatives of @xmath584 , @xmath837 , of the form for all @xmath254 such that @xmath838 ( see @xcite ) .
using the fact that @xmath584 satisfies and @xmath126 s coefficients are bounded , an analogous estimate is obtained for a single @xmath32 derivative of @xmath584 .
evan randles : department of mathematics , university of california , los angeles , los angeles ca 90025 .
e - mail : [email protected] + laurent saloff - coste : department of mathematics , cornell university , ithaca ny 14853 .
e - mail : [email protected] | we consider a class of homogeneous partial differential operators on a finite - dimensional vector space and study their associated heat kernels .
the heat kernels for this general class of operators are seen to arise naturally as the limiting objects of the convolution powers of complex - valued functions on the square lattice in the way that the classical heat kernel arises in the ( local ) central limit theorem .
these so - called positive - homogeneous operators generalize the class of semi - elliptic operators in the sense that the definition is coordinate - free .
more generally , we introduce a class of variable - coefficient operators , each of which is uniformly comparable to a positive - homogeneous operator , and we study the corresponding cauchy problem for the heat equation . under the assumption that such an operator has hlder continuous coefficients , we construct a fundamental solution to its heat equation by the method of e. e. levi , adapted to parabolic systems by a. friedman and s. d. eidelman . though our results in this direction are implied by the long - known results of s. d. eidelman for @xmath0-parabolic systems ,
our focus is to highlight the role played by the legendre - fenchel transform in heat kernel estimates .
specifically , we show that the fundamental solution satisfies an off - diagonal estimate , i.e. , a heat kernel estimate , written in terms of the legendre - fenchel transform of the operator s principal symbol
an estimate which is seen to be sharp in many cases . _ dedicated to professor rodrigo bauelos on the occasion of his 60th birthday . _
* keywords : * semi - elliptic operators , quasi - elliptic operators , @xmath0-parabolic operators , heat kernel estimates , legendre - fenchel transform . + * mathematics subject classification : * primary 35h30 ; secondary 35k25 . |
since einstein s relativistic theory tell that the matter can convert into energy , the possibility of converting energy into matter , i.e. , the electron - positron pair , as predicted by dirac in quantum electrodynamics @xcite , has attracted a great deal of interest@xcite . in presence of a static and uniform electric field , the quantum electrodynamic ( qed ) vacuum may break down and decay into electron - positron pairs due to a quantum tunneling effect @xcite .
the critical schwinger field is @xmath0 , which can accelerate the electron to an energy of the order of its rest mass on its compton wavelength @xmath1 , where @xmath2 is the electron mass .
starting from the works of brezin and popov et.al @xcite , the schwinger mechanism was generalized to time dependent fields @xcite , where another mechanism may be responsible for the pair creation .
if the frequency of the alternating field exceeds the gap @xmath3 , electrons in dirac sea can transit to positive states and pairs are triggered .
experimentally , pairs can be generated by the relativistic heavy - ion collisions@xcite or the collision of an intense laser pulse and a 46 gev electron beam@xcite , but pairs created from pure laser light has not been observed until now .
recently , various numerical approach were developed to deal with the time dependent dirac equation@xcite , the klein paradox@xcite , the zitterbewegung@xcite , and the pair production process @xcite .
the one - dimensional well potential , specially , for its simplicity , is studied extensively@xcite .
the super critical well potential has bound states embedded in the negative continuum can cause spontaneous electron - positron pair creation .
theoretical investigations are expected to make the physics of the creation clear and predict a higher generation rate . however , the pauli exclusion principle will block further creation once the bound states are occupied , resulting a asymptotic saturation behavior@xcite .
motivated by this requirement and a better understanding of the pair creation process in the one - dimensional well potential , we examine the pair creation in a well with its width or depth oscillating . by oscillating the width and depth , the transfer channels for population
are opened and closed alternately .
the electrons confined in the well will be released and the pauli block become invalid .
this can lead to a non - vanishing production rate , which means that pairs can be pumped inexhaustibly form the well .
this paper is organized as follows . in sec .
we present the model and the numerical method we employed .
the well potential is set to be oscillating in two modes , the width oscillating mode and the depth oscillating mode .
the energy spectrum is plotted as a function of the width or the depth . in sec
.. we discuss the pair production process in both two modes . the time evolution of pair number , spacial density and pumping rate are studied
we also investigate the adiabatic limit of the oscillating . in the last section we give a brief summary .
in one dimension , the time evolution of the heisenberg field operator @xmath4 is given by the dirac equation ( without spin , for simplicity in this paper ) @xcite @xmath5\hat{\psi}\left(z,\ , t\right).\label{eq : dirac_eq}\ ] ] @xmath6 are pauli matrices , @xmath7 is the speed of light in vacuum , @xmath8 is the external potential . the atomic units ( [ a.u . ] )
is used in this paper : @xmath9 , @xmath10 , @xmath11 is fine - structure constant , compton wave length of electron is @xmath12 .
the hamiltonian of the system is @xmath13 $ ] .
we define the potential as @xmath14.\label{eq : vzt}\ ] ] @xmath15 is the width of potential edge ( a measure of the width of the electric field ) , and we set @xmath16 .
the numerical box size is set to @xmath17 .
the potential width @xmath18 and the depth @xmath19 ( positive , but note that the potential @xmath8 is negative in the center and zero elsewhere ) are set to two modes : ( 1 ) the * w - oscillating mode * : @xmath20 is constant , @xmath21 $ ] ; ( 2 ) the * v - oscillating mode _ _ : _ _ * @xmath22 is constant , @xmath23 $ ] . in this paper
we assume @xmath24 and @xmath25 , then @xmath18 ( or @xmath19 ) varies as a sine function between zero and its upper boundary @xmath26 ( or @xmath27 ) , and the turning on process is from zero point with first order derivative equal to zero . in the following numerical simulation , we choose the total evolution time to be the period ( @xmath28 or @xmath29 ) of the oscillating @xmath30 or @xmath31 multiples an integer , to make the potential turning off is finished with first order derivative equal to zero .
, as @xmath22 increasing , the bound states dive into the dirac sea at @xmath32(in units of @xmath33 , the electron compton wavelength ) .
( b ) , @xmath34 , as @xmath20 increasing , the bound states dive into the dirac sea at @xmath35(in units of @xmath36 ) .
[ fig : spectrum ] ] the numerically energy spectrum of the total hamiltonian of finite - size ( length , for one dimension ) are presented in fig .
[ fig : spectrum ] for varying @xmath22 and @xmath20 , which can be a schematic of the real one - dimensional system . in fig .
[ fig : spectrum ] , we show the critical width or depth , and the behavior of diving of the bound states into the negative continuum . for example , if @xmath37 , there are bound states embedded and then pair can be spontaneously triggered only when @xmath38 . in recent years , numerical quantum field
theoretical approach @xcite has been established to overcome the single particle picture described by quantum mechanics and the mathematical difficulty of quantum electrodynamics . in this section
we will briefly review this method and describe that how do we deal with the model in this paper .
the field operator can be expressed in terms of the electron annihilation and positron creation operators as@xcite @xmath39 in which @xmath40 is the solution of the filed - free dirac hamiltonian ( @xmath41 ) , @xmath42 is the time dependent solution of the dirac equation , and the term @xmath43 denotes the summation over all states with positive ( negative ) energy .
the eigenstates of the filed - free hamiltonian are @xmath44 where @xmath45 , and @xmath46 respectively . the time dependent single particle wave function @xmath47 can be got by introducing the time - evolution operator @xmath48 , @xmath49 where @xmath50 denotes the dyson time ordering operator . in this paper
, we use the numerical split operator technique @xcite , then @xmath51 with @xmath52 practically , since the derivation ( the momentum operator ) can be implemented by replacing the operator @xmath53 with its value @xmath54 in momentum space , the evolution operation has the following form @xmath55\mathcal{f}w\left(t\right),\\ e^{-idth_{z}}w\left(t\right ) & = & \left[\cos\left(v\left(t\right)dt\right)\right.\nonumber \\ & & \left .- i\sin\left(v\left(t\right)dt\right)\right]w\left(t\right),\end{aligned}\ ] ] where @xmath54 is momentum , @xmath56 and @xmath57 is fourier transformation ( inverse fourier transformation ) .
then , after the time dependent field operator @xmath4 can be calculated , the number , the spacial distribution of electrons created from the vacuum ( defined as @xmath58 , @xmath59 ) are obtained from the positive part of the field operator , @xmath60 where @xmath61 .
the pair number @xmath62 is equal to the electron number @xmath63 .
the spacial distribution of the created positrons can be written as @xmath64 the total positron number @xmath65 is equal to the electron number @xmath63 .
we can also get it from the negative part of the field operator by compute the number and spacial distribution of the holes . in this paper
we use this expression ( eq . ) to reduce the computational cost , because @xmath66 has been calculated in eq .. furthermore , we can neglect the larger part of the momentum ( @xmath67 is far greater than @xmath68 and @xmath69 ) in the numerical simulation , for its contribution to the matrix element @xmath70 is very small . in the following ,
the number of spatial points is @xmath71 , and we only take @xmath72 discrete momentum into account . based on the projection of the field operator onto the the field - free electronic states in this method and the definition of electron and positron in the dirac hole theory , in this paper we will present physical quantities for all time and focus on the moments when the field is absent .
for a well potential of depth @xmath20 , if @xmath73 , the positive continuum and negative continuum can not overlap .
but for a super - critical depth , @xmath74 , the domain @xmath75 exist , and bound states in the well are possible ( which we call bound states embedded in the negative continuum ) : their wave function do not decrease exponentially out the well , but join a continuum wave of the same energy @xmath76 out the well , hence the wave function has a non - zero probability outside .
an empty bound state will spontaneously be occupied by an electron ( two , if the spin is considered ) from the filled dirac sea , and the hole ( identified as positron ) will travel away from the well to infinity @xcite .
this is the picture of spontaneous creation of electron - positron pair . for a static well potential , electrons will fill the embedded bound states , and
the pauli principle will prevent further pair creation
. the number of pair created should be the number of bound states which meet these conditions . for a time dependent potential , the situation is more complicated . in paper @xcite ,
the effect of open and close a pair - creation channel was studied .
the well depth is fixed at @xmath77 , while the width @xmath22 varies between @xmath78 and @xmath79 . for @xmath80 , there is one bound state embedded in the dirac sea , and there are two for @xmath81 . after enough time for saturation , the pair number will increase as one more channel is opened , but do not decrease as one of the two channels is closed .
the reason is that the annihilation of the pair need the electron and positron to be in the same place , which is not satisfied because the electron remains in the well while the positrons have left the creation zone and escaped to the opposite direction .
naturally , one can propose that if the channel is opened and closed periodically , can this mechanism will lead to a continuously pair creation ?
moreover , for fixed @xmath22 and varying @xmath20 , since the diving behavior is similar ( fig .
[ fig : spectrum ] ) , will something similar happen ?
motivated by these questions , we construct two oscillating modes as described in sec . .
a. results and discussion are as follows . , @xmath77 ; ( b ) , v - oscillating mode , * @xmath82 * , @xmath34 .
the frequency @xmath83 and @xmath84 are in units of @xmath36 .
the dash line represent the time @xmath85 when the positrons arrive the boundary , @xmath86 .
the triangles denote pair number when the field is absent .
the dot line just link these triangles.[fig : n_t ] ] using the method presented in sec . .
b , we graph the time evolution of the pair number defined as eq.([eq : n_pair ] ) for both w - oscillating and v - oscillating mode in fig . [
fig : n_t ] .
the width frequency @xmath83 and depth frequency @xmath83 are in units of @xmath36 , and their values are assumed to be relative low , comparing to the gap @xmath87 , so that the photon absorption mechanism is not valid .
the total time is @xmath88 and the period is @xmath89 or @xmath90 .
the dot line represent the time @xmath91 when the particles arrive the boundary , @xmath86 . since @xmath24 and @xmath25 , if the time is an integer multiples of the period ( @xmath28 or @xmath29 ) , the system hamiltonian degenerate to a field free one .
the triangles in fig .
[ fig : n_t ] denote the pair number when the field is absent .
* w - oscillating mode : * in fig .
[ fig : n_t ] ( a ) , we illuminate the total number of pairs as a function of time for @xmath92 , @xmath93 , @xmath94 , @xmath95 . the depth @xmath20 is fixed at @xmath77 .
the width @xmath22 varies between @xmath24 and @xmath96 , corresponding zero and three bound states embedded .
when @xmath97 , there are also eight bound states exist in the gap , which can be associate with the pair creation@xcite .
when @xmath92 , the width @xmath22 can only finish one cycle in the total time @xmath98 .
@xmath99 begin to arise before @xmath100 , corresponding @xmath101 , when the first bound state dive into the negative continuum .
the reason is the non - adiabatic varying width , and @xmath99 will begin to arise precisely at the time when @xmath101 in the adiabatic case ( @xmath102 , see the discussion below ) .
@xmath99 increases as more bound states dive in , and reach its maximum @xmath103 at @xmath104 , between @xmath105 and @xmath106 , at which time the third and the second bound state were pulled out the dirac sea . undergoing the particle - antiparticle annihilation ,
@xmath99 decreases but remains an appreciable value @xmath107 at the end . in the latter half of this cycle ,
the embedded bound states depart from the dirac sea , return to the positive continuum , and become scattering states .
the released positrons are reflected by the numerical box boundary , come back to the interaction region and will affect the pair generation after . though the effect is weak when @xmath92 , it is non - ignorable when , i.e. , @xmath108 ( see fig . [
fig : imagesc_w ] for details ) . for @xmath109 and @xmath108
, @xmath22 can finish four and eighteen cycles in the total time and the pair number are @xmath110 , @xmath111 at the end . for @xmath112
, @xmath22 can finish one and eight cycles , respectively . in each cycle , the positrons are repulsed by the electric field to the infinity once they were generated , while the electrons are limited in the well when the field is strong enough and extruded out as the well is turning off , avoiding the inevitable pauli block in the non - varied static well construction .
the non - synchronous ejection prevent the annihilation and lead to a high production rate .
the next cycle starts from field free and is independent on the previous cycle . in fig .
[ fig : n_t ] , the dot line link the triangles which denote the pair number when the field is absent . we can find that the pair generation denoted by the dot line is linearly depend on time for low frequency @xmath83 , for @xmath112 .
if the system length @xmath113 is infinite and there is no reflection at the boundary , the pairs can be pumped inexhaustibly with a constant production rate from the well . even for @xmath114 , there is nonlinear effect at the beginning , the generation rate become stable soon . due to the finite period @xmath28 and the bound states in the gap ,
particle generation and ejection process is not monotonic with the increase of the frequency @xmath83 , see fig .
[ fig : n_t ] ( a ) , however , ignoring the reflection , if the w - oscillating frequency @xmath83 is very small , we can expect a linear dependent of final pair number on the frequency . *
v - oscillating mode : * the number of pairs @xmath99 as a function of time are presented in fig . [
fig : n_t ] ( b ) , for @xmath115 , @xmath93 , @xmath94 and @xmath95 .
the width @xmath22 is fixed at @xmath34 , while the depth varies between @xmath25 and @xmath82 , corresponding zero and three bound states embedded .
there are also eight bound states exist in the gap when @xmath116 . for @xmath92 , the first bound state dive in at @xmath117 ,
at which time there are already @xmath118 pair generated .
the first bound state depart the negative continuum after the second and the third one , at @xmath119 , when @xmath99 reach its maximum @xmath120 .
finally , there are @xmath121 pairs survived at @xmath122 . for @xmath123 , @xmath94 , @xmath95 ,
the pair number at the end are @xmath124 @xmath125 , @xmath126 . instead of pulling and pushing the walls of the well in w - oscillating mode , in this mode
it is the rising and falling bottom of the well that control the bound states diving in and departing from the negative continuum .
it is also the non - synchronous ejection of the positrons and electrons which dominant the pumping process .
the dot line here indicate a linear relation between the pair number and time .
the final number is not monotonic depending on the frequency @xmath84 , and we can also expect a linear dependent of final pair number on @xmath84 when @xmath84 is very small . note that although the two modes has the same beginning and ending parameters , the generation rate in the w - oscillating mode is much higher . , the three dimensional diagrams for entire time and the waterfall figures for field free moments ( the time indicated by triangles in fig .
[ fig : n_t ] ( a ) ) , for electron spacial density ( a , c ) and positron spacial density ( b , d ) .
the thicker curve in sub - figure ( c , d ) mark the last cycle before the positron arrive the boundary .
the well potential @xmath127 with @xmath77 , @xmath34 , are included on the bottom for comparison .
all other parameters are the same as fig .
[ fig : n_t ] ( a).[fig : imagesc_w ] ] in last section we discussed the total pair number as a function of time in different oscillating frequency for both modes . to show the pumping process explicitly , we compute the time evolution of spacial density of electrons and positrons ( eq . [ eq : n_z_e ] and eq.[eq : n_z_p ] ) for @xmath108 and @xmath128 respectively . in fig .
[ fig : imagesc_w ] , for w - oscillating mode , @xmath108 , we plot the the time evolution of spacial density of electrons and positrons ( sub - figure ( a ) and ( b ) ) .
specially , for the moments when the field are zero , denoted by the triangles in fig . [ fig : n_t ] ( a ) , these quantities are plotted in the waterfall figures , fig .
[ fig : imagesc_w](c , d ) . for v - oscillating mode , @xmath128 , similar diagram
are presented in fig .
[ fig : imagesc_v ] . for comparison
, the well potential @xmath127 with wide and depth equal to the upper boundary of the two modes , @xmath129 , @xmath130 , are included on the bottom .
these figures clearly show the process how are the particles pumped from the well and spread in the numerical box .
, the three dimensional diagrams for entire time and the waterfall figures for field free moments ( the time indicated by triangles in fig .
[ fig : n_t ] ( b ) ) , for electron spacial density ( a , c ) and positron spacial density ( b , d ) .
the thicker curve in sub - figure ( c , d ) mark the last cycle before the positron arrive the boundary .
the well potential @xmath127 with @xmath77 , @xmath34 , are included on the bottom for comparison .
all other parameters are the same as fig .
[ fig : n_t ] ( b ) .
[ fig : imagesc_v ] ] since @xmath108 , the period of the width oscillating is @xmath131 .
before positrons arrive the boundaries , the width can finish eight cycles .
if we detect the particle population at the boundary , we can find that positrons arrive the boundary first , at @xmath132 , in conformity to the estimation @xmath133 .
electrons arrive the boundary at @xmath134 , about one period ( @xmath28 or @xmath29 ) later than the positrons .
we can see that the particles reflected by the boundary come back to the interaction region , and may cause non - ignorable effect , i.e. , the non - linearity of the last three triangles in the dot line in fig .
[ fig : n_t](a ) , @xmath108 . comparing with
the rising and falling bottom of the well , more work is done by the wall of the well in the case of opening and closing the well . in the w - oscillating mode ,
the wavefront of the particles are more abrupter and regular . in energy space ,
higher energy modes are excited , and the spectrum show periodic structure with @xmath94 between each peak . in the v - oscillating mode , electrons are lifted and released naturally .
less work is done and only low momentum mode are excited , the rate of electrons in the well region ( @xmath135 ) is more larger .
also , we can see the absent of interferences both in or out the well , as discussed in @xcite .
, sub - figure a , c ) and v - oscillating mode ( @xmath128 , sub - figure b , d ) , particles in the well ( @xmath136 ) and the pumping rate @xmath137 as a function of time .
the triangles denote the time when field is absent and the dot line link them .
the blue triangles denote electron and the red denote positron .
all parameters are the same as fig .
[ fig : imagesc_w ] and fig .
[ fig : imagesc_v ] , respectively.[fig : pump_rate ] ] in the v - oscillating mode , it turns out that the electrons is more inclined to gather in the well region ( defined as @xmath135 ) than it in w - oscillating mode .
we can integrate the spacial density @xmath138 in this region and get the particle number in the well , @xmath139 .
for the pumping process in last section , @xmath140 are graphed in fig .
[ fig : pump_rate](a , b ) . in w - oscillating mode , as time
increasing , @xmath141 increase to a constant @xmath142 quickly , while @xmath143 to a constant @xmath144 .
but in w - oscillating mode , @xmath141 keep increasing while @xmath143keep zero .
the reason is positrons can be generated in the well region in w - oscillating mode , while the wall ( the electric field ) shut the door upon positrons in v - oscillating mode . in a pumping process
, the pumping rate is vitally important and can be defined as @xmath145 , where @xmath146 , as shown in fig .
[ fig : pump_rate](c , d ) . in both modes , at the end of the first cycle , when @xmath147 or @xmath29 , nearly all the electrons are limited in the well region , while positron are ejected . in v - oscillating mode , since all the generated positrons are ejected and kept out of the well , the pump rate directly become @xmath148 . for electron in the v - oscillating mode , or electron and positron in w - oscillating mode , in the long time limit ,
@xmath149 come to @xmath148 as @xmath150 , where @xmath151 depends on the saturation number of particles in the well and the number of particles can be generated in each cycle . in fig .
[ fig : n_t ] , for @xmath92 and @xmath115 , there are @xmath107 and @xmath121 pairs survived at the end @xmath122 .
we have proposed that in low frequency limit , the pairs survived finally should equal to three , the maximum number of embedded bound states swept in one cycle of each mode . in fig .
[ fig : adiabatic ] , ignoring the reflection , for each frequency , the total time is chosen equal to the oscillating period for both modes , so that the oscillation can only finish one cycle .
the final number of pairs survived as a function of the upper boundary of the oscillating width ( @xmath26 ) and depth ( @xmath27 ) are presented . in the adiabatic limit ,
a sub - critical well potential can not trigger pairs .
as the width or depth increasing , the bound states in the gap dive into the negative continuum successively , the potential become super - critical .
pairs can be generated and saturated to the number of embedded bound states .
however , as the width and depth decreasing , bound states depart the negative continuum successively and the generated pairs can not annihilate because of the non - synchronous ejection . finally , the number of pairs survived in this cycle is equal to the maximum number of bound states embedded .
this maximum number is a function of the upper boundary of the two oscillating cycle ( @xmath26 or @xmath27 ) . ; ( b ) , v - oscillating mode , * * @xmath34 .
the total time @xmath152 is chosen equal to the oscillating period .
@xmath26 is in units of @xmath33 , @xmath152 is in units of @xmath153 , @xmath69 and @xmath27 is in units of @xmath36 . [
fig : adiabatic ] ] for a low frequency , the curve which indicate the final number of pairs vs. @xmath26 or @xmath27 is like a flight of stairs . as the frequency become lower , the rising edge of the stairs become more sharper . as shown in fig .
[ fig : adiabatic ] , in the limit @xmath154 , the the rising edge of the stairs will precisely locate at the points where the bound states dive into the negative continuum .
these points are @xmath32(in units of @xmath33 ) , and @xmath35(in units of @xmath36 ) , as illuminated in fig .
[ fig : spectrum ] .
the gaps between bound states in the positive and negative continuum in v - oscillating mode are smaller than that in w - oscillating mode . to achieve a quasi - adiabatic ( finite @xmath28 or @xmath29 ) simulation
, @xmath29 should be larger than @xmath28 to build a similar stairs .
now , if the two quasi - adiabatic oscillating cycle repeate periodically , we can expect a linear increasing pair number , i.e. , for fig .
[ fig : adiabatic](a ) , @xmath155 , the final pair number will be @xmath156 times the number of the cycles .
in this work , we have constructed a toy model , one - dimensional well potential with its width and depth oscillating , and studied the electron - positron creation .
since the bound states diving behavior in the energy spectrum are similar when sweeping the depth or width , the physical process are similar in these two modes . we find that the non - synchronous ejection of particles prevent the particle annihilation , break the pauli block effect in a static super - critical well potential , and lead to a high constant production rate .
the width oscillating mode can deliver more energy to particles and is more efficient in pumping pairs than the depth oscillating mode .
the time evolution of spacial density illustrate the particles pumping from the well and the spreading of them in the numerical box . in a quasi - adiabatic case ,
pair number as a function of upper boundary of the oscillating , will reveal the diving of the bound states .
this can be expected to detect the energy structure of a complicated potential . in order to reduce the computing cost
, we neglect the larger part of the discrete momentum in the numerical simulation . on the other hand , with the same number of discrete momentum
, the number of spatial points can be larger to describe the details of the potential .
the simulation in this paper is done on a personal stand - alone computer . in this algorithm ,
the time evolution of each negative eigenstate can be done on a single cpu , hence the computation can be paralleled easily .
further more , if the second order spatial derivative in the hamiltonian are done by finite difference approximations instead of fourier transformation@xcite , more lager one - dimensional , even two dimensional system can be simulated through paralleling the algorithm on memory shared parallel computers .
this work is supported by national basic research program of china ( 973 program ) ( grants no .
2013cba01502 , no .
2011cb921503 , and no .
2013cb834100 ) , the national natural science foundation of china ( grants no . 11374040 , no . 11274051 , and 11475027 ) .
d. l. burke , r. c. field , g. horton - smith , j. e. spencer , d. walz , s. c. berridge , w. m. bugg , k. shmakov , a. w. weidemann , c. bula , k. t. mcdonald , e. j. prebys , c. bamber , s. j. boege , t. koffas , t. kotseroglou , a. c. melissinos , d. d. meyerhofer , d. a. reis , and w. ragg , phys .
lett . * 79 * , 1626 ( 1997 ) .
g. r. mocken and c. h. keitel , j. comput .
199 , 558 ( 2004 ) .
g. r. mocken and c. h. keitel , comput .
. commun . * 178 * , 868 ( 2008 ) .
m. ruf , h. bauke , and c. h. keitel , j. comp .
phys . * 228 * , 9092 ( 2009 ) .
p. krekora , q. su , and r. grobe , phys .
lett . * 92 * , 040406 ( 2004 ) .
christopher c. gerry , q. su , and r. grobe , phys .
a * 74 * , 044103 ( 2006 ) .
p. krekora , k. cooley , q. su , and r. grobe , phys .
lett . * 95 * , 070403 ( 2005 ) .
y. liu , m. jiang , q. z. lv , y. t. li , r. grobe , and q. su , phys .
rev . a * 89 * , 012127 ( 2014 ) .
suo tang , bai - song xie , ding lu , hong - yu wang , li - bin fu , and jie liu , phys .
a * 88 * , 012106 ( 2013 ) . | in this paper we show that electron - positron pairs can be pumped inexhaustibly with a constant production rate from the one - dimensional well potential with oscillating depth or width .
bound states embedded in the the dirac sea can be pulled out and pushed to the positive continuum , become scattering states .
pauli block , which dominant the saturation of pair creation in the static super - critical well potential , can be broken by the ejection of electrons .
we find that the width oscillating mode is more efficient that the depth oscillating mode . in the adiabatic limit , pair number as a function of upper boundary of the oscillating , will reveal the diving of the bound states . |
s.e.r .- l . thanks r.f . berger and t. birol for valuable discussions .
work at the molecular foundry was supported by the office of science , office of basic energy sciences , of the u.s .
department of energy , and laboratory directed research and development program at the lawrence berkeley national laboratory under contract no .
de - ac02 - 05ch11231 . t.r . acknowledges support from the scidac program on excited state phenomena in energy materials .
f.b . acknowledges the enhanced eurotalent program and the france berkeley fund for supporting his sabbatical leave in uc berkeley .
this research used resources of the national energy research scientific computing center , which is supported by the office of science of the u.s .
department of energy .
51ifxundefined [ 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 * * , ( ) link:\doibase 10.1038/nature08566 [ * * , ( ) ] link:\doibase 10.1038/nmat2750 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1038/ncomms1111 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.235102 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.035119 [ * * , ( ) ] link:\doibase 10.1103/physrevb.93.075204 [ * * , ( ) ] link:\doibase 10.1063/1.4798241 [ * * , ( ) ] link:\doibase 10.1038/nature12582 [ * * , ] link:\doibase 10.1103/physrevlett.107.257602 [ * * , ( ) ] link:\doibase 10.1016/j.ijhydene.2005.10.005 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/acs.inorgchem.5b01042 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.205202 [ * * , ( ) ] link:\doibase 10.1103/physrevb.92.045414 [ * * , ( ) ] link:\doibase 10.1143/jjap.47.5516 [ * * , ( ) ] link:\doibase 10.1016/j.commatsci.2014.09.026 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevb.82.085203 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1016/j.cpc.2009.07.007 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.88.085117 [ * * , ( ) ] link:\doibase 10.1063/1.2404663 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1016/j.cpc.2011.12.006 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevlett.107.146804 [ * * , ( ) ] link:\doibase 10.1080/095008300176308 [ * * , ( ) ] link:\doibase 10.1103/physrevb.68.184111 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/acs.inorgchem.5b00102 [ * * , ( ) ] link:\doibase 10.3938/jkps.56.490 [ * * , ( ) ] link:\doibase 10.1103/physrevb.79.045109 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrev.140.a651 [ * * , ( ) ] link:\doibase 10.1063/1.1415766 [ * * , ( ) ] link:\doibase 10.1016/s0022 - 4596(03)00289 - 5 [ * * , ( ) ] link:\doibase 10.1134/s1063783410100124 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.95.176404 [ * * , ( ) ] @noop * * , ( ) | the ruddlesden - popper ( rp ) homologous series sr@xmath0ti@xmath1o@xmath2 provides a useful template for the study and control of the effects of dimensionality and quantum confinement on the excited state properties of the complex oxide srtio@xmath3 .
we use _ ab initio _ many - body perturbation theory within the @xmath4 approximation and the bethe - salpeter equation approach to calculate quasiparticle energies and absorption spectrum of sr@xmath0ti@xmath1o@xmath2 for @xmath5 and @xmath6 .
our computed direct and indirect optical gaps are in excellent agreement with spectroscopic measurements .
the calculated optical spectra reproduce the main experimental features and reveal excitonic structure near the gap edge .
we find that electron - hole interactions are important across the series , leading to significant exciton binding energies that increase for small @xmath7 and reach a value of 330 mev for @xmath8 , a trend attributed to increased quantum confinement .
we find that the lowest - energy singlet exciton of sr@xmath9tio@xmath10 ( @xmath8 ) localizes in the 2d plane defined by the tio@xmath9 layer , and explain the origin of its localization .
2.5 in 0.5 in # 1 strontium titanite ( srtio@xmath3 ) is a prototypical abo@xmath3 perovskite that , in pure , doped or strained form , displays important dielectric , optoelectronic , and transport properties , such as ferroelectricity @xcite , robust photocatalysis @xcite , superconductivity @xcite and high electron mobility at low temperatures @xcite .
atomically thin two - dimensional layers in srtio@xmath3 heterostructures display exotic electronic effects , such as an emergent two - dimensional electron gas @xcite , magnetism @xcite , and metal - insulator transitions @xcite . recently , anomalous excitonic phenomena have been observed in the optical spectrum of cubic srtio@xmath3 @xcite .
these effects , previously overlooked , arise due to strong coulomb interactions between excited electrons and holes in the absence of strong screening , and can be enhanced through dimensionality and quantum confinement .
fundamental understanding of many - body electronic interactions in complex oxides and their interplay with structural degrees of freedom can lead to the design of novel functionalities and the manipulation of optical properties through electrostatic boundary conditions , size effects , or layered heterostructure .
the homologous sr@xmath0ti@xmath1o@xmath2 ruddlesden - popper ( rp ) series is a well - studied class of materials where , via stoichiometry , one can systematically alter confinement effects on electronic structure . as shown in fig .
[ fig : fig1](a ) , the @xmath7-th member of the rp series , which takes up the tetragonal _ i4/mmm _ structure , consists of a periodic vertical stacking of alternating sro and tio@xmath9 layers with an additional sro rocksalt layer every @xmath7 srtio@xmath3 perovskite unit cells along the stacking direction .
as @xmath7 decreases , the three - dimensional corner - sharing network of octahedra of cubic srtio@xmath3 transforms into the quasi two - dimensional layered structure of sr@xmath9tio@xmath10 ( @xmath8 ) with oxygen octahedra connected only in the @xmath11 plane ( fig .
[ fig : fig1](b ) ) ; the resulting quantum well , defined by the region of connected octahedra , leads to localization due to the modification of coulomb interactions along the vertical direction and the associated quantum confinement , with significant potential consequences for excited - state properties , as we show here .
( a ) structure of the @xmath7-th member of the rp series .
the distortion @xmath12 is defined as the distance between the sr ( ti ) and o atoms along the vertical direction .
sr , ti and o atoms in orange , light - blue and white , respectively .
( b ) rp series as a function of @xmath13 for @xmath14 . ] in this work , we perform _ ab initio _ many - body perturbation theory ( mbpt ) calculations on the rp series and investigate the effect of dimensionality and quantum confinement on quasiparticle energies and absorption spectrum of the complex oxide srtio@xmath3 .
several members of the rp series ( @xmath15 and @xmath6 ) have been experimentally stabilized through molecular beam epitaxy and under epitaxial strain @xcite . while most of the interest in the rp series has focused on its dielectric @xcite and ferroelectric properties @xcite , several members of the series ( @xmath16 ) display photocatalytic activity @xcite . additionally , recently - synthesized halide rp phases show promise in photovoltaic applications @xcite . earlier density functional theory ( dft )
calculations for the rp series @xcite captured the qualitative trend of the experimental indirect optical gaps .
however , these studies neglected electron - hole interactions and excitonic effects , and proper description of quasiparticle and optical properties requires higher level of theory , as used here .
quasiparticle and optical spectra of selected members of the sr@xmath0ti@xmath1o@xmath2 series are obtained with _ ab initio _ mbpt within the @xmath4 approximation @xcite and the bethe - salpeter equation ( bse ) approach @xcite ; which has been shown to successfully describe the closely related compound tio@xmath9 @xcite . in this method ,
direct ( @xmath17 ) and indirect ( @xmath18 ) quasiparticle band gaps are computed with @xmath4 , where the green s function _ g _ and screened potential _
w _ are constructed from mean - field energies and wave functions from dft .
direct optical gaps ( @xmath19 ) are extracted from the bse solution through the imaginary part of the electronic dielectric function @xmath20 , using the @xmath4 calculations as input .
indirect optical gaps ( @xmath21 ) are estimated by subtracting the exciton binding energy ( @xmath22 ) from indirect @xmath4 band gaps , neglecting electron - phonon effects .
first - principles dft calculations are performed within the local density approximation ( lda ) @xcite using the abinit code @xcite .
our abinit calculations use a plane wave basis energy cut - off of 820 ev , monkhorst - pack * k*-point grids @xcite , and the recently proposed ` oncvpsp ` norm - conserving pseudo potentials @xcite , including semicore states ( @xmath23 for sr and @xmath24 for ti ) in the valence .
hybrid functional calculations , within the heyd - scuseria - ernzerhof ( hse06 ) exchange - correlation functional @xcite , are performed with the vasp code @xcite .
our vasp calculations use an energy cut - off of 450 ev and projected augmented wave pseudopotentials @xcite .
electronic and optical properties are computed at the experimental lattice constants @xcite , but with optimized internal atomic positions .
single shot @xmath4 ( @xmath25 ) and @xmath25/bse calculations are performed with berkeleygw code @xcite .
we use dft - lda mean - field calculations as a starting point , 200 bands per atom , an screened coulomb energy cut - off of 190 ev ( 54 ev ) for the self - energy ( electron - hole kernel ) , and the generalized plasmon - pole model ( gpp ) @xcite of godby - needs @xcite .
extensive convergence studies in the number of bands and * k*-point grids show that @xmath25/bse band gaps are converged within 0.1 ev .
further details of the calculations are included in the supplemental information ( si ) . ]
epitaxial strain has been shown to systematically modify the band gap of srtio@xmath3 @xcite . in the case of the rp series , prior spectroscopic measurements were performed on thin films systems for which the in - plane lattice constant of the substrate ( @xmath26 ) induces a small ( @xmath27 1 % ) compressive ( tensile ) strain in the experimental ( theoretical ) structure of each @xmath7 .
full structural relaxation of @xmath16 and @xmath6 with dft - lda underestimate experimental lattice constants of free - standing compounds by @xmath281% , as is typical for the lda ( see table i in si ) . as shown in fig .
[ fig : fig1](a ) , the symmetry of the rp structure allows an off - centering or _ distortion _
@xmath12 of the sr ( ti ) atoms in the sro ( tio@xmath9 ) layers . for @xmath8
, we find @xmath29 , in agreement with previous calculations @xcite and experiments @xcite . for large @xmath7 ( @xmath30 )
, the distortion converges to @xmath31 and @xmath32 next to the sro layer , and decreases rapidly toward the center of the perovskite layers at fixed @xmath7 ( see fig . 1 in si ) .
( a ) lda band structure for @xmath5 and @xmath33 .
the arrow points to the low - lying conduction states of @xmath34 .
surface of constant charge density ( violet ) at @xmath35 for ( b ) cbm and ( c ) cbm@xmath36 of @xmath34 , and ( d ) cbm of @xmath34 in the absence of distortion @xmath37 .
atom color code is as in fig .
[ fig : fig1 ] . ]
we start by considering the single - particle electronic properties of the rp series as a function of @xmath7 .
fig .
[ fig : fig2](a ) shows the lda band structure of @xmath5 and @xmath6 along equivalent paths in the corresponding brillouin zones ( see table ii and fig . 2 in si ) .
direct ( @xmath35 @xmath38 @xmath35 ) and indirect ( @xmath39 @xmath38 @xmath35 for @xmath40 , @xmath41 @xmath38 @xmath35 for @xmath33 ) band gaps are composed of transitions primarily between o @xmath42 and ti @xmath43 states . for @xmath40 , the conduction band minimum ( cbm ) and the valence band maximum ( vbm )
are degenerate : the cbm is at both @xmath35 and @xmath44 , whereas the vbm is at @xmath45 and @xmath39 .
the degeneracy is due to the reduced dimensionality of the system .
as @xmath7 increases , the dispersion of conduction bands decreases along the @xmath46 symmetry line . for @xmath33 , we recover the band structure of srtio@xmath3 .
computed direct and indirect lda band gaps are shown as a function of @xmath13 at the bottom of fig .
[ fig : fig3 ] . compared to spectroscopic measurements ( error bars ) @xcite , direct and indirect lda band gaps display two important differences .
first , band gaps are severely underestimated by @xmath281.5 ev , as is usual for lda kohn - sham states .
although , indirect lda band gaps reproduce the experimental trend of indirect optical band gaps @xcite , direct lda band gaps deviate from experimental trends at small @xmath7 , where quantum confinement effects are expected to play an important role .
second , band gaps display a discontinuity of @xmath280.2 ev for large @xmath7 , which is not observed in experiments .
different choices of semi - local functionals or lattice constants ( e.g. experimental , relaxed or epitaxially strained ) display the same discontinuity , up to a rigid shift of @xmath280.1 ev .
the band gap discontinuity for large @xmath7 arises from the low - lying conduction states that are significant for large finite @xmath7 ( @xmath30 ) but are absent for @xmath33 ( see fig .
[ fig : fig2](a ) ) .
these states arise due to the distortion of the two ti atoms adjacent to the extra sro layers and the associated splitting of the local ti @xmath47 states . as shown in fig .
[ fig : fig2](b ) for @xmath34 , the cbm wavefunction for large finite @xmath7 corresponds to two ti @xmath48 states localized at the tio@xmath9 layers next to the extra sro layers .
the bands above the low - lying conduction states ( cbm@xmath36 ) are formed by @xmath49 ti d@xmath50 states that localize at the center of the perovskite layers ( see fig .
[ fig : fig2](c ) ) .
the energy difference between the cbm and cbm+2 converges to @xmath51 ev at large @xmath7 and leads to the discontinuity .
the distortions at the extra sro layers localize the cbm ( vbm ) orbital at the edges ( center ) of the perovskite layers , leading to the quasi two - dimensional properties of the rp series .
if the distortions are removed from the structure of @xmath7 by artificially aligning the sr and ti atoms with the o atoms , the cbm becomes a degenerate set of @xmath7 ti @xmath48 states and the cbm wavefunction recovers to an unconfined uniform distribution of @xmath48 states along the vertical direction ( see fig .
[ fig : fig2](d ) ) .
note that if the distortions are removed from the entire series , lda band gaps display a linear trend as a function of @xmath13 that deviates from experiments at large @xmath7 ( see lda@xmath52 in fig .
[ fig : fig3 ] ) . ( a )
direct and ( b ) indirect band gaps as a function of @xmath13 using different levels of theory .
experimental values are taken from ref .
@xcite ( black solid error bars ) and table [ table : optical ] ( red dash error bars ) . ] before discussing our _ ab initio _ mbpt results , we compare our lda results with hybrid functional calculations . as a function of @xmath13 ( fig . [
fig : fig3 ] ) , hse06 band gaps are in good agreement with measured optical band gaps .
however , this agreement , while good , is fortuitous , since the generalized kohn - sham system is formally unable to describe interacting two - particle states , e.g. excitons .
trends in hse06 band gaps are well approximated by a rigid shift of lda band gaps , and in particular , underestimate by @xmath53 ev the experimental direct optical band gap of @xmath8 , where electron - electron and electron - hole interaction effects are expected to be important .
lcc l + + this work + @xmath25(gpp)/lda + @xmath25(ff)/lda + prev .
work + @xmath25/lda + @xmath25/lda + @xmath25/lda+u + @xmath25/lda + expts .
+ cc c srtio@xmath3 + cc c @xmath17 + + 3.90 + 3.86 + + 3.76 + 3.80 + 3.77 + + 3.30 + & c @xmath18 + + 3.57 + 3.52 + + + 3.40 + 3.28 + 3.32 + + & c sr@xmath9tio@xmath10 + cc c @xmath17 + + 4.17 + 4.21 + + + + + + + & c @xmath18 + + 4.02 + 4.06 + + + + + + + l ref .
+ + + + + + @xcite + @xcite + @xcite + @xcite + @xcite + [ table : electronic ] we now turn to our results with _ ab initio _ mbpt .
table [ table : electronic ] shows fundamental band gaps for cubic srtio@xmath3 and @xmath8 calculated with @xmath25/lda .
our @xmath25 calculations for cubic srtio@xmath3 are in line with previous @xmath4 results ( by @xmath54 ev ) but overestimate photoemission measurements ( by @xmath55 ev ) .
full - frequency @xmath4(ff ) calculations for cubic srtio@xmath3 and @xmath8 are well approximated by @xmath4(gpp ) , and we therefore use the gpp for the rest of the series . as a function of @xmath13 ( fig .
[ fig : fig3 ] ) , direct ( indirect ) @xmath25 band gaps increase monotonically by @xmath280.3 ev ( 0.5 ev ) , with a steeper slope than hse06 band gaps . @xmath25
band gaps overestimate experimental optical band gaps due to the absence of electron - hole interactions . for @xmath8 ,
the measured direct ( indirect ) optical gap is overestimated by @xmath280.3 ev ( 0.5 ev ) .
lcc l + + this work + @xmath25/bse + prev .
work + @xmath25/bse + expts .
+ + + + + + + cc c srtio@xmath3 + cc c @xmath19 + + 3.73 + + 3.54 + 3.80 + 3.63 + + 3.75 + + + 3.90 + & c @xmath21 + + 3.40 + + + 3.20 + 3.15 + 3.20 + 3.25 + 3.10 + 3.29 + + & c sr@xmath9tio@xmath10 + cc c @xmath19 + + 3.84 + + + + 3.86 + + + + + + & c @xmath21 + + 3.69 + + + + 3.48 + + + 3.40 + + + l ref .
+ + + + + @xcite + @xcite + @xcite + @xcite + @xcite + @xcite + @xcite + @xcite + [ table : optical ] table [ table : optical ] shows @xmath25/bse direct optical gaps for cubic srtio@xmath3 and @xmath8 as obtained from an exact diagonalization solution of the bse .
these calculations correctly include the electron - hole attraction that reduces the optical gap relative to the quasiparticle ( fundamental ) band gap .
our direct bse optical gaps for srtio@xmath3 are in good agreement with previous calculations and experiments .
the large overestimation of @xmath25 for @xmath8 is compensated by a large increase of exciton binding energy , which almost doubles from @xmath33 ( 170 mev ) to @xmath8 ( 330 mev ) , an increase that can be attributed to quantum confinement .
note that the increase of exciton binding energy is not explained by trends in effective mass and electronic dielectric constant as a function of @xmath7 ( see si ) . as a function of @xmath13 ( fig .
[ fig : fig3 ] ) , @xmath25/bse band gaps extracted from the first feature ( peak or shoulder for @xmath40 , first excitation for @xmath33 ) of the @xmath25/bse optical spectrum are in excellent agreement ( by @xmath54 ev ) with measured optical gaps . the best agreement is obtained for small @xmath7 direct gaps , where indirect phonon - assisted transitions are irrelevant .
we find relatively large electron - hole interaction across the entire series , the exciton binding energy remains large for several members of the series : @xmath56 ( 330 mev ) and @xmath34 ( 300 mev ) , since the cbm remains localized at the tio@xmath9 layers next to the extra sro layers .
( a ) absorption spectra for sr@xmath0ti@xmath57o@xmath2 ( @xmath5 and @xmath6 ) with ( solid green ) and without ( dashed red ) electron - hole interactions .
the black arrows show the bse band gap .
experiments ( dashed violet ) are taken from ref .
@xcite .
( b ) top and ( c ) side view of the exciton wave function of @xmath8 ( violet ) and the position of the hole in the oxygen atom ( black dot ) . ]
we focus now in the properties of the @xmath25/bse spectrum as a function of @xmath7 .
fig .
[ fig : fig4](a ) shows the experimental and computed absorption spectra for @xmath5 and @xmath6 with ( @xmath25/bse ) and without ( @xmath25/rpa ) the inclusion of electron - hole interactions .
the @xmath25/bse spectra correctly reproduces the position and magnitude of the experimental onset and reveal excitonic structure .
the spectrum of @xmath8 displays two - dimensional characteristics and strong excitonic peaks ; there is a first peak at 3.84 ev , a weak second peak at 4.00 ev , and a strong third peak at 4.40 ev . from direct examination of the lowest - energy bse solution ,
the first peak arises primarily from direct transitions to the ti @xmath48 conduction band .
the large energy difference ( 0.5 ev ) between the dominant first and third peak is explained by the large energy difference ( 0.9 ev ) between the cbm and the rest of the conduction bands .
the first peak merges with the rest of the spectrum at a critical layer thickness @xmath58 , and the third peak becomes the spectrum onset at @xmath33 . as @xmath7 increases , there is a large transfer of spectral weight toward small energies due to the decrease of dispersion along the @xmath46 line ( see fig .
[ fig : fig2 ] ( a ) ) . at the critical layer thickness @xmath59 ,
transitions to the cbm at the sro layers become negligible compared to transitions to the cbm@xmath36 at the center of the perovskite layers . as already suggested in ref .
@xcite , the volume of the increasingly large region of connected octahedra dominates over the thin sro extra layers , and the spectrum approaches the spectrum of @xmath33 .
experimentally , low - lying conduction states and excitonic effects become negligible due to limited resolution ( see fig .
[ fig : fig4 ] ( a ) ) and optical gaps converge continuously with increasing @xmath7 , explaining the absence of band gap discontinuity in the optical measurements ( see fig .
[ fig : fig3 ] ( a ) ) .
the increase of exciton binding energy at small @xmath7 implies a spatial localization of the exciton of @xmath8 . as a means of visualizing the lowest singlet exciton , fig .
[ fig : fig4](b)-(c ) shows the top and side views of an isosurface probability amplitude of finding an electron given a hole is fixed near the o atom .
the exciton displays charge - transfer character and localizes , in the 2d plane defined by the lone tio@xmath9 layer .
the real space wave function envelope of the exciton resembles the @xmath60 state of the 2d hydrogenic model and has an average radius of @xmath61 @xcite , with @xmath62 the bohr radius .
our calculations show that the exciton binding energy is insensitive to epitaxial strain : while direct and indirect band gaps decrease linearly by @xmath63 ev from tensile ( @xmath64% ) to compressive ( @xmath65% ) strain , the exciton binding energy varies by less than 10 mev in the same strain range . as @xmath7 increases , the exciton localizes on the tio@xmath9 layers next to the extra sro layer due to the distortions next to the sro layers . in summary ,
we have used dft with different functionals , and the @xmath25/bse approach to compute quasiparticle and optical spectra of selected members of the sr@xmath0ti@xmath1o@xmath2 series .
our @xmath25/bse direct and indirect optical gaps for the rp series are in excellent agreement with measured optical gaps when electron - hole interactions are included .
we find that quantum confinement , induced by the extra sro layers in the rp series , produces a larger increase in quasiparticle band gaps , relative to trends in optical gaps .
the exciton binding energy is significant and increases for small @xmath7 , reaching a value of 330 mev .
the @xmath25/bse spectra reproduce the main experimental features and reveal excitonic structure near the gap edge .
our work suggests that the rp series may be a useful framework for the study of dimensionality and quantum confinement in optoelectronic properties of complex oxides . |
in the favored cold dark matter cosmology , present - day galaxies were assembled hierarchically from smaller building blocks at earlier cosmic times . since all nearby galaxies with stellar spheroids are observed to host nuclear smbhs @xcite , hierarchical merging leads inevitably to the formation of smbh binaries @xcite .
if the binary lifetime exceeds the typical time between mergers , then some galactic nuclei should contain systems of three or more smbhs .
these systems are particularly interesting as they often lead to the ejection of one of the bhs at a speed comparable to the galactic escape velocity @xcite . in massive elliptical galaxies
the typical speeds are @xmath2 , far greater than attainable through gravitational radiation recoil @xcite . spatially resolved pairs of nuclei have been observed in a few active galaxies .
the most famous example is ngc 6240 , an ultraluminous infrared galaxy ( ulirg ) in which two distinct active galactic nuclei ( agn ) are clearly seen in hard x - rays at a projected separation of @xmath01 kpc @xcite .
@xcite observed a variable uv source , possibly a second active nucleus , at a projected separation of @xmath3 pc from the primary nucleus in the spiral galaxy ngc 4736 , which shows signs of a recent merger . @xcite
have detected what is thought to be an smbh binary at a projected separation of just 7.3 pc in the radio galaxy 0402 + 379 , through multi - frequency radio observations using the very long baseline array ( vlba ) .
we begin by discussing the theory of how such systems evolve , and the conditions under which they might acquire a third bh .
when two galaxies merge , their dense nuclei sink to the center of the merger product by dynamical friction . as the nuclei spiral in
, tidal forces gradually strip the two smbhs of their surrounding stars and dark matter . in mergers between galaxies of comparable mass ,
the bhs are able to come together and form a bound smbh binary on a timescale of order @xmath4 yrs .
the binary continues to harden by dynamical friction until it reaches a separation of order @xmath5 known as the `` hardening radius '' ( e.g. @xcite ) . here
@xmath6 is the reduced mass of the two bhs with masses @xmath7 and @xmath8 , @xmath9 is the velocity dispersion of the stars beyond the binary s sphere of gravitational influence , @xmath10 is the binary mass ratio @xmath11 , and @xmath12 is the total mass of the binary . for smaller separations
the binary looks like a point mass to the distant stars contributing to dynamical friction , but close stellar encounters preferentially harden the binary and so dominate further energy loss .
only stars on nearly radial orbits , with periapsis distances of order the binary separation , can extract energy from ( `` harden '' ) the binary in this stage .
these stars undergo strong 3-body interactions with the binary and escape its vicinity with speeds comparable to the black holes orbital speed . in the low - density nuclei of large elliptical galaxies , the total mass in stars on such `` loss cone '' orbits is small compared to the mass of the binary . furthermore the two - body stellar relaxation time
is long compared to a hubble time , so once the stars initially on loss cone orbits are cleared out , the loss cone remains empty @xcite .
since the binary must eject of order its own mass per @xmath13-folding in its semi - major axis , the system stops hardening around @xmath14 unless some other mechanism causes sufficient mass flux through the binary .
if the binary reaches a separation around @xmath15^{1/4 } \hspace{2pt } \cdot \notag \\ & \left ( \frac{\tau_{gw}}{10^{10 } \hspace{2pt } { \rm yrs } } \right)^{1/4 } f^{-1/4}(e ) \hspace{6pt } { \rm pc},\end{aligned}\ ] ] where @xmath13 is the orbital eccentricity of the binary and @xmath16 , then it can coalesce on a timescale @xmath17 through gravitational radiation @xcite . to get from @xmath14
to @xmath18 it must bridge a gap @xmath19^{3/4}\ ] ] by some mechanism other than stellar - dynamical friction or gravitational radiation .
the question of whether and how it crosses this gap has become known as the `` final parsec problem '' @xcite .
in many galaxies there probably are alternative mechanisms for crossing the gap .
when gas - rich galaxies merge , tidal torques channel large amounts of gas into the central @xmath20 pc @xcite .
the gas may lose energy through radiation and angular momentum through viscous torques , and is therefore not subject to a loss cone problem .
using smoothed particle hydrodynamics simulations @xcite compute a merger time of order @xmath21 yrs in an environment typical of the central regions of ulirgs , which are thought to be gas - rich galaxies caught in the act of merging @xcite .
the nuclei of galaxies are also observed to contain numerous massive perturbers ( mps ) such as star clusters , molecular clouds , and possibly intermediate - mass black holes ( imbhs ) .
these objects scatter stars into the loss cone much more efficiently than other stellar mass objects , since the relaxation rate scales as the perturber mass for a fixed mass density of perturbers .
@xcite extended the fokker - planck loss cone formalism @xcite to accomodate a spectrum of perturber masses and account for relaxation by rare close encounters with mps .
they show that the population of known mps in the nucleus of the milky way is sufficient to bring a @xmath22 m@xmath23 bh binary to @xmath18 in @xmath24 yrs , and it is reasonable to expect similar perturber populations in other star - forming spiral galaxies .
the final parsec problem is often mentioned as a caveat when predicting the smbh coalescence signal in low - frequency gravitational wave detectors such as the upcoming laser interferometer space anntena ( lisa ) .
however the lisa event rate is expected to be dominated by small galaxies at high redshift @xcite , where the gas content and central densities tend to be high and the relaxation times short .
for this reason _ the stalling problem is probably not a significant concern for the lisa smbh coalescence signal_. on the other hand bh ejections by gravitational radiation recoil @xcite may play an important role in the high - redshift coalescence rate .
the long - term survival of smbh binaries is likewise unlikely in the gas - rich cores of quasars and ulirgs .
however none of the gap - crossing mechanisms discussed so far are likely to reduce the coalescence time below a hubble time in mergers between giant , gas - poor elliptical galaxies .
@xcite show that a significant fraction of stars on `` centrophilic '' orbits in a triaxial potential can greatly increase the mass flux into the loss cone .
some non - axisymmetric potentials can also excite bar instabilities that cause rapid mass flow through the binary and efficient coalescence @xcite .
however a central smbh can disrupt box orbits and induce axisymmetry in the inner regions of a triaxial galaxy @xcite , and it is uncertain how often these geometry - specific mechanisms bring the coalescence time below a hubble time . one can naively assess the likelihood of coalescence by considering the `` full '' and `` empty '' loss cone hardening times , @xmath25 and @xmath26 , in the nuclei of various galaxies assuming a spherical and isotropic distribution function .
@xmath25 is the hardening time assuming every star kicked out of the loss cone is instantly replaced , while @xmath26 is the time assuming stellar two - body relaxation to be the only replenishing mechanism .
in small , dense galaxies @xmath27 yrs and @xmath28 yrs while in the lowest - density cores of giant ellipticals and cd galaxies @xmath29 yrs and @xmath30 yrs @xcite . while the empty loss cone rate is difficult to believe in any galaxy given at least some clustering on scales larger than 1 m@xmath23 , it also seems difficult to approach the full loss cone rate if there is no gas around and no strong radial bias in the stellar distribution . from this point of view
the stalling of binaries seems unlikely in small galaxies but probable in low - density , gas - poor ellipticals .
if some binaries do survive for around a hubble time , then the hierarchical buildup of galaxies will inevitably place three or more smbhs in some merging systems .
an inspiralling satellite affects the evolution of a binary smbh even long before it sinks to the center , by perturbing the large - scale potential and scattering stars into the loss cone .
we may estimate the extent of this effect as a function of satellite mass and distance from the center of the host galaxy using a rough but simple argument due to @xcite .
the change in velocity necessary to deflect a star at radius @xmath10 into the loss cone is @xmath31 , where @xmath32 is the characteristic specific angular momentum of stars on loss cone orbits @xcite , @xmath33 is the smbhs radius of influence , and @xmath34 is the binary separation .
the dynamical time at this radius is @xmath35 , so the acceleration required to scatter a star into the loss cone is roughly @xmath36 . equating this with the tidal acceleration caused by the satellite , @xmath37 , where @xmath38 is the satellite s radius , yields @xmath39 , or with @xmath40 , @xmath41^{1/3 } r.\ ] ] the @xmath38-dependence of @xmath42 reflects the tidal stripping of the satellite as it spirals inward .
equation defines a critical radius @xmath10 , outside of which the satellite can deflect stars into ( and out of ) the loss cone in one dynamical time .
the mass flux through the binary induced by the satellite is then approximately @xmath43 where @xmath44 is the density of the host galaxy at radius @xmath10 and @xmath45 is the geometrical factor accounting for the fraction of stars on loss cone orbits as a function of radius @xmath38 , assuming an isotropic distribution function @xcite . for a fixed satellite mass and distance
, we can then define a `` binary feeding '' timescale by @xmath46 to determine whether the scattering of stars into the loss cone by the satellite is sufficient to harden the binary enough to prevent a close 3-body encounter before the intruder arrives at the galactic center , we must compare @xmath47 with the timescale on which the satellite spirals in by dynamical friction . in the approximation of slow inspiral
we may write the dynamical friction timescale as @xmath48},\ ] ] where @xmath49 .
@xmath50 in equation is computed from @xmath51 , where @xmath52 is the mass of the host galaxy enclosed within radius @xmath38 .
@xmath53 is the satellite mass contained within the tidal truncation radius obtained from a simple point mass approximation , @xmath54^{1/3 } r$ ] ( this slightly underestimates @xmath55 as the satellite approaches the center of the host ) . in fig .
1 we plot @xmath56 and @xmath47 as a function of @xmath38 for a satellite with one third the stellar mass of the host , which contains a binary with @xmath57 m@xmath23 .
both host and satellite are modelled as hernquist profiles @xcite , with their masses and effective radii set by observed scaling relations .
the details of the galactic model are described further in 2.1 - 2 .
, on which an inspiralling satellite scatters mass into the loss cone of an smbh binary , with the dynamical friction timescale , @xmath56 , on which the satellite spirals in .
upper ( red dashed ) line : @xmath47 computed from equation ; lower curve : @xmath56 computed from equation .
both timescales are plotted as a function of the satellite s distance from the center of the host galaxy . for this plot
we chose a binary mass of @xmath58 m@xmath23 and merger mass ratio of 3:1 in the stars .
the galactic model is discussed in the text . ]
since @xmath47 remains about an order of magnitude above @xmath56 throughout the inspiral , this simple calculation makes it plausible that the binary survives the merger process and undergoes close triple interactions with the infalling smbh .
the tidal approximation ( as well as our treatment of dynamical friction ) breaks down as the satellite approaches @xmath59 , so the plot is cut off at a separation of @xmath20 pc , when the satellite still has @xmath04 @xmath13-foldings to go to reach @xmath14 .
however this final stage of the inspiral is found to proceed very rapidly in n - body simulations @xcite .
the merger s effect on the binary may be dominated by violent relaxation or collective effects such as a bar instability @xcite , in which case our two - body approach does not capture its essence .
the evolution of the core distribution function under the influence of a major merger is an intriguing open problem for simulators . after the third bh becomes bound to the binary ( but still before the onset of close 3-body interactions ) another hardening mechanism may become important . if the angle of inclination @xmath60 of the outer binary ( formed by the intruder and the inner binary center - of - mass ( com ) ) exceeds a critical angle @xmath61 , then the quadrupolar perturbation from the intruder induces eccentricity oscillations through a maximum @xcite @xmath62 since the gravitational radiation rate increases sharply toward high eccentricities , these `` kozai oscillations '' can greatly enhance the radiation , possibly causing the binary to coalesce before it can undergo strong 3-body interactions with the intruder @xcite .
general relativistic precession can destroy the kozai resonance ( e.g. @xcite ) , but @xcite find that this does not happen for @xmath63^{1/3 } \sqrt{\frac{1-e_{in}^{2}}{1-e_{out}^{2}}},\ ] ] where @xmath64 and @xmath65 are the semi - major axes of the inner and outer binaries , @xmath66 is the outer binary mass ratio , and @xmath67 and @xmath68 are the inner and outer eccentricities .
this leaves a window of about a factor of 10 in @xmath69 in which the kozai mechanism can operate before unstable 3-body interactions begin .
the actual enhancement of the gravitational radiation rate of course depends on the amount of time spent at high eccentricity , but one may place an upper limit on the importance of kozai oscillations by computing the radiation timescale if the inner binary spends all of its time at @xmath70 . the orbit - averaged power radiated by gravitational radiation
is given by @xmath71 @xcite , where @xmath72 is the semi - major axis and @xmath13 is the eccentricity . in fig .
2 we plot contours of the gravitational radiation time @xmath73 in the @xmath72-@xmath60 plane by putting @xmath70 into equation , for an equal - mass @xmath74 m@xmath23 binary . m@xmath23 ) .
the cosine of the initial inclination angle is plotted on the horizontal axis , and the inner binary semi - major axis is plotted on the vertical axis .
contours are plotted for gravitational radiation timescales of @xmath75 , @xmath76 , @xmath77 , and @xmath78 yrs , if the binary were to stay at maximum eccentricity throughout the whole oscillation cycle .
the horizontal lines indicate the hardening radius and the separation such that a circular binary would coalesce on a @xmath79 yr timescale . ]
this may seem like a gross overestimate of the gravitational radiation rate , especially since the shape of the kozai oscillations is in fact such that the binary spends more time near @xmath80 than near @xmath70 .
however since @xmath17 is so strongly dominated by periapsis passages at @xmath81 , the shift in the contours for a realistic high-@xmath13 duty cycle is only modest .
see @xcite for comparison with a detailed study of radiation enhancement by kozai oscillations in binaries with initial @xmath82 yrs . for a binary at @xmath14 , kozai oscillations can induce coalescence within @xmath76 yrs in @xmath83% of cases assuming @xmath84
is uniformly distributed . in the remainder of cases
the inner binary may survive until the outer binary shrinks to the point of unstable 3-body interactions .
if the intruder comes close enough before it causes sufficient hardening of the ( inner ) binary , then a strong 3-body encounter takes place .
strong encounters are characterized by a significant transfer of energy between the binary s internal degrees of freedom and the com motion of the binary and third body .
when the intruder is slow relative to the binary s orbital speed @xmath85 , energy typically flows from the inner binary to the outer components , so that the binary is more strongly bound after the encounter .
this is one manifestation of the negative specific heat characteristic of gravitationally bound systems .
the encounter ends in the escape of one of the three bodies , usually the lightest , from the system at a speed comparable to @xmath85 .
when the lightest body @xmath86 escapes , momentum conservation requires that the binary com recoil in the opposite direction with a speed smaller by a factor @xmath87 .
it is instructive to compare the expected ejection velocities of the binary and @xmath86 with the typical galactic escape velocity . for a circular binary with @xmath88 , the binding energy at the hardening radius
is @xmath89^{3/2}$ ] erg .
the binding energy at the radius where @xmath90 yrs is @xmath91^{5/4}$ ] erg .
the mean energy @xmath92 harvested from the binary in close encounters with slow intruders is about @xmath93 , though the median @xmath92 is somewhat lower @xcite .
energy conservation implies that the escaper leaves the system with kinetic energy @xmath94 $ ] while the binary leaves with @xmath95 $ ] in the system com frame . for an equal mass binary with @xmath96 m@xmath23 ,
this gives ejection velocities of @xmath97 and @xmath98 for the binary at @xmath14 , and @xmath99 and @xmath100 for the binary at @xmath18 .
any nonzero eccentricity of the binary will increase the semi - major axis corresponding to a fixed @xmath17 , lowering the ejection velocities for the binary at @xmath18 .
also any deviation from equal masses will result in a smaller fraction of the extracted energy being apportioned to the binary and a smaller binary recoil velocity .
the typical escape velocity for galaxies hosting @xmath101 m@xmath23 bhs is around 1500 @xmath102 , accounting for both the stars and the dark matter .
from these numbers , it appears that single escapes will be fairly common as repeated encounters harden the binary to @xmath103 .
however accounting for realistic mass ratios and eccentricities ( the first 3-body encounter tends to thermalize the eccentricity even if it starts off circular ) , binary escapes should be rare .
since the binary must come near the escape velocity to remain outside the nucleus for a significant amount of time , we do not expect triple interactions to empty many nuclei of bhs .
we will quantify these statements with our triple - bh simulations .
the formation of triple smbh systems through inspiral of a merging satellite leads to a rather specific initial configuration .
the three bhs start off as a bound `` hierarchical triple , '' consisting of an inner binary with @xmath104 and a more widely separated outer binary with semi - major axis @xmath65 .
for very large @xmath69 we expect hierarchical triples to exhibit very regular behavior ; in this case the third body sees the inner binary as a point mass and the system essentially consists of two independent ( inner and outer ) binaries .
however as @xmath69 approaches unity , secular evolution gives way to chaotic 3-body interactions in which the orbits diverge and the system becomes subject to escape of one its components .
@xcite derive a criterion for the stability of 3-body systems based on an analogue with the problem of binary tides .
the most distant intruder orbit at which unstable interactions can begin is reliably estimated by @xmath105^{2/5},\ ] ] where @xmath106 is the periapsis separation of the outer binary , @xmath64 is the semimajor axis of the inner binary , @xmath107 is the outer binary mass ratio , and @xmath68 is its eccentricity .
this criterion has great practical importance due to the high numerical cost of unnecessarily following weak hierarchical systems .
it specifies an optimal starting point for our simulations , which aim to study strong interactions in 3-body systems starting off as hierarchical triples . naively one might expect a strong 3-body encounter following a merger with a galaxy hosting a binary , so long as the intruder does not induce coalescence of the binary before it reaches the center .
however the stability criterion implies a condition for close interactions much more stringent than this . to undergo a chaotic encounter with the inner binary
, the intruder must reach the stability boundary before the outer binary hardens and stalls .
a triple system covers somewhat more stellar phase space than a binary of the same size , but not by much for a stable hierarchical system .
this means that the merger process can not cause the binary to harden by more than around an @xmath13-folding for a nearly circular , equal - mass system before the intruder arrives at the center .
though the order - of - magnitude estimates in the previous section make this plausible , further study is needed to determine the likelihood of unstable triple interactions in realistic merger situations .
an eccentric outer binary relaxes the criterion somewhat , but dynamical friction tends to circularize the orbits of satellites with moderate initial eccentricities before they reach the nucleus @xcite .
we therefore assume near - circular initial orbits and begin each simulation from a weakly hierarchical configuration .
triple smbh systems in galactic nuclei were first considered by @xcite , who computed an extensive series of newtonian 3- and 4-body orbits , and compared the slingshot ejection statistics to the observed structure of extragalactic radio sources .
@xcite included a gravitational radiation drag force in the 3-body dynamics .
he showed that this perturbing force could in some cases yield much higher ejection velocities than would be possible in newtonian gravity , with associated bursts of gravitational waves .
the more complex problem of three or four smbhs coming together in the hierarchical merging process and interacting in a galactic potential was first addressed by @xcite and @xcite , who experimented with a variety of initial bh configurations . @xcite
studied binary - binary scattering in galactic nuclei using initial conditions ( ics ) based on extended press schechter theory @xcite .
@xcite followed the formation of triple bh systems in halo merger trees tracking the hierarchical buildup of smbhs from @xmath108 m@xmath23 seeds in high-@xmath9 peaks at @xmath109 . using a simple analytic prescription for the ejection velocities
, they inferred the presence of a large population of smbhs and imbhs wandering through the halos of galaxies and intergalactic space .
@xcite performed the first full n - body simulations of equal - mass triple bh systems embedded in stellar bulges , an important contribution to our understanding of galactic nuclei .
because of the large computation time required for each run , they could not statistically sample the highly varied outcomes of the 3-body encounters as the previous authors did .
in this paper we study the dynamics of repeated triple - smbh interactions in galactic nuclei . between close encounters
we follow the wandering bhs through the galaxy as their orbits decay by dynamical friction .
we use physically - motivated initial bh configurations and mass distributions , and updated galactic models characteristic of the low - density , massive elliptical galaxies in which smbh binaries are most likely stall . we include both a stellar and a dark matter component , with the stellar spheroid fixed to lie on the observed @xmath110 and @xmath111 relations @xcite .
the close encounters are treated using a ks - regularized bulirsch - stoer integrator provided by sverre aarseth @xcite .
the inner density profile is updated throughout the simulations to roughly account for core heating by dynamical friction and stellar mass ejection .
gravitational radiation losses are modelled as a drag force determined by the relative coordinates and velocities of each pair .
each simulation takes only a few minutes to run , so we can try a variety of distributions of ics and statistically sample the outcomes for each .
we use this algorithm to study a variety of consequences of the ongoing encounters , such as the merging efficiency of bh pairs , the time spent wandering at various distances from the galactic center , the distribution of final sizes and eccentricities of the binaries remaining in the galaxy after a steady state has been reached , and the extent of the core scouring caused by the triple smbh systems . aside from the motivating order - of - magnitude calculations in previous sections , this paper does not address the question of _ whether _ close triple smbh systems form in galactic nuclei .
we start our simulations from a state that the system must reach shortly before the onset of unstable 3-body interactions _ assuming _ that they occur , and proceed to derive the subsequent evolution .
our results may be used to argue for or against the occurrence of triple systems in real galaxies , as observations support or disfavor the signatures that we derive . in 2
we describe our model and code methods . in
3 we present the results of our study , and in 4 we discuss these results and conclude .
to get a physically motivated distribution of bh mass ratios , we associate the formation of the inner and outer binaries with the last two major mergers in the history of the galactic halo hosting the triple system .
we use extended press - schechter theory @xcite to calculate the probability distributions of the halo formation times and progenitor masses , and randomly select the parameters of the previous two mergers from these distributions .
we then assign a bh to each progenitor halo using a simple prescription based on the assumption of a flat galactic rotation curve .
@xcite derive the instantaneous halo merger rate , @xmath112}{[1-\sigma^{2}(m_{f})/\sigma^{2}(m_{1})]^{3/2}}.\end{gathered}\ ] ] this equation gives the probability , per unit time per unit mass of @xmath113 , of a given halo of mass @xmath114 merging with another halo of mass @xmath113 to form a product of mass @xmath115 @xmath116 @xmath117 at time @xmath118 . here @xmath119 is the present - day variance of the linear density field on mass scale @xmath120 , @xmath121 where @xmath122 is the power spectrum of density fluctuations today , @xmath123 is a tophat window function , and @xmath38 is related to @xmath120 through @xmath124 , the volume times the present - day matter density .
@xmath122 is related to the primordial power spectrum through the transfer function @xmath125 , which encapsulates the suppression of perturbations on small scales due to radiation pressure and damping over the history of the universe .
for @xmath125 we adopt the standard fitting formulae of @xcite . for the linear growth function
@xmath126 we use the approximation @xmath127 \left [ 1+\frac{\omega_{\lambda}(z)}{70 } \right ] } , \ ] ] good to within a few percent for all plausible values of @xmath128 and @xmath129 @xcite .
@xmath130 is the growth function for an einstein - de sitter universe , @xmath131 $ ] is the matter density ( normalized to the critical density ) as a function of redshift , and we take @xmath132 assuming the rest of the density is in the form of a cosmological constant .
@xmath133 has the weak redshift dependence @xcite @xmath134.\ ] ] we adopt the cosmological parameters obtained from three years of data collection by the wilkinson microwave anisotropy probe ( wmap ) , @xmath135 , @xmath136 , @xmath137 , @xmath138 , and @xmath139 @xcite .
since the merger rate diverges as @xmath140 @xmath141 0 , applications of the formula that track individual merging halos must employ a cutoff mass ratio @xmath140 @xmath142 @xmath143 , such that all mergers below @xmath143 are treated as smooth _ accretion _ rather than as discrete mergers ( see @xcite for further discussion ) .
the instantaneous rate of accretion onto a halo of mass @xmath120 at redshift @xmath144 is @xmath145 to get the growth history ( `` accretion track '' ) of a halo of mass @xmath146 at time @xmath147 due to accretion since the last merger , one need only solve the differential equation @xmath148 @xmath116 @xmath149 $ ] , subject to the initial condition @xmath150
@xmath116 @xmath146 .
we integrate this equation backward in time using a 4th - order runge - kutta method to get the accretion tracks of the halos in our simulations . since we are interested in bh binary formation , we loosely associate @xmath143 with the halo mass ratio such that tidal stripping of the satellite would prevent the eventual merging of the two nuclei . n - body simulations of galaxy mergers place this mass ratio in the range @xmath143 @xmath0 @xmath151 , depending on the density and orbital parameters of the satellite @xcite . hence our canonical choice is @xmath143 @xmath116 0.3 , and we also try values of @xmath152 ( runs d1 ) and 0.5 ( runs d5 ) , the latter being the halo mass that corresponds to a stellar mass ratio of @xmath0 3:1 in our prescription . following @xcite , we write the probability , per unit time , of a halo with mass @xmath115 at time @xmath118 arising from a merger with a smaller halo of mass between @xmath120 and @xmath153 ( the `` capture rate '' ) as @xmath154 \hspace{2pt } \frac{n(m , t)}{n(m_{f},t ) } \hspace{2pt } dm,\end{gathered}\ ] ] the eps merger rate excluding halos below the threshold @xmath143 , and weighted by the number of mass @xmath120 halos per unit halo of mass @xmath115 . the rate at which halos of mass @xmath115 form through all mergers at time @xmath118 is @xmath155 the probability distribution function ( pdf ) of formation times of halos with mass @xmath146 at time @xmath147 is @xmath156 \hspace{2pt } e^{-\int_{t}^{t_{0 } } r_{f}[m(t'),t ' ] dt'}.\ ] ] given a formation time @xmath157 and corresponding mass @xmath158 along the past accretion track of @xmath146 , the mass of the larger progenitor @xmath114 is distributed according to @xmath159 = \frac{2g(m_{1},m)}{\int_{m\delta_{m}/(1+\delta_{m})}^{m/(1+\delta_{m } ) } g(m',m ) dm'},\ ] ] where @xmath160^{-3/2}.\ ] ] by choosing formation times and progenitor masses randomly according to and , we capture the stochasticity of the intervals between mergers above @xmath143 , but treat merging below this threshold only in the mean .
see @xcite , @xcite , and @xcite for further details and derivations of and .
3 shows the distribution of formation times , progenitor masses , and accretion tracks for a present - day @xmath161 m@xmath23 halo for @xmath162 0.1 , 0.3 , and 0.5 , and the accretion tracks for 1 , 2.3 , and 5 @xmath163 m@xmath23 halos with @xmath143 fixed at 0.3 .
all accretion tracks are normalized to the present - day mass @xmath146 .
note the insensitivity of the shape of these tracks to @xmath146 , as expected for masses above the critical mass @xmath164 .
halo at @xmath165 .
_ upper right : _ pdf of masses of the larger progenitor of the same halo given by equation , normalized to the mass of the merger product , @xmath166 .
_ lower left : _ past accretion tracks of a present - day @xmath167 halo back to @xmath168 , normalized to the mass at @xmath165 . _
lower right : _ normalized accretion tracks for three different halo masses . ]
our algorithm for generating the bh masses is illustrated schematically in fig .
4 . for each run
we begin with a halo of mass @xmath146 @xmath116 @xmath161 @xmath169 at time @xmath147 @xmath142 @xmath170 , choose it s formation time @xmath171 randomly according to equation , and find the mass @xmath172 along its accretion track at that time .
the mass @xmath173 is assigned to the dark matter halo hosting the triple bh system , and the physical time for the run to end if other termination conditions are not met first is set to @xmath174 . to explore the dependence of the results on the absolute mass scale
, we also try beginning with a @xmath175 @xmath169 halo ( runs h1 ) .
we model the halo as a hernquist profile @xcite , which is identical to an nfw profile @xcite in its inner regions if the scale radius @xmath176 is related to the nfw scale radius by @xmath176 @xmath116 @xmath177}$ ] , where @xmath178 is the halo concentration defined by @xmath179 .
the hernquist model falls off as @xmath180 instead of @xmath181 far outside @xmath176 @xcite .
the virial radius @xmath182 is given by @xmath183^{1/3 } h^{-1 } { \rm kpc},\end{gathered}\ ] ] where @xmath184 @xmath116 @xmath185 - 39[\omega_{m}(z)-1]^{2}$ ] @xcite , and @xmath178 roughly follows the median relation from the @xmath186cdm simulations of @xcite , @xmath187^{0.13}/(1+z)$ ] .
the @xmath144 dependence of @xmath188 and @xmath178 nearly cancel to make @xmath176 depend only weakly on redshift , so we simply use the @xmath165 relation between @xmath189 and @xmath176 in our simulations .
we choose the mass @xmath114 of the larger progenitor of @xmath173 randomly according to equation , and assign a mass @xmath113 @xmath116 @xmath190 to the smaller progenitor . before the merger the larger progenitor is assumed to have hosted a bh binary , while the smaller one hosted a single bh . repeating the procedure used for @xmath146
, we assign formation times @xmath191 and @xmath192 to @xmath114 and @xmath113 using equation , and choose progenitor masses @xmath193 , @xmath194 , @xmath195 , and @xmath196 according to equation . having constructed a set of progenitor halos , we now need a bh - halo relation @xmath197 to complete our algorithm .
we obtain such a relation by equating the halo virial velocity @xmath198 to the circular velocity @xmath199 of the stellar spheroid , and using empirical @xmath200 and @xmath201 correlations to connect @xmath199 to @xmath202 , similar to the approaches in @xcite and @xcite . combining @xmath203^{1/6 } { \rm km \hspace{3pt } s^{-1 } } \end{gathered}\ ] ] @xcite with @xmath199 @xmath204 @xmath205^{0.84}$ ] @xmath102 @xcite and @xmath206 @xmath204 @xmath207^{1/4.02}$ ] @xcite , we arrive at the relation @xmath208 where @xmath209^{-1/2}$ ] . in our canonical runs we set the masses of the inner binary members to @xmath210 and @xmath211 , and that of the intruding bh to @xmath212 .
note that in this prescription the intruder is usually lighter than the heavier binary member , so that most of the 3-body interactions result in an exchange . to examine the effect of more interactions without exchange ,
we try choosing @xmath213 $ ] for the intruder mass in runs ( mx ) . as there is neither a direct causal relationship between @xmath202 and @xmath189 predicted by theory @xcite nor a tight correlation directly observed between these two variables , and
we know that identical halos may host galaxies of different morphologies and occupation numbers , @xmath197 should be taken with something of a grain of salt .
nevertheless it is a useful way to generate simple but physically - motivated bh mass distibutions when no information other than the halo mass is available . for the halo @xmath146 hosting the bh triple system randomly from equation .
given @xmath214 and @xmath215 , we select two progenitor masses @xmath114 and @xmath113 according to equation , assign the binary to the larger one and the third bh to the smaller one .
we repeat this process going back one step further in the `` merger tree '' to get the masses of the binary constituents . ]
we make one final modification to the set of bh masses used in our simulations .
if the outer binary s hardening radius lies outside the stability boundary given by equation with @xmath216 , then the decay of the outer orbit is expected to stall before a strong encounter can begin . to roughly account for this
we exclude all ics where @xmath217 @xmath218 @xmath219 .
the final distribution of bh mass ratios is shown in fig . 5 for @xmath1620.1 , 0.3 , and 0.5 . .
_ lower panel : _
@xmath220 , where @xmath221 is the mass of the lightest bh and @xmath222 is the sum of the masses of the other two bhs . ] in the upper panel we plot the inner binary mass ratios , while the lower panel shows the distribution of @xmath220 , where @xmath221 is the mass of the lightest bh and @xmath222 is the sum of the masses of the other two bhs .
this ratio determines the binary recoil speed when the lightest bh is ejected from the system .
the total bh mass is typically @xmath24 m@xmath23 in our canonical runs . to complete the galactic model
we surround the bh system by a smooth stellar potential superimposed on the dark matter halo .
the stars are modelled using the `` @xmath223-models '' of @xcite , with a sharp break to shallower slope @xmath224 added at @xmath225 : @xmath226 our canonical model is the @xmath227 ( hernquist ) profile , and we also try @xmath228 ( runs sc ) to explore the effect of a steeper inner profile and higher central density ( @xmath229 for @xmath228 vs. @xmath230 for the hernquist profile at the bh radius of influence ) . @xmath231 and @xmath232 were initialized to reflect the cusp destruction caused by the inspiralling bhs in reaching their initial configuration , and were updated throughout the simulation to account for the continued core heating and mass ejection .
our algorithm for updating the core is described further in 3.5 .
the parameters @xmath120 and @xmath72 in the @xmath223-models were set based on the tight correlations observed between smbh mass @xmath202 and stellar bulge mass @xcite and velocity dispersion @xcite .
@xcite found the relation @xmath233 between @xmath202 and the virial mass @xmath234 of the stellar bulge , where @xmath235 is the half - light radius and @xmath236 is the effective bulge velocity dispersion .
they set @xmath237 ( @xmath238 would be 8/3 for an isothermal sphere ) to get an average ratio of unity between @xmath239 and the dynamically measured masses @xmath240 of galaxies with more direct stellar - dynamical mass determinations @xcite .
@xmath236 is typically measured over either a circular aperture of radius @xmath241 @xcite or a linear aperture out to @xmath235 @xcite
@xcite discuss the essential agreement between the velocity dispersions measured in these two ways .
thus for each model ( @xmath242 ) we compute the projected radius @xmath243 containing half the integrated surface brightness ( assuming a constant mass - to - light ratio ) , and velocity dispersion @xmath244 at radius @xmath241 . here
@xmath245 and @xmath246 are constants depending on the density profile ( see @xcite for relevant formulae ) .
the parameter @xmath120 is then chosen to satisfy @xmath247
@xmath248 @xmath249 . for the hernquist model with @xmath250 and @xmath251 , @xmath252 .
the scale radius is then obtained from @xmath253 , where @xmath254 is the velocity dispersion computed from the @xmath110 relation of @xcite . in each of these relations
@xmath202 is set to the total mass of the triple bh system . in a perfectly smooth , spherically symmetric galactic potential , bhs ejected on distant radial orbits return directly to the center to interact strongly with any other nuclear black holes .
since real galaxies are clumpy and triaxial , the interaction will more realistically be delayed until the orbit of the ejected bh decays by dynamical friction . to mitigate this problem we flattened the @xmath223-models by adding two low - order spherical harmonic terms to the spherical potential @xcite : @xmath255 where @xmath256 is the potential of the spherical @xmath223-model , @xmath257 , and @xmath258 . since near - sphericity in the inner regions
is probably a necessary prerequisite for the survival of the inner binary for of order a hubble time until the next merger @xcite , the parameters @xmath259 were chosen to give a spherical profile near the galactic center , and axis ratios approaching 1.3 and 1.5 for @xmath260 .
a similar triaxial modification was applied to the dark matter halo , and the relative orientation of the stellar and halo potentials was chosen randomly . by misaligning their axes
we eliminate any artificial stable orbits ( e.g. along the long axis of an ellipsoid ) near which ejected bhs tend to return on a perfectly radial orbit to the center .
this triaxial modification had the desired effect of preventing frequent strong encounters at periapsis on distant orbits , but had little influence on the global outcome statistics .
we assume that the 3-bh system starts off as a hierarchical triple on the verge of unstable 3-body interactions .
in our canonical runs we initialize the inner binary semi - major axis @xmath64 to @xmath14 . to study the effect of varying @xmath64 we also try runs with @xmath261 ( runs ba ) and @xmath262 ( runs sa ) . the outer binary semi - major axis @xmath65 is set by the stability criterion of @xcite , equation
the initial eccentricity of the inner ( outer ) binary was chosen uniformly between 0.0 and 0.2 ( 0.3 ) , in accordance with the low eccentricities found in galaxy merger simulations where dynamical friction tends to circularize the orbits of satellites as they spiral inward @xcite .
the three euler angles of the intruder s orbital plane were chosen randomly relative to the reference plane of the binary orbit , as was the phase of the initial periapsis of the binary .
both orbits were always started at periapsis ; since many orbital periods elapse before unstable interactions begin , the relative phase is effectively randomized in any case . having defined an initial configuration of three bhs embedded in a stellar+dark matter potential , we next describe how we evolve the system forward in time .
we treat the close 3-body encounters using sverre aarseth s _ chain _ code , an implementation of the n - body regularization technique of mikkola and aarseth @xcite .
the masses are first ordered so that neighbors in the chain are the dominant two - body interactions , then the ks - transformation @xcite is applied to neighboring pairs .
this transformation eliminates the singularity at @xmath263 in newtonian gravity and transforms the equations of keplerian motion to the simple harmonic oscillator equation @xcite .
external perturbing forces of arbitrary strength depending on the coordinates , velocities , and/or time are simply incorporated into the formulation ( though of course singularities in these perturbing forces need not be eliminated by the change of variables ) .
we use this to add a galactic potential ( 2.1 - 2 ) , a gravitational radiation back - reaction force , and a stellar - dynamical friction force on the intruding bh .
the regularized equations of motion are integrated using the bulirsch - stoer ( bs ) method @xcite based on romberg extrapolation . for unperturbed sinusoidal motion , the bs integrator requires only two or three timesteps per orbital period ! when the binary and third body are far apart we switch to two - body motion ( of the single bh and binary com ) using a 4th - order runge - kutta ( rk4 ) method .
we simultaneously evolve the binary semi - major axis and eccentricity using orbit averaged equations , @xmath264 dt$ ] and @xmath265 dt$ ] , where @xmath266 and @xmath267 are the contributions from stellar interactions and gravitational radiation .
the timesteps are adaptively controlled with a simple step - doubling scheme : at each step the 14 numbers @xmath268 are all required to remain the same to within an error @xmath269 @xmath116 @xmath270 under doubling of the step size . to avoid wasting computation time
when any of these values approach zero , we accept agreement to @xmath271 decimal places as an alternative criterion for convergence . for the calculations reported in this paper we set @xmath272 .
the relative perturbation to the binary from the third body at apoapsis , @xmath273 is used to decide which integration method to use at any given time .
here @xmath274 is the apoapsis distance between @xmath275 and @xmath276 , @xmath86 is the intruder mass , and @xmath277 is the distance of the intruder from the binary com .
we switch to two - body rk4 integration each time @xmath278 falls below @xmath279 and call the _ chain _ code again when @xmath278 reaches @xmath280 .
we choose different @xmath278 thresholds for beginning and ending close encounters to prevent overly frequent toggling between the two methods .
when @xmath281 bhs remain in the simulation ( after coalescence of the inner binary or escape of one or more bhs from the galaxy ) , we primarily use the rk4 integrator , but call the _ chain _ code to treat very close two - body encounters . since chain regularization is defined only for three or more bodies , we add a light and distant `` dummy '' particle when using this method for two - body motion . during the two - body motion we declare the single bh or binary ( remnant ) to have escaped if its distance from the galactic center exceeds 500 kpc and its specific energy @xmath282 exceeds @xmath283 , the energy needed to escape from @xmath284 to infinity .
we declare the binary to have coalesced during a close encounter when ( i ) @xmath285 , where @xmath286 is the schwarzchild radius of the larger member of the pair , or ( ii ) @xmath287 @xmath288 @xmath289 and @xmath287 @xmath288 50 yrs while @xmath290 , where @xmath291 is the current outer binary dynamical time . during the rk4 integration we require that @xmath287 @xmath288 50 yrs or @xmath292 at periapsis , where @xmath293 are the schwarzchild radii of the two binary members . upon coalescence
we replace the pair with a single body of mass @xmath222 and the com position and velocity .
a run ends when ( a ) only one smbh remains in the galaxy and it has settled to the center of the potential by dynamical friction ; ( b ) two bhs remain and have formed a hard binary at the galactic center ; ( c ) the physical time exceeds @xmath294 , the current age of the universe minus the halo formation time ; ( d ) all bhs have escaped the galaxy ; or ( e ) the physical time spent in a call to _ chain _ exceeds a maximum allowed time @xmath295 .
the last condition is added to avoid spending too much computation time on very long close encounters . during the two - body evolution
we apply a dynamical friction force given by chandrasekhar s formula @xcite , @xmath296}{v^{2 } } \hat{v},\ ] ] where @xmath297 , to the single bh and binary com .
the factor in square brackets @xmath204 @xmath298 for @xmath299 and @xmath204 @xmath300 for @xmath301 .
we take @xmath302 , 1 \right\}\ ] ] for the coulomb logarithm , where @xmath38 is the bh s distance from the galactic center . for @xmath303 in equation
we use @xmath304 $ ] , effectively capping the density at its value at the bh radius of influence , @xmath305 , when the bhs pass through the core .
the semi - major axis @xmath72 of the binary also evolves under stellar - dynamical friction as it wanders through the galaxy .
however chandrasekhar s formula applied separately to the binary constituents does not give a good description of this evolution , since the hard binary loses energy through close 3-body encounters with stars , while equation relies on the assumption that the energy loss is dominated by weak two - body encounters .
we approximate the evolution of @xmath72 using a formulation for the decay rate of a hard , massive binary in a uniform and isotropic sea of stars developed in @xcite and @xcite .
the formulation was calibrated with an extensive series of 3-body scattering experiments in @xcite and tested against n - body simulations in @xcite .
the binary decay rate is given by @xmath306 where the hardening rate @xmath307 can be approximated by the empirical fitting function @xcite @xmath308^{1/2}}.\ ] ] here @xmath309 is the characteristic velocity distinguishing the hard binary regime
stars with @xmath310 can not be easily captured into bound orbits and preferentially harden the binary in close encounters . in our simulations
the binary com is often speeding through the stellar medium at @xmath311 after an energetic ejection , so the stellar medium looks `` hotter '' in its frame of reference . to account for this
we replace @xmath9 in equations and with @xmath312 , a good approximation since @xmath307 is not very sensitive to the shape of the distribution function ( e.g. @xmath313 for a maxwellian vs. @xmath314 for a uniform velocity distribution ) . for @xmath303 in equation
we took @xmath304 $ ] as we did for the drag on the com .
we ignored the mild eccentricity evolution @xmath315 , which is shown in @xcite to be far weaker than that predicted by chandrasekhar s formula for hard eccentric binaries .
when the amplitude of oscillation of one of the two masses falls below @xmath316 , we stop integrating its motion and place it at rest at the galactic center until the second body returns to within a distance of twice the break radius , @xmath317 .
if the settled mass is the binary , then we also stop updating its semi - major axis for stellar hardening , assuming that it clears out its loss cone and stalls once it stops moving about the nucleus and encountering new stars . since the total mass in loss cone stars is small compared to the bh mass in the low - density galaxies that we consider , to good approximation the binary stalls as soon as the replenishing mechanism ( motion ) shuts off . during close encounters between the three bhs an orbit - averaged prescription for stellar - dynamical friction
is not feasible .
however the triples are still marginally stable at the boundary given by equation , so we apply a drag force given by chandrasekhar s formula with @xmath318 to the intruder at the beginning of each run . at the onset of chaotic interactions in the first encounter ( defined loosely by the first time the closest pair is not formed by the original binary members ) this perturbation
is shut off , and it remains off in all later close encounters .
fortunately the chaotic interactions occur on timescales very short compared to a dynamical friction time , so it is valid to neglect stellar dissipation during close encounters .
gravitational radiation is modelled using the @xmath319 $ ] post - newtonian ( 2.5pn ) back - reaction acceleration computed by @xcite , evaluated in the two - body com frame ( e.g. gultekin et al .
2006 ) , @xmath320 } + \vec{v } { \big [ } -\frac{6g(m_{1}+m_{2})}{r}-2v^{2 } { \big ] } { \big \}}.\end{gathered}\ ] ] @xmath321 and @xmath322 are the relative positions and velocities of the two masses .
we sum the force linearly over all pairs , a valid approximation provided the perturbations from the third body and other external tidal forces are instantaneously small at periapsis .
when averaged over a complete orbit , equation is equivalent to the @xcite equations for the binary semi - major axis and eccentricity , [ pet ] @xmath323 however when @xmath324 comes close to one on hyperbolic orbits , so that @xmath325 , @xmath326 as given by equation becomes _ positive _ , though we know physically that gravitational waves can only carry energy away from the system . to give the correct answer averaged over an orbit ,
this positive contribution must be cancelled by extra energy loss near periapsis , making the equation potentially sensitive to numerical error .
this effect is much less pronounced in the @xcite form than in other expressions derived for the radiation back - reaction acceleration - they derived the formula specifically for practical use on the problem of two point masses ( see appendix of @xcite and references therein ) .
for computational ease we neglect the lower - order 1 - 2pn terms ( precession of the periapsis ) in the post - newtonian expansion . though much larger in magnitude than the radiation reaction force , these terms are unimportant in the statistical sense because they conserve the intrinsic properties of the system , such as energy @xcite .
we need not concern ourselves with relativistic precession destroying the kozai resonance since the semi - major axis ratio given by equation is much smaller than that of equation .
one way to establish the reliability of our integration methods is to test them on problems with known solutions .
6 shows an example on the two - body problem with gravitational radiation .
@xmath116 @xmath21 yrs at the beginning of each integration .
_ left : _ semi - major axis evolution for initial eccentricities of @xmath327 @xmath116 0.0 , 0.5 , 0.9 , and 0.99 ( bottom to top ) .
_ right : _ eccentricity evolution for @xmath327 @xmath116 0.5 , 0.9 , and 0.99 .
the curves indistinguishability demonstrates the reliability of all three methods . _
lower panels : _ hyperbolic orbits with impact parameters @xmath328 set to 80% and 120% of the critical value for gravitational radiation capture , computed using the dd acceleration in the _ chain _ code .
_ left : _ 0.8@xmath329 ; bh is captured .
_ right : _
1.2@xmath329 ; bh is not captured .
the blue asterisks are points along the newtonian trajectory ( without gravitational radiation ) .
the deviation from the newtonian trajectory after periapsis can be seen in both plots , even though the energy remains positive in the latter . ]
the upper panels show the evolution of the semi - major axis @xmath72 and eccentricity @xmath13 of four decaying elliptical orbits , computed using ( a ) our rk4 integrator and equation , with an error tolerance of @xmath330 , ( b ) the _ chain _ code and equation , with @xmath331 , and ( c ) the @xcite equations . in each case
the initial semi - major axis @xmath332 was chosen to give a gravitational radiation timescale of @xmath333 yrs , and the four curves ( from bottom to top ) are for eccentricities of 0.0 , 0.5 , 0.9 , and 0.99 .
the agreement of the three computation methods demonstrates the reliability of both the rk4 integrator and our implementation of the _ chain _ code in handling dissipative forces .
the lower panels show two hyperbolic orbits with periapsis distances around 30 times the schwarzchild radius @xmath286 of the larger bh , computed using equation in _ chain_. the rk4 integrator was found to fail some tests on very close approaches from hyperbolic orbits with gravitational radiation , so we treat all such approaches using the regularized _ chain _ code in our runs , even during the unperturbed binary evolution .
the blue asterisks are points along the newtonian orbits while the red solid lines show the trajectories with gravitational radiation .
there is a simple analytic expression for the maximum periapsis distance for gravitational radiation capture from a hyperbolic orbit , @xmath334^{2/7},\ ] ] where @xmath275 and @xmath276 are the masses of the two bodies and @xmath335 is their relative velocity at infinity .
the orbit on the lower left begins at 80% of the critical impact parameter and the incoming bh is captured . on the right
the intruding bh starts at 120% of the critical impact parameter and is not captured , though the deviation from the newtonian trajectory due to the energy radiated at periapsis can be seen on the way out .
we tried iterating over impact parameters close to the critical value and found that the code reproduces equation to within a part in @xmath78 for periapsis distances @xmath336 , and to within a part in @xmath337 for @xmath338 .
we also evaluated the performance of the code by repeating our canonical set of 1005 runs with a static inner profile to check the precision of energy conservation . in fig .
7 we histogram the energy errors , computed as @xmath339 - e_{f}}{e_{0}-e_{f } } \right|,\ ] ] where @xmath340 and @xmath341 are the initial and final energies , and the two terms in the sum under the integral are the work done by dynamical friction and gravitational radiation during the current stage of the code . in the upper panel
we separately plot the errors for close 3-body encounters , rk4 integration of the unperturbed binary motion ( `` far '' ) , and close two - body encounters computed with _ chain _ during the unperturbed binary evolution .
the plot includes all code stages where the energy dissipated was at least @xmath342 in code units , or about a part in @xmath343 of the initial binding energy of the system .
the black ( heavy ) histogram shows the errors for close encounters normalized to the _ initial _ energy instead of the dissipated energy in the denominator of equation , since the energy dissipated was very small in many close encounters . in the lower panel
we combine the energy errors from the various code stages to get an effective energy error for each entire run , @xmath344 we had to combine the separate errors to obtain @xmath345 since the galactic potential is handled slightly differently during different stages of the code , e.g. the triaxial modification is applied only during the rk4 integration . in a large majority of cases
@xmath345 falls between @xmath346 and @xmath347 , and energy is conserved to better than a part in @xmath348 in every run .
the excellent energy conservation gives us confidence in the robustness of our integration methods .
we begin with an overview of the outcomes of our 3-body simulations . in subsequent sections we focus on various effects in more detail .
our data consists of eight sets of 1005 runs , each sampling a different distribution of the ics .
a set of 1005 runs took anywhere from @xmath04 to @xmath030 hours to finish on five 2.0 ghz opteron processors , depending on the ics . in our canonical runs ( cn ) , we chose @xmath349 for the threshold merger mass ratio , modelled the stellar bulge as a hernquist ( @xmath350 ) profile , started off the inner binary at @xmath14 , and generated the ics from a @xmath161 m@xmath23 halo at @xmath165 . in each of the remaining runs we varied one of these assumptions .
runs d1 and d5 used @xmath351 and @xmath352 to explore the effects of widening or narrowing the range of bh mass ratios . in runs mx we assigned a mass @xmath213 $ ] instead of @xmath212 to the intruding bh , as discussed in 2.1 . in runs ba and
sa we started off the inner binary at @xmath353 and @xmath354 instead of at @xmath14 .
we initialized the stellar bulge to an @xmath228 profile in runs sc , to explore the effect of a steeper inner cusp .
finally in runs h1 we generated the ics from a @xmath175 m@xmath23 halo at @xmath165 , for total bh masses of @xmath355 m@xmath23 , about an order of magnitude lower than in our canonical runs .
table 1 summarizes the outcomes .
.summary of outcomes for eight different distributions of the ics .
see text for explanation of the entries . [
cols="<,<,<,<,<,<,<,<,<,<",options="header " , ] [ obsmbh ] for the rest of the galaxies we used the @xcite @xmath356 relation to estimate the bh masses in galaxies with @xmath357 , and the @xcite @xmath110 relation for those with @xmath358 . showing both plots gives an idea of how much the mass deficits vary with @xmath202 estimator
. in set cn @xmath011% of the runs resulted in cores with @xmath359 . in some rare runs where the binary was ejected to a large distance and then brought back by dynamical friction
, we obtained even higher mass deficits ( up to @xmath360 ) .
the binary is more efficient at core scouring than the single bh since it is more massive .
if each independent binary inspiral adds @xmath361 to the mass deficit , then the mean @xmath362 enhancement of @xmath361 due to the triple encounters is equivalent to one extra merger in the system s history .
an @xmath362 one standard deviation above the mean is equivalent to two extra mergers .
triple - smbh systems in galactic nuclei produce a range of phenomena and signatures rather different from those expected if no more than two smbhs occupy them at a time .
we have developed an efficient numerical method for following the evolution of 3-body systems in the centers of galaxies , and used it to explore the outcomes of such encounters in massive elliptical galaxies at low redshift .
we find a high efficiency of smbh coalescence due to the encounters , providing a `` last resort '' solution to the final parsec problem .
there is , however , a caveat in extending this result immediately to all bh masses .
if we define @xmath363 to be the binary semi - major axis where escape of one bh first becomes likely ( @xmath364 where @xmath365 is a factor of order 10 for a hernquist profile ) , then since @xmath366 we have @xmath367 if @xmath222 obeys the @xmath110 relation @xmath368 , so @xmath369 .
in other words , at smaller bh masses , the lightest bh is more likely to escape the galaxy before driving the binary to coalescence by gravitational radiation . by focusing on massive galaxies
we have chosen the systems where the binary is _ least _ likely to coalesce by other means ( e.g. gas or massive perturbers ) , and _ most _ likely to coalesce in the next merger with the help of 3-bh interactions .
we may address the efficiency of triple - induced coalescence in much smaller - mass systems in future studies .
we find that close triple encounters can produce a population of high - eccentricity binaries , whose gravitational radiation signal could potentially be observable by lisa .
such signals originate from kozai oscillations in hierarchical triples at high initial inclinations and highly eccentric binaries formed following distant ejections . as the eccentricity increases , the radiation spectrum peaks at progressively higher harmonics of the fundamental frequency , approaching a nearly flat spectrum as @xmath370 @xcite .
a circular @xmath371 m@xmath23 bh binary remains below the band of frequencies ( @xmath372 hz ) detectable by lisa throughout its inspiral , but the occurrence of high - eccentricity coalescences could extend lisa s sensitivity into this mass range , or lengthen the duration of its sensitivity to @xmath373 m@xmath23 events .
a highly eccentric binary produces a `` spiky '' waveform that looks quite different from that of a circular system ( see fig . 7 in @xcite ) .
gravitational radiation `` spikes '' at very close approaches during chaotic 3-body interactions could also produce radiation bursts detectable by lisa .
if triple encounters are indeed limited to massive systems at low redshift , then the importance of these considerations is limited by the expected event rate in this mass range , assuming efficient coalescence .
this rate is highly uncertain , ranging from @xmath374 @xcite to @xmath375 @xcite depending on the merger and bh population model adopted .
if 3-bh systems occur in other contexts , e.g. imbhs in galactic nuclei or star clusters , then the phenomena we have discussed may be observationally relevant even if the high - mass smbh event rate is low .
a detailed look at the gravitational waveforms expected from 3-body encounters and their expected detection rates is an interesting topic for a future study .
the slingshot ejections in triple encounters produce a population of `` wandering '' smbhs in and outside the halos of galaxies . in systems that have undergone several major dry mergers ( e.g. cd galaxy systems )
, one might expect a few such ejected smbhs to be floating in the vicinity .
as of yet , no probable way of observing these wandering bhs has been proposed . in principle one
can imagine a star bound to the ejected smbh entering a giant phase and overflowing its roche lobe , producing some accretion onto the smbh and an observable flare .
single ejections could also in principle affect bh - bulge correlations such as the @xmath110 relation , but since it is the lightest bh that gets ejected this effect would fall well within the observed scatter in the correlations for just one or two ejection events .
triple interactions in galactic nuclei can have a large effect on the expected properties of stable smbh binaries in the local universe . while many models of binary formation predict mostly circular binaries around @xmath14 , 3-body encounters produce binaries at all eccentricities .
they also create a population of stalled binaries at separations significantly smaller than @xmath14 but still larger than @xmath18 , as does any partial gap - crossing mechanism .
better measurements and statistics on the mass deficits in cored elliptical galaxies may provide clues on the history of the nuclear smbh activity in these systems .
triple bh encounters produce a highly scattered distribution of core sizes , with mass deficits up to @xmath376 higher than expected for successive binary coalescences .
the apparent peak at mass deficits of @xmath377 times the nuclear bh mass in observed cores may very tentatively hint that multiple - bh encounters are not the norm in these systems .
this signature of binary or multiple - bh activity is appealling because ( a ) its duty cycle is the lifetime of the galaxy ; ( b ) it is present whether binary pairs stall or coalesce ; and ( c ) it can be observed even in the complete absence of radiative activity , such as disk accretion or jet production .
however the interpretation of galaxy cores is complicated by multiple mergers , the possibility of partial stellar cusp regeneration from traces of cold gas , and observational complications such as projection effects in nonspherical galaxies and optimizing the fitting / extrapolation algorithm to best represent the mass deficit .
there is a need for theoretical studies on the cores produced by smbh mergers in triaxial galaxies , since triaxility seems to be the most likely candidate for a gap - crossing mechanism in dry mergers between gas - poor , giant ellipticals .
inferring the nuclear histories of galaxies from their observed core properties will likely be a topic of much interest in the future .
we would like to thank suvendra dutta for technical help at the parallel computing center of the institute for theory and computation ( itc ) , and sverre aarseth and seppo mikkola for making their n - body algorithms and codes available .
we are also grateful to scott hughes , michael eracleous , marta volonteri , and fred rasio for useful discussions .
this research was supported in part by an fqxi grant and harvard university funds . | galaxies with stellar bulges are generically observed to host supermassive black holes ( smbhs ) .
the hierarchical merging of galaxies should therefore lead to the formation of smbh binaries .
merging of old massive galaxies with little gas promotes the formation of low - density nuclei where smbh binaries are expected to survive over long times .
if the binary lifetime exceeds the typical time between mergers , then triple black hole ( bh ) systems may form .
we study the statistics of close triple - smbh encounters in galactic nuclei by computing a series of 3-body orbits with physically - motivated initial conditions appropriate for giant elliptical galaxies .
our simulations include a smooth background potential consisting of a stellar bulge plus a dark matter halo , drag forces due to gravitational radiation and dynamical friction on the stars and dark matter , and a simple model of the time evolution of the inner density profile under heating and mass ejection by the smbhs .
we find that the binary pair coalesces as a result of repeated close encounters in @xmath085% of our runs , and in @xmath015% of cases a new eccentric binary forms from the third smbh and binary remnant and coalesces during the run time . in about 40% of the runs the lightest bh
is left wandering through the galactic halo or escapes the galaxy altogether , but escape of all three smbhs is exceedingly rare .
the triple systems typically scour out cores with mass deficits @xmath01 - 2 @xmath1 their total mass , which can help to account for the large cores observed in some massive elliptical galaxies , such as m87 .
the high coalescence rate , prevalence of very high - eccentricity orbits , and gravitational radiation `` spikes '' during close encounters in our runs , may provide interesting signals for the future laser interferometer space antenna ( lisa ) .
[ firstpage ] black hole physics cosmology : theory galaxies : elliptical and lenticular , cd galaxies : interactions galaxies : nuclei methods : numerical |
the present eps database consists of a rather heterogeneous sample of planets , showing great variety in all the measured quantities .
statistical analysis provides a necessary means to find correlations among various physical parameters involved in planetary formation and evolution .
nevertheless , we caution that statistical analysis may not be able to disclose important relationships due to the complex and -mostly unknown- interplay of the involved parameters . in order to overcome such problems
, we performed a global statistical analysis of the eps , using a novel approach .
the underlying idea is to find groups of _ similar _ eps and to distinguish different ep groups on the basis of their _ diversity_. in this work , the concept of similarity and diversity among eps is quantified by means of a distance measure in the multifold space of physical parameters .
this goal has been achieved with the aid of principal component analysis and hierarchical cluster analysis . in the present analysis we followed the procedure used by @xcite , and updated in @xcite .
the database used in this paper is updated to july 9@xmath0 , 2008 .
we restricted our analysis to those eps having measurements for five _ input variables _ , that is : planetary mass ( @xmath1 ) , semimajor axis ( @xmath2 ) , eccentricity ( @xmath3 ) , stellar mass ( @xmath4 ) and metallicity ( [ fe / h ] ) .
of 308 eps ( including jupiter ) , 252 were finally selected for the analysis .
the purpose of our analysis is to identify planets which are similar with respect to the 5-fold space of the input variables . as a result
, 6 robust ep clusters have been identified . before focusing on hps
, we briefly outline the general nature of the clusters .
+ cluster c1 is characterized by sub- to jovian - like @xmath1 , @xmath5 au , and super - solar [ fe / h ] .
cluster c2 has sub - jovian @xmath1 , @xmath5 au , sub - solar @xmath4 and sub - solar [ fe / h ] .
both clusters c1 and c2 also have low mean eccentricity .
cluster c3 is the least populated cluster ( 14 eps ) and probably has no strong significance except the fact that it contains many peculiar eps that for different reasons have been rejected by the other clusters .
they are mostly jovian mass planets , having a remarkably super - solar [ fe / h ] .
cluster c4 has mostly jovian mass planets , a relatively large @xmath2 , orbiting solar mass stars with a widespread [ fe / h ] distribution characterized by sub - solar values ( jupiter belongs to this cluster ) .
cluster c5 has super - jovian mass , @xmath6 au , super - solar @xmath4 and sub - solar [ fe / h ] .
the same holds for cluster c6 , except for its super - solar [ fe / h ] and its higher average eccentricity .
all the input variables have an important role in defining the clusters , in particular @xmath2 and [ fe / h ] . beyond the general traits of the solutions , which will be described in more detail elsewhere , we focus in this paper on the hps defined as those having a period less than 12 days ( see later for further comments on this selection ) . according to the adopted definition ,
69 hps are present in the database .
the main results of the taxonomy is that hps have been split into two main groups , belonging to clusters c1 and c2 .
in addition , there are a few outliers placed in other clusters . one of the most interesting features about the hps , is the shape of their mass distribution ( see fig . [ mp_a_hist ] ) .
it has two peaks .
these peaks are thought to be real , even if the shape of the present distribution is severely affected by observational biases that make the discovery of low mass hps difficult . in the literature ,
the bodies belonging to the lower mass peak are referred to as hot neptunes ( hns ) while those belonging to the higher mass peak as hot jupiters ( hjs ) .
theoretical models also predict a double peaked distribution @xcite , where hns are expected to be much more abundant than hjs . + it is interesting that our taxonomy splits the majority of hps into two different clusters ( except for a few outliers ) .
most of the hps belonging to the peak at @xmath7 are placed in cluster c1 ( fig . [ mp_a_hist ] ) .
hps of the cluster c2 have a broad and flatter mass distribution , and it contains most of the hns . concerning the semimajor axis distribution , similar considerations hold .
hps of cluster c1 are strongly clustered at @xmath8 ( @xmath9 au ) ; while those of cluster c2 have a flatter @xmath2 distribution ( fig .
[ mp_a_hist ] ) .
the two groups of hps also have clearly distinctive traits in terms of properties of stellar metallicity and -to a lesser extent- to stellar mass .
these variables are important because they account for the environment where the planets formed .
figure [ fe_ms_hist ] shows a remarkable result : hps of cluster c1 have super - solar [ fe / h ] , while those of c2 have a sub - solar [ fe / h ] . notice that this result is also valid for the whole of c1 and c2 , and not only for their hps .
a similar , but less pronounced , result also holds for the stellar mass : hps of c1 have mostly super - solar @xmath4 ; while those of c2 have mostly sub - solar @xmath4 ( fig .
[ fe_ms_hist ] ) . + the hps of cluster c1 and c2 also have other distinctive traits .
we found some significant intracluster correlations which hold for one cluster but not for the other .
these are the correlations @xmath10 , @xmath11[fe / h ] , @xmath12 ( see fig . [ ae ] ) .
the semimajor axis of the hps of cluster c1 strongly correlate with @xmath3 , while this correlation is absent for cluster c2 . on the other hand , hps of cluster c2 exhibit a strong correlation of @xmath11[fe / h ] and @xmath12 , while cluster c1 does not ( fig .
these two plots in turn clearly show that hns of cluster c1 are very different from the few belonging to cluster c2 .
the latter have remarkable sub - solar [ fe / h ] and sub - solar @xmath4 . + another important point is why some hps have been placed in clusters other than c1 and c2 .
they are : hd73256b , hd68988b , hat - p-7b and hd118203b ( c3 ) ; xo-4b , hat - p-6b and hd162020b ( c4 ) ; wasp-14b , corot - exo-3b and xo-3b ( c5 ) ; hat - p-2b ( c6 ) .
it is not clear if they are real outliers or if they are misplaced by the clustering algorithm .
however , a close look at their properties shows that the main characteristic of these hps is that they have high @xmath1 and high @xmath4 . among the outliers , the most massive ( @xmath13 ) hps are present , namely hat - p-2b , wasp-14b , corot - exo-3b , xo-3b and hd162020b ( see fig .
[ mp_a_hist ] , left panel ) .
notice that the massive hd41004bb is placed instead in c2 , due to the very low stellar mass .
in the present database , the combination of high @xmath1 and high @xmath4 is rather unusual , and this suggests that these hps may be real outliers .
of course , if more hps having such characteristics are discovered , it is possible that they may be grouped into a third class of hps .
finally , it is also important to note that previous results are robust to changes in the selection of hp periods , at least in the range from 10 d to 30 d.
our present theoretical understanding of the formation of hps is based on two models : planetary migration within a protoplanetary disk @xcite and planet - planet scattering followed by tidal circularization @xcite .
planetary migration has been shown to be an efficient mechanism to produce hps .
migration can occur when planetary embryos are still embedded within the disk ( type i ) , or when they are large enough to open a gap in the disk ( type ii ) . according to the present state of the art
, the migration seems the best candidate for hp formation .
on the other hand , early works on planet - planet scattering showed that the probability for the scattering model to produce hps was very low .
however , @xcite found that the kozai mechanism enhances the probability significantly .
therefore , it is possible that the scattering contributes to the formation of hot planets as well as type ii migration , although the latter may be a main channel
. + type ii migration is more efficient for moderate mass planets since the planets have to be massive enough to clear a gap in the disk , but not too massive to efficiently exchange angular momentum and energy with the disk itself .
previous simulations have shown that the formation of hps via gravitational scattering among planets and the subsequent kozai mechanism combined with tidal dissipation is more likely for dynamically active systems of multiple planets , typically containing three or more gas giants ( e.g. * ? ? ?
the formation of many giant planets is preferred in high dust surface density disks @xcite .
dust surface density scales as @xmath14}m_d$ ] , where @xmath15 is the total disk mass . the latter scales , according to theoretical models , as @xmath16 , where @xmath17 @xcite .
therefore high stellar masses and high - metallicity disks favor the onset of a scattering phase .
it must also be noted that smaller planets tend to be scattered inward during the scattering phase and that the tidal circularization is more efficient for planets with small mass and large physical radius .
the detailed final orbital configuration of the hps may vary according to several parameters . however , there are a number of general features that can be outlined .
+ in the case of type ii migration , the hps final position is close to the location of the disk s inner edge .
this is placed near the corotation radius ( namely the distance where the keplerian period matchs the star spin period ) , where disk material accretes onto the stellar poles following the magnetic field lines .
the spin period for young stars having @xmath18 may vary considerably , from 1 d up to 20 d or more @xcite .
therefore , taking also into account a wide variety of disk parameters , the final location of hps formed by type ii migration tends to be spread , in the range from 0.01 to 0.1 au . concerning the eccentricity , at the end of the migration phase low @xmath3 values
are expected .
recent simulations @xcite suggest that , when hps reach the stellar magnetosphere cavities , they may further evolve to smaller @xmath2 . for moderate - to - high @xmath1 , the @xmath3 also increases . on the other hand ,
the location of hps formed by scattering is determined by the location where the tidal force is effective .
the tidal strength depends on several parameters ( planetary radius , planetary mass , etc ) but the final location tends to be in the range 0.03 - 0.08 au , for typical parameter values .
the eccentricity of the inner planet is excited to values close to unity by close scattering and via the subsequent kozai mechanism .
the resultant small pericenter distance enables the planet s eccentricity and semi - major axis to be decreased by the tidal dissipation and moderate eccentricities can remain in some cases @xcite .
we show this in fig .
[ scatter_ae ] .
the outcome of the simulations vary according to tide efficiency . for the simulations shown here
, we followed the tidal evolution of test planets , according to @xcite , for a time span of @xmath19 yr .
a large number of planetary radii , planetary masses and stellar masses have been considered , choosen at random within the following intervals : @xmath20 , @xmath21 , @xmath22 .
the initial eccentricity is randomly chosen from the distribution obtained by @xcite .
figure [ scatter_ae ] ( left panel ) shows that at the end of the scattering and tidal evolution phases the eccentricity and semimajor axes are correlated ( compared to the observed distribution in fig .
[ ae ] ) , while the right panel shows that the resulting semimajor axis distribution is peaked .
the value of the peak depends on the strength of the tide , and for the values used here it is peaked at @xmath23 au , as observed in cluster c1 ( compare with the observed distribution in fig .
[ mp_a_hist ] ) .
the two main processes of hp formation produce different orbital distributions . on the basis of our taxonomic analysis
, we identify two types of hps , which may retain the footprints of these two different formation processes . in this respect ,
hps of cluster c2 and c1 may have been formed by migration and scattering , respectively .
this scenario is supported by a number of facts .
first of all the @xmath2 distribution and the strong @xmath10 relationship for cluster c1 , which do not hold for cluster c2 .
moreover , also the @xmath1 distribution of the two clusters support this conclusion : higher mass bodies are located in cluster c1 , indicating more massive protoplanetary disks , which in turn is confirmed by high [ fe / h ] and high @xmath4 . in this case , the detected hps are expected to be the least massive for each system , since in a multibody scattering the least massive planets are more effectively pushed inward .
the orbits of the most massive planets are , however , only slightly affected by the scattering phase , therefore they tend to stay close to their formation regions and therefore on relatively large @xmath2 .
although hjs originated via the scattering process are expected to be accompanied by outer companions , the latter would be beyond the detectable limit by present surveys .
we find no significant difference between c1 and c2 about the multiplicity of planets , but we caution that this result is affected by low number statistics .
some jupiter - like mass hps are also present in cluster c2 , but in this case they may be the most massive bodies formed in these systems , given also the moderate - to - low @xmath4 and low [ fe / h ] .
another interesting point is that most of the hns belong to cluster c2 .
+ therefore , if we extrapolate directly the percentage of hps belonging to c1 and c2 into the efficiency of formation of the two processes , we end up with 50% of hps formed via scattering and 30% via migration .
the remaining 20% are outliers and may be formed either way .
these numbers , however , have to be taken with caution , since some degree of mixing between the two clusters is expected . on the other hand , from a theoretical point of view ( see discussion in previous sections ) type ii is expected to be more efficient in producing close - in eps . this fact is not in contradiction with our findings since it is possible that planets migrating inward by type ii may stop before becoming hps
. this would be the case , for instance , if the gas in the disk dissipates before the planet reaches the magnetospheric cavity . in this respect
, it is interesting that many giant eps exist in the range 0.1 - 3 au and that their semimajor axis distribution is well explained by the type ii migration model @xcite .
+ an alternative scenario is that the two groups of hps were formed by the same process , and the cluster analysis splits them on the basis of their diversity in the input variables . in this case
, the most distinctive variable would be the metallicity .
formation in low metallicity , moderate - to - low @xmath4 , environments would have produced the hps of cluster c2 .
on the other hand , high [ fe / h ] and moderate - to - high @xmath4 would have produced the hps of cluster c1 .
the combinations of these diversities would have also produced the observed differences in the @xmath1 and @xmath2 distribution of the two groups .
although this is a possibility , we think the peculiar traits of cluster c1 and c2 clearly shown the influences of the two formation mechanisms . | the extrasolar planets ( eps ) so far detected are very different to the planets in our own solar system .
many of them have jupiter - like masses and close - in orbits ( the so - called hot planets , hps ) , with orbital periods of only a few days . in this paper
, we present a new statistical analysis of the observed eps , focusing on the origin of the hps . among the several hp formation mechanisms proposed so far , the two main formation mechanisms are type ii migration and scattering . in both cases , planets form beyond the so - called snow - line of the protoplanetary disk and then migrate inward due to angular momentum and energy exchange with either the protoplanetary disk or with companion planets .
although theoretical studies produce a range of observed features , no firm correspondence between the observed eps and models has yet been established . in our analysis , by means of principal component analysis and hierarchical cluster analysis , we find convincing indications for the existence of two types of hps , whose parameters reflect physical mechanisms of type ii migration and scattering .
[ firstpage ] planets and satellites : formation planetary systems : formation planetary systems : protoplanetary discs . |
the quantum two - time green function technique has provided a very powerful tool in condensed matter physics for exploring the equilibrium and transport properties of a wide variety of many - body systems . within this framework ,
the equation of motion method ( emm ) and the spectral density method ( sdm ) allow to obtain reliable approximations to treat typically unperturbative problems @xcite . the pioneering introduction of the two - time green functions and the emm in classical statistical mechanics by bogoljubov and sadovnikov @xcite , has opened the concrete possibility to describe classical and quantum systems on the same footing .
nextly , a classical version of the sdm ( csdm ) has been also formulated in extensive @xcite and nonextensive @xcite classical statistical mechanics and applied to classical magnetic chains @xcite .
the last method , which seems to present many advantages with respect to the most conventional emm @xcite , offers a robust instrument for systematic and well tested approximations @xcite to explore the macroscopic properties of classical many - body systems as well .
unfortunately , both the mentioned methods have not received the due consideration in the classical context and further developments and applications are desiderable . along this direction , in this paper we apply the csdm to investigate the thermodynamic properties of a highly nontrivial @xmath7-dimensional classical spin-@xmath1 heisenberg ferromagnet with long - range interactions decaying as @xmath2 ( @xmath5 ) with the distance @xmath8 between spins in the presence of an external magnetic field .
the choice of this model , besides its intrinsic theoretical interest , is motivated also by its relevance in the description of many materials @xcite of experimental and technological importance and , in particular , in the understanding of the critical behavior of magnetic systems when long - range exchange interactions are believed to play an important role . in general , realistic microscopic magnetic models with long - range interactions
are difficult to be studied in a reliable way .
since the exact solution for the spin-@xmath9 quantum heisenberg chain with an inverse - square exchange @xmath10 and without an external field was found independently by haldane @xcite and shastry @xcite , only a limited lot of information has been acquired about the low - temperature properties , the existence of long - range order ( lro ) and the related critical behavior for different values of the decaying exponent @xmath11 and the dimensionality @xmath0 .
recently , the critical properties at finite temperature of @xmath0-dimensional quantum heisenberg models , with interactions of the type here considered ( here often named long - range spin models ) have been studied using microscopic techniques @xcite and monte carlo simulations @xcite .
a summary of the known features sounds as follows .
the one- and two - dimensional long - range quantum spin-@xmath9 heisenberg ferromagnets in absence of an external magnetic field were investigated by nakano and takahashi using the so called modified spin - wave theory @xcite and the schwinger - boson mean - field approximation @xcite .
further information were derived for the @xmath0-dimensional case by means of the emm for the two - time green functions using the tyablikov decoupling procedure @xcite .
monte carlo simulations for the two - dimensional quantum spin-@xmath9 heisenberg model have been also performed for @xmath12 @xcite .
this scenario , has been recently enriched by an extension @xcite of the mermin - wagner theorem @xcite for the existence of ferromagnetic ( fm ) lro at finite temperature in quantum heisenberg and xy models in @xmath13 dimensions with @xmath2- and oscillatory- interactions .
classical long - range spin - s heisenberg fm models have attracted great attention , too .
it has been proved that lro exist in @xmath13 dimension when @xmath3 @xcite and is destroyed at all finite temperatures for @xmath14 @xcite .
similar results were obtained for the spherical model @xcite and the present scenario of the critical properties is largely based on renormalization group calculations for the classical @xmath15-vector model @xcite .
classical long - range antiferromagnetic ( afm ) models have been studied less extensively .
the available rigorous results @xcite suggest orientational disorder at all finite temperatures when @xmath14 , but no theorem exists entailing existence or absence of lro for @xmath3 .
monte carlo simulations have been also performed for both @xmath13-dimensional classical fm ( for @xmath16 @xcite ) and afm ( for @xmath17 and @xmath18 with @xmath19 and @xmath20 respectively @xcite ) heisenberg long - range models .
the results confirm that fm - lro survives at finite temperature provided @xmath3 and allow to conjecture that no afm - lro exists at all finite temperatures for @xmath5 .
spin - wave studies @xcite agree with last conjecture but no definitive statement can be drawn at the present stage . in a quite recent work @xcite the thermodynamics and the critical properties of the classical long - range spin-@xmath1 heisenberg fm chain in the presence of an external field
have been systematically studied using the csdm to lowest order of approximation .
the results are in good agreement with previous available analytical and numerical investigations . as mentioned before , here we use the same method to explore a @xmath0-dimensional heisenberg fm model .
as we shall see , a rich phase diagram appears where all the above mentioned scenario is reproduced and extended .
the paper is organized as follows . in sec .
2 we introduce the model and the moment equations ( me s ) for the spectral density ( sd ) to lowest order of approximation .
the main low - temperature properties are studied in sec . 3 and the existence of lro in different regions of the @xmath21-plane
is also shown in a transparent way .
is devoted to the critical properties and the low - temperature paramagnetic susceptibility of the model . finally , in sec .
5 , some concluding remarks are drawn .
a classical @xmath0-dimensional spin-@xmath1 heisenberg ferromagnet with long - range interactions is described by the hamiltonian : @xmath22 here , @xmath23 is the number of sites of the hypercubic lattice with unitary spacing , @xmath24 are the classical spins , @xmath25 is the external magnetic field and the spin - spin interaction , in view of the thermodynamic limit as @xmath26 , is assumed to be @xmath27 , where @xmath28 and @xmath29 measures the strength of the coupling .
the extreme case @xmath30 corresponds to the standard nearest - neighbor interaction while the mean field approximation is obtained when @xmath31 ( replacing @xmath32 by @xmath33 ) . for this type of interaction , the thermodynamical limit @xmath26
is well defined only for @xmath5 , while for @xmath34 the ground state of the system has an infinite energy per particle as @xmath26 and the conventional statistical mechanics can not be directly applied .
the classical spins model ( [ eq:1model ] ) can be appropriately described by the set of @xmath35 canonical variables @xmath36 where @xmath37 is the angle between the projection of the spin vector @xmath38 in the ( @xmath39)-plane and the @xmath40 axis . for practical calculations we find convenient to introduce the new variables @xmath41 , so that @xmath42
. then , the hamiltonian can be rewritten as @xmath43 involving the fourier components of the spins and the exchange interaction defined by : @xmath44 where @xmath45 denotes a wave vector in the @xmath0-dimensional fourier space .
the sum in eq .
( [ eq:2model ] ) is restricted to the first brillouen zone ( @xmath46 ) of the lattice .
the poisson brackets for the spin fourier components relevant for us , are : @xmath47 in the context of the csdm @xcite , all the thermodynamical properties of the model can be derived using the sd defined by : @xmath48 where the classical dynamical variables depend on time through the conjugate canonical coordinates @xmath49 .
here @xmath50 , @xmath51 is the liouville operator and @xmath52 denotes a canonical ensemble average . according to the spirit of the csdm
, we try to determine it to the lowest order in the form @xcite : @xmath53 where the unknown parameters @xmath54 and @xmath55 are to be calculated solving the first two moment equations ( me s ) : @xmath56 with @xmath57 the magnetization per spin . to close this system
, we should express all the unknown quantities in terms of the sd .
one can show@xcite that the transverse correlation function @xmath58 , appearing on the right hand side of eq .
( [ eq : me2 ] ) , can be exactly expressed in terms of the sd by the relation : @xmath59 then , taking into account the one-@xmath60 ansatz ( [ eq : sd_delta ] ) , we have @xmath61 on the contrary , the longitudinal correlation function @xmath62 can not be exactly related to the spectral density @xmath63 .
so , to close the system of me s ( [ eq : me1]),([eq : me2 ] ) we should solve another moment problem by introducing a new sd .
the simplest way to avoid this difficulty is to resort to the decoupling @xmath64 , which means to neglect the correlations between the fourier components of the @xmath65 .
of course , the approximation is appropriate to describe thermodynamic regimes with finite magnetization as under near saturation conditions . in sec .
4 we will introduce a different procedure which allows us to describe also , in a reliable way , regimes with zero or near zero magnetization . as final step one must express the magnetization @xmath66 in terms of the @xmath63 .
this is not a simple problem for a classical spin system , also when @xmath67 ( for which the exact relation @xmath68 exists in the quantum counterpart ) .
however , as shown in refs . @xcite within the spirit of the sdm , one can use the expression : @xmath69 which is appropriate for all thermodynamic regimes and reduces , correctly , to the near saturation relation @xmath70 arising from the identity @xmath71 with @xmath72 . then , from eq .
( [ eq : s+s- ] ) , as a consequence of eq .
( [ eq : s+s-_sd ] ) and the ansatz ( [ eq : sd_delta ] ) , eq .
( [ eq : m2 ] ) yields : @xmath73 with the above ingredients , eqs .
( [ eq : me1])-([eq : me2 ] ) become a closed system to be solved self - consistently . by introducing conveniently the dimensionless variables @xmath74 , @xmath75 , @xmath76 and @xmath77 , the me s reduce to : @xmath78 with @xmath79 in the thermodynamic limit @xmath26
one must replace the sum @xmath80 by an integral @xmath81 in the @xmath0-dimensional @xmath45-space and hence eqs.([eq : se1 ] ) and ( [ eq : sigma1 ] ) can be rewritten as : @xmath82 with @xmath83 the solution of the problem is complicated and one must consider asymptotic regimes for obtaining explicit results or use numerical calculations .
moreover , we can note that these equations have physical meaning only if @xmath84 @xcite .
hence , they can not describe the critical behavior of the system or a paramagnetic phase in zero external field . as we will see , in sec .
3 this problem can be simply overcome with an appropriate modification of eq .
( [ eq : r(k ) ] ) .
we first examine analytically the low temperature solution of eqs .
( [ eq : se_r])-([eq : r(k ) ] ) .
the expressions of @xmath85 and @xmath86 to the first order in the reduced temperature @xmath87 , which allow us to capture some relevant aspects of the low - temperature physics of the model ( [ eq:1model ] ) , are given by : @xmath88 where @xmath89 at first , we assume @xmath90 so that no convergency problem for the above integrals in the @xmath45-space occurs .
to calculate analytically these integrals , one needs an explicit expression of the function @xmath91 which , unfortunately , can not be obtained in terms of elementary functions in the whole @xmath46 , for arbitrary values of the parameter @xmath5 .
nevertheless , for sufficiently small values of the external magnetic field , the dominant contribution to the integrals in eqs.([eq : omg_lowt ] ) and ( [ eq : sigma_lowt ] ) arises from the low wave - vector excitations as close to the critical temperature ( see sec . 4 ) .
then , we can obtain an explicit estimate of @xmath85 and @xmath92 assuming the dominant behavior of @xmath91 in the @xmath46 as @xmath93 , provided that the coefficients of @xmath94 in eqs .
( [ eq : omg_lowt ] ) and ( [ eq : sigma_lowt ] ) remain finite . keeping this in mind
, one can show @xcite that for @xmath5 we have for @xmath95 the low-@xmath96 expansions : @xmath97 the explicit expressions of the coefficients @xmath98 , @xmath99 and @xmath100 , which depend in a cumbersome way on the dimensionality @xmath0 , the exponent @xmath11 and a wave - vector cut - off @xmath101 related to the geometrical definition of @xmath46 , are inessential at this stage and will be omitted .
however , for case @xmath20 , they will be explicitly given in sec . 4 where a comparison with some analytical and monte carlo predictions is performed .
taking into account eq .
( [ eq : omega_lowk ] ) , the integral @xmath102 can be explicitely estimated and we have the following low - temperature representation for the reduced magnetization @xmath103 ^{-1 } , & p = d+2 \\
\frac{1}{\overline{h}}\text { } _ { 2}f_{1}\left ( 1,\frac{d}{2},1+\frac{d}{2};-% \frac{b_{d}\lambda ^{2}}{\overline{h}}\right ) , & p >
d+2 \end{array } \right . \label{eq : sigma_explicit}\ ] ] where@xmath104 is the hypergeometric function , @xmath105 and @xmath106 is the gamma function .
an analogous calculation of @xmath107 in the expression ( [ eq : omg_lowt ] ) for @xmath108 is rather complicated .
however , an explicit estimate of @xmath108 based on the expansions ( [ eq : omega_lowk ] ) is irrelevant for next developments .
the low - temperature susceptibility @xmath109 can be now easily obtained from eq .
( [ eq : sigma_explicit ] ) by derivation with respect to the reduced magnetic field @xmath110 . for the reduced susceptibility @xmath111 we have : @xmath112 ^{-2 } ,
& p = d+2 \\ \frac{1}{\overline{h}^{2}}\text { } _ { 2}f_{1}\left ( 2,\frac{d}{2},1+\frac{d}{2}% , -\frac{b_{d}\lambda
^{2}}{\overline{h}}\right ) , & p > d+2 .
\end{array } \right . \label{eq : chi}\ ] ] the above low-@xmath113 expressions have a physical meaning for @xmath94 and @xmath110 in the near saturation regime ( @xmath114 ) and also for @xmath115 when long range order occurs .
as mentioned before , the integrals in eqs .
( [ eq : omg_lowt ] ) and ( [ eq : sigma_lowt ] ) , and hence the functions in eqs .
( [ eq : sigma_explicit])-([eq : chi ] ) , could diverge in the limit @xmath115 for particular values of the exponent @xmath116 and the dimensionality @xmath0 of the lattice . when this is not the case , from eq .
( [ eq : sigma_lowt ] ) a spontaneous magnetization at a finite temperature should arise signaling the occurrence of lro . of course , the low-@xmath45 behaviors ( [ eq : omega_lowk ] ) determine the constraints for convergency of the integrals involved in the low-@xmath113 expressions when @xmath115 .
it is easy to see that , in the limit @xmath115 , the integral ( [ eq : funci(h ) ] ) converges only for @xmath3 with @xmath4 and for @xmath5 with @xmath6 .
then , for these values of @xmath11 and @xmath0 , a spontaneous magnetization @xmath117 exists and hence lro occurs at small but finite temperature , with the result : @xmath118 where @xmath119 is a finite quantity whose estimate can be immediately obtained from eq .
( [ eq : sigma_explicit ] ) . on the contrary , for @xmath120 with @xmath121 , the integral ( [ eq : funci(h ) ] ) diverges as @xmath115 and no finite solution for @xmath85 exists at @xmath122 .
this means that , for these values of @xmath116 and @xmath7 , no lro occurs at finite temperature .
the global situation is schematically shown in fig . 1 . here
we also distinguished the domains where , as we will show in the next sec .
4 , the system exhibits a critical behavior like for a heisenberg model with a nearest - neighbor exchange coupling ( short - range interaction ( sri ) regime ) and the long - range nature of interactions ( lri ) becomes effective . in conclusion
, our explicit low - temperature results suggest that a transition to a fm phase at finite temperature occurs in the domains of the ( @xmath123)-plane where lro exists . in the remaining domains a different scenario takes place with absence of a phase transition .
these predictions are quite consistent with the recent extension @xcite of the known mermin - wagner theorem @xcite to quantum spin models with lri s of the type here considered .
other thermodynamic properties can be also obtained within the framework of csdm when the solutions of the me s are known .
indeed , for our spin model , one can easily show that the internal energy @xmath124 and free energy @xmath125 per spin are given by @xcite : @xmath126 \text { , } \label{eq : u } \\
f(t , h ) & = f_{0}+\frac{1}{n}\int_{0}^{j}\frac{dj^{\prime } } { j^{\prime } } % \left\langle h_{i}\right\rangle \left ( j^{\prime } \right ) \nonumber \\ & = f_{0}+\frac{1}{2n^{2}}\sum_{\mathbf{k}}\frac{j\left ( \mathbf{k}\right ) } { j% } \int_{0}^{j}dj^{\prime } \left [ \left\langle s_{\mathbf{k}}^{+}s_{-\mathbf{k}% } ^{-}\right\rangle _ { j^{\prime } } + \left\langle s_{\mathbf{k}}^{z}s_{-\mathbf{% k}}^{z}\right\rangle _ { j^{\prime } } \right ] \text{. } \label{eq : f}\end{aligned}\ ] ] here , @xmath127 is the interaction part of the spin hamiltonian , @xmath128 is the free energy per spin for a magnetic model without interactions and @xmath129 denotes a canonical average as a function of the interaction strength @xmath130 .
as we see , within our approximations all the quantities in eqs .
( [ eq : u ] ) and ( [ eq : f ] ) can be expressed in terms of the transverse sd @xmath63 and hence all the relevant thermodynamic quantities of our classical spin model can be evaluated . in particular , for the reduced internal energy @xmath131 we have : @xmath132 then , in the low temperature limit , from ( [ eq : sigma_explicit ] ) we find : @xmath133 where the explicit expression of the term @xmath134 can be obtained in a straightforward but tedious way .
hence , for the reduced specific heat @xmath135 , as expected for a classical spin model @xcite , we have : @xmath136
as we mentioned before , the decoupling @xmath137 is suitable for regimes with nonzero magnetization ( @xmath138 ) .
hence , the basic equations used in the previous section do not allow us to explore near - zero magnetization domains in the phase diagram as the critical region . to overcome this difficulty ,
one is forced to find a more appropriate decoupling procedure for the longitudinal correlation function which allows us to obtain self - consistent me s appropriate for describing regimes when @xmath139 and preserves also the simplicity of the one @xmath60-function ansatz for the transverse sd @xmath63 . a possible and successful solution to this problem
was suggested several years ago for spin models with short - range interactions @xcite .
these studies showed that a suitable decoupling procedure when the magnetization approaches to zero ( see also ref .
@xcite for the quantum counterpart ) consists in writing ( see eq .
( [ eq : me2 ] ) ) : @xmath140 \left\langle s_{\mathbf{k}^{\prime } } ^{z}s_{-\mathbf{k}^{\prime } } ^{z}\right\rangle \nonumber \\ & \simeq \frac{1}{n}\sum_{\mathbf{k}^{\prime } } \left [ j(\mathbf{k}^{\prime } ) -j(\mathbf{k}-\mathbf{k}^{\prime } ) \right ] \left\ { \left\langle s_{\mathbf{% k}}^{z}\right\rangle \left\langle s_{-\mathbf{k}}^{z}\right\rangle -\frac{1}{% 2}\left ( 1-\frac{\left\langle s_{0}^{z}\right\rangle ^{2}}{n^{2}s^{2}}% \right ) \left\langle s_{\mathbf{k}^{\prime } } ^{+}s_{-\mathbf{k}^{\prime } } ^{-}\right\rangle \right\ } \text{. } \label{eq : decoupling}\end{aligned}\ ] ] as we see , with eq .
( [ eq : decoupling ] ) only the sd @xmath141 is involved and , inserting in the me ( [ eq : me2 ] ) , one finds for the reduced dispersion relation @xmath108 the same formal expression ( [ eq : se_r ] ) but with @xmath142 replaced by : @xmath143 } { \overline{\omega } _ { \mathbf{k}^{\prime } } \omega ^{\left ( p\right ) } \left ( \mathbf{k}\right ) } \text{. } \label{eq : r(k)_new}\ ] ] notice that , the effect of the decoupling ( [ eq : decoupling ] ) corresponds essentially to achieve in eq .
( [ eq : r(k ) ] ) the transformation @xmath144 @xcite .
now , the new me s can be properly used for an estimate of the main critical properties of our classical spin model when it exhibits lro and of the low temperature paramagnetic susceptibility in the remaining domains of the ( @xmath123)-plane when no lro exists . in the limit @xmath115 with @xmath145 , the system of eqs .
( [ eq : se_r])-([eq : sigma ] ) with @xmath142 given by eq .
( [ eq : r(k)_new ] ) , become @xmath146 where @xmath147 the reduced critical temperature @xmath148 ( with @xmath149 ) can be determined by imposing the condition @xmath150 and hence by solving the self consistent equation : @xmath151 for an explicit estimate of @xmath148 , one can again calculate the integrals in eqs .
( [ eq : omg2])-([eq : q(t ) ] ) assuming for @xmath95 the dominant contribution as @xmath93 and solving our self - consistent equations by iteration . to first level of iteration
we find @xmath152 then , eq . ( [ eq : find_tc ] )
yields : @xmath153 where @xmath119 is given by eq .
( [ eq : sigma_explicit ] ) .
thus , a critical temperature exists when @xmath119 is finite and hence in the domains of the ( @xmath123)-plane where lro takes place ( see fig .
1 ) , as expected . of course
the estimate for the critical temperature can be systematically improved by calculating the integral @xmath154 using next @xmath96-powers in the expansions ( [ eq : omega_lowk ] ) . here
we consider explicitly the two - dimensional case for @xmath155 also for a comparison with recent analytical and monte carlo results . in this case , the coefficients in the expansions ( [ eq : omega_lowk ] ) are given by @xcite : @xmath156 } \\
b_{2}=2^{p-2}\zeta \left ( \frac{p}{2}-1\right ) \left [ \zeta \left ( 1-\frac{p% } { 2},\frac{1}{4}\right ) -\zeta \left ( 1-\frac{p}{2},\frac{3}{4}\right ) \right ] \end{array } \right .
\text { , \ } 2<p<4\ ] ] where @xmath157 is the generalized riemann zeta function and @xmath158 is the ordinary riemann zeta function .
then , for the integral @xmath159 , in the case of interest @xmath155 , we have : @xmath160 , \ ] ] and hence : @xmath161 } .\ ] ] the critical temperature as a function of @xmath11 ( @xmath155 ) for @xmath20 and @xmath67 is plotted in fig . 2 and compared with the corresponding results recently obtained for the quantum heisenberg model by nakano and takahashi , using a modified spin - wave ( sw ) theory @xcite and the two - time green function emm within the tyablikov decoupling @xcite , and by vassiliev et al . with a monte carlo simulation @xcite .
our result appears to be consistent with the ones obtained for the quantum counterpart in view of the known feature that a quantum spin model can be reasonably approximated by a classical one only in the large-@xmath1 limit @xcite . from eqs .
( [ eq : omg2])([eq : q(t ) ] ) it is easy to obtain also the behavior of @xmath162 as @xmath163 . with @xmath164 and @xmath165 ,
we get ( for all values of @xmath11 and @xmath0 in the domains where a phase transition is allowed ) : @xmath166 with @xmath167 next , for the reduced paramagnetic susceptibility , defined as @xmath168 in the limit @xmath115 with @xmath169 , eqs .
( [ eq : se_r])-([eq : sigma ] ) with the new expression ( [ eq : r(k)_new ] ) for @xmath142 reduce to @xmath170 with @xmath171 . as usual , an estimate of @xmath172 as @xmath173 can be obtained assuming the low-@xmath45 behavior ( [ eq : omega_lowk ] ) for @xmath91 with the aim to find a solution of eqs .
( [ eq : x1 ] ) and ( [ eq : x2 ] ) such that @xmath174 as @xmath175 .
then , we find : @xmath176 with @xmath177 and @xmath178 the symbol @xmath179 , here and below , denotes the main @xmath180-dependence with a logarithmic correction ( for instance , @xmath181 $ ] ) .
we now determine the behavior of the reduced magnetization @xmath85 along the critical isotherm as @xmath115 .
starting from the basic me s ( [ eq : se_r])([eq : sigma ] ) it is immediate to see that , for small values of @xmath110 and @xmath85 , the equation for the critical isotherm has the form : @xmath182 where @xmath183 and @xmath184 is determined by eq .
( [ eq : x2 ] ) with @xmath185 replaced by @xmath186 and @xmath187 .
( [ eq : qc1 ] ) can be solved numerically but an reliable estimate of the reduced magnetization @xmath85 as @xmath188 can be simply obtained , as usual , assuming in eqs .
( [ eq : qc1 ] ) and ( [ eq : qc2 ] ) the dominant contribution of @xmath91 as @xmath189 . with this assumption and the definition @xmath190 for the isotherm critical exponent @xmath191 , eqs . ( [ eq : qc1 ] ) and ( [ eq : qc2 ] ) yield : @xmath192 and @xmath193 where @xmath194 characterizes the behaviour @xmath195 @xmath196 .
finally , we calculate the critical exponent @xmath116 for the specific heat in the domains of @xmath11 and @xmath0 where the transition to a fm phase occurs .
it is important to note that we can not use the expression ( [ eq : u2 ] ) for the reduced internal energy per spin which has been obtained assuming the decoupling @xmath197 .
rather , we must use the general relation ( [ eq : u ] ) with the decoupling procedure ( [ eq : decoupling ] ) appropriate near the critical point . bearing this in mind
, the reduced internal energy per spin near the critical point assumes the form : @xmath198 where now @xmath108 involves the quantity @xmath199 given by eq .
( [ eq : r(k)_new ] ) . working for @xmath200 , @xmath169 with @xmath201 as @xmath175 , setting @xmath202 in eq .
( [ eq : u_ch ] ) and taking into account eq .
( [ eq : x1 ] ) for @xmath203 with @xmath204 evaluated as * * k**@xmath205 , we have for the reduced zero - field specific heat @xmath206 : @xmath207 this interesting expression allows us to determine the specific heat critical exponent @xmath116 . indeed , with @xmath208 or @xmath209 $ ] , eq .
( [ eq : ch0 ] ) yields : @xmath210 , \text { } & \alpha = 0_{\ln } \text { . } \end{array } \right . \label{ch_alpha}\ ] ] with these definitions and the values of the exponent @xmath211 determined before one can immediately obtain the desidered values of @xmath116 .
we find : @xmath212 in the domain where lri s are active , and @xmath213 in the sri regime .
one can easily check that the critical exponents @xmath214 , @xmath116 , @xmath211 , and @xmath60 calculated above satisfy the well known scaling laws @xmath215 , @xmath216 and @xmath217 for all values of @xmath11 and @xmath0 which allow a transition to the fm phase . besides , they coincide with those obtained by nakano and takahashi @xcite for the quantum counterpart , consistently with the universality hypothesis @xcite . the low temperature behavior of the susceptibility in absence of lro , _
i.e. _ for @xmath14 and @xmath4 ( see fig .
1 ) , is given by eqs .
( [ eq : x1 ] ) and ( [ eq : x2 ] ) but bearing in mind that now @xmath219 .
following a procedure similar to the one used in the preceding section for obtaining the behavior of the paramagnetic susceptibility as @xmath220 , in the low temperature limit , we obtain : @xmath221 , \quad & p=2d \\
\overline{t}^{-\frac{p - d}{p-2d}},\quad & 2d < p<2+d \\
\overline{t}^{-\frac{2}{2-d}}\left ( \ln \left [ \frac{1}{t}\right ] \right ) ^{% \frac{d}{2-d } } & p=2+d \\ \overline{t}^{-\frac{2}{2-d}},\quad & p>2+d . \end{array } \right .
\label{eq : chi_tdiv}\ ] ] the previous behaviors generalize those obtained in ref .
@xcite for @xmath19 .
it is worth noting that for case @xmath16 one finds a pure exponential divergence as @xmath222 as for one - dimensional model @xcite .
this result , although consistent with that one obtained within the modified spin - wave theory @xcite and the emm for the two - time green functions @xcite for the quantum counterpart , does not reduce , for @xmath19 , to the expression derived by haldane @xcite which contains a factor proportional to @xmath223 .
so , one must expect a @xmath0-dependent power law factor in @xmath94 in eq .
( [ eq : chi_tdiv ] ) , which corrects the pure low - temperature exponential divergence .
the simple lowest - order approximation in the csdm , here used , is not able to capture this important physical aspect .
however , within the spirit of the sdm @xcite , one can hope to improve systematically this result working to higher order approximations .
in this work we have applied the sdm , within the framework of the classical statistical mechanics ( csdm ) , to study the thermodynamic properties of a classical @xmath0-dimensional heisenberg fm model with lri s decaying as @xmath2 ( @xmath5 ) in the presence of an external magnetic field .
to lowest order in the csdm , the most relevant magnetic quantities have been obtained analytically in the low - temperature regime under near saturation conditions as a functions of @xmath0 and @xmath11 .
the fm lro at finite temperature has been shown to occurs in a wide region of the ( @xmath123)-plane ( see fig .
1 ) where a transition to a fm phase takes place decreasing the temperature .
the thermodynamic regimes with near zero and zero magnetization have been explored on the basis of a proper modification in the me s of the transverse sd .
so we have estimated the critical temperature and the main critical properties of the model , beyond the mean field approximation , in the domains of the ( @xmath123)-plane where a fm ordered phase is expected .
the critical exponents , here determined as a function of @xmath11 and @xmath0 , coincide with those ones for the corresponding quantum model .
this constitutes a microscopic check of the irrelevance of quantum fluctuations , consistent with the universality hypothesis in the theory of critical phenomena .
in contrast , as expected , our calculations show that the nonuniversal parameters , such as the critical temperature , depends on the classical or quantum nature of the model under study , especially for small values of the spin @xmath1 as explicitly shown by a comparison of our critical temperature estimates with analytical studies and monte carlo simulations previously achieved for the quantum counterpart .
finally , we have determined the low - temperature behavior of the paramagnetic susceptibility for @xmath14 with @xmath121 when no lro occurs .
the rich scenario here obtained in a unified and consistent way for a nontrivial spin model , shows clearly the potentiality and the effectiveness of the sdm also in treating classical many - body systems .
it is indeed relevant feature that , already to the lowest - order approximation , the csdm is able to capture the essential physics of the model , consistent with monte carlo simulations @xcite and exact results @xcite , and to obtain results beyond the tyablikov - like approximation for arbitrary values of dimensionality @xmath0 and the decaying exponent @xmath5 . besides , it offers the possibility to achieve systematically higher - order approximations with the aim to improve the results here obtained and to study also the damping of the oscillations on the same footing @xcite . in conclusion , due to the great experience acquired in the quantum many - body theory , we believe that the csdm @xcite and the formalism of the two - time green functions in classical statistical mechanics @xcite constitute a promising tool to explore equilibrium and transport properties of a wide variety classical many - body ( not only magnetic ) systems .
further developments and applications along this directions are desiderable . )
-plane when a ferromagnetic long - range - order ( lro ) exists and is absent ( no lro ) . the dashed line @xmath224 separates the domains where long - range interaction ( lri)- and short - range interaction ( sri)-regimes occur .
the dashed region ( @xmath225 ) corresponds to a nonextensive thermodynamics .
, width=340 ] as a function of the decaying exponent @xmath11 for a two - dimensional spin-@xmath9 heisenberg ferromagnet .
the full line ( @xmath226 ) represents the estimate here obtained .
the dashed ( @xmath227 ) and ( @xmath228 ) lines refer to green function ( gf ) method @xcite and modified spin - wave ( sw ) theory @xcite results .
the dots depict the quantum monte carlo ( @xmath229 ) predictions @xcite .
, width=377 ] | the thermodynamic properties of a classical @xmath0-dimensional spin-@xmath1 heisenberg ferromagnet , with long - range interactions decaying as @xmath2 and in the presence of an external magnetic field , is investigated by means of the spectral density method in the framework of classical statistical mechanics .
we find that long - range order exists at finite temperature for @xmath3 with @xmath4 and for @xmath5 with @xmath6 , consistently with known theorems .
besides , the related critical temperature is determined and a study of the critical properties is performed . , and classical heisenberg model , long - range interactions .
05.20.-y , 75.10.hk , 64.60.fr |
according to our knowledge , gamma - ray bursts ( grbs ) are the most powerful mysterious explosions the universe has ever seen since the big bang . with swift satellite , since november 20 , 2004 , we have a tool , which can solve the gamma - ray burst mystery .
we define two samples from the swift dataset @xcite : the sample of grbs without measured redshift @xmath1 ( 189 grbs ) , and the sample with measured @xmath1 ( 97 grbs ) .
the swift catalogue consists of the name of grb , its bat duration @xmath2 , bat fluence at range 15 - 150 kev , bat peak flux at range 15 - 150 kev , and redshift .
the sample covers the period november 2004 - december 2007 ; the first / last event is grb041227 / grb071227 .
we have studied both samples separately and also together as the whole sample ( 286 grbs ) .
the first evidence about the existence of three subgroups of grbs came from the @xmath0 fitting of the durations of batse dataset @xcite .
we proceed here identically .
we study the whole sample and the two ones . on the x - axis
there are the bins ( intervals ) of decimal log @xmath2 , and on the y - axis there is the number of grbs in each interval .
the number of bins is 10 and the bins define a histogram . we fitted the histogram .
the whole sample of the grbs has 276 grbs with the measured duration .
the best fit of the whole sample with one single gaussian curve gives : @xmath3 = 1.42 ( with mean @xmath2 = 26.30s ) , @xmath4 = 0.94 , @xmath0 = 53.54 .
the goodness of the fit with 8 @xmath5 ( degrees of freedom ) gives the rejection on the 99.99% significance level @xcite .
the fit with the sum of two gaussian curves gives : @xmath6 = -0.27 ( @xmath2(1 ) = 0.54s ) , @xmath7 = 0.94 , @xmath8 = 1.57 ( @xmath2(2 ) = 37.15s ) , @xmath9 = 0.56 , @xmath10 = 0.12 ( 12% of the grbs belongs to the short grbs ) , @xmath0 = 6.72 . here
@xmath5 = 5 and we obtain an excellent fit with the significance level 50% .
the fit with the sum of three gaussian curves gives : @xmath6 = 0.30 ( @xmath2(1 ) = 2.00s ) , @xmath7 = 1.27 , @xmath12 = 0.20 , @xmath8 = 0.82 ( @xmath2(2 ) = 6.61s ) , @xmath9 = 0.07 , @xmath13 = 0.12 , @xmath14 = 1.72 ( @xmath2(3 ) = 52.48 ) , @xmath15 = 0.46 , @xmath0 = 3.82 , we obtain an excellent fit for @xmath5 = 2 , because the significance level is only 78% .
the decreasing of @xmath16 = 2.89 is not statistically significant @xcite , and hence the introduction of the third intermediate subgroup is not necessary .
the fit is shown on the figure 3 .
the sample with measured redshifts contains 94 grbs with measured duration .
the best fit of the sample with one single gaussian curve gives : @xmath3 = 1.43 ( @xmath2 = 26.92s ) , @xmath4 = 0.87 , @xmath0 = 12.59 .
the goodness of the fit with 8 @xmath5 does not reject the hypothesis of one single gaussian curve .
the rejection is only on the 85% significance level .
the fit with the sum of two gaussian curves gives : @xmath6 = -0.46 ( @xmath2(1 ) = 0.35s ) , @xmath7 = 0.51 , @xmath8 = 1.52 ( @xmath2(2 ) = 33.11s ) , @xmath9 = 0.62 , @xmath10 = 0.06 ( 6% of the grbs belongs to the short grbs ) , @xmath0 = 3.45 .
here we obtain an excellent fit with the significance level 97% @xcite , thus the fit can not be rejected .
the fit with the sum of three gaussian curves gives : @xmath6 = -0.71 ( @xmath2(1 ) = 0.19s ) , @xmath7 = 0.70 , @xmath12 = 0.09 , @xmath8 = 0.51 ( @xmath0(2 ) = 3.24s ) , @xmath9 = 0.01 , @xmath13 = 0.01 , @xmath14 = 1.56 ( @xmath2(3 ) = 36.31s ) , @xmath15 = 0.54 , @xmath0 = 3.16 , it is an excellent fit with @xmath5 = 2 , because the significance level is only 80% .
the decreasing of @xmath16 = 0.29 is not statistically significant @xcite , also the introduction of the third intermediate subgroup is not necessary .
the fit is shown on the figure 2 .
the sample without measured redshifts contains 182 grbs with measured duration .
the best fit of the sample with one single gaussian curve gives : @xmath3 = 1.40 ( @xmath2 = 25.12s ) , @xmath4 = 0.95 , @xmath0 = 43.23 .
the goodness of the fit with 8 @xmath5 gives the rejection on the 99.99% significance level .
the fit with the sum of two gaussian curves gives : @xmath6 = -0.73 ( @xmath2(1 ) = 0.19s ) , @xmath7 = 0.71 , @xmath8 = 1.54 ( @xmath2(2 ) = 34.67s ) , @xmath9 = 0.56 , @xmath10 = 0.09 ( 9% of the grbs belongs to the short grbs ) , @xmath0 = 3.42 .
here we obtain an excellent fit , which can not be rejected , @xcite .
the fit with the sum of three gaussian curves gives : @xmath6 = -0.70 ( @xmath2(1 ) = 0.20s ) , @xmath7 = 0.70 , @xmath12 = 0.09 , @xmath8 = 0.51 ( @xmath2(2 ) = 3.24s ) , @xmath9 = 0.01 , @xmath13 = 0.01 , @xmath14 = 1.56 ( @xmath2(3 ) = 36.31s ) , @xmath15 = 0.34 , @xmath0 = 3.16 , hence is an excellent fit with @xmath5 = 2 , because the significance level is only 80%.the decreasing of @xmath16 = 0.26 is not statistically significant @xcite , and then the introduction of the third intermediate subgroup is not necessary .
the fit is shown on the figure 1 .
because the article @xcite described the existence of the third ( intermediate ) subgroup of grbs in the batse database by the @xmath0 fitting of the duration , we worked out an identical procedure on the existing swift database .
contrary to the batse grbs ( @xcite ) , the swift grbs do not require any introduction of the third intermediate subgroup in contrast to @xcite ; the hypothesis is that the short and intermediate subgroups should form one single subclass .
this study was supported by the gauk grant no.46307 , by the otka grant no.t48870 , by the grant agency of the czech republic , grants no .
205/08/h005 , and by the research program msm0021620860 of the ministry of education of the czech republic .
the authors appreciate valuable discussion and help of j. pa . | a sample of 286 gamma - ray bursts , detected by swift satellite , is studied statistically by the @xmath0 test and the student t - test , respectively .
the short and long subgroups are well detected in the swift data .
but no intermediate subgroup is seen .
the non - detection of this subgroup in the swift database can be explained , once it is assumed that in the batse database the short and the intermediate subgroups form a common subclass .
address = astronomical institute of the charles university , v holeovikch 2 , prague , czech republic address = astronomical institute of the charles university , v holeovikch 2 , prague , czech republic |
natural language is an evolving system whose present structure can doubtlessly be considered a product of long history of self - organization @xcite . like for many other self - organized systems known in nature ,
the observables associated with language , being , for example , written texts or spoken messages , reveal quite sophisticated dynamics .
any language sample by no means is an amorphous mixture of symbols ( letters , phonemes , morphemes , words , etc . ) but rather a highly organized sequence in which particular symbols are ordered according to specific rules most of which are defined by the language grammar . since the existence of grammar is an emergent phenomenon @xcite , language can be counted among the complex systems @xcite .
the grammatical rules together with the information content impose on the language elements relations which can be most easily expressed in a form of network where , for instance , words are expressed by nodes and their relations by edges .
some earlier attempts along this way were presented in refs .
@xcite for english , portuguese and chinese . here
we show a few results that were obtained for english and polish and for different types of texts ( literary or scientific ) .
our analysis was based on texts samples written in two languages : english and polish .
both belong to the indo - european family , but to different groups : the west - germanic and the west - slavic group , respectively .
their grammar therefore significantly differs , most notably in the existence of rich inflection of words in polish as compared to a rather residual one in english .
however , in the present work we do not deal with the semantic analysis , but restrict our study to a statistical analysis of word adjacency .
as regards the english part , we analyze two groups of texts .
the first one comprises the literary texts represented by 9 works of prose ( `` ulysses '' and `` finnegans wake '' by j. joyce , `` alice s adventures in wonderland '' by l. carroll , `` adventures of huckelberry finn '' by m. twain , `` pride and prejudice '' by j. austen , `` oliver twist '' by c. dickens , `` secret adversary '' by a. christie , `` adventures of sherlock holmes '' and `` study in scarlet '' by a. conan doyle ) , 4 dramas by w. shakespeare ( `` hamlet '' , `` macbeth '' , `` winter tale '' , and `` romeo and juliet '' ) , 61 poems by o. wilde , and 25 poems by t.s .
the second group comprises the scientific texts represented by selected works of e. witten @xcite , e.g.d .
cohen @xcite , s. weinberg @xcite , and p.w .
anderson @xcite , as well as by three long reviews by d. sornette @xcite , r. albert and a .-
barabsi @xcite , and j. kwapie and s. drod @xcite .
somewhere at the interface of these two groups , there is `` a brief history of time '' by s. hawking and `` the emperor s new mind : concerning computers , minds , and the laws of physics '' by r. penrose representing popular science .
the polish language was represented by the novels : `` lalka '' ( `` the doll '' ) by b. prus , `` bramy raju '' ( `` the gates of paradise '' ) by j. andrzejewski , `` dolina issy '' ( `` the issa valley '' ) by c. miosz , `` cesarz '' ( `` the emperor : downfall of an autocrat '' ) by r. kapuciski , `` dzienniki gwiazdowe '' ( `` the star diaries '' ) by s. lem , `` ferdydurke '' by w. gombrowicz , the epic `` pan tadeusz '' ( `` sir thaddeus '' ) by a. mickiewicz , the only polish translation of `` ulysses '' done by m. somczyski , 35 poems by c. miosz , and 99 poems by w. szymborska .
all the texts were filtered in order to remove some of the punctuation marks ( all except the ones that can functionally end sentences : the periods , the colons and semicolons , the question and exclamation marks ) as well as the non - word sequences like numbers .
the so - preprocessed texts were subject to further analysis .
although language and language samples are traditionally subject to purely qualitative analysis in the fields of humanities , the language samples consist of symbolic sequences , which can be easily subject to quantitative analysis .
historically , the beginning of quantitative analysis of natural language is usually associated with the name of g.k .
zipf , who was the first to carry out an extensive study of word frequencies in written texts in a few different languages @xcite , despite that in fact he also had known predecessors like j .- b .
estoup @xcite and e.l .
thorndike @xcite who did some research in the same direction but far less extensive than the zipf s and without any significant impact on science .
the main result attributed to zipf is his eponymous law stating that number @xmath0 of occurrences of words ordered according to their relative frequency in text samples is , roughly , inverse proportional to their rank @xmath1 . for the english language , it is : @xmath2 while for other languages @xmath3 can also be slightly smaller or larger than 1 . @xmath4 stands here for an empirical proportionality constant equal to @xmath5 , where @xmath6 is the total number of words in a sample ( a sample s length ) . for single texts samples like books ,
the above power - law relation is usually well preserved for medium ranks ( e.g. , @xmath7 ) in the majority of cases , but for the lowest and the highest ranks it breaks down leading to flattening of @xmath8 for the most frequent words and to its faster decline for the least frequent ones .
as regards the multi - piece unions of text samples ( corpora ) , the scaling ( [ eqn::zipf ] ) holds up to the ranks of a few thousand and above them another scaling region with a larger value of @xmath3 appears @xcite .
the latter phenomenon can be interpreted as a division of the vocabulary into two sections : the basic vocabulary containing words that are shared by almost all the text samples , thus common to all people , and the specialist vocabulary that can be subject - specific and author - specific @xcite .
an interesting property of the zipf law is its similarity to a number of other relations that can be found in different and sometimes very distant fields : city population , scientific paper citations , earthquake magnitude , and many more @xcite . as regards the rank - frequency relation for text samples , fig .
[ fig::zipf](a ) shows such a plot for `` ulysses '' .
it is notable for its uniquely broad range of ranks ( 3 decades ) for which a power - law scaling holds , which is extremely rarely equalled by other pieces of texts .
for example , a similar plot for a brief history of time " in fig .
[ fig::zipf](b ) reveals scaling valid for only 2 decades .
this means that the vocabulary volume of this book is smaller than it would be expected from the power - law relation holding over all the ranks . on contrary ,
`` finnegans wake '' ( fig .
[ fig::zipf](c ) ) possesses extremely diverse vocabulary and the rarest words are overrepresented leading to breaking of the zipf - like relation in the opposite direction as compared to the hawking s book .
such a situation does not surprise us , however , since `` finnegans wake '' is known to be a highly experimental piece of text comprising words from many languages .
next , fig .
[ fig::zipf](d ) shows a rank - frequency plot obtained for the polish text of `` pan tadeusz '' .
a well - fitted power - law function with @xmath9 describes the plot , whose slope is much smaller than for typical english texts and even for typical polish ones ( @xmath10 @xcite ) , while it is characteristic for the works of a. mickiewicz @xcite .
characterization of a text sample by means of the zipf plot is informative in respect to vocabulary volume of an author and mutual relations between the most common and other words , but it is insensitive to any kind of correlations possibly present in the sample . in order to incorporate correlations into our analysis , we create network representations of each of the text sam2ples studied here .
we choose such a representation in which different words are regarded as different network nodes .
one type of interesting correlations that can be quantified in this way is the adjacency relation between pairs of words .
two words are considered related and their nodes linked by an edge if they are the nearest neighbours at least once in a sample . for a given word
, there are two possible relations with its neighbours : precedence and succession .
the former is when a neighbour precedes the considered word , while the latter is in the opposite case . in this context
, we study two networks : the precedence ( left - side neighbourhood ) network and the succession ( right - side neighbourhood ) network .
we decided to consider only the neighbours that belong to the same sentence and neglect the inter - sentence adjacency .
this does not influence our results , however : a preliminary analysis carried on a few text samples showed that there is no qualitative difference of the results between these cases .
this , of course , might be a consequence of a much smaller number of such inter - sentence pairs ( roughly , less than 10% of all pairs ) . by construction
, our networks can be either binary or weighted . in the former case ,
we consider two nodes to be linked by an edge if the respective words are neighbours at least once in a text but we do not pay attention to how many times they neighbour each other .
in contrast , in the latter case , we may count the number of such occurrences and attribute a corresponding weight to each edge .
examples of both situations are presented in fig .
[ fig::hedgehog ] , where a binary network ( left ) and a minimal spanning tree calculated from a weighted network ( right ) are created for an exemplary piece of text . to begin with ,
let us calculate a cumulative distribution @xmath11 of node degrees @xmath12 for different texts .
this is one of the most informative quantities since it allows one to detect a hierarchical and scale - free structure of a given network @xcite .
since such distributions for weighted networks carry basically the same information as the zipf plots , we restrict our analysis to the binary networks only .
[ fig::literary.english.node.degrees ] exhibits cumulative distributions @xmath11 for selected literary texts in english .
interestingly , although there are clear differences between the distributions for different texts , all the texts studied ( including other not shown here ) reveal the scale - free or almost scale - free dependence for some range of @xmath12 , with the scaling exponents @xmath13 being in agreement with the results from other studies of different systems @xcite .
it happens for some texts that the precedence and the succession networks visibly differ from each other .
we do not inspect this issue in more detail , but a source of this difference might be either author - specific or text - specific . in fig .
[ fig::literary.polish.node.degrees ] , the @xmath11 distributions are shown for selected polish literary texts . for prose ( ( a)-(d ) ) , the scale - free slopes of these distributions are even better visible than for the english texts in fig .
[ fig::literary.english.node.degrees ] . the same refers to poems except the overrepresentation of nodes with small @xmath12 in the case of polish poetry ( fig . [ fig::literary.polish.node.degrees](e)-(f ) ) .
this overrepresentation probably stems for the fact that poetry , which needs a specific rhythm , imposes strong restrictions on the words that can be used in particular places . in order to compare statistical properties of node degrees for different types of texts and the two languages , from the texts considered in this work , we create five separate corpora containing english prose , english poetry , english scientific texts , polish prose , and polish poetry .
then for each corpus , we calculate a node degree cumulative distribution @xmath11 .
[ fig::corpora.node.degrees ] shows these distributions along with their power - law slopes .
regarding prose , the distribution for the english language has smaller slope than its counterpart for the polish language .
however , the picture for poetry looks different : the node degree distribution is steeper in the case of english .
roughly , as regards the corpora of polish prose and of polish poetry , the distributions look similar , which is not the case for their english counterparts .
@xmath11 for the corpus of scientific texts written in english reveals the most steep slope with @xmath14 .
this is not surprising , however , since many scientific texts are full of mathematics and related formal names and expressions , which make the vocabulary poorer than in the case of literary works , which do not have any vocabulary restrictions ( see fig .
[ fig::corpora.node.degrees](a ) ) .
as regards @xmath11 for the individual scientific papers , some of them do not reveal any trace of scaling while other are clearly scale - free .
this strongly depends on the relative amounts of standard description and strict mathematical language : the less mathematics is there , the better scaling can be observed .
our results for both languages indicate that the adjacency networks in both representations are strongly non - democratic with a clear hierarchy of hubs . indeed , fig .
[ fig::hierarchy ] confirms this conclusion by showing both global and local hubs with large values of @xmath12 surrounded by clouds of peripheral nodes with @xmath15 .
other topological properties of the networks can be characterized by their spatial extension and inclination of nodes to form clusters .
the former can be quantitatively described by the average characteristic path length @xmath16 expressing the average node - to - node distance . for a binary network
it is defined by : @xmath17 where @xmath18 is the length of the shortest path connecting the nodes @xmath19 and @xmath20 ( the length of each path is defined as a number of edges this path passes through ) .
magnitude of @xmath16 and , especially , its dependence on @xmath21 is different for different network types .
it typically grows fast for regular lattices and chains ( @xmath22 ) , moderately fast for random networks of the erds - rnyi type ( @xmath23 ) , the small - world networks ( @xmath24 ) and for the barbasi - albert networks ( @xmath25 ) @xcite , slowly for the ultrasmall networks ( @xmath26 ) @xcite , while for densely connected networks it can roughly be independent of @xmath21 . for our networks , values of this quantity ( for the complete texts ) belong to the interval : @xmath27 , which generally falls into the small - world networks regime .
however , the asymptotic behavior of @xmath28 is significantly different from the small - world one since for @xmath29 , @xmath16 is decreasing function of @xmath21 ( figure [ fig::path.nodes ] ) .
this is because , for large @xmath21 , the vocabulary volume @xmath30 ( here , @xmath31 ) used for writing a text grows much more slowly than the length @xmath6 of the text ( which is expressed by a general relation @xmath32 , where @xmath33 - the so - called heaps law ) and this leads to increasing density of edges .
the clustering coefficient @xmath34 for an undirected binary network is expressed by : @xmath35 where @xmath36 are binary edges ( equal to 1 for the existing edge and 0 otherwise ) .
its value is typically small for the erds - rnyi networks ( @xmath37 ) and for the barbasi - albert networks ( @xmath38 ) @xcite , while it is large ( and independent of @xmath21 ) for the small - world networks of the watts - strogatz type @xcite .
for all the texts except the highly mathematical papers by e. witten and s. weinberg , values of @xmath34 are in the interval @xmath39 , but unlike the shortest path length , here we observe rather clear separation between polish and english novels ( including the shakespeare s drama ) : @xmath40 for the polish ones and @xmath41 for the english ones . the poetry in both languages tends to have smaller clustering coefficient than any other considered piece of text : @xmath42 and @xmath43 .
these results can be seen in a scatter plot in figure [ fig::path.vs.coeff ] , which shows values of the clustering coefficient and the shortest path length for all the texts considered in this work .
indeed , the polish and the english texts occupy different regions of the ( @xmath34,@xmath16 ) plane with the polish ones being characterized by smaller @xmath34 . for english
, one can also see that scientific texts may have different properties than the literary texts , especially if they contain much mathematics ( witten , weinberg ) .
if they are more descriptional than mathematical , their properties can resemble the properties of literary texts ( sornette , hawking , penrose ) .
it is interesting to note that unlike other network types mentioned above , for our adjacency networks , @xmath34 is an increasing function of @xmath21 and , typically , its dependence is either power - law @xmath44 ( at least for large values of @xmath21 ) with the scaling index @xmath45 or not far from power - law .
a similar power - law behavior can be seen for @xmath46 , where @xmath6 is the text length , but this is not surprising due to the already - mentioned heaps law : @xmath47 .
figure [ fig::clust.power.law]a shows two examples of such power - law behavior for one english and one polish text . as it was the case with @xmath34 ,
the english and the polish texts have distinct values of @xmath48 , with the latter being significantly larger .
calculated values of this index for the texts that show clear power - law dependence of @xmath46 are collected in figure [ fig::clust.power.law]b .
although @xmath48 can not be estimated for all the texts from figure [ fig::path.vs.coeff ] , the separation of the languages is clear also here .
the observed behavior of @xmath28 and @xmath49 suggests that the networks considered in this work have their own specific structure that clearly distinguishes them from the most well - known network structures .
they can not be counted as small - world networks even though the average characteristic path length is relatively short .
they also differ from the barbasi - albert networks despite the fact that some of they show the scale - free structure and from the random ( erds - rnyi ) ones .
we presented several results from our quantitative study of statistical and network properties of literary and scientific texts written in two substantially different languages : english and polish .
we transformed the text samples into word - adjacency networks defined by the nodes representing individual words and the edges representing pairs of directly neighbouring words .
for the majority of the studied literary texts in both languages , the corresponding networks revealed the scale - free structure , while this was rarely the case for the scientific texts .
we also showed that there are differences in node degree distributions between prose and poetry , especially in english .
poetry has a detectable steeper distribution s slope than has prose .
the slope for scientific texts is even steeper than for poetry , but this can be explained by typically poorer vocabulary in the former case . despite these differences , all the network representations of texts were hierarchical with a few important hubs and the majority of less important nodes .
no qualitative and quantitative difference between the languages was noticed in this respect .
this picture changed completely if we looked at other network statistics like the clustering coefficient and the average shortest path length .
the english texts appear to possess more clustered structure , while the polish ones were less clustered .
this result was attributed to differences in grammar of both languages , which was also indicated in the zipf plots .
our results suggest that the word - adjacency networks can not fully be described by any of the erds - rnyi , the watts - strogatz , and the barbasi - albert models even though these networks exhibit certain characteristics of the latter two models .
such networks will be a subject of our forthcoming study .
99 l. da fontoura costa , o.n .
oliveira jr . , g. travieso , f. aparecido rodrigues , p.r .
villas boas , l. antiquera , m. pahlares viana , l.e .
correa da rocha , analyzing and modeling real - world phenomena with complex networks : a survey of applications , adv .
phys . * 60 * , 329 ( 2011 ) e. witten , ( 1 ) analytic continuation of liouville theory ; ( 2 ) khovanov homology and gauge theory ; ( 3 ) knot invariants from four - dimensional gauge theory ; ( 4 ) fivebranes and knots ; ( 5 ) a new look at the path integral of quantum mechanics ; ( 6 ) an algebraic construction of boundary quantum field theory ; ( 7 ) the omega deformation , branes , integrability , and liouville theory ; ( 8) analytic continuation of chern - simons theory ; ( 9 ) geometric langlands and the equations of nahm and bogomolny ; ( 10 ) geometric langlands from six dimensions ; all papers available from _ http://www.arxiv.org _ e.d.g .
cohen , ( 1 ) jet - like tunneling from a trapped vortex ; ( 2 ) stick - slip motion of solids with dry friction subject to random vibrations and an external field ; ( 3 ) path integral approach to random motion with nonlinear friction ; ( 4 ) steady state work fluctuations of a dragged particle under external and thermal noise ; ( 5 ) anomalous fluctuation relations ; ( 6 ) fluctuation properties of an effective nonlinear system subject to poisson noise ; ( 7 ) entropy , probability and dynamics ; ( 8) properties of nonequilibrium steady states : a path integral approach ; ( 9 ) gaussian approximation to single particle correlations at and below the picosecond scale for lennard - jones and nanoparticle fluids ; ( 10 ) nonequilibrium steady state thermodynamics and fluctuations for stochastic systems ; all papers available from _ http://www.arxiv.org _ s. weinberg , ( 1 ) quantum contributions to cosmological correlations ; ( 2 ) effective field theory for inflation ; ( 3 ) a tree theorem for inflation ; ( 4 ) non - gaussian correlations outside the horizon ii : the general case ; ( 5 ) living with iinfinities ; ( 6 ) effective field theory , past and future ; ( 7 ) asymptotically safe inflation ; ( 8) six - dimensional methods for four - dimensional conformal field theories ; ( 9 ) pions in large - n quantum chromodynamics ; ( 10 ) collapse of the state vector ; all papers available from _ http://www.arxiv.org _ p.w .
anderson , ( 1 ) simple explanation of fermi arcs in cuprate pseudogaps : a motional narrowing phenomenon ; ( 2 ) a fermi sea of heavy electrons ( a kondo lattice ) is never a fermi liquid ; ( 3 ) a gross - pitaevskii treatment for supersolid he ; ( 4 ) incoherent tunneling amplitude in high-@xmath50 cuprates ; ( 5 ) transport anomalies of the strange metal : resolution by hidden fermi liquid theory ; ( 6 ) beyond the fermi liquid paradigm : hidden fermi liquids ; ( 7 ) personal history of my engagement with cuprate superconductivity ; ( 8) hidden fermi liquid : self - consistent theory for the normal state of high-@xmath50 superconductors ; ( 9 ) the ground state of the bose - hubbard model is a supersolid ; ( 10 ) theory of supersolidity ; all papers available from _ http://www.arxiv.org _ a. orczyk , zjawisko zoonoci a struktura jzyka polskiego : analiza leksykologiczna wybranych rde literackich
i pozaliterackich narzdziami fizyki statystycznej , master thesis ( in polish ) , wfiis agh ( 2008 ) | we present results from our quantitative study of statistical and network properties of literary and scientific texts written in two languages : english and polish .
we show that polish texts are described by the zipf law with the scaling exponent smaller than the one for the english language .
we also show that the scientific texts are typically characterized by the rank - frequency plots with relatively short range of power - law behavior as compared to the literary texts .
we then transform the texts into their word - adjacency network representations and find another difference between the languages . for the majority of the literary texts in both languages
, the corresponding networks revealed the scale - free structure , while this was not always the case for the scientific texts .
however , all the network representations of texts were hierarchical .
we do not observe any qualitative and quantitative difference between the languages .
however , if we look at other network statistics like the clustering coefficient and the average shortest path length , the english texts occur to possess more clustered structure than do the polish ones .
this result was attributed to differences in grammar of both languages , which was also indicated in the zipf plots .
all the texts , however , show network structure that differs from any of the watts - strogatz , the barbasi - albert , and the erds - rnyi architectures . |
bose - einstein condensates ( becs ) located in different minima of an external potential created by magnetic or light forces and coupled by tunneling through a potential barrier have been the host to many exciting developments and discoveries in recent years @xcite .
phenomena explored include analogs of the josephson effect in double or multiple quantum - well structures @xcite , gap solitons of repulsive becs @xcite , and quantum phase transitions @xcite
. often these systems are modeled by considering just one mode per potential minimum and their linear coupling provided by tunnling through a separating barrier .
these simplified models , which are usually labelled as two - mode or multiple - mode models , variants of the bose - hubbard model , or the discrete nonlinear schrdinger equation , are tailored to describe certain properties or aspects of the dynamics of the many - body system under investigation . although there is an abundance of literature on such models @xcite ,
there still appear to exist misconceptions about the nature of the effective model parameters , especially the sign of the tunnel coupling , as only few works attempt to calculate such parameters based on a more complete theoretical treatment @xcite .
the sign of the tunnel coupling bears special significance in systems where the tunneling appears over an extended ( at least one - dimensional ) region of space .
such systems have recently been analysed by bouchoule @xcite and kaurov and kuklov @xcite , who studied two parallel tunnel - coupled cigar - shaped becs . in another recent work , lesanovsky and
von klitzing studied the stability of tunnel - coupled annular becs @xcite .
the latter paper points to an interesting dynamical instability leading to the spontaneous formation of angular - momentum fluctuations .
we will show further below that the sign on the tunnel coupling bears consequences on the nature and stability of the stationary states found in the mean - field treatment of tunnel - coupled becs . specifically , we find that the system studied by lesanovsky and von klitzing has a stable ground state for any value of the tunnel coupling and repulsive interactions .
the instability of an _ excited _ state with @xmath0 phase difference between the condensates can be interpreted in terms of the familiar snake instability @xcite .
the ground state of a rotating co - planar double - ring system is discussed in ref .
@xcite .
we examine the stationary states of double - ring becs in sec .
[ sec : neg ]
. a careful analysis of the sign of the tunnel coupling used in effective models for becs in double - well traps follows in sec .
[ sec : sign ] .
conclusions are presented in sec .
[ sec : concl ] .
the calculation performed in ref .
@xcite starts from a number of generally reasonable assumptions . under the condition that radial excitations of the vertically stacked annular becs are suppressed by the trapping potentials and
the only mechanism for coupling the two systems is via tunneling through a potential barrier , the gross - pitaevskii equation for the two - mode spinor wave function @xmath1 specialises to @xmath2 here @xmath3 is the condensate wave function for atoms in the upper ( lower ) ring .
our equation ( [ chieqs ] ) agrees with eq .
( 2 ) of ref .
@xcite , except that we explicitly indicate the negative sign of the tunnel coupling .
we give detailed reasons for the relevance of the sign of the tunnel coupling below in sec .
[ sec : sign ] , where we also show that the tunnel coupling is indeed negative . at this point
, we only note that the tunnel coupling @xmath4 was assumed to be positive in ref .
@xcite ( see their fig . 1 ) , in contradiction to our findings .
the most general form of the polar - angle - dependent wave function can be written as a fourier series , @xmath5 . inserting this _ ansatz _ into eq .
( [ chieqs ] ) and equating coefficients of the orthogonal fourier components , we find @xmath6 this result differs from the corresponding eq .
( 3 ) in ref . @xcite in the tunnel coupling and the non - linear term . as a first approximation
, it is reasonable to assume that only the @xmath7 mode is occupied in each of the two annuli .
straightforward calculation yields the new ground and excited state of the coupled - annuli system , which are the symmetric and antisymmetric superpositions of single - well states having chemical potential @xmath8 , respectively .
@xmath9 is defined in terms of the equal number of atoms @xmath10 in each well as in ref .
in order to study the stability of these states , finite but small amplitudes in the @xmath11 modes are assumed : @xmath12 \quad .\ ] ] here the subscript @xmath13 distinguishes perturbations to the ground and excited states , respectively .
inserting the perturbation ( [ peransatz ] ) into eq .
( [ alphaeqs ] ) and linearising in the small amplitudes @xmath14 yields @xmath15 the upper ( lower ) sign refers to the symmetric ground ( antisymmetric excited ) state .
crucial differences between our eq .
( [ bdgeqsfin ] ) and eq .
( 5 ) in ref .
@xcite result in markedly different excitation spectra .
we find that both the symmetric ( ground ) state and antisymmetric ( excited ) state share one branch , @xmath16 whose frequency is independent of the tunnel coupling .
this was also found in ref .
in contrast to these authors , however , we find that the second branch differs for the two states : @xmath17 clearly , @xmath18 is always real for repulsive becs ( @xmath19 ) , implying stability of the symmetric ( ground ) state of the coupled annular condensates .
in contrast , the antisymmetric ( excited ) state will become unstable for @xmath20 , signified by @xmath21 becoming imaginary in this range .
our own numerical simulations of the time evolution of the antisymmetric state seeded with a small amount of noise show the development of angular - momentum josephson junctions similar to those shown in figs . 2 and 3 of ref .
@xcite . in attractive condensates where @xmath22 , imaginary solutions of @xmath23 for @xmath24 indicate the well - known modulational instability towards the formation of localized peaks ( bright solitons ) in the individual rings .
for the symmetric state , @xmath18 does not add new instabilities ( with imaginary solutions for @xmath25 ) .
the antisymmetric state , however , is further destabilised by the tunnel coupling due to imaginary frequencies of @xmath21 at @xmath26 .
in our analysis so far we have assumed that the sign of the coupling constant @xmath4 is negative .
this lead to the symmetric state with @xmath27 and @xmath28 with @xmath29 being the ground state .
let us now briefly consider the consequences of the ( hypothetical ) case of a positive coupling constant @xmath30 .
the analysis of sec [ sec : neg ] can be carried out the same way as before , with the difference that @xmath31 should be replaced by @xmath32 in all formulae .
it is easily seen that , in this case , the antisymmetric state with @xmath33 will be the ground state . since the sign change also affects eq .
( [ eqn : unstable ] ) , we find the antisymmetric state being stable ( for @xmath19 ) and the symmetric one becoming unstable .
however , since the roles of these states have changed , we still find that the ground state is stable for repulsive becs .
energy difference @xmath34 between the lowest antisymmetric and symmetric eigenstates of a quadratic - plus - quartic double - well potential , plotted as a function of the dimensionless effective interaction strength @xmath35 .
the fact that @xmath36 indicates that the symmetric ( node - less ) state remains the ground state even in the limit where the atoms interact strongly .
double - well parameters [ see eq .
( [ eq : dwell ] ) ] are @xmath37 and @xmath38 ( solid curve ) , 0.02 ( dashed curve ) , 0.05 ( dot - dashed curve).,width=288 ] in order to determine the correct sign and value of the coupling constant @xmath4 appearing in eq .
( [ chieqs ] ) , we briefly revisit the derivation of this model . for the purpose of finding @xmath4 , the azimuthal degree of freedom in the double - ring model of ref .
@xcite is irrelevant and it suffices to consider the problem of a bec in a one - dimensional ( 1d ) double - well potential , as in refs .
generalisation to multiple wells and different geometries ( coupled cigars or pancakes ) are straightforward .
different derivations of effective two - mode models have been presented in the literature @xcite .
the goal of a two - mode model is generally to correctly describe the ground and low - lying excited states of the system .
the quantity that is obtainable from the 1d model and carries unambiguous information about the sign of the tunnel coupling is the energy difference @xmath39 between the antisymmetric state with one node and the node - less symmetric state . in the simplest case ,
the tunnel coupling @xmath4 is determined from the single - particle linear schrdinger equation .
this approach is commonly used when deriving the fully quantum - mechanical bose - hubbard model @xcite and was the basis of ref .
@xcite . in this case both the sign and the value of @xmath4 are completely independent of particle number or interaction strength .
the node theorem of quantum mechanics @xcite guarantees that the node - less symmetric state in a one - dimensional double - well potential must be the ground state , thus @xmath36 and consequently , the correct sign of @xmath4 is negative . in a more general class of models based on mean - field theory ,
the parameters of the two - mode model are chosen in order to reproduce @xmath40 as found from a one - dimensional gp equation .
the ordering of eigenvalues of the gp equation by the number of nodes in the wave function is now no longer guaranteed by the node theorem of linear quantum mechanics and we are not aware of a non - linear generalization of this theorem . however , we find by numerical calculation that the ordering is preserved under repulsive interactions . the main result of this section is the dependence of @xmath40 on the non - linear interaction strength @xmath35 , shown in fig .
[ fig : sas ] . as can be seen from fig .
[ fig : sas ] , the presence of a repulsive non - linear interaction does not change the sign of @xmath40 and therefore @xmath4 remains negative .
we now present details of our calculation . starting from the three - dimensional gp equation for a bec in a double - well or double - ring trap and employing a separation ansatz , an effective 1d equation describing the dynamics in the direction perpendicular to the potential barrier can be derived : @xmath41 \phi(\xi ) \ , .\ ] ] here the energy scale @xmath42 and length scale @xmath43 defined by the trap are used as units for all energies and the spatial coordinate , respectively , and the condensate wave function @xmath44 is normalized to unity .
we introduced the dimensionless interaction strength @xmath45 , where @xmath46 denotes the number of atoms in the trap and @xmath47 is the effective 1d interaction strength @xcite . to be specific
, we use the double - well potential @xmath48 where @xmath49 parameterizes the barrier height between the two wells centered at @xmath50 .
it is straightforward to solve eq .
( [ eq:1dgpe ] ) with the potential ( [ eq : dwell ] ) and find the lowest symmetric and antisymmetric eigenstates as well as their respective energies @xmath51 and @xmath52 .
figure [ fig : wavefunct ] shows typical results obtained for low and high interactions strengths , respectively . as is apparent from the figure , the higher repulsive interaction strength is associated with more strongly delocalized double - well wave functions , indicating an effectively stronger tunnel coupling .
this can be explained simply by noting that the nonlinear interaction energy for the two condensate fractions in each well shifts up their respective energies , thus effectively lowers the barrier and brings the condensates closer together . as a result , the effective tunnel coupling increases .
most importantly , the energy difference between the lowest symmetric and antisymmetric eigenstates remains positive for any strength of repulsive interactions , which implies that the sign of the tunnel coupling @xmath4 entering eq .
( [ chieqs ] ) is negative .
reference @xcite predicts a dynamical instability of a repulsively interacting bec in a double - ring trap against angular momentum fluctuations . we have carefully revisited the analysis of ref .
@xcite and have recalculated the elementary excitation spectrum .
this leads us to a different conclusion that makes physical sense .
the ground state of a non - rotating condensate in the double - ring configuration is stable against spontaneous angular momentum oscillations .
however , the antisymmetric state with its circular node between the two annular quantum wells can be viewed as the analog of a stationary 2d dark soliton , which is known to have a dynamical instability towards the formation of local vorticity ( " snake instability ) @xcite . | we revisit recent claims about the instability of non - rotating tunnel coupled annular bose - einstein condensates leading to the emergence of angular - momentum josephson oscillation [ phys .
rev .
lett . * 98 * , 050401 ( 2007 ) ] .
it was predicted that all stationary states with uniform density become unstable in certain parameter regimes . by careful analysis
, we arrive at a different conclusion .
we show that there is a stable non - rotating and uniform ground state for any value of the tunnel coupling and repulsive interactions .
the instability of an excited state with @xmath0 phase difference between the condensates can be interpreted in terms of the familiar snake instability .
we further discuss the sign of the tunnel coupling through a separating barrier , which carries significance for the nature of the stationary states .
it is found to always be negative for physical reasons . |
deep surveys of the extragalactic sky have been made at x - ray ( hasinger et al .
1998 ; cowie et al . 2001 ) , optical ( lilly et al.1996 ; madau et al .
1996 ; steidel et al .
1999 ) , infra - red ( goldschmidt et al . 1997 ; rowan - robinson et al . 1997 ; elbaz et al .
1999 ) , sub - mm ( hughes et al . 1998 ; barger et al . 1999 ; blain et al . 1999 ; barger et al . 2000 ; scott et al .
2000 ) , as well as radio wavelengths ( windhorst et al .
1984 ; windhorst et al . 1985
; condon & mitchell 1984 ; donnelly et al .
1987 ; fomalont et al . 1991 ; windhorst et al.1993 ; richards et al . 1998 ; richards et al . 1999 ; prandoni et al.2001 ) .
much of the interest in these surveys of weak objects concern the evidence for star formation and agn evolution in galaxies in the early universe , at redshifts in the range 1 to 3 ( condon & yin 1990 ; rowan - robinson et al .
1993 ; cram 1998 ; mobasher et al . 1999 ; steidel et al . 1999 ; haarsma et al .
2000 ) .
the radio observations are unique in that at radio wavelengths it is possible to peer through the gas and dust that obscures the nuclear regions of galaxies at other wavelengths .
the radio observations also have sufficient angular resolution to distinguish between emission that is driven by star formation and that driven by agn .
although star forming activity in the early universe is perhaps most readily observed at sub - millimeter wavelengths , current sub - millimeter instruments do not have sufficient angular resolution to avoid confusion due to blending of nearby sources , so high resolution radio observations are needed to uniquely identify optical counterparts .
moreover , because of the large negative k - correction , dusty galaxies observed at sub - mm wavelengths all have high redshifts , whereas the radio observations are sensitive to a wide range of redshift and a mix of star - burst and agn activity .
as the most sensitive radio telescope available for these kinds of observations , the very large array ( vla ) has been used for many deep surveys @xcite .
deep radio source surveys have also been made with the australian telescope compact array @xcite , and with the westerbork synthesis radio telescope @xcite , but with less sensitivity and poorer resolution than possible with the vla .
these observations show that below levels of about 10 mjy , the number of radio sources increases more rapidly than the number between 1 mjy and 500 mjy and are composed of a different population of sources .
in this paper we report on the results of new 8.4 ghz vla surveys which cover two fields , each of diameter @xmath12 to the 8% sensitivity level of the on - axis position .
one field , located at @xmath13 + 43@xmath1 , which we will designate as the sa13 field , was one of the hubble space telescope medium deep survey ( mds ) key projects , observed in 1992 - 3 @xcite .
these optical observations , which were made before hst refurbishment , reach a limiting sensitivity of i=25.5 mag .
the extensive complementary vla observation detected sources as faint as @xmath3jy at 8.4 ghz . a preliminary account of the hst identifications in this field @xcite , hereinafter designated as paper i ,
was based on the first half of these observations which were made at lower resolution , and included only radio sources located in the small 2.5 ( arcmin)@xmath14 field of view of the hst .
we have also used these same vla observations of this field to examine the small scale fluctuations in the cosmic microwave background radiation @xcite . this field has also been observed with the vla at 1.4 ghz , covering a much larger area , with additional kpno optical imaging @xcite , and with the merlin / vlbi national facility of the university of manchester at 1.4 ghz @xcite a second field , located at @xmath15 + 50@xmath1 and designated at the hercules field , was previously imaged with wfpc2 in a 48-orbit exposure , and reaches a 10-@xmath16 point source sensitivity of about 27.7 mag in b and v and 26.8 in i @xcite .
twelve known objects in this field , including three radio - weak agn , have spectroscopic redshifts z near 2.4 ( pascarelle et al . 1996a ; pascarelle et al .
1996b ; pascarelle et al .
1998 ; pascarelle et al .
2001 ) . a wfpc2 image made with a medium band redshifted
ly@xmath17 filter indicates the presence of a number of other compact galaxy candidates with redshifts near 2.4 .
the vla image of this field reached a limiting flux density of only @xmath6jy , but provides improved statistics for radio sources above this level .
in @xmath182 of this paper we describe the vla observations , calibrations and imaging for the two fields .
the radio and optical parameters for the sources are given in @xmath183 , and detailed radio / optical images with discussions are given in @xmath184 , along with discussions of the identifications .
in addition , using data from five deep 8.4-ghz surveys , we discuss the spectral properties , the source density and the optical properties of the micro - jansky radio sources .
further discussion is given in the final section .
the sa13 field was selected from among the hst medium deep survey fields to be devoid of strong radio sources , which would limit the sensitivity of a deep radio survey .
this field , centered at @xmath19 and @xmath20 , epoch j2000 , was observed with the vla for a total of 190 hours , of which we obtained 159 hours of good data ( more detail given below ) .
these data included 84 hours in the most compact d - configuration ( maximum baseline 1 km ) observed between 1993 october and 1994 january , and 75 hours in the c - configuration ( maximum baseline 3 km ) observed between 1994 november and 1995 january .
the earlier discussion in paper i of the optical counterparts in this field was based on the 1993/1994 d - configuration observations , but with the preliminary new c - configuration data .
the hercules field was observed for 12 hours in the vla c - configuration in 1996 february .
the center of this field is at @xmath21 and @xmath22 , epoch j2000 .
the observations were made at two frequencies , 8.415 ghz and 8.465 ghz , each with dual circular polarization with a bandwidth of 50 mhz . at this frequency the full - width half - power ( fwhp ) beamwidth of the vla 25-m antennas is @xmath23 .
we examined the area out to the 8% power level , at a radius of @xmath24 from the field center .
this corresponds to a solid angle of @xmath25 sr for each field .
each observing session lasted 8 - 10 hours and was split into 30-minute segments which included 27 minutes on the survey field and 3 minutes on a nearby calibrator source .
for the sa13 field , we used the unresolved calibrator source , j1244 + 4806 ( @xmath26 , @xmath27 ; j2000 ) , and for the hercules field , the calibrator source j1658 + 4749 ( @xmath28 , @xmath29 ; j2000 ) .
the position grids for the surveys are tied to these calibrators , both of which have a position known to better than @xmath30 .
the amplitude calibration was derived from observations of 3c286 made once per day ; we assumed a flux density of 5.19 jy at 8.44 ghz .
the data from each observing session were edited for occasional short periods of interference , excessive noise in individual correlators or other technical problems , antenna shadowing , and inclement weather conditions . in the sa13 field ,
less than one percent of the data was lost to interference and technical problems , but we had to discard about 30 hours of data due to poor weather , primarily snow , during 4 of the 27 observing sessions , each of 6 to 8 hours duration .
both fields were imaged and deconvolved with the aips tasks imagr using a pixel separation @xmath31 over a field of view of @xmath32 in order to detect radio sources well outside of the primary beam field of view which could seriously affect the image quality @xcite .
the visibility data were weighted ( aips terminology of robust=1 ) to obtain high signal to noise with low sidelobe levels and good resolution . for the sa13 field which contained both c and d configuration data ,
the full - width at half - power ( fwhp ) resolution was @xmath33 .
we also made an image with a resolution of @xmath34 to help resolve possible source blends .
for the hercules field with only c - configuration data , the resolution was @xmath35 .
the measured rms noise in the sa13 field was @xmath2jy , which makes this the most sensitive radio image yet obtained .
for the hercules field the rms noise was @xmath36jy .
the strongest source in the sa13 field has a measured map flux density of @xmath37jy .
this is less than one hundred times the thermal receiver noise limit , so dynamic range limitations are not an important factor . the hercules field contained a known ` bright ' radio source of 6 mjy at the field center , 53w002 @xcite , which was the target of the deep hst study @xcite .
artifacts from the 6-mjy source limited the dynamic range of the image , but self - calibration with a solution interval of 3 min had sufficient signal to noise to improve the calibration , and produce an image was signal - to - noise limited the completeness limit of a survey is that level above which there is near certainty that any source is real .
as discussed in previous vla papers on radio surveys with comparable resolution , crowdedness and field of view , a source with a peak flux density on the image @xmath38-@xmath16 has a probability of @xmath39 of being a true detection @xcite .
this percentage includes the effects of receiver noise and side - lobe contamination from faint sources .
the most negative peak flux density on the image is also a good indication of the detection limit @xcite .
we found that detection levels of @xmath3jy and @xmath40jy level for the sa13 and hercules fields , respectively , met both of the above criteria .
we also confirmed 19 of the 20 sources detected in paper i using similar criteria . in order to search for possible extended radio sources with low surface brightness
, we convolved the images to @xmath41 resolution for sources above the detection level of the highest resolution image , but above that for a lower resolution image .
one such extended source was found in the sa13 field ( source 27 in table 1 ) .
the radio images for the entire sa13 and the hercules fields are shown in figures 1 and 2 , respectively . in figure 3
we show the image of the central part of the sa13 field where the radio sensitivity is greatest .
the figure shows the excellent quality of the radio image , the precise alignment of most of the radio / optical identifications and the obvious faintness of the identifications for some of the sources .
the sources labeled with a number are given in table 1 .
seven weaker radio sources which are coincident with an optical object are labeled with a letter and given at the end of table 1 .
parameters for the 50 sources labeled in the sa13 field and the 10 sources labeled in the hercules field have entries listed in tables 1 and 2 , respectively .
thirty - four sources in the sa13 field were found in the complete sample with a peak flux density greater than @xmath3jy and within @xmath24 of the field center where the sensitivity drops to 8% of the on - axis sensitivity .
sixteen additional sources , not in the complete sample , are also included in the table .
all instrumental and resolution effects used to obtain the source parameters have been described elsewhere @xcite .
table 1 is organized as follows : columns 1 and 2 show the source numbers used in this paper and in paper i , respectively , to identify the sources more conveniently .
column 3 gives the source name .
an asterisk before the name indicates the fifteen sources which are * not * in the complete sample .
column 4 shows the total sky flux density and error estimate after correction for all instrumental and resolution effects including the primary beam attenuation of the telescopes .
column 5 shows the signal - to - noise on the @xmath33-resolution image and column 6 gives the deconvolved source angular size or limit in arcseconds .
the right ascension and declination , with rms errors are given in columns 7 and 8 , respectively .
the remaining four columns list the optical identification type and integrated magnitude in three bands .
further descriptions of the optical data are given in the next section .
eight faint radio sources , which are below our formal completeness level but have probable identifications with optical counterparts , are listed at the end of table 1 , and labeled @xmath42 through @xmath43 .
the probability that these radio sources is real is less than 30% based on the radio flux density alone .
however , their near coincidence with an optical counterpart increases their reality to 80% ; hence , only one of the eight sources is likely to be bogus @xcite .
there is good agreement between the source list in table 1 and that in paper i based on about one - half as much data , and at lower resolution . of the 20 sources above the previous detection level of @xmath44jy , 16 are in the complete sample of table 1 .
the four remaining sources from paper i are just below our @xmath3jy completion limit ; three of these objects are identified with galaxies .
the source list for the hercules field is given in table 2 .
the resolution is @xmath35 and the completeness limit is @xmath6jy .
ten sources are listed ; six of these are in the complete sample .
columns 1 through 7 are similar in content to that in table 1 . only two of the radio sources have been identified from an hst image which have been described elsewhere @xcite ; the optical information is given in column 8 .
the identification of radio sources in the sa13 field is based on two sets of optical material .
the hst - mds wf / pc images observed with two filters : v - band ( 5420 ) and i - band ( 8920 ) , respectively were used for the identification of radio sources near the center of the sa13 radio field . the calibration and processing of this optical data
was discussed in paper i. the identification limits are v=26.8 mag and i=25.8 mag with fwhp resolution of @xmath45 . for radio sources outside the mds field of view , identifications were made with the kpno 4-m telescopes with b - band ( 4420 ) and i - band ( 8920 ) filters .
the limiting magnitudes here were b=26.0 and i=25.5 mag , with seeing of about fwhm @xmath46 .
the observational details are discussed elsewhere @xcite .
the i - magnitude scale of the kpno images was calibrated to agree with that of the mds , with an estimated error of 0.2 mag .
the b - magnitude scale of the kpno images were calibrated using the extrapolation of the mds i- and v - magnitude scale , assuming a power law extrapolation of the intensity of the the flat - spectrum ( blue ) objects .
this extrapolation is accurate to an estimated error of 0.4 mag .
the precise registration of the kpno and the mds images with the radio grid was determined from the alignment of the high quality radio / optical coincidences to an accuracy of @xmath45 . only zero and first order corrections to the optical image were needed .
confidence in the registration of the images at this accuracy is important for the proper interpretation of the identifications .
the integrated optical magnitudes for all identifications in the sa13 field are given in table 1 .
the galaxy type is given in column 9 and the b- , v- and i - magnitudes are given in columns 10 , 11 and 12 , respectively .
the galaxy types are : g = galaxy of unknown morphology ; g?=uncertain identification of a very faint object ; ef = empty field ; g / b = binary galaxies or closely spaced galaxies , with the magnitude given for the objects containing the radio emission ; q = quasar : ell = elliptical galaxy ; sp = spiral galaxy . in figures 4a and
4b we show postage - stamp plots for all 42 radio sources in the sa13 field listed in table 1 .
in contrast to figure 3 , the green contour lines in figures 4a and 4b show the optical i - band distribution , while the location of the position of the radio source is indicated by the small black ellipses in each plot .
each field is @xmath47 on a side and centered at the position of the radio source .
most of the radio sources are unresolved ( typically @xmath48 ) and the black ellipses indicates the 1- and 2-@xmath16 limits for the position of the radio centroid .
all sources have a kpno i - band image fwhm @xmath46 seeing and the 22 sources nearest the field center also have an mds i - band image with fwhp @xmath45 resolution .
( see paper i for a discussion of the removal of the chromatic aberration from the pre - refurbished hst images . )
0.3truein * comments on the sa13 identifications : * * * no 1 .
j131157 + 423910 : * detected only on the i - band kpno image with i=23.5 mag .
the optical identification is too faint to be classified . * * no 2 .
j131157 + 423610 : * empty field . possible detection of an i=26 mag object . * * no 3 .
j131203 + 424030 : * detected only on the i - band kpno image with i=24.3 mag .
the identification is too faint to be classified . *
j131203 + 423331 : * identified with an isolated galaxy , somewhat extended in pa=@xmath49 . * * no 5 . j131205
+ 423851 : * empty field . * * no 6 .
j131207 + 423945 : * empty field . * * no 7 .
j131209 + 424217 : * identified with an isolated galaxy , possibly extended in pa=@xmath50 . *
j131211 + 424053 : * the radio emission is identified with the southern pair of two circular galaxies separated by @xmath51 .
the identified galaxy has ( b - i ) color = 3.7 mag and a size of about @xmath52 .
the northern galaxy , also circular , has about the same angular size , but is not unusually red .
its radio flux density is @xmath53jy , but lies below the formal detection level . * * no 9 .
j131213 + 423333 : * a weak radio source , not in the complete sample , which is identified with a bright galaxy with a faint companion @xmath52 to the north . another elliptically - shaped galaxy about @xmath34 south
is extended towards the radio source and may be interacting with the bright galaxy . because of the faintness of the radio source , the radio - optical position offset is uncertain , * * no 10 .
j131213 + 424129 : * identified with a bright galaxy with an extensive , blue halo . *
j131213 + 423706 : * a faint smudge is present on both the kpno and mds images with i@xmath54 mag . * * no 12 .
j131214 + 423821 : * a faint radio source below the completeness level is identified with a galaxy . both this galaxy and a fainter galaxy , about @xmath51 nw , are elongated in the same direction .
the radio emission is extended about @xmath51 , about the size of the galaxy disk .
the light profile of the galaxy is exponential , suggesting two interacting galaxies @xcite . * * no 13 .
j131214 + 423730 : * identified with a faint blue galaxy . it is too faint to be classified further . * * no 14 .
j131215 + 423702 : * identified with a bright , distorted galaxy with z=0.322 ( paper i )
. there may be a bridge of optical emission between this galaxy and a fainter galaxy @xmath51 to the west .
the galaxy shows a bulge population and a strong @xmath55 profile .
it is relatively blue in color and shows narrow emission lines . @xcite . * * no 15 .
j131215 + 423901 : * identified with a quasar with z=2.561 ( paper i ) . the radio source is slightly extended with an angular size of @xmath51 . *
j131216 + 423921 : * empty field associated with the bright radio source .
another faint radio source , about @xmath56 south with a flux density of @xmath57jy ( source b in table 1 ) , is probably identified with a i=24 galaxy .
the two sources are probably unrelated . * * no 17 . j131217 + 423912 : * identified with the brighter of a pair of two overlapping galaxies .
the fainter galaxy , @xmath5 to the nw is 1.5 mag fainter and is clearly distorted .
the velocity profile of the galaxy is exponential @xcite and the galaxy is very red . * * no 18 .
j131217 + 423930 : * empty field . * * no 19 .
j131218 + 423807 : * a faint radio source , not in the complete sample , is probably identified with a faint 25 mag galaxy or galaxies .
a brighter 23-mag galaxy about @xmath34 to the west appears unrelated . *
j131218 + 423843 : * identified with the brighter pair of two nearly overlapping galaxies . the fainter galaxy , @xmath5 to the se is 1.5 mag fainter and is is more distorted than the brighter galaxy , which is very red . *
j131219 + 423608 : * identified with a distorted galaxy .
the radio source is extended @xmath58 in position angle @xmath59 . from the higher resolution observations at 1.4 ghz @xcite ,
the radio emission is composed of two small - diameter sources whose positions are indicated by the two black ellipses in the image . since the two radio components lie in about the same orientation as the optical object , they are probably both associated with the galaxy .
the cause of this non - nuclear radio emission is unknown .
the radio objects may be a small compact double source , resembling a fr-1 morphology , or they maybe associated with individual galactic objects far from the galactic center . * * no 22 .
j131219 + 423631 : * the higher resolution observations at 1.4 ghz suggest that this extended radio source is probably composed of two separated radio sources @xcite .
the fainter radio component to the nw is coincident with a bright , elliptically - shaped galaxy .
the brighter radio component to the se is unidentified with a limit of i=26 mag .
we believe these two radio components are unrelated , although the unidentified source could be associated with a galactic component in the halo of the bright galaxy .
the light profile of the galaxy is exponential , indicating a disk - dominated galaxy @xcite . * * no 23 .
j131220 + 423923 : * empty field .
however , this radio sources is only @xmath60 west of the bright 18-m galaxy associated with j131221 + 423923 ( no .
27 ) @xcite .
the question arises if this radio source is associated with a galactic component in the halo of this bright galaxy ? * * no 24 .
j131220 + 423704 : * empty field . * * no 25 .
j131220 + 424029 : * empty field . * * no 26 .
j131220 + 423535 : * empty field . * * no 27 .
j131221 + 423923 : * identified with an 18-m barred spiral galaxy with z=0.302 ( paper i ) .
the radio source is about @xmath60 in size , contained within the disk of the galaxy .
the galaxy disk is distorted ; it is blue in color and there are narrow emission lines @xcite .
* * no 28 .
j131221 + 423722 : * identified with an 18-m , edge - on , barred spiral galaxy with z=0.180 ( paper i ) .
the galaxy has a relatively blue color and shows weak emission lines .
more details are given elsewhere @xcite .
* * no 29 .
j132121 + 423827 : * identified with the middle optical peak in a chain of three or four galaxies comprising one of the most distorted optical regions associated with a detected radio source . the light profile of the galaxy is exponential , indicating a disk galaxy . @xcite .
* * no 30 .
j131222 + 423813 : * identified with a quasar with z=2.561 , i.e. the same redshift for j131315 + 423901 ( no 15 ) .
they are probably in the same distant cluster or group of objects . with a separation of @xmath61 they are unlikely to be a gravitational - lens pair . * * no 31 .
j131223 + 423908 : * identified with the brighter of a pair of galaxies , separated by @xmath33 . the brighter galaxy is somewhat distorted and the fainter galaxy is slightly extended towards the brighter .
the radio emission is also slightly extended , although well within the disk of the brighter galaxy .
the velocity profile of the galaxy is exponential and its color is very red @xcite . * * no 32 .
j131223 + 423712 : * identified with the brighter of a pair of overlapping galaxies , separated by @xmath52 .
two other overlapping galaxies lie @xmath60 to the east , and all four may be connected by a faint optical bridge .
the radio emission is coincident with the galactic nucleus of the bright galaxy and is @xmath62 in angular size .
higher resolution observations at 1.4 ghz @xcite give a size of @xmath46 in pa=@xmath59 . this radio source may be associated with an interacting system of galaxies .
the redshift of both bright galaxies is 0.401 ( paper i ) .
the galaxy containing the radio source has little significant bulge population , contains weak emission lines and is rather red @xcite .
* * no 33 .
j131223 + 423525 : * identified with a bright , somewhat elliptical galaxy with a large distorted halo . * * no 34 . j131224 + 423804 : * a faint radio source , not in the complete sample , which is identified with a faint object or objects . however , the identification is too faint to be further categorized . * * no 35 .
j131225 + 424656 : * possibly identified with a i=25 mag object , although the radio / optical offset is larger than expected .
* * no 36 .
j131225 + 424103 : * a relatively bright radio source which lies @xmath52 south of the i=22 mag ir galaxy and @xmath34 north of an i=20 mag asymmetric galaxy . the radio source may be associated with a faint extension south from the ir galaxy with i=25.5 mag .
the faint optical extensions associated with the two bright galaxies indicate they are probably interacting with each other . * * no 37 .
j131225 + 423941 : * a bright radio source which may be associated with an object just at the detection limit .
it is @xmath46 north of an i=23.5 mag galaxy .
this radio source lies near the mds field edge .
* * no 38 .
j131226 + 424227 : * identified with an i=18 mag , slightly elliptical galaxy .
the radio source is not in the complete sample . * * no 39 .
j131227 + 423800 : * identified with a i=18mag edge - on galaxy which is @xmath60 sse of an i=17 mag elliptical galaxy ( the contours are distorted near this galaxy ) .
these two galaxies overlap , although there is no appreciable distortion ; hence , their interaction is not certain .
the light profile of the galaxy is exponential and its color is unusually red , possibly a dusty , disk galaxy @xcite . * * no 40 .
j131232 + 424038 : * identified with an i=24.8 mag object which may be extended in the east - west direction . the identification is too faint to be further classified . *
j131236 + 424027 : * this radio source is displaced in pa=@xmath63 from the nucleus of this i=18 mag galaxy .
it is not in the complete radio sample so the four - sigma radio / optical offset might be spurious .
the galaxy is elliptical and symmetric with little distortion .
however , the galaxy orientation is also in pa=@xmath63 , which somewhat supports the radio / optical offset . * * no 42 .
j131239 + 423911 : * identified with an i 24.5 mag galaxy which is too faint to be further classified . *
* weak source a. j131213 + 423946 : * identified with an i=20.0 mag circular galaxy .
another galaxy with i=22.5 mag lies @xmath34 to the south - west . * * weak source b. j131213 + 423826 : * identified with a faint galaxy which is too faint to be further classified . * * weak source c. j131215 + 423913 : * identified with a diffuse , extended galaxy . * * weak source d. j131216 + 423920 : * radio source is @xmath52 east of an i=23.8 mag , somewhat extended , galaxy . * * weak source e. j131219 + 423932 : * radio source near an i=24.0 mag galaxy .
brighter galaxies lie @xmath52 to the south - west and @xmath51 to the north - east . * * weak source f. j131221 + 423836 : * companion galaxy lies @xmath51 east - north - east . * * weak source g. j131225 + 423742 : * bright elliptical galaxy with i=18.0 mag .
radio source may be extended in the same direction . * * weak source h. j131225 + 423907 : * identified with a slightly extended i=22.5 mag galaxy .
we have gathered data from five deep vla integrations at 8.4 ghz in order to study the radio spectral properties , the optical properties and the source count of the micro - jansky radio population .
parameters for the five surveys made over the past decade are summarized in table 3 , which includes references to lower frequency radio data and optical data as well .
column 5 gives the integration time in each field , and column 6 gives the complete survey detection level .
column 7 , 8 and 9 show the total number of sources listed in each survey , the number of sources in the complete sample above the detection level , and the number of sources expected based on the best source count described below ; respectively .
although the five catalogs contain a total of 201 sources , only 89 of them are in a complete flux - limited 8-ghz sample , and only these are listed in table 4 .
the table contains the source properties that are discussed in this section .
the last column indicates the detection of x - ray emission from the sources in the hubble and sa13 fields @xcite accurate spectral indices of the 89 sources have been determined by comparing our results to observations at frequencies lower than 8.4 ghz ( see references at the end of table 3 ) .
only seven of them are lower limits ( one in the sa13 field and six in the hubble fields ) . in figure 5
we show four histogram comparison of the spectral index distributions associated with various radio and optical properties . in figure 5a
we show the spectral index distribution for all 89 sources and separately for the 34 sources in the sa13 field .
there is no significant difference between the two distributions . in the total distribution , the spectral range of the seven sources which were not detected above twice the 1.4-ghz noise level upper limits on the spectral index , are indicated by the arrows in the histogram .
since the observations at the various frequencies were made at different times , source variability could affect the spectral index calculation , but variability of micro - jansky sources is not common @xcite .
about one - quarter of the sources have a spectral index in the range @xmath64 .
the distribution drops precipitously at lower spectral indices ( steeper spectra ) , but more slowly for high spectral indices ( flatter spectra ) .
the median value is about @xmath65 .
only a few objects have an inverted spectral index with @xmath66 , although some sources with only a 1.4 ghz flux density upper limit could fall in this category .
the spectral distribution shows a higher proportion of steep spectrum objects than reported from less sensitive , 8.4-ghz micro - jansky surveys where the median spectral index was @xmath67 @xcite .
the reason for this difference is illustrated in figure 5b where the spectral distribution is shown for 46 sources with s@xmath68jy and for 43 weaker sources with s@xmath69jy .
_ the lower flux density sources are more likely to have a steep spectral index than the high flux density sources . _
the likelihood that the distribution of two samples chosen randomly would differ by at least this much is only 18% .
this change explains why a flatter average spectrum was found in the less sensitive , earlier , 8.4 ghz surveys .
possible causes of the change of spectral properties below @xmath6jy are given in @xmath186 .
figure 5c shows that those objects which are fainter than i=25.5 mag are predominantly steep radio spectrum objects .
although there are only 17 sources in this faint galaxy sample ; no object has a spectral index flatter than @xmath70 and most are steeper than @xmath71 .
figure 5d is a plot of the spectral index distribution for the 41 galaxies which have measured colors . as a discriminatory of the color for the sa13 field we used the ( b - i ) magnitudes ; for the hubble field we used the [ i-(h+k)/2 ] magnitudes . in this sample of 41 sources ,
19 galaxies are significantly reddened and their spectral index distribution is also shown .
the slight difference in the radio spectral index between the two groups suggests that fewer of reddened objects have a flat spectrum .
this trend was also found from vla observations at 5 ghz and 1.4 ghz and from the canadian - france redshift survey @xcite .
this correlation , although weaker than that in figure 5c , is compatible since reddened objects would tend to have relatively faint i - band emission .
nine of the sources in table 4 are identified with x - ray sources @xcite .
two are in the sa13 field .
in contrast , seven x - ray identifications have been made in the hubble field and these associations show no trends in radio spectral index or optical counterpart , and are identified with galaxies of various types .
the density of sources ( source count ) is generally specified as @xmath72 where @xmath73 is the number of sources per solid angle per flux density interval @xmath74 at a flux density @xmath75 . for a flat space which
is uniformly filled with radio sources and ignoring the attenuation effects of redshift , the slope @xmath76 . in order to determine the two parameters , @xmath77 and @xmath78 , from the 89 source sample determined from five surveys with different sensitivities
, we used a modification of the maximum likelihood method ( mlm ) originally developed for a uniformly obtained sample @xcite .
for a range of test models , we chose reasonable parameters , @xmath77 and @xmath78 , and calculated the relative probability that this model would produce the observed flux density distribution of the 89 sources , using the appropriate constraints for each survey .
the best fit model was the one with the highest probability , and the error estimate was determined by the set of models that were at least 37% ( 1/e ) as likely as the best model .
the number of micro - jansky sources , @xmath73 ( arcmin@xmath79jy@xmath80 ) , is given by @xmath81 the corresponding integral count , @xmath82 ( arcmin@xmath83 is @xmath84 where n is the number of sources above a flux density @xmath75 in micro - jansky . the value , 40 @xmath9jy , is the mean flux density of the sources , and in this form , the two parameters @xmath77 and @xmath78 are nearly orthogonal . above @xmath85jy the source density is 0.1 ( arcmin)@xmath11 , or @xmath86 sources sr@xmath80 . at the minimum detection level of these surveys , @xmath3jy , there are 0.64 sources ( arcmin)@xmath11 , or @xmath87 sources sr@xmath80 .
extrapolation to weaker flux density with the same slope gives 6 sources ( arcmin)@xmath11 above @xmath88jy , corresponding to an average source separation of @xmath89 .
the density of sources is still far from the natural confusion limit when sources begin to blend with each other .
if @xmath90 is the proportion of sky covered with emission , significant blending occurs for @xmath91 . if we assume an average source size of @xmath52 .
then then @xmath92 . extrapolating to @xmath88jy and @xmath93jy gives @xmath94 and @xmath95 , respectively .
unless the average angular size of most sources is less than @xmath96 , radio source blending will become a problem below about @xmath97jy .
recent results from merlin , however , shows that some faint radio sources contain a significant part of their emission within an angular size of @xmath4 @xcite . the plot of the _ integral _ source count at 8.4 ghz is shown in figure 6 .
the ordinate scale has been normalized to the euclidean integral slope of @xmath98 . at any point in the plot , the ordinate value times s@xmath99 gives the number of sources ( arcmin)@xmath11 with a flux density greater than @xmath100jy . the count derived from the mlm calculation ,
describe above , is shown by the solid line .
this line extends to @xmath88jy , although we can not exclude that the slope of the counts change below our detection limit of @xmath3jy .
the plotted points at the micro - jansky levels have been obtained by gridding the data in table 4 into five flux density bins .
the results of this binned data are shown in table 5 .
the mlm fit is clearly compatible with the gridded data , as it should be . the number of sources in each of the five bins listed in column 4 agrees reasonably well with the number predicted by the mlm fit , given in column 5 .
we have added in the count of sources at 8.4 ghz at high flux densities @xcite .
there are no extensive surveys of sources at 8.4 ghz of intermediate flux density sources between one and eight milli - janskys .
the source count at 8.4 ghz derived at micro - jansky levels is consistent with that at the milli - jansky and jansky levels .
extrapolation of the micro - jansky count falls slightly under the milli - jansky points , the excess being produced by the rapid evolution of the brightest radio sources in the sky .
the dotted line shows the best fit count obtained from the lynx and cepheus fields only @xcite which contained only 20 sources to a detection level of @xmath101jy . the ` old ' count and
the ` new ' count differ at the 1.4-@xmath16 level near @xmath85jy
. there may be several reasons for the discrepancy : first , the difference may be real and reflect the anisotropic distribution of sources in the sky on scales of tens of arcmin .
such differences between the number of sources among radio fields have already been noted by others at various frequencies ( windhorst et al .
1985 ; windhorst et al . 1993
; hopkins et al . 1998 ; hopkins et al . 1999 ; richards 2000 ) .
second , the optical and radio selection criteria for the various fields were not the same .
the hubble and sa13 fields were deliberately selected to be radio quiet above 0.5 mjy , whereas the hercules field contained a known radio source of about 10 mjy .
the optical constraints for the hdf selection ( no relatively bright objects ) may also be relevant .
however , since most of the radio sources and statistics come from sources which are in the 10 to 200 @xmath9jy flux density range at 8 ghz , the systematic effect of the avoidance of radio fields with sources stronger than 500 @xmath9jy is unclear .
third , the lynx and cepheus fields were observed with lower resolution than the hubble and sa13 fields .
although the effects of source blending were considered for the lynx and cepheus fields @xcite , the blending may be more important than originally thought .
the average angular size of the sources in these two fields were also larger than in other fields , and perhaps a consequence of the blending of weak sources .
regardless of the reasons for the somewhat different count of sources in the lynx and cepheus fields , the new results are based on more than four times the number of sources found in five different areas and with higher resolution . when extrapolated to @xmath88jy ,
the new counts give a density of @xmath102 ( arcmin)@xmath11 , compared with the previous value of @xmath103 ( arcmin)@xmath11 . this difference is about 1.8-@xmath16 and should not be regarded as significant .
the average sky brightness at 8.4 ghz , contributed by sources with flux densities above @xmath3jy is @xmath104k . extrapolating to weaker sources
, the sky brightness will increase as s@xmath105 , assuming a constant slope @xmath78 .
using @xmath106 , the estimated sky brightness from discrete sources increases by only a factor of 2 for a decrease of 1000 in flux density .
hence , the sky brightness at 8.4 ghz contributed by discrete sources down to the nano - jansky level is unlikely to exceed 0.001 k unless there is a new population of weaker sources , allowing the slope @xmath78 to become more negative ( i.e. steeper ) than @xmath107 .
figure 7 shows the distribution of the i - magnitudes of the 63 sources in the complete samples in the sa13 field plus the hubble deep and flanking fields . the density of sources with radio identifications and optical magnitudes between i=17.5 and 20.5 mag is about 30% of the density of all galaxies within this magnitude range @xcite . for galaxies fainter than i=20.5 mag
the distribution remains relatively constant per magnitude interval to i=26.0 mag , with about 20% of the sources unidentified below this magnitude .
this distribution is similar to that found in other investigations of the identification of weak radio sources @xcite .
analysis of several hst fields shows that 34%@xmath1089% of all galaxies brighter than i=23.3 mag are in binary systems .
this is much higher than the 7% found for the galaxy population brighter than i=15 mag , which comprise the local population @xcite .
thus , galaxy interaction appears to be a strong function of cosmological time , and such galaxy interactions may be the main cause of the strong evolution of radio sources with redshift @xcite .
this connection between galaxy interaction and radio emission is supported in these data . for the complete sample of 63 radio sources in the sa13 and hubble fields ,
37 are identified with galaxies brighter than i=23.3 mag .
seventeen of these identifications are classified as _
g / b _ meaning that at least one other galaxy lies within @xmath60 of the radio galaxy , possibly indicating galaxy interactions .
thus , 46% of these radio identifications are associated with multiple systems .
this percentage is somewhat larger than the 34% of galaxies that are found in multiple systems below @xmath109 mag , and suggests that the radio emission is even more snhanced by galaxy interactions .
a more complete analysis of the optical properties of the radio sources in the sa13 and hubble fields is given elsewhere @xcite .
we have reported the results from two new 8.4 ghz vla surveys of micro - jansky radio sources . in one of the fields , sa13 ,
the rms noise is @xmath2jy and we have detected 49 sources of which 34 are in a complete sample with a flux density above @xmath3jy .
six sources were detected above @xmath6jy in the hercules field . in tables 1 and 2
we present the radio catalog for these two fields .
comments and a detailed radio / optical comparison are given for 42 sources in the sa13 field using deep hst and kpno images .
the results from these two fields double the number of 8.4-ghz radio sources now known at micro - jansky levels . in the second part of this paper
, we have compiled a list of 89 sources in a flux density complete sample from five deep vla surveys at 8.4 ghz .
this source list is dominated by observations of the sa13 field ( detection level of @xmath3jy ) and the hubble deep field plus the flanking fields ( detection level of @xmath110jy ) .
we have used other optical and radio data in order to determine the properties of these weak radio sources .
the count of radio sources follows an approximate power - law with a slope @xmath111 from 10 mjy to less than @xmath112jy , with good continuity to the count at the jansky level ( see figure 6 ) .
there are no significant differences in the number of sources detected in each field , as shown by the comparison of columns 8 and 9 in table 3 , which indicate the number of sources detected versus the number expected from the best fit to the count .
the average sky brightness temperature contributed by the micro- and nano - jansky radio sources is extrapolated to be less than 0.001 k as long as the slope @xmath78 does not become steeper than @xmath107 .
most of the sources are unresolved with @xmath33 or even @xmath34 resolution .
for the 63 sources in the hubble and sa13 samples , 15 sources ( 24% ) are larger than about @xmath51 , and only four sources are larger than @xmath60 .
thus , the angular size of the micro - jansky sources at 8.4 ghz appears statistically somewhat smaller than the angular size at 1.4 ghz @xcite .
the median spectral index of these 15 extended sources is @xmath113 , steeper than than our general sample and consistent with an increasing angular size with lower frequency .
the correlation of large angular size with steeper spectral index is seen for all radio sources , regardless of their flux density . the combination of emission from opaque small - diameter components and large components showing an aging synchrotron ( steep ) spectrum probably cause this difference in the correlation of angular size and spectral index at the two different frequencies .
higher resolution observations are needed in order to resolve the majority of the micro - jansky radio sources @xcite .
the radio sources are identified with a variety of galaxy types .
only two of the radio sources are identified with quasars and none with stars .
we estimate that about 30% of all galaxies brighter than i=21 mag can be detected at the @xmath112jy level at 8.4 ghz . for fainter magnitudes ,
the radio source identifications are spread relatively evenly , per magnitude interval , with about 20% of the radio sample fainter than i=26 mag .
the centroid of the radio emission is nearly always located within @xmath4 of the galaxy nucleus .
the few exceptions in the sa13 field are : j131220 + 423923 is an empty field which is displaced @xmath60 from an i=18-mag galaxy ; j131219 + 423608 contains two radio components each displaced about @xmath52 on opposite sides of a i=22 mag galaxy nucleus and is possibly a small double radio source ( see notes ) ; j131225 + 424103 is displaced @xmath52 south of an i=22 mag galaxy ( see notes ) ; j131239 + 423911 is displaced @xmath96 from the galaxy centroid , but well within the disk component .
the likelihood that a micro - jansky radio source will be associated with a galaxy , between i=18 to 22 mag with at least one neighbor closer than @xmath60 , is nearly 50% .
this percentage is somewhat larger than the 34% of all such galaxies which are in groups .
this suggests that radio source luminosity and/or density evolution is enhanced somewhat within galaxy groups which may be interacting .
the spectral index distribution of the micro - jansky radio sources at 8.4 ghz shows one clear trend ; the spectral index distribution steepens for sources below @xmath6jy ( see figure 5b ) .
somewhat less conclusively , the spectral index distribution may also steepen for sources with optical counterparts fainter than i=25.5 mag ( see figure 5c ) .
the change in the source count and spectral properties of radio sources fainter than 1 mjy is well - documented . below this level , star - forming galaxies begin to dominate the source population over agn .
the cause for a further change in spectral properties less than @xmath6jy may be associated with the different evolution of agn and star - burst galaxies , but a definitive answer must await for additional observations of the higher energy photons emitted from these fainter radio sources , and measurement of their radio size and internal complexity .
both the agn and the star - burst phenomena produce a wide range in spectral index , and can not be used as a good discriminator between the two types .
a steepening of the spectral index below @xmath6jy could arise in several ways : ( 1 ) the opaque emission ( always with flat spectrum ) from agn components in weak sources may be a smaller proportion of the total emission ; ( 2 ) the star - burst phenomenon may produce steeper radio spectra for the weak sources , possibly because of synchrotron aging associated with larger magnetic fields in the weaker , distant sources .
the national radio astronomy observatory is a facility of the national science foundation , operated under cooperative agreement by associated universities , inc .
the space telescopes science institute is operated by associated universities for research in astronomy , inc .
, under contract to nasa .
rbp was supported in part by nsf grant ast96 - 16971 and by the keck northeast astronomy consortium .
ear was supported by a hubble fellowship . 1 & & j131157@xmath114423910 & 109.3 @xmath108 8.1 & 14.8 & @xmath115 & 13 11 57.499 @xmath108 0.02 & 42 39 10.16 @xmath108 0.21 & g & @xmath11626.0 & @xmath117 & @xmath118 23.5 + 2 & & j131157@xmath114423630 & 196.6 @xmath108 10.2 & 23.2 & @xmath115 & 13 11 57.903 @xmath108 0.01 & 42 36 30.15 @xmath108 0.14 & ef & @xmath11626.0 & @xmath117 & @xmath119 25.5 + 3 & & * j131203@xmath114424030 & 213.2 @xmath108 8.7 & 36.7 & @xmath115 & 13 12 03.092 @xmath108 0.01 & 42 40 30.68 @xmath108 0.10 & g & @xmath11725.5 & @xmath117 & @xmath118 24.3 + 4 & & j131203@xmath114423331 & 886.1 @xmath108 60.9 & 15.6 & @xmath115 & 13 12 03.508 @xmath108 0.02 & 42 33 31.72 @xmath108 0.20 & g & @xmath11724.0 & @xmath117 & @xmath118 22.2 + 5 & & j131205@xmath114423851 & 12.4 @xmath108 2.6 & 5.0 & @xmath120 & 13 12 05.596 @xmath108 0.06 & 42 38 51.15 @xmath108 0.60 & ef & @xmath11626.0 & @xmath117 & @xmath119 25.5 + 6 & & j131207@xmath114423945 & 14.0 @xmath108 2.8 & 5.4 & @xmath121 & 13 12 07.645 @xmath108 0.06 & 42 39 45.87 @xmath108 0.56 & ef & @xmath11626.0 & @xmath117 & @xmath119 25.5 + 7 & & j131209@xmath114424217 & 93.4 @xmath108 17.8 & 5.8 & @xmath121 & 13 12 09.068 @xmath108 0.05 & 42 42 17.49 @xmath108 0.52
& g & @xmath11725.5 & @xmath117 & @xmath118 23.4 + 8 & & j131211@xmath114424053 & 717.0 @xmath108 21.9 & 183.0 & @xmath115 & 13 12 11.029 @xmath108 0.01 & 42 40 53.66 @xmath108 0.05 & g / b & @xmath11724.5 & @xmath117 & @xmath118 20.8 + 9 & & * j131213@xmath114423555 & 12.2 @xmath108 2.6 & 4.7 & @xmath120 & 13 12 13.303 @xmath108 0.07 & 42 35 55.76 @xmath108 0.64 &
g / b & @xmath11721.7 & @xmath117 & @xmath118 19.5 + 10 & & j131213@xmath114424129 & 48.4 @xmath108 5.6 & 9.2 & @xmath121 & 13 12 13.446 @xmath108 0.04 & 42 41 29.44 @xmath108 0.33 & ell & @xmath11721.9 & @xmath117 & @xmath118 18.7 + 11 & & j131213@xmath114423706 & 9.4 @xmath108 1.7 & 5.6 & @xmath121 & 13 12 13.858 @xmath108 0.06 & 42 37 06.50 @xmath108 0.53 & g ?
& @xmath117 & @xmath11626.8 & @xmath118 25.5 + 12 & 1 & * j131214@xmath114423821 & 6.5 @xmath108 1.6 & 4.3 & @xmath122 & 13 12 14.518 @xmath108 0.08 & 42 38 21.86 @xmath108 1.02 & g / b & @xmath11723.7 & @xmath11723.4 & @xmath118 21.8 + 13 & 2 & * j131214@xmath114423730 & 6.2 @xmath108 1.6 & 4.4 & @xmath120 & 13 12 14.666 @xmath108 0.09 & 42 37 30.96 @xmath108 1.13 & g & @xmath117 & @xmath11726.3 & @xmath118 25.2 + 14 & 3 & j131215@xmath114423702 & 13.4 @xmath108 1.7 & 7.8 & @xmath121 & 13 12 15.130 @xmath108 0.04 & 42 37 02.70 @xmath108 0.39 & g / b & @xmath11721.4 & @xmath11720.7 & @xmath118 19.7 + 15 & 4 & j131215@xmath114423901 & 26.6 @xmath108 1.8 & 14.9 & @xmath123 & 13 12 15.280 @xmath108 0.02 & 42 39 01.16 @xmath108 0.21 & q & @xmath11718.6 & @xmath11718.4 & @xmath118 17.8 + 16 & 5 & j131216@xmath114423921 & 29.5 @xmath108 2.0 & 15.4 & @xmath121 & 13 12 16.083 @xmath108 0.02 & 42 39 21.47 @xmath108 0.20 & ef & @xmath117 & @xmath11626.8 & @xmath119 25.8 + 17 & 6 & j131217@xmath114423912 & 20.1 @xmath108 1.8 & 11.7 & @xmath115 & 13 12 17.181 @xmath108 0.03 & 42 39 12.15 @xmath108 0.26 & g / b & @xmath11722.5 & @xmath11722.2 & @xmath118 20.0 + 18 & 7 & j131217@xmath114423930 & 15.5 @xmath108 1.9 & 8.6 & @xmath121 & 13 12 17.601 @xmath108 0.04 & 42 39 30.49 @xmath108 0.35 & ef & @xmath117 & @xmath11626.8 & @xmath119 25.8 + 19 & 8 & * j131218@xmath114433907 & 6.4 @xmath108 1.6 & 4.3 & @xmath120 & 13 12 18.311 @xmath108 0.09 & 42
39 07.79 @xmath108 1.07 & g ? & @xmath117 & @xmath11626.8 & @xmath118 25.2 + 20 & 9 & j131218@xmath114423843 & 25.0 @xmath108 1.7 & 14.8 & @xmath124 & 13 12 18.437 @xmath108 0.02 & 42 38 43.92 @xmath108 0.21 & g / b & @xmath11724.0 & @xmath11723.1 & @xmath118 20.3 + 21 & & j131219@xmath114423608 & 26.4 @xmath108 2.4 & 11.4 & @xmath125 & 13 12 19.838 @xmath108 0.03 & 42 36 08.91 @xmath108 0.27 & g & @xmath11723.8 & @xmath117 & @xmath118 22.1 + 22 & 10 & j131219@xmath114423831 & 11.6 @xmath108 1.6 & 6.7 & @xmath126 & 13 12 19.932 @xmath108 0.05 & 42 38 31.39 @xmath108 0.45 & g & @xmath11723.0 & @xmath11722.2 & @xmath118 20.4 + 23 & 12 & j131220@xmath114423923 & 14.5 @xmath108 1.9 & 7.7 & @xmath121 & 13 12 20.016 @xmath108 0.04 & 42 39 23.97 @xmath108 0.39 & ef ? & @xmath117 & @xmath11626.8 & @xmath119 25.8 + 24 & 11 & j131220@xmath114423704 & 13.9 @xmath108 1.7 & 8.4 & @xmath121 & 13 12 20.185 @xmath108 0.04 & 42 37 04.02 @xmath108 0.36 & ef & @xmath117 & @xmath11626.8 & @xmath119 25.8 + 25 & & j131220@xmath114424029 & 33.7 @xmath108 3.0 & 11.5 & @xmath115 & 13 12 20.236 @xmath108 0.03 & 42 40 29.67 @xmath108 0.26 & ef & @xmath11626.0 & @xmath117 & @xmath119 25.5 + 26 & & j131220@xmath114423535 & 20.4 @xmath108 3.0 & 6.7 & @xmath121 &
13 12 20.871 @xmath108 0.05 & 42 35 35.33 @xmath108 0.45 & ef & @xmath11626.0 & @xmath117 & @xmath119 25.5 + 27 & 13 & j131221@xmath114423923 & 11.5 @xmath108 1.9 & 5.4 & @xmath124 & 13 12 21.109 @xmath108 0.06 & 42 39 23.49 @xmath108 0.55 & sp & @xmath11720.2 & @xmath11719.2 & @xmath118 18.0 + 28 & 15 & j131221@xmath114423722 & 10.0 @xmath108 1.7 & 5.9 & @xmath121 & 13 12 21.397 @xmath108 0.05 & 42 37 22.91 @xmath108 0.51 & sp & @xmath11720.0 & @xmath11719.0 & @xmath118 17.8 + 29 & 16 & j131221@xmath114423827 & 18.5 @xmath108 1.7 & 10.2 & @xmath121 & 13 12 21.839 @xmath108 0.03 & 42 38 27.55 @xmath108 0.30
& g / b & @xmath117 & @xmath11724.3 & @xmath118 22.7 + 30 & 17 & j131222@xmath114423813 & 10.4 @xmath108 1.6 & 5.6 & @xmath127 & 13 12 22.428 @xmath108 0.06 & 42 38 13.47 @xmath108 0.54 & q & @xmath11721.3 & @xmath11721.0 & @xmath118 19.6 + 31 & 18 & j131223@xmath114423908 & 19.3 @xmath108 2.0 & 9.2 & @xmath128 & 13 12 23.298 @xmath108 0.04 & 42 39 08.35 @xmath108 0.33
& g / b & @xmath11722.6 & @xmath11721.6 & @xmath118 19.3 + 32 & 19 & j131223@xmath114423712 & 27.7 @xmath108 2.0 & 15.2 & @xmath115 & 13 12 23.693 @xmath108 0.02 & 42 37 12.09 @xmath108 0.20 & g / b & @xmath11721.3 & @xmath11720.6 & @xmath118 18.8 + 33 & & j131223@xmath114423525 & 65.5 @xmath108 4.1 & 17.7 & @xmath115 & 13 12 23.999 @xmath108 0.02 & 42 35 25.57 @xmath108 0.18 & ell & @xmath11722.4 & @xmath117 & @xmath118 18.9 + 34 & 20 & * j131224@xmath114433804 & 6.2 @xmath108 1.6 & 4.2 & @xmath120 & 13 12 24.133 @xmath108 0.09 & 42
38 04.89 @xmath108 1.02 & g & @xmath117 & @xmath11725.5 & @xmath118 24.0 + 35 & & j131225@xmath114423656 & 12.9 @xmath108 2.1 & 6.5 & @xmath121 & 13 12 25.201 @xmath108 0.05 & 42 36 56.66 @xmath108 0.47 & g ? & @xmath11626.8 & @xmath117 & @xmath118 25.0 + 36 & & j131225@xmath114424103 & 34.9 @xmath108 4.8 & 6.7 & @xmath121 & 13 12 25.228 @xmath108 0.05 & 42 41 03.44 @xmath108 0.45 & g ?
& @xmath11626.0 & @xmath117 & @xmath118 25.5 + 37 & 23 & j131225@xmath114423941 & 200.3 @xmath108 6.5 & 79.3 & @xmath115 & 13 12 25.743 @xmath108 0.01 & 42 39 41.58 @xmath108 0.06 & g / b & @xmath11626.5 & @xmath117 & @xmath118 25.5 + 38 & & * j131226@xmath114424227 & 318.3 @xmath108 25.3 & 13.0 & @xmath115 & 13 12 26.287 @xmath108 0.03 & 42 42 27.43 @xmath108 0.24 & ell & @xmath11720.9 & @xmath117 & @xmath118 17.7 + 39 & 24 & j131227@xmath114423800 & 25.6 @xmath108 2.3 & 11.6 & @xmath115 & 13 12 27.532 @xmath108 0.03 & 42 38 00.23 @xmath108 0.26 & g / b & @xmath11722.1 & @xmath11722.0 & @xmath118 19.8 + 40 & & j131232@xmath114424038 & 165.7 @xmath108 8.9 & 20.5 & @xmath120 & 13 12 32.797 @xmath108 0.02 & 42 40 38.42 @xmath108 0.15 & g & @xmath11626.0 & @xmath117 & @xmath118 24.8 + 41 & & * j131236@xmath114424027 & 54.6 @xmath108 11.5 & 4.8 & @xmath115 & 13 12 36.044 @xmath108 0.07 & 42 40 27.66 @xmath108 0.63 & g & @xmath11720.2 & @xmath117 & @xmath118 18.3 + 42 & & j131239@xmath114423911 & 66.2 @xmath108 12.7 & 5.1 & @xmath115 & 13 12 39.705 @xmath108 0.06 & 42 39 11.32 @xmath108 0.60 & g & @xmath11625.0 & @xmath117 & @xmath118 24.5 + a & & * j131213@xmath114423932 & 4.2 @xmath108 1.7 & 2.5 & @xmath129 & 13 12 13.521 @xmath108 0.26 & 42 39 32.16 @xmath108 1.33 & g & @xmath117 & @xmath117 & @xmath118 20.0 + b & & * j131213@xmath114423826 & 4.8 @xmath108 1.6 & 2.7 & @xmath129 & 13 12 13.583 @xmath108 0.22 & 42 38 26.68 @xmath108 1.13 & g & @xmath117 & @xmath117 & @xmath118 24.5 + c & & * j131215@xmath114423913 & 6.7 @xmath108 1.6 & 3.8 & @xmath120 & 13 12 15.847 @xmath108 0.12 & 42 39 13.22 @xmath108 0.70 & g & @xmath117 & @xmath117 & @xmath118 23.2 + d & & * j131216@xmath114423920 & 4.8 @xmath108 1.6 & 2.8 & @xmath130 & 13 12 16.521 @xmath108 0.25 & 42 39 20.40 @xmath108 2.20 & g & @xmath117 & @xmath117 & @xmath118 23.8 + e & & * j131219@xmath114423932 & 5.0 @xmath108 1.9 & 2.9 & @xmath130 & 13 12 19.092 @xmath108 0.22 & 42 39 32.20 @xmath108 1.80 &
g / b & @xmath117 & @xmath117 & @xmath118 24.0 + f & & * j131221@xmath114423836 & 4.8 @xmath108 1.8 & 2.8 & @xmath127 & 13 12 21.794 @xmath108 0.13 & 42 38 36.96 @xmath108 1.40 & g & @xmath117 & @xmath117 & @xmath118 22.4 + g & & * j131225@xmath114423742 & 5.5 @xmath108 1.8 & 3.3 & @xmath131 & 13 12 25.065 @xmath108 0.17 & 42 37 42.33 @xmath108 1.50 & g & @xmath117 & @xmath117 & @xmath118 18.0 + h & & * j131225@xmath114423907 & 5.0 @xmath108 1.8 & 3.0 & @xmath129 & 13 12 25.549 @xmath108 0.15 & 42 39 07.01 @xmath108 1.60 & g & @xmath117 & @xmath117 & @xmath118 22.5 + 1 & j171349@xmath114501610 & 1136 @xmath108 63 & 20.3 & @xmath115 & 17 13 49.077 @xmath108 0.02 & 50 16 10.49 @xmath108 0.16 + 2 & j171354@xmath114501547 & 153 @xmath108 22 & 6.7 & @xmath121 & 17 13 54.160 @xmath108 0.05 & 50 15 47.70 @xmath108 0.45 + 3 & * j171407@xmath114501547 & 41 @xmath108 8 & 4.9 & @xmath120 & 17 14 07.032 @xmath108 0.06 & 50 15 47.76 @xmath108 0.61 + 4 & j171411@xmath114501602 & 37 @xmath108 7 & 5.0 & @xmath120 & 17 14 11.951 @xmath108 0.06 & 50 16 02.42 @xmath108 0.60 & obj 18 , 23.5 m g / b + 5 & j171414@xmath114501530 & 6482 @xmath108 20 & 0.9 & @xmath132 & 17 14 14.754 @xmath108 0.00 & 50 15 30.46 @xmath108 0.05 & 53w002 , 22.1 m g / b + 6 & j171415@xmath114501535 & 49 @xmath108 7 & 8.5 & @xmath120 & 17 14 15.219 @xmath108 0.01 & 50 15 35.45 @xmath108 0.15 + 7 & j171416@xmath114501817 & 495 @xmath108 22 & 30.4 & @xmath115 & 17 14 16.771 @xmath108 0.01 & 50 18 17.10 @xmath108 0.11 + 8 & * j171424@xmath114501339 & 62 @xmath108 13 & 4.7 & @xmath120 & 17 14 24.613 @xmath108 0.07 & 50 13 39.78 @xmath108 0.64 + 9 & * j171436@xmath114501329 & 194 @xmath108 42 & 4.5 & @xmath120 & 17 14 36.222 @xmath108 0.07 & 50 13 29.56 @xmath108 0.67 + 10 & * j171442@xmath114501640 & 505 @xmath108 99 & 4.9 & @xmath120 & 17 14 42.629 @xmath108 0.07 & 50 16 40.09 @xmath108 0.62 + cepheus & 1988 - 89 & 03 10 00 & 80 00 00 & 30 & 20.3 & 36 & 6 & 10.2 & 1,2,3 + lynx & 1989 - 90 & 08 41 40 & 44 45 00 & 63 & 12.8 & 46 & 14 & 17.1 & 1,4 + hercules & 1996 & 17 14 15 & 50 15 30 & 12 & 35.0 & 10 & 6 & 5.5 & 0 + sa13 & 1994 - 95 & 13 12 17 & 42 38 05 & 159 & 7.5 & 50 & 34 & 30.9 & 0,5,6 + hubble & 1996 - 97 & 12 36 49 & 62 12 58 & 139 & 9.0 & 60 & 29 & 25.2 & 7,8,9 + lrrlrc j030755 + 801008 & 1221 & @xmath1331.0 & & @xmath117 + j030852 + 801409 & 449 & @xmath1340.2 & & @xmath117 + j030902 + 800955 & 54 & @xmath1340.2 & & @xmath117 + j030924 + 801159 & 34 & @xmath1330.4 & & @xmath117 + j031005 + 800824 & 52 & @xmath1330.2 & & @xmath117 + j031101 + 800943 & 293 & @xmath1330.6 & & @xmath117 + j084120 + 444453 & 58 & @xmath1330.5 & & @xmath117 + j084121 + 444318 & 289 & @xmath1330.7 & & @xmath117 + j084126 + 444648 & 62 & @xmath1330.6 & & @xmath117 + j084133 + 444544 & 20 & @xmath1330.7 & & @xmath117 + j084134 + 444409 & 50 & @xmath1330.2 & & @xmath117 + j084135 + 444442 & 21 & @xmath1330.8 & & @xmath117 + j084136 + 444443 & 32 & @xmath1330.6 & & @xmath117 + j084141 + 444509 & 32 & @xmath1330.6 & & @xmath117 + j084141 + 444452 & 35 & @xmath1331.0 & & @xmath117 + j084142 + 444613 & 110 & @xmath1330.6 & & @xmath117 + j084142 + 444718 & 115 & @xmath1330.2 & & @xmath117 + j084144 + 444534 & 62 & @xmath1330.8 & & @xmath117 + j084150 + 444545 & 31 & @xmath1331.1 & & @xmath117 + j084155 + 444640 & 78 & @xmath1330.9 & & @xmath117 + j123632 + 621105 & 22 & @xmath1350.1 & sp & @xmath11720.1 + j123634 + 621212 & 56 & @xmath1330.7 & g / b & @xmath11719.3 + j123634 + 621240 & 53 & @xmath1330.7 & g / b & @xmath11723.5 + j123637 + 621135 & 18 & @xmath1350.2 & sp & @xmath11718.2 + j123640 + 621010 & 29 & @xmath1330.4 & g & @xmath11725.0 + j123641 + 621142 & 19 & @xmath1330.3 & g / b & @xmath11722.7 + j123642 + 621331 & 80 & @xmath1330.9 & g ? & @xmath11725.8 & yes + j123642 + 621545 & 54 & @xmath1330.5 & g & @xmath11722.0 + j123644 + 621249 & 10 & @xmath1350.4 & g / b & @xmath11721.9 & yes + j123644 + 621133 & * 599 & @xmath1330.3 & ell & @xmath11720.5 & yes + j123646 + 621448 & * 25 & @xmath1330.8 & g / b & @xmath11726.0 + j123646 + 621445 & 13 & @xmath1331.0 & g & @xmath11723.9 + j123646 + 621404 & 190 & @xmath1180.0 & sp & @xmath11720.0 & yes + j123649 + 621313 & 22 & @xmath1330.7 & g / b & @xmath11721.8 & yes + j123651 + 621030 & 26 & @xmath1330.7 & sp & @xmath11720.7 + j123651 + 621221 & 17 & @xmath1330.7 & g & @xmath11727.8 & yes + j123652 + 621444 & 185 & @xmath1180.1 & ell & @xmath11719.4 + j123653 + 621139 & 15 & @xmath1330.8 & sp & @xmath11722.3 + j123655 + 621311 & 12 & @xmath1350.2 & g / b & @xmath11723.1 & yes + j123657 + 621455 & 15 & @xmath1350.2 & g & @xmath11722.8 + j123700 + 620908 & * 67 & @xmath1330.9 & g & @xmath11726.3 + j123701 + 621146 & * 30 & @xmath1330.7 & g / b & @xmath11725.4 + j123707 + 611408 & * 29 & @xmath1330.3 & g & @xmath11726.4 + j123708 + 621056 & * 26 & @xmath1330.4 & sp & @xmath11720.4 + j123708 + 621246 & 20 & @xmath1350.1 & off & @xmath11724.4 + j123711 + 621331 & * 31 & @xmath1330.7 & g / b & @xmath11723.1 + j123716 + 621512 & * 145 & @xmath1330.4 & g / b & @xmath11720.4 + j123721 + 621129 & 677 & @xmath1330.3 & ef & @xmath11627.0 + j123725 + 621128 & * 613 & @xmath1331.2 & g & @xmath11724.1 + j131157 + 423910 & 109 & @xmath1330.3 & g & @xmath11723.5 + j131157 + 423630 & 197 & @xmath1330.6 & ef & @xmath11625.5 + j131203 + 423331 & 886 & @xmath1330.3 & g & @xmath11722.2 + j131205 + 423851 & 12 & @xmath1330.7 & ef & @xmath11625.5 + j131207 + 423945 & 14 & @xmath1330.7 & ef & @xmath11625.5 + j131209 + 424217 & 93 & @xmath1330.4 & g & @xmath11723.4 + j131211 + 424053 & 717 & @xmath1330.6 & g / b & @xmath11720.8 + j131213 + 424129 & 48 & @xmath1180.0 & ell & @xmath11718.7 + j131213 + 423706 & 9 & @xmath1330.8 & g ? & @xmath11725.5 + j131215 + 423702 & 13 & @xmath1330.4 & g / b & @xmath11719.7 + j131215 + 423901 & * 27 & @xmath1330.5 & q & @xmath11717.8 & yes + j131216 + 423921 & 30 & @xmath1330.9 & ef & @xmath11625.8 + j131217 + 423912 & 20 & @xmath1330.6 & g / b & @xmath11720.0 + j131217 + 423930 & 16 & @xmath1330.9 & ef & @xmath11625.8 + j131218 + 423843 & * 25 & @xmath1330.9 & g / b & @xmath11720.3 + j131219 + 423608 & * 26 & @xmath1330.9 & g & @xmath11722.1 + j131219 + 423831 & * 12 & @xmath1330.8 & g & @xmath11720.4 + j131220 + 423923 & 15 & @xmath1330.7 & ef & @xmath11625.8 + j131220 + 423704 & 14 & @xmath1331.0 & ef & @xmath11625.8 + j131220 + 424029 & 34 & @xmath1330.3 & ef & @xmath11625.5 + j131220 + 423535 & 20 & @xmath1330.6 & ef & @xmath11625.5 + j131221 + 423923 & * 12 & @xmath1330.8 & sp & @xmath11718.0 + j131221 + 423722 & 10 & @xmath1330.9 & sp & @xmath11717.8 + j131221 + 423827 & 19 & @xmath1330.8 & g / b & @xmath11722.7 + j131222 + 423813 & 10 & @xmath1330.5 & q & @xmath11719.6 & yes + j131223 + 423908 & * 19 & @xmath1330.8 & g / b & @xmath11719.3 + j131223 + 423712 & 28 & @xmath1330.8 & g / b & @xmath11718.8 + j131223 + 423525 & 66 & @xmath1330.1 & ell & @xmath11718.9 + j131225 + 423656 & 13 & @xmath1350.1 & g ? &
@xmath11725.0 + j131225 + 424103 & 35 & @xmath1330.7 & g ?
& @xmath11725.5 + j131225 + 423941 & 200 & @xmath1330.7 & ef & @xmath11625.5 + j131227 + 423800 & 26 & @xmath1330.9 & g / b & @xmath11719.8 + j131232 + 424038 & 166 & @xmath1330.8 & g & @xmath11724.8 + j131239 + 423911 & 66 & @xmath1330.6 & g & @xmath11724.5 + j171349 + 501610 & 1138 & @xmath118 & & @xmath117 + j171354 + 501547 & 153 & @xmath118 & & @xmath117 + j171411 + 501602 & 37 & @xmath118 & g / b & @xmath11723,5 + j171414 + 501530 & 6480 & @xmath1331.1 & g / b & @xmath11722.1 + j171415 + 501535 & 49 & @xmath118 & & @xmath117 + j071416 + 501817 & 495 & @xmath118 & & @xmath117 + crcrrc 7.5 to 19.0 & 13.7 & 65 & 21 & 26.6 & @xmath136 + 9.0 to 30.0 & 24.5 & 174 & 20 & 15.6 & @xmath137 + 30.0 to 65.0 & 44.9 & 317 & 20 & 24.8 & @xmath138 + 65.0 to 600.0 & 195.0 & 558 & 21 & 29.4 & @xmath139 + @xmath140 & 1675.0 & 653 & 7 & 3.1 & @xmath141 + | we present the results from two radio integrations at 8.4 ghz using the vla .
one of the fields , at 13@xmath0 + 43@xmath1 ( sa13 field ) , has an rms noise level of @xmath2jy and is the deepest radio image yet made .
thirty - four sources in a complete sample were detected above @xmath3jy and 25 are optically identified to a limit of i=25.8 , using our deep hst and ground - based images .
the radio sources are usually located within @xmath4 ( typically 5 kpc ) of a galaxy nucleus , and generally have a diameter less than @xmath5 .
the second field at 17@xmath0 + 50@xmath1 ( hercules field ) has an rms noise of @xmath6jy and contains 10 sources .
we have also analyzed a complete flux density - limited sample at 8.4 ghz of 89 sources from five deep radio surveys , including the hubble deep and flanking fields as well as the two new fields .
half of all the optical counterparts are with galaxies brighter than i=23 mag , but 20% are fainter than i=25.5 mag .
we confirm the tendency for the micro - jansky radio sources to prefer multi - galaxy systems .
the distribution of the radio spectral index between 1.4 and 8.4 ghz peaks at @xmath7 ) , with a median value of @xmath8 .
the average spectral index becomes steeper ( lower values ) for sources below @xmath6jy , and for sources identified with optical counterparts fainter than i=25.5 mag .
this correlation may suggest that there is an increasing contribution from star - burst galaxies compared to active galactic nuclei ( agns ) at lower radio flux densities and fainter optical counterparts .
the differential radio count between 7.5 and 1000 @xmath9jy has a slope of @xmath10 and a surface density of 0.64 sources ( arcmin)@xmath11 with flux density greater than @xmath3jy . |
the study of hadron spectroscopy has made significant progress during the past decade benefiting from the experimental observations of a lot of new states in various processes . in particular , in the heavy flavor sector , a large number of the so - called @xmath9 , @xmath10 , @xmath11 states have been candidates for exotic hadrons which may contain more complicated quark - gluon structures than the conventional quark model picture . among all those exotic candidates ,
the @xmath0 is undoubtedly one of the most mysterious states and has initiated a lot of experimental and theoretical studies since its observation by babar collaboration in the @xmath12 channel in 2005 @xcite .
it was later confirmed by belle @xcite and cleo - c collaboration @xcite in the same decay channel , but evaded " from observation in its decays into open charm channels @xcite which is unexpected for conventional charmonium states above the @xmath13 threshold .
nevertheless , in the @xmath14 value measurement the inclusive cross sections for @xmath1 annihilations appear to have a dip instead of a peak in the vicinity of the @xmath0 mass region .
another important observation which makes @xmath0 peculiar is that the successful potential quark model calculations did not predict a vector state in such an energy region and the conventional @xmath3 states have been assigned to other structures @xcite .
such surprising properties have initiated a lot of theoretical interpretations of its nature in the literature , such as vector hybrid candidate refs .
@xcite , hadro - quarkonium @xcite , tetraquark state @xcite , and hadronic molecules of @xmath2 @xcite , @xmath15 @xcite , or @xmath16
@xcite , or non - resonance explanation due to state interferences @xcite . in ref .
@xcite @xmath0 was proposed to be a conventional charmonium @xmath17 state .
interestingly , the lattice qcd ( lqcd ) calculation indicates that the hybrid vector charmonium is located in the mass region near 4.3 gev which makes the @xmath0 a possible hybrid state corresponding to the lqcd spectrum @xcite . in the recent lqcd study of ref .
@xcite a spatially extended hybrid - like operator is applied and yields a rather small leptonic decay width of less than 40 ev .
it makes the measurement of the leptonic decay width an interesting observable for probing its internal structures .
the recent observation of the charged charmonium states @xmath18 at the mass of @xmath0 in @xmath19 with high statistics at besiii @xcite once again attracted attention from the community and provoked further studies of the @xmath0 .
the mass of the @xmath18 is close to the @xmath20 threshold . with @xmath21
it decays via an @xmath22 wave into @xmath20 and @xmath23 with large coupling to @xmath20 which makes it a reasonable candidate for hadronic molecule of @xmath20 nevertheless , its production from the @xmath0 decays provides important information for the structure of @xmath0 and a detailed studies of the @xmath0 should be crucial for a self - consistent picture for those puzzling phenomena arising in this energy region . in ref .
@xcite , it was proposed that the production of @xmath18 in @xmath24 should be a strong signal for @xmath0 being a @xmath2 hadronic molecule .
one notices that the @xmath2 is the first two - body open charm @xmath22-wave threshold in the vector charmonium channel with narrow final states and the @xmath0 is less than 30 mev below the @xmath2 threshold .
it should be pointed out that the mass and width of @xmath0 were extracted by a breit - wigner fit to the @xmath19 cross sections @xcite .
since the cross section lineshape is not symmetric at the two sides of the peak this treatment may cause significant uncertainties with the resonance parameters .
in fact , the asymmetric lineshape can be regarded as an indication of the nontrivial nature of the @xmath0 of which the propagator can not be parameterized by a simple breit - wigner .
this forms one of the motivations in this study that a proper treatment of the renormalization effects due to the meson loops will be essential for extracting the pole information . recall that we have been emphasizing that a strong @xmath22-wave interactions between @xmath2 can lead to a dynamically generated pole structure near the @xmath2 threshold and the @xmath0 should be the natural candidate @xcite . to firm up such a scenario ,
various properties and observables were investigated in the molecular picture . in ref .
@xcite , it was studied that the @xmath0 as a @xmath2 molecule will have sizeable decay rate into @xmath25 given that @xmath26 is also a hadronic molecule of @xmath27 one interesting question arising from the @xmath2 molecule picture is that the @xmath28 should be the narrow state in the @xmath29 and @xmath30 mixing for the @xmath31 states .
the data show that the narrow @xmath28 will be dominated by the configuration with the light quark angular momentum @xmath32 coupled by the orbital angular momentum and antiquark spin and parity
. this will be a pure configuration either by taking the heavy quark spin symmetry ( hqss ) limit or if @xmath28 is one of the eigenstate from the @xmath29 and @xmath30 ideal mixing .
it was then pointed out in ref .
@xcite , in the hqss limit the coupling of the heavy @xmath3 in an @xmath22 wave of @xmath33 while the light quark degrees of freedom with the spin and orbital angular momentum coupling to @xmath34 is forbidden in the spin decomposition of @xmath2 this observation brings concerns on the production of @xmath0 in @xmath1 annihilations if it is indeed a hadronic molecule of @xmath2 however , as shown by refs .
@xcite , the hqss breaking in the charmonium mass region should be expected and it is sufficient to account for the cross sections observed for @xmath0 .
our studies in this work will further clarify the hqss breaking effects and make the molecular picture dynamically self - contained . in brief , we will follow the molecular picture of refs .
@xcite and explore the @xmath0 production mechanism in @xmath1 annihilations .
we will try to accommodate the short - distance production mechanism coherently in the molecular picture using weinberg s compositeness theorem @xcite .
this will quantify the hqss breaking effects in explicit calculation and this treatment should be able to tell us more about the structure of the @xmath0 .
as follows , we first present the full propagator of the @xmath0 in the framework of the non - relativistic effective field theory ( nreft ) in sec .
[ nreft - propagator ] . in sec .
[ lineshape - study ] we present the formalism for the study of the cross section lineshape of @xmath4 and the width of @xmath35 .
we then investigate the hqss breaking effects in the @xmath6 model as an independent check of the hqss breaking in sec .
[ 3p0-model ] .
a brief summary is given in sec .
[ summary ] .
in our scenario @xmath0 is treated as an @xmath36wave molecule of @xmath37 containing a small charmonium component @xmath38 , i.e. @xmath39 . for the @xmath40 scattering ( the conjugate part is implicated ) , the contact interaction is also introduced in addition to the bare pole structure .
the process for the decay of @xmath41 and scattering of @xmath42 are illustrated in fig .
the corresponding lagrangian can be written as @xmath43 where the real coefficient @xmath44 is the coupling constant of the contact interaction . in the molecular scenario
the non - relativistic propagator of the @xmath0 can be expressed as @xmath45}\nonumber\\ & \equiv & \frac{1}{2}\frac{i}{e - m_0-\sigma_1(e ) } , \end{aligned}\ ] ] where @xmath46 is the bare mass of @xmath0 and @xmath47 comes from the sum of the infinite bubble loops in the @xmath40 rescattering .
the one loop function is defined as the following : @xmath48 in dimensional regularization with the @xmath49 subtraction scheme , the loop integral @xmath50 is finite for @xmath51 and one obtains @xmath52 where @xmath53 is the reduced mass .
the bare mass @xmath46 should be renormalized to the physical mass @xmath54 , which gives the physical pole position in the real axis of the energy plane via @xmath55 . by expanding the denominator of the propagator near the physical pole position of @xmath54
, the propagator can be expressed as @xmath56 where @xmath57 and @xmath58 $ ] is the wavefunction renormalization constant .
note that @xmath59 is the probability of finding @xmath0 in a @xmath37 molecular state .
it can then be related to the molecular component of the @xmath0 wavefunction via @xmath60 .
alternatively , the renormalization constant @xmath11 defines the probability of finding the @xmath0 in a non - molecular state .
a natural assumption is that @xmath11 can be related to the compact @xmath3 component in the @xmath0 wavefunction and at leading order one expects @xmath61 .
it should be emphasized that this renormalization will allow constraints on the production of @xmath0 via the short - distance component , e.g. in the @xmath1 annihilations .
this also implies that @xmath62 should be one of the important decay channels of @xmath0 .
however , notice that the @xmath63-wave decay of @xmath5 is suppressed .
the partial width of @xmath64 is not expected to be sizeable .
a detailed study is necessary for a coherent picture for the nature of @xmath0 .
we also introduce a constant width @xmath65 in the propagator to account for contributions from decay channels other than the @xmath37 then the complete propagator of @xmath0 reads @xmath66 where @xmath54 and @xmath65 are parameters to be fitted by experimental data for the cross section lineshapes of @xmath19 and @xmath67 . as studied in ref .
@xcite , the asymmetric cross section lineshape of @xmath19 turns out to be selective for the determination of the pole position @xmath54 and @xmath65 .
thus , by fitting the cross section lineshapes of @xmath19 and @xmath67 , but allowing an overall scale parameter , we can determine @xmath68 gev and @xmath69 gev @xcite .
we note that the pole corresponding to @xmath0 is close to the threshold of @xmath37 when considering the pole value of @xmath28 .
this follows the approach of ref .
@xcite , namely , the renormalization constant @xmath11 in the numerator of eq .
( [ propagator - full ] ) was absorbed into the vertex coupling constant as an overall scale parameter .
this is because that the pole is fitted at @xmath70 with @xmath71 , i.e. @xmath72 .
it means that the pole position does not explicitly depend on the renormalization constant .
one notices that @xmath69 gev accounts for the sum of contributions from non-@xmath37 channels .
they include the hidden charm decays , i.e. @xmath12 , @xmath67 , and @xmath15 , and also open charm decays that do not go through the @xmath37 this part of open charm decays reflect the hqss breaking effects for which we will try to quantify in this work later .
it should be noted that the recent experimental measurements of besiii @xcite suggest that the cross sections for @xmath19 , @xmath67 , and @xmath15 are compatible .
if these channels saturate the total width of @xmath0 , it will lead to relatively small leptonic decay width for @xmath35 @xcite .
this is also a challenge in understanding the nature of @xmath0 .
as follows , we will combine the production mechanism to investigate the relation between the renormalization constant @xmath11 and its implication of relative decay strength for @xmath0 to different decay channels .
the full propagator of eq .
( [ propagator - full ] ) and decomposition of the @xmath0 wavefunction of eq .
( [ decomp ] ) will help us to separate out the short - distance compact @xmath3 component and long - distance molecular component , and also provide constraint on its leptonic decay width .
the production and decay of @xmath0 in @xmath1 annihilations can go through two steps , i.e. the compact @xmath3 component is first created by the virtual photon and then the physical @xmath0 can decay either via the molecular @xmath37 component or the compact @xmath3 into final @xmath62 it can then be illustrated by fig .
[ y - decay ] where ( a ) and ( b ) are decay processes through the molecular component while ( c ) and ( d ) are through the compact @xmath3 which are originated from the hqss breaking .
although there are various interpretations for the @xmath18 , we treat it as an @xmath36wave molecule of @xmath74 in our framework .
similar to @xmath0 the propagator of the @xmath18 can be written as @xmath75+i\gamma^{\text{non}-\bar{d}d^*}/2 } \ , \end{aligned}\ ] ] with @xmath76 where @xmath77 , @xmath78 and @xmath79 gev@xmath80 is the coupling constant of @xmath18 to @xmath74 @xcite .
the contact interaction of @xmath81 is introduced by the term with coupling @xmath82 in eq .
( [ zcprop ] ) .
however , by fitting the @xmath20 invariant mass spectrum it shows that the value of @xmath82 is negligibly small and so we take @xmath83 in the following analysis . given that a compact @xmath3 component is present in the wavefunction of @xmath0 , it would be a natural mechanism to produce the @xmath0 in the @xmath1 annihilations .
the coupling of @xmath0 to the virtual photon can be approximated by the vector meson dominance ( vmd ) model , which is described by the following lagrangian for the compact component coupling to the virtual photon : @xmath84 where @xmath85 is the bare decay constant defined for the compact wavefunction of @xmath0 .
it should be addressed that the compact @xmath3 in an @xmath22 wave should decouple to the @xmath37 molecular component in the hqss limit @xcite . however , since the charm quark is not heavy enough the hqss breaking is anticipated and there will be several possible mechanisms leading to mixing between the compact @xmath3 and molecular @xmath37 : * the presence of the compact @xmath3 component may be produced by the coupled channel between @xmath37 and @xmath86 which allows a small @xmath22-wave coupling for @xmath87 since the @xmath88 state is very broad such a contribution including the @xmath18 pole , i.e. contributions from fig .
[ y - decay ] ( c ) and ( d ) , can be parameterized by an @xmath22-wave coupling amplitude for @xmath89 as demonstrated in ref .
@xcite : @xmath90 where @xmath91 is the invariant mass square of @xmath92 and @xmath93 in the final state and @xmath94 and @xmath95 are real numbers to be fitted by the experimental data .
this parametrization @xcite respects watson theorem and is equivalent to introducing a contact interaction and a pole term to the transition amplitude as illustrated by fig .
[ y - decay ] ( c ) and ( d ) , respectively .
* the coupled channel between @xmath37 and @xmath86 will lead to the hqss breaking and result in the deviation of the @xmath29 and @xmath30 mixing from the ideal mixing .
then , a small @xmath22-wave coupling for @xmath0 to the physical @xmath37 will be allowed .
this mechanism will also affect the decay of @xmath5 where a small @xmath22-wave coupling will contribute to the amplitude of eq .
( [ s - wave - para ] ) .
this scenario can be examined by the quark model calculation and we will come back to this point later in this work .
the above considerations can be recognized by the full amplitude of @xmath73 which is expressed as @xmath96 \times\frac{h^\prime}{f_\pi}(3p_\pi^ip_\pi^j-\overrightarrow{p}_\pi^2\delta^{ij})\big{\}}\epsilon_{d^*}^j\nonumber \\ & \equiv&\bar{v}(p_1)(\gamma^i)u(p_2)\frac{ie^2m^2_y}{f_ye^2}\frac{1}{2}\frac{iz}{e - m_y - z\widetilde{\sigma}_1(e)+i\gamma^{\text{non}-\bar{d}d_1}/2}\nonumber \\&&\times \big{\{}a_s\delta^{ij}+(\frac{iy^{\text{bare}}}{\sqrt{2}})a_d(e , m_{\bar{d}d^*},m_{d^*\pi } ) \times\frac{h^\prime}{f_\pi}(3p_\pi^ip_\pi^j-\overrightarrow{p}_\pi^2\delta^{ij})\big{\}}\epsilon_{d^*}^j,\end{aligned}\ ] ] where @xmath97 , @xmath98 and @xmath99 is the triangle loop integral , @xmath100 in the above equation , @xmath101 is the @xmath22-wave strength and @xmath102 is the @xmath63-wave strength via the intermediate @xmath31 while @xmath103 and @xmath104 are the incoming momenta of the electron and positron , respectively .
this expression is different from the treatment of ref .
@xcite where the wavefunction renormalization constant @xmath11 in the numerator has been absorbed into an overall scale parameter . in eq .
( [ trans - amp ] ) the renormalization constant is explicitly included based on the compositeness theorem .
one notices that the effective coupling for @xmath0 to @xmath37 and a virtual photon can be renormalized as @xmath105 and @xmath106 , respectively . the parameters in eq .
( [ trans - amp ] ) includes the bare coupling @xmath107 for @xmath0 to @xmath37 , bare decay constant @xmath108 for @xmath0 to virtual photon , and parameters @xmath94 and @xmath95 for the small @xmath22-wave contributions .
note that the renormalization constant @xmath11 is a function of @xmath107 . for the @xmath0 being a @xmath109 molecule , the value of @xmath11 which representing the non-@xmath37 component will take the zero limit .
it means that the production of @xmath0 in @xmath1 annihilations will be suppressed .
alternatively , if there is a sizeable non-@xmath37 component inside @xmath0 , its production in @xmath1 annihilations will be enhanced .
however , its coupling to @xmath37 will be suppressed by the compositeness theorem .
such a constraint can provide a self - consistent check of the molecular scenario .
the @xmath110 coupling @xmath111 can be determined by @xmath5 via the effective coupling @xcite : @xmath112+h.c.\end{aligned}\ ] ] note that the neutral and charged @xmath28 have similar total widths , i.e. @xmath113 mev and @xmath114 mev @xcite . meanwhile ,
the partial width for @xmath5 has not been precisely measured .
therefore , the total width will give the upper limit of the coupling constant @xmath111 if one assumes that the total width of @xmath28 is saturated by the @xmath115 decay . by applying the pole mass and non-@xmath40 width from ref .
@xcite , we fit the besiii data @xcite for the angular distribution of pion which recoils the @xmath116 threshold peak in the @xmath1 c.m .
frame , the invariant mass spectrum of the @xmath117 , and the belle data @xcite for the cross section for @xmath118 to extract the couplings .
the fitting results are illustrated in figs .
[ fractional]-[sigma ] , respectively .
similar to what found in ref .
@xcite , as shown by fig .
[ fractional ] the angular distribution of the recoiled pion provides a strong constraint on the @xmath22-wave component .
a flat angular distribution of the pion recoiling the threshold enhancement in the @xmath74 spectrum can be produced with a dominant @xmath63-wave and relatively small @xmath22-wave amplitudes .
actually , the improved data from ref .
@xcite appear to have tilted behavior deviating from a pure @xmath22-wave angular distribution .
this could be an indication of presence of non-@xmath22 component .
one can see in fig .
[ dsigma ] that away from the @xmath116 threshold enhancement the @xmath22-wave component becomes important .
in fact , the integrated cross section is dominated by the @xmath22-wave contributions which will be the dominant source of partial widths of @xmath0 near threshold . but with the increase of the c.m .
energy , the @xmath63-wave contributions will become dominant . this feature can be seen clearly in fig .
[ sigma ] .
the combined contributions generate the non - trivial cross section lineshape which appear to be consistent with the belle results although there are still large uncertainties with the data @xcite
. future precise measurement of the cross section lineshape at besiii would be important for confirming this phenomenon .
m. ablikim _ et al . _
[ besiii collaboration ] , phys .
rev . lett .
* 112 * ( 2014 ) 2 , 022001 [ arxiv:1310.1163 [ hep - ex ] ] . m. ablikim _ et al . _
[ besiii collaboration ] , phys .
d * 92 * ( 2015 ) no.9 , 092006 doi:10.1103/physrevd.92.092006 [ arxiv:1509.01398 [ hep - ex ] ] . | we study the @xmath0 production mechanism in @xmath1 annihilations in the framework of hadronic molecules and investigate the consequence of such a picture in different decay channels . in the hadronic molecule picture the @xmath0
is described as a mixture state composed of a long - ranged @xmath2 molecule state and a compact @xmath3 component .
we show that the compositeness relation can still provide a reasonable constraint on the wavefunction renormalization parameter due to the dominance of the molecular component .
such a mechanism can be regarded as a natural consequence of the heavy quark spin symmetry ( hqss ) breaking .
this study elaborates the molecular picture for the @xmath0 in the @xmath1 annihilations and affirms that the cross section lineshape of @xmath4 in the vicinity of the @xmath0 should have a nontrivial behavior .
in this framework we predict that the upper limit of the @xmath0 leptonic decay width is about 500 ev .
we also investigate the coupling for @xmath5 in the @xmath6 quark model and examine the possible hqss breaking effects due to the deviation from the @xmath7 and @xmath8 ideal mixing .
this will in turn provide a constraint on the hqss breaking coupling for the @xmath0 to @xmath2 via its @xmath3 component . |
in the paper @xcite pinchuk gave an example of a polynomial map @xmath0 with a non - vanishing jacobian such that @xmath1 is not a global diffeomorphism .
let us recall his construction .
we define the following auxiliary polynomials in variables @xmath2 , @xmath3 : @xmath4 and a polynomial @xmath5 then @xmath6 where @xmath7 and @xmath8 are given by @xmath9 our aim is to give a geometric description of pinchuk s map . as in @xcite
we find the set of points at which @xmath1 is not proper .
it allows us to divide the image of @xmath1 into sets with constant multiplicity of fibers .
finally we illustrate how @xmath1 transforms the real plane to these sets .
let @xmath10 be a continuous map of locally compact spaces .
we say that the mapping @xmath11 is not proper at a point @xmath12 , if there is no a neighborhood @xmath13 of a point @xmath3 such that the set @xmath14 is compact .
the set @xmath15 of points at which the map @xmath11 is not proper indicates how the map @xmath11 differs from a proper map .
in particular the restriction of @xmath11 from @xmath16 to @xmath17 is proper .
this notion was studied in @xcite in the complex case and in @xcite , where @xmath15 is called an asymptotic variety .
let @xmath18 be a curve given by @xmath19 [ p1 ] the set of points at which @xmath1 is not proper is the curve @xmath20 .
the set @xmath21 is a smooth curve having three connected components . furthermore * for every @xmath22 @xmath23 , * for every @xmath24 @xmath25 , * @xmath26 .
a behavior of @xmath1 is illustrated in figure below .
a set @xmath27 has 4 connected components .
the marked ones are mapped diffeomorphically to a marked area above the curve @xmath20 .
two unmarked regions are mapped diffeomorphically to an area below @xmath20 . indicated connected components of a curve @xmath21
are mapped to indicated parts of the curve @xmath20 , respectively .
a similar description of pinchuk s map was given in @xcite .
the author characterizes there another mapping from the class discovered by pinchuk .
note that @xmath20 is not an algebraic curve although it has a polynomial parameterization .
it follows from the proof of lemma [ l4 ] that the zariski closure of @xmath20 in @xmath28 consists of @xmath20 and a point @xmath29 , where @xmath30 is a complex solution of an equation @xmath31 .
this shows that theorem 3 in @xcite , where it is claimed that the asymptotic variety is algebraic , is partially false .
a polynomial map @xmath32 is another counterexample .
an asymptotic variety of this map is a right half line @xmath33
. moreover points ( b ) and ( c ) in theorem 3 can be removed .
we refer the interested reader to @xcite , theorem 3.3 .
our approach is based on the following observation due to zbigniew jelonek .
[ j1 ] let @xmath34 be a complexification of @xmath1 .
then the set @xmath35 of points at which @xmath1 is not proper is contained in the set @xmath36 of points at which the mapping @xmath37 is not proper .
* easy exercise .
@xmath38 the next result concerning complex polynomial maps was proved in @xcite , @xcite : [ j2 ] let @xmath39 be a dominant polynomial map .
then the set @xmath15 of points at which @xmath11 is not proper consists of a finite number ( possibly 0 ) of affine algebraic curves . to prove theorem [ p1 ] we find the set of points at which complex pinchuk s map is not proper and then apply proposition [ j1 ] .
[ lc ] @xmath40 .
[ l2 ] the following relations hold for @xmath41 , @xmath42 , @xmath11 , @xmath8 : @xmath43 where @xmath44 . * proof . *
the formula ( [ e5 ] ) follows immediately from ( [ e1 ] ) .
multiplying equation ( [ e2.2 ] ) by @xmath45 gives @xmath46 . by ( [ e5 ] )
we have @xmath47 . substituting the right hand side into above formula
we obtain ( [ e6 ] ) . to solve this system for a given @xmath49 it suffices to find roots of @xmath50 where @xmath51 is a monic polynomial of degree 6 with respect to @xmath52 .
therefore a number of solutions of ( [ e7 ] ) equals a number of roots of a polynomial @xmath50 . for every @xmath53 with @xmath54
there is a unique pair @xmath49 such that @xmath55 satisfies ( [ e7 ] ) . namely @xmath56 observe that ,
generically , the pinchuk mapping is a composition of a polynomial map @xmath57 and a map @xmath58 given by ( [ e72 ] ) .
we precise this idea in lemmas [ l5 ] and [ l51 ] .
first we need some preparation .
[ l3 ] the restriction of @xmath60 from @xmath61 to @xmath62 is a homeomorphism .
a curve @xmath63 has two algebraic components : @xmath64 and @xmath65 .
the restriction of the polynomial function @xmath41 to @xmath66 is injective and @xmath67 for @xmath68 .
* we claim that @xmath69 . indeed , suppose that @xmath70 and @xmath71 .
substituting @xmath72 into ( [ e5 ] ) gives @xmath73 .
hence @xmath74 and by ( [ e1 ] ) @xmath75 .
therefore @xmath76 contrary to our assumption .
now we describe a curve @xmath63 .
assume that @xmath79 .
then by ( [ e1 ] ) @xmath80 or @xmath81 .
we obtain by ( [ e1 ] ) in the first case @xmath82 , @xmath83 , @xmath75 and in the second case @xmath84 , @xmath85 , @xmath86 .
this finishes the proof .
fix @xmath106 lying in the complement of above curves and take a neighborhood @xmath13 of @xmath107 such that @xmath108 is a compact set disjoint from @xmath109 .
let @xmath110 be the set of all @xmath111 such that @xmath55 satisfies ( [ e7 ] ) for some @xmath112 .
by continuity of roots ( see @xcite , proposition 1.5.5 ) @xmath110 is a compact set .
moreover , it follows from lemma [ l6 ] that @xmath113 .
hence @xmath114 .
since @xmath115 , we conclude by lemma [ l3 ] that @xmath116 is compact .
therefore the map @xmath37 is proper at the point @xmath107 .
now we show that the polynomial @xmath50 has no multiple factor in a ring @xmath117 $ ] .
assume that it has .
then @xmath118 has also a multiple factor .
since @xmath119 can be rewritten as @xmath120 the only candidate for a multiple factor is @xmath52 .
but @xmath52 does not divide @xmath121 contrary to our assumption .
our next task is to show that lines @xmath124 and @xmath125 are not contained in @xmath36 .
we will use the following property ( see @xcite ) : _ if @xmath126 , then @xmath37 is proper at a point @xmath49 .
_ here @xmath127 denotes a point in a general position , so @xmath128 .
we have @xmath129 .
thus for a generic @xmath130 a system ( [ e7 ] ) has 4 solutions @xmath131 ( @xmath132 ) such that @xmath133 and one double solution @xmath134 . by [ l5 ] and
[ l51 ] we get @xmath135 . hence by the above mentioned property @xmath136 .
we show in the same way that @xmath137 . from [ j2 ] it follows that the set @xmath36 is either a curve @xmath138 or is empty .
the latter is impossible because in this case real pinchuk s map would be proper and thus it would be a diffeomorphism .
* suppose to the contrary that there are @xmath141 such that @xmath142 .
an equation @xmath143 gives @xmath144 . combining this with equality @xmath145 we get @xmath146 .
hence @xmath147 which contradicts the choice of @xmath148 and @xmath149 .
[ l1 ] let @xmath152 be a polynomial function which is a local diffeomorphism .
then the function @xmath153 is lower semi - continuous .
moreover , @xmath154 is proper at a point @xmath155 if and only if @xmath156 is locally constant at @xmath157 .
* proof of theorem [ p1 ] . *
it follows from [ j1 ] and [ lc ] that @xmath35 is a subset of @xmath158 . since an asymptotic variety does not have isolated points ( see @xcite , theorem 3 ) , @xmath159 . from the topological point of view
@xmath160 is an embedded line .
hence @xmath161 has 2 connected components .
we shall calculate a multiplicity of fibers in each of them .
the curve @xmath20 cuts a line @xmath162 transversely at points @xmath163 and @xmath164 .
hence the points @xmath165 and @xmath166 lie on opposite sides of @xmath20 .
one checks that polynomials @xmath167 and @xmath168 have 2 real roots ( e.g. we can use sturm sequence ) .
hence @xmath169 and by lemma [ l1 ] @xmath23 for every @xmath22 .
now we show that @xmath170 .
we have @xmath171 . the only real root of this polynomial is @xmath96 .
since by [ l5 ] @xmath172 does not belong to the set @xmath173 we have @xmath174 .
we check similarly that @xmath175 . it remains to compute @xmath176 for @xmath177 .
let @xmath178 be a discriminant of the polynomial @xmath50 with respect to @xmath52 . using any computer symbolic algebra program one can check that the polynomial @xmath179 is nonzero at a point @xmath180 .
hence @xmath179 does not vanish on @xmath20 but a finite number of points .
fix @xmath181 for which @xmath182 .
we have shown before that @xmath50 has two real roots for @xmath183 .
therefore by continuity of roots the system ( [ e7 ] ) has two real solutions at a point @xmath107 . by [ l6 ] and
[ l4 ] one of these solutions is @xmath184 .
thus @xmath185 .
the restriction of @xmath1 from @xmath21 to @xmath189 is a diffeomorphism .
hence the curve @xmath21 has @xmath190 connected components homeomorphic to a line and its complement @xmath27 has @xmath191 connected components .
the set @xmath192 has two connected components @xmath193 , @xmath194 .
write @xmath195 where @xmath196 are connected components of the set @xmath197 .
the mappings @xmath198 ( @xmath68 , @xmath199 ) induced by @xmath1 are proper and unramified hence they are topological coverings .
since sets @xmath200 ( @xmath68 ) are simply connected , these mappings are homeomorphisms . since @xmath23 for @xmath22 , @xmath197 consists of 2 connected components for @xmath68 .
it follows that @xmath201 , therefore @xmath202 . | sergey pinchuk found a polynomial map from the real plane to itself which is a local diffeomorphism but is not one - to - one .
the aim of this paper is to give a geometric description of pinchuk s map . |
the concept of current - carrying edge channels accounts for the magnetotransport properties of a two - dimensional electron gas ( 2deg ) in a high magnetic field , both in the integer@xcite and fractional@xcite quantum hall regime .
although the initial theoretical studies have used a noninteracting picture of edge channels , a considerable effort has recently been devoted to understand the effects of electron - electron interactions in the _ integer _ quantum hall regime.@xcite at present there are two incompatible pictures for the electronic ground state of edge channels in the integer quantum hall regime . in the hartree - fock approximation
the ground state wave function is a single slater determinant , which corresponds to occupation numbers zero and one.@xcite in this description the edge state is incompressible and the electron density drops rapidly at the edges , on a length scale which is typically of the order of the magnetic length .
although the hartree - fock approximation is widely used for the description of quantum dots and wires,@xcite it has been challenged on the grounds that is does not take the global electrostatics into account properly.@xcite in the electrostatic description given in ref .
a more gradual variation of the electron density at the edge is found energetically favorable . here
the typical length scale for the density drop at the edge is of the order of the depletion length , which is much larger than the magnetic length . in this picture
the electrons at the edge can screen the electrostatic confinement potential by a slow density variation and the edge states are therefore compressible .
however , this description completely neglects the exchange interaction .
a transition from incompressible to compressible edge states was already qualitatively discussed in ref . , where a spontaneous transition from an unpolarized ( _ i.e. _ equal occupation for both spin levels ) to a polarized ( different occupation for different spin levels ) hartree - fock ground state was found for a critical confinement strength .
it was speculated that a development to the electrostatic regime takes place as the confinement strength is decreased .
more recently the transition between smoothly and abruptly varying density distributions has been studied by chamon and wen.@xcite on the basis of few particle exact solutions they predict formation of compressible edge states when the strength of the confinement potential is reduced beyond a certain point .
their approach is difficult to extend to quantum wires or dots with a large number of electrons . in the present work we interpolate between the different regimes of applicability of both hartree - fock and electrostatic pictures by extending the thomas - fermi approach used in ref . to a thomas - fermi-_dirac _ like treatment of the exchange effects in strong magnetic fields .
first results of the present approach have been published in ref . .
very recently ferconi and vignale@xcite have studied the ground state energies and densities of a quantum dot in an arbitrary magnetic field , taking into account exchange - correlation effects by employing a kohn - sham scheme of current - density functional theory.@xcite for small quantum dots ( 2 or 3 electrons ) their method yields an accuracy better than 3% when compared with exact results . for a larger number of electrons and high magnetic fields
we find that the kohn - sham formalism encounters serious problems . in sec .
[ sec : dft ] we describe our implementation of density - functional theory in strong magnetic fields as applied to quantum wires and derive expressions for the density profile and single - particle potentials of the ground state . in sec .
[ sec : results ] we present results of the numerical calculations and we investigate both the accuracy of our theory and the relation with possible experiments .
two experimentally relevant applications are studied in sec .
[ sec : applications ] ; the influence of a plane of constant potential parallel to the 2deg on the electronic ground state and the ground state properties of a quantum dot in a strong magnetic field .
section [ sec : conclusions ] summarizes our conclusions .
let us consider first a quantum wire of the strictly two dimensional electron gas along the @xmath0 axis in the @xmath1 plane ( fig .
[ fig : system ] ) .
perpendicular to the plane a strong uniform magnetic field is applied . also present
is a uniform positive background charge which ensures global neutrality .
we use periodic boundary conditions in the @xmath0-direction and adopt the landau gauge so that @xmath2 and @xmath3 . in the high magnetic field limit considered here only the lowest ( spin - up and spin - down ) landau levels are occupied .
the wire is confined in the @xmath4 direction by a parabolic confinement potential : @xmath5 where @xmath6 is the dielectric constant and @xmath7 is a dimensionless parameter .
we will ignore here the difference between the bare magnetic length @xmath8 and the magnetic length @xmath9 which is renormalized by the parabolic confinement potential : @xmath10 where @xmath11 is the cyclotron frequency with @xmath12 the effective electron mass and @xmath13 characterizes the parabolic confinement potential @xmath14 .
since we treat the problem numerically , the assumption of a parabolic form for the confinement is not essential : it is chosen here because it is widely used in the literature .
all lengths will be given in units of @xmath15 throughout the paper unless otherwise indicated .
we use density - functional theory in strong magnetic fields to find the electronic ground state of the system.@xcite without a magnetic field , the total energy of the system would be a functional of the electron density only . however , a magnetic field , apart from giving rise to zeeman splitting , causes orbital currents to flow in the electron gas , even when the system is in thermodynamic equilibrium . as a consequence ,
the total energy is now a functional of the density @xmath16 , the spin density @xmath17 and the paramagnetic current density @xmath18.@xcite we do not take into account correlation effects , which means that we operate strictly in the integer quantum hall regime , since correlation is responsible for the energy gaps that cause the fractional quantization . due to the fact that the correlation part of the energy is disregarded and exchange does not depend on the current density the total energy functional depends only on the total and the spin density or equivalently , since the spin - quantization axis is well defined by the strong magnetic field , on the density of spin up and spin down electrons . in the kohn - sham scheme of density - functional theory the ground state densities @xmath19 of the interacting electron system
are expressed in terms of a set of @xmath20 kohn - sham orbitals @xmath21 : @xmath22 where @xmath23 denotes up or down spin respectively and @xmath20 is the number of electrons with spin @xmath24 .
these orbitals satisfy the kohn - sham equations : @xmath25 with @xmath26 the effective one - particle potential : @xmath27;{\bf r } ) + v^{\sigma}_{x}([n^{\uparrow},n^{\downarrow } ] ; { \bf r } ) , \label{veff}\ ] ] where @xmath28 is the hartree potential and @xmath29 the exchange potential . here
we have disregarded the exchange - correlation vector potential @xmath30 which gives only a very small contribution to the total energy.@xcite the kohn - sham ground state wave function is a slater determinant of the @xmath20 lowest kohn - sham orbitals : @xmath31 it can be shown ( see sec .
[ sec : exp ] ) that in the limit of high magnetic fields the mixing between landau level wave functions vanishes .
consider , _
e.g. _ , the case of a single occupied , spin resolved landau level . for large magnetic fields the kohn - sham ground state wave function , eq .
( [ ksground ] ) , reduces to a slater determinant of lowest landau level wave functions with integer filling . as a consequence
the scheme can not describe fractional filling in the extreme quantum limit we are interested in .
the reason for the failure of this procedure is the fact that the crucial assumption of the kohn - sham scheme , namely that the interacting @xmath32-representable densities are also non - interacting @xmath32-representable,@xcite breaks down in the high magnetic field limit .
it is not possible to extend the kohn - sham scheme to fractional filling since in that case the effective potential , @xmath26 , is no longer a unique functional of the total and spin density .
however , at sufficiently high magnetic fields the basis wave functions are known to be just landau level wave functions .
we may then use a variational scheme which allows fractional filling and in which @xmath33 $ ] is a unique functional of the total and spin density .
the equilibrium density distribution @xmath34 which minimizes the total energy @xmath35 of the system can be found by solving : @xmath36}{\delta n^{\uparrow}(x)}=
\frac{\delta e[n^{\uparrow},n^{\downarrow}]}{\delta n^{\downarrow}(x ) } = \mu , \label{chempot}\ ] ] where @xmath37 is the chemical potential and the total energy of the quantum wire is given by : @xmath38 = t[n ] + e_{c}[n ] + e_{h}[n ] + e_{z}[n^{\uparrow},n^{\downarrow } ] + e_{x}[n^{\uparrow},n^{\downarrow } ] .
\label{totenergy}\ ] ] the total energy , eq .
( [ totenergy ] ) , consists of five contributions : the first term on the right hand side is the kinetic energy which we define as a functional of the total density : @xmath39 = \sum_{n \sigma } \int_{-\infty}^{\infty } dx \ ; n_{n}^{\sigma}(x)\ ; ( n+\frac{1}{2 } ) \hbar \omega_{c } , \label{kinen}\ ] ] where @xmath40 labels the landau levels .
the densities @xmath41 are the partial densities for a given spin direction @xmath24 and landau level @xmath40 . the total electron density is found by summing over all occupied landau levels and over spin directions : @xmath42 and is for sufficiently high magnetic fields given by : @xmath43 where @xmath44 is the quantum number of an electron with momentum @xmath45 , @xmath46 are the single particle bulk landau level wave functions and @xmath47 is the local filling factor for electrons with spin @xmath24 in landau level @xmath40 . for well behaved
confinement potentials the partial densities can be deduced from the total densities @xmath48 and @xmath49 . for these potentials
the total filling factor for a given spin direction , @xmath50 , lying between @xmath51 and @xmath52 , always consists of @xmath51 completely filled landau levels and a partially filled one .
this means that the kinetic energy ( and also the exchange energy , see eqs .
( [ exchen ] ) and ( [ genldaexch ] ) ) is still a functional of the _ total _ and not the partial densities .
the second contribution is the confinement energy which is given by : @xmath53 = \int_{-\infty}^{\infty } dx \ ; n(x)\ ; e v_{c}(x ) .
\label{confen}\ ] ] the electrostatic hartree energy is : @xmath54 = \int_{-\infty}^{\infty } \;dx \ ; n(x)\ ; e v_{h}[n ] , \label{harten}\ ] ] where the hartree potential , @xmath55 $ ] , is itself a functional of the total density .
for the strictly two dimensional electron gas ( see ref . for the case of a quasi - two dimensional electron gas ) it is given by : @xmath56=- \frac{e}{2 \pi \epsilon l_{b}}\;\int_{-\infty}^{\infty } dx ' \;n(x')\;\ln{\left|x - x'\right|}. \label{hartree}\ ] ] the fourth term denotes the zeeman energy : @xmath57 = \frac{1}{2 } g \mu_{b } b_{0}\;\int_{-\infty}^{\infty}\;dx \left\ { n^{\uparrow}(x ) - n^{\downarrow}(x ) \right\ } , \label{zeemanen}\ ] ] where @xmath58 is the bare land factor .
our only concern left is the explicit form of the last term in eq .
( [ totenergy ] ) , which is the exchange - energy functional . in the local density approximation ( lda )
it reads : @xmath59 = \int_{-\infty}^{\infty } dx \;\left\ { n^{\uparrow}(x)\ ; \epsilon_{x}(n^{\uparrow}(x ) ) + n^{\downarrow}(x ) \ ; \epsilon_{x}(n^{\downarrow}(x ) ) \right\}. \label{exchen}\ ] ] here @xmath60 denotes the one - particle exchange energy of the homogeneous 2deg with ground state densities @xmath48 and @xmath49 , which is magnetic field dependent : in the absence of a magnetic field , it is proportional to @xmath61 ( ref . ) , but in the magnetic quantum limit considered here , the local exchange energy per electron is : @xmath62 where the coefficients @xmath63 describe the exchange coupling between landau levels @xmath40 and @xmath64
. they can be found by calculating the exchange energy per particle of an extended @xmath40 landau level system.@xcite for two occupied levels , _
e.g. _ , they are @xmath65 , @xmath66 and @xmath67 .
the local density approximation is clearly only justified when the density variations are small on a characteristic length scale , which in our case is the magnetic length . in practice , however , it often turns out to be a useful and accurate tool even in cases in which this condition is not met.@xcite because we expect the results for higher landau levels to be qualitatively the same , we will in the following restrict ourselves to the magnetic quantum limit , for which only the lowest landau level is occupied . in the space of the lowest ( spin - polarized ) landau level ,
the kinetic energy per particle is constant and may be disregarded .
the validity of this approximation is discussed in sec .
[ sec : exp ] .
the density is in this case given by : @xmath68 where we have replaced the sum over @xmath69 in eq .
( [ genconv ] ) by an integral and substituted the lowest landau level wave functions .
this relation shows that in the extreme quantum limit there exists a one to one correspondence between the density and the filling factor which enables us to use the latter as the variational function in the determination of the ground state of the system ( see sec .
[ sec : resdis ] ) .
the exchange energy per electron in the lowest landau level ( @xmath70 ) reduces to : @xmath71 _ i.e. _ the one - particle exchange energy for a given spin direction is just proportional to the corresponding electron density . on substituting eqs .
( [ hartree ] ) , ( [ convolution ] ) and ( [ ldaexch ] ) in eq . ( [ totenergy ] ) for the total energy and performing the functional derivatives , eqs .
( [ chempot ] ) become : [ potentials ] @xmath72}{\delta n^{\uparrow}(x ) } & = & e v_{c}(x ) + e v_{h}[n ] + 2 \epsilon_{x}[n^{\uparrow } ] + \frac{1}{2 } g \mu_{b } b_{0 } = \mu , \\
\frac{\delta e[n^{\uparrow},n^{\downarrow}]}{\delta n^{\downarrow}(x ) } & = & e v_{c}(x)+e v_{h}[n ] + 2 \epsilon_{x}[n^{\downarrow } ] - \frac{1}{2 } g \mu_{b } b_{0 } = \mu.\end{aligned}\ ] ] the numerical treatment of these coupled equations and the results will be presented in the next section .
as described above we obtain the ground state occupation numbers by numerically minimizing the total energy with respect to the filling factors , while keeping the total number of electrons constant .
we discretize the filling factor and impose the boundary conditions : @xmath73 . the numerical algorithm that minimizes the total energy uses a sequential quadratic programming method and is very stable .
numerical integration of the singular integrand in eq .
( [ hartree ] ) requires some care . by sampling the integrand equidistantly and symmetrically around the divergence , @xmath74 , and by sampling enough points to avoid oscillatory behavior in the resulting integral .
an advantage of the present numerical scheme is that it can be very easy generalized to different confinement potentials .
it should be noted , however , that the choice of confinement potential is not completely arbitrary . for _
e.g. _ a hard wall potential there is a stronger mixing with higher landau levels near the edges and for such a confinement eqs .
( [ convolution ] ) and ( [ ldaexch ] ) may be no longer accurate
. with the approach outlined above we can obtain , to a very good approximation , both the electrostatic solution ( by neglecting the exchange term ) and the hartree - fock solution as calculated by dempsey _ et al._@xcite ( by forcing the filling factors to be integer valued ) .
therefore we expect our approach to give a good description of the intermediate regime between these two extremes . in fig .
[ fig : fillfact ] ground state occupation numbers obtained with different methods are plotted as a function of the confinement strength @xmath7 .
[ fig : fillfact](a ) shows the purely electrostatic solution , the hartree - fock solution is plotted in fig .
[ fig : fillfact](b ) and the solution including exchange in the local density approximation is shown in fig .
[ fig : fillfact](c ) .
the small incompressible regions found in the electrostatic solution are caused by zeeman splitting .
the calculations were performed using a magnetic field of 7.2 t which for filling factor @xmath75 in the bulk corresponds to a zero - field bulk density of @xmath76 @xmath77 .
for gaas the static dielectric constant @xmath78 and the bare land factor @xmath79 .
the filling factors are plotted for a constant number of electrons and confinement strengths @xmath80 0.035 , 0.041 and 0.047 respectively .
converted to energy level spacings of a parabolic confinement potential of the form : @xmath14 these values correspond to @xmath81 mev , 3.8 mev and 4.1 mev respectively . as can be seen from fig .
[ fig : fillfact](c ) the effect of exchange on the electrostatic solution is a reduction of the compressible regions because it favors integer filling .
however , a comparison with fig .
[ fig : fillfact](b ) shows that for soft confinement potentials the hartree - fock approximation no longer holds , _
i.e. _ the electrostatic interaction overcomes the tendency of exchange to form an incompressible ground state .
note , however , that the solution for @xmath82 in fig . [ fig : fillfact](c ) is almost identical to the corresponding hartree - fock solution shown in fig .
[ fig : fillfact](b ) , which shows that for confinement potentials that are strong enough the solution is forced into the integer filling regime , in agreement with the qualitative picture given in ref . .
note that this transition from incompressible to compressible state is a genuine correlation effect since it corresponds to the mixing of different single - determinant configurations , in spite of our exchange - only potential .
it is clear from fig .
[ fig : fillfact](c ) that the width of the incompressible region is of the same order of magnitude as the width of the outermost compressible strip , even for the relatively wide strips we consider here ( approximately 420 nm for a magnetic field of 7.2 t ) .
this in contrast to the electrostatic description of edge states shown in fig .
[ fig : fillfact](a ) , where the width of the incompressible region is always much smaller than that of the compressible one . however , for soft confinement potentials the solution including exchange does not deviate dramatically from the purely electrostatic one , which obviously implies that the hartree - fock approach breaks down in this regime . only a strong confinement potential can cause incompressible edge channels , although in fig .
[ fig : fillfact](c ) the confinement is still not strong enough to reduce the splitting to that of the bare zeeman splitting . in the case of even harder confinement ( @xmath83 , _ i.e. _ @xmath84 mev )
we find that the splitting indeed reduces to this minimum value ( not shown ) .
note that the entire range from the electrostatic to the ( unpolarized ) hartree - fock regime can be realized in realistic confinement potentials with @xmath85 ranging from 3.4 mev to 5.3 mev .
confinement potentials with level spacings of a few mev have been realized@xcite so it should be possible to test these results by experiments . a possible method to stiffen the confinement
experimentally is fabricating the 2deg closer to the surface of the algaas / gaas heterostructure .
in essence , our method to include exchange is using an position dependent @xmath58-factor that depends on the local density . a simple approximation to this method would be to use the enhanced @xmath58-factor corresponding to a single occupied spin level.@xcite this method does increase the width of the incompressible strip but in general does not reproduce the qualitative and quantitative features obtained with our method.@xcite in fig .
[ fig : potentials ] the calculated one - particle potentials in the wire are plotted for both spin directions .
these potentials correspond to the ground state density distribution of fig .
[ fig : fillfact](c ) with @xmath86 which is also included in the figure .
[ fig : potentials](a ) shows the electrostatic potential , which consists of the hartree and confinement potential .
also plotted are the exchange potential and the total potential for the majority - spin electrons ,
_ i.e. _ electrons occupying the lowest spin level . in fig .
[ fig : potentials](b ) the same can be seen for the minority - spin electrons . in the figures
the constant zeeman term has been omitted for simplicity so that the total potential is just the sum of the electrostatic and exchange potential .
[ fig : potentials](a ) shows that for a majority spin electron the energetically most favourable position is near the edges .
this is because these electrons are forced by exchange to form an electrostatically unfavourable density profile in which no screening of the confinement potential is possible , _
i.e. _ the density of majority spin electrons is constant in the bulk of the wire .
this in contrast to the minority spin electrons , for which the potential in the bulk is flat .
this is due to the fact that these electrons still have the freedom to form a density distribution which can screen the external potential without being influenced much by the exchange interaction .
the typical voltage drop from the edge to the middle of the wire has experimentally realistic values of @xmath87 300 mv for the bare confinement potential and @xmath87 30 mv for the selfconsistent potential .
we also performed calculations of the widths of compressible and incompressible regions as a function of the confinement strength .
the results are visible in fig .
[ fig : phasetr ] where the widths of the outermost incompressible region ( i ) and innermost compressible region ( c ) have been plotted as a function of @xmath7 .
if no incompressible strip is present in the middle of the wire , c is defined as half the width of the total compressible middle region ( see inset ) . in the figure
the points do not always lie on the smooth curve which serves as a guide for the eye .
this is due to the spacing between sampling points ( 0.2 @xmath15 ) , which imposes a upper bound on the accuracy of the calculated widths .
the compressible strip shrinks rapidly as the confinement is increased but if the width has reached a value of roughly a few magnetic lengths , the further decrease becomes very slow and in practice we always find a small but finite compressible region .
it is clear , however , that a hartree - fock treatment should give good results for hard confinements where small compressible regions exist .
we recover the spin - polarizing transition of ref . since the incompressible strip between spin - up and spin - down channels ( i in fig . [ fig : phasetr ] ) drops to zero at a certain critical value of the confinement strength ( neglecting zeeman splitting ) .
both the value for the critical confinement strength and the overall shape of the curve agree well with the results of ref . .
an important consequence of exchange on the distribution of electrons for experiments is the increased separation between the compressible regions .
this causes a strong decrease in inter - edge channel scattering because of the reduced overlap of the edge - channel wave functions , which leads to an increase in the spin - flip equilibration length of the edge channels .
this in contrast to the assumption of an edge channel separation of the order of one magnetic length in ref . .
using a typical edge channel separation of a few magnetic lengths in accordance with the present results , khaetskii s theory@xcite would give a much longer equilibration length .
the exchange enhanced channel separation has also consequences for the two - terminal magnetoconductance of a narrow channel or point contact . in ref .
, chklovskii _ et al .
_ discuss the conductance for these systems in the framework of their electrostatic description of the channels .
they obtain conductance quantization but , due to the small width of the incompressible regions , the calculated plateau widths are much smaller than those experimentally observed .
they attribute this discrepancy between theory and experiment to the presence of disorder .
however , we propose as an alternative explanation the electronic exchange interaction which we find to strongly enhance the width of the insulating strips .
in this subsection we study the accuracy of our theory and how our present results are altered if we take into account realistic features which do not change the qualitative results but are important to describe experimental situations .
first we investigate the validity of eq .
( [ convolution ] ) for high magnetic fields where mixing with higher landau levels was disregarded . to this end
we use a parabolic confinement @xmath14 and a density distribution of the form : @xmath88 , where @xmath89 is the width of the wire .
the 2deg in the presence of the uniform magnetic field and parabolic confinement potential is our unperturbed system and we treat the hartree potential as a perturbation .
the energy levels of the unperturbed system are @xmath90 , where @xmath91 is the renormalized cyclotron frequency and @xmath92 is the renormalized electron mass : @xmath93 .
the wave functions are the bulk landau level wave functions with effective magnetic length @xmath94.@xcite according to first order perturbation theory , the ground state wave function , @xmath95 , of the electron at the edge , with quantum number @xmath96 , is approximately : @xmath97 where we have used the hartree potential as the perturbation . a straightforward calculation of the coefficient of @xmath98 ( which of course depends on @xmath89 ) shows that for @xmath99 , a magnetic field of 5 t and a typical value of @xmath100 mev for the parabolic confinement , the mixing with the first landau level wave function is approximately 10% and that this value decreases as @xmath101 with increasing magnetic field .
we have to conclude that the dip in the electrostatic , _ i.e. _ hartree and confinement , potential at the edges which is responsible for the spin - polarizing transition and ultimately for the compressible edge state , can not be compensated by mixing with higher landau levels , _
i.e. _ that our approximation is allowed in the high magnetic field limit considered here .
by performing a similar calculation it is possible to show that the mixing due the exchange - correlation potential used by ferconi and vignale@xcite is only a few percent .
another approximation which gives rise to quantitative deviations from our theory is the assumption that the electron gas is strictly two dimensional . in reality
the electron wave functions also extend in the @xmath102-direction and a form factor @xmath103 in the coulomb interaction , @xmath104 , takes this effect into account .
calculations of edge channel splitting including a form factor in the hartree - fock approximation have already been calculated by rijkels and bauer.@xcite by comparison with ref .
we estimate the values of @xmath7 to decrease by about 0.004 if a form factor would be included into our calculations .
this shifts the range of @xmath85 in which both the electrostatic and the hartree - fock regime are unreliable to lower energies , _ i.e. _ to 3.2 mev-5.1 mev .
the present theory can be easily generalized to a variety of systems . in the next subsections we consider two special cases , which are experimentally relevant . as a first application we have carried out calculations for a system like the one considered in sec . [
sec : system ] , but now in the presence of an infinite plane of constant potential at a distance @xmath105 from the 2deg , which may represent a real back gate or a plane of not fully depleted donors parallel to the 2deg .
this gate can be easily included into the calculations by adding the potential of a mirror charge distribution of opposite sign located at a distance @xmath106 from the electron gas : @xmath107 in fig .
[ fig : nubackgate ] we have plotted the solutions that minimize the total energy of the system including the back gate as a function of the distance @xmath105 to the wire for a given number of electrons and a fixed confinement potential . it is clear from the figure that moving the back gate closer to the 2deg has the effect of reducing the width of the compressible regions .
this is to be expected because the back gate screens the long - range coulomb interaction in the electron gas , thus effectively reducing the importance of the direct interaction relative to the exchange interaction and thereby forcing the solution for small distances @xmath105 into the hartree - fock regime .
it should be noted that the description of the back gate given here does not apply for a quantum wire defined by a depletion gate . in that case
the back gate would also screen the charge on the depletion gates , thus increasingly altering the confinement potential as the back gate is moved closer to the 2deg . in that case
a fully self - consistent calculation is necessary .
furthermore , for our results to be correct , the distance from the gate to the 2deg should not be too small since in that case also the short - range exchange interaction would be influenced by the gate .
a system that is very similar to the strip previously considered is a quantum dot in a high magnetic field , especially when the number of electrons in the dot is large .
there are two main differences . in the first place we use the symmetric gauge to describe the dot . in this gauge
the wave functions in the lowest landau level , labeled by the quantum number of angular momentum , are @xmath108 where @xmath109 .
the second difference is that because the system is finite the filling factor @xmath110 is no longer labeled by the continuous variable @xmath69 but by the discrete quantum number @xmath111 . as a consequence the radial symmetric density is now given by : @xmath112 where @xmath113 is the filling factor for a given angular momentum @xmath111 and spin direction @xmath24 .
we impose as boundary conditions that @xmath114 and that the total charge in one spin level @xmath115 is an integer . due to the circular symmetry of the dot
the hartree potential is in this case given by : @xmath116 an analysis analogous to the one presented in sec .
[ sec : exp ] shows that for high magnetic fields ( approximately 5 t ) the mixing with higher landau levels due to the hartree potential and the exchange - correlation potentials used in ref .
is only a few percent . as a consequence ,
fractional occupation numbers are essential , which can not be described by the kohn - sham scheme of ref . .
we have calculated the ground state of a dot consisting of 40 electrons using a parabolic confinement potential @xmath117 with varying strength @xmath7 in a magnetic field of 5.0 t. to find the ground state for a given magnetic field we vary the distribution of electrons among the occupied spin levels while keeping the total number of electrons constant .
the results are plotted in fig .
[ fig : nuqdot ] .
the ground states are similar to those of the strip except for the fact that the middle region is wider .
this is due to the fact that the wave functions are centered around the radii @xmath118 , where @xmath111 is the angular momentum index , and not around equidistant points like in the strip . in ref .
a selfconsistent calculation was performed for the density profile of a quantum dot containing 39 electrons using a modified hartree form for the electrostatic electron - electron interaction : @xmath119 thus taking into account the finite @xmath102-extent @xmath120 of the wave functions and a back gate at distance @xmath105 .
again we want to extend these calculations to include exchange but we must keep in mind that a finite @xmath121 also reduces the exchange interaction by a form factor : @xmath122 where @xmath121 is given in units of the magnetic length and @xmath123x@xmath124 denotes the complementary error function . a calculation using interaction eq .
( [ veemceuen ] ) and the local exchange contribution reduced by the form factor eq .
( [ formfactor ] ) gives a ground state which differs from that of mceuen _ et al .
_ in the sense that the widths of the incompressible ( compressible ) regions are larger ( smaller ) in our solution .
we have calculated the addition spectrum of dots consisting of 37 and 38 electrons in order to determine the effect of exchange on the inter - level transitions that are responsible for the oscillatory behavior of the addition energy . here
we take into account the fact that the magnetic length is renormalised by the magnetic field , eq .
( [ renmaglen ] ) , in order to compare the energies at different magnetic fields .
the addition energy @xmath125 needed for adding an extra electron to a dot of @xmath40 electrons is : @xmath126 where @xmath127 is the total energy given by eq .
( [ totenergy ] ) .
the results are depicted in fig .
[ fig : addition ] .
it should be mentioned that our results are very close to the hartree - fock solutions .
this is due to the fact that the combination of the back gate and exchange force the solution into the integer filling regime .
the inset of fig .
[ fig : addition ] shows the energy dependence of different dot configurations with magnetic field for a dot containing 39 electrons .
the ground state is formed by the configuration with lowest energy .
the numbers in the inset represent the number of electrons in the majority ( _ i.e. _ lowest ) spin level for the indicated curve . as the magnetic field increases , the degeneracy of the spin levels is increased and at certain magnetic fields an electron can jump from the upper to the lower level , causing the kinks in the ground state energy .
when all electrons are in the lowest spin level no transitions can occur and the curves for both the total and addition energy are smooth .
recent calculations@xcite and measurements@xcite indicate that at higher magnetic fields than the ones shown in fig .
[ fig : addition ] an edge reconstruction occurs , resulting in an extra kink in the curve for the addition energy in the regime where only one spin level is occupied . performing a calculation with filling factors that are restricted to integer values , we are able to reproduce this feature
. however , for the small confinement strength @xmath7 at which this transition occurs , the fractional - filling solution is not so well converged . in spite of the fact that we find evidence for the edge reconstruction
, the numerical accuracy of our solution does not allow decisive conlusions . in comparison with the results obtained in ref .
the oscillations found in our calculations have a larger period and amplitude by a factor of approximately 1.5 .
the increase in amplitude of the oscillations can be qualitatively understood in terms of the capacitance model for the island proposed by evans _ et al._@xcite and is due to the increase of the width of the incompressible strip in the picture including exchange .
this results in a decrease of the capacitance between the inner and outer compresible regions where the extra electron can be added and hence in an increase in the peak - to - peak amplitude . the larger period can be explained in a similar fashion : exchange effects reduce the widths of the compressible strips and thus the capacitances @xmath128 and @xmath129 between the respective compressible strips and the back gate which results in an increased period of the oscillations compared to the electrostatic treatment of mceuen _ et al._. compared with their experimental data , the period of the oscillations found in our calculations is about 0.3 t too large whereas their calculations give a period which is 0.25 t too small .
however , these values depend sensitively on the calculated ground state density since the total energy , with a typical value around 1 ev , has to be calculated with an absolute accuracy of at least 0.1 mev in order to resolve the oscillations and therefore small numerical deviations from the true ground state would drastically influence the addition energies .
we have used density - functional theory in strong magnetic fields to investigate the effects of the electron - electron interaction on edge states in the integer quantum hall regime .
we have included exchange in the local density approximation . in this approximation smooth density distributions corresponding to fractional filling at the edges
can be treated , which is beyond a hartree - fock treatment of edge channels.@xcite to describe fractional occupation could otherwise only be achieved by exact diagonalization or configuration interaction calculations.@xcite we have found that the width of the edge channels is strongly reduced due to exchange effects and that the width of the incompressible strips is of the order of several magnetic lengths , which should have a strong influence on the spin - flip equilibration length and the two - terminal conductance of a point contact .
we predict a range of confinement potentials @xmath85=3.2 mev-5.1 mev in which the entire regime from electrostatic to hartree - fock is covered .
furthermore we have calculated ground state properties of the system in the presence of a back gate parallel to the quantum wire .
we established an increasingly important role for the exchange interaction as the back gate is closed in on the 2deg .
as a second application we have calculated the ground state of a quantum dot in a strong magnetic field .
we found that the results are similar to those found for the quantum wire .
we compared addition energies for the dot with the selfconsistent calculations of mceuen _ et al._@xcite and found that the oscillations in the addition energies have larger amplitudes and periods in our calculations .
the authors would like to thank sasha khaetskii , yuli nazarov and lex rijkels for valuable discussions and a.h .
macdonald for a reprint of ref . .
this work is part of the research program of the `` stichting voor fundamenteel onderzoek der materie '' ( fom ) , which is financially supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek '' ( nwo ) .
halperin , phys .
b * 25 * , 2185 ( 1982 ) .
beenakker , phys .
lett . * 64 * , 216 ( 1990 ) .
a.h . macdonald , phys .
lett . * 64 * , 220 ( 1990 ) .
wen , phys .
lett . * 64 * , 2206 , ( 1990 ) .
a.h . macdonald , s.r .
eric yang , and m.d .
johnson , aust . j. phys . *
46 * , 345 ( 1993 ) .
chklovskii , b.i .
shklovskii , and l.i .
glazman , phys .
b * 46 * , 4026 ( 1992 ) .
chklovskii , k.a .
matveev , and b.i .
shklovskii , phys .
b * 47 * , 12605 ( 1993 ) .
j. dempsey , b.y .
gelfand , and b.i .
halperin , phys .
* 70 * , 3639 ( 1993 ) . c. de c. chamon and x.g .
wen , phys .
b , * 49 * , 8227 ( 1994 ) .
stoof and g.e.w .
bauer , proceedings of the 11th int .
on high magnetic fields in semiconductor physics , semi mag 94 , cambridge , ma , usa , 8 - 12 august 1994 ( world scientific , singapore ) , to be published .
m. ferconi and g. vignale , phys .
b * 50 * , 14772 ( 1994 ) . g. vignale and m. rasolt , phys .
* 59 * , 2360 ( 1987 ) ; phys .
b * 37 * , 10685 ( 1988 ) .
e.g. _ , r.m .
dreizler and e.k.u .
gross , _ density functional theory .
an approach to the quantum many - body problem _ , ( springer , berlin , 1990 ) .
l. rijkels and g.e.w .
bauer , phys .
b * 50 * , 8629 ( 1994 ) .
f. stern , phys .
lett . * 30 * , 278 ( 1973 ) . j.m .
kinaret and p.a .
lee , phys .
b * 42 * , 11768 ( 1990 ) .
mceuen , e.b .
foxman , j. kinaret , u. meirav , m.a .
kastner , n.s .
wingreen , and s.j .
wind , phys .
b * 45 * , 11419 ( 1992 ) .
ashoori , h.l .
stormer , j.s .
weiner , l.n .
pfeifer , k.w .
baldwin , and k.w .
west , phys .
lett . * 71 * , 613 ( 1993 ) .
janak , phys . rev .
* 178 * , 1416 ( 1969 ) ; t. ando and y. uemura , j. phys .
japan * 37 * , 1044 ( 1974 ) . calculating the ground state in this manner
does increase the width of the incompressible region and it slightly reduces the width of the edge channels but the effects are smaller than in those obtained with our method .
khaetskii , phys . rev .
b * 45 * , 13777 ( 1992 ) .
o. klein , c. de c. chamon , d. tang , d.m .
abusch - magder , u. meirav , x .-
wen , m.a .
kastner , and s.j .
wind , phys .
lett . * 74 * , 785 ( 1995 ) .
evans , l.i .
glazman , and b.i .
shklovskii , phys .
rev.b * 48 * , 11120 ( 1993 ) . | we study the competition between the exchange and the direct coulomb interaction near the edge of a two - dimensional electron gas in a strong magnetic field using density - functional theory in a local approximation for the exchange - energy functional .
exchange is shown to play a significant role in reducing the spatial extent of the compressible edge channel regions obtained from an electrostatic description .
the transition from the incompressible edge channels of the hartree - fock picture to the broad , compressible strips predicted by electrostatics occurs within a narrow and experimentally accessible range of confinement strengths .
+ + pacs numbers : 73.20.dx ; 71.10.+x , 71.45.gm + + submitted to phys .
rev . |
superconducting qubits @xcite are very well suited for the task of solid - state quantum computation , because two of the most obvious decoherence sources in solid - state systems are supressed : quasiparticle excitations experience an energy gap and phonons are frozen out at low temperatures @xcite .
the computational hilbert space is engineered using josephson tunnel junctions that are characterized by two competing energy scales : the josephson coupling of a junction with critical current @xmath0 , @xmath1 , and the charging energy @xmath2 of a single cooper pair on the geometric capacitance @xmath3 of the junction . here
@xmath4 is the superconducting flux quantum .
there is a variety of qubit proposals classified by the ratio of this energies .
whereas another contribution in this volume @xcite focuses on the case of charge qubits , @xmath5 , this contribution is motivated by flux qubit physics , @xmath6
. however , most of the discussion has its counterpart in other superconducting setups as well . .
the circuits a)-c ) are connected to filtering and electronics ( not drawn ) ] specifically , we discuss a three junction qubit @xcite , a micrometer - sized low - inductance superconducting loop containing three josephson tunnel junctions ( fig .
[ figexp ] ) . by applying an external flux
@xmath7 a persistent supercurrent can be induced in the loop . for values where @xmath7 is close to a half - integer number of flux quanta , two states with persistent currents of opposite sign
are nearly degenerate but separated by an energy barrier .
we will assume here that the system is operated near @xmath8 .
the persistent currents in the classically stable states have here a magnitude @xmath9 .
tunneling through the barrier causes a coupling between the two states , and at low energies the loop can be described by a hamiltonian of a two state system @xcite , @xmath10 where @xmath11 and @xmath12 are pauli matrices .
the two eigenvectors of @xmath11 correspond to states that have a left or a right circulating current and will be denoted as @xmath13 and @xmath14 .
the energy bias @xmath15 is controlled by the externally applied field @xmath7 .
we follow @xcite and define @xmath16 as the tunnel splitting at @xmath17 , such that @xmath18 with @xmath19 the tunnel coupling between the persistent - current states .
this system has two energy eigen values @xmath20 , such that the level separation @xmath21 gives @xmath22 . in general @xmath23
is a function of @xmath24 .
however , it varies on the scale of the single junction plasma frequency , which is much above the typical energy range at which the qubit is operated , such that we can assume @xmath23 to be constant for the purpose of this paper . in the experiments
@xmath7 can be controlled by applying a magnetic field with a superconducting coil at a distance from the qubit and for local control one can apply currents to superconducting lines , fabricated on - chip in the vicinity of the qubit .
the qubit s quantum dynamics will be controlled with resonant microwave pulses ( i. e. by rabi oscillations ) . in recent experiments
the qubits were operated at @xmath25 or @xmath26 @xcite .
the numerical values given in this paper will concentrate on the former case . at this point
, there is a good trade - off between a system with significant tunneling , and a system with @xmath11-like eigenstates that can be used for qubit - qubit couplings and measuring qubit states @xcite .
the qubit has a magnetic dipole moment as a result of the clockwise or counter - clockwise persistent current the corresponding flux in the loop is much smaller than the applied flux @xmath27 , but large enough to be detected with a squid .
this will be used for measuring the qubit states . for our two - level system eq .
( [ 2lsham ] ) , this means that both manipulation and readout couple to @xmath11 .
consequently , the nyquist noise produced by the necessary external circuitry will couple in as flux noise and hence couple to @xmath11 , giving @xmath28 a small , stochastically time - dependent part @xmath29 .
operation at @xmath26 has the advantage that the flux noise leads to less variation of @xmath30 . in the first experiments
@xcite this has turned out to be crucial for observing time - resolved quantum dynamics . here , the qubit states can be measured by incorporating the qubit inside the dc - squid loop . while not working that out in detail , the methods that we present can also be applied for the analysis of this approach .
this also applies to the analysis of the impact of electric dipole moments , represented by @xmath12 . with @xmath31 ,
these couple much less to the circuitry and will hence not be discussed here .
as the internal baths are well suppressed , the coupling to the electromagnetic environment ( circuitry , radiation noise ) becomes a dominant source of decoherence .
this is a subtle issue : it is not possible to couple the circuitry arbitrarily weakly or seal the experimental setup , because it has to remain possible to control the system .
one rather has to engineer the electromagnetic environment to combine good control with low unwanted back - action .
any linear electromagnetic environment can be described by an effective impedance @xmath32 .
if the circuit contains josephson junctions below their critical current , they can be included through their kinetic inductance @xmath33 , where @xmath34 is the average phase drop across the junction .
the circuitry disturbs the qubit through its johnson - nyquist noise , which has gaussian statistics and can thus be described by an effective spin - boson model @xcite . in this model
, the properties of the oscillator bath which forms the environment are characterized through a spectral function @xmath35 , which can be derived from the external impedance .
note , that other nonlinear elements such as tunnel junctions which can produce non - gaussian shot noise are generically _ not _ covered by oscillator bath models .
as explained above , the flux noise from an external circuit leads to @xmath36 in eq .
( [ 2lsham ] ) .
we parametrize the noise @xmath37 by its power spectrum @xmath38 thus , from the noise properties calculated by other means one can find @xmath35 as was explained in detail in @xcite . in this contribution , we would like to outline an alternative approach pioneered by leggett @xcite , where @xmath35 is derived from the classical friction induced by the environment . in reality , the combined system of squid and qubit will experience fluctuations arising from additional circuit elements at different temperatures , which can be treated in a rather straightforward manner .
we study a dc - squid in an electrical circuit as shown in fig.[figexp ] .
it contains two josephson junctions with phase drops denoted by @xmath39 .
we start by looking at the average phase @xmath40 across the read - out squid . analyzing the circuit with kirchhoff rules
, we find the equation of motion @xmath41 here , @xmath42 is the dynamical variable describing the circulating current in the loop which is controlled by the flux , @xmath43 is the bias current imposed by the source , @xmath44 is the admittance in parallel to the whole squid and @xmath45 its fourier transform .
the squid is described by the junction critical currents @xmath46 which are assumed to be equal , and their capacitances @xmath3 .
we now proceed by finding a static solution which sets the operation point @xmath47 and small fluctuations around them , @xmath48 .
the static solution reads @xmath49 where @xmath50 is the effective critical current of the squid . linearizing eq .
[ eq : tiltwashboard ] around this solution and fourier - transforming , we find that @xmath51 where @xmath52 is the effective impedance of the parallel circuit consisting of the @xmath53 , the kinetic inductance of the squid and the capacitance of its junctions .
neglecting self - inductance of the squid and the ( high - frequency ) internal plasma mode , we can straightforwardly substitute @xmath54 and split it into @xmath55 set by the externally applied flux @xmath56 through the squid loop and @xmath57 where @xmath58 is the mutual inductance between qubit and the squid and @xmath59 is the circulating current in the qubit as a function of the junction phases , which assumes values @xmath60 in the classically stable states . in order to analyze the backaction of the squid onto the qubit in the two - state approximation , eq .
( [ 2lsham ] ) , we have to get back to its full , continuous description , starting from the classical dynamcis .
these are equivalent to a particle , whose coordinates are the two independent junction phases in the three - junction loop , in a two - dimensional potential @xmath61 the details of this equation are explained in @xcite .
@xmath62 is the capacitance matrix describing the charging of the josephson junctions in the loop , @xmath63 contains the josephson energies of the junctions as a function of the junction phases and @xmath64 . is the ciculating current in the squid loop . the applied flux through the qubit @xmath65 is split into the flux from the external coil @xmath66 and the contribution form the squid .
using the above relations we find @xmath67 where @xmath68 is the non - fluctuating back - action from the squid .
from the two - dimensional problem , we can now restrict ourselves to the one - dimensional subspace defined by the preferred tunneling direction @xcite , which is described by an effective phase @xmath69 .
the potential restricted on this direction , @xmath70 has the form of a double well @xcite with stable minima situated at @xmath71 . in this way , we can expand @xmath72 . approximating the phase - dependence of the circulating current as @xmath73 where @xmath74 the circulating current in one of the stable minima of @xmath69 , we end up with the classical equation of motion of the qubit including the backaction and the friction induced from the squid @xmath75 \varphi \nonumber \\ & & = -\partial_\varphi u_{\rm 1d}(\varphi,\phi_{x , q}+\delta\phi_{\rm cl}).\end{aligned}\ ] ] from this form , encoded as @xmath76 we can use the prescription given in @xcite and identify the spectral function for the continuous , classical model as @xmath77 . from there
, we can do the two - state approximation for the particle in a double well @xcite and find @xmath35 in analogy to @xcite @xmath78 from @xmath35 , we can analyze the dynamics of the system by studying the reduced density matrix , i.e. the density matrix of the full system where the details of the environment have been integrated out , by a number of different methods . the low damping limit , @xmath79 for all frequencies , is most desirable for quantum computation .
thus , the energy - eigenstates of the qubit hamiltonian , eq . ( [ 2lsham ] ) , are the appropriate starting point of our discussion . in this case
, the relaxation rate @xmath80 ( and relaxation time @xmath81 ) are determined by the environmental spectral function @xmath82 at the frequency of the level separation @xmath21 of the qubit @xmath83 where @xmath84 is the temperature of the bath .
the dephasing rate @xmath85 ( and dephasing time @xmath86 ) is @xmath87 with @xmath88 .
these expressions have been derived in the context of nmr @xcite and recently been confirmed by a full path - integral analysis @xcite . in this paper ,
all rates are calculated for this regime . for performing efficient measurement
, one can afford to go to the strong damping regime .
a well - known approach to this problem , the noninteracting blip approximation ( niba ) has been derived in ref.@xcite .
this approximation gives good predictions at degeneracy , @xmath89 . at low @xmath90
it contains an artifact predicting incoherent dynamics even at weak damping . at high bias , @xmath91 and at strong damping
, it becomes asymptotically correct again .
we will not detail this approach here more , as it has been extensively covered in @xcite .
if @xmath35 is not smooth but contains strong peaks the situation becomes more involved : at some frequencies , @xmath35 may fall in the weak and at others in the strong damping limit . in some cases ,
whern @xmath92 holds at least for @xmath93 with some @xmath94 , this can be treated approximately : one can first renormalize @xmath95 through the high - frequency contributions @xcite and then perform a weak - damping approximation from the fixed - point hamiltonian .
this is detailed in ref .
@xcite . in the general case
, more involved methods such as flow equation renormalization @xcite have to be used .
from eq . [ jwsquid ] we see that engineering the decoherence induced by the measurement apparatus essentially means engineering @xmath32 .
this includes also the contributions due to the measurement apparatus . in this section
, we are going to outline and compare several options suggested in literature .
we assume a perfect current source that ramps the bias current @xmath43 through the squid .
the fact that the current source is non - ideal , and that the wiring to the squid chip has an impedance is all modeled by the impedance @xmath53 .
the wiring can be engineered such that for a very wide frequency range the impedance @xmath53 is on the order of the vacuum impedance , and can be modeled by its real part @xmath96 .
it typically has a value of @xmath97 .
it has been suggested @xcite to overdamp the squid by making the shunt circuit a simple resistor @xmath98 with @xmath99 .
this is inspired by an analogous setup for charge qubits , @xcite .
following the parameters given in @xcite , a squid with @xmath100na at @xmath101 biased at @xmath102na , we find @xmath103 .
together with @xmath104 1ff , this means that the squid is overdamped if @xmath105 . using eq.[jwsquid ] , we find that this provides an ohmic environment with drude - cutoff , @xmath106 where @xmath107 and @xmath108 . using the parameters from ref .
@xcite , @xmath109 , we find @xmath110 and @xmath111 .
thus , for our range of parameters ( which essentially correspond to weak coupling between squid and qubit ) , one still has _
low _ damping of the qubit from the ( internally overdamped ) environment at reasonable shunt resistances down to tens of ohms .
for such a setup , one can apply the continuous weak measurement theory as it is outlined e.g. in @xcite .
this way , one can readily describe the readout through measurement of @xmath32 which leaves the system on the superconducting branch . if one desires to read out the state by monitoring the voltage at bias currents _ above _ the @xmath112 , our analysis only describes the pre - measurement phase and at least shows that the system is hardly disturbed when the current is ramped .
-shunted dc - squid ( a ) and the @xmath113-shunted dc - squid ( b ) .
the squid is modeled as an inductance @xmath114 .
a shunt circuit , the superconducting capacitor @xmath115 or the @xmath116-@xmath115 series , is fabricated on chip very close to the squid . the noise that couples to the qubit results from johnson - nyquist voltage noise @xmath117 from the circuit s total impedance @xmath32 .
@xmath32 is formed by a parallel combination of the impedances of the leads @xmath118 , the shunt and the squid , such that @xmath119 , with @xmath120 for the circuit ( a ) ] for the @xmath121-shunted squid ( a ) and the @xmath113-shunted squid ( b ) , and corresponding @xmath122 in ( c ) and ( d ) respectively . for comparison , the dashed line in ( c )
shows a simple ohmic spectrum , @xmath123 with exponential cut off @xmath124 = 0.5 ghz and @xmath125 = 0.00062 .
the parameters used here are @xmath126 na and @xmath127 mk .
the squid with @xmath128 = 200 na is operated at @xmath129 and current biased at 120 na , a typical value for switching of the @xmath121-shunted circuit ( the @xmath113-shunted circuit switches at higher current values ) .
the mutual inductance @xmath130 = 8 ph ( i. e. @xmath131 ) .
the shunt is @xmath115 = 30 pf and for the @xmath113 shunt @xmath132 10 @xmath133 .
the leads are modeled by @xmath134 @xmath135 next , we consider a large superconducting capacitive shunt ( fig .
[ figsqcir]a , as implemented in refs .
the @xmath121 shunt only makes the effective mass of the squid s external phase @xmath136 very heavy .
the total impedance @xmath137 and @xmath35 are modeled as before , see fig .
[ figsqrez ] . as limiting values
, we find @xmath138 we can observe that this circuit is a weakly damped @xmath139-oscillator and it is clear from ( [ gmix ] ) and ( [ jwsquid ] ) that one should keep its resonance frequency @xmath140 , where @xmath141 has a maximum , away from the qubit s resonance @xmath142 .
this is usually done by chosing @xmath143 . for a @xmath121-shunted circuit with @xmath144 ,
this yields for @xmath145 @xmath146 the factor @xmath147 indicates a natural cut - off for @xmath148 , which prevents the ultraviolet divergence @xcite and which in much of the theoretical literature is introduced by hand . using eq .
[ gmix ] , we can directly analyze mixing times @xmath81 vs @xmath149 for typical sample parameters ( here calculated with the non - approximated version of @xmath150 ) , see ref .
@xcite for details .
the mixing rate is then @xmath151 . with the @xmath121-shunted circuit it seems possible to get @xmath81 values that are very long .
they are compatible with the ramp times of the squid , but too slow for fast repetition rates .
for the parameters used here they are in the range of 15 @xmath152 .
while this value is close to the desired order of magnitude , one has to be aware of the fact that at these high switching current values the linearization of the junction as a kinetic inductor may underestimate the actual noise . in that regime ,
phase diffusion between different minima of the washboard potential also becomes relevant and changes the noise properties @xcite .
as an alternative we will consider a shunt that is a series combination of a capacitor and a resistor ( fig . [ figsqcir]b ) ( @xmath113-shunted squid ) .
the @xmath113 shunt also adds damping at the plasma frequency of the squid , which is needed for realizing a high resolution of the squid readout ( i. e. for narrow switching - current histograms ) @xcite .
the total impedance @xmath153 of the two measurement circuits are modeled as in fig .
[ figsqcir ] . for
the circuit with the @xmath113 shunt @xmath154 the difference mainly concerns frequencies @xmath155 , where the @xmath121-shunted circuit has a stronger cutoff in @xmath156 and thereby a relaxation rate , that is several orders lower than for the @xmath113-shunted circuit .
given the values of @xmath35 from fig .
[ figsqrez ] one can directly see from the values of that an @xmath113-shunted circuit with otherwise similar parameters yields at @xmath157 10 ghz relaxation times that are about four orders of magnitude shorter .
so far , we have applied our modeling only to single qubits . in order to study entanglement in a controlled way and to eventually perform quantum algorithms ,
this has to be extended to coupled qubits .
there is a number of ways how to couple two solid - state qubits in a way which permits universal quantum compuation .
if the qubit states are given through real spins , one typically obtains a heisenberg - type exchange coupling . for other qubits ,
the three components of the pseudo - spin typically correspond to physically completely distinct variables . in our case
, @xmath158 corresponds to the flux through the loop whereas @xmath159 are charges .
consequently , one usually finds ising - type couplings .
the case of @xmath160 coupling , i.e. coupling by a component which is orthogonal to all possible single - qubit hamiltonians , has been extensively studied @xcite , because this type is straightforwardly realized as a _
tunable _ coupling of charge qubits @xcite .
we study the generic case of coupling the `` natural '' variables of the pseudospin to each other , which can be realized in flux qubits using a switchable superconducting transformer @xcite , but has also been experimentally utilized for coupling charge qubits by fixed capacitive interaction @xcite .
we model the hamiltonian of a system of two qubits , coupled via ising - type coupling .
each of the two qubits is described by the hamiltonian eq .
( [ 2lsham ] ) .
the coupling between the qubits is described by @xmath161 that represents e.g. inductive interaction .
thus , the complete two - qubit hamiltonian in the absence of a dissipative environment reads @xmath162 for two qubits , there are several ways to couple to the environment : both qubits may couple to a common bath such as picked up by coupling elements @xcite .
local readout and control electronics coupling to individual qubits @xcite can be described as coupling to two uncorrelated baths . in analogy to the procedure described above , one can determine the spectral functions of these baths by investigating the corresponding impedances . in the case of two uncorrelated baths , the full hamiltonian reads @xmath163 @xmath164 are collective coordinates of the bath . in the case of two qubits coupling to one common bath we model our two qubit system in a similar way with the hamiltonian @xmath165 where @xmath166 is a collective bath
coordinate similar to above .
we can derive formulae for relaxation and dephasing rates similar to eqs.([gmix ] ) and ( [ gphase ] ) .
our hilbert space is now four - dimensional .
we label the eigenstates as @xmath167 .
we chose @xmath168 to be the singlet state @xmath169 , which is always an eigenstate @xcite whereas @xmath170 are the energy eigenstates in the triplet subspace , which are typically _ not _ the eigenstates of @xmath171 .
as we have 4 levels , we have 6 independent possible quantum coherent oscillations , each of which has its own dephasing rate , as well as 4 relaxation channels , one of which has a vanishing rate indicating the existence of a stable thermal equilibrium point .
the expressions for the rates , although of similar form as in eqs .
( [ gmix ] ) and ( [ gphase ] ) are rather involved and are shown in @xcite .
figure [ fig_temp_dependence_1 ] displays the dependence of typical dephasing rates and the sum of all relaxation rates @xmath172 on temperature for the case @xmath173 with @xmath174 .
the rates are of the same magnitude for the case of one common bath and two distinct baths . if the temperature is increased above the roll off point set by the intrinsic energy scales , @xmath175 , where @xmath176ghz , the increase of the dephasing and relaxation rates follows a linear dependence , indicating that the environmental fluctuations are predominantly thermal . , @xmath28 and @xmath177 are all set to @xmath178 and the bath is assumed to be ohmic @xmath179 .
the upper panel shows the case of one common bath , the lower panel the case of two distinct baths . at the characteristic temperature of approximately @xmath180
the rates increase very steeply ] as a notable exception , in the case of one common bath the dephasing rates @xmath181 go to zero when the temperature is decreased while all other rates saturate for @xmath182 .
this can be understood as follows : the singlet state @xmath168 is left invariant by the hamiltonian of coupled qubits in a common bath , eq .
( [ hop_1bath ] ) , i.e. it is an energy eigenstate left unaffected by the environment .
superpositions of the singlet with _ another _ eigenstate are usually still unstable , because the other eigenstate generally suffers from decoherence .
however , the lowest - energy state of the triplet subspace @xmath183 can not decay by spontaneous emission and flip - less dephasing vanishes at @xmath184 , hence the dephasing rate between eigenstates @xmath168 and @xmath183 vanishes at low temperatures , see fig .
[ fig_temp_dependence_1 ] .
as shown in @xcite , there can be more `` protected '' transitions of this kind if the qubit parameters are adjusted such that the symmetry between the unperturbed qubit and the coupling to the bath is even higher , e.g. at the working point for a cphase operation .
the rates derived in the previous section are numerous and do strongly dependend on the tunable parameters of the qubit .
thus , they do not yet allow a full assesment of the performance as a quantum logic element .
a quantitative measure of how well a two - qubit setup performs a quantum logic gate operation are the gate quality factors introduced in @xcite : the fidelity , purity , quantum degree and entanglement capability .
these factors characterize the density matrices obtained after attempting to perform the gate operation in a hostile environment , starting from all possible initial conditions @xmath185 . to form all possible initial density matrices needed to calculate the gate quality factors
, we use the 16 unentangled product states @xmath186 , @xmath187 defined @xcite according to @xmath188 , ( @xmath189 ) , with @xmath190 , @xmath191 , @xmath192 , and @xmath193 .
they form one possible basis set for the superoperator @xmath194 which describes the open system dynamics such that @xmath195 @xcite .
the cnot gate is implemented using rectangular dc pulses and describing dissipation through the bloch - redfield equation as described in @xcite .
the fidelity is defined as @xmath196 the fidelity is a measure of how well a quantum logic operation was performed . clearly , the fidelity for the ideal quantum gate operation is equal to 1 .
the second quantifier is the purity @xmath197\textrm{,}$ ] which should be 1 in a pure and @xmath198 in a fully mixed state .
the purity characterizes the effects of decoherence .
the quantum degree measures nonlocality .
it is defined as the maximum overlap of the resulting density matrix after the quantum gate operation with the maximally entangled bell - states @xmath199 . for
an ideal entangling operation , e.g. the cnot gate , the quantum degree should be 1 .
it has been shown @xcite that all density operators that have an overlap with a maximally entangled state that is larger than the value @xmath200 @xcite violate the clauser - horne - shimony - holt ( chsh ) inequality and are thus non - local .
the entanglement capability @xmath201 is the smallest eigenvalue of the partially transposed density matrix for all possible unentangled input states @xmath186 .
( see below ) .
it has been shown @xcite to be negative for an entangled state .
this quantifier should be -0.5 , e.g. for the ideal cnot , thus characterizing a maximally entangled final state . in fig .
[ gqftdep ] , the deviations due to decoherence of the gate quality factors from their ideal values are shown .
similar to most of the rates , all gate quality factors saturate at temperatures below a threshold set by the qubit energy scales .
the deviations grow linearily at higher temperatures until they reach their theoretical maximum . comparing the different coupling scenarios , we see that at low temperatures , the purity and fidelity are higher for the case of one common bath , but if temperature is increased above this threshold , fidelity and purity are approximately equal for both the case of one common and two distinct baths .
this is related to the fact that in the case of one common bath all relaxation and dephasing rates vanish during the two - qubit - step of the cnot ( see @xcite for details ) , due to the special symmetries of the hamiltonian , when temperature goes to zero as discussed above .
still , the quantum degree and the entanglement capability tend towards the same value for both the case of one common and two distinct baths .
this is due to the fact that both quantum degree and entanglement capability are , different than fidelity and purity , not defined as mean values but rather characterize the `` best '' possible case of all given input states . in the recent work by thorwart and hnggi @xcite ,
the cnot gate was investigated for a @xmath202 coupling scheme and one common bath .
they find a pronounced degradation of the gate performance with gate quality factors only weakly depending on temperature .
if we set the dissipation and the intrinsic energy scale to the same values as in their work , we also observe only a weak decrease of the gate quality factors for both the case of one common bath and two distinct baths in the same temperature range discussed by thorwart and hnggi . however , see fig .
[ gqftdep ] , overall we find substantially better values .
this is due to the fact that for @xmath203 coupling , the hamiltonian does _ not _ commute with the coupling to the bath during the two - qubit steps of the pulse sequence , i.e. the symmetries of the coupling to the bath and the inter - qubit coupling are not compatible . to @xmath204 . in all cases
the dotted line indicates the upper bound set by the clauser - horne - shimony - holt inequality ] the dotted line in fig .
[ gqftdep ] shows that already at comparedly high temperature , about 20 qubit energies , a quantum degree larger than @xmath206 can be achieved . only then
, the clauser - horne - shimony - holt inequality is violated and non - local correlations between the qubits occur as described in @xcite .
thus , even under rather modest requirements on the experimental setup which seem to be feasible with present day technology , it appears to be possible to demonstrate nonlocality and entanglement between superconducting flux qubits .
it has been outlined , how one can model the decoherence of an electromagnetic environment inductively coupled to a superconducting flux qubit .
we have exemplified a procedure based on analyzing the classical friction induced by the environment for the specific case of the read - out squid .
it is shown that the squid can be effectively decoupled from the qubit if no bias current is applied .
the effect of the decoherence on relaxation and dephasing rates of single qubits has been discussed as well as the gate performance of coupled qubits .
we have shown that by carefully engineering the impedance and the symmetry of the coupling , one can reach excellent gate quality which complies with the demands of quantum computation .
mooij , t.p .
orlando , l. levitov , l. tian , c.h .
van der wal , and s. lloyd , science * 285 * , 1036 ( 1999 ) ; t.p .
orlando , j.e .
mooij , l. tian , c.h .
van der wal , l.s .
levitov , s. lloyd , and j.j .
mazo , phys .
b * 60 * , 15398 ( 1999 ) .
l. tian , l.s .
levitov , c.h .
van der wal , j.e .
mooij , t.p .
orlando , s. lloyd , c.j.p.m .
harmans , and j.j .
mazo in i. kulik and r. elliatiogly , _
quantum mesoscopic phenomena and mesoscopic devices in microelectronics _
( kluwer , dordrecht , 2000 ) , 429 . | on the way to solid - state quantum computing , overcoming decoherence is the central issue . in this contribution
, we discuss the modeling of decoherence of a superonducting flux qubit coupled to dissipative electronic circuitry .
we discuss its impact on single qubit decoherence rates and on the performance of two - qubit gates .
these results can be used for designing decoherence - optimal setups .
quantum computation is one of the central interdisciplinary research themes in present - day physics @xcite .
it promises a detailed understanding of the often counterintuitive predictions of basic quantum mechanics as well as a qualitative speedup of certain hard computational problems . a generic ,
although not necessarily exclusive , set of criteria for building quantum computers has been put forward by divincenzo @xcite .
the experimental realization of quantum bits has been pioneered in atomic physics , optics and nmr . there
, the approach is taken to use microscopic degrees of freedom which are well isolated and can be kept quantum coherent over long times .
efficient controls are attached to these degrees of freedom . even though these approaches are immensely succesful demonstrating elementary operations , it is not evident how they can be scaled up to macroscopic computers .
solid - state systems on the other hand have proven to be scalable in present - day classical computers .
several proposals for solid - state based _ quantum _ computers have been put forward , many of them in the context of superconductors @xcite .
as solid - state systems contain a macroscopic number of degrees of freedom , they are very sensitive to decoherence . mastering and
optimizing this decoherence is a formidable task and requires deep understanding of the physical system under investigation .
recent experimental success @xcite suggests that this task can in principle be performed . in this contribution
, we are going to study decoherence of superconducting qubits coupled to an electromagnetic environment which produces johnson - nyquist noise .
we show , how the decoherence properties can be engineered by carefully designing the environmental impedance . we will discuss how the decoherence affects the performance of a cnot operation . |
let @xmath0 be a coxeter system with an involutive automorphism @xmath1 .
a _ twisted identity _ is an element of the form @xmath9 for @xmath10 . in other words ,
the set @xmath3 of twisted identities is the orbit of the identity element under the twisted conjugation action of @xmath11 on itself .
our terminology appeared in @xcite and stems from the fact that when @xmath12 , the only twisted identity is the identity element @xmath13 .
as will be demonstrated in section [ se : motivation ] , the study of @xmath3 is motivated by its appeareance in a variety of situations .
for example , certain orbit decompositions of symmetric varieties have a close relationship with the subposet @xmath14 of the bruhat order on @xmath11 which is induced by @xmath3 @xcite .
also , the bruhat order on @xmath11 itself appears as a special case of @xmath14 ; see example [ ex : cox ] .
this article is chiefly devoted to @xmath14 .
the questions that we strive to answer emerge from a context which we now briefly describe .
a _ twisted involution _ in @xmath11 is an element which is sent to its inverse by @xmath1 .
denote by @xmath15 the set of twisted involutions
. clearly , @xmath16 let @xmath17 denote the subposet of the bruhat order on @xmath11 induced by @xmath18 .
it is a fact that @xmath19 is graded with rank function given by the coxeter length @xmath20 .
furthermore , a fundamental result due to bjrner and wachs @xcite asserts that ( the order complex of ) every ( open ) interval in @xmath19 is a pl sphere .
recent results on @xmath21 produce a similar picture .
it is graded with rank function @xmath22 , where @xmath23 is the _ twisted absolute length _ @xcite . moreover ,
every interval in @xmath21 is a pl sphere @xcite . in the spirit of the above description , we pose the following problems : * for which @xmath0 and @xmath1 is @xmath14 a graded poset ?
* describe the topology of the intervals of @xmath14 .
we do not know the complete solution to either of the problems .
our main results on @xmath14 are these partial answers : * in theorem [ th : graded ] it is shown that if @xmath5 never is of odd order for @xmath4 unless @xmath24 is a fixed point of @xmath1 , then @xmath14 is graded with rank function @xmath7 . for example
, this is always the case if @xmath11 is of odd rank and its coxeter graph is a tree . by way of contrast , there exist @xmath0 and @xmath1 such that @xmath14 is not graded ; see example [ ex : ungraded ] .
* under the same conditions on @xmath5 as above , every interval in @xmath14 is either @xmath8-acyclic ( i.e. has trivial reduced integral homology ) or a pl sphere .
the latter case occurs precisely when the interval coincides with an interval in @xmath21 .
this is theorem [ th : topology ] .
dropping these conditions , the homotopy type of an interval can be computed in certain special cases ( theorems [ th : collapsible ] and [ th : recursive ] ) .
in addition to the above results , we also investigate what we call the _ poincar series _ of @xmath3 .
this is the rank generating function of @xmath14 whenever it is graded .
specifically , we provide a simple necessary condition for an intriguing factorisation phenomenon to occur and demonstrate that this condition also is sufficient in the context of finite coxeter groups .
the general case is left open .
the structure of the remainder of the paper is as follows . in the next section
we gather preliminaries on poset topology and coxeter groups , including some new material on twisted involutions , that we need in the sequel . in section [ se : motivation ] we then give an account of various natural contexts where @xmath3 and @xmath14 appear . after that
, we turn to the study of @xmath14 in section [ se : bruhat ] .
section [ se : poincare ] is concerned with poincar series considerations .
finally , we mention several open problems and give further comments in section [ se : problems ] .
say that a poset is _ locally finite _ if every interval is finite .
a locally finite poset equipped with a minimum element is _ graded _ if in any given interval all maximal chains have the same cardinality . with any finite poset @xmath25 , we associate the _ order complex _ @xmath26 .
it is the ( abstract ) simplicial complex whose simplices are the chains in @xmath25 .
whenever we make topological statements about @xmath25 , we have the corresponding properties of @xmath26 in mind .
we make no notational distinction between an abstract simplicial complex and its geometric realisation . given a regular cw complex @xmath27 , we define the _ face poset _
@xmath28 to be the set of cells in @xmath27 ordered by inclusion . by convention
, we include the empty cell as the minimum element in @xmath28 .
a simplicial complex is a _ pl ( piecewise linear ) ball _ if it has a common subdivision with a simplex .
similarly , it is a _ pl sphere _ if it has a common subdivision with the boundary of a simplex . in particular ,
a pl sphere is of course homeomorphic to a sphere ( and similarly for balls ) .
we now review elements of forman s discrete morse theory @xcite .
its formulation in terms of matchings , to which we adhere , is due to chari @xcite . for unexplained terminology from combinatorial topology as well as
further background , we refer to @xcite .
let @xmath27 be a finite regular cw complex .
a _ matching _ on the face poset @xmath28 is an involution @xmath29 , for some @xmath30 such that for all @xmath31 , either @xmath32 or @xmath33 , where @xmath34 is the covering relation in @xmath28 . in other words , @xmath35 is nothing but a graph - theoretic matching of the hasse diagram of @xmath28 .
the unmatched elements , i.e. the members of @xmath36 , are called the _
critical cells_. the matching @xmath35 is _ acyclic _
if whenever we have a sequence in @xmath37 of the form @xmath38 with @xmath39 , it holds that @xmath40 . a nice way to interpret
this condition is as follows .
if , in the hasse diagram of @xmath26 , we direct the matching edges upwards and the others downwards , then this directed graph is acyclic iff @xmath35 is an acyclic matching . without loss of generality , we will always assume that the empty cell is not critical , i.e. that @xmath37 includes the minimum element to include it .
there is always a matching partner for the empty cell available , because no acyclic matching can match all @xmath41-dimensional cells with @xmath6-dimensional ones . ] .
our acyclic matching @xmath35 determines a way to collapse @xmath27 onto a ( possibly non - regular ) _ critical complex _
@xmath42 consisting of the critical cells together with the vertex which was matched with the empty cell . in the process , incidences among the cells may change . in some situations , however , this is not a problem .
[ th : forman ] if @xmath35 is an acyclic matching on @xmath28 which is complete ( i.e.has no critical cells ) , then @xmath27 is collapsible , in particular contractible .
theorem [ th : forman ] has the following consequence , which happens to be well - suited for some of our applications .
[ co : filter ] suppose @xmath35 is an acyclic matching on @xmath28 with set of critical cells @xmath43 . if @xmath43 is an order filter ( i.e. @xmath44 , @xmath45 ) , then @xmath27 is homotopy equivalent to the complex @xmath42 obtained from @xmath27 by collapsing the complex of non - critical cells to a point .
a standard result in topology asserts that for a cw complex @xmath46 with contractible subcomplex @xmath47 , the quotient map @xmath48 is a homotopy equivalence . since @xmath43 is an order filter
, @xmath49 is a subcomplex of @xmath27 .
this subcomplex is contractible by theorem [ th : forman ] .
the reader is assumed to be familiar with the basics of coxeter group theory . here ,
bits and pieces are reviewed in order to agree on notation .
we refer to the textbooks @xcite and @xcite for further details . henceforth , let @xmath0 be a coxeter system with @xmath50 .
the coxeter length function is @xmath51 . if @xmath52 and @xmath53 , the word @xmath54 is called a _ reduced expression _ for @xmath55 . here , and in what follows , symbols of the form @xmath56 are always elements in @xmath57 .
we make no notational distinction between a word in the free monoid over @xmath57 and the element in @xmath11 which it represents ; we rely on the context to clarify our intentions .
given coxeter generators @xmath58 , we let @xmath59 denote the ( possibly infinite ) order of @xmath60 .
this information is gathered in the _ coxeter graph _
whith vertex set @xmath57 and an edge labelled @xmath59 connecting @xmath24 and @xmath61 if and only if they do not commute ( i.e. if and only if @xmath62 ) .
as is customary , omitted edge labels are understood to equal @xmath63 .
the set of _ reflections _ in @xmath11 is @xmath64 .
the _ absolute length _
@xmath65 is the minimal @xmath66 such that @xmath67 for some @xmath68 .
given @xmath10 , we define its _ ( right ) descent set _ by @xmath69 for @xmath70 , let @xmath71 be the standard parabolic subgroup of @xmath11 generated by @xmath72 . if @xmath71 is finite , it has a longest element which is denoted by @xmath73 . in @xmath71 , it is characterised by the fact that @xmath74 .
following tradition , we write @xmath75 .
we now define the bruhat order . among several equivalent definitions ,
the one which follows is probably best suited for our purposes .
a disadvantage is that it is not obvious that what it defines is indeed a partial order .
[ de : bruhat ] the _ bruhat order _ on @xmath11 is the partial order defined by @xmath76 if and only if for every reduced expression @xmath77 for @xmath78 there exist @xmath79 such that @xmath80 . by an automorphism of @xmath0 we mean a group automorphism of @xmath11 which preserves @xmath57 .
such an automorphism is determined by the corresponding automorphism of the coxeter graph .
let @xmath1 be an involutive automorphism of @xmath0 .
recall from the introduction that the set of _ twisted involutions _ in @xmath11 is @xmath81 observe that @xmath82 is the set of ordinary involutions .
the combinatorics of @xmath15 can be described in terms of `` reduced expressions '' in a way which is remarkably similar to that of @xmath11 itself .
we now proceed to review the parts of this theory that we will use in the sequel .
most of this appeared in @xcite , but some results are new .
define a set of symbols @xmath83 .
there is an action of the free monoid @xmath84 on the set @xmath11 defined by @xmath85 and @xmath86 .
abusing notation , we will write @xmath87 for @xmath88 , where @xmath89 is the identity . the relevance of all this is the following : the orbit of @xmath13 under the @xmath84-action is @xmath15 , i.e. the twisted involutions are the elements of the form @xmath87 . for @xmath90 , the _ rank _
@xmath91 is the smallest @xmath66 such that @xmath92 for some @xmath93 .
then , the expression @xmath94 is called a _
reduced @xmath95-expression _ for @xmath55 .
it follows from ( * ? ? ?
* theorem 4.8 ) that @xmath21 , the bruhat order on twisted involutions , is graded with rank function @xmath7 .
more precisely , it was shown that the rank function is @xmath96 , where @xmath23 is the _ twisted absolute length _ , but it follows from the proof that this function coincides with @xmath7 .
the twisted absolute length was defined differently in @xcite , but here is a description which is more convenient for our purposes : [ pr : elltwist ] suppose @xmath87 is a reduced @xmath95-expression for @xmath97 . then , @xmath98 is the number of indices @xmath99 = \{1 , \dots , k\}$ ] that satisfy @xmath100 = @xmath101 .
let @xmath102 be the asserted number ( a priori , it depends on the choice of reduced @xmath95-expression ) .
one can check that @xmath103 for all @xmath97 and @xmath4 ( see ( * ? ? ?
* lemma 3.4 ) ) .
furthermore , by ( * ? ? ? * lemma 3.8 ) , @xmath104 .
the construction of the @xmath84-action therefore implies @xmath105 thus , @xmath106 .
as the terminology suggests , @xmath107 ; see @xcite .
either by a simple induction argument based on proposition [ pr : elltwist ] or as an immediate consequence of ( * ? ? ?
* definition 4.5 ) , we conclude that the twisted identities are the twisted involutions of vanishing twisted absolute length : @xmath108 a useful consequence is that @xmath109 whenever @xmath110 and @xmath111 .
twisted involutions have reduced @xmath95-expressions of a convenient form , as shown by the next lemma which is due to springer @xcite .
[ le : springer ] any @xmath112 with @xmath113 has a reduced @xmath95-expression of the form @xmath114 with the property that for some @xmath115 , @xmath116 is a reduced @xmath95-expression for the longest element @xmath73 in a @xmath1-invariant standard parabolic subgroup @xmath71 , @xmath70 , with @xmath117 .
the following fundamental lemma will be put to use repeatedly throughout the paper .
it is completely analogous to the corresponding property in @xmath11 ( due to deodhar ( * ? ? ?
* theorem 1.1 ) ) and explains why @xmath21 and @xmath19 are so similarly behaved
. the first two parts are ( * ? ? ?
* lemma 3.9 ) .
together , they imply the third .
let @xmath4 and @xmath118 with @xmath119 .
suppose @xmath111 . then , 1 .
3 . @xmath122 .
next , we show that @xmath21 , like @xmath19 , can be described in terms of subwords .
the result extends richardson and springer s ( * ? ? ?
* corollary 8.10 ) which later was shown to coincide with @xmath21 in @xcite . ] from finite to general coxeter groups ( and adjusts it to our definition of the @xmath84-action which differs somewhat from the one given in @xcite ) .
[ th : subword ] let @xmath123 be given . then , @xmath124 in the bruhat order if and only if for every reduced @xmath95-expression @xmath87 for @xmath78 there exist @xmath125 such that @xmath126 .
we begin with the `` only if '' direction , so suppose @xmath127 and choose a reduced @xmath95-expression @xmath87 for @xmath78 .
if @xmath128 , then the lifting property shows that @xmath129 . by induction on the rank of @xmath78 , we are done . otherwise , if @xmath130 , we have @xmath131
. it follows , again by induction , that @xmath132 for some @xmath133 .
acting by @xmath134 yields the desired result .
now consider the `` if '' part of the assertion .
assume @xmath87 is a reduced @xmath95-expression for @xmath78 and let @xmath135 for some @xmath79 . given @xmath136 $ ] , let @xmath137 and @xmath138 , where @xmath139 ( with @xmath140 , @xmath141 ) .
assuming by induction that @xmath142 , we either have @xmath143 or ( by the lifting property ) @xmath144 .
thus , @xmath145 .
henceforth , let @xmath0 be a finitely generated coxeter system with an involutive automorphism @xmath1 . recall that @xmath1 is determined by an automorphism of the coxeter graph .
it should be clear by now that the goal of this paper is to study certain properties of @xmath3 . in this section
, we motivate this task by indicating some situations in which @xmath3 naturally shows up .
_ consider a connected reductive linear algebraic group @xmath146 ( over @xmath147 , say ) .
let @xmath148 be a borel subgroup and @xmath149 a maximal torus . given an involutive automorphism @xmath150 which preserves @xmath151 and @xmath148 , let @xmath152 be the fixed point subgroup . by means of left translations , @xmath148 acts on the symmetric variety @xmath153 .
this gives rise to a finite number of orbits that may be ordered by containment of their zariski closures .
richardson and springer @xcite studied this poset @xmath154 by defining an order preserving map @xmath155 . here ,
the underlying coxeter group @xmath11 is the weyl group @xmath156 ( where @xmath157 is the normaliser of @xmath151 in @xmath146 ) and @xmath1 is induced by @xmath158 .
_ in general , @xmath159 is neither injective nor surjective .
however , @xmath3 is always contained in the image .
moreover , @xmath159 produces an isomorphism @xmath160 in certain interesting cases . for instance , with @xmath161 , we may define @xmath158 so that @xmath162 as in ( * ? ? ?
* example 10.4 ) .
the corresponding poset @xmath154 , governing the orbit decomposition of @xmath163 , is then isomorphic to @xmath14 , where @xmath11 is the symmetric group @xmath164 and @xmath1 is given by conjugation with the longest element in @xmath11 ( i.e. the reverse permutation @xmath165 ) .
figure [ fi : a5 ] displays this poset for @xmath166 . [ ex : cox ] _ suppose @xmath1 is the involution @xmath167 given by @xmath168 , where @xmath11 is any coxeter group . observe that @xmath169 .
hence , we have a bijection @xmath170 in this case .
furthermore , it is clear from definition [ de : bruhat ] that this bijection gives a poset isomorphism @xmath171 .
therefore , ( bruhat orders on ) twisted identities generalise ( bruhat orders on ) coxeter groups .
_ [ ex : quotient ] _ let @xmath172 denote the fixed point subgroup of @xmath1 .
it is known ( @xcite ) that @xmath172 itself is a coxeter group with the following canonical set of coxeter generators : @xmath173 _ observe that @xmath174 . in other words , there is a bijection between @xmath3 and the set of cosets @xmath175 .
thus , @xmath3 can be thought of as a quotient of coxeter groups .
[ ex : conjugation ] _ if @xmath11 is finite , it contains a unique longest element @xmath176 .
then , @xmath177 defines an involutive automorphism of @xmath0 .
in this situation , @xmath178 for some @xmath10 .
this means that @xmath179 if and only if @xmath180 and @xmath176 are conjugate .
now , ( left or right ) multiplication by @xmath176 is an antiautomorphism of @xmath19 .
the poset @xmath14 is therefore isomorphic to the dual of the subposet of @xmath19 induced by the conjugacy class of the longest element . _ in the special case when @xmath11 is the symmetric group @xmath164 , the conjugacy class of @xmath176 consists of all fixed point free involutions of @xmath181 $ ] .
equivalently , if we think of a permutation in terms of its disjoint cycle decomposition , the conjugacy class of @xmath176 corresponds to the set of complete matchings on @xmath182 elements .
@xmath14 in this case gives ( the dual of ) a bruhat order on matchings .
we mention in this context that the literature already contains a `` bruhat order '' on matchings defined by deodhar and srinivasan @xcite .
their poset is , however , strictly weaker than the dual of @xmath14 , although both posets are graded with the same rank function ; see remark [ rem : poin1 ] .
we now turn to the core of the paper , namely the study of @xmath14 . to begin with , we observe that it sometimes coincides with the more familiar @xmath21 .
[ pr : coincide ] if there exists no @xmath4 such that @xmath183 and @xmath24 are conjugate , then @xmath184 .
notice that @xmath185 if and only if @xmath186 .
thus , if @xmath24 and @xmath183 are not conjugate , we have @xmath187 for all @xmath55 . now , the assertion is a consequence of proposition [ pr : elltwist ] and the remarks that follow it .
the hypothesis of proposition [ pr : coincide ] means that for all @xmath4 , every path from @xmath24 to @xmath183 in the coxeter graph of @xmath11 contains an edge with an even label ; see ( * ? ? ?
* exercise 1.16 ) .
for example , proposition [ pr : coincide ] applies when @xmath188 with @xmath1 being the unique non - trivial automorphism .
we will find that @xmath3 is particularly well - behaved if @xmath1 never flips an edge with an odd label in the coxeter graph of @xmath11 .
say that @xmath1 has the _ no odd flip ( nof ) property _ if @xmath189 is even or infinite for all @xmath190 . for example
, if @xmath11 is finite and irreducible , @xmath1 has the nof property unless @xmath11 is of type @xmath191 or @xmath192 , @xmath193 , and @xmath1 is the unique non - trivial automorphism .
we now proceed to prove a series of lemmata that put restrictions on the behaviour of elements of twisted absolute length @xmath6 when @xmath1 has the nof property .
informally put , these restrictions ensure that intervals in @xmath14 never can be `` sparse enough '' not to inherit the gradedness from @xmath21 . [
le : length1 ] suppose @xmath11 is irreducible and finite . if @xmath194 , then @xmath11 is either the rank one group @xmath195 or the dihedral group @xmath196 for some odd integer @xmath197 . if @xmath198 , the twisted absolute length coincides with the absolute length .
hence , @xmath176 is a reflection .
it follows , for example by the classification of finite coxeter groups , that @xmath199 or @xmath200 , @xmath197 odd .
now , assume @xmath201 .
obviously , @xmath98 and @xmath202 always have the same parity .
in particular , @xmath203 is odd . by inspection of coxeter graphs , the only groups with @xmath203 odd that admit a nontrivial @xmath1 are @xmath196 , @xmath197 odd , and @xmath204 , @xmath205 .
these groups have @xmath206 as unique nontrivial automorphism ; see ( * ? ? ?
* exercise 4.10 ) .
therefore , @xmath207 is a bijection which sends @xmath21 to the dual of @xmath208 . in particular
, the top element @xmath176 has the same rank in @xmath21 as in @xmath208 .
this implies @xmath209 , i.e. that @xmath176 is again a reflection .
hence , @xmath11 is the dihedral group @xmath196 for some odd @xmath197 .
[ le : length2 ] let @xmath11 be finite and assume @xmath1 has the nof property . if @xmath194 , then @xmath176 has a reduced @xmath95-expression beginning with @xmath210 for some @xmath1-fixed @xmath4 .
the twisted absolute length clearly is additive over direct products of @xmath1-invariant coxeter systems .
thus , @xmath11 decomposes as @xmath211 where @xmath212 , @xmath213 , @xmath71 is irreducible and @xmath70 is invariant under @xmath1 .
choose a reduced @xmath95-expression @xmath87 for @xmath214 .
lemma [ le : length1 ] shows that either @xmath215 or @xmath216 , @xmath197 odd . in the former case ,
@xmath217 for some @xmath1-fixed @xmath4 .
this implies that @xmath176 has the reduced @xmath95-expression @xmath218 . in the latter situation , @xmath73 has the reduced @xmath95-expression @xmath219 ( @xmath220 letters ) , where @xmath221 .
the hypothesis on @xmath189 means that @xmath24 and @xmath61 are @xmath1-fixed .
hence , @xmath222 is a reduced expression for @xmath176 of the desired form .
[ le : cover ] suppose @xmath1 has the nof property .
then , given @xmath223 with @xmath224 , @xmath78 covers at most one twisted identity in @xmath21 . by lemma [ le : springer ] , @xmath224 implies @xmath225 with @xmath226 for some @xmath1-invariant @xmath70 satisfying @xmath212 . invoking lemma [ le : length2 ] , we may assume @xmath227 .
in other words , @xmath78 has a reduced @xmath95-expression of the form @xmath228 , where @xmath24 is fixed by @xmath1 .
each reduced @xmath66-letter subword of @xmath228 except at most one begins with @xmath210 , hence represents an element of non - vanishing twisted absolute length .
the exception is @xmath87 ( if this expression is reduced ) which , by theorem [ th : subword ] , is the only candidate for a twisted identity covered by @xmath78 .
next , we state our main result on gradedness . as a notable special case , @xmath14 is graded whenever the coxeter graph of @xmath11 is a tree containing a @xmath1-fixed node
. the nature of the rank function should not be surprising . indeed ,
if @xmath229 with @xmath230 , then @xmath231 is an unrefinable chain in @xmath14 .
hence , the rank function necessarily is @xmath7 whenever @xmath14 is graded .
[ th : graded ] if @xmath1 has the nof property , then @xmath14 is graded with rank function @xmath7 .
since we know that @xmath21 is graded with rank function @xmath7 , it is enough to show that every interval @xmath232\subseteq { \mathrm{br}}({\iota(\theta)})$ ] contains a chain of length @xmath233 . in order to get a contradiction ,
assume the theorem is false .
choose a minimal interval @xmath232 \subseteq { \mathrm{br}}({\iota(\theta)})$ ] which does not contain a chain of length @xmath233 and such that @xmath234 is minimal among all such intervals .
pick @xmath235 .
we must have @xmath236 ; otherwise @xmath232 $ ] would not be minimal . by the lifting property , @xmath237 .
minimality of @xmath234 shows that @xmath238 $ ] contains some @xmath239 which covers @xmath240 in @xmath21 .
using the lifting property once again , we conclude that @xmath241 covers @xmath242 in @xmath21 .
minimality of @xmath232 $ ] implies @xmath243 .
thus , @xmath244 and @xmath245 .
now , @xmath241 covers two distinct twisted identities , namely @xmath246 and @xmath242 .
this contradicts lemma [ le : cover ] , and the proof is complete .
unfortunately , theorem [ th : graded ] does not give the full picture .
for example , one readily checks that with @xmath11 being any dihedral group and @xmath1 the non - trivial automorphism , @xmath14 is graded , although this is not predicted by theorem [ th : graded ] .
at this point one may very well suspect that @xmath14 is always graded .
this is not true , as shown by the next example .
[ ex : ungraded ] _ suppose @xmath247 , the affine weyl group corresponding to @xmath248 .
its coxeter generating set is @xmath249 , and @xmath250 .
define the involution @xmath1 by @xmath251 , @xmath252 and @xmath253 .
_ consider the twisted identities @xmath254 and @xmath255 . as an interval in @xmath21 , the open interval
@xmath256 contains two elements , namely @xmath257 and @xmath258 , neither being a twisted identity .
thus , @xmath259 and @xmath260 are unrefinable chains in @xmath14 .
hence , it is not graded .
our attention is now turned to the topology of intervals in @xmath14 .
the first results are valid without any restriction on @xmath11 or @xmath1 .
[ th : collapsible ] consider an interval @xmath232 \subseteq { \mathrm{br}}({\iota(\theta)})$ ] . if @xmath261 for some @xmath235 , then @xmath256 is collapsible .
let @xmath262 .
our strategy is to invoke theorem [ th : forman ] after defining a complete acyclic matching @xmath35 on the face poset @xmath263 .
suppose @xmath264 is a chain in @xmath265 , and define @xmath266 .
now , let @xmath267 be given by @xmath268 observe that @xmath269 because @xmath270 .
the lifting property ensures that @xmath271 is a chain in @xmath265 . since @xmath272
, we have @xmath273 and therefore @xmath274 .
hence , @xmath35 is a complete matching . to prove acyclicity of @xmath35 ,
suppose we have four chains @xmath275 , where @xmath34 is the covering relation in @xmath25 and @xmath276 . by construction , @xmath277 and @xmath278 for some @xmath279 .
in fact , @xmath280 ; otherwise we would have @xmath281 , implying @xmath282 .
we therefore conclude that @xmath283 .
in particular , @xmath284 has strictly fewer elements with @xmath24 as a descent than @xmath264 does .
this shows the impossibility of a sequence @xmath285 with @xmath286 . in other words
, @xmath35 is acyclic .
the strategy employed in the preceding proof is not quite applicable if @xmath287 , @xmath288 , because chains that contain @xmath240 would not have a well - defined matching partner . leaving these chains unmatched
, however , allows us to conclude a result which is useful in recursive arguments .
[ th : recursive ] consider an interval @xmath232 \subseteq { \mathrm{br}}({\iota(\theta)})$ ] . if @xmath289 for some @xmath290 , then @xmath256 is homotopy equivalent to the suspension of @xmath291 .
just as in the proof of theorem [ th : collapsible ] , we define a matching on @xmath263 , where @xmath292 .
this time , there will be critical cells .
define @xmath293 now , arguing exactly as in the proof of theorem [ th : collapsible ] we have a complete acyclic matching @xmath294 .
the set of critical cells is the order filter @xmath295 which consists of the cells in the open star of @xmath240 .
corollary [ co : filter ] therefore implies that @xmath296 is homotopy equivalent to the star of @xmath240 with the entire link of @xmath240 collapsed to a single point .
the complex obtained in this way is homotopic to the suspension of the link of @xmath240 , and this link is @xmath297 .
using theorem [ th : recursive ] , we can always compute the homotopy type of an interval in terms of that of another interval .
the problem is that if @xmath298 , we are only able to express @xmath256 in terms of larger intervals .
this makes it difficult to set up an inductive argument since we may not reach a reasonable base case .
to rectify the situation , we pay the price of restricting @xmath11 and @xmath1 .
let us call an interval @xmath232\subseteq { \mathrm{br}}({\iota(\theta)})$ ] _ full _ if @xmath299 .
[ le : full ] suppose @xmath1 has the nof property . if an interval @xmath232\subseteq { \mathrm{br}}({\iota(\theta)})$ ] is full and @xmath300 , then either @xmath261 or @xmath301 $ ] is full ( implying , in particular , @xmath302 )
assume , in order to deduce a contradiction , that @xmath287 and @xmath301 $ ] is not full .
choose @xmath303 $ ] minimal with the property that @xmath304 ; it exists since every element in @xmath301\setminus [ u , v]$ ] is of the form @xmath305 , @xmath303 $ ] .
we have @xmath306 and @xmath307 .
since @xmath308 and intervals in @xmath21 are pl spheres , @xmath309 must cover some element @xmath310 with @xmath311 . by the lifting property , @xmath312 . since @xmath232 $ ] is full , @xmath313 .
minimality of @xmath55 now implies @xmath314 .
hence , @xmath309 covers more than one twisted identity , contradicting lemma [ le : cover ] .
[ le : matching ] suppose @xmath1 satisfies the nof property .
let @xmath315 for @xmath316 and suppose @xmath317 for some @xmath318 . if @xmath319 , then @xmath320 .
we employ induction on @xmath321 .
the assertion is vacuously true if @xmath322 , because in this case the hypotheses imply that @xmath323 covers both @xmath240 and @xmath78 , contradicting lemma [ le : cover ] .
now suppose @xmath324 .
since @xmath14 is graded , we can choose @xmath325 such that @xmath55 covers @xmath242 and @xmath326 . if @xmath327 , we are done .
otherwise , the lifting property implies that @xmath305 covers @xmath240 . as in the @xmath328 case above
, @xmath304 is impossible , so @xmath187 . by the induction assumption , @xmath329 .
hence , @xmath320 .
[ th : topology ] assume @xmath1 has the nof property .
consider an interval @xmath232\subseteq { \mathrm{br}}({\iota(\theta)})$ ] .
if it is full , then @xmath256 is a pl sphere of dimension @xmath330 . otherwise , @xmath256 is @xmath8-acyclic . if @xmath232 $ ] is full , the assertion follows immediately from the corresponding result on @xmath21 , namely (
* corollary 4.6 ) .
suppose @xmath232 $ ] is not full , and let @xmath292 .
proceeding by induction on @xmath234 , we choose @xmath235 .
there are two cases depending on whether or not @xmath24 also is a descent of @xmath242 .
* case i , @xmath288 .
* if @xmath261 , theorem [ th : collapsible ] shows that @xmath265 is collapsible , in particular contractible and @xmath8-acyclic . by lemma
[ le : full ] , this is always the case if @xmath301 $ ] is full .
therefore , we may assume that @xmath301 $ ] is not full and @xmath287 . by the induction assumption
, @xmath331 is @xmath8-acyclic .
let @xmath263 . given a chain @xmath332 ,
define @xmath333 ; this set is nonempty since it contains @xmath242 .
let @xmath334 and define @xmath335 by @xmath336 lemma [ le : matching ] together with the lifting property proves that @xmath337 . observing that @xmath273
, we conclude that @xmath35 is a matching on @xmath25 with set of critical cells @xmath295 .
an argument completely analogous to the acyclicity part of the proof of theorem [ th : collapsible ] shows that @xmath35 is an acyclic matching .
exactly as in the proof of theorem [ th : recursive ] , this implies that @xmath265 is homotopy equivalent to the suspension of the @xmath8-acyclic complex @xmath331 . since a complex and its suspension
have isomorphic reduced homology groups up to an index shift , @xmath265 is @xmath8-acyclic .
* case ii , @xmath338 . * by lemma [ le : full ]
, @xmath238 $ ] is not full , and therefore @xmath339 is @xmath8-acyclic .
combining case i with theorem [ th : recursive ] , we have the following homotopy equivalences : @xmath340 where @xmath341 denotes suspension .
hence , the suspension of @xmath265 is @xmath8-acyclic .
therefore , @xmath265 is @xmath8-acyclic , too .
from a combinatorial point of view , the _ poincar series _ of @xmath11 is simply the length generating function : @xmath342 analogously , we may define the poincar series of @xmath3 to be @xmath343 in particular , @xmath344 is the rank generating function for @xmath14 whenever it is graded .
since @xmath3 is , in some sense , a quotient of coxeter groups ( cf.example [ ex : quotient ] ) , one may naively hope that its poincar series is a quotient of the series of the corresponding groups ; let us say that @xmath344 _ factors through @xmath1 _ if @xmath345 .
with @xmath346 and @xmath1 given by conjugation with the longest element , for example , one obtains @xmath347 , whereas @xmath348 .
thus , in this case , @xmath344 does not factor through @xmath1 .
intriguingly , though , factorisation does occur in several situations , as we shall see below .
first , we give a necessary condition for the poincar series of @xmath3 to factor through @xmath1 .
the condition is that @xmath1 ( seen as a coxeter graph automorphism ) is not allowed to flip edges unless they are labelled @xmath349 .
[ pr : factor ] if @xmath345 , then @xmath350 for all @xmath4 .
suppose equality holds , and consider the coefficient of the linear term on both sides .
on the left hand side , this coefficient is @xmath351 .
the elements of length @xmath6 in @xmath172 ( i.e. the coxeter generators ) are the longest elements in the finite parabolic subgroups of the form @xmath352 . in @xmath3 ,
the elements of rank @xmath6 are the twisted identities of the form @xmath210 , @xmath4 .
now , @xmath210 is a twisted identity if and only if @xmath353 .
moreover , for @xmath354 , @xmath355 if and only if @xmath356 and @xmath357 . as a consequence ,
the linear coefficient of the right hand side is @xmath358 this is equal to @xmath351 if and only if @xmath350 for all @xmath4 .
it turns out that in the finite setting , proposition [ pr : factor ] actually gives a characterisation of the @xmath359 for which the factorisation phenomenon occurs .
( the label @xmath349 , of course , can not occur in this case ) .
[ th : poincare ] suppose @xmath11 is finite
. then , @xmath344 factors through @xmath1 if and only if @xmath360 for all @xmath4 , i.e. if and only if @xmath24 commutes with @xmath183 for all @xmath361 .
suppose @xmath70 is minimal such that @xmath362 and @xmath363 whenever @xmath364 , @xmath365 .
then , either @xmath71 is irreducible or @xmath71 is isomorphic to the direct product of two isomorphic irreducible coxeter groups that are interchanged by @xmath1 as in example [ ex : cox ] . to prove the theorem it therefore suffices to check that it holds in the setting of example [ ex : cox ] and for the finite irreducible groups that satisfy the hypotheses . to begin with ,
return to the setup in example [ ex : cox ] .
observe that @xmath366 .
furthermore , the rank of @xmath367 is @xmath202 , so that @xmath368 . since @xmath369
, we have @xmath370 .
it remains to consider the finite irreducible groups .
if @xmath371 , the assertion is trivial . if not , the groups that satisfy the criteria are @xmath372 , @xmath373 and @xmath374 . * type a. * consider the symmetric group case @xmath375 with @xmath1 given by conjugation with the longest element @xmath176 ( the reverse permutation ) .
let us verify that @xmath344 factors through @xmath1 .
it is known ( * ? ? ?
* supplementary problem 1.24 ) that the fixed point free involutions @xmath376 in @xmath164 have the following generating function : @xmath377 where @xmath378 denotes the number of inversions in @xmath379 , which is the same as the coxeter length of @xmath379 . since @xmath380 ( example [ ex : conjugation ] ) , @xmath381 for all @xmath382 and @xmath383 for all @xmath110 , we obtain @xmath384 it is well - known , and straightforward to check , that @xmath385 , the hyperoctahedral group of rank @xmath386 .
now , @xmath387 and @xmath388 thus , @xmath389 in this situation .
* type d. * let @xmath390 with the coxeter generators being labelled as described in the coxeter graph below .
( 150,60)(0,0 ) ( 10,50 ) ( -1,50)@xmath391 ( 10,10 ) ( 0,4)@xmath392 ( 30,30 ) ( 30,34)@xmath393 ( 60,30 ) ( 56,34)@xmath394 ( 130,30 ) ( 126,34)@xmath395 ( 10,50)(1,-1)20 ( 10,10)(1,1)20 ( 30,30)(1,0)30 ( 60,30)(1,0)10 ( 130,30)(-1,0)10 ( 90,29)@xmath396 define @xmath1 by @xmath397 , @xmath398 and @xmath399 for all @xmath400 .
it is not hard to realise that @xmath3 consists precisely of the elements of the form @xmath401 for @xmath402 $ ] ( if @xmath403 , we interpret this as the identity element @xmath13 ) . in particular ,
@xmath404 . noting that @xmath405 ( whose poincar series was presented in the type @xmath46 case above ) and @xmath406 we again conclude that @xmath344 factors through @xmath1 . * type e. * when @xmath1 is the unique non - trivial automorphism of the coxeter graph of @xmath374 , the hasse diagram of @xmath14 is displayed in figure [ fi : e6 ] . inspecting it ,
one obtains the rank generating function and verifies that @xmath407 factors through @xmath1 .
here , @xmath408 .
[ rem : poin1 ] _ return to the type a case @xmath375 with @xmath1 given by conjugation with @xmath176 .
then , @xmath344 coincides with the rank generating function for deodhar and srinivasan s `` bruhat order '' on @xmath376 ( * ? ? ? * theorem 1.3(i ) ) which was mentioned in example [ ex : conjugation ] .
in fact , although formulated differently in @xcite , the definition of their rank function @xmath409 is readily seen to yield @xmath410 so that @xmath411 where the last equality is valid since @xmath344 is a symmetric polynomial in this case . _
we have already commented on the fact that , although theorem [ th : graded ] does not predict it in all cases , @xmath14 is graded whenever @xmath11 is dihedral . in fact , the only instances of non - graded @xmath14 that we know of follow from example [ ex : ungraded ] and simple variations thereof is increased .
similarly , @xmath412 can be increased to any odd number @xmath413 with simple adjustments . ] .
ultimately , of course , we would like a complete characterisation of the @xmath0 and @xmath1 for which @xmath14 is graded .
a step in this direction would be to resolve the following two problems .
consider the setup of theorem [ th : topology ] . by a result of whitehead @xcite , a collapsible pl manifold is a pl ball ( see e.g. ( * ? ? ?
* corollary 3.28 ) ) . by induction on the rank
, we may assume that the proper subintervals of @xmath256 are pl spheres or balls , hence that @xmath256 is a pl manifold .
thus , if we were somehow able to replace `` @xmath8-acyclic '' by `` collapsible '' in the conclusion of theorem [ th : topology ] , we would have a proof of the next conjecture .
[ co : pl ] suppose @xmath1 has the nof property .
consider an interval @xmath232 \subseteq br({\iota(\theta)})$ ] .
if it is full , then @xmath256 is a pl sphere of dimension @xmath330 . otherwise , @xmath256 is a pl ball of the same dimension .
the conditions on @xmath0 and @xmath1 which guarantee that @xmath14 is graded in theorem [ th : graded ] are exactly those for which theorem [ th : topology ] asserts that the intervals are pl spheres or @xmath8-acyclic ( namely , @xmath1 should satisfy the nof property ) .
this immediately leads to the next question .
the mbius function of an interval coincides with the reduced euler characteristic of its order complex .
thus , a negative answer to question [ qu : homotopy ] would imply a negative answer to the following : whenever @xmath11 is finite , @xmath21 and @xmath19 contain a unique top element @xmath176 .
this element is , however , usually not a twisted identity .
indeed @xmath14 does not always possess a unique maximal element .
[ pr : maximal ] if @xmath11 is irreducible and finite , then @xmath14 has more than one maximal element if and only if @xmath11 is of type @xmath418 or @xmath419 , for some positive integer @xmath197 , and @xmath1 is the unique non - trivial automorphism . if @xmath198 , @xmath13 is the only twisted identity .
consulting the classification of finite irreducible coxeter groups , we find that the groups that admit a nontrivial @xmath1 are @xmath420 , @xmath373 , @xmath374 , @xmath421 and @xmath422 .
furthermore , this @xmath1 is unique in all cases ( except @xmath423 , but the various choices are then equivalent ) .
the dihedral groups @xmath422 are easy to handle .
types @xmath424 and @xmath425 are covered by theorem [ th : poincare ] whereas the type @xmath426 assertion follows from proposition [ pr : coincide ] .
it remains to understand type @xmath420 . here
, one may study the bruhat order on the conjugacy class of @xmath176 as described in example [ ex : conjugation ] .
it is straightforward to check that the minimal elements in this class are the possible products of @xmath427 mutually commuting coxeter generators . if @xmath386 is odd , there is only one such product ; otherwise there are @xmath428 of them .
it is known that @xmath19 always is directed . a proof using the lifting property of @xmath19 is given in @xcite . employing instead the lifting property of @xmath21
, it follows in exactly the same way that @xmath21 is directed , too .
thus , in addition to the cases provided by proposition [ pr : maximal ] , we may immediately conclude that @xmath14 is directed whenever the hypothesis of proposition [ pr : coincide ] is satisfied .
[ pr : directed ] assume @xmath11 is infinite .
suppose there is a partition @xmath429 with @xmath430 and @xmath431 such that the elements of @xmath432 commute pairwise for @xmath433 .
then , @xmath14 is directed .
recall that a _ coxeter element _ is a product of the coxeter generators ( in any order ) .
speyer @xcite has shown that @xmath434 for all @xmath435 and an arbitrary coxeter element @xmath264 whenever @xmath11 is infinite .
therefore , given @xmath436 , a large enough power of any coxeter element is an upper bound for @xmath78 and @xmath55 in @xmath19 .
observe that @xmath437 is a coxeter element in @xmath11 .
moreover , @xmath438 .
thus , @xmath14 is directed . in section [
se : poincare ] , in order to answer the finite case version of question [ qu : poincare ] affirmatively , we resorted to a case by case computation which did not shed much light on the situation . a combinatorial proof , type independent if possible , would be much preferred .
specifically , what we are looking for is this : given that @xmath344 factors through @xmath1 , construct a bijection @xmath440 such that @xmath441 , where @xmath442 denotes the length of @xmath443 in terms of the canonical coxeter generators of @xmath172 ( cf .
example [ ex : quotient ] ) .
recalling from example [ ex : quotient ] that we may identify @xmath3 with the set of cosets @xmath444 , there is of course a natural bijection @xmath445 defined by sending @xmath446 to @xmath447 , but in general it does not have the desired properties . | given a coxeter system @xmath0 equipped with an involutive automorphism @xmath1 , the set of _ twisted identities _
is @xmath2 we point out how @xmath3 shows up in several contexts and prove that if there is no @xmath4 such that @xmath5 is of odd order greater than @xmath6 , then the bruhat order on @xmath3 is a graded poset with rank function @xmath7 given by halving the coxeter length . under the same condition , it is shown that the order complexes of the open intervals either are pl spheres or @xmath8-acyclic . in the general case
, contractibility is shown for certain classes of intervals .
furthermore , we demonstrate that sometimes these posets are not graded . for the poincar series of @xmath3 , i.e. its generating function with respect to @xmath7 , a factorisation phenomenon is discussed . |
type ii supernovae ( sne ) emerge from shock - driven explosion of massive ( @xmath12 ) stars initiated by core collapse @xcite .
in particular , plateau type ii sne ( sne iip ) result from core collapse of supergiants that have massive hydrogen - rich envelope .
the light curves of sne iip are characterized by a constant luminosity plateau lasting about 80 - 120 days after explosion @xcite .
recently , the progenitors of a few sne iip have been directly identified on pre - explosion images @xcite , revealing that these are @xmath13 supergiants , close to the low - mass theoretical limit of the core collapse process .
sn 2004dj in ngc 2403 is an outstanding sn discovered by itagaki @xcite on july 31 , 2004 ( jd 2453218 ) , because it is one of the brightest and closest supernova observed ever .
based on its first optical spectrum @xcite sn 2004dj was quickly classified as a `` normal '' type iip event . unfortunately , it was discovered during the plateau phase ( @xmath14 month after explosion ) , thus , the maximum of the light curve could not be covered . due to its proximity ( @xmath15 mpc ) @xcite , sn 2004dj was also detected in radio @xcite , infrared @xcite and x - ray bands @xcite .
@xmath16-band optical photometry and high - resolution @xmath17 spectroscopy obtained during the plateau phase was published by @xcite .
@xcite presented @xmath18 light curve and full optical spectra in the nebular phase .
they estimated the amount of @xmath19ni as @xmath20 , and pointed out an asymmetric , bipolar distribution of the ejected nickel from modeling the nebular @xmath17 profile .
the optical light curve observed in the batc ( intermediate - band ) system was also studied by @xcite .
they concluded that sn 2004dj was a typical sn iip that ejected a @xmath21 envelope and synthesized @xmath20 of radioactive nickel .
particular interest was paid on the association of sn 2004dj with the compact cluster sandage 96 which is thought to be the host of the progenitor star @xcite .
@xcite and @xcite analysed the spectral energy distribution ( sed ) of sandage 96 ( based on archival ccd photometry ) , and concluded that s96 is indeed a compact cluster with age of @xmath22 myr and total stellar mass of @xmath23 ( the uncertainty is due to the different reddening determination ) . from population synthesis ,
the initial mass of the progenitor star of sn 2004dj was estimated as @xmath24 , which is in the upper part of the mass range of other sne iip progenitors ( see above ) . in this paper we present time - resolved optical photometry and spectroscopy of sn 2004dj covering its first year after discovery .
our primary goal is to derive a reasonable distance to this sn and its host galaxy ngc 2403 by using all available information .
the new observational data are presented in section 2 and 3 . in section 4
we apply a variant of the expanding photosphere method ( epm ) and the sne iip standard candle method ( scm ) for distance determination , and critically compare the results with other distance estimates of the host galaxy .
based on the improved distance , we estimate some of the physical properties of sn 2004dj and s96 in section 5 .
finally , section 6 summarizes our results .
[ cols="<,^,^,^,^,^,^,^,^",options="header " , ] the resulting parameters corresponding to the minimum of @xmath7 are listed in table 13 and plotted in fig . 19 .
because the metallicity of ngc 2403 at the position of s96 is probably subsolar ( @xmath25o / h@xmath26 = -0.24 $ ] , @xcite ) , we have used the models with @xmath27 and @xmath28 metallicities from both grids . comparing these results with the published ones ( shown in the last three rows of table 13 ) , it is visible that our analysis resulted in the same kinds of solutions as the previous studies : a `` young '' population with @xmath8 myr and an `` old '' population with @xmath29 myr .
the `` old '' solutions are very similar to the ones given in the previous studies .
the `` young '' solution , however , has considerably less age .
it is interesting that this solution comes out using both model grids , regardless of metallicity ( but with different reddenings and cluster masses , of course ) .
based on the @xmath7 values , the `` young '' solution from the bc03 grid gives the best description of the observed data .
however , we agree with @xcite that if the observed sed is undersampled , such as the one presented here , it may be misleading to accept or reject solutions based on @xmath7 values only .
thus , we conclude that , at present , there are three probable solutions for the sed fitting of s96 : the old one corresponds to @xmath30 myr , the young one with @xmath31 myr , and a solution with age between these two , @xmath32 myr .
each solutions have their own combinations of cluster reddening and total mass , between the values of @xmath33 and @xmath34 . due to the correlations between these parameters
, it is not possible to assign a unique solution for all these quantities simultaneously . regarding the potential progenitor of sn 2004dj ,
it is seen in table 13 that the sn mass ( determined from the padova isochrones , @xcite ) ranges from @xmath35 to @xmath36 , depending on age .
the uncertainty of the age determined from sed fitting is at least 10 % @xcite , implying an error of the sn - mass of about @xmath37 for the young- and @xmath38 for the old solution .
it is interesting that the young ( 8 myr ) solution results in significantly higher mass than all the sne iip progenitor masses detected so far .
the highest mass of a detected sn type ii progenitor is that of sn 1999ev @xcite with @xmath39 .
this is similar to the progenitor mass of sn 2004et , being @xmath40 @xcite and sn 1999gi with @xmath41 @xcite .
contrary to these , the progenitors of sn 2003gd @xcite and sn 2005cs @xcite , also detected on pre - explosion images , are found within the mass range of @xmath42 .
these two recent discoveries have led to a conclusion that perhaps all type iip sne emerge from relatively less massive progenitors , so that @xmath43 .
this is supported by the non - detection of the progenitors of sn 1999em , sn 1999gi and sn 2001du @xcite giving @xmath44 .
however , the possible masses of the progenitor of sn 2004dj determined above do not fully support this scenario . if the @xmath45 myr solution of s96 is indeed valid , then the progenitor mass of sn 2004dj must have been higher than @xmath46 .
this is significantly higher than the masses found for the other progenitors , even for sn 1999ev ( see above ) .
since the other two solutions for s96 give progenitor masses that are consistent with the mass range of the others , it can not be stated that sn 2004dj had definitely a more massive progenitor , but it can not be ruled out at present . in principle , the cluster masses in table 13 imply that there might be at least a few stars with @xmath11 when sn 2004dj exploded . from the integration of the salpeter imf , the number of stars with @xmath11 is @xmath47 , which gives @xmath48 for @xmath49 .
the actual number of massive stars in s96 may , of course , differ from this value , but statistically it is possible that at least a few such stars are indeed there .
@xcite estimated that s96 contains about 12 red and perhaps 2 - 3 blue supergiants , based on their sed parameters .
the spectrophotometric observations of the remainder of s96 after the sn faded away would help to clarify the picture .
comparing the observations of the co vibrational bands in the mid - ir with theoretical explosion models combined with radiative transfer calculations , @xcite presented evidence for a red supergiant with @xmath50 as the likely progenitor of sn 2004dj .
again , this is at the upper end of the detected progenitor masses for type ii sne , close to the prediction of the @xmath51 myr cluster solution by @xcite .
@xcite estimated the parameters of the pre - supernova star using the formulae given by @xcite .
these formulae relate the length of the plateau @xmath52 , the middle - plateau absolute magnitude @xmath53 and expansion velocity at the photosphere @xmath54 to the explosion energy ( @xmath55 ) , ejected mass ( @xmath56 ) and progenitor radius ( @xmath16 ) .
@xcite derived @xmath57 days , @xmath58 mag and @xmath59 kms@xmath60 , and they got @xmath61 erg , @xmath62 and @xmath63 . from the more detailed analysis presented in this paper , we estimate these parameters as @xmath64 days ( sect .
2.1.1 , allowing @xmath65 days for the peak duration before the plateau , see @xcite ) , @xmath66 mag and @xmath67 kms@xmath60 ( sect . 4.2 and 4.3 ) .
these parameters result in @xmath68 erg , @xmath69 and @xmath70 .
it is interesting that the ejected mass is , again , close to @xmath71 , although with high uncertainty .
it is concluded that the presently available information suggest that the progenitor mass of sn 2004dj is @xmath72 .
the sed fitting of s96 may suggest a cluster age as low as 8 myr , indicating @xmath73 .
this massive progenitor may also be suspected from theoretical sn models .
the conclusions of this paper are summarized as follows : * new @xmath0 photometric observations of sn 2004dj are presented .
the date of explosion is estimated as june 30 , 2004 ( jd @xmath74 ) , about 1 month before discovery .
the plateau phase lasted about @xmath75 days after explosion .
the most probable value of the reddening toward sn 2004dj is @xmath76 mag @xcite .
the progenitor cluster s96 is also detected on pre - explosion frames . *
the new , plateau - phase optical spectra reveal great similarity between sn 2004dj and sn 1999em , a typical sn iip .
radial velocities from metallic lines and @xmath77 are presented .
the spectral features are successfully modeled with the spectrum synthesis code _ synow _ with @xmath78 k as the photospheric temperature .
the observed p cyg profile of @xmath77 can be better described by an increased emission of the photosphere relative to the blackbody intensity at the rest wavelength of @xmath77 .
a single nebular spectrum confirms the @xmath77 asymmetry reported by @xcite . *
the distance to sn 2004dj is estimated from two methods applicable for sne iip .
the epm distance is @xmath79 mpc , while the scm gives @xmath80 mpc .
these are in good agreement with other distance estimates to the host galaxy ngc 2403 , based on cepheids and tf - relation , although the cepheid - distance is slightly less than the sn - distances .
the averaged distance is @xmath81 mpc , which is consistent with all available information . *
using the updated distance , the nickel mass of sn 2004dj is calculated as @xmath82 from the shape of the tail light curve .
the progenitor mass has been estimated from fitting theoretical seds to the observations of s96 .
the solutions are similar to those of @xcite and @xcite , except that a `` young '' solution corresponding to @xmath8 myr is also possible .
such a young cluster age would mean that the mass of the progenitor was @xmath11 .
the sn - masses of the other solutions are less than @xmath83 , being in better agreement with other sne iip progenitors .
this research has been supported by hungarian otka grants no.ts 049872 and t042509 .
the authors are grateful to the directors and staff of konkoly observatory , szeged observatory ( hungary ) , david dunlap observatory ( canada ) and f. l. whipple observatory ( usa ) for generously allocating telescope time .
they also want to express their warm thanks to prof .
david branch and his group members at university of oklahoma for providing access and support to the synow code .
thanks are also due to an anonymous referee for the very thorough report and numerous suggestions that helped us to improve the paper . the nasa astrophysics data system , the simbad and ned databases , the canadian astronomy data centre and the supernova spectrum archive ( suspect ) were used to access data and references .
the availability of these services are gratefully acknowledged . | new @xmath0 photometry and optical spectroscopy of the type iip supernova 2004dj in ngc 2403 , obtained during the first year since discovery , are presented .
the progenitor cluster , sandage 96 , is also detected on pre - explosion frames .
the light curve indicates that the explosion occured about 30 days before discovery , and the plateau phase lasted about @xmath1 days after that .
the plateau - phase spectra have been modelled with the @xmath2 spectral synthesis code using h , na i , ti ii , sc ii , fe ii and ba ii lines . the sn distance is inferred from the expanding photosphere method ( epm ) and the standard candle method ( scm ) applicable for sne iip .
they resulted in distances that are consistent with each other as well as earlier cepheid- and tully - fisher distances .
the average distance , @xmath3 mpc is proposed for sn 2004dj and ngc 2403 .
the nickel mass produced by the explosion is estimated as @xmath4 .
the sed of the progenitor cluster is reanalysed by fitting population synthesis models to our observed @xmath0 data supplemented by @xmath5 and @xmath6 magnitudes from the literature .
the @xmath7-minimization revealed a possible `` young '' solution with cluster age @xmath8 myr , and an `` old '' solution with @xmath9 - @xmath10 myr .
the `` young '' solution would imply a progenitor mass @xmath11 , which is higher than the previously detected progenitor masses for type ii sne .
[ firstpage ] stars : evolution supernovae : individual ( sn 2004dj ) galaxies : individual ( ngc 2403 ) |
at the beginning of the twentieth century , the atomistic theory of matter was not yet fully accepted by the scientific community .
while searching for phenomena that would prove , beyond any doubt , the existence of atoms , einstein realized that `` _ ... according to the molecular - kinetic theory of heat , bodies of microscopically - visible size suspended in a liquid will perform movements of such magnitude that they can be easily observed in a microscope ... _ '' , as he wrote in his celebrated paper in 1905 @xcite . in this work ,
devoted to explain the irregular motion of browinan particles on theoretical grounds , einstein argued that the motion of these small bodies has a diffusive character .
moreover , he discovered an important relation involving the diffusion coefficient @xmath1 , the fluid viscosity @xmath2 , the particles radius @xmath3 ( having assumed spherical particles ) , avogadro s number @xmath4 , the temperature @xmath5 and the gas constant @xmath6 : @xmath7 this relation can be employed , and actually had been , to determine experimentally the avogadro s number @xcite .
indeed , the diffusion coefficient can be measured by monitoring the growth , with time @xmath8 , of the particle displacement @xmath9 , which is expected to behave as @xmath10 einstein relation ( [ diff ] ) , that may be seen as the first example of the fluctuation - dissipation theorem @xcite , allowed for the determination of avogadro s number and gave one of the ultimate evidences of the existence of atoms .
einstein s theoretical explanation of bm is based on the intuition that the irregular motion of a brownian particle is a consequence of the huge number of collisions per unit time with the surrounding fluid molecules . since einstein s approach , diffusion and irregular phenomena were commonly associated to the presence of many degrees of freedom .
the effects of the disregarded degrees of freedom on an observed small part of a system can be either studied directly ( as initiated by smoluchowski @xcite ) or modeled by means of stochastic dynamics ( as proposed by langevin @xcite ) . from the latter point of view , bm provided the first and main stimulus to the building of the modern theory of stochastic processes .
after the ( re)discovery of deterministic chaos @xcite , it was clear that also fully deterministic and low dimensional systems can give rise to erratic seemingly random motions , practically indistinguishable from those produced by a stochastic process .
this implied an affective unpredictability of chaotic systems and the need for a probabilistic description also of a strictly deterministic world .
the success in understanding the basic mechanisms for the onset of chaos , and the wealth of interesting phenomena occurring in low dimensional systems hinted at optimistic expectations about the possibility of a systematic deterministic approach to irregular natural phenomena .
this rised a rapid development of time series analysis with the idea to demonstrate the deterministic character of many irregular phenomena .
nowadays we are aware of the limits of this optimistic programme @xcite , and we know that a definite answer on the deterministic or stochastic character of experimental signals is impossible .
however , some tools of time series analysis , such as the entropy analysis at varying the scale of resolution , are very useful to characterize important features of complex systems . among the recent developments in this context
, we can mention the experiment by gaspard et al .
@xcite on the motion of a brownian particle .
the debate @xcite around the possible theoretical interpretation of the experiment is a clear indication of how , one century after the seminal einstein s work , bm continues to be a subject of intricate and fascinating discussions . beyond its undoubted importance for applications in many natural phenomena
, deterministic chaos also enforces us to reconsider some basic problems standing at the foundations of statistical mechanics such as , for instance , the applicability of a statistical description to low dimensional systems .
in addition , the combined effects of noise and deterministic evolution can generate highly nontrivial and rather intriguing behaviours .
as an example , we just mention the stochastic resonance @xcite and the role of colored noise in dynamical systems.@xcite the aim of this paper is a discussion on the viable approaches to characterize and understand the dynamical ( microscopic ) character of bm ( sect .
ii ) . in particular , we shall focus on the distinction between chaos and noise from a data analysis and on conceptual aspects of the modeling problem ( sect .
moreover , we shall investigate and discuss about the basic microscopic ingredients necessary for bm as , for instance , the possibility of genuine bm in non - chaotic deterministic systems ( sect .
iv ) . finally , we conclude ( sect .
v ) with a discussion on the role of chaos in statistical mechanics .
einstein s work on bm is based on statistical mechanics and thermodynamical considerations applied to suspended particles , with the assumption of velocity decorrelation ( molecular chaos ) . instead , one of the first attempts to develop a purely dynamical theory of bm dates back to langevin @xcite that , as himself writes , gave `` _ ...
a demonstration _ [ of einstein results ] _ that is infinitely more simple by means of a method that is entirely different . _ '' langevin considers the newton equation for a small spherical particle in a fluid , taking into account that the stokes viscous force it experiences is only a mean force . in one direction , say e.g. the @xmath11-direction , one has : @xmath12 where @xmath13 is the mass of the particle . in the r.h.s .
the first term is the stokes viscous force .
@xmath14 is a fluctuating random force which models the effects of the huge number of impacts with the surrounding fluid molecules , responsible for the thermal agitation of the particle . in statistical mechanics terms
, this corresponds to molecular chaos . with the assumption that the force @xmath14 is a gaussian , time uncorrelated random variable , the probability distribution functions ( pdf ) for the position and velocity of the brownian particle
can be exactly derived @xcite . in particular
, the pdf of the position , at long times , reduces to the gaussian distribution in agreement with einstein s result .
langevin s work along with that of ornstein and uhlenbeck @xcite are at the foundation of the theory of stochastic differential equations .
the stochastic approach is however unsatisfactory being at the level of a phenomenological description .
the next theoretical challenge toward the building of a dynamical theory of brownian motion is to understand its microscopic origin from first principles .
a very early attempt was made in 1906 by smoluchowski , who tried to derive the large scale diffusion of brownian particles starting from the microscopic description of their collisions with the fluid molecules @xcite .
a renewed interest on the subject appeared some years later , when it was realized that even purely deterministic systems composed of a large number of particles produce macroscopic diffusion , at least on finite time scales .
these models had an important impact in the justification of brownian motion theory and , more in general , in deriving a consistent microscopic theory of irreversibility .
some of these works considered chains of harmonic oscillators of equal masses @xcite , while others @xcite analyzed the motion of a heavy impurity linearly coupled to a chain of equal mass oscillators . for an infinite number of oscillators
, the momentum of the heavy particle behaves as a genuine stochastic process described by the langevin equation ( [ langevin ] ) .
when their number is finite , diffusion remains an effective phenomenon lasting for a ( long but ) limited time .
soon after the discovery of dynamical chaos , it was realized that also simple low dimensional deterministic systems may exhibit a diffusive behavior . in this framework , the two - dimensional lorentz gas @xcite , describing the motion of a free particle through a lattice of hard round obstacles , provided the most valuable example .
particle trajectories can be ballistic ( with very few collisions in the case of infinite horizon ) or chaotic as a consequence of the convexity of the obstacles . in the latter case , at large times ,
the mean square displacement from the particle initial condition grows linearly with time .
lorentz system is closely related to the sinai billiard @xcite , which can be obtained from the lorentz gas by folding the trajectories into the unitary lattice cell .
the extensive study on billiards has shown that chaotic behavior might usually be associated to diffusion in simple low dimensional models , supporting the idea that chaos was at the very origin of diffusion . however , more recently ( see e.g. ref . )
it has been shown that even non - chaotic deterministic systems , such as a bouncing particle in a two - dimensional billiard with polygonal but randomly distributed obstacles ( wind - tree ehrenfest model ) , may exhibit a diffusion - like behavior ( see sect .
iv ) . deterministic diffusion is a generic phenomenon present also in simple chaotic maps on the line . among the many contributions we mention the work by fujisaka , grossmann @xcite and geisel @xcite .
a typical example is the @xmath15 discrete - time dynamical system : @xmath16 + f(x(t)-[x(t ) ] ) \
, , \label{eq : chaos}\ ] ] where @xmath17 ( the position of a point - like particle ) performs diffusion in the real axis .
the bracket @xmath18 $ ] denotes the integer part of the argument .
@xmath19 is a map defined on the interval @xmath20 $ ] that fulfills the following properties : + i ) the map , @xmath21 ( mod @xmath22 ) is chaotic . + ii ) @xmath19 must be larger than @xmath22 and smaller than @xmath23 for some values of @xmath24 , so there exists a non vanishing probability to escape from each unit cell ( a unit cell of real axis is every interval @xmath25 $ ] , with @xmath26 ) .
+ iii ) @xmath27 , where @xmath28 and @xmath29 define the map in @xmath30 and @xmath31 $ ] respectively .
this anti - symmetry condition with respect to @xmath32 is introduced to avoid a net drift .
a very simple and much studied example of @xmath14 is @xmath33 \end{array}\right .
\label{eq : chaos2}\end{aligned}\ ] ] where @xmath34 is the control parameter .
it is useful to remind the link between diffusion and velocity correlation , i.e. the taylor - kubo formula , that helps in understanding how diffusion can be realized in different ways . defining @xmath35 as the velocity correlation function , where @xmath36 is the velocity of the particle at time @xmath8 .
it is easy to see that for continuous time systems ( e.g. eq.([langevin ] ) ) @xmath37 standard diffusion , with @xmath38 , is always obtained whenever the hypotheses for the validity of the central limit theorem are verified : + i ) the variance of the velocity must be finite : @xmath39 \ii ) the decay to zero of the velocity correlation function @xmath40 at large times should be faster than @xmath41 . in discrete - time systems , the velocity @xmath36 and the integration of @xmath42
are replaced by the finite difference @xmath43 and by the quantity @xmath44 respectively .
condition i ) is justified by the fact that having an infinite variance for the velocity is rather unphysical .
it should be noted that this requirement is independent of the microscopic dynamics under consideration : langevin , deterministic chaotic or regular dynamics .
condition ii ) , corresponding to the request of molecular chaos , is surely verified for the langevin dynamics where the presence of the stochastic force entails a rapid decay of @xmath40 . in deterministic regular systems , such as the many oscillator model , the velocity decorrelation comes from the huge number of degrees of freedom that act as a heat bath on a single oscillator . while in the ( non - chaotic ) ehrenfest wind - tree model decorrelation originates from the disorder in the obstacle positions , the situation is more subtle for deterministic chaotic systems . in fact , even if nonlinear instabilities generically lead to a memory loss and , henceforth , to the validity of the molecular - chaos hypothesis , slow decay of correlation , e.g. @xmath45 with @xmath46 , may appear in very intermittent systems @xcite .
when this happens , condition ii ) is violated , and superdiffusion , @xmath47 , is observed .
though interesting , superdiffusion is a quite rare phenomenon .
moreover , usually , small changes of the control parameters of the dynamics restore standard diffusion .
therefore , also for chaotic systems we can state that the `` rule '' is the standard diffusion and the `` exception '' is the superdiffusion @xcite .
we end this section by asking whether is it possible to determine , by the analysis of a brownian particle , if the microscopic dynamics underlying the observed macroscopic diffusion is stochastic , deterministic chaotic or regular ?
inferring the microscopic deterministic character of brownian motion on an experimental basis would be attractive from a fundamental viewpoint .
moreover it could provide further evidence to some recent theoretical and numerical studies @xcite . before discussing a recent experiment@xcite in this direction , we must open the `` pandora box '' of the longstanding and controversial problem of distinguishing chaos from noise in signal analysis @xcite .
the first observation is that , very often in the analysis of experimental time series , there is not a unique model of the `` system '' that produced the data .
moreover , even the knowledge of the `` true '' model might not be an adequate answer about the character of the signal . from this point of view
, bm is a paradigmatic example : in fact it can be modeled by a stochastic as well as by a deterministic chaotic or regular process . in principle
a definite answer exist .
if we were able to determine the maximum lyapunov exponent ( @xmath48 ) or the kolmogorov - sinai entropy ( @xmath49 ) of a data sequence , we would know without uncertainty whether the sequence was generated by a deterministic law ( @xmath50 ) or by a stochastic one ( @xmath51 ) .
nevertheless , there are unavoidable practical limitations in computing such quantities . those are indeed defined as infinite time averages taken in the limit of arbitrary fine resolution .
but , in experiments , we have access only to a finite , and often very limited , range of scales and times .
however , there are measurable quantities that are appropriate for extracting information on the signal character .
in particular , we shall consider the @xmath52-entropy per unit time @xcite @xmath53 that generalizes the kolmogorov - sinai entropy . in a nutshell , while for evaluating @xmath49 one has to detect the properties of a system with infinite resolution , for @xmath54 a finite scale ( resolution ) @xmath0 is requested .
the kolmogorov - sinai entropy is recovered in the limit @xmath55 , i.e. @xmath56 .
this means that if we had access to arbitrarily small scales , we could answer the original question about the character of the law that generated the recorded signal .
even if this limit is unattainable , still the behavior of @xmath53 provides a very useful scale - dependent description of the nature of a signal .
the @xmath0-entropy was originally introduced in the context of information theory by shannon @xcite and , later , by kolmogorov @xcite in the theory of stochastic processes .
an operative definition is as follows .
one considers a continuous variable @xmath57 , that represents the state of a @xmath58-dimensional system , and one introduces the vector @xmath59 which lives in @xmath60 and is a portion of the trajectory discretized in time with step @xmath61 .
then the phase space @xmath62 is partitioned using hyper - cubic cells of side @xmath0 .
the vector @xmath63 is coded into the word , of length @xmath13 , @xmath64 where @xmath65 labels the cell in @xmath62 containing @xmath66 . for bounded motions ,
the number of visited cells ( i.e. the alphabet ) is finite . under the hypothesis of stationarity ,
the probabilities @xmath67 of the admissible words @xmath68 are obtained from the time evolution of @xmath63 .
the @xmath52-entropy per unit time , @xmath69 is then defined by @xcite : @xmath70 where @xmath71 is the @xmath13-block entropy : @xmath72 and @xmath73 . + it is worth remarking a few important points .
a rigorous mathematical procedure@xcite would require to take the infimum over all possible partitions with elements of size smaller than @xmath0 .
the kolmogorov - sinai entropy is obtained in the limit of small @xmath0 @xmath74 note that in deterministic systems @xmath75 and henceforth @xmath49 do not depend on the sampling time so that ( [ eq:2 - 5 ] ) can be in principle used with any choice of @xmath61 @xcite .
however , in practical computations , where the rigorous definition is not applicable , the specific value of @xmath61 is important and @xmath75 may also depend on the used norm .
for very small @xmath0 , no matter of the norm , the correct value of the kolmogorov - sinai entropy is usually recovered .
indeed , when the partition is very fine , usually , it well approximates a generating partition .
it is worth reminding that the kolmogorov - sinai entropy is a dynamical invariant , i.e. independent of the used state representation ( [ eq:2 - 1 ] ) . in deterministic systems
the following chain of inequalities holds @xcite @xmath76 where the summation is over all positive lyapunov exponents .
a system is chaotic when @xmath77 and regular when @xmath78 .
typically , one observes that @xmath53 attains a plateau @xmath49 , below a resolution threshold , @xmath79 , associated to the smallest characteristic length scale of the system .
instead , for @xmath80 due to ( [ eq : ineq ] ) @xmath81 , in this range the details of the @xmath0-dependence may be informative on the large scale ( slow ) dynamics of the system ( see e.g. refs . ) . in stochastic signals
@xmath82 , but for any @xmath83 , @xmath53 is finite and a well defined function of @xmath0 and @xmath61 .
the nature of the dependence of @xmath53 on @xmath0 and @xmath61 provide a characterization of the underlying stochastic process ( see refs . ) .
for an important and wide class of stochastic processes @xcite an explicit expression for @xmath54 can be given in the limit @xmath84 .
this is the case of stationary gaussian processes characterized by a power spectrum @xmath85 , with @xmath86 , for which @xcite : @xmath87 the case @xmath88 , corresponding to the power spectrum of a brownian signal , would give @xmath89 .
some stochastic processes , such as _ e.g. _ time uncorrelated and bounded ones , are characterized by a logarithmic divergence below a critical @xmath79 , which may depend on @xmath61 . in experiments , usually , only a scalar variable @xmath90 can be measured and moreover the dimensionality of the phase space is not known . in these cases it is reconstructed by delay embedding technique @xcite , where the vector @xmath63 is build as @xmath91 , now in @xmath92 .
this is a special instance of ( [ eq:2 - 1 ] ) .
-entropy for the map ( [ eq : mappa ] ) with @xmath93 computed by the standard techniques @xmath94 $ ] at @xmath95 ( @xmath96 ) , @xmath97 ( @xmath98 ) and @xmath99 ( @xmath100 ) and different block length ( @xmath101 ) .
the boxes give the entropy computed with @xmath95 by using periodic boundary condition over @xmath102 cells .
the straight lines correspond to the two asymptotic behaviors , @xmath103 and @xmath104 . ] then to determine the entropies @xmath105 , very efficient numerical methods are available @xcite ( the reader may find an exhaustive review in refs . ) . here , avoiding technicalities , we just mention some subtle points which should be taken into account in data analysis .
+ first , if the information dimension of the attractor for a given system is @xmath106 then , to have a meaningful measure of the entropy , the embedding dimension @xmath13 has to be larger than @xmath106 .
second , as above mentioned , the plateau @xmath107 appears only below a critical @xmath79 , meaning that it is possible to distinguish a deterministic signal from a random one only for @xmath108 .
however , one should be aware of the fact that the finiteness of the data set imposes a lower cut - off scale @xmath109 below which no information can be extracted from the data ( see ref . ) .
also in stochastic signals , there exists a lower critical cut - off @xmath109 due to the finiteness of the data set , and often , as mentioned at the end of the previous subsection , one has logarithmic divergences below @xmath79 .
since , this is also what happens in general for @xmath110 , the interpretation of the results requires much attention @xcite .
+ therefore , if @xmath13 is not large enough and/or @xmath0 is not small enough ( or in the lack of a good estimation of the important range of scales for the different behaviors ) one may obtain misleading results .
+ another important problem concerns the choice of @xmath61 .
if @xmath61 is much larger or much shorter than the characteristic time - scale of the system at the scale @xmath0 , then the correct behavior of the @xmath0-entropy @xcite can not be properly recovered .
+ to exemplify the above difficulties let us consider the map @xmath111 which , for @xmath112 , is chaotic ( similarly to ( [ eq : chaos ] ) ) and displays large scale diffusion . on the basis of the previous discussion
, the @xmath0-entropy is expected to behave as @xmath113 where @xmath48 is the lyapunov exponent and @xmath1 is the diffusion coefficient .
the typical problems encountered in numerically computing @xmath75 can be appreciated in fig .
[ fig:1 ] .
first notice that the threshold @xmath114 . as for the importance of the choice of @xmath61 , note that the diffusive behavior @xmath115 is roughly obtained only by considering the envelope of @xmath116 evaluated at different values of @xmath61 .
the reason for this is as follows .
the natural sampling interval would be @xmath117 , however this choice requires considering larger and larger embedding dimensions @xmath13 at increasing @xmath0 .
indeed , a simple dimensional argument suggests that the characteristic time of the system is determined by its diffusive behaviour @xmath118 .
if we consider for example , @xmath119 and the typical values of the diffusion coefficient @xmath120 , the characteristic time , @xmath121 , is much larger than the elementary sampling time @xmath95 . on the other hand
, the plateau at the value @xmath49 can be recovered only for @xmath122 , even if , in principle , any value of @xmath61 could be used .
+ the above difficulties can be partially overcome by means of a recently introduced method based on exit times @xcite .
the main advantage of this approach is that it is not needed to fix _ a priori _ @xmath61 , because the `` correct '' @xmath61 is automatically selected .
we are now ready to discuss the experiment and its results reported in ref . .
in this experiment , a long time record ( about @xmath123 data points ) of the motion of a small colloidal particle in water was sampled at regular time intervals ( @xmath124 s ) with a remarkable high spatial resolution ( @xmath125 nm ) .
to our knowledge , this is the most accurate measurement of a bm .
the data were then processed by means of standard nonlinear time - series analysis tools , i.e. the cohen - procaccia method @xcite , to compute the @xmath0-entropy .
this computation shows a power - law dependence @xmath115 . actually , similarly to what displayed in fig .
[ fig:1 ] , this behavior is recovered only by considering the envelope of the @xmath53-curves , for different @xmath61 s .
however , unlike to fig . [ fig:1 ] , no saturation @xmath126 is observed in the small @xmath0-region because of the finiteness of data set and resolution as well . from the previous discussion , this can be understood as the fact that @xmath79 of the observed system is much smaller than the smallest detectable scale extracted from data and therefore the @xmath127-entropy can not be properly estimated .
nevertheless , from the chain of inequalities ( [ eq : ineq ] ) and by _ assuming _ from the outset that the system dynamics is deterministic , the authors deduce , from the positivity of @xmath75 , the existence of positive lyapunov exponents in the system .
their conclusion is thus that microscopic chaos is at the origin of the macroscopic diffusive behavior .
+ as pointed out by several works , a few points need to be considered in the data analysis of the aforementioned experiment , namely : the huge amount of involved degrees of freedom ( brownian particle and the fluid molecules ) ; the impossibility to reach high enough ( spatial and temporal ) resolution ; the limited amount of data points .
+ the limitation induced by the finite resolution is particularly relevant to the experiment . even if one assumes that the number of data points and the embedding dimension are large enough , the impossibility to see a saturation to a constant value , @xmath128 , prevents any conclusion about the character of the analyzed signal .
for example , whenever the the analysis of fig .
[ fig:1 ] would be restricted to the region with @xmath129 , then discerning whether the data were originated by a chaotic system or by a stochastic process would be impossible .
in fact in both cases the behavior @xmath89 would have been observed . ) with @xmath130 and random phases uniformly distributed in @xmath131 $ ] .
the numerically computed diffusion constant is @xmath132 .
( b ) time record obtained with a continuous random walk ( [ eq : rw ] ) with the same value of the diffusion constant as in ( a ) .
in both cases data are sampled with @xmath133 , i.e. @xmath134 data points . ]
+ as for the number of degrees of freedom , we recall that , for a correct evaluation of the entropy of the microscopic dynamics , a very high embedding dimension should be used , in practice @xmath135 . for a fluid ( with @xmath136 molecules )
the necessary number of points is of course prohibitive .
moreover , as stressed by grassberger and schreiber @xcite , when the number of degrees of freedom is so high that it can be considered practically infinite there is an additional difficulty related to the definition of entropy and lyapunov exponents , which become norm dependent .
+ furthermore , the limited amount of data severely affects even our ability to recognize if the signal is deterministic but of zero entropy ( i.e. regular ) .
this has been pointed out by dettmann et al .
@xcite , who have shown that the same entropic analysis of ref . , applied to the ehrenfest wind - tree model ( see next section ) reproduced results very similar to those extracted in the brownian experiment .
this model is deterministic and non - chaotic .
in fact if the time records were long enough to see the periodic nature of the signal , and the embedding dimension high enough to resolve the system manifold , the measured entropy would have been zero .
+ computed with the grassberger - procaccia algorithm using using @xmath134 points from the time series of fig .
[ fig:2 ] .
we show the results for embedding dimension @xmath137 .
the straight - lines show the @xmath138 behavior . ]
the following example serves as a clue to better understand these points .
let us consider two signals , the first generated by a continuous random walk : @xmath139 where @xmath2 is a zero mean gaussian variable with @xmath140 , and the second obtained as a superpositions of fourier modes : @xmath141 the coordinate @xmath17 in eq .
( [ eq : mazur ] ) , upon properly choosing the frequencies @xcite and the amplitudes ( e.g. @xmath142 ) , describes the motion of a heavy impurity in a chain of @xmath143 linearly coupled harmonic oscillators .
we know @xcite that @xmath17 performs a genuine bm in the limit @xmath144 .
for @xmath145 the motion is periodic and regular , nevertheless for large but finite times it is impossible to distinguish signals obtained by ( [ eq : rw ] ) and ( [ eq : mazur ] ) ( see fig . [ fig:2 ] ) .
this is even more striking looking at the computed @xmath0-entropy of both signals ( see fig . [ fig:3 ] ) .
+ the results of fig . [ fig:3 ] along with those by dettman et al .
@xcite suggest that , also by assuming the deterministic character of the system , we are in the practical impossibility of discerning chaotic from regular motion . + from the above discussion , one may have reached a very pessimistic view on the possibility to detect the `` true '' nature of a signal by means of data analysis only .
however the scenario is different when the question about the character of a signal remains restricted only to a certain interval of scales . in this case , in fact , it is possible to give an unambiguous classification of the signal character based solely on the entropy analysis and free from any prior knowledge on the system / model that generated the data . indeed
, we can define stochastic / deterministic behavior of a time series on the basis of the absence / presence of a saturation plateau @xmath146 in the observed range of scales .
moreover the behaviour of @xmath53 as a function of @xmath52 provides a very useful `` dynamical '' classification of stochastic processes @xcite .
one has then a practical tool to classify the character of a signal as deterministic or stochastic without referring to a specific model , and is no longer obliged to answer the metaphysical question , whether the system that produced the data was a deterministic or a stochastic @xcite one . +
this is not a mere way to escape the original question .
indeed , it is now clear that the maximum lyapunov exponent and the kolmogorov - sinai entropy are not completely satisfactory for a proper characterization of the many faces of complexity and predictability of nontrivial systems , such as intermittent systems or with many degrees of freedom ( e.g. , turbulence ) @xcite .
+ in the literature the reader may find several methods developed to distinguish chaos from noise .
they are based on the difference in the predictability using prediction algorithms rather than estimating the entropy @xcite or they relate determinism to the smoothness of the signal @xcite .
all these methods have in common the necessity to choose a certain length scale @xmath0 and a particular embedding dimension @xmath13 , therefore they suffer the same limitations of the entropy analysis presented here .
with all the provisos concerning its interpretation , gaspard s and coworkers @xcite experiment had a very positive role not only in stimulating the discussion about the chaos / noise distinction but also in focusing the attention on deep conceptual aspects of diffusion . in this context , from a theoretical point of view , the study of chaotic models exhibiting diffusion and their non - chaotic counterpart contributed to a better understanding of the role of chaos on macroscopic diffusion . in lorentz gases ,
the diffusion coefficient is related , by means of periodic orbits expansion methods @xcite , to chaotic indicators such as the lyapunov exponents .
this suggested , for certain time , that chaos was or might have been the basic ingredient for diffusion .
however , as argued by dettmann and cohen @xcite , even an accurate numerical analysis based on the @xmath0-entropy has no chance to detect differences in the diffusive behavior between a chaotic lorentz gas and its non - chaotic counterpart , such as the wind - tree ehrenfest s model . in the latter model , particles ( wind )
scatter against square obstacles ( trees ) randomly distributed in the plane but with fixed orientation .
since the reflection by the flat edges of the obstacles can not produce exponential separation of trajectories , the maximal lyapunov exponent is zero and the system is not chaotic . in this case
the relation between the diffusion coefficient and the lyapunov exponents is of course nullified .
the result of ref . implies thus that chaos may be not indispensable for having deterministic diffusion .
the question may be now posed on what are the necessary microscopic ingredients to observe deterministic diffusion at large scales . in the wind - tree ehrenfest s model ,
most likely , the disorder in the distribution of the obstacles is crucial . in particular
, one may conjecture that a finite spatial entropy density , @xmath147 , is necessary to the diffusion . so that deterministic diffusion may be a consequence either of a non - zero `` dynamical '' entropy ( @xmath148 ) in chaotic systems or of a non - zero `` static '' entropy ( @xmath149 ) for non - chaotic systems .
this is key - point , because someone can argue that a deterministic infinite system with spatial randomness can be interpreted as an effective stochastic system , but this is probably a `` matter of taste '' .
with the aim of clarifying this point , we consider now a spatially disordered non - chaotic model @xcite , which is the one - dimensional analog of a two - dimensional non - chaotic lorentz system with polygonal obstacles .
it has the advantage that both the case with finite and zero spatial entropy density can be investigated .
let us start with the map defined by eqs .
( [ eq : chaos ] ) and ( [ eq : chaos2 ] ) , and introduce some modifications to make it non - chaotic
. one can proceed as exemplified by fig .
[ fig:4 ] , that is by replacing the function ( [ eq : chaos2 ] ) on each unit cell by its step - wise approximation that is generated as follows . defining the macroscopic slope
is set to @xmath150 .
half domain @xmath151 $ ] is divided into @xmath152 micro - intervals of random size .
the map on @xmath153 $ ] is obtained by applying the antisymmetric transformation with respect to the center of the cell @xmath154 . ] the first - half of @xmath155 is partitioned in @xmath156 micro - intervals @xmath157 , @xmath158 , with @xmath159 . in each interval
the map is defined by its linear approximation @xmath160 where @xmath161 is ( [ eq : chaos2 ] ) evaluated at @xmath162 .
the map in the second half of the unit cell is then determined by the anti - symmetry condition with respect to the middle of the cell .
the quenched random variables @xmath163 are uniformly distributed in the interval @xmath151 $ ] , i.e. the micro - intervals have a _ random _ extension .
further they are chosen independently in each cell @xmath155 ( so one should properly write @xmath164 ) .
all cells are partitioned into the same number @xmath156 of randomly chosen micro - intervals ( of mean size @xmath165 ) .
the modification of the continuous chaotic system is conceptually equivalent to replacing circular by polygonal obstacles in the lorentz system @xcite .
the steps with unitary slope are indeed the analogous of the flat boundaries of the polygon . while the discontinuities in @xmath166 , allowing for a moderate dispersion of trajectories , play a role similar to the vertex of the polygon that splits a narrow beam of particles hitting on it .
since @xmath166 has slope @xmath22 almost everywhere , the map is no longer chaotic , violating the condition i ) ( see sect . [
sec : diff ] ) . for @xmath167 ( i.e. @xmath168 ) the continuous chaotic map ( [ eq : chaos ] ) is recovered .
however , this limit is singular and as soon as the number of intervals is finite , even if extremely large , chaos is absent .
it has been found @xcite that this model still exhibits diffusion in the presence of both quenched disorder and a quasi - periodic external perturbation @xmath16 + f_{\delta}(x(t)-[x(t ) ] ) + \varepsilon \cos(\alpha t)\ , .
\label{eq : nochaos}\ ] ] the strength of the external forcing is controlled by @xmath169 and @xmath170 defines its frequency , while @xmath171 indicates a specific quenched disorder realization .
the diffusion coefficient @xmath1 has been numerically computed from the linear asymptotic behavior of the mean square displacement .
the results , summarized in fig .
[ fig:5 ] , show that @xmath1 is significantly different from zero only for values @xmath172 . for @xmath173 ,
@xmath1 exhibits a saturation close to the value of the chaotic system ( horizontal line ) defined by eqs .
( [ eq : chaos ] ) and ( [ eq : chaos2 ] ) .
the existence of a threshold @xmath174 is not surprising . due to the staircase nature of the system ,
the perturbation has to exceed the typical discontinuity of @xmath166 to activate the `` macroscopic '' instability which is the first step toward the diffusion .
data collapsing , obtained by plotting @xmath1 versus @xmath175 , in fig . [ fig:5 ] confirms this argument .
these findings are robust and do not depend on the details of forcing .
therefore , we have an example of a non - chaotic model in the lyapunov sense by construction , which performs diffusion . on the external forcing strength @xmath169 .
different data relative to a number of cell micro - intervals @xmath176 , @xmath177 and @xmath178 are plotted vs the natural scaling variable @xmath175 to obtain a collapse of the curves .
horizontal line represents the result for chaotic system ( [ eq : chaos],[eq : chaos2 ] ) . ]
now the question concerns the possibility that the diffusive behavior arises from the presence of a quenched randomness with non zero spatial entropy per unit length . to clarify this point , similarly to ref .
, the model can be modified in such a way that the spatial entropy per unit cell is forced to be zero , and see if the diffusion still persists .
this programme can be accomplished by repeating the same disorder configuration every @xmath143 cells ( _ i.e. _ @xmath179 ) , ensuring a zero entropy for unit length .
looking at the diffusion of an ensemble of walkers it was observed that diffusion is still present with @xmath1 very close to the expected value ( as in fig .
[ fig:5 ] ) .
however , a close analysis reveals the presence of weak average drift @xmath180 , that vanishes approximately as @xmath181 for large @xmath143 .
this suggests that , at large times , @xmath182 and the ballistic motion should overcome diffusion .
however , the crossover time @xmath183 , at which the motion switches from diffusive to ballistic , diverges with @xmath143 as @xmath184 , so for very large but finite @xmath143 , the ballistic regime is not observed in simulations .
finally , it should be considered that the value of @xmath180 depends on the realization of the randomness , and after averaging over the disorder the drift becomes zero .
indeed the behavior @xmath185 indicates a self - averaging property for large @xmath143 .
therefore , we can conclude that the system displays genuine diffusion for a very long times even with a vanishing ( spatial ) entropy density , at least for sufficiently large @xmath143 .
these results along with those by dettmann and cohen @xcite allows us to draw some conclusions on the fundamental ingredients for observing deterministic diffusion ( both in chaotic and non - chaotic systems ) . *
an instability mechanism is necessary to ensure particle dispersion at small scales ( here small means inside the cells ) . in chaotic systems
this is realized by the sensitivity to the initial condition . in non - chaotic systems
this may be induced by a finite size instability mechanisms .
also with zero maximal lyapunov exponent one can have a fast increasing of the distance between two trajectories initially close @xcite . in the wind - tree ehrenfest model
this stems from the edges of the obstacles , in `` stepwise '' system ( [ eq : chaos2],[model_1],[eq : nochaos ] ) from the jumps . * a mechanisms able to suppress periodic orbits and therefore to allow for a diffusion at large scale .
it is clear that the first requirement is not very strong while the the second is more subtle . in systems with `` strong chaos '' , all periodic orbits are unstable and , so , it is automatically fulfilled . in non - chaotic systems , such as
the non - chaotic billiards studied by dettmann and cohen and the map ( [ eq : chaos2],[model_1],[eq : nochaos ] ) , the stable periodic orbits seem to be suppressed or , at least , strongly depressed , by the quenched randomness ( also in the limit of zero spatial entropy ) .
we note , that , unlike the two dimensional non - chaotic billiards , in the 1-d system ( [ eq : chaos2],[model_1],[eq : nochaos ] ) , the periodic orbits may survive to the presence of disorder , so we need the aid of a quasiperiodic perturbation to obtain their destruction and the consequent diffusion .
before summarizing the results of this article , we believe that is conceptually important to comment about the relevance of chaos in statistical mechanics approaches .
the statistical mechanics @xcite had been funded by maxwell , boltzmann and gibbs for systems with a very large number of degrees of freedom without any precise requirement on the microscopic dynamics , apart from the assumption of ergodicity .
after the discovery of deterministic chaos it becomes clear that also in systems with few degrees of freedom statistical approaches are necessary .
but , even after many years , the experts do not agree yet on the fundamental ingredients which should ensure the validity of the statistical mechanics .
the spectrum of points of view is very wide , ranging from the landau ( and khinchin @xcite ) belief on the main role of the many degrees of freedom and the ( almost ) complete irrelevance of ergodicity , to the opinion of who , as prigogine and his school @xcite , considers chaos as the basic ingredient .
we strongly recommend the reading of ref . for a detailed discussion of irreversibility .
this work discusses the `` orthodox '' point of view ( based on the role of the large number of degrees of freedom , as stressed by boltzmann @xcite ) re - proposed by lebowitz @xcite and the following debate on the role of deterministic chaos @xcite .
here we focus only on the aspects related to diffusion problems ( and some related aspects , e.g. conduction ) . by means of the powerful method of periodic orbits expansion , in systems with very strong chaos ( namely hyperbolic systems ) , it has been shown that there exists a close relation between transport properties ( e.g. viscosity , thermal and electrical conductivity and diffusion coefficients ) and indicators of chaos ( lyapunov exponents , ks entropy , escape rate ) .
these aspects are , e.g. , discussed in refs . .
at a first glance , the existence of such relations seem to give evidence against the `` anti dynamical '' point of view of landau and khinchin . however , it may be incorrect to employ those results to obtain definite answers valid for generic systems .
in fact it seems to us that there are rather clear evidences that chaos is not a necessary condition for the validity of some statistical behavior @xcite . beyond the problem of diffusion in non - chaotic systems ,
it is worth mentioning the interesting results of lepri et al .
@xcite showing that the gallavotti - cohen formula @xcite , originally proposed for chaotic systems , holds also in some non - chaotic model . moreover , recently li and coworkers @xcite studied the transport properties in quasi - one - dimensional channels with triangular scatters . in such systems ,
the maximal lyapunov exponent is zero because of the flatness of triangle sides .
however , numerical simulations show that , when the scatterers are placed at random ( or their height is random ) , the fourier heat law remains valid .
another interesting non - chaotic model exhibiting the fourier heat conduction is the simple one - dimensional hard - particle system with alternating masses @xcite .
for a recent review on heat conduction in one dimension see ref . ) . these and many other examples prove that the heat conduction is present also in system without microscopic chaos .
this is a further indication that microscopic chaos is not the unique possible source of a macroscopic transport in a given dynamical system .
finally let us briefly summarize the main items of this articles .
the problem of distinguishing chaos from noise can not receive an absolute answer in the framework of time series analysis .
this is due to the finiteness of the observational data set and the impossibility to reach an arbitrary fine resolution and high embedding dimension .
however , this restriction is not necessarily negative , and we can classify the signal behavior , without referring to any specific model , as stochastic or deterministic on a certain range of scales . diffusion may be realized in stochastic and deterministic systems .
in particular , in the latter case ,
chaos is not a prerequisite for observing diffusion and , more in general , nontrivial statistical behaviors .
we gratefully thank d. del - castillo - negrete , o. kantz and e. olbrich who recently collaborated with us on the issues discussed in this paper .
we thank g. lacorata and l. rondoni for very useful remarks on the manuscript .
f. cecconi acknowledges the financial support by firb - miur rbau013lse-001 . | one century after einstein s work , brownian motion still remains both a fundamental open issue and a continous source of inspiration for many areas of natural sciences .
we first present a discussion about stochastic and deterministic approaches proposed in the literature to model the brownian motion and more general diffusive behaviours .
then , we focus on the problems concerning the determination of the microscopic nature of diffusion by means of data analysis . finally , we discuss the general conditions required for the onset of large scale diffusive motion . * brownian motion ( bm ) played a fundamental role in the development of molecular theory of matter , statistical mechanics and stochastic processes .
remarkably , one century after einstein s work , bm is still at the origin of scientific discussions as testified by a recent experiment performed to detect a trace of deterministic chaotic sources on macroscopic diffusion .
several authors , which discussed the results of such an experiment , argued that the possibility to discern experimentally between a deterministic chaotic and noisy dynamics , at the microscopic level , is severely limited by subtle technical and conceptual points .
however , the remarks raised by the scientific community have gone over the criticism and have led to a deeper understanding of the role of chaos in the diffusion . * after a short historical introduction to bm , we focus on the dynamical conditions to observe macroscopic diffusion .
in particular , we discuss the technical and conceptual limits in distinguishing , by means of data analysis , the deterministic or stochastic nature of diffusion .
a main tool for that is the @xmath0-entropy .
part of the discussion is devoted to the problem of macroscopic diffusion in deterministic non - chaotic dynamics . |
the original parking problem of knuth can be stated as follows . consider a parking lot with @xmath0 spaces , identified with the cyclic group @xmath1 .
initially the parking lot is empty , and @xmath2 cars in a queue arrive one by one .
car @xmath3 tries to park on a uniformly distributed space @xmath4 among the @xmath0 possible , independently of other cars , but if the space is already occupied , then it tries places labeled @xmath5 until it finally finds a free spot to park .
as cars arrive , blocks of consecutive occupied spots are forming .
it appears that a _ phase transition _ occurs at the stage where the parking lot is almost full , more precisely when the number of free spots is of order @xmath6 .
indeed , while the largest block of occupied spots is of order @xmath7 with high probability as long as @xmath8 , a block of size approximately @xmath0 is present ( while the others are of order at most @xmath9 ) with high probability when @xmath10 . in the meanwhile
, precisely when @xmath11 is of order @xmath12 with @xmath13 , a clustering phenomenon occurs as @xmath14 decays .
the behavior of this clustering process has been studied precisely by chassaing and louchard @xcite .
it turns out that the process of the relative sizes of occupied blocks is related to the so - called _ standard additive coalescent _
@xcite .
the model originates from a problem in computer science : spaces in the parking lot should be thought of as elementary memory spaces , each of which can be used to store elementary data ( cars ) .
roughly , our aim in this work is to investigate the more general situation where one wants to store larger files , each requiring several elementary memory spaces . in other words ,
single cars are replaced by _
caravans _ of cars , i.e. several cars may arrive simultaneously at the same spot . in this direction , it will be convenient to consider a continuous version of the problem , that goes as follows .
let @xmath15 be a sequence of positive real numbers with sum @xmath16 , and @xmath17 , @xmath18 distinct locations on the unit circle @xmath19 .
imagine that @xmath18 drops of paint with masses @xmath20 , fall successively at locations @xmath21 . each time
a drop of paint falls , we brush it clockwise in such a way that the resulting painted portion of @xmath22 is covered by a unit density of paint .
so at each step the drop of paint is used to cover a new portion of the circle and the total length of the painted part of the circle when @xmath23 drops have fallen is @xmath24 . in this
setting , drops of paint play the role of caravans , and the painted portion of the circle corresponds to occupied spots in the parking lot . more precisely , we consider an increasing sequence @xmath25 of open subsets of @xmath26 , starting from @xmath27 and ending at @xmath28 , which can be thought of as the successive painted portions of the circle . given @xmath29 and the location @xmath30 from where the @xmath31-th drop of paint will be brushed , we paint as many space as possible to the right of @xmath30 with the quantity @xmath32 of paint , without covering the already painted parts , i.e. the blocks of @xmath29 .
alternatively , we break the @xmath31-th caravan into several pieces , so that to fill as much as possible the holes left by @xmath33 after @xmath30 , when reading in clockwise order .
the last car to park arrives at some location @xmath34 , and we let @xmath35 be the union of @xmath29 and the arc between @xmath30 and @xmath34 , see figure [ fig : park ] .
more formal definitions will come in sect .
[ sec : bridge ] . in particular , @xmath29 is a disjoint union of open intervals and @xmath36 .
let @xmath37 be the sequence of the lebesgue measures of the connected components of @xmath38 , ranked by decreasing order .
it will be convenient to view @xmath39 as an infinite sequence , by completing with an infinite number of zero terms . now consider the following random problem .
let @xmath40 be a random variable with finite expectation @xmath41 $ ] .
we say that @xmath42 whenever @xmath43 has a finite second moment @xmath44 $ ] . for @xmath45
, we say that @xmath46 whenever @xmath47 for some @xmath48 .
this implies that @xmath43 is in the domain of attraction of a spectrally positive stable random variable with index @xmath49 , and we stress that our results can be extended under this more general hypothesis ; ( [ tail ] ) is only intended to make things easier .
we suppose from now on that @xmath46 for some @xmath50 $ ] , and take a random iid sample @xmath51 of variables distributed as @xmath43 , and independently of this sequence , iid uniform random variables on @xmath52 , @xmath53 . for @xmath54 , set @xmath55 so by the elementary renewal theorem , @xmath56 . then introduce the sequence @xmath57 defined by @xmath58 so the terms of @xmath59 sum to @xmath60 .
following chassaing and louchard @xcite , we are interested in the formation of macroscopic painted components in the limit when @xmath61 tends to @xmath62 , at times close to @xmath63 , i.e. when the circle is almost entirely painted .
specifically , we let @xmath64 for @xmath65 defined as above with the data @xmath66 .
observe that @xmath67 $ ] decreases when @xmath68 increases , and therefore , in order to investigate the formation of painted components , we should consider the process @xmath69 backwards in time .
this is what we shall do in theorem 1 , using the exponential time change @xmath70 . before describing our main result ,
let us first recall some features of the additive coalescent .
the additive coalescent @xmath71 is a markov process with values in the infinite ordered simplex @xmath72 endowed with the uniform distance , whose evolution is described formally by : given that the current state is @xmath73 , two terms @xmath74 and @xmath75 , @xmath76 , of @xmath73 are chosen and merge into a single term @xmath77 ( which implies some reordering of the resulting sequence ) at a rate equal to @xmath77 .
a version @xmath78 of this process defined for times describing the whole real axis is called _
eternal_. we refer to @xcite for background .
as shown in @xcite , eternal additive coalescents can be encoded by certain bridges .
specifically , let @xmath79 be a cdlg real - valued process with exchangeable increments , such that @xmath80 .
suppose further that @xmath81 has infinite variation and no negative jumps a.s .
then @xmath81 attains its overall infimum at a unique location @xmath82 ( which is uniformly distributed on @xmath83 $ ] ) , and @xmath81 is continuous at @xmath82 .
consider the so - called _ vervaat transform _ which maps the bridge @xmath81 into an excursion @xmath84 defined by @xmath85 where the addition @xmath86 is modulo @xmath16 .
finally , we let for @xmath87 @xmath88 and introduce @xmath89 as the random element of @xmath90 defined by the ranked sequence of the lengths of the constancy intervals of the process @xmath91 . here , a constancy interval means a connected component of the complement of the support of the stieltjes measure @xmath92 .
finally , if we define @xmath93 , then @xmath94 is an eternal additive coalescent ( see section [ subsec : extreme ] for comments and details ) . in this work ,
eternal additive coalescent associated to certain remarkable bridges will play a key role .
more precisely , we write @xmath95 for the eternal additive coalescent @xmath71 constructed above when @xmath96 is a standard brownian bridge ; so that @xmath97 is the so - called standard additive coalescent ( cf .
@xcite ) . next ,
for @xmath98 , we denote by @xmath99 a standard spectrally positive stable lvy process with index @xmath49 , that is @xmath100 has independent and stationary increments , no negative jumps , and @xmath101 we call _ standard stable loop
_ a _ loop _ and not a _ bridge _ to avoid a possible confusion : even though @xmath102 starts from @xmath62 , ends at @xmath62 and has exchangeable increments , it does not have the same law as the stable process @xmath100 conditioned on @xmath103 ! ] of index @xmath49 the process @xmath102 defined by @xmath104 we finally write @xmath105 for the eternal additive coalescent @xmath71 constructed above when the bridge @xmath81 is the standard stable loop of index @xmath49 .
we are now able to state our main result .
[ convaddcoal ] the process @xmath106 converges as @xmath107 in the sense of weak convergence of finite - dimensional distributions to some process @xmath108 . the exponential time - changed process @xmath109 is an eternal additive coalescent ; more precisely : \(i ) when @xmath110 , @xmath109 is distributed as @xmath111 \(ii ) when @xmath112 , @xmath109 is distributed as @xmath113 it might be interesting to discuss further the role of the parameter @xmath49 and the interpretation in terms of phase transition .
as it was already mentioned , the renewal theorem entails than the number of drops of paint needed for the complete covering is @xmath114 , a quantity which is not sensitive to @xmath49 .
it is easy to show that for every @xmath115 , there are no macroscopic painted components when only @xmath116 $ ] drops of paint have fallen , so the phase transition ( i.e. the number of drops which is needed for the appearance of macroscopic components ) occurs for numbers close to @xmath117 .
more precisely , the regime for the phase transition is of order @xmath118 ; so the phase transition occurs closer to @xmath117 when @xmath49 is larger .
we would like also to stress that one - dimensional distributions of the limiting additive coalescent process @xmath119 depend on @xmath49 , but not its semigroup which is the same for all @xmath50 $ ] .
a heuristic explanation might be the following : the number of drops needed to complete the covering once the phase transition has occurred is too small ( of order @xmath120 ) to observe significant differences in the dynamics of aggregation of macroscopic painted components .
* remark . *
our model bears some similarity with another parking problem on the circle , where drops of paints fall uniformly on the circle and then are brushed clockwise , but where overlaps are now allowed ( some points may be covered this way several times ) , call it the `` random covering of an interval '' problem .
however , as showed in @xcite , this last model has very different asymptotics from those of the parking problem , as it turns out that the random covering of an interval is related to kingman s coalescent rather than the additive coalescent .
a shared feature is that the phase transition of the random covering problem appears also when the circle is almost completely covered , but for example the different fragments are ultimately finite in number rather than infinite .
we also mention yet another parking problem , first considered by rnyi ( see @xcite ) . in
can be formulated as follows : caravans with size @xmath61 are placed on @xmath26 ( the original work rather considers @xmath121 ) one after another , but the locations @xmath74 where cars park are chosen uniformly among spaces that do not induce overlaps and splitting of caravans , i.e. so that the length of the arc from @xmath74 to @xmath122 is exactly @xmath61 .
this is done until no uncovered sub - arc of @xmath26 with size @xmath123 remains .
this process does not involve coalescing blocks of cars , and one is rather interested in the properties of the random number of cars that are able to park .
the method in @xcite relies on an encoding _ parking _ function which is shown to be asymptotically related to a function of standard brownian bridge , and a representation of the standard additive coalescent due to bertoin @xcite .
our approach to theorem [ convaddcoal ] is close in spirit to that of @xcite , and uses the representation of eternal additive coalescent that we presented above ; we briefly sketch it here .
first , we encode the process @xmath124 by a bridge with exchangeable increments in sect .
[ sec : bridge ] . in sect.[sec : conv ] , we show that this bridge converges to some bridge with exchangeable increments that can be represented in terms of the standard brownian bridge ( for @xmath110 ) or the standard stable loop ( for @xmath112 ) .
theorem [ convaddcoal ] then follows readily .
the rest of this work is organized as follows . in section
[ sec : bridge ] we provide a representation of the painted components in terms of a bridge and its vervaat s transform .
the convergence of these bridges when @xmath61 tends to @xmath62 is established in section [ sec : conv ] , and that of the sequence of the sizes of the painted components in section [ convxeps ] .
section [ sec : discrete ] is devoted to a brief discussion of the analogous discrete setting ( i.e. knuth s parking for caravans ) , and finally some complements are presented in section [ sec : compl ] .
we develop a representation of the parking process with the help of bridges with exchangeable increments , which is crucial to our study .
let us first give the proper definition the of sequence @xmath125 of the introduction .
we identify the circle @xmath22 with @xmath52 and write @xmath126 for the canonical projection .
if @xmath127 is a measurable subset of @xmath26 ( identified with @xmath52 ) , let @xmath128 be its repartition function defined by @xmath129\cap a)$ ] for @xmath130 , where @xmath131 is lebesgue measure .
also , extend @xmath128 on the whole real line with the formula @xmath132 .
given @xmath29 for some @xmath133 , let @xmath134 notice that the arc @xmath135 oriented clockwise from @xmath30 to @xmath136 has length @xmath137 .
then let @xmath35 be the interior of the closure of @xmath138 .
the point in taking the closure and then the interior is that we consider that two painted connected components of @xmath22 that are at distance @xmath62 constitute in fact a single painted connected component .
define @xmath139 and @xmath140 in @xmath141 , so @xmath142 is a cdlg function ( right - continuous with left - limits ) on @xmath143 .
consider it as a function on @xmath26 by letting @xmath144 where @xmath145 is the element of @xmath146 .
the quantity @xmath147 can be thought of as the quantity of cars of the @xmath31-th caravan that try to park at @xmath148 .
see figure [ fig : park2 ] .
we consider the _ profile _ @xmath149 of the parking at step @xmath150 , so @xmath151 is the total quantity of cars that have tried ( successfully or not ) to park at @xmath148 ( with the convention that @xmath152 ) before the @xmath31-th caravan has arrived .
[ h ] for @xmath153 , \(i ) the set @xmath29 is the interior of the support of @xmath154 .
\(ii ) @xmath155 .
\(iii ) @xmath156 jumps at times @xmath157 with respective jump magnitudes @xmath158 , and has a drift with slope @xmath159 on its support.that is , if @xmath160\subseteq\supp(h^{{{\bf p}}}_i)$ ] , @xmath161 properties ( i ) and ( iii ) are easily shown using a recursion on @xmath3 and splitting the behavior of @xmath142 on @xmath162 and @xmath163 .
we give some details for ( ii ) . for @xmath164 ,
notice that by definition @xmath122 can not be a point of increase of @xmath165 , i.e. a point such that @xmath166 for every @xmath54 .
therefore , @xmath167 and @xmath168 for @xmath169 . since it follows by continuity of @xmath165 that @xmath170 , ( ii ) is proved .
consider the _ bridge function _ : @xmath171 which starts from @xmath172 and ends at @xmath173 .
we extend @xmath174 to a function on @xmath175 by setting @xmath176 . for any @xmath177
, it is easily seen using ( iii ) in lemma [ h ] that @xmath178 } ( h^{{{\bf p}}}_i(v-)+b^{{{\bf p}}}_i(u))\right).\ ] ] suppose @xmath179 is such that @xmath180 ( here @xmath181 ) , call such a number a _ last empty spot_. by ( ii ) , lemma [ h ] , the set of last empty spots is not empty since it contains @xmath182 . on the other hand , by ( i ) in the same lemma ,
the support of @xmath183 is the closure of @xmath184 which has measure @xmath16 , hence it is @xmath26 . by ( iii ) , we conclude by letting @xmath185 that @xmath186 for @xmath187 , so for @xmath188 , @xmath189 necessarily since @xmath183 is non - negative .
this implies that the last empty spots are those @xmath179 s such that @xmath190 .
we choose one of them by letting @xmath191:b^{{{\bf p}}}_m(x-)=\inf_{u\in[0,1]}b^{{{\bf p}}}_m(u)\},\ ] ] the first location when the infimum of @xmath192 is reached .
we have proved [ v ] for any @xmath193 , @xmath194 } b^{{{\bf p}}}_i(u).\ ] ] recall that we are interested in @xmath39 , the ranked sequence of the lengths of the interval components of @xmath38 , where @xmath38 can be viewed as the painted portion of the circle after @xmath3 drops of paint have fallen , or the set of occupied spots after the @xmath3-th caravan has arrived .
lemma [ h](i ) enables us to identify @xmath38 as the interior of support of the function @xmath154 , and since the lebesgue measure of the interval components of the interior of the support of @xmath156 is not affected by a cyclic shift , we record the following simple identification [ lamb ] for every @xmath195 , @xmath39 coincides with the ranked lengths of the intervals of constancy of the function @xmath196 } b^{{{\bf p}}}_i(u)\,,\qquad x\in [ 0,1].\ ] ]
we now consider a rescaled randomized version of the bridges introduced above .
let @xmath197 , where @xmath198 is obtained as above with data @xmath199 , and these quantities are introduced in the introduction .
so for @xmath200 @xmath201 because @xmath202 . recall that @xmath203 denotes the standard brownian bridge , and @xmath102 the standard stable loop with index @xmath49 as defined in ( [ pontstable ] ) .
[ convbeps ] as @xmath107 , the bridge @xmath204 converges weakly on the space @xmath205 of cdlg paths endowed with skorokhod s topology , to a bridge with exchangeable increments @xmath206 . more precisely : \(i ) if @xmath110 then @xmath81 is distributed as @xmath207 .
\(ii ) if @xmath45 , then @xmath81 is distributed as @xmath208 the proof of lemma [ convbeps](ii ) will use the following well - known representation : @xmath209 where @xmath210 is a sequence of i.i.d .
uniform@xmath121 r.v.s , @xmath211 is the ranked sequence of the atoms of a poisson measure on @xmath212 with intensity @xmath213 , and these two sequences are independent .
more precisely , the series in the right - hand side does not converge absolutely , but is taken in the sense @xmath214 where the limit is uniform in the variable @xmath148 , a.s .
this representation follows immediately from the celebrated lvy - it decomposition , specified for the stable process @xmath100 , as the process of the jumps of the latter is a poisson point process on @xmath215 with intensity @xmath216 .
see also kallenberg @xcite .
following kallenberg @xcite , we represent the jump sizes of the bridge @xmath204 by the random point measure @xmath217 by theorem 2.3 in @xcite , we have to show : @xmath218 and @xmath219 where the convergence is in law with respect to the weak topology on measures on @xmath220 , and in ( [ ii ] ) , @xmath211 is the ranked sequence of the atoms of a poisson measure on @xmath212 with intensity @xmath213 . case ( i ) is easier to treat . indeed , notice that the total mass of @xmath221 is @xmath222 since @xmath223 , the law of large numbers gives @xmath224 .
now let @xmath225 so to prove ( [ i ] ) , it suffices to show that @xmath226 in probability .
notice that @xmath227 .
let @xmath228 and @xmath229 .
then @xmath230 the second term converges to @xmath62 since @xmath231 a.s .
for the first term , notice that @xmath232 taking logarithms and checking that @xmath233 as @xmath107 ( which holds since @xmath234<\infty$ ] ) , we finally obtain that @xmath235 .
this completes the proof of ( [ i ] ) .
now we turn our attention to ( [ ii ] ) .
it suffices to show that for every function @xmath236 , say of class @xmath237 with bounded derivative @xmath238 see for instance section ii.3 in le gall @xcite . in this direction ,
recall from the classical formula for poisson random measures that @xmath239 to start with , we observe from the renewal theorem that @xmath240 converges to @xmath62 in probability as @xmath241 , so in ( [ laplace ] ) , we may replace @xmath221 by @xmath242 next , for every @xmath243 , we consider the random measure @xmath244 again , by the ( elementary ) renewal theorem , @xmath245 in probability , so for every @xmath228 , the event @xmath246[encadre ] has a probability which tends to @xmath16 as @xmath241 .
now @xmath247 taking logarithms , we have to estimate @xmath248 by ( [ tail ] ) and dominated convergence , we see that the preceding quantity converges as @xmath241 towards @xmath249 taking @xmath250 , using ( [ encadrem ] ) and letting @xmath251 tend to @xmath62 , we see that ( [ laplace ] ) holds , which completes the proof of the statement .
in this section , we deduce theorem [ convaddcoal ] from lemmas [ lamb],[convbeps ] . recall the definition of the bridge @xmath174 in section [ sec : bridge ] . for @xmath252 ,
let @xmath253 be the bridge @xmath254 with data @xmath255 , so @xmath256 .
let also @xmath257 be the left - most location of the infimum of @xmath204 , and @xmath258 the vervaat transform of @xmath204 .
by lemma [ lamb ] , @xmath259 coincides with the ranked sequence of lengths of constancy intervals of the infimum process of @xmath260 where the constant @xmath261 has no effect and is added for future considerations .
[ drift ] for every @xmath87 , the difference @xmath262 converges in probability for the uniform norm to the pure drift @xmath263 as @xmath107 .
recall from the renewal theorem that @xmath264 in probability as @xmath265 .
therefore , we might start the sum appearing in the statement from @xmath266 .
now , the sequences @xmath267 and @xmath268 have the same distribution . up to doing the substitution ,
lemma [ drift ] for fixed @xmath269 is therefore a simple application of the strong law of large numbers .
the conclusion is obtained by standard monotonicity arguments . as a consequence of lemmas [ convbeps ] , [ drift ] , and the fact that @xmath270 is continuous
, the process @xmath271 converges in the skorokhod space to @xmath272 where @xmath273 is the vervaat transform of the limiting bridge @xmath81 which appears in lemma [ convbeps ] , @xmath82 being the a.s .
unique location of its infimum .
now letting @xmath274 be the infimum process of @xmath275 and @xmath89 be the decreasing sequence of lengths of constancy intervals of @xmath274 , we have [ convintervalles ] the process @xmath276 converges to @xmath277 in the sense of weak convergence of finite - dimensional marginals .
the technical point is that skorokhod convergence of @xmath278 to @xmath279 , though it does imply convergence of respective infimum processes , does not _ a priori _ imply that of the ranked sequence of lengths of constancy intervals of these processes .
however , this convergence does hold because for every @xmath87 , if @xmath280 is such a constancy interval , then @xmath281 for @xmath282 , a.s .
see e.g. lemmas 4 and 7 in @xcite .
this proposition proves theorem [ convaddcoal ] .
indeed , recall from lemma [ convbeps ] that @xmath283 , where @xmath284 and for @xmath98 @xmath285 then plainly , @xmath286 , and hence the limiting process @xmath287 is distributed as @xmath288 .
in situations involving parking problems , it may be more natural to consider discrete parking lots , i.e. @xmath1 instead of the unit circle , and caravans with integer sizes , e.g. as in knuth s original parking problem .
each caravan chooses a random spot , uniform on @xmath1 , and tries to park at that spot .
studying the frequencies of blocks of cars fits with our general framework by taking @xmath43 with integer values , @xmath289 and @xmath290 .
rename by @xmath291 the former quantity @xmath63 ( the number of caravans ) .
let @xmath292 so @xmath293 would be @xmath294 in the notation above .
the analog of lemma [ drift ] is still true when replacing @xmath293 by @xmath295 , without essential change in the proof .
thus to obtain the very same conclusions as in the preceding sections , it suffices to check a result similar to lemma [ convbeps ] .
namely , we must prove that @xmath296 in the skorokhod space as @xmath297 .
now it is easy to check that a.s .
, @xmath298 for every @xmath299 $ ] , because no @xmath300 is rational a.s .
therefore , it suffices to check that @xmath301 converges to @xmath81 in distribution for the skorokhod topology on @xmath205 . up to using skorokhod s representation theorem ,
this is done by taking @xmath302 and @xmath303 in the next lemma .
[ sko ] let @xmath304 be a sequence of functions converging in @xmath205 to @xmath305 .
for @xmath306 let also @xmath307 be a right - continuous non - decreasing function ( not necessarily bijective ) from @xmath83 $ ] to @xmath83 $ ] , such that the sequence @xmath308 converges to the identity function uniformly on @xmath83 $ ] .
then @xmath309 in @xmath205 .
first consider the case @xmath310 for every @xmath0 .
fix @xmath54 .
let @xmath311 be the right - continuous inverse of @xmath307 defined by @xmath312:\kappa_n(y)>x\}.\ ] ] it is easy to prove that @xmath313 for every @xmath148 . since @xmath305 is cdlg , one may find @xmath314 such that the oscillation @xmath315 for @xmath316 , where @xmath317 since @xmath307 approaches the identity , for @xmath0 large we may assume @xmath318 for @xmath316 . define a time - change @xmath319 ( i.e. an increasing bijection between @xmath83 $ ] and @xmath83 $ ] ) by interpolating linearly between the points @xmath320
. now let @xmath321 $ ] .
suppose @xmath322 for some @xmath316 , and notice that @xmath323 .
therefore , @xmath324 belongs to @xmath325 as well as @xmath326 by definition of @xmath327 , and @xmath328 else , one must have @xmath329 or @xmath330 , and the result is similar .
finally , doing the same reasoning for @xmath331 converging to @xmath62 slowly enough gives the existence of some time - changes @xmath327 converging to the identity uniformly such that @xmath332}|f(\kappa_{n}(x))-f(\lambda_{n}(x))|\leq 2\eps_{n}$ ] , hence giving convergence of @xmath333 to @xmath305 in the skorokhod space . in the general case , for every @xmath334 let @xmath319 be a time - change such that @xmath319 converges to the identity as @xmath297 and @xmath335 converges to @xmath305 uniformly .
take @xmath336 . then @xmath337 uniformly , so it suffices to show that @xmath338 in @xmath205 , which is done by the former discussion .
in particular , we recover and extend a certain number of results from @xcite .
in this section , we would like to provide some information on the eternal additive coalescents @xmath339 for @xmath340 , which appear in theorem [ convaddcoal ] . to start with
, we should like to specify the representation of @xmath339 as a mixture of so - called _ extreme _ eternal additive coalescents ( @xcite , @xcite ) . in this direction , let us first consider a sequence @xmath341 of non - negative numbers satisfying @xmath342 and @xmath343 following kallenberg @xcite we associate to @xmath344 a _ bridge with exchangeable increments _
@xmath345 where @xmath346 denotes a sequence of iid uniform variables and @xmath347 is an independent standard brownian bridge .
we write @xmath348 for the eternal additive coalescent associated to the bridge @xmath349 as explained in the introduction and call such @xmath348 extreme . according to (
* theorem 15 ) , every eternal version of the additive coalescent @xmath71 can be obtained as a mixing of shifted versions of extreme eternal additive coalescents @xmath348 , i.e. @xmath71 can be expressed in the form @xmath350 with @xmath351 random .
equivalently , @xmath71 can be viewed as the eternal additive coalescent constructed in the introduction from the bridge with exchangeable increments @xmath352 . as observed by aldous and pitman @xcite ,
the mixing variables @xmath351 can be recovered from the initial behavior of @xmath71 : @xmath353 in the case of the standard stable loop @xmath102 with @xmath112 , recall from the lvy - it decomposition that @xmath354 and @xmath355 is the ranked sequence of the atoms of a poisson random measure on @xmath212 with intensity @xmath356 . in particular , @xmath357 has the law of a ( positive ) stable variable with index @xmath358 and @xmath359 is such that the sequence of squares @xmath360 is distributed according to the poisson - dirichlet law @xmath361 ; see pitman and yor @xcite .
we also stress that every coalescent @xmath348 can be obtained as a limit of appropriate caravan parking problems , which are quite natural given the results of @xcite .
precisely , suppose that a sequence of probabilities @xmath362 satisfying @xmath363 is given , and satisfies @xmath364 for a sequence @xmath344 as described above , and where @xmath365 when @xmath366 .
for every @xmath0 , let @xmath367 be a uniform permutation on @xmath368 .
consider the parking problem where the caravans which try to park successively have magnitudes @xmath369 .
let @xmath53 be independent uniform@xmath121 random variables independent of @xmath367 , so we may consider the bridge with exchangeable increments @xmath370 kallenberg s theorem shows that under the asymptotic assumptions on @xmath371 , @xmath293 converges in distribution to the bridge @xmath372 defined above .
the key to this lemma is to show that @xmath376 in probability as @xmath297 .
the result is then obtained via the so - called `` weak law of large numbers for sampling without replacement '' : if @xmath377 is a sequence with sum @xmath68 satisfying @xmath378 as @xmath297 , and if @xmath367 is a uniform permutation on @xmath379 , then for every rational @xmath380 $ ] , @xmath381 in probability ( in fact in @xmath382 ) .
the result in probability remains true if @xmath377 is random with sum @xmath68 , and @xmath383 in probability .
one concludes that the process @xmath384 converges in probability to @xmath385 for the uniform norm by a monotonicity argument .
the lemma is then proved by letting @xmath386 ( note that this last term is @xmath387 , which goes to @xmath62 ) .
so let us show ( [ maxzero ] ) .
to this end , let @xmath388 , then @xmath389 in probability , since @xmath390\sim\sigma({{\bf p}}^n)^{-1}(1-\rho)$ ] goes to infinity ( notice @xmath391 ) while @xmath392\sim e[x^{\rho}_{n}]^2 $ ] , as a simple computation shows .
therefore , @xmath393 in probability .
consequently , for any @xmath394 , the quantity @xmath395 goes to @xmath16 , so @xmath396 in probability .
but then , for any @xmath54 , if @xmath397 is such that @xmath398 , then @xmath399 for @xmath0 large . up to taking @xmath0 even larger , with probability close to @xmath16 , @xmath400 for @xmath401 and therefore @xmath402 , hence ( [ maxzero ] ) .
[ thetacoal ] as @xmath297 , under the asymptotic regime ( [ regime ] ) , the process @xmath405 converges in the sense of weak convergence of finite - dimensional marginals to the time - reversed eternal additive coalescent @xmath406
. it would also be interesting to determine the marginal laws of the fragmentation @xmath407 .
the task seems quite difficult if started from the description of @xmath408 in terms of lengths of constancy intervals of vervaat transform of bridges , because excursion theory seems powerless here , unlike in @xcite . in particular
, the fact that the fragmentation is based on stable loops and not stable bridges impedes the application of results of miermont @xcite on additive coalescents based on bridges of certain l ' evy processes .
another way to start the exploration is to use the representation of fragmentation processes @xmath409 described in the preceding section with the help of inhomogeneous continuum random trees ( icrt ) discussed in @xcite .
in particular , it is easy to obtain the first moment of a size - biased pick from a ( random ) positive sequence @xmath410 with sum @xmath411 a.s .
is a random variable of the form @xmath412 , where @xmath413 . ]
@xmath414 from the sequence @xmath89 for any fixed @xmath68 , as follows .
let us recall the basic facts on the icrt@xmath415 construction of @xmath416 .
the icrt can be viewed via a _ stick - breaking construction _ as the metric completion of the positive real line @xmath417 endowed with a non standard metric .
precisely , suppose we are given the following independent random elements : * a poisson process @xmath418 on the octant @xmath419 , with intensity @xmath420 , so in particular @xmath421 is a poisson process with intensity @xmath422 , * a sequence of independent poisson processes @xmath423 with respective intensities @xmath424 .
we distinguish the points @xmath425 as _ joinpoints _ , while @xmath426 are called _
cutpoints_. if @xmath251 is a cutpoint , let @xmath427 be its associated joinpoint , i.e. @xmath428 . by the assumption on @xmath344 , it is a.s .
possible to arrange the cutpoints by increasing order @xmath429 .
we then construct a family @xmath430 of `` reduced trees '' as follows .
cut the set @xmath212 into `` branches '' @xmath431 $ ] , where by convention @xmath432 .
let @xmath433 be the segment @xmath434 $ ] , endowed with the usual distance @xmath435 .
then given @xmath436 , we obtain @xmath437 by adding the branch @xmath438 $ ] somewhere on @xmath439 , and we plant the left - end @xmath440 on the joinpoint @xmath441 ( since a.s .
@xmath442 , the point @xmath443 is indeed an element of @xmath439 ) .
precisely , @xmath444 $ ] and @xmath445 if @xmath446 , @xmath447 if @xmath448 $ ] , and @xmath449 if @xmath450,y\in{\cal r}(k)$ ] .
as the distances @xmath451 are compatible by definition , this defines a random metric space @xmath452 such that the restriction of @xmath453 to @xmath439 is @xmath451 , we call its metric completion @xmath454 the icrt@xmath415 , its elements are called _ vertices_. the point @xmath455 is distinguished and called the _
root_. one can see that @xmath454 is an @xmath175-tree , i.e. a complete metric space such that for any @xmath456 there is a unique simple path @xmath457 $ ] from @xmath148 to @xmath145 , which is isometric to the segment @xmath458 $ ] , i.e. is a geodesic .
moreover , it can be endowed with a natural measure @xmath459 which is the weak limit as @xmath297 of the empirical measures @xmath460 .
this measure is non - atomic and supported on _ leaves _ , i.e. vertices @xmath461 such that @xmath462\setminus\{y\}$ ] for any @xmath463 .
non - leaf vertices form a set called the _ skeleton_.
a second natural measure is the lebesgue measure @xmath14 on @xmath454 , i.e. the unique measure such that @xmath464)=d(x , y)$ ] for any @xmath465 , and this measure is supported on the skeleton .
now for each @xmath68 consider a poisson measure on @xmath454 with atoms @xmath466 , with intensity @xmath467 , so the different processes are coupled in the natural way as @xmath68 varies , i.e. @xmath466 increases with @xmath68 .
these points disconnect the tree into a forest of disjoint connected tree components , order them as @xmath468 by decreasing order of @xmath459-mass .
then the process @xmath469 of these @xmath459-masses has same law as @xmath416 . a size - biased pick from this sequence of masses
is then obtained as the @xmath459-mass of the tree component at time @xmath68 that contains an independent @xmath459-sample , conditionally on @xmath470 .
therefore , if @xmath471 denotes such a size - biased pick , @xmath472 $ ] is the probability that two independent @xmath459-samples @xmath473 belong to the same tree component of the cut tree , i.e. that no atom of the poisson measure at time @xmath68 falls in the path @xmath474 $ ] , and hence it equals @xmath475 $ ] .
it turns out @xcite that @xmath476 has same law as the length @xmath477 of the first branch ( i.e. the length of @xmath433 ) .
it is easy to see ( see also @xcite ) that this branch s length has law @xmath478 in our setting , recall that the random sequence @xmath479 is related to that of the atoms @xmath480 of a poisson measure on @xmath212 with intensity @xmath481 by ( [ equt ] ) and ( [ equth ] ) .
observe that @xmath482 , and since we must take a poisson process with intensity @xmath483 on the skeleton of the icrt(@xmath479 ) , terms @xmath484 cancel out and @xmath485=e[e^{-t\eta}]$ ] where @xmath486\\ & = & \exp\left(-\int_0^{\infty}\frac{\alpha(\alpha-1)\d x}{\gamma(2-\alpha ) x^{1+\alpha } } ( 1-\exp(-rx+\log(1+rx)))\right)\\ & = & \exp(-(\alpha-1)r^{\alpha}),\end{aligned}\ ] ] which is a weibull distribution .
this gives ( at least in principle ) the first moment @xmath487= \int_0^{\infty}\alpha(\alpha-1)r^{\alpha-1}\exp\left(-t r-(\alpha-1 ) r^{\alpha}\right)\d r.\ ] ] in principle , this method could be used for the computation of moments of higher order , where the length @xmath488 of the first branch would simply be replaced by the total length @xmath489 of @xmath439 .
unfortunately , the distribution of @xmath489 is complicated for @xmath490 , and seems intractable for higher @xmath491 s ( @xcite ) . | we consider a generalized version of knuth s parking problem , in which caravans consisting of a random number of cars arrive at random on the unit circle .
then each car turns clockwise until it finds a free space to park . extending a recent work by chassaing and louchard @xcite
, we relate the asymptotics for the sizes of blocks formed by occupied spots with the dynamics of the additive coalescent . according to the behavior of the caravan s size tail distribution ,
several qualitatively different versions of eternal additive coalescent are involved .
* keywords : * parking problem , additive coalescent , bridges with exchangeable increments . *
m.s.c .
code : * 60f17 , 60j25 . |
the _ voronoi diagram _ of points @xmath0 in @xmath1-dimensional euclidean space decomposes @xmath2 into @xmath3 convex polyhedra , called _
voronoi domains_. the domain of @xmath4 is the set of points @xmath5 for which @xmath4 minimizes the euclidean distance : @xmath6 for points in general position , the geometrically realized nerve ( the straight - line dual in @xmath7 ) decomposes the convex hull of the points into simplices , called the _
delaunay triangulation_. it is a particular _ geometric triangulation _ of the points , which is a simplicial complex whose vertices are the given points and whose underlying space is convex .
delaunay triangulations are used in numerous applications and often preferred over other geometric triangulations .
the question arises why the delaunay triangulation is better than others . in the plane ,
the advantages of delaunay triangulation are sometimes rationalized by the max - min angle criterion @xcite .
it requires that the diagonal of every convex quadrangle should be chosen to maximize the minimum of the six angles in the two triangles making up the quadrangle @xcite .
it follows that the delaunay triangulation lexicographically maximizes the non - decreasing sequence of angles .
triangulations of the same finite collection of points in @xmath8 dimensions can have different length sequences of angles . instead of comparing them directly
, we may consider a functional , which assigns a real number to every triangulation . in this paper , we consider functionals that are defined for @xmath1-simplices and we assign the sum of the values over all @xmath1-simplices to the triangulation .
for example , instead of the angle sequence we may consider the functional that maps a triangulation to the sum of minimum angles of @xmath1-simplices . in @xmath7 ,
this functional attains its minimum for the delaunay triangulation .
other functionals that attain their minimum for the delaunay triangulation in the plane have been studied in @xcite and generalized from finite collections to ( infinite ) delone sets in @xcite .
for example , the rajan functional is defined by mapping a triangle @xmath9 with edges of length @xmath10 to @xmath11 times the area of @xmath9 .
there is an intuitive geometric interpretation obtained by lifting a point with coordinates @xmath12 and @xmath13 to @xmath14 , which is a point on the standard paraboloid in @xmath15 .
similarly , we lift @xmath9 to a triangle in space , namely the one spanned by the three lifted vertices .
the rajan functional maps @xmath9 to the volume between the paraboloid and the lifted triangle , of course restricted to the vertical prism over @xmath9 .
the rajan functional of a geometric triangulation is therefore the volume between the paraboloid and the lifted triangulation . since the delaunay triangulation corresponds to taking the convex hull of the lifted points @xcite , it is easy to see that the delaunay triangulation minimizes this functional .
the focus in this paper is a related concept , which we refer to as the _ voronoi functional _ ; see also @xcite . for a triangle @xmath9 with acute angles ,
it is the volume below the standard paraboloid and above the tangent planes that touch the paraboloid in the lifted vertices of @xmath9 , again restricted to the vertical prism above @xmath9 . the definition can be extended to triangles with non - acute angles and to @xmath1-simplices , as we will explain in the body of this paper .
our main results are a characterization of the voronoi functional for triangulations in @xmath7 in terms of squared euclidean distances to the given points , and a proof that the voronoi functional attains its maximum for the delaunay triangulation ; see the voronoi cell decomposition and the voronoi optimality theorems in section [ sec3 ] . while the voronoi functional
is defined for topological triangulations of the given points , which allow for folding along edges , the optimality of the delaunay triangulation holds only within the smaller class of geometric triangulations .
similarly , we have a counterexample to the optimality of the delaunay triangulation among the geometric triangulations of points in three and higher dimensions . [
[ outline . ] ] outline .
+ + + + + + + + section [ sec2 ] defines the voronoi functional for geometric and topological triangulations of points in @xmath7 .
section [ sec3 ] introduces the circumcenter map , which is instrumental for the interpretation of the voronoi functional as a volume in @xmath15 , and it proves our two main results . section [ sec4 ] gives counterexamples to extending the optimality result .
section [ sec5 ] concludes the paper .
we introduce the voronoi functional in three steps , beginning with the relatively easy case of an acute triangle .
[ [ acute - case . ] ] acute case .
+ + + + + + + + + + + let @xmath16 be the vertices of an acute triangle , @xmath9 , in @xmath7 .
for a point @xmath17 , we write @xmath18 for the vertex among the three that is nearest to @xmath17 , and we define @xmath19 to interpret @xmath20 geometrically , we introduce the unit paraboloid as the graph of @xmath21 defined by @xmath22 . setting @xmath23 , we note that the graph of @xmath24 is the tangent plane that touches the paraboloid in the point @xmath25 .
the vertical distance between the paraboloid and the plane above a point @xmath26 is @xmath27 subdividing @xmath9 into regions of constant nearest vertex , we get three quadrangles . integrating the squared distance to the vertex over each quadrangle , and adding the results , we get @xmath20 as the volume between the paraboloid and the upper envelope of the three tangent planes ; see figure [ fig : voronoi - acute ] . before extending @xmath20 to obtuse triangles
, we develop an algebraic formula in the acute case .
we use the notation of figure [ fig : voronoi - acute ] , assuming @xmath7 is spanned by the first two coordinate axes of @xmath15 , and @xmath28 is the common origin of @xmath7 and @xmath15 . without loss of generality
, we may assume that @xmath29 is the circumcenter of @xmath9 .
let @xmath30 be the area of the triangle , and @xmath31 the volume of the triangular prism bounded from below by @xmath9 and from above by the triangle with vertices @xmath32 .
letting @xmath33 be the radius of the circumcircle , we have @xmath34 , and therefore @xmath35 . writing @xmath36 , @xmath37 , and @xmath38 for the quadrangles subdividing the acute triangle , we let @xmath39 be the volume of the quadrangular prism between @xmath40 and @xmath41 , and similar for @xmath42 and @xmath43 . unlike suggested by figure [ fig : voronoi - acute ] , the point @xmath44 is below @xmath29 , so that a portion of each quadrangular prism is below the horizontal coordinate plane . the notion of volume we use
is signed , which means that @xmath39 is the ( unsigned ) volume of the portion above @xmath7 minus the ( unsigned ) volume of the portion below @xmath7 .
writing @xmath45 , @xmath46 , and @xmath47 for the lengths of the three edges , we set @xmath48 this is known as the _ rajan functional _ , which is the volume between the paraboloid and the triangle @xmath49 .
it is now easy to express the voronoi functional by subtracting the rajan functional and volumes of the three quadrangular prisms from the volume of the triangular prism : @xmath50 .
after half a page of trigonometric calculations , we get @xmath51 see appendix b. [ [ obtuse - case . ] ] obtuse case .
+ + + + + + + + + + + + to extend the formula to the case of an obtuse triangle , it is convenient to further subdivide each quadrangle into two triangles by drawing the edges between @xmath29 and the three vertices . recalling that @xmath52 is the midpoint of the edge from @xmath53 to @xmath54 , we set @xmath55 and similarly for the other five triangles in the subdivision . in the acute case
, we can rewrite to get @xmath56 let now @xmath9 be obtuse and use to define the voronoi functional of @xmath9 . assuming the angle at @xmath57 exceeds @xmath58 , we get @xmath59 see appendix a. this has a geometric interpretation , which we illustrate in figure [ fig : fold ] .
in particular , two of the six triangles subdividing @xmath9 have negative orientation , namely @xmath60 and @xmath61 , and we record the volume of the corresponding triangular prisms with a minus sign .
the reason for the negative orientation is that @xmath29 lies outside the triangle and , in the illustrated case , on the opposite side of the edge from @xmath53 to @xmath54 .
besides the fold - over interpretation of @xmath20 , we will use the interpretation as the integral over a difference between two squared distances . call a vertex
_ visible _ from @xmath17 if the line segment connecting @xmath17 to the vertex does not intersect @xmath9 other than in this vertex .
writing @xmath62 for the nearest vertex of @xmath9 that is visible from @xmath17 , we have @xmath63 for points
@xmath17 inside @xmath9 , @xmath62 is not defined and we set @xmath64 .
this way we get a formula that is correct both for acute and for obtuse triangles .
indeed , in the acute case , we have @xmath65 for all points
@xmath17 outside @xmath9 , which implies that agrees with . in the obtuse case
, we have @xmath66 for all points @xmath17 in the shaded region shown in figure [ fig : fold ] . in this case , @xmath20 is not the volume between the paraboloid and the three tangent planes restricted to the triangular prism , but rather something smaller than this volume . [ [ in - the - large . ] ] in the large .
+ + + + + + + + + + + + + we extend the voronoi functional by taking the sum over all triangles of a triangulation . specifically , letting @xmath67 be a triangulation of @xmath68 , we set @xmath69 what do we mean by a triangulation of @xmath70 ?
most important is the _ delaunay triangulation _ that consists of all triangles spanned by three points of @xmath70 such that no point of @xmath70 is enclosed by the circumcircle of the triangle . assuming the points are in general position
by which we mean that no four points lie on a common circle the delaunay triangulation is well defined and unique . a more general notion is a _ geometric triangulation _ , which is a simplicial complex in @xmath7 whose vertex set is @xmath70 and whose underlying space is @xmath71 .
clearly , the delaunay triangulation is a geometric triangulation of @xmath70 , and in many ways , it is the most natural and most interesting geometric triangulation of the point set .
more general yet is a _ topological triangulation _ , which is a simplicial complex homeomorphic to a disk whose vertex set is @xmath70 .
in contrast to a geometric triangulation , the triangles of a topological triangulation may intersect as we think of them abstractly , worrying primarily about how they are connected .
every geometric triangulation is also a topological triangulation but not the other way round .
if no three points of @xmath70 are collinear , then every triangle of a topological triangulation maps to a geometric triangle , whose orientation may be positive or negative .
we interpret accordingly , namely that the sign of @xmath20 is the same as that of the orientation of the triangle .
we give a complete description of the voronoi functional for delaunay triangulations , and we prove that among the geometric triangulations of a finite set in @xmath7 , the delaunay triangulation maximizes the voronoi functional .
[ [ the - circumcenter - map . ] ] the circumcenter map .
+ + + + + + + + + + + + + + + + + + + + + let @xmath67 be a geometric triangulation of @xmath68 .
the _ barycenter _ of a simplex @xmath72 is the point @xmath73 that is the average of the vertices of @xmath74 .
a _ flag _ a sequence of simplices in @xmath67 such that each simplex is a proper face of its successor .
the _ barycentric subdivision _ of @xmath67 is the simplicial complex , @xmath75 , whose vertices are the barycenters of the simplices in @xmath67 and whose simplices correspond to flags in @xmath67 ; see figure [ fig : sd ] .
every @xmath72 has a unique smallest circle that passes through its vertices .
we call the center of this circle the _ circumcenter _ , @xmath76 , and its radius the _ circumradius _ , @xmath77 , of the simplex .
we introduce the _ circumcenter map _ and the _ height map _ , @xmath78 each the piecewise linear extension of a vertex map : the first such that @xmath79 , and the second such that @xmath80 , for every @xmath72 . as illustrated in figure [ fig : sd ]
, the circumcenter map distorts the barycentric subdivision so it aligns with the voronoi diagram , with the exception of occasional fold - edges caused by obtuse triangles .
the height map lifts the image of the circumcenter map to @xmath15 . to see how it does it , let @xmath81 be a vertex of @xmath67 , and consider its star in @xmath75 . a triangle , @xmath82 , in this star
is spanned by the barycenters @xmath83 , @xmath84 , @xmath85 , with @xmath86 an edge , and @xmath87 a triangle in @xmath67 .
the images under @xmath88 are the point , @xmath89 , the midpoint of the edge , @xmath90 , and the circumcenter of the triangle , @xmath91 .
the corresponding heights are @xmath92 , @xmath93 , and @xmath94 .
we see that the combined map , @xmath95 , sends the entire star of @xmath53 to the plane that is the graph of @xmath96 .
similarly , the star of @xmath57 is sent to the graph of @xmath97 , and the star of @xmath54 is sent to the graph of @xmath98 .
the stars are glued along shared boundary pieces , which must therefore lie in the common intersection of the planes .
indeed , we have @xmath99 and @xmath100 .
we use the two maps to recast the voronoi functional as an integral .
write @xmath101 and @xmath102 for the signed areas of a triangle and its image under the circumcenter map . with this
, we have @xmath103 the ratio in front of the integral captures the area distortion experienced by @xmath104 , and it is negative iff @xmath88 reverses the orientation of the triangle .
geometrically , we interpret @xmath105 as the signed volume between the paraboloid and the lifted copy of @xmath106 . finally , @xmath107 is the sum of the @xmath105 , over all triangles @xmath108 . [
[ inclusion - exclusion . ] ] inclusion - exclusion .
+ + + + + + + + + + + + + + + + + + + + in the case in which @xmath109 is the delaunay triangulation of @xmath68 and all angles are acute , @xmath110 has a very appealing geometric interpretation as the volume of the body between the paraboloid and the upper envelope of the graphs of @xmath96 , @xmath81 , restricted to within the vertical prism over the convex hull of @xmath70 .
this interpretation applies more generally , namely as long as the angles opposite to the convex hull edges are acute . to explain why this is true , we consider an interior vertex @xmath53 ; see figure [ fig : stars ] on the left .
let @xmath87 and @xmath111 be two triangles and @xmath112 the corresponding edge of the voronoi polygon . letting @xmath113
be the midpoint of the shared edge , @xmath114 , we have @xmath115 and @xmath116 as images under @xmath88 of two edges in @xmath75 . if the angles at @xmath57 and @xmath117 inside the mentioned triangles are acute , then @xmath113 lies in the interior of the voronoi edge , the triangles @xmath118 and @xmath119 both have the same orientation as their preimages under @xmath88 , and they decompose the triangle @xmath120 . to describe the other case , we consider the triangles @xmath111 and @xmath121 with corresponding edge @xmath122 of the voronoi polygon in figure [ fig : stars ] .
if the angle at @xmath123 is obtuse , then the angle at @xmath54 is necessarily acute , and the midpoint @xmath124 of @xmath125 does not lie on the voronoi edge .
instead , @xmath124 lies on the line of @xmath122 so that @xmath126 has the same orientation as its preimage under @xmath88 , and @xmath127 has the opposite orientation . subtracting the integral over @xmath127 from the integral over @xmath126 , we get again the integral over @xmath128 . adding the signed contributions of all triangles in the star
, we get the integral over voronoi polygon .
in contrast to , we formulate the claim by integrating over all points of the _ image _ of a triangle @xmath129 .
we therefore replace the ratio of the determinants by its sign : let @xmath53 be an interior vertex in the delaunay triangulation of a finite set @xmath68 .
then @xmath130 where @xmath131 is the voronoi polygon of @xmath53 , and the sum is over all triangles @xmath104 in the star of @xmath53 inside the barycentric subdivision of the delaunay triangulation . to generalize the lemma to boundary vertices , we may clip the necessarily infinite voronoi polygons to within the convex hull of @xmath70 . with this modification , the lemma
holds provided the angles opposite to boundary edges are acute .
indeed , the signed integrals over the triangles add up to the integral over the cone from @xmath53 to the finite edges of the voronoi polygon plus the finite pieces of the infinite edges that end at the midpoints of the boundary edges @xmath132 and @xmath86 ; see figure [ fig : stars ] on the right . to summarize , we let @xmath133 be the point in @xmath70 minimizing the distance to @xmath17 .
let @xmath134 be the delaunay triangulation of a finite set @xmath68 .
if all angles opposite to edges of the convex hull of @xmath70 are acute , then @xmath135 [ [ difference - of - squared - distances . ] ] difference of squared distances .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + to shed light on the general case , assume that @xmath87 is a triangle with boundary edge @xmath86 and obtuse angle at @xmath54 .
the star of @xmath53 inside @xmath136 contains a linear sequence of triangles . removing the first triangle and the last
, the signed areas add up to the area of the cone of @xmath53 over the finite edges of its voronoi polygon .
since the angle at @xmath54 is obtuse , the cone extends outside the convex hull of @xmath70 , but that extension is covered by the negatively oriented first triangle in the linear sequence .
this triangle covers more , so we have a remaining negative contribution , which is the integral over the region of points @xmath17 outside @xmath71 for which the nearest vertex in @xmath70 is not the nearest vertex on the boundary of @xmath71 ; see figure [ fig : stars ] on the right . to write this more succinctly , we recall that @xmath18 and @xmath62 denote the nearest vertex and the nearest visible vertex of @xmath9 to @xmath26 .
similarly , @xmath133 is the nearest point in @xmath70 to @xmath17 , and we write @xmath137 for the nearest visible vertex of @xmath71 to @xmath17 . for @xmath138 ,
@xmath137 is not defined and we set @xmath139 in this case .
let @xmath134 be the delaunay triangulation of a finite set @xmath68 . then @xmath140 proof .
write @xmath141 , and consider first the case in which all angles opposite to convex hull edges are acute . by
, @xmath110 is the integral , over all points of @xmath71 , of @xmath142 .
since @xmath137 of a point in the convex hull is not defined , this is the same as the integral of @xmath143 , still only over the convex hull of @xmath70 . for a point @xmath144
, we have @xmath145 by assumption on the angles .
it follows that the contribution to the integral outside the convex hull vanishes , which implies the claimed equation in this special case . to extend the equation to the general case
, we note that next to each convex hull edge but outside the convex hull , we get a region of points where the contributions to the nearest visible vertices do not cancel ; see figure [ fig : stars ] on the right . any two such regions are disjoint , and the net effect within each region is the described difference between the squared distances to the nearest point of @xmath70 and the nearest visible vertex on the convex hull boundary . the claimed equation follows .
[ [ optimality - of - the - delaunay - triangulation . ] ] optimality of the delaunay triangulation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we prove that the delaunay triangulation maximizes the voronoi functional .
more than that , we show that the delaunay triangulation maximizes the functional locally , at every point of the plane . to explain this , define @xmath146 where the sum is over all triangles @xmath9 of a geometric triangulation @xmath67 of @xmath70 . with this notation
, we get @xmath147 from .
let @xmath134 be the delaunay triangulation and @xmath67 a geometric triangulation of a finite set @xmath68
. then @xmath148 .
we prove optimality by showing @xmath149 for all @xmath26 .
we have @xmath141 from the voronoi delaunay theorem , so it suffices to show @xmath150 .
consider first a point @xmath138 .
for the triangle that contains @xmath17 , we have @xmath151 , which is positive . for every other triangle
, we have @xmath152 .
let @xmath153 , and consider the sequence of triangles @xmath154 constructed as follows .
if @xmath17 lies inside one of the triangles in the star of @xmath155 , then @xmath156 is this triangle , and @xmath157 . otherwise , there is a triangle @xmath158 such that the line segment from @xmath155 to @xmath17 crosses one of its edges .
let @xmath159 be the nearest visible vertex of @xmath156 , and repeat the construction substituting @xmath159 for @xmath155 to get @xmath160 , and so on until @xmath161 contains @xmath17 or we formed a cycle . in the former case , the sum of squared distances is @xmath162 the right hand side evaluates to @xmath163 .
the contribution of the triangles not in this sequence is non - positive , which implies @xmath164 .
if in the latter case the cycle goes around the triangle that contains @xmath17 , then it must contain a vertex , @xmath165 , that is further from @xmath17 than the vertices of the containing triangle , and in particular further than the closest vertex of that triangle , @xmath166 .
we stop the process at this vertex and get @xmath167 this time , the right hand side evaluates to @xmath168 . the triangle that contains @xmath17 is the only one with a positive contribution , which is @xmath169 , so that we again get @xmath170 , as desired . we still need to show that if the vertices @xmath171 form a cycle , then it goes around the point @xmath17 . to derive a contradiction
, we assume the cycle does not go around @xmath17 , as in figure [ fig : cycle ] . at each point @xmath166 of the cycle ,
we let @xmath172 be the counterclockwise rotation angle between the direction of the cycle at @xmath166 and the vector @xmath173 .
this is a multivalued function , so we pick a branch by choosing its value at the starting point inside @xmath174 and requiring that @xmath175 be continuous along any edge , with right - continuous jumps at the vertices .
the jump is positive at a counterclockwise turn and negative at a clockwise turn , never exceeding @xmath176 in absolute value . assuming the cycle is oriented in a counterclockwise order
, @xmath175 grows by @xmath177 when we go around once .
indeed , because @xmath17 lies outside the cycle , the continuous changes total to @xmath29 , and since the cycle is simple , the jumps total to @xmath177 .
let @xmath171 be the vertex at which the jump changes @xmath175 from a value less than @xmath177 to a value at least @xmath177 .
as illustrated in figure [ fig : cycle ] , the line passing through @xmath178 and @xmath171 separates @xmath17 from @xmath179 . but
this implies that the edge connecting @xmath178 to @xmath171 crosses the edge of the triangle @xmath180 opposite to @xmath171 .
this contradicts that both edges belong to @xmath67 and concludes the proof that the cycle must go around @xmath17 . to extend the argument to points @xmath17 outside the convex hull of @xmath70
, we note that @xmath181 and @xmath182 are both non - positive .
let @xmath33 be the region of points with @xmath183 and note that it is bounded .
we can therefore add vertices and triangles on the outside , maintaining that we still have a delaunay triangulation , such that @xmath33 is completely covered by the added triangles .
adding the same triangles to @xmath67 , we get two new triangulations , @xmath184 and @xmath185 , and we have @xmath186 for every @xmath187 using the above argument .
since we add the same triangles to @xmath134 and to @xmath67 , we have @xmath188 , which implies @xmath149 also for the points @xmath187 . for the remaining points , @xmath189
, we have @xmath190 by definition of @xmath33 , and @xmath191 because @xmath17 lies outside the underlying space of @xmath67 .
we thus get @xmath149 for all points @xmath26 , as claimed .
in this section , we show that the optimality of the delaunay triangulation among all geometric triangulation of a finite point set does not generalize if we extend the family to all topological triangulations . furthermore
, we show that even without this extension , the optimality of the delaunay triangulation does not extend to three dimensions .
consider the triangulation in figure [ fig : folded ] on the left .
all triangles are acute , which implies that this is the delaunay triangulation , @xmath134 , of the eight points .
the topological triangulation , @xmath67 , of the same eight points is obtained by exchanging the positions of points @xmath192 and @xmath193 , moving the edges and triangles along to preserve all incidences ; see figure [ fig : folded ] on the right .
the pentagons @xmath194 and @xmath195 are covered only once by @xmath67 , while @xmath196 is covered three times .
we prove @xmath197 by computing @xmath198 for every point @xmath26 .
all triangles in @xmath67 are acute , which implies that the only non - zero terms come from triangles that contain @xmath17 .
first consider @xmath17 inside @xmath87 , and note that the case @xmath17 inside @xmath199 is symmetric . here
@xmath17 is covered by @xmath87 , which has negative orientation , and by two triangles with positive orientation , namely one of @xmath200 and @xmath201 and one of @xmath202 and @xmath203 . in every case , @xmath181 is the sum of the squared distances to the nearest vertices of these triangles : @xmath204 , in which @xmath205 we have @xmath206 because @xmath17 is closer to @xmath54 than to @xmath123 and @xmath207 .
similarly , @xmath208 .
this implies @xmath209 with strict inequality on a set of positive measure .
next consider @xmath17 inside the region @xmath194 , and note that the case of @xmath17 inside @xmath195 is symmetric .
the region is covered only once , by triangles @xmath200 , @xmath210 , @xmath211 , all of which have positive orientation .
since all three triangles are acute , @xmath181 is the squared distance of the closest vertex of the pentagon @xmath212 , while @xmath182 is the squared distance to the closest vertex of @xmath194 .
point @xmath57 is further from all points within the pentagon than @xmath54 , which again implies @xmath213 .
all triangles are acute , so @xmath214 outside @xmath215 . in summary , @xmath213 for all points @xmath26 , and
the inequality is strict on a set of positive measure .
hence , @xmath216 as claimed
. we may generalize the voronoi functional to three and higher dimensions , defining it as the sum , over all simplices in the barycentric subdivision , of the integral of the signed squared distance : @xmath217 where @xmath53 is the unique vertex of @xmath104 that is also a vertex of @xmath67 . in @xmath218 dimensions ,
the circumcenter map moves two vertices of a tetrahedron in the barycentric subdivision , possibly inverting a tetrahedron twice and thus returning it to positive orientation .
this explains why there may be triangulations for which the voronoi functional exceeds that of the delaunay triangulation .
[ [ double - fold - over . ] ] double fold - over .
+ + + + + + + + + + + + + + + + + we exhibit a tetrahedron for which there are points not contained in the tetrahedron that have a positive contribution to the integral of the squared distance to the nearest vertex . in @xmath7 , a triangle with this property
does not exist .
figure [ fig : badtet ] shows the tetrahedron , @xmath219 , together with the center of the circumsphere , @xmath220 .
the faces @xmath221 , @xmath222 are isosceles triangles with three acute angles each . in contrast , @xmath223 , @xmath224 are isosceles triangles with obtuse angles at @xmath57 and at @xmath117 , and we show the centers of their circumcircles : @xmath225 and @xmath226 .
the circumcenter map leaves the four vertices and the midpoints of the edges fixed , and it moves the barycenters of the four triangles and of the tetrahedron to new locations . as a consequence , some of the images of the @xmath227 triangles in the barycentric subdivision of the boundary of @xmath219 change their orientation within their respective planes . because of the obtuse angles at @xmath57 and at @xmath117 , these are the four triangles that share the midpoint of @xmath114 .
the other @xmath228 triangles preserve their orientation .
the image of the barycenter of the tetrahedron is @xmath29 , the planes of @xmath223 and @xmath224 separate @xmath229 from @xmath29 , while the planes of @xmath221 and @xmath222 do not separate the two points .
it follows that @xmath230 tetrahedra in the barycentric subdivision preserve their orientation , while @xmath231 tetrahedra reverse their orientation . just like in @xmath7 , where we have regions outside an obtuse triangle whose points contribute negatively to the voronoi functional ( see figure [ fig : fold ] )
, we have regions outside @xmath219 whose points contribute positively to the voronoi functional .
specifically , there is such a point @xmath17 near @xmath123 inside the cone of @xmath29 over the blue triangle in figure [ fig : badtet ] .
it satisfies @xmath232 , and its contribution to the voronoi functional is @xmath233 , which is positive .
the argument used to prove the optimality of the delaunay triangulation in @xmath7 does therefore not apply . [
[ counterexample . ] ] counterexample .
+ + + + + + + + + + + + + + + inspired by the construction in figure [ fig : badtet ] , we use a numerical method for estimating the voronoi functional to find points in @xmath15 for which the delaunay triangulation does not maximize the functional .
figure [ fig : counter ] illustrates one such example .
there are six point with coordinates @xmath234 whose convex hull is an octahedron .
the delaunay triangulation decomposes the octahedron into the four tetrahedra sharing the edge @xmath235 .
we compare it with the triangulation , @xmath67 , whose four tetrahedra share the edge @xmath114 .
the values of the voronoi functional are @xmath236 which shows that @xmath134 does not maximize the functional .
as a by - product of the voronoi optimality theorem , we have the optimality of the delaunay triangulation for the functional that takes the sum over all triangles of the squared circumradius times the area . more generally , we define @xmath237 where @xmath238 is the circumradius of @xmath9 .
indeed , @xmath239 so we conclude that @xmath240 attains its minimum at the delaunay triangulation .
similarly , we can prove that @xmath241 attains its minimum at the delaunay triangulation .
writing @xmath242 , we have @xmath243 and therefore @xmath244 .
when we flip an edge to turn a geometric triangulation into the delaunay triangulation , the sum of @xmath245 can not increase , which implies the delaunay triangulation is optimal
. it would be interesting to prove optimality for all @xmath246 .
it might be interesting to generalize the voronoi functional to points with weights ; see e.g. @xcite .
the inclusion - exclusion argument presented in section [ sec3 ] extends to the weighted case , but is it true that the generalized voronoi functional is maximized by the delaunay triangulation for weighted points in @xmath7 ? 15 extremal properties of delaunay triangulations .
_ trudy isa ras _ * 46 * ( 2009 ) , 174187 . a spectral characterization of the delaunay triangulation .
_ comput . aided geom .
design _ * 27 * ( 2010 ) , 295300 . sur la sphre vide .
nauk sssr , otdelenie matematicheskikh i estestvennykh nauk _ * 7 * ( 1934 ) , 793800
. geometry of positive quadratic forms .
_ uspekhi mat .
nauk _ * 3 * ( 1937 ) 1662 .
functionals on triangulations of delaunay sets .
_ moscow math .
j. _ * 14 * ( 2014 ) , 491504 .
_ geometry and topology for mesh generation .
_ cambridge univ . press , cambridge , england , 2001 .
three - dimensional triangulations from local transformations .
_ siam j. sci .
statist .
* 10 * ( 1989 ) , 718741 .
the delaunay triangulation maximizes the mean inradius . _ in _ `` proc .
6th canad .
geom . , 2994 '' , 201206 .
software for @xmath247 surface interpolation . in _ mathematical software iii _ , ed . : j. r. rice , academic press , new york , 1977 , 161194 .
properties of the delaunay triangulation . _ in _ `` proc .
13th ann .
geom . , 1997 '' , 424426 . about optimality of delaunay triangulations . in _ geometry ,
topology , algebra and number theory , applications _ , conf .
dedicated to 120th anniversary of b. n. delone , ( 2010 ) , 166167 .
problems in plane geometry m. : mccme , 2006 .
optimality of the delaunay triangulation in @xmath248 .
_ discrete comput .
* 12 * ( 1994 ) , 189202 .
minimal roughness property of the delaunay triangulation .
_ comput .
aided geom . design _ * 7 * ( 1990 ) , 489497 .
locally equiangular triangulations .
. j. _ * 21 * ( 1978 ) , 243245 .
nouvelles applications des parametres continus a la theorie des formes quadratiques .
_ j. reine angew .
_ * 34 * ( 1908 ) , 198287 .
as an alternative to the proof in section [ sec3 ] , we can show the optimality of delaunay triangulations for the voronoi functional via flips .
more precisely , it is sufficient to show that the voronoi optimality theorem holds for any convex quadrangle .
since the delaunay triangulation may be derived from any triangulation through a sequence of flips , this will imply that the theorem is true for all finite sets @xmath68 .
let @xmath134 and @xmath67 be the delaunay and the non - delaunay triangulations of four points in convex position in @xmath7 .
then @xmath249 for all @xmath26 .
let @xmath250 be the four points in convex position , @xmath26 , and assume @xmath251
. then 1 .
the edge @xmath252 belongs to the delaunay triangulation , because there is a circle ( centered at @xmath17 ) such that @xmath253 are on or inside , and @xmath254 are on or outside the circle , 2 .
the points @xmath113 and @xmath124 do not lie inside the triangle @xmath255 , else they would be closer to @xmath17 than @xmath256 or @xmath257 .
[ a ] implies that if @xmath252 is a diagonal of the quadrangle , then @xmath134 consists of the triangles @xmath258 .
the remainder of the proof is a case analysis , which we illustrate in figure [ fig : cases ] .
case i : : @xmath17 is inside the quadrangle ; see figure [ fig : cases ] , @xmath259st row .
+ case i.1 ; ; @xmath252 is a side of the quadrangle .
then @xmath260 because the values for the two triangles are either @xmath261 and @xmath29 , or @xmath262 and @xmath263 .
; ; @xmath252 is a diagonal . using [ a ] , we get @xmath264 , which is greater than or equal to @xmath265 .
case ii : : @xmath17 lies outside , and only two vertices of the quadrangle are visible from @xmath17 . + case ii.1 ; ; @xmath252 is a side ; see figure [ fig : cases ] , @xmath266nd row .
we have four subcases and @xmath260 in each .
+ case ii.1.a : : both @xmath256 and @xmath257 are visible from @xmath17 .
then @xmath267 for all four triangles .
case ii.1.b : : @xmath256 is visible and @xmath257 is invisible .
then @xmath267 for all four triangles .
case ii.1.c : : @xmath257 is visible and @xmath256 is invisible . in both triangulations ,
the values for the two triangles are @xmath29 and @xmath268 .
case ii.1.d : : both @xmath256 and @xmath257 are invisible . for one triangulation the values for the two triangles are @xmath268 and @xmath269 , while for the other triangulation the values are @xmath270 and @xmath29 .
case ii.2 ; ; @xmath252 is a diagonal ; see figure [ fig : cases ] , @xmath218rd row .
we have two subcases and @xmath271 in each .
by [ a ] we know which triangulation is delaunay .
+ case ii.2.a : : @xmath256 is visible and @xmath257 is invisible .
then @xmath272 and @xmath273 .
case ii.2.b : : @xmath257 is visible and @xmath256 is invisible . then @xmath274 and @xmath275 .
case iii : : @xmath17 is outside the quadrangle and three vertices of the quadrangle are visible from @xmath17 .
+ case iii.1 ; ; @xmath252 is a side ; see figure [ fig : cases ] , @xmath276th row .
we have three subcases and @xmath260 in each .
+ case iii.1.a : : @xmath256 is invisible . by [ b ] there is only one choice for @xmath257 so that the line through @xmath17 and @xmath256 separates @xmath257 from @xmath254 .
both triangulations contain one triangle with value @xmath29 and the other with @xmath263 .
case iii.1.b : : @xmath257 is invisible . as in case iii.1.a , there is only one choice for @xmath256 , and @xmath277 for all four triangles .
case iii.1.c : : @xmath256 and @xmath257 are both visible .
then @xmath278 for all four triangles .
case iii.2 ; ; @xmath252 is a diagonal ; see figure [ fig : cases ] , @xmath279th row .
we have two subcases and @xmath271 in each . by [ a ] we know which triangulation is delaunay .
+ case iii.2.a : : @xmath256 is invisible
. @xmath280 and @xmath281 .
case iii.2.b : : @xmath257 is invisible . then @xmath272 and @xmath282 .
case iii.2.c : : that @xmath256 and @xmath257 are both visible is impossible because it would contradict [ b ] .
the case analysis is exhaustive , which implies @xmath283 , as claimed .
we remark that a flip can have only two possible outcomes : it leaves the value of @xmath284 the same , or it increases this value by @xmath285 .
the latter case happens iff the two vertices closest to @xmath17 form a diagonal .
in this section , we derive formula , both for the acute and obtuse cases .
we will be working in @xmath15 and denote the three cartesian coordinates by @xmath286 .
the voronoi functional is invariant under planar isometries , so we may assume that the circumcenter of the triangle is at the origin .
the points @xmath32 thus have a common @xmath287-coordinate , which is @xmath288 , and the @xmath287-coordinate of @xmath44 is @xmath289 .
we denote the angles of @xmath87 by @xmath290 and the area of @xmath87 by @xmath291 .
we begin with the formula from section [ sec2 ] : @xmath292 we divide the quadrangle @xmath40 into two triangles , @xmath293 and @xmath294 , and similarly for @xmath42 and @xmath43 . the triangles adjacent to side @xmath86 are congruent , and we denote both of them by @xmath295 . similarly , two of the other triangles are denoted by @xmath296 , and two by @xmath297 .
now we rewrite : @xmath298 since @xmath299 is a prism , @xmath300 .
to compute @xmath301 , @xmath302 , @xmath303 , we use the following result .
let @xmath304 be a triangle with vertices @xmath305 , @xmath306 , and the plane @xmath304 perpendicular to the @xmath307-plane .
then we have @xmath308 @xmath309 , in which @xmath310 . proof .
we draw the altitude @xmath311 to divide @xmath304 into @xmath312 and @xmath313
. then we can find the area under this triangle for each plane perpendicular to the @xmath17-axis and integrate it : @xmath314_0^{r\cos^2{\phi } } \\ & = r^4 \tan{\phi } \left ( \tfrac{2}{3 } \cos^6{\phi } - \tfrac{1}{2 } \cos^4{\phi } \right ) .
\end{aligned}\ ] ] similarly , @xmath315 .
writing @xmath316 , we get @xmath317 which evaluates to @xmath318 , as claimed .
writing @xmath319 , we apply the lemma to all six triangles to obtain @xmath320 finally , we combine it with and : @xmath321 we use the identities @xmath322 where @xmath323 is the half - perimeter of @xmath87 , and @xmath324 is its inradius ( for details see ( * ? ? ? * , b ) ) .
@xmath325 the sequence of calculations for the obtuse triangle is almost the same .
assume that @xmath326 .
then the formula is transformed to : @xmath327 in this case , @xmath328 .
therefore , all further calculations are completely the same as in the acute case .
ll @xmath68 & finite set of points + @xmath20 & voronoi functional of triangle + @xmath21 & unit paraboloid + @xmath24 & tangent plane + @xmath16 & vertices + @xmath329 & barycenters + @xmath330 & lifted points + + @xmath72 & simplex in triangulation + @xmath108 & triangle in barycentric subdivision + @xmath331 & barycenter , circumcenter , circumradius + @xmath332 & circumcenter map + @xmath333 & height map + + @xmath334 & nearest , nearest visible vertex + @xmath335 & nearest , nearest visible point + @xmath134 & delaunay triangulation + @xmath336 & sequence of triangles + @xmath337 & local voronoi functional
* section [ sec1 ] : introduction .
* section [ sec2 ] : the voronoi functional . * * @xmath20 . * * @xmath338 . * * @xmath107 . * section [ sec3 ] : geometric triangulations . * * _ circumcenter map_. * * _ height map_. * * interior cancellation lemma . *
* voronoi cell decomposition theorem . *
* voronoi optimality theorem . * section [ sec4 ] : non - optimality .
* * section [ sec41 ] : topological triangulations .
* * section [ sec42 ] : beyond two dimensions .
* section [ sec5 ] : discussion . * * @xmath339 . * section [ appa ] : voronoi optimality by flipping . * * voronoi optimality theorem for flips . * section [ appb ] : calculations . * * angle lemma . | we introduce the voronoi functional of a triangulation of a finite set of points in the euclidean plane and prove that among all geometric triangulations of the point set , the delaunay triangulation maximizes the functional . this result
neither extends to topological triangulations in the plane nor to geometric triangulations in three and higher dimensions . |
quantum chromodynamics ( qcd ) is the theory of the hadronic interactions .
it is an elegant theory whose full non perturbative solution has escaped our knowledge since its formulation more than 30 years ago.@xcite the theory is asymptotically free@xcite and confining.@xcite a particularly good test of our understanding of the nonperturbative aspects of qcd is to study particles where the gauge field plays a more important dynamical role than in the standard hadrons . in particular glueballs , bound states of gluons , represent such a scenario .
the glueball spectrum has attracted much attention since the early days of qcd.@xcite the interest in this subject is related to the significant progress in the understanding of the properties of such states within qcd , as well as , in the new possibilities for their identification in modern experiments .
historically the investigation of the glueball properties started in the bag model by jaffe and johnson.@xcite they found many glueball states with different quantum numbers lying in the mass interval 1000 - 2000 mev .
they emphasized that one should expect rather small widths for such states because their decays in conventional hadrons violate the okubo - zweig - iizuka ( ozi ) rule.@xcite after this pioneering work the study of glueballs was carried out by using various versions of constituent models , by exploiting the qcd sum rule approach and by performing lattice qcd calculations .
glueballs have not been an easy subject to study due to the lack of phenomenological support and therefore much debate has been associated with their properties .
the main achievement of these approaches is the understanding of the deep relation between the properties of the glueball states and the structure of the qcd vacuum . besides
, they provide a determination of the spectrum both in gluodynamics , the theory with just gluons and no quarks , and in qcd .
however , in ( unquenched ) qcd , the results of several calculations for the spectrum are still not universally accepted , in particular , for the lowest lying glueballs.@xcite from the phenomenological point of view it has become clear by now that it is difficult to single out which states of the hadronic spectrum are glueballs because we lack the necessary knowledge to determine their decay properties .
moreover the strong expected mixing between glueballs and quark states leads to a broadening of the possible glueball states which does not simplify their isolation .
the wishful sharp resonances which would confer the glueball spectra the beauty and richness of the baryonic and mesonic spectra are lacking .
this confusing picture has led to a loss of theoretical and experimental interest in these hadronic states .
however , it is important to stress , that if they were to exist they would be a beautiful and unique consequence of qcd . at the present ,
several candidates for the low mass glueballs with quantum numbers @xmath0 , @xmath1 , @xmath2 and @xmath3 are under discussion .
@xcite in this review we will discuss the modern development in glueball spectroscopy from various perspectives .
in section [ sec : lattice ] we will summarize the results that lattice techniques have obtained for the spectrum , both in the pure gauge theory and in the unquenched calculations . in section [ sec : models ] we present a review on constituent models .
section [ sec : sumrules ] is dedicated to discuss qcd sum rules . in section [ sec : prodecay ] the production and decay mechanisms of glueballs in hadronic reactions are discussed . in section [ sec : qgp ] the peculiarities of glueball production and behavior in quark - gluon plasma ( qgp ) is considered .
section [ sec : other ] is dedicated to present two open topics the relation between the pomeron and glueballs and glueball - quarkonium mixing . finally in section [ sec : conclusions ] we extract the main conclusions of our analysis and try to foresee future developments .
gluon self - couplings in qcd suggest the existence of glueballs , bound states of mainly gluons .
investigating glueball physics requires an intimate knowledge of the confining qcd vacuum and it is well known that such properties can not be obtained using standard perturbative techniques . to handle the nonperturbative regime of qcd
, one can resort to numerical methods , known as lattice qcd .
lattice qcd needs as input the quark masses and an overall scale , conventionally given by @xmath4 .
then any green function can be evaluated by taking average of suitable combinations of lattice fields in vacuum samples .
this allows masses and matrix elements , particularly those of weak or electromagnetic currents , to be studied .
one limitation of the lattice approach is in exploring hadronic decays because the lattice , using euclidean time , has no concept of asymptotic states .
lattice qcd was originally formulated by wilson @xcite and is a clever implementation of the qcd dynamics using a finite difference formalism .
the starting point is the correlation function in a discrete euclidean space @xmath5 where @xmath6 is the fermion action and @xmath7 the pure gauge action .
the continuum limit is controlled by the input parameter @xmath8 ( @xmath9 is the number of color ) . by varying the inverse lattice
coupling @xmath10 we vary the lattice spacing @xmath11 .
the fermion action can be integrated out exactly in eq .
to produce the fermion determinant .
the determinant describes the dynamics of the sea quarks . in
quenched qcd calculations , the determinant is set to a constant.@xcite the physics is extracted from the fit @xmath12 @xmath13 are the energy eigenstates and @xmath14 the corresponding masses . in practice , one has to choose the operator @xmath15 which best overlaps with the lowest - lying glueball in the channel of interest .
@xmath15 is thus expanded in a basis with well - defined symmetry properties under the octahedral group and the variational coefficients are determined by monte - carlo simulations.@xcite the spectrum in a box with periodic boundary conditions includes not only single glueball states , but also states consisting of several glueballs and torelons ( gluon excitations which wrap around the toroidal lattice ) .
fortunately , torelons are found to overlap only weakly with single glueball states.@xcite since ( classical ) gluodynamics is dimensionless , its observables will be also dimensionless .
masses are usually expressed in terms of the string tension @xmath16 or the hadronic scale parameter @xmath17 defined through the static potential between quarks @xmath18_{r = r_0}=1.65$].@xcite their values are very close and usually read @xmath19 in lattice calculations there are errors arising from the finite size of the lattice spacing @xmath11 and the finite lattice volume .
but for small enough @xmath11 we expect the continuum limit to be approached as @xmath20 with a constant @xmath21.@xcite the pure gauge spectrum of quenched qcd was investigated initially by morningstar and peardon in an anisotropic lattice.@xcite they used different spacings for the spatial @xmath22 and for the temporal @xmath23 extensions with the ratio @xmath24 .
this technique allows to control better the inherent errors induced by the lattice .
morningstar and peardon identified 13 glueballs below 4 gev . in order to distinguish single from multiple glueball states , they determined approximately the locations of the two - particles states using the mass estimation of the lowest few particles . in their estimates of these locations they assumed that the two glueballs do not interact and that the threshold for their production is given by the energy @xmath25 where @xmath26 and @xmath27 are the masses of the single glueballs .
all states lying below the two - glueball threshold are then single glueball states .
these authors pointed out that they can not rule out a single glueball interpretation for higher states .
they do not find any hint that their states are torelon pairs .
finite volume effects are quite under control .
when going from a @xmath28 lattice to a @xmath29 lattice , the fractional changes on the mass @xmath30 are less than a few percent and consistent with zero .
the largest effect of these errors occur in the @xmath31 and @xmath32 representations of the octahedral group .
they are the main cause of uncertainties in the @xmath33 and @xmath34 glueballs .
the proximity of the two glueball thresholds and the finite volume errors on the @xmath35 lead the authors to withhold judgment on whether or not this level is a single glueball . .
final continuum - limit glueball mass estimates @xmath36 .
when a unique @xmath37 interpretation for a state can not be made , the other possibilities are indicated in the second column .
states whose interpretation requires further study are indicated by a dagger . in column 3 , the first error is the statistical uncertainty from the continuum - limit extrapolation and the second is the estimated uncertainty from the anisotropy . in the final column , the first error comes from the combined uncertainties in @xmath38 , the second from the uncertainty in @xmath39 mev . [ tab : finalmasses ] [ cols="^,^,^,^",options="header " , ] 0.3 cm the status of glueball observation would become clearer with a combination of data of @xmath40 and @xmath41 machines with large statistics , however since the glueball width into two - photons is small , this would require very high luminosity .
it has been pointed out that heavy - ion colliders , due to their large center of mass energy , allow access to the regge region @xmath42 and hence to the production of glueballs in peripheral collisions through photon - photon collisions as well as double pomeron exchange @xcite .
distinctive features can be derived from the decays of the glueballs since they are flavour singlets : decays to @xmath43 identify a flavour octet and radiative decays of glueballs are forbidden . however , these arguments have to be taken with care since mixing of a glueball with mesons , having the same quantum numbers , can occur and would dilute any selection rule .
an interesting idea recently proposed is that of chiral suppression.@xcite if chiral symmetry breaking in glueball decay is dominated by quark masses , then the coupling of a spin zero glueball to light @xmath44 pairs is chirally suppressed .
an interesting consequence is that radiative j/@xmath45 decay becomes a filter for new physics in the @xmath46 channel , since at leading order radiative decays to spin zero light quark mesons are suppressed while radiative decays to @xmath46 glueballs , hybrid , and four quark states are not suppressed .
this idea is not without controversy .
@xcite the glueball groundstate @xmath47 would be an ideal state to find since it has many fundamental connotations.@xcite in gluodynamics , as we have seen , the situation that arises from lattice calculations is clear and the masses of the scalar glueballs are large @xmath48 gev .
however , when sea quarks are considered no firm conclusion about the scalar spectrum can be drawn .
the theoretical calculations based on qcd sum rules and/or low energy theorems lead to contradictory results .
its properties , _
i.e. _ , mass , decay channels and widths still differ among the various approaches .
experimentally there are too many isoscalar , scalar mesons between 0.6 and 1.75 gev to be explained by the naive quark model alone . from the theoretical point of view
there has been a lot of debate about the structure of these states .
let us summarize here the different theoretical explanations for the spectrum .
jaffe obtains the @xmath49 and the @xmath50 in the bag model as cryptoexotic @xmath51 states.@xcite the p - wave @xmath44 scalar nonet is likely to lie in the region of the other spin - triplet p - wave nonets , with isoscalars roughly between @xmath52 and @xmath53 mev .
@xcite dominguez and paver,@xcite bordes , pearrocha and gimnez,@xcite and kisslinger and johnson,@xcite obtain , using low energy theorems and/or sum rule calculations , a low lying and narrow glueball ( mass @xmath54 700 mev and @xmath55 100 mev ) .
vento using low energy theorems has also proposed the existence of a low mass ( @xmath56 mev ) and almost sterile scalar glueball.@xcite narison and collaborators,@xcite using sum rules , get a broader ( 200 - 800 mev ) , heavier ( 700 - 1000 mev ) scalar glueball in this range .
forkel,@xcite obtains a scalar glueball at 1250 @xmath57 200 mev with a large width ( @xmath58300 mev ) . between @xmath59 and 1750 mev
there are three @xmath60 states : @xmath61 , @xmath62 , and @xmath63 . in the analysis of vento @xcite
two scalar glueballs appear in this range , an intermediate mass glueball ( @xmath64 mev ) and another one in the upper range ( @xmath65 mev ) . the narrow state at 1500 mev , discovered in @xmath66 annihilation by crystal barrel ( cb ) is considered by amsler and close a good candidate for the glueball groundstate@xcite while chanowitz considers the @xmath63 as the scalar groundstate@xcite . finally , that the scalar glueball is shared by both resonances is also contemplated.@xcite
relativistic heavy - ion collisions might be a tool to produce glueballs .
it is conceivable that glueball production becomes a dominant part in central nucleon collisions .
two scenarios will be analyzed : * * quark gluon plasma phase * : one expects that at some point above a certain critical temperature a plasma of quarks and gluons , named quark - gluon plasma ( qgp ) is formed.@xcite this phase is characterized by a large amount of thermal gluons . as the qgp cools down the gluons can form singlet configurations via the color interaction . * * strong coulomb phase * : a recent formulation of the dynamics in the region above the transition temperature @xmath67 , based on a description of recent experiments in ultra relativistic heavy ion collisions,@xcite states that , despite de - confinement , the color coulomb interaction between the constituents is strong and a large number of binary ( even color ) bound states , with a specific mass pattern , are formed.@xcite this phase we call strong coulomb phase ( scp ) .
the qgp phase occurs at a much higher temperature @xmath68 when the bound states dissolve .
let us describe some attempts to find signatures arising from qgp .
the basic idea is that during the hadronization process in this gluon rich environment the gluons combine due to the strong force into glueballs .
as the qgp cools further and transforms into hadronic matter , these glueballs decays into conventional hadrons giving rise to signatures of their existence .
it has been claimed that the existence of glueballs alters the @xmath69 ratio in the final state.@xcite it is conceivable that glueball production becomes dominant in central nuclear collisions since the existence of qgp provides a gluon rich environment especially at high energy density . in this scenario
the lowest mass glueballs are copiously produced .
particular decay modes @xmath70 and @xmath71 have been investigated@xcite and enhancements associated with possible glueball production observed .
search strategies , including dipion production have been also proposed.@xcite estimates within thermal models for the multiplicities of scalar glueballs in central au - au collisions at present and future experimentally available energies , _
i.e. _ from ags to lhc , have been presented.@xcite for the experimental identification of glueballs one can use the decay modes @xmath72 , @xmath73 and @xmath74 . despite of small branching ratio
, the @xmath75 channel has the important advantage that photons have practically no rescattering in the hadronic medium .
this analysis leads to maximal multiplicities for the glueball of @xmath76 of the @xmath15 meson multiplicity .
even larger yields are expected in the case of explosive hadronization of the quark gluon plasma.@xcite let us now turn to the second scenario . from its inception two decades ago , the high - t phase of qcd commonly known as the quark - gluon plasma,@xcite was described as a weakly interacting gas of `` quasiparticles''(quarks and gluons ) .
indeed , at very high temperatures asymptotic freedom causes the electric coupling to be small and the qgp to be weakly interacting . at intermediate temperatures of few times the critical temperature @xmath77 of immediate relevance to current experiments , there is new and growing evidence that the qgp is not weakly coupled . in this region qcd seems to be close to _ a strongly coupled coulomb regime _
, with an effective coupling constant @xmath78 0.5 - 1 and multiple _ bound states _ of quasiparticles.@xcite this phase we call strong coulomb phase .
this description is not universally accepted since some lattice calculations do not to find bound states above the transition.@xcite also , one must be aware that other explanations of the data have been presented.@xcite in order to clarify these issues it is interesting to find clear experimental observables that would shed some light into the discussion .
-0.0 cm 0.5 cm 0.5 cm kochelev and min consider the properties of scalar and pseudoscalar glueballs in scp.@xcite in an effective lagrangian approach , based on the low - energy qcd theorems , they find out that scalar glueball remains massive above deconfinement temperature ( see fig .
[ glueball ] ) . at the same time , pseudoscalar glueball changes its properties in qgp in a drastic way . indeed , this glueball becomes massless at @xmath79 and therefore it can contribute strongly to , the bulk properties of the plasma ( see fig . [ glueball ] ) .
they demonstrate that the disappearance of pseudoscalar glueball mass above the deconfinement temperature and its strong coupling to gluons gives the rise to the sign change of the gluon condensate in the pure @xmath80 gauge theory as observed in the lattice calculations at @xmath81 ( see fig .
[ glueball]).@xcite the strong nonperturbative coupling of the glueball to the gluons leads to the conjecture that one might expect that the role of very light pseudoscalar glueball in qgp must be quite similar to the role played by the massless pion in nuclear matter below deconfinement temperature . in scp despite de - confinement the color coulomb interaction between the constituents is strong and a large number of binary ( even color ) bound states , with a specific mass pattern , are formed.@xcite with this input , the scenario envisaged by vento for gluodynamics , _
i.e. _ the theory with only gluons goes as follows.@xcite the strong coulomb phase is crowded with gluon bound states and the lightest is the scalar glueball , labelled @xmath82 .
as one moves towards the dilution limit , the binding energy of these states decreases , the gluon mass increases , and therefore the color and singlet bound states increase their mass softly until the gluons are liberated forming a liquid.@xcite however , as the system cools towards the confining phase , color and singlet states decay into the conventional low lying glueballs , in particular @xmath82 .
thus the number of @xmath82 s becomes large .
this reasoning generalizes to qcd since in the scp the multiplicity of glueball channels is larger than in the confined phase .
the ratio of glueball to meson channels goes from 1 to 8 below the phase transition to 1 to 2 above.@xcite thus the number of scalar glueballs is much larger in scp than in the cold world . as the fireball cools a large number " of gluonic bound states decay by gluon emission into @xmath82 s .
the emitted gluons form new bound states of lower mass due to the strong color coulomb interaction .
as we approach the confinement region the mass of the color bound states increases and it pays off to make multiparticle color singlet states , which decay by rearrangement into ordinary color singlet states . since the coupling is strong and the phase space is large , these processes take place rapidly . thus in no time , close to the phase transition temperature @xmath67 , a large number of scalar glueballs populate the hadronic liquid .
0.5 cm 0.5 cm let us now analyze the experimental signal of these phenomenon .
if we assume that the @xmath83 is the o(4 ) partner of the @xmath84 in the chiral symmetry realization of qcd , its mass decreases when approaching the phase transition , becoming degenerate with the pion at @xmath67 ( see fig .
[ scp ] ) . beyond @xmath67 , in the scp , chiral symmetry is restored , and @xmath84 and @xmath83 remain degenerate for @xmath85 .
thus in the scp the @xmath83 can only decay in @xmath75 for obvious kinematical reasons .
the glueball @xmath82 does not vary its mass in this region appreciably .
thus even before we reach @xmath67 , the mixing between @xmath82 and @xmath83 disappears ( see fig .
[ scp ] ) and @xmath82 becomes stable around @xmath67 .
however , in the scp the mass of the @xmath83 increases and in a certain region of @xmath86 it again becomes degenerate with @xmath82 and mixing is restored .
thus the physical @xmath82 is able to decay , once the @xmath83 component is attained , only to @xmath75 .
the enhancement in the number of @xmath82s with respect to the hadronic phase arises because of the larger population of glueballs in the scp , as described above , and because these particles are stable in the medium against the dominating hadronic decays ( see fig .
[ scp ] ) .
thus a clear signal for the existence of a low mass scalar glueball and a confirmation of the scp scenario would be two @xmath75 peaks corresponding to @xmath82 and the @xmath83-meson as shown in fig .
[ fits ] .
0.5 cm the investigation on jets in the relativistic heavy ion collisions at rhic provides a deep insight into the properties of the quark - gluon plasma.@xcite one of the important rhic discoveries is the jet quenching phenomenon coming from the partonic energy loss in qgp . in the conventional approach to
the jet quenching the perturbative ( pqcd ) type of energy loss is taken into account by the channels , elastic and radiative , of one - gluon exchange between the jet and the massless gluons and quarks.@xcite however , the large quark - gluon rapidity density @xmath87 which is needed to describe the rhic jet quenching data within this approach , seems to be in contradiction with the restriction @xmath88 coming from the measured final entropy density @xmath89.@xcite furthermore , the lattice calculations show that even at very high temperature gluons and quarks still interact strongly in qgp .
@xcite -0.0
cm 0.5 cm as seen above , it has been suggested that the glueballs , the bound states of gluons , can exist above deconfinement temperature and may play an important role in the dynamic of strongly interacting qgp.@xcite in particular , it has been suggested that a very light pseudoscalar glueball can exist in qgp and might be responsible for the residual strong interaction between gluons.@xcite the lattice results showing a change of sign of the gluon condensate@xcite and a small value of the topological susceptibility above @xmath77 can be explained in the glueball picture as well .
furthermore , one expects that the suppression of the mixing between glueballs and quarkonium states in the qgp leads to a smaller width for the former as compared to the vacuum.@xcite this property opens the possibility for a clear separation of the glueball and the quark states in heavy ion collisions .
such separation is rather difficult in other hadron reactions due to existence of strong glueball - quarkonium mixing in the vacuum .
min and kochelev made an estimate of the energy loss induced by interaction of an energetic parton , which was produced in the hard scattering of two heavy ion s partons , with glueballs in the hot quark - gluon plasma.@xcite they showed that such contribution leads to a significant energy loss as can be seen from fig .
in particular , for the gluon jet such contribution is about a few gev / fm and approximately twice larger than the perturbative elastic loss.@xcite it should be pointed out that more than one half of contribution to the gluon energy loss comes from interaction of gluon with the light pseudoscalar glueball in qgp .
thus in spite of the fact that for the quark jet the glueball contribution is smaller than perturbative elastic loss , it can not be neglected in comparison with latter one . the existence of such light bound state of gluons above @xmath67 is crucial for the understanding of the large observed partonic energy loss in qgp .
therefore , not only pqcd type of energy loss but also glueball - induced loss , arising from existence of scalar and pseudoscalar glueballs in qgp , are important for the understanding of the rhic results such as the jet quenching .
it is well known that in the many high energy reactions with small momentum transfer the exchange by the highest - lying regge trajectory , called as a soft pomeron , gives the dominant contribution .
this exchange carries vacuum quantum numbers and has very peculiar properties in comparison with the usual regge pole trajectories as the @xmath90 , @xmath84 , _ etc_. from the analysis of the varions cross sections
it follows that the pomeron trajectory has a linear behavior @xmath91 the pomeron does not seem to be related to the usual mesons since the latter have usually lower intercepts and very different slopes . there has been a long - standing speculation that the physical particles on the trajectory might be glueballs .
meyer and teper investigated the pure gauge spectrum on a lattice to check its compatibility with a linear trajectory.@xcite they calculated the lightest states @xmath92 since the trajectory has an even signature .
their masses ( within errors bars ) are consistent with the pomeron trajectory
. these studies of quarkless qcd helps to understand the close relation between the pomeron and the pure glue states , but it is worth mentioning that the physical particles should probably mix with quarkonia .
the linearity of the trajectory , supported by these arguments , favors flux tube and stringy pictures as discussed in the constituent model section , at least for ( even)@xmath93 glueballs .
most of our review has been dedicated to the study of pure glueball states .
we have assumed in our discussion that glueballs are physical objects , which is fine for the purpose of the developments thus far .
but in order to establish contact with reality we have to discuss in some detail mixing , an idea which has been mentioned in passing on several occasions .
we first discuss the mathematics of mixing in a two state model and then apply the reasoning to the physical reality in the isosinglet scalar sector .
mixing arises when the total hamiltonian is not diagonal in the glueball and @xmath95 fock space . in this case
the physical states come out as superpositions of glueballs and @xmath96 states .
let us assume that we have a pure glueball , we label @xmath82 , and a @xmath95 state , we label @xmath83 , which have the same quantum numbers and , for simplicity , also the same mass @xmath97 , in a certain approximation .
we next relax this approximation and the hamiltonian in the reduced @xmath82 and @xmath83 fock space becomes @xmath98 the diagonal basis of this hamiltonian can be presented as , @xmath99 where the tilde labels the physical particles and @xmath100 is the mixing angle .
the masses of the physical particles become @xmath101 where @xmath102 and @xmath103 in fig .
[ 2statemixmass ] we represent the masses of the physical states as a function of the mixing angle @xmath100 .
the curves separate possible mass regions .
the curves show that the two state mixing scenario for positive @xmath104 leads to a light " glueball and heavier meson .
the opposite result can be obtained with a negative @xmath104 .
the procedure can be generalized to more states and to different initial masses as we next see .
this mechanism has been used to try to disentangle the complex isoscalar sector .
vento has shown that a low mass , almost invisible isosinglet scalar glueball , could hide in the low energy spectrum due to small mixing with the @xmath49 , and pointed out the possible existence of other glueballs components among the remaining isosinglet scalars @xmath61 , @xmath62 and @xmath63.@xcite close _ et al . _
have discussed the content of the the isosinglet scalar mesons @xmath61 , @xmath62 and @xmath63 in an attempt to discover which of them is dominantly a scalar glueball.@xcite these authors have suggested that @xmath62 is primarily a scalar glueball , due partly to the fact that @xmath62 , discovered in @xmath105 annihilation at lear , has decays to @xmath106 and @xmath107 which are relatively large compared to that of @xmath108@xcite and that the earlier quenched lattice calculations@xcite predict the scalar glueball mass to be around @xmath109 mev .
furthermore , because of the small production of @xmath108 in @xmath63 decay compared to that of @xmath110 , they claim that the @xmath63 is primarily @xmath111 dominated .
in contrast , the smaller production rate of @xmath110 relative to @xmath108 in @xmath61 decay leads to the conjecture that @xmath61 is governed by the non - strange light quark content . based on these observations ,
they have proposed a flavor - mixing scheme to consider the glueball and @xmath112 mixing in the neutral scalar mesons @xmath63 , @xmath62 and @xmath61.@xcite fits to the measured scalar meson masses and their branching ratios of strong decays have been performed in several@xcite leading to a mixing matrix of the form @xmath113 where @xmath114 and @xmath115 denote the quarkonium states @xmath116 and @xmath117 , and @xmath118 denotes the pure scalar glueball state .
thus for close _
the @xmath62 is composed primarily of a scalar glueball .
another analysis has been carried out by cheng _ et
al._.@xcite two lattice results are employed as the starting point ; one is the approximate su(3 ) symmetry in the scalar sector above 1 gev for the connected insertion part without @xmath112 annihilation,@xcite and the other is the scalar glueball mass at 1710 mev in the quenched approximation.@xcite in the su(3 ) symmetry limit , @xmath62 becomes a pure su(3 ) octet and is degenerate with @xmath119 , while @xmath61 is mainly an su(3 ) singlet with a slight mixing with the scalar glueball which is the primary component of @xmath63 .
these features remain essentially unchanged even when su(3 ) breaking is taken into account .
the observed enhancement of @xmath120 production over @xmath121 in hadronic j/@xmath45 decays and the copious @xmath63 production in radiative j/@xmath45 decays lend further support to the prominent glueball nature of @xmath63 .
furthermore , chiral suppression@xcite is advocated to obtain the following mixing matrix , @xmath122 therefore for cheng _
_ it is the @xmath63 is the particle composed mostly of a glueball state .
thus a definitive conclusion on this problem is still lacking . finally we would like to mention that the quarkonium - glueball mixing gives also a strong influence to the properties of the lowest mass pseudoscalar glueball.@xcite
the leitmotif of this review has been that the study of glueballs , states where the gauge field plays an important dynamical role , is important to understand the nonperturbative behavior of qcd .
this study requires precise and abundant experimental input , which despite a lot of effort has not achieved a level of understanding that allows a unique theoretical interpretation .
the lack of clarity arises from the fact that the theoretical developments are not able to determine in a well defined manner how the production and decay properties of glueballs are distinct from those of conventional mesons .
but the future is bright for several reasons .
new energy domains will be studied by besiii and in the future by @xmath123anda , and therefore the possibility of producing oddballs , glueballs with exotic quantum numbers will open . moreover , at low energy , both crystal barrel and in the future gluex will produce light mesons with a level of statistics , and precision for analysis , that will be able to separate exotic from non exotic behaviors and quark model nonets from particles extraneous to them . in this process
the new theoretical developments will be crucial to guide and/or interpret the results .
in this review we have discussed three major approaches , lattice qcd , qcd based constituent models and qcd sum rules , with a short excursion into the ads / qcd formalism , which at present leads ultimately to a constituent model type of description . due to the lack of observable states ,
the lattice qcd results have played the role of experimental data .
lattice qcd is a powerful technique , especially for the determination of masses .
the spectrum of the pure gluodynamics ( the pure gauge theory ) , equivalent to the so called quenched approximation of lattice qcd , is well established and the calculational errors are under control .
however the various treatments of quark loops ( unquenched lattice qcd ) and the effects of the mixing of glueballs with mesons are still a matter of debate and no firm conclusions can be drawn for the real world . moreover ,
important features of the study would be the determination of the decay properties of the various glueball candidates , and lattice qcd being a theory described in euclidean space - time does not possess the capability of describing asymptotic states and therefore of determining decay properties .
however , a full lattice qcd calculation with unlimited and precise calculational power should produce the experimental spectrum , if as we strongly believe qcd is the theory of the hadronic interactions , and moreover , if one can extend lattice qcd into minkowski space - time one would be able to determine the decay properties .
given the lack of experimental and detailed theoretical knowledge , models turn out to be an interesting laboratory to test ideas of the various perturbative and nonperturbative mechanisms in the theory .
we have discussed in some detail constituent gluon models for glueballs from different perspectives .
most of the pioneer work considered glueballs as strongly bound states of two ( or three ) heavy spin-1 gluons .
the comparison with lattice qcd shows that the spectrum obtained by these approaches does not correspond to lattice qcd .
however , if one implements a formalism dealing with only transverse gluons , the spurious states ( induced by the gluonic longitudinal components ) disappear and the hierarchy in the spectrum of lattice qcd is recovered .
these improvements have been developed up to now only in the two - gluon sectors and the generalization to negative @xmath124-parity is still lacking .
it remains for the future to find out which constituent glueball model confirms the lattice qcd spectrum for the higher states .
the dynamics in these models is described , in general , by a linear plus coulomb potentials .
this dynamics , however , does not remove the degeneracy of the scalar and pseudoscalar glueballs .
the candidate instanton induced interaction , which arises naturally in qcd and which has very particular properties under spin - parity , is the natural candidate to do the job . in the quenched approximation , the instanton induced interaction is attractive in the scalar and repulsive in the pseudoscalar channels , and equal in magnitude .
this is the precise behavior needed to lift the degeneracy in the constituent models .
the qcd sum rules offer the possibility to describe mixing and sea quarks effects in glueball states .
the method is rather technical and requires many ingredients ( condensates , instantons , topological charge screening , ... ) .
this complexity has led to many different calculations , with different results for the lowest lying states .
we refer the reader to the literature for the various attempts to describe the low mass glueball spectrum .
in here , we present the general formalism and some details of a specific calculation , in which the masses are obtained from pure gluonic currents , however , sea quarks effects are included ( _ via _ the quark condensate ) .
glueballs should be produced preferably in gluon - rich processes .
we described briefly the main examples : j/@xmath45 decays , central production @xmath125 annihilation,@xmath126 fusion and photoproduction .
the realization of a new generation of experiments , besiii , @xmath123anda and gluex , provide hope for new exciting developments in this field .
the ultra - relativistic heavy ion experiments is another experimental scenario where one expects that glueballs might play a role .
it is clear that the transition to a different phase of confinement , be it the strong coulomb phase or the quark gluon plasma , precludes that the confinement properties of the bound systems will change and with it their physical behavior .
there is no consensus on the glueball properties above the critical temperature and this has led to the description of several scenarios which can be very interesting from the point of view of observation .
rhic has produced large amount of date at high temperature , as will very soon do alice .
on the other hand fair will probe the high density region . given these circumstances
it is clear that the aim is to performed realistic analysis which are are able to disentangle the right from the wrong ideas .
finally , the strong expected mixing between glueballs and quark states leads to a broadening of the possible glueball states which does not simplify their isolation and their theoretical description .
the wishful sharp resonances which would confer the glueball spectra the beauty , richness and simplicity of the conventional baryonic and mesonic spectra are lacking . we hope that in the future and at higher energies the situation changes and we are able to have isolated exotic states .
it is , however , important to stress , that in any case gluebals are a beautiful and unique consequence of qcd .
the study of glueballs is intimately related to the quantitative understanding of confinement in qcd , since understanding confinement requires an understanding of the soft gluonic field responsible for binding the hadrons and the structure of the qcd vacuum .
we have emphasized that this has been the goal of the various studies , _ i.e. _ to get a clearer picture of how qcd behaves at relatively low momentum and how this behavior changes as the temperature or the density increase .
our review shows that we have learned much but we have yet not achieved the goal .
we foresee a bright future in this respect since a new experimental era of the study of qcd in many fronts is opening up . to find a description , which is able to describe all the various phenomena in a unique framework , is our task , and when achieved
will imply that we finally understand qcd .
the venture is even more exciting since maybe ads gravity might come to our help , a possibility unthinkable a few years back .
we would like to thank philippe de forcrand , alexander dorokhov , hilmar forkel , sergo gerasimov and dong - pil min for useful discussions .
vincent mathieu thanks the i.i.s.n .
( belgium ) for financial support .
nikolai kochelev was partly supported by belarus - jinr grant .
this work was done vicente vento was on a sabbatical from the university of valencia at the ph - th at cern , whose members he thanks for their hospitality .
vicente vento was supported by mecyt - fpa2007 and by mec - movilidad pr2007 - 0048 .
s. okubo , _ phys .
lett . _ * 5 * , 1975 ( 1963 ) ; g. zweig , in _ development in the quark theory of hadrons _ , edited by d.b .
lichtenberg and s.p .
rosen ( hadronic press , massachusetts , 1980 ) ; j. iizuka , _ prog .
suppl . _ * 37 * , 38 ( 1966 ) . c. amsler and f. e. close , _ phys .
b _ * 353 * ( 1995 ) 385 [ arxiv : hep - ph/9505219 ] .
m. s. chanowitz , _ int . j. mod
. phys . a _ * 21 * ( 2006 ) 5535 [ arxiv : hep - ph/0609217 ] .
f. giacosa , t. gutsche , v. e. lyubovitskij and a. faessler , _ phys . rev .
d _ * 72 * ( 2005 ) 114021 [ arxiv : hep - ph/0511171 ] .
c. j. morningstar and m. j. peardon , _ phys .
d _ * 60 * , 034509 ( 1999 ) [ arxiv : hep - lat/9901004 ] .
h. b. meyer and m. j. teper , _ phys .
b _ * 605 * ( 2005 ) 344 [ arxiv : hep - ph/0409183 ] .
h. b. meyer , arxiv : hep - lat/0508002 . y. chen _ et al .
_ , _ phys .
d _ * 73 * ( 2006 ) 014516 [ arxiv : hep - lat/0510074 ] .
b. lucini and m. teper , _ jhep _ * 0106 * , 050 ( 2001 ) [ arxiv : hep - lat/0103027 ] .
c. j. morningstar and m. j. peardon , _ phys .
d _ * 56 * ( 1997 ) 4043 [ arxiv : hep - lat/9704011 ] . c. mcneile and c. michael [ ukqcd collaboration ] , _ phys .
d _ * 63 * ( 2001 ) 114503 [ arxiv : hep - lat/0010019 ] .
w. a. bardeen , a. duncan , e. eichten , n. isgur and h. thacker , _ phys .
d _ * 65 * ( 2001 ) 014509 [ arxiv : hep - lat/0106008 ] . c. michael , _ nucl .
b _ * 327 * ( 1989 ) 515 .
a. szczepaniak , e. s. swanson , c. r. ji and s. r. cotanch , _ phys .
lett . _ * 76 * ( 1996 ) 2011 [ arxiv : hep - ph/9511422 ] .
a. p. szczepaniak and e. s. swanson , _ phys .
b _ * 577 * ( 2003 ) 61 [ arxiv : hep - ph/0308268 ] .
g. s. bali , _ phys .
d _ * 62 * ( 2000 ) 114503 [ arxiv : hep - lat/0006022 ]
. t. schafer and e. v. shuryak , _ phys .
_ * 75 * ( 1995 ) 1707 [ arxiv : hep - ph/9410372 ] .
a. b. kaidalov and yu .
a. simonov , _ phys .
_ * 63 * ( 2000 ) 1428 [ _ yad . fiz . _ * 63 * ( 2000 ) 1428 ] [ arxiv : hep - ph/9911291 ] . a. b. kaidalov and yu .
a. simonov , _ phys .
b _ * 636 * ( 2006 ) 101 [ arxiv : hep - ph/0512151 ] .
v. mathieu , c. semay and f. brau , _ eur .
j. a _ * 27 * ( 2006 ) 225 [ arxiv : hep - ph/0511210 ] .
m. cardoso and p. bicudo , arxiv:0807.1621 [ hep - lat ] .
r. c. brower , s. d. mathur and c. i. tan , _ nucl .
b _ * 587 * ( 2000 ) 249 [ arxiv : hep - th/0003115 ] . h. boschi - filho and n. r. f. braga , _ jhep _ * 0305 * ( 2003 ) 009 [ arxiv : hep - th/0212207 ] .
h. boschi - filho , n. r. f. braga and h. l. carrion , _ phys .
d _ * 73 * ( 2006 ) 047901 [ arxiv : hep - th/0507063 ] .
e. v. shuryak , _ nucl .
b _ * 203 * , 116 ( 1982 ) .
a. l. zhang and t. g. steele , _ nucl .
phys . a _ * 728 * , 165 ( 2003 ) [ arxiv : hep - ph/0304208 ] . s. narison , _ phys . rev .
d _ * 73 * , 114024 ( 2006 ) [ arxiv : hep - ph/0512256 ] .
j. i. latorre , s. narison and s. paban , _ phys .
b _ * 191 * , 437 ( 1987 ) .
g. hao , c. f. qiao and a. l. zhang , _ phys .
b _ * 642 * , 53 ( 2006 ) [ arxiv : hep - ph/0512214 ] .
l. j. reinders , h. rubinstein and s. yazaki , _ phys .
* 127 * ( 1985 ) 1 .
b. v. geshkenbein and b. l. ioffe , _ nucl . phys .
b _ * 166 * , 340 ( 1980 ) .
a. e. dorokhov and n. i. kochelev , _ z. phys .
c _ * 46 * , 281 ( 1990 ) .
t. schafer and e. v. shuryak , _ rev .
phys . _ * 70 * , 323 ( 1998 ) [ arxiv : hep - ph/9610451 ] .
h. j. lee , n. i. kochelev and v. vento , _ phys .
b _ * 610 * , 50 ( 2005 ) h. j. lee , n. i. kochelev and v. vento , _ phys .
d _ * 73 * , 014010 ( 2006 ) h. j. lee and n. i. kochelev , _ phys .
b _ * 642 * , 358 ( 2006 ) .
a. e. dorokhov , n. i. kochelev and yu .
a. zubov , _ int .
j. mod .
a _ * 8 * ( 1993 ) 603 .
d. diakonov , _ prog .
phys . _ * 51 * , 173 ( 2003 ) n. i. kochelev , _ phys .
nucl . _ * 36 * , 608 ( 2005 ) [ _ fiz .
yadra _ * 36 * , 1157 ( 2005 ) ] .
b. l. ioffe , _ prog .
phys . _ * 56 * , 232 ( 2006 ) .
m. delia , a. di giacomo and e. meggiolaro , _ phys .
b _ * 408 * ( 1997 ) 315 [ arxiv : hep - lat/9705032 ] .
a. a. migdal and m. a. shifman , _ phys .
b _ * 114 * ( 1982 ) 445 .
j. r. ellis and j. lanik , _ phys .
lett . b _ * 150 * ( 1985 ) 289 . g. b. west , arxiv : hep - ph/9608258 ; g.b .
west , _ phys .
_ * 77 * ( 1996 ) 2622 ; _ nucl .
* 54*a ( 1997 ) 353 . v. vento , _ phys .
d _ * 73 * ( 2006 ) 054006 [ arxiv : hep - ph/0401218 ] . c. a. dominguez and n. paver , _ z. phys .
c _ * 31 * ( 1986 ) 591 .
j. bordes , v. giminez and j. a. penarrocha , _ phys .
b _ * 223 * ( 1989 ) 251 .
l. s. kisslinger and m. b. johnson , _ phys .
b _ * 523 * ( 2001 ) 127 [ arxiv : hep - ph/0106158 ] . s. narison , _ nucl .
suppl . _
* 121 * ( 2003 ) 131 [ arxiv : hep - ph/0208081 ] . h. g. dosch and s. narison , _ nucl .
. suppl . _
* 121 * ( 2003 ) 114 [ arxiv : hep - ph/0208271 ] . h. forkel , _ phys .
d _ * 71 * , 054008 ( 2005 ) [ arxiv : hep - ph/0312049 ] .
p. pascual and r. tarrach , _ phys .
b _ * 113 * , 495 ( 1982 ) .
h. forkel , _ phys . rev .
d _ * 64 * , 034015 ( 2001 ) [ arxiv : hep - ph/0005004 ] .
d. harnett and t. g. steele , _ nucl .
phys . a _ * 695 * , 205 ( 2001 ) [ arxiv : hep - ph/0011044 ] .
n. kochelev and d. p. min , _ phys .
d _ * 72 * , 097502 ( 2005 ) m. c. tichy and p. faccioli , arxiv:0711.3829 [ hep - ph ] .
d. harnett , k. moats and t. g. steele , arxiv:0804.2195 [ hep - ph ] .
s. narison , _ nucl .
b _ * 509 * , 312 ( 1998 ) [ arxiv : hep - ph/9612457 ] .
s. narison , nucl .
a * 675 * ( 2000 ) 54c [ arxiv : hep - ph/9909470 ] .
p. r. page , arxiv : hep - ph/0107016 .
@xmath123anda collaboration : http:/www - panda.gsi.de/ .
d. robson , _ nucl .
b _ * 130 * ( 1977 ) 328 .
d. m. asner _
_ , arxiv:0809.1869 ; bes collaboration : http://bes.ihep.ac.cn/ . n. i. kochelev , arxiv : hep - ph/9902203 .
gluex collaboration : http://www.gluex.org/ .
a. a. natale , _ phys .
b _ * 362 * ( 1995 ) 177 [ arxiv : hep - ph/9509280 ] .
a. j. schramm , _
j. phys .
g _ * 25 * ( 1999 ) 1965 [ arxiv : hep - ph/9907217 ] .
m. bashkanov , _ aip conf .
* 619 * ( 2002 ) 525 .
m. chanowitz , _ phys .
_ * 95 * ( 2005 ) 172001 [ arxiv : hep - ph/0506125 ] . k. t. chao , x. g. he and j. p. ma , _ phys . rev .
* 98 * ( 2007 ) 149103 [ arxiv:0704.1061 [ hep - ph ] ] . m. s. chanowitz , _ phys .
* 98 * ( 2007 ) 149104 [ arxiv:0704.1616 [ hep - ph ] ] .
r. l. jaffe , _ phys .
d _ * 15 * ( 1977 ) 267 . c. amsler and f. e. close , _ phys .
d _ * 53 * ( 1996 ) 295 [ arxiv : hep - ph/9507326 ] .
a. abele _ et al . _ [ crystal barrel collaboration ] , _ eur .
j. c _ * 19 * ( 2001 ) 667 .
e. v. shuryak , _ phys .
* 61 * ( 1980 ) 71 .
phys . a 757 ( 2005 ) 1- 283 , _
first three years of operation of rhic .
_ e. v. shuryak and i. zahed , _ phys .
d _ * 70 * ( 2004 ) 054507 [ arxiv : hep - ph/0403127 ] . c. s. gao , x. q. li and w. lu , _ phys .
c _ * 52 * ( 1995 ) 421 .
s. kabana and p. minkowski , _ phys .
b _ * 472 * ( 2000 ) 155 [ arxiv : hep - ph/9907570 ] .
minkowski , s. kabana and w. ochs , arxiv : hep - ph/0011040 .
i. n. mishustin , l. m. satarov and w. greiner , _ j. phys .
g _ * 32 * ( 2006 ) l59 [ arxiv : hep - ph/0606251 ] . v. koch , a. majumder and j. randrup , _ phys .
* 95 * ( 2005 ) 182301 [ arxiv : nucl - th/0505052 ] . f. karsch , s. ejiri and k. redlich , _ nucl . phys .
a _ * 774 * ( 2006 ) 619 [ arxiv : hep - ph/0510126 ] . c. ratti , s. roessner , m. a. thaler and w. weise , _ eur
. phys .
j. c _ * 49 * ( 2007 ) 213 [ arxiv : hep - ph/0609218 ] .
n. kochelev and d. p. min , _ phys .
b _ * 650 * ( 2007 ) 239 [ arxiv : hep - ph/0611250 ] .
d. e. miller , _ phys .
* 443 * ( 2007 ) 55 [ arxiv : hep - ph/0608234 ] .
v. vento , _ phys .
d _ * 75 * ( 2007 ) 055012 [ arxiv : hep - ph/0609219 ] .
j. sollfrank and u. heinz , _ z. phys .
_ * c65 * ( 1995 ) 111 ; a. drago , m. gibilisco and c. ratti , _ nucl .
phys . _ * a742 * ( 2004 ) 165 ; b .- j .
schaefer , o. bohr and j. wambach , _ phys .
* d65 * ( 2002 ) 105008 .
s. wicks and m. gyulassy , _ j. phys .
g _ * 34 * ( 2007 ) s989 [ arxiv : nucl - th/0701088 ] .
b. muller and j. l. nagle , _ ann .
nucl . part .
* 56 * ( 2006 ) 93 [ arxiv : nucl - th/0602029 ] . b. alles , m. delia and a. di giacomo , _ nucl .
b _ * 494 * ( 1997 ) 281 [ erratum - ibid .
b * 679 * ( 2004 ) 397 ] [ arxiv : hep - lat/9605013 ] .
a. peshier , _ phys .
* 97 * ( 2006 ) 212301 [ arxiv : hep - ph/0605294 ] . c. amsler and f.e .
close , _ phys .
b _ * 353 * ( 1995 ) 385 ; f.e .
close and a. kirk , _ phys .
b _ * 483 * ( 2000 ) 345 .
f.e . close and q. zhao , _ phys .
d _ * 71 * ( 2005 ) 094022 . c. amsler
_ , _ phys .
b _ * 342 * ( 1995 ) 433 ; phys .
b * 340 * ( 1994 ) 259 .
c. michael and m. teper , _ nucl .
b _ * 314 * , 347 ( 1989 ) .
li , x. liu , and x.q .
zeng , _ phys .
d _ * 73 * ( 2006 ) 051502 ; _ ibid .
d * 73 * ( 2006 ) 114026 . h. y. cheng , c. k. chua and k. f. liu , _ phys .
d _ * 74 * ( 2006 ) 094005 [ arxiv : hep - ph/0607206 ] . | glueballs are particles whose valence degrees of freedom are gluons and therefore in their description the gauge field plays a dominant role .
we review recent results in the physics of glueballs with the aim set on phenomenology and discuss the possibility of finding them in conventional hadronic experiments and in the quark gluon plasma . in order to describe their properties
we resort to a variety of theoretical treatments which include , lattice qcd , constituent models , ads / qcd methods , and qcd sum rules .
the review is supposed to be an informed guide to the literature .
therefore , we do not discuss in detail technical developments but refer the reader to the appropriate references . |
following the ejection of half or more of their mass via isotropic , slowy expanding winds , agb stars evolve into planetary nebulae ( pne ) , which , surprisingly show a diverse range of aspherical ( e.g. , bipolar and multipolar ) morphologies ( e.g. , sahai & trauger 1998 , sahai , morris & villar 2011a ) . studies of pre - planetary nebulae ( ppns ) , objects in transition between the agb and planetary nebula ( pn ) evolutionary phases , are critical for characterising the physical processes responsible for this dramatic transformation . sahai & trauger ( 1998 ) proposed that fast collimated outflows or jets , operating during the ppn and/or very late agb phase , are the primary agents for the dramatic change in the mass - loss geometry and dynamics during the agb - to - pn evolutionary phase .
however , the physical mechanism for producing these fast outflows remains a mystery .
high - angular - resolution interferometric ( sub)millimeter - wave observations are the best way to quantitatively probe the fast outflow s dynamics and energetics crucial information for theoretical models ( e.g. , akashi & soker 2013 ) , and detailed numerical hydrodynamical simulations ( e.g. , lee & sahai 2003 , balick et al . 2013 ) , for pn shaping .
such observations have resulted in the discovery of a handful of extreme - outflow " ppns objects in which the molecular outflows reach speeds in excess of @xmath0100 , e.g. , boomerang nebula ( sahai et al . 2013 ) , iras22036 + 5306 ( sahai et al .
2006 ) , iras19374 + 2359 ( s '
anchez contreras et al .
2013 ) , iras16342 - 3814 ( imai et al . 2012 ) and hd101584 ( olofsson et al . 2015 ) .
detailed studies of such extreme objects are likely to provide the best motivation for , and most stringent tests of , theoretical models to explain their origin ( e.g. , blackman & lucchini 2014 ) . in this letter , we report ( sub)millimeter - line observations of the ppn iras08005 - 2356 ( i08005 ) , which clearly reveal it to be an extreme - outflow ppn .
early co j=21 observations by hu et al .
( 1994 ) resulted in the marginal detection of a weak , broad line . i08005
s f5ie central star v510pup ( slijkhuis et al . 1991 ) may have made a recent transition from ejecting oxygen - rich material to carbon - rich material ( bakker et al .
its morphological classification is bo*(0.55 ) ( sahai et al . 2007 ) , i.e. , it has a bipolar morphology ( resolved via hst imaging , ueta et al .
2000 ) with lobes open at their ends and a central star seen at 0.55 .
its optical spectrum reveals the presence of a prominent h@xmath5 emission line with very broad wings ( fwzi@xmath7 ) and a p - cygni type blue - shifted absorption feature ( s ' anchez contreras et al .
2008 : scetal08 ; also slijkhuis et al . 1991 , klochkova & chentsov 2004 ) . its estimated distance ranges between 2.85kpc ( oppenheimer et al . 2005 :
obs05 ) and 34kpc ( klochkova & chentsov 2004 ) ; we adopt a value of 3kpc .
the 1.3 mm and 0.87 mm interferometric observations were obtained with the submillimeter array ( sma ) at mauna kea , hawaii .
bandpass calibration was performed using observations of 3c279 . at 1.3 ( 0.87)mm ,
complex gain calibration was obtained from observations of the quasars 0750 + 125 and 0730 - 116 ( 0747 - 331 and 0826 - 225 ) , and flux calibration was obtained from observations of callisto ( europa ) .
additional observing parameters are listed in table[obs - parm ] .
sma data were calibrated using the mir - idl packagecqi / mircook.html ] , and images were made using the miriad software .
data cubes were obtained with a velocity resolution smoothed to 10 per channel ( to increase the s / n in each channel ) .
natural weighting was used to produce all images .
single - dish observations of the @xmath1co j=21 , 32 and 43 , @xmath2co j=21 , and sio ( v=0 ) j=54 and 65 line emission were obtained at smt during 2014 , nov / dec and 2015 jan .
telescope pointing was frequently checked on vycma , and is estimated to be better than a small fraction of the beam .
the weather was generally good , with system temparatures in the range @xmath8k at 1.3 mm and @xmath9k at 0.7 and 0.8 mm .
linear baselines were subtracted from the spectra shown . we have assumed main - beam efficiency factors of 0.76 , 0.66 , and 0.62 for the 230.6 , 345.8 , and 461.0ghz observations ( edwards et al . 2014 ) .
the optical images were taken with the nasa / esa hubble space telescope in go programs i d 6364 ( pi : m. bobrowsky ) and 6366 ( pi : s. trammell ) , using the pc on wfpc2 , and extracted from the hubble legacy archive .
i08005 shows an hourglass morphology in the hst f439w , f555w , and f675w images ( fig.[i08005hst ] ) the lobes , separated by a narrow dark lane , flare out from the central star s location , and attain a cylindrical shape .
the se lobe is significantly brighter than the nw one . a narrow , slightly - curved feature of linear extent 038 ,
is seen extending from the central star in the se lobe ( see inset ) .
the feature is oriented at a @xmath10 , slightly different from the pa of the nebula symmetry axis , @xmath11 .
the total linear extent of the nebula along its long axis , as seen in the f555w image , is @xmath12 ( @xmath13 cm ) .
we detected the co j=21 , 32 and sio j=54 lines with the sma .
the co j=32 and sio j=54 have significantly lower signal - to - noise , and the @xmath2co j=21 is tentatively detected .
the ( spatially ) integrated sma co j=21 line profile ( using a @xmath14 diameter circular aperture ) shows a very broad profile ( fig.[cosmt ] ) covering a total velocity extent of about 350 , i.e. , from @xmath15 .
the smt co j=21 line profile ( fig.[cosmt ] ) is similar to the sma one ( a narrow emission feature at @xmath16due to the presence of an unrelated line - of - sight interstellar cloud , has been removed from the profile by interpolation ) .
a gaussian fit to the profile gives a central velocity @xmath17 , and a fwhm of 196 .
we have divided the velocity range spanned by the co j=21 line into three intervals on each side of the central velocity the extreme high - velocity or ev component ( blue : @xmath18 , red : @xmath19 ) , a medium - velocity or mv component ( blue : @xmath20 , red : @xmath21 ) , and a low - velocity lv component ( blue : @xmath22 , red : @xmath23 ) .
the sma co j=21 emission in the blue and red - shifted parts of the ev , mv , and lv components show separations of about @xmath24 , @xmath25 , @xmath26 , respectively , along the nebular symmetry axis ( fig.[blured]a ) .
these results imply the presence of a fast , bipolar outflow with a ( roughly ) linear velocity - gradient , directed along the nebular symmetry axis , and a linear extent of @xmath27 .
since agb mass - loss is typically spherical , with outflow velocities of 1015 , we conclude that the outflow observed in co j=21 emission is a fast , collimated post - agb outflow , typical of the ppn evolutionary phase .
we made similar plots from the sma data cubes for co j=32 ( fig.[blured]b ) and sio j=54 ( not shown ) , dividing the emission velocity - range into two halves only ( blue : @xmath28 , red : @xmath29 ) due to the lower s / n in these lines , and found that these show a separation of @xmath30 roughly along the nebular axis .
this separation is equal to that derived from our co j=21 data for the same red and blue outflow - velocity ranges , implying that the sio outflow s linear extent is likely comparable to that of co. we also detected the co j=43 , j=32 and @xmath2co j=21 as well as the sio ( v=0 ) j=54 and 65 lines with the smt , but with lower s / n ( fig.[cosmt ] ) .
we have compared the total co j=21 line fluxes from our sma and smt data , as these have the highest s / n , and find that these are consistent within 15% , i.e. , within typical calibration uncertainties ) we conclude that the sma data do not suffer from any significant flux losses . continuum images were obtained after removing spectral regions containing line emission , using the miriad task uvlin . no continuum emission was detected from i08005 . using line - free channels , we find a @xmath31 noise of 1.05 ( 1.2 ) mjy / beam in the usb ( lsb ) at 1.3 mm and 11 ( 18 ) mjy / beam in the usb ( lsb ) at 0.87 mm .
we now determine the physical properties of the high - velocity bipolar outflow ( e.g. , scalar momentum , mass , and age ) from an analysis of the molecular - line data .
the observed ratios of the co j=21 , 32 and j=43 fluxes ( in k ) are @xmath32 , which imply corresponding non - beam diluted flux ratios of @xmath33 , since the source is unresolved by the smt beams of @xmath34 , @xmath35 , and @xmath36 in these lines .
we first make a simple emission model assuming a spherical outflow at a uniform excitation temperature equal to the kinetic temperature , @xmath37 , and carry out a least - squares fit to the line fluxes , varying @xmath37 , the outer radius ( @xmath38 ) , and mass of the emitting region ( @xmath39 ) .
we find @xmath40 and @xmath41k .
however , the derived @xmath37 is rather low compared to energy of the co j=4 level ( 55.4k ) , and modeling with the non - lte radex code ( van der taak et al .
2007 ) shows that the j=43 excitation temperature , @xmath42 , is lower than @xmath37 by about ( 2025)% due to the relatively low co line optical depths in our model ( @xmath43 ) and the average density , @xmath44@xmath45 , implied by the emitting region s mass and volume .
thus , the resulting , corrected model @xmath46 , too low to fit the data . using radex
, we find that the average density required to bring @xmath42 closer to @xmath37 in order to fit the observed source brightness temperature ratios , is @xmath47@xmath45 ; the required @xmath48k , giving a source size of @xmath49 , comparable to the co source - size estimated from the sma data .
the model source size is similar to the size derived from the hst image , implying that the co emission likely comes from the dense walls of the bipolar lobes seen in scattered light .
the co column density is , @xmath50@xmath51 ( and the j=21 , 32 , and 43 lines have optical depths of 0.48 , 0.44 , 0.20 ) . assuming that the average emitting column is equal to the source radius , we find a co - to - h@xmath52 abundance ratio of , @xmath53 , in reasonable agreement with the value typically assumed for ppns , @xmath54 ( e.g. , bujarrabal et al .
2001 ) . assuming a spherical emitting volume
, the mass is , @xmath55 .
the @xmath1co/@xmath2co abundance ratio is , f(@xmath1c/@xmath2c)=9.6 , from fitting the @xmath1co/@xmath2co j=21 line flux ratio corrected for the different beam - dilutions ( 8.1 ) .
the above values of @xmath39 and f(@xmath1c/@xmath2c ) are lower limits since this is our minimum - mass " model models with higher values of @xmath56 are allowed .
however , the total mass increases more slowly than the average density , since in models with higher values of @xmath56 , the emitting region s size is smaller . for example , if we use the dust mass , @xmath57=0.0019@xmath58 derived by obs05 from a detailed 2d radiative transfer model of i08005 s sed , scale it to d=3kpc , and adopt a typical gas - to - dust ratio for oxygen - rich agb stars , @xmath59 , we get @xmath60 . for this value of @xmath39 , we need a co model with @xmath61@xmath45 , a factor @xmath62 higher than in our minimum - mass model ; @xmath63k , and @xmath64@xmath51 . since the co j=21 optical depth is higher ( @xmath65 ) , the abundance ratio is higher , f(@xmath1c/@xmath2c)=17 .
we calculate the scalar momentum using the formulation described in bujarrabal et al .
( 2001 ) . using our minimum - mass model , we find @xmath66gcms@xmath67 for an inclination angle of the nebular axis to the sky - plane , @xmath68 ( obs05 ) .
the kinetic energy in the outflow is @xmath69erg .
these values of @xmath70 and @xmath71 lie near the upper end of the range for ppns , @xmath72 gcms@xmath67 and @xmath73erg ( bujarrabal et al .
2001 ) . this outflow can not be driven by radiation pressure because the lobes dynamical ( expansion ) time - scale , @xmath74yr ( from dividing the model co shell - size by the co j=2 - 1 fwhm line - width ) is much smaller than that required by radiation pressure to accelerate the observed bipolar outflow to its current speed , @xmath75 yr , given i08005 s luminosity of @xmath76 at d=3kpc ( using the value derived by obs05 , @xmath77 . ) the mass - loss rate in the outflow is @xmath78 .
the observed smt sio j=65 to 54 line flux ratio is @xmath79 ( with about @xmath80% uncertainty ) , implying an intrinsic flux ratio ( i.e. , corrected for beam - dilution ) of @xmath81 , since the source is unresolved by the smt beams of @xmath82 and @xmath83 in these lines . using radex
, we find that for @xmath84k , the average @xmath85 density in the sio emission region is , @xmath86@xmath45 in order to fit the observed value of @xmath87 .
the fractional sio abundance is @xmath88 , close to the maximum value assuming cosmic abundances ( @xmath89 ) , and comparable to the observed maximum circumstellar sio abundance in oxygen - rich agb stars ( @xmath90 : sahai & bieging 1993 ) .
if the sio emission comes from shocked gas at a higher kinetic temperature than that for co , the minimum density required is higher .
evidence of very fast outflows in i08005 comes from three independent probes : ( i ) co data ( presented here ) , ( ii ) h@xmath5 spectrum in scetal08 , and ( iii ) oh maser emission ( vla mapping ) reported in zijlstra et al .
the oh maser features cover the velocity range of @xmath91 , i.e. , roughly the range covered by red half of the co j=21 profile .
in contrast to oh and co emission , the h@xmath5 absorption probes atomic gas .
since the h@xmath5 absorption feature is not spatially resolved in the ground - based long - slit spectra , it is difficult to directly establish its relationship to the fast molecular outflow , however the close agreement between the terminal outflow velocities derived for these indicates that they are closely associated .
the atomic gas may be located inside the lobes and may constitute unshocked material of an underlying jet , and/or the interface between the latter and the lobe walls , as proposed for iras22036 + 5306 ( scetal06 ) .
the jet - like feature seen in the hst image ( fig.[i08005hst],inset ) may represent the precessing jet s signature close to its launch - site .
the high - speed bipolar outflow in i08005 may be driven by an accretion disk .
bakker et al .
( 1997 ) find numerous narrow , double - peaked , chromospheric emission lines from neutral and singly ionized metals , and propose that these might arise in an accretion disk .
we conjecture that the broad h@xmath5 wings seen towards this object might arise as a result of raman - scattering of ly@xmath6 emission generated by such a disk a process that produces an h@xmath5 profile with a width that is a factor 6.4 larger than the ly@xmath6 width and a @xmath92 wing profile .
the very wide ( @xmath93 ) h@xmath5 line wings in i08005 show this shape , furthermore a weak emission feature around 6830 with @xmath94 , corresponding to raman - scattering of the 1032 component of the ovi doublet at @xmath95 is also seen ( scetal08 ) ; the 7088 feature that corresponds to the 1038 component is a factor 4 weaker , and too weak to be visible .
other line - broadening mechanisms include electron scattering and emission from a rotating disk .
arrieta & torres - peimbert ( 2003 ) find that electron scattering requires extreme densities ( e.g. , @xmath96 @xmath45 in m2 - 9 ) making that an implausible mechanism .
keplerian rotation in a disk around the central star is too low to directly account for the broad line - width in i08005 ( and ppne in general : sahai et al .
2011b ) taking the radius for the central post - agb star of i08005 to be @xmath97 ( slijkhuis et al . 1991 ) and a nominal stellar mass of 1 , we find the maximum rotation speed is @xmath98 , too low for generating the extreme line - wings of the h@xmath5 profile , even with the factor 6.4 increase provided by raman - scattering .
we therefore conclude that the accretion disk in i08005 must be around a much smaller star , e.g. , a main - sequence companion .
the raman - scattering likely occurs in the innermost regions of the fast neutral wind seen via its blue - shifted absorption feature signature in the h@xmath5 profile . in this case , following sahai et al .
( 2011b ) , we scale from the minimum scattering column density , @xmath99 @xmath51 needed to achieve the raman conversion efficiency necessary for producing the very broad observed line - widths ( see fig.1 of lee & hyung 2000 ) , to find that @xmath100(@xmath101au ) ( d/3kpc ) ( @xmath102/100 ) ( @xmath103@xmath51 ) , where @xmath104 is the radius of the neutral wind where the raman - scattering occurs .
@xmath102 is set equal to @xmath105 of the broad co j=21 line profile , and similar to the neutral outflow s average outflow - velocity derived by scetal08 .
the mass - loss rate requirement is easily met in i08005 since obs05 derive a mass - loss rate of @xmath106 for a collimated fast wind in the 24 - 780au region around the central star ,
i08005 is very similar to other extreme - outflow oxygen - rich ppns , iras22036 + 5306 ( sahai et al .
2006 ) and hd101584 ( olofsson et al . 2015 , sivarani et al . 1999 ) in the properties of its collimated fast outflow ( expansion - velocity , scalar - momentum , and kinetic - energy ) , and its h@xmath5 emission profile ( very broad wings , p - cygni absorption ) .
but in striking contrast to these objects , it appears to lack mm and submm continuum emission from its central region
the ratio of its 1.3 mm to 60micron flux is @xmath107 ( @xmath108 ) , compared to @xmath109 and @xmath110 for hd101584 and iras22036 + 5306 , respectively .
both ppns , as well as the disk - prominent " sub - class of young post - agb objects ( dpagb objects : sahai et al .
2011c ) , which unlike ppns , show little or no extended nebulosity , and have central stars that are radial - velocity binaries ( e.g. , van winckel et al . 2008 ) , emit relatively strong mm / submm continuum emission from their central regions .
this emission has been attributed to the presence of substantial masses of cool , mm - sized grains ( sahai et al .
2011c , gielen et al .
2007 , de ruyter et al . 2006 ) .
their origin is not understood at present but is potentially a key probe of important mass - ejection processes occurring during the late - agb and post - agb evolutionary phases , especially those that lead to the formation of large dusty equatorial disks or torii .
for example , in hd101584 , where a binary companion has been found from radial - velocity variations ( bakker et al .
1997 ) , olofsson et al .
( 2015 ) propose a scenarion in which both the formation of the central , equatorially - dense mass structure and the collimated outflow result from a common - envelope ( ce ) event but the latter does not release enough energy to drive the mass - ejection , and another mechanism augments or even dominates it .
the apparent lack of a substantial mass of material in the equatorial waist of i08005 suggests that perhaps it did not undergo a ce event , and its collimated outflow ( and collimated outflows in ppns generally ) may be launched differently , possibly from an accretion disk as we have suggested earlier .
i08005 is thus a key post - agb object for further detailed study .
for example , alma can be used to probe its compact central region , in order to search for weak mm / submm continuum emission that may be present but was below our sensitivity limit , and for the presence of gas ( and its kinematics ) associated with its dusty waist .
we thank ken young and ray blundell for granting filler time on the sma .
r.s . is thankful for partial financial support for this work from a nasa/ adp and ltsa grant .
r.s.s contribution to this paper was carried out at jpl , california institute of technology , under a contract with nasa .
akashi , m. , & soker , n. 2013 , , 436 , 1961 arrieta , a. , & torres - peimbert , s. 2003 , , 147 , 97 bakker , e. j. , van dishoeck , e. f. , waters , l. b. f. m. , & schoenmaker , t. 1997 , , 323 , 469 balick , b. , huarte - espinosa , m. , frank , a. , et al .
2013 , , 772 , 20 blackman , e. g. , & lucchini , s. 2014 , , 440 , l16 bujarrabal , v.,castro - carrizo , a. , alcolea , j. , & s ' anchez contreras , c. 2001 , , 377 , 868 de ruyter , s. , van winckel , h. , maas , t. et al .
2006 , a&a , 448 , 641 edwards , s. et al .
1987 , , 321 , 473 edwards , j. l. , cox , e. g. , & ziurys , l. m. 2014 , , 791 , 79 gielen , c. , van winckel , h. , waters , l. b. f. m. , min , m. , & dominik , c. 2007 , , 475 , 629 hu , j. y. , te lintel hekkert , p. , slijkhuis , f. , et al .
1994 , , 103 , 301 imai , h. , chong , s. n. , he , j .- h . , et al .
2012 , , 64 , 98 klochkova , v. g. , & chentsov , e. l. 2004 , astronomy reports , 48 , 301 lee , h .- w .
, & hyung , s. 2000 , , 530 , l49 lee , c .- f . , & sahai , r. 2003 , , 586 , 319 olofsson , h. , vlemmings , w. h. t. , maercker , m. , et al .
2015 , , 576 , l15 olofsson , h. , carlstrom , u. , eriksson , k. , gustafsson , b. , & willson , l. a. 1990 , , 230 , l13 oppenheimer , b. d. , bieging , j. h.,schmidt , g. d. , gordon , k. d. , misselt , k. a. , & smith , p. s. 2005 , apj , 624 , 957 sahai , r. , & bieging , j. h. 1993 , , 105 , 595 sahai , r. , claussen , m. j. , schnee , s. , morris , m. r. , & snchez contreras , c. 2011c , , 739 , l3 sahai , r. , morris , m. r. , & villar , g. g. 2011a , , 141 , 134 sahai , r. , snchez contreras , c. , morris , m. r. , & claussen , m. 2011b , asymmetric planetary nebulae 5 conference , 11 sahai , r. , morris , m. , snchez contreras , c. , & claussen , m. 2007 , , 134 , 2200 sahai , r. & trauger , j.t .
1998 , aj , 116 , 1357 sahai , r. , vlemmings , w. h. t. , huggins , p. j. , nyman , l .- .
, & gonidakis , i. 2013 , , 777 , 92 sahai , r. , young , k. , patel , n. a. , snchez contreras , c. , & morris , m. 2006 , , 653 , 1241 sivarani , t. , parthasarathy , m. , garca - lario , p. , manchado , a. , & pottasch , s. r. 1999 , , 137 , 505 skopal , a. 2006 , , 457 , 1003 slijkhuis , s. , de jong , t. , & hu , j. y. 1991 , , 248 , 547 snchez contreras , c. , martin , s. , & sahai , r. 2013 , _ highlights of spanish astrophysics vii _ , eds . : j.c .
guirado , l.m .
lara , v. quilis , and j. gorgas .
, 671 snchez contreras , c. , sahai , r. , gil de paz , a. , & goodrich , r. 2008 , , 179 , 166 van der tak , f.f.s . , black , j.h . , schoier , f.l . , jansen , d.j . ,
van dishoeck , e.f . , 2007 ,
a&a 468 , 627 ueta , t. , meixner , m. , & bobrowsky , m. 2000 , , 528 , 861 van winckel , h. , deroo , p. , gielen , c. , et al.2008 , evolution and nucleosynthesis in agb stars , aip conference proceedings , 1001 , 349 zijlstra , a. a. , chapman , j. m. , te lintel hekkert , p. , likkel , l. , comeron , f. , norris , r. p. , molster , f. j. , & cohen , r. j. 2001 , mnras , 322 , 280 llllllllll lines & freq.range & @xmath111 & array ( # ants . ) & time & beam , pa & t@xmath112 & epoch + & [ ghz ] & [ khz ] & & [ hr ] & @xmath113 & [ k ] & yy / mm / dd + @xmath1co(2 - 1 ) & usb : 228.85232.85 & 812.5 & compact(6 ) & 1 & @xmath114,20.8 & 100230 & 13/12/11 + continuum & 228.9229.6,229.9230.2 & ... & ... & ... & ... & ... & ... + & 230.8232.9 & ... & ... & ... & ... & ... & ... + + @xmath2co , c@xmath115o(2 - 1 ) & lsb : 216.85220.85 & ... & ... & ... & ... & ... & ... + sio(5 - 4),v=0 & & & & & & & + continuum & 216.9217.0,217.3219.4 & ... & ... & ... & ... & ... & ... + @xmath1co(3 - 2 ) & lsb : 342.0346.0 & 812.5 & extended(7 ) & 2 & @xmath116,33.5 & 240400 & 14/04/13 + continuum & 342.0345.4 & ... & ... & ... & ... & ... & ... + + continuum & usb : 354.0358.0 & ... & ... & ... & ... & ... & ... + -2 in .
the diagonal linear features are diffraction spikes due to the bright central source in the relatively long - exposure f555w image .
inset is a false - color view ( linear stretch : colorbar shows counts / s from 0 to 210 ) of the central @xmath117 region ( dashed box ) in f439w , highlighting the curved jet - like feature emanating from the central source .
, title="fig:",width=755 ] -3 in axis ) have been shifted vertically for clarity , after scaling as follows : sio j=65 and j=54(@xmath118 ) , @xmath1co j=43 and 21(@xmath119 ) , & @xmath2co j=2 - 1(@xmath120 ) .
the velocity ranges of the hv , mv , & lv components are marked in the sma panel .
the sma @xmath1co j=21 spectrum , shifted below @xmath121 , shows the total flux divided by the smt conversion factor of 44.5 jy / k , then scaled @xmath119 , for comparison with the smt spectrum . the apparent difference in the smt and sma co j=2 - 1 line - profiles at
@xmath122 is probably due to non - linear baseline structure that is not removed by the linear baseline subtraction .
, title="fig:",width=680 ] -0.7 in ) ranges ( a ) j=21 , ( _ top _ ) blue : @xmath123 , red : @xmath124 ; ( _ middle _ ) blue : @xmath125 , red : @xmath126 ; ( _ bottom _ ) blue : @xmath127 , red : @xmath128 .
( b ) j=32 , blue : @xmath129 , red : @xmath130 .
contour levels for co j=21 ( 32 ) are : minimum=50 ( 46)% , step=10 ( 15)% of the peak intensities . the central source in the hst image ( marked with a cross ) , is offset ( 0041,04 ) relative to the phase - center ( at 0,0 with j2000 coordinates ra=08:02:40.7 , dec=-24:04:43.0 ) , and consistent with the bipolar - outflow center , within the combined 1-@xmath131 uncertainty , @xmath0(@xmath13204,@xmath13204 ) , of the absolute hst and smt astrometry.,title="fig:",width=680 ] | we report interferometric mapping of the bipolar pre - planetary nebula iras08005 - 2356 ( i08005 ) with an angular - resolution of @xmath015 , using the submillimeter array ( sma ) , in the @xmath1co j=21 , 32 , @xmath2co j=21 and sio j=54 ( v=0 ) lines .
single - dish observations , using the smt 10-m , were made in these lines as well as in the co j=43 and sio j-65 ( v=0 ) lines .
the lines profiles are very broad , showing the presence of a massive ( @xmath3 ) , extreme high - velocity outflow ( @xmath4 ) directed along the nebular symmetry axis derived from the hst imaging of this object .
the outflow s scalar momentum far exceeds that available from radiation pressure of the central post - agb star , and it may be launched from an accretion disk around a main - sequence companion .
we provide indirect evidence for such a disk from its previously published , broad h@xmath5 emission profile , which we propose results from ly@xmath6 emission generated in the disk followed by raman - scattering in the innermost regions of a fast , neutral wind . |
chiral perturbation theory ( @xmath2pt ) is a very powerful tool to describe various @xmath3 decays in which long distance contributions are expected to dominate . for example , the decay mode @xmath4 has been observed recently , and compared with @xmath2pt @xcite .
the measured branching ratio and @xmath5 momentum spectrum are found to be consistent with the predictions , after fitting one free parameter contained in the theory .
the neutral counterpart , @xmath6 , is another decay mode , where detailed studies have been performed . in this case ,
the lowest order @xmath7 calculation @xcite does not reproduce the measured branching ratio @xcite . extending the calculation to the next - to - leading order @xmath8 @xcite , and adding a vector meson contribution @xcite
, the prediction is now in good agreement with the branching ratio as well as distinct @xmath9 spectrum .
we note that an effective coupling constant ( @xmath10 ) , a free parameter in the vector meson contribution , has been determined by the measurements with experimental errors @xcite .
the as yet unobserved decay mode @xmath11 can provide another testing ground for @xmath2pt . since theoretical ingredients are same for both @xmath6 and @xmath11 modes , a straightforward extension from @xmath6 provides definite prediction for a branching ratio and @xmath12 spectrum @xcite .
in particular , the branching ratio is calculated to be @xmath13 .
thus experimental study of this mode is important to test the theoretical framework of @xmath2pt . other interests in this decay mode stem from its close relationship to the mode @xmath14 , which has been of much attention as a possible channel to observe direct @xmath15-violation .
first , @xmath11 is expected to have a much larger branching ratio than @xmath14 , and hence can be an experimental background in a soft photon region .
second , there also exists a @xmath15-conserving amplitude in @xmath14 via two - photon intermediate states ; this can be in principle determined from a detail analysis of @xmath6 @xcite .
further understanding of the @xmath6 amplitude , which can be checked by the @xmath11 mode , is thus essential . in this article ,
we report on an experimental search for the decay mode @xmath11 conducted with a proton synchrotron at high energy accelerator research organization ( kek ) .
the data for the @xmath11 mode were recorded simultaneously with the experiment which has established a new decay mode @xmath16 @xcite . since the experimental set - up was described already in ref .
@xcite , it is briefly mentioned here for convenience .
the @xmath17 beam was produced by focusing 12-gev / c primary protons onto a 60-mm - long copper target .
its divergence was @xmath184 mrad horizontally and @xmath1820 mrad vertically after a series of collimators embedded in magnets .
the set - up started with a 4-m - long decay volume .
it was followed by a charged particle spectrometer consisting of four sets of drift chambers and an analyzing magnet with an average horizontal momentum kick of 136 mev / c .
a threshold cherenkov counter ( gc ) with pure n@xmath19 gas at 1 atm was placed inside of the magnet gap to identify electrons .
for the present decay mode , we obtained an average electron efficiency of 94% with a pion - rejection factor of 350 by adjusting software cuts in the off - line analysis . a pure csi electromagnetic calorimeter played a crucial role in this analysis .
it was located at the downstream end of the spectrometer , and consisted of 540 crystal blocks , each being 70 mm by 70 mm in cross section and 300 mm ( @xmath20 16x@xmath21 ) in length .
it was configured into two banks of 15 ( horizontal ) @xmath22 18 ( vertical ) matrix .
its energy and position resolutions were found to be approximately 3% and 7 mm for 1-gev electrons , respectively .
the trigger for the present mode was designed to select events with @xmath23 2 electron - like tracks and @xmath23 3 cluster candidates in the calorimeter .
it was formed with information from gc and csi together with trigger scintillator hodoscopes interspersed between the chambers and calorimeter .
in reconstructing @xmath11 events , major backgrounds are expected come from both @xmath24 and @xmath25 modes , where @xmath26 denotes the dalitz decay @xmath27 . among them
, @xmath24 is an intrinsic background and can not be removed .
it is instead used as a normalization mode by positively identifying the dalitz decay .
the @xmath25 mode may become background when two ( or more ) photons fuse into one in the calorimeter , and/or its vertex is reconstructed incorrectly to give false invariant masses .
care was taken in this analysis to enhance position resolution of decay vertex and purity of photon clusters .
this effort was found useful also to reject backgrounds originating from external conversions and hadron ( mostly neutron ) interactions . in the actual off - line analysis
, we first selected events containing two tracks with a common vertex in the beam region inside the decay volume .
we then requested events to have @xmath23 5 clusters in the calorimeter . here
the cluster was defined as @xmath28 csi blocks around the local maximum with the total energy deposit greater than 200 mev ( @xmath18 3.5 nsec timing window ) .
particle species were then determined . a charged track which could project onto a csi cluster
was called a matched track .
an electron ( or positron ) was identified as a matched track with 0.9 @xmath29 e / p @xmath29 1.1 , where e was an energy measured by the calorimeter and p was a momentum determined by the spectrometer , respectively .
also gc hits in the corresponding cells were requested .
clusters which did not match with any charged tracks were treated as photon candidates .
event topology was then checked .
we requested events to contain exactly one @xmath30-pair and three photon candidates , consistent with the @xmath11 topology .
no additional activities , such as an extra track , gc hit , or cluster with an energy above @xmath2060 mev , was allowed in the detector track segment upstream of the magnet could be projected onto a neutral cluster with an energy below 400 mev , we simply ignored this activity . ] .
the probability of over - veto was estimated using @xmath24 and @xmath31 reconstructed events , and was found to be about 12% ( common to both @xmath11 and @xmath24 modes ) .
two quality cuts were imposed at this stage .
one was a cluster shape cut ; it examined mainly a cluster s transverse energy profile , and checked whether or not it was consistent with a single photon .
we used a large sample of @xmath32 reconstructed events to characterize actual photon showers , and found the cut efficiency to be 95% per a single photon cluster .
the other quality cut was imposed on an @xmath33-track opening angle ( @xmath34 ) .
larger opening angle resulted generally in better vertex position resolution , which in turn led to better invariant mass resolution and background rejection .
we employed a monte carlo simulation to study background rejection power , especially for the @xmath25 mode , and determined to demand @xmath35 20 mrad .
as a final step in the general event selection , we imposed two loose kinematical cuts to reduce sample size . they were @xmath36 400 mev / c@xmath37 and @xmath38 100 mrad@xmath39 , where @xmath40 represents the angle of the reconstructed @xmath41 momentum with respect to the line connecting the target and vertex .
having selected candidate events , we scrutinized each event from the viewpoint of kinematical variables .
first of all , we calculated @xmath42 invariant mass for three possible combinations .
the solid - line histogram shown in fig.[fig : gamma - gamma ] is the distribution for all combinations while the shaded one for the combination closest to @xmath43 .
we retained all combinations which satisfied @xmath44 2.5@xmath45 background events , estimated by a monte carlo simulation , became less than one inside the final signal box ( see below ) .
] , where @xmath45 being the observed mass resolution of 5.1 mev / c@xmath37 .
we then calculated the angle ( @xmath46 ) between reconstructed @xmath47 and @xmath48 momentum vectors in the @xmath17 rest frame with an observed @xmath49 energy . ] .
events originating from @xmath11 or @xmath24 should satisfy @xmath50 in this frame .
the dots with error bars in fig.[fig : cos - star ] show the @xmath51 distribution ; a clear peak of events at @xmath50 can be seen .
the histogram in fig.[fig : cos - star ] is monte carlo data for the @xmath25 mode , in which the flux was normalized by the observed @xmath24 events ( see below ) . to select signals
, we requested events to satisfy a collinearity cut , @xmath52 , as shown by the arrow in fig.[fig : cos - star ] .
we now identify @xmath24 events . if more than one @xmath42 combination within an event satisfied the @xmath53 cut , we selected the one for which the quantity , @xmath54 became minimum . here
@xmath55 is the observed @xmath47 mass resolution for the dalitz decay mode ( see below for the actual value ) .
fig.[fig : p0p0d](a ) shows a scatter plot of @xmath56 vs @xmath57 after selecting the @xmath42 combination with @xmath58 . to select @xmath24 events further
, @xmath38 30 mrad@xmath39 was demanded .
fig.[fig : p0p0d](b ) shows a scatter plot of @xmath56 vs @xmath59 after this cut .
a clear cluster of @xmath24 events can be seen in the expected region of @xmath60 and @xmath61 .
fig.[fig : p0p0d](c ) is the projection onto the @xmath59 axis , and fig.[fig : p0p0d](d ) is onto the @xmath56 axis with a @xmath62 mass cut ( see below ) . from these projections , we found @xmath62 and @xmath17 mass resolutions to be @xmath63 4.5 mev / c@xmath37 and @xmath64 16 mev / c@xmath37 , respectively .
our final signal box , shown by the rectangle in fig.[fig : p0p0d](b ) , was defined by @xmath653@xmath66 and @xmath67 3@xmath68 .
after all the cuts , 49 events remained .
we estimated the number of @xmath25 backgrounds in the signal region by a monte carlo simulation , and found to be less than one .
we are now in a position to look for the @xmath11 mode .
first we rejected @xmath24 events ; if an event satisfied both @xmath69 5@xmath45 and @xmath70 5@xmath66 for any @xmath71 combinations , then the event was discarded .
note that we employed the looser kinematical cut of 5@xmath45(@xmath66 ) to exclude possible @xmath24 events .
then we rejected @xmath25 backgrounds . in this case , an event containing @xmath62 whose transverse momentum was consistent with @xmath25 ( p@xmath72 139 mev / c ) was excluded .
the remaining events are shown in a scatter plot of @xmath56 vs @xmath57 in fig.[fig : p0p0p0d](a ) .
we are left with no events inside our signal box defined by @xmath38 30 mrad@xmath37 and @xmath67 3@xmath73 was 11.6 mev / c@xmath37 .
this was the value of the monte carlo mass resolution ( 8 mev / c@xmath37 ) for @xmath11 times the ratio of the observed ( @xmath74 mev / c@xmath37 ) to monte carlo ( 11 mev / c@xmath37 ) resolutions for @xmath24 . ] .
the background events still remaining in the low mass region of @xmath75 460 mev / c@xmath37 were found to originate mostly from the @xmath25 mode .
the projection of events with @xmath38 30 mrad@xmath37 onto the @xmath56 axis is shown in fig.[fig : p0p0p0d](b ) .
the solid - line histogram is for the data , the shaded one for monte carlo events ( sum of @xmath24 and @xmath25 ) and the dotted one for @xmath11 monte carlo events . from the monte carlo simulation ,
whose flux was normalized by the observed @xmath24 signals , we expected 1.1 backgrounds ( 0.45 from @xmath24 and 0.66 from @xmath25 ) to remain in the signal box .
the branching ratio is calculated by @xmath76 where @xmath77 , @xmath78 and @xmath79 denote acceptance , efficiency and observed number of events , respectively .
the detector acceptances depend on the matrix elements : we employed a theoretical model given by ref .
@xcite for @xmath11 and kroll - wada spectrum for the dalitz decay @xcite . the actual acceptances ( including the hardware trigger conditions ) were determined by monte carlo simulations and were found to be @xmath80 for @xmath11 and @xmath81 for @xmath24 , respectively .
the detector efficiency was found to be practically same for both modes . however , some of the quality and kinematical cuts caused efficiency differences . since the opening angle of @xmath30-tracks was different for the two modes , it caused efficiency difference in both the vertex reconstruction and the track opening angle cut .
these efficiencies were found to be 70% ( vertex ) and 89% ( @xmath34 ) for @xmath11 mode while 62% and 81% for @xmath24 mode , respectively .
next , the collinearity ( @xmath82 ) cut produced 25% inefficiency for the signal mode while negligibly small loss for the normalization mode .
the @xmath62 mass cut was unique to @xmath24 ; its efficiency was found to be 96% ( 3@xmath83 cut ) . for the signal mode we rejected @xmath24 events ; it caused 9% inefficiency at the specific @xmath84 invariant mass region .
we also rejected @xmath25 events with the @xmath85 inclusive veto ; it resulted in 7% inefficiency . combining other efficiencies together , we found the final acceptance and efficiency ratio @xmath86 to be 0.670 . inserting the known branching ratios @xcite into @xmath87 ,
the single event sensitivity was obtained to be @xmath88 , where the error represents the statistical uncertainty .
the upper limit on the branching ratio was determined to be @xmath89 in which the statistical error of the normalization mode was taken into account @xcite . in summary
, we performed an experimental search for the @xmath11 mode .
we observed no events and set a 90% confidence level upper limit of @xmath1 for its branching ratio .
this is the first published experimental result on this decay mode .
we wish to thank professors h. sugawara , s. yamada , s. iwata , k. nakai , and k. nakamura for their support and encouragement .
we also acknowledge the support from the operating crew of the proton synchrotron , the members of beam channel group , computing center and mechanical engineering center at kek .
, y.m and m.s acknowledge receipt of research fellowships of the japan society for the promotion of science for young scientists . 99 p. kitching _ et al .
lett . * 79 * 4079 ( 1997 ) .
g. ecker , a. pich and e. de rafael , phys .
b * 189 * 363 ( 1987 ) .
barr _ et al .
b * 242 * 523 ( 1990 ) , b * 284 * 440 ( 1992 ) .
v. papadimitriou _ et al .
d * 44 * 573 ( 1991 ) .
a. alavi - harati _ et al .
_ , hep - ex/9902029 . l. cappiello , g. dambrosio and m. miragliuolo , phys . lett .
b * 298 * 423 ( 1993 ) ; + a.g .
cohen , g. ecker and a. pich , phys .
b * 304 * 347 ( 1993 ) .
t. morozumi and h. iwasaki , prog .
. phys . * 82 * 371 ( 1989 ) ; + p. heiliger and l.m .
sehgal , phys .
d * 47 * 4920 ( 1993 ) ; + g. dambrosio and j. portoles , nucl .
b * 492 * 417 ( 1997 ) .
donoghue and f. gabbiani , phys .
d * 56 * 1605 ( 1997 ) .
donoghue and f. gabbiani , phys .
d * 51 * 2187 ( 1995 ) .
y. takeuchi _ et al .
b * 443 * , 409 ( 1998 ) .
t. nomura _ et al .
b * 408 * , 445 ( 1997 ) ; + t. nomura , kyoto university ph.d thesis ( 1998 ) unpublished . n.m .
kroll and w. wada , phys . rev . *
98 * , 1355 ( 1955 ) .
particle data group , c. caso _ et al .
j. c * 3 * , 1 ( 1998 ) .
cousins and v.l .
highland , nucl .
instrum . meth .
a * 320 * , 331 ( 1992 ) . | we report on results of a search for the decay mode @xmath0 conducted by the e162 experiment at kek .
we observed no events and set a 90% confidence level upper limit of @xmath1 for its branching ratio .
this is the first published experimental result on this decay mode .
1em _ pacs : _ 13.20.eb , 14.40.aq , , , , , , , , , , , |
odd - electron radical complexes like ho@xmath0cl@xmath3 and ho@xmath0h@xmath2o are of tremendous importance throughout chemistry and in related fields , such as in radiation chemistry , atmospheric chemistry , environmental chemistry , and cell biology@xcite . in particular , the behavior of anions in droplets is of critical importance to understanding atmospheric chemistry .
common sense suggests that anions are less perfectly screened near a water - air interface , and so have lower concentration there .
recent classical molecular dynamics ( md ) simulations have shown just the opposite@xcite , creating considerable controversy on this point .
since anions are strongly quantum mechanical , it is logical to check traditional md simulations , using only classical force fields , against _ ab initio _ molecular dynamics ( aimd ) calculations , which use density functional theory ( dft ) to generate the potential energy surfaces ( pes ) .
unfortunately , standard dft approximations have issues for such systems@xcite .
several studies show dft approximations predict two minima in the ground - state pes .
one of the minima is a hydrogen - bonding structure , while the other is a two - center three - electron interacting hemi - bonding structure@xcite .
many dft studies predict the hemi - bonding structure as the global minimum of [ ho@xmath0cl(h@xmath2o)@xmath4 ^ -$ ] complex@xcite .
this is attributed to the infamous self - interaction error@xcite , because aimd studies with self - interaction corrected ( sic ) methods agree with pes scans with high - level ( beyond dft ) quantum - chemical methods , showing no hemi - bonding configuration in the ground state@xcite .
hemi - bonding configurations have been observed in experiments@xcite , but high - level quantum - chemical studies show that these are excited - state rather than ground - state configuration@xcite . thus , high - level quantum chemical calculations reveal that the ground - state pes has only one minimum , which is the hydrogen - bonding structure .
the hemi - bonding structure is relatively overstabilized in dft because three electrons are incorrectly delocalized over two atoms rather than having localized electron configurations .
essentially , the aimd studies raise more questions than they answer . however , recent work@xcite produces a much simpler and more general picture of such errors , and suggests alternative solutions . in any approximate dft calculation ,
the density is found by self consistently solving the kohn - sham ( ks ) equations , using an approximate ks potential derived from the energy approximation .
this procedure finds the density that minimizes the energy approximation for the given system .
the final output energy is of the approximate energy functional evaluated at the approximate density . for most ks - dft calculations using standard approximations
, the ks potential appears to be of very poor quality@xcite . in particular , the eigenvalues are usually too shallow by several ev@xcite .
dft approximations are designed to yield good energies , but this does not automatically imply good potentials .
functional derivatives are determined by how well the approximation performs for small variations _
away _ from the density of interest . nevertheless , usually the self - consistent density is rather good , because the approximate ks potential is rather close to a simple shift of the exact ks potential .
( highlighting this point , a recent approximation yields highly accurate energies with truly terrible potentials@xcite . )
this then raises a simple issue .
one can think of two distinct sources of error in such a calculation : one of the error due to the energy approximation itself , and the other error in the corresponding approximate density . in most dft calculations ,
it is ( correctly ) believed that the first ( functional ) error dominates .
but in many interesting cases that are specific to the system and the approximation , the error in the approximate density can be unusually large , so that it contributes more to the total energy error than would usually be the case .
in such cases of density - driven errors , use of a more accurate density should reduce the final error substantially .
a small homo - lumo gap ( @xmath5 ) in the original approximate dft calculation is often a strong indicator of a large density - driven error@xcite . in many cases , with standard dft approximations , a hartree - fock ( hf )
calculation provides a sufficiently accurate density to garner much of the improvement in energetics .
we dub such a calculation , which is quite general , density corrected dft ( dc - dft ) , meaning any dft calculation in which the density is _ not _ found self consistently , but by some other method that substantially reduces the density - driven error .
we note that a better term for self - interaction error is delocalization error , which has been recently much studied by cohen et al@xcite .
our analysis complements theirs . in that language , we are able to determine if a given delocalization error is density - driven or not .
if the former , a better density will improve the energy ; if the latter , as in the case of stretched h@xmath6 , it will not .
we expect cases like dft calculation on stretched he@xmath6 to be normal , so the density - driven error would be small compared to the functional error@xcite . in the present work , we apply this scheme to a variety of commonly - used standard approximations in dft , applied to the ho@xmath0cl@xmath1 complex .
we use the hf density as our more accurate density .
we use the notation a - b , where a indicates the method for finding the density , and b the method for finding the energy .
thus hf - dft implies using a hf density , but evaluating its energy using dft approximation . in fig .
[ hoclpesfunc2 ] we show the near - perfect agreement of many commonly - used functionals for the potential energy surface of ho@xmath0cl@xmath1 , once they are evaluated on hf densities .
we find that not only does hf - dft always produce a pes with the correct global minimum , but that the difference in pes s between different approximations becomes negligible and is , up to a constant , essentially identical to the ccsd(t ) pes .
thus , with this elementary correction , the pes of such systems changes from being a disappointing failure to being a resounding success for approximate dft .
there is always a price to be paid for progress , and in this case , there are two ( not so hidden ) challenges to taking advantage of this improved performance .
the first is that , if the density is no longer the minimizer for the given approximate energy functional , many basic theorems , such as the hellmann - feynman theorem , no longer apply , and many of these are used in standard dft codes .
but the way to calculate forces has already been detailed in a pioneering work@xcite which pointed out that hf - dft generically improves most reaction barriers over self - consistent dft , and implemented it in the aces code .
the second is that aimd simulations are often performed with plane waves enforcing periodic boundary conditions , and solving the hf equations can be computationally expensive in such implementations@xcite .
however , some simpler scheme available in those codes , such as self - interaction corrected local density approximation ( lda - sic)@xcite , might yield densities sufficiently accurate for the purpose while using e.g. , generalized gradient approximations ( gga ) for the energy evaluation . alternatively with the recent success of hybrid functionals such as hse06@xcite in predicting fundamental gaps
, it might be straightforward to run a hf calculation under these conditions .
but all these are beyond the scope of the present paper .
the aim here is to show just how much of an improvement is possible with dc - dft .
we begin with some theory , which also includes the historical background .
since this idea is so elementary , it has been touched upon several times in the distant and recent past .
we also explain how this fits in with all the present attempts to go beyond standard approximations , including using exact exchange in dft@xcite , _ ab initio _
dft@xcite , and several many - body approaches using ks orbitals@xcite .
next we show a variety of results for the complex at hand , using several functionals , basis - sets , and implicit solvent models .
we also demonstrate similar results for the ho@xmath0h@xmath2o radical , showing that the effect is not specific to anionic species , but is much more general .
in any ks - dft calculation , only a small fraction of the total energy need be approximated as a functional of the density , namely the exchange - correlation ( xc ) energy . in reality
, we always use spin densities , but for the present purposes , we suppress the spin index . the self - consistent solution of the ks equations has been designed to deliver the density that minimizes the total energy in the ks scheme@xcite : e [ ] = t+ u [ ] + v [ ] + e where the @xmath7 indicates an approximation . the functionals @xmath8 $ ] , @xmath9 $ ] , @xmath10 $ ] , @xmath11 $ ] , and @xmath12 $ ] are the total energy , kinetic energy , external potential energy , hartree - energy and xc energy functionals , respectively@xcite . the energy error in a dft calculation is defined as e = e [ ] - e [ ] where @xmath13 is the approximate self - consistent density .
we may write this error as the sum of two contributions .
we call the first the _ functional _ error .
it is the energy error made by the functional evaluated on the exact density , and comes entirely from the xc approximation : e_f = e [ ] - e [ ] = e - e .
the _ density - driven _
error is the energy difference generated by having an approximate density : e_d = e[]-e[n ] , so that the total energy error is the sum of these two : e = e_f + e_d .
this separation applies to _ any _ approximate dft calculation , not just a ks calculation of electronic structure .
but here we focus exclusively on the latter , since we wish to use this as a tool to analyze chemical calculations using ks - dft . note
that this elementary breakdown is hardly a breakthrough .
most developers and many users of dft have thought along these lines or come across this in some calculation or context of theory development .
what _ is _ new is how far this elementary step can be taken in analyzing all practical dft calculations , i.e. , those using approximate functionals , and how to improve many of these .
the first point to note is that for most calculations using the ks scheme and modern approximations to xc , such as a gga or a global hybrid , the densities are remarkably accurate .
the above tool allows us to specifically quantify this accuracy , by measuring it in terms of its effect on the quantity we almost exclusively care about , namely the ground - state energy . in any calculation , if @xmath14 , any error in the density is irrelevant for practical purposes .
for example , it has long been known that the density with standard approximations is highly inaccurate at large distances from the nuclei , due to the highly inaccurate homo in such calculations@xcite . however , in most cases , this inaccuracy produces only a very small density - driven error in the energy , so such approximations remain accurate for ground - state energies .
moreover , dft approximations produce far more accurate ionization energies via total energy differences than via orbital energy differences .
next , we consider some popular application , such as the calculation of a bond length with dft . since the bond length is extracted as the minimum of the total energy , the contributions to the error can be split into functional - driven and density - driven , and the two compared .
thus one extracts the density - driven contribution to a given property of a given system with a given approximate functional .
if this error is small or negligible compared to the actual error , we classify such a calculation as _
normal_. we believe the vast majority of dft calculations fit into this category , and we gave several examples in our previous work , including cases like stretched h@xmath6 cation@xcite .
in fact , in many circumstances of method development , there is an underlying assumption that the calculation _ is _ normal . with a new approximate functional
, it can often be the case that the functional derivative is demanding to calculate .
thus , often a lower - level , more standard approximation is used to calculate orbitals in the ks equations , and the new approximation is tested on those orbitals .
if the calculation is normal , this ( almost ) guarantees that only a small error is made by this procedure , and the change upon self - consistency will be negligible . our main interest , naturally , will be in those calculations where the density - driven error is a significant fraction of the total .
we denote such calculations as _ abnormal_. in such cases , a more accurate density will reduce the error ( assuming no accidental cancellation of functional- and density - driven errors ) .
we can also state just how much more accurate that density need be : enough to make the density - driven error small relative to the functional error .
in such cases , correcting the density in a dft calculation greatly reduces the error ; hence our name for this method .
while this separation scheme can be applied to any approximate dft calculation , we are presently focused on the infamous self - interaction error inherent in standard gga and global hybrid calculations . in earlier work , we found that many cases of such errors ( small anions , underestimated transition barriers , and incorrect dissociation limits ) were in fact density - driven@xcite .
in such cases , the density was particularly poorly described , and a simple hf density was sufficient to drive out the density - driven error .
note that this does not mean the hf density is especially good . in the vast majority of cases ( the normal ones ) , the hf density is worse than the self - consistent dft density@xcite , and hf - dft is worse than self - consistent dft , as we have shown in cases like stretched h@xmath15 in our previous work@xcite . in these normal cases ,
self - consistent dft is usually sufficient enough for getting accurate energies despite having incorrect potentials@xcite .
but for abnormal systems , the self - consistent dft density is especially poor in a very systematic way , a way that is largely fixed by hf . as we showed in ref .
, an unusually small ks homo - lumo gap , @xmath16 , in the dft calculation indicates a likely abnormal calculation .
this means the self - consistent solution is unusually sensitive to small changes in the potential , so that an error in the xc contribution to that potential can produce an unusually large effect on the density . in the case of atomic anions , this becomes extreme : the homo is positive if the basis set is used to hold it in , and zero in the basis - set converged limit@xcite .
this leads to very poor densities , missing a significant fraction of an electron , and large density - driven errors .
hf densities are a major improvement in such cases .
.energy decomposition of h@xmath3 and he . [ cols="^,^,^,^,^,^,^ " , ] [ 2eenergy ] [ 2edenpot ] to illustrate the idea , we consider the simplest possible case , the he atom .
we begin by applying our analysis to a hf calculation of the system . for two spin - unpolarized electrons ,
a hf calculation is equivalent to a ks - dft calculation with e= - u[]/2 .
in fact , the total error in such a calculation is simply the ( quantum - chemical ) definition@xcite of the negative of the correlation energy : e = e- e[n ] = - e. in table i , we list the different errors in hf and pbe calculations , for both he and h@xmath1 . we see that , for hf applied to he , the density - driven error is minuscule ( 0.05 mh ) , i.e. , about 0.1% of the functional error .
this says that , for this problem , the hf density is extremely accurate , and essentially all the error comes from the missing correlation energy .
such a calculation could be considered ultra - normal .
next , we repeat the analysis with the pbe approximation@xcite . here
the total error is smaller ( about 11 mh ) , and the density - driven error is -1 mh . because this is still only of order 10% , this remains a normal calculation , just like the hf calculation .
but in fig .
[ hedenpot ] , we show the corresponding densities and ks potentials for the pbe and hf calculation .
although the densities are identical to the eye , we see that the ks potential of the pbe calculation is far too shallow .
this is typical of all approximate dft calculations , and leads to ks eigenvalues that are far too shallow(-0.58 ev instead of -0.903 ev ) .
nevertheless , we emphasize that these are normal calculations , and the error in the potential produces very little error in the density , and so relatively small density - driven errors .
we also emphasize that normality ( or otherwise ) is a characteristic of a particular calculation ( i.e. , the system and the approximation together ) . , but for h@xmath1.,width=240 ] now we repeat this exercise for the @xmath17 ion .
it is often said that h@xmath1 is ` correlation bound ' , meaning that in hf theory , h@xmath1 is not stable . in a hf calculation , the ground - state energy of h is lower than that of h@xmath1 .
nonetheless , we can still converge the calculation and obtain the ground - state energy .
the correlation energy remains about the same , but the density - driven error is much larger . in fig .
[ h - denpot ] , we can see the error in the density by eye . nonetheless , this calculation is also normal , with a density - driven error that is only about 3% of the total .
it is the functional error that makes h@xmath1 correlation bound .
a very different story is seen for the pbe calculation on the same system .
in fact , only by allowing a fraction of an electron ( about 0.3 ) to escape the system can the calculation be properly converged at all@xcite . here
the density - driven error is more than 10 times larger than the magnitude of the functional error .
the density itself is very dramatically different from the exact one , and the pbe ks potential is not only too shallow , but is actually positive , and the eigenvalue is exactly zero .
this is a very abnormal calculation .
evaluation of the pbe approximation on the exact density removes the density - driven error , and so drops the total energy error by an order of magnitude . even using the hf density is sufficient to produce a much highly accurate electron affinity of h@xcite .
we end this section by noting that the functional errors and density - driven errors have opposite signs .
thus , in a normal calculation , application of the approximate functional on the exact density will _ increase _ the energy error ( albeit only slightly ) .
we have found this in all our calculations so far , which suggests this is typical behavior .
thus we do _ not _ recommend universally using more accurate densities than the self - consistent density .
only when a calculation is abnormal do we suggest such a procedure . in the early days of dft , it was often easier to use a hf code to find densities and evaluate xc approximations on those densities , again because such calculations were assumed to be normal@xcite . when dft began to become popular in chemical applications in the early 1990 s , this mode of testing approximations , called hf - dft , was used in calculations@xcite .
but very quickly , the computational and conceptual advantages of self - consistency led to self - consistent dft calculations .
moreover , the hf - dft results were not systematically compared to self - consistent dft calculations , except in some pioneering works which suggested that in difficult cases , the hf - dft may yield more accurate answers@xcite . the present paper may be considered as a fuller exploration and quantification of those early results .
what does this analysis say about the many attempts to improve energetics in other ways , such as _ ab initio _
dft@xcite , random - phase approximation ( rpa ) @xcite , density - matrix functionals , etc ?
our analysis explains several key features .
the first is that , despite having very wrong - looking xc potentials , and hence bad ks potentials , the effect of these errors on the energy via the density is minimal ( a more detailed explanation is given in ref . ) .
thus despite legions of papers reporting the very incorrect energy eigenvalues of ks potentials with approximate functionals , especially the highest occupied one which , for the exact functional , matches the negative of the ionization potential , dft with these approximations continues to be very heavily used , because the energetics are unaffected .
second , we now have a tool for quantifying the energetic error due to the density error .
this allows us to ask questions such as when do we need to improve the density , in order to improve the energy . for example , dft calculations using so - called exact exchange ( exx ) have much more accurate ks potentials than those of standard approximations , yet usually worse energetics .
this is because their functional errors usually outweigh the reduction in the density - driven error .
obviously , our present answer is a purely pragmatic one . in specific circumstances ,
the density becomes sufficiently poor as to be the major source of error .
this shows that only in certain circumstances does this problem need to be addressed , and if ways could be found to avoid the poor self - consistent densities in such cases , a variety of apparent dft errors would be avoided .
we begin with the pes of the ho@xmath0cl@xmath1 complex .
the pes was scanned by changing the cl - o distance ( @xmath18 ) and cl - o - h angle ( @xmath19 ) , as indicated in fig .
[ hoclpesfunc1](a ) .
the o - h bond length was fixed at 1.0 .
we performed single - point energy calculations with dft , hf - dft , and coupled - cluster method ( ccsd(t ) ) on each geometry .
we used gga functionals ( pbe@xcite and blyp@xcite ) , hybrid functionals ( pbe0@xcite and b3lyp@xcite ) and a double - hybrid functional with empirical dispersion correction ( b2plyp - d)@xcite .
we did these both in the gas phase and with implicit solvent . for implicit solvent calculations
, we use the conductor - like screening model ( cosmo)@xcite within the turbomole suite@xcite .
we used dunning s augmented correlation - consistent basis sets with x zeta functions ( aug - cc - pvxz , x = 2 - 3 , denoted as avxz for simplicity from now on)@xcite .
contour plots of the ho@xmath0cl@xmath1 complex pes evaluated with various methods are shown in fig .
[ hoclpesfunc1 ] . for this simple complex
, we compare dft results with ccsd(t ) , which we take as a benchmark .
we notice several drastic failures of dft approximation in these calculations .
the worst qualitative failure is that the minimum of the gga pes is not at 0@xmath20 , but is closer to 30@xmath20 .
this is an incorrect hemi - bonding arrangement , attributed to the strong self - interaction error of the extra electron in the literature@xcite .
we also note that the contours of the pes are quite incorrect in shape everywhere in the plane we have plotted . finally , the gga pes is too negative overall ( blue everywhere ) indicating it is essentially useless for performing aimd simulations of anions .
while pbe is very popular for many materials simulations and static quantum chemical calculations , in fact , most aimd simulations do not use pbe but other gga s instead@xcite .
this is because some key attributes of thermal simulations of water are incorrectly described by pbe .
a popular alternative is blyp , even though this is rarely used in regular quantum chemical calculations ( unlike its hybrid off - spring , b3lyp ) . but a glance at fig .
[ hoclpesfunc1 ] ( c ) shows that blyp is almost identical to pbe for this purpose , and suffers all the same difficulties . on the other hand , hybrid density functionals , which add some fraction of hf to the gga form ,
usually improve energetics@xcite and partially correct self - interaction error .
so in fig .
[ hoclpesfunc1 ] ( e ) and ( f ) we plot the results with pbe0 and with b3lyp , which is the most popular functional in quantum chemistry .
we see that indeed there is great improvement .
the minimum is now correctly at alignment , the surfaces are not entirely blue , and the shape is roughly correct . as we have shown in fig .
[ hoclpesfunc2 ] , the results by evaluating dft energies on hf densities are striking . in every case , we get essentially identical results throughout the plane of the pes . all minima are in the correct locations , no curves are too blue , and all the details are correct .
the results are so consistent that we can draw several important conclusions . *
such good agreement confirms the theory behind dc - dft .
all these dft calculations are _ abnormal _ , and the error is greatly reduced by using a better density .
it also confirms that the hf density is sufficiently more accurate than the self - consistent dft density for these calculations to produce much more accurate energies . * for this problem , we no longer need the benchmark defined by ccsd(t ) . the extreme level of consistency between so many different dft approximations imply that all such calculations are yielding a very accurate answer .
* while the hybrid functionals definitely improve over the gga s , the primary effect is the improvement in the self - consistent density due to the hf component in the energy .
in fact , evaluated on a sufficiently accurate density , there is no need to use a hybrid functional ( but of course , finding the hf density is relatively expensive in aimd calculations ) . ,
of ho@xmath0cl@xmath1 complex for several approximate functionals.,width=240 ] next we consider the gap in the approximate self - consistent ks calculations .
part of the dc - dft theory is that an abnormal system should have an unusually small ks gap , suggesting that its density is unusually inaccurate@xcite .
[ hoclgapfunc ] shows @xmath16 for the ho@xmath0cl@xmath1 complex .
each point corresponds to the energy minimum of each @xmath19 , i.e. , using @xmath18 that gives the lowest binding energy for given @xmath19 .
since we scanned @xmath18 = 2.5 @xmath7 4.5 for calculations , we excluded @xmath21 70@xmath20 from this figure where the energy minimum was located at @xmath18 = 2.5 or @xmath18 = 4.5 .
the gga methods clearly have small @xmath16 ( less than 1 ev ) which is consistent with the large density - driven error in these methods . in the case of hybrid methods ,
the @xmath16 is a mixture of a hf gap and a ks gap rather than pure ks gap , so it may not be as good as an indicator for density - driven errors .
the hybrid @xmath16 are not as small as in gga , but still less than 2 ev , which explains the moderate density - driven error compared to gga methods
. as some dft calculations with @xmath16 even as high as 2.5 ev have large density - driven error@xcite , calculations with @xmath16 below 2 ev should be suspected of being abnormal .
to gain more insight and quantitative understanding , we show energy curves along @xmath18 = 3.0 in fig .
[ hoclfunc30r ] .
the binding energy @xmath22 is defined as @xmath23 - ( e[\cdot\mbox{oh } ] + e[\mbox{cl}^{-}]$ ] ) , where @xmath24 $ ] , @xmath25[@xmath0oh ] , and @xmath25[cl@xmath3 ] is the energy of ho@xmath0cl@xmath3 complex , oh radical , and cl@xmath3 anion respectively .
the dissociation energy @xmath27 is defined as @xmath28 , where @xmath29 is the energy minima on @xmath18 = 3.0 for the given calculation method .
we see very clear patterns . in fig .
[ hoclfunc30r](a ) , the gga s ( blue and green ) produce incorrect minima at @xmath19 = 20@xmath20 .
hybrid methods ( red and orange ) are much more accurate near the minimum , but show increasing error as @xmath19 gets larger .
the b2plyp - d curve is highly accurate almost everywhere . on the other hand , in fig .
[ hoclfunc30r](b ) , all the gga and hybrid curves line up almost perfectly , once evaluated on hf densities .
the small remaining deviation among them is near the minimum , where the pbe methods ( pbe and pbe0 ) are most accurate . in any case , all are slightly shifted above the accurate ccsd(t ) curve .
finally , we decompose the energy error into density - driven and functional errors for both the ggas and hybrids in fig .
[ hoclfunc30r](c ) .
gga methods show large density - driven error for all regions , maximizing at @xmath19 = 130 @xmath7 140@xmath20 , while hybrid methods have less but still significant density - driven error maximizing at @xmath19 = 120 @xmath7 130@xmath20 .
functional errors stay almost constant for every @xmath19 .
the evaluated density changes as the geometry changes and the functional used is left unchanged , resulting in this independence of functional error to geometry . we have also calculated pes for both pbe and b3lyp , self consistently and in hf - dft , using a smaller basis , namely , avdz .
these are qualitatively and quantitatively almost identical to those with avtz , showing basis set convergence , and that avdz may be sufficient for most purposes for these calculations .
this is illustrated in the energy error decomposition for pbe in fig .
[ hoclbas30ang2](a ) , where the shifts from one basis to the next are tiny compared to all other energy error contributions .
the @xmath16 of both methods in fig .
[ hoclbas30ang2](b ) are also similar .
next , we show what happens when we use a more modern and more accurate approximate functional for this problem .
the b2plyp - d functional is a double hybrid functional combined with empirical dispersion parameters@xcite . in conventional hybrid functionals , hf exchange is added as the non - local exchange contribution .
in addition to this , b2plyp - d has the non - local perturbation correction added for the correlation part by second - order perturbation theory .
this is based on _ ab initio _ kohn - sham perturbation theory ( ks - pt2 ) by grling and levy@xcite . due to the large fock exchange fraction , self - interaction error
is greatly reduced , while the side effects of having large fock exchange , such as incomplete static correlation , are alleviated by the second - order perturbation in the correlation@xcite .
this leads to excellent results in many cases@xcite , including two - center three - electron bonding in radical complexes@xcite , which makes the method a great choice of benchmark for this work . in fig .
[ hoclpesb2plyp](a ) , this approximation is doing an excellent job of reproducing the pes everywhere , on its _ self - consistent density ! _
thus this functional is sufficiently accurate for this problem that this is a _
normal _ calculation .
sure enough , when we repeat the calculation using the hf density , as shown in fig .
[ hoclpesb2plyp](b ) , the pes worsens .
this strongly suggests that the b2plyp - d self - consistent density is better than the hf density here .
thus this functional can be used for this problem without modification , so long as the user can afford to evaluate it , and should not be density corrected . but
all the cruder older approximations yield abnormal results and need correction .
dauria _ et al._@xcite performed aimd simulations on the ho@xmath0cl@xmath1 complex in explicit water solvents where the minimum appeared to be a hemi - bonding structure ( @xmath19 = 80@xmath20 ) with a standard approximate functional blyp , while self - interaction corrected blyp ( blyp - sic ) gave a hydrogen - bonding minimum structure on @xmath19 = 0@xmath20 .
they then scanned the gas phase ho@xmath0cl@xmath1 complex pes along @xmath19 = 0@xmath20 and @xmath19 = 80@xmath20 based on those observations . as we observed in our gas phase calculation , the true minimum of gas phase ho@xmath0cl@xmath1 complex lies somewhere between @xmath19 = 0@xmath20 and 80@xmath20 .
to look in more depth at solvation effects , we show contour plots of ho@xmath0cl@xmath1 complex in implicit water solvent in fig .
[ hoclpessolv ] . for both pbe and b3lyp calculations ,
the minimum is clearly a hemi - bonding structure with @xmath19 = 80@xmath20 , in contrast to the gas phase calculation , where the global minimum was at @xmath19 = 20@xmath20 .
the fock exchange in the hybrid functional indeed has some effect , producing a second local minimum along the hydrogen - bonding region ( @xmath19 = 0@xmath20 ) , yet did not correct the overstabilization of the hemi - bonding structure , resulting in the wrong global minimum . on the other hand ,
the sole minimum of both dc - dft calculations is the hydrogen - bonding structure for both gas phase and implicit water calculations . unlike the self - consistent dft results , the pes are quite similar regardless of functional , which was a trait
also observed in gas phase pes , and no sort of local minimum is shown in the hemi - bonding region .
this shows the accuracy of dc - dft is on par with sic - dft with far less computational cost , at least for non - periodic cases , even in the presence of implicit solvent .
finally , we would like to mention that there are some works that show ks - dft greatly underestimates the redox potential of oh@xmath0/oh@xmath1 and cl@xmath0/cl@xmath1 in explicit solvent while this is not the case in implicit solvent simulations@xcite . even though it is possible that there is some difference between our results and the explicit solvent results
, our findings matched the explicit solvent result unlike the oh@xmath0/oh@xmath1 and cl@xmath0/cl@xmath1 case .
we believe this discrepancy is due to the difference between a anion - radical complex and a lone radical , alleviating the effect of extended states of explicit solvent .
h@xmath30o complex .
( b ) comparison of the pes scan for ho@xmath0h@xmath2o complex using various methods .
binding energies are plotted against @xmath31 .
each point is using the minimum energy geometry for given @xmath31.,width=412 ] h@xmath2o complex .
each point is using the minimum energy geometry for given @xmath31.,width=288 ] to confirm that the performance of dc - dft is not restricted to anion complexes , we also look at a neutral radical complex .
we evaluated the pes of the ho@xmath0h@xmath2o complex using dft , hf - dft and ccsd(t ) with avtz basis set .
pbe and blyp functionals are used in dft calculations .
we used the same parameters used by chipman@xcite , depicted in fig . [ hoh2opescurve](a ) .
the evaluated pes is depicted in figs .
[ hoh2opescurve ] and [ hoh2opes ] .
the binding energy @xmath22 here is defined as @xmath32ho@xmath0h@xmath2o ] - ( e[@xmath0oh ] + e[h@xmath2o ] ) , where @xmath33ho@xmath0h@xmath2o ] , @xmath34oh ] , and @xmath33h@xmath2o ] is the energy of the ho@xmath0h@xmath2o complex , oh radical , and h@xmath2o molecule , respectively . in fig .
[ hoh2opescurve ] , each point indicates the minimum energy possible for a given @xmath31 .
we chose to scan between @xmath35 to @xmath36 and @xmath37 to @xmath38 .
the @xmath35 to @xmath36 region is where the hemi - bonding geometry was observed in excited state calculations , while the latter is where the global minimum of the ground state was discovered@xcite .
the ccsd(t ) results reproduce chipman s result , where the global minimum is in the hydrogen - bonding region of @xmath39 . on the other hand , despite having a minimum in the hydrogen region , both dft methods clearly have the global minimum in the hemi - bonding region of @xmath40 . now ,
as one can expect from the ho@xmath0cl@xmath1 complex results , the hf - dft curve successfully resembles the ccsd(t ) results , having the global minimum in the hydrogen - bonding region of @xmath41 . also ccsd(t ) and hf - dft results have no minimum in the hemi - bonding region , so the @xmath31 value with the lowest energy is 60@xmath42 .
we scanned through @xmath18 and @xmath43 on the @xmath31 values that give minimum energy for each method and region in fig .
[ hoh2opes ] .
once again , the pes of hf - pbe looks like the pes of ccsd(t ) with an energy shift in both hemi - bonding and hydrogen - bonding region , while pbe has a clearly different pes in the hemi - bonding region . fig .
[ hoh2oerror ] shows the error decomposition of the pbe calculation .
as expected , calculations in the hemi - bonding region exert a strong density - driven error . in the hydrogen - bonding region ,
the density - driven error is quite small compared to the hemi - bonding region , but @xmath16 in both regions is still quite small ( @xmath16 = 1.09 ev at the hemi - bonding minimum , 0.97 ev at the hydrogen - bonding minimum ) .
approximate dft typically suffers from severe self - interaction error in calculations of odd - electron radical complexes@xcite .
this work shows that the density correction in dc - dft , even using simple hf densities , can often give more accurate results than dft using self - consistent densities in these types of calculations .
to explain this in a systematic way , we showed _ any _ approximate calculations can be classified into one of the two types . in _
normal _ calculations , the functional error dominates , while in _ abnormal _ calculations , the density - driven error is larger than the functional error .
we illustrated this using simple two - electron systems , namely h@xmath1 anion and he atom .
approximate dft was abnormal in h@xmath1 showing severe density - driven errors , while all other cases were normal . in these normal cases , the density - driven error was negligible , despite having very wrong - looking ks potentials . by this analysis
, we stressed that dc - dft is likely to give better results than approximate dft with self - consistent densities only for abnormal calculations , and not for normal calculations .
we presented pes of an anion radical complex , e.g. , the ho@xmath0cl@xmath1 complex , using various common approximate functionals including gga functionals and hybrid functionals and even more modern functionals like b2plyp - d , with both self - consistent dft and dc - dft . both gga and hybrid functionals behaved poorly self consistently and only using the highly accurate b2plyp - d self - consistent density was sufficient for getting accurate pes . on the other hand , dc - dft gave identical pes regardless of the approximate functional used , and gave correct global minima and pes slopes , showing gga and hybrid approximate functional calculations are abnormal calculations .
using the concept introduced from our previous letter@xcite , we showed a very small @xmath16 can be used as an indicator of abnormality .
we checked the calculations with different basis sets where we have similar results for both avdz and avtz basis set .
we also showed even calculations with implicit solvent have similar tendencies , where self - consistent gga and hybrid densities give poor results , predicting hemi - bonding structures as the global minimum .
pes evaluated from dc - dft were once more identical , independent of the functional used .
finally , we examined the validity of dc - dft for neutral radical complexes by evaluating pes of ho@xmath0h@xmath2o .
self - consistent dft predicted the hemi - bonding structure as the global minimum , while dc - dft correctly predicted the hydrogen - bonding structure as the global minimum .
these results and the @xmath16 showed the abnormality in these self - consistent dft calculations .
dc - dft can be used as a simple cure of abnormality , i.e. , strong density - driven error especially driven from self - interaction , which has less computational cost and is free of empirical parameters compared to various sic methods .
we must mention the hf density we used in this work may not be appropriate for all cases , including cases with strong spin - contamination@xcite , or periodic boundary conditions .
nonetheless , one can use dc - dft with any other source of accurate densities in cases where the hf density is not suited .
additionally , we expect this method to give promising results for various problems that are challenging for approximate dft like reaction barriers and dissociation@xcite , when the errors are density - driven .
we thank eberhard engel for the use of his atomic oep code and cyrus umrigar for the exact two electron system data from qmc calculation .
this work was supported by the global research network grant ( no .
nrf-2010 - 220-c00017 ) and the national research foundation [ 2012r1a1a2004782 ( e. s. ) ] .
k.b . acknowledges support under nsf che-1112442 .
m - c thanks the fellowship of the bk 21 program , and sanghyeon lee for his effort . 65ifxundefined [ 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 _
( , ) link:\doibase 10.1021/cr030453x [ * * , ( ) ] @noop _ _ ( , ) link:\doibase 10.1063/1.555805 [ * * , ( ) ] link:\doibase 10.1080/10643380500326564 [ * * , ( ) ] link:\doibase 10.1080/10643380601163809 [ * * , ( ) ] @noop _ _ , ed .
( , ) link:\doibase 10.1021/jp077669d [ * * , ( ) ] link:\doibase 10.1063/1.1630017 [ * * , ( ) ] link:\doibase 10.1039/b311840a [ * * , ( ) ] link:\doibase 10.1063/1.1468640 [ * * , ( ) ] link:\doibase 10.1021/jz3015293 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1021/jp044751p [ * * , ( ) ] link:\doibase 10.1039/b501603 g [ * * , ( ) ] link:\doibase 10.1021/jp2063386 [ * * , ( ) ] link:\doibase 10.1039/j19690000446 [ ( ) ] link:\doibase 10.1021/jp964097 g [ * * , ( ) ] link:\doibase 10.1021/jp903625k [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.073003 [ * * , ( ) ]
link:\doibase 10.1103/physreva.50.3827 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.80.3 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4834075 [ * * , ( ) ] link:\doibase 10.1063/1.4770226 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2741248 [ * * , ( ) ] link:\doibase 10.1016/j.cplett.2011.12.017 [ * * , ( ) ] link:\doibase 10.1103/physrevb.48.5058 [ * * , ( ) ] link:\doibase 10.1103/physrevb.23.5048 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2404663 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1398093 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.1453958 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.4755818 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.2977789 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.3043729 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.093003 [ * * , ( ) ] link:\doibase 10.1002/qua.24259 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1063/1.3590364 [ * * , ( ) ] link:\doibase 10.1039/c3cp52547c [ * * , ( ) ] @noop * * , ( ) @noop `` , '' in link:\doibase 10.1021/bk-1996 - 0629.ch003 [ _ _ ] , , chap . , pp . @noop
* * , ( ) link:\doibase 10.1103/physreva.38.3098 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1002/qua.560440828 [ * * , ( ) ] @noop * * ( ) link:\doibase 10.1063/1.463977 [ * * , ( ) ] \doibase doi:10.1063/1.477479 [ * * , ( ) ] @noop * * , ( ) @noop * * , ( ) \doibase http://dx.doi.org/10.1063/1.2148954 [ * * , ( ) ] link:\doibase 10.1002/jcc.20495 [ * * , ( ) ] link:\doibase 10.1039/p29930000799 [ ( ) ] @noop @noop * * , ( ) http://link.aip.org/link/?jcp/98/1358/1 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1021/jp960669l [ * * , ( ) ] link:\doibase 10.1103/physrevb.47.13105 [ * * , ( ) ] link:\doibase 10.1103/physreva.50.196 [ * * , ( ) ] link:\doibase 10.1021/ar700208h [ * * , ( ) ] link:\doibase 10.1039/b704725h [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.cplett.2010.11.062 [ * * , ( ) ] | standard density functional approximations often give questionable results for odd - electron radical complexes , with the error typically attributed to self - interaction . in density corrected density functional theory ( dc - dft ) , certain classes of density functional theory calculations are significantly improved by using densities more accurate than the self - consistent densities .
we discuss how to identify such cases , and how dc - dft applies more generally . to illustrate , we calculate potential energy surfaces of ho@xmath0cl@xmath1 and ho@xmath0h@xmath2o complexes using various common approximate functionals , with and without this density correction .
commonly used approximations yield wrongly shaped surfaces and/or incorrect minima when calculated self consistently , while yielding almost identical shapes and minima when density corrected .
this improvement is retained even in the presence of implicit solvent .
_ c # 1*#1 * |
electronic systems in the mesoscopic regime are ideal testing - grounds for investigating the quantum - to - classical transition from a microscopic coherent world ( where quantum interference effects prevail ) to a macroscopic classical world @xcite . on one hand , their size is intermediate between macroscopic and microscopic ( atomic ) systems , on the other hand , today s experimental control over their design and precision of measurement allows one to investigate them in regimes ranging from almost fully coherent to purely classical @xcite
. the extent to which quantum coherence is preserved in these systems is usually determined by the ratio @xmath7 of the dephasing time @xmath8 to some relevant classical time scale @xmath9 .
for instance , @xmath9 can be the traversal time through one arm of a two - path interferometer @xcite , or the average dwell time spent inside a quantum dot @xcite . in a given experimental set - up ,
@xmath8 can often be tuned from @xmath10 ( quantum coherent regime ) to @xmath11 ( purely classical regime ) by varying externally applied voltages or the temperature of the sample .
coherent effects abound in mesoscopic physics , the most important of them being weak - localization , universal conductance fluctuations and aharonov - bohm interferences in transport , as well as persistent currents @xcite .
the disappearance of these effects as dephasing processes are turned on has raised lots of theoretical @xcite and experimental @xcite interest .
focusing on transport through ballistic systems , dephasing is usually investigated using mostly phenomenological models of dephasing @xcite , the most successful of which are the voltage - probe and dephasing - lead models @xcite . in these models ,
a cavity is connected to two external , l ( left ) and r ( right ) transport leads , carrying @xmath12 and @xmath13 transport channels respectively .
dephasing is modeled by connecting a third `` fictitious '' lead to the system , with a voltage set such that no current flows through it on average .
electrons leaving the system through this third lead are thus re - injected at some later time , with a randomized phase ( and randomized momentum ) .
these models of dephasing present the significant advantage that the standard scattering approach to transport can be applied as in fully coherent systems , once it is properly extended to account for the presence of the third lead . using random matrix theory ( rmt )
, the voltage / dephasing probe models@xcite predict an algebraic suppression of the weak - localization contribution to the conductance ( in units of @xmath14 ) @xcite , @xmath15 ^ 2 } \ ; [ 1+\tau_{\rm d}/\tau_\phi]^{-1 } , \label{eq : dephasing - wl - without - ehrenfest } \end{aligned}\ ] ] where @xmath16 ^ 2 $ ] is the weak - localization correction in the absence of dephasing . similarly , universal conductance fluctuations become @xcite @xmath17 ^ 4 } \ ; [ 1+\tau_{\rm d}/\tau_\phi]^{-2 } , \label{eq : dephasing - ucf - without - ehrenfest } \end{aligned}\ ] ] and are thus damped below their value @xmath18 ^ 2 $ ] ( in units of @xmath19 ) for fully coherent systems with ( @xmath20 ) or without ( @xmath21 ) time reversal symmetry . for a fictitious lead connected to a two - dimensional cavity ( a lateral quantum dot ) via a point contact of transparency @xmath22 and carrying @xmath23 channels one has @xmath24 , with the electron mass @xmath25 and the area @xmath26 of the cavity .
similarly , the dwell time through the cavity is given by @xmath27 .
the dephasing and voltage probe models account for dephasing at the phenomenological level only , without reference to the microscopic processes leading to dephasing . at sufficiently low temperature
, it is accepted that the dephasing arises dominantly from electronic interactions , which , in diffusive systems , can be well modeled by a classical noise potential @xcite .
remarkably enough , this approach reproduces the rmt results of eqs . and with @xmath8 set by the noise power .
these results are moreover quite robust in diffusive systems .
they are essentially insensitive to most noise - spectrum details , and hold for various sources of noise such as electron - electron and electron - phonon interactions , or external microwave fields .
for this reason it is often assumed that dephasing is system - independent , and exhibits a character of universality well described by the rmt of transport applied to the dephasing / voltage probe models @xcite . according to the bohigas - giannoni - schmit surmise @xcite ,
closed chaotic systems exhibit statistical properties of hermitian rmt @xcite in the short wavelength limit .
opening up the system , transport properties derive from the corresponding scattering matrix , which is determined by both the hamiltonian of the closed system and its coupling to external leads @xcite .
it has been shown that for not too strong coupling , and when the hamiltonian matrix of the closed system belongs to one of the gaussian ensembles of random hermitian matrices , the corresponding scattering matrix is an element of one of the circular ensembles of unitary random matrices @xcite .
one thus expects that , in the semiclassical limit of large ratio @xmath28 of the system size to fermi wavelength , transport properties of quantum chaotic ballistic systems are well described by the rmt of transport .
this surmise has recently been verified semiclassically @xcite .
the regime of validity of rmt is generally bounded by the existence of finite time scales , however , and it was noticed by aleiner and larkin that , while the dephasing time @xmath29 gives the long time cut - off for quantum interferences , a new _ ehrenfest time _
scale appears in quantum chaotic system in the deep semiclassical limit , which determines the short - time onset of these interferences @xcite .
the ehrenfest time @xmath30 corresponds to the time it takes for the underlying chaotic classical dynamics to stretch an initially localized wavepacket to a macroscopic , classical length scale . in open cavities ,
the latter can be either the system size @xmath31 or the width @xmath32 of the opening to the leads .
accordingly , one can define the closed cavity , @xmath33 $ ] , and the open cavity ehrenfest time , @xmath34 $ ] @xcite .
the emergence of a finite @xmath30 strongly affects quantum effects in transport , and recent analytical and numerical investigations of quantum chaotic systems have shown that weak localization @xcite and shot noise @xcite are exponentially suppressed @xmath35 $ ] in the absence of dephasing ( @xmath36 ) .
interestingly enough , the deep semiclassical limit of finite @xmath30 sees the emergence of a quantitatively dissimilar behavior of weak - localization and quantum parametric conductance fluctuations , the latter exhibiting no @xmath30-dependence in the absence of dephasing @xcite .
these results are not captured by rmt , instead one has to rely on quasiclassical approaches @xcite or semiclassical methods @xcite to derive them .
the behavior of quantum corrections to transport at finite @xmath30 in the presence of dephasing was briefly investigated analytically in ref .
[ ] , for a model of classical noise with large angle scattering , and numerically in ref .
[ ] , for the dephasing lead model with a tunnel barrier . intriguingly enough , the two approaches delivered the same result , that quantum effects are exponentially suppressed @xmath37 $ ] .
this suggested that dephasing retains a character of universality even in the deep semiclassical limit .
more recent investigations have however showed that at finite ehrenfest time , eq . ( [ eq : dephasing - wl - without - ehrenfest ] ) becomes @xcite ( see also ref . [ ] ) @xmath38 ^ 2 } \ ; \frac{\exp[-({\tau_{\rm e}^{\rm cl}}/{\tau_{\rm d}}+ \tilde{\tau}/\tau_\phi)]}{1+\tau_{\rm d}/\tau_\phi } , \label{eq : wl - with - ehrenfest } \end{aligned}\ ] ] with a strongly system - dependent time scale @xmath3 . ref .
[ ] showed that , for the dephasing lead model , @xmath39 in terms of the transparency @xmath22 of the contacts to the leads , which provides theoretical understanding for the numerical findings of ref .
[ ] . if however one considers a system - environment model , where the environment is mimicked by electrons in a nearby closed quantum chaotic dot , one has @xmath40 , where @xmath41 , \label{eq : tau_xi } \end{aligned}\ ] ] in terms of the correlation length @xmath5 of the inter - dot interaction potential . on the experimental front , an exponential suppression @xmath42
$ ] of weak - localization with temperature has been reported in ref .
[ ] . taking @xmath43 as for dephasing by electronic interactions in two - dimensional diffusive systems , this result was interpreted as the first experimental confirmation of eq .
( [ eq : wl - with - ehrenfest ] ) .
there is no other theory for such an exponential behavior of weak - localization , however , the temperature range over which this experiment has been performed makes it unclear whether the ballistic @xcite , @xmath44 , or the diffusive dephasing time determines the ehrenfest time dependence of dephasing ( see the discussion in ref . [ ] ) .
[ cols="<,^,^,^",options="header " , ] in the present article , we amplify on ref . [ ] and extend the analytical derivation of eq .
( [ eq : wl - with - ehrenfest ] ) briefly presented there .
we investigate three different models of dephasing and show that the suppression of weak - localization corrections to the conductance is strongly model - dependent .
first , we consider an external environment modeled by a capacitively coupled , closed quantum dot . we restrict ourselves to the regime of pure dephasing , where the environment does not alter the classical dynamics of the system .
second , we discuss dephasing by a classical noise field .
third , following ref .
[ ] , we provide a semiclassical treatment of transport in the dephasing lead model .
for these three models , we reproduce eq .
( [ eq : wl - with - ehrenfest ] ) and derive the exact dependence of @xmath3 on microscopic details of the models considered .
all our results are summarized in table [ table1 ] . the outline of this article goes as follows . in section [ sys - env - sec ] , we present the treatment of the system - environment model , focusing in particular on the construction of a new scattering approach to transport that incorporates the coupling to external degrees of freedom .
we apply this formalism to a model of an open quantum dot coupled to a second , closed quantum dot .
we present a detailed calculation of the drude conductance and the weak - localization correction , including coherent - backscattering , which explicitly preserves the unitarity of the @xmath45-matrix , and hence current conservation .
this calculation is completed by a derivation of the fano factor , showing that , in the pure dephasing limit , shot noise is insensitive to dephasing to leading order . in section [
sect : classical ] , we present a model of dephasing via a classical noise field ( such as microwave noise ) .
we consider classical johnson - nyquist noise models of dephasing due to electron - electron interactions within the system , and dephasing due to charge fluctuations on nearby gates . in section [ dephas - sec ]
we present a trajectory based semiclassical calculation of conductance in the dephasing lead model , both for fully transparent barriers and tunnel barriers .
we also comment on dephasing in multiprobe configurations . finally , section [ num - sec ] is devoted to numerical simulations confirming our analytical results .
summary and conclusions are presented in section [ concl - sec ] , while technical details are presented in the appendix .
in the scattering approach to transport , the system is assumed fully coherent and all dissipative processes occur in the leads @xcite .
apart from its coupling to the leads , the system is isolated . here
we extend this formalism to include coupling to external degrees of freedom in the spirit of the standard theory of decoherence .
the coupling to an environment can induce dephasing and relaxation . here , we restrict ourselves to pure dephasing , where the system - environment coupling does not induce energy , nor momentum relaxation in the system . in semiclassical language , we assume that classical trajectories supporting the electron dynamics are not modified by this coupling .
the starting point of the standard theory of decoherence is the total density matrix @xmath46 that includes both system and environmental degrees of freedom @xcite .
the observed properties of the system alone are contained in the reduced density matrix @xmath47 , obtained from @xmath46 by tracing over the environmental degrees of freedom .
this procedure is probability conserving , @xmath48=1 $ ] , but it renders the time - evolution of @xmath49 non - unitary , and in particular , the off - diagonal elements of @xmath49 decay with time .
this can be quantified by the basis independent purity @xcite , @xmath50 \le 1 $ ] , which remains equal to one only in the absence of environment .
we generalize this standard approach to the transport problem .
schematic of the system - environment model .
the system is an open quantum dot that is coupled to an environment in the shape of a second , closed dot.,width=264 ] we consider two capacitively coupled chaotic cavities as sketched in fig .
[ fig : env_model ] .
the first one is the system ( sys ) , an open , two - dimensional quantum dot , ideally connected to two external leads .
the second one is a closed quantum dot , which plays the role of an environment ( env ) .
the two dots are capacitively coupled , and in particular , they do not exchange particles .
thus current through the system is conserved .
we require that the size of the contacts between the open system and the leads is much smaller than the perimeter of the system cavity but is still semiclassically large , so that the number of transport channels satisfies @xmath51 .
this ensures that the chaotic dynamics inside the dot has enough time to develop , @xmath52 , with the classical lyapunov exponent @xmath53 .
electrons in the leads do not interact with the second dot .
few - electron double - dot systems have recently been the focus of intense experimental efforts @xcite .
parallel geometries , of direct relevance to the present work , have been investigated in refs .
the total system is described by the following hamiltonian @xmath54 inside each cavity the chaotic dynamics is generated by the corresponding one - particle hamiltonian @xmath55 .
we only specify that the capacitive coupling potential @xmath56 is a smooth function of the distance between the particles .
it is characterized by its magnitude @xmath57 and its correlation length @xmath5 such that its typical gradient is @xmath58 .
physically , @xmath5 is determined by the electrostatic environment of the system , such as electric charges on the gates defining the dots and the amount of depletion of the electrostatic confinement potential between the gates and the inversion layer in semiconductor heterostructures . generally speaking , @xmath57 and @xmath5 are independent parameters and can have different values in different systems , and might even be tuned by applying external backgate voltages on a given system . in the standard scattering approach ,
the transport properties of the system derive from its @xmath59 scattering matrix @xcite @xmath60 which we write in terms of transmission ( @xmath61 ) and reflection ( @xmath62 , @xmath63 ) matrices . from @xmath64 , the system s dimensionless conductance ( conductance in units of @xmath14 )
is given by @xmath65 to include coupling to an environment in the scattering approach , we need to define an extended scattering matrix @xmath66 that includes the external degrees of freedom .
this is formally done in appendix [ extends ] , and our starting point is eq .
( [ conductance ] ) for the case of an initial product density matrix @xmath67 , with @xmath68 , @xmath69 , and @xmath70 , the initial density matrix of the environment .
we define the conductance matrix @xmath71 { \mathbb s}^{\dagger } \right ) \right\vert m \right \rangle,\ ] ] where @xmath72 stands for the trace over the environmental degrees of freedom . from this matrix ,
the dimensionless conductance is then given by , @xmath73 and eq . ( [ conductance ] ) reads @xmath74 eq .
( [ conductance_gen ] ) is the generalization of the laudauer - bttiker formula in the presence of an external environment .
it constitutes the backbone of our trajectory - based semiclassical theory of dephasing .
the semiclassical derivation of the one particle scattering matrix has become standard @xcite .
once we introduce the environment we deal with a bipartite problem , here we use the two - particle semiclassical propagator developed in the framework of entanglement and decoherence @xcite .
the extended scattering matrix elements can be written as , @xmath75 where we take a @xmath76dimensional , one - particle system ( throughout what follows we take @xmath77 ) , and a @xmath78dimensional , @xmath79-particle environment . at this point
, @xmath66 depends on the coordinates of the environment and is given by a sum over pairs of classical trajectories , labeled @xmath80 for the system and @xmath81 for the environment .
the classical paths @xmath80 and @xmath82 connect @xmath83 ( on a cross - section of lead l ) and @xmath84 ( anywhere in the volume occupied by the environment ) to @xmath85 ( on a cross - section of lead r ) and @xmath86 ( anywhere in the volume occupied by the environment ) in the time @xmath87 .
for an environment of @xmath79 particles in @xmath78 dimensions , @xmath88 is a @xmath89 component vector . in the regime of pure dephasing ,
these paths are solely determined by @xmath90 and @xmath91 .
each pair of paths gives a contribution weighted by the square root @xmath92 of the inverse determinant of the stability matrix @xcite , and oscillating with one - particle ( @xmath93 and @xmath94 , which include maslov indices ) and two - particle ( @xmath95 $ ] ) action integrals accumulated along @xmath80 and @xmath81 .
we insert eq .
( [ sc - smatrix ] ) in eq .
( [ conductance_gen ] ) , sum over channel indices with the semiclassical approximation @xcite @xmath96 .
for the environment we make the random matrix ansatz that @xmath97 , where @xmath98 is the environment phase - space volume .
the dimensionless conductance then reads @xmath99 this is a quadruple sum over classical paths of the system ( @xmath80 and @xmath100 , going from @xmath101 to @xmath85 ) and the environment ( @xmath82 and @xmath102 , going from @xmath103 to @xmath88 ) with action phases , @xmath104 /\hbar , \label{phis}\\ \phi_{\rm env } & = & \big [ s_{\gamma } \left ( { \bf q},{\bf q}_{\rm 0 } ; t \right)- s_{\gamma^{\prime}}\left ( { \bf q},{\bf q}_{\rm 0 } ; t^{\prime } \right ) \big ] /\hbar , \label{pshie}\\ \phi_{{\cal u } } & = & \big [ { \cal s } _ { \gamma,\gamma } ( y , y_0 ; { \bf q},{\bf q}_0;t ) - { \cal s } _ { \gamma^{\prime},\gamma^{\prime}}(y , y_0 ; { \bf q},{\bf q}_0;t^{\prime } ) \big ] /\hbar .
\label{phiu } \end{aligned}\ ] ] we are interested in quantities averaged over variations in the energy or cavity shapes . for most sets of paths the phase of a given contribution will oscillate wildly with these variations , so the contribution averages to zero . in the semiclassical limit , eq .
( [ sc - cond ] ) is thus dominated by terms which satisfy a stationary phase condition ( spc ) , i.e. where the variation of @xmath105 has to be minimized . in the regime of pure dephasing , individual variations of @xmath106 , @xmath107 and @xmath108
are uncorrelated .
they are moreover dominated by variations of @xmath106 and @xmath107 , on which we therefore enforce two independent spc s .
the dominant contributions that survive averaging are the diagonal ones .
they give the drude conductance . indeed setting @xmath109 and @xmath110
straightforwardly satisfies spc s over @xmath106 and @xmath107 .
these two spc s require @xmath111 and lead to an exact cancellation of all the phases @xmath112 .
the dimensionless drude conductance is given by @xmath113 from here on , the calculation proceeds along the lines of ref .
the main idea is to relate semiclassical amplitudes with classical probabilities .
this is done by the introduction of two sum rules that express the ergodic properties of open cavities , eq . ( [ classprob_open ] ) , and of closed ones , eq .
( [ classprob_close ] ) , @xmath114_{\gamma } \ ! & = & \ ! \ !
\int_{-\frac{\pi}{2}}^{\frac{\pi}{2 } } { \rm d } \theta_{0}{\rm d } \theta\ , p_{\rm sys } ( { \bf y } , { \bf y}_0 ; t ) \left [ \cdots \right]_{{\bf y}_0 } , \ , \label{classprob_open } \\ \hskip -6 mm \sum_{\gamma } a_{\gamma}^2 \left [ \cdots \right]_{\gamma } \ ! & = & \ ! \ ! \int { \rm d } { \bf p}_0 \,{\rm d } { \bf p}\ , \tilde{p}_{\rm env } ( { \bf q } , { \bf q}_0 ; t ) \left [ \cdots \right]_{{\bf q}_0}.\ , \label{classprob_close } \end{aligned}\ ] ] here , @xmath115 , and @xmath116 and @xmath117 are the classical probability densities . for the system ,
we need to take into account the fact that particles are injected , which is why the classical probability density must be multiplied with the initial system momentum @xmath118 along the injection lead @xcite .
the phase points @xmath119 and @xmath120 are at the boundary between the system and the leads .
in contrast @xmath121 and @xmath122 are inside the closed environment cavity .
the momenta are integrated over the entire environment phase - space , while @xmath117 will always contain a @xmath123-function which restricts the final energy to equal the initial one , ( i.e. @xmath124 if all @xmath79 environment particles have the same mass ) .
the average of @xmath125 over an ensemble of systems or over energy gives a smooth function .
for a chaotic system we write @xmath126 likewise , the average of @xmath127 gives @xmath128 where @xmath129 is the size of the hypersurface in the environment s phase - space defined by @xmath124 ( for @xmath130 , @xmath131 where @xmath132 is the area of the environment ) .
inserting eqs .
( [ classprob_open ] ) , ( [ classprob_close ] ) , ( [ classprobaves ] ) and ( [ classprobavee ] ) into eq . ( [ drudecond ] ) , we can perform all integrals using @xmath133 and @xmath134 .
then , since @xmath135 , we recover the classical drude conductance , @xmath136 the leading - order weak - localization corrections to the conductance were identified in refs .
[ ] as those arising from trajectories that are paired almost everywhere except in the vicinity of an encounter .
an example of such a trajectory is shown in fig .
[ fig : weakloc ] . at the encounter , one of the trajectories ( @xmath100 )
intersects itself , while the other one ( @xmath80 ) avoids the crossing .
thus , they travel along the loop they form in opposite directions . for chaotic ballistic systems in the semiclassical limit , only pairs of trajectories with small crossing angle @xmath137 contribute significantly to weak - localization . in this case , trajectories remain correlated for some time on both sides of the encounter , the correlated region indicated in pink in fig .
[ fig : weakloc ] . in other words ,
the smallness of @xmath137 requires two minimal times , @xmath138 to form a loop , and @xmath139 in order for the legs to separate before escaping into different leads . in the case of an hyperbolic dynamics one estimates @xcite , @xmath140 , \\
\label{tw } t_{\rm w}(\epsilon ) & \simeq & \lambda^{-1 } \ln [ \epsilon^{-2 } ( w / l)^2 ] .
\end{aligned}\ ] ] as long as the system - environment coupling does not generate energy and/or momentum relaxation , the presence of an environment does not significantly change this picture .
however it does lead to dephasing via the accumulation of uncorrelated action phases , mostly along the loop , when @xmath80 and @xmath100 are more than a distance @xmath5 apart .
this is illustrated in fig .
[ fig : weakloc ] for the case when @xmath5 is less than @xmath32 .
we define a new timescale @xmath141 as twice the time between the encounter and the start of the dephasing , @xmath142.\ ] ] dephasing occurs mostly in the loop part . however if @xmath143 and @xmath144 , dephasing starts before the paths reach the encounter .
we discuss this point in more detail below in section [ env - discus ] .
( color online ) a semiclassical contribution to weak - localization for the system - environment model .
the paths are paired everywhere except at the encounter , where one path crosses itself at angle @xmath137 , while the other one does not ( going the opposite way around the loop ) . here
we show @xmath145 , so the dephasing ( dotted path segment ) starts in the loop ( @xmath146 ) .
, width=264 ] in the absence of dephasing each weak - localization contribution accumulates a phase difference @xmath147 @xcite . in the presence of an environment ,
an additional action phase difference @xmath148 is accumulated . incorporating this additional phase into the calculation of weak - localization
does not require significant departure from the theory at @xmath149 .
we extend the theory of ref .
[ ] to account for this additional phase .
we follow the same route as for the drude conductance , but now consider the pairs of paths described in section [ sect : weak - loc - env ] above , and shown in fig .
[ fig : weakloc ] , while the environment is still treated within the diagonal approximation , @xmath150 . the sum rule of eq .
( [ classprob_close ] ) still applies . though the sum over system paths is restricted to paths with an encounter , we can still write this sum in the form given in eq .
( [ classprob_open ] ) , provided the probability @xmath151 is restricted to paths which cross themselves . to ensure this we write @xmath152 here
, we use @xmath153 , @xmath154 $ ] for phase - space points inside the cavity , while @xmath155 lies on the lead as before .
we then restrict the probabilities inside the integral to trajectories which cross themselves at phase - space positions @xmath156 with the first ( second ) visit to the crossing occurring at time @xmath157 ( @xmath158 ) .
we can write @xmath159 and set @xmath160 . then
, the weak - localization correction to the dimensionless conductance in the presence of an environment is given by , [ eq : gwl - integral ] @xmath161 \big\langle f ( { \bf y}_{0 } , \epsilon ) \big\rangle , \qquad \label{eq : gwl - integral - a } \end{aligned}\ ] ] with , @xmath162 .
\end{aligned}\ ] ] comparison with eq .
( 34 ) of ref .
[ ] shows that the effect of the environment is entirely contained in the last line of eq .
( [ eq : f ] ) . at the level of the diagonal approximation for the environment , @xmath150 ,
one has @xmath163 , \nonumber \\
\end{aligned}\ ] ] where @xmath164 and @xmath165 parametrize the trajectories of the system and of the environment respectively .
we note that in the absence of coupling , @xmath166 , the integral over the environment is one , and we recover the weak - localization correction of an isolated system ( cf .
@xmath167 of ref .
[ ] ) . to evaluate eqs .
( [ eq : gwl - integral ] ) , we need the average effect of the environment on the system , after one has traced out the environment . for a single measurement , this average is an integral over all classical paths followed by the environment , starting from its initial state at the beginning of the measurement .
we therefore also average over an ensemble of initial environment states , or an ensemble of environment hamiltonians , which corresponds to performing many measurements .
for compactness we define @xmath168 as this integral over environment paths and the ensemble averaging over the environment , @xmath169_{{\bf q}_0 } \big \rangle \ , .
\end{aligned}\ ] ] without loss of generality we assume that for all @xmath170 , the interaction @xmath171 is zero upon averaging over all @xmath88
. we can ensure an arbitrary interaction fulfills this condition by moving any constant term in @xmath56 into the system hamiltonian ( these terms do not lead to dephasing ) . since the environment is ergodic , we have @xmath172 now we use the chaotic nature of the environment to give the properties of the correlation function @xmath173 .
* correlation functions typically decay exponentially fast with time in chaotic systems , with a typical decay time related to the lyapunov exponent @xcite .
the precise functional form @xmath174 of the temporal decay of the coupling correlator depends on details of @xmath175 and @xmath56 , however for all practical purposes , it is sufficient to know that it decays fast , and we approximate it by a @xmath123-function , @xmath176 . * we argue that the spatial correlations of @xmath56 for two different system paths at the same @xmath177 also decay , because the averaging over many paths , and many initial environment states act like an average over @xmath88 .
we define @xmath178 as the functional form of this decay of spatial correlations , with @xmath179 .
the precise form of @xmath180 depends on details of @xmath181 , @xmath175 and @xmath56 .
in particular , the typical length @xmath5 of this decay is of order the scale on which @xmath171 changes between its maximum and minimum value . given these arguments , we have @xmath182 then , @xmath183 \nonumber \\ & & \qquad\times \big [ { \cal u}\big({\bf r}_\gamma(\tau_1),{\bf q}_\gamma(\tau_1)\big)- { \cal u}\big({\bf r}_{\gamma'}(\tau_1),{\bf q}_\gamma(\tau_1)\big ) \big ] \big\rangle_{\rm env } \nonumber \\ & = & 2 { \langle { \cal u}^2 \rangle \over \lambda_{\rm env } } \int_0^t { { \rm d}}\tau \big[1 - k(|{\bf r}_{\gamma'}(\tau)-{\bf r}_\gamma(\tau)|/\xi)\big ] \ ,
. \end{aligned}\ ] ] we further make the following step - function approximation for @xmath184 , @xmath185 where the euler @xmath186-function is one ( zero ) for positive ( negative ) arguments . in principle , this is unjustifiable for @xmath187 , however since the paths diverge exponentially from each other , the time during which @xmath188 is of order @xmath189 , while dephasing happens on a timescale @xmath8 which is typically of order the dwell time , @xmath190 .
thus the step - function approximation of @xmath180 will have corrections of order @xmath191 , which we therefore neglect .
once we have made the approximation in eq .
( [ eq : f - approx ] ) , we see that non - zero contributions to @xmath192 come from regions where the distance between @xmath80 and @xmath100 is larger than @xmath5 . we are now ready to calculate dephasing for those system paths shown in fig .
[ fig : weakloc ] .
as defined above , @xmath157 and @xmath158 are the two times at which the path @xmath100 crosses itself .
dephasing acts on the loop formed by @xmath100 and , as just argued , it acts once the distance between @xmath80 and @xmath100 is greater than @xmath5 , i.e. in the time window from @xmath193 to @xmath194 , where @xmath195 is given in eq .
( [ txi ] ) .
we average eq .
( [ eq : f ] ) over the environment and use the central limit theorem to evaluate the action phase due to the coupling between system and environment , @xmath196 \big\rangle_{\rm env } & = & \exp \big[-{{\textstyle{\frac{1}{2}}}}\langle\delta \phi^2_{\cal u}\rangle_{\rm env } \big ] \nonumber \\ & = & \exp\left [ - ( t_2-t_1-t_{\xi})/\tau_{\phi}\right ] , \label{detphiu } \end{aligned}\ ] ] where the dephasing rate is @xmath197 given that @xmath198 is defined in eq .
( [ eq : env - average ] ) , we can substitute eq .
( [ detphiu ] ) directly into eqs .
( [ eq : gwl - integral ] ) .
we thereby reduce the problem to an integral over system paths which is almost identical to the equivalent integral for @xmath149 . assuming phase space ergodicity for the system , we get the probabilities @xmath199 with @xmath200 being the real space volume occupied by the system ( the area of the cavity ) . at this point
the integral is the same as without dephasing , except that during the time @xmath201 , the inverse dwell time is replaced by @xmath202 .
thus when evaluating the @xmath203-integral , we get the extra prefactor @xmath204/(1 + \tau_{\rm d}/\tau_{\phi})$ ] compared with the equivalent integral result without dephasing .
thus we have @xmath205 since @xmath206 with @xmath207 defined in eq .
( [ eq : tau_xi ] ) , the @xmath208 dependence of the @xmath209-term drops out .
this means that @xmath210 simply differs from its value without dephasing by a constant factor , @xmath211 .
thus the integral over @xmath208 in eq .
( [ eq : gwl - integral - a ] ) is identical to the one in the absence of the environment , and takes the form@xcite @xmath212 $ ] , where we have assumed @xmath213 .
the substitution @xmath214 immediately yields a dimensionless integral and an exponential term , @xmath215 ( neglecting as usual @xmath216$]-terms in the logarithm in @xmath217 ) . from this analysis
, we find that the weak - localization correction is given by @xmath218,\ ] ] where @xmath219 is the weak - localization correction at finite-@xmath220 in the absence of dephasing , @xmath221 .
\end{aligned}\ ] ] we see that the dephasing of weak localization is not exponential with the ehrenfest time , instead it is exponential with the @xmath6-independent scale @xmath207 given in eq .
( [ eq : tau_xi ] ) . in all cases where @xmath5 is a classical scale ( i.e. of similar magnitude to @xmath222 rather than @xmath6 )
we see that @xmath207 is much less than the ehrenfest time , @xmath223 . in such cases the exponential term in eq .
( [ eq : gwl - env ] ) is much less noticeable that the universal power - law suppression of weak - localization .
we show explicitly that our semiclassical method is probability - conserving , and thus current - conserving , also in the presence of dephasing .
we do this by calculating the leading - order quantum corrections to reflection , showing that they enhance reflection by exactly the same amount that transmission is reduced .
there are two leading - order off - diagonal corrections to reflection .
the first one reduces the probability of reflection to arbitrary momenta ( weak - localization for reflection ) , while the second one enhances the probability of reflection to the time - reversed of the injection path ( coherent - backscattering ) .
the distinction between these two contributions is related to the correlation between the path segments when they hit the leads . for coherent - backscattering contributions ,
these segments are correlated ( see fig . [
fig : cbs ] ) , but for weak - localization contributions , they are not .
the derivation of the weak - localization for reflection @xmath224 is straightforward and proceeds in the same way as the derivation for @xmath225 given above , replacing the factor @xmath226 by @xmath227 .
we thus get , @xmath228,\ ] ] where @xmath229 \ ; n_{\rm l}^2 /(n_{\rm
l } + n_{\rm r})^2 $ ] is the finite-@xmath220 correction in the absence of dephasing .
( color online ) a semiclassical contribution to coherent backscattering for the system - environment model .
it involves paths which return to close , but anti - parallel to themselves at lead @xmath31 .
the two solid paths are paired ( within @xmath32 of each other ) in the cross - hatched region . here
we show @xmath145 , so the dephasing ( dotted path segment ) starts in the loop ( @xmath146 ) . in the basis
parallel and perpendicular to @xmath80 at injection the initial position and momentum of path @xmath80 at exit are @xmath230 , @xmath231 and @xmath232 .
, width=264 ] we next calculate the contributions to coherent - backscattering , extending the treatment of ref .
[ ] to account for the presence of dephasing . as before ,
the environment is treated in the diagonal approximation .
the coherent - backscattering contributions correspond to trajectories where legs escape together within @xmath233 of the encounter .
such a contribution is shown in fig .
[ fig : cbs ] .
the correlation between the system paths at injection and exit induces an action difference @xmath234 not given by the richter - sieber expression .
it is convenient to write this action difference in terms of relative coordinates at the lead ( rather than at the encounter ) .
the system action difference is then @xmath235 where the perpendicular difference in position and momentum are @xmath230 and @xmath236 . as with weak - localization
, we can identify three timescales , @xmath237 , associated with the time for paths to spread to each of three length scales , @xmath238 .
however unlike for weak - localization we define these timescales as a time measured from the lead rather than from the encounter .
thus we have @xmath239 \end{aligned}\ ] ] with @xmath240 .
writing the integral over @xmath241 as an integral over @xmath242 and using @xmath243 , the coherent - backscattering contribution is @xmath244 \big\langle f^{\rm cbs } ( { \bf y}_{0 } ) \big\rangle \ , .
\qquad \end{aligned}\ ] ] after integrating out the environment in the same manner as for weak - localization , we get @xmath245 \nonumber\\ & = & \frac{n_{\rm l}p_{\rm f } \sin \theta_0 } { \pi(n_{\rm l}+n_{\rm r } ) } { \exp\big[- ( t'_{\rm l } -t'_w/2)/\tau_{\rm d } - ( t'_{\rm l}-t'_\xi)/\tau_\phi \big ] \over 1 + \tau_{\rm d } /\tau_\phi}\ , . \ \
\ \nonumber \end{aligned}\ ] ] now we can proceed as for @xmath149 , pushing the momentum integral s limits to infinity , and evaluating the @xmath246integral over the range @xmath32 , with the help of an euler @xmath82-function .
we finally obtain @xmath247,\ ] ] in terms of @xmath248 \ ;
n_{\rm l } /(n_{\rm
l } + n_{\rm r})$ ] , the finite-@xmath220 coherent backscattering contribution in the absence of dephasing .
hence @xmath249 for all values of @xmath207 and @xmath8 , and our approach is probability- and thus current - conserving .
so far , we have only considered cases where we make a diagonal approximation for the environment . on the face of it
this seems a little unreasonable .
for instance , if the system and environment are of similar sizes then one would expect that diagonal contributions for the system and weak - localization for the environment would be as important as the contributions calculated above .
since the environment is a closed cavity , one would naively think that the weak - localization contribution for the environment should be calculated in a similar manner to the form - factor in refs .
[ ] . in the absence of coupling , the environment part of eq .
( [ sc - cond ] ) would then be @xmath250 where @xmath251 is a number of order one , and @xmath252 is the environment level - spacing ( for a two - dimensional environment containing a single particle , @xmath253 ) .
the first term above comes from the diagonal approximation used throughout this article , while the second term is a weak - localization correction .
this correction becomes of order the system s weak - localization correction on the timescale @xmath254 , so there is a priori no reason to neglect it .
the environment part of our calculation differs however from the form - factor in that it corresponds to the time - evolution of the environment during the time it takes for a particle to be transported across the system .
therefore , the sum in eq .
( [ eq : env - weak - loc ] ) is not restricted to periodic orbits , and the unitarity of the environment s time - evolution imposes that @xmath255 .
furthermore , unitarity must be preserved even in the presence of a finite-@xmath56 , as long as there is no exchange of particles between system and environment .
we thus conclude that we do not need to consider the weak - localization type corrections to the environment evolution because they cancel . throughout this article we assumed that the system - environment coupling is weak enough not to modify the classical paths in the system .
formally , this assumption can be rigorously justified by invoking theorems on structural stability @xcite .
however , care should be taken in extrapolating our results to the limit @xmath256 , since the force on the particle is the gradient of the interaction potential , @xmath257 .
we therefore estimate the minimum @xmath5 for which we can legitimately assume that classical system paths are left unchanged by the system - environment coupling
. this will give the bound on the regime of validity of our approach .
( color online ) dephasing of weak - localization when @xmath258 , and hence @xmath195 is negative , see eq .
( [ txi ] ) .
the dephasing starts and ends in the `` legs '' rather than the loop . in the language of disordered systems
this means the dephasing affects the diffusons as well as the cooperon . , width=264 ] to see significant dephasing we need @xmath259 , so we can not take the interaction strength to zero , instead we require that @xmath260 , see eq .
( [ dephi_rate ] ) .
this induces a typical force @xmath261 on the particle . to see if this noisy force significantly modifies the paths in the vicinity of the encounter , we compare it with the relative force of the chaotic system hamiltonian on the particle at the encounter .
since the perpendicular extension of the encounter is @xmath262 , and the duration of the encounter is of order the lyapunov time @xmath263 , the system force goes like @xmath264 . estimating @xmath265 as is typical of chaotic billiards , the ratio of the noisy force to the system force becomes @xmath266^{-1/2 } $ ] .
thus one can ignore the modifications of the classical paths due to the coupling to the environment , as long as @xmath267^{1/2 } .
\end{aligned}\ ] ] we see that @xmath5 can easily be less than the typical encounter size @xmath262 ( remember that @xmath268 is always assumed ) .
thus our method is not only applicable for @xmath5 up to the system size , where dephasing happens only in the loop .
it is also applicable for @xmath5 smaller than the encounter size , in which case the time @xmath195 is negative , and dephasing occurs in part of the legs as well as the whole of the loop , see fig . [
fig : negative - tau_xi ] .
finally we caution the reader that the whole semiclassical method used in this article relies on the lead width being greater than the encounter size , this requires that @xmath269 , thus we _ can not _ access the regime @xmath270 , which is dominated by stochastic diffraction at the leads . when the temperature of the electrons in the leads coupled to a chaotic system is taken to zero , there is no thermal noise in the current through the device .
however there is still the intrinsically quantum noise which originates from the wave - like nature of the electrons .
this zero - temperature noise is known as shot noise @xcite . in the absence of dephasing
, shot noise has been well - studied using rmt @xcite , quasi - classical field theory @xcite , and semiclassical methods @xcite .
it is generally argued that the shot noise is unaffected by the presence of an environment which causes dephasing but not heating of the electrons the regime of _ phase - breaking _ of ref .
[ ] . this belief is founded on the fact that ( i ) the dephasing - lead model gives a dephasing - independent shot noise , ( ii ) kinetic equations in which interference effects are ignored give the same shot noise as full quantum calculations . here
we show explicitly that , under the assumption that the system - environment coupling does not heat up the current - carrying electrons , indeed , the coupling to the environment does not affect shot noise .
we start with the formula for the zero - frequency shot noise power through a system coupled to an environment .
this formula is derived in appendix [ appendix : current - noise ] , and is given as @xmath271 in eq .
( [ shot - noise - final ] ) .
we use eq .
( [ sc - smatrix ] ) to write each matrix element as sums over classical paths .
this gives us a sum over eight paths 4 system paths and 4 environment paths as sketched in fig .
[ fig : shot - noise ] .
the system paths are as follows @xmath272 the sums over lead modes and the trace over the environment density matrix are performed in the same manner as for the conductance , [ see above eq .
( [ sc - cond ] ) ] , which results in @xmath273 here , @xmath274 , @xmath275 and we absorbed all maslov indices into the actions .
similarly , @xmath276 , @xmath277 , and @xmath278 . as argued above , in the regime of pure dephasing , @xmath106 , @xmath107 and @xmath279 are uncorrelated .
we thus first pair up the system paths to minimize @xmath106 , this pairing is the same as it would be in the absence of the environment ( compare fig .
[ fig : shot - noise ] to fig . 1 of ref .
we see from the construction of eq .
( [ shot - noise - final ] ) , that all paths reach the encounter at the same time@xcite , @xmath280 .
( color online ) sketch of typical trajectories which contribute to shot noise in the presence of an environment . in both ( a ) and ( b ) the system ( environment ) paths
are sketched above ( below ) the time axis . in ( a )
we show a contribution which will survive system averaging , because the system paths are paired almost everywhere with an encounter at time @xmath280 .
there is no constraint on the environment paths , as yet . for simplicity
we show only system paths @xmath281 and @xmath282 , with path @xmath283 and @xmath284 being the same as @xmath281 and @xmath282 except that they cross at the encounter . depending on the choice of @xmath280 with respect to the other timescales , @xmath285 ,
this pair of system paths could represent any of the contributions in fig . 1 of ref .
[ ] . in ( b ) additional constraints are imposed on the @xmath286 s , after one has integrated over all initial and final positions of the environment .
this integration removes all contributions from _ environment _ propagation when the _ system _ paths are paired .
, width=302 ] now we make the crucial observation that , for any given set of system paths , we have @xmath287 for times greater than @xmath280 .
thus for times greater than @xmath280 we can write the sum over @xmath288 in the second line of eq . ( [ eq : noise1 ] ) as @xmath289^ * { \cal k}_{\rm env } ' ( { \bf q}_1,t_1;{\bf q}_1',t_1 ' ) \qquad \label{eq : kprimes } \end{aligned}\ ] ] where we set @xmath290 to get the second line .
we define @xmath291 as the propagator for the environment evolving under an effective time - dependent potential , @xmath292 , which is the sum of @xmath293 and the potential term in @xmath175 . since both propagators in eq .
( [ eq : kprimes ] ) evolve under the same potential ( because @xmath294 ) , we can use the basic property of propagators@xcite that @xmath295^ * { \cal k}_{\rm env } ' ( { \bf q}_1,t_1;{\bf q}'',t_1 ' ) \nonumber \\ & & \qquad \qquad = { \delta}({\bf q}''-{\bf q } ' ) \label{eq : pairing - propagators } \end{aligned}\ ] ] to integrate out these propagators for times greater than @xmath280 . in their place
we have a constraint that path @xmath296 and @xmath297 must be the same at time @xmath280 .
this is sketched in fig .
[ fig : shot - noise ] ; the paths @xmath296 and @xmath297 to the right of the encounter in fig .
[ fig : shot - noise]a are replaced in fig .
[ fig : shot - noise]b by the constraint that paths @xmath296 and @xmath297 meet at time @xmath280 .
note that we can not use eq .
( [ eq : pairing - propagators ] ) to integrate out paths @xmath296 and @xmath297 for arbitrary times before @xmath280 because the system paths @xmath281 and @xmath283 are then different enough that the potential @xmath292 will be different for the two propagators in eq .
( [ eq : kprimes ] ) .
however we can use the same argument to integrate out the pair @xmath298-@xmath299 after time @xmath280 , and to integrate out the pairs @xmath296-@xmath299 and @xmath297-@xmath298 _ before _ the time @xmath280 .
after all these pairs are replaced by @xmath300-functions , we get the situation shown in fig . [
fig : shot - noise]b . focusing on @xmath5 much greater than the encounter size
, the above method can be used to integrate out the environment paths for all @xmath301 and all @xmath302 .
this leaves a single point ( the environment state at @xmath303 ) to integrate over . doing this
we see that eq .
( [ eq : noise1 ] ) reduces to @xmath304 this is identical to the shot noise formula in the absence of an environment .
thus we have completely removed the environment from the problem without affecting the shot noise of the system at all . to calculate the shot noise now ,
one simply needs to follow the derivation for a system without an environment in ref .
the result is most conveniently written in terms of the fano - factor , @xmath305 , which is the ratio of the shot noise to the poissonian noise @xmath306 , where @xmath307 is the average current .
one gets @xmath308 .
\label{eq : fano } \end{aligned}\ ] ] this is of course independent of the coupling to the environment . as a final comment , we note that above we kept only the leading @xmath309 $ ] term in the shot noise .
there is a hierarchy of weak - localization - like corrections @xmath310 $ ] , @xmath311 to this result@xcite , which are suppressed by dephasing in much the same way as the weak - localization correction to conductance .
thus for @xmath312 , we can expect the environment to cause a cross - over from the result in ref .
[ ] to the result in eq .
( [ eq : fano ] ) with @xmath313 .
hence in the classical limit of wide leads ( @xmath314 ) the environment s effect is negligible , however for narrow leads ( @xmath315 ) the environment s effect may be significant .
we have shown that , to capture the effect of dephasing on weak localization , it is sufficient to treat the environment at the level of the diagonal approximation .
we thus observe that , in the semiclassical limit of short wavelength , @xmath316 , a quantum chaotic environment has the same dephasing effect on weak - localization as the equivalent _ classical _ chaotic environment . because correlations typically decay exponentially fast in classical hyperbolic systems , this makes the effect of this classical environment very similar to a classical noise field with a suitably chosen spatial and temporal correlation function . in this section
we show that the conclusions that we draw for a quantum environment can also be drawn for a classical noise field .
one common experimental example of such a field is microwave radiation , applied to the chaotic dot either by accident or on purpose @xcite .
a second example , is the common theoretical treatment of electron - electron interactions as a source of classical ( johnson - nyquist ) noise @xcite .
we add a new term to the system hamiltonian of the form , @xmath317 .
we assume this term is weak enough that it does not affect the classical paths , but strong enough to modify the phase acquired along such paths .
the phase difference for a pair of paths contributing to the conductance is @xmath318 , where @xmath106 is given in eq .
( [ phis ] ) and @xmath319/\hbar .
\end{aligned}\ ] ] we now assume that the noise is gaussian distributed with @xmath320 here , @xmath321 gives the form of the spatial decay of the correlation function ( on a scale @xmath5 ) , and @xmath322 gives the form of the temporal decay of the correlation function ( on a scale which we call @xmath323 to make the analogy with the notation in section [ sect : weak - loc - env ] ) .
we can now follow the derivation in section [ sect : weak - loc - env ] ) by replacing @xmath324 with @xmath325 throughout .
we assume that the correlations in time are short enough to be treated as white - noise - like , @xmath326 , and that the spatial correlations decay fast enough that we can justify the approximation in eq .
( [ eq : f - approx ] ) .
this directly leads to the same result for weak - localization as in eq .
( [ eq : gwl - env ] ) , where now @xmath327 here we consider the noise generated by electron - electron interactions in a ballistic chaotic system .
in such a system , dephasing is caused by noise with momenta ( @xmath328-vectors ) larger than the inverse system size , @xmath329 .
thus the dephasing processes are the same as those for ballistic motion in disordered systems ( @xmath328-vectors greater than inverse mean free path ) .
such processes were first studied in ref . [ ] for spinless electrons , while more recently ref .
[ ] explored the full crossover from ballistic to diffusive motion for electrons with spin .
the effect of a finite ehrenfest time on such dephasing was considered in ref .
[ ] , which found that the dephasing rate in the vicinity of the encounter has a logarithmic dependence on the perpendicular distance between paths .
this led them to observe that the electron - electron interaction in a 2-dimensional ballistic system dephases weak - localization exponentially with the ehrenfest time .
we repeat their derivation here and show that @xmath330 , \label{eq : dephasing - e - e - interaction } \end{aligned}\ ] ] where @xmath331 is given by eq .
( [ eq : gwl - env ] ) with @xmath5 equaling a thermal length scale @xmath332 . ref .
[ ] neglected the @xmath331-term in the exponent since it is often small . in our qualitative derivation of this result
, we treat the electron - electron interaction as classical - noise , rather than using the perturbative field theory approach in refs .
our approach is similar in spirit to those for diffusive systems @xcite .
the screened electron - electron interaction gives a correlation function of the form @xcite @xmath333 where we assume @xmath334 , so the energy difference between a particle with momentum @xmath335 and @xmath336 is @xmath337 .
the factor of @xmath338 comes from the imaginary part of the screened coulomb interaction , is due to the polarization bubble and corresponds to the fluctuations of the electron sea at momentum and energy @xmath339 .
the @xmath123-function ensures that energy and momentum are conserved in the interaction between the system and the environment .
the function @xmath340 gives the weight of environment modes excited at energy @xmath341 . at temperature @xmath342
it is typically of the form @xmath343^{-1}$ ] .
it is convenient to write @xmath344 in terms of components parallel and perpendicular to the relevant classical path , i.e. parallel / perpendicular to @xmath336 .
then , @xmath345^{1/2 } } \times j({\omega } ) , \label{eq : v - e - e } \end{aligned}\ ] ] and we have [ eq : noise - correlator - d ] @xmath346/\hbar}\ \langle v^2_{{\bf \delta p},{\omega } } \rangle \\ & \propto & \int { { \rm d}}{\omega}j({\omega } ) \exp \big [ { { \rm i}}2{\omega}(t'-t)/\hbar \big ] \nonumber \\ & & \times { \rm re } \left[\int_{\hbar / l}^{p_{\rm f } } { { \rm d}}\delta p_\perp { \exp [ { { \rm i}}\delta p_\perp \delta r_\perp/\hbar ] \over [ \delta p_\perp^2 + ( { \omega}/v_{\rm f})^2]^{1/2 } } \right ] .
\label{eq : q - perp - int } \end{aligned}\ ] ] here , @xmath347 .
to get the second line , we wrote @xmath348 in the basis parallel and perpendicular to @xmath336 , we then inserted eq .
( [ eq : v - e - e ] ) , evaluated the @xmath349-integral and noted that @xmath350 .
( [ eq : noise - correlator - d ] ) expresses the correlator of the coulomb interaction along classical trajectories in the form convenient for our semiclassical approach , eq .
( [ eq : classical - noise - correlation ] ) . at first sight
the form of the interaction in eqs .
( [ eq : v - e - e1],[eq : v - e - e ] ) does not appear to correspond to a classical noise - field , since it is a function of the momentum , @xmath351 , of _ system _ paths .
the correlator must however be evaluated on weak localization loops , in which case one can use the fact that @xmath352 throughout the encounter region to perform the fourier transform .
thus the interaction term in eq .
( [ eq : q - perp - int ] ) is equivalent to a classical - noise field which is a single function of @xmath353 and @xmath177 for all @xmath354 .
this function must simply be chosen such that the integral over @xmath355 reduces to eq .
( [ eq : q - perp - int ] ) .
the time - dependent part of the correlator is given by the @xmath356-integral in eq .
( [ eq : q - perp - int ] ) .
we assume that the temperature is high enough , @xmath357 , that the correlation time becomes shorter than the time of flight @xmath358 through the cavity .
accordingly , we treat the noise as @xmath123-correlated in time , and set @xmath359 from now on .
we next investigate the properties of the real part of the @xmath360-integral in eq .
( [ eq : q - perp - int ] ) , giving the spatial dependence of the correlator .
we write it as @xmath361^{1/2 } } \right ] \ , , \end{aligned}\ ] ] where @xmath362 is the distance a system particle will travel on the timescale that the @xmath341-energy component of the noise fluctuates . for the ballistic model of the e - e interactions to be valid
we need that @xmath363 , but we assume that @xmath364
. we can easily evaluate this integral in the following regimes @xcite , @xmath365 , & \quad & \hbox{for } { \delta}r_\perp \ll \lambda_{\rm f } , \\
\ln [ l_{\omega}/ { \delta}r_\perp ] , & \quad & \hbox{for } \lambda_{\rm f } \ll { \delta}r_\perp \ll l_{\omega } , \\ 0 , & \quad & \hbox{for } l_{\omega}\ll { \delta}r_\perp \sim l , \end{array}\right .
\end{aligned}\ ] ] where we neglected all @xmath216$]-terms ( for example the result for @xmath366 is actually @xmath367 \lesssim { \cal o}[1]$ ] ) .
the crossover between these regimes is smooth .
we thus conclude that @xmath321 , as defined by eq .
( [ eq : classical - noise - correlation ] ) , becomes @xmath368 \over \ln [ l_{\omega}/\lambda_{\rm f } ] } \end{aligned}\ ] ] this function does _ not _ decay fast as @xmath369 grows , and thus it _ can not _ be treated in the manner we do elsewhere in this article , i.e. we can not write an equation analogous to eq .
( [ eq : f - approx ] ) .
instead we note that the dephasing rate in the vicinity of the encounter ( where @xmath370 ) goes like @xmath371 = \ln [ { \delta}r_\perp/\lambda_{\rm f}]$ ] . while the dephasing rate in the loop , @xmath2 , goes like @xmath372 \sim \ln [ l_{\omega}/\lambda_{\rm f } ] $ ] .
now we note that the integral over @xmath341 is dominated by @xmath373 , thus we can define @xmath332 and write @xmath374 = \ln [ l_{\rm t}/\lambda_{\rm f } ] - \ln [ { \omega}/k_{\rm b } t]$ ] .
we can neglect the second term as it is much smaller than the first , and thereby replace @xmath375 with @xmath376 in the above formulae . in this way we reproduce the result in ref .
[ ] , that the dephasing rate in the vicinity of the encounter is @xmath377 \over \ln [ l_{\rm t}/\lambda_{\rm f } ] } \end{aligned}\ ] ] while @xmath378 .
this is rather different from the systems considered elsewhere in this article where the dephasing rate is approximately zero for @xmath379 and approximately constant for @xmath380 .
we now calculate the effect of such a @xmath369-dependent dephasing rate .
we note that close to the encounter , @xmath381 where @xmath382 is the time measured from the encounter .
we split the dephasing into two contributions .
the first contribution is where @xmath383 ( here dephasing is time - independent , at the rate @xmath2 ) , the second is where the paths have @xmath384 ( here dephasing is time - dependent ) .
the boundary between the two contributions is at @xmath385 , where we define @xmath386 $ ] .
the lower bound on the second contribution ( @xmath387 ) is at time @xmath388 $ ] .
then the exponent induced by the dephasing is @xmath389}^{t_{\rm t}/2 } { { { \rm d}}\tau \over \tilde{\tau}_\phi \big({\delta}r_\perp \big ) } \nonumber \\ & = & \ ! - { t_2-t_1 - t_{\rm t}({\epsilon } ) \over \tau_\phi } -{\lambda^{-1 } \ln[l_{\rm t}/\lambda_{\rm f}]\over \tau_\phi } \qquad \end{aligned}\ ] ] where to evaluate the integral we defined @xmath390 .
the first term ( which comes from the dephasing in the loop ) alone would give no exponential term in the dephasing .
the integral of that term over @xmath391 gives it a form @xmath392 giving an exponent like in eq .
( [ eq : gwl - env ] ) with @xmath393 .
however we also have the second term which gives dephasing in the vicinity of the encounter , we can write it in terms of an ehrenfest time using @xmath394 = \tau_{\rm e}^{\rm cl } -{{\textstyle{\frac{1}{2}}}}\tau_{l_{\rm t}}$ ] .
summing the two terms we find that the exponential term in the dephasing goes like @xmath395 , with @xmath396 .
this is the result which we gave in eq .
( [ eq : dephasing - e - e - interaction ] ) and was found in ref .
[ ] ( neglecting the @xmath331-term ) . because @xmath6 is the scale of friedel oscillations
, one might have expected that electron - electron interactions give a noise with a correlation length @xmath270 , which would lead to a suppression @xmath397 $ ] of weak localization , instead of @xmath398 $ ] .
the factor of 2 difference between the correct result and this naive argument , is due to the fact that all scales ( i.e. all values of @xmath328 ) contribute to the noise induced by the electron - electron interactions .
sketch of a typical chaotic quantum dot in a two - dimensional electron gas ( 2deg ) .
the dot is defined by metallic gates which are biased to deplete the 2deg everywhere except in the regions defining the leads and the dot .
these gates are a distance @xmath399 above the 2deg ( with @xmath400 ) . at finite temperature
, the electrons at the surface of these metallic gates will fluctuate , leading to noise which will be felt by the electrons in the chaotic quantum dot .
, width=302 ] our aim here is to show that electron - electron interactions do _ not _ automatically lead to dephasing which is an exponential function of the ehrenfest time .
it only happens when the @xmath401-integral is divergent at its upper limit , with an upper cut - off of order @xmath402 .
if the @xmath401-integral is cut - off by some other length scale , then the dephasing of weak - localization will be independent of the ehrenfest time .
the example we consider is noise in a two - dimensional system due to electron - electron interactions between the system and the gates . a typical experimental set - up
is sketched in fig .
[ fig:3d - gates ] .
the gates are bulk metal , the electrons in the system will feel fluctuations of electrons at the surface of the gates .
one can expect that these fluctuations at the surface of the gate are sufficiently well - confined to two - dimensions ( by screening in the bulk of the gate ) , that they will cause a noise field in the system of a form similar to eq .
( [ eq : v - e - e1 ] ) . however the fact that the distance between the gates and the chaotic system is @xmath399 means that the natural upper cut - off on the @xmath401-integral will be @xmath403 and not @xmath402 .
assuming @xmath404 , we replace @xmath402 by @xmath403 throughout the derivation in section [ sect : e - e-2d ] , and find that @xmath405 , \end{aligned}\ ] ] with @xmath406 and hence @xmath407 $ ] .
the dephasing rate here is @xmath408 . in systems in which dephasing is dominated by thermal system - gate interactions
, we therefore expect a dephasing that is @xmath30-independent and algebraic in @xmath8 for @xmath409 . in real experiments the gates are typically much more disordered than the chaotic system ,
thus we can easily have a situation in which the thermally excited modes in the gates are diffusive or even localized by a charge trap at the edge of a gate in which case their noise field will be similar to that in ref . [ ] .
this modifies the integrand of the @xmath401-integral , however the upper cut - off will still be @xmath403 , therefore dephasing will again be @xmath30-independent . in general , dephasing is due to a combination of e - e interactions _ within _ the system and the coulomb coupling between the system and external charge distributions in gates and other reservoirs of charges . which of these sources of dephasing dominates is determined by microscopic details which we do not discuss here , in particular the temperatures and mean - free - paths of both system and gates @xcite .
in its simplest formulation the dephasing lead model consists of adding a fictitious lead @xmath410 to the cavity .
this is illustrated in fig .
[ fig : lead_model ] .
contrary to the two real leads l , r , the potential voltage on lead @xmath410 is tuned such that the net current through it is zero .
every electron that leaves through lead @xmath410 is replaced by one with an unrelated phase , leading to a loss of phase information without loss of current . in this situation
the conductance from l to r is given by @xcite @xmath411 where @xmath412 is the conductance from lead @xmath25 to lead @xmath413 in the absence of a voltage on lead @xmath410 .
we separate the drude and weak - localization parts of @xmath412 , @xmath414,\,\ ] ] where the drude contribution , @xmath415 , is @xmath416 $ ] and the weak - localization contribution , @xmath417 , is @xmath418 $ ] and @xmath419 is the total number of channels in this three terminal geometry .
we expand @xmath420 for large @xmath421 and collect all @xmath416$]-terms ( drude contributions ) and all @xmath418$]-terms ( weak - localization contributions ) to get @xmath422 with @xmath423 these equations form the basis of our semiclassical derivation of weak localization in the dephasing lead model .
we first consider the case of a dephasing lead perfectly coupled to the cavity , and then move on to consider a dephasing lead with a tunnel barrier of transparency @xmath424 .
we finally discuss multiple dephasing leads .
schematic of the dephasing - lead model .
the system is a open quantum dot with an extra lead ( lead 3 ) , whose voltage is chosen to make zero current flow ( on average ) in that lead .
this extra lead thereby causes dephasing without loss of particles .
, width=158 ] the drude conductance and weak - localization correction from lead @xmath25 to lead @xmath413 in a three - lead cavity are @xmath425 \end{aligned}\ ] ] where @xmath426 , in terms of @xmath427 , the time of flight across the system .
we substitute these results into eq .
( [ 3leadmodeldr ] ) and eq .
( [ 3leadmodelwl ] ) , and write the answer in terms of the dwell time in the two lead ( l and r ) geometry , @xmath190 , and the dephasing rate , @xmath2 , which we define as the decay rate to lead @xmath410 , @xmath428 we have @xmath429 , and from this we find the drude conductance and weak - localization correction , @xmath430 .
\end{aligned}\ ] ] here @xmath431 and @xmath432 are the results for a two lead cavity in the absence of dephasing , eqs .
( [ eq : drudecond ] ) and ( [ eq : gwl0 ] ) the weak - localization correction with a dephasing lead has a similar structure to that with a real environment .
however here the time scale involved in the additional exponential suppression contains no independent parameter analogous to @xmath5 .
we could have expected that the width of the dephasing lead would play a role similar to @xmath5 .
however this turns out not to be the case , instead the fermi wavelength appears in place of @xmath5 , so the time scale in the additional exponential suppression is the ehrenfest time , @xmath220 . in the absence of tunnel barrier , we can go further and calculate conductance fluctuations at almost no extra cost . from eqs .
( [ 3leadgeo ] ) and ( [ splitdrude - wl ] ) , we get the following expression for the variance of the conductance to order @xmath418 $ ] @xmath433 in the universal regime , ref .
[ ] gives , to order @xmath418 $ ] in the presence of time - reversal symmetry , @xmath434 ref .
[ ] showed that eq .
( [ ucf : baranger ] ) remains valid even at finite @xmath435 .
inspection of their calculation for @xmath436 shows that the same conclusion also applies to @xmath437 , and thus eq .
( [ covar : baranger ] ) still holds independently of @xmath435 . together with eq .
( [ tdandtphi ] ) , this straightforwardly leads to @xmath438 in the universal regime , this result was previously derived in ref .
[ ] . we thus conclude that , in the dephasing lead model without tunnel - barrier , conductance fluctuations exhibit the universal behavior of eq .
( [ varg_double_lorentzian ] ) .
below , we confirm this result numerically . putting a tunnel - barrier on the dephasing lead @xmath410
is attractive because one can avoid the local character of the dephasing lead model by considering a wide third lead with an almost opaque barrier @xcite .
additionally , this is the model studied numerically in ref .
[ ] in the context of conductance fluctuations .
weak - localization and shot noise in this model have been considered within the trajectory - based semiclassical approach in ref .
[ ] , and here we only mention the main results . according to ref .
[ ] , when all leads are connected to the cavity via tunnel - barriers , with the barrier on lead @xmath25 having transparency @xmath439 $ ] , the drude conductance between lead @xmath25 and @xmath413 , @xmath440 , and the weak - localization correction , @xmath417 , are @xmath441 .
\quad \label{gijwl } \end{aligned}\ ] ] here , @xmath442 and @xmath443 are the single path and the paired paths survival times respectively , @xmath444 is the width of lead @xmath413 , @xmath445 and @xmath446 .
now we assume that @xmath447 so only the dephasing lead has a tunnel barrier . substituting the drude and weak - localization contributions into eq .
( [ 3leadmodelwl ] ) we find that @xmath448 .
\qquad \end{aligned}\ ] ] the argument in the exponent in eq .
( [ eq : gwl3lead ] ) has a simple physical meaning .
it is the probability that a path survives throughout the paired - region ( @xmath449 on either side of the encounter ) without escaping into lead @xmath410 , multiplied by the probability to survive the extra time @xmath450 unpaired without escaping into lead @xmath410 ( to close a loop of length @xmath220 ) .
the first probability is @xmath451 $ ] while the second is @xmath452 $ ] .
we note that if we consider a nearly opaque barrier , the relevant time scale involved in the exponent is @xmath453 .
thus by tuning the opacity of the barrier , we can vary the exponential contribution to dephasing from @xmath398 $ ] to @xmath454 $ ] , but we can not remove the exponent . in particular , we can not mimic dephasing due to a real environment with @xmath455 , eq .
( [ eq : gwl - env ] ) , since it has only a power - law dephasing . there is to date no theory for conductance fluctuations at finite ehrenfest time in the presence of tunnel barriers , and constructing such a theory would require a formidable theoretical endeavor .
numerically , ref .
[ ] observed that , in contrast to the dephasing lead model without tunnel - barrier , conductance fluctuations in the presence of a dephasing lead with a tunnel barrier are exponentially damped @xmath37.$ ] the @xmath413 probe dephasing model consists of adding @xmath413 fictitious leads to the cavity ( labeled @xmath456 ) in addition to lead @xmath457 , @xmath458 .
the voltage on each supplementary lead is tuned so that the current it carries is zero . without loss of generality we defined @xmath459
, then we get the set of equations @xmath460 where the superscript-@xmath342 indicates the transpose .
the column - vectors @xmath461 , @xmath462 have an @xmath463th element given by the current or voltage ( respectively ) for the dephasing lead @xmath464 .
the column - vectors @xmath465 and @xmath466 have an @xmath463th element given by @xmath467 and @xmath468 , respectively .
finally @xmath469 has matrix elements given by @xmath470_{ij } & = & n_i { \delta}_{ij } - t_{ij } \\ & = & \left [ { \textstyle \sum_{k\neq j}}t_{kj}\right ] { \delta}_{ij } - t_{ij}(1-{\delta}_{ij } ) \ , , \end{aligned}\ ] ] where again @xmath471 . substituting @xmath462 from eq .
( [ eq : multi - dephasing - b ] ) into eq .
( [ eq : multi - dephasing - a ] ) and using @xmath472 gives us the conductance from @xmath457 to @xmath458 as @xmath473 thus finding @xmath420 requires the inversion of @xmath469 .
this is cumbersome , so instead we present a simple argument to extract only the information we are interested in the nature of the exponential in the dephasing .
we argue that whatever the formula for conductance for @xmath413 dephasing leads is , we can expand it in powers of @xmath79 and collect the @xmath474$]-terms to get a formula for weak - localization of the form @xmath475 here , the sum is over all dephasing leads . to get the prefactors
@xmath476 we would have solved the full problem by inverting @xmath469 , however we can already see that , to leading order , they are combinations of drude conductances and thus independent of the ehrenfest time .
in contrast all the weak - localization contributions contain an exponential of the form [ @xmath477 and @xmath478 are defined below eq .
( [ gijwl ] ) ] @xmath479 \ , .
\end{aligned}\ ] ] thus defining @xmath2 as the rate of escaping into any of the dephasing leads , so @xmath480 , we see that @xmath225 decays with an exponential @xmath481 , \end{aligned}\ ] ] where we define @xmath482 such that @xmath483 .
we have just shown that multiple dephasing leads cause an exponential suppression of the weak - localization which is qualitatively similar to that caused by a single dephasing lead .
the exponent is proportional to the ehrenfest time , and contains no independent parameter analogous to @xmath5 .
because of the slow , logarithmic increase of @xmath30 with the size @xmath484 of the hilbert space , the ergodic semiclassical regime @xmath485 , @xmath52 is unattainable by standard numerical methods . we therefore follow refs . [ ] and consider the open kicked rotator model .
the system is described by the time - dependent hamiltonian @xmath486 the kicking strength @xmath184 drives the dynamics from integrable ( @xmath487 ) to fully chaotic ( @xmath488 ) .
the lyapunov exponent in the classical version of the kicked rotator is given by @xmath489 .
in most quantum simulations , however , one observes an effective lyapunov exponent @xmath490 instead that is systematically smaller than @xmath53 by as much as 30% @xcite .
two parameters , @xmath491 and @xmath492 , are introduced to break two parities and drive the crossover from the @xmath20 to the @xmath21 universality class @xcite , corresponding to breaking the time reversal symmetry .
we quantize the hamiltonian of eq .
( [ krot ] ) on the torus @xmath493 $ ] , by discretizing the momentum coordinates as @xmath494 , @xmath495 . a quantum representation of the hamiltonian of eq .
( [ krot ] ) is provided by the unitary @xmath496 floquet operator @xmath57 , which gives the time evolution for one iteration of the map . for our specific choice of the kicked rotator , the floquet operator has matrix elements @xmath497 } \\ & \times & \sum_m e^{2 \pi
i m(l - l')/m } e^{-(imk/2\pi ) \cos[2\pi ( m - m_0)/m ] } , \nonumber \end{aligned}\ ] ] with @xmath498 and @xmath499 .
the hilbert space size @xmath484 is given by the ratio @xmath28 of the linear system size to the fermi wavelength .
to investigate transport , we open the system by defining @xmath500 contacts to leads via absorbing phase - space strips @xmath501 $ ] , @xmath502 . with @xmath503
, we construct a @xmath504 scattering matrix from the floquet operator @xmath57 as @xcite @xmath505^{-1 } u { \mathcal p}^\dagger.\ ] ] the @xmath506 projection matrix @xmath507 , which describes the coupling to the leads , has matrix elements @xmath508 where @xmath509 $ ] gives the transparency of the contact to the external channel @xmath510 , and @xmath511 denotes the ensemble of cavity modes coupled to the external leads . below we focus on perfectly transparent contacts , @xmath512 , @xmath513 . the conductance ( [ 3leadmodeldr ] ) in the dephasing lead model is obtained from eq .
( [ smatrix ] ) with @xmath514 leads .
the transport leads carry @xmath515 channels , which defines the dwell time through the system as @xmath516 .
the dephasing time @xmath517 is defined by the number @xmath23 of channels carried by the third , dephasing lead , and the ehrenfest time is given by @xmath518.\ ] ] we extend the kicked rotator model to account for the coupling to external degrees of freedom .
the exponential increase of memory size with number of particles forces us to focus on an environment modeled by a single chaotic particle .
therefore , we follow ref . [ ] and consider two coupled kicked rotators ( with @xmath519 ) @xmath520 in this model , the interaction potential @xmath56 acts at the same time as the kicks , which facilitates the construction of the @xmath45-matrix . for this particular choice of interaction ,
the correlation length @xmath521 so that one expects a universal behavior of dephasing , as in eq .
( [ eq : dephasing - wl - without - ehrenfest ] ) .
the quantum representation of the coupled hamiltonian is a unitary @xmath522 floquet operator .
we open the system ( and not the environment ) to two external leads by means of extended projectors @xmath523
. a straightforward generalization of eq .
( [ smatrix ] ) defines a @xmath524 extended scattering matrix , from which we evaluate the conductance via eq .
( [ conductance ] ) .
we focus on the symmetric situation where the two leads carry the same number @xmath79 of channels and average our data over a set of pure but random initial environment density matrices @xmath525 .
we estimate the dephasing time from the @xmath526-induced broadening of two - particle levels in the corresponding closed two - particle kicked rotator , @xmath527 , with @xmath528 , the effective planck constant @xcite .
a ) magnetoconductance curves @xmath529 for the double kicked rotator model ( see text ) with @xmath530 ( @xmath531 ) , @xmath532 , @xmath533 and hilbert space sizes @xmath534 , @xmath535 .
different symbols correspond to different dephasing times @xmath536 ( @xmath537 , circles ) , @xmath538 ( @xmath539 , squares ) , @xmath540 ( @xmath541 , diamonds ) , @xmath542 ( @xmath543 , downward triangles ) and @xmath544 ( @xmath545 , upward triangles ) .
data are averaged over @xmath546 different lead positions , each with @xmath546 different quasi - energies and @xmath547 different initial environment states .
b ) magnetoconductance curves for the open kicked rotator with transparent dephasing lead ( see text ) with @xmath548 , @xmath532 and hilbert space size @xmath549 .
different symbols correspond to different dephasing times @xmath536 ( circles ) , @xmath550 ( squares ) , @xmath551 ( diamonds ) , @xmath552 ( downward triangles ) and @xmath553 ( upward triangles ) .
data are averaged over @xmath554 different lead positions , each with @xmath555 different quasi - energies.,width=302 ] ( color online ) amplitude @xmath556 of the weak localization correction to the conductance as a function of @xmath557 for the double kicked rotator model ( circles ) and the open kicked rotator with transparent dephasing lead ( squares ) . the black line gives the universal algebraic behavior @xmath558 , and the red line is a guide to the eye , including an exponential decay with @xmath559 , on top of the universal decay .
, width=302 ] to investigate weak - localization , we follow the same procedure as in ref . [ ] of taking a constant nonzero @xmath491 , while varying @xmath492 ( which hence plays the role of a magnetic field ) .
the obtained magnetoconductance in the absence of dephasing is lorentzian @xcite , @xmath560 with @xmath561 . in fig .
[ fig : weakloc_num ] we compare the suppression of weak - localization for the system - environment kicked rotator [ panel a ) , top ] and the dephasing lead kicked rotator [ panel b ) , bottom ] . for both models
, we show five magnetoconductance curves , corresponding to five different ratios @xmath562 .
all curves exhibit the expected lorentzian behavior vs. @xmath563 , however , the amplitude of the magnetoconductance is reduced as @xmath562 is reduced .
this allows one to extract the @xmath8-dependence of @xmath225 . for the system - environment model
, we found no significant departure from our analytical prediction , eq .
( [ eq : gwl - env ] ) with @xmath564 ( since the interaction in eq .
( [ krotcoupled ] ) has @xmath521 ) .
the same behavior is observed for the dephasing lead model , as long as @xmath562 is large , however , compared to the system - environment kicked rotator , the damping of magnetoconductance accelerates as @xmath565 becomes comparable to or smaller than one . when this regime is reached
, magnetoconductance curves for the dephasing lead model lies significantly below those of the system - environment model , even when the latter corresponds to shorter dephasing times ( compare in particular the upward triangles in both panels of fig .
[ fig : weakloc_num ] ) .
this behavior is further illustrated in fig .
[ dg(2.4 ) ] , where we plot the amplitude @xmath566 of the weak localization corrections to the conductance as a function of @xmath557 . the data for the double kicked rotator nicely line up on the universal algebraic behavior @xmath558 without any fitting parameter .
this is clearly not the case for the dephasing lead model , where an additional , exponential dependence on @xmath8 emerges .
we attribute this to the exponential damping factor @xmath37 $ ] of eq .
( [ eq : weakloc_3lead ] ) . the data presented in figs . [
fig : weakloc_num ] and [ dg(2.4 ) ] for @xmath567 exhibit a weak dependence on @xmath568 $ ] only , with @xmath569 , which we attribute to terms of order one in eq .
( [ taueorder1 ] ) .
all collected data ( including some that we do not present here ) thus confirm qualitatively if not quantitatively the validity of eq .
( [ eq : weakloc_3lead ] ) for the dephasing lead model .
conductance fluctuations were studied numerically in ref . [ ] for the dephasing lead model with a tunnel barrier of low transparency , and an exponential damping @xmath570 $ ] was reported . instead , here we consider a model in which the dephasing lead is transparently coupled to the system . in fig .
[ fig : ucf ] , we show data for @xmath571 against @xmath562 , which is varied only by varying the width of the third , dephasing lead .
there are four data sets ( empty symbols ) corresponding to a fixed classical configuration at different stages in the quantum - classical crossover , i.e. with increasing ratio @xmath572 $ ] .
as @xmath484 increases , so does @xmath573 , however no change of behavior of @xmath571 is observed .
these data are compared to a fifth set obtained in the universal regime , @xmath574 , and the universal prediction of eq .
( [ varg_double_lorentzian ] ) ( dashed line ) .
these data confirm our analytical result , eq .
( [ varg_double_lorentzian ] ) , that @xmath571 exhibits no ehrenfest time dependence for the dephasing lead model with perfectly transparent contacts .
variance of the conductance vs. @xmath557 for the open kicked rotator with @xmath575 and @xmath576 ( empty symbols ) , transparently coupled to a dephasing lead .
different symbols correspond to different hilbert space sizes ( and hence different @xmath220 ) @xmath577 ( squares , @xmath578 ) , @xmath579 ( diamonds , @xmath580 ) , @xmath581 ( upward triangles , @xmath582 ) and @xmath583 ( downward triangles , @xmath584 ) .
additional data for @xmath585 , @xmath586 and @xmath581 are also shown ( full circles , @xmath587 ) .
the dashed line shows the universal behavior of eq .
( [ eq : dephasing - ucf - without - ehrenfest ] ) . unlike for weak - localization
( see fig .
[ fig : weakloc_num ] ) and for the dephasing lead model with partial transparency @xcite , the behavior of @xmath588 remains universal and shows no noticeable dependence on @xmath589 .
data are averaged over 50 different quasi - energies and from 50 ( for @xmath590 ) to 500 ( for @xmath591 and 512 ) different lead positions.,width=302 ]
we have investigated the dephasing properties of open quantum chaotic system , focusing on the deep semiclassical limit where the ehrenfest time is comparable to or larger than the dwell time through the system .
we treated three models of dephasing . in the first one ,
the transport system is capacitively coupled to an external quantum chaotic system . for that model
, we developed a new scattering formalism , based on an extended scattering matrix @xmath66 , including the degrees of freedom of the environment .
transport properties are extracted from @xmath66 , once the environment has been traced out properly . in that model
, we find that , in addition to the universal algebraic suppression @xmath592 with the dwell time @xmath190 through the cavity and the dephasing rate @xmath2 , weak - localization is exponentially suppressed by a factor @xmath593 $ ] , with a new time scale @xmath207 depending on the correlation length of the coupling potential between the system and the environment .
the second model we treated is that of dephasing due to a classical noise field . we show that the new time scale @xmath207 plays the same role here as in the system - environment model .
we then consider a classical johnson - nyquist noise model of electron - electron interactions . we show that @xmath270 ( and so @xmath207 equals the ehrenfest time ) when dephasing is dominated by electron - electron interactions within the system , but that @xmath594 when dephasing is dominated by interactions between electrons in the system and those in a gate , a distance @xmath399 away .
the third model we treated is the dephasing lead model .
we found a similar exponential suppression of weak - localization . to our surprise , however , it is the fermi wavelength , not the dephasing - lead s width , which plays a role similar to @xmath5 in that model .
this inequivalence between the dephasing lead model and dephasing due to a real environment or classical noise can be most clearly seen in a situation where the interaction with a real environment has @xmath595 .
then the environment induces only power - law dephasing , which is impossible to mimic with a dephasing lead . for the dephasing lead model
, we also showed analytically and numerically that conductance fluctuations exhibit only power - law dephasing if the connection between the cavity and the dephasing lead is perfectly transparent .
this is to be contrasted with the exponential dephasing observed for the dephasing lead model with tunnel barrier , and reflects the fact that the presence of tunnel barriers violates a sum rule otherwise preserving the universality of conductance fluctuations vs. @xmath435 @xcite .
related results have been obtained for conductance fluctuations in ref .
[ ] , where different behaviors have been predicted for external sources which give similar dephasing as our environment model and internal source of dephasing which qualitatively reproduce the prediction of ref . [ ] .
we thank i. aleiner , p. brouwer , m. bttiker and m. polianski for useful and stimulating discussions . while working on this project , c. petitjean was supported by the swiss national science foundation , which also funded the visits to geneva of p.jacquod and r.whitney .
p. jacquod expresses his gratitude to m. bttiker and the department of theoretical physics at the university of geneva for their hospitality .
this work was initiated in the summer of 2006 at the aspen center for physics .
here we extend the scattering approach to transport to account for those environmental degrees of freedom which couple to the system being studied .
we follow the lines of the derivation of the expression for noise presented in ref . [ ] , focusing on the two - terminal configuration .
the following derivation is valid in the limit of pure dephasing , when there is no energy / momentum exchange between system and environment . in the presence of an environment , the current operator at time @xmath177 on a cross - section deep inside of lead @xmath596 ( where there is no system - environment interaction ) reads @xmath597 \times { \mathbb i}_{\rm env}. \end{aligned}\ ] ] the second quantized operators @xmath598 and @xmath599 create and destroy incoming and outgoing system particles respectively .
since the environment particles carry no current , the current operator acts as the identity operator @xmath600 in the environment sub - space .
we could write @xmath600 in terms of second quantized operators , however for our present purpose , it is more convenient to write it as @xmath601 as in ref .
[ ] we now back evolve the outgoing - states into incoming - states , this time with an @xmath45-matrix which also depends on the coordinates of the environment , @xmath602 here , @xmath603 gives the transmission amplitude from channel @xmath604 in lead @xmath605 with energy @xmath606 to channel @xmath413 in lead @xmath607 with energy @xmath608 , while simultaneously , the environment evolves from @xmath103 to @xmath88 .
we can set @xmath609 , because throughout this article we only consider the regime of pure dephasing . using ( [ extended - s ] )
we rewrite the current operator as @xmath610 .
\end{aligned}\ ] ] we next rewrite the first line of eq .
( [ eq : current - op ] ) as @xmath611 finally , we write the current operator in terms of the initial environment density - matrix @xmath612 , @xmath613 where we defined the reduced current operator as the current is obtained by taking the expectation value of the current operator over the system , using @xmath615 where @xmath616 is the fermi function in lead @xmath605 .
the unitarity of @xmath66 implies , @xmath617 we use this latter equality to rewrite the kronecker @xmath123 s in eq .
( [ operator - contractions ] ) .
finally the current in the left lead is @xmath618 \eta_{\rm env}({\bf q}_0 , { \bf q}'_0).\ ] ] in the limit of zero temperature in the leads , and assuming that the scattering matrix is not too strongly energy dependent , eq .
( [ current ] ) leads to the linear conductance @xmath619 with scattering matrices to be evaluated at the fermi energy . from eqs ( [ current ] ) and ( [ conductance ] ) , we see that both current and conductance are obtained by tracing over the environmental degrees of freedom of the square of the extended scattering matrix . besides this prescription , these two equations are extremely similar to their counterpart in the standard scattering approach to transport
. we also note that conductance fluctuations can be obtained by squaring eqs .
( [ current ] ) and ( [ conductance ] ) .
it is legitimate to expect that a complex environment such as the chaotic system considered in this article has a complicated initial wavefunction , which , under ensemble averaging , is uncorrelated with itself on all scales greater than the environment wavelength .
this justifies us treating the initial environment state as @xmath620 .
we follow similar steps as in the previous section to calculate the zero - frequency current noise .
however now we have two current operators , and hence two creation and two annihilation operators for the system @xcite .
the environment has two @xmath600 operators , each of which we write in the form @xmath621 to get the current time - correlator as @xmath622 the reduced current operator @xmath623 is given in eq .
( [ reduced - current - operator ] ) .
the zero - frequency noise power is obtained from the product of the deviations from the average current at times @xmath624 and @xmath177 .
consequently it is proportional to @xmath625 there is only one trace over the environment here because we assume we measure the current as a function of time in a given experiment ( with a given initial @xmath626 ) , and then extract the average current ( and the deviations from it ) from that data set .
we need to take the following expectation value of products of creation / annihilation operators over the system@xcite @xmath627 , \end{aligned}\ ] ] where the minus ( plus ) sign stands for fermions ( bosons ) . from this
we finally get the zero - frequency noise power @xmath628 + [ 1\mp f_\alpha(e ) ] f_\beta(e ) \big ) \
\eta_{\rm env}({\bf q}_{01},{\bf q}'_{01 } ) . \end{aligned}\ ] ] all the relevant information for shot noise is contained in the diagonal @xmath629 , @xmath596 .
shot noise is obtained by calculating this latter expression in the limit of low temperature but finite voltage bias @xmath630 between the two leads . in that case
, it is easily checked that the contribution to @xmath631 arising from the first term on the right - hand side of eq .
( [ btensor ] ) does not contribute , and one gets @xmath632 + [ 1\mp f_\alpha(e ) ] f_\beta(e ) \big ) \eta_{\rm env}({\bf q}_{01},{\bf q}'_{01}).\qquad \end{aligned}\ ] ] we finally assume a slow dependence of @xmath66 on @xmath608 , in which case the integral over the energy is easily performed , giving a factor @xmath633 .
for a two - lead device ( l , r ) , we find that @xmath634 b.l .
altshuler , a.g .
aronov , and d. khmelnitsky , j. phys .
c * 15 * , 7367 ( 1982 ) ; b.l .
altshuler and a.g .
aronov , in _ electron - electron interactions in disordered systems _ , a.l .
efros and m. pollak eds .
, north holland ( 1985 ) .
s. chakravarty and a. schmid , phys . rep . *
140 * , 195 ( 1986 ) .
baranger and p.a .
mello , waves in random media * 9 * , 105 ( 1999 ) .
brouwer and c.w.j .
beenakker , phys .
b * 55 * , 4695 ( 1997 ) ; * 66 * , 209901(e ) ( 2002 ) .
aleiner , b.l .
altshuler , and m.e .
gershenson , waves in random media * 9 * , 201 ( 1999 ) . h. fukuyama and e. abrahams , phys .
b * 27 * , 5976 ( 1983 ) .
i.l . aleiner and a.i .
larkin , phys .
b * 54 * , 14423 ( 1996 ) . b.n .
narozhny , g. zala , and i.l .
aleiner , phys .
b * 65 * , 180202(r ) ( 2002 ) m. bttiker , phys . rev .
b * 33 * , 3020 ( 1986 ) .
de jong and c.w.j .
beenakker physica a * 230 * , 219 ( 1996 ) m.g .
vavilov and i.l .
aleiner , phys .
b * 60 * , r16311 ( 1999 ) .
g. seelig and m. bttiker , phys .
b * 64 * , 245313 ( 2001 ) . m.l .
polianski and p.w .
brouwer , j. phys . a : math gen . * 36 * , 3215 ( 2003 ) . c. texier and g. montambaux , phys .
b * 72 * , 115327 ( 2005 ) .
j. tworzydlo , a. tajic , h. schomerus , p.w .
brouwer , and c.w.j .
beenakker , phys .
lett . * 93 * , 186806 ( 2004 ) .
a. altland , p.w .
brouwer , and c. tian , phys .
. lett . * 99 * , 036804 ( 2007 ) . c. petitjean , p. jacquod and r.s .
whitney , jetp lett .
* 86 * , 736 ( 2007 ) . strictly speaking , dephasing and voltage probe models are fully equivalent in the one channel limit
only , however , they also lead to the same results in the limit of low temperature and frequency .
see h. frster , p. samuelsson , s. pilgram , m. bttiker , phys .
b * 75 * , 035340 ( 2007 ) . at finite @xmath30 , the ergodic hypothesis breaks down and sample - to - sample conductance fluctuations are dominated by classical effects .
they lose their universality and increase by orders of magnitude as @xmath435 becomes sizeable .
see refs .
@xcite .
more precisely , the encounter introduces a typical length scale @xmath636 , that corresponds to the perpendicular distance between the two paths in the vicinity of the encounter .
assuming hyperbolic dynamics , we get @xmath637 , and the exact definition of a typical cut - off time is given by @xmath638 $ ] , with @xmath639 .
this was not the case in ref .
[ ] where we used the time - independence of the problem to shift paths @xmath640 in time , so that all paths entered the cavity at the same time .
such a shift is unhelpful when the environment is present , so we do not do it here . for @xmath641
, we evaluate the integral by approximating @xmath642 . for @xmath643
, we split the integral into two at an arbitrary value @xmath644 , where @xmath645 . for the integral from @xmath646 to @xmath644
we approximate @xmath642 , while for the integral from @xmath644 to @xmath647 we approximate the denominator by @xmath648 . in the latter case
we get an exponential integral , @xmath649 , whose properties at @xmath650 and @xmath651 are well - known . the arbitrary cut - off , @xmath644 , drops out of the final result , as it should . finally for @xmath652 , we can use integration - by - parts to perform an asymptotic expansion for large @xmath653 .
this has been first discussed in : p.g .
silvestrov , j. tworzydlo , and c.w.j .
beenakker , phys .
e * 67 * , 025204(r ) ( 2003 ) .
for instance , numerical investigations of dynamical echoes observe a @xmath490 that is smaller than @xmath654 because @xmath655 \rangle \ne \exp [ - \langle\lambda \rangle t]$ ] , see e.g. : c. petitjean , d.v .
bevilaqua , e.j .
heller , and ph .
jacquod , phys .
98 * , 164101 ( 2007 ) . | we investigate the effect of dephasing / decoherence on quantum transport through open chaotic ballistic conductors in the semiclassical limit of small fermi wavelength to system size ratio , @xmath0 .
we use the trajectory - based semiclassical theory to study a two - terminal chaotic dot with decoherence originating from : ( i ) an external closed quantum chaotic environment , ( ii ) a classical source of noise , ( iii ) a voltage probe , i.e. an additional current - conserving terminal .
we focus on the pure dephasing regime , where the coupling to the external source of dephasing is so weak that it does not induce energy relaxation .
in addition to the universal algebraic suppression of weak localization , we find an exponential suppression of weak - localization @xmath1 $ ] , with the dephasing rate @xmath2 .
the parameter @xmath3 depends strongly on the source of dephasing . for a voltage probe ,
@xmath3 is of order the ehrenfest time @xmath4 $ ] .
in contrast , for a chaotic environment or a classical source of noise , it has the correlation length @xmath5 of the coupling / noise potential replacing the fermi wavelength @xmath6 .
we explicitly show that the fano factor for shot noise is unaffected by decoherence .
we connect these results to earlier works on dephasing due to electron - electron interactions , and numerically confirm our findings . |
n - body simulations are one of the most important tools by which contemporary theoretical cosmologists try to investigate the evolution of the large scale structure of the universe . due to the long - range nature of the gravitational force ,
the number of particles required to reach a significant mass resolution is a few orders of magnitude larger than those allowed even by present - day state - of - the - art massively parallel supercomputers ( hereafter mpp ) .
work- and data - sharing programming techniques are customary tools exploited in the development of many parallel implementations of the n - body problem @xcite @xcite @xcite .
the most popular algorithms are generally based on grid methods like the @xmath0 .
the main problem with this method lies in the fact that the grid has typically a fixed mesh size , while the cosmological problem is inherently highly irregular , because a highly clustered large - scale structure develops starting from a nearly homogeneous initial distribution . on the other hand
the barnes & hut ( 1986 ) oct - tree recursive method is inherently adaptive , and allows one to achieve a higher mass resolution than grid - based methods when clusters of galaxies form . because of these features , however
, the computational problem can easily run into unbalance causing a performance degradation .
for this reason , we have undertaken a study of the optimal work- and data - sharing distribution for our parallel treecode .
+ our work- and data - sharing parallel tree - code ( hereafter wdsh - ptc ) is based on this algorithm tree scheme , which we have modified to run on a shared - memory mpps @xcite @xcite .
we have adopted the cray research corporation craft environment @xcite to share work and data among the pes involved in the run .
+ to optimize the performances of the wdsh - ptc in order to run simulations with a very high number of particles , we have carried out a study on the optimal data distribution in the t3d global memory @xcite .
the obtained results allow us to determine an optimal strategy for the dynamic load balance ( dlb ) , that is the main purpose of this work .
generally all tests and measurements were carried out using a @xmath1 value equal to 0.8 , considering this value as the optimal value for the simulations that we would like to run .
the barnes - hut tree algorithm is a @xmath2 procedure to compute the gravitational force through a hierarchical subdivision of the computational domain in a set of cubic nested regions .
the evolution of the system involves the advancement of the trajectories of all particles , and this is carried out through a discrete integration of the trajectories of each particle . at each timestep
the force and the trajectory of each body are updated . in practice one
does not adopt the `` nude '' gravitational potential , in order to avoid the formation of binary systems , but a potential smoothed on a sufficiently small scale .
a more detailed discussion on the bh tree method can be found in @xcite . + for our purposes , we can distinguish three main phases in each timestep .
the first is the tree_formation ( hereafter tf ) phase where the tree structure is built starting from the whole computational domain included in a cubic region , that is the `` root - cell '' ( i.e. the zero level ) of the tree . using the orb ( orthogonal recursive bisection ) technique , at the begining the root - cell
is subdivided into 8 cubic nested regions ( subcells ) each including a portion of the computational domain .
this subdivision creates the second level of the tree .
the orb technique is then recursively applied to each new subcell , so that several levels of the tree are formed until all the final cells contain at most one body .
tree cells containing more than one body are called `` internal cells '' ( icells ) , and those containg only one body are called `` terminal cells '' ( fcells ) . for each icell the total mass , the center of mass and the quadrupole moment are computed .
+ the second phase is the force_compute ( hereafter fc ) , during which the forces acting on each body of the system are computed . in the tree_inspection ( hereafter ti ) subphase , for each body one makes an `` interaction list '' containing pointers to cells with which the particle will interact , formed according to the following criteria . starting from the root - cell the @xmath3 ratio is compared with a threshold parameter @xmath1 ( the opening angle parameter ) , `` c '' being the cell - size and `` d '' the distance between the particle and the cell center of mass .
+ if the @xmath3 ratio is smaller than @xmath1 , the cell is considered as a single region that contributes a force component to the particle , and its subcells are not further investigated .
otherwise one checks all the subcells using the same criterion , until one arrives at cells containing only one particle . at the end , all the tree cells satisfying the criterion form the `` body interaction list '' for the given body .
+ in general , using the value @xmath1 = 1 , the acceleration approximation has an error of about 1% .
typical values of @xmath1 for cosmological simulations are in the range 0.5 - 1.2 : the larger the @xmath1 the smaller the length of the interaction list .
the average length of the `` interaction list '' is proportional to @xmath4 , so that the total computation complexity in a serial code based on the barnes - hut tree code scales as @xmath5 .
after the ti subphase , using the interaction list , the acc_components ( hereafter ac ) subphase is executed and the particle acceleration is computed . + at the end of the fc phase there is a synchronization point followed by the update_position phase , the last phase of the timestep , when the bodies positions are updated ( fig .
the tests carried out on the original serial n - body code , kindly provided to us by dr .
l. hernquist , confirm the results shown by salmon ( 1990 ) concerning the complexity .
in particular the total time spent in the tf phase ranges from 7% to 10% of the overall time according to the particles initial condition ; i.e. , uniform or clustered distributions .
whereas the ti ranges from 65% to 70% , the ac subphase last for the remaining 20% - 25% of the total timestep .
the wdsh - ptc uses the work - sharing technique in order that all pes cooperate in the tf phase , during which the tree is formed and the cell properties are computed , and the synchronization is implicit @xcite .
+ during the fc phase each pe computes the acceleration components for each body in asynchronous mode and only at the end of the phase an explicit barrier statement is set .
+ the wdsh - ptc parallelism level reached is very high , more than 90% of the work is performed in a parallel region as shown by apprentice , a `` performance analysis tool '' designed for cray mpp systems that allows one to investigate the effective performances reached by a code . with various pool configurations of pes ranging from 16 to 128 , our results show that the most time - consuming phases ( tf , ti and ac ) are executed in a parallel regime . + in the following sections we will discuss the performances obtained using different ways for data distribution in the memory and how a strategy of dynamical load balancing ( hereafter dlb ) can be devised .
several strategies to share data in the t3d pes memory @xcite @xcite were investigated to find the best data distribution , in order to maximize the number of computed particles per second .
we have considered the two main kinds of data , tree data ( cells pointers and cells properties ) and body data , and we have observed the code performances by varying the tree data and the body data distribution .
+ tests were carried out , fixing the constraint that each pe executes the fc phase only for all bodies residing in the local memory .
a bodies data distribution ranging from contiguous blocks ( coarse grain : cdir$ shared pos(:block , : ) ) to a fine grain distribution ( cdir$ shared pos(:block(1 ) , : ) ) was adopted .
we studied different tree data distributions ranging from assigning to contiguous blocks a number of cells equal to the expected number of internal cells ( ntotcell ) , as described in j. salmon ( 1990 ) ( coarse grain : cdir$ shared pos_cell(block(:ntotcell / n$pes ) , : ) ) , to a simple fine grain distribution ( cdir$ shared pos_cell(:block(1 ) , : ) ) . + all the tests were performed for two different set of initial conditions , namely uniform and clustered distribution having @xmath6 particles each .
the tests were carried out using from 16 to 128 pes , and in tab .
1 we report only the most significant results obtained with 128 pes and using coarse grain and fine grain data distribution , although the same trend was obtained using 16 and 64 pes . the data ( i.e. particles and cells ) distribution greatly affects the overall code performance and an accurate study has to be carried out to obtain high gain from the mpp machines for this kind of code .
one possible approach was adopted by @xcite , and is based on the so - called `` locally essential tree '' ( let ) introduced by @xcite , where each pe builds a `` local tree '' for the particles assigned to it , and the force acting on each particle is then computed after all the pes have builded a let assembling together the pointers to all the cells from other pes which do contribute ( according to the above mentioned @xmath1-criterion ) to the force on all their particles .
the main problem with this approach is that the memory requirements grow very quickly with the number of particles @xmath7 .
we have then preferred to keep a single tree shared among all the pes and to look for the optimal distribution of tree s cells and particles .
+ the tree data distribution greatly affects the overall code performance and the distribution must be thoroughly carried out to obtain high gain from the mpp machines for this kind of code .
our results show that the best tree data distribution is obtained using a block factor equal to 1 , degenerate in the second dimension ( cdir$ shared pos_cell(:block(1 ) , : ) ) .
this is in accordance with what we expected , and in order to understand this point we notice that there are two aspects of the problem to be considered .
+ the first is related to the cell inspection performed during the fc phase . during the tf phase , cells properties are determined level by level , starting from level 0 ( containing only the root cell ) and descending down to deeper levels of depth l ( each l level has @xmath8 cells ) .
the cells are numbered progressively starting from the root cell . considering that all the cells belonging to the first four low levels generally include a large part of the domain , they are inspected by each pe @xmath9 times during each timestep because that is the average number of bodies that each pe has to treat .
a fine grain tree data distribution involves that each pe have the same number of cells ( @xmath10 ) and cells belonging to low levels are distributed over the pes local memory . using this kind of tree data distribution
, we obtain that for each pe the execution time of the fc phase is almost the same , because all the pes spend on average the same amount of time spent in the tree data access .
results in tab .1 show that , a coarse grain tree data distribution increases the duration of the tf phase and the number of particles per second , executed in a timestep , up to a factor of five .
this is due to the network contention to access to `` critical resources '' . in the case of a coarse grain tree data distribution ,
all the cells belonging to the first levels are located in the first pe ( or in the first two pes ) , and all pes access them at the same time , during the fc phase . +
another aspect is related to the highly dynamical evolution of tree s properties .
each timestep produces a different tree , making it very difficult to determine rules for an optimal tree data distribution which can minimize the access time on the t3d toroidal architecture .
because of the overhead due to data redistribution during a run , as well as the fact that the block factor power of two ( imposed by craft in the t3d ) we deem it inconvenient to further examine this point .
so it is possible to conclude that , as shown in tab .
1 , a fine grain tree data distribution should be used for this kind of codes . bodies are labelled in such a way that close bodies have adjacent numbers , and the properties are shared among the pes using the craft directive cdir$ shared pos(:block(n ) , : ) n ranging between @xmath11 and @xmath9 .
+ the fine grain bodies data distribution ( bf ) is obtained using a block factor @xmath12 ; i.e. , bodies are shared among the pe but there is no spatial relation in the body set residing in the same pe local memory .
the medium grain bodies data distribution ( bm ) is obtained using a block factor @xmath13 . using this kind of distribution
each pe has two data block of bodies properties residing in the local memory , each block having a close bodies set . at the end
the coarse grain bodies data distribution ( bc ) is obtained using a block factor @xmath14 ; i.e. , each pe has one close data set block of bodies residing in the local memory . in any case , each pe executes the fc phase only for those bodies residing in the local memory .
the results reported in tab . 1 show that the best bodies data distribution , having the highest code performance in terms of particles per second , is obtained using the block factor n = nbodies / n$pes as expected .
the most time - consuming subphases are the ti and the ac . during the first phase the body interaction list is formed , containing some tree cells and close bodies .
generally , tree cells are shared among all pes , whereas bodies are residing in the same pe or in the nearest pes .
this fact reduces the access time of the bodies properties included in the interaction list .
therefore the obtained results confirm that by distributing the close bodies in the same pe ( coarse grain ) we obtain the best performance .
+ this effect depends on the order of the bodies so that nearest bodies have nearest numbering in the arrays containing bodies properties . if necessary
, a sorting of the array can be performed , to obtain higher performances .
as stated above , the best choice is to have a fine grain tree data distribution and a coarse grain bodies data distribution . as emphasized by the unbalance factor in tab .
1 , it is very important to adopt a strategy allowing to increase the load balance and obtain higher performances . at the beginning @xcite
, we adopted a dlb technique based on the concet of `` pe executor '' , i.e. , the pe executing the fc phase for the body .
the pe executor was re - assigned at each time step in order to balance the load .
+ although this usually brings some advantages , sometimes the overhead due to the pe executor re - assignment may greatly reduce the usefulness of this scheme . here
we present the results of a new dlb strategy , that allows us to avoid any large overhead , because no explicit control mechanism is necessary . the total time spent in a parallel region @xmath15
, can be considered as the sum of the following terms + @xmath16 where @xmath17 is the number of processors executing the job , @xmath18 is the time spent in the serial portion of the code ( a typical master region ) , @xmath19 is the time spent by a single processor ( @xmath20 ) to execute the parallel region , @xmath21 the overhead time due to the remote data access and to the synchronization points , and @xmath22 is a costant .
+ & pe # & p / sec & fc phase & t - step & uf + 1mun_tf_bf & 128 & 4129 & 230.05 & 249.5 & 4.22 + 1mcl_tf_bf & 128 & 3832 & 250.32 & 268.81 & 4.57 + 1mun_tf_bm & 128 & 3547 & 270.51 & 290.45 & 5.90 + 1mcl_tf_bm & 128 & 3308 & 291.63 & 312.26 & 6.32 + 1mun_tf_bc & 128 & 4875 & 186.31 & 205.32 & 4.14 + 1mcl_tf_bc & 128 & 4490 & 203.37 & 222.72 & 4.38 + 1mun_tc_bc & 128 & 837 & 1051.93 & 1230.0 & 16.33 + 1mcl_tc_bc & 128 & 750 & 1173.24 & 1373.4 & 17.62 + tab .1 ll legend : & 1mun - 1 million of particles in uniform initial conditions ; + & 1mcl - 1 million of particles in clustered initial conditions ; + & tf - tree fine grain data distribution + & tc - tree coarse grain data distribution + & bf - bodies fine grain data distribution + & bm - bodies medium grain data distribution + & bc - bodies coarse grain data distribution + + & pe # : pes number + & p / sec : particle per second + & fc : total fc pashe duration in second + & t - step : timestep duration in second + & uf : unbalance factor ( variance of fc phase duration ) + + in the fc phase , there are no serial regions , so the @xmath19 term is proportional to the length of the interaction list needed to compute the force acting on each body , the interaction list average length being @xmath23 . using a coarse grain subdivision ,
each pe has a block of close bodies in the local memory ( @xmath24 ) ; in a uniform distribution initial condition , the pes having extreme numeration in the pool of available pes , have residing bodies near the border of the computational domain .
owing to the lack of bodies besides the border line of the computational domain ( i.e. these bodies have a smaller interaction list than bodies in the center of the domain ) , the pes having extreme numeration have a lower load at each timestep .
this kind of effect may be enhanced , if a clustered initial condition is used .
+ the @xmath21 term depends principally on the latency time , on the bandwidth of the internal network , on the code synchronization point , and on the network contention . when the number of pes involved in the simulation increases the data dispersion on the t3d torus increases .
2 we plot the total length of the interaction list , and the total number of cells and bodies forming the interaction list , using @xmath25 .
these data were obtained making several tests with uniform and clustered initial conditions and the results are in accordance with plots reported in salmon ( 1990 ) .
the number of internal cells included in the interaction list ranges from @xmath26 to @xmath27 of its total length .
the tree cells have a fine grain distribution as stated above , and then data access to these elements increases with raising pe number which means the @xmath21 term in ( 1 ) increases . + the figs . 3 , 4 and 5 report the total time ( @xmath15 in ( 1 ) ) in seconds spent in the fc phase ( t_fc ) by each pe , for different values of the total length of the interaction list ( tl_il ) , using 16 , 32 and 64 pes , respectively with 1 million of clustered particles @xmath25 .
tl_il value is the sum of all the interaction lists obtained during the fc phase performed by the single pe .
6 shows a similar result obtained by running a test with only 1 pe on the t3d machine .
in this case all bodies and tree cells properties are located in the pe local memory .
the results show a linear dependence between tl_il and the @xmath19 term . a comparison between data reported in fig . 6 and figs .
3 , 4 and 5 leads us to the following consideration .
when 16 pes are used ( fig .
3 ) a relationship between the t_fc and tl_il terms may be found .
the overhead time @xmath21 , ranging from 60% up to 70% of the @xmath28 term in ( 1 ) , can be calculated from the difference between data reported in figs .
3 and 6 . when the pe number increases the relationship
is lost as shown fig .
the @xmath21 increases as the pe number increases , whereas the code performances increase when the load balance is optimized .
+ the adopted technique is to perform a load redistribution among the pes so that all pes have the same load in the fc phase .
we force each pe to execute this phase _ only _ for a fixed portion nb_lp of the bodies residing in the local memory .
the nb_lp value is calculated as @xmath29 the p_lp being a constant ranging from @xmath30 to @xmath11 .
the fc phase for all the remaining bodies @xmath31 is executed from all the pes that have finished the fc phase for the nb_lp bodies . no correlation between the pe memory , where the body properties are residing , and the pe , executing the fc phase for it , is found .
+ if p_lp=1 all pes execute the fc phase only for bodies residing in the local memory .
@xmath32 is equal to 0 and the body locality is totally involved in the overall performances , on the contrary if p_lp=0 , @xmath33 ( nb_lp = 0 ) , the pes execute only nfree bodies and the locality is not taken into account . + a p_lp value lower than 1 gives an automatic dynamic load balance mechanism to the system ; i.e. , each pe works until all the nfree bodies are computed . on the other hand ,
if p_lp value is equal to 0 , it gives the maximum load balance and the maximum degree of automatism in the fc phase .
+ several tests were performed with p_lp value ranging from 0 to 1 and pes ranging from 16 to 128 , using several initial conditions from 1000 up to 2 million particles uniform and clustered .
7 - 10 show the results obtained only with a high ( @xmath34 ) number of particles , but the results obtained with smaller numbers of particles show the same trend .
+ using the reported results and fixing the pe number , it is possible to determine the p_lp value allowing the best code performances . in particular , note that for a simulation using a high number of pes ( more than 32 ) , it is convenient to fix the p_lp value near to @xmath30 , that is maximize the load balance among the pes rather than the bodies locality .
this is due to the network contention of the system that becomes relevant for the remote data access , mainly for the data access time of the tree cells .
+ on the other hand , using a lower pe number ( from 16 to 32 ) we observe a different effect .
we have a lower data dispersion among the pes , so the remote loads are fewer and have a shorter access time , thus the code performance takes advantage of the data locality . from figs .
7 - 10 it is possible to fix a p_lp value combining the `` load remote '' effect and the `` data locality '' effect to maximize the computed number of particles per second , thus improving the code performances .
+ the figures show that , fixing the pes number and the particles number , the same p_lp value gives the best performance both in uniform and clustered conditions .
this means that it is possible to fix a p_lp value and this can be usefully adopted during all the running time : it is not necessary to recompute the p_lp value to have good performances .
hence an automatic load balance mechanism is found without recalculating the p_lp value and without introducing any overhead time to obtain a good dynamic load balance .
as stated in @xcite the memory occupancy of this kind of n - body code is lower than the n - body local essential tree - based codes .
the total memory occupancy of wdsh - ptc , at present , allows to execute large simulation with more than 80 million of particles on the cineca t3e machine having 128 pes each with 16 mword ram .
moreover the total memory occupancy may be reduced in order to run larger simulations with a little degradation of code performances .
11 shows the code speed ( preliminary data ) in number of particles per second , computed at each timestep , vs. the number of pes , using two @xmath1 values : @xmath35 and @xmath36 , for a clustered configuration of @xmath37 particles , using the periodic boundary conditions and adopting a _ grouping _
strategy @xcite , @xcite , @xcite i.e. building an interaction list for a group of bodies ( included in a cell ) .
+ the results obtained using the wdsh - ptc code , at present , give performances comparable to those obtained with different approaches like let @xcite , with the advantage of avoiding the let and an excessive demand for memory .
the study carried out on the cray t3d machines at the cineca ( casalecchio di reno , italy ) , allows us to propose a criterion for the optimal data distribution of bodies and tree cells properties . a strategy for
the _ automatic _ dynamic load balance has been described , which does not introduce a significant overhead .
the results of this work will allow us to obtain , in the next future , a wdsh - ptc version for the cray t3e system , using the hpf - craft and the shmem library .
the new version will include an enhanced grouping strategy and periodic boundary conditions , and will allow us to run large simulations with very high performances @xcite . using the cineca visualization laboratory experience and resources
, we plan also to develop an ad hoc visualization package for the scientific visualization of the simulation results we would like to thank the cineca for the supercomputing time grant , mr .
s. bassini for the availability of the cineca support and dr .
l. calori , both from cineca , and dr .
f. dur from silicon graphics for useful suggestions and comments .
a. pagliaro wish to tank the epcc staff . v. antonuccio - delogu , u. becciani , _
`` a parallel tree n - body code for heterogeneous clusters '' _ , in : j. dongarra and j. wasniewsky , eds.,_parallel sciebtific computing - para 94 _ ( springer verlag : 1994 ) , 17 | n - body algorithms for long - range unscreened interactions like gravity belong to a class of highly irregular problems whose optimal solution is a challenging task for present - day massively parallel computers . in this paper
we describe a strategy for optimal memory and work distribution which we have applied to our parallel implementation of the barnes & hut ( 1986 ) recursive tree scheme on a cray t3d using the craft programming environment .
we have performed a series of tests to find an _
optimal data distribution _ in the t3d memory , and to identify a strategy for the _ dynamic load balance _ in order to obtain good performances when running large simulations ( more than 10 million particles ) .
the results of tests show that the step duration depends on two main factors : the data locality and the t3d network contention .
increasing data locality we are able to minimize the step duration if the closest bodies ( direct interaction ) tend to be located in the same pe local memory ( contiguous block subdivison , high granularity ) , whereas the tree properties have a fine grain distribution . in a very large simulation , due to network contention ,
an unbalanced load arises . to remedy this
we have devised an automatic work redistribution mechanism which provided a good dynamic load balance at the price of an insignificant overhead . , , , , , and |
emission line stars ( els ) range from young to evolved stars ( ttauri , herbig ae / be , wr , planetary nebulae , etc ) , from hot to cool stars ( classical be star , oe , supergiant star , mira ceti , flare stars , etc ) . among the els , here ,
we focus on the classical be stars .
they are non - supergiant b type stars , which have displayed at least once emission lines in their spectra , mainly in the balmer series of the hydrogen .
the emission lines come from a circumstellar decretion disk formed by episodic matter ejection of the central star .
it appears that the be phenomenon is related to fast rotation and probably additional properties such as non - radial pulsation or magnetic fields . for a comprehensive review of
be stars in the milky way we refer the reader to * ? ? ?
it seems also that the low metallicity plays a role @xcite : at low metallicity , typical of the small magellanic cloud ( smc ) , the stellar radiatively driven winds are less efficient than at high metallicity ( typical of the milky way [ mw ] ) , thus the mass loss is lower and the stars keep more angular momentum . as a consequence ,
b - type stars rotate faster in the smc / lmc than in the mw @xcite .
it is then expected that the metallicity has also an effect on the be - phenomenon itself as reported by * ? ? ? * or * ? ? ?
* , while the evolutionary phase could also play a role @xcite . in most cases ,
the study of open clusters was done by using photometric observations , in the mw .
the work by * ? ? ?
* is a typical example . to test these issues and improve the comparisons between smc and mw , spectroscopic observations of stars in open clusters have to be done . in the smc
, a survey of emission line objects was performed by * ? ? ?
* using photographic plates . however , they were not able to study the stars in open clusters but mainly in the field .
this paper deals with our slitless spectroscopic survey of els and oe / be / ae stars in smc open clusters , while in the whole smc 3 million spectra were obtained .
to increase the number of open clusters studied in the smc ( 1 to 6 in the previous photometric studies of * ? ? ?
* or * ? ? ?
* ) , to improve the statistics , and to quantify the evolution of the rates of be stars to b stars with decreasing metallicity , we performed a slitless spectroscopic survey of the smc .
the observations were obtained on september 25 , 2002 with the eso wide field imager ( wfi , see * ? ? ?
* ) at the 2.2 m mpg telescope located at la silla in chile .
we used its slitless spectroscopic mode with the r50 grism and a narrow filter centered on h@xmath0 to reduce the crowding of observed areas .
we recall that this kind of instrumentation is not sensitive to the diffuse ambient nebulosities and does not allow weak emission lines to be found . with 10-minutes exposures with wfi , it is possible to detect emission lines with ew@xmath110 or with a relative intensity to the continuum equal to 2 down to v magnitudes around 17.5 . for fainter stars , due to the noise in the spectra , it is only stronger emission lines are within reach . as an example of this kind of observations ,
see the study by * ? ? ?
* in the mw ngc6611 open cluster and the eagle nebula .
the ccd image treatment was performed using iraf tasks .
the spectra extraction was done using the sextractor code @xcite .
the treatment of spectra and search for stars with emission were done by using the album code ( see * ? ? ?
details about the samples used in this study are given in table [ tab1 ] .
.details and comparison of samples used in this study [ cols="^,^,^",options="header " , ]
after the data - reduction , the objects were sorted by categories ( definite or candidate els , absorption stars ) , and the astrometry of the stars was performed with an accuracy of 0.5 with the astrom package of .
+ we then cross - matched our catalogues with ogle photometric catalogues in smc open clusters from * ? ? ?
* , * ? ? ?
* , and * ? ? ?
* , in order to obtain the b , v , i magnitudes of the stars , the reddening , and the age of open clusters . with this information , the absolute magnitudes and the derreddened colour indices of the stars were derived .
several open clusters are found with high ratios of be to b stars , occasionally even higher than in ngc330 , an open cluster already known for its rich be content . in order to abstract from the large variations from one open cluster to another in their frequency of
be stars , the stars were grouped in a global sample .
we classified the 4300 stars in our sample by using the calibration provided by * ? ? ? * .
fig.[figs1 ] shows the corresponding global hr colour / magnitude diagrammes . from these results
it seems that the smc sample is complete for the els till spectral types b3-b4 , while for absorption stars it is complete till b5 .
+ to highlight and quantify a potential trend of the rates of be stars to b stars with decreasing metallicity , we compared these ratios by spectral - type categories with the study of * ? ? ? * in the mw .
the same calibration for the classification of the stars was used in the mw .
the comparison of the rates is shown in fig .
[ figs3 ] . down to the completness limit of our study of els
, one can see that the ratios are several times higher in the smc than in the mw .
thus , there is an impact of the metallicity on the number of be stars probably corresponding to the increase of the rotational velocities in the smc in comparison with the mw .
+ the distribution of be stars by spectral type categories in the smc and mw was also studied . in the mw , the results from * ? ? ?
* in open clusters , and from * ? ? ?
* in the field , were used .
the result is shown in fig .
[ figs4 ] .
it appears that the maximum of be stars is reached at the spectral type b2 in the smc and mw .
there is also another peak at b5-b6 , which can be seen in the mw studies ( complete for the whole b - main sequence ) .
the first peak corresponds to the maximum of the emission intensity , the second one to the combined effects of the decrease of the emission intensity and the increase of the initial mass function of stars towards late type stars .
+ from preliminary results it appears that predominantly young open clusters + ( age@xmath2100myears ) are found to contain classical be ( cbe ) stars , but depending on the types of the stars , certain cbe stars could have reached the terminal - age main sequence ( tams ) .
for example in the case of b0e stars , the tams is reached in @xmath310 myears . some other old open clusters ( age@xmath4100myears ) host cbe stars . from these results
, it seems that certain cbe could be born as cbe stars and some others appear during their evolution .
this point needs more investigations star by star .
preliminary studies have shown a trend of the increase of the fraction of be stars with decreasing metallicity .
however , up to now , the studies were only performed on a very limited sample of open clusters ( less than 10 in the smc ) and with photometric data . using the eso / wfi in its slitless spectroscopic mode
, we observed 84 open clusters in the smc .
thanks to different codes and ogle data , we were able to find and classify the emission line stars and absorption stars ( @xmath34300 stars ) .
the ratios of be stars to b stars in the smc and mw were studied .
the comparison allows to quantify the increase of the number of be stars with decreasing metallicity .
be stars in the smc are 2 to 4 times more abundant than in the mw depending on the spectral types .
it seems also that early be stars follow the same distribution in the smc and mw with a maximum at the spectral type b2 . about the stellar phases at which be stars
appear , from our preliminary results , it seems that certain be stars could be born as be stars , while others could appear during the main sequence depending also probably on the metallicity and spectral types of stars .
c.m . acknowledges funding from the esa / belgian federal science policy in the framework of the prodex program ( c90290 ) .
thanks support from eso s dgdf 2006 , the iaus soc / loc for the iaus grant , and the fnrs for the travel grant . | at low metallicity , b - type stars show lower loss of mass and , therefore , angular momentum so that it is expected that there are more be stars in the magellanic clouds than in the milky way .
however , till now , searches for be stars were only performed in a very small number of open clusters in the magellanic clouds . using the eso / wfi in its slitless spectroscopic mode
, we performed a halpha survey of the large and small magellanic cloud .
eight million low - resolution spectra centered on halpha were obtained . for their automatic analysis
, we developed the album code . here , we present the observations , the method to exploit the data and first results for 84 open clusters in the smc .
in particular , cross - correlating our catalogs with ogle positional and photometric data , we classified more than 4000 stars and were able to find the b and be stars in them .
we show the evolution of the rates of be stars as functions of area density , metallicity , spectral type , and age . |
in spite of its long history as a basic ferromagnetic metal characterization tool , the theory of the anomalous hall effect ( ahe ) continues to be a subject of confusion and debate.@xcite theoretical descriptions of the dc ahe invariably involve long and complex calculations and usually do not lead to simple general results with transparent interpretations .
theories of the ahe normally focus on particular simple model hamiltonians , and ignore interactions apart from mean - field exchange potentials that encode magnetic order . even with these simplifications , the ahe _ problem _ tends to be difficult because the effect usually follows mainly from density - matrix linear response that is off - diagonal in bloch state band indices , _
i.e. _ from induced interband coherence rather than simply from changes in bloch state occupations .
farraginous theoretical results have followed from the application of different methods to the same models .
rigorous approaches based on green s function techniques , like the keldysh or kubo - streda formalisms , have the advantage of being systematic but can be technically more difficult to implement and are often not physically transparent .
renewed interest in reaching consensus on a general theory of the ahe has been fueled in part by the realization of diluted magnetic semiconductor ( dms ) ferromagnets which have strong spin - orbit interactions , large anomalous hall effects , and variable carrier densities and magnetic properties .
dms ferromagnets can serve as a testing ground for the different theoretical predictions concerning the anomalous hall effect.@xcite interest has also been increased by the ( phenomenological ) demonstration that a relatively simple _ intrinsic _ contribution that is a momentum - space berry phase band - structure property dominates the ahe in many ferromagnets @xcite . from one point of view
, the main reason for the physical opaqueness of formal microscopic approaches to the ahe is that the hall current is much weaker than the current parallel to the electric field .
no hall contribution appears in usual theories of the dc conductivity which use gaussian disorder potential distribution models and evaluate conductivity to the leading order of the small parameter @xmath0 where @xmath1 is the disorder scattering length .
higher order terms must therefore be considered - and these are unfortunately plentiful .
bookkeeping becomes a challenge .
additionally typical metallic or semiconducting ferromagnets have many occupied bands and anisotropic band structures , making it difficult in general to obtain analytical solutions to the relevant transport equations .
the semiclassical description of transport coefficients through a boltzmann equation does lead to a hall contribution if skew scattering is accounted for in the collision term .
there are however , other contributions to the hall effect that are ignored in such a theory including anomalous velocity and the side jump effects .
the presence of these additional contributions is directly linked to induced nonzero interband elements of the density matrix , in other words , the interband coherence , as recognized since the work of kohn and luttinger . in a previous paper several of us
demonstrated @xcite that by utilizing the fermi golden rule and a newly derived gauge invariant expression for the coordinate shift due to a scattering event one can construct a semiclassical , physically transparent description of all contributions to the anomalous hall effect and obtain the same final results as derived in a more complicated way using the microscopic perturbative approach of luttinger.@xcite the major drawback of this semiclassical approach is the fact that the identification of contributions to the ahe is not systematic ; it is not obvious for example that other contributions of the same size are not being neglected and therefore , the assertion that we do include all main contributions must be verified by comparison to other approaches since the semiclassical approach lacks the rigor of the microscopic approaches .
the main goal of the present paper is to provide a more detailed explanation of the semiclassical approach and to link it with a more systematic and formally rigorous approach based on the kubo - streda formula .
the various different contributions that have been identified in the semiclassical approach , can , for the particular case of two - dimensional dirac bands , all be directly linked to particular feynman - diagram subsums in the kubo - streda formalism .
the correspondence is summarized in fig .
[ mainfig ] , in which the diagrammatic expansion of the kubo - streda formalism in the chiral ( band - eigenstate ) basis is summarized . in fig .
[ mainfig ] feynman diagrams are grouped in the three major contributions that arise naturally in the semiclassical approach : intrinsic , side - jump , and skew - scattering . '' .
the subsets of diagrams that correspond to specific terms in the semiclassical boltzmann formalism are indicated .
a detailed discussion of each set of diagrams is given in sec .
[ comp].,width=16 ] we organize the rest of the article as follows . in sec .
[ mh ] we introduce the model hamiltonian for which we demonstrate the equivalence of the two approaches . in sec .
[ sbe ] we provide a formal description of both techniques and discuss the approximations involved in their application . in sec .
[ comp ] we discuss different contributions identified in the boltzmann approach : intrinsic , side - jump , and skew scattering contributions , and show their equivalence to partial sums of diagrams in the kubo - streda formalism . in sec . [ missing ] we discuss some diagrams that appear in the kubo expansion beyond our approximations .
finally we summarize our results in sec .
[ conclusion ] .
although most of the discussion in the following sections is general and applicable to electronic structure in any number of dimensions , we choose as a concrete example the massive 2d dirac model hamiltonian with randomly placed weak @xmath2-function - like spin - independent impurities .
we have chosen this model due to its simplicity which permits the derivation of compact analytical expressions that facilitate comparison between the semiclassical boltzmann approach and the microscopic kubo approach .
the impurity free 2d dirac hamiltonian is @xmath3 where @xmath4 and @xmath5 are pauli matrices .
( in the case of graphene , this degree - of - freedom represents the two sites per cell on a honeycomb lattice . ) here and throughout the text we take @xmath6 .
this hamiltonian breaks time - reversal symmetry and therefore has a non - zero hall conductivity .
kubo formula calculations of the dc - limit of the hall conductivity for this hamiltonian have been already performed@xcite in the self - consistent non - crossing approximation with the following result which can be applied to the charge and spin hall conductivities in metallic graphene : @xmath7 - \frac { e^2v_1 ^ 3 } { 2\pi n_i v_0 ^ 4 } \frac{\ ( vk_f)^4 \delta } { ( 4\delta^2+(vk_f)^2)^2 } , \label{sxy2}\end{aligned}\ ] ] where @xmath8 is the fermi momentum and the parameters @xmath9 and @xmath10 characterizing the disorder distribution as defined below . the non - zero mass @xmath11 opens a gap in the spectrum and splits the dirac band into sub - bands above and below the gap with dispersions @xmath12 where @xmath13 and the labels @xmath14 distinguish bands with positive and negative energies . in what follows , we will assume that the fermi energy is positive and the fermi level lies above the gap , as illustrated in fig .
[ cone1 ] .
the chiral basis states that diagonalize the hamiltonian in eq .
( [ dh1 ] ) are @xmath15 , where @xmath16 @xmath17 , and @xmath18 . in the eigenstate representation ( eq.([basis ] ) ) , the disorder and velocity operators have the following matrix form @xmath19 @xmath20 and @xmath21 where @xmath22 orbital disorder matrix element .
we consider the model of randomly located @xmath2-function scatterers : @xmath23 with @xmath24 random and strength distributions satisfying @xmath25 , @xmath26 and @xmath27 .
this model is different from the standard white noise disorder model in which only the second order cumulant is nonzero ; @xmath28 where @xmath29 is the impurity concentration and other correlators are either zero or related to this correlator by wick s theorem .
the deviation from white noise in our disorder model is quantified by @xmath30 , and is necessary to capture part of the skew scattering contribution to the anomalous hall effect .
note however that within this model , as we will show in later sections , there are other contributions to the skew scattering that arise from the gaussian disorder distribution which turn out to be independent of disorder strength , e.g. see sec .
[ skew_section ] . in what follows we use @xmath31/@xmath32 for the upper / lower chiral band indices .
for example , from eq .
( [ sx1 ] ) @xmath33 $ ] or from eq .
( [ vkk1 ] ) @xmath34 . working in the eigenstate basis
, we will refer to the velocity matrix elements @xmath35 and @xmath36 as the diagonal velocities and to @xmath37 and @xmath38 as the off diagonal or interband velocities .
there is a substantial literature on the application of boltzmann equation concepts to ahe theory .
however , stress was usually placed only on one of the many possible mechanisms for a hall current . in this section
we briefly review this approach , incorporating some new insights from recent work.@xcite the semi - classical boltzmann equation ( sbe ) describes the evolution of the electron distribution function as though electrons were classical particles labeled by a band index and living in the crystal s momentum space . from a quantum mechanical point of view , it is clear that this approach is not universally applicable because it violates the heisenberg uncertainty principle by having a momentum - distribution function at each point in space .
a rigorous treatment of the electronic state evaluation in the presence of disorder should generally consider the entire density matrix , including off - diagonal elements .
formal justifications of the boltzmann description are usually made in terms of the properties of wave packets that have well defined average momentum , center of mass coordinate , spin , intrinsic angular momentum , _ etc_.@xcite the distribution functions then acquire the meaning of the probability distribution of wave packets that behave in many respects like classical particles.@xcite the semi - classical distribution function evolves both due to hydrodynamic particle fluxes and due to collisions with impurities .
quantum mechanically , the elementary scattering process is described by the scattering matrix from one state to another in the presence of a perturbing potential . in the semi - classical approach
, scattering can be accompanied by changes of coordinate .
as we discuss below , these have to be taken into account when constructing a kinetic equation for the semi - classical distribution function that captures the ahe .
the golden rule connects the classical and quantum descriptions of a scattering event.@xcite it shows how the scattering rate @xmath39 between states with different quantum numbers @xmath40 and @xmath41 is related to the so called @xmath42-matrix elements : @xmath43 the scattering @xmath42-matrix is defined as @xmath44 where @xmath45 is the impurity potential operator and @xmath46 is the eigenstate of the full hamiltonian @xmath47 that satisfies the lippman - schwinger @xmath48 for weak disorder one can approximate the scattering state @xmath46 by a truncated series in powers of @xmath49 and take @xmath50 as the combined ( momentum , band ) index of the eigenstate @xmath51 of the unperturbed hamiltonian @xmath52 : @xmath53 for example the @xmath42-matrix up to the second order in @xmath45 is @xmath54 substituting eq .
( [ tmatrix2 ] ) into eq .
( [ wt ] ) one can find the scattering rate up to the 3rd order in the disorder strength @xmath55 where @xmath56 @xmath57 the skew scattering contribution to the hall effect follows from the asymmetric part of the scattering rate:@xcite @xmath58 since @xmath59 is symmetric , the leading contribution to @xmath60 appears at order @xmath61 .
the @xmath62 contribution to the scattering rate @xmath63 contains symmetric and antisymmetric parts .
the symmetric part is not essential since it only renormalizes the second order result of eq .
( [ om2 ] ) . the sum in parentheses in eq .
( [ om3 ] ) can be rewritten in the form @xmath64 the first term is symmetric under the exchange of indexes @xmath65 ( note that this needs the fact that @xmath66 due to the delta - function in eq .
( [ om3 ] ) ) , hence this sum does not contribute to the asymmetric part of the scattering amplitude and one can concentrate on the second one : @xmath67 with the superscript @xmath68 meaning that this is the antisymmetric part of the scattering rate calculated at order @xmath69 . in a semi - classical description of wavepacket motion in a crystal
, scattering produces both a change in the direction of motion and a separate coordinate shift . in the lowest order born approximation for a scalar disorder potential one can derive an expression for the coordinate shift which accompanies scattering between two band states .
the shift does not depend explicitly on the type of impurity and can be expressed in terms of initial and final states only : @xmath70 .
\label{delr4b}\ ] ] where @xmath71 $ ] is the phase of the complex number @xmath72 and @xmath73 the topological interpretation of eq .
( [ delr4b ] ) has been explained in ref . .
the first two terms in eq .
( [ delr4b ] ) have been known for a long time .
the last term was first derived only recently in ref . but is an essential contribution which makes the expression for the coordinate shift gauge invariant .
( [ wt ] ) and ( [ delr4b ] ) contain the quantum mechanical information necessary to write down a semi - classical boltzmann equation that takes into account both the change of momentum and the coordinate shift during scattering in a homogeneous crystal in the presence of a driving electric field @xmath74 . keeping only terms up to the linear order in the electric field the boltzmann equation reads@xcite @xmath75 , \end{array } \label{beint}\ ] ] where @xmath76 is the usual velocity @xmath77 note that if only elastic scatterings with static impurities are responsible for the collision term , the rhs .
( [ beint ] ) is linear in @xmath78 and not in @xmath79 .
this property follows from the fact that heizenberg time evolution equations for creation or annihilation operators in a noninteracting electron system are linear and can be mapped to the schrdinger equation for amplitudes in a single particle system .
( for further discussion of this point see appendix b in ref . .
) this boltzmann equation has the standard form except for the coordinate shift effect which is taken into account in the last term in the collision integral on the rhs of eq .
( [ beint ] ) .
this term appears because the kinetic energy of a particle in the state @xmath41 before scattering into the state @xmath40 is smaller than @xmath80 by the amount @xmath81 .
the collision term does not vanish in the presence of an electric field when the occupation probabilities @xmath82 are replaced by their thermal equilibrium values because @xmath83 .
the last term in the collision integral in eq .
( [ beint])@xcite accounts for this interplay between coordinate shifts and spatial variation of local chemical potential in the presence of an electric field . the total distribution function @xmath78 in the steady state ( @xmath84 ) can be written as the sum of the equilibrium distribution @xmath85 and non - equilibrium contributions , @xmath86 where we have split the non - equilibrium contribution into two terms @xmath87 and @xmath88 which solve independent self - consistent time - independent equations:@xcite @xmath89 and @xmath90 ( the label _ adist _ stands for anomalous distribution . ) a standard approach to solving these equations in 2d is to look for the solution of eq .
( [ bbeint3 ] ) in the form ( see e.g. ref . ): @xmath91 where @xmath92 is the unit vector in the out - of - plane direction .
we will assume that the transverse conductivity is much smaller than the longitudinal one so that @xmath93 .
one then finds by the direct substitution of eq .
( [ solg ] ) into eq .
( [ beint ] ) that @xmath94 where @xmath95 and @xmath96 and @xmath97 are the angles between @xmath76 or @xmath98 and the x - axes .
this completes the solution of eq .
( [ bbeint3 ] ) for @xmath87 .
we will show in sec.iii.b.1 how a similar _ ansatz _ can also solve eq .
( [ ganl ] ) for the anomalous distribution @xmath99 . to find the current induced by an electric field and hence the conductivity we need to derive an appropriate expression for the velocity of semiclassical particles , in addition to solving the sbe for the state occupation probabilities . in considering the ahe , in addition to the band state group velocity @xmath100 one
should also take into account velocity renormalizations due to the accumulations of coordinate shifts after many scattering events ( the side - jump effect ) and due to band mixing by the electric field ( the anomalous velocity effect ) : @xmath101 the second term in eq . ( [ tovel ] ) captures changes in the speed at which a wave packet moves between scattering events under the influence of the external electric field @xcite and @xmath102 is the berry curvature of the band @xcite @xmath103 .
\label{berrcurv}\ ] ] the last term in eq .
( [ tovel ] ) is due to the accumulation of coordinate shifts after many scatterings . combining eqs .
( [ ffgg ] ) and ( [ tovel ] ) we obtain the total current @xmath104 from ( [ jtotal ] ) we can extract the off - diagonal hall conductivity which is naturally written as the sum of four contributions : @xmath105 the first term is the so called intrinsic contribution @xmath106 which is due to the anomalous velocity of free electrons under the action of the electric field @xmath107 this term is called the intrinsic contribution because it is not related to impurity scattering , _
i.e. _ its evaluation does not require knowledge of the disorder present in the system .
the intrinsic ahe is completely determined by the topology of the bloch band .
the next two contributions follow from coordinate shifts during scattering events : @xmath108 is the conductivity due to the anomalous correction to the distribution function while the direct side - jump contribution @xmath109 is due to the side - jump velocity , _
i.e. _ due to the accumulation of coordinate shifts after many scattering events . since coordinate shifts are responsible both for @xmath110 and for @xmath111 there is , unsurprisingly , an intimate relationship between those two contributions .
we will demonstrate their equality by evaluating @xmath110 below .
because of this relationship , in most of the literature , @xmath110 is usually considered to be part of the side - jump contribution , i.e. @xmath112 . here we will distinguish between the two when connecting the sbe approach to the kubo - streda formalism . finally , @xmath113 is the skew scattering contribution @xmath114 the skew scattering contribution is independent of the coordinate shift and of the anomalous velocity .
it is non - zero for an isotropic energy dispersion @xmath80 only when the scattering rate @xmath63 is asymmetric .
writing the transverse velocity in the form @xmath115 , then using eq .
( [ taus ] ) and substituting eq .
( [ solg ] ) for @xmath87 into eq .
( [ jtotal_sk ] ) one finds @xmath116 at zero temperature this expression simplifies to @xmath117 there is potentially one more contribution which has not been included in eq .
( [ jttt ] ) and which involves the product of @xmath88 and the side jump velocity @xmath118 .
such a term does not contribute to the hall conductivity in the present model calculation of isotropic scattering and band structure .
it may however , in principle , give rise to a non - zero contribution given an appropriate anisotropic scattering and anisotropic fermi distribution .
we do not consider this term further here . in sec .
iv we evaluate each of these four contributions for the particular example of the 2d dirac model .
a fully quantum mechanical formally exact expression for the conductivity is provided by the linear response kubo theory .
there are various equivalent formulations of this approach . in one version
the conductivity at zero temperature @xmath42 is expressed in terms of causal green s functions .
this formalism has previously been applied to the ahe in an electron system with rashba spin - orbit coupling.@xcite in the present paper we work mainly with another version of the kubo formula , derived specially for the the dc conductivity .
its advantage is that one should not perform the @xmath119 limit at the end of calculations .
the starting point in this approach is the bastin formula:@xcite @xmath120,\end{aligned}\ ] ] which can be manipulated to give the kubo - streda formula for the dc @xmath121 hall conductivity @xmath122 where : @xmath123 @xmath124 @xmath125 = ec \left .
\frac{\partial n(\epsilon)}{\partial b } \right|_{\epsilon=\epsilon_f , b=0 } \label{sigmaii}\end{aligned}\ ] ] here the subscript @xmath126 indicates a disorder configuration average .
the last contribution , @xmath127 , was first derived by streda .
@xcite in these equations @xmath128 is an external magnetic field that couples only to the orbital degrees of freedom , and @xmath129 is the integrated density of states of the disordered system and @xmath130 are the retarded and advanced green s functions evaluated at the fermi energy . every term in @xmath131 depends on products of retarded green s functions only or products of advanced green s functions only .
since every such term has poles on the same side of the imaginary plane , disorder corrections to this contribution are small because they contain at least one ( small ) factor @xmath132 which is not compensated by energy and momentum integrations .
hence only the disorder free part of @xmath131 is important in the weak disorder limit , _
i.e. _ this contribution is zeroth order in the parameter @xmath133 .
it can also be shown by a similar argument that in general @xmath134 , is of order @xmath135 and can be neglected in the weak scattering limit.@xcite thus , important disorder effects are contained only in @xmath136 . in what follows we neglect @xmath134 and retain only the clean limit @xmath131 contribution .
the following paragraphs concern only @xmath136 .
the effect of disorder on the disorder - configuration averaged green s function is conveniently captured by the use of the t - matrix , defined by the integral equation @xmath137 where @xmath138 . from this
one obtains @xmath139 upon disorder averaging we obtain @xmath140 to linear order in @xmath29 this translates to @xmath141 with @xmath142 being the fourier transform of the single impurity potential , which in the case of delta scatterers is simply @xmath9 .
within the t - matrix formalism in the normal spin - basis we will take @xmath143 for simplicity since , as it will be shown when considering the kubo formalism within the chiral basis , only a couple of diagrams contribute to part of the skew scattering for @xmath144 and can therefore its contribution can be computed separately ( see sec .
[ skew_section ] ) . replacing the bare green s function by the self consistent born approximation green s function , eq .
( [ g ] ) , in the expression for @xmath136 does not capture all the contributions to lowest order in @xmath135 .
one must also compute the ladder diagram correction to the bare velocity vertex defined by @xmath145 , where @xmath146 satisfies @xmath147 the explicit calculation of this vertex correction in the @xmath148 eigenstate basis is described in appendix [ vertex_calculation ] .
the final result for the 2d dirac model hamiltonian is @xmath149 where @xmath150 . incorporating this result in @xmath136 we obtain for @xmath151 @xmath152 \\ \\
= -\frac{e^2 \cos\theta}{4\pi}\left(1+\frac{3\sin^2\theta}{(1 + 3\cos^2\theta ) } + \frac{4\sin^2\theta}{(1 + 3\cos^2\theta)^2}\right ) \end{array } \label{eq48}\ ] ] this is equivalent to the part of eq .
( [ sxy2 ] ) that is independent of @xmath10 . in order to understand more fully the relationship between this result and the more physically transparent sbe approach
, we need to analyze the kubo approach in the chiral band eigenstate basis . as described in the remaining part of this section , we are able to make the connection very clear .
we have also obtained the same result using the time ordered green s functions formalism .
the equivalence of the two results is shown in detail in appendix [ dugaev ] . in the chiral bloch eigenstate basis ,
specified in eq .
( [ basis ] ) , the disorder free green s functions in momentum space can be written in the form @xmath153 from a calculational point of view the bare green s function is simpler in this basis , but we must pay the price in the disorder matrix of the hamiltonian which will contain , as seen in sec .
ii , intraband ( @xmath154 , @xmath155 ) and interband ( @xmath156 , @xmath157 ) matrix elements . in the weak disorder limit
, the energy width of bloch state spectral peaks is smaller than the gap , allowing us to ignore direct interband scattering .
this does not mean , however , that interband matrix elements of the disorder potential are not important and can be disregarded .
in the rest of the text we will show that interband disorder matrix elements , and the interband coherence they imply , are in fact crucial to fully explain disorder contributions to the ahe . although interband scattering probability remains zero , interband disorder matrix elements produce virtual transitions that mix states in the two bands in a way which is ultimately relevant . when evaluating diagrams in the chiral basis , disorder lines associated with interband and intraband scattering processes can be distinguished .
their separate contributions will be easily linked to the semiclassical interpretation of the scattering process .
as we will see below , the relevant vertex correction infinite subsum involves only intraband elements @xmath154 and does not produce unusual effects in the sense that it only renormalizes the quantum life time and the diagonal velocity vertex , @xmath158 .
in contrast , vertex correction diagrams which contain @xmath156 , @xmath157 and therefore connect green s functions with different band indices , do not have to be summed to infinite order and do capture the unusual transport physics which leads to ahe contributions .
only a few low order ( in terms of @xmath135 ) diagrams , which depend on @xmath156 or @xmath157 , have to be included in the calculations of @xmath159 .
note that these terms still contain the renormalized @xmath160 velocity vertex but the interband component does not contribute to this renormalization as we will see below .
we consider first all possible diagrams that do not contain an interband part of the disorder potential .
this part of the calculation is essentially the familiar standard calculation because the green s functions remain diagonal in band indexes and the calculations are reduced to single band ones . in this case
the non - crossing approximation leads to a finite life time and the diagonal velocity vertex is renormalized after the summation of all non - crossing disorder paring lines .
only the imaginary part of the self - energy is important in the @xmath161 limit .
the self - energy diagram that contributes to its diagonal element for the gaussian disorder model is shown in fig .
[ selfen1 ] . because @xmath162 has a negligible imaginary part one can keep only the @xmath163 part of the green s function in calculations of @xmath164 , i.e. @xmath165 .
hence , the renormalized retarded green s function is given by a standard result @xmath166 where @xmath167 note that we have so far ignored the off - diagonal part of the self - energy which will contribute later when considering the diagrammatic expansion of the conductivity in powers of @xmath135 .
it is convenient to organize the calculation in this way because the off - diagonal parts do not need to be partially summed to infinite order and only a few terms contribute to the desired order .
every disorder line that connects retarded and advanced green s functions adds a factor @xmath168 multiplied by a product @xmath169 . while the former factor gives a power of @xmath135 , the product of two green s functions gives a @xmath170 contribution if the pair of green s functions over which the momentum sum is performed are in the same band .
thus one can approximate@xcite @xmath171 since @xmath172 is proportional to @xmath173 this product of green s functions compensates the small factor due to disorder vertexes .
therefore terms @xmath174 and @xmath175 are of the same order and one should perform the summation over all ladder diagrams that include only green s functions in the positive energy band .
these ladder diagrams simply renormalize the diagonal matrix element of the velocity vertex operator .
the bare diagonal velocity is @xmath176 .
the vertex equation is @xmath177 eq .
( [ veq2 ] ) for the vertex correction is represented graphically in fig .
[ fvert2 ] .
this equation has a solution of the form @xmath178 where @xmath179 , @xmath180 @xmath181 and @xmath182 is the fermi velocity @xmath183 . combining this result with eq .
( [ inouegg ] ) gives the useful weak - scattering identity @xmath184 after summing the series of diagrams that contain only diagonal disorder vertices one can look at the effect of the inter - band disorder matrix elements @xmath185 and @xmath157 .
any disorder average line with this component still adds a small factor @xmath132 , however now it is followed by at least one green s function in the lower band @xmath186 or @xmath187 .
the new green s function products that accompany the interband disorder interaction lines , products like @xmath188 , _ etc .
_ , no longer can be considered as large .
they are at most zeroth order in the parameter @xmath135 in the final expression rather than being proportional to @xmath170 .
since any part of a diagram containing either @xmath186 or @xmath187 adds a factor of the small parameter @xmath135 , these diagrams do not have to be summed to infinite order .
it is enough to consider only the finite set of them that contribute at order @xmath189 , the set of diagrams with only one @xmath186 or @xmath187 line .
the full set of such diagrams in the non - crossing approximation is shown in fig .
[ mainfig ] and in individual sets in figs .
[ fintr ] , [ fsj ] , [ fadist ] , and [ om6 ] .
of course , to find the total conductivity , one should consider them together with any number of intra - band disorder averaging lines which are already included by replacing @xmath190 by @xmath160 .
after discussing the two formalisms we are now ready to demonstrate their equivalence and to identify the set of chiral basis diagrams that correspond to the various physical effects that have been identified in the more physically transparent semiclassical treatment .
it is important to note that the same result is obtained by doing the calculation in the normal spin basis , as shown in appendix [ vertex_calculation ] , but in that case a given feynman diagram can contain contributions from more than one of the effects identified in the semiclassical theory .
calculations in the band eigenstate basis are necessary to interpret the sbe physics because the sbe approach starts by ignoring interband coherence and adds coherence effects by hand by positing side - jump and anomalous velocity effects . according to eq .
( [ jtotal_i ] ) , to find the part of the hall conductivity not related to impurity scattering one should first find the berry curvature of both bands . in two dimensions ,
only the z - component of the berry curvature is meaningful . from eqs .
( [ basis ] ) and ( [ berrcurv ] ) for the dirac hamiltonian we find @xmath191 where @xmath14 in @xmath192 refers to upper / lower bands .
since the intrinsic contribution comes from the whole fermi sea , both the totally filled negative energy and the partially filled positive energy bands contribute to it . substituting eq .
( [ fpm ] ) into eq .
( [ jtotal_i ] ) we find @xmath193 which coincides with the first term in eq .
( [ sxy2 ] ) .
since the intrinsic contribution to the ahe , as identified in the semiclassical approach , does not depend on impurities , it is expected that this contribution must correspond to the kubo formula contribution from diagrams that do not involve any disorder lines , _
i.e. _ the bubble diagram conductivity with all green s functions bare . in fig .
[ fintr ] we show the diagrams from @xmath194 that are non - zero even without disorder .
the unperturbed green s function is diagonal in the chiral basis given by eq .
( [ basis ] ) ; the band label in the diagram is `` @xmath31 '' for @xmath195 and `` @xmath32 '' for @xmath196 .
the contribution to the hall current due to these diagrams is @xmath197 surprisingly this result coincides with the semiclassical result obtained by evaluating eq .
( [ sxyintr1 ] ) even though we have not yet included the disorder free part of @xmath127 , which is also non - zero in general and must be included as a part of the intrinsic conductivity .
it turns out however that for the case of the 2d dirac model @xmath127 vanishes when the fermi level is above of the gap . to see that we first observe that @xmath127 is exactly the sum of two contributions ,
one of which is the same as the intrinsic conductivity given by the sbe eq .
( [ sxyintr1 ] ) while the other is equal to the intrinsic part of @xmath198 but with the opposite sign : @xmath199 details of the calculation which lead to this result are presented in appendix [ sigmaii_app ] .
this result is of course not surprising when we recognize that eq .
( [ sxyintr1 ] ) expresses the linear response of a clean system , _
i.e. _ the full @xmath159 in the absence of disorder under the influence of an accelerating electric field ; since @xmath159 must equal @xmath200 it follows that @xmath201 .
this observation also means that there is no special intrinsic contribution arising from the fermi surface , where occupation numbers change because of acceleration , because the corresponding terms in @xmath198 and @xmath127 cancel and the final result coincides with the semiclassical prediction eq .
( [ sxyintr1 ] ) , coming from all electrons _ below _ the fermi surface .
we have already calculated the sum of the two integrals in eq .
( [ sigmaiifin1 ] ) and @xmath202 ( see eqs .
[ sxyintr1 ] , [ sxyint ] ) and showed that they have the same value .
it follows that in the metallic regime , when @xmath203 , @xmath127 is identically zero . in the case of the 2d dirac hamiltonian ,
the fermi surface contribution @xmath204 vanishes when the chemical potential is in the gap ( @xmath205 ) , but @xmath206 ( in units @xmath6 ) .
thus @xmath207 is responsible for the quantum hall effect in the dirac band.@xcite since direct inter - band scattering is not energetically allowed in the weak disorder limit of our model , we consider only the coordinate shift effect of scattering in the upper band : @xmath208 where @xmath209 .
evaluation of eq .
( [ delr44 ] ) for the positive energy band leads to the following expression after noting that the absolute magnitude of the momentum is conserved : @xmath210 where @xmath211 is the berry curvature of the upper band , defined in eq .
( [ fpm ] ) .
the average side - jump velocity can be obtained by multiplying the transition rate and the side - jump associated with a particular transition as in eq .
( [ jtotal_sj ] ) : @xmath212 where we have used that in the lowest born approximation @xmath213 .
the side - jump velocity does not produce any contribution to the total current in equilibrium , but in an external electric field the nonequilibrium correction to the distribution function , eq .
( [ solg ] ) , results in a non - zero net current . disregarding a small antisymmetric part
the distribution function correction is @xmath214 , where @xmath215 .
it follows that the side - jump contribution to the sbe hall conductivity is @xmath216 the second side - jump effect discussed in section iii follows from the change of energy of the scattered particle after the coordinate shift at a scattering event in the presence of an external electric field .
it results in the anomalous correction to the distribution function that can be determined from eq.([ganl ] ) . in our case
@xmath217 substituting @xmath218 from eq .
( [ sj1 ] ) and looking for a solution of eq .
( [ ganlb ] ) in the form @xmath219 we find that @xmath220 .
the corresponding contribution to the hall conductivity is @xmath221 for the 2d dirac hamiltonian we find that @xmath222 , in an agreement with ref . and , where we showed that in the absence of direct interband scatterings these contributions must be identical .
thus the total side - jump related conductivity is just twice the one in eq .
( [ vsj44 ] ) : @xmath223 this result coincides with the second term in the total conductivity eq .
( [ sxy2 ] ) . explains the equality of two contributions to hall current.,width=340 ] if we look at the expression for the conductivity @xmath224 , we find two velocity operators . tracing the history of how this expression was derived , the operator @xmath225 appeared when we were looking for its average @xmath226 and the operator @xmath227 appeared from the coupling of the charge current to the electric field .
thus the right velocity vertex in feynman diagrams is associated with the immediate effect of the electric field , such as the acceleration of wave packets ( due to the coupling of electric field to the diagonal velocity ) or the mixing of states from different bands ( due to the coupling to the off - diagonal velocity ) . in turn , the left velocity vertex shows what average velocity is finally induced due to this effect if this diagram is nonzero .
this observation allows us to look for the diagrams responsible for various contributions to the ahe .
the side jump current is due to the accumulation of transverse coordinate shifts .
these shifts are anomalous , _
i.e. _ they arise from inter - band coherence induced by impurities .
in the operator language , they appear because the expectation value of the off diagonal part of the velocity operator becomes nonzero on average .
hence the diagrams responsible for the side jump effect are expected to have off diagonal elements @xmath228 or @xmath229 on their left side . in the sbe
the side jump velocity becomes nonzero on average only after the electric field induces the nonequilibrium correction to the distribution function @xmath230 .
that correction is due to the coupling of the electric field to the diagonal part of the velocity .
hence side jump diagrams should have only diagonal velocity @xmath174 on their right hand side .
there are four different diagrams , shown in fig .
[ fsj ] with such properties .
their calculation , using eq .
( [ rengr ] ) and ( [ vyren ] ) , leads to the same result as the one in eq .
( [ vsj3 ] ) .
the band labels at the right vertex in these diagrams are near the fermi energy in the occupied band .
the identification of the diagrams which contain the contribution obtained in the sbe due to the anomalous distribution that arises from the side - jump , @xmath231 , can be identified by similar reasoning .
for these contributions , the electric field is coupled to the coordinate shift ( and hence to the off - diagonal velocity ) .
the coupling changed the band energy upon scattering , leading to the anomalous correction to the distribution function .
after this the average of the diagonal velocity from such a distribution correction becomes nonzero . hence the corresponding diagrams should have the off diagonal velocity on the right and the diagonal velocity on the left .
there are again four such diagrams in fig .
[ fadist ] whose evaluation leads to the same result as eq .
( [ vsj44 ] ) .
symmetry relations between the feynman diagrams shown in fig .
[ fsj ] and in fig .
[ fadist ] explain why the two contributions are equal to each other .
however , within the semiclassical formalism , the seemingly exact relation between the two contributions is not clear to us .
we emphasize once again that here only four diagrams for each of the two contribution have been evaluated to obtain the result of order @xmath232 and not a partial infinite sum .
( this counting assumes that @xmath160 has already been evaluated separately . )
we also emphasize that although the explicit demonstration of equivalence between these specific feynman diagrams and the side - jump calculations which we have identified on physical grounds is for the dirac model alone , the considerations on which the identification has been made is general , and we believe that the identification is generally valid .
the skew scattering contribution originates microscopically from asymmetry in the scattering rate .
this asymmetry can be estimated for known disorder scattering sources by truncating the born series expansion of the @xmath42-matrix at low order . as discussed in sec .
[ sbe ] , in the lowest born approximation the scattering rate is symmetric with respect to an exchange of indexes .
the first non - zero contribution appears at the next ( 3rd ) order . in eq .
( [ om3 ] ) the terms in parentheses represent the absolute value of the squared @xmath42-matrix element @xmath233 to the order @xmath69 . in fig .
[ fskew3 t ] we show the diagrams representing both such terms from eq .
( [ om3 ] ) .
this contribution vanishes for a gaussian ( white noise ) disorder distribution .
however , one should generally keep this contribution , even if @xmath10 is small , because it leads to a parametrically different and hence experimentally distinguishable contribution to the ahe . at 4-th order in @xmath234
the number of terms that contribute to the asymmetric scattering rate , and hence skew scattering , is non - zero even for gaussian disorder models . in fig .
[ fskew4 t ] we show schematically all the terms that contribute at 4-th order in the non - crossing white noise approximation . for example , the diagram in fig .
[ fskew4t](a ) corresponds to the expression @xmath235
.,width=188 ] in the white noise non - crossing approximation at 4th order in the disorder potential.,width=245 ] the summation of terms in figs .
[ fskew3 t ] and [ fskew4 t ] leads to the following expression for the asymmetric part of the scattering rate : @xmath236 where @xmath237 } \sin(\phi-\phi')\delta ( \epsilon_{{\bf k}}^+ -\epsilon_{{\bf k'}}^+ ) \label{om33}\ ] ] @xmath238^{3/2 } } \sin(\phi-\phi')\delta(\epsilon^+_{{\bf k}}- \epsilon^+_{{\bf k ' } } ) .
\label{om44}\ ] ] the dependence of @xmath239 and @xmath240 on the impurity concentration @xmath29 provides valuable information on the physics behind asymmetric scattering .
although @xmath239 is proportional to @xmath29 and is therefore due to separate incoherent contributions from all impurities , the 4th order gaussian contribution @xmath240 is proportional to @xmath241 and can be interpreted as due to interference between different scattering centers . given the scattering rate
one can calculate the transverse scattering time and the hall current using eqs .
( [ taus ] ) and ( [ cond1 ] ) .
the skew scattering contribution to the hall conductivity is then @xmath242 ^ 2 } - \frac { e^2v_1
^ 3 } { 2\pi n_i v_0 ^ 4 } \frac{\ ( vk_f)^4 \delta } { ( 4\delta^2+(vk_f)^2)^2 } \end{array } \label{jt}\ ] ] this result coincides with the last two terms in eq .
( [ sxy2 ] ) .
note also that the second term in eq .
( [ jt ] ) is inversely proportional to the impurity concentration @xmath29 and thus the transverse conductivity due to this skew scattering becomes large in the clean limit whenever @xmath30 .
in contrast , the first term in eq .
( [ jt ] ) appears even for the gaussian disorder correlations and it is independent of the impurity concentration .
this is a result that has not been previously noticed : _ the skew scattering contribution , as defined through the collision term in the sbe , can give a result independent of the impurity concentration and disorder strength_. of course this contribution is subdominant if disorder is weak .
we have shown above that intrinsic and side - jump related diagrams contain at least one off - diagonal velocity operator element , @xmath243 .
skew scattering is , from this conceptual point of view , the most conventional contribution to the ahe since it depends only on velocity operator matrix elements that are diagonal in the band indices , @xmath244 , or more accurately its renormalized version @xmath245 . in figs .
[ fskew ] and [ om6 ] we show respectively the leading order diagrams for non - gaussian correlations and for gaussian disorder in the non - crossing approximation .
the correspondence between calculations of skew scattering contributions based on the kubo formalism and the sbe becomes obvious when we compare figs .
[ fskew ] and [ om6 ] with terms contributing to the asymmetric scattering rate in figs .
[ fskew3 t ] and [ fskew4 t ] . comparing both sets of figures
shows that the central part in figs .
[ fskew ] and [ om6 ] coincides with the formal perturbation expansion of the asymmetric part of @xmath233 .
evaluating the diagrams in figs .
[ om6 ] and [ fskew ] , using the expressions for the renormalized diagonal velocities @xmath246 given by eq .
( [ vyren ] ) , one obtains the first and second terms in eq .
( [ jt ] ) respectively .
this final result completes our comparison showing the equivalence between the sbe and the kubo formalism .
in this paper we have considered only diagrams without crossing disorder paring lines .
this approximation is self - consistent and justified by ward identities but incomplete .
there is no obvious reason why crossing diagrams like the one in fig .
[ fcross ] are small in comparison with others .
it is expected , however , that crossing diagrams are parametrically different from others and thus the self - consistent non - crossing approximation can be justified in many cases .
another type of terms that we dropped from our discussion is due to higher order non - gaussian correlations .
an example is shown in fig .
[ fskew4 ] .
this diagram leads to current that behaves as @xmath247 on the impurity concentration .
although formally it is small in comparison with similar diagrams from fig .
[ fskew ] , when many different types of impurities are involved , the 3rd cumulant of the disorder potential can be suppressed because different types of impurities may contribute to it with different signs .
in contrast , the 4th order cumulants remain of the same sign and thus can compete with the 3rd one .
we are not aware of any work on crossing diagrams and the 4th order non - gaussian skew scattering contributing to the ahe .
we have compared two different techniques commonly used for anomalous hall current calculations .
one is based on the semiclassical boltzmann equation ( sbe ) and on the heuristic identification of side - jump and anomalous velocity unconventional contributions .
the second approach is systematic linear response theory , which requires evaluation of a kubo formula .
the sbe lacks a systematic derivation in a controlled expansion . in particular , it is not completely obvious that all unconventional effects have been identified or that these effects represent independent contributions that should be added together .
it is based on the assumption that electrons in the diffusive regime can be treated as classical particles and only the collision term and the expression for the velocity need to be calculated quantum mechanically .
this assumption fails , for example , when localization effects are important .
the semiclassical approach , however , has several advantages , such as the attractive transparency of its physical interpretation and its applicability beyond the linear response approximation .
it is well known that in the theory of the longitudinal conductivity , the sbe is equivalent to the self - consistent non - crossing approximation in the kubo formalism .
we have extended this correspondence between sbe and kubo approaches by identifying the feynman diagrams that correspond to the unconventional side - jump and anomalous velocity effects that are important for the anomalous hall effect .
we have established a one to one correspondence between both techniques in leading order of the small parameter @xmath133 .
this direct correspondence between particular diagrams and various side - jump and anomalous velocity effects applies only when working in the kubo formalism with the chiral exact bloch eigenstate representation .
the identification is based in part on the physical transparency of the sbe ahe contributions , has been verified by explicit calculation for the 2d dirac model , and is completely general .
this connection between the two approaches has not been directly established in previous literature , to our knowledge , which contains many farraginous results which often appear to be inconsistent with each other .
in addition , in performing this comparison with the skew scattering contribution , we have discovered that there is a contribution that semiclassically is also related to the asymmetric part of the collision term kernel , but which leads to a contribution independent on the impurity concentration and is non - zero even for gaussian correlations .
the identification of an impurity - independent skew scattering contribution has not been made in previous literature .
although this contribution to the hall current is independent of impurity concentration , like the side - jump and intrinsic contributions , it has a distinct physical meaning and thus has generally different dependence on other parameters , e.g. @xmath248 , as one can see from our results for the 2d dirac hamiltonian . the kubo formalism and the sbe
should be considered as complementary techniques .
the former provides a rigorous justification for the latter .
the latter provides simple semiclassical interpretations of the predictions of the former , which is important in developing intuition that can be applied to more complicated problems .
we demonstrated in this work for the first time how very different approaches of the hall current calculation converge to the same expressions .
we hope that the observations made in this paper will help increase the consistency of various publications devoted to disorder contributions in the ahe .
the authors greatly acknowledge insightful conversations with j. inoue , t. nunner , n. nagaosa , s. onoda , and g. bauer .
this work was supported by the nsf under the grant dmr-0547875 , by the onr under grant onr - n000140610122 , by the kitp nsf grant phy99 - 07949 , by grant no .
poci / fis/58746/2004 in portugal , by the polish state committee for scientific research under grant no .
2 p03b 053 25 , by stcu grant no .
3098 in ukraine , by the welch foundation and by doe grant de - fg03 - 02er45958 .
jairo sinova is a research corporation cottrell scholar .
in this appendix we discuss the procedure to calculate the vertex correction to the velocity operator in the presence of disorder in the normal spin basis .
this ladder vertex corrections is defined by @xmath145 @xmath249 we decompose the operators in the pauli matrix representation @xmath250 , @xmath251 , @xmath252 , where the coefficients can be read from the expression for @xmath253 : @xmath254 where @xmath255 , @xmath256 , @xmath257 , and @xmath258 . we can write a closed equation for the vector @xmath259 after multiplying the equation by @xmath260 from the left and taking the trace : @xmath261 where the vector @xmath262 and matrix @xmath263 are defined as @xmath264,\,\,\,\ , a_{ab}\equiv \frac{n_i v_0 ^ 2}{v}\sum_{{\bf k}}{\rm tr}[\sigma_a
g \sigma_b g ] \label{a_c}\ ] ] here @xmath265 through @xmath266 and we reserve @xmath267 to label @xmath268 .
simplifications of these expressions can be obtained through the use of the following relations @xmath269&= & { \rm tr}[(\delta_{ii'}+i\epsilon_{ii'k}\sigma_k)(\delta_{jj'}+i\epsilon_{jj'k'}\sigma_k')]\nonumber\\ & & = 2(\delta_{ii'}\delta_{jj'}-\delta_{ij}\delta_{j'i'}+\delta_{ij'}\delta_{ji'}),\end{aligned}\ ] ] @xmath270={\rm tr}[i\epsilon_{ijk}\sigma_k \sigma_{j'}]= 2i\epsilon_{ijj'}.\ ] ] for the case of the driac model hamiltonian @xmath271 and @xmath272 .
also @xmath273 and @xmath274 do not have any angular dependence .
the only non - zero terms in @xmath263 and @xmath275 are : @xmath276\equiv
v b\nonumber\\ c_y^y&= & \frac{2n_i v_0 ^ 2 } { v}\sum_{{{\bf k}}}v(g_0^{\rm ret}g_0^{\rm adv}-g_z^{\rm ret}g_z^{\rm adv})\equiv v a\nonumber\\ \nonumber\\a_{00}&=&\frac{n_i v_0 ^ 2}{v}\sum_{{{\bf k } } } ( g_+g_++g_-g_-)\nonumber\\ a_{z0}&=&a_{0z}=\frac{4n_i v_02}{v}\sum_{{{\bf k}}}{\rm re}[g_zg_0]\nonumber\\ a_{xx}&=&a_{yy}=va\nonumber\\ a_{zz}&=&\frac{2n_i v_0 ^ 2}{v}\sum_{{{\bf k}}}(g_0g_0+g_zg_z - g_xg_x - g_xg_x)\nonumber\\ a_{xy}&=&-a_{yx}=vb\nonumber\\ \ ] ] this gives a block diagonal form to the matrix a so in what follows and since @xmath277 we will ignore the @xmath278 terms since irrespective of their value @xmath279 .
the main step we have left to do is to compute @xmath72 and @xmath280 defined above : @xmath281 where we have defined @xmath282 . because @xmath283 and since @xmath284 for @xmath285 and @xmath286 for @xmath287 we can write this using the fact that @xmath288 : @xmath289 for @xmath280 we have @xmath290= -\frac{4n_i v_0 ^ 2 } { v}\sum_{{{\bf k } } } \frac{-\gamma h+\epsilon_f\tilde\gamma}{d_+d_-}\nonumber\\ & = & -4n_i v_0 ^ 2 \left(\frac{\pi\nu_+}{\gamma_+(4\lambda_+^2+\gamma_-^2)}+\right . \nonumber\\&&\left .
\frac{\pi\nu_-}{\gamma_-(4\lambda_+^2+\gamma_+^2)}\right)(-\gamma h+\epsilon_f\tilde\gamma)\nonumber\\ & = & \frac{4 \gamma
cos\theta}{(1+\cos^2\theta)\lambda}+\gamma o(\frac{\gamma^2}{\epsilon_f^2})\,\,\,{\rm for}\ , |\epsilon_f|>h\end{aligned}\ ] ] where we have used the combination @xmath291 .
the matrix a reads ( the quadrant of x and y ) : @xmath292 and putting this back in eq .
[ deltav ] we finally get the renormalized vertex : @xmath293
the off - diagonal dc conductivity can be calculated using the kubo formula representation in terms of causal green s functions in the limit @xmath294 .
we start from @xmath295 with the green s functions which include a self energy related to the scattering from impurities .
the result of integration over @xmath296 can be presented in the following form @xmath297 v_x \frac{\epsilon + \omega + v\bsig \cdot { \bf k}+\delta \sigma _ z - i\gamma _ 0+i\gamma _
+ + i\gamma _ 1\sigma _ z } { \omega
( 2\epsilon + \omega -i\gamma _
-+i\gamma _ + ) } v_y\\ \\ \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \frac{\epsilon + v\bsig \cdot { \bf k}+\delta \sigma _ z - i\gamma _ 0+i\gamma _
+ + i\gamma _ 1\sigma _ z } { 2\epsilon -i\gamma _ -+i\gamma _ + } \\ \\ + \left [ f(\epsilon ) -f(\epsilon + \omega ) ) \right ] \upsilon _ x \frac{\epsilon + \omega
+ v\bsig \cdot { \bf k}+\delta \sigma _ z+i\gamma _ 0+i\gamma _ + -i\gamma _ 1\sigma _ z } { ( \omega + 2i\gamma _ + ) ( 2\epsilon + \omega + i\gamma _ -+i\gamma _ + ) } v_y \\ \\ \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , \frac{\epsilon + v\bsig \cdot { \bf k}+\delta \sigma _ z - i\gamma _ 0+i\gamma _ + + i\gamma _ 1\sigma _ z } { 2\epsilon -i\gamma _ -+i\gamma _ + } \ } \end{array } \label{c2}\ ] ] where @xmath298 , note that the renormalization of the vertex @xmath299 is included only in the last term .
even though the 5th term contains the integration near the fermi level , it should include the unrenormalized vertex .
this is because the corresponding green s functions come with different signs of @xmath300 and @xmath296 only in the last term . for definiteness
, we assumed that the fermi level is located in the upper band , @xmath301 .
the first four terms in ( [ c2 ] ) correspond to the contribution of states below the fermi energy , @xmath302 , which does not have the vertex corrections and is not affected by impurities .
the 1st and 2nd terms are associated with the filled lower dirac band , whereas the 3rd and the 4th are due to the upper band . in the limit of @xmath294 it gives @xmath303 the 5th and 6th terms in ( [ c2 ] )
give two contributions to @xmath304 .
we can identify them as @xmath305 and @xmath306 , respectively , because they are related to integrals of @xmath307 ( i.e. , with @xmath308 in a corresponding product of causal functions ) and @xmath309 [ i.e. , with @xmath310 in @xmath311 . for @xmath312
we find from ( [ c2 ] ) @xmath313 \end{array}\ ] ] as we see , the second term in ( [ c4 ] ) gives different results in the limits of @xmath314 and @xmath315 ( pure and `` dirty '' cases ) .
in the pure case we should also take @xmath316 .
then , as follows from ( [ c4 ] ) , in the pure limit we get @xmath317 .
it should be noted that each of the terms @xmath318 and @xmath319 is nonzero but they exactly cancel each other . for example , calculating @xmath319 we find @xmath320 . in the limit of @xmath314
, we find from ( [ c4 ] ) @xmath321\ ] ] where we used the notation @xmath322 . after substituting the values of @xmath280 , @xmath126 , and @xmath323 we find @xmath324\ ] ]
where @xmath325 .
calculating integral ( [ c3 ] ) we find @xmath326 combining it with ( [ c5 ] ) we obtain finally @xmath327 which coincides with eq.([sxy2 ] ) for @xmath143 .
in calculating @xmath131 it is useful to consider separately integrations over positive and negative @xmath328 , i.e. @xmath329 , where with our simplifications @xmath330 we can get rid of derivative over @xmath328 by integration by parts and noticing that for @xmath331 and positive chemical potential follows that @xmath332 .
thus @xmath333 further simplifications come from the fact that off - diagonal elements of the velocity operator are related to interband matrix elements of the coordinate operator @xmath334)^{+-}=ia_x^{+- } ( \epsilon^{+}_{{\bf k_f}}-\epsilon^{-}_{{\bf k}})$ ] where @xmath335 etc .
such elements have the property @xmath336=f^{\pm}$ ] , where @xmath337 is the berry curvature of the upper / lower sub - band .
after this calculations of @xmath340 is analogous , @xmath341 however for positive @xmath328 follows that @xmath342 and thus extra terms will appear after integration by parts .
@xmath343 where @xmath344 and @xmath345 } \end{array } \label{ibig3}\ ] ] comparing ( [ ibig3 ] ) and ( [ sxyint ] ) we find that they differ only by sign thus we can write that | the anomalous hall effect ( ahe ) is a consequence of spin - orbit coupling in a ferromagnetic metal and related primarily to density - matrix response to an electric field that is off - diagonal in band index .
for this reason disorder contributions to the ahe are difficult to treat systematically using a semi - classical boltzmann equation approach , even when weak localization corrections are disregarded . in this article
we explicitly demonstrate the equivalence of an appropriately modified semiclassical transport theory which includes anomalous velocity and side jump contributions and microscopic kubo - streda perturbation theory , with particular unconventional contributions in the semiclassical theory identified with particular feynman diagrams when calculations are carried out in a band - eigenstate representation .
the equivalence we establish is verified by explicit calculations for the case of the two - dimensional ( 2d ) dirac model hamiltonian relevant to graphene . |
we study the traveling salesman problem in the shortest path metric completion ( _ graphic tsp _ ) of cubic as well as subcubic graphs .
these two cases played a crucial role in some recent developments on _ graphic tsp _
@xcite , @xcite , @xcite , @xcite , @xcite ) .
we shed some light on their inapproximability status and prove explicit approximation hardness bounds of 1153/1152 for the cubic graphic tsp and 685/684 for the subcubic case .
we design new @xmath0-regular gadget amplifier construction yielding the above bounds , and establish also new inapproximability bounds for the cubic and subcubic instances of the ( 1,2)-tsp of 1141/1140 and 673/672 , respectively .
the inapproximability bounds for the ( 1,2)-tsp improve over the bounds of 1291/1290 and 787/786 claimed in @xcite ( see also @xcite ) .
our proof method in this paper depends on improved amplifier construction and two transparent and direct reduction stages , firstly proving approximation lower bounds for the cubic and subcubic instances of the ( 1,2)-tsp , and then connecting it , in a special way , to the cubic and subcubic instances of the graphic tsp .
we call an instance of ( 1,2)-tsp cubic and subcubic if the graph induced by the all weight 1 edges is cubic and subcubic , respectively .
the paper is organized as follows . in section
[ sec : prelim ] , we give some basic definitions . in section [ sec : approx ] , we review some connections between the approximability of the graphic tsp and the ( 1,2)-tsp . in section [ sec : mainresults ] , we formulate our main results , whereas in section [ sec : technique ] , we describe the techniques used in our proofs . in section [ sec : hybrid ]
, we introduce a bounded occurrence constraint satisfaction problem called the hybrid problem . in section [ sec:12tspmax5 ]
, we describe the reduction given in @xcite from the hybrid problem to the ( 1,2)-tsp . in section [ sec:12tspsubcub ] and [ sec:12tspcub ]
, we introduce modular gadget constructions and prove explicit approximation hardness bounds for the ( 1,2)-tsp in subcubic and cubic graphs , respectively .
the inapproximability results for the graphic tsp in cubic and subcubic graphs are given in section [ sec : graphtsp ] . in section [
sec : summary ] , we summarize our results .
given an arbitrary connected undirected graph @xmath1 , we consider the shortest path metric completion @xmath2 and define the graphic tsp problem for @xmath3 as the standard tsp on the metric instance @xmath2 .
equivalently , the graphic tsp is the problem of finding a smallest eulerian spanning multi - subgraph of @xmath3 .
we are interested here in special cases of the above problem for cubic ( 3-regular ) and subcubic ( maximum degree 3 ) .
both cases are known to be @xmath4-hard in exact setting , as the hamiltonian cycle problem is @xmath4-hard for the @xmath0-regular graphs ( cf .
@xcite ) , it can be reduced to both ( 1,2)-tsp and graphic tsp on cubic graphs . in order to describe a ( 1,2)-tsp instance , it is sufficient to specify the edges of weight one . by constructing a graph @xmath1 , the distance of the vertices @xmath5 and @xmath6
is defined to be 1 if @xmath7 and 2 otherwise .
to compute the cost of a tour , it is enough to consider the parts of the tour traversing edges of @xmath3 .
we call a vertex , in which the tour leaves or enters @xmath3 an _ endpoint_. in addition , a vertex with the property that the tour both enters and leaves @xmath3 in that vertex is called _ double endpoint _ , and we count it as two endpoints . if @xmath8 is the number of vertices and @xmath9 is the total number of endpoints , the cost of the tour is @xmath10 since every edge of weight two corresponds to two endpoints . on the other hand , every tour with cost @xmath11 has exactly @xmath9 endpoints .
the graphic tsp for cubic and subcubic graphs is of special interest because of its connection to the famous @xmath12 conjecture on the integrality gap of the metric tsp ( cf .
@xcite , @xcite ) .
recently , the first polynomial time approximation algorithms with approximation factor @xmath12 for the above problem on cubic and subcubic graphs were designed @xcite and @xcite .
this was slightly improved beyond 4/3 bound for the case of @xmath13-connected cubic graphs @xcite .
there was also a remarkable progress on general graphic tsp ( @xcite , @xcite , @xcite ) leading to the approximation factor @xmath14 , cf .
seb and vygen @xcite .
the ( 1,2)-tsp can be viewed as a special case of the graphic tsp . to see this
, we simply augment the subgraph induced by all weight 1 edges in an instance of the ( 1,2)-tsp by a new vertex @xmath15 and add all edges connecting the original vertices with that vertex @xmath15 .
thus , the explicit approximation lower bound of @xmath16 for general ( 1,2)-tsp is also the inapproximabiltity bound for the general graphic tsp .
it is also known that the factor @xmath17 of christofides algorithm @xcite for the general metric tsp is tight for the graphic tsp on cubic graphs . the best up to now approximation factor for (
1,2)-tsp is @xmath18 @xcite ( see also @xcite ) . in this paper , we attack both cubic and subcubic ( 1,2)-tsp and graphic tsp , and will use inherent connections between that problems .
we prove the following explicit inapproximability results .
[ thm:12subcubic ] the subcubic ( 1,2)-tsp is @xmath4-hard to approximate to within any factor less than @xmath19 .
[ thm:12cubic ] the cubic ( 1,2)-tsp is @xmath4-hard to approximate to within any factor less than @xmath20 . for subcubic and cubic instances of the graphic tsp , we prove the following .
[ thm : gr : cubic ] the graphic tsp on subcubic graphs is @xmath4-hard to approximate to within any factor less than @xmath21 . [
thm : gr : subcubic ] the graphic tsp on cubic graphs is @xmath4-hard to approximate to within any factor less than @xmath22 .
the method and techniques of the paper use new modular constructions of simulating gadgets and are built upon some ideas of @xcite and @xcite .
the underlying constructions and their correctness arguments are presented in the subsequent sections .
we start with defining the following hybrid problem ( cf .
@xcite , see also and @xcite ) . given
a system of linear equations mod 2 containing n variables , @xmath23 equations with exactly two variables , and @xmath24 equations with exactly three variables , find an assignment to the variables that satisfies as many equations as possible .
the following result is due to berman and karpinski @xcite .
[ ssphybridsatz ] for every constant @xmath25 and @xmath26 , there exists instances of the hybrid problem @xmath27 with @xmath28 variables , @xmath29 equations with exactly two variables , and @xmath30 equations of the form @xmath31 such that : @xmath32 each variable occurs exactly three times .
@xmath33 it is @xmath4-hard to decide whether there is an assignment to the variables that leaves at most @xmath34 equations unsatisfied , or else every assignment leaves at least @xmath35 equations unsatisfied .
@xmath36 an assignment to the variables in @xmath27 can be transformed in polynomial time into an assignment satisfying all @xmath29 equations with two variables without decreasing the total number of satisfied equations in @xmath27 .
the instances of the hybrid problem produced in theorem [ ssphybridsatz ] have an even more special structure , which we are going to describe . for this
, we are going to introduce the : given a systems @xmath37 of linear equations mod @xmath13 with exactly @xmath0 variables in each equation , find an assignment that maximizes the number of satisfied equations in @xmath37 . + for the , hstad @xcite gave an optimal inapproximability result stated below .
[ thm : max - e3-lin ] for every @xmath38 , there exists a constant @xmath39 and instances of the with @xmath40 equations such that : + @xmath32 each variable in the instance occurs at most @xmath39 number of times . + @xmath33 it is @xmath4-hard to distinguish whether there is an assignment satisfying all but at most @xmath41 equations , or every assignment leaves at least @xmath42 equations unsatisfied . in the following ,
we describe briefly the reduction given in @xcite from the to the hybrid problem and give the proof of theorem [ ssphybridsatz ] . for this ,
let us first recall some definitions ( see also [ bk03 ] ) .
let @xmath3 be a graph and @xmath43 .
we say that @xmath3 is a @xmath44-regular amplifier for @xmath45 if the following two conditions hold : * all vertices in @xmath45 have degree @xmath46 and all vertices in @xmath47 have degree @xmath44 . * for every non - empty subset @xmath48
, we have the condition that @xmath49 where @xmath50 .
we call @xmath45 the set of contact vertices and @xmath47 the set of checker vertices . amplifier graphs are used for proving hardness of approximation for constraint satisfaction problems , in which every variable occurs a bounded number of times .
berman and karpinski @xcite gave a probabilistic argument on the existence of @xmath0-regular amplifiers .
in particular , they constructed a very special amplifier graph , which they called wheel amplifier .
a _ wheel amplifier _ @xmath51 with @xmath52 contact vertices is constructed as follows .
we first create a hamiltonian cycle on @xmath53 vertices with edge set @xmath54 .
then , we number the vertices @xmath55 and select uniformly at random a perfect matching @xmath56 of the vertices whose number is not a multiple of @xmath57 .
the vertices in the matching are our checker vertices and the remaining vertices are our contacts
. the set @xmath58 defines the edge set of @xmath51 .
it is not hard to see that the degree requirements are satisfied .
berman and karpinski @xcite gave a probabilistic argument to prove that with high probability the above construction indeed produces a @xmath0-regular amplifier graph .
[ thm : wheelamp ] with high probability , wheel amplifier are @xmath0-regular amplifier .
let us proceed and give the proof of theorem [ ssphybridsatz ] .
let @xmath38 be a constant and @xmath37 an instance of the , in which the number of occurences of each variable is bounded by @xmath39 .
for a fixed @xmath26 , we can negate some of the literals such that all equations in the instance @xmath37 are of the form @xmath59 , where @xmath60 are variables or negated variables . for a variable @xmath61 in @xmath37 ,
we denote by @xmath62 the number of occurences of @xmath61 in @xmath63 . for each @xmath61
, we create a set of @xmath64 new variables @xmath65 .
in addition , we construct a wheel amplifier @xmath66 on @xmath67 vertices with @xmath62 contacts .
since @xmath62 is bounded by a constant , it can be accomplished in constant time . in the remainder
, we refer to _ contact _ and _ checker _ variables as @xmath68 , whose corresponding index @xmath69 is a contact and checker vertex in @xmath66 , respectively .
let us now define the equations of the new instance @xmath27 of the hybrid problem .
for each edge @xmath70 , we create @xmath71 and refer to equations of this form as _ matching equations_. on the other hand , for each edge @xmath72 in @xmath73 , we introduce @xmath74 .
equations of the form @xmath75 with @xmath76 and @xmath77 are called _ cycle equations _ , whereas @xmath78 is the cycle border equation .
finally , we replace the @xmath79-th occurrence of @xmath61 in @xmath37 by the contact variable @xmath80 , where @xmath81 .
accordingly , we have @xmath82 equations with three variables in @xmath27 , @xmath83 equations with two variables and each variable appears in exactly @xmath0 equations .
we call an assignment to @xmath84 as _ consistent _ if for @xmath85 , we have that @xmath86 for all @xmath87 $ ]
. a consistent assignment to the variables of @xmath27 is an assignment that is consistent for each @xmath84 . by using standard arguments and the amplifier constructed in theorem [ thm : wheelamp ] , we are able to transform an assignment to the variables of @xmath27 into a consistent one without decreasing the number of satisfied equations and the proof of theorem [ ssphybridsatz ] follows .
in this section , we describe the reduction constructed in @xcite from the hybrid problem to the ( 1,2)-tsp . in particular , this construction can be used to prove the following theorem .
[ thm:12gen : ks12 ] the ( 1,2)-tsp is @xmath4-hard to approximate to within any factor less than @xmath88 . in the following ,
we describe briefly the reduction from the hybrid problem to the ( 1,2)-tsp and refer for more details to @xcite and @xcite .
+ + let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath83 equations with two variables and @xmath82 equations with two variables .
in order to simulate the variables of @xmath27 , we introduce for each variable @xmath90 the corresponding parity gadget @xmath91 displayed in figure [ fig:12tsp:1 ] ( a ) . if we start in @xmath92 or @xmath93 , there are two ways to traverse this gadget visiting every vertex only once . in the following ,
we refer to those traversals as _
@xmath94-traversals _ , which are defined in figure [ fig:12tsp:1 ] ( b ) and ( c ) .
c|c|c & & + ( a ) parity gadget @xmath91 & ( b ) @xmath95-traversal & ( c ) @xmath96-traversal the idea of the parity gadgets is that any tour in the instance of the ( 1,2)-tsp can be transformed into a tour , which uses only @xmath94-traversals of all parity gadgets that are contained as a subgraph in @xmath89 without increasing its cost .
the @xmath94-traversal of the parity gadget defines the value that we assign to the variable associated with the parity gadget . for each equation
, we have a specific way to connect the parity gadgets that are simulating the variables of the underlying equation .
let us start with the construction for matching equations .
+ + * matching equations : * given a matching equation @xmath97 in @xmath27 with @xmath98 and the cycle equations @xmath99 and @xmath100 , we connect the associated parity gadgets @xmath101 , @xmath102 , @xmath103 , @xmath104 and @xmath105 as displayed in figure [ fig:12tsp:3 ] .
* equations with three variables : * for equations with three variables @xmath106 in @xmath27 , we use the graph @xmath107 displayed in figure [ fig:12tsp:2 ] . engebretsen and karpinski @xcite introduced this graph and proved the following statement .
[ prop : ek06gadgettsp ] there is a simple path from @xmath108 to @xmath109 in figure [ fig:12tsp:2 ] containing the vertices @xmath110 and @xmath111 if and only if an even number of parity gadgets is traversed .
we now explain how we connect the parity gadgets for @xmath61 and @xmath112 with @xmath107 : let us assume that @xmath113 and @xmath114 are equations in @xmath27 .
we denote the parity gadgets that appear in @xmath107 as @xmath115 , @xmath116 and @xmath117 .
then , we connect @xmath91 and @xmath102 with @xmath115 via edges @xmath118 , @xmath119 .
furthermore , we add @xmath120 to connect @xmath91 and @xmath102 .
if @xmath61 appears negated in the equation with three variables , we create @xmath121 and @xmath122 and @xmath123 .
* cycle border equations : * for each wheel @xmath124 with @xmath125 $ ] , we introduce three vertices @xmath126 , @xmath127 and @xmath128 , which are connected via @xmath129 .
let @xmath130 be the associated set of variables of @xmath124 .
then , we connect @xmath128 with @xmath131 and @xmath132 .
in addition , we add @xmath133 and @xmath134 .
+ for the last wheel , we introduce the path @xmath135 , where @xmath136 is the starting vertex of the graph @xmath137 simulating an equation with three variables .
the graphs corresponding to equations with three variables are connected via vertices @xmath138 , where @xmath139 is the first vertex of the path @xmath140 .
this is the whole description of the corresponding graph @xmath141 .
we are going to prove one direction of the reduction and prove the following lemma .
[ lem : gen12:easy ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equations with two variables and @xmath82 equations with three variables and @xmath143 an assignment that leaves @xmath144 equations in @xmath27 unsatisfied for a constant @xmath145 .
then , there is a tour in @xmath89 with cost at most @xmath146 . according to theorem [ ssphybridsatz ]
, we may assume that all variables associated to a wheel take the same value under @xmath143 .
our tour in @xmath89 starts in @xmath147 and traverses @xmath148 .
then , we use the @xmath149-traversals of the parity gadgets corresponding to the variables of the wheel @xmath150 until we enter the vertex @xmath151 . for each wheel , we use the corresponding traversal defined by the assignment . finally , we get to the vertex @xmath136 , which belongs to the graph @xmath152 .
we refer to this part of the tour as the _
inner loop_. in the remaining part of the tour , we are going to traverse the graphs corresponding to equations with three variables .
if an odd number of parity gadgets was visited in the inner loop , we can find a hamiltonian path in @xmath107 . in the other case
, we have to introduce two endpoints . in the _ outer loop _ of the tour , we visit all gadgets corresponding to equations with three variables . accordingly , our tour has cost at most @xmath153 . in the following ,
we briefly describe the other part of the reduction given in @xcite .
+ + let us first introduce the notion of consistent tours .
we call a ( 1,2)-tour @xmath154 in @xmath89 _ consistent _ if all parity gadgets in @xmath89 are visited by @xmath155 using a @xmath94-traversal . in order to ensure that we can restrict ourselves to consistent ( 1,2)-tours ,
the following statement can be proved .
[ lem:12gen : cons ] let @xmath154 be a tour in @xmath89 .
for every parity gadget @xmath156 in @xmath89 , it is possible to convert efficiently @xmath154 into a tour @xmath157 in @xmath89 , that uses a @xmath94-traversal of @xmath156 , without increasing the cost . due to the following lemma
, we can construct efficiently an assignment if we are given a consistent tour in @xmath89 .
[ lem:12gen : hard ] let @xmath154 be a consistent tour in @xmath89 with cost @xmath158 for some constant @xmath159 . then , it is possible to construct efficiently an assignment that leaves at most @xmath160 equations in @xmath27 unsatisfied .
we are ready to give the proof of theorem [ thm:12gen : ks12 ] .
let @xmath37 be an instance of the max - e3lin2 problem with @xmath8 variables and @xmath161 equations . for all @xmath162 ,
there exists a constant @xmath163 such that if we create @xmath163 copies of each equation , we get an instance @xmath164 with @xmath165 equations and @xmath8 variables with @xmath166 . from @xmath164
, we generate an instance @xmath27 of the hybrid problem consisting of @xmath8 wheels , @xmath83 equations with two variables and @xmath82 equations with three variables .
finally , we construct the associated instance @xmath141 of the ( 1,2)-tsp . given an assignment @xmath143 to the variables of @xmath27 leaving @xmath167 equations unsatisfied with @xmath168 , according to lemma [ lem : gen12:easy ] , there is a tour with length at most @xmath169 . on the other hand ,
if we are given a tour @xmath157 in @xmath141 with cost @xmath170 , it is possible to transform @xmath157 in polynomial time into a consistent tour @xmath154 without increasing the cost by applying lemma [ lem:12gen : cons ] to each parity gadget in @xmath89 .
moreover , due to lemma [ lem:12gen : hard ] , we are able to construct efficiently an assignment , which leaves at most @xmath171 equations in @xmath27 unsatisfied . according to theorem [ ssphybridsatz ]
, we know that for all @xmath172 , it is @xmath4-hard to decide whether there is a tour with cost at most @xmath173 or all tours have cost at least @xmath174 , for some @xmath175 which depends only on @xmath176 and @xmath177 .
by appropriate choices for @xmath178 and @xmath179 , the ratio between these two cases can get arbitrarily close to @xmath88
in this section , we are going to define a new outer loop of the construction from the previous section in order to obtain an instance of the ( 1,2)-tsp in subcubic graphs .
+ + the gadgets simulating equation with three variables in the construction given in @xcite contain vertices with degree 5 .
we are going to replace these gadgets by cubic graphs which we will specify later on . in order to understand the cubic gadgets ,
we first describe a reduction from the to the max-2in3sat problem .
the reduction is straightforward : given an equation of the form @xmath180 , we create three clauses @xmath181 , @xmath182 and @xmath183 . note that if we are given an assignment to @xmath184 and @xmath15 that satisfies the equation mod 2 , then , it is possible to find an assignment to @xmath185 , @xmath186 and @xmath187 that satisfies all three corresponding clauses . in the other case
, we find assignments to @xmath185 , @xmath186 and @xmath187 that make at most two clauses satisfied .
c|c & + @xmath188 modular view of the graph @xmath189 & @xmath190 detailed view of @xmath189 in the next step , we are going to design a gadget that simulates the predicate 2in3sat .
this gadget is displayed in figure [ fig:3 ] ( a ) . the boxes can be viewed as modules , which will be replaced with a parity gadget or a graph with similar properties ( see figure [ fig:2 ] ) . any graph with less vertices and the properties of a parity gadget will lead to improved inapproximability factors for the corresponding problems . note that the graph in figure [ fig:3 ] ( b ) has degree at most @xmath0 .
we are going to prove the following lemma .
there is a hamiltonian path from @xmath191 to @xmath192 in the graph displayed in figure [ fig:3 ] ( a ) if and only if 2 edges with modules are traversed .
there are three possibilities to enter the vertex @xmath193 .
therefore , a hamiltonian path in @xmath189 contains @xmath32 @xmath194 , @xmath33 @xmath195 or @xmath196 @xmath197 . in the case @xmath32 , we are forced to use @xmath198 and then , either @xmath199 and @xmath200 or @xmath201 and @xmath202 . in the case
@xmath33 , we first note that we can not use @xmath203 . due to the degree condition
, we have to use @xmath204 .
the only remaining vertex with degree one is @xmath205 and has to be connected to @xmath191 . in case @xmath196
, we may argue similarly to case @xmath33 .
as for the next step , we introduce a gadget that simulates @xmath206 displayed in figure [ fig:4 ] .
we see that in order to get from the vertex @xmath136 to @xmath207 , we simply use the edge @xmath208 or the three edges which are connecting the two parity gadgets .
c|c & + @xmath188 modular view of the graph @xmath209 & @xmath190 detailed view of @xmath209 we are ready to describe the construction that simulates the equation @xmath210 : we create three copies of the gadget @xmath189 , denoted by @xmath211 , @xmath212 and @xmath213 , to simulate @xmath214 , @xmath215 and @xmath216 . for each @xmath217 $ ] , the vertex set of @xmath218
is defined by @xmath219 . in order to connect those three copies ,
we add the edge @xmath220 for each @xmath221 $ ] . in the next step ,
we create three copies of the gadget @xmath209 , denoted by @xmath222 , @xmath223 and @xmath224 , to simulate @xmath206 , @xmath225 and @xmath226 . for each @xmath217 $ ] , the vertex set of @xmath227
is defined by @xmath228 .
again , we connect those three copies by adding @xmath229 for each @xmath221 $ ] and we also create @xmath230 in order to connect @xmath231 with @xmath222 .
the whole construction is illustrated in figure [ fig:9 ] .
finally , we connect the graphs that we introduced by parity gadgets as follows : for each graph @xmath232 , we create two parity gadgets and connect them to the graph @xmath233 corresponding the clause , in which the variable @xmath234 with @xmath235 appear ( see figure [ fig:10 ] for a detailed view ) .
the parity gadgets , which are associated to the variables @xmath236 , @xmath237 and @xmath15 , are attached to @xmath233 with @xmath238 similarly as in the construction described in section [ sec:12oldconstr ] for the graph @xmath239 .
hence , the parity gadget is also connected to the graph which is associated to the wheel @xmath240 , where @xmath241 .
given an instance @xmath27 of the hybrid problem , we refer to the corresponding instance of the ( 1,2)-tsp in subcubic graphs as @xmath242 .
we are going now to construct a tour from a given assignment and prove the following lemma .
[ lem:12sub : easy ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equation with two variables , @xmath82 equations with three variables and @xmath143 an assignment that leaves at most @xmath171 equations unsatisfied . then , there is a tour in @xmath242 with cost at most @xmath243 given the assignment @xmath143 , we define the inner loop of the tour in @xmath242 in the same way as in lemma [ lem : gen12:easy ] .
this means that some of the parity gadgets which are connected to gadgets simulating equations with three variables may have been traversed in the inner loop of the tour . in the outer loop of the tour ,
if the assignment satisfies the underlying equation @xmath210 , then there is a hamiltonian path traversing all graphs corresponding to @xmath214 , @xmath215 , @xmath216 , @xmath206 , @xmath225 and @xmath226 . for each satisfied equation with three variables , we associate the cost @xmath244 . if the underlying equation is not satisfied , we have to introduce two endpoints .
thus , we associate the cost @xmath245 . summing up , we obtain a tour in @xmath242 with cost at most @xmath246 given a tour in @xmath242 , we are going to construct an assignment to the variables of the corresponding instance @xmath27 of the hybrid problem and prove the following lemma .
[ lem:12sub : hard ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equations with two variables , @xmath82 equations with three variables and @xmath154 a tour in @xmath242 with cost @xmath247 .
then , it is possible to construct efficiently an assignment that leaves at most @xmath171 equations in @xmath27 unsatisfied . in the first step ,
we convert the underlying tour in @xmath242 into a consistent one without increasing its cost .
this is done by applying lemma [ lem:12gen : cons ] to each parity gadget in @xmath242 . in the second step
, we use the same @xmath94-traversals of the parity gadgets in the inner loop of the tour which enables us to construct a tour in the corresponding instance @xmath141 with cost at most @xmath248 finally , we apply lemma [ lem:12gen : hard ] and compute efficiently an assignment that leaves at most @xmath171 equations in @xmath27 unsatisfied .
we are ready to give the proof of theorem [ thm:12subcubic ]
. given @xmath27 an instance of the hybrid problem consisting of @xmath8 wheels , @xmath83 equations with two variables and @xmath82 equations with three variables , we construct in polynomial time the associated instance @xmath242 of the ( 1,2)-tsp .
given an assignment @xmath143 to the variables of @xmath27 leaving @xmath167 equations unsatisfied with @xmath168 , then , according to lemma [ lem:12sub : easy ] , it is possible to find a tour with cost at most @xmath249 .
on the other hand , if we are given a tour @xmath157 in @xmath242 with cost @xmath250 , due to lemma [ lem:12sub : hard ] , we are able to construct efficiently an assignment to the variables of @xmath27 , which leaves at most @xmath171 equations in @xmath27 unsatisfied . similarly to the proof of theorem
[ thm:12gen : ks12 ] , for a constant @xmath251 , we may assume that @xmath252 holds . according to theorem [ ssphybridsatz ] , we know that for all @xmath172 , it is @xmath4-hard to decide whether there is a tour with cost at most @xmath253 or all tours have cost at least @xmath254 , for some @xmath175 that depends only on @xmath176 and @xmath177 .
by appropriate choices for @xmath178 and @xmath177 , the ratio between these two cases can get arbitrarily close to @xmath19 .
this section is devoted to the proof of theorem [ thm:12cubic ] .
given an instance @xmath27 of the hybrid problem with @xmath8 wheels , @xmath256 equations with two variables and @xmath82 equations with three variables , we construct the corresponding graph @xmath242 . in order to convert the instance @xmath242 of the ( 1,2)-tsp in subcubic graphs into an instance @xmath255 of the ( 1,2)-tsp in cubic graphs ,
we replace all vertices with degree exactly two by a path in which all vertices will have degree exactly three .
let us describe this in detail : let @xmath257 be a vertex with degree two in @xmath242 , which is connected to @xmath236 and @xmath237 . replace @xmath257 with the path @xmath258 .
in addition , we add edges @xmath259 , @xmath260 , @xmath261 and @xmath262 . by applying this modification to each vertex of degree exactly two ,
we create a cubic graph and refer to it as @xmath255 .
a modified parity gadget is displayed in figure [ fig:2 ] @xmath188 .
the corresponding traversals are defined in figure [ fig:2 ] @xmath190 and @xmath263 .
c|c|c & & + @xmath188 modified parity gadget & @xmath190 @xmath95-traversal & @xmath263 @xmath96-traversal the following lemma enables us to construct a tour in @xmath255 given an assignment @xmath143 to the variables of the corresponding instance @xmath27 of the hybrid problem with a certain cost that depends on the number on unsatisfied equations in @xmath27 by @xmath143 .
[ lem:12cub : easy ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equation with two variables , @xmath82 equations with three variables and @xmath143 an assignment that leaves @xmath264 equations unsatisfied for some @xmath265 .
then , it is possible to construct efficiently a tour in @xmath255 with cost at most @xmath266 basically , we use the same tour as constructed in lemma [ lem:12sub : easy ] for the graph @xmath242 with the difference that instead of traversing a vertex @xmath257 of degree exactly two in @xmath242 , we have to use the path @xmath267 consisting of @xmath0 more vertices .
thus , if we have given a tour @xmath157 in @xmath242 , that was constructed according to lemma [ lem:12sub : easy ] , we have to add @xmath268 ( for each equation with two variables ) , @xmath269 ( for each equation with three variables ) , and @xmath270 ( for each wheel ) to the cost of @xmath157 and obtain a tour in @xmath255 with cost at most @xmath271 and the proof of lemma [ lem:12cub : easy ] follows .
we are going to prove the other direction of the reduction and give the proof of the following lemma .
[ lem:12cubhard ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equation with two variables , @xmath82 equations with three variables and @xmath154 a tour in @xmath255 with cost @xmath272 .
then , it is possible to construct efficiently an assignment that leaves at most @xmath167 equations in @xmath27 unsatisfied .
let @xmath154 be a tour in @xmath255 with cost @xmath273 .
we are going to show that we can convert efficiently @xmath155 into a tour @xmath274 in @xmath242 with cost @xmath275 .
for this , we consider the path @xmath276 in @xmath255 , where @xmath277 corresponds to the vertex @xmath278 of degree exactly two in the instance @xmath242 . as we want to contract the path @xmath279 into one vertex
, we will ensure that the ( 1,2)-tour is using either the path @xmath280 or @xmath281 .
let us assume that either @xmath111 or @xmath282 is an endpoint , say @xmath111 .
clearly , it implies that there is another endpoint in @xmath283 or @xmath111 is a double endpoint .
we delete all edges of weight 1 that the tour is using and are incident on @xmath111 and @xmath282 .
then , we add @xmath284 , @xmath285 and @xmath286 to connect @xmath287 and @xmath110 by edges of weight 1 .
note that this transformation decreased the total number of endpoints and the cost of the ( 1,2)-tour . by applying this transformation successfully to each such path @xmath279
, we obtain a tour which is using the complete path that corresponds to a vertex of degree 2 in the instance @xmath242 without increasing the cost of the tour . by contracting each path @xmath279 into the vertex @xmath278
, it yields a ( 1,2)-tour in @xmath242 with cost at most @xmath288 . finally , we apply lemma [ lem:12sub : hard ] and obtain an assignment that leaves at most @xmath289 equations in @xmath27 unsatisfied .
analogously to the proof of theorem [ thm:12subcubic ] , we combine lemma [ lem:12cub : easy ] with lemma [ lem:12cubhard ] and obtain theorem [ thm:12cubic ] .
in this section , we are going to give the proof of theorem [ thm : gr : cubic ] and theorem [ thm : gr : subcubic ] .
let @xmath27 be an instance of the hybrid problem .
we first construct the corresponding instances @xmath255 and @xmath242 of the ( 1,2)-tsp in cubic and subcubic graphs , respectively .
each gadget @xmath290 in @xmath242 is replaced by the graph @xmath291 displayed in figure [ fig:5 ] .
we refer to this construction as the graph @xmath292 . in order to obtain an instance of the graphic tsp on cubic graphs
, we use the modified parity gadgets in @xmath291 and denote this instance as @xmath293 .
c|c & + @xmath188 modular view of the graph @xmath294 & @xmath190 detailed view of @xmath294 let us prove one direction of the reductions .
[ lem : grsub : easy ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equation with two variables , @xmath82 equations with three variables and @xmath143 an assignment that leaves at most @xmath171 equations unsatisfied . then , there is a tour in @xmath292 and in @xmath293 with cost at most @xmath295 and @xmath296 , respectively .
let us start with the description of the tour in @xmath292 .
as for the inner loop , we use the same tour as in lemma [ lem:12sub : easy ] .
note that we traversed only edges with weight 1 in the inner loop of the tour in @xmath242 . in the outer loop
, we can not use the same shortcuts as in the ( 1,2)-tsp , since in some cases the weight of an edge can be greater than 2 . to ensure that the cost traversing a gadget corresponding to an equation with three variables increases only by one if the equation is unsatisfied by the assignment , we will use the following trick : consider an equation of the form @xmath297 that is simulated by @xmath214 , @xmath215 , @xmath216 , @xmath206 , @xmath225 and @xmath226 . if we have an assignment that satisfies @xmath297 , then there is also an assignment that satisfies all @xmath298 associated predicates .
furthermore , we see that in the other case , we can find an assignment that satisfies all predicates except exactly one equation with two variables .
in particular , it implies for a tour traversing the gadget @xmath291 simulating @xmath299 that if @xmath300 and @xmath301 holds , we use @xmath302 and @xmath302 , respectively . on the other hand , assuming @xmath303 , we traverse either @xmath304 or @xmath305 .
thus , we use the edge @xmath306 twice increasing the cost only by 1 .
summarizing , given an assignment leaving @xmath307 equations unsatisfied , we find a tour in @xmath242 with cost at most @xmath308 and a tour in @xmath292 with cost at most @xmath295 , since we have to take into account the small detour and add @xmath309 to the cost . under the same conditions , we find a tour in @xmath255 with cost at most @xmath310 and a tour in @xmath293 with cost at most @xmath311 .
we now give the other direction of the reductions and prove the following lemma .
[ lem : grsub : hard ] let @xmath27 be an instance of the hybrid problem with @xmath8 wheels , @xmath142 equation with two variables , @xmath82 equations with three variables , @xmath154 a tour in @xmath292 with cost @xmath312 and @xmath157 a tour in @xmath293 with cost @xmath313 . by using either @xmath154 or @xmath157 , it is possible to construct efficiently an assignment that leaves at most @xmath171 equations in @xmath27 unsatisfied .
let us consider a tour @xmath154 in @xmath292 with cost @xmath312 .
we interpret @xmath154 as a ( 1,2)-tour in @xmath292 with cost at most @xmath312 . in the first step ,
we convert the underlying tour in @xmath292 into a consistent one without increasing its cost by applying lemma [ lem:12gen : cons ] to each parity gadget in @xmath292 . in the second step
, we use the same @xmath94-traversals of the parity gadgets in the inner loop which enables us to construct a tour in the corresponding instance @xmath242 with cost at most @xmath314 . finally , we apply lemma [ lem:12sub : hard ] and construct an assignment leaving at most @xmath315 equations in @xmath27 unsatisfied .
analogously , if we have given a tour in @xmath292 with cost @xmath313 , we convert it into a ( 1,2)-tour without increasing its cost . by applying the contractions defined in lemma [ lem:12cubhard ] , we obtain a ( 1,2)-tour in @xmath292 with cost at most @xmath312 , for which we already know how to construct an assignment with the desired properties . by combining lemma [ lem : grsub :
easy ] and lemma [ lem : grsub : hard ] , we obtain immediately theorem [ thm : gr : cubic ] and theorem [ thm : gr : subcubic ] .
as mentioned before the explicit inapproximability bound of 535/534 ( @xcite,@xcite ) for the ( 1,2)-tsp carries through to the graphic tsp .
we summarize here ( table [ fig : inapproxresults ] ) the results of the paper . .
inapproximability bounds for the instances of ( 1,2)-tsp and graphic tsp . [ cols="^,^,^ " , ]
we provided new explicit inapproximability bounds for cubic and subcubic instances of ( 1,2)-tsp and graphic tsp .
the important question is to improve the explicit inapproximability bounds on those instances significantly .
a bottleneck in our constructions , especially for the cubic case , are the parity gadgets .
using the modularity of the constructions , any improvement of the costs of the parity gadgets will lead to improved inapproximability bounds for the corresponding problems .
the current best upper approximation bound for general cubic instances of graphic tsp is 4/3 ( cf .
@xcite ) . for the special case of 2-connected cubic graphs ,
the bound was recently improved to ( 4/3 - 1/61236 ) @xcite .
how about further improving those bounds ? how about improving the general upper bound of 8/7 @xcite for cubic instances of the ( 1,2)-tsp ?
we thank leen stougie and ola svensson for a number of interesting discussions . | we prove explicit approximation hardness results for the graphic tsp on cubic and subcubic graphs as well as the new inapproximability bounds for the corresponding instances of the ( 1,2)-tsp .
the proof technique uses new modular constructions of simulating gadgets for the restricted cubic and subcubic instances .
the modular constructions used in the paper could be also of independent interest . |
optical network design has evolved over time to meet different challenges : high bandwidth , deployment flexibility , multiuser requirements , etc .
other characteristics were considered secondary at the time and added later , as their need or convenience became more pressing .
security has been one of these secondary requirements .
it was usually taken for granted in optical networks due to the technical difficulty of spying the signal carried by an optical fiber @xcite .
however , technological advances and transformations in the structure of the network changed the panorama and this is no longer true @xcite . on one hand , sensitive detectors and small transponders are nowadays able to perform , at a fraction of the cost and size , the operations that previously needed rack - sized equipment . on the other , certain network architectures simply does not lend itself to security @xcite . as an example , downstream signals in a gigabit passive optical network ( gpon ) arrive to all users and are only dismissed by a well behaved optical network terminator ( ont ) .
nothing prevents them to actually record the signals .
classically , there are several ways to provide security in a network .
the most used one is just to add a cryptographic layer independent of the optical network layer that ciphers all the communications in the network . in a typical
setting , a session key is exchanged in some way , for example using asymmetric ( public - key ) encryption and its underlying infrastructure ( e.g. rsa and certification authorities ) or by having exchanged previously a physical storage media with a pool of keys .
then , the exchanged key is used for a given maximum amount of time or of ciphered data size .
when the pool is exhausted , a new one must be exchanged . in high security settings ,
no single channel is considered safe enough and a mixture of several methods ( physical and rsa , and at different times using different paths ) are used .
while these are well - known techniques , they also have their drawbacks .
for instance , the security of rsa is still an unproven assumption .
although its exponentially difficult nature has still to be really challenged , the continuous growth in power of the algorithms and computers make for a constant revision of the recommended key size .
what once was considered secure during the age of the earth , was actually broken in 17 years @xcite .
the recommendations for security have steadily grown from a few hundred bits to 2048 bits @xcite length or even close to 15 kbits for certain operations that require a level of security equal to symmetric - key algorithms @xcite .
these are even bigger when long term security ( @xmath0 years ) is required .
hence , if a technology is able to produce a continuous supply of high quality symmetric keys all over a network , it would be a welcome addition to its capabilities . in some cases as valuable as bandwidth itself . in this regard ,
the purpose of this paper is to explore a recent technology , namely quantum key distribution ( qkd ) , to provide security in an optical network and , in particular , to propose some means to overcome its main limitation in distance through the application of ideas from network coding .
quantum key distribution @xcite allows two legitimate parties to generate a secret key even in the presence of an eavesdropper .
the key is known only to the parties , since the information leaked to an eavesdropper can be bounded as tightly as they want . hence , qkd is a technology able to distribute information - theoretic secure keys .
the measurement of a quantum system in an unknown state modifies it ( except if the same basis is used for measurement and coding state preparation ) , thus allowing to detect the signals suspected to have been read by an eavesdropper .
no computational complexity assumptions are needed .
just the laws of physics and , of course , common assumptions in cryptographic scenarios ( e.g. the eavesdropper can not control the devices at his will ) .
a qkd protocol requires a quantum channel and an authenticated classical channel .
the quantum channel is a communication channel supporting the transmission of quantum signals that are typically encoded as qubits : two states quantum systems such as the horizontal / vertical polarization states of a single photon .
an optical fiber performs very well as a qubit carrier , hence a purely optical network is a very good candidate to implement a quantum network , i.e. a network based on the exchange of quantum signals .
qkd technology is delicate , since it must deal with signals at the quantum level .
however , it has reached the point where devices able to work unattended during months are commercially available @xcite . as in any media ,
photons suffer an exponential attenuation while propagating through the fiber . in conventional communications , this is usually solved by using an amplifier , but in quantum communications this can not be done , as amplifying a signal is just a measurement made to clone it , something forbidden at the quantum level by the `` no cloning '' principle : if a signal is unknown ( i.e. we do not know for sure whether the qubit codes `` 1 '' or `` 0''the single photon is in a horizontal or vertical state of polarization , for example ) the copy can not be made exact @xcite .
a measurement made to obtain information results in an increase of the error rate in the signal , something that allows the detection of an eavesdropper .
optical fiber is at the core of today s data networks because of its capability to support the bandwidth demand , relatively cheap manufacture , easy deployment and long life .
the bulk of the communications is carried by optical fiber .
optical networks are pervasive nowadays , from long haul to fiber - to - the - home . its bandwidth capacity has put much pressure on the electronics in order to keep pace with the raw transmission rate .
bandwidth , increase in reliability and low power consumption have made preferable to keep as much of the network in the purely optical domain .
therefore , most of the networks being deployed today are passive optical networks ( pon ) .
these can work under many different schemes , like the aforementioned gpon or wdm - pon , but all of them have the characteristic that no active elements perturb the optical path at least under typical metropolitan area distances .
thus , it is possible to create a direct , uninterrupted , optical path among two nodes in the network that is also able to support a quantum channel for qkd .
this allows to seamlessly integrate qkd in optical networks , where a quantum channel can be established between any two points in the network . in this regard ,
a flurry of activity has started in the field @xcite ; nevertheless , the distance limitation is still a problem .
current distance records are around 260 km for a link with experimental superconducting detectors @xcite .
this means the capability to tolerate around 50 db of losses , while commercial implementations are still not capable of going beyond 20 db .
if we take into account the insertion losses of typical optical components ( e.g. filters , multiplexers , splitters , etc .
) , this barely allows to span a single access network km . ] . to alleviate these problems , specific networks devised for qkd
have been proposed @xcite . however , these are not really practical since , on one hand , the cost of using and deploying special infrastructures in populated environments , like in cities , is prohibitively high and , on the other , it does not solve the fundamental problem of limited reach .
only quantum repeaters @xcite would allow to extend the reach without limits , but these are far beyond any practical technology today . beyond quantum repeaters ,
we are forced to use trusted repeaters to overcome these losses . a trusted nodes network @xcite links two places through a series of shorter links that create a secret key .
the key produced in the first link is relayed to the destination by encoding it with the keys created in the intermediate ones .
therefore , the key is known by all intermediate nodes , making the key secure only as long as all these are trusted .
if a spy is successful attacking one , all the key is known .
since this is the only practical possibility , all network testbeds deployed up to date ( e.g. @xcite ) make use of them in their design .
the typical trusted node has a complex design @xcite , it includes several qkd devices ( one to complete the pair needed in each link ) and a computer to do all the associated key management , etc . which makes it even harder to certify @xcite to any security level .
the reliance on all these intermediate nodes makes trusted repeater networks expensive and not acceptable by many users .
we propose here a new approach to alleviate the reliance condition on trusted repeaters and apply it to the case of optical networks using standard components .
our work is based on the new paradigm provided by network coding @xcite .
the introduction of network coding by ahlswede was a complete revolution in network theory : simple flow processing by the nodes allows to improve different scenarios in terms of throughput , needed resources and security .
here we use the idea of network codes to reduce the dependence on the trusted repeaters .
we introduce conditional trust structures that guarantee that , as long as there is no cooperation between specified sets of _ weakly - trusted repeaters _ , the distributed key remains fully secret .
the name underscores the fact that some assumptions are no longer needed .
the paper is organized as follows .
section [ sec : network - coding ] reviews the basics of network coding and the wiretap network model that support our proposal .
we have written the paper without assuming any previous knowledge about network coding , in consequence , this section is to a certain extent self - contained and should allow the reader to grasp the fundamental principles of network coding . in section [ sec : repeaters ] we discuss the limitations of qkd and the need for intermediate repeaters .
these repeaters are currently fully trusted , but using network coding principles we can impose a lighter trust constraint on the repeaters and achieve the same information theoretic security . in section [ sec :
scenarios ] we discuss the integration of weakly - trusted repeaters on passive optical networks and in section v we describe some particular scenarios .
finally , we summarize the discussion and draw some future improvements in section [ sec : conclusions ] .
network coding is a paradigm where the intermediate nodes , instead of simply forwarding the incoming flows through the outgoing paths ( according to some routing algorithm ) , distribute a function of the inputs through each outgoing path .
it was shown in @xcite that linear combinations of the inputs suffice to maximize multicast transmissions .
more generally , linear network codes allow to improve on several other aspects of common networks , in particular their security @xcite .
the application of network coding in optical networks has been widely studied in recent works .
for instance , in @xcite the authors improve the performance , robustness and reliability of unicast and multicast optical networks , while @xcite focus on the new generation of passive optical networks commonly deployed in commercial infrastructures .
let us consider a network over a directed acyclic multigraph @xmath1 defined by the tuple @xmath2 , where @xmath3 is the set of vertices or nodes in the graph and @xmath4 the set of directed edges or links .
we denote the adjacency in the network graph of the node @xmath5 to the edge @xmath6 using the notation @xmath7 .
two nodes in the graph , @xmath8 , can communicate if there exist an edge @xmath6 such that @xmath9 and @xmath10 . for convenience ,
we simplify the network model allowing every link to transmit a symbol taking values in the discrete finite alphabet @xmath11 .
note that by allowing multiple edges between two nodes we can generalize the model to links with different capacities .
the network serves a subset of nodes in the graph called source nodes @xmath12 .
every source @xmath13 generates a random message @xmath14 taking uniformly values from the discrete alphabet @xmath15 .
we call @xmath16 the message jointly generated by all sources and consequently taking values in @xmath17 the direct product of all @xmath18 .
in contrast to forward routing ( see fig . [
fig : network - codes ] ) , routers in the network coding paradigm are allowed to output a function of the incoming flows . if we restrict the functions to linear combinations of the inputs , we can easily deduce that they also represent linear combinations of the source messages .
we associate every edge @xmath6 in the graph with @xmath19 a mapping from @xmath17 to @xmath11 .
finally , let @xmath20 be the set of user nodes , a subset of nodes in the graph .
a user @xmath21 aims to receive with no error @xmath22 the messages sent by @xmath23 , a specific subset of @xmath12 .
we will denote , abusing notation , by @xmath24 and @xmath25 the random symbol sent through the edge @xmath6 and the messages reaching the user @xmath26 , respectively . before we proceed to formalize the security framework ,
let us describe a network consisting of a single qkd link between two users ( see fig . [
fig : qkd - link ] ) . logically , we can consider it to be composed by : ( i ) a private or secure link , in which a random key @xmath27 is exchanged , and ( ii ) a public channel in which a message @xmath28 is sent encrypted with a one time pad between @xmath28 and @xmath27 .
this is equivalent to having a private channel that a source @xmath29 can use to send @xmath28 to a user @xmath26 .
therefore , we will consider every link in a qkd network to be a private link between its neighboring nodes .
this restricts eavesdropping to the intermediate network nodes ; only a curious router can gain access to network messages .
this ability to extend the traditional security perimeter to also cover the communication channel between two qkd nodes is the consequence of the laws of quantum physics and is the key attribute of qkd .
however , this property can not be extended to classical repeaters .
in essence , any classical repeater node in a chain of quantum links gets meaningful information @xcite . in order to prevent the curious routers from accessing information
, we can create extended source messages by adding randomness to the source messages .
formally , the messages are drawn from the direct product of the source alphabet @xmath17 , and a random key alphabet @xmath30 .
the negative effect of the extended source messages is , however , a reduction in the achievable transmission rate , since the part devoted to the key transport increases the security but detracts from the information bandwidth . and a public channel in which the key can help the source send a secret message @xmath31 to the user .
it is equivalent to considering both as a single private link connecting the source and the user . ]
let us consider a set of @xmath32 independent eavesdroppers .
every @xmath33 may receive the messages traversing a fixed collection of nodes , or eavesdropping pattern @xmath34 , in order to recover a subset of the source message @xmath35 . in consequence
an eavesdropper has access to @xmath36 , the messages traversing @xmath34 .
note that the elements in @xmath37 , i.e. eavesdroppers , are elements of the power set of @xmath3 and in consequence potentially overlapping .
we say that the intermediate nodes in the network graph are weakly trusted repeaters ( wtr ) to reflect the assumption that no further cooperation with the eavesdroppers is performed .
following @xcite , a network code is admissible over this _ eavesdrop network _ model if every user node @xmath26 can recover @xmath22 and the information that every eavesdropper @xmath38 holds about @xmath35 does not reduce its entropy : @xmath39 and @xmath40 : @xmath41 where @xmath42 and @xmath43 are shannon s entropy and shannon s conditional entropy , respectively @xcite .
these two conditions are called the secure and decodable conditions .
note that this is a generic definition , the special case in which there is one source , one user and the eavesdropper is interested in the whole message is just one of many possible configurations . a simple scenario that offers immediate gain for qkd using the network coding approach is shown in fig .
[ fig : qkd - simple ] . in this scenario
two parties , the source @xmath29 and the user @xmath26 , exchange a secure key relying in two intermediate nodes , @xmath44 and @xmath45 as depicted in fig .
[ fig : qkd - simple ] .
the source generates a secret message @xmath16 and a secret key @xmath46 both taking values over the finite field @xmath47 . if either @xmath44 or @xmath45 tries to get any information about @xmath16 it is easy to verify that @xmath48 where @xmath49 and @xmath50 are the sets of extended messages traversing @xmath44 and @xmath45 , respectively .
wishes to send a message @xmath28 to the user @xmath26 ( possibly a secret key ) in the presence of @xmath44 and @xmath45 , two intermediate nodes that eavesdrop their incoming and outgoing links . if @xmath44 and @xmath45 limit their eavesdropping activities to non cooperative eavesdropping they have no information about the source message @xmath51 , where @xmath52 is a random message from the random alphabet @xmath30 . ]
the previous scenario can be used to enable multicast key distribution , as shown in fig .
[ fig : qkd - multi ] , without providing any further information to the intermediate routers : the extra links joining @xmath44 and @xmath45 with the second user @xmath53 replicate the links with @xmath54 .
distributes the same secret key to two different users @xmath54 and @xmath53 . ] consider now the scenario depicted in fig .
[ fig : qkd - mac ] . in this scenario , proposed by chan _
_ in @xcite , four nodes ( @xmath55 , @xmath56 , @xmath54 and @xmath53 ) exchange keys pairwise ( @xmath57 between @xmath55 and @xmath53 , and @xmath58 between @xmath56 and @xmath54 ) relying in one randomizing node @xmath44 and one centralized node @xmath45 . in other words , @xmath54 and @xmath53 should be able to recover @xmath58 and @xmath57 , respectively , but not @xmath57 and @xmath58 .
in effect , the users recover the desired message by adding the incoming flows and @xmath59 it should be noted that @xmath60 , @xmath61 but @xmath62 .
that is , the network code is admissible if @xmath45 aims to recover either @xmath63 or @xmath64 but not both . and @xmath56 , and two users @xmath54 and @xmath53 .
the sources , @xmath55 and @xmath56 , transmit @xmath57 and @xmath58 to the users , @xmath53 and @xmath54 , respectively , while no information is leaked to the intermediate nodes or the remaining users . ] in the analysis presented in the previous sections the adversaries have been described as passive eavesdroppers : they are assumed to be able to read the information stream , but not to modify it .
this kind of bound on the eavesdropping power might be a concern because qkd , in absence of classical repeaters , provides their users with stronger security . however , this limitation on the eavesdropper is not actually a problem . in general ,
when dealing with the distribution of messages over a network , the security is studied against @xmath65-bounded adversaries , i.e. adversaries in control of at most @xmath65 nodes .
these so called byzantine adversaries are allowed to listen in their incoming links and output any message on their outgoing links . a network with @xmath65-bounded adversaries is said to be secure if it provides the sources and users with perfect secrecy and perfect resiliency .
perfect secrecy is achieved if the adversaries get no meaningful information , in an information theoretic way , about the exchanged messages .
perfect resiliency means that @xmath65-bounded adversaries are unable to stop the sources from reliably transmitting the messages to the users .
et al . _
@xcite showed that if a network with one - way links has at least @xmath66 node disjoint paths between a source and a user , the source can transmit messages with perfect secrecy and perfect resiliency ( in the presence of @xmath65-bounded adversaries ) .
in the network coding community jaggi _
et al . _
@xcite showed , surprisingly , that if there are @xmath66 node disjoint paths a source can transmit secure messages with rate @xmath67 , where @xmath68 is the network capacity .
that is , the only rate reduction from full capacity is @xmath65 and no extra penalty is paid .
resiliency is a desirable property for qkd networks since a qkd link is in itself non resilient , i.e. qkd links are not protected from denial of service attacks .
however , with current qkd networks dealing with small sets of nodes , the need for @xmath66 node disjoint paths is a strong constraint : for instance , if the network is to be secured against any 3-bounded adversary , it has to provide 10 disjoint paths .
_ propose in @xcite a weaker form of security : the network is secure if it provides with perfect secrecy and message authenticity .
message authenticity is achieved if either the message reaches the user uncorrupted or the user and the source are aware that it is corrupted .
a message authentication schema is proposed such that in a network with @xmath69 disjoint paths , it can still provide message authenticity if there is at least one uncorrupted path , the network can provide message authenticity .
their schema is said to be unforgeable because no intermediate node , or set of nodes , outside from the nodes belonging to the uncorrupt path have any information about the exchanged messages .
this same kind of restriction can be imposed to the network code scenarios ; eavesdropping patterns can be defined such that the eavesdroppers can tap any node except for nodes belonging to one uncorrupted path . with these eavesdropping patterns
the authentication mechanism can be applied with the same security .
qkd devices use qubits as their information carriers .
these qubits are described mathematically as vectors in a two dimensional hilbert space , while in practice , any two dimensional quantum system can be used to encode a qubit . when qubits are used in a quantum communication system , such as a qkd device ,
photons are the usual choice for the physical realization .
properties like polarization , phase or even time can be used to encode a qubit into a single photon , making them a flexible choice that are also reasonably easy to transmit , detect and manipulate : laser diodes , avalanche photo diodes , beam splitters , modulators , etc . , are all the optical components needed .
photon qubits are transmitted through either free space ( air ) or optical fiber , being the latter the logical choice for communication networks . as any signal propagating through an absorbing medium
, photons suffer an exponential attenuation when traveling through the fiber .
this losses are harmful for the extremely sensitive qkd devices , which are made to transmit and manipulate single photons . moreover
, the interaction of the photon qubit with the environment is actually indistinguishable from an eavesdropper manipulation , thus rendering the signal unusable for cryptographic purposes .
this includes any attempt to amplify the signal , which is basically an interaction with the qubit in order to know its state and copy it in many photons , something that can not be done without introducing an error , the same error that is used to rule out the existence of an eavesdropper and that forms the basis of a qkd protocol . under these conditions ,
a qkd device is limited to use fully transparent optical networks without active devices such as amplifiers or electro - optical converters and , even in this case , its maximum reach is limited .
there are many factors that hinder the performance of a qkd system : far from ideal single - photon detectors or emitters are the two main ones ; but even if they could be made perfect and working at a very high speed rate , practical limits of qkd systems would be ultimately set by the absorptions in the quantum channel .
a system working at 10 ghz and with detectors 100% efficient would have a maximum practical reach of 500 km using the best fiber available and working in 1550 nm , the most transparent window .
as an example of the practical qkd limits , we calculated the secret key rate using two different fiber based qkd systems , gys @xcite and clavis @xcite , and present them in fig .
[ fig : secret - key - rate ] .
gys was selected because is a laboratory implementation with typical components and parameters that has been widely used in the literature as a benchmark , while clavis is a qkd system already available in the market .
the latter is the development variant of the i d quantique vectis system , used to provide commercial grade security based on qkd .
the specific parameters for the systems are provided in the caption of fig .
[ fig : secret - key - rate ] .
although it is typical to present the secret - key rate as a function of the distance in km , we show it as a function of the losses , which is the meaningful figure when using qkd in networks instead of a direct point to point link .
this allows us to show ( grayed out areas ) the absorptions of common network components .
this highlights the fact that in real optical networks they are the limiting factor .
[ fig : secret - key - rate ] depicts the key exchange of two nodes that are located in different access networks of a metropolitan area network ( its structure is detailed in section [ sec : scenarios ] ) .
the absorptions are presented in the same order in which a real qubit would find them .
following the standard procedures for computing the secret key rate , the quantum bit error rate is roughly approximated as a function of the losses and the dark count rate which is assumed to be constant ( @xmath70 in the gys experiment and @xmath71 for the clavis system ) . since losses reduce the noise to signal ratio , the secret key rate in fig .
[ fig : secret - key - rate ] can also be regarded as a function of the quantum bit error rate .
the secret key rate shows the amount of secret key extracted per transmitted qubit . to calculate the secret key rate per second , it has to be multiplied by the emitter frequency ( @xmath72 mhz for current commercial qkd systems ) .
as expected , the secret key rate decreases exponentially with the losses up to a point where it goes to zero rather abruptly .
this effect has nothing to do with the optical network , but with the way in which the secret key has to be extracted . when losses are higher , the signal to noise ratio is worse and the quantum bit error rate grows . since in cryptography one has to deal with the worst possible case , all errors have to be attributed to an attacker .
more quantum bit error rate means , then , that in the last step in a qkd communication , known as privacy amplification , a high number of bits have to be discarded to account for the presence of an hypothetical eavesdropper . in any case , fig .
[ fig : secret - key - rate ] clearly shows that there is a maximum number of losses that can be tolerated .
this is approximately 20 db for the commercial system .
this loss budget , that in a point to point link could amount to 80 km of standard optical fiber , is not enough to cross ( without repeaters ) a typical metropolitan area network .
it should be noted that continuous variable qkd is an alternative technology for using in future telecom networks since it is intrinsically more adapted to wdm @xcite , and although it has classically been considered highly limited in distance , recent experiments show that this technology is also capable of achieving the aforementioned 80 km @xcite .
new types of qkd devices could also be well suited for network integration @xcite .
it is to be noted that tolerance to losses and high bit rates are complementary magnitudes , since high speed systems would be , in general , able to reach further away with a useful key rate .
db / km for the fiber ( both devices are transmitting at the 1550 nm window ) .
losses due to network devices are depicted using a shadowed region .
note that the different regions are plotted in the order in which an hypothetical photon qubit would find them if it is connected through an access network homed into a passive metropolitan ring and crosses two core nodes before being dropped to its final destination . ]
modern optical networks have embraced the full optical domain model , hence they are capable of transporting quantum signals .
there have been many studies trying to integrate qkd in optical networks @xcite .
however , the distance limitation still persists . as mentioned in the introduction , there are only two ways to overcome this .
the best one is to build a repeater or amplifier in the quantum regime .
this is a possibility not ruled out by theory and with good experimental progress @xcite but one that , by all accounts , is still many years in the future .
the other one is to build a classical repeater that is actually a measurement device that relays the keys produced in the first quantum link . to extend the reach of qkd ,
an intermediate classical repeater can be used ( see fig . [
fig : qkd - simple - real ] ) .
it is to be noted that a qkd link is composed of an emitter side and a receiver side .
the emitter is typically made up of a laser diode while the receiver contains the single - photon detectors , which are based on avalanche photo diodes .
thus , a classical repeater , @xmath65 , is composed of a receiver and an emitter ( other pairing possibilities exist , but here we stick with the most simple one )
. the receiver links with the previous emitter in the chain and the emitter with the forthcoming receiver ( in the figure , @xmath29 and @xmath26 , respectively ) .
the receiver is used to establish a quantum channel different from the quantum channel built up with the emitter .
both quantum channels create , independently , keys @xmath73 and @xmath74 with the same length .
then , through the public authenticated channel , a message @xmath28 is relayed securely computing a one - time pad with the qkd keys created before ( the logical scheme of this process is depicted in fig . [ fig : qkd - simple - logic ] ) .
this process can continue with any number of intermediate links , but each repeater @xmath65 will know the message being transmitted .
the secure message @xmath28 can be used later as a key to cipher future communications .
unfortunately , it only takes one corrupted node to completely spoil the security of the system .
this full trust condition is not acceptable in many cases .
a message is information theoretically secure if the entropy of the message is not reduced with respect to a well - defined adversary and scenario . in qkd
the eavesdropper is allowed to perform any attack allowed by quantum physics .
however , the laboratories of the legitimate parties are assumed to be trusted and their devices well known .
recent studies show that it is possible to extract a non zero , though significantly smaller , secret key without making any assumption on the device characterization .
this model is known as device independent qkd @xcite .
the secret key distilled by a device independent qkd protocol improves the security of a key distilled by a qkd protocol only in the sense that it reduces the set of hypothesis for information theoretic security .
if we use trusted repeaters to extend the reach of qkd the key is still information theoretically secure provided that the eavesdropper is limited to quantum attacks and the ( well - characterized ) laboratories and devices of bob and alice as well as all the intermediate repeaters are outside of eve s control .
however , the full disclosure implied by trusted repeaters might be too strong in many scenarios . in these situations ,
trusted repeaters are not a valid option .
in contrast , we can only weakly trust the intermediate nodes and consider that some might try to recover information of the secret key or message , i.e. the setting discussed in section [ sec : network - coding ] . at design time
the set of tappable nodes and possible associations is established .
the set can be delimited in a rather precise fashion , e.g. any set of @xmath75 nodes can be tapped , etc .
then , the users can search for a secure network code that fulfills their trust requirements .
nonetheless , the weakly trusted repeater paradigm does not increase the security with respect to trusted repeaters in a quantifiable way , what it offers is the possibility to modulate the number of intermediate repeaters that a user is willing to trust .
it is in the same sense that device independent qkd improves on qkd that the weakly trusted repeaters paradigm improves on the security of the trusted repeaters scenario .
explicit code constructions in the general wiretapping model is an open problem @xcite .
however , in the single source scenario , secure network code constructions are fairly well known @xcite and , in simple cases like the examples from section [ sec : network - coding ] , they can be discovered by inspection .
hence , we can readily use this formalism to reduce the dependence on the intermediate nodes .
assuming a network where nodes can be deployed at will , we can implement directly the first example from section [ sec : network - coding ] ( see fig . [
fig : qkd - simple - wtr ] ) .
the distance limitation of a qkd emitter at the source @xmath29 is denoted by a gray sector
. this limitation can be overcome by introducing two disjoint paths between the source , @xmath29 , and the user , @xmath26 .
in contrast to the trusted repeaters solution , different messages are sent through each path , such that here the key is not fully disclosed to any of the intermediate nodes .
, the grey sector marks the distance limitation of the qkd system with emitter source at @xmath29 . ] unfortunately , despite its benefits , practical optical networks do not have this deployment ease .
instead , nodes have to be arranged in specific structures to fit different geographical , cost , deployment , etc . constraints .
the next scenario focuses on one of these kind of optical networks : metropolitan area networks . as already mentioned in the introduction ,
optical networks are the preferred technology for commercial telecommunication networks .
for instance , in metropolitan areas , optical fiber and passive technology is widely available , making possible to establish an optical path among two nodes .
here we focus on metropolitan area networks using only passive technology , since due to the moderate absorptions , optical paths do not need amplification .
this means that they are not disrupted at the quantum level , hence being suitable for qkd .
metropolitan area networks span from ten to several hundred kilometers .
[ fig : pon-1 ] shows the typical design of such networks : a central ring - shaped backbone connected to peripheral access networks .
the backbone is composed of multiple nodes , separated by tens of kilometers , that connect to the optical line termination ( olt ) of the access networks .
these nodes use different kinds of optical add - drop multiplexers ( oadm ) , for instance reconfigurable oadm , to route signals to the correct access network depending on their wavelength .
starting at the olt , the access networks follow a point to multipoint topology in order to serve nodes , known as optical network units ( onu ) , located in the same zone .
the preferred choice for access networks is passive technology , thus the signal is distributed among onu via passive devices like splitters o arrayed waveguide gratings ( awg ) .
the typical distance between the backbone node and the onu is a few kilometers @xcite . using this structure
, metropolitan area networks can accommodate many users via multiplexing techniques such as wavelength division or time division multiplexing ( wdm and tdm respectively ) .
the use of qkd in optical network is further complicated if quantum and classical signals are multiplexed . since at the quantum level we are dealing with single photon signals ,
any leak of the classical signals , typically 100 db stronger , will spoil the quantum channel .
although there are studies demonstrating this coexistence ( e.g. @xcite ) , its effect basically amounts to increase the noise and , thus , reduce the maximum distance / rate achievable .
being the only purpose of this section to show the benefits of weakly trusted repeaters in optical networks , we prefer to keep the network in the quantum regime and avoid these problems .
thus we assume the network is being used only for qkd purposes .
an authenticated classical channel is supposed to be available among any nodes . to secure a metropolitan optical network
, a quantum channel has to be created among any two nodes of the access networks .
therefore , first we connect an emitter to the end of an access network . looking at fig .
[ fig : secret - key - rate ] and table [ tab : insertion - losses ] , we realize that losses , due to fiber and network components , do not allow to directly plug the receiver in a different access network .
intermediate nodes are needed .
possible locations are the immediate backbone node or its closest neighbors . in the associated graph
, this scheme generates a bipartite graph with emitters placed at the end of the access network and receivers at the backbone nodes .
each emitter has several outgoing links : to the receiver in its own backbone node and the neighboring ones .
the type of qkd device selected for each node is not arbitrary .
receivers are more expensive and difficult to maintain due to the single - photon detectors , hence they are kept at the telco installations .
.insertion losses [ cols="<,<,<",options="header " , ] this shows that telecom networks not only suit qkd , but they are flexible enough to provide several alternative paths among two nodes ( at least in metro areas ) .
therefore , the network coding approach described in section [ sec : network - coding ] can be used .
this implies that a commercial optical network with improved security and resilience , as compared to the traditional scheme , can be designed using qkd and wtr .
a simple scenario where two parties communicate through two intermediate nodes can be derived from the theoretical framework described in section [ sec : network - coding ] . in this scenario
the parties are able to exchange a secret - key under weak trust assumptions .
this simplicity also facilitates the implementation of secure optical networks , like the ones described in the previous section . here
we propose several network prototypes to demonstrate that secure telecom pon can be implemented using qkd and wtr in metropolitan area networks ( see fig .
[ fig : pon-1 ] ) .
a first prototype is depicted in fig .
[ fig : qkd - pon-1 ] .
emitters , labeled as @xmath76 , are connected to user nodes in an access network , and receivers , labeled as @xmath77 , are connected to backbone nodes . in a backbone node ,
several passive wdm components are included to route the quantum signals .
the objective is to reach from an emitter at least two different receivers in order to have several paths for improved security .
a coarse wdm oadm ( cwdm ) is used to route signals within a passband to the corresponding access network . add and drop ports , @xmath78 and @xmath79 in the figure ,
are connected to a dense wdm oadm ( dwdm ) .
the function of this second multiplexer is to filter signals with the wavelengths associated to a receiver . in this wavelength addressing scheme , quantum signals can reach a receiver from two different directions .
both dwdm oadm are connected to a band - pass filter , @xmath80 , using the reflected and filtered ports .
the common port of the @xmath80 filter connects the backbone with the access network , thus routing signals from the access network to the correct destination within the backbone ring . in the access network
several emitters are connected to the splitter .
fiber lengths are assumed to be within the typical distances for metropolitan area networks . even though this prototype works under the wdm paradigm ,
a wavelength multiplexer can not be used because of the backbone cwdm oadm . instead , an splitter is used , as in gpon . in this way
, multiple emitters can communicate simultaneously with different receivers , because each receiver is addressed using a different wavelength .
in particular , in this prototype each emitter can communicate with three receivers , all of them within the range of current commercial qkd systems ( fig .
[ fig : secret - key - rate ] ) : the receiver in the immediate backbone node and both neighboring backbone nodes .
for instance , in fig .
[ fig : qkd - pon-1 ] , @xmath81 is able to communicate with @xmath82 , @xmath83 and @xmath84 .
since communications between farther nodes are unfeasible because of the absorptions , wavelengths for the filters @xmath80 and dwdm oadm can be used repeatedly all over the network .
this reduces the number of required wavelengths and simplifies the network construction .
an example to illustrate the operation mode of this prototype is as follows .
assume that @xmath85 and @xmath81 can not communicate directly because of the absorptions and then they want to exchange a key using weakly trusted repeaters , thus they need at least two intermediate nodes : 1 .
@xmath85 transmits at 1540 nm .
the signal is filtered by @xmath80 and it is dropped at the dwdm oadm , reaching then @xmath82 . the total loss budget , according to table [ tab : insertion - losses ] , is @xmath86 db .
2 . @xmath85 transmits at 1545 nm .
the signal is filtered by @xmath80 , passes the dwdm oadm , exits the backbone node , is dropped by the cwdm oadm and is dropped again by the dwdm oadm in front of @xmath83 .
the total loss budget is @xmath87 db .
@xmath81 transmits at 1545 nm . the signal is reflected by @xmath80 and dropped by the dwdm oadm in order to reach @xmath83 .
the total loss budget is @xmath86 db .
@xmath81 transmits at 1540 nm .
the signal is reflected by @xmath80 , it passes through the dwdm oadm , exits the backbone node , is dropped by the cwdm oadm and dwdm oadm , and then reaches @xmath82 .
the total loss budget is @xmath87 db . the logical diagram of the communication is presented in fig .
[ fig : qkd - scenario - nc-1 ] .
note how it reflects the network coding structure of fig .
[ fig : qkd - simple ] . using the secure keys exchanged through qkd
, a secret message @xmath28 can be exchanged between @xmath85 and @xmath81 through authenticated classical channels .
no information is disclosed to the intermediate ( repeater ) nodes : @xmath82 and @xmath83 .
they act as wtr and they gain no information about the message @xmath28 although they know either @xmath88 or @xmath89 . this simple case can be modified to exchanges between farther nodes .
for instance , fig .
[ fig : qkd - scenario - nc-2 ] shows the case where @xmath85 exchanges a key with @xmath90 .
nodes connected to the same splitter can be handled in a similar way .
, either the quantum signal has to travel the full ring in the forward direction till reaching the node before or a second ring running in parallel but in the backward direction must be used . ] fig . [
fig : qkd - pon-2 ] shows a second network prototype .
this has the advantage of requiring less components than the prototype depicted in fig .
[ fig : qkd - pon-1 ] .
the band - pass filter @xmath80 used to route signals within the backbone is replaced by a circulator , such that signals can now be transmitted in only one direction .
in such a configuration , an emitter can reach only the receiver attached to its immediate backbone node and the first backbone node to the right .
this limitation can be avoided in two ways : ( i ) assuming that the backbone network is a closed ring , such that any emitter can exchange keys with at least one forward receiver .
thus , two disjoint paths can be drawn using the full ring .
( ii ) using an additional identical backbone network to transmit signals in the opposite direction . note that in this prototype only one dwdm oadm is required , but this advantage disappears if we set up a second network to communicate in the opposite direction .
is not able to directly connect beyond @xmath83 ) .
this is not an option with current commercial qkd systems , since their loss budget is as much 20 db , but with the expected new generation , able to withstand 30 db , this advantage except for the increase in throughput between nearest neighbor nodes would be lost . in this case , other schemes like the first two prototypes can be used . ]
a third network prototype is shown in fig .
[ fig : qkd - pon-3 ] .
this improves in terms of the number of users , resources and loss budget .
note that reducing the losses we are also increasing the final secret - key rate in a qkd system , thus , increasing the overall wtr transmission rate too .
the circulator is replaced by an splitter , solving the problem of backward communications .
cwdm oadm are removed since emitters and distant receivers can not exchange secure keys because of the absorptions .
this could be a disadvantage for the next generation of qkd systems which are able to withstand at least 30 db . in this case ,
other schemes like the first two prototypes can be used to exchange keys between distant nodes . as a result ,
backbone nodes require only three components : splitters , oadm and receivers .
now , the splitter in the access network can also be replaced by an awg used as wavelength optical multiplexer .
since the awg is able to multiplex n wavelengths ( channels ) into a single fiber , it increases the number of users up to 40 with less insertion losses . in a basic setup
each user is connected to an output port of the awg .
each one communicates using a particular wavelength .
however , because of the periodicity of the awg , lower and upper spectrum bands are also multiplexed .
thus , each user is actually able to communicate using more than one wavelength : the wavelength within the spectrum band and the corresponding periodic wavelengths .
this allows to address multiple receivers from a single emitter . in this case
, care must be taken in that the passband of the oadm coincides with a band of the awg .
the use of several bands in an awg is a common solution in commercial optical networks for instance to separate downstream from upstream signals @xcite .
although all the examples comply with the 20 db loss budget of current commercial qkd systems , a better tolerance to losses and higher bit rates are expected for the next generation of qkd systems @xcite .
this will allow to put qkd devices in farther away locations and increase the number of disjoint paths , limited to two in the prototypes , thus improving the throughput and security of the network .
moreover , we limit ourselves to the pon realm in order to not have to contend with any active switching elements .
if these are included in the network design , they open new possibilities at the expense of having to deal with routing algorithms that might include other security threats . a rough estimate of the cost , as a function of the number of components , can be used to briefly compare the three proposed prototypes .
on one hand , the third prototype is the one with fewer main network components per branch : an optical divisor ( awg ) , a dwdm oadm connecting the receivers , and a splitter to connect the access and backbone network . on the other ,
the second prototype requires and additional component since the access network is connected to the backbone using a circulator and a cwdm oadm instead of the splitter .
thus , the third prototype is slightly better than the second relative to the number of components .
finally , the first prototype is clearly a more expensive scenario with five main network components per branch .
in this paper we introduce weakly trusted repeaters for quantum key distribution networks . the objective is to overcome the distance limitation of qkd with current technology , while improving on the traditional trusted repeater model used in these cases .
we formalize these wtr using a network coding approach and construct information theoretically secure scenarios , assumed given specific trust structures .
in particular , that there is at least one non - malicious path among the several disjoint paths used . securing networks with wtr reduces the strong full trust dependence that is assumed for traditional trusted repeaters which , in turn , improves the security of applications and services relying on them .
wtr can directly found its niche in organizations with private networks and high security needs , e.g. , telecom companies , banks , military institutions or government agencies . in these cases , although nodes belong to the organization , they can not be fully trusted ( possible eavesdropping in the future ) .
wtr can also be useful in networks infrastructure where each path belongs to a different owner .
in contrast to the private network situation , here all paths are initially weakly trusted .
moreover , we have also shown practical scenarios based on optical networks , and detailed implementations with standard components of typical telecom networks . in these networks ,
wtr can be used along with qkd to exchange secure keys between two users .
in contrast with existing proposals , the structure of wtr is more simple and homogeneous , hence facilitating its industrialization .
these scenarios can be easily extended to exchange secret keys among a higher number of users and through more intermediate nodes . considering a higher number of disjoint paths
would certainly bring more flexibility to the trust structures .
this approach would , as well , open the door to other important topics such as the robustness and resiliency of these networks .
this work has been partially supported by the project quantum information technologies madrid ( quitemad , i d .
p2009/esp-1594 ) , funded by _ comunidad autnoma de madrid _ and , hybrid quantum networks ( hyqunet , i d .
tec2012 - 35673 ) , funded by _
ministerio de economa y competitividad _ , spain .
k. shaneman and s. gray , `` optical network security : technical analysis of fiber tapping mechanisms and methods for detection prevention , '' in _
milcom 2004 , ieee military communications conference _ , vol . 2 , 2004 , pp .
711716 .
d. gutierrez , j. cho , and l. g. kazovsky , `` tdm - pon security issues : upstream encryption is needed , '' in _
ofc / nfoec 2007 , conference on optical fiber communication and the national fiber optic engineers conference _ , 2007 ,
l. g. kazovsky , s .- w .
wong , v. gudla , p. t. afshar , s .- h .
yen , s. yamashita , and y. yan , `` challenges in next - generation optical access networks : addressing reach extension and security weaknesses , '' _ iet optoelectron .
_ , vol . 5 , no . 4 ,
pp . 133143 , 2011 .
d. atkins , m. graff , a. k. lenstra , and p. c. leyland , `` the magic words are squeamish ossifrage , '' in _ proceedings of the 4th international conference on the theory and applications of cryptology : advances in cryptology _ , 1995 , pp
. 263277 .
d. stucki , m. legr , f. buntschu , b. clausen , n. felber , n. gisin , l. henzen , p. junod , g. litzistorf , p. monbaron , l. monat , j .- b .
page , d. perroud , g. ribordy , a. rochas , s. robyr , j. tavares , r. thew , p. trinkler , s. ventura , r. voirol , n. walenta , and h. zbinden , `` long - term performance of the swissquantum quantum key distribution network in a field environment , '' _ new j. phys .
_ , vol . 13 , no . 12 , p. 123001
, 2011 .
p. jouguet , s. kunz - jacques , t. debuisschert , s. fossier , e. diamanti , r. allaume , r. tualle - brouri , p. grangier , a. leverrier , p. pache , and p. painchault , `` field test of classical symmetric encryption with continuous variables quantum key distribution , '' _ opt .
20 , no . 13 , pp . 1403014041 , 2012 .
p. toliver , r. j. runser , t. e. chapuran , j. l. jackel , t. c. banwell , m. s. goodman , r. j. hughes , c. g. peterson , d. derkacs , j. e. nordholt , l. mercer , s. mcnown , a. goldman , and j. blake , `` experimental investigation of quantum key distribution through transparent optical switch elements , '' _ ieee photonics technol .
_ , vol . 15 , no . 11 , pp .
16691671 , 2003 .
d. lancho , j. martnez , d. elkouss , m. soto , and v. martn , `` qkd in standard optical telecommunications networks , '' in _
1st international conference on quantum communication and quantum networking _ , vol .
36 , 2010 , pp . 142149 .
w. maeda , a. tanaka , s. takahashi , a. tajima , and a. tomita , `` technologies for quantum key distribution networks integrated with optical communication networks , '' _ ieee j. sel .
quantum electron .
_ , vol . 15 , no . 6 , pp .
15911601 , 2009 .
t. e. chapuran , p. toliver , n. a. peters , j. jackel , m. s. goodman , r. j. runser , s. r. mcnown , n. dallmann , r. j. hughes , k. p. mccabe , j. e. nordholt , c. g. peterson , k. t. tyagi , l. mercer , and h. dardy , `` optical networking for quantum key distribution and quantum communications , '' _ new j. phys .
_ , vol . 11 , no . 10 , p. 105001
, 2009 .
kitayama , m. sasaki , s. araki , m. tsubokawa , a. tomita , k. inoue , k. harasawa , y. nagasako , and a. takada , `` security in photonic networks : threats and security enhancement , '' _ j. lightwave technol .
_ , vol . 29 , no . 21 , pp .
32103222 , 2011 .
m. peev , c. pacher , r. allaume , c. barreiro , j. bouda , w. boxleitner , t. debuisschert , e. diamanti , m. dianati , j. f. dynes , s. fasel , s. fossier , m. frst , j .- d .
gautier , o. gay , n. gisin , p. grangier , a. happe , y. hasani , m. hentschel , h. hbel , g. humer , t. lnger , m. legr , r. lieger , j. lodewyck , t. lornser , n. ltkenhaus , a. marhold , t. matyus , o. maurhart , l. monat , s. nauerth , j .- b .
page , a. poppe , e. querasser , g. ribordy , s. robyr , l. salvail , a. w. sharpe , a. j. shields , d. stucki , m. suda , c. tamas , t. themel , r. t. thew , y. thoma , a. treiber , p. trinkler , r. tualle - brouri , f. vannel , n. walenta , h. weier , h. weinfurter , i. wimberger , z. l. yuan , h. zbinden , and a. zeilinger , `` the secoqc quantum key distribution network in vienna , ''
_ new j. phys .
_ , vol . 11 , no . 7 , p. 075001
, 2009 .
m. sasaki , m. fujiwara , h. ishizuka , w. klaus , k. wakui , m. takeoka , s. miki , t. yamashita , z. wang , a. tanaka , k. yoshino , y. nambu , s. takahashi , a. tajima , a. tomita , t. domeki , t. hasegawa , y. sakai , h. kobayashi , t. asai , k. shimizu , t. tokura , t. tsurumaru , m. matsui , t. honjo , k. tamaki , h. takesue , y. tokura , j. dynes , a. dixon , a. sharpe , z. yuan , a. shields , s. uchikoga , m. legr , s. robyr , p. trinkler , l. monat , j. page , g. ribordy , a. poppe , a. allacher , o. maurhart , t. lnger , m. peev , and a. zeilinger , `` field test of quantum key distribution in the tokyo qkd network , '' _ opt . express _ , vol .
19 , no . 11 , pp . 1038710409 , 2011 .
s. jaggi , m. langberg , s. katti , t. ho , d. katabi , m. medard , and m. effros , `` resilient network coding in the presence of byzantine adversaries , '' _ ieee trans .
inf . theory _
54 , no . 6 , pp .
25962603 , 2008 .
l. salvail , m. peev , e. diamanti , r. allaume , n. ltkenhaus , and t. lnger , `` security of trusted repeater quantum key distribution networks , '' _ journal of computer security _ ,
18 , no . 1 , pp . 6187 , 2010 .
p. jouguet , s. kunz - jacques , a. leverrier , p. grangier , and e. diamanti , `` experimental demonstration of continuous - variable quantum key distribution over 80 km of standard telecom fiber , '' in _
qcrypt 2012 , 2nd annual conference on quantum cryptography _ , 2012 .
a. ruiz - alba , j. mora , w. amava , a. martnez , v. garca - muoz , d. calvo , and j. capmany , `` microwave photonics parallel quantum key distribution , '' _ ieee photonics j. _ , vol . 4 , no . 3 , pp .
931942 , 2012 .
park , c .- h .
jeong , h .- j .
park , j .- g .
ahn , and k .- h .
song , `` fiber - to - the - home services based on wavelength - division - multiplexing passive optical network , '' _ j. lightwave technol . _ , vol .
22 , no . 11 , pp . 25822591 , 2004 .
d. stucki , n. walenta , f. vannel , r. t. thew , n. gisin , h. zbinden , s. gray , c. r. towery , and s. ten , `` high rate , long - distance quantum key distribution over 250 km of ultra low loss fibres , '' _
new j. phys .
_ , vol . 11 , no . 7 , p. 075003
, 2009 .
a. r. dixon , z. l. yuan , j. f. dynes , a. w. sharpe , and a. j. shields , `` gigahertz decoy quantum key distribution with 1 mbit / s secure key rate , '' _ opt .
express _ ,
16 , no .
1879018979 , 2008 .
n. namekata , h. takesue , t. honjo , y. tokura , and s. inoue , `` high - rate quantum key distribution over 100 km using ultra - low - noise , 2-ghz sinusoidally gated ingaas / inp avalanche photodiodes , '' _ opt . express _
19 , no . 11 , pp .
1063210639 , 2011 . | in this paper we explore how recent technologies can improve the security of optical networks .
in particular , we study how to use quantum key distribution ( qkd ) in common optical network infrastructures and propose a method to overcome its distance limitations .
qkd is the first technology offering information theoretic secret - key distribution that relies only on the fundamental principles of quantum physics .
point - to - point qkd devices have reached a mature industrial state ; however , these devices are severely limited in distance , since signals at the quantum level ( e.g. single photons ) are highly affected by the losses in the communication channel and intermediate devices . to overcome this limitation , intermediate nodes ( i.e. repeaters )
are used .
both , quantum - regime and trusted , classical , repeaters have been proposed in the qkd literature , but only the latter can be implemented in practice . as a novelty ,
we propose here a new qkd network model based on the use of not fully trusted intermediate nodes , referred as _
weakly trusted repeaters_. this approach forces the attacker to simultaneously break several paths to get access to the exchanged key , thus improving significantly the security of the network .
we formalize the model using network codes and provide real scenarios that allow users to exchange secure keys over metropolitan optical networks using only passive components .
moreover , the theoretical framework allows to extend these scenarios not only to accommodate more complex trust constraints , but also to consider robustness and resiliency constraints on the network .
network coding ; quantum key distribution ; passive optical networks ; trusted repeaters . |
quasinormal spectrum of black holes has been extensively investigated during recent years for a great variety of black hole backgrounds and fields , because it is an important characteristic for observation of the gravitational waves @xcite , stability analysis @xcite , ads / cft calculations of temperature green functions @xcite .
special attention has been paid to perturbations of a scalar field @xcite , as a simplest model , when the influence of the spin of the field is neglected . when one considers the charged black hole , the scalar electrodynamic can successfully model the interaction of the charged field with the electromagnetic background of the black hole .
therefore the calculation of the quasinormal modes of charged fields , initiated in @xcite for charged scalar field in the reissner - nordstm and dilaton backgrounds , was continued in the further research @xcite . in particular , in @xcite
the quasinormal modes of the massive charged scalar field were found with the wkb accuracy . in @xcite and @xcite
it was shown that the quasinormal modes , corresponding to the charged scalar field , decay quicker than those of the neutral field unless the black hole charge is larger than some large near extremal value .
this is opposite to the behavior at asymptotically late times , characterized by the so - called `` tail '' decay , when the charged scalar field decays slower , and therefore , dominates at asymptotically late times @xcite .
yet , the quasinormal frequencies calculated for reissner - nordstrm black holes in @xcite with the help of the wkb method need better accuracy and can not be trusted near the extremal limit , especially for the scalar case , because the effective potential is frequency dependent in this case , and wkb equation for the qn frequency must be solved together with a frequency dependent equation determining the position of the maximum of the effective potential .
on the other hand , one has much richer physical situation , when one takes into consideration all the relevant parameters , such as black hole angular momentum and the cosmological term , i.e. when one starts from the kerr - newman - de sitter ( knds ) black hole as a gravitational background . in this paper
we achieve both the above mentioned aims : to find qn modes with very high accuracy , by using the convergent frobenius expansion , and to consider the most general relevant black hole solution of general relativity , knds solution .
the latter gives us dependence of the qn spectrum on a great number of parameters : charge of the black hole @xmath2 , charge of the field @xmath4 , normalized angular momentum of the black hole @xmath1 , the cosmological term @xmath3 .
the paper is organized as follows : sec .
ii gives some basic formulas on knds metric and on radial wave equation for charged massless scalar and dirac fields , and also discusses the system of units we used for showing the qnms .
iii reviews the obtained numerical results . in the conclusion
we summarize the obtained results .
in the boyer - lindquist coordinates the kerr - newman - de sitter metric has the form @xcite
@xmath7 ^ 2 + \frac{\delta_r}{(1+\alpha)^2 \rho^2}(dt - a\sin^2\theta d\varphi)^2,\ ] ] where @xmath8 the electromagnetic background of the black hole is given by the four - vector potential @xmath9 the charged scalar and dirac fields in curved space - time are described by the following equations of motion : @xmath10 @xmath11 where @xmath4 is the charge of particles and @xmath12 is the electromagnetic potential of the background . some manipulation with scalar and dirac equations allow in the some separable form ( see @xcite and references therein ) .
the existence of the killing vectors @xmath13 , @xmath14 , implies the exponential harmonics of the following form @xmath15 , @xmath16 ) . after the separation of the angular , radial and time variables @xmath17
one can obtain the equation for the angular part @xcite @xmath18 { } \frac{d}{dx } ( 1+\alpha x^2 ) ( 1-x^2 ) \frac{d}{dx } + \lambda - s ( 1-\alpha ) + \frac{(1+\alpha)^2}{\alpha } \xi^2 - 2 \alpha x^2 - \frac{(1+\alpha)^2 m^2}{(1+\alpha x^2 ) ( 1-x^2 ) } - \frac{(1+\alpha ) ( s^2 + 2smx)}{1-x^2}\nonumber \\ & + & \frac{1+\alpha}{1+\alpha x^2 } \left [ \ 2s ( \alpha m - ( 1+\alpha ) \xi ) x - \frac{(1+\alpha)^2}{\alpha } \xi^2 -2 m ( 1+\alpha ) \xi + s^2 \
\right]\bigg\}s_{s}(x ) = 0\label{eqn : sx}\end{aligned}\ ] ] where @xmath19 is the spin of the field ( @xmath20 for the scalar and dirac field respectively ) , @xmath21 , @xmath22 , and @xmath23 is the separation constant ( for the non - rotating case @xmath24 , where @xmath25 is positive ( half)integer multipole number ) .
the angular equation can be solved numerically with respect to @xmath23 for each value of @xmath26 by using the three - term recurrence relation @xcite .
the equation for the radial part is @xcite @xmath27 { } \bigg\ { \ \delta_r^{-s}\frac{d}{dr}\delta_r^{s+1}\frac{d}{dr } + \frac{1}{\delta_r}\left(k^2 - is k \frac{d\delta_r } { dr } \right ) % \makebox[-5 mm ] { } + 4\imo s(1+\alpha)\omega r -\frac{2\alpha}{a^2}(s+1)(2s+1 ) r^2 + 2s(1-\alpha)-2 i s q q -\lambda \ \bigg\ } r_s(r ) = 0 , \label{eqn : rr}\ ] ] where @xmath28(1+\alpha)- q q r$ ] .
generally , this equation has five regular singularities : the event horizon @xmath29 , the internal horizon @xmath30 , the cosmological horizon @xmath31 , the spatial infinity and one more singularity at @xmath32 .
one should note , that @xmath33 and @xmath34 are roots of equation @xmath35 . in the limit of
@xmath36 one has , @xmath37 and , therefore , the spacial infinity becomes an irregular singularity @xcite . by the definition ,
the qnms are eigenfrequencies @xmath26 of ( [ eqn : rr ] ) which satisfy the following boundary conditions ( b.c . )
: @xmath38 represents purely ingoing waves at the event @xmath39 horizon and purely outgoing waves at the cosmological horizon @xmath40 ( or the spacial infinity if @xmath41 ) .
now we introduce the new function , which is regular at these two points if the qnm b.c .
are satisfied @xmath42 where @xmath43 .
the appropriate frobenius series are @xmath44 where @xmath45 . substituting ( [ eqn : sing ] ) and ( [ eqn : fb ] ) into ( [ eqn : rr ] )
, one can obtain the @xmath46-term recurrence relation for the coefficients @xmath47 @xmath48 where @xmath46 depends on the black hole parameters .
for the particular schwarzschild case @xmath49 , but for the general case under consideration @xmath46 is higher than 3 .
we decrease the number of terms in the recurrence relation using the _ gaussian eliminations _ : @xmath50 & & c_{j , i}^{(k)}(\omega ) = c_{j , i}^{(k+1)}(\omega)-\frac{c_{k , i}^{(k+1)}(\omega)\ , c_{j-1,i-1}^{(k)}(\omega)}{c_{k-1,i-1}^{(k)}(\omega)}\ , .
\nonumber%\label{gausselimination}\end{aligned}\ ] ] after one finds @xmath51 numerically , he can solve the equation with _ infinite continued fraction _
( see @xcite for more details ) @xmath52 since we can find the separation constant @xmath23 for each particular value of @xmath26 , ( [ invcf ] ) allows to find qnms with the desired precision .
this technique of the qn spectrum calculation was proposed by leaver @xcite . [ cols=">,>,^,^,^,^,^,^",options="header " , ]
in this work , with the help of an accurate convergent frobenius method , we performed an extensive calculation of quasinormal modes of charged scalar and dirac fields for kerr - newman - de sitter black holes and have analyzed the dependence of the qn spectrum upon the great variety of parameters of the black holes @xmath2 , @xmath0 , @xmath1 , of @xmath3-term , and of the field parameters @xmath4 and @xmath19 .
this generalizes a number of previous works when only some of the parameters were considered non - vanishing .
the model we considered might be successful , when considering the interaction of the charged fields with the electromagnetic background of rotating black holes .
99 k. d. kokkotas and b. g. schmidt , living rev .
relativity * 2 * , 2 ( 1999 ) .
r. a. konoplya and a. zhidenko , nuclear physics b777 , 182 ( 2007 ) [ arxiv : hep - th/0703231 ] g. t. horowitz and v. e. hubeny , phys . rev .
d * 62 * , 024027 ( 2000 ) [ arxiv : hep - th/9909056 ] ; d. t. son and a. o. starinets , arxiv:0704.0240 [ hep - th ] ; a. o. starinets , phys . rev .
d * 66 * , 124013 ( 2002 ) [ arxiv : hep - th/0207133 ] ; v. cardoso and j. p. s. lemos , phys .
d * 63 * , 124015 ( 2001 ) [ arxiv : gr - qc/0101052 ] ; r. a. konoplya , phys . rev .
d * 68 * , 124017 ( 2003 ) [ arxiv : hep - th/0309030 ] ; v. cardoso , r. konoplya and j. p. s. lemos , phys .
d * 68 * , 044024 ( 2003 ) [ arxiv : gr - qc/0305037 ] ; g. michalogiorgakis and s. s. pufu , jhep * 0702 * , 023 ( 2007 ) [ arxiv : hep - th/0612065 ] . s. musiri , s. ness and g. siopsis , phys .
d * 73 * , 064001 ( 2006 ) [ arxiv : hep - th/0511113 ] ; i. amado , c. hoyos , k. landsteiner and s. montero , arxiv:0706.2750 [ hep - th ] .
h. r. beyer , commun .
phys . * 221 * , 659 ( 2001 ) [ arxiv : astro - ph/0008236 ] ; l. e. simone and c. m. will , class .
quant . grav .
* 9 * , 963 ( 1992 ) ; a. ohashi and m. a. sakagami , class .
* 21 * , 3973 ( 2004 ) ; r. a. konoplya and a. v. zhidenko , phys .
b * 609 * , 377 ( 2005 ) [ arxiv : gr - qc/0411059 ] ; r. a. konoplya and a. zhidenko , phys .
d * 73 * , 124040 ( 2006 ) [ arxiv : gr - qc/0605013 ] ; a. zhidenko , phys .
d * 74 * , 064017 ( 2006 ) [ arxiv : gr - qc/0607133 ] ; a. zhidenko , class . quant .
* 23 * , 3155 ( 2006 ) [ arxiv : gr - qc/0510039 ] ; e. berti and k. d. kokkotas , phys .
d * 67 * , 064020 ( 2003 ) [ arxiv : gr - qc/0301052 ] . c. ma , y. gui , w. wang and f. wang , arxiv : gr - qc/0611146 ; r. konoplya , phys .
d * 71 * , 024038 ( 2005 ) [ arxiv : hep - th/0410057 ] ; r. a. konoplya and e. abdalla , phys .
d * 71 * , 084015 ( 2005 ) [ arxiv : hep - th/0503029 ] ; r. a. konoplya and r. d. b. fontana , arxiv:0707.1156 [ hep - th ] ; a. lopez - ortega , gen .
grav . * 38 * , 1565 ( 2006 ) [ arxiv : gr - qc/0605027 ] .
r. a. konoplya , phys .
d * 66 * , 084007 ( 2002 ) [ arxiv : gr - qc/0207028 ] .
w. zhou and j. y. zhu , int .
j. mod .
d * 13 * , 1105 ( 2004 ) [ arxiv : gr - qc/0309071 ] .
r. a. konoplya , phys .
b * 550 * , 117 ( 2002 ) [ arxiv : gr - qc/0210105 ] .
x. he and j. jing , nucl .
b * 755 * , 313 ( 2006 ) [ arxiv : gr - qc/0611003 ] .
j. jing , phys .
d * 72 * , 027501 ( 2005 ) [ arxiv : gr - qc/0408090 ] .
s. hod and t. piran , phys .
d * 58 * , 024018 ( 1998 ) [ arxiv : gr - qc/9801001 ] ; s. hod and t. piran , phys .
d * 58 * , 024017 ( 1998 ) [ arxiv : gr - qc/9712041 ] ; h. suzuki , e. takasugi and h. umetsu , prog .
. phys .
* 100 * ( 1998 ) 491 [ arxiv : gr - qc/9805064 ] .
a. zhidenko , phys .
d * 74 * ( 2006 ) 064017 [ arxiv : gr - qc/0607133 ] .
e. w. leaver , proc .
a * 402 * , 285 ( 1985 ) .
b.f . schutz and c.m .
will , astrophys .
291 , l33 ( 1985 ) ; s. iyer and c.m .
will , phys .
d 35 , 3621 ( 1987 ) ; r. a. konoplya , j. phys . stud .
* 8 * , 93 ( 2004 ) ; r. a. konoplya , phys .
d * 68 * , 024018 ( 2003 ) [ arxiv : gr - qc/0303052 ] ; h. onozawa , t. mishima , t. okamura and h. ishihara , phys .
d * 53 * , 7033 ( 1996 ) [ arxiv : gr - qc/9603021 ] .
r. a. konoplya and a. zhidenko , jhep * 0406 * , 037 ( 2004 ) [ arxiv : hep - th/0402080 ] .
v. cardoso and j. p. s. lemos , phys .
d * 67 * , 084020 ( 2003 ) [ arxiv : gr - qc/0301078 ] . | we find the quasinormal modes of the charged scalar and dirac fields in the background of the rotating charged black holes , described by the kerr - newman - de sitter solution .
the dependence of the quasinormal spectrum upon the black hole parameters mass @xmath0 , angular momentum @xmath1 , charge @xmath2 , as well as on values of the @xmath3-term and a field charge @xmath4 is investigated .
special attention is given to the near extremal limit of the black hole charge .
in particular , we find that for both scalar and dirac fields , charged perturbations decay quicker for @xmath5 and slower for @xmath6 for values of black holes charge @xmath2 less than than some threshold value , which is close to the extremal value of charge and depend on parameters of the black holes . |
conformal structures on the 2-torus @xmath1 ( elliptic curves ) are classified by a moduli space , which is the quotient of the upper half - plane @xmath3 by the action of the modular group @xmath2 by linear fractional transformations .
in other words , equivalence classes of elliptic curves correspond to orbits of @xmath2 on @xmath3 .
since @xmath4 is not a smooth manifold , it is often easier to study properties of elliptic curves in terms of the _ action _ of @xmath2 on the smooth manifold @xmath3 .
this note concerns the analogous question for _ complete affine structures _ on @xmath1 .
complete affine structure _ on a manifold @xmath5 is a representation of @xmath5 as the quotient of affine space @xmath6 by a discrete group of affine transformations .
equivalence classes of such structures on @xmath1 identify with the orbits of @xmath2 on @xmath0 for the standard linear action of @xmath2 on @xmath0 .
however , unlike the action on @xmath3 , this action is _ not proper _ and the quotient space is even more badly behaved . in particular
it fails to be hausdorff and since the action of @xmath2 on @xmath0 is ergodic @xcite , it also enjoys a nonstandard borel structure @xcite , @xcite .
the coordinates in @xmath3 , parametrizing elliptic curves , are defined by the periods of holomorphic 1-forms .
analogously , we use periods of parallel 1-forms to define coordinates on the deformation space of complete affine structures on @xmath1 .
these coordinates define a natural differentiable structure on this deformation space , under which it is diffeomorphic to @xmath0 .
( using a different approach baues @xcite noticed that this deformation space is homeomorphic to @xmath0 . ) whereas every smooth deformation of elliptic curves is induced by a _
map into @xmath3 , the analogous result fails for the deformation space of complete affine structures on @xmath1 with respect to the coordinates given by the periods of parallel 1-forms .
we provide an example of a smooth one - parameter deformation of an euclidean 2-torus , in fact an affine fibration , which corresponds to a continuous curve in the deformation space which is _ not _ smooth in the origin .
the complete affine structures on @xmath1 were classified by kuiper @xcite . in this paper
we describe the _ moduli space _ of such structures .
as is often the case in moduli problems , the moduli space can be made more tractable by introducing extra structure in this case a _ marking _ , and the moduli problem for marked structures is simpler .
equivalence classes of marked structures form a space homeomorphic to the plane @xmath0 , which we call the _ deformation space . _ the moduli space
consisting of equivalence classes of structures ( without marking ) is then the orbit space of @xmath0 by the standard linear action of @xmath2 . that the deformation space of complete affine structures on @xmath1 is even hausdorff is in itself surprising .
( the larger deformation space of ( not necessarily complete ) affine structures on @xmath1 is _ not _ hausdorff . )
moreover , in our setup the group actions defining the moduli problem are not reductive , and hence the usual techniques of geometric invariant theory do not apply . indeed ,
the affine structures discussed here all come from affine structures on abelian lie groups which are invariant under multiplication , and the resulting automorphisms of the affine structure form a unipotent group .
+ benzecri @xcite ( see also milnor @xcite ) proved that the only compact orientable surface which admits affine structures is the two - torus @xmath1 . the classification of affine structures on @xmath1 up to affine diffeomorphism , initiated by kuiper @xcite , was completed independently by furness - arrowsmith @xcite and nagano - yagi @xcite . according to kuiper , the complete affine structures on the 2-torus fall into two types .
quotients of @xmath7 by translations we call _ euclidean structures , _ since they admit compatible euclidean ( or flat riemannian ) metrics .
the levi - civita connection of a compatible metric coincides with the flat connection associated to the affine structure .
the euclidean structures fall into a single affine equivalence class , which corresponds to the origin in @xmath0 .
the other structures are more exotic , and do not correspond to levi - civita connections of a ( possibly indefinite ) riemannian metric .
the non - riemannian structures can be characterized as follows .
let @xmath5 be a marked non - riemannian affine 2-torus .
there is a parallel area form @xmath8 which we normalize to have area @xmath9 .
then there exists a vector field @xmath10 , polynomial of degree @xmath9 , such that the covariant derivative @xmath11 is a nonzero parallel vector field . since
parallel vector fields on @xmath5 are determined up to multiplication by a nonzero scalar , we can normalize @xmath12 by requiring that @xmath13 then the interior product @xmath14 is a parallel @xmath9-form on @xmath5 , uniquely determined by the affine structure of @xmath5 .
the cohomology class @xmath15 \in h^1(t^2;{{\mathbb r}})\ ] ] determines the equivalence class of the marked affine manifold @xmath5 .
as non - euclidean structures converge to the euclidean structure , this cohomology class approaches zero , and the period mapping extends to a homeomorphism of the deformation space of _ all _ complete marked affine structures on @xmath1 with @xmath0 .
the case when the periods are _ rationally related , _ that is when @xmath16 $ ] lies in @xmath17 , is particularly notable . in this case
the parallel vector field @xmath12 has closed trajectories , and the homology class of a trajectory is poincar dual to @xmath16 $ ] .
equivalently , @xmath18 integrates to an affine fibration @xmath19 , where @xmath20 is given a complete affine structure .
+ for every complete affine structure on @xmath1 there is a commutative group operation on @xmath1 for which multiplication is affine .
thus every complete affine 2-torus @xmath5 is a group object in the category of affine manifolds .
such objects correspond to associative algebras , providing an alternative approach to the classification . from this viewpoint ,
a marked complete affine structure on @xmath1 corresponds to a commutative associative nilpotent @xmath21-algebra , with a chosen basis corresponding to the marking .
similarly matsushima @xcite showed that the set of all complex affine structures on a fixed _ complex torus _ forms an affine algebraic variety , by identifying each structure with the associative algebra which is formed by the holomorphic vector fields of the torus under the covariant differentiation of the affine structure .
as a related fact the deformation spaces of complete affine structures on ( also higher dimensional ) real tori are homeomorphic to a real algebraic set @xcite . + in figure 1 - 9 we show various tilings of the plane which are obtained from affine group actions corresponding to different values of the period class .
the software used for creating these pictures is available from w. goldman at the experimental geometry lab of the university of maryland .
let @xmath6 denote real affine @xmath22-space with an orientation .
let @xmath23 denote its group of orientation - preserving affine automorphisms .
we use the standard linear representation .
@xmath6 is the affine hyperplane @xmath24 and @xmath23 the subgroup of @xmath25 consisting of matrices @xmath26 where @xmath27 and @xmath28 .
the affine action is defined by @xmath29 an _ affine structure _ on an oriented smooth manifold @xmath5 is a maximal atlas of coordinate charts into @xmath6 such that on overlapping coordinate patches the transformation is locally the restriction of an element of @xmath23 .
the manifold @xmath5 together with an affine structure is called an _ affine manifold . _
a local diffeomorphism @xmath30 between affine manifolds is _ affine _ if it is related to affine isomorphisms by the local affine coordinate charts .
given a local diffeomorphism @xmath31 , an affine structure @xmath5 determines a unique affine structure on @xmath32 such that @xmath31 is an affine map . thus covering spaces , and the universal covering space @xmath33 in particular , of affine manifolds
are themselves affine manifolds .
we shall assume for convenience that our affine manifolds are path - connected and oriented .
let @xmath5 be an affine manifold .
the affine coordinate atlas on @xmath5 globalizes to a _ developing map _ @xmath34 from the universal covering @xmath35 into @xmath6 .
this affine map determines the affine structure on @xmath5 , and is uniquely determined ( for a given affine structure ) by its restriction to a coordinate patch .
conversely , any affine chart on a coordinate patch extends to a unique developing map .
let @xmath5 be an affine manifold and @xmath36 .
affine germ _ at @xmath37 is an equivalence class of affine diffeomorphisms @xmath38 where @xmath39 is a neighborhood of @xmath37 in @xmath5 and @xmath40 is open .
affine diffeomorphisms @xmath41 and @xmath42 define the same germ if and only if they agree on @xmath43 .
clearly @xmath23 acts simply transitively on the set of affine germs at @xmath36 .
since an affine germ determines a unique developing map , the developing map of a affine manifold is unique up to composition with an affine transformation @xmath44 .
deck transformations of @xmath45 act by affine automorphisms of @xmath35 and define the _ affine holonomy representation _
@xmath46 for which @xmath47 is equivariant .
the _ developing pair _
@xmath48 is unique up to the action of @xmath23 where @xmath23 acts by left - composition on @xmath47 and by conjugation on @xmath49 .
an affine manifold is _ complete _ if its developing map is a diffeomorphism .
equivalently , an affine manifold is _ complete _ if it is affinely diffeomorphic to a quotient space of @xmath6 by a discrete group @xmath50 of affine transformations acting properly on @xmath6 .
affine crystallographic group _ is a discrete subgroup of @xmath23 acting properly with compact quotient .
the classification of complete affine manifolds up to affine diffeomorphism is equivalent to the classification of affine crystallographic groups up to affine conjugacy .
our goal is to construct a space whose points represent inequivalent affine structures on a given closed manifold @xmath51 . unfortunately such spaces are typically non - hausdorff quotients of singular spaces which are not smooth manifolds .
thus we replace our goal with finding a smooth manifold ( or at least a smoothly stratified space ) with a group action , whose orbits represent affine diffeomorphism classes of affine structures on @xmath51 . to compare different affine manifolds , define a _ marked affine manifold _ as an orientation - preserving diffeomorphism @xmath52 where @xmath5 is an oriented affine manifold .
the group @xmath53 of all orientation - preserving diffeomorphisms @xmath54 acts by composition with @xmath31 on the set of affine manifolds which are marked by @xmath51 : @xmath55 two affine manifolds @xmath56 and @xmath57 marked by @xmath51 are considered isomorphic if there exists an affine diffeomorphism @xmath58 such that @xmath59 , they are _ isotopic _ if there exists an affine diffeomorphism @xmath58 such that @xmath60 and @xmath61 are related by a diffeomorphism in the identity component @xmath62 of @xmath53 .
let @xmath63 be a basepoint , @xmath64 the corresponding universal covering space , and @xmath65 its group of deck transformations .
let @xmath56 be an affine manifold marked by @xmath51 . an affine germ @xmath66 $ ] at @xmath67 determines a developing pair @xmath48 , which is defined on @xmath68 .
this development pair depends only on the affine isomorphism class of the pair @xmath69\big)$ ] .
let @xmath70 denote the set of all affine isomorphism classes of pairs @xmath69\big)$ ] .
one can alternatively think of @xmath70 as the set of developing pairs @xmath48 .
since the developing map @xmath47 determines the affine holonomy @xmath49 , one can further regard @xmath70 as the set of developing maps of affine structures on @xmath51 .
affine holonomy defines a map @xmath71 which is invariant under the subgroup @xmath72 of @xmath62 fixing @xmath73 , since the elements of @xmath72 induce the identity map on @xmath74 .
these spaces are topologized as follows .
give @xmath70 the @xmath75-topology of developing maps : a sequence of developing maps converges if and only if it and all of its derivatives converge uniformly on compact subsets .
give @xmath76 the quotient topology .
if @xmath77 is finitely generated , then @xmath78 is a real affine algebraic set , and has the classical topology .
a sequence @xmath79 converges if and only if for all finite subsets @xmath80 , and all compact subsets @xmath81 , the restrictions of @xmath82 to @xmath83 converge uniformly for @xmath84 . in these topologies , @xmath85 is continuous .
thurston @xcite was the first to realize that @xmath85 defines a local homeomorphism @xmath86 that is , @xmath85 is an open map and two nearby developing maps with identical holonomy are isotopic by a basepoint - preserving isotopy .
see @xcite for further discussion of this general fact . in general , the holonomy map @xmath87 is not injective , even if @xmath51 is compact .
the simplest example occurs when @xmath88 and the holonomy has cyclic image ( covering spaces of hopf manifolds ) .
see smillie@xcite , sullivan - thurston @xcite and goldman @xcite for more exotic examples .
however , the restriction of @xmath87 to _ complete structures _ is injective and identifies complete structures with their holonomy representations .
since @xmath77 is discrete , a representation @xmath89 defines a _ proper action _ if for all compact @xmath90 , the set @xmath91 is finite . if furthermore @xmath49 defines a _ free action , _
then @xmath77 is torsionfree .
in particular the quotient @xmath92 is hausdorff , the quotient map @xmath93 is a covering space , and @xmath92 inherits a smooth structure from @xmath6 .
let @xmath94 denote the subset of @xmath78 consisting of proper actions @xmath49 such that the quotient @xmath92 is diffeomorphic to @xmath51 .
let @xmath95 denote the subgroup of @xmath96 which induces the identity on @xmath97 .
[ prop : hol ] let @xmath98 denote the subspace of @xmath70 corresponding to complete affine structures .
then the restriction @xmath99 is a homeomorphism .
we sketch the proof .
let @xmath100 .
a diffeomorphism @xmath101 defines a complete structure on @xmath51 with holonomy @xmath49 , showing surjectivity of @xmath87 . if @xmath102 is a marked affine structure on @xmath5 such that the holonomy @xmath49 is the holonomy of a marked complete affine structure @xmath103 then the development map @xmath104 descends to a local diffeomorphism @xmath105 inducing the isomorphism @xmath106 defined by @xmath49 .
since @xmath51 is closed , the local diffeomorphism is a covering space and since @xmath107 is an isomorphism , @xmath108 is a diffeomorphism .
the composition @xmath109 induces the identity on @xmath110 .
this completes the sketch .
evidently @xmath85 is @xmath23-equivariant , where @xmath23 acts by composition of the affine germ on @xmath70 and by conjugation on @xmath78 .
the space @xmath111 of isomorphism classes of marked affine manifolds @xmath56 identifies with the quotient @xmath112 , and @xmath85 defines a continuous map @xmath113 note that @xmath114 acts on @xmath78 by right composition on homomorphisms . furthermore @xmath85 is equivariant with respect to the @xmath96-action on @xmath70 and the action on @xmath78 defined by the homomorphism @xmath115 the space of isotopy classes of marked complete affine manifolds @xmath56 identifies with the quotient @xmath116 and is called the _ deformation space_. the @xmath96-action on @xmath117 factors through the discrete group @xmath118 and the homeomorphism @xmath87 of proposition [ prop : hol ] induces a surjective map @xmath119 the first space is the _ moduli space , _ consisting of affine diffeomorphism classes of complete affine manifolds which are diffeomorphic to @xmath51 . in this paper
we show that when @xmath120 , then this space identifies naturally with the quotient of @xmath0 by @xmath2 . in general ,
neither @xmath111 nor @xmath121 is hausdorff .
two - dimensional hopf manifolds provide counterexamples . namely the quotients @xmath122 are affine 2-tori .
all the structures with @xmath123 are equivalent , but are not equivalent to @xmath124 .
in particular the equivalence class of @xmath125 contains the equivalence class of @xmath124 in its closure .
the conventions in this paper were chosen to simplify the discussion of our specific example : complete affine structures on @xmath1 .
in general , one may to want to work with markings which are only homeomorphisms or homotopy equivalences , or with manifolds which are not oriented ( or nonorientable ) .
we have chosen to work in the oriented differentiable category for convenience , thereby avoiding as many topological technicalities as possible .
similarly one may want to avoid the discussion of base - point preserving diffeomorphisms in discussing the mapping class group ( see 2 ) , but in the interest of brevity we have chosen the present approach to avoid the general complications .
also in the specific simple case @xmath88 which is treated in this paper , we have that @xmath94 coincides with the subset of @xmath78 consisting of _ all _ proper actions of @xmath77 , and the condition on the quotient space is actually superfluous .
in fact , if @xmath126 then @xmath127 must be diffeomorphic to the 2-torus from the classification of surfaces .
( see @xcite for further discussion , and generalization to higher - dimensional examples . )
moreover , since @xmath1 is a surface , @xmath128 , see 2 .
an affine structure is equivalent to a flat torsionfree affine connection . recall that an _ affine connection _ is a covariant differentiation operation on vector fields @xmath129 which is @xmath21-bilinear , and for @xmath130 , satisfies @xmath131 ( where @xmath132 denotes the directional derivative of @xmath31 with respect to @xmath133 ) .
the connection is torsionfree if and only if @xmath134\ ] ] and flat if and only if @xmath135}.\ ] ] conversely any such affine connection arises from an atlas of locally affine coordinates as above .
( compare @xcite . )
covariant differentiation extends to other tensor fields by enforcing identities such as the following .
for example , if @xmath136 is a @xmath9-form , and @xmath137 , then @xmath138 is the @xmath9-form defined by : @xmath139 for all @xmath140 . if @xmath141 are @xmath9-forms , then @xmath142 ( and linearity ) defines @xmath143 for any exterior @xmath144-form @xmath136 .
a tensor field @xmath136 is _ parallel _ if @xmath145 for all @xmath140 .
equivalently , in local affine coordinates the coefficients of @xmath136 are constants .
more generally @xmath136 is _ polynomial of degree @xmath146 _ if and only if @xmath147 for all @xmath148 .
let @xmath51 be the two - torus @xmath1 , that is the quotient space of @xmath0 by the action of the integer lattice @xmath149 by translations .
the quotient map @xmath150 is a universal covering space , thereby providing global euclidean coordinates for @xmath68 .
let the image of the origin @xmath151 be the basepoint @xmath152 .
a developing map for @xmath1 is thus a local diffeomorphism @xmath153 which is equivariant with respect to the translation action of @xmath154 on @xmath0 , and an affine action defined by a homomorphism @xmath155 : @xmath156 where @xmath157 for @xmath158 and @xmath159 generate the fundamental group @xmath160 .
the _ mapping class group _ of @xmath1 is particularly tractable .
every orientation - preserving diffeomorphism @xmath161 is isotopic to a basepoint - preserving diffeomorphism , and every isotopy between basepoint - preserving diffeomorphisms can be deformed to a basepoint - preserving isotopy . furthermore the natural map @xmath162 corresponds to an isomorphism @xmath163 see 6.4 of stillwell @xcite for discussion .
@xmath2 is explicitly realized by _ linear _
diffeomorphisms of @xmath1 as follows .
a matrix @xmath164 defines a linear automorphism of @xmath0 which normalizes the action of integer translations @xmath165 .
thus @xmath166 induces a diffeomorphism of the quotient @xmath167 , which preserves the basepoint @xmath73 .
thus the extension @xmath168 is split .
since translations of @xmath0 are affine transformations , @xmath165 acts as an affine crystallographic group and @xmath1 inherits a complete affine structure from the structure of @xmath0 ( its universal covering ) as a vector space .
the generators @xmath169 are translations by linearly independent vectors in @xmath0 spanning a lattice in @xmath0 .
furthermore any two lattices in @xmath0 are equivalent by a linear automorphism of @xmath0 , so different lattices in @xmath0 define affinely diffeomorphic structures .
this affine structure is called the _ euclidean structure _ since it supports more refined geometric structures modelled on euclidean geometry .
( of course , most linear automorphisms will not preserve this more refined metric structure . ) the affine equivalence class of the euclidean structure will be the origin in the deformation space . the mapping class group @xmath2 fixes this affine equivalence class
. however , @xmath1 carries many other non - equivalent affine structures .
these structures are not subordinate to riemannian metrics , so we call them _ non - riemannian . _
all the structures arise from simply transitive unipotent affine actions , and we briefly review the general theory . by smillie
@xcite or fried - goldman - hirsch @xcite , any proper affine action of an abelian ( or , more generally , nilpotent ) group is _ unipotent .
_ that is , the linear parts are unipotent linear transformations .
equivalently the corresponding elements of @xmath170 are unipotent .
furthermore a subgroup consisting of unipotent elements is conjugate to the upper triangular unipotent subgroup of @xmath170 . if @xmath171 is an affine crystallographic group which is _ nilpotent ,
_ then there exists a unique simply connected _ unipotent _ subgroup @xmath172 of @xmath23 containing @xmath50 which acts simply transitively on @xmath6 .
indeed , @xmath172 is the _ algebraic hull _ of @xmath50 in @xmath23 , that is , the zariski closure of @xmath50 in @xmath23 .
let @xmath173 denote the lie algebra of @xmath172 .
the exponential map @xmath174 is a diffeomorphism . for any @xmath175 ,
the composition @xmath176 is a _ polynomial diffeomorphism _ of the vector space @xmath173 with the affine space @xmath6 , where .
@xmath177 denotes evaluation at @xmath178 .
hence any two unipotent affine crystallographic actions of a group @xmath50 are conjugate by a polynomial diffeomorphism of @xmath6 ( @xcite , theorem 1.21 ) .
furthermore this implies that the de rham cohomology of the quotient @xmath179 is computable from the cohomology of polynomial exterior differential forms on @xmath5 ( @xcite ) . since @xmath50-invariant polynomial tensor fields on @xmath6
are invariant under the algebraic hull @xmath172 , every tensor at the origin @xmath175 extends uniquely to a @xmath50-invariant polynomial tensor field on @xmath6 .
conversely , any two @xmath50-invariant polynomial tensor fields which agree at @xmath37 are equal .
thus the space of polynomial vector fields on @xmath5 identifies with the tangent space @xmath180 , the space of polynomial @xmath181 on @xmath5 identifies with the dual space @xmath182 . since a unipotent subgroup of @xmath23 preserves volume ,
the top - dimensional cohomology is represented by a _
parallel volume form . _ here is an explicit construction for @xmath183 .
every unipotent subgroup is conjugate to a subgroup of the multiplicatively closed affine subspace @xmath184 where @xmath185 is the subalgebra of strictly upper triangular matrices .
the algebra @xmath186 satisfies @xmath187 , and thus the exponential and logarithm maps are given by : @xmath188 and @xmath189 respectively .
if @xmath190 , then @xmath191 taking @xmath192 to be real extends this to a definition of the unique one - parameter subgroup in @xmath193 containing the cyclic group @xmath194 . in particular if @xmath195 generate an action of @xmath165 , then @xmath196 for @xmath197 .
taking @xmath198 defines the extension of a unipotent @xmath165-action to a unipotent @xmath0-action .
( this is a special case of the _ malcev extension _ of a finitely generated torsionfree nilpotent group to a 1-connected nilpotent lie group .
see malcev @xcite , or chapter ii of raghunathan @xcite for the general construction . ) for the application to complete affine manifolds with virtually polycyclic fundamental group , see 1 of @xcite .
let @xmath199 .
the mappings @xmath200 define a one - parameter subgroup of _ polynomial diffeomorphisms _ of @xmath7 .
for @xmath201 , let @xmath202 denote translation by @xmath203 : @xmath204 @xmath205 conjugates translations to the affine action @xmath206 defined by : @xmath207 in the standard linear representation , this affine representation equals : @xmath208 the polynomial one - parameter group @xmath205 on @xmath7 interacts with three linear one - parameter groups on on @xmath7 : * _ dilations _ defined by @xmath209 for @xmath210 ; * _ shears _ defined by @xmath211 for @xmath212 ; * _ homotheties _ defined by @xmath213 for @xmath214 .
dilations and shears commute with @xmath205 , while homotheties conjugate the various elements of the one - parameter group @xmath215 : @xmath216 implies that if @xmath123 , then the group @xmath217 is affinely conjugate to @xmath218 . when @xmath123 , different lattices in @xmath0 give inequivalent affine structures ( unlike the euclidean case @xmath219 ) .
the map @xmath205 is a _ polynomial developing map _ for this structure .
every complete affine structure on @xmath1 arises from this construction .
denote the group of orientation - preserving linear automorphisms of @xmath0 by @xmath220 .
given @xmath221 and @xmath199 , we define a marked affine manifold @xmath56 as follows .
represent @xmath88 as @xmath222 , where @xmath160 acts by integer translations .
then @xmath205 defines a developing map @xmath223 and a holonomy homomorphism @xmath224 @xmath225 the corresponding affine manifold @xmath5 is the quotient @xmath127 and the marking @xmath31 is the diffeomorphism induced by @xmath226 above .
the malcev construction extends every proper affine action of @xmath165 on @xmath7 to a simply transitive affine action of @xmath0 on @xmath7 . by kuiper @xcite , furness - arrowsmith @xcite , nagano - yagi @xcite , fried - goldman @xcite , this subgroup is conjugate to the subgroup @xmath227 the affine representations @xmath228 define simply transitive affine actions and the evaluation map @xmath229 is the developing map for an _ invariant affine structure _ on the abelian lie group @xmath172 .
the automorphisms of the affine structure of @xmath172 which normalize the multiplications in @xmath172 are induced by the normalizer @xmath230 of the image of @xmath172 in @xmath231 .
this normalizer is generated by two commuting one - parameter subgroups @xmath232 of dilations and shears @xmath233 defined in and respectively .
when @xmath219 , the corresponding normalizer is larger : in this case the normalizer of the translation group is the full affine group .
every automorphism of the vector group @xmath0 is an element of @xmath234 .
it follows that the two pairs @xmath235 ( where @xmath158 ) determine the same point in @xmath236 if and only if : * @xmath237 ; or * @xmath238 for @xmath158 and @xmath239 for some @xmath210 and @xmath212 . an oriented non - riemannian complete affine 2-torus @xmath5 admits a unique parallel area form @xmath8 of area @xmath9 , as well as a unique parallel @xmath9-form @xmath18 satisfying the conditions of the following lemma .
we call @xmath18 the _ canonical parallel @xmath9-form _ on @xmath5 .
[ lem : conditions ] let @xmath5 be a 2-torus with a non - riemannian complete affine structure and parallel 2-form @xmath8 such that @xmath240 then there exists a unique parallel vector field @xmath12 such that @xmath241 for a polynomial vector field @xmath10 and @xmath242 there is a unique parallel @xmath9-form @xmath18 such that @xmath243 where @xmath244 denotes interior multiplication .
furthermore @xmath10 is unique up to addition of a constant multiple of @xmath12 . choose coordinates so that @xmath245 where @xmath246 is a lattice .
dilations distort area by @xmath247 so replacing @xmath50 by @xmath248 , we may assume that @xmath249 induces an area form on @xmath5 with area @xmath9 .
the @xmath50-invariant polynomial vector fields on @xmath7 are @xmath250-invariant , and @xmath251 is a basis for the @xmath50-invariant polynomial vector fields on @xmath7 . every parallel vector field is a constant multiple of @xmath252 .
furthermore @xmath253 satisfy : @xmath254 let @xmath255 be the vector fields on @xmath5 induced by @xmath256 respectively .
this establishes existence .
every polynomial vector field is of the form @xmath257 and determines a parallel vector field @xmath258 which satisfies @xmath259 thus the coefficient in is @xmath260 and conditions uniquely determine @xmath261 .
finally uniquely determines @xmath18 .
let @xmath56 be a marked complete affine 2-torus such that @xmath5 is non - riemannian .
let @xmath18 be its canonical parallel @xmath9-form .
let @xmath262 be the standard basis for @xmath160 .
the pair @xmath263 where @xmath264 for @xmath158 , is an invariant of @xmath56 .
call @xmath265 the _ period class _ of @xmath56 .
if @xmath5 is euclidean , define @xmath265 to equal @xmath266 .
suppose that @xmath267 .
since the images of @xmath268 generate @xmath269 , the cohomology class @xmath16\in h^1(m;{{\mathbb r}})$ ] is zero and @xmath18 is exact .
thus @xmath270 for a smooth function @xmath271 .
since @xmath18 is nowhere vanishing , @xmath272 is a submersion , contradicting compactness of @xmath5 .
suppose that @xmath56 is a non - riemannian marked torus , defined by parameters @xmath276 , @xmath277 , as in @xmath278 .
since @xmath205 is area - preserving @xmath279 induces an area form on @xmath5 which has total area @xmath280 .
therefore @xmath8 is the area form on @xmath5 induced by @xmath281 a basis for the @xmath282-invariant polynomial vector fields on @xmath7 is @xmath283 and we compute @xmath284 for which @xmath285 induces polynomial vector fields @xmath255 satisfying . define @xmath286 so that @xmath287 .
thus implies @xmath288 so the canonical vector field , and canonical parallel @xmath9-form on @xmath5 are represented by @xmath289 respectively .
the period class for @xmath290 $ ] is : @xmath291 _ @xmath292 is surjective : _ we construct an inverse @xmath293 to @xmath292 as follows .
let @xmath294 .
define @xmath295 where @xmath296 as usual .
then @xmath297 is a polynomial developing map , where @xmath298 is the orthogonal matrix @xmath299 when @xmath300 .
( compare . ) explicitly , note also that @xmath306 .
hence the corresponding holonomy representation @xmath307 depends smoothly on @xmath308 .
furthermore @xmath309 equals the conjugate of @xmath310 by @xmath311 , where @xmath312 by , the period class of @xmath313 equals @xmath308 .
thus the composition @xmath314 is the identity map on @xmath0 , and @xmath292 is surjective .
note also that @xmath293 is surjective . indeed
, since every element @xmath315 decomposes as a product of an orthogonal and an upper - triangular matrix , @xmath316 implies that every element of @xmath236 may be represented by a holonomy representation with parameters @xmath317 , where @xmath318 is an orthogonal matrix .
therefore , ( [ qi ] ) shows that @xmath293 is a continuous bijection onto @xmath236 . _
@xmath292 is continuous : _ it follows from ( [ eq : univdev ] ) that the set @xmath319 is closed in the space of development maps @xmath320 , and hence @xmath236 carries the quotient topology from @xmath321 .
since @xmath293 is a bijection , @xmath321 is actually homeomorphic to @xmath236 .
write @xmath322 . since @xmath323{{\partial^2 \over { \partial y}^2 } \psi^2_{{\mathsf{k } } } }
\ ; \mbox { and } \ ; k_2 = \sqrt[3]{{\partial^2 \over { \partial x}^2 } \psi^1_{{\mathsf{k } } } } \ ; , \ ] ] @xmath324 is continuous . therefore , @xmath292 is a continuous function on the deformation space as well . _
@xmath292 is injective : _ let @xmath325 $ ] be a point in in @xmath236 .
if @xmath267 , then by lemma [ lem : nonexact ] , @xmath5 must be euclidean .
all euclidean structures correspond to a single point in @xmath236 .
explicitly , the action of the modular group @xmath2 on @xmath327 is given as follows .
let @xmath56 be an affine manifold marked by @xmath88 , and @xmath328 the diffeomorphism of @xmath1 which corresponds to @xmath329 .
then the action of @xmath166 on @xmath236 is represented by @xmath330 .
we assume that @xmath332 is of the form @xmath333 , for some linear map @xmath334 with determinant @xmath9 . by ( [ eq : periodformula ] ) , @xmath335 , where @xmath336 is the second row vector of @xmath337
. then @xmath338 and @xmath339 .
we construct a _
differentiable family _ of complete affine tori over the deformation space : for @xmath340 , we let @xmath341 denote the holonomy homomorphism for the developing map @xmath342 ( see proof of theorem [ thm : periods ] ) .
we let @xmath165 act on @xmath343 via @xmath344 by , @xmath165 acts differentiably ( in fact by polynomial diffeomorphisms ) , and the quotient space @xmath345 is a differentiable manifold , and a torus - bundle over @xmath0 via the projection onto the first factor .
this bundle is trivial as a differentiable torus bundle .
the diffeomorphism @xmath346 induces an explicit trivialization @xmath347 the bundle @xmath348 admits a system of locally trivializing charts in @xmath343 which define an affine structure on the fibers , giving a differentiable family of affine manifolds over @xmath0 ( see @xcite for the definition of differentiable family of _ complex _ manifolds ) .
each fibre of the corresponding differentiable family @xmath349 is a marked complete affine torus which represents its base - point in the deformation space .
not every differentiable family of complete affine tori is induced by a _ smooth _ map into the deformation space .
indeed , for @xmath350 , let us consider the development map @xmath351 defined by @xmath352 then @xmath353 , @xmath350 , constitutes a differentiable family of developing maps , and there is a corresponding differentiable family of affine manifolds @xmath354 where each fibre @xmath355 , @xmath350 , is a complete affine torus marked by @xmath353 .
( since the parameter @xmath356 enters linearly , the family @xmath357 defines an affine fibration , see @xcite . )
the corresponding curve @xmath358 is a line in @xmath0 , parametrized by @xmath359 it is not smooth in @xmath360 .
canary , d.b.a .
epstein , p. green , _ notes on notes of thurston _ , analytical and geometric aspects of hyperbolic space , london mathematical society lecture note series * 111 * , cambridge university press , ( 1984 ) . | periods of parallel exterior forms define natural coordinates on the deformation space of complete affine structures on the two - torus .
these coordinates define a differentiable structure on this deformation space , under which it is diffeomorphic to @xmath0 .
the action of the mapping class group of @xmath1 is equivalent in these coordinates with the standard linear action of @xmath2 on @xmath0 . |
the complementarity principle has traditionally been addressed in order to illustrate the role of observer in quantum mechanics .
one of the most intriguing aspect of the issue is the possibility of delayed - choice experiments @xcite .
according to it , the experimenter may decide for instance to display wavelike aspects of an atomic system , long after it has been forced into a particle - like behavior @xcite .
if the atomic state is entangled with a memory system available to the observer , i.e. , a which - way detector ( wwd ) , the experimenter may decide to postpone his decision even after the atom has been detected .
interference fringes can be recovered , provided the which - way information ( wwi ) is erased from the wwd @xcite .
an experimental demonstration of a quantum eraser have been given using entangled photon pairs @xcite , using single photon entanglement between spatial mode and polarization state @xcite , and using atom interferometry @xcite .
the great progress in cavity quantum electrodynamics offers the possibility of designing a quantum detector to retain which - way information for rather long time after the system has been detected @xcite . on the other hand
, microwave resonators can couple strongly to rydberg atoms . as a matter of fact , high - finesse microwave resonators have been used to generate einstein - podolsky - rosen pairs of atoms @xcite and in the construction of quantum logic - gates @xcite .
thus , quantum optical ramsey interferometers are good candidates for complementarity experiments .
several models of cavity - qed methods for demonstrating complementarity have been proposed .
some of them supplement the young double - slit experiment with wwd based on dispersive atom - field interaction @xcite .
other proposal uses two atomic beams as wwd of each other using a cavity field in order to entangle them @xcite .
dispersive cavity - qed experiments were reported in @xcite and more recently in @xcite , where the quantum - classical limit of complementarity have been explored using a cavity field continuously tuned from microscopic to macroscopic regimes . a simplified model for quantum erasing
have been found in @xcite .
we present in this paper a novel quantum - erasing scheme : the quantum self - eraser , based on a different strategy .
the innovative idea behind is to use time reversibility to undo the interactions that deposited the wwi into the wwd .
thus , we launch back the same atom ( or a velocity reversal replica of it ) to act as the eraser atom ( erason ) .
controlling time reversibility , we can let the erason absorb or erase its own wwi that was deposited previously on the wwds .
this scheme will emphasize the role of entanglement swapping in quantum erasing operation .
moreover , the phase difference @xmath0 between the interfering alternatives is mapped into the non - local phase of a superposition between the two macroscopically separated cavity - field states .
the self - eraser probes this phase , allowing for a joint demonstration of complementarity and the generation of a non - local coherent macroscopic superposition . finally , we analyze the relation between the violation of temporal bell inequalities as expression of the establishment of macroscopic coherence , and the erasability of the wwi in the quantum self - eraser .
the paper is organized as follows . in section 2 ,
complementarity in a quantum optical ramsey interferometers ( qori ) is discussed in order to set up the notation . in section 3 , we prepare the stage for the discussion showing how the qori scheme can be used to prepare in an unitary fashion a non - local field state .
section 4 is devoted to probe of the phase of the macroscopic superposition via quantum self - erasing .
section 5 studies the violation of temporal bell inequalities .
finally , we end up with a discussion and a summary of our results .
in this section , we describe a quantum optical ramsey interferometer ( qori ) .
we can summarize the system as follows .
it consists in an atomic interferometer , as depicted in figure 1a .
the interfering ways are realized on the internal states of a rydberg two - level atom , that we denote by @xmath1 . before entering into the interferometer
, the atoms are prepared in the upper state @xmath2 a microwave resonator provides the quantized beam splitter ( bs ) as well as the wwd . following the strategy of @xcite , the phase shifter ( ps ) is provided by an electrostatic external field applied in the region between the cavities .
the differential stark shift between upper and lower levels induces a relative phase @xmath0 that enables an external control of the off - diagonal elements of the density matrix .
subsequently , a classical microwave -@xmath3 pulse provides the beam merger ( bm).the combined interferometer action performs a transformation from the initial state @xmath4 to the final state @xmath5 where the operators @xcite @xmath6 have been introduced , @xmath7 are the pauli spin matrices of the two - level atom , and @xmath8 is the initial state of the wwd , i.e. , the cavity - field inside the resonator . here
@xmath9 and @xmath10are annihilation and creation operators of the cavity field mode , @xmath11 is the vacuum rabi frequency of the atom - field interaction , and @xmath12 the interaction time , i.e. , the time of flight of the atom through the cavity . in the derivation of ( [ final state qd ] ) resonant interaction has been assumed , and cavity damping and spontaneous atomic decay have been neglected .
the final state of the wwd is obtained after tracing out the atomic degree of freedom .
thus @xmath13 where @xmath14 , @xmath15 are the detector s two final states corresponding to each way . the wwd can be regarded as a memory system , with two internal pointer states correlated to the interfering ways , given by @xmath16 in ( [ rhodfinal ] ) .
the quantities @xmath17 are the probabilities for taking each way after the bs .
thus , the asymmetry of the bs is measured by the predictability @xmath18 the final state of the exit atom is measured by means of state selective field ionization techniques in @xmath19 .
quantum optical ramsey fringes are exhibited in the probability of detecting the outgoing atoms in a definite internal state , in case complementarity allows for them .
for instance , @xmath20 the probability of detecting the quanton in the upper ( lower ) state can be calculated to yield @xmath21 where the contrast factor @xmath22 is given by the expression @xmath23 the averages in this equation are taken over the initial field prepared inside the cavity , and can be calculated after an appropriated expansion of the state in the photon number bases .
the stage is now ready for the analysis of complementarity in the interferometer .
complementarity demands that distinguishability of the ways must be followed by lost of coherence in the visibility @xmath24 of the interference pattern in ( [ paa ] ) .
a measure of the distinguishability @xmath25 of the ways in a two - ways interferometer have been given in @xcite , quantifying the maximum potential wwi that can be available to the experimenter .
two sources of wwicontribute to @xmath26 .
one is the a - priori wwi the experimenter has about the ways , stemming from the asymmetric preparation of the beam splitter , and thus , quantified by the predictability @xmath27 in ( [ p ] ) .
the other source of wwi depends on the quantum quality
@xmath28 of the wwd , i.e. , its ability to trace down the ways of the two - level system ( quanton ) via quantum correlations .
recently , a formalism has been developed @xcite which allows one to separate both contributions , even in the case where beam splitter ( bs ) and wwd are provided by the same physical interaction @xcite . assuming pure state preparation in ( [ initial state qd ] ) , we have @xmath29 as can be seen from the above equation , a high - quality @xmath30 wwd implies full distinguishability @xmath31 no matter the value of @xmath32 and viceversa .
in particular , a perfect - quality wwd will prevent any fringes from been displayed at the output port of the interferometer , for any value of the predictability @xmath27 .
in fact , for pure state preparation the following equality holds @xcite @xmath33 the visibility can be rapidly calculated from ( [ paa ] ) to be @xmath34 @xcite , where the contrast factor @xmath22 ranging from 0 to 1 is given in ( [ c ] ) .
fringes are degraded by the potential availability of wwi stemming from the introduction of a quantum wwd into the interferometer .
thus , the observation of an interference pattern in the detection probability of the atom depends on the state preparation of the cavities .
if they are unable to acquire which - way information ( @xmath35 ) , quantum interference is observable due to the indistinguishability of the path leading to the same final state .
on the other hand , if we prepare the cavities to act as which - way detectors , for instance in a fock state ( @xmath30 ) , then no interference effects are observable ( @xmath36 ) , as ensued by complementarity @xcite . as a matter of fact
, @xmath24 has been measured in a recent experiment @xcite for different preparation of the initial cavity field state , showing the transition between both complementary situations .
consider now the initial cavity field prepared in the vacuum state @xmath37 .
the relevant quantities of this section can be rapidly calculated to yield @xmath38 no interference pattern can be measured at the output port of the interferometer , since the cavity - field stores full wwi about the alternatives taken by the atom ( @xmath39 ) .
the upper ( lower ) way is perfectly correlated to the no - photon ( one photon ) state of the cavity field . in order to restore the interference pattern , the wwi must be erased so it is no longer available to the experimenter .
this can be done by launching a second rydberg atom through the cavity , in order to absorb the `` which - way '' photon stored in the cavity field .
an interference pattern can be builded in the correlated detection probability of both quanton and erason atoms @xcite .
we follow a different strategy in order to achieve full quantum erasing which is based in time - reversal invariance . in order to preserve the time symmetry property of closed systems ,
the second microwave pulse providing the beam merger ( bm ) must be quantized too .
we can use a qori scheme described in @xcite .
it consists in two high - q microwave resonator successively transversed by a beam of monoenergetic two - level atoms , at such a low rate that only one atom is present in both cavities at a time ( see figure [ fig2]-a ) .
the symbol a will be used in the sequel to label the quanton .
thus , we denote the upper and lower levels by @xmath40 . before entering the first cavity
the atoms are prepared in the upper level @xmath41we choose now an asymmetrical design for the cavities , as depicted in figure [ fig2](a ) .
it consists of two asymmetrical cavities ( different lengths ) , such that @xmath42 where @xmath43 ( i = 1,2 ) is the interaction time of the atom in the cavity m@xmath44 the cavities are prepared initially to be empty , i.e. , the state vector of the combined cavities - atom system is given by @xmath45 assuming resonant interaction , the probability from the upper - to - lower level transition in each cavity is given by the square of the sine in eq .
( [ 2.4 ] ) . under this preparation ,
the atoms is left with certainty in its lower state after passage of the second cavity . in this energy transferring situation , there are alternative routes along which the final lower level state can be reached .
different routes are realized by changing the photon number in different cavities .
these paths , sketched in figure [ fig2](a ) , will be called the
upper- and lower - paths .
it should be noticed that the conditional dynamics on the alternative path transitions in m@xmath46 is due to the mixing of two different state manifolds of the jaynes - cummings interaction . in the case
the atom enters m@xmath46 in the upper level , it will undergo a @xmath47-pulse leaving a photon in the cavity for the upper path . on the other hand , if the atom enter in the lower level no transitions occur since the atomic - cavity system coincides with the ground state of the interaction hamiltonian .
after atom a has exited the cavity m@xmath46 at time t@xmath48 , the state of the system reads @xmath49 where s@xmath50 is given in ( [ 2.4 ] ) and @xmath51 . after the passage of a single atom
the cavities have been prepared in a non - local field state , i.e. , in a coherent superposition of correlated states of macroscopically separated quantum systems .
it should be noticed that the stated preparation give by eq .
( [ 2.6 ] ) is completely unitary .
after the passage of the pumping atom , the cavity is left , with unit probability , in a macroscopic superposition with an externally controllable non - local coherence . due to the asymmetric design of the resonators , the phase difference @xmath0 of the interfering alternatives
is mapped , in an unitary fashion , into the non - local phase of the superposition state of the two wwds .
note that we have two wwds , m@xmath52 and m@xmath53 , both with @xmath54 .
also , the predictability in m@xmath53 is @xmath55 , so the probability of detecting atom a in the lower level will exhibit a featureless dependence on @xmath56 however , as will be shown in the next section , fringes of unit visibility can be extracted from a suitable measurement of the wwd , provided that full wwi is erased from both cavity fields .
we launch backwards the quanton to erase its own wwi . for instance
, gravity could play the role of an atomic mirror for the quanton @xcite .
alternatively , we can take advantage of the factorization property in ( [ 2.6 ] ) in order to simplify the experimental setup . as a matter of fact , after t@xmath57 a subsequent measure of the quanton 's internal state can be performed without introducing projection noise .
we can use as the erason a replica of the rydberg two - level atom , with upper and lower states that we denote by @xmath58 from a second beam e ( see figure [ fig2]-b ) .
thus , the detection of the system - atom in d@xmath48 can eventually be used to trigger the excitation laser preparing the erason in the lower level @xmath59 the erason is sent through both cavities in reversed order with the same velocity magnitude as that selected for the quanton , and thus , it can be regarded as a velocity reversal version of the quanton . after calculating the coherent transients developed by the erason
, the state of the entangled system is @xmath60\mid 0,0\rangle \mid e\rangle _ { e } + [ c_{1}^{{}}s_{1}^{{}}-s_{1}^{{}}c_{1}^{{}}e^{-2i\phi } ] \mid 1,0\rangle \mid g\rangle _ { e } \}\otimes \mid b\rangle _ { a}\ , \label{2.7}\ ] ] where t@xmath61 is the final time at which the erason leaves the cavity m@xmath50 .
the contributions to the amplitude coefficients in ( [ 2.7 ] ) can be regarded as labels for the four possible paths of the global system . before reaching m@xmath50
, the paths of the returning erason perfectly correlate with those of the quanton .
thus , a relative phase of 2@xmath0 is accumulated between the upper and lower atomic paths . at this stage ,
the role of the erason is to probe the which - way information stored in the cavity m@xmath46 , using time reversal symmetry to transfer it to its own internal state .
however , when the atom crosses cavity m@xmath50 , a rotation of the bloch vector is performed such that each of the two possible outcomes in the subsequent measurement of state of the erason is correlated with both upper and lower paths . due to the asymmetric design of the interferometer , wwi or quantum erasure
can be obtained by the experimenters by means of letting the erason cross one or both cavities , respectively .
after detection of the quanton , the experimenter may decide at will , in a delayed - choice fashion , to display either particle or wavelike aspects of the quantum system , deciding how many cavities the erason will be allowed to fly through . the correlations between the quanton and the final state of the erason
after crossing both cavities are given by the final detection probabilities @xmath62 after many repetitions of the experiment , the detection probabilities ( [ 3.2 ] ) will exhibit fringes and antifringes upon variation of the phase @xmath63 . within repetitions ,
the cavities should be reinitialized to the vacuum state .
it should also be noticed that the erasure could be performed at any time after t@xmath48 within the lifetime of the cavity , and in particular after the atom a has been detected @xcite . as in the usual quantum eraser ,
fringes and antifringes add destructively , and the featureless pattern @xmath64 is recovered when both erason detection alternatives are summed up . on the other hand , in spite the factorization property of the final state of the quanton , we would like to remark that the self - eraser do perform quantum erasing operation .
the non - local phase is constructed on the interfering alternatives of the quanton , i.e. , its wave - like properties , transferred to the cavity fields via entanglement swapping .
thus , the full recovery of the interference fringes in eqs .
( [ 3.2 ] , [ 3.2bis ] ) demands the complete erasure of wwi about the particle - like properties of the quantum stored in the wwds , as imposed by complementarity . as discussed in the previous section
, the system - atom prepares the cavity fields in a non - local state . in order to demonstrate the generation of the macroscopic superposition ,
interference effects sensitive to the non - local coherence must be measured , in order to distinguish this state from an incoherent mixture .
the ramsey scheme converts the non - local phase into a population difference .
thus , according to ( [ 3.2 ] ) , a simple population measurement can be used to probe non - local phase of the cavity fields in ( [ 2.6 ] ) .
observation of fringes in the detection probability of the erason demonstrates the coherent character of the macroscopic superposition ( [ 2.6 ] ) . the visibility of the interference pattern implicit in @xmath65 of eqs .
( [ 3.2 ] ) is computed to be @xmath66 in the case m@xmath50 is prepared to act also as a @xmath47-pulse , @xmath67 and no fringes are obtained as consequence of the single path situation .
perfect visibility fringes are obtained for @xmath68 .
this situation correspond to a symmetrical mixing ( @xmath69 ) of the correlation between eraser and system paths in m@xmath50 . @xmath70 and @xmath71 , given in eqs .
( [ 3.2 ] ) are plotted in figure [ fig3 ] as a function of the induced phase @xmath0 and of the single - cavity transition probability s@xmath72 , showing both visibility limits .
perfect visibility is obtained for atomic velocities matching the condition @xmath73 where l@xmath50 is the length of cavity m@xmath50 .
note that eq .
( [ 3.7 ] ) is compatible with the @xmath47-pulse preparation of m@xmath46 , provided this cavity is built twice as long as m@xmath74 further insight into the system can be gained by means of calling upon time reversal symmetry .
the erason is designed to act as the velocity reversal version of the quanton .
in fact , time reversal invariance is used in m@xmath46 in order to swap the wwi from the cavity field to the internal state of the atom .
however , the passage coursed by the erason can not be regarded as a complete time reversal of the quanton passage : the open phase shifter breaks the symmetry , allowing for different alternatives in the final detection probability .
indeed , the relative phase induced by the ps , adds up to 2@xmath0 after the erason passage , instead of canceling each other . only in the latter case
, the whole interferometer would exhibit complete time reversal invariance and the erason would end up in the initial state of the quanton .
this can be seen from eqs .
( [ 3.2 ] ) where , consistently with time reversal invariance , @xmath75 for @xmath0 @xmath76 .
a signature of quantum coherence can be found in the violation of temporal bell inequalities of the type of @xcite .
we show in this section that the quantum self - eraser violates a temporal bell inequalities for certain ranges of the rabi phase .
according to realistic descriptions , dynamical variables possesses definite values at definite times .
following @xcite , a stochastic process for a two - state system can be defined as the dichotomic random variable @xmath77 assuming the values @xmath78 when the system is on upper ( lower ) state at time @xmath79 .
let us consider now the quantity @xmath80 where @xmath81 are the two - time autocorrelation function @xmath82 and we take @xmath83 as the entering and exit time of the quanton through m@xmath84 ( @xmath85 ) , and @xmath86 as the exit time of the returning erason after crossing m@xmath52 , i.e. , @xmath87 ( see figure [ fig2 ] ) .
any two - state stochastic process must satisfy the inequality @xcite
@xmath88 on the other hand , from the quantum mechanical point of view , the autocorrelation function given by @xmath89 can be calculated for the quantum self - eraser to yield @xmath90 inserting ( [ kaes ] ) into ( [ delta ] ) , we obtain @xmath91 which coincides with the value derived for a single cavity preparation in @xcite . this coincidence can be understood taking into account that for any two times @xmath92 between @xmath93 and @xmath94 we have @xmath95 due to the perfect correlations provided by the @xmath96-pulse in m@xmath97 the quantity ( [ marshall ] ) is plotted in figure [ fig4 ] , showing violation of the inequality ( [ -1delta ] ) for certain ranges of the vacuum rabi phase @xmath98 as can be seen in this plot , there is no violation at @xmath99 ( for which @xmath100 which is precisely the point of full recovery of the interference fringes in the detection probability of the erason .
it is also important to remark that bell experiments involving rydberg atoms are especially of interest , since they can close the communication and the detection loopholes @xcite .
a proposal to measure bell s inequality violation with a rydberg atom sequentially interacting with two classically driven cavities has been given in @xcite .
on the other hand , rydberg atoms are mesoscopic systems , which are of interest to study deviations from macroscopic realism @xcite .
thus , the proposed self - eraser interferometer posses a wide play - ground to test the role of measurement in quantum mechanics at the mesoscopic level .
a scheme for a atomic beam quantum self - eraser has been presented .
in contrast to previous models , the present proposal is based on time reversal invariance . this property demands an experimental setup in which the quanton- ( erason- ) atoms pump ( probe ) the cavities in a symmetrical fashion .
the present proposal uses an asymmetric design of the microwave resonators : the second resonator allowing one to factorize the state of the exit atom from the state of the cavities .
thus , we can send another atom from a reversed beam , to play the role of the self - eraser atom , provided it is send as a velocity reversal of the pump atom . on the other hand ,
the factorizing conditions extend our view about quantum erasing operation .
interference fringes are recovered directly : we do not need to correlate the measurements on the detectors @xmath19 and @xmath101 although in an unconventional fashion @xcite , the erason performs quantum erasing operation .
in fact , due to the experimental configuration , the actual full recovery of the interference pattern demands the quantum erasure of wwi from the wwds , as required by complementarity . at the point of maximum erasure
, we have @xmath102 , allowing for a full recovery of the interference pattern ( @xmath103 ) .
the self - eraser highlights the role of complementarity in the recovery of interference fringes via entanglement swapping . also resulting from the asymmetric design of the resonators , the scheme allows one to prepare the correlated cavity fields in a macroscopic superposition state in an unitary fashion @xcite .
when the quanton leaves the second cavity , the phase information accumulated by the atom in the interference region is kept in the cavity fields as the non - local coherence between the two components of its superposition state .
this coherence is transferred back to the erason as probed in the fringe structure of its final state .
observation of an interference pattern in the final detection probability of the erason can be used for a joint demonstration of quantum erasing operation and the generation of non - local superposition in the correlated field state of the macroscopically separated cavity systems .
violation of temporal bell inequalities provides a criterion for a signature of the establishment of macroscopic quantum coherence . on the other hand ,
the quantum self - eraser is built on entanglement swapping of macroscopic coherence .
thus , it is interesting to analyze how both manifestations of quantum coherence are mutually related .
we have shown that the quantum self - eraser violates temporal bell inequalities . however , this violation is not directly connected to quantum erasing capability .
in fact , the point of maximum erasability is outside of the parameter range leading to violation .
thus , the quantum self - eraser appears as an useful playground to study the relation between non - locality , macroscopic realism and complementarity . j.
was initially supported by cic from universidad michoacana de san nicols de hidalgo , and the promep program of the public education secretary ( sep ) in mxico .
wheeler , in _ mathematical foundations of quantum theory , _ edited by a. marlow ( academic , new york , 1978 ) , p.9 .
e.t . jaynes in _ foundations of radiation theory and quantum electrodynamics , _ edited by a.o .
barut ( plenum , new york , 1980 ) , pp .
m. o. scully and k. druhl , phys . rev .
* a25 , * 2208 ( 1982 ) b.g .
englert , phys .
* 77 , * 2154 ( 1996 ) .
w.k . wooters and w.h .
zurek , phys . rev . *
d19 * , 473 ( 1979 ) , r. glauber , _ new techniques and ideas in quantum measurement theory_. annals of the new york academy of sciences , vol .
480 , p. 336
( 1986 ) and g. jaeger , a. shimony and l. vaidman , physical review a 51 : 54 -67 ( 1995 ) .
j. martnez - linares and j. vargas medina , _ duality in an asymmetric quantum optical ramsey interferometers_. proceedings of the 8th international conference on squeezed states and uncertainty relations , ed
. h. moya - cessa ( rinton press , new jersey , in press ) | a scheme for an atomic beam quantum self - eraser is presented .
the proposal is based on time reversal invariance on a quantum optical ramsey fringes experiment , where a realization of complementarity for atomic coherence can be achieved .
it consists of two high finesse resonators that are pumped and probed by the same atom .
this property relates quantum erasing with time reversal symmetry , allowing for a full quantum erasing of the which - way information stored in the cavity fields .
the outlined scheme also prepares and observes a non - local state in the fields of the resonators : a coherent superposition between correlated states of macroscopically separated quantum systems .
the proposed scheme emphasizes the role of entanglement swapping in delayed - choice experiments .
finally , we show that the quantum self - eraser violates temporal bell inequalities and analyze the relation between this violation and the erasability of which - way information . |
why are the neutrinos much lighter than the charged leptons , but not absolutely massless ? it is universally recognized that this can be simply and elegantly understood in an @xmath1 effective theory .
the most general , gauge - invariant interactions of dimension less than six , which can lead to masses for the known leptons from the vacuum expectation value ( vev ) of a higgs doublet , are : @xmath2 where @xmath3 and @xmath4 are the lepton doublet and singlet fields , and @xmath5 is the higgs doublet .
the dimensionless matrix of yukawa couplings , @xmath6 , has a hierarchy of eigenvalues to describe the masses of the charged leptons .
such a hierarchy could result by promoting the couplings to fields , @xmath7 , which acquire vevs to sequentially , spontaneously break the flavor symmetry @xmath8 .
such a flavor symmetry will also result in a certain structure for the neutrino mass matrix via @xmath9 .
the mass scale @xmath10 is the cutoff of the low - energy effective theory .
the crucial point is that if this cutoff is very large , for example the planck or gauge coupling unification scale , then the neutrino masses are very small .
while the charged lepton masses are linear in the higgs vev , @xmath11 , the neutrino masses are quadratic : @xmath12 the power of this effective field theory approach is that no assumption need be made about the full theory at or above the scale @xmath10 .
the only assumption is that the low - energy theory is the most general allowed by its symmetries .
nevertheless , there is a very simple theory which does lead to the dimension 5 operator of ( [ eq : leff ] ) .
right - handed neutrino fields , @xmath13 , are introduced , with majorana masses @xmath10 and couplings to the lepton doublets : @xmath14 where @xmath10 and @xmath15 are mass matrices . on integrating out the heavy neutrinos
, the well - known see - saw mechanism gives @xmath16 in eq .
( [ eq : leff ] ) by @xmath17 @xcite . in supersymmetric theories ,
there is a very important reason for questioning this simple view of neutrino masses : the low - energy effective theory must contain more fields than the leptons , higgs boson and their superpartners . in particular , there are two higgs doublet superfields , @xmath18 and @xmath19 , and there is another sector of the theory which spontaneously breaks supersymmetry , and triggers electroweak symmetry breaking . at first sight
these additions seem irrelevant to the question of neutrino masses , but closer inspection reveals new opportunities .
an important objection to the minimal supersymmetric standard model is that it is not the most general low - energy effective theory consistent with @xmath20 gauge symmetry .
the gauge symmetry allows a bare mass term for the higgs fields , @xmath21_f$ ] , giving the expectation @xmath22 , which removes the higgs from the low - energy theory .
this is known as the `` @xmath0 problem '' in supersymmetric theories . in this paper
we take supersymmetry to be broken in a hidden sector , at the intermediate scale @xmath23 , via fields @xmath24 : @xmath25 , where @xmath26 is the planck mass .
the supersymmetry breaking is communicated to the standard model by supergravitational interactions , so that the cutoff for the low - energy effective field theory is @xmath26 .
this gives rise to the superpartner masses at the weak scale in the usual way . in this case , the above `` @xmath0 problem '' is easily solved by introducing a further global symmetry , @xmath27 , and dividing the light matter superfields into two types .
there are fields which are chiral with respect to the gauge interactions , such as @xmath3 and @xmath4 , which are guaranteed to be massless until the gauge symmetry is broken , and there are fields which are vector - like , such as @xmath28 , which are kept massless only via @xmath27 .
furthermore , the fields @xmath29 which break the flavor symmetry do not break @xmath27 the vector - like fields acquire mass only from supersymmetry breaking . there is a subset of the @xmath24 fields , which we call @xmath30 , that transform non - trivially under @xmath27 .
this ensures that @xmath0 is of order the supersymmetry breaking scale , as it must be for electroweak symmetry breaking to occur successfully @xcite .
in particular , the operator @xmath31_d \label{eq : mu}\ ] ] leads to @xmath32 , where @xmath33 is the vev of the highest component of @xmath30 . the right - handed neutrino fields , @xmath13 , like @xmath28
, are vector - like with respect to the gauge interactions , hence the crucial question becomes how they transform under @xmath27 .
if they are also vector - like with respect to @xmath27 , they will acquire a large mass , whence the see - saw proceeds via ( [ eq : l ] ) as usual .
alternatively , they may be protected from acquiring a large mass by @xmath27 , in which case they will appear in the low - energy effective theory .
the question of neutrino masses now becomes much richer , since it is necessary to study all possible interactions of @xmath3 and @xmath13 in the low - energy theory .
in particular , mass terms can be induced via interactions with higgs vevs and with @xmath30 vevs by analogy with ( [ eq : mu ] ) , the right - handed neutrinos may acquire mass via the operator @xmath34_d \label{eq : nn}\ ] ] giving the right - handed neutrinos a majorana mass of order the weak scale , @xmath11 .
the see - saw mechanism can still be operative if the yukawa couplings are suppressed .
for example , if one of the @xmath30 fields acquires an @xmath35 component vev at the intermediate scale , this occurs via @xmath36_f .
\label{eq : xlnh}\ ] ] in general , the coefficients of these higher dimensional operators are understood to depend on flavor symmetry breaking , and are functions of @xmath37 .
the see - saw now gives light majorana masses : @xmath38 the usual result !
even simpler possibilities occur : the @xmath27 quantum numbers may prevent a right - handed majorana mass to very high order , so that the dominant mass term is dirac . for example
the operator @xmath39_d \label{eq : xdaggerlnh}\ ] ] dominates either if the operator ( [ eq : xlnh ] ) is forbidden by @xmath27 or if @xmath30 does not have an @xmath35 component vev , and gives light dirac neutrinos of mass @xmath40 . in section two
, we study the general low - energy effective theory for the interactions of @xmath3 and @xmath13 with @xmath18 and @xmath30 , with a view to studying the interesting forms for the neutrino mass matrices .
the above majorana and dirac cases are considered further .
what are the consequences of our proposal that neutrino masses are suppressed relative to charged lepton masses by supersymmetry breaking factors ?
while the most immediate consequence is that it opens up a new class of models for neutrino masses , the most important consequence may be that supersymmetric phenomenology can be drastically altered .
this is largely due to the possibility that the right - handed sneutrino , @xmath41 , can now be at the weak scale and couple via a new @xmath35-type interaction : terms arising from ( [ eq : a ] ) with the light dirac neutrinos whose masses are generated by ( [ eq : xdaggerlnh ] ) , we must require that the @xmath35 component vev of @xmath30 is small , @xmath42 . ]
@xmath36_f \supset v \ ; \tilde{l } \tilde{n } h_u .
\label{eq : a}\ ] ] the structure of this interaction is investigated in section 3 , its consequences for the sneutrino mass spectrum is studied in section 4 , and its consequences for the lightest sneutrino as the cosmological dark matter in section 5 .
we find two interesting cases where this occurs , and each case predicts a characteristic signal in upcoming experiments to directly detect halo dark matter . the @xmath35-term interaction @xmath43 can significantly alter the decay branching ratios for the charged and neutral higgs bosons in supersymmetric theories .
this is studied in section 6 , where we find that , for certain ranges of parameters , the decays @xmath44 and @xmath45 can be the dominant decay modes .
further consequences for collider phenomenology , arising from the @xmath35-term changing the decay chain of @xmath46 , are discussed in section 7 .
finally , in section 8 we study the rare lepton flavor violation implied by our mechanism for neutrino mass generation .
we begin by considering the low - energy effective theory which describes the interactions of the leptons @xmath3 and @xmath13 with the higgs doublet @xmath18 and the fields @xmath30 which spontaneously break both supersymmetry and the global symmetry group @xmath27 .
we impose @xmath47-parity , which changes the sign of the @xmath3 and @xmath13 superfields , as well as the superspace coordinate @xmath48 , but leaves @xmath18 and @xmath30 unchanged . expanding in powers of @xmath49 , @xmath50_f \\ \nonumber & + & { 1 \over m } \left ( \left[{c_{5,1 } } xlnh_u + { c_{5,2 } } ( lh_u)^{2}+{c_{5,3}}n^4 + { c_{5,4 } } ( xn)^2\right]_f + \left[{c_{5,5 } } x^{\dagger } nn\right]_d \right ) \\ \nonumber & + & { 1 \over m^2 } \left ( \left[{c_{6,1 } } x^{\dagger } lnh_u + { c_{6,2 } } x x^\dagger nn \right]_d+\left[{c_{6,3 } } x^{3 } n^{2}\right]_f + \ldots \right ) \label{eqnarray : operators}\end{aligned}\ ] ] here and below , the energy scale @xmath10 is the ultraviolet cutoff of the low - energy theory , such as the planck scale or the gut - scale .
we have included all possible operators even though many may be excluded by the global symmetry @xmath27 , depending on the model .
the bare mass @xmath51_f$ ] is always forbidden by @xmath27 and is not shown .
likewise , the mass terms @xmath52 are forbidden by @xmath47-parity .
the flavor structure is not shown explicitly there are three @xmath3 fields and one or more @xmath13 field , and in general the coefficients @xmath53 are power series in flavor symmetry breaking parameters @xmath54 .
there are other dimension six operators , such as @xmath55_d$ ] , but these will not affect the structure of the model nor the phenomenology , so we will not discuss them .
it is possible that lepton - number violating dimension seven operators have important consequences , and we will discuss some of the consequences within the context of dark matter . before we consider particular symmetries ,
let us explore which combinations of operators are phenomenologically interesting .
in later sections , we will exhibit particular symmetries which realize these scenarios . since @xmath13 must appear in combination with some of the @xmath56 superfields , @xmath57 .
if we allow @xmath58 , when the @xmath56 get @xmath33 component vevs , this term will generate an intermediate scale mass for the @xmath13 scalar , @xmath59 , and this case is thus of less phenomenological interest .
furthermore , precisely the same term must be omitted in the mssm ( @xmath60 ) , and its omission here seems very natural . for these reasons we will take @xmath61.-component vev of @xmath30 is zero , one might also think that @xmath62 must be zero for another reason .
if one allowed such a large supersymmetry breaking mass for @xmath13 , such that the fermion was present in the weak - scale theory , but the scalar was not , one might worry that loop effects would destabilize the @xmath63 hierarchy .
however , all dangerous diagrams that appear are suppressed by small yukawa couplings and are harmless . ]
the structure of the theory can vary , depending on a few elements .
in particular , if @xmath30 develops an @xmath35 component vev , the size of the yukawa couplings will be different .
there are three different natural values for the @xmath35 component : @xmath10 , @xmath64 and zero .
if it is @xmath10 , then the yukawa couplings from the @xmath65 term in ( 10 ) are order one , which is phenomenologically unacceptable .
we consider the other two alternatives in greater detail
. we will begin by considering situations with only one generation . if the @xmath35 component of @xmath30 is zero , but @xmath30 does gain an @xmath33 component , we generate yukawas @xmath66_d = \frac{f}{m}\left [ lh_un\right]_f .
\label{eq : yukawas1}\ ] ] if we assume that @xmath67 sets @xmath68 in ( 10 ) , so that there is no majorana mass for the right - handed neutrino , then we have generated a yukawa of the order @xmath69 .
when the higgs field takes on a vev , we then have a mass for the neutrino @xmath70 .
this is astonishing , because we now have a naturally light _
neutrino , with a mass of the correct size to explain the observed phenomena associated with neutrino mass could easily be @xmath71 , or some other scale , in which case the yukawas are larger . ] .
if this is correct , then experiments studying neutrino mass have _ already _ begun to probe the structure of supersymmetry breaking ! a small yukawa coupling can , in principle , at least , be understood with the froggatt - nielsen mechanism @xcite , through the ratio @xmath63 of the vev of a supersymmetry conserving spurion over some higher mass scale .
such a vev could arise from dimensional transmutation , or , as recently discussed by @xcite , radiative symmetry breaking , which was employed by @xcite to achieve such light sterile states .
the possibility of using susy breaking operators to generate light sterile states was initially explored by @xcite . however , in contrast to these theories , the present class of models features a natural mechanism for generating weak scale @xmath35 terms through the operator is generated in the global susy limit , supergravity effects will generate a small shift in the @xmath35 component , giving @xmath72 @xcite .
thus , the large @xmath35 terms and the small dirac neutrino masses can be generated from the operator of ( [ eq : aterms1 ] ) alone @xcite . ]
@xmath73 operators like ( [ eq : yukawas1 ] ) and ( [ eq : aterms1 ] ) could be selected by a symmetry @xmath74 , with fields @xmath30 and @xmath75 , with charges @xmath76 , and @xmath77 , respectively .
@xmath4 , @xmath3 and @xmath13 have charges @xmath78 , @xmath79 and @xmath80 , respectively . with these charges ,
the only operators of dimension six or less allowed are @xmath81 the superfields @xmath30 and @xmath75 could acquire @xmath33 component vevs , but no @xmath35 component vevs if embedded in an oraifeartaigh model . given a superpotential @xmath82 the minimum of the scalar potential will occur with @xmath83 . here
@xmath84 and @xmath85 have charges @xmath86 and @xmath87 , respectively .
two linear combinations of @xmath88 , @xmath89 and @xmath90 will have positive masses at tree level , while the third independent combination will get its mass in the one - loop effective potential , stabilizing @xmath91 .
note that the presence of the superpotential term @xmath92_f$ ] generates a contribution to the yukawas of the same order of magnitude as that from @xmath93_d$ ] .
note also that this breaks @xmath74 , but preserves @xmath94 , which is the ordinary lepton number symmetry .
we will refer to this scenario , in which right - handed neutrinos couple with suppressed yukawas , but have no majorana masses , as `` dirac masses from supersymmetry breaking , '' or `` sdirac '' for short .
an alternative to generating light dirac neutrino masses from supersymmetry breaking is to instead generate light majorana masses .
we begin by considering the operators @xmath95 such terms could be justified by an r symmetry , where n has r charge @xmath96 , @xmath3 and @xmath5 both have r charge @xmath97 , and @xmath30 has r charge @xmath98 . if @xmath30 takes on an @xmath35 component vev @xmath99 , as well as an @xmath33 component vev , @xmath100 and @xmath33 components of the same order of magnitude under the dynamical assumption that a constant appearing in the kahler potential is negative @xcite .
] , then the second term in ( [ eq : smajops ] ) generates a weak scale @xmath35 term , but now generates a yukawa coupling to the higgs as well roughly of the size @xmath101 . if this were the end of the story , then we would have a ( dirac ) neutrino mass @xmath102 .
however , the first term of ( [ eq : smajops ] ) will now generate a majorana mass for @xmath13 of the order @xmath103 , yielding a lr neutrino mass matrix @xmath104 after integrating out the heavy @xmath13 fermion , we are left with a majorana mass for the neutrino with a size @xmath105 , again reproducing the see - saw result .
could have been generated with nonzero @xmath106 as well .
however , the scalars would then have a supersymmetry breaking mass squared @xmath107 .
thus , for the same reasons we took @xmath61 , we do not consider the case where @xmath67 allows nonzero @xmath106 . ]
we will refer to this scenario , in which the right - handed neutrinos have yukawas @xmath108 and weak scale majorana masses , as `` majorana mass from supersymmetry breaking '' , or smajorana for short .
in section [ sec : bigsec ] , we concerned ourselves simply with the origin of neutrino mass itself , but did not address the additional question of what determines the structures of these masses when we include additional generations . as was discussed in the introduction
, we are considering a scenario in which the global symmetry of the theory is @xmath109 .
@xmath27 must include some symmetry to keep the higgs doublets and right - handed neutrinos light , and @xmath8 may include symmetries such as @xmath110 which relate the different generations .
the key feature of our model is that the supersymmetry breaking spurions also contain charges under @xmath27 .
when these spurions acquire @xmath33 , and possibly @xmath35 component vevs , they break @xmath27 .
of course , they need not be charged merely under @xmath27 , but potentially under some larger group @xmath5 , where @xmath111 . in the most minimal framework , @xmath112 and would contain only those symmetries which are necessary to suppress the @xmath0 term and the right - handed neutrino masses , for instance @xmath113 in section [ sec : sdirac ] , or the @xmath47 symmetry in section [ sec : smajorana ] . with such an assumption , the textures of the neutrino mass matrices and the @xmath35 terms would be determined by supersymmetry preserving elements of the theory .
for instance , in the sdirac scenario , the couplings would be given by @xmath114 where @xmath115 and @xmath116 are supersymmetry _ preserving _ but flavor breaking spurions .
as such , an explanation of structure of the yukawas and @xmath35 terms is beyond the scope of our scenarios .
however , because of this , in the presence of a flavor symmetry , we are able to easily relate the structure of @xmath117 and @xmath118 . however , in the smajorana case , the couplings will be given by @xmath119 here , the yukawas are precisely the same as the @xmath35 terms , but due to the potential mixing of the @xmath13 s , it is impossible to necessarily relate the @xmath35 term matrix to the neutrino mass matrix obtained by the see - saw mechanism .
we consider this to be the minimal scenario , in which @xmath5 is as small as possible , so to speak .
however , @xmath5 can easily be much larger .
indeed , people have investigated the possible effects of baryon and lepton number violation operators from supersymmetry breaking @xcite . a priori
, there is no reason why we should reject the possibility that @xmath120 . if that were the case , we would write the sdirac couplings as @xmath121 now the spurions @xmath30 and @xmath75 carry generation indices themselves !
we assume that charged fermion yukawas are generated in a supersymmetry preserving sector of the theory .
since the flavor structure of @xmath30 and @xmath75 is entirely determined in the supersymmetry breaking sector , they need not be aligned with those of the supersymmetry preserving yukawas .
consequently , a large mixing between @xmath122 and @xmath123 is natural . of course
, the small mixing between @xmath124 and this heavy state ( @xmath125 as required by chooz @xcite ) , must be viewed as somewhat of an accident , but not necessarily a fine - tuning .
this is similar to the anarchy proposal of @xcite , except that here we need not relate the yukawas of @xmath126 and @xmath127 , and a hierarchy of eigenvalues could still possibly occur in @xmath30 .
consequently , even for sdirac neutrinos , anarchic aspects of the theory are reasonable .
whether this is compatible with supersymmetric flavor changing constraints is an important question .
we will address this further in section [ sec : lfv ] .
it is important to note that we do not need three right - handed neutrinos for the cases of sections [ sec : sdirac ] and [ sec : smajorana ] .
the presence of just two @xmath13 states is enough to generate either two massive dirac or two massive majorana neutrinos .
the remaining neutrino is simply a massless weyl neutrino . in a certain sense
, this is more minimal than with three @xmath13 s , but the phenomenology is largely unchanged .
one limit of this could be that there is , in fact , only one @xmath13 .
here , one might generate majorana masses for the neutrinos through an ordinary @xmath128 see - saw , while a sdirac or ssterile @xmath13 would contribute a fourth mass eigenstate , resulting in four majorana neutrinos . given appropriate @xmath8 charges , other
, more exotic possibilities may exist , such as combinations of sdirac and smajorana neutrinos .
in the mssm , the sneutrino and charged slepton masses are intimately related : @xmath129 where @xmath130 is the soft scalar mass for the left - handed sleptons . for @xmath131 , @xmath132 and the @xmath133-term splitting pushes the sneutrino mass down and the charged slepton mass up .
the present experimental bound @xmath134gev still allows for light sneutrinos due to this splitting .
however , if in the future it becomes established that @xmath135 is very large , much of the phenomenology associated with light sneutrinos will be ruled out in the mssm . in our model , with light right - handed sneutrinos , the story is quite different , both because the @xmath35 terms provide an additional source of splitting between the sneutrino and charged slepton masses , and because the right - handed sneutrino mass is not linked to slepton masses by gauge invariance .
thus , even if @xmath130 is quite large it is still possible to have significant change in phenomenology that would otherwise be absent . to better understand the spectrum
, we consider a single sneutrino generation with mass - squared matrix @xmath136 given that @xmath130 , @xmath137 and @xmath35 are independent parameters , this matrix can have very different eigenvalues .
we plot the mass spectra for various choices of @xmath130 and @xmath137 as a function of @xmath35 in figure [ fig : spectra ] .
an independent lower bound on the sneutrino mass in the mssm , @xmath138 gev , comes from the measurement of the invisible with of the @xmath24 , and is also altered by the addition of right - handed sneutrinos .
the lightest sneutrino in our model is a superposition of left and right - handed states : @xmath139 if this state is light enough to be produced in @xmath24 decays , its contribution to the @xmath24 width is given by @xmath140 where @xmath141 is the @xmath24 width to ordinary neutrinos .
if we take the current lep limit of @xmath142 @xcite , the @xmath143 factor allows us to evade the bounds regardless of mass provided @xmath144 , which is a very mild constraint .
although @xmath130 , @xmath137 and @xmath35 are independent parameters , we can gain some intuition for their sizes from their renormalization group running .
in fact , it is somewhat natural to have @xmath145 in the low - energy theory .
the running of @xmath146 and @xmath147 is governed by @xmath148 @xmath149 where @xmath150 . since @xmath59 is a standard model singlet ,
there are no gaugino loops to drive its mass upward as we run the energy scale down from @xmath26 to @xmath151 .
likewise , there are new , sizeable loop diagrams arising from the @xmath35 terms ( figure [ fig : aloop ] ) , which push the soft masses down .
however , two states ( @xmath152 and @xmath153 ) can propagate in the loop contributing to @xmath137 , while only one ( @xmath154 ) can propagate in the loop contributing to @xmath130 , pushing @xmath155 down faster than @xmath156 .
in summary , the mass matrix can have two _ very _ different mass eigenstates , and there can be particles that couple very much like @xmath153 , but with suppressed couplings and masses unrelated to our expectations from the mssm . the lightest is likely to be dominantly @xmath59 , such that its mass is not restricted by @xmath24 decay data .
one of the appealing features of r - parity conserving supersymmetric theories with gravity - mediated supersymmetry breaking is that the lightest superpartner ( lsp ) is a good candidate for dark matter .
searches for superheavy hydrogen have ruled out a charged lsp , leaving the neutralino and the sneutrino as candidates for dark matter .
a number of direct searches for dark matter have been carried out @xcite which have essentially excluded sneutrino dark matter in the mssm unless @xmath157 gev . however , as we have already discussed , measurements of the invisible width of the @xmath24 exclude such a light sneutrino . within our framework
, the @xmath24 width provides only a mild constraint , and we are free to explore the possibility of a light sneutrino dark matter candidate . a second , equally important
point is that the @xmath158 suppression of the light sneutrino coupling to the @xmath24 boson greatly reduces the sneutrino - nucleon cross section , making it possible even for a heavier @xmath46 to evade direct detection .
finally , if we include lepton number violation , the lightest sneutrino can not scatter elastically via @xmath24 exchange @xcite , further diminishing the limits from direct searches .
sneutrino dark matter requires the presence of the @xmath35-term of ( [ eq : a ] ) , and hence is linked to the other phenomenology of the interaction
. we can determine the relic density of light sneutrinos ( @xmath159 gev ) through standard methods @xcite .
the dominant annihilation process is through @xmath160-channel neutralino exchange .
if the neutral wino exchange dominates , we find @xmath161 where @xmath162 is the normalized hubble parameter .
we find it highly significant that for wino masses of roughly @xmath163 and angles roughly a half of the limit from the invisible @xmath24 width , we have a cosmologically interesting amount of sneutrino dark matter .
we show in figure [ fig : relicdensity]a the relic density of sneutrino dark matter considering all annihilation processes .
if this scenario is correct , and @xmath164 is the dark matter , then the mixing angle must be near the limit from the invisible @xmath24 width measurements , making a future detection possible .
in particular , such a sneutrino would almost certainly be seen in the upcoming cresst experiment @xcite .
relic sneutrinos captured by the sun will annihilate into neutrinos that can induce upward - going - muon events on earth @xcite .
the flux of these muons is constrained to be less than @xmath165 @xmath166 s@xmath167 @xcite . for a 10 gev sneutrino with @xmath168
, we calculate a flux , using the formulae of @xcite , that is roughly three times this , assuming that all neutrinos produced are muon - type when they reach the earth , the rate for sneutrino capture by the sun is dominated by higgs exchange and the flux can be much larger . here
we assume @xmath169 gev , so that the capture rate is dominated by @xmath24 exchange . ] ( we find a much smaller flux due to capture by the earth itself ) . the actual muon flux could be suppressed depending on the flavor of the lsp sneutrino and on neutrino oscillation parameters
; for instance , for an electron - type sneutrino and the small angle msw solution to the solar neutrino problem , the flux would be smaller by roughly a factor of a thousand . for heavier sneutrinos ,
the strongest current direct detection limit comes from cdms @xcite , which , under the assumption of @xmath170 scaling , constrains the nucleon - relic cross section to be less than @xmath171 @xmath172 for relic masses of @xmath173 gev ) .
the cross section for ordinary sneutrino - nucleus scattering is @xmath174 where @xmath0 is the sneutrino - nucleus reduced mass . in our framework this cross section comes with an additional @xmath175 suppression , implying that for @xmath176 much larger than the nucleon mass @xmath177 , the cdms constraint can be evaded by requiring @xmath178 taking @xmath179 and @xmath180 for ge@xmath181 gives @xmath182 .
the dama collaboration @xcite has reported a positive signal corresponding to a relic - nucleon cross section of roughly @xmath183 @xmath172 and a relic mass @xmath184 gev in our framework this range in cross section corresponds approximately to @xmath185 . in figure
[ fig : relicdensity2 ] , we plot contours of @xmath186 for @xmath187 gev and a bino mass of 200 gev . for this choice of parameters , and for large enough @xmath35 ,
the dominant annihilation processes in the early universe are s - channel higgs exchange into @xmath188 and @xmath24 pairs , which have cross sections proportional to @xmath189 rather than @xmath175 .
these are also the dominant annihilation processes for sneutrinos trapped in the sun .
this is important because the alternative process is annihilation directly into neutrinos via t - channel neutralino exchange , which would likely lead to a much larger signal at indirect detection experiments .
assuming that @xmath190 of all neutrinos produced in the sun are muon - type upon reaching the earth , we find that indirect detection constrains @xmath191 for the parameters we have chosen , comparable to the cdms constraint .
the interesting relic abundances indicated in figure [ fig : relicdensity2 ] lead us to conclude that the prospects for sneutrino dark matter with @xmath192 gev are quite interesting in our model .
as previously explored @xcite , the presence of lepton number violation changes the limits from direct searches for dark matter significantly . in our model
, lepton number violation can reside in the @xmath193 term in the lagrangian .
such a term could easily arise from dimension seven operators in the lagrangian , such as @xmath194 the presence of this lepton number violation splits the cp - even and odd states @xmath195 and @xmath196 .
however , the coupling to the @xmath24 is off diagonal , i.e. , @xmath197 .
consequently , for large enough @xmath198 , the lsp sneutrino can not scatter off nuclei via @xmath24-exchange , eliminating constraints arising from cdms , dama , and the heidleberg - moscow ge experiment @xcite .
more precisely , the scattering is kinematically forbidden if @xmath199 , where @xmath200 is the mass of the target nucleus , and @xmath201 for virialized halo particles on average .
for example , taking @xmath202 gev and a ge target , we require @xmath203 . since @xmath204 , this corresponds to @xmath205 , which is of the order of what we expect from ( [ eq : splitop ] ) may be required to prevent indirect detection due to sneutrino capture and annihilation in the sun .
here we simply assume that @xmath206 is large enough to evade indirect detection as well . ] .
the effects of lepton number violation in dark matter have been previously explored @xcite . however , in the model previously proposed , there were no singlet sneutrinos , so the mass splitting @xmath206 was required to be adequately large so as to suppress coannihilation between @xmath207 and @xmath196 via @xmath88-channel @xmath24 exchange . in our model , this process is further suppressed by @xmath208 in the cross section , so that even with small mass splittings from dimension seven or higher operators , we can still generate a cosmologically interesting abundance .
unlike @xcite , we now have the @xmath209 coupling , which yields an extra contribution to the scattering of the lightest sneutrino off of nuclei via higgs exchange .
the coupling of the higgs to nucleons is larger than just that from scattering off of valence quarks @xcite , but it is still quite small . using the numerical value for the higgs - nucleon coupling from @xcite , we find that the sneutrino - nucleon cross section obtained from higgs exchange alone is @xmath210 for broad ranges of parameters this cross section falls well below the current direct detection limits .
for instance , it is quite reasonable to consider @xmath211 even for values of @xmath212 larger than those that allow one to evade cdms in the lepton - number conserving case .
future experiments @xcite should be able to probe an additional three orders of magnitude down from the present constraint , giving a significant probe of sneutrino dark matter over a considerable range of parameters , for both the lepton - number conserving and lepton - number violating cases . since the dominant annihilation process in the early universe is @xmath88-wave , there is little dependence of the relic abundance on the sneutrino mass given that the sneutrino is relatively light ( @xmath213 gev ) .
for these relatively light sneutrinos , figure [ fig : relicdensity]a is still qualitatively applicable . for
larger sneutrino masses , @xmath24 and higgs pole effects or production of @xmath214 and @xmath24 pairs can be relevant .
the sneutrino relic density is shown in figure [ fig : relicdensity]b for @xmath187 gev and @xmath215 20 gev .
the lepton number violating mass @xmath216 can induce radiative corrections to the neutrino mass through neutralino loops . if the splitting @xmath206 is too large , the possibility exists of generating neutrino masses radiatively which are large enough to affect the overall analysis .
such a possibility will be explored elsewhere @xcite . for our purposes here , we limit ourselves to the case where the mixing between @xmath59 and @xmath217 is small enough to suppress this radiatively generated mass ( which is why a relatively small value for @xmath35 is taken in figure [ fig : relicdensity]b ) .
we conclude that the possibility of evading direct detection through lepton number violation leads to another interesting version of sneutrino dark matter in our framework
. moreover , the elastic scattering of sneutrinos from nuclei via higgs exchange is just below the current limits , and potentially detectable at upcoming dark matter searches .
the unsuppressed @xmath218 coupling in our scenario can lead to interesting collider phenomena . if kinematically allowed , @xmath219 is typically the dominant decay mode for the light higgs .
there is a similar situation in the the mssm @xcite : provided the sneutrinos are sufficiently light and that @xmath220 is not too close to 1 , the @xmath221 coupling proportional to @xmath222 leads to a partial width into sneutrinos that is larger than that into @xmath223 by two orders of magnitude . assuming that the sneutrinos decay invisibly into @xmath224 ( or that the sneutrino itself is the lsp ) , a light higgs that decays dominantly into sneutrinos would be very difficult to discover at lhc , leaving the nlc the opportunity for discovery through the process @xmath225 . in the mssm , the @xmath24 width measurement ,
the theoretical bound @xmath226 gev , and the relation between the @xmath46 and @xmath227 masses constrain the region of parameter space in which the light higgs can decay into sneutrinos .
for example , if in the future it becomes established that @xmath228 gev , the decay @xmath229 will be ruled out in the mssm . in the present scenario ,
however , the sneutrino mass spectrum is expected to be quite different from that in the mssm , as discussed in section [ sec : mm ] .
even if @xmath146 ( and therefore @xmath230 ) is quite large , it is still possible for the light higgs decay into sneutrinos to be kinematically allowed . to explore this possibility quantitatively
, we consider a single generation of sneutrinos with the mass matrix of equation ( [ eq : massmatrix ] ) , whose four free parameters are @xmath146 , @xmath147 , @xmath231 , and @xmath35 . for simplicity
we consider the case in which the splitting between @xmath232 and @xmath233 is generated by rg running from the gut scale to the weak scale , and adopt @xmath234 , with @xmath235 , the universal gaugino mass , set to 100 gev .
the region of @xmath236 parameter space in which the @xmath24 width constraint is met and @xmath237 holds are displayed in figure [ fig : hdecay ] for @xmath238 gev and @xmath239 ( the plot is very similar for high @xmath231 ) .
as expected , there is a band in parameter space that yields invisible higgs decays : for the region shown , the window for @xmath35 is roughly 10 gev at fixed @xmath146 .
the band persists for large @xmath146 , with the window for @xmath35 scaling as @xmath240 .
the @xmath218 coupling can also alter the details of charged higgs decays .
if @xmath241 , one can look for the charged higgs through the process @xmath242 , with one or both of the top quarks decaying into @xmath243 ( @xmath244 ) .
the standard analysis exploits the fact that the charged higgs is coupled most strongly to @xmath245 ( in contrast to the universally coupled @xmath214 ) , and so if produced should lead to a surplus of @xmath246 s .
this analysis has been applied at the tevatron to establish lower bounds on the charged higgs mass for @xmath247 and @xmath248 @xcite .
the region of intermediate @xmath231 will be only partially accessible to run ii of the tevatron , but should be covered entirely at the lhc @xcite .
similarly , the higgs search at lhc for high @xmath231 employs the decay of heavy neutral higgs states @xmath249 , which can be suppressed due to the decay modes into sneutrinos .
it has already been pointed out that if the charged higgs decays into susy particles , the analysis changes drastically @xcite .
if the decay into a charged slepton and a sneutrino is kinematically allowed , the mssm lagrangian term @xmath250 ensures that @xmath251 dominates over the yukawa - coupling - induced decay into @xmath252 for small @xmath231 . even for large @xmath231 , it is still possible for the decay into @xmath253 to dominate due to the coupling @xmath254 ( of course , @xmath0 and @xmath255 must not take on values that push the lightest charged slepton mass below the experimental bound ) . in this case , the excess @xmath246 s produced will have lower energy than when produced directly via @xmath256 .
an unsuppressed @xmath218 coupling introduces the added possibility of @xmath257 decays .
if kinematically allowed , this is another process that can dominate over the direct decay into fermions for small @xmath231 . once again
, kinematical considerations for this decay are modified from the mssm decays both because of the additional mass splitting between the sneutrinos and charged sleptons , and because the @xmath24-width constraint does not apply to a sneutrino mass eigenstate that is chiefly right - handed .
thus it is conceivable that @xmath257 is the only susy decay mode allowed .
another important difference is that if one supposes that the flavor structure of the @xmath35 coupling is similar to that of the neutrino masses , then one expects the charged higgs to decay into @xmath258 and @xmath259 with similar probabilities , leading to an excess of both @xmath0 s and @xmath246 s over @xmath127 s .
is kinematically allowed for all flavors . ] even if @xmath260 , it is still possible for @xmath261 to be a dominant decay , for small to intermediate @xmath231 .
the width is proportional to @xmath262 , to be compared with @xmath263 $ ] for @xmath264 . to obtain a competitive rate requires @xmath265 to be sizable , which is most easily achieved kinematically when @xmath266 itself is large . in this regime , decays into other susy particles are also likely to be important .
the unsuppressed @xmath35 terms can have other interesting consequences for collider phenomenology , both due to their effect on the particle spectrum , and because of the trilinear scalar vertex itself . here
we briefly consider a few examples , first for a visibly decaying @xmath46 , and second for an invisible @xmath46 .
we have already seen that it is natural in our scenario to have large mass splittings among the various sneutrino states , so it is easily conceivable that there will be sneutrinos in both categories . as in the standard framework of the mssm
, sneutrino decays into @xmath267 and @xmath268 , if kinematically allowed , lead to final states with , e.g. , @xmath269 , @xmath270 , or @xmath271 . for example , possible decay chains include might play in these decay chains .
for instance , @xmath272 might decay invisibly into @xmath273 , as discussed below . ]
@xmath274 and @xmath275 , followed by @xmath276 , @xmath277 , and @xmath278 .
a possible signal for sneutrino pair production at the nlc is thus @xmath279 .
sneutrino pair production should be distinguishable from neutralino pair production due to the different angular distributions and the different @xmath280 dependences at threshold .
as far as this signal is concerned , the distinctive feature of our model is the admixture of the gauge singlet @xmath41 in the sneutrino mass eigenstate . decomposing the mass eigenstate as @xmath281 ,
the sneutrino pair production rate will be suppressed by a factor @xmath175 relative to its mssm value . by performing a scan in energy
one would be able to see the @xmath279 signal turn on for an isolated sneutrino mass eigenstate .
then , knowing the masses and mixings of the charginos and neutralinos , one could infer from the measured rate the value of @xmath212 , and demonstrate that the sneutrino produced is only partly left handed .
if a heavier @xmath282 state is sufficiently split from a lighter @xmath283 , the unsuppressed @xmath35 term induces the decay @xmath284 , providing an interesting new way to produce higgs particles .
one might wonder how efficient this method of producing higgs would be at the lhc , through cascade decays such as @xmath285 , @xmath286 , @xmath284 .
the production cross section of squarks and gluinos at the lhc depends sensitively on their masses .
taking @xmath287 gev , the cross section is @xmath288 nb at @xmath289 tev @xcite , roughly fifty times larger than the cross section for gluon fusion higgs production at that energy for @xmath290 gev in the decoupling limit @xcite , the regime we consider here .
if @xmath291 gev , the cross sections are comparable . assuming that @xmath35 is sizable and that @xmath292 , the question of whether or not there is an appreciable branching fraction for the cascades to produce @xmath293 depends largely on @xmath294 . for simplicity consider the case of a bino - like @xmath295 and wino - like @xmath272 and @xmath296 , and take @xmath282 and @xmath283 to be essentially left- and right - handed , respectively . if @xmath297 , then @xmath282 will never be produced in the cascades , because the colored particles decay via @xmath298 .
if @xmath299 , then the branching fraction for producing @xmath293 is the product of three probabilities : first , the probability of the gluino / squark decaying into a neutralino or chargino heavier than @xmath282 ; second , the probability of that gaugino decaying into @xmath300 rather than into a lighter gaugino or @xmath301 ; and third , the probability that @xmath282 decays into @xmath293 rather than @xmath302 . ) and br(@xmath303 ) are suppressed if @xmath283 is essentially right - handed . ]
none of these probabilities is likely to be smaller than @xmath304few if @xmath299 , so in this case the rate for higgs production from @xmath282 decay at the lhc could easily be comparable to or even much larger than that due to @xmath305 . moreover , the cascade products ( additional jets , and an energetic lepton from @xmath306 decay , for instance ) , allow for detection via th @xmath307 mode , so that the signal is further enhanced relative to @xmath308 by a factor of a thousand .
for example , suppose that @xmath309 gev .
is forced to be much heavier , @xmath310 gev , in order for @xmath311 to be satisfied . in this case
the squark and gluino production cross section is comparable , at best , to the @xmath312 cross section . ]
then the first probability is @xmath313 , because @xmath314 couples to the bino , while @xmath315 prefers winos .
the second probability is also @xmath313 , because @xmath316 and @xmath296 are equally likely to produce @xmath317 and @xmath282 and do not couple to @xmath295 in the limit that it is pure bino .
the third probability is determined by @xmath318 , and can be as large as @xmath319 .
so in this case , the branching fraction for producing @xmath293 could be @xmath320 , leading to higgs production rate of about ten times larger than that from gluon fusion ( not just five times larger , since each strong production event gives two sparticles that can potentially produce a higgs ) .
in fact , in this case the rate of higgs _ pair _ production through @xmath282 decay is as large as the rate for @xmath312 .
these pair production events would lead to striking final states @xmath321 , with the invariant masses of both @xmath322 pairs equal to @xmath323 .
note that even if @xmath324 gev , the rate for cascade higgs production is only down from the @xmath312 rate by a factor of @xmath325 , and would still likely allow for discovery because the signal is not suppressed by the @xmath326 branching ratio .
even more striking is the possibility that @xmath282 and @xmath283 are the two lightest supersymmetric particles , with @xmath327 . in this case
, every squark and gluino produced yields a higgs particle in its cascade .
so for @xmath287 gev , gluon fusion would account for only one in every @xmath328 100 higgs particles produced at the lhc .
production of @xmath282 s and their subsequent decay could also be an interesting source of higgs particles at the nlc .
the rate of higgs production through @xmath329 is typically at least an order of magnitude lower than that due to @xmath330 and @xmath331 fusion for @xmath332 gev @xcite .
however , @xmath333 could lead to sizable higgs _ pair _ production .
for example , for @xmath332 gev and @xmath334 gev , @xmath335 compared to a cross section of @xmath336 fb for the double higgs - strahlung process @xmath337 in the decoupling regime , for @xmath338 gev @xcite . especially if @xmath282 is lighter than all gauginos except for a bino - like state , it is easy to choose @xmath146 , @xmath147 , and @xmath35 so that @xmath339 is not more than an order of magnitude suppression ( there is even the possibility , as mentioned above , that that @xmath340 )
. thus a possible signature of our scenario is an excess of events with missing energy plus two @xmath223 pairs whose invariant masses equal the higgs mass , beyond the number expected from double higgs - strahlung followed by @xmath341 .
the motivation for considering this case in our scenario is that the @xmath35 terms suppress the masses of the lighter sneutrinos , making it more likely than in standard schemes that some @xmath46 s can either only decay invisibly , or not decay at all .
let us assume that @xmath295 and @xmath46 are the two lightest supersymmetric particles .
one immediate question is whether the clean trilepton signal from @xmath342 production at hadron machines remains , since @xmath272 has access to the invisible 2 body decay mode @xmath343 .
however , provided @xmath344 , @xmath272 can also decay through the visible two body mode @xmath345 . if the branching ratio for this decay is not too small , the trilepton signal survives , because @xmath296 decays into @xmath346 and possibly @xmath347 , if the later is kinematically accessible .
especially interesting is the particular case where only the lightest @xmath46 mass eigenstate is lighter than @xmath296 , and @xmath348 .
the flavor of the lepton produced in @xmath349 is correlated with the flavor of the light @xmath46 .
moreover , it is reasonable to expect the lightest sneutrino to be coupled to the largest @xmath35 term , and so , assuming that the flavor structure for the @xmath35 terms resembles that of the neutrino masses ( the connection between the two is most immediate in the sdirac case ) , the lightest @xmath46 is likely to be mixture of @xmath350 and @xmath351 . in this case , @xmath313 of the leptons produced in the @xmath352 decays are @xmath0 s , while very few @xmath127 s are produced , leading to roughly 7 @xmath0 s for every 4 @xmath127 s in the trilepton signal , assuming that the @xmath272 decays produce equal numbers of each lepton flavor
. an invisible @xmath283 state can be produced along with a heavier , visibly decaying @xmath282 at @xmath353 colliders through @xmath88-channel @xmath24 exchange ( and @xmath160-channel chargino exchange for @xmath354 ) , provided that @xmath355 is not too small .
the decays of the heavier @xmath46 would lead to the signal @xmath356 at the nlc .
if the masses of heavier , visibly decaying sneutrinos have already been established through pair production , it should be possible to measure the mass of a light , invisible @xmath46 using the energy spectrum endpoints for the leptons produced in this process . as mentioned above , if @xmath357 , then the only two - body decay for @xmath272 is into @xmath273 , so that both @xmath46 and @xmath272 decay invisibly .
the process @xmath358 has been shown to be a feasible means for detecting the presence of these extra carriers of @xmath359 at the nlc @xcite .
finally , suppose that @xmath46 and @xmath317 , rather than @xmath295 and @xmath46 , are the lightest supersymmetric particles .
thus , @xmath360 , and @xmath361 decays visibly into @xmath362 .
meanwhile , the nlsp @xmath317 has only three - body decays , into @xmath363 and @xmath364 . in this case , a signal for slepton pair production is @xmath365 , a characteristic signature for chargino pair production ( although the two cases are distinguishable by their angular distrubutions , for instance ) @xcite .
in our framework , lepton flavor violating contributions to @xmath366 can arise at tree level due to the same spurion(s ) responsible for the dirac neutrino masses and @xmath35 terms .
this possibility exists in each of the three scenarios discussed in section [ sec : bigsec ] , but the connection between the flavor structure of @xmath366 and that of the neutrino masses is most direct for the sdirac case , which we therefore consider here for simplicity .
suppose that @xmath30 , a chiral superfield with @xmath367 , has the appropriate flavor structure to induce dirac neutrino masses via @xmath368 then one can also write down the lagrangian term @xmath369 giving potentially large lepton - flavor violating contributions to @xmath232 . as discussed in section [ sec : flavstruc ] , @xmath30 could alternatively be a product of multiple spurions , some of which conserve flavor and break supersymmetry , and others which do the opposite .
there are additional contributions to @xmath232 from the spurion that generates the @xmath35 terms .
we will assume that the flavor structure of this spurion is identical to that of @xmath30 : this is especially likely in the case that the supersymmetry - breaking piece of @xmath30 is flavor conserving .
the contributions of ( [ eq : lfvmsq ] ) might lead , for example , to slepton oscillation signals at the nlc @xcite or , as considered here , to rare lepton decays .
of course , similar contributions arise in more standard schemes as well : a flavor breaking spurion @xmath370 that generates majorana neutrino masses through @xmath371_f$ ] can also appear in @xmath372_d$ ] , where @xmath24 breaks supersymmetry but not flavor . if the lepton - flavor violating contributions to @xmath366 are not screened by much larger universal contributions , then a generic flavor structure for @xmath373 in ( [ eq : lfvmsq ] ) , or for @xmath6 in the standard case , leads to unacceptably large flavor violating signals .
on the other hand , not every structure for @xmath373 or @xmath6 leads to a phenomenologically acceptable neutrino mass matrix . in light of this ,
we briefly consider forms for @xmath373 motivated by neutrino phenomenology , and estimate the flavor changing signals they induce .
we will not have specific flavor symmetries in mind that motivate the textures we will consider .
moreover we will estimate only the lepton flavor violating signals due to the non - universal contributions to @xmath366 , and make the simplifying assumption that all other potential sources of flavor violation ( for instance , an @xmath374 matrix that is not aligned with @xmath375 ) vanish .
for our discussion we will assume that one dirac neutrino is hierarchically heavier than the others , and we will also take there to be three @xmath13 states , although this is not an important assumption . using our freedom to choose a basis for the @xmath13 s , we consider the leading order flavor structure @xmath376 with @xmath377 and @xmath280 comparable .
that is , there is large @xmath378 mixing as indicated by super - kamiokande , but @xmath379 has only a small component in the heavy state , to satisfy the chooz bound @xcite . since we assume that ( [ eq : lfvmsq ] ) gives contributions to @xmath232 as large as the universal ones , the form taken for @xmath373 suggests that the 23 entry of @xmath232 will be comparable in size to the diagonal entries . in this case
the branching ratio for the process @xmath380 is near the current experimental limit for slepton masses near 100 gev @xcite .
another potential signal is @xmath381 .
the size of the branching ratio depends on the higher order contributions to @xmath373 and is highly model dependent . in the abelian flavor symmetry models considered in @xcite , with
right handed neutrinos integrated out above the flavor scale , both large angle msw and vacuum oscillation solutions to the solar neutrino problem generally lead to too large a rate for @xmath382 , and even models compatible with the small angle msw solution force the slepton masses above @xmath328 500 gev .
here we do not construct flavor models for light dirac neutrinos but instead simply consider the texture @xmath383 with only the order of magnitude of each entry , and not its precise value , indicated .
one finds in this case that @xmath384 has eigenvalues @xmath385 , @xmath386 , and @xmath387 , and mixing angles @xmath388 , @xmath389 , and @xmath390 .
thus this case is most likely to correspond to either large angle msw or vacuum oscillation solutions to the solar neutrino problem . for vacuum oscillations ,
we take @xmath391 .
since the 12 and 13 entries of @xmath384 are both of order @xmath392 , we obtain the order of magnitude relation @xmath393 where @xmath394 is a typical slepton mass , the lightest neutralino is taken to be photino - like , and @xmath395 @xcite .
this case is thus safe as far as @xmath396 is concerned .
depending on the assortment of assumed order unity factors we have ignored , it is still possible , for light sparticle masses , that the branching ratio will be accessible to future experiments . on the other hand , for the large angle msw solution we should take @xmath397 , leading to @xmath398 where we again take @xmath395
. thus @xmath381 forces the sparticle masses to be heavy .
one should keep in mind that these estimates have been obtained ignoring other possible sources of lepton flavor violation , and for a particular texture for @xmath373 .
one choice for the higher order entries in @xmath373 that is compatible with the small angle msw solution to the solar neutrino problem is @xmath399 leading to eigenvalues for @xmath384 @xmath400 , @xmath386 , and @xmath387 , and mixing angles @xmath388 , @xmath401 , and @xmath402 . choosing @xmath403 leads to acceptable masses and mixings . since the 12 and 13 entries of @xmath384
are order @xmath386 , @xmath404 is suppressed by roughly a factor of @xmath405 relative to the large angle msw case .
if large universal contributions to @xmath366 are present , they will suppress the effects of the flavor violating contributions induced by @xmath373 at tree level .
however , even if we ignore ( [ eq : lfvmsq ] ) entirely and take a universal form for @xmath366 at , say , the gut scale , we still obtain potentially interesting flavor violating signals .
this is because the @xmath35 terms generate non universal contributions to @xmath366 radiatively : @xmath406 this effect has been studied in @xcite in the context of see - saw theories with a high scale for the right - handed neutrinos . in these models ,
the dirac neutrino yukawa couplings are sizable and generate additional non - universal contributions , leading to @xmath407 for a universal scalar mass @xmath408 and a right handed neutrino scale @xmath409 .
notice that while in our framework only the @xmath35 term contributions are present and not those from the yukawa couplings , the logarithm is larger than in the see - saw case because the right handed neutrinos remain in the effective theory down to low energies . to calculate rates for flavor changing processes due to ( [ eq : nonu ] ) for a given set of mssm parameters
, one needs to know the @xmath35 matrix .
this ambiguity is at the same level as in @xcite , where the dirac neutrino yukawa couplings are unknown : the size of the largest coupling is fixed by atmospheric neutrino data only once the scale of the right handed neutrinos is specified ( the @xmath35 terms are chosen proportional to the yukawas , with the scale set by the universal gaugino mass ) . setting @xmath410 gev ,
the authors of @xcite considered textures for @xmath6 based on abelian symmetries and found promising signals for @xmath411 and @xmath380 for significant portions of parameter space .
if we take @xmath412 in ( [ eq : nonu ] ) , then our framework yields very similar signals to @xcite for a given choice of susy parameters and a given form for @xmath6 .
the relationship @xmath413 , where @xmath11 is of the order of weak scale and @xmath10 an ultraviolet cutoff , has been tremendously successful in describing the small mass of the neutrino . whether arising from planck - scale suppressed operators in an effective theory , or from a particular realization such as the see - saw mechanism , this has been generally interpreted to signify the presence of lepton - number violating physics at the scale @xmath10 , well above the reach of laboratory high - energy physics .
this belief is predicated upon the idea that the coupling of the neutrino to the lepton - number violating sector of physics is order one , as might be expected in a gut see - saw , for instance .
the likelihood of a light dirac neutrino has been discounted for decades . given the observed mass scales for neutrinos from solar and atmospheric neutrino data
, we would need a yukawa coupling at @xmath414 or smaller which appears difficult to understand when the smallest known yukawa , @xmath375 , is @xmath415 .
there is a possibility to explain the needed small yukawa coupling as a consequence of a new flavor symmetry @xmath8 broken only very slightly . in previous efforts , the factorization of the symmetry group into @xmath416 has been extended to the factorization of the symmetry breaking itself : vevs which break @xmath8 are supersymmetry preserving , while vevs breaking susy ( @xmath417 ) are @xmath8 conserving . if this is not the case , however , the yukawa coupling can have an additional suppression factors in powers of @xmath418 .
the lightness of the higgs doublets in supersymmetric theories ( @xmath419 ) suggests that such a factorization of symmetry breaking is inadequate .
there should exist an additional symmetry group @xmath27 which is broken in the supersymmetry breaking sector .
given that the higgs is kept light by @xmath27 , we may ask whether there might be other particles such as right - handed neutrinos , singlet under the standard model , also kept light by @xmath27 .
conventional & sdirac ( @xmath420 ) & see - saw & & & & eft & @xmath387 & @xmath177 & & & & & @xmath23 & & @xmath177 & & & & @xmath421 & @xmath422 & & @xmath177 & & & @xmath423 & & & @xmath422 & @xmath177 & & @xmath424 & @xmath425 & @xmath426 & @xmath425 & @xmath426 & @xmath425 & @xmath422 one immediate consequence is that the physics responsible for the relationship @xmath427 is not occuring at the scale @xmath10 , and , in particular , that the mass of the standard model singlet state may be much lighter than @xmath10 even as light as @xmath428 itself .
the numerous new possibilities , employing only vevs in integer powers of @xmath429 , are summarized in table [ tb : scenarios ] .
in particular , there are interesting possibilities of sdirac ( a light dirac neutrino ) and smajorana ( a light majorana neutrino with a weak - scale right - handed neutrino ) scenarios in a single generation .
the idea that @xmath27 , which protects @xmath177 , is broken in the supersymmetry breaking sector is by no means purely philosophical . in theories in which @xmath27 is broken by a supersymmetry preserving vev
, we typically expect _ all _ couplings of @xmath13 supermultiplet to be highly suppressed . on the other hand
, the case @xmath27 is broken in the supersymmetry breaking sector immediately invites the possibility of unsuppressed @xmath35 terms for right - handed scalar neutrinos , which radically alter the phenomenology of this scenario relative to previous ones .
for instance , higgs physics can be modified drastically , both in production and decay .
the mass spectrum of sleptons is dramatically altered and the presence of a light @xmath430 sneutrino is permitted .
collider signatures can be changed dramatically . with or without lepton
number violation , the sneutrino is revived as a dark matter candidate .
all of these things are easily realized if @xmath27 is broken by supersymmetry breaking vevs .
furthermore , if the supersymmetry breaking sector breaks flavor symmetries that are also broken in a separate supersymmetry conserving sector , we have a potentially new understanding of the large mixing in the neutrino sector .
since the vevs of the fields @xmath30 which break supersymmetry need not be aligned with flavor violating vevs that preserve supersymmetry , there is no reason to expect the mass eigenstates of neutrinos to be aligned with those of charged fermions , although they may still have a hierarchical structure .
when viewed from the perspective of the @xmath0 problem , such a scenario is exceedingly natural .
the presence of unsuppressed @xmath35 terms provide not only exciting phenomenology , but also the promise that this scenario can be tested in the near future .
while experiments will provide the ultimate test of these ideas , this framework provides exciting possibilities for connections between what previously have seemed separate elements of supersymmetric theories .
0.25 in * acknowledgements * 0.15 in this work was supported in part by the director , office of science , office of high energy and nuclear physics , division of high energy physics of the u.s .
department of energy under contract de - ac03 - 76sf00098 and in part by the national science foundation under grant phy-95 - 14797 .
5 t. yanagida , in proc . of the workshop on the unified theory and baryon number in the universe , eds . o. sawada and a. sugamoto ( kek , tsukuba , 1979 ) 95
; m. gell - mann , p. ramond , r. slansky , in supergravity , ed . by f. van nieuwenhuizen and d. freedman ( amsterdam , north holland , 1979 ) 315 | an alternative to the conventional see - saw mechanism is proposed to explain the origin of small neutrino masses in supersymmetric theories .
the masses and couplings of the right - handed neutrino field are suppressed by supersymmetry breaking , in a way similar to the suppression of the higgs doublet mass , @xmath0 .
new mechanisms for light majorana , dirac and sterile neutrinos arise , depending on the degree of suppression .
superpartner phenomenology is greatly altered by the presence of weak scale right - handed sneutrinos , which may have a coupling to a higgs boson and a left - handed sneutrino .
the sneutrino spectrum and couplings are quite unlike the conventional case - the lightest sneutrino can be the dark matter and predictions are given for event rates at upcoming halo dark matter direct detection experiments .
higgs decays and search strategies are changed .
copious higgs production at hadron colliders can result from cascade decays of squarks and gluinos .
# 1#2 # 1#2#3_phys .
lett . _
* b#1 * ( # 2 ) # 3 # 1#2#3_z .
phys . _
* c#1 * ( # 2 ) # 3 # 1#2#3_phys . rev .
lett . _ * # 1 * ( # 2 ) # 3 # 1#2#3_rev .
mod .
phys . _ * # 1 * ( # 2 ) # 3 # 1#2#3_phys .
rep . _ * # 1 * ( # 2 ) # 3 # 1#2#3_phys .
rev .
_ * d#1 * ( # 2 ) # 3 # 1#2#3_nucl .
phys . _
* b#1 * ( # 2 ) # 3 # 1_d # 1_f # 1_a #
1#1 ucb - pth-00/18 + lbnl-46180 + hep - ph/0006312 + + .5 in * small neutrino masses + from supersymmetry breaking * .50 in nima arkani - hamed , lawrence hall , hitoshi murayama , david smith and neal weiner 0.05 in _ department of physics + university of california , berkeley , california 94720 _ 0.05 in and 0.05 in _ theoretical physics group + ernest orlando lawrence berkeley national laboratory + university of california , berkeley , california 94720 _ .5 in .5 in |
the dark energy survey data management ( desdm ) is the part of the dark energy survey project @xcite that will transfer , process and distribute the data generated by survey s camera decam @xcite .
desdm is a large , scalable system led by the national center for supercomputing applications at the university of illinois at urbana - champaign ( ncsa / uiuc ) consisting of : 1 . an archive system for different levels of data .
2 . scientific codes to process raw data .
3 . database to support calibration , provenance and data analyses .
4 . web portals providing process control and easy access to images and catalogs . 5 . hardware platforms for execution and storage . as of august 2011 ,
the desdm team consists of around 20 computing professionals and physicists from several institutions around the world , each providing their own expertise to its development . in this contribution
we describe the basic design of the desdm and make particular emphasis on how the raw data in the form of ccd images is reduced to its final form as science - ready catalogs .
we also present the testing campaigns that the system is undergoing and the outlook in the short term , in the context of the des project . for more information on other aspects of desdm , see also @xcite @xcite .
the main functions of the system are schematically summarized in fig .
[ fig : desdm_diagram ] .
the design is driven by the science requirements document of the des project , flowed down to the technical level .
additional requirements include the timely and reliable processing and the archiving of the data .
the functions of the desdm are the following : 1 .
* transfer .
* raw images must be transferred from the telescope site at the rate of approximately 100 mbps .
this rate takes into account that 360 science and associated calibration exposures are produced each night , totalling @xmath0 gb of data that have to be transported in less than 18 hours , to allow for nightly processing and feedback . some overhead is included to consider possible network outages .
recent studies indicate normal data delivery will occur in near real time .
this step is taken care of by the national optical astronomy observatory data transport system through a microwave downlink off the mountain and then by network to ncsa .
2 . * processing . *
as the data arrive at ncsa , it is ingested into the system , and sent to high performance computing ( hpc ) resources on xsede @xcite and possibly the open science grid @xcite .
data are staged using using gridftp and globusonline @xcite .
the pipeline containing the parallelization , scientific codes ( see section [ sec : algorithms ] ) and quality assurance is executed for the night s images .
archive and distribution .
* results are then returned to the ncsa primary archive , and releases built .
data files in releases are replicated to the secondary archive at fermilab and tertiary archives at collaborator s sites .
data are also released generally to the collaboration .
catalogs of objects from coadded images and single epoch images , as well as other meta - data are served to the collaboration using an ncsa - provided oracle rac system . after a proprietary period
, data will be released to the community using the same methods .
we expect the system will evolve to use vao protocols and tools in time for the public release . ]
in this section we describe what algorithms are involved in the reduction of the raw data once the process that sends it to the hpc platform has been started .
the basic event in an astronomical observation is the _
exposure_. the decam camera is exposed to the night sky for @xmath1 seconds , generating a file of slightly less than 1 gb in size , which is written in fits format . an exposure ( see fig.[fig : sample_image ] for an example ) consists of 62 ccd images of a part of the sky , covering a total solid angle of 3 square degrees , showing multiple sources and instrumental effects . around 300 scientific images are generated per night , for different pointings , together with about 60 calibration images .
the former have to be corrected for instrumental effects ( detrending ) as well as calibrated for the absolute position ( astrometry ) and absolute flux ( photometry ) . ]
this pipeline contains several astronomy modules that together remove the decam instrumental signatures .
it includes the following steps : * exposure segmentation and crosstalk correction .
the exposure is divided into 62 ccd images and crosstalk between them is accounted for , using the crosstalk coefficients measured from the calibration dataset @xmath2 where @xmath3 stands for the image of a particular ccd and @xmath4 is the crosstalk coefficient describing the fraction of image @xmath5 appearing in image @xmath6 .
* image correction .
this involves correcting for the pedestal or bias level of the ccds , eliminating non - imaging sections of the exposure , eliminating ghost images from multiple scattered light in the optics , removing illumination and fringing effects and correcting for pixel - to - pixel sensitivity variations ( flat - fielding ) .
thermal noise from electrons is also treated but is usually negligible at the camera s working temperature . in this step additional calibration images ( taken during the same night , every season or during commissioning )
are required and must be part of the job submitted to the hpc resource .
after this process , raw data has been turned into reduced images with associated maps containing information on bad pixels and the weights of the image at each particular pixel ( inverse variance ) .
this step requires the identification of positional standards in the exposure , i.e. , stars with very well - known positions in celestial coordinates .
the coordinates of the brightest sources in the exposure are extracted as well ( using the sextractor @xcite package ) in the exposures s own reference system in _
x_-_y _ coordinates . knowing an approximate initial solution ( provided by the telescope s control system ) a match of both catalogs can be performed , as well as the fitting of the best transformation parameters from the image _
x_-_y _ system to the celestial reference system ( absolute system in spherical coordinates ) .
this part is performed by the scamp @xcite software .
additionally , the effect of the distortion caused by the optics towards the edges of the field , has to be included when calculating the astrometric solution .
the actual digital counts observed in each pixel of the detrended image , have to be translated into physical flux units . in order to make this conversion , on every night in which conditions are sufficiently good ( moonless , stable skies ) specific calibration star fields will be imaged .
for these stars , the flux is known or can be obtained ( @xmath7 ) and this can be compared to the total light measured in the image ( @xmath8 ) .
this relationship is called the photometric equation and contains several unknowns which are ccd and filter dependent : @xmath9 where @xmath10 is the photometric zeropoint , representing the normalization value of the image ; @xmath11 is an instrumental coefficient known as _ color term _ which takes into account the shape of the response of the ccd to light in different filters ( @xmath12 ) ; and @xmath13 is the atmospheric extinction coefficient which considers the increased absorption by the atmosphere at the airmass denoted by @xmath14 ( dependent on sky angle ) .
measuring multiple reference stars in the ccds for different filters and angles in the sky provides us with enough equations to solve the system , thus allowing us to find the photometric solution ( values of these constants and their errors ) . in order to reach the scientific requirements of des in terms of depth ,
it is necessary to perform a process called image coaddition .
as the name implies , this is the combination of several overlapping single - epoch images in a given filter to improve the signal to noise of real sources in the image .
moreover , if the different images thus co - added are slightly offset from one another , this procedure has the added bonus of improving the photometric calibration with a more robust determination of the solution , as the photometric solution will incorporate information from different parts of the sky simultaneously .
another advantage is the possibility of eliminating transient effects from the final coadded image such as satellite trails or cosmic rays , as they will show up in only one of the images being added and are easily identifiable .
prior to the coaddition itself , it is necessary to transform the flux values in the individual overlapping ccd pixels into a uniform pixel grid in which the coaddition can be performed ( _ remapping _ ) .
to do this , artificial _ tiles _ in the sky are created , one degree on a side , and one coadded image per band is produced for every tile . for each single - epoch image
, it is determined to which tiles it contributes to ( using swarp @xcite ) .
the photometric solution ( in particular , the zeropoints ) is re - evaluated for these new images .
there are two main caveats to be pointed out : * the color terms of the ccds , which are simple to take into account in single - epoch images , are not easily combined if they vary across the field .
a solution is currently under implementation ; * the point spread function ( psf ) changes within an image and from one image to another , due to varying conditions of the exposure and quality of the sky .
therefore the psf of the coadded image is subject to discontinuous jumps . to deal with the second point above
, a psf homogenization procedure has been developed .
the model psf and its variation has been computed for each image during nightly processing using the psfex @xcite package .
we define the target psf to be used as a circular moffat @xcite function with a full width at half - maximum which is the median of the seeing distribution one in the whole set of images contributing to the coadd .
going back to the individual component images with psf @xmath15 , we find the kernel @xmath16 which minimizes the difference with respect to this median target psf @xmath17 .
@xmath18 where @xmath19 are the elements of a polynomial basis in @xmath20 .
the kernel elements are stored and during the homogenization process , they are recovered to convolve with the image .
once every image has been treated this way , the coaddition will not introduce any discontinuities from the psf variations .
image coaddition takes place off - season due to its increased cpu requirements with respect to the nightly processing .
coadd construction is to be carried out multiple times depending on the specific needs of each science working group , which have different requirements in the balance between depth and source morphology .
for instance , the weak lensing group would be interested in no coaddition at all given their stringent requirements on the shape measurement ( though they will benefit from multiple imaging of the same objects , improving the solution of the extracted shear , see section [ sec : wl ] ) .
the final step in the processing consists in transforming the coadded images in different filters into single object entries in a catalog .
desdm uses the sextractor software in this step too , running over a master image which uses coadd images in all filters to detect where the sources are , and then extracting the relevant information from single - filter coadded images .
an object is identified as such when the convolution of the psf with the image is above a certain local background estimation ( see @xcite for details ) . in this step , it is important to approach the deblending of sources .
this is currently being done by producing several isophotal layers for each object and at each layer where two light-islands join making the decision on whether to merge them into the same object using as a criterion the relative integrated intensity between the branch and full object .
the type of information extracted in this step is positional , photometric and morphological , besides identification and other bookkeeping variables . concerning the astrometry ,
the barycenter for each object is derived using several estimators .
the most reliable one makes use of an iterative calculation through a gaussian window .
this is transformed to celestial coordinates using the astrometric solution found during nightly processing .
photometry is measured using several methods : * simple flux counting inside a fixed circular aperture for several apertures ( in the arcseconds range ) ; * using an elliptical aperture adjusted to the morphological properties of the object .
from the second order moments of the object , we would find the elongation and orientation of the ellipse representing it .
the ellipse scaling factor is derived from the first order moment of the radial distribution @xcite ; * using a fit to the measured psf shape ( suitable for stars ) ; * using a one- or two- component model convolved with a local model of the psf ( exponential and/or spheroidal , suitable for galaxies ) . among the morphological measurements , currently there are two variables appraising the deviation of the object from a point - like shape , therefore providing a star - galaxy separation handle .
one of them relies on a previously trained neural network and is the well - known stellarity parameter class_star from sextractor .
the other arises from the calculation of a discriminant function measuring the deviation of the image from the psf shape .
most of the variables in the catalogs have their corresponding errors and quality flags to guide the selection of sources for analysis .
additional columns are included in the final catalog , and these are produced in pipelines running over images and catalogs , to produce the final coadded catalogs .
two of these are briefly described in the following sections .
an additional catalog is built to identify transient objects , as described in section [ sec : diff_imaging ] .
redshifts in the des will be estimated from the photometric information of the objects in different filters .
traditionally , the approach to this has been to either find the best fit of the spectral energy distribution to a collection of templates for different types of galaxies at different redshifts , or use a neural network trained with spectroscopic information ( see @xcite and @xcite for examples of both ) .
currently the default estimation in desdm uses the latter method , using ten input magnitudes ( fixed circular and automatic elliptical apertures , described above , for five filters each ) . in fig .
[ fig : photoz ] a comparison of photometric redshifts versus spectroscopic ( true ) ones is shown , as obtained in one of the latest data challenges ( section [ sec : dcs ] ) .
there is a large code comparison project within a specific photo - z science working group . ]
the weak lensing probe will require measuring the distortion in the shapes of galaxies to extract the shear produced by gravity from the intervening matter between the source and observer .
it will require a very precise measurement of the local psf shape independently of the standard determination used in the pipeline .
this is done using bright isolated stars and additional instrument data .
the psf is interpolated with a polynomial throughout the image , and it is deconvolved from each galaxy .
the shear will be extracted using all the images where the object shows up ( in multiple filters ) to provide a more robust determination .
the des will contain two modes of operation : the survey mode in which a wide area of the sky will be scanned several times for each filter over the course of the five - year survey ; and a time - domain survey where particular regions will be observed repeatedly on short time - scales ( @xmath21 weekly ) in search of transient phenomena ( from which supernovae of type ia have to be identified ) . from every reduced image coming from the nightly processing of these regions ,
a template is extracted from the coadd corresponding to that position .
this template is subtracted from the reduced image , and all objects remaining above a certain threshold are cataloged .
the desdm also plans to automate the generation of a polygon - based survey mask which would encode information on which single epoch images contributed to the coadded image at a given point and track the coadded magnitude depth , coadded color terms , regions blocked by saturated stars and any other information which is relevant to be included as a survey map rather than in a source - by - source basis .
the mask is generated offline using the mangle @xcite software , which is currently being incorporated to the pipeline .
during the execution of the pipelines , in - built quality assurance modules make sure that the images and catalogs are good enough for scientific analysis and diagnose possible problems .
some sample results are shown in fig.[fig : photometry ] and fig.[fig : astrometry ] .
a broader approach to the testing and validation of the desdm is the data challenge ( dc ) process , in which a small sample of the survey is simulated in detail and fed to the pipelines to generate realistic images and catalogs .
the scale for the sample ranges from a single 0.6 square degree tile to 200 square degrees corresponding to about 10 nights of observations .
it starts with the creation of galaxy catalogs stemming from an n - body simulation@xcite and detailed models of the milky way galaxy for the star component @xcite .
these are merged and fed to an image simulator which includes atmospheric and instrumental effects .
the resulting images serve as inputs for the desdm , as if they had been really observed at the telescope .
this whole process is a joint effort of the stanford , brazil and barcelona teams , for the catalog side , and fermilab , for the image simulation aspect .
the resulting catalogs are then examined by members of the desdm and science working groups .
several of these dcs have taken place during the project s lifetime .
currently the testing approach consists on large data challenges every 6 months to verify the compliance with science requirements , while maintaining a 2-week cycle with the generation of small samples for quick feedback to the developers .
in addition , the system is being used on real images of the blanco cosmology survey @xcite which are proving to be very valuable for the development of the algorithms .
the desdm system has been fully developed and all elements are in place to process the large dataset that will be generated by the des camera .
current testing results point to requirements being met in terms of astrometry , photometry and depth .
others such as completeness and star - galaxy separation are compliant up to shallower magnitudes though improvements are in the works . in parallel , a community pipeline is being developed using essentially the same structure and algorithms but with the goal of addressing non - des user needs , given that two - thirds of the year the instrument will be available to the general astronomical community with access to noao facilities .
funding for the des projects has been provided by the u.s .
department of energy , the u.s
. national science foundation , the ministerio of ciencia y educacin of spain , the science and technology facilities council of the united kingdom , the higher education funding council for england , the national center for supercomputing applications at the university of illinois at urbana - champaign , the kavli institute for cosmological physics at the university of chicago , financiadora de estudos e projetos , fundao carlos chagas filho de amparo pesquisa do estado do rio de janeiro , conselho nacional de desenvolvimento cientfico e tecnolgico and the ministrio da cincia e tecnologia , the deutsche forschungsgemeinschaft and the collaborating institutions in the dark energy survey .
the collaborating institutions are argonne national laboratories , the university of california at santa cruz , the university of cambridge , centro de investigaciones energticas , medioambientales y tecnolgicas - madrid , the university of chicago , university college london , des - brazil , fermilab , the university of edinburgh , the university of illinois at urbana - champaign , the institut de cincies de lespai ( ieec / csic ) , the institut de fsica daltes energies , the lawrence berkeley national laboratory , ludwig - maximilians universitt and the associated excellence cluster universe , the university of michigan , the national optical astronomy observatory , the university of nottingham , the ohio state university , the university of pennsylvania , the university of portsmouth , universidade federal do rio grande do sul , slac , stanford university , the university of sussex and texas a&m university . 99 f.abdalla these proceedings .
j.hao these proceedings .
j.mohr et al .
observatory operations : strategies , proceses , and systems ii .
proceedings of the spie 7016(2008)70160l k.kotwani et al .
8th international workshop on middleware for grids , clouds and e - science mgc ( 2010 ) http://mgc2010.lncc.br/ n.wilkins-diehr et al .
41(11)(2008)32 r.pordes et al . journal of physics : conference series 78(2007)012057 i.foster ieee internet computing may - june(2011)70 s.arnouts & e.bertin a&as 117(1996)393 e.bertin astronomical data analysis software and systems xv .
astronomical society of the pacific conference series 351(2006)112 e.bertin et al .
astronomical data analysis software and systems xi .
astronomical society of the pacific conference series 281(2002)228 e.bertin astronomical data analysis software and systems xx .
astronomical society of the pacific conference series 442(2011)435 a.f.j.moffat a&a 3(1969)455 r.g.kron apjs 43(1980)305 n.bentez apj 536(2000)571 h.oyaizu et al .
apj 674(2008)768 m.swanson et al .
mnras 387(2008)1391 http://space.mit.edu/@xmath21molly/mangle/ m.busha et al .
( 2011 ) in preparation b.rosetto et al .
aj 141(2011)185 j.mohr http://www.usm.uni-muenchen.de/@xmath21jmohr/bcs | the dark energy survey ( des ) is a project with the goal of building , installing and exploiting a new 74 ccd - camera at the blanco telescope , in order to study the nature of cosmic acceleration .
it will cover 5000 square degrees of the southern hemisphere sky and will record the positions and shapes of 300 million galaxies up to redshift 1.4 .
the survey will be completed using 525 nights during a 5-year period starting in 2012 .
about o(1 tb ) of raw data will be produced every night , including science and calibration images .
the des data management system has been designed for the processing , calibration and archiving of these data .
it is being developed by collaborating des institutions , led by ncsa . in this contribution
, we describe the basic functions of the system , what kind of scientific codes are involved and how the data challenge process works , to improve simultaneously the data management system algorithms and the science working group analysis codes . |
integrable equations exhibit many extraordinary properties like infinitely many conservation laws , multi- hamiltonian structures , soliton solutions etc .
many integrable equations in 1 + 1 dimensions like kdv , mkdv , harry - dym , boussinesq equations possess interesting conformal properties as well @xcite .
they can be written in terms of the so - called independent conformal invariants of the function @xmath2 : [ eq1 ] p_1 & = & + p_2 & = & \{;x}- + here @xmath3 denotes the schwarz derivative .
a quantity is called conformally invariant if it is invariant under the mbius transformation [ eq2 ] & & , .
for example , the kdv equation [ eq3 ] u_t+au_xxx+3uu_x=0 ( @xmath4 is a constant ) can be written in a schwarzian form , i.e. in terms of the conformal invariants ( [ eq1 ] ) as @xmath5 or [ eq3b ] + a\{;x}=0 where [ eq3a ] u = a\{;x}. the kdv and camassa - holm ( ch ) @xcite equations arise also as equations of the geodesic flow for the @xmath6 and @xmath7 metrics correspondingly on the bott - virasoro group @xcite .
the conformal properties of these equations and their link to the bott - virasoro group originate from the hamiltonian operator [ eq4 ] d_3a^3+u(x ) + u(x ) where @xmath8 .
@xmath9 is also known as the third bol operator @xcite and is the conformally covariant version of the differential operator @xmath10 , see also @xcite .
@xmath9 defines a poisson bracket @xcite [ eq5 ] \{f , g}=((a^3+u(x ) + u(x ) ) ) .
dx and kdv , ch and many other equations @xcite can be written in the form [ eq6 ] u_t=\{u , h } for some hamiltonian @xmath11 .
on the other hand , suppose for simplicity that @xmath12 is @xmath13 periodic in @xmath14 , i.e. [ eq7 ] u(x)=^_-l_ne^inx- then the fourier coefficients @xmath15 close a classical virasoro algebra with respect to the poisson bracket ( [ eq5 ] ) : [ eq8 ] \{l_n , l_m}=(n - m)l_n+m+a(n^3-n)_n+m,0 therefore , it is natural to expect that equations which can be written in the form ( [ eq6 ] ) exibit interesting conformal properties , and this is the case indeed @xcite . in what follows we will concentrate on the camassa - holm and the associated camassa - holm equation .
the camassa - holm equation @xcite @xmath16 describes the unidirectional propagation of shallow water waves over a flat bottom @xcite .
ch is a completely integrable equation @xcite , describing permanent and breaking waves @xcite .
its solitary waves are stable solitons @xcite .
ch arises also as an equation of the geodesic flow for the @xmath7 metrics on the bott - virasoro group @xcite .
the equation ( [ eq9 ] ) admits a lax pair @xcite [ eq10 ] v_xx&=&(-(m+))v + [ eq11 ] v_t&=&-(+u - a)v_x+v where [ eq12 ] m = u - u_xx . in order to find the schwarz form for the ch equation we proceed as follows .
let @xmath17 and @xmath18 be two linearly independent solutions of the system ( [ eq10 ] ) , ( [ eq11 ] ) and let us define [ eq13 ] = then , from ( [ eq11 ] ) it follows that [ eq14 ] = -u+(a- ) according to the theorem 10.1.1 from @xcite due to ( [ eq10 ] ) we also have [ eq15 ] \ { ; x}=2 m + a- from ( [ eq12 ] ) , ( [ eq14 ] ) and ( [ eq15 ] ) we obtain the schwarz - camassa - holm ( s - ch ) equation : [ eq16 ] ( 1-^2)+\{;x}=a- since @xmath19 is an arbitrary constant , one can take @xmath20 ( if @xmath21 ) and then the s - ch equation ( [ eq16 ] ) acquires the form @xmath22 or [ eq17 ] ( 1-^2)+a\{;x}=0 applying the hodograph transform [ eq17a ] x , tt , x to the s - ch ( [ eq17 ] ) and using the transformation properties of the schwarzian derivative @xcite [ eq17b ] \{;x}=-_x^2\{x ; } we obtain the following integrable deformation of the harry dym equation for the variable @xmath23 : [ eq17c ] v_t+v^2[v(v_^-1(v^-1)_t)_]_=av^3v _ the equations ( [ eq9 ] ) and ( [ eq17 ] ) with @xmath24 are not equivalent as a matter of fact ( [ eq17 ] ) implies ( [ eq9 ] ) , cf .
it is often convenient to think that the lax operator belongs to some lie algebra , and the corresponding jost solution- to the corresponding group .
thus the relation between @xmath12 and @xmath25 ( see ( [ eq13 ] ) ) resembles the relation between the lie group and the corresponding lie algebra , as pointed out in @xcite .
more precisely , the following proposition holds : [ p1 ] let @xmath25 be a solution of ( [ eq16 ] ) .
then @xmath26 and @xmath27 are solutions of ( [ eq10 ] ) , ( [ eq11 ] ) with @xmath28 _ proof _ : it follows easily by a direct computation .
note that the proposition [ p1 ] is consistent with ( [ eq13 ] ) . from proposition [ p1 ]
it follows the general solution of ( [ eq10 ] ) , ( [ eq11 ] ) is [ eq18 ] v= where @xmath29 and @xmath30 are two arbitrary constants , not simultaneously zero .
note that the expression ( [ eq18 ] ) is covariant with respect to the mbius transformation ( [ eq2 ] ) , i.e. under ( [ eq2 ] ) , the expression ( [ eq18 ] ) transforms into an expression of the same form but with constants [ eq19 ] aa=,bb=.
an inverse scattering method , which can be applied directly to the spectral problem ( [ eq10 ] ) is not developed completely yet .
however , the solutions of the ch ( in parametric form ) can be obtained by an implicit change of variables , which maps ( [ eq10 ] ) to the well known spectral problem of the kdv hierarchy .
this change of variables leads to the so - called associated camassa - holm equation ( ach ) @xcite [ eq20 ] p_t=p^2f _ x , f= ( p)_xt- . indeed , the ach is related to the ch equation ( [ eq9 ] ) through the following change of variables : [ eq21 ] p= , f=-- [ eq22 ] dx=p dx+ ( -+)p dt , dt=dt .
in what follows we will omit the dash from the new variables @xmath31 and @xmath32 and we will write simply @xmath14 and @xmath33 instead . the ach equation ( [ eq20 ] ) is related to the kdv hierarchy @xcite since it may be written in the form [ eq23 ] u_t=-2p_x , u=- ( ) and the lax pair for ( [ eq23 ] ) is [ eq24 ] v_xx+(u-)v=0 , + [ eq25 ] v_t=(pv_x - p_xv ) .
the second part of ( [ eq23 ] ) is the ermakov - pinney equation for @xmath34 ( for a known @xmath35 ) @xcite . in order to obtain the schwarzian form of the ach equation
, we introduce again @xmath36 as in ( [ eq13 ] ) , where now @xmath17 and @xmath18 are two linearly independent solutions of the system ( [ eq24 ] ) , ( [ eq25 ] ) . then , by analogous arguments , from ( [ eq24 ] ) it follows [ eq26 ] u= \{;x}+ , and from ( [ eq25 ] ) [ eq27 ] p=. using the relation between @xmath34 and @xmath35 given in the first part of ( [ eq23 ] ) , ( [ eq26 ] ) and ( [ eq27 ] ) we obtain the schwarz - ach ( s - ach ) equation @xmath37 , or [ eq28 ] \{;x}_t+()_x=0 .
now one can easily check by a direct computation the validity of the following let @xmath25 be a solution of s - ach ( [ eq28 ] ) .
then @xmath26 and @xmath27 are solutions of ( [ eq24 ] ) , ( [ eq25 ] ) with [ eq28a ] p=. as a corollary we find that the general solution of ( [ eq24 ] ) , ( [ eq25 ] ) , expressed through the solution of ( [ eq28 ] ) is [ eq29 ] v= where @xmath29 and @xmath30 are two arbitrary constants , not simultaneously zero . as we know the expression ( [ eq29 ] )
is covariant under the mbius transformations ( [ eq2 ] ) .
a transformation , which connects the solution @xmath12 of one equation to the solution @xmath38 of the same ( or another ) equation is called a bcklund transform ( bt ) .
such type of transformations were first discovered in relation to the sine - gordon equation by a. bcklund around 1883 .
usually bt depends on an arbitrary parameter which is called bcklund parameter . the lax pair ( [ eq24 ] ) , ( [ eq25 ] ) has the following important property @xcite , namely the bt for the ach : if @xmath39 is a solution of the lax pair ( [ eq24 ] ) , ( [ eq25 ] ) with potentials @xmath35 and @xmath34 , then @xmath40 is a solution of ( [ eq24 ] ) , ( [ eq25 ] ) with potentials [ eq30 ] & = & u+2(v)_xx , + [ eq31 ] & = & p-(v)_xt . from ( [ eq29 ] ) and ( [ eq31 ] ) it now follows that the bt for the solution ( [ eq28a ] ) of the ach equation can be expressed through the solution of the s - ach equation as follows : [ eq32 ] = -()_xt .
since the constants @xmath29 , @xmath30 are not simultaneously zero , one can always factor out one of the two constants , i.e. ( [ eq32 ] ) obviously depends only on the ratio of these two constants ( which is the bcklund parameter in this case ) .
we conclude this section with the following remark for the kdv equation .
it is not difficult to see through similar considerations , that the bcklund transform for the solution ( [ eq3a ] ) of the kdv equation ( [ eq3 ] ) is [ eq33a ] = a\{;x}+4a ( ) _ xx , where @xmath25 is the solution of the s - kdv equation ( [ eq3b ] ) .
then the choice @xmath41 in ( [ eq33a ] ) corresponds to the exotic bcklund transform found by galas in @xcite and mentioned in @xcite .
the solutions of the ach and s - ach equations can be constructed , following the scheme for the construction of the solutions of the kdv hierarchy @xcite .
let @xmath42 be @xmath43 different solutions of the lax pair ( [ eq24 ] ) , ( [ eq25 ] ) with potentials @xmath35 and @xmath34 , taken at @xmath44 respectively and let @xmath17 and @xmath18 be two linearly independent solutions for some value @xmath19 .
consider the wronskian determinants w&=&w(v_1,v_1;1 , , v_1;m_1 , v_2 , v_2;1 ,
, v_2;m_2 , , v_n , v_n;1 , , v_n;m_n ) + w_k&= & w(v_1,v_1;1 , , v_1;m_1 , , v_n , v_n;1 , , v_n;m_n ,
v_k ) , k=1,2 , + [ eq33 ] where @xmath45 , @xmath46 are arbitrary nonnegative integers and the wronskian determinant of @xmath47 functions @xmath48 is defined by [ eq34 ] w(_1 , _ 2, _n)=a , a_ij= , i , j=1,2, ,n .
the following generalization of the crum theorem is valid for the kdv hierarchy and in particular for the ach equation @xcite : [ p4 ] @xmath49 , @xmath50 are two linearly independent solutions of the lax pair ( [ eq24 ] ) , ( [ eq25 ] ) with potentials [ eq35 ] & = & u+2(w)_xx , + [ eq36 ] & = & p-(w)_xt
. from proposition [ p4 ] it follows immediately that [ eq37 ] = = , = are the solutions of the s - ach ( [ eq28 ] ) , which correspond to the potentials = , p=. in addition to the solutions mentioned in @xcite one may construct also the so - called negaton and positon solutions ( which are singular ) and mixed soliton - positon - negaton solutions . for a detailed discussion on the positon and negaton solutions of kdv
we refer to @xcite .
let us return to ( [ eq24 ] ) with @xmath51 , i.e. @xmath52
. then it has the form [ eq280 ] v_xx+e(h,)v=0 , e(h,)=-(+ ) .
the type of the solutions of ( [ eq280 ] ) clearly depends on the sign of the constant @xmath53 . if @xmath54 the independent solutions of ( [ eq280 ] ) and ( [ eq25 ] ) are [ eq281 ] v_1 ( ) & = & ( ( x - ht+x_1 ( ) ) ) , + [ eq282]v_2 ( ) & = & ( ( x - ht+x_2 ( ) ) ) , where @xmath55 are arbitrary functions .
the simplest choice for @xmath56 in ( [ eq35 ] ) , ( [ eq36 ] ) is @xmath57 or @xmath58 ( wronskian of order one ) , which corresponds to the _ one - soliton solution_. the case @xmath59 corresponds to the _ two - soliton solution _ , i.e. this solution describes the interaction of two solitons .
however , according to ( [ eq33 ] ) it is possible to merge two or more identical solitons , taking [ eq283 ] w&=&w(v_1,v_1;1 , , v_1;m_1 ) , or [ eq284 ] w&= & w(v_2,v_2;1 , , v_2;m_2 ) . in physical terms @xmath53
is the energy of the soliton .
since the energy of the solitons is negative , ( [ eq283 ] ) is called _ negaton _ of type @xmath60 and order @xmath61 , and ( [ eq284 ] ) is called _ negaton _ of type @xmath62 and order @xmath63 , cf .
the analogous considerations resulting from the following solutions of ( [ eq280 ] ) , [ eq285 ] v_1 ( ) & = & ( ( x - ht+x_1 ( ) ) ) , + [ eq286]v_2 ( ) & = & ( ( x - ht+x_2 ( ) ) ) ( @xmath64 ) , whose energy is positive , @xmath65 leads naturally to the definition of _ positon _ of type @xmath60 and order @xmath61 via ( [ eq283 ] ) and ( [ eq36 ] ) , and _ positon _ of type @xmath62 and order @xmath63 , ( [ eq284 ] ) and ( [ eq36 ] ) ; cf . @xcite . due to ( [ eq33 ] ) it is possible to construct solutions , which describe the interaction between solitons , positons and negatons .
i.e. the interaction between @xmath43 positons ( of orders @xmath66 ) and @xmath47 solitons is given by the _
@xmath43-positon @xmath47 soliton solution _
@xmath67 of the ach ( [ eq20 ] ) which can be constructed from ( [ eq36 ] ) taking the following solutions of the lax pair ( [ eq24 ] ) , ( [ eq25 ] ) : [ eq38 ] v_i= ( ( x - h_it+y_i(_i))),i=1,
n + [ eq39 ] _ j = ( ( x - h_n+jt+x_j))+ + c_j(-(x - h_n+jt+x_j)),j=1, n , where @xmath68 are arbitrary functions of @xmath19 , @xmath69 , @xmath70 are arbitrary constants and @xmath71 for @xmath72
. then ( cf .
@xcite ) [ eq40 ] _ np_n(x , t)&= & + h&-&(w(v_1,v_1;1 , , v_1;m_1 , ,
v_n , v_n;1 , , v_n;m_n,_1, ,_n))_xt .
+ as an example , we provide the one - positon solution of order one for the ach , see also fig .
[ 1positon ] : [ eq41 ] _ 1p_0(x , t)=h- where @xmath73 , @xmath74 , @xmath4 is an arbitrary constant ; [ eq42 ] z(x , t)&=&2(x - h_1t+a _ 1 ) , + r(x , t)&=&(h^2x+_1(h^2 - 2_1)(ht - a ) ) .
the author is grateful to prof .
a. constantin for helpful discussions and to the referees for valuable suggestions .
the author also acknowledges funding from the science foundation ireland , grant 04/br6/m0042 .
fokas , p.j .
olver , p. rosenau , a plethora of integrable bi - hamiltonian equations , in : algebraic aspects of integrable systems : in memory of irene dorfman , a.s .
fokas and i.m .
gelfand , eds . , progress in nonlinear differential equations , vol .
26 , birkhauser , boston , 1996 , pp . | integrable equations exhibit interesting conformal properties and can be written in terms of the so - called conformal invariants .
the most basic and important example is the kdv equation and the corresponding schwarz - kdv equation .
other examples , including the camassa - holm equation and the associated camassa - holm equation are investigated in this paper .
it is shown that the bcklund transform is related to the conformal properties of these equations .
some particular solutions of the associated camassa - holm equation are discussed also . *
pacs : * 05.45.yv , 02.30.ik * key words : * schwarz derivative , conformal invariants , lax pair , integrable systems , solitons , positons , negatons . * conformal properties and bcklund transform for the associated camassa - holm equation + * * rossen ivanov@xmath0 * -.3 cm c -1 cm @xmath1_school of mathematics , trinity college , _
+ _ dublin 2 , ireland _ + _ tel : + 353 - 1 - 608 2898 _ + _ fax : + 353 - 1- 608 2282 _
+ + _ @xmath0e - mail : [email protected]_ + + -.8 cm |
observations from galactic to cosmological scales leave little room to explanations for dark matter ( dm ) other than that it is provided by a new type of particle .
the most studied option for this particle is a weakly interacting massive particle ( wimp ) .
the reason for this is sometime referred to as the `` wimp miracle '' : for a particle of mass of the weak scale ( @xmath0100 gev ) , the annihilation cross section is given approximately by : @xmath1 and at the same time @xmath2 , from considering the relic density of a stable particle produced thermally in the early universe .
wimps are searched for in a variety of ways : by particle production at accelerators , by searching for signals of nuclear recoils ( direct detection ) and last but not least by searching for a signal from secondary products of wimp annihilation or decay , in particular gamma - rays .
+ spectral signatures in gamma - rays can be classified in smoking gun signals and ambiguous signals .
the smoking gun signals are mainly strong spectral features , such as bumps from virtual internal brems - strahlung @xcite or lines from loop processes , e.g. @xcite @xcite , ambiguous signals are due to continuum emission from pion decay , e.g. @xcite or hard power - laws ( e.g. from final state radiation , e.g. @xcite ) .
+ often constraints in the annihilation cross - section , mass plane of wimps are presented under simplifying assumptions on the final state .
specifically , annihilation into quarks is presented together with annihilation into leptons .
annihilation into quarks yields spectral shapes which are relatively independent of flavor ( e.g. @xcite ) , which is why constraints assuming say 100 % branching fraction into @xmath3 can be taken to be representative for a large class of models . a more accurate approach is provided by considering underlying models providing the wimp , e.g. supersymmetry , as is done in in a fit combining several observables in e.g. @xcite@xcite or considering a large set of individual models , e.g. @xcite .
constraints presented here are for predominant annihilation into quarks . + gamma - rays are detected in space by pair conversion telescopes , such as the fermi large area telescope ( fermi - lat)@xcite and on ground by imaging air cherenkov telescopes ( iact ) , such as hess @xcite , veritas @xcite and magic @xcite , and water cherenkov telescopes , such as milagro @xcite .
most relevant for searches of dark matter are the fermi - lat and iacts , which are complementary in energy coverage .
fermi - lat is mostly sensitive below @xmath4(100 gev ) and is a survey instrument with large field of view ( @xmath0 2.4 sr ) , but with modest effective area ( @xmath5m@xmath6 ) .
iacts are pointing instruments are sensitive mostly to energies above @xmath4(100 gev ) , they have a small field of view ( @xmath0 0.003 sr ) , but effective areas of @xmath7m@xmath6 .
the number of photons created by dm particle annihilation is proportional to the square of the dm density along the line of sight .
n - body simulations show that our visible galaxy is embedded in a much larger roughly spherical halo , which however shows a lot of substructure , e.g. @xcite .
the density is largest at the center of the halo , which is why the galactic centre should provide the largest signal from dm annihilation .
substructure reveals itself by close - by dwarf galaxies , but n - body simulations usually predict more abundant substructure than observed by `` barionic condensates '' , i.e. the existence of dark matter - only satellites is conjectured .
in addition , integrating over density enhancements may yield a galactic diffuse contribution .
the most interesting extragalactic source - type is galaxy clusters , that might yield a signal due to strong substructure enhancement , e.g. @xcite .
diffuse extragalactic emission , originating from integration over all dm halos throughout the universe is another viable channel .
table [ tab : targets ] summarizes the targets .
+ while consideration of dm structure provides guidance regarding possible targets for searches , the remaining ignorance about the density distribution constitutes the most important uncertainty in constraining the particle properties of dm .
the impact of this ignorance depends on the target : for dwarf galaxies relatively robust density profile estimates can be obtained based on measured stellar motions , whose uncertainties affect constraints by factors of a few @xcite or somewhat less for the combined source analysis @xcite . for
the galactic halo and galaxy clusters uncertainties can be several orders of magnitude , inferred mainly from n - body simulations with relatively weak or indirect experimental constraints .
if the very inner part of the galaxy is excluded from the analysis , these uncertainties are somewhat reduced ( factors of a few , see e.g. @xcite ) . + the present status of the most relevant constraints is summarized in figure [ fig : status ] .
note however that none of them ( except the ( fermi - lat combined dwarf analysis , see below ) includes the uncertainties in the dark matter density .
.usual targets for searches for particle dark matter - and some comments .
[ cols= " < , > , < " , ] about 20 dwarf spheroidal galaxies , satellites of our own galaxy have been targeted by the main gamma - ray experiments .
fermi - lat presented observations of 14 dwarf galaxies with dm constraints for about 2/3 of them based on availability of reasonable estimates of the dm distribution based on stellar data ( see figure [ fig : status ] ) @xcite .
h.e.s.s presented constraints from the sagittarius dwarf @xcite and canis major @xcite and sculptor and carina @xcite , magic for willman i and draco @xcite @xcite and segue i @xcite .
veritas for bootes i , draco , ursa minor and willman i @xcite and a using a deep ( @xmath0 50h ) exposure for segue i @xcite .
the typical exposure of the dwarfs for the ground - based experiments is at about 10 - 20 hours , which can be compared to an equivalent 1000 hours ( 11 month of survey mode ) of fermi , which therefore provides comparable constraints even at 1 tev assumed particle mass .
however , it should be emphasized that in the region above 1 tev the iact will be more sensitive than the fermi - lat , thus truly complementary to fermi - lat .
constraints derived from single dwarf analyses are about one order of magnitude larger than the interesting cosmological bench mark @xmath8 .
it is also noteworthy , that it has recently been claimed that the h.e.s.s .
constraint might be over - optimistic by almost one order of magnitude @xcite , due to updated and more accurate estimates of its dm density . +
a particular interesting analysis approach is the stacking of dwarf galaxies , which due to the universality of the dm spectrum in dwarf galaxies , can be performed as a combined likelihood ( `` likelihood stacking '' ) analysis @xcite .
not only is the statistical power increased , but also the impact of individual uncertainties in dark matter density , is reduced .
this approach resulted in the strictest limits on annihilation cross - section , yet arguably most robust , excluding the canonical cosmological cross - section up to masses of about 30 gev . as mentioned previously , the analysis presented in @xcite constitutes a step forward in one other aspect :
the uncertainties in the dm density have been included in the derivation of the constraints in a statistically well established fashion ( `` profile likelihood '' ) .
the galactic centre ( gc ) is expected to be one of the strongest sources for gamma - ray radiation due to dm annihilation . at the same time , the gc is crowded with conventional gamma - ray sources : hess and fermi - lat sources at the gc are consistent with each other and known sources @xcite@xcite@xcite@xcite .
has used the spectrum of the gc sources ( which is inconsistent with being dominantly due to dm ) , to constrain dm models @xcite , which allows constraining models with extreme properties ( for example those having a large contribution from virtual internal bremsstrahlung ) and under assumptions of very cuspy profiles .
+ the more promising approach is to exclude the very center from the analysis .
however , for gev energies the modeling of the diffuse emission in the inner galaxy poses a challenge , which is why fermi - lat has not presented dark matter constraints for the inner galaxy yet . for galactic diffuse emission excluding the inner 15@xmath9 @xmath10 15@xmath9 ,
an analysis attempting to quantitatively take into account the uncertainties in the cosmic ray induced galactic diffuse emission obtaining competitive constraints has been presented in @xcite . + at iact energies the systematics of the diffuse emission become less important .
the analysis presented in @xcite applies an on - off technique , where ( as usual in iacts ) the background is determined by off - source observations . in this particular case ,
the off source region is defined within the field of view , at distances greater @xmath11 kpc from the galatctic center .
the signal region excludes the very center , i.e. distances smaller than @xmath0 0.05 kpc .
this excludes not only the strong galactic centre sources , but also makes the constrain relatively insensitive ( within a factor 2 ) to the assumptions on dm profile .
the background estimate ( except for instrumental systematics ) is sensitive to gradients in galactic diffuse emission predictions only , which should be negligible at these energies . as the gc is observed for almost 100 hours , and the expected signal is expected to be comparably large , the obtained constraints obtained are up to one order of magnitude better than for dwarfs , and the observation of the gc halo seems to be the most promising venue for iact for the moment
. + the cherenkov telescope array ( cta ) will be the next generation imaging air cherenkov telescope @xcite .
cta will cover over 3 decades in energy , from a few tens of gev up to several tens of tev .
the energy resolution and pointing resolution will be improved by a factor 2 to 3 .
the field of view will be up to 10 degrees ( as compared to about 5 degrees for present day instruments ) .
a comprehensive study of ctas capabilities with respect to fundamental physics has been presented in @xcite . as is the case for present day experiments
, the most sensitive target will be galactic center halo .
the forecast is that the thermal wimp cross - section will be constrained from @xmath0 200 gev up to several tev , see figure [ fig : cta_forecast ] .
dark matter lines have been suggested as a smoking gun signal for dm roughly 15 years ago @xcite .
constraints on line emission have been presented using egret @xcite data , fermi - lat @xcite@xcite and h.e.s.s .
data @xcite@xcite .
however , very recently , the claim of an indication of line emission in fermi - lat data @xcite @xcite has drawn considerable attention also at this conference @xcite @xcite . using an analysis technique similar to @xcite , but doubling the amount of data as well as optimizing the region of interest for signal over square - root of background
, @xcite found a ( trial corrected ) 3.2 @xmath12 significant excess at a mass of @xmath0 130 gev that , if interpreted as a signal would amount to a cross - section of about @xmath13 .
the signal is concentrated on the galactic centre with a spatial distribution consistent with an einasto profile @xcite .
this is marginally compatible with the upper limit presented in @xcite .
the result was later confirmed by @xcite , who even claimed a ( trial corrected ) 5@xmath12 detection significance .
+ the main challenge to this claim is the fact that the gamma - ray signal from the earth s limb shows an excess of similar significance at the same energy , see e.g. weniger ( this conference ) .
this effect appears after an appropriate zenith cut is applied that takes account for the fact that the fermi - lat is usually not head - on exposed to the earth limb .
gamma - rays in the earth limb are caused by cosmic - ray interaction in the earth atmosphere and their spectrum is featureless .
the sample therefore constitutes a useful control sample for ( not only ) spectral feature searches .
the community is now impatiently waiting for a verdict from the fermi - lat collaboration .
the main problem at this point is the limited statistics in both the gc and limb sample .
the fermi - lat has scheduled increase in statistics for the limb sample , by weekly 3 hour long earth limb observations .
+ should an instrumental origin of the signal be ruled out ( or results are inconclusive ) , fermi - lat could alter its observation mode to increase exposure to the galactic centre @xcite , on the other hand independent confirmation will be necessary .
both the operational hess ii iact and the future satellite mission gamma-400 @xcite seem to be sensitive enough to provide confirmation @xcite . for indirect detection ,
constraining annihilation cross - section , gamma - rays are the most attractive probe , providing targets with high signal flux expectation and relatively well constrained astrophysical uncertainties .
as this conference is about gamma - rays , the detailed discussion of results of direct searches and accelerator searches ( as well as other indirect probes ) is out of scope .
however , it is worth pointing out the complementarity of the different probes once again .
this complementarity is ultimately model - dependent , but at least for one of the most attractive candidates for a wimp theory ( supersymmetry ) it is apparent . in figure
[ fig : complementarity1 ] , taken from @xcite , it is obvious that there is little correlation between scattering cross section and annihilation cross - section in phenomenological minimal supersymmetry : direct detection and neutrinos from the sun constrain the model space approximately orthogonal to indirect detection by gamma - rays .
+ another example concerns collider searches for particles beyond the standard model .
collider detection and subsequent measurements of the sparticle mass spectrum and splitings of a supersymmetric model providing dark matter might be used to calculate annihilation cross - sections and relic density @xcite . in the framework of phenomenological minimal supersymmetry ,
however , it is quite conceivable that only after a combination with indirect detection experiments the solution providing dm can be identified @xcite .
since the launch of the fermi - lat indirect detection of dark matter with gamma - rays has reached sensitivities that allow to probe the most interesting parameter space of the wimp paradigm .
the crucial target here are dwarf spheroidal galaxies that are analysed in a combined fashion .
this gain in sensitivity is naturally accompanied by a number of detection claims or indications , the most recent being a line feature at a wimp mass of @xmath0 130 gev at the galactic centre . in the near future ,
ut is now up to the fermi - lat instrument team , as well as possible hess ( within a timescale of about 1 - 2 years ) to confirm or refute this claim .
+ iacts have gained ground .
the constraints obtained from observations of the galactic centre halo constitute a large step forward in terms of constraining wimp parameter space above masses of about a tev .
predictions for cta , employing a galactic center halo analysis , together with forecasts on the fermi - lat constraints obtainable after 10 years arrive at the exciting possibility that the cosmological benchmark cross - section could be probed from masses of a few gev to about 10 tev .
+ as the popularity of the wimp as dark matter candidate rests in large part on the `` wimp miracle '' , it is fair to claim that a non detection in the coming 10 years will imply a serious challenge to the paradigm . or to put it more optimistically : a discovery within the next 10 years is certainly in the cards .
the author thanks lars bergstrm and joakim edsj for useful discussions .
christian farnier is thanked for providing the model points in figures 1 and 2 .
jc is a research fellow of the royal swedish academy of sciences financed by a grant of the knud and alice wallenberg foundation .
p. scott , j. conrad , j. edsjo , l. bergstrom , c. farnier and y. akrami , jcap * 1001 * , 031 ( 2010 ) [ arxiv:0909.3300 [ astro-ph.co ] ] .
j. ripken , j. conrad and p. scott , jcap * 1111 * ( 2011 ) 004 [ arxiv:1012.3939 [ astro-ph.he ] ] .
m. ackermann _ et al .
_ [ fermi - lat collaboration ] , phys . rev .
lett . * 107 * ( 2011 ) 241302 [ arxiv:1108.3546 [ astro-ph.he ] ] .
a. abramowski _ et al .
_ [ h.e.s.s .
collaboration ] , phys .
* 106 * ( 2011 ) 161301 [ arxiv:1103.3266 [ astro-ph.he ] ] .
e. a. baltz , m. battaglia , m. e. peskin and t. wizansky , phys .
d * 74 * ( 2006 ) 103521 [ hep - ph/0602187 ] .
g. bertone , d. g. cerdeno , m. fornasa , l. pieri , r. ruiz de austri and r. trotta , phys .
d * 85 * ( 2012 ) 055014 [ arxiv:1111.2607 [ astro-ph.he ] ] .
q. yuan , b. yue , b. zhang and x. chen , jcap * 1104 * ( 2011 ) 020 [ arxiv:1104.1233 [ astro-ph.co ] ] .
m. ackermann _ et al .
_ [ fermi lat collaboration ] , phys . rev .
d * 85 * ( 2012 ) 083007 [ arxiv:1202.2856 [ astro-ph.he ] ] . | in this contribution i review the present status and discuss some prospects for indirect detection of dark matter with gamma - rays .
thanks to the fermi large area telescope , searches in gamma - rays have reached sensitivities that allow to probe the most interesting parameter space of the weakly interacting massive particles ( wimp ) paradigm . this gain in sensitivity
is naturally accompanied by a number of detection claims or indications , the most recent being the claim of a line feature at a dark matter particle mass of @xmath0 130 gev at the galactic centre , a claim which requires confirmation from the fermi - lat collaboration and other experiments , for example hess ii or the planned gamma-400 satellite .
predictions for the next generation air cherenkov telescope , cherenkov telescope array ( cta ) , together with forecasts on future fermi - lat constraints arrive at the exciting possibility that the cosmological benchmark cross - section could be probed from masses of a few gev to a few tev .
consequently , non - detection would pose a challenge to the wimp paradigm , but the reached sensitivities also imply that optimistically a detection is in the cards .
+ address = oskar klein centre , physics department , stockholm university |
the green s functions of qcd are the fundamental building blocks of hadron phenomenology @xcite .
their infrared behaviour is also known to contain essential information about the realisation of confinement in the covariant formulation of qcd , in terms of local quark and gluon field systems .
the landau gauge dyson - schwinger equation ( dse ) studies of refs .
@xcite established that the gluon propagator alone does not provide long - range interactions of a strength sufficient to confine quarks .
this dismissed a widespread conjecture from the 1970 s going back to the work of marciano , pagels , mandelstam and others .
the idea was revisited that the infrared dominant correlations are instead mediated by the faddeev - popov ghosts of this formulation , whose propagator was found to be infrared enhanced .
this infrared behaviour is now completely understood in terms of confinement in qcd @xcite , it is a consequence of the celebrated kugo - ojima ( ko ) confinement criterion .
this criterion is based on the realization of the unfixed global gauge symmetries of the covariant continuum formulation . in short ,
two conditions are required by the ko criterion to distinguish confinement from coulomb and higgs phases : ( a ) the massless single particle singularity in the transverse gluon correlations of perturbation theory must be screened non - perturbatively to avoid long - range fields and charged superselection sectors as in qed .
( b ) the global gauge charges must remain well - defined and unbroken to avoid the higgs mechanism . in landau gauge , in which the ( euclidean ) gluon and ghost propagators
, @xmath0 are parametrised by the two invariant functions @xmath1 and @xmath2 , respectively , this criterion requires @xmath3 the translation of ( b ) into the infrared enhancement of the ghost propagator ( 2b ) thereby rests on the ghost / anti - ghost symmetry of the landau gauge or the symmetric curci - ferrari gauges . in particular , this equivalence does not hold in linear covariant gauges with non - zero gauge parameter such as the feynman gauge .
as pointed out in @xcite , the infrared enhancement of the ghost propagator ( 2b ) represents an additional boundary condition on dse solutions which then lead to the prediction of a conformal infrared behaviour for the gluonic correlations in landau gauge qcd consistent with the conditions for confinement in local quantum field theory .
in fact , this behaviour is directly tied to the validity and applicability of the framework of local quantum field theory for non - abelian gauge theories beyond perturbation theory .
the subsequent verification of this infrared behaviour with a variety of different functional methods in the continuum meant a remarkable success .
these methods which all lead to the same prediction include studies of their dyson - schwinger equations ( dses ) @xcite , stochastic quantisation @xcite , and of the functional renormalisation group equations ( frges ) @xcite .
this prediction amounts to infrared asymptotic forms @xmath4 for @xmath5 , which are both determined by a unique critical infrared exponent @xmath6 with @xmath7 . under a mild regularity assumption on the ghost - gluon vertex @xcite
, the value of this exponent is furthermore obtained as @xcite @xmath8 the conformal nature of this infrared behaviour in the pure yang - mills sector of landau gauge qcd is evident in the generalisation to arbitrary gluonic correlations @xcite : a uniform infrared limit of one - particle irreducible vertex functions @xmath9 with @xmath10 external gluon legs and @xmath11 pairs of ghost / anti - ghost legs of the form @xmath12 when all @xmath13 , @xmath14 . in particular , the ghost - gluon vertex is then infrared finite ( with @xmath15 ) as it must @xcite , and the non - perturbative running coupling introduced in @xcite via the definition @xmath16 approaches an infrared fixed - point , @xmath17 for @xmath18 .
if the ghost - gluon vertex is regular at @xmath19 , its value is maximised and given by @xcite @xmath20 comparing the infrared scaling behaviour of dse and frge solutions of the form of eqs .
( [ infrared - gh_gl ] ) , it has in fact been shown that in presence of a single scale , the qcd scale @xmath21 , the solution with the infrared behaviour ( [ kappaz = kappag ] ) and ( [ genir ] ) , with a positive exponent @xmath22 , is unique @xcite .
because of its uniqueness , it is nowadays being called the _ scaling solution_. this uniqueness proof does not rule out , however , the possibility of a solution with an infrared finite gluon propagator , as arising from a transverse gluon mass @xmath23 , which then leads to an essentially free ghost propagator , with the free massless - particle singularity at @xmath24 , _
i.e. _ , @xmath25 for @xmath5 .
the constant contribution to the zero - momentum gluon propagator , @xmath26 , thereby necessarily leads to an infrared constant ghost renormalisation function @xmath2 .
this solution corresponds to @xmath27 and @xmath28 .
it does not satisfy the scaling relations ( [ kappaz = kappag ] ) or ( [ genir ] ) .
this is because in this case the transverse gluons decouple for momenta @xmath29 , below the independent second scale given by their mass @xmath23 .
it is thus not within the class of scaling solutions considered above , and it is termed the _ decoupling solution _ in contradistinction @xcite .
the interpretation of the renormalisation group invariant ( [ alpha_minimom ] ) as a running coupling does not make sense in the infrared in this case , in which there is no infrared fixed - point and no conformal infrared behaviour . without infrared enhancement of the ghosts in landau gauge ,
the global gauge charges of covariant gauge theory are spontaneously broken .
within the language of local quantum field theory the decoupling solution can thus only be realised if and only if it comes along with a higgs mechanism and massive physical gauge bosons .
the schwinger mechanism can in fact be described in this way , and it can furthermore be shown that a non - vanishing gauge - boson mass , by whatever mechanism it is generated , necessarily implies the spontaneous breakdown of global symmetries @xcite .
early lattice studies of the gluon and ghost propagators supported their predicted infrared behaviour qualitatively well . because of the inevitable finite - volume effects , however , these results could have been consistent with both , the scaling solution as well as the decoupling solution . recently
, the finite - volume effects have been analysed carefully in the dyson - schwinger equations to demonstrate how the scaling solution is approached in the infinite volume limit there @xcite . comparing these finite volume dse results with latest @xmath30 lattice data on impressively large lattices @xcite , corresponding to physical lengths of up to @xmath31 fm in each direction , finite - volume effects appear to be ruled out as the dominant cause of the observed discrepancies with the scaling solution .
the lattice results are much more consistent with the decoupling solution which poses the obvious question whether there is something wrong with our general understanding of covariant gauge theory or whether we are perhaps comparing apples with oranges when applying inferences drawn from the infrared behaviour of the lattice landau gauge correlations on local quantum field theory ? the latter language is based on a cohomology construction of a physical hilbert space over the indefinite metric spaces of covariant gauge theory from the representations of the becchi - rouet - stora - tyutin ( brst ) symmetry . but
do we have a non - perturbative definition of a brst charge ?
the obstacle is the existence of the so - called gribov copies which satisfy the same gauge - fixing condition , _
i.e. _ , the lorenz condition in landau gauge , but are related by gauge transformations , and are thus physically equivalent .
in fact , in the direct translation of brst symmetry on the lattice , there is a perfect cancellation among these gauge copies which gives rise to the famous neuberger @xmath32 problem .
it asserts that the expectation value of any gauge invariant ( and thus physical ) observable in a lattice brst formulation will always be of the indefinite form @xmath32 @xcite and therefore prevented such formulations for more than 20 years now .
in present lattice implementations of the landau gauge this problem is avoided because the numerical procedures are based on minimisations of a gauge fixing potential w.r.t .
gauge transformations . to find absolute minima is not feasable on large lattices as this is a non - polynomially hard computational problem .
one therefore settles for local minima which in one way or another , depending on the algorithm , samples gauge copies of the first gribov region among which there is no cancellation . for the same reason ,
however , this is not a brst formulation .
the emergence of the decoupling solution can thus not be used to dismiss the ko criterion of covariant gauge theory in the continuum .
from the finite - volume dse solutions of @xcite it follows that a wide separation of scales is necessary before one can even hope to observe the onset of an at least approximate conformal behaviour of the correlation functions in a finite volume of length @xmath33 .
what is needed is a reasonably large number of modes with momenta @xmath34 sufficiently far below the qcd scale @xmath35 whose corresponding wavelengths are all at the same time much shorter than the finite size @xmath33 , @xmath36 it was estimated that this requires sizes @xmath33 of about @xmath37 fm , especially for a power law of the ghost propagator of the form in ( [ infrared - gh_gl ] ) to emerge in a momentum range with ( [ scales ] ) .
a reliable quantitative determination of the exponents and a verification of their scaling relation ( [ kappaz = kappag ] ) on the other hand might even require up to @xmath38 fm @xcite . as an alternative to the brute - force method of using ever larger lattice sizes for the simulations
might therefore be to ask what one observes when the formal limit @xmath39 is implemented by hand .
this should then allow to assess whether the predicted conformal behaviour can be seen for the larger lattice momenta @xmath34 , after the upper bound in ( [ scales ] ) has been removed , in a range where the dynamics due to the gauge action would otherwise dominate and cover it up completely .
therefore , the ghost and gluon propagators of pure @xmath30 lattice landau gauge were studied in the strong coupling limit @xmath40 in @xcite . in this limit , the gluon and ghost dressing functions tend towards the decoupling solution at small momenta and towards the scaling solution at large momenta ( in units of the lattice spacing @xmath41 ) as seen in figure [ beta0props ] .
the transition from decoupling to scaling occurs at around @xmath42 , independent of the size of the lattice .
the observed deviation from scaling at @xmath43 is thus not a finite - size effect .
the high momentum branch can be used to attempt fits of @xmath44 and @xmath45 in ( [ infrared - gh_gl ] ) and the data is consistent with the scaling relation ( [ kappaz = kappag ] ) . with some dependence on the model used to fit the data , good global fits
are generally obtained for @xmath46 , with very little dependence on the lattice size . for the scaling solution one would expect the running coupling defined by ( [ alpha_minimom ] ) to approach its constant fixed - point value in the strong - coupling limit , and this is indeed being observed for the scaling branch @xcite : the numerical data for the product ( [ alpha_minimom ] ) levels at @xmath47 for large @xmath48 . as expected for an exponent @xmath22 slightly smaller than the value in ( [ kappa_c ] ) , see @xcite , this is just below the upper bound given by ( [ alpha_c ] ) , @xmath49 for @xmath30 .
when comparing various definitions of gauge fields on the lattice , all equivalent in the continuum limit , one furthermore observes that neither the estimate of the critical exponent @xmath50 nor the corresponding value of @xmath51 are sensitive to the definition used @xcite .
this is in contrast to the decoupling branch for @xmath52 , which is very sensitive to that definition .
different definitions , at order @xmath53 and beyond , lead to different jacobian factors .
this is well known from lattice perturbation theory where , however , the lattice slavnov - taylor identities guarantee that the gluon remains massless at every order by cancellation of all quadratically divergent contributions to its self - energy .
the strong - coupling limit , where the effective mass in ( [ decoupling ] ) behaves as @xmath54 , therefore shows that such a contribution survives non - perturbatively in minimal lattice landau gauge .
this contribution furthermore depends on the measure for gauge fields whose definition from minimal lattice landau is therefore ambiguous .
one might still hope that this ambiguity will go away at non - zero @xmath55 , in the scaling limit . while this is true at large momenta , it is not the case in the infrared , at least not for commonly used values of the lattice coupling such as @xmath56 or @xmath57 in @xmath30 , as demonstrated in @xcite .
it would obviously be desirable to have a brst symmetry on the lattice which could then provide lattice slavnov - taylor identities beyond perturbation theory . in principle , this could be achieved by inserting the partition function of a topological model with brst exact action into the gauge invariant lattice measure . because of its topological nature , this gauge - fixing partition function @xmath58 will be independent of gauge orbit and gauge parameter .
the problem is that in the standard formulation this partition function calculates the euler characteristic @xmath59 of the lattice gauge group which vanishes @xcite , @xmath60 neuberger s @xmath32 problem of lattice brst arises because we have then inserted zero instead of unity ( according to the faddeev - popov prescription ) into the measure of lattice gauge theory . on a finite lattice ,
such a topological model is equivalent to a problem of supersymmetric quantum mechanics with witten index @xmath61 . unlike the case of primary interest in supersymmetric quantum mechanics , here we need a model with non - vanishing witten index to avoid the neuberger @xmath32 problem . then however , just as the supersymmetry of the corresponding quantum mechanical model
, such a lattice brst can not break .
in landau gauge , with gauge parameter @xmath62 , the neuberger zero , @xmath63 , arises from the perfect cancellation of gribov copies via the poincar - hopf theorem .
the gauge - fixing potential @xmath64 $ ] for a generic link configuration @xmath65 thereby plays the role of a morse potential for gauge transformations @xmath66 and the gribov copies are its critical points ( the global gauge transformations need to remain unfixed so that there are strictly speaking only @xmath67sites@xmath68 factors of @xmath69 in ( [ eulerc ] ) ) .
the morse inequalities then immediately imply that there are at least @xmath70 such copies in @xmath71 on the lattice , or @xmath72 in compact @xmath73 , and equally many with either sign of the faddeev - popov determinant ( _ i.e. _ , that of the hessian of @xmath64 $ ] ) .
the topological origin of the zero originally observed by neuberger in a certain parameter limit due to uncompensated grassmann ghost integrations in standard faddeev - popov theory @xcite becomes particularly evident in the ghost / anti - ghost symmetric curci - ferrari gauge with its quartic ghost self - interactions @xcite .
due to its riemannian geometry with symmetric connection and curvature tensor @xmath74 for @xmath71 , in this gauge the same parameter limit leads to computing the zero in ( [ eulerc ] ) from a product of independent gauss - bonnet integral expressions , @xmath75 for each site of the lattice .
this corresponds to the gauss - bonnet limit of the equivalent supersymmetric quantum mechanics model in which only constant paths contribute @xcite .
the indeterminate form of physical observables as a consequence of ( [ gaussbonnet ] ) is regulated by a curci - ferrari mass term .
while such a mass @xmath10 decontracts the double brst / anti - brst algebra , which is well - known to result in a loss of unitarity , observables can then be meaningfully defined in the limit @xmath76 via lhospital s rule @xcite .
the 0/0 problem due to the vanishing euler characteristic of @xmath71 is avoided when fixing the gauge only up to the maximal abelian subgroup @xmath77 because the euler characteristic of the coset manifold is non - zero .
the corresponding lattice brst has been explicitly constructed for @xmath30 @xcite , where the coset manifold is the 2-sphere and @xmath78 .
this indicates that the neuberger problem might be solved when that of compact @xmath73 is , where the same cancellation of lattice gribov copies arises because @xmath79 . a surprisingly simple solution to this problem is possible , however , by stereographically projecting the circle @xmath80 which can be achieved by a simple modification of the minimising potential @xcite .
the resulting potential is convex to the above and leads to a positive definite faddeev - popov operator for compact @xmath73 where there is thus no cancellation of gribov copies , but @xmath81 , for @xmath82 gribov copies . as compared to the standard lattice landau gauge for compact @xmath73
their number is furthermore exponentially reduced .
this is easily verified explicitly in low dimensional models .
while @xmath82 grows exponentially with the number of sites in the standard case as expected , the stereographically projected version has only @xmath83 copies on a periodic chain of length @xmath84 and @xmath85 on a @xmath86 lattice of size @xmath87 in coulomb gauge , for example , and in both cases their number is verified to be independent of the gauge orbit . the general proof of @xmath81 with stereographic projection which avoids the neuberger zero in compact @xmath73 @xcite follows from a simple example of a nicolai map @xcite . applying the same techniques to the maximal abelian subgroup @xmath77 , the generalisation to @xmath88 lattice gauge theories is possible when the odd - dimensional spheres @xmath89 , @xmath90 , of its parameter space
are stereographically projected to @xmath91 . in absence of the cancellation of the lattice artifact
gribov copies along the @xmath73 circles , the remaining cancellations between copies of either sign in @xmath71 , which will persist in the continuum limit , are then necessarily incomplete , however , because @xmath92 . for @xmath30
this program is straightforward .
one replaces the standard gauge - fixing potential @xmath64 $ ] of lattice landau gauge by @xmath93 $ ] , via gauge - transformed links @xmath94 , where @xmath95 = 4 \sum_{x,\mu}\left(1-{\mbox{\small $ { \displaystyle \frac{1}{2}}$ } } { \operatorname{tr}}u^g_{x\mu}\right ) \quad \mbox{and } \quad \label{modlg } \widetilde v_u[g ] = -8 \sum_{x,\mu } \ln\left(\frac{1}{2}+\frac{1}{4}{\operatorname{tr}}u^{g}_{x,\hat{\mu } } \right)\;.\ ] ] the standard and stereographically projected gauge fields on the lattice are defined as @xmath96 the gauge - fixing conditions @xmath97 and @xmath98 are their respective lattice divergences , in the language of lattice cohomology , @xmath99 and @xmath100 .
a particular advantage of the non - compact @xmath101 is that they allow to resolve the modified lattice landau gauge condition @xmath98 by hodge decomposition .
this provides a framework for gauge - fixed monte - carlo simulations which is currently being developed for the particularly simple case of @xmath30 in 2 dimensions . in the low - dimensional models mentioned above it can furthermore be verified explicitly that the corresponding topological gauge - fixing partition function is indeed given by @xmath102 as expected from @xmath103 .
the proof of this will be given elsewhere .
comparisons of the infrared behaviour of qcd green s functions as obtained from lattice landau gauge implementations based on minimisations of a gauge - fixing potential and from continuum studies based on brst symmetry have to be taken with a grain of salt .
evidence of the asymptotic conformal behaviour predicted by the latter is seen in the strong coupling limit of lattice landau gauge where such a behaviour can be observed at large lattice momenta @xmath104 . there
the strong coupling data is consistent with the predicted critical exponent and coupling from the functional approaches .
the deviations from scaling at @xmath52 are not finite - volume effects , but discretisation dependent and hint at a breakdown of brst symmetry arguments beyond perturbation theory in this approach .
non - perturbative lattice brst has been plagued by the neuberger @xmath32 problem , but its improved topological understanding provides ways to overcome this problem .
the most promising one at this point rests on stereographic projection to define gauge fields on the lattice together with a modified lattice landau gauge .
this new definition has the appealing feature that it will allow gauge - fixed monte - carlo simulations in close analogy to the continuum brst methods which it will thereby elevate to a non - perturbative level .
r. alkofer and l. von smekal , phys .
* 353 * , 281 ( 2001 ) .
l. von smekal , r. alkofer and a. hauck , phys .
rev . lett . * 79 * , 3591 ( 1997 ) .
l. von smekal , a. hauck and r. alkofer , annals phys .
* 267 * , 1 ( 1998 ) .
r. alkofer and l. von smekal , nucl .
phys . a * 680 * , 133 ( 2000 ) .
lerche and l. von smekal , phys .
d * 65 * , 125006 ( 2002 ) .
d. zwanziger , phys .
d * 65 * , 094039 ( 2002 ) .
j. m. pawlowski , d. f. litim , s. nedelko and l. von smekal , phys .
lett . * 93 * , 152002 ( 2004 ) ; aip conf .
proc . *
756 * , 278 ( 2005 ) .
r. alkofer , c. s. fischer and f. j. llanes - estrada , phys .
b * 611 * , 279 ( 2005 ) .
j. c. taylor , nucl .
b * 33 * , 436 ( 1971 ) . c. s. fischer and j. m. pawlowski , phys .
d * 75 * , 025012 ( 2007 ) . c. s. fischer , a. maas and j. m. pawlowski , arxiv:0810.1987 [ hep - ph ] .
n. nakanishi and i. ojima , _ covariant operator formalism of gauge theories and quantum gravity _ ,
27 of lecture notes in physics , world scientific ( 1990 ) . c. s. fischer , a. maas , j. m. pawlowski and l. von smekal , annals phys .
* 322 * , 2916 ( 2007 ) ; pos * lat2007 * , 300 ( 2007 ) . a. sternbeck , l. von smekal , d. b. leinweber and a. g. williams , pos * lat2007 * , 340 ( 2007 ) .
a. cucchieri and t. mendes , pos * lat2007 * , 297 ( 2007 ) .
h. neuberger , phys .
b * 175 * , 69 ( 1986 ) ; _ ibid .
_ * 183 * , 337 ( 1987 ) .
a. sternbeck and l. von smekal , arxiv:0811.4300 [ hep - lat ] .
a. sternbeck and l. von smekal , pos * lattice2008 * , 267 ( 2008 ) . m. schaden , phys . rev . * d59 * , 014508 ( 1998 ) . l. von smekal , m. ghiotti and a. g. williams , phys .
d * 78 * , 085016 ( 2008 ) .
d. birmingham , m. blau , m. rakowski , g. thompson , phys .
rept . , 129 ( 1991 ) .
l. von smekal , d. mehta , a. sternbeck , a. g. williams , pos * lat2007 * , 382 ( 2007 ) . | the infrared behaviour of qcd green s functions in landau gauge has been focus of intense study .
different non - perturbative approaches lead to a prediction in line with the conditions for confinement in local quantum field theory as spelled out in the kugo - ojima criterion .
detailed comparisons with lattice studies have revealed small but significant differences , however . but are nt we comparing apples with oranges when contrasting lattice landau gauge simulations with these continuum results ?
the answer is yes , and we need to change that .
we therefore propose a reformulation of landau gauge on the lattice which will allow us to perform gauge - fixed monte - carlo simulations matching the continuum methods of local field theory which will thereby be elevated to a truly non - perturbative level at the same time . *
landau gauge qcd : functional methods versus lattice simulations international conference on _ selected problems of modern theoretical physics _ ( spmtp08 ) , bogoliubov laboratory of theoretical physics , dubna , russia , 23 27 june , 2008 . ] * lorenz von smekal + _ centre for the subatomic structure of matter , school of chemistry and physics , + the university of adelaide , sa 5005 , australia _ |
a prompt , 9-th magnitude , optical flash has recently been detected ( @xcite ) accompanying the gamma ray burst ( grb ) grb990123 .
the most natural explanation of this flash is emission from a reverse shock propagating into fireball ejecta shortly after it interacts with surrounding gas ( @xcite ) .
although optical - uv emission from the reverse shock accompanying , or following shortly after , the @xmath0-ray emission has been predicted ( @xcite ) based on the simplest fireball models for gamma - ray bursts ( grbs ) , the intensity of optical emission could not have been reliably predicted due to the uncertainty in reverse shock parameters .
the observations of grb990123 suggest that the electron and magnetic field energy fractions in the reverse shock are similar to those in the forward shock propagating into the surrounding gas and producing the long term afterglow .
this in turn implies that strong optical flashes accompanying @xmath0-ray emission is a generic grb characteristic .
this is consistent with previous non - detection of optical flashes ( @xcite ) given the wide grb luminosity function ( @xcite ) and the fact that grb990123 is in the top 0.3% of the batse brightness distribution ( @xcite ) .
there is evidence in several grb afterglows for significant dust extinction , which may imply an association of grbs with star forming regions ( @xcite ) .
it has been shown that if grbs indeed reside in such environment , the ionizing x - ray and uv afterglow radiation may lead to time dependent ( on hour time scale ) absorption ( @xcite ) and emission ( @xcite ) line features . if h@xmath11 is present near the grb , conspicuous 11101650absorption will be imprinted on the spectrum , followed by uv fluorescence ( @xcite ) . on longer time scales , up to @xmath12 yr , grb photo - ionization may lead to indicative recombination line features , which may allow identification of grb remnants in nearby galaxies ( @xcite ) . here , we discuss dust sublimation by the optical - uv flash accompanying the grb . since grb990123 was an exceptionally bright burst , with exceptionally high intrinsic @xmath0-ray luminosity , we first discuss in 2 the optical - uv flash itself to obtain the scaling of optical - uv luminosity with burst energy , and to estimate the prompt optical - uv emission at energies above the @xmath13 ev energy of the prompt emission observed by akerlof et al .
we show that if electron and magnetic field energy fractions in the reverse shock are similar among different bursts , then an optical - uv flash of luminosity @xmath14 , where @xmath2 is the beaming solid angle , is expected for a typical grb .
dust destruction by thermal sublimation is discussed in 3 , and other possible destruction mechanism are considered in 4 .
it should be emphasized that the analysis of the physics of dust destruction is independent of the model for optical - uv emission , and therefore of the discussion presented in 2 .
however , we present numerical results for dust destruction based on the characteristic optical - uv luminosity derived in 2 . in 5 we discuss dust infra - red emission .
the implications of our results are discussed in 6 .
in fireball models of grbs ( see @xcite for a recent review ) , the energy released by an explosion is converted to kinetic energy of a thin baryonic shell expanding at an ultra - relativistic speed . after producing the grb , the shell impacts on surrounding gas , driving an ultra - relativistic shock into the ambient medium .
after a short transition phase , the expanding blast wave approaches a self - similar behavior ( @xcite ) , where the expansion lorentz factor decreases with radius as @xmath15 .
the initial interaction of fireball ejecta with surrounding gas produces a reverse shock which propagates into and decelerates the fireball ejecta .
transition to self - similar behavior occurs on a time scale comparable to the reverse shock crossing time of the ejecta .
the long term afterglow is produced by the forward , expanding shock that propagates into the surrounding gas
. this shock continuously heats fresh gas and accelerates relativistic electrons , which produce the observed radiation through synchrotron emission .
the most natural explanation of the optical flash is that it is due to synchrotron emission of electrons which are accelerated by the reverse shock , during the transition to self - similar behavior .
once the reverse shock crosses the ejecta , the ejecta expand and cool adiabatically .
thus , emission from the fireball ejecta is suppressed after the transition to self - similar expansion .
since the optical flash is produced when the reverse shock crosses the ejecta , the plasma emitting the radiation expands with a lorentz factor which is close to that given by the blandford - mckee self - similar solution , @xmath16 , where @xmath17 is the fireball energy and @xmath18 is the surrounding gas number density .
the characteristic time over which radiation emitted by the fireball at radius @xmath19 is observed by a distant observer is @xmath20 ( @xcite ) .
the plasma lorentz factor during optical flash emission @xmath21 where @xmath22 erg , @xmath23 s is the observed duration , @xmath24 , and @xmath25 .
transition to self - similar expansion occurs on time scale @xmath26 comparable to the longer of the two time scales set by the initial conditions : the ( observed ) grb duration @xmath27 and the ( observed ) time @xmath28 at which the self - similar lorentz factor equals the original ejecta lorentz factor @xmath29 , @xmath30 .
that is , @xmath31 .
\label{delta_t}\ ] ] during the transition , the unshocked fireball ejecta propagate at the original expansion lorentz factor , @xmath32 , and the lorentz factor of plasma shocked by the reverse shock in the rest frame of the unshocked ejecta is @xmath33 . if @xmath34 then @xmath35 , the reverse shock is relativistic , and the lorentz factor associated with the random motion of protons in the reverse shock is @xmath36 .
if @xmath37 then @xmath38 , and the reverse shock is not relativistic .
nevertheless , the following argument suggests that the reverse shock speed is not far below @xmath39 , and that the protons are therefore heated to relativistic energy , @xmath40 .
the comoving time , measured in the fireball ejecta frame prior to deceleration , is @xmath41 .
the expansion lorentz factor is expected to vary across the ejecta , @xmath42 .
such variation would lead to expansion of the ejecta , in the comoving frame , at relativistic speed .
thus , at the deceleration radius , @xmath43 , the ejecta width exceeds @xmath44 .
since the reverse shock should cross the ejecta over a deceleration time scale , @xmath45 , the reverse shock speed must be close to @xmath39 .
we therefore conclude that the lorentz factor associated with the random motion of protons in the reverse shock is approximately given by @xmath46 for both @xmath38 and @xmath35 .
for protons shocked by the forward shock @xmath47 , and therefore the ratio between thermal energy per proton in the reverse and forward shocks is @xmath48 .
below we use this relation to derive the emission characteristics of the reverse shock by scaling the exact analytic results given for the forward shock emission in @xcite .
if the fraction of thermal energy carried by electrons , @xmath49 , and magnetic field , @xmath50 , is similar in the forward and reverse shocks , then the frequency of peak synchrotron emission from the reverse shock is smaller than that of the forward shock by a factor @xmath51 .
this is due to the fact that the energy density behind the reverse and forward shocks are similar , so that similar @xmath50 implies similar magnetic field strength in both regions , while similar @xmath49 implies @xmath52 . using eq .
( 10 ) of @xcite for the forward shock peak frequency , we find that the reverse shock emission peaks at a frequency @xmath53 ( measured at the grb redshift ) . here
@xmath49 and @xmath50 are the values relevant for the reverse shock , which in general may differ from those of the forward shock .
the numerical values we have used are those characteristic of the forward shock ( @xcite ) . as demonstrated below , the cooling time of electrons in the reverse shock , radiating in the optical - uv range ,
is long compared to the fireball expansion time . in this case , the peak synchrotron intensity is proportional to the product of magnetic field strength and number of radiating electrons ( and independent of the electron lorentz factor ) .
the number of radiating electrons in the reverse shock is larger than in the forward shock by a factor @xmath54 .
this can be deduced from the following considerations .
the proton random lorentz factor in the reverse shock is @xmath54 times smaller than that in the forward shock .
since the energy density in both regions is similar , the density of protons , and therefore of electrons , in the reverse shock is higher than that in the forward shock by the same factor . in addition , the width of the shocked fireball ejecta is similar to that of the shell of shocked surrounding gas , since both shocks propagate relativistically in the shocked plasma frame .
thus , if @xmath50 is similar in the reverse and forward shock , the peak synchrotron intensity @xmath55 is higher in the reverse shock by a factor @xmath54 . using eq .
( 11 ) of @xcite we find for the reverse shock @xmath56 for a flat universe with zero cosmological constant , and @xmath57 .
the decay of grb990123 optical flux implies an electron energy distribution @xmath58 with @xmath59 ( @xcite ) , for which the intensity at @xmath60 is @xmath61 .
thus , the optical [ @xmath62 hz ] intensity is @xmath63 for the parameters of grb990123 ,
@xmath64 s and @xmath65 erg ( based on the @xmath0-ray fluence @xmath66 ) , we obtain @xmath67 jy , approximately twice the peak observed flux .
thus , the observed optical flash can be naturally explained by the simplest fireball model , provided the reverse shock parameters @xmath49 and @xmath50 are similar to those implied for the forward shock by afterglow observations , @xmath68 and @xmath69 .
if this is typical , than for a typical grb , with @xmath70 erg and @xmath71 s , we find @xmath72 where @xmath73 is the frequency at the redshift of the grb .
the form ( [ l_nu ] ) is valid for frequencies @xmath73 larger than the peak frequency @xmath74 , and smaller than the frequency @xmath75 above which emission is dominated by electrons for which the cooling time is shorter than the dynamic time .
since the energy density in the reverse and forward shock regions is similar , @xmath75 ( at the grb redshift ) is given by @xmath76 at frequencies @xmath77 the spectrum steepens to @xmath78 .
strong optical - uv emission requires , similar to grb @xmath0-ray production , large initial lorentz factor , @xmath79 : eq .
( [ delta_t ] ) implies that @xmath80 for @xmath81 , and the optical - uv luminosity given by eq .
( [ l_nu ] ) drops as @xmath82 .
this implies that the plasma emitting the optical - uv photons must also be expanding at high lorentz factor , @xmath83 [ see eqs .
( [ delta_t],[gamma ] ) ] , and strong optical - uv emission may be confined , similar to @xmath0-rays , to a cone around the line of sight of opening angle @xmath84 .
this may be the case if , e.g. , the fireball is a jet of opening angle @xmath85 .
a jet of finite opening angle expands as if it were a conical section of a spherical fireball , as long as @xmath86 .
thus , the analysis presented above is valid for a jet - like fireball . in this case
, the energy @xmath17 in the above equations should be understood as the energy that the fireball would have carried if it were spherically symmetric , and the optical - uv , as well as @xmath0-ray , emission is confined to a small solid angle @xmath87 with optical - uv luminosity given by @xmath88 . it should be noted that prompt optical emission has been observed for only 1 grb to date , and therefore may not be typical of grbs .
indeed , fireballs in low density environments with @xmath89 would not be expected to produce strong prompt emission [ see eq .
( [ eq : f_v ] ) ] . however , we note from eq .
( [ eq : f_v ] ) that optical flashes are expected for typical grbs with @xmath90 and @xmath91 , and should not be limited to unusually luminous grbs such as grb990123 .
as discussed in 3.3 below , in dense regions radiation with @xmath92 ev will largely go into photoionizing h and h@xmath11 , and photons in the 7.513.6 ev range will primarily be absorbed by h@xmath11 , which is rovibrationally excited by ultraviolet pumping ( draine 1999b ) .
we therefore first discuss , in 3.1 and 3.2 , sublimation of dust grains under the assumption that dust heating is dominated by @xmath93 ev photons , and discuss the contribution of 7.550 ev photons to dust sublimation in 3.3 .
using equations ( [ l_nu ] ) and ( [ nu_c ] ) , we find that for typical grb parameters , the prompt luminosity in the 17.5 ev range is @xmath94 and the 7.550 ev prompt luminosity is @xmath95 .
since our simple analysis overestimates the flux of grb990123 by a factor @xmath96 , we will take the typical 17.5 ev luminosity to be @xmath97 , and the 7.550 ev luminosity to be @xmath98 .
consider a grain of radius @xmath99 located at a distance @xmath19 from a transient source of radiation radiating a 17.5 ev power @xmath100 into a solid angle @xmath2 .
if the radiation from the grb is `` beamed '' into @xmath101 , we will consider only dust grains within the beam .
note that we expect the beaming of optical - uv emission from the forward shock to be similar to that of gamma - ray emission .
the grain temperature @xmath102 is determined by @xmath103 where @xmath104 is the effective optical depth for attenuation of the optical - uv flash , @xmath105 is the density of the grain material , @xmath106 is the mean atomic mass , @xmath107 is the chemical binding energy per atom , @xmath108 is the usual planck - averaged absorption efficiency , and @xmath109 is the absorption efficiency factor averaged over the 17.5 ev spectrum of the optical - uv flash . for the grain radii @xmath110 expected in dense clouds , we will assume @xmath111 since we are interested in energy depositions large enough to sublime grains , the heat capacity of the grain has been neglected in equation ( [ eq : balance ] ) .
the sublimation rate can be approximated by @xmath112 guhathakurta & draine ( 1989 ) have estimated @xmath113 , @xmath114 for mg@xmath11sio@xmath115 , and @xmath116 , @xmath117 for graphite .
we adopt @xmath118 , @xmath119 , and @xmath120 as representative values for refractory grains .
if we assume the grain temperature @xmath102 is approximately constant over the time @xmath121 , then the condition for the grain to be completely sublimed during this time would be @xmath122 , where @xmath123 } \approx 2300 { \rm k } \left [ 1 + 0.033\ln\left ( \frac{a_{-5 } } { \delta t_{1 } } \right ) \right ] \label{eq : t_c}\ ] ] where @xmath124 .
equivalently , the grain survival time at temperature @xmath102 is @xmath125{\rm\ , s } \quad .
\label{eq : t_sur}\ ] ] the infrared emissivity is quite different for graphite and silicate materials ( draine & lee 1984 ; draine 1999a ) .
for the temperature range of interest for dust sublimation , @xmath126 , we approximate @xmath127 with @xmath128 and 0.3 for astronomical silicate and graphite , respectively .
we adopt equation ( [ eq : q_t ] ) with @xmath129 as representative of refractory grain material . for @xmath132 , so that sublimation cooling dominates over radiative cooling , the grain temperature is @xmath133\quad , \label{eq : t_sub}\ ] ] where @xmath134 . for @xmath135 radiative
cooling dominates and the grain temperature is @xmath136 where @xmath137 is the instantaneous attenuation of the 17.5 ev flash by intervening absorption .
if attenuation by intervening material can be neglected , grains are completely sublimed out to a destruction radius @xmath138 where @xmath139 is the radius where the unattenuated flash can heat grains to the critical temperature @xmath140 . for our nominal parameters , eq .
( [ eq : t_c ] ) and ( [ eq : t_r = s ] ) give @xmath141 , so that radiative cooling dominates at @xmath142 , and the critical distance @xmath139 is given by @xmath143 in the optically - thin limit , then , the optical - uv flash from the grb will destroy dust _ in the beam _ out to a substantial distance . while the 17.5 ev photons from a grb may be capable of destroying dust out to @xmath144 pc in the optically - thin limit , in dusty regions ( such as molecular clouds ) attenuation of the radiation by dust grains before they are destroyed will limit the grain destruction to a smaller radius .
the dust optical depth is a function @xmath145 of both space and time , as the grb flash `` burns '' its way through the cloud .
the attenuation of @xmath517.5 ev photons is dominated by dust . to estimate the effect of high optical depth
, we will assume that only a single dust type is present .
let @xmath146 be the dust destruction radius : all grains at @xmath147 are destroyed by the heating effects of the optical - uv flash .
we approximate the 17.5 ev emission from the grb as a rectangular pulse . at radii @xmath148
, the _ leading _ edge of the optical - uv pulse is attenuated by the dusty medium through which it propagates , but the _ trailing _ edge of the pulse is unattenuated since it propagates through a dustless medium , and we are neglecting the effects of gas - phase absorption . rather than solve for @xmath145 , we will simplify the problem by assuming that the effects of extinction can be approximated as primarily a _ narrowing _ of the optical pulse , retaining a rectangular profile .
the problem then reduces to determination of a function @xmath149 , the fraction of the flash energy which is absorbed by dust interior to radius @xmath19 , and survival of grains at radius @xmath19 when irradiated by a radiation field @xmath150 for a time @xmath151 .
the function @xmath149 then satisfies @xmath152\over \delta t}\ ] ] where @xmath153 is the survival time of a grain irradiated by the unattenuated radiation field . in the unattenuated portion of the optical pulse
, grains are heated to temperatures given by equations ( [ eq : t_sub],[eq : t_rad ] ) with @xmath154 .
the dust destruction radius @xmath146 is then determined by the condition @xmath155\delta t$ ] .
figure [ fig : r_d ] shows @xmath146 as a function of cloud density @xmath156 for a standard dust - to - gas ratio @xmath157 , for several different values of the grb 17.5 ev luminosity ( @xmath158 ) , duration ( @xmath159 ) , and characteristic dust grain radius ( @xmath160 ) .
the radius @xmath161 for `` typical '' grb parameters is shown as the heavy curve .
we see that if a grb occurred in a dusty region , the optical - uv flash from the grb can clear out a substantial amount of dust which lies in the beam .
radiation with @xmath92 ev may ionize h and h@xmath11 . neglecting dust opacity ,
the prompt flash will be able to photoionize a mass @xmath162 or out to a radius @xmath163 where @xmath164 , and we approximate the flash by a rectangular pulse of duration @xmath165 . we have taken the typical 13.650 ev luminosity to be @xmath166 . the dust destruction radius , @xmath161 , is compared with @xmath167 in figure [ fig : r_d ] . at densities for which @xmath168 , typically @xmath169 ( characteristic of star - forming regions ) , our neglect of absorption of @xmath170 photons by dust in estimating @xmath167 is justified , because the dust will be destroyed by absorption of @xmath171 photons prior to arrival of most of the ionizing photons at a given point in the cloud . at these densities , @xmath172 photons
are fully absorbed by the gas in a region smaller than the dust destruction zone , thus justifying our neglect of ionizing radiation when estimating @xmath161 .
since the number of 7.513.6 ev photons is comparable to the number of @xmath170 photons , this also shows that , for densities @xmath173 , the 7.513.6 ev photons will be mainly absorbed by h@xmath11 , and can be neglected for purposes of dust destruction . at lower densities , where @xmath174 , some fraction of @xmath175 ev radiation would also contribute to dust sublimation . at these densities the grain temperature near @xmath161
is determined by radiative cooling and therefore @xmath176 , where @xmath177 is the uv luminosity available for dust heating . since @xmath178 , the contribution of @xmath175 ev radiation may increase @xmath161 by up to @xmath179 .
because of the large fluence of energetic photons , dust destruction could , in principle , also result from extreme ionization of the dust grain .
high degrees of ionization could result in fission of the grain , or emission of individual ions from the grain surface by the process known as `` ion field emission '' .
for an approximately spherical grain of radius @xmath99 , charged to a potential @xmath180 , the tensile stress averaged over a cross section @xmath181 is @xmath182 .
if the maximum tensile stress which the grain material can support is @xmath183 , then the potential gradient and grain charge @xmath184 are limited by @xmath185 @xmath186 we note that if a grain with @xmath187 fissioned into two halves , each of the fragments would have @xmath188 and there would therefore be no further fragmentation unless additional ionization took place .
a grain contains @xmath189 electrons .
if the mean atomic number is @xmath190 and the photoionization cross section per electron is @xmath191 , then substantial grain destruction by this process would require a fluence @xmath192 .
grain fission would therefore occur within a radius @xmath193 where we have taken the fluence @xmath194 .
note that this process changes the grain size distribution and therefore affects the optical extinction curve , but grain fission alone would not appreciably reduce the ultraviolet extinction , and might even increase it .
ideal materials have tensile strengths @xmath195 , so that a coulomb explosion would not take place until the surface electric field reaches @xmath196 .
however , in the laboratory electric fields exceeding @xmath197 are observed to result in `` ion field emission '' , where individual ions are emitted from the sample ( muller & tsong 1969 ) . as a result , if the grain tensile strength @xmath198 , intense irradiation by @xmath199 photons will first cause the grain to charge up to @xmath200 , and each subsequent ionization will result in emission of an ion ( assuming that electron capture is negligible during the @xmath201 s of the gamma ray burst ) .
if the mean atomic number is @xmath190 and the photoionization cross section per electron is @xmath191 , then substantial grain destruction by this process would require a fluence @xmath202 .
grain destruction by ion field emission would therefore occur only within a radius @xmath203 ion field emission is evidently much less important than sublimation .
for clouds of mass @xmath204 extending to @xmath205 pc , most of the optical - uv energy absorbed by sublimated dust is re - radiated in the infra - red , typically around @xmath206 m .
this is due to the fact that at distances larger than @xmath207 pc radiative cooling of dust grains dominates over sublimation cooling [ see eqs .
( [ eq : t_rad],[eq : t_r = s ] ) ] , and to the fact that for @xmath204 grains are heated to the temperature @xmath208 k required for complete sublimation out to a distance @xmath209 pc [ see eq . ( [ eq : t_c ] ) , fig
. 1 ] ( for @xmath210 , optical photons are completely absorbed at distances @xmath211 pc , where sublimation cooling dominates and only a small fraction of the absorbed energy is re - radiated ) . at radii where radiation cooling dominates , the grain temperature drops approximately as @xmath212 .
the strong dependence of grain survival time @xmath213 on temperature , eq .
( [ eq : t_sur ] ) , then implies a sharp transition , i.e. over a distance @xmath214 , between the region at @xmath148 where @xmath213 is much smaller than the flash duration @xmath215 , to the region at @xmath216 , where @xmath217 .
since the energy radiated by grains is proportional to @xmath218 , infra - red emission of sublimated dust is dominated by emission from grains just outside @xmath146 .
if the optical depth for uv photons due to dust at @xmath219 is @xmath220 , then most of the flash energy would be absorbed by dust and re - radiated in the infra - red .
most of the infra - red radiation would escape the cloud , and may therefore be detected , if the infra - red optical depth is not high , @xmath221 . for @xmath222 ,
the requirements @xmath220 and @xmath221 may be written as @xmath7 . in order to estimate the dust infra - red luminosity in the case where @xmath221 and @xmath220 , let us first assume that optical - uv flash emission is beamed into a small solid angle around the line of sight , @xmath223 .
the observed duration of the infra - red emission is then @xmath224 , where @xmath225 pc is the radius out to which grains are heated to @xmath226 k. this may be written in the form @xmath227 day , which is valid for @xmath228 as well as for @xmath229 .
the infra - red luminosity is given by the ratio of the flash energy absorbed by dust , @xmath230 erg , and the observed duration @xmath231 , @xmath232 .
the luminosity of the prompt optical - uv emission accompanying grb @xmath0-ray emission is given by eqs .
( [ l_nu ] ) and ( [ delta_t ] ) . for typical grb parameters
we expect an optical - uv flash with 17.5 ev luminosity @xmath233 , assuming isotropic emission
. such a uv flash can destroy dust by sublimation out to an appreciable distance , @xmath209 pc ( see figure [ fig : r_d ] ) , and may clear the dust out of @xmath234 of molecular cloud material , where @xmath2 is the solid angle into which the optical - uv emission is beamed , and where dust is sublimed . if grb sources indeed lie in dusty regions , then the extinction would decrease with time during prompt optical - uv emission , over tens of seconds .
detection of such time dependent extinction would provide strong constraints on the grb environment .
the destruction of dust implies that existing , or future , observations of not - heavily - reddened fireballs are not inconsistent with grbs being associated with star formation .
we have shown in 5 that if the optical depth due to dust beyond @xmath6 pc is of order unity , most of the uv flash energy is absorbed and re - radiated in the infra - red , typically at @xmath8 m .
the resulting infra - red luminosity , @xmath9 , extends over an apparent duration of @xmath10 day . for grbs at @xmath235 , therefore , k - band photometry may reveal thermal emission from dust grains .
in fact , such emission may already have been observed in grb970228 ( @xcite ) and grb980326 ( @xcite ) . in both cases ,
a deviation from a power law decline of optical flux , which at early time is consistent with synchrotron emission from shock accelerated electrons , is observed at @xmath236 d delay . as the flux drops below @xmath237jy ,
a new infra - red emission component is revealed , with a flux @xmath238jy between @xmath239 m and @xmath206 m for grb970228 and @xmath240jy at @xmath241 m for grb980326 . in both cases ,
the spectrum is modified at this time to @xmath242 at @xmath243 m .
the infra - red flux is of the same order of magnitude estimated for dust grain emission , and the spectrum is consistent with dust emission peaking at @xmath244 m ( at the source redshift ) , provided grb980326 is at redshift @xmath245 .
we note that for the typical parameters adopted in this paper , dust emission is expected to peak at somewhat longer wavelength , @xmath246 m .
however , since the grain properties are not well known , dust emission can not be ruled out as an alternative to the proposal that the `` excess '' emission is due to a supernova ( @xcite ) we note also that the non - detection of optical emission from grb980326 at @xmath247 d , implies , under the dust emission hypothesis , beaming of the optical - uv flash to @xmath248 , consistent with the interpretation that the rapid , @xmath249 , flux decline is due to the fireball being a jet of small opening angle ( @xcite ) .
we have shown in 2 that strong optical - uv emission requires , like grb @xmath0-ray production , large initial expansion lorentz factor , @xmath79 , which also implies that the plasma emitting the optical - uv flash must be expanding with @xmath83 .
thus , if the fireball is a jet of finite opening angle , @xmath250 , then both @xmath0-ray and optical - uv emission will be confined to a small solid angle @xmath87 . in this case , dust would be evaporated only within a narrow cone around the line of sight .
a jet - like fireball expands as a conical section of a spherical fireball , with @xmath251 , as long as @xmath252 .
after deceleration to @xmath253 , the jet expands sideways , its opening angle increasing to @xmath254 and @xmath255 ( @xcite ) . at this stage radiation reaching us must travel a distance @xmath256 through gas which was not exposed to the initial flash , and which will still contain dust . for @xmath257 , and using @xmath258 , we have @xmath259 .
thus , a significant increase in extinction over time would be observed if @xmath260 .
confinement of @xmath0-ray and strong optical - uv emission to a small solid angle is not limited to the case of a jet - like fireball .
it may also arise if @xmath29 , the initial expansion lorentz factor , is anisotropic .
consider a fireball carrying similar energy per unit solid angle in all directions , with @xmath29 a decreasing function of angle with respect to the line of sight , such that @xmath261 for @xmath262 . in this case ,
optical - uv ( and @xmath0-ray ) emission would be suppressed at angles @xmath262 .
the isotropic fireball energy per unit solid angle implies that , after a transition phase , the fireball would approach spherical expansion , with @xmath251 . at this stage ,
most of the radiation detected at time @xmath263 by a distant observer is produced by fireball plasma within a narrow ring of radius @xmath264 around the line of sight , where the fireball radius @xmath19 and expansion lorentz factor @xmath265 are related to @xmath263 through @xmath266 ( @xcite ) .
thus , here too a significant increase in reddening is expected once @xmath267 drops below @xmath268 . for
@xmath269 most of the radiation reaching us must pass through a path length @xmath270 of gas which was not exposed to the initial flash .
finally , it should be noted that although optical flashes are expected for typical grbs [ with @xmath90 and @xmath271 , see eq .
( [ eq : f_v ] ) ] , prompt optical emission has been observed for only 1 grb to date and therefore may not accompany all grbs .
indeed , fireballs in low density environments with @xmath89 would not be expected to produce strong prompt emission [ see eq .
( [ eq : f_v ] ) ] .
in addition , if the fireball initial lorentz factor @xmath272 , reverse shock emission may be shifted to photon energies above 7.5 ev [ see eq .
( [ nu_m ] ) ] , where most photons are absorbed by h and h@xmath11 .
akerlof , c. w. , _ et al .
_ 1999 , gcn 205 blandford , r. d. , & mckee , c. f. 1976 , phys .
fluids 19 , 1130 bloom , j. s. , _ et al .
_ 1999 , astro - ph/9905301 bttcher , m. _ et al .
_ 1998 , astro - ph/9809156 djorgovski , s. g. , _ et al .
_ 1999 , gcn circular no .
289 draine , b.t .
1999a , http://www.astro.princeton.edu/@xmath5draine/dust/dust.diel.html draine , b.t .
1999b , submitted to apj ( astro - ph/99997232 ) draine , b.t . ,
& lee , h .-
1984 , , 285 , 89 fruchter , a. s. , _ et al .
_ 1999 , , 516 , 683 galama , t. j. , _ et al .
_ 1999 , astro - ph/9907264 ghisellini , g. _ et al .
_ 1998 , astro - ph/9808156 granot , j , piran , t. & sari , r. 1998 , astro - ph/9808007 gruzinov , a. & waxman , e. 1999 , ap .
j. 511 , 852 guhathakurta , p. , & draine , b.t .
1989 , , 345 , 230 hogg , d. w. & fruchter , a. s. , submitted to ap . j. ( astro - ph/9807262 ) .
kouveliotu , c. _ et al .
_ 1999 , nature , submitted krumholtz , m. , thorsett , s. e. , & harrison , f. a. , ap .
j. * 506 * , l81 ( 1998 ) .
kulkarni , s. r. , _ et al .
_ 1998 , nature 395 , 35 loeb , a. & perna , r. 1998 , , 503 , l35 mau , s. & mo , h. j. 1998 , a&a 339 , l1 mszros , p. & rees , m. 1997 , ap .
j. 476 , 232 mszros , p. & rees , m. 1999 , submitted to mnras , ( astro - ph/9902367 ) mszros , p. , rees , m. & papathanassiou , h. 1994 , , 432 , 181 metzger , m. _ et al .
_ 1997 , iauc 6676 muller , e.w . , & tsong , t.t .
1969 , field ion microscopy ( new york : american elsevier ) paczyski , b. 1998 , ap .
j. 494 , l45 panaitescu , a. & mszros , p. 1998
, , 501 , 772 park , h. s. _ et al .
_ 1997 , , 490 , l21 perna , r. , & loeb , a. 1998 , , 501,467 perna , r. , raymond , j. , & loeb , a. 1999 , astro - ph/9904181 piran , t. 1998 , to appear in phys .
reports ( atro - ph/9810256 ) reichart , d. e. 1999 , astro - ph/9906079 rhoads , j. e. 1999 , submitted to apj ( astro - ph/9903399 ) sari , r. , & piran , t. 1999a , submitted to apj ( astro - ph/9901338 ) sari , r. , & piran , t. 1999b , submitted to apj ( astro - ph/9902009 ) waxman , e. 1997a , ap .
j. 485 , l5 waxman , e. 1997b , ap .
j. 489 , l33 waxman , e. 1997c , ap .
j. 491 , l19 wijers , r. a. m. j. , & galama , t. j. 1998 , astro - ph/98050341 | the prompt optical flash recently detected accompanying grb990123 suggests that , for at least some grbs , @xmath0-ray emission is accompanied by prompt optical - uv emission with luminosity @xmath1 , where @xmath2 is the solid angle into which @xmath0-ray and optical - uv emission is beamed .
such an optical - uv flash can destroy dust _ in the beam _ by sublimation out to an appreciable distance , @xmath3 pc , and may clear the dust out of as much as @xmath4 of molecular cloud material on an apparent time scale of @xmath5 ten seconds .
detection of time dependent extinction on this time scale would therefore provide strong constraints on the grb source environment .
dust destruction implies that existing , or future , observations of not - heavily - reddened fireballs are not inconsistent with grbs being associated with star forming regions . in this case , however , if @xmath0-ray emission is highly beamed , the expanding fireball would become reddened on a @xmath51 week time scale .
if the optical depth due to dust beyond @xmath6 pc from the grb is @xmath7 , most of the uv flash energy is converted to infra - red , @xmath8 m , radiation with luminosity @xmath9 extending over an apparent duration of @xmath10 day .
dust infra - red emission may already have been observed in grb970228 and grb980326 , and may possibly explain their unusual late time behavior .
# 1#2 epsfbox 16 you need to input epsf ; i ll do it for you epsf = # 2 1 # 1 1 16cant open # 1 1 |
the experimental data on kaons and other mesons containing strange quarks are both abundant and rather precise [ ] .
these particles are expected to play an important role in meson vacuum phenomenology and at non - zero temperatures , most notably in the restoration of the chiral symmetry that is spontaneously [ ] and explicitly [ ] broken in vacuum .
their structure - at least in the scalar sector - is ambiguous , as in the case of the non - strange mesons [ ] .
+ open questions regarding the strange mesons can be addressed , for example , within the linear sigma model [ ] , as has been done , e.g. , in ref .
[ ] . however , our approach is different in comparison to ref . [ ] in that our calculations are based on a globally chirally invariant @xmath1 lagrangian that contains the scalar and pseudoscalar but also vector and axial - vector degrees of freedom . in this paper
, we briefly outline our calculations and present some first results .
the paper is organized as follows : in sec . 2 we present the model lagrangian and its implications and in sec .
3 we summarize our results .
the lagrangian of the @xmath1 model with global chiral invariance has an analogous form as the corresponding linear sigma model lagrangian for @xmath2 [ ] : @xmath3-m_{0}^{2 } \mathrm{tr}(\phi^{\dagger}\phi)-\lambda_{1}[\mathrm{tr}(\phi^{\dagger}% \phi)]^{2 } -\lambda_{2}\mathrm{tr}(\phi^{\dagger}\phi)^{2}\nonumber\\ & -\frac{1}{4}\mathrm{tr}[(l^{\mu\nu})^{2}+(r^{\mu\nu})^{2}]+\frac{m_{1}^{2}% } { 2 } \mathrm{tr}[(l^{\mu})^{2}+(r^{\mu})^{2}]+\mathrm{tr}[h(\phi+\phi ^{\dagger})]\nonumber\\ & + c(\det\phi+\det\phi^{\dagger})-2ig_{2}(\mathrm{tr}\{l_{\mu\nu}[l^{\mu } , l^{\nu}]\ } + \mathrm{tr}\{r_{\mu\nu}[r^{\mu},r^{\nu}]\})\nonumber\\ & -2g_{3}\left [ \mathrm{tr}\left ( \left\ { \partial_{\mu}l_{\nu}+\partial_{\nu}l_{\mu } \right\ } \{l^{\mu},l^{\nu}\}\right ) + \mathrm{tr } \left ( \left\ { \partial_{\mu}% r_{\nu } + \partial_{\nu}r_{\mu } \right\ } \{r^{\mu},r^{\nu}\}\right ) \right ] \nonumber\\ & + \frac{h_{1}}{2}\mathrm{tr}(\phi^{\dagger}\phi)\mathrm{tr}[(l^{\mu})^{2 } + ( r^{\mu})^{2}]+h_{2}\mathrm{tr}[(\phi r^{\mu})^{2}+(l^{\mu}\phi)^{2 } ] + 2h_{3}\mathrm{tr}(\phi r_{\mu}\phi^{\dagger}l^{\mu } ) .
\label{lagrangian}%\end{aligned}\ ] ] where @xmath4{ccc}% \frac{(\sigma_{n}+a_{0}^{0})+i(\eta_{n}+\pi^{0})}{\sqrt{2 } } & a_{0}^{+}% + i\pi^{+ } & k_{s}^{+}+ik^{+}\\ a_{0}^{-}+i\pi^{- } & \frac{(\sigma_{n}-a_{0}^{0})+i(\eta_{n}-\pi^{0})}% { \sqrt{2 } } & k_{s}^{0}+ik^{0}\\ k_{s}^{-}+ik^{- } & { \bar{k}_{s}^{0}}+i{\bar{k}^{0 } } & \sigma_{s}+i\eta_{s}% \end{array } \right ) \label{phi}\ ] ] is a matrix containing the scalar and pseudoscalar degrees of freedom and @xmath5{ccc}% \frac{\omega_{n}^{\mu}+\rho^{\mu0}}{\sqrt{2}}+\frac{f_{1n}^{\mu}+a_{1}^{\mu0}% } { \sqrt{2 } } & \rho^{\mu+}+a_{1}^{\mu+ } & k_{\star}^{\mu+}+k_{1}^{\mu+}\\ \rho^{\mu-}+a_{1}^{\mu- } & \frac{\omega_{n}^{\mu}-\rho^{\mu0}}{\sqrt{2}}% + \frac{f_{1n}^{\mu}-a_{1}^{\mu0}}{\sqrt{2 } } & k_{\star}^{\mu0}+k_{1}^{\mu0}\\ k_{\star}^{\mu-}+k_{1}^{\mu- } & { \bar{k}}_{\star}^{\mu0}+{\bar{k}}_{1}^{\mu0 } & \omega_{s}^{\mu}+f_{1s}^{\mu}% \end{array } \right ) , \nonumber \\
r^{\mu } : = \frac{1}{\sqrt{2}}\left ( \begin{array } [ c]{ccc}% \frac{\omega_{n}^{\mu}+\rho^{\mu0}}{\sqrt{2}}-\frac{f_{1n}^{\mu}+a_{1}^{\mu0}% } { \sqrt{2 } } & \rho^{\mu+}-a_{1}^{\mu+ } & k_{\star}^{\mu+}-k_{1}^{\mu+}\\ \rho^{\mu-}-a_{1}^{\mu- } & \frac{\omega_{n}^{\mu}-\rho^{\mu0}}{\sqrt{2}}% -\frac{f_{1n}^{\mu}-a_{1}^{\mu0}}{\sqrt{2 } } & k_{\star}^{\mu0}-k_{1}^{\mu0}\\ k_{\star}^{\mu-}-k_{1}^{\mu- } & { \bar{k}}_{\star}^{\mu0}-{\bar{k}}_{1}^{\mu0 } & \omega_{s}^{\mu}-f_{1s}^{\mu}% \end{array } \right ) \label{lr}\end{aligned}\ ] ] are , respectively , the left - handed and right - handed matrices containing the vector and axial - vector degrees of freedom . furthermore , @xmath6 is the covariant derivative ; @xmath7 and @xmath8 are , respectively , the left - handed and right - handed field strength tensors ; the term tr@xmath9 $ ] [ @xmath10 , @xmath11 , @xmath12 explicitly breaks chiral symmetry due to nonzero quark masses , and the term @xmath13 describes the @xmath14 anomaly [ ] . + as in ref .
[ ] , in the non - strange sector , we assign the fields @xmath15 and @xmath16 to the pion and the @xmath17 counterpart of the @xmath18 meson , @xmath19 . the fields @xmath20 and @xmath21 represent the @xmath22 and @xmath23 vector mesons , respectively , and the fields @xmath24 and @xmath25 represent the @xmath26 and @xmath27 mesons , respectively . in the strange sector
, we assign the @xmath28 fields to the kaons ; the @xmath29 field is the strange contribution to the physical @xmath18 and @xmath30 fields ; the @xmath31 , @xmath32 , @xmath33 , and @xmath34 fields correspond to the @xmath35 , @xmath36 , @xmath37 , and @xmath38 mesons , respectively . in accordance with ref . [ ] , we assign the scalar kaon @xmath39 to the physical @xmath40 state .
the assignment of the non - strange scalar states depends on the results for their masses ( see next paragraph ) . + in order to implement spontaneous symmetry breaking
, we shift the @xmath41 and @xmath42 fields by their respective vacuum expectation values @xmath43 and @xmath44 , resulting in @xmath45-@xmath46 and @xmath47-@xmath48 mixing [ ] as well as in @xmath29-@xmath32 , @xmath39-@xmath33 , and @xmath28-@xmath34 mixing .
these are removed as described in ref .
we observe consequently the mixing between the @xmath41 ( @xmath49 , with @xmath50 standing for non - strange quarks ) and @xmath42 ( @xmath51 ) states , yielding two scalar states that are respectively predominantly of @xmath52 or @xmath53 nature [ note that the lagrangian ( [ lagrangian ] ) also yields @xmath45 - @xmath29 mixing ] . in this paper
, it suffices to say that our preliminary results , in the limit where the large-@xmath54 suppressed parameter @xmath55 is set to zero , show the existence of a predominantly @xmath52 state with a mass between 1.0 and 1.3 gev and of a predominantly @xmath53 state with mass between 1.3 and 1.6 gev , depending on the choice of parameters .
our work concerning these ( and other ) issues is still ongoing and will allow us to determine the parameters in the model more exactly so that the masses of the two mentioned scalars will become more precisely determined [ ] .
note that our @xmath1 model possesses no free parameters due to the preciseness of the experimental data [ ] .
we have presented an @xmath1 linear sigma model with ( axial-)vector degrees of freedom and global chiral invariance .
our preliminary results suggest that the notion of the scalar states above 1 gev as quarkonia may hold also when the @xmath2 lagrangian of ref .
[ ] is extended to three flavors .
clearly , the parameters in our model still need to be calculated precisely and , regarding the scalar sector , one needs to include the tetraquark and glueball fields in order to study their mixing with the scalar quarkonia .
0 j. goldstone , nuovo cim . * 19 * , 154 ( 1961 ) ; j. goldstone , a. salam and s. weinberg , phys .
rev . * 127 * , 965 ( 1962 ) .
g. t hooft , phys .
rept . * 142 * , 357 ( 1986 ) .
d. parganlija , f. giacosa and d. h. rischke , arxiv:1003.4934 [ hep - ph ] . | we outline the extension of the globally chirally invariant @xmath0 linear sigma model with vector and axial - vector degrees of freedom to @xmath1 .
we present preliminary results concerning the scalar meson masses . |
observational cosmology has come a long way since the early days of being `` the science of two numbers '' .
the revolution could be traced back to the discovery of the cosmic microwave background ( cmb ) in 1964 by penzias and wilson @xcite , which won them the nobel prize in 1978 , and confirmed an earlier prediction of george gamow @xcite .
gamow had realized that the observed abundance of light nuclei can be explained from a hot uniform big bang phase , which is the inevitable beginning of an expanding universe in general relativity .
however , it also predicts a relic background of radiation with @xmath0 k , which is reasonably close to the current observed value of @xmath1 k ( e.g. , * ? ? ?
gamow s triumph could be arguably thought of as the beginning of physical cosmology , as it bore all the marks of scientific method as we know it : a self - consistent theory is devised based on observations ( cosmic expansion and stellar elemental abundances ) , which makes predictions for completely different observables ( microwave background radiation ) , that are later confirmed .
the subsequent successes of observational cosmology only further confirmed the predictions of the big bang paradigm , and its general relativistic framework . in particular , the observation of large scale correlations in galaxy surveys , such as cfa , las campanas , 2df , and the sloan digital sky survey confirmed the growth of gravitational instability expected in a friedmann - robertson - walker ( frw ) space - time dominated by non - relativistic matter .
observation of anisotropies in the cmb angular maps by the cosmic background explorer differential microwave radiometer ( cobe dmr ) experiment connected the gravitational instability of structures today to their linear seeds in the early universe @xcite .
a decade later , cmb observations reached their next milestone with the wilkinson microwave anisotropy probe ( wmap ) , which measured the anisotropy power spectrum down to @xmath2 degrees with unprecedented precision @xcite , and verified consistency with the concordance cosmological model @xcite . yet another leap in sensitivity
is expected from planck satellite in early 2013 . in the mean time , geometrical tests in the late - time cosmology , in particular high redshift supernovae ia as standard candles @xcite , and baryonic acoustic oscillations ( bao ) in the spatial correlation of galaxies as standard rulers @xcite , confirmed a consistent picture for cosmic expansion history which appeared to have started accelerating some 5 billion years ago due to a mysterious dark energy component .
these successes have led many to call the current era , the era of `` precision cosmology '' . indeed , a whole battery of cosmological observations , ranging from lyman-@xmath3 forest in quasar spectra on scales smaller than @xmath4 mpc , to the statistics of galaxies and galaxy clusters on scales of @xmath5 mpc , and cmb anisotropies on scales of @xmath6 mpc are well - described by a six - parameter model ( e.g. , see * ? ? ?
* ) , often dubbed as concordance or @xmath7cdm cosmological model ( where @xmath7 and cdm stand for cosmological constant and cold dark matter , which comprise most of the energy in the present day universe ) .
these parameters include the present day densities of baryons , dark matter , and dark energy , as well as the amplitude and power of the initial spectrum of cosmological fluctuations ( assuming that it is a power - law ) . with the compilation of independent constraints from different cosmological observations ,
the statistical errors on these parameters have shrunk quickly over the past decade .
most of these parameters are now measured with 2 - 3% statistical precision ( e.g. , * ? ? ?
* ) , even though it is sometimes hard to quantify the systematic errors on these measurements . the renaissance of observational cosmology has also been recognized by the greater physics community , who have awarded two nobel prizes in physics to this discipline in the past six years : the nobel prize in 2006 was awarded to john mather and george smoot for discovery of blackbody spectrum and anisotropy of the cmb . the nobel prize in 2011
was awarded to saul perlmutter , adam riess , and brian schmidt for the discovery of dark energy ( or cosmological constant , the @xmath7 in @xmath7cdm ) by study of magnitude - redshift relation of high redshift supernovae ia .
the two discoveries , however , were of very different nature : the first confirmed two of the most fundamental predictions of standard big bang model , while the latter revived an apparently redundant but enigmatic parameter in general relativity . as i will discuss below ,
this dichotomy is reminiscent of the status of modern cosmology today .
there is a famous quote from lev landau , the great russian physicist of the 20th century , which says : `` cosmologists are often in error , but seldom in doubt '' ! unfortunately , landau passed away in 1968 , from the complications of a car accident that he was involved in 6 year earlier , so i wonder whether he ever had a chance to reflect on penzias and wilson s discovery of the cmb . maybe , if he had , or if he had lived for another decade or so to witness the onset of the renaissance in observational cosmology , he might have revisited his original scepticism .
nevertheless , one often wonders whether there is some wisdom in the words of such great minds that would survive beyond the ages . for me , this rang particularly true in october 1998 , when two of the most influential figures in modern cosmology , james peebles and michael turner debated `` the nature of the universe '' in the main auditorium of smithsonian s national museum of natural history in washington , dc .
the 1998 debate subtitle was `` cosmology solved ? '' , and their points of view appeared on the arxiv shortly afterwards @xcite . despite being the early days of precision cosmology ( which was a term also coined by turner ) , and only a few months from the discovery of dark energy from supernovae ia observations , turner was very optimistic that the basic tenets of @xmath7cdm cosmology , along with inflation , will survive further scrutiny .
on the other hand , peebles was more cautious : `` we have a well defined , testable , and so far quite successful theoretical description of the expansion : the relativistic friedmann - lemaitre cosmological model .
the observational successes of this model are impressive but i think hardly enough for a convincing scientific case . ''
it appears that , as we discussed above , the influx of observational data over the ensuing decade has validated turner s vision of precision cosmology .
however , there is a price for this success which is often ignored . more than 99 percent of today s energy content of the universe in the concordance cosmological model is either unidentified , or invisible to us ( e.g. , see * ? ? ?
* ; * ? ? ?
the most enigmatic component of @xmath7cdm is @xmath7 or the so - called dark energy , which comprises around 73% of the energy of the universe today .
we will discuss the mysteries of @xmath7 ( often called the cosmological constant problem ) in the next section , as it is intimately connected to the quantum theories of high energy physics and even a quantum theory of gravity . the next biggest contributor is cold dark matter ( or cdm ) which makes up around 23% of the cosmic energy budget .
the most popular candidates for cdm are in the form of elementary particles : either weakly interacting massive particles ( or wimp s ) or very light scalar particles , hypothesized to resolve the strong cp problem , known as axions . however
, none of the efforts to find non - gravitational evidence for these particles have managed to conclusively detect these particles .
therefore , it remains a possibility that a more bizarre candidate , such as a modification of gravity , could explain the same observations . while none of the proposed alternatives to cdm have enjoyed a similar phenomenological success in explaining both the early and late universe observations ( e.g. , * ? ? ?
* ) , apparent failures of cdm in matching observations on small scales may point to a more complex possibility ( e.g. , * ? ? ?
even most of the standard model particles ( often referred to as baryons ) , which comprise the remaining 5% of the energy of the universe , are expected to lie in a tenuous intergalactic medium , which has remained largely invisible to us .
attempts to account for these baryons in representative samples of the universe , found in large galaxy clusters , has been controversial , and arguably misses up to 30%-40% of the cosmic baryonic budget ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . finally ,
inflation , a period of rapid exponential expansion in the very early universe @xcite , which is often credited for generating a nearly spatially flat cosmology with a nearly scale - invariant spectrum of cosmological fluctuations , is plagued by questions of empirical falsifiability .
it turns out that while natural expectations from ( certain ) inflationary models are consistent with these observations , it is very easy to introduce arbitrarily large modifications to inflationary predictions by modifying inflationary initial conditions , and/or adding extra physics to the inflationary model . in the absence of a(n
established ) uv - complete model of high energy physics ( which includes quantum gravity ) , it is hard to argue whether such modifications might ( or might not ) be natural ( e.g. , * ? ? ?
further complication is introduced by the possibility of eternal inflation , where predictions are plagued by the infamous `` measure problem '' , which we will come back to in section [ falsify ] .
as a careful reader might have noticed , we have mentioned quantum gravity more than once in our introduction .
this is not a coincidence , as we will see in the next section .
the search for a consistent quantum theory that includes gravity ( or geometrodynamics ) as a component is as old as both general relativity ( which made geometry and gravity synonymous ) and quantum mechanics ( see * ? ? ? * for an overview of the history of quantum gravity ) . by now
, it has become quite clear that a quantization of einstein s theory of relativity , while well - behaved as an effective field theory , is non - renormalizable and thus fails ( at least as a perturbative theory ) as gravitons approach planck energy ( @xmath8 gev ) .
it is easy to see this on dimensional grounds : considering small perturbations around minkowski background , @xmath9 , in natural units ( @xmath10 ) , the gr action can be written as : s_gr ~-d^4x ( h+_2 h^2+_3 h^3 + ... ) h,[s_gr ] where @xmath11 s represent dimensionless constants , we have abbreviated the tensorial structure of the equation , and ignored additional subtleties in dealing with gauge symmetries .
considering the zero point fluctuations of @xmath12 on energy scale @xmath13 , from the free or quadratic action ( first term in equation [ s_gr ] ) we have : h^2 _ e 8m^-2_p ^^-1(e ) ~(em_p)^2 , [ h2_lorentz ] where , in the last step , we have used the lorentz - invariant ( or minkowski space ) dispersion relation for gravitons : @xmath14 .
it is now quite clear that as @xmath13 approaches @xmath15 , the perturbative expansion in eauation ( [ s_gr ] ) breaks down . in quite the same way that @xmath16 and @xmath17 gauge bosons in modern electroweak theory
cured the non - renormalizability of fermi s low energy 4-point weak interaction , various attempts at a theory of quantum gravity have mainly comprised of coming up with `` more fundamental '' degrees of freedom , that can be included in a renormalizable or finite theory .
such degrees of freedom could be fundamental strings ( e.g. , * ? ? ?
* ) , spin networks @xcite , discrete causal sets ( e.g. , * ? ? ?
* ) , or other discrete theories in space or space - time that resemble general relativity in the continuum limit .
alternatively , it has been proposed that gr might be `` asymptotically safe '' , i.e. it has a non - perturbative but well - defined quantization at arbitrarily high energies @xcite .
however , for most experimental physicists , approaching energies comparable to planck energy is little more than a distant fantasy .
the most powerful accelerators on earth miss planck energy by 15 orders of magnitude , while ultra high energy cosmic rays are still 9 orders of magnitude short of @xmath15 .
therefore , the majority of physicists may not be much disturbed by the limitations of einstein s theory of gravity .
unfortunately , astrophysicists do not enjoy such luxury ! as first proved by hawking and penrose in a series of singularity theorems @xcite , general relativity predicts its own demise ! in particular , the end state of massive stars are in singularities , where temperatures exceed planck energy .
millions of these singularities just happen to live in our own galaxy , although they are expected to be shrouded by the event horizons of astrophysical black holes . while strong theoretical motivations exist for a cosmic censorship conjecture , which hides singularities behind the event horizons , there is no guaranty that this will be the case in a theory of quantum gravity .
in fact , it is widely believed that event horizons do not exist in a full theory of quantum gravity , although the minimal quantum effects ( such as hawking radiation , * ? ? ?
* ) are far from being observable for astrophysical black holes .
more seriously , as we approach big bang in our past , the temperature rises to planck energy and beyond .
therefore , any consistent theory of cosmological initial conditions has to include quantum gravity , which impacts scalar and tensor gravitational degrees of freedom directly probed in observational cosmology ( and potentially gravitational wave detectors ) . yet ,
the most dramatic challenge of quantum gravity for cosmology does not come from planck - scale physics .
quite to the contrary , it comes on scales that should see very little quantum corrections to general relativity . like quantum gravity itself
, this issue also dates back to the early days of the development of quantum mechanics . as early as 1920 s
, pauli had recognized the tremendous amount of energy in the zero - point fluctuations of the electromagnetic field @xcite . while this energy density is divergent , apparently , he had regulated the divergence by taking the classical radius of electron as a momentum cut - off .
if all this energy were to gravitate according to einstein s theory of relativity , the universe would curve so much that it could not fit the lunar orbit , let alone the solar system , or the rest of the galaxy !
this is now recognized as the `` old cosmological constant ( cc ) problem '' .
a more careful computation of vacuum energy involves introducing a lorentz - invariant regulator , which suggests that a natural value for vacuum energy is roughly given by the sum of the fourth power of the particle masses in the theory : _
vac ~_i m^4_i . for standard model , the sum is clearly dominated by the most massive particle , i.e. top quark with @xmath18 gev . if the energy @xmath19 were to gravitate , the entire universe would have been smaller than a centimetre in size !
classical contributions from the higgs potential can change this expectation by order unity , but short of a conspiracy , it is extremely unnatural to expect a cancelation between different contributions to @xmath20 to 1 part in @xmath21 , in order to be consistent with the observed size of the universe ( @xmath22 gpc ) . because of the clear pathology in the above estimates , physicists ignored the old cc problem for the better part of the 20th century .
more conscious theorists speculated a yet unknown symmetry that would lead to cancellations of different contributions to the vacuum energy .
while examples of such symmetries , such as conformal symmetry or supersymmetry , exist , they all seem to be violated by the mass terms in standard model , and thus , as we argued , stop short of curing the old cc problem by some 60 orders of magnitude .
in an influential review article @xcite , steven weinberg reviewed different approaches to the cc problem , outlining why each approach fails to solve the problem , while stopping short of dismissing any approach completely .
he further speculated that , if all other approaches fail , anthropic considerations , which require a @xmath7 suitable for existence of intelligent life , will predict a value just within reach of cosmological observations .
as we discussed in the previous sections , this was indeed verified at the turn of the century with the discovery of cosmic acceleration from high redshift supernovae .
the modern cc problem is sometimes divided into three problems : 1 . *
the old cc problem : * why is nt @xmath7 as big as its natural scale in particle physics ?
the new cc problem : * why does the observed vacuum energy have such an un - naturally small but non - vanishing value ? 3 . * the coincidence problem : * why do we happen to observe vacuum density to be so close to matter density , even though their ratio can vary by up to 120 orders of magnitude during the cosmic history ? let us reiterate that these are still problems ( or puzzles ) in quantum gravity , as they appear when we couple gravity to quantum mechanical theories , even though they concern physics far below the planck scale .
other than the _ gravitational aether _
model that we will discuss in section [ aether ] , the only known `` solution '' to the cc problem comes from the anthropic considerations , or its modern incarnations in the string landscape and/or eternal inflation .
in fact , as we mentioned above , one may argue that the non - vanishing @xmath7 was `` predicted '' based on these considerations . in the next section ,
we discuss to what extent this can be compared to predictions in other scientific theories .
presumably the best ( and the worst ! ) way to define physics is by the set of problems that are tackled by practicing physicists .
nevertheless , a careful study of the history of meaningful progress in science , as often done by philosophers , may reveal common features that may not be immediately obvious to practicing scientists .
probably the most influential philosopher of science of the 20th century was karl popper , who coined the term _ critical rationalism _ to describe his philosophy of a scientific theory @xcite .
popper argues that scientific theories are in fact measured by their falsifiability , as opposed to their verification , as no amount of positive tests can in fact verify a theory .
however , a single counter - example suffices to rule out ( or falsify ) a theory .
more significantly , a scientific theory needs to be falsifiable , i.e. there should exist observations and/or experiments that can potentially falsify the theory . in practice ,
falsifiable theories that survive more tests than their competitors will be considered more successful .
so , do anthropic considerations provide a falsifiable solution to the cc problem ? to answer this , it is interesting to recount how developments in string theory and cosmology culminated in the anthropic principle at the turn of the century .
one of the most puzzling features of the standard big bang scenario was how points in the universe that have never been in casual contact have the same temperature to 1 part in @xmath23 .
this is known as the `` horizon problem '' , which was later solved by cosmic inflation , as we discussed in the previous section @xcite . what inflation does is to stretch a small causally connected patch exponentially , across many hubble horizons .
so , points that appear causally disconnected today were parts of the same hubble patch at the beginning of inflation .
however , it was later realized that for many ( if not most ) successful inflationary models , inflation never ends !
while the success of standard big bang theory requires inflation to end before big bang nucleosynthesis , there are always regions in field space that never stop inflating in these model . since the physical volume of these regions is exponentially bigger than those that have stopped inflating , most of the volume of the universe will always be inflating .
this is known as eternal inflation , and has a significant anti - copernican implication : if our cosmology emerges from eternal inflation , we can not live in a typical region of the universe ( e.g. , * ? ? ?
the second development was when research in string theory , which is the most popular contender for a theory of quantum gravity , failed to single out any particular vacuum for the theory .
in fact , it was argued that string theory might have as many as @xmath24 viable vacua ( or a landscape of vacua ) , with each vacuum having a very different low energy physics ( e.g. , * ? ? ? * ) .
the combination is now straightforward : no matter where you start the universe in the string landscape , assuming that it permits eternal inflation , there will be an infinite time to populate all the other landscape vacua via quantum tunneling . of course
, most of these regions will not be hospitable to humans ( and presumably other intelligent life ) .
therefore , we can use anthropic principle , to pick the region where cosmology and particle physics allows humans to live .
in particular , this region can not have a very big positive or negative @xmath7 , as neither allow enough time for galaxies , stars , planets , and life ( as we know them ) to form @xcite .
besides the nebulosity of the notion of `` intelligent life '' , one of the problems with this interpretation is that all the predictions are now probabilistic .
however , unlike quantum mechanics , where we compute probabilities for finite ensembles , the eternally inflating ensembles are inherently infinite in size .
this is known as the `` measure problem '' ( which we referred to earlier ) , as you could find very different probabilities , depending on how you regulate ( or cut off ) your infinite ensembles .
the second problem is what we started this section with , i.e. falsifiability . since most of the string landscape exists beyond our cosmological horizon , or at energies much higher than those accessible in accelerators
, it is really hard to test ( or falsify ) its existence .
of course , we might get lucky and see glimpses of another vacuum ( e.g. , * ? ? ? * ) , but for the most part , it has been hard to come up with ways to falsify this paradigm ( but see * ? ? ? * for a notable possible exception ) .
we should emphasize that string theory and inflation are the direct results of extending _ locality _ and _ unitarity _ , the underlying principles of relativity and quantum mechanics , to gravity and cosmology .
my personal point of view is that their failure in coming up with a falsifiable cosmological model ( and most notably a falsifiable solution to the cc problem ) , is cause to revisit these sacred principles of 20th century physics .
we will discuss this next .
it has been long recognized that one of the ways to deal with infinities in quantum gravity ( and even infinities in quantum field theory ) , is to break lorentz symmetry ( see e.g. , * ? ? ?
the reason is that lorentz group so(3,1 ) , unlike e.g. , the rotation group so(3 ) , is non - compact , and thus has an infinite volume .
therefore , the sum over the intermediate states for many quantum mechanical calculations ( for rates or energy shifts ) yield infinities . for renormalizable theories
, these infinities can be absorbed in renormalization of a finite number of parameters that can be fixed empirically .
however , this is not the case for non - renormalizable theories , such as gravity , which require renormalizing an infinite number of parameters , rendering the theory non - predictive .
it is easy to see how violating lorentz symmetry can cure this problem @xcite . going back to our quantized gravitons in section [ qg ] , we can see that using @xmath25 in equation ( [ h2_lorentz ] ) yields : h^2 _ e 8m^-2_p ^^-1(e ) ~(mm_p)^2 ( em ) .
therefore , as long as @xmath26 , the theory remains perturbative , even for energies far beyond @xmath15 .
the anisotropic scaling of space and time in this theory ( as @xmath27 ) is known as lifshitz symmetry .
a more realistic scenario is an interpolation between lorentz symmetry at low energies , and lifshitz symmetry at high energies : ( k ) = k + .[omega_k ] even though lorentz symmetry is approximately recovered in the ir , there is still a single preferred frame in which equation ( [ omega_k ] ) can be valid , as the dispersion relation is not invariant under lorentz transformation .
this amounts to a revival of `` gravitational aether '' , as an additional component to the geometric structure of einstein s gravity .
moreover , equation ( [ omega_k ] ) violates the _ locality _ of relativistic theories , as localized perturbations can travel arbitrarily fast .
note that this already resolves the `` horizon problem '' , which was one of the original motivations for cosmological inflation .
moreover , the dispersion relation @xmath25 leads to a scale - invariant spectrum of cosmological fluctuations , provided that it can be converted to curvature perturbations at late times @xcite .
in other words , with lifshitz symmetry we can potentially kill two birds with one stone : make gravity renormalizable and generate the correct statistical distribution for cosmological fluctuations
. however , more detailed studies are necessary to quantify the phenomenological implications of hoava - lifshitz cosmology as an alternative to cosmic inflation . in the next section ,
we provide a _ second _
motivation for aether , based on a falsifiable solution to the cc problem .
the old cc problem can be quantified as the pathologically large contribution to @xmath20 in the stress tensor @xmath28 on the right hand side of einstein equations : t _ = _ vac g
_ + excitations .
we thus see that if only the traceless part of @xmath28 appeared on the right hand side of einstein equations , gravity would not be sensitive to @xmath20 , which could potentially resolve ( at least ) the ( old ) cc problem .
let us write this as @xcite : ( 8g ) g _
= t_-g_t^_+t_. [ grav_aether ] the reason we added @xmath29 to the right hand side of equation ( [ grav_aether ] ) is that , thanks to bianchi identities and energy - momentum conservation , einstein and stress tensors have zero divergence .
however , @xmath30 is not generically a constant , which implies that consistency of equation ( [ grav_aether ] ) requires : ^t _ = _ t^_. [ aether_cons ] here , we call @xmath29 `` gravitational aether '' , which can be interpreted as an additional fluid ( or degree of freedom ) of this new theory of gravity .
of course , the predictions of the theory depend our choice for @xmath29 . given that equation ( [ aether_cons ] )
provides 4 equations , they completely specify the evolution of a perfect fluid with a given equation of state .
if we want to avoid introducing additional dimensionful scales to the theory , the aether equation of state will be set by a constant : @xmath31 .
it turns out that only two possibilities are consistent with conditions of stability and element abundances from big bang nucleosynthesis : @xmath32 or @xmath33 @xcite .
the former possibility leads to the so - called unimodular gravity , which has been discussed by several authors , including einstein himself ( e.g. , see * ? ? ?
* ) . in this case
, aether does not have any dynamical degree of freedom
. however , equation ( [ aether_cons ] ) can be solved to show that the cc re - emerges in the theory as an integration constant .
the second possibility is more novel and interesting .
one may dismiss @xmath33 due to superluminal propagation , as sound speed is @xmath34 for aether .
while , as we argued in the previous section , this should not necessarily scare us , the case of @xmath35 , or the incompressible limit is particularly interesting .
it can be argued that , in this limit , sound waves in aether can not be excited , and thus there is no superluminal signal propagation .
one way to see this is that any phonon of finite wavelength has an energy of @xmath36 , which implies that we need infinite energy to excite aether phonons .
furthermore , similar to the case of @xmath32 , @xmath37 does not have any _ independent _ dynamical degree of freedom , even though it does have a velocity , and thus specifies a preferred frame at each point in space ( relativistic dynamics of irrotational incompressible fluids , otherwise known as cuscuton , has been studied in @xcite , and @xcite ) .
notice that the `` gravitational aether '' theory , as we just specified with @xmath37 , i.e. ( 8g ) g _ = t_-g_t^_+ p(u_u_-g_),[full_aether ] has no additional parameter ( or independent dynamical degree of freedom ) , compared to einstein s gravity .
so could it be consistent with all the precision and cosmological tests of gravity ?
probably the sharpest prediction of the aether theory is that effective gravitational constant in friedmann equation becomes dependent on the matter equation of state : h^2= _ m , g_eff = ( 1+w_m ) g_n,[cosmo ] where @xmath38 , is the matter equation of state , and @xmath39 is newton s gravitational constant . in particular
, this predicts that gravitational constant during radiation era was 33% larger than in the matter era : @xmath40 .
( [ fig_bbn ] ) and table ( [ tab - constraints ] ) summarize the big bang nucleosynthesis and cmb+late time cosmology constraints on this ratio @xcite .
we see that some datasets , namely @xmath41li , ( most ) cmb experiments , and lyman-@xmath3 forest in quasar spectra prefer ratios close to aether prediction , while others are closer to gr ( @xmath42 ) , or can not distinguish between the two .
interestingly , however , all the best fits are at @xmath43 .
the influx of observational data , and in particular cmb power spectrum from the planck satellite over the next year , should dramatically improve these constraints , and hopefully confirm or rule out aether predictions conclusively .
lines for d / h , @xmath44(or @xmath45he ) and @xmath46li / h are shown @xcite .
the upper and lower horizontal dashed lines indicate gr and gravitational aether predictions , respectively.,scaledwidth=75.0% ] .summary of the constraints on @xmath47 and the associated @xmath48 confidence intervals for different combinations of observational data @xcite .
here , wmap , act , and spt are different cmb experiments , bao stands for baryonic acoustic oscillations , while sne and hubble refer to measurements of distance - redshift relation from supernovae ia , and hubble constant .
ly-@xmath3 refers to measurements of lyman-@xmath3 forest absorption distribution in the spectra of distant quasars .
the last two rows take helium abundance @xmath44 as a free parameter , while the rest fix it to its value set by galactic observations . [
cols="<,^",options="header " , ] [ tab - constraints ] it can be shown that the simple scaling of effective newton s constant with matter equation of state ( equation [ cosmo ] ) is still valid for inhomogenous situations , provided that : u_=u _ , w_m = const .
g_eff = ( 1+w_m ) g_n , i.e. if the matter equation of state is constant , all the solutions in gr also satisfy the equations in the gravitation aether theory with a renormalized gravitational constant , if aether moves with matter .
let us first ignore the gravitational effect of local vorticity in matter and aether flows . in this regime , the flow of aether is completely fixed by matter , and we find that the only effect of aether is to renormalize newton s constant by a factor of @xmath49 .
@xcite show that none of the current precision tests of gravity constrain this effect , as it involves probing the internal structure of objects with near - relativistic pressure .
we will discuss the case of neutron stars in the next section .
coming back to the effect of vorticity , it turns out that rotational motion of aether is essentially decoupled from matter .
therefore , there is no reason for aether to rotate within rotating bodies .
assuming an irrotational aether will then boost the gravitomagnetic effect sourced by local vorticity , by 33% , which is currently consistent at @xmath50 with lageos and gpb experiments @xcite . as we mentioned above
, the internal structure of objects with relativistic pressure is expected to be significantly different in the aether theory .
the only known ( close to equilibrium ) astrophysical objects that have this property are neutron stars . in fig .
( [ fig : mreos ] ) , we show the mass - radius relation for gr and aether theories , for two widely used nuclear equations of state @xcite .
most notably , we see that as gravity gets stronger with relativistic pressure , the maximum allowed mass of neutron stars ( the so - called oppenheimer - volkov limit ) decreases in aether theory .
this is already close to ruling out the theory , for the most massive neutron star with a reliable mass measurement : @xmath51 @xcite .
however , the uncertainty in the nuclear equations of states may prohibit drawing definite conclusions from such observations . probably the most speculative , and yet most fascinating , feature of the aether theory is how it couples to astrophysical black holes . as we discussed above ,
the exciting physics of singularities of black holes is hidden behind their event horizons in general relativity .
in fact , astrophysical black holes are expected to have only two numbers ( or hairs ) associated with them : mass and angular momentum .
however , this does not need to be the case for a different theory of gravity , such as the aether theory . by solving the static spherically symmetric aether equations in vacuum , @xcite
show that the aether pressure , no matter how small at infinity , blows up close to the horizon of black holes .
this is not too surprising ; one expects the same thing to happen for other types of matter .
of course , the reason this is pathological for regular matter is that we do not expect matter to sit at rest close to the horizon , but rather it would fall through .
the story , however , is different for an incompressible fluid , as fluid inside the light horizon can communicate with the outside .
one can show that aether pressure would still blow up even in a dynamical situation , just inside the horizon in a collapsing star ( saravani , afshordi & mann , in preparation ) .
what @xcite propose instead is that this singularity is regulated by quantum gravity effects .
the vacuum metric in the presence of aether is given by @xcite : ds^2= ( 1-r_s / r)^2-(1-r_s / r)^-1 dr^2-r^2d^2,[bh_metric ] where @xmath52 is the schwarzschild radius , and @xmath53 is the aether pressure far away from the black holes .
while @xmath54 is an analytic function with a closed form , it is particularly illuminating to consider it close to @xmath55 : f(r ) = r_s^2\{-2(r / r_s-1)^-1/2+o}. we notice that unlike schwarzschild black hole , the gravitational redshift @xmath56 approaches a maximum value at @xmath57 : 1+z_max = -(8g_n p_0 r_s^2)^-1 , while the metric is not defined for @xmath58 by analytically continuing metric ( [ bh_metric ] ) in terms of proper radial distance . ] .
if we assume quantum gravity effects set a maximum gravitational redshift of planck energy divided by hawking temperature ( which is equivalent to assuming that aether only becomes important within a planck length of the horizon ) , the aether pressure away from the black hole is fixed to : p_0 = - = p_obs , ( m_bh 85 m_)^-3,[aether_de ] i.e. for stellar black holes , which happen to make ( by number or mass ) the majority of astrophysical black holes in our universe , this criterion fixes the pressure of aether to be comparable to the observed pressure of dark energy .
given that the average mass of astrophysical black holes evolves with redshift , one can predict an evolution for dark energy , or the dark energy equation of state ( see fig .
[ fig - obsmstar ] ) .
, in units of @xmath59 as a function of redshift .
different lines represent different astrophysical scenarios of black hole formation . * top panel : * the prediction of these scenarios for the effective dark energy equation of state @xmath60 , given that aether pressure scales as @xmath61 , which can be compared to constraints from cosmology .
the shaded area shows the region currently excluded at 68% confidence level for this parameter , as measured from cosmological observations @xcite . ] however , extending the analysis of @xcite to more realistic situations , i.e. including multiple moving black holes , in the presence of matter , has proven incredibly challenging .
this is not too surprising , as one needs to solve time - dependent non - linear partial differential equations on vastly disparate scales ( ranging from schwarzschild to hubble radii ) , and is practically impossible , in lieu of ( a yet to be found ) appropriate approximation scheme . until that is done , which can then provide falsifiable predictions for the black hole - dark energy connection in the theory , equation ( [ aether_de ] )
remains little more than a(n extremely suggestive ) numerical coincidence .
in this article , i have provided a very subjective overview of successes and failures of the standard cosmological model , especially in its relation to the fundamental physics of gravity . while observational cosmology is undergoing its renaissance , deep mysteries continue to haunt our theoretical understanding of the ingredients of the concordance cosmological model . in my opinion ,
the most enigmatic of these puzzles is the cosmological constant ( cc ) problem : the apparently extremely fine - tuned quantum vacuum energy , its relation to the observed dark energy , and possibly a more fundamental theory of quantum gravity .
fuelled by parallel theoretical invention of eternal inflation and string landscape at the end of the last century , the anthropic `` solution '' to these puzzles is gaining more traction among the theoretical physics community .
personally , i find this to be an alarming development .
this is partly due to ( near ) lack of falsifiability in these paradigms , which is a minimum standard for our scientific theories , as i discussed in section [ falsify ] .
yet another source of worry is extrapolating well - established physical principles far beyond the regime in which they are tested , which happens both in eternal inflation and string theory .
however , my most serious concern is that a general acceptance of an unfalsifiable framework will inevitably stifle a search for alternative solutions to these fundamental puzzles .
if this happens , and observational probes of dark energy fail to find any deviation from a cosmological constant , we might very well enter a long period of intellectual stagnation in our understanding of the universe , with no clear end in sight ( to borrow terminology from christopher stubbs , * ? ? ?
. given that a century of scientific enquiry to solve problems of quantum gravity , particularly the cc problem , have failed to come up with a falsifiable solution , it stands to reason that it might be time to revisit some of the fundamental principles of 20th century physics , i.e. locality ( or lorentz symmetry ) , and/or unitarity . in this article ,
i gave two arguments , based on renormalizability of gravity and the cc problem , for why we might have to give up lorentz symmetry to come up with a falsifiable solution .
i then outlined different predictions of the theory , with varying degrees of robustness , which can rule out or confirm the theory in comparison with gr , potentially as early as next year .
alternative approaches that take a similar point of view in regards to lorentz symmetry include einstein - aether theories @xcite , horava - lifshitz gravity @xcite , and shape dynamics @xcite . a typical objection to lorentz - violating theories
is that nothing prevents quantum corrections from introducing order unity violation of lorentz symmetry at low energies ( e.g. , * ? ? ?
so why does particle physics seem to obey lorentz symmetry to such precision ? here
is an argument for why expectation of @xmath62 lorentz violation might be too naive : imagine that aether is described by the gradient of a light scalar field @xmath63 which has a canonical kinetic term .
if @xmath64 , then the field will be slowly rolling down its potential , and thus @xmath65 specifies a preferred frame , which coincides with the cosmological comoving hypersurfaces . a typical lagrangian which spontaneously breaks lorentz symmetry for field @xmath66
can be written as : = ( ) ^2- m_^2 ^ 2++ ( ) ^2- m_^2 ^ 2 .
[ lagrangian ] given that the kinetic energy of the aether field @xmath63 should be less than the critical density of the universe , we have : ^2 < m_p^2h^2 , which puts an upper limit on the lorentz - violation for the @xmath66 field : c
_ = < ( m_p h ^2)^2 , given that @xmath7 is the energy cut - off of the effective field theory in equation ( [ lagrangian ] ) , i.e. the theory is only valid for @xmath67 , which puts an upper limit on the lorentz - violation for @xmath66 particles at energy @xmath13 , assuming the validity of the effective field theory : c _ < ( m_p h e^2)^2 ~(e ) ^-4 .
we see that , already for energies as low as @xmath68 mev ( which is the mass of electron , and the typical energy of solar or reactor neutrinos ) , the lorentz violation should be less than @xmath69 , and is further constrained for more energetic particles .
this observation suggests that expectations of large violations of lorentz symmetry might be too naive within a consistent effective field theory framework , which include gravity in a cosmological spacetime .
another objection , specific to the gravitational aether model ( equation [ full_aether ] ) is that an action principle that could lead to these equations is so far non - existent .
however , an action is only necessary if we want to quantize gravity , while the field equations ( [ full_aether ] ) , assuming that they can be consistently solved , are sufficient at the level of classical and semi - classical gravity .
presumably , more structure will be necessary to define a quantum theory that reduces to the gravitational aether in the classical regime .
finally , i should mention previous attempts to decouple quantum vacuum from gravity , which have provided much of the inspiration for this work .
these include massive gravity and degravitation @xcite , cascading gravity ( e.g. , * ? ? ?
* ) , and supersymmetric large extra dimensions ( e.g. , * ? ? ?
however , to the best of my knowledge , none of these frameworks have been developed well enough to make concrete cosmological predictions ( at least in the regime that they address the cc problem ) .
let me close this article by stating the obvious , that physics is an empirical science .
while the bulk of activity in theoretical physics and astrophysics is driven by ( for lack of a better word ) fashion , the credibility of a theory is ultimately judged by its concrete predictions against nature , _ not _ its popularity , mathematical elegance , parsimony , etc .
do you hold your favourite theories to this standard ? !
i would like to first thank the astronomical society of india for awarding me its 2008 vainu bappu gold medal .
moreover , i am indebted to my students and collaborators siavash aslanbeigi , michael balogh , brendan foster , farbod kamiab , kazunory kohri , chanda prescod - weinstein , and georg robbers , who are responsible for most of the results that are reviewed in this article .
i am supported by the university of waterloo and the perimeter institute for theoretical physics .
research at perimeter institute is supported by the government of canada through industry canada and by the province of ontario through the ministry of research & innovation .
afshordi n. , chung d. j. h. , doran m. , geshnizjani g. , 2007a , phrvd , 75 , 3509 afshordi n. , lin y .- t .
, nagai d. , sanderson a. j. r. , 2007b , mnras , 378 , 293 alpher r. a. , bethe h. , gamow g. , 1948 , phrv , 73 , 803 | einstein s theory of general relativity is the benchmark example for empirical success and mathematical elegance in theoretical physics .
however , in spite of being the most successfully tested theory in physics , there are strong theoretical and observational arguments for why general relativity should fail .
it is not a question of if , but rather a question of where and when ! i start by recounting the tremendous success in observational cosmology over the past three decades , that has led to the era of precision cosmology .
i will then summarize the pathologies in einstein s theory of gravity , as the cornerstone of standard cosmological model .
attempts to address these pathologies are either inspired by mathematical elegance , or empirical falsifiability .
here , i provide different arguments for why a falsifiable solution should violate lorentz symmetry , or revive `` gravitational aether '' .
deviations from einstein s gravity are then expected in : 1 ) cosmological matter - radiation transition , 2 ) neutron stars , 3 ) gravitomagnetic effect , 4 ) astrophysical black holes , and their potential connection to dark energy , and 5 ) early universe , where the predictions are ranked by their degree of robustness and falsifiability .
[ firstpage ] cosmology : theory gravitation dark energy neutron stars black hole physics
early universe |
in recent years considerable attention has been focussed on the structural , magnetic and electron transport properties of perovskite oxides of the type r@xmath0a@xmath1mno@xmath2 ( where @xmath10 is a rare - earth element , @xmath11 a divalent alkaline - earth element ) .
this interest was caused by observation of an extremely large negative magnetoresistance in these compounds @xcite , which was called colossal magnetoresistance ( cmr ) .
along with fundamental importance for condensed matter physics , this phenomenon also offers applications in advanced technology .
therefore the problem of cmr continues to be topical .
the doped manganites undergo a phase transition with decreasing temperature from a paramagnetic insulating state into a highly conducting ferromagnetic phase .
it can be said that this insulator - metal transition occurs approximately simultaneously with a paramagnetic - ferromagnetic transition ( at least in good quality crystals ) .
the external magnetic field shifts the transition temperature @xmath8 ( which is usually near room temperature in well ordered samples with @xmath12 to higher temperature producing the cmr ( see reviews in refs . ) .
the most pronounced cmr effect was found in la@xmath0ca@xmath1mno@xmath2 films with @xmath13 .
the undoped compounds from this series ( lamno@xmath2 and camno@xmath2 ) are antiferromagnetic insulators . in the intermediate range of doping
( @xmath14 ) la@xmath0ca@xmath1mno@xmath2 is a ferromagnetic conductor at low temperature .
the ferromagnetic state is believed to be due to the appearance of mn@xmath15 ions with substitution of la@xmath16 by a divalent cation .
it can be assumed that ferromagnetism results from the strong ferromagnetic exchange betweeen mn@xmath16 and mn@xmath15 .
the appearance of such an interaction can be qualitatively explained within the double - exchange ( de ) model @xcite .
this model , however , can not alone explain either the huge drop in resistance at the transition , or the real nature of the insulating state at @xmath17 and , therefore , the conductivity mechanism in this state .
thus , additional physical processes have been invoked to explain the insulating state and insulator - metal transition . among them
are lattice ( polaron ) effects @xcite and the possibility of phase separation into charge - carrier - poor and charge - carrier - rich regions@xcite .
the conductivity of la@xmath0ca@xmath1mno@xmath2 with @xmath18 is determined by holes which appear as the result of replacement of trivalent la by divalent atoms .
the de model is based on the assumption that the holes in doped manganites correspond to mn@xmath15 ions arising among the regular mn@xmath16 ions due to doping .
however another point of view exists@xcite that the holes go on oxygen sites .
the experimental data on this point are contradictory .
there is experimental evidence ( see ref . and references therein ) that holes doped into lamno@xmath2 are mainly of mn @xmath19 character . on the other hand experimental studies described in refs .
give evidence that the charge carriers responsible for conduction in doped manganites have significant oxygen @xmath20 character .
this is just one example illustrating that to date there is no consensus in the scientific community about the basic transport properties of doped manganites .
it may be inferred , therefore , that the understanding of these properties is far from completion and that further experimental and theoretical investigations of this matter are necessary .
it is well known that doped manganites of the same chemical composition but with different degrees of crystal lattice disorder show quite different transport and magnetic properties .
the disorder can be altered either with variation of sample preparation conditions ( for example , substrate temperature and post - annealing at film preparation ) or using radiation damage@xcite . with increasing disorder the resistivity peak temperature @xmath21 and the curie temperature @xmath8 decrease , while the magnetoresistance increases . in understanding the nature of cmr
the influence of disorder of the crystal lattice is one of the important points and should be taken into account together with spin , lattice and other effects .
this communication is concerned mainly with this problem .
the object of investigation was thin - films la@xmath0ca@xmath1mno@xmath2 with @xmath22 .
the disorder was enhanced by irradiating the films at room temperature with high - energy ( @xmath4 mev ) electrons .
this high energy of the incident electrons makes it possible to produce a uniform distribution of damage defects , without any significant variation of defect concentration as a function of depth ( all incident electrons go through the film ) .
in contrast to low energy ion irradiation , no interstitial implanted impurity ions can remain in the film for electron irradiation to produce inhomogeneity .
similarly , in contrast to very high energy ion irradiation , electron irradiation in our study does not produce extended defects , such as cascades and clusters .
this facilitates the interpretation of the experimental results . at the low damage level in this experiment
, however , the electron radiation damage may indeed be quite similar to damage induced by very low level , intermediately high - energy ion irradiation .
the maximal electron fluence in this study was about @xmath5 @xmath6 .
the calculated quantity of displacements per atom ( dpa ) is about @xmath7 .
this comparatively small radiation damage results in an appreciable increase in film resistivity .
it was found that the relative resistivity increase in the ferromagnetic metallic state ( below curie temperature @xmath8 ) was much greater than in the insulating state ( above @xmath8 ) .
such a small amount of radiation damage should not induce any noticeable resistance variations in ordinary ferromagnetic or non - ferromagnetic metals . at the same time any large influence of electron irradiation with the above - mentioned fluence on the @xmath8 and the magnitude of the colossal magnetoresistance was not observed .
possible reasons for this unusual behavior for the doped manganites are discussed .
the la@xmath0ca@xmath23mno@xmath24 films were prepared by physical vapor codeposition of la , ca and mn from three separate , independently controlled sources , similar to the technique for preparation of ca - ba - cu oxide precursors for growth of oriented tl@xmath25ca@xmath25ba@xmath25cu@xmath24o thin films@xcite .
the deposition was performed in 10@xmath26 torr of oxygen onto laalo@xmath24 substrates heated to about 600@xmath27c .
la and mn were evaporated from alumina crucibles heated with a tungsten filament , and ca was evaporated from a knudsen cell .
post deposition anneals of the films at 900@xmath27c in flowing oxygen improved the cmr behavior and produced well ordered films .
the composition of the film was determined by microprobe analysis of an unannealed film deposited simultaneously onto a fused quartz substrate .
the films were also characterized by x - ray diffraction and ac susceptibility measurements .
agreement among the values of @xmath8 determined by the real part @xmath28 of the susceptibility and @xmath21 determined by both measurements of the resistivity and the imaginary part @xmath29 of the susceptibility confirm that the films have good chemical and magnetic homogeneity based on the scheme proposed by araujo - moreira , @xmath30@xcite .
further details of the preparation technique and characterization are presented elsewhere@xcite .
although a sensitive magnitometer was not available for magnetization measurements with these films , ac susceptibility was measured , both for unirradiated and irradiated films . in each case
the onset of the sharp increase in the real part of the susceptibility @xmath28 and the sharp peak in the imaginary part of the susceptibility @xmath29 coincide within experimental error with the value of @xmath31 .
the sharp increase in the low frequency @xmath28 ( @xmath32 hz ) data presumably corresponds to the magnetic transition temperature @xmath33 .
representative data for an unirradiated film is presented in fig . 4 of ref . .
data for @xmath28 and @xmath29 for one of the films irradiated in this study ( not shown ) has much less noise and provides clear evidence that @xmath33 and @xmath31 coincide , both for the unirradiated and irradiated films in this study .
this is not unexpected , however , since ion irradiation studies have shown@xcite that for high quality films with small lattice damage , these two temperatures are essentially the same , but for much higher lattice damage @xmath31 will be at much lower temperature than @xmath33 . throughout this paper reference
will be made to @xmath31 , but , since @xmath31 and @xmath33 are essentially identical , the conclusion from these experiments apply to both equally well . during the electron irradiation
the films were in a special holder which was cooled with running water and a powerful fan .
the estimated overheating above room temperature during the irradiation was no more than @xmath34 c. two film samples were investigated ( x=0.27 and 0.36 ) .
these films ( with thicknesses about 300 nm ) were prepared under nearly the same conditions .
the resistance of the films was measured using a standard four - probe technique .
an applied magnetic field ( up to 20 koe ) was perpendicular to the film plane and to the direction of current .
the results obtained were nearly the same for both films and will be illustrated by the data from the x=0.36 sample .
the transport properties of this film in its initial state ( before irradiation ) correspond to the usual behavior of cmr films ( fig.1 and 2 ) .
namely , the temperature dependence of resistance @xmath35 has a maximum ( peak ) at @xmath36 k ( the maximum is rather smeared ) .
below @xmath21 ( which for these manganite samples is always in the vicinity of the curie point @xmath8 ) the temperature behavior of the resistance is metallic in character .
the resistance @xmath37 at @xmath21 is about 1315 @xmath38 ( this corresponds to the resistivity @xmath39
@xmath38 cm ) ; whereas , already at @xmath40 k the resistance @xmath41 is much less ( 178 @xmath38 ) , and at @xmath42k the resistance has decreased to @xmath43 @xmath38 ( @xmath44 @xmath38 cm ) .
we have taken @xmath45/r(h)$ ] at a magnetic field @xmath46 koe as a measure of the magnetoresistance .
it can be seen from fig.3 that @xmath47 has its maximum value ( about 66 % ) at a characteristic temperature @xmath48 k ( @xmath49 is also near @xmath8 for these manganites ) .
after the first irradiation with a fluence @xmath50 @xmath6 the above mentioned parameters have changed to the following values : @xmath51 k , @xmath52 @xmath38 , @xmath53 k , @xmath54 % , @xmath55 , @xmath56 ( figs . 2 and 3 ) .
after a second irradiation ( the total fluence after two irradiations is about @xmath5 @xmath6 ) these parameters are : @xmath57 k , @xmath58 @xmath38 , @xmath59 k , @xmath60 % , @xmath61 , @xmath62
. it can be seen from these results that the electron irradiation has produced a rather large effect on film resistance .
the film resistance in the paramagnetic insulating state ( above @xmath21 ) has increased over 25 % .
more striking is the change in @xmath10 in the ferromagnetic state at low temperature : @xmath63 is tripled by the electron irradiation . at the same time
( taking into account the experimental errors ) there is no substantial changes of the cmr characteristics : the values of @xmath21 , @xmath49 ( and thus @xmath8 ) decrease only about 5 k ; whereas , the magnitude of the magnetoresistance @xmath47 remains practically unchanged . in discussion and analysis of the results obtained it is important to determine the degree of radiation damage produced by the electron irradiation in our study .
the types of defects produced by electron irradiation are frenkel pairs , i.e. isolated vacancies and interstitials .
the atomic displacement cross sections by fast electrons and the corresponding values of dpa for all elements ( la , ca , mn , and o ) of the sample have been calculated taking into account the exact chemical composition of the film and using the well - known fundamental concepts of such type of relativistic calculations @xcite and the cascade calculational procedures outlined in ref . together with the ratios of the mott to the rutherford cross section @xmath64 .
the results of this type of calculation depend essentially on the specified value of the threshold energy @xmath65 ( an atom which receives energy @xmath66 will be displaced certainly from its lattice site@xcite ) which was chosen to be @xmath67 ev for all ions , the typical value of @xmath65 in common use for this type of calculation . at the total fluence @xmath68 @xmath6
the calculations result in the following values of dpa for the chemical elements which comprise this film : @xmath69 ( la ) , @xmath70 ( ca ) , @xmath71 ( mn ) , @xmath72 ( o ) . the total dpa is about @xmath73 .
one should not take these values literally . as mentioned above
, the output of such calculations depends essentially on the values of energy @xmath65 , which are obscure and which may be quite different for the different constituent elements .
nevertheless , we believe , based on previous studies@xcite , the calculation results should be correct at least to the order of the magnitude .
the experimental results correlate , at least qualitatively , with the de model @xcite . in this model
the ferromagnetic coupling between pairs of mn@xmath16 and mn@xmath15 ions through the oxygen ions is also responsible for the metallic properties of the manganites .
the electron hopping amplitude @xmath74 from site @xmath75 to site @xmath76 is given by @xmath77 where @xmath78 is a material - dependent constant , @xmath79 is the angle between the directions of two ionic spins .
it can be seen from eq . [ 1 ] that in the de model a clear connection exists between electron transport and magnetic order , i.e. the electron conduction is a function of magnetic order .
the angle @xmath79 decreases below @xmath8 or in a magnetic field .
this may be a possible reason for cmr .
the disorder ( for example , vacancies ) must reduce the coupling between the mn@xmath16o
mn@xmath15 ions and , therefore , the probability of electron transfer .
this must cause the increase in resistivity . at the same time the disorder should influence the ferromagnetic order ( the curie temperature @xmath8 must go down ) . therefore , the increase of disorder must induce simultaneously an increase of resistance and decrease of @xmath8 in the de model , that qualitatively corresponds to our results and the results of previous studies with ion - irradiated manganites@xcite .
it is usually assumed that @xmath78 in eq . [
1 ] is a constant for all lattice cells , which can be correct only in perfect crystals .
it was taken into account in ref . that in disordered crystals @xmath78 is a position - dependent quantity and denotes a static disorder . the numerical simulations in ref .
in the frame of the model for disorder - induced polaron formation@xcite have shown that increasing static disorder decreases the values of @xmath74 and leads to a metal - insulator transition as observed in refs . . the general approach of ref . ( to look beyond the de model and
take into consideration additional important effects ) seems to be quite fruitful .
the proper consideration and interpretation of the irradiation - disorder influence is possible , however , only if the exact conduction mechanisms in the insulating and high - conducting ferromagnetic regimes of the doped manganites are known . at the moment
there is still no clear enough understanding of these mechanisms .
nevertheless the experimental and theoretical achievements in this matter in the last years@xcite enable such an attempt .
some general observations should be noted .
the magnitude of the resistance increase near and above @xmath21 ( about 25 % ) at first sight does not arouse great surprise , since semiconductors with a very small concentration of charge carriers are generally very sensitive to irradiation that produces displacement atoms in the crystal .
the irradiation defects quite often cause the reduction of charge carrier concentration and mobility@xcite .
the charge carrier concentration in doped manganites is not , however , very small .
based on the chemical doping the charge carrier concentration in la@xmath0ca@xmath1mno@xmath2 ( @xmath3 ) should be about 0.33 holes per unit cell , a density of carriers @xmath80 @xmath81 ( for the cubic cell with lattice parameter about 0.385 nm ) . in hall - effect studies of this compound
it was found that in ferromagnetic state below @xmath33 the charge - carrier density should be in the range 0.85 - 1.9 per unit cell@xcite . even higher value ( 2.4 )
was found in ref . for la@xmath82(ca , pb)@xmath83mno@xmath24 .
the reasons for such high values ( which deviate much from the nominal doping level ) is not clear at present@xcite . because of this we will assume that charge - carrier density in the ferromagnetic state corresponds roughly to 0.33 holes per unit cell ( @xmath80 @xmath81 ) . in the paramagnetic state
not all the dopants contribute to the charge carrier density .
part of the doped holes may be localized@xcite .
indeed , it follows from the hall - effect measurements above @xmath33 that in the paramagnetic insulating state the charge - carrier density is much lower , namely , in the range from 0.004 to 0.5 holes per unit cell@xcite .
we can rather safely assume that charge - carrier density below @xmath33 decreases by at least a factor of five .
this corresponds approximately to the value @xmath84
@xmath81 which can be used for numerical evaluations . in the case of a semiconductor with activated conductivity due to a band gap or mobility edge
this value appears to be too high to understand how the @xmath7 dpa can produce this rather appreciable ( about 25 % ) resistivity increase .
indeed , it is easy to see that even if each of the displaced ions produces a trap for the mobile charge carrier , the traps can lead to localization of only about @xmath85 carrier/@xmath86 which is much less than estimated carrier density
. therefore , the explanation based on the reduction in charge carrier density , which is quite usual for semiconductors@xcite , can not explain the observed irradiation induced resistance increase for these manganites .
it is even more difficult to explain how such a low dpa can induce the observed threefold increase in the resistivity in the metallic ferromagnetic state at low temperature ( fig .
it is known@xcite , that 1% of displaced atoms
( that is 0.01 dpa ) in the noble metals like au , ag or cu result in a change of resistivity of about 1 @xmath87 cm .
such small changes practically could not be experimentally distinguished for these rather high - resistance manganite films .
it follows that additional assumptions which take into account the peculiarities of the insulating and metallic states and the nature of the charge carriers in doped manganites are needed to explain the experimental results . for the insulating state of the doped
manganites it is essential to take into account the polaronic nature of charge carriers in them ( see refs . and references therein ) .
the introduction to polaron physics and the main references can be found , for example , in ref . .
it is rather commonly assumed that the conductivity of doped manganites above @xmath8 is determined by small polarons@xcite .
the exact nature of these small polarons is now the object of intensive theoretical and experimental studies .
the different kinds of lattice or magnetic polarons are considered .
it is widely accepted that at decreasing temperature in the region of @xmath8 the crossover from localized small polarons to itinerant large polarons takes place@xcite .
this point of view has found experimental support@xcite . according to the definition@xcite
, a lattice polaron is the unit consisting of the `` self - trapped '' ( localized ) charge carrier , together with its induced lattice deformation .
the polaron is called small when the spatial extent of the wave function of the trapped charge carrier is comparable with the separation of next - neighbor ions .
the polaron radius @xmath88 for small polarons in doped manganites is estimated to be about 0.5 nm@xcite .
small polarons have a large effective mass ( 10 - 100 larger than mass of free electron ) and can move by tunneling or thermally activated hopping .
the mobility of the small polaron is very low because the charge carrier movement includes the displacements of atoms surrounding it . for any conductor
the conductivity @xmath89 is given by the general relation @xmath90 , where @xmath91 is density of carriers and @xmath92 is mobility .
in contrast to band semiconductors in which @xmath91 can depend on temperature in a thermally activated way , the density of carriers is assumed to be constant with temperature for polaronic conductors . at fairly high temperatures @xmath93 ( where @xmath94 is the debye temperature ) in the adiabatic limit@xcite ( which is assumed to be true for the doped manganites@xcite ) it is the small polaron mobility that is activated and the resistivity @xmath95 is given by @xmath96 where @xmath97 is the activation energy , with @xmath98 the polaron binding energy and @xmath99 the overlap integral ; @xmath100 is the hopping distance , and @xmath101 is the optical - phonon frequency .
[ 2 ] is true in the dilute , noninteracting limit , when the density of carriers is far less than the density of equivalent hopping sites@xcite .
it may be assumed as in ref . , that in doped manganites all the carriers form polarons . in this case with the above - estimated value of charge carrier density in the insulating paramagnetic state ( @xmath84 @xmath81 ) the mean distance @xmath102 between the trapped charge carriers is @xmath103 nm . since it is assumed@xcite in the general case that the hopping distance @xmath100 is equal to a lattice constant , the noninteracting limit is quite justified for these doped manganites .
for the value of dpa in this study ( about @xmath104 ) the mean distance @xmath105 between the damage lattice sites is about 6 nm . in ref .
a much larger dpa ( about 0.01 ) was produced by ion irradiation .
this resulted in a tenfold increase in the resistivity in the insulating state , as compared to the approximately @xmath106 increase shown in fig . 2 for electron irradiation .
in that experiment the length @xmath105 would be approximately 1.0 nm .
the effect of radiation damage in the insulating state of doped manganites can be understood , at least qualitatively , by taking into account the small - polaronic nature of charge carriers .
two main sources of radiation influence on small polaron conduction in doped manganites can be seen .
first , according to ref .
, for the crystals with not too strong an electron - lattice interaction it is quite possible that some appreciable number of carriers would be quasifree rather than small polarons .
this should be true for the doped manganites since many experimental and theoretical studies indicate@xcite the coexistence of localized and itinerant carriers in a rather wide temperature range near @xmath8 . in this case
the disorder can convert some of the available quasifree states to small - polaron states@xcite .
that is , disorder reduces the strength of the electron - lattice coupling needed to stabilize the global small - polaron formation . defects and impurities serve as centers for electron localization and small - polaron formation .
this explanation is supported by the numerical simulations in ref . .
this mechanism of the disorder - induced conductivity decrease may be dominant near @xmath8 . in ion - irradiation experiments@xcite
much larger dpa values ( up to 0.01 and more ) have been produced which have resulted in an increase in resistance in the insulating paramagnetic state by one ( and sometimes two ) order of magnitude .
this effect is accompanied by an increase in the activation energy @xmath107 ( see eq .
[ 2 ] ) and a decrease in peak temperature @xmath21 . in this case , especially at temperatures rather far above @xmath8 , it is not possible to explain the resistance increase only by the transformation of available quasi - free carriers to small polarons .
these results demonstrate that the disorder influences directly the charge - carrier hopping and leads to a decrease in the charge - transfer probability .
there appear to be no specific theoretical treatments of this problem for small - polaron hopping .
it is known that at high temperatures polaron jumps occur when electron energies associated with the initial and final sites ( these energies are determined by a configuration of lattice atoms ) are equal@xcite .
maybe disorder affects these so called coincidence events in such way that it leads to a decrease in transfer probability .
it should be taken into account also the possible influence of anderson localization@xcite .
it is evident that more experimental and theoretical efforts are needed to clarify this problem .
the foregoing discussion indicates that an adequate consideration of radiation - damage effects on conductivity is possible only in the frame of a rather strictly determined conduction mechanism and charge - carrier nature .
unfortunately , no determination has been made for the ferromagnetic high - conducting state of doped manganites well below @xmath8 .
at least one assumption for this state is , however , clear : the charge carriers at low temperatures can be considered to be quasifree .
it has been argued@xcite that the charge carriers in this state are itinerant large polarons .
the polaron of this type@xcite moves without thermal activation and behaves like a heavy particle ( with mass in 2 - 4 times larger than mass of free electron ) .
another possibility is that the doped manganites below @xmath8 are just degenerate semiconductors@xcite . in any case the doped manganites in the ferromagnetic state with a minimal resistivity of about 100 @xmath87 cm should be considered as some kind of `` bad '' metal , like heavily doped semiconductors or amorphous metals . for such conductors
it is quite difficult ( and sometimes of no use ) to estimate a value of the electron mean - free path @xmath108 and consider the decrease of @xmath108 under influence of irradiation - induced disorder .
indeed , for a fermi velocity @xmath109 m / s ( as was calculated in ref . for la@xmath110ca@xmath111mno@xmath2 )
the use of the quasifree - electron relation @xmath112 with @xmath91 about 1.0 hole per unit cell and @xmath113 given by the mass of a free electron , gives @xmath114 nm . with an effective mass @xmath115 , @xmath116 nm . the films in this experiment are not single - crystal , but they do consist of rather large grains with a size near 0.5 @xmath92 m .
therefore , the `` intrinsic '' value of @xmath108 within the grains determined in this model should be larger .
it is inconceiveable , however , that such considerations with a mean distance @xmath105 between the damage sites of about 6 nm could explain the threefold increase in the resistivity of such a rather `` bad '' metal with an electron mean - free path on the order of 1 nm . the unusual magnetic behavior of the doped manganites suggests a possible phenomenological explanation of the large effect of small radiation damage on the resistance in the ferromagnetic metallic regime .
irradiation not only leads to lattice disorder that can lead to elastic electron scattering as in normal non - ferromagnetic metals , but it also perturbes the long - range ferromagnetic order . in the manganites
the conductivity increases with the enhancement of ferromagnetic order .
indeed , that is the source of the huge resistivity decrease at the paramagnetic - ferromagnetic transition and the cmr .
below @xmath33 an unusual correlation between resistivity and magnetization @xmath117 has been reported @xcite .
for example , in ref .
the following experimental relation between @xmath118 and @xmath117 for the la@xmath119ca@xmath120mno@xmath2 films was found @xmath121 where @xmath122 and @xmath123 are sample - dependent parameters . at present
there is no clear theoretical understanding of this correlation between @xmath118 and @xmath124 .
it is generally accepted that the increase in @xmath124 should lead to delocalization of the holes and to the increase in hole mobility . in any case , however , it is clear that doped manganites are not conventional ferromagnetic metals even well below @xmath8 , and that electronic transport in them is influenced to a high degree by magnetic order@xcite .
a reasonable hypothesis is that the dominant effect of irradiation on the resistivity of the doped manganites at low temperature in the ferromagnetic phase comes primarily from the disruption of long range magnetic order , perhaps through the magnetoelastic coupling that produces magnetostriction .
indirect evidence for this is provided by the observation that ion irradiation induces a considerable decrease in the saturation magnetization value @xmath125@xcite .
for example , in ref . for an ion irradiation dose which has resulted in the nearly same dpa ( @xmath126 ) as in the present study , the saturation magnetization decreased by about @xmath127
. further indirect evidence of the influence of disorder effects on @xmath125 is provided by the three - fold decrease in @xmath125 with only a small shift in @xmath33 that was associated with a decrease in grain size from 110 to 20 nm in bulk samples@xcite .
unfortunately , the additional experimental facilities needed to test this hypothesis were not available for this experiment , but its discussion may lead to future tests of the hypothesis and generate new interest in irradiation damage studies as a way to probe the fundamental nature of conduction in these exotic materials . in conclusion , the high - energy electron irradiation effect on the transport properties of la@xmath0ca@xmath1mno@xmath2 films ( @xmath22 ) has been investigated .
comparatively small electron fluences used in this study do not have any substantial influence on the curie temperature @xmath8 or the magnitude of the magnetoresistance . at the same time
these fluences result in an appreciable increase in film resistivity in both the insulating paramagnetic state and especially in the highly conductive ferromagnetic state .
the relative resistivity increase in the metallic ferromagnetic state ( below @xmath8 ) was found to be much ( an order of magnitude ) greater than that in the insulating paramagnetic state .
this behavior is quite different from that associated with non - magnetic metals and semiconductors and can be understood in the high - temperature regime qualitatively by taking into account the polaronic nature of manganite s conductivity above and near @xmath8 .
a possible explanation for the low temperature behavior has been suggested , but it must be tested with magnetization measurements that were not available to the present experiments . the work at texas a&m is supported by the robert a. welch foundation ( a-0514 ) , the texas advanced technology program ( grant no .
010366 - 141 ) , and the texas center for superconductivity at the university of houston ( tcsuh ) .
bib and dgn acknowledge support by nato scientific division ( collaborative research grant no .
the authors wish to thank i. lyuksyutov , v. pokrovsky and e. l. nagaev for useful discussions and c. w. chu at tcsuh for x - ray diffraction of the samples .
the authors are especially grateful to s. m. seltzer of the national institute of standards and technology for the calculation of the mott / rutherford ratio @xmath64 for the la - ca - mn - o compounds and to r. m. stroud for copies of her work before publications .
e - mail : [email protected] e - mail : [email protected] r. von helmont , j. wecker , b. holzapfel , l. schultz , and k. samwer , phys .
* 71 , * 2331 ( 1993 ) .
s. jin , t. h. tiefle , m. mccormack , r. a. fastnacht , r. ramesh , and l. h. chen , science * 264 , * 413 ( 1994 ) ; s. jin , m. mccormack , t. h. tiefel , and r. ramesh , j. appl .
* 76 , * 6929 ( 1994 ) .
e. l. nagaev , physics
uspekhi * 39 , * 781 ( 1996 ) .
j. m. d. coey , in _ aspects of modern magnetism _ , edited by f. c. pu , y. j. wang and c. h. shang ( world scientific , singapore , 1996 ) , ch
a. p. ramirez , j. phys . ,
matter * 9 , * 8171 ( 1997 ) .
d. i. khomskii and g. a. sawatzky , solid state commun .
* 102 , * 87 ( 1998 ) . c. zener , phys
* 82 , * 403 ( 1951 ) .
p. w. anderson and h. hasegawa , phys .
* 100 , * 675 ( 1955 ) .
p. g. de gennes , phys
* 118 , * 141 ( 1960 ) .
a. j. millis , p. b. littlewood , and b. i. shraiman , phys .
* 74 , * 5144 ( 1995 ) ; a. j. millis , r. mueller , and b. i. shraiman , phys .
b * 54 , * 5389 ( 1996 ) ; _ ibid _ 5405 ; a. j. millis , nature * 392 , * 147 ( 1998 ) .
l. p. gorkov , physics - uspekhi , * 41 , * 589 ( 1998 ) .
a. s. alexandrov and a. m. bratkovsky , preprint , cond - mat/9806030 . m. croft , d. sills , m. greenblatt , c. lee , s .- w .
cheong , k.v .
ramanujachary , and d. tran , phys .
b * 55 , * 8726 ( 1997 ) .
t. saitoh , a , bocquet , t. mirokawa , h. namatome , a. fujimori , m. abbade , y. takeda , and m. takano , phys . rev .
b * 51 , * 13942 ( 1995 ) . h. l. ju , h .- c . sohn , and kannan m. krishnan , phys .
* 79 , * 3230 ( 1997 ) .
chen , v. talyansky , c. kwon , m. rajeswari , r. p. sharma , r. ramesh , t. venkatesan , j. melngailis , z. zhang , and w. k. chu , appl .
* 69 , * 3089 ( 1996 ) .
m. l. wilson , j. m. byers , p. c. dorsey , j. s. horwitz , d. b. chrisey , and m. s. osofsky , j. appl . phys . * 81 , * 4971 ( 1997 ) .
r. m. stroud , v. m. browning , j. m. byers , d. b. chrisey , w. w. fuller - mora , k. s. grabowski , j. s. horwitz , j. kim , d. l. knies , and m. s. osofsky , in materials research society symposium proceeding , vol . 474 , p.185 ( materials research society , 1997 ) ; preprint ( submitted to phys .
lett ) ; v. m. browning , r. m. stroud , w. w. fuller - mora , j. m. byers , m. s. osofsky , d. l. knies , k. s. grabowski , d. koller , j. kim , d. b. chrisey and j. s. horwitz , j. appl . phys .
* 83 , * 7072 ( 1998 ) .
s. p. ogale , k. ghosh , .
y. gu , r. shreekala , s. r. shinde , m. downes , m. rajeswari , r. p. sharma , r. l. greene , t. venkatesan , r. ramesh , ravi bathe , s. i. patil , r. ravikumar , s. k. arora , and g. k. mehta , j. appl .
* 84 , * 6255 ( 1998 ) .
p. s. wang and d. g. naugle , materials letters * 29 * , 205 ( 1996 ) .
f. m. araujo - moreira , m. rajeswari , a. goyal , k. gosh , v. smolyaninova , t. venkatesan , c. l. lobb , and r. greene , appl .
* 73 , * 3456 ( 1998 ) .
k. d. d. rathnayaka , a. parasiris , c. andrews , j. chepin , a. k. bhatnagar , a. c. dumar , and d. g. naugle , materials letters * 37 , * 21 ( 1998 ) .
f. seitz and j. s. koehler , in _ solid state physics , _ edited by f.
seitz and d. turnbull ( academic , new york , 1956 ) , vol.2 , p.305 .
g. j. dienes and g. h. vineyard , radiation effects in solids ( interscience , new york , 1957 ) .
o. s. oen , us atomic energy comission report ornl-4897 ( 1973 ) .
d. emin and m. n. bussac , phys . rev . * b49 , * 14290 ( 1994 ) . j. w. corbett , electron radiation damage in semiconductors and metals , _ solid state physics , _ edited by f.
seitz and d. turnbull ( academic , new york , 1966 ) , suppl.7 .
g. j. snyder , m. r. beasley , t. h. geballe , r. hiskes , and s. dicarolis , appl .
* 69 , * 4254 ( 1996 ) .
g. jacob , f. martin , w. westenburg , and h. adrian , phys . rev .
* b57 , * 10252 ( 1998 ) .
s. h. chun , m. b. salamon , and p. d. han , j. appl .
* 85 , * 5573 ( 1999 ) .
j. e. nez - regueiro , d. gupta , and a. m. kadin , j. appl . phys .
* 79 , * 5179 ( 1996 ) .
k. li , r. cheng , x. chen , j. fan , x. cao , and y. zhang , j. appl . phys . * 84 , * 1467 ( 1998 ) .
h. rder , j. zang , and a. r. bishop , phys .
* 76 , * 1356 ( 1996 ) ; a. r. bishop and h. rder , in _ optical and magnetic materials _ edited by w. a. gambling and o. kahn , current opinion in solid state & materials science , v.2 , no 2 , april 1997 , p.244
. m. r. ibarra , j. m. de teresa , j. magn .
magn . matter . * 177 - 181 , * 846 ( 1998 ) .
m. f. hundley , j. j. neumeier , r. h. heffner , q. x. jia , x. d. wu , and j. d. thompson , j. appl .
* 79 , * 4535 ( 1996 ) ; m. f. hundley , j. j. neumeier , r. h. heffner , q. x. jia , x. d. wu , and j. d. thompson , in materials research society symposium proceeding , vol .
474 , p.167 ( materials research society , 1997 ) .
s. j. billinge , r. d. difrancesco , g. h. kwei , j. j. neumeier , and j. d. thompson , phys .
lett . , * 77 , * 715 ( 1996 ) ; m. jaime , h. t. hardner , m. b. salamon , m. rubinstein , p. dorsey , and d. emin , phys .
lett . , * 78 , * 951 ( 1997 ) .
m. ziese and c. srinitiwarawong , phys . rev .
* b58 , * 11519 ( 1998 ) .
i. g. austin and n. f. mott , adv . phys . * 18 , * 41 ( 1969 ) ; a. s. alexandrov and n. mott , polarons and bipolarons ( world scientific , singapore , 1995 ) ; t. holstein , ann . phys .
* 8 , * 325 ( 1959 ) ; d. emin and t. holstein , ann . phys .
* 53 , * 439 ( 1969 ) ; d. emin , phys . today , p. 34 , june 1992 .
in our discussion we will assume that small or large polarons , as well as lattice or magnetic polarons are charge carriers according to the definitions in ref . .
at the present time in some scientific papers the term `` large polaron '' is sometimes used as a name for ferromagnetic droplets or clusters resulting from phase - separation or inhomogeneity effects in the doped manganites .
a. lanzara , n. l. saini , m. brunelli , f. natali , a. bianconi , p. g. radaelli , and s .- w .
cheong , phys .
* 81 , * 878 ( 1998 ) ; k. h. kim , j. h. jung , and t. w. noh , phys .
lett . * 81 , * 1517 ( 1998 ) ; s. yoon , h. l. liu , g. scholler , s. l. cooper , p. d. han , d. a. payne , s .- w .
cheong , and z. fisk , phys . rev .
* b58 , * 2795 ( 1998 ) ; a. machida , y. morimoto , and a. nakamura , phys . rev .
* b58 , * r4281 ( 1998 ) .
d. c. worledge , l. miville , and t. h. geballe , j. appl .
* 83 , * 5913 ( 1998 ) . c. h. booth , f. bridges , g. h. kwei , j. m. lawrence , a. l. cornelius , and j. j. neumeier , phys .
* 57 , * 10440 ( 1998 ) .
d. louca , t. egami , e. l. brosha , h. rder , and a. r. bishop , phys . rev .
* b56 , * r8475 ( 1997 ) . i. f. lyuksyutov and v. pokrovsky , modern phys . lett .
* b13 , * 379 ( 1999 ) .
w. e. pickett and d. j. singh , phys . rev .
* b55 , * r8642 ( 1997 ) .
r. h. heffner , l. p. le , m. f. hundley , j. j. neumeier , g. m. luke , k. kojima , b. nachumi , y. j. uemura , d. e. maclaughlin , and s - w .
cheong , phys .
* 77 , * 1869 ( 1996 ) .
r. d. snchez , j. rivas , c. vzquez - vzquez , a. lpez - quintella , m. t. causa , m. tovar , and s. oseroff , appl .
* 68 , * 134 ( 1996 ) . | the effect of crystal lattice disorder on the conductivity and colossal magnetoresistance in la@xmath0ca@xmath1mno@xmath2 ( @xmath3 ) films has been examined .
the lattice defects are introduced by irradiating the film with high - energy ( @xmath4 mev ) electrons with a maximal fluence of about @xmath5 @xmath6 .
this comparatively low dose of irradiation produces rather small radiation damage in the films .
the number of displacements per atom ( dpa ) in the irradiated sample is about @xmath7
. nevertheless , this results in an appreciable increase in the film resistivity .
the percentage of the resistivity increase in the ferromagnetic metallic state ( below the curie temperature @xmath8 ) was much greater than that observed in the insulating state ( above @xmath8 ) .
at the same time irradiation has much less effect on @xmath8 or on the magnitude of the colossal magnetoresistance .
a possible explanation of such behavior is proposed .
# 1@xmath9#1 |
channel polarization was introduced in @xcite for binary input discrete memoryless channels as a coding technique to construct codes called polar codes for data transmission .
polar codes are capable of achieving the ` symmetric capacity ' of any binary input channel , using low - complexity encoding and decoding algorithms . in terms of the block - length @xmath0
, polar codes can be encoded and decoded in complexity @xmath1 and achieve a block error probability that decays roughly like @xmath2 .
the latter result was shown in @xcite .
the aim of this note is to extend these results of @xcite to dmcs with @xmath3-ary inputs for any finite integer @xmath4 . to that end , we recall the polarization construction and outline how the results above were shown .
given a binary input channel @xmath5 with @xmath6 define its _ symmetric capacity _ as @xmath7 @xmath8 is nothing but the mutual information developed between the input and the output of the channel when the input is uniformly distributed . in @xcite ,
two independent copies of @xmath9 are first combined and then split so as to obtain two _ unequal _ binary input channels @xmath10 and @xmath11 .
the channel @xmath12 describes two uses of the channel @xmath9 , @xmath13 the input @xmath14 to the channel @xmath15 are put in one - to - one correspondence with @xmath16 via @xmath17 , @xmath18 , thus obtaining the _ combined _ channel @xmath19 described by @xmath20 the _ split _ is inspired by the chain rule of mutual information : let @xmath21 be random variables corresponding to their lowercase versions above . if @xmath22 are independent and uniformly distributed , then so are @xmath23 and consequently , on the one hand , @xmath24 the split channels @xmath10 and @xmath11 describe those that occur on the right hand side of the equation above : @xmath25 so that @xmath26 and @xmath27 . the polarization construction given in @xcite
is obtained by a repeated application of @xmath28 . since both @xmath10 and @xmath11 are binary input channels , one can obtain @xmath29 , @xmath30 , @xmath31 , and @xmath32 .
after @xmath33 levels of application , one obtains @xmath34 channels @xmath35 .
the main observation in @xcite is that these channels polarize in the following sense : [ prop : bipolar ] for any @xmath36 , @xmath37 in other words , except for a vanishing fraction , all the channels obtained at level @xmath33 are either almost perfect , @xmath38 , or almost pure noise , @xmath39 .
as the equality @xmath40 leads by induction to @xmath41 , one then concludes that the fraction of almost perfect channels approaches the symmetric capacity .
this last observation is the basis of what lets @xcite conclude that polar codes achieve the symmetric capacity .
we give here a new proof of this proposition because it will readily generalize to the @xmath3-ary input case we will discuss later . before we embark on this proof , we introduce the bhattacharyya parameter for a binary input channel @xmath5 , defined by @xmath42 the relationship between @xmath43 , @xmath44 , @xmath45 and @xmath8 is already discussed in @xcite , where the following is shown : [ lem : iz ] 1 .
@xmath46 , 2 .
@xmath47 , 3 .
@xmath48 , 4 .
@xmath49 .
proposition [ prop : bipolar ] was proved in @xcite for the binary case ( @xmath50 ) using lemma [ lem : iz ] .
unfortunately , lemma [ lem : iz ] does not generalize to the non - binary case ( @xmath51 ) .
the following alternate proof of proposition [ prop : bipolar ] uses less stringent conditions that can be fulfilled for all @xmath52 . [
lem : conv ] suppose @xmath53 , @xmath54 are i.i.d . , @xmath55-valued random variables with @xmath56 defined on a probability space @xmath57 . set @xmath58 as the trivial @xmath59-algebra and set @xmath60 , @xmath61 to be the @xmath59-field generated by @xmath62 .
suppose further that two stochastic processes @xmath63 and @xmath64 are defined on this probability space with the following properties : * @xmath65 takes values in the interval @xmath66 $ ] and is measurable with respect to @xmath60 .
that is , @xmath67 is a constant , and @xmath65 is a function of @xmath68 . *
@xmath69 is a martingale .
* @xmath70 takes values in the interval @xmath66 $ ] and is measurable with respect to @xmath60 . *
@xmath71 when @xmath72 .
* for any @xmath73 there exists @xmath36 such that @xmath74 implies @xmath75 .
then , @xmath76 exists with probability 1 , @xmath77 takes values in @xmath78 , and @xmath79 .
the almost sure convergence of @xmath65 to a limit follows from @xmath80 being a bounded martingale .
once it is known that @xmath77 is @xmath78-valued it will then follow from the martingale property that @xmath81=i_0 $ ] .
it thus remains to prove that @xmath77 is @xmath78-valued .
this , in turn , is equivalent to showing that for any @xmath82 , @xmath83 since for any @xmath84 , the event @xmath85 is included in the event @xmath86 and since by property ( i&t.1 ) there exists @xmath36 such that @xmath87 where @xmath88 it suffices to prove that @xmath89 for any @xmath36 .
this is trivially true for @xmath90 .
therefore , it suffices to show the claim for @xmath91 .
given such a @xmath92 , find a positive integer @xmath93 for which @xmath94 .
this choice of @xmath93 guarantees that if a number @xmath95 $ ] is squared @xmath93 times in a row , the result lies in @xmath96 . for @xmath61
define @xmath97 as the event that @xmath98 , i.e. , @xmath97 is the event that there are @xmath93 consecutive @xmath99 s in the sequence @xmath100 starting at index @xmath33 . note that @xmath101 , and that @xmath102 is a collection of independent events . the borel
cantelli lemma thus lets us conclude that the event @xmath103 has probability 1 , and thus @xmath104 .
we will now show that @xmath105 is empty , from which it will follow that @xmath89 . to that end , suppose @xmath106 .
since @xmath107 , there exists @xmath108 such that @xmath109 whenever @xmath110 .
but since @xmath111 there exists @xmath112 such that @xmath113 , and thus @xmath114 which contradicts with @xmath115 .
the proof of lemma [ lem : conv ] uses property ( t.2 ) only in the way that repeated squarings of a number in @xmath116 will eventually fall outside @xmath116 .
thus , condition ( t.2 ) may be replaced by any other that has this property .
e.g. , conditioned on @xmath60 , _ at least one _ of the two values of @xmath117 satisfies @xmath118 for a nondecreasing @xmath119 having the property that for any @xmath36 , there exists @xmath93 such that @xmath120 . here
@xmath121 denotes @xmath93-fold composition of @xmath119 .
let @xmath122 be i.i.d .
, @xmath55-valued random variables taking the two values with equal probability , as in lemma [ lem : conv ] .
define @xmath123 and @xmath124 these processes satisfy the conditions of lemma [ lem : conv ] : ( i.1 ) is trivially true with @xmath125 ; the martingale property ( i.2 ) follows from @xmath40 ; ( t.1 ) is again trivially true ; ( t.2 ) follows from lemma [ lem : iz](i ) ; ( i&t.1 ) follows from lemma [ lem : iz](iii ) and ( iv ) .
thus , the process @xmath65 converges with probability 1 to a @xmath78-valued random variable .
this implies that @xmath126 note that the distribution of @xmath62 is the uniform distribution on @xmath127 .
thus , @xmath128 and proposition [ prop : bipolar ] follows .
the following lemma was proved in @xcite .
[ lem : rate ] suppose that the processes @xmath129 , @xmath80 and @xmath130 , in addition to the conditions ( i.1 ) , ( i.2 ) , ( t.1 ) , ( t.2 ) and ( i&t.1 ) in lemma [ lem : conv ] , also satisfy * for some constant @xmath131 , @xmath132 when @xmath133 .
* for any @xmath73 there exists @xmath36 such that @xmath134 implies @xmath135 .
then , for any @xmath136 @xmath137 note that in the proof of proposition [ prop : bipolar ] the random variable @xmath70 denotes the bhattacharyya parameter of a randomly chosen channel after @xmath33 steps of polarization .
therefore , lemma [ lem : rate ] states that after @xmath33 steps of polarization , almost all ` good ' channels will have bhattacharyya parameters that are smaller than @xmath138 for any @xmath139 , provided that @xmath33 is sufficiently large .
since the bhattacharyya parameter is an upper bound to the error probability of uncoded transmission , this implies that , at any fixed coding rate below @xmath125 , the block error probability @xmath140 of binary polar codes under successive cancellation decoding will satisfy @xmath141 when the block - length @xmath142 is sufficiently large .
in this section we will show how the transformation @xmath143 ( and consequently @xmath144 ) and the definition of @xmath43 can be modified so that the hypotheses of lemmas [ lem : conv ] and [ lem : rate ] are satisfied when the channel input alphabet is not binary .
this will establish that the new transformation satisfies equation , leading to the conclusion that @xmath3-ary polar codes achieve symmetric capacity . that the error probability behaves roughly like
@xmath2 will also follow . to that end , let @xmath3 denote the cardinality of the channel input alphabet @xmath145 and define @xmath146 as the symmetric capacity of a channel @xmath9 .
we will take the base of the logarithm in this mutual information equal to the input alphabet size @xmath3 , so that @xmath147
. for any pair of input letters @xmath148 , we define the bhattacharyya distance between them as @xmath149 here , the notation @xmath150 should be interpreted as denoting the channel obtained by restricting the input alphabet of @xmath9 to the subset @xmath151 .
we also define the average bhattacharyya distance of @xmath9 as @xmath152 the average bhattacharyya distance upper bounds the error probability of uncoded transmission : [ prop : mlerrorbound ] given a @xmath3-ary input channel @xmath9 , let @xmath140 denote the error probability of the maximum - likelihood decoder for a single channel use
. then , @xmath153 let @xmath154 denote the error probability of maximum - likelihood decoding when @xmath155 is sent .
we have , @xmath156 therefore the average error probability is bounded as @xmath157 [ ratereliability ] we have the following relationships between @xmath8 and @xmath43 . @xmath158
\label{eq : ratereliability4 } i(w ) & \le \log ( q/2 ) + ( \log2)\sqrt{1-z(w)^2}\\[.5ex ] \label{eq : ratereliability2 } i(w ) & \le 2(q-1)(\log e)\sqrt{1-z(w)^2}. \end{aligned}\ ] ] proof is given in the appendix .
we will see that when the input alphabet size @xmath3 is a prime number , polarization can be achieved by similar constructions to the one for the binary case . for this purpose
, we will equip the input alphabet @xmath145 with an operation ` @xmath159 ' so that @xmath160 forms a group .
( this is possible whether or not @xmath3 is prime . )
we will let @xmath161 denote the identity element of @xmath160 .
in particular , we may assume that @xmath162 and that ` @xmath159 ' denotes modulo-@xmath3 addition .
note that when @xmath3 is prime , this is the only group of order @xmath3 . as in the binary case
, we combine two independent copies of @xmath9 , by choosing the input to each copy as @xmath163 we define the channels @xmath10 and @xmath11 through @xmath164 where again @xmath165 .
the main result of this section is the following : [ thm : prime ] the transformation described in and polarizes all @xmath3-ary input channels in the sense of proposition [ prop : bipolar ] , provided that @xmath3 is a prime number .
the rate of polarization under this transformation is the same as in the binary case , in the sense that the block error probabilities of polar codes based on this transformation satisfy . to prove theorem [ thm : prime ] we first rewrite @xmath43 as @xmath166 where we define @xmath167
we also define @xmath168 we will use the following lemma in the proof . [
lem : zmax ] given a channel @xmath9 whose input alphabet size @xmath3 is prime , if @xmath169 , then @xmath170 for all @xmath36 .
let @xmath171 be such that @xmath172 , and note that @xmath173 implies @xmath174 for a given @xmath175 define @xmath176 for all @xmath177 .
the triangle inequality states that @xmath178 or equivalently , that @xmath179 on the other hand , since @xmath3 is prime , the input alphabet can be written as @xmath180 for any @xmath181 and @xmath175 .
hence , applying inequality repeatedly yields @xmath182 for all @xmath183 , which implies @xmath184 the proof is similar to the one for the binary case : let @xmath185 be i.i.d .
@xmath55-valued random variables taking the two values with equal probability .
define the random processes @xmath186 and @xmath187 with @xmath125 and @xmath188 .
it suffices to show that @xmath80 and @xmath130 satisfy the conditions of lemmas [ lem : conv ] and [ lem : rate ] : conditions ( i.1 ) , ( i.2 ) , and ( t.1 ) hold trivially .
also , by and in proposition [ ratereliability ] , for any @xmath73 there exists @xmath36 such that @xmath189 furthermore , it follows from lemma [ lem : zmax ] that for any @xmath36 @xmath190),\ ] ] from which ( i&t.1 ) follows . to show ( t.2 ) , we write @xmath191 which implies @xmath192 , or equivalently @xmath71 when @xmath193 .
similarly , one can bound @xmath194 as @xmath195 thus we have @xmath196 , which implies ( t.3 ) .
finally , ( i&t.2 ) follows from and the relation @xmath197 .
the proof of lemma [ lem : zmax ] , and hence of theorem [ thm : prime ] , depends critically on the assumption that @xmath3 is a prime number , and does not extend trivially to the case of composite input alphabet sizes .
in fact , it is possible to find channels that the transformation given in the previous section will not polarize : consider the quaternary - input channel @xmath198 defined by the transition probabilities @xmath199@xmath200 , with @xmath201 . if @xmath9 is combined / split using the transformation described in and , where @xmath159 denotes modulo-@xmath202 addition , then the channels @xmath11 and @xmath10 are statistically equivalent to @xmath9 . therefore @xmath203 . for the general case ,
our first attempt at finding a polarizing transformation is to let @xmath204 where ` @xmath159 ' denotes the group operation , and @xmath205 is a fixed permutation on @xmath145 . in this case
one can compute easily that @xmath206 to be able to mimic the proof of proposition [ prop : bipolar ] one would want that @xmath46 . however , as the value of the inner sum above may depend on @xmath207 , the equality @xmath46 will not necessarily hold in general . as we will see , however , the average value of the above @xmath45 over all possible choices of @xmath205 is @xmath208 .
for this reason , it is appropriate to think of a randomized channel combining / splitting operation , where the randomness is over the choice of @xmath205 . to accomodate this randomness ,
again let @xmath209 denote the independent and uniformly distributed inputs , and let @xmath210 be chosen uniformly at random from the set of permutations @xmath211 , independently of @xmath209 , and revealed to the receiver .
set @xmath212 observe that @xmath213 and that we may define the channels @xmath214 and @xmath215 so that the terms on the right hand side equal @xmath216 and @xmath217 : @xmath218 where @xmath219 .
the transformation described in , , and polarizes all discrete memoryless channels @xmath9 in the sense of proposition [ prop : bipolar ] .
as in the binary case , we will let @xmath185 be i.i.d .
, @xmath55-valued random variables taking the two values with equal probability , and define @xmath220 with @xmath125 and @xmath221 .
we will prove the theorem by showing that the processes @xmath80 and @xmath130 satisfy the conditions of lemma [ lem : conv ] .
since ( i.1 ) , ( i.2 ) , ( t.1 ) are readily seen to hold , and ( i&t.1 ) is implied by inequalities and in proposition [ ratereliability ] , we only need to show ( t.2 ) . to that end
observe that @xmath222 note that for any @xmath223 the value of @xmath224 is equal to @xmath43 , and for any @xmath225 , the value of @xmath226 also equals @xmath43 .
thus , @xmath46 .
as @xmath43 upper bounds the error probability of uncoded transmission ( cf
. proposition [ prop : mlerrorbound ] ) , in order to bound the error probability of @xmath3-ary polar codes it suffices to show that the hypotheses of lemma [ lem : rate ] hold .
since ( i&t.2 ) is already implied by , it remains to show ( t.3 ) : [ prop : z- ] for the transformation described in , , and , we have @xmath227 proof is given in the appendix . we have seen that choosing the transformation @xmath144 in a random fashion from a set of transformations of size @xmath228 yields @xmath46 , leading to channel polarization . in particular , for each @xmath9 there is at least one transformation with @xmath229 .
therefore , randomness is needed only in order to find such transformations at code construction stage , and not for encoding / decoding . in a channel polarization construction of size @xmath0 , there are @xmath230 channels ( @xmath9 , @xmath10 , @xmath11 , @xmath231 , @xmath232 , etc . ) in the recursion tree of code construction . for each channel @xmath233 residing in any one of the @xmath234 internal nodes of this tree , we need to find a suitable permutation @xmath205 such that @xmath235 .
thus , the total complexity of finding the right permutations scales as @xmath236 , in the worst case where all @xmath228 permutations are considered .
recall that polar code construction also requires determining the _ frozen _ coordinates , which is a task of complexity @xmath237 at best .
so , the order of polar code construction complexity is not altered by the introduction of randomization .
the transformation @xmath238 described above uses a random permutation to satisfy @xmath46 .
this amount of randomness over a set of size @xmath228 is in general not necessary , randomization over a set of size @xmath239 is sufficient : if the random permutation @xmath210 that defines is chosen uniformly over the set of permutations for which @xmath161 is a fixed point , the resulting transformation yields @xmath46 and thus is polarizing .
a more significant reduction in randomness can be attained when the input alphabet @xmath145 can be equipped with operations @xmath240 to form an algebraic field this is possible if and only if @xmath3 is a prime power .
a random variable taking only @xmath241 values is sufficient in this case .
( we have already seen that no randomization is needed when @xmath3 is prime . ) to see this , pick @xmath242 to be uniformly distributed from the non - zero elements @xmath243 of @xmath145 , reveal it to the receiver and set @xmath244 as was above we have @xmath245 provided that we define @xmath246 and @xmath247 as @xmath248 the transformation described in , , and polarizes all @xmath3-ary input channels in the sense of proposition [ prop : bipolar ] , provided that @xmath3 is a prime power .
again , we only need to show that @xmath46 . to that end
observe that @xmath249 writing @xmath250 , and @xmath251 , we can rewrite the above as @xmath252 noting that for any fixed @xmath171 , the sum @xmath253 equals @xmath43 , and that the sum @xmath254 also equals @xmath43 yields @xmath46 .
when the field is of odd characteristic ( i.e. , when @xmath3 is not a power of two ) , a further reduction is possible : since @xmath255 is invariant under @xmath256 , one can show that the range of @xmath242 can be reduced from @xmath243 to only half of the elements in @xmath243 , by partition @xmath243 into two equal parts in one - to - one correspondence via @xmath257 , and picking one of the parts as the range of @xmath242 .
it is easy to show that choosing @xmath242 uniformly at random over this set of size @xmath258 will also yield @xmath46 .
when the input alphabet size @xmath3 is not prime , an alternative multi - level code construction technique can be used in order to avoid randomness : consider a channel @xmath9 with input alphabet size @xmath259 , where @xmath260 s are the prime factors of @xmath3 . when the input @xmath261 to @xmath9 is uniformly distributed on @xmath145 ,
one can write @xmath262 , where @xmath263 s are independent and uniformly distributed on their respective ranges @xmath264 . defining the channels @xmath265 through @xmath266 it is easily seen that @xmath267 having decomposed @xmath9 into @xmath268 , one can polarize each channel @xmath269 separately .
the order of successive cancellation decoding in this multi - level construction is to first decode all channels derived from @xmath270 , then all channels derived from @xmath271 , and so on . since the input alphabet size of each channel is prime , no randomization is needed .
a channel @xmath9 is said to be _ equidistant _ if @xmath272 is constant for all pair of distinct input letters @xmath273 and @xmath274 . these are channels with a high degree of symmetry .
in particular , if a channel @xmath9 is equidistant , then so are the channels @xmath11 and @xmath10 created by the deterministic mapping @xmath275 . by similar arguments to those in section
[ subsec : prime ] , it follows that this mapping polarizes equidistant channels , regardless of the input alphabet size . in all of the above
, the input letters of the channel under consideration were used with equal frequency .
this was sufficient to achieve the symmetric channel capacity .
however , in order to achieve the true channel capacity , one should be able to use the channel inputs with non - uniform frequencies in general .
the following method , discussed in @xcite , shows how to implement non - uniform input distributions within the polar coding framework
. given two finite sets @xmath145 and @xmath276 with @xmath277 , any distribution @xmath278 on @xmath145 for which @xmath279 is an integer for all @xmath273 can be induced by the uniform distribution on @xmath276 and a deterministic map @xmath280 .
given a channel @xmath5 , and a distribution @xmath278 as above , we can construct the channel @xmath281 whose input alphabet is @xmath276 and @xmath282 .
then @xmath283 is the same as the mutual information developed between the input and output of the channel @xmath9 when the input distribution is @xmath278 .
consequently , a method that achieves the symmetric capacity of any discrete memoryless channel , such as the channel polarization method considered in this paper , can be extended to approach the true capacity of any discrete memoryless channel by taking @xmath278 as a rational distribution approximating the capacity achieving distribution .
( in order to avoid randomization , one may use prime @xmath108 in the constructions . )
although the discussion above has been restricted to channels with _
discrete _ input and output alphabets , it should be clear that the results hold when the output alphabet is continuous , with minor notational changes . in the more interesting case of channels with continuous input alphabets possibly with input constraints , such as the additive gaussian noise channel with an input power constraint we may readily apply the method of section [ subsec : nonuniform ] to approximate any desired continuous input distribution for the target channel , and thereby approach its capacity using polar codes .
the work of e. arikan was supported in part by the scientific and technological research council of turkey ( tbitak ) under contract no .
107e216 , and in part by the european commission fp7 network of excellence newcom++ ( contract no .
216715 ) . the right hand side ( r.h.s . ) of equals the channel parameter known as _
symmetric cutoff rate_. more specifically , it equals the function @xmath284 defined in gallager ( * ? ? ?
* section 5.6 ) with @xmath285 taken as the uniform input distribution .
it is well known ( and shown in the same section of @xcite ) that the cutoff rate can not be greater than @xmath8 .
this completes the proof of .
let @xmath288 where @xmath285 is the uniform distribution on @xmath145 and @xmath289 clearly we have @xmath290 . by the chain rule , @xmath291 .
now , simple calculations show that @xmath292 and @xmath293 equal the two terms that appear on the right side of .
( intuitively , @xmath294 are the input and output of @xmath9 and @xmath295 is a side information of value @xmath296 supplied by a genie to the receiver . )
note that the summation in can be written as the expectation @xmath297 $ ] where @xmath295 ranges over all distinct pairs of letters from @xmath145 with equal probability .
next , use the form of for @xmath50 ( which is already established in @xcite ) to write @xmath297 \le \log(2)\ , e\left[\sqrt{1-z(w_{\{x_1,x_2\}})^2}\right]$ ] .
use jensen s inequality on the function @xmath298 , which is concave for @xmath299 , to obtain @xmath300 \le \sqrt{1-e[z(w_{\{x_1,x_2\}})]^2}$ ] . since @xmath301 $ ]
, this completes the proof of . for notational simplicity we will let @xmath302 . first note that @xmath303 where @xmath304 is the kullback - leibler divergence . each term in the above summation can be bounded as @xmath305 in the above , the first inequality follows from the relation @xmath306 , and the second inequality is due to @xmath307 .
the @xmath308 distance on the right hand side of can be bounded , using the triangle inequality , as @xmath309 also , it was shown in ( * ? ? ?
* lemma 3 ) that @xmath310 combining the inequalities above , we obtain @xmath311 where the last step follows from the concavity of the function @xmath312 for @xmath313 .
we will show the stronger result that @xmath318 for all @xmath205 , which will imply proposition [ prop : z- ] since @xmath319 . to prove the upper bound on @xmath320 , we write @xmath321 note that @xmath322 for any @xmath323 .
therefore the r.h.s . of
( [ eqn : z2 - 1 ] ) is equal to @xmath43 .
also , note that the innermost sum over @xmath324 in is upper bounded by @xmath325 .
therefore , is upper bounded by @xmath326 . alternatively , noting that for any fixed @xmath327 @xmath328 \\ & \leq q + q(q-1 ) z(w),\end{aligned}\ ] ] we have @xmath329 this in turn implies @xmath330 .
e. arikan , `` channel polarization : a method for constructing capacity - achieving codes for symmetric binary - input memoryless channels , '' _ ieee trans .
inform . theory _
it-55 , pp .
30513073 , july 2009 . | channel polarization , originally proposed for binary - input channels , is generalized to arbitrary discrete memoryless channels .
specifically , it is shown that when the input alphabet size is a prime number , a similar construction to that for the binary case leads to polarization .
this method can be extended to channels of composite input alphabet sizes by decomposing such channels into a set of channels with prime input alphabet sizes .
it is also shown that all discrete memoryless channels can be polarized by randomized constructions .
the introduction of randomness does not change the order of complexity of polar code construction , encoding , and decoding .
a previous result on the error probability behavior of polar codes is also extended to the case of arbitrary discrete memoryless channels .
the generalization of polarization to channels with arbitrary finite input alphabet sizes leads to polar - coding methods for approaching the true ( as opposed to symmetric ) channel capacity of arbitrary channels with discrete or continuous input alphabets .
capacity - achieving codes , channel polarization , polar codes . |
relativistic jets are now believed to be a common occurrence in black hole x - ray binaries ( e.g. corbel 2004 , fender 2004 ) .
the optically thin synchrotron spectra of the discrete ejection events ( the so - called superluminal jets ) imply that the emission becomes much fainter at higher frequencies . for that reason , such events
were only observed at radio frequencies .
however , sams et al .
( 1996 ) reported the detection of extended infrared emission in after a massive ejection event that could be of synchrotron nature .
the most extreme case has been the detection of moving and decelerating x - ray ( and radio ) relativistic jets in the microquasar a few years after the ejection event and on over large distances ( corbel et al .
2002 ; tomsick et al . 2003a ; kaaret et al .
the detection of optically thin synchrotron x - ray emission from discrete ejection events implies in - situ particle acceleration up to several tev .
this acceleration may be caused by interaction of the jets with the interstellar medium ( ism ) .
it now appears that relativistic jets emit throughout the entire electromagnetic spectrum . but are the jets of unique due to special physical conditions or should we expect to observe similar x - ray jets in other x - ray novae ( xne )
? if jets containing tev particles are common amongst xne , then there may be interesting consequences .
the high energy particles could produce a distinctive signature in the cosmic - ray flux ( heinz & sunyaev 2002 ) or produce neutrinos if the jets contain protons ( distefano et al .
furthermore , the study of x - ray jets in microquasars could add an additional bridge between the physics of jets in microquasars and those from supermassive black holes .
interestingly , a connection with gamma - ray burst has also been drawn by wang , dai & lu ( 2003 ) , who showed that the evolution of the eastern x - ray jet of was consistent with the emission from adiabatically expanding ejecta heated by a reverse shock following its interaction with the ism . the sensitivity of current x - ray missions ( in particular _
chandra _ and _ xmm - newton _ ) is sufficient to enable new discoveries in this emerging field and therefore provides further clues regarding the physical nature of these high energy phenomena .
h was discovered with ariel 5 in august 1977 ( kaluzienski & holt 1977 ) and has been precisely localized by heao-1 ( doxsey et al .
it was originally proposed to be a black hole candidate ( bhc ) by white & marshall ( 1983 ) .
h has probably been active several times during the past decades with activity observed in 1984 by exosat ( reynolds et al .
1999 ) and in 1996 by ttm ( emelyanov et al . 2000 ) . in march 2003 ,
integral detected new activity from igr j17464@xmath13213 ( revnivtsev et al .
2003 ) that was later found to correspond to h. after its reactivation in 2003 , a radio counterpart was found with the vla by rupen , mioduszewski & dhawan ( 2003a ) and a bright radio flare ( likely associated with a massive ejection event ) was observed on 2003 april 8 ( mjd 52738 ) by rupen , mioduszewski & dhawan ( 2003b ) . during outburst , h went through several x - ray states with properties typical of bhc .
low and high frequency quasi - periodic oscillations have been detected in h by _ rxte _ ( homan et al . 2005 ; remillard et al . 2005 ) .
_ chandra _ observations , during the 2003 outburst , revealed the presence of narrow and variable absorption lines in the x - ray spectrum of h , possibly related to an ionized outflow ( miller et al . 2005 ) .
the 2003 active phase ( as observed by rxte / asm ) ended around late november 2003 , but a new 4-month outburst started in july 2004 ( swank 2004 ) with a later reactivation at radio frequencies ( rupen , mioduszewski & dhawan 2004 ) .
in this paper , we present an analysis of _ chandra _ x - ray and atca radio observations of h from november 2003 through june 2004 that led to the discovery of large scale radio jets on each side of the bhc h. more importantly , we also report the discovery of x - ray emission is very reminiscent of the large scale decelerating relativistic jets observed in ( corbel et al .
2002 ) . in 2
, we describe the detection and localization of the radio and x - ray sources . in 3 , we discuss the jet kinematics and the nature of the jet emission mechanism .
lllccc date & mjd & source & & angular + & & & r.a . &
decl . & separation + 2003/11/12 & 52955.86 & eastern jet & 17 46 15.978 & -32 14 01.08 & 4.63 @xmath2 0.30 + 2003/11/30 & 52973.72 & eastern jet & 17 46 15.995 & -32 14 00.15 & 4.91 @xmath2 0.30 + 2003/12/09 & 52983.45 & eastern jet & 17 46 16.005 & -32 14 00.83 & 4.98 @xmath2 0.30 + 2003/12/20 & 52994.49 & eastern jet & 17 46 16.026 & -32 14 00.67 & 5.25 @xmath2 0.30 + 2004/02/13 & 53049.40 & eastern jet & 17 46 16.056 & -32 14 00.40 & 5.65 @xmath2 0.50 + & & western jet & 17 46 15.274 & -32 14 00.78 & 4.30 @xmath2 1.00 + 2004/04/07 & 53103.22 & + 2004/06/01 & 53158.15 & + mjd : modified julian date ( observation midtime ) .
the angular separation is based on the _ chandra _ position of the black hole h. lllccccc & & & & + date & mjd & source & & spectral index + & & & 1384 & 2368 & 4800 & 8640 & + 2003/11/12 & 52955.86 & eastern jet & ... & ... & 1.10 @xmath2 0.10 & 0.48 @xmath2 0.11 & @xmath11.41 @xmath2 0.42 + 2003/11/30 & 52973.72 & eastern jet & ... & ... & 1.55 @xmath2 0.09 & 0.87 @xmath2 0.09 & @xmath10.98 @xmath2 0.20 + 2003/12/09 & 52983.45 & eastern jet & ... & ... & 1.86 @xmath2 0.09 & 0.84 @xmath2 0.09 & @xmath11.35 @xmath2 0.20 + 2003/12/20 & 52994.49 & eastern jet & ... & ... & 2.37 @xmath2 0.06 & 1.34 @xmath2 0.07 & @xmath10.97 @xmath2 0.10 + 2004/02/13 & 53049.40 & eastern jet & 0.62 @xmath2 0.15 & 0.41 @xmath2 0.15 & 0.33 @xmath2 0.05 & @xmath3 0.21 & @xmath10.49 @xmath2 0.22 + & & western jet & @xmath3 0.45 & @xmath3 0.45 & 0.14 @xmath2 0.05 & @xmath3 0.21 & n.a .
+ 2004/04/07 & 53103.22 & eastern jet & ... & ... & @xmath3 0.18 & ... & n.a .
+ 2004/06/01 & 53158.15 & eastern jet & @xmath3 0.45 & @xmath3 0.30 & @xmath3 0.12 & ... & n.a .
+ mjd : modified julian date ( observation midtime ) .
radio upper limits are given at the 3 sigma level .
fluxes for the western jet are only given for the day of its unique atca detection , otherwise the upper limits can be deduced from the error bars ( one sigma ) or upper limits on the flux densities of the eastern jet .
following the transition of h to the hard x - ray state ( tomsick & kalemci 2003 ) on 2003 october 20 ( mjd 52933 ) , we initiated a series of radio observations with the australia telescope compact array ( atca ) starting on 2003 november 12 ( mjd 52955 ) .
the atca synthesis telescope is located in narrabri , new south wales , australia and consists of an east - west array with six 22 m antennas .
a total of seven radio observations were carried out with the atca , mostly at 4800 mhz and 8640 mhz with occasional observations at 1384 mhz and 2496 mhz ( see tables 1 and 2 for details ) .
the amplitude and band - pass calibrator was pks 1934@xmath1638 , and the antenna s gain and phase calibration , as well as the polarization leakage , were derived from regular observations of the nearby ( 5.8 away ) calibrator pmn 1729@xmath1373 .
the editing , calibration , fourier transformation , deconvolution , and image analysis were performed using the miriad software package ( sault & killen 1998 ) . no radio emission is detected at the position of h in any of our atca observations .
however , we notice the presence of a new radio source ( fig .
1 ) about 4.6 to the east of h , immediately in the first atca observation .
the following observations indicate that the new source increased in flux ( see tables 1 and 2 and fig .
the position of the source also changed over time , moving away from the bhc h. we discuss the motion of the source in a later section taking into account the additional constraints from the _ chandra _ observations .
the light - curve of this source , combining all our atca observations ( at 4800 mhz and 8640 mhz ) , is presented in figure 2 .
this source was brightening from our first observations until the end of 2003 . the single detection on 2004 february 13 ( mjd 53049 ) at a much ( almost a factor ten ) fainter level points to a very fast decay . the maximum of radio emission was likely in january 2004 .
the rise of radio emission at 4800 mhz is gradual , whereas there may be some variations during the rise at 8640 mhz .
we fitted an exponential rise and a power - law rise to the radio data ( at both frequencies ) and found that both fits are equally acceptable .
the time - scales are consistent for both radio frequencies .
the 1/_e_-folding time of the exponential rise is 49.4 @xmath2 4.8 days at 4800 mhz and 40.0 @xmath2 6.8 days at 8640 mhz .
the index of the power - law rise is 4.9 @xmath2 0.5 at 4800 mhz and 6.0 @xmath2 1.0 at 8640 mhz . due to the fast decay of radio emission
, we can not accurately constrain its decay rate . however
, we can give some limits that clearly indicate that the decay is faster than the rise . assuming a maximum at the time when we detected the source at its brightest level ( on 2003 december 20 ) , and taking the 4800 mhz detection during the decay , we deduced an 1/_e_-folding time of the exponential decay of 27.9 @xmath2 2.2 days or an index of the power - law decay of 10.2 @xmath2 0.8 .
we emphasize that these limits ( an upper limit on the 1/_e_-folding time and a lower limit on the index of the power - law ) are quite conservative and that the true decay during the early part of 2004 is likely faster than indicated by these numbers .
the light - curve at 8640 mhz is consistent with that at 4800 mhz , but very few data points are available . we searched for linearly polarized radio emission when the eastern source was at its brightest level ( i.e. on 20 december 2003 ) .
linear polarization at the level of 9.9 @xmath2 2.9% with a mean polarization angle of 51.9 @xmath2 8.3 is detected at 4800 mhz . at 8640 mhz ,
we find an ( 3 sigma ) upper limit of @xmath4 on the polarization .
this limit is weak because the 8640 mhz flux density is weaker than at 4800 mhz .
in addition to the new source to the east of h , we note that a radio source is detected to the west of h ( figure 3 ) , almost symmetrically placed relative to the eastern source , only on 2004 february 13 with a flux density of 0.14 @xmath2 0.05 mjy at 4800 mhz . despite its weak flux density , the _ chandra _ data on the same day ( figs . 4 and 5 ) confirm the existence of an x - ray source at this position and therefore strengthen the reality of the western radio source . as both of these radio sources appear to be moving away from the bhc ,
we conclude that they are related to the h system .
the properties are very reminiscent to the large scale relativistic radio and x - ray jets of ( corbel et al .
2002 ) after 2000 .
it is therefore likely that these evolving radio sources represent the action of previously ejected plasma on the ism .
as discussed below , the most likely ejection date is around 2003 april 8 when a major radio flare was observed by rupen et al .
. we will now refer to these sources as the eastern and western jets , based on their location with respect to h. we report in tables 1 and 2 their positions ( from radio observations ) , angular separation from h , as well as their flux densities . as h
is not detected at radio frequencies , we used its _
chandra _ position as defined in the next section . following the detection of the eastern moving radio source and its strong similarities with the large scale x - ray jets in ( corbel et al .
2002 ; tomsick et al . 2003a ; kaaret et al .
2003 ) , we triggered our target of opportunity proposal for the _ chandra x - ray observatory _
( weisskopf et al . 2002 ) to search for an x - ray counterpart to the radio jets .
the _ chandra _
observations were significantly delayed relative to the trigger because h passed near the sun and was not observable . as indicated previously , the eastern jet was already decaying when the _ chandra _ observations were performed ( see fig . 2 for their scheduling relative to the radio light - curve ) .
h was observed with _
on 2004 february 12 ( mjd 53048 ) , march 24 ( mjd 53089 ) and march 27 ( mjd 53092 ) using the advanced ccd imaging spectrometer spectroscopic array ( acis - s ; bautz et al .
in all observations , the target was placed on one of the back - illuminated acis chips ( s3 ) with the acis - s operated in imaging mode . for the first observation , only the s3 chip was read out and a 1/2 sub - array mode was used to limit pile - up . for the latter two observations , the source was known to be a lower flux state and the full acis - s imaging mode array was used .
we produced 0.38 kev acis images using the `` level 2 '' event lists from the standard data processing ( ascds v7.2.1 ) using the _ chandra _ interactive analysis of observations ( ciao ) software package version 3.0.2 and calibration data base ( caldb ) version 2.26 .
we constructed light curves with all valid events on the s3 chips to search for time of high background .
only weak background flares were found ( and removed ) for the february observation , otherwise the count rate appears uniform .
the total useful exposure obtained was 17796 s on february 12 ( obs . 1 ) , 28363 s on march 24 ( obs .
2 ) , and 40037 s on march 27 ( obs . 3 ) .
we searched for x - ray sources in each 0.38 kev image using _ wavdetect _ ( freeman et al .
2002 ) , the wavelet - based source detection routine in ciao .
for all three _ chandra _ observations , an x - ray source is found at the location of h , the eastern jet , and western jet ( figs 4 and 5 ) .
all three sources appear aligned and therefore provide further evidence for a connection with h. we check the _ chandra _ _ wavdetect _ position of each source by calculating the source s centroid using the 0.3 - 8 kev events from a 4 @xmath5 4 pixels ( 2@xmath5 2 ) region centered on the it wavdetect positions .
this region contains all events likely related to the source along with a small number of background events .
the recalculated positions of h , the eastern and western jets were in good agreement with the _ wavdetect _ positions ; the differences were less than @xmath6 .
the position angle of the eastern jet relative to h is 89.0 @xmath2 1.5 , while the position angle of the western jet is 91.7 @xmath2 1.8 . to obtain the best constraint on the position of each detected source ( especially h , which is not detected in our radio images )
, we registered the _
images with an infrared image from the 2 micron all - sky survey ( 2mass ) following the procedure described in tomsick et al .
( 2003b ) .
we use the two longest _ chandra _ observations ( # 2 and 3 ) and inspected each observation separately .
we restricted our search for x - ray and infrared sources within a 4 arcmin radius circle centered on the position of h as reported by _
wavdetect_. we cross - correlated the _
chandra _ source positions with the 2mass sources in the field .
we find that 5 _ chandra _ sources have 2mass sources within the _ chandra _ pointing uncertainty of 0.6 for observation 2 and 9 for observation 3 .
the accuracy of the 2mass source positions is 0.2 ( 90% confidence ) . for these sources , the angular separation between the 2mass and _ chandra _ positions ranges from 0.07 to 0.22 for observation 2 and from 0.06 to 0.46 for observation 3 .
given the surface density of 2mass sources , there is a 0.11% probability that a match with the largest separation is spurious for observation 2 , and 0.50% for observation 3 .
we used the full set of matches to register each x - ray image .
the average 2mass to _ chandra _ differences are @xmath7 in r.a . and @xmath8 in decl . for observation 2 and @xmath9 in r.a
. and and @xmath10 for observation 3
. we estimated the location of h as the average of the positions measured in the two _ chandra _ images after registration to the 2mass positions by performing the above indicated shifts .
his found to be located at r.a .
= 17h 46 m 15.s613 , decl . = @xmath13214 0.95 with a total uncertainty of 0.10 ( including the 0.2 systematics due to the registration to 2mass images and the statistical uncertainties on the _ chandra _ _ wavdetect _ position ) .
the difference in position of the eastern and western jets , after the registering process , between the two _ chandra _ observations ( only separated by 2.5 days ) were less than 0.10 ( i.e. within the total uncertainty ) .
this precise _ chandra _ position of h is consistent within uncertainties ( a difference of 0.25 ) with the recently refined vla radio position ( rupen , mioduszewski , & dhawan 2004 ) .
we used the position of h to register _ chandra _ observation # 1 on the 2mass frame , for which shifts of @xmath10.25 @xmath2 0.17 arcsecond ( in r.a . ) and @xmath10.32 @xmath2 0.16 arcsecond ( in decl . )
are needed .
these shifts are smaller than the _
_ absolute astrometric of @xmath11 at 90% confidence .
we report in table 3 the location of h , the western and eastern jets , as well as the angular separation between the jets and the bhc for all three _ chandra _ observations .
we note that the angular separation does not depend on the image registration .
cllclccccc obs . &
date & mjd & exposure & source & & angular + # & 2004- & & time ( s ) & & r.a . & decl . &
uncertainty & separation + 1 & 02 - 12 & 53048.0 & 17796 & h & 17 46 15.613 & -32 14 0.95 & 0.10 & n.a .
+ & & & & eastern jet & 17 46 16.094 & -32 14 0.92 & 0.19 & 6.10 @xmath2 0.20 + & & & & western jet & 17 46 15.263 & -32 14 1.44 & 0.24 & 4.47 @xmath2 0.30 + 2 & 03 - 24 & 53088.9 & 28363 & h & & n.a .
+ & & & & eastern jet & 17 46 16.142 & -32 14 0.76 & 0.17 & 6.72 @xmath2 0.20 + & & & & western jet & 17 46 15.243 & -32 14 1.05 & 0.16 & 4.70 @xmath2 0.20 + 3 & 03 - 27 & 53091.5 & 40037 & h & & n.a .
+ & & & & eastern jet & 17 46 16.135 & -32 14 0.83 & 0.11 & 6.63 @xmath2 0.20 + & & & & western jet & 17 46 15.243 & -32 14 1.09 & 0.15 & 4.70 @xmath2 0.20 + mjd : modified julian date ( exposure midtime ) . the angular separation between the jets and h 1743@xmath1322
is based on the _ chandra _ position of the black hole h 1743@xmath1322 as estimated in observation # 2 and 3 during the 2mass/_chandra _ registration process .
the uncertainties on the position include systematics due to the registration process and the statistical uncertainties on the _ wavdetect _ position .
however , the angular separation does not depend on the image registration .
we extracted an energy spectrum in the 0.38 kev energy range for the eastern and western jets in all three _ chandra _ observations using ciao v3.0.2 tools and we fitted these spectra using xspec v11.3.0 .
we used a circular source extraction region with a radius of 2.3 , resp .
1.2 , and we extracted background spectra from an annulus with an inner radius of 13 and an outer radius of 22 , resp . 19.5 for the eastern and western jets respectively .
these regions were centered on the jet positions as given by _
wavdetect_. due to the low numbers of source counts ( at maximum of 24 counts ) , we used the w statistic for fitting ( wachter , leach & kellog 1979 ; arnaud , in prep . ) the un - binned spectra , this is adapted from the cash statistic ( cash 1979 ) and works with background subtracted spectra and low count rate .
these spectra are adequately fitted with a power - law model including interstellar absorption .
we fixed the equivalent hydrogen absorption column density , n@xmath12 , to the constant value measured in _
_ observations of h during its 2003 outburst ( miller et al . 2005 ) , i.e. 2.3 @xmath5 10@xmath13 @xmath14 .
for the eastern jet , the best - fit photon index is 1.67 @xmath2 0.90 , 1.78 @xmath2 1.10 and 2.03 @xmath2 0.90 for _ chandra _ observations # 1 , 2 and 3 respectively .
refitting these three datasets simultaneously ( allowing the normalization to vary ) led to a photon index of 1.83 @xmath2 0.54 ( with 90% confidence errors ) .
we did the same for the western jet , giving photon indexes of 2.5 @xmath2 2.5 , 2.2 @xmath2 1.0 and 1.6 @xmath2 1.0 for _ chandra _ observations # 1 , 2 and 3 respectively and a photon index of 1.9 @xmath2 0.8 ( with 90% confidence errors ) for the combined spectrum .
we fixed the power - law photon index ( see section 3.2.2 ) to a value of 1.6 to obtain measurements of the x - ray flux .
this value is extracted from the fit to the radio / x - ray spectral energy distribution .
we measure 0.38 kev absorbed fluxes for the eastern jet of ( 21.3 @xmath2 4.9 ) @xmath5 10@xmath15 , ( 10.0 @xmath2 2.6 ) @xmath5 10@xmath15 , and ( 9.4 @xmath2 2.3 ) @xmath5 10@xmath15 ergs s@xmath16 @xmath14 for observations 1 , 2 , and 3 , respectively
. concerning the western jet , the 0.38 kev absorbed fluxes are ( 3.0 @xmath2 2.1 ) @xmath5 10@xmath15 , ( 6.4 @xmath2 2.6 ) @xmath5 10@xmath15 , and ( 6.2 @xmath2 1.8 ) @xmath5 10@xmath15 ergs s@xmath16 @xmath14 for observations 1 , 2 , and 3 , respectively .
the unabsorbed fluxes , in the same band , are 2.0 higher than the absorbed values quoted above .
the quoted errors are based on the numbers of source and background counts and poisson statistics .
figure 6 shows the time variation of the 0.38 kev unabsorbed flux of the eastern and western jets .
the x - ray emission of the eastern jet is decaying , whereas the emission from the western jet is consistent with a slightly rising source ( or even constant ) .
we fitted an exponential or a power - law decay to the eastern jet and we found that both fits are equally acceptable .
the 1/_e_-folding time of the exponential decay is 53.5 @xmath2 19.6 days , whereas the index of the power - law decay is 6.2 @xmath2 2.3 .
the constraints are rather weak for the western jet , the 1/_e_-folding time of the exponential rise is 58.2 @xmath2 46.3 days and the index of the power - law rise is 5.7 @xmath2 4.5 .
we note that the western jet was likely detected at 4.8 ghz during the first _
chandra _ observation ( figure 3 ) , but unfortunately no radio observation was possible during the time of the second and third _ chandra _ observations .
however , we did perform a atca observation about 2 weeks after the third _ chandra _ observations but failed to detect radio emission from the western jet ( with similar sensitivity to its detection in february , see table 2 ) .
this indicates that the peak of emission from the western jet was likely sometime between february and the beginning of april 2004 , i.e. roughly 2 or 3 months after the peak of emission from the eastern jet . at this peak ,
the x - ray ( 0.3 8 kev band ) luminosity ( for a distance of 8 kpc ) of the eastern jet was of the order of more than 3 @xmath5 10@xmath17 ergs s@xmath16 , i.e. a level consistent with the x - ray emission of some bhcs while in their quiescent state ( e.g. garcia et al .
2001 ) . therefore ,
if the production of x - ray jets is more common than previously thought as we suggest here , some of the reported measurements of the quiescent x - ray emission of bhc may be contaminated by jet emission , particularly for observations made using instruments with worse angular resolution than _
chandra_. the western jet of was clearly extended with a leading peak and a trailing tail extended up to 5 toward the black hole ( kaaret et al .
2003 ) . to study the morphology of the x - ray jets in h , we decomposed the images along axes parallel and perpendicular to the jet axis defined as a line extending through the black hole candidate and the jets .
following the procedure described in kaaret et al .
( 2003 ) , we calculated the displacement of each x - ray event parallel and perpendicular to this axis .
all photons with energies in the range 0.38 kev and within @xmath18 of the jet axis in the perpendicular direction are included .
for each observation , we compared the morphology of the black hole candidate h , assumed to be point - like , to each of the detected jets , except for observation 1 in which the western jet had too few counts for a meaningful comparison .
we carried out the morphology comparison using a kolmogorov - smirnov ( ks ) test to permit comparison of unbinned position data .
we shifted the jet positions to match the black hole candidate position using the positions derived from _ wavdetect_. we compared the morphology for events lying with @xmath18 along the jet axis of each source position .
we find no significant evidence for spatial extent along or perpendicular to the jet axis for any jet in any observation .
we also combined the data from observations 2 and 3 ( fig .
7 ) and still did not find any significant evidence for spatial extent of either jet .
the ks - test probabilities that the black hole and jet samples are drawn from the same parent distribution range from 0.11 to 0.96 . to place an upper bound on the size of the western jet along the jet axis , we calculated the standard deviation of the displacements of events along the jet axis from the _ wavdetect _ position of the jet .
we find a value of @xmath19 .
this is inconsistent with the standard deviation calculated for the black hole candidate ( 0.48 @xmath2 0.02 ) by only @xmath20 , so , again , there is no strong evidence for spatial extent and the value should be taken as an upper limit on the jet size . in order to increase the statistics for the comparison source
, we decided to compare the x - ray jet profiles to the profiles for the source pg 1634@xmath21706 ( observation i d 1269 ) , which is used to calibrate the acis point - spread function .
again , the ks test does not indicate significantly that the bhc h or its associated x - ray jets are extended along or perpendicular to the jet axis .
the positions of the eastern and western jets change between the various radio and x - ray observations with the jets are moving away from the bhc h. in order to quantify this motion , we have calculated the angular separation between h and each jet . for all
_ chandra _ and atca observations , we used the position of h as determined in the previous section .
figure 8 shows the angular separation between h and each jet as a function of time . in this figure ,
the origin of time ( day 0 ) is the day ( 2003 april 8 = mjd 52738 ) in which a major radio flare , and hence a massive plasma ejection , was observed ( rupen et al .
2003b ) .
this figure clearly illustrates that the eastern jet is moving away from h. the _ chandra _ and atca positions of the eastern jet are consistent with a linear increase of the angular separation versus time and a linear fit to the angular separation implies a proper motion of 15.2 @xmath2 1.8 mas day@xmath16 .
the _ chandra _
locations are consistent with a continuation of the motion from the atca observations .
the data for the western jet are only weakly inconsistent with a fixed position .
a linear fit to the angular separation for the western jet gives a proper motion of 6.7 @xmath2 5.2 mas day@xmath16 .
extrapolation of the linear fit for the eastern jet ( for which we have the best statistics ) implies zero separation on mjd 52652 @xmath2 38 ( i.e. around 2003 january 13 ) .
this date is well before the initial detection of h by integral on 2003 march 21 ( mjd 52719 ; revnivtsev et al .
2003 ) . with the rxte / asm data ,
it is not possible to assess the x - ray activity of h around mjd 52652 .
however , observations of several black hole systems ( corbel et al . 2004 ; fender et al .
2004 ) has revealed that the massive ejection events occurred later the during the outburst at a period consistent with the transition from the intermediate state to the steep power - law state .
again , the bright radio flare observed by rupen et al .
( 2003b ) on 2003 april 8 ( mjd 52737 ) is consistent with this interpretation , as illustrated by e.g. the hardness evolution of the x - ray spectra ( capitanio et al . 2005 ) . for the rest of this paper
, we therefore consider the time origin of the jets as the day of the bright radio flare ( day zero in figure 8) . with the linear fit to the motion of the eastern jet
, we predict an angular separation of 1.29 @xmath2 0.55 at the time of the first bright radio flare .
if the velocity of the eastern jet was constant , we should have obtained a zero separation on this day . whereas the deviation from zero is not highly significant ( 2.4 sigma or 98% confidence level )
, this could imply that the jets were intrinsically more relativistic at the time of ejection and decelerated gradually later on . this would be another similarity with the x - ray jets of ( corbel et al .
more frequent radio and/or x - ray observations would have been necessary to better constrain this motion . the proper motion of the eastern jet appears higher that of the western jet . also , the x - ray flux of the western jet continues to rise and the eastern jet is decaying in both the radio and x - rays .
the western jet is not detected at radio frequency during the april and june 2004 atca observations , despite a sensitivity similar to the february 2004 observation ( table 2 ) .
this could suggest that the rise of emission from the western jet has stopped and that the peak of its emission occurred sometimes between february and april 2004 , i.e. few months after the peak observed for the eastern jet .
this probably indicates that the eastern jet is approaching and the western jet is receding . as there is no strong indication of a decelerating jet in h
, we first consider the kinematics of the jets under the assumption of an intrinsically symmetric ballistic ejection ( e.g. see fender et al .
1999 for the formalism ) . from the first detection of the jets ( atca on mjd 52956 for the eastern jet and _ chandra _ on mjd 53092 for the western jet ) , we estimate their average proper motions between ejection on 2003 april 8 and their first detections .
we deduced an average proper motion of 21.2 @xmath2 1.4 mas day@xmath16 for the eastern jet and 13.3 @xmath2 0.6 mas day@xmath16 for the western jet . from this , we derive @xmath22 , with @xmath23 the true bulk velocity and @xmath24 the angle between the axis of the jet and the line of sight .
this immediately gives a maximum angle to the axis of the jets of @xmath25 and a minimum velocity of @xmath26 .
we can also infer a maximum distance to h of 10.4 @xmath2 2.9 kpc .
to date , the distance to h is basically unknown , however its location toward the galactic bulge could possibly imply a galactic center location , which would be consistent with the above upper limit . if we assume a source distance of 8 kpc ( distance to the galactic center )
, we derive an intrinsic velocity of the ejection of @xmath27 and an angle of @xmath28 for the axis of the jets . at this distance
the apparent velocity of the eastern jet ( @xmath29 ) is 0.98 and for the western jet we deduce an apparent velocity ( @xmath30 ) of 0.61 .
this clearly indicates the presence of a significantly relativistic outflow in h , almost a year after the ejection event .
if the jets have been gradually decelerated since the ejection event , the average proper motions can not be used to constrain the jets kinematics . in that case
, we can still study their properties using the 2004 measurements of the proper motions ( 15.2 @xmath2 1.8 mas day@xmath16 for the eastern jet and 6.7 @xmath2 5.2 mas day@xmath16 for the western jet ) .
we again assume a source distance of 8 kpc , keeping in mind that this is a major source of uncertainty in the derived parameters .
at this distance the apparent velocity of the eastern jet ( @xmath31 ) is 0.71 @xmath2 0.08 and for the western jet we deduce an apparent velocity ( @xmath30 ) of 0.31 @xmath2 0.24 .
the true bulk velocity , @xmath32 , is defined as @xmath33 , with @xmath24 equal to the angle between the axis of the jet and the line of sight .
this function has a minimum at @xmath34 = tan@xmath16(1/@xmath35 ) = 54 with our measured apparent velocity for the approaching jet ( again for a distance of 8 kpc ) .
this translates into a minimum true bulk velocity of @xmath36 = 0.57 @xmath2 0.05 , still indicating the presence in 2004 of a significantly relativistic flow in h. we define the radio flux density s@xmath37 @xmath38 @xmath39 , with @xmath40 the radio spectral index and @xmath41 the frequency .
we note a difference in the properties of radio emission during the rise versus decay and the decay phases .
the radio spectra are consistent with being optically thin ( i.e. @xmath42 ) all the time .
we note that the spectra are very steep during the rise ( see table 2 , @xmath43 1 ) , whereas the single radio detection during the decay is more typical of optically thin synchrotron emission ( @xmath40 = @xmath10.49 @xmath2 0.22 ) .
the detection of a significant level of polarization ( when the sensitivity allowed it ) is consistent with the radio emission being optically thin synchrotron emission .
the broadband spectra ( see below ) accurately constrains the spectral index and is consistent with a classical synchrotron spectral index of @xmath44 .
this index from the broadband sed on 2004 february 12 - 13 is significantly different from the average radio spectral index during the rise , which is @xmath40 = @xmath11.18 @xmath2 0.13 .
the difference of spectral indices during the rise is probably not related to the emission beeing resolved out at high frequency by the atca , as the four radio observations during the rise were performed in various configurations of the telescope .
in addition , we note that both jets are not resolved by _
chandra _ ,
and therefore they should appear as point sources to atca . whereas the synchrotron slope is almost a textbook example during the decay , the spectral index during the rise is likely an intrinsic property of the process that accelerates the emitting particles during the collision of relativistic plasma with the local medium .
this may possibly be explained as followed : the rising phase represents the propagation of the jets through a dense medium associated with a mechanism that accelerate particles to high energy , whereas the decaying phase represents propagation through a less dense medium and possibly without any further re - acceleration of particles in the jets .
the emission properties during the decay phase would then be governed by synchrotron and/or adiabatic losses of the relativistic particles .
the steep spectrum during the rise may eventually shed light on the mechanism that accelerates particles to high energies .
the eastern and western jets are simultaneously detected at radio and x - ray frequencies on 2004 february 12 - 13 .
figure 9 shows the broadband spectral energy distribution of the jets at this epoch .
the radio fluxes are from table 2 , and we convert the unabsorbed 0.38 kev flux to a flux density at a pivot x - ray energy defined by allowing the x - ray photon index to vary ( fig .
. a combined fit of the radio and x - ray data result in a spectral index of -0.64 @xmath2 0.02 , which is typical of synchrotron emission .
the same is true for the western jet , giving a radio / x - ray spectral index of -0.70 @xmath2 0.05 on 2004 february 12 - 13 .
the detection of linear polarization ( @xmath45 10% ) from the eastern jets of h , as well as the spectral index , strongly favor synchrotron emission as the physical origin of the radio emission in the jets .
the fact that the x - ray emission is consistent with an extrapolation of the radio power - law suggests that the synchrotron emission extends up to x - rays .
we can therefore derive the minimum energy associated with the eastern jet on 2004 february 12 - 13 ( the jet for which we have the best constraints ) under the assumption of equipartition between the magnetic and electron energy densities ( longair 1994 ) .
we assume the case of ballistic ejection ( section 3.1.1 ) , corresponding to an angle between the jet axis and the line of sight of @xmath24 = 73 and a bulk velocity @xmath46 = 0.79 .
we use a frequency range of 1.4 @xmath47 hz to 2 @xmath48 hz with a spectral index of 0.6 and a flux density of 2.4 njy at the pivot energy ( figure 9 ) of 2.7 kev .
the major uncertainty for this calculation is the estimate of the volume of the emitting region .
for that purpose , we use the same method as in tomsick et al .
( 2003a ) by assuming that the emitting region is a section of a cone with its vertex at the compact object .
an upper limit can be obtained using the _ chandra _ constraint on the source size ( fwhm @xmath49 1.4 ) as calculated in section [ xmorph ] .
the opening angle of the cone is not well constrained , so we consider the upper limit from the _ chandra _ observations of 12 and also 1as in ( kaaret et al .
2003 ) , giving a volume of the emitting region of about 4 @xmath50 @xmath51 ( for an opening angle of 12 ) or 3 @xmath52 @xmath51 ( for 1 ) . for this volume range ,
the corresponding minimum energy required is about 1.6 @xmath53 to 1.4 @xmath54 erg and the associated magnetic field for which the energy in relativistic particles equals the magnetic energy is in the range 0.2 to 0.8 mg .
the kinetic energy of a pure electron / positron plasma is @xmath55 to 9 @xmath53 erg .
if there is one proton per electron , then the kinetic energy of the electron / proton plasma is about 2 @xmath53 to 1.1 @xmath56 erg .
the associated mass of this electron / proton plasma would be of the order of 1@xmath57 to 4 @xmath58 g. for a typical mass accretion rate of 10@xmath59 g s@xmath16 , this amount of material could be accumulated in 1000 to 4000 s. as the mass outflow rate is likely much lower , an accumulation time of the order of a day would be obtained if we assume a few percents efficiency , this could be consistent with the timescale of the initial radio flare .
the lorentz factor of the x - ray synchrotron emitting electrons would then be of the order of 1 to 3 @xmath60 ( taking a pivot energy of 2.7 kev and assuming equipartition ) , giving synchrotron lifetime cooling of the orders of 2 to 50 years , i.e. much longer than the lifetime of the x - rays jets of h. it is interesting to note ( keeping in mind the limitation on the volume estimate ) that all the above derived quantities are of the same order of magnitude as those derived for ( corbel et al .
2002 ; tomsick et al .
2003a ; kaaret et al .
2003 ) . in the above
, the x - ray emission seems consistent with an extrapolation of the synchrotron radio spectrum .
however , it is of some interest to see if some other emission mechanism , such as inverse compton or synchrotron self - compton , could also be viable at high energy .
assuming that only the radio emission is of synchrotron origin lead to an equipartition magnetic field of 0.2 to 0.7 mgauss ; from this we deduce a range for the magnetic energy density of the order of 2 @xmath61 to 2 @xmath62 erg @xmath63 .
the synchrotron photon energy density is then estimated , similarly to the case ( tomsick et al .
2003a ) , to be of the order of 6 @xmath64 to 2 @xmath65 erg @xmath63 .
the synchrotron photon density is well below the magnetic energy density , implying that ssc is very likely not important at x - ray frequencies . as argued in tomsick
( 2003a ) , inverse - compton emission from the interstellar radiation field and the cosmic microwave background are unlikely because the associated energy density is well below the magnetic energy density .
all of these considerations point to a coherent picture which is that the x - ray emission associated with the jets of h is synchrotron emission and therefore very high energy ( @xmath0 10 tev ) particles are produced in those jets . with this discovery of transient x - ray jets in h
, we can now come to the idea that such events may be a common occurrence associated with any x - ray binary in outburst .
indeed , after the original discovery of x - ray jets in ( corbel et al .
2002 ) , a relativistic ejection in has also been observed to later develop into a large scale jet , possibly related to the interaction with the ism ( gallo et al .
however , the western large scale jet of has only been observed at radio frequencies ( e. gallo , private communication ) , possibly due its fast decay rate .
indeed , taking the radio flux density quoted in gallo et al .
( 2004 ) , we can estimate the time - scale associated to the two knots in the large scale jets of assuming an exponential or a power - law decay .
the 1/_e_-folding times of the exponential decays are 47.3 @xmath2 8.3 days and 57.0 @xmath2 14.7 days , whereas the indeces of the power - law decays are 5.9 @xmath2 1.0 and 4.9 @xmath2 1.0 , respectively for knot a and knot b. these timescales are consistent with those of h , but much faster than those of ( kaaret et al .
the x - ray observation of was 6 months after the last detection of the western large scale jet at radio frequencies ( gallo et al .
2004 ) and any x - ray emission likely decayed away before the observation .
it is unclear why the decay could be faster in some sources , but this may be related to local ism conditions .
also , as outlined by wang et al .
( 2003 ) , the ( already ) fast decay of the x - ray emission from the eastern x - ray jet in was not consistent with a forward shock propagating through the ism . on the contrary , if the emission was driven by a reverse shock following the interaction of the ejecta with the ism , then the rapidly fading x - ray emission could be the synchrotron emission from adiabatically expanding ejecta . with a `` much '' faster decay in h or even , it may also be possible that the emission in these two cases is also related to a reverse shock moving back into the ejecta . in that case
, it would be interested to see if a reverse shock model ( as in wang et al . 2003 ) could also reproduce the flux and spectral evolution that we observed during the rise of emission of the jets of h. also the fast transition between the rising and fading phases should also be explained .
in addition , we would like to mention the strong similarities with the neutron star system sco x1 , which shows the ballistic motion of two radio lobes located on each side of the compact star ( fomalont et al .
2001a , b ) . however , the sco x1 lobes are powered by a continuous and unseen ultra relativistic beam of energy ( fomalont et al .
2001a , b ) , possibly related to the fact that sco x1 is always close to the eddington luminosity for a neutron star .
similarly , as discussed before , the bright radio flare in h on 2003 april 8 took place around the transition to a steep power - law state , i.e. the state which is believed to be close to the eddington luminosity .
otherwise , h , like most x - ray transients , is usually believed to be in a state of low accretion rate ( quiescence ) , and even if a large fraction of the accretion energy could be carried out by an outflow ( compact jet ) in this accretion regime ( fender 2004 ) , the total amount of energy would still be much less than in the brighter x - ray state .
therefore , as we do not have any indication for a continuous and persistent powerful beam of energy in h , we believe that the evolution of the large scale jets in h and is the result of one ( or several in a short period ) inpulsive ejection event(s ) which later interact with the ism . as we have calculated above , many of the parameters ( magnetic field , particle energy , etc .
) derived for h are consistent with those obtained for , and therefore strengthen the similarities between these two sources . here , we point to another similarity . when first detected , both jets in h were at the same angular distance ( @xmath45 4 arcsec = 0.16 pc for a distance of 8 kpc ) from the black hole ( however , we note that the eastern jets may have been active some time before our first atca observations ) , i.e. the plasma originally ejected ( probably around 2003 april 8) by the black hole has traveled the same distance before brightening again , possibly due to the collision with denser ism .
similarly , in the jets were observed to brighten at the same angular distance ( @xmath45 22 arcsec = 0.5 pc for a distance of 5.3 kpc ) from the black hole .
this could indicate that these two black holes lie in a cavity or that the jets propagate through an evacuated channel pre - existing to the outburst ( e.g. heinz 2002 ) and that they
turn on again when they hit a denser ism phase . whereas either the ejection or the ism was not symmetric or homegeneous in the case of ( kaaret et al .
2003 ) , the evolution of the radio emission of both jets may also suggest a similar case in h. indeed , on 2004 february 13 , the western jet was at an angular separation of 4.47 @xmath2 0.30 and with the kinematics of the jets as outlined previously , it should have passed through the 4.63 to 5.25 arcsecond region during the last two radio observations . if the ejection was symmetric and the ism homegeneous , the western jet should have been brighter at radio frequencies in april - june than in february 2004 .
the non - detection of the western jet in the last two atca observation may therefore suggest a hint of asymmetry in the ism density or in the ejection .
however , a more detailled lightcurve would have been necessary to confirm this .
the australia telescope is funded by the commonwealth of australia for operation as a national facility managed by csiro .
_ rxte_/asm results are provided by xte / asm team at mit .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
we thank the referee for constructive comments .
sc acknowledges useful discussions with s. heinz .
pk acknowledges partial support from nasa chandra grant go4 - 5039x .
jat acknowledges partial support from nasa chandra grant go3 - 4040x . | we report on the formation and evolution of two large - scale , synchrotron - emitting jets from the black hole candidate h following its reactivation in 2003 . in november 2003
after the end of its 2003 outburst , we noticed , in observations with the australia telescope compact array , the presence of a new and variable radio source about 4.6 to the east of h , that was later found to move away from h. in february 2004 , we detected a radio source to the west of h , symmetrically placed relative to the eastern jet . in 2004 , follow - up x - ray observations with _
chandra _ led to the discovery of x - ray emission associated with the two radio sources .
this likely indicates that we are witnessing the interaction of relativistic jets from h with the interstellar medium causing in - situ particle acceleration .
the spectral energy distribution of the jets during the decay phase is consistent with a classical synchrotron spectrum of a single electron distribution from radio up to x - rays , implying the production of very high energy ( @xmath0 10 tev ) particles in those jets .
we discuss the jet kinematics , highlighting the presence of a significantly relativistic flow in h almost a year after the ejection event . |
although there are some thirty galaxies in which h@xmath0o masers are observed ( certain others have oh masers ) , there is little in the way of theory " other than for the few where there is evidence for disk structure . even for these , almost all of the theoretical activity has been directed toward ngc4258 where the highly refined observations of 22 ghz water masers in a near ideal , thin keplerian disk provide the best opportunity for applying and testing ideas .
my discussion of the theory and extraglactic masers " is thus largely a summary of theory directed toward the masers in a single galaxy but it is a paradigm of considerable importance , and the ideas involve the basic description of molecular accretion disks near active galactic nuclei .
analysis of the structure of this disk utilizing information from maser observations potentially provides opportunities to assess the mass accretion rate which , when related to the energy emitted from this nucleus in other wavebands ( x - ray , infrared ) , bears on current issues concerning active galactic nuclei especially , on the possibility for advection dominated accretion flows . such analysis also may help clarify the nature of the long sought mechanism for viscosity in accretion disks .
further , the masing disk in ngc4258 provides arguably the best evidence for massive black hole at the nucleus of an active ( or at least mildly active ) galaxy .
see moran , greenhill , & herrnstein ( 1999 ) for a summary of the observational data about extragalactic h@xmath0o masers .
masing by the water molecule at 22 ghz involves states ( 6@xmath1 and 5@xmath2 ) which lie some 600k above the ground state .
elevated temperatures and relatively high gas densities are thus required to create strong 22 ghz masers .
gas kinetic temperatures of 400k and hydrogen densities of 10@xmath3 @xmath4 are considered representative for galactic water masers , though somewhat lower temperatures ( 300k ) and densities ( 10@xmath5 @xmath4 ) may be adequate in circumnuclear disks where the dimensions of the masing regions can be larger . since the earliest calculations ( schmeld , strelnitskii , & muzylev 1976 ) , galactic water masers other than the circumstellar variety are generally believed to be associated with shock waves .
shock waves almost certainly create some of the environments for the extragalactic masing of water , as well .
however , the accretion disks are likely to be relatively quiescent . further
, strong x - ray emission is typical at the nuclei of these galaxies .
hence , the elevated temperatures in the gas of the accretion disk that are necessary for masing have seemed most likely to be a result of heating by x - rays , though shock waves ( maoz & mckee 1998 ) and even viscous heating ( desch , wallin , & watson 1998 ) can not readily be dismissed . the chemistry and excitation for masing in a dense gas on which x - rays with intensities that are likely in circumnuclear environments has been computed by neufeld , maloney & conger ( 1994 ) . as in galactic water masers ,
thermal collisions elevate the water molecules to a range of excited states .
the inversion results from the ensuing radiative / collisional cascade .
escape of a fraction of the infrared , spectral line radiation that is involved in the cascade is essential to prevent the thermalization of the populations of the masing states , and hence the quenching of the masing . in standard calculations including those of neufeld
( 1994 ) , the escape ( and hence the maser power ) is limited by the trapping due to reabsorption of the infrared radiation by the water molecules themselves ( see figure 1 ) .
however , the maser power can be increased significantly by utilizing deguchi s ( 1981 ) recognition that absorption of the infrared radiation by dust grains can serve as an alternative to escape for preventing the thermalization of the masing transition .
the results of calculations ( collison & watson 1995 ) for the pumping that include the influence of this absorption by dust grains in the circumnuclear masing environments created by x - rays are shown in figure 1 .
it is essential that the temperature of the grains remain below that of the gas .
this requirement is satisfied for a wide range of likely conditions ( desch et al .
although the potentially important issue of time variations in the flux from extragalactic masers is beyond the scope of this discussion , it is noteworthy that changes in the temperatures of the dust grains may play a role in understanding the short time scales for the changes ( neufeld 2000 ; gallimore et al .
> from the viewpoint of a distant observer in the plane of a masing keplerian disk that is uniformly pumped in a ring between inner radius r@xmath6 and outer radius r@xmath7 , the optical depths for unsaturated masing are largest toward the center of the disk and for rays at the sides of the disk with impact parameters that are near ( but slightly less than ) the outer radius of the disk .
the gradients in velocity are a minimum in these directions .
the emission at the sides arises mainly from locations along the rays where the rays are nearly tangent to the keplerian orbits ( i.e. , near the midlines ) .
thus , three symmetrical peaks in the spectrum and three images extended along a line in the sky are considered indicative of a disk .
if the masing is highly saturated , the optical depths at the sides will be largest for rays with impact parameters near the inner radius of the masing ring and will occur at doppler velocities that are essentially equal to the velocities of the keplerian orbits that are tangential to these rays . toward the center
, the doppler velocities of the most intense rays at each impact parameter will vary linearly with the impact parameter .
this variation is proportional to the rotational velocity at the outer edge of the ring if the masing is unsaturated , and is proportional to the rotational velocity at the inner edge if the masing is highly saturated .
in addition to exhibiting characteristics similar to the foregoing , the doppler velocities of a number of microfeatures within the central emission of ngc4258 are observed to change linearly with time ( baan , haschick , & peng 1993 ) . as a result
, the diverse data from ngc4258 can be interpreted as defining a masing disk viewed nearly edge - on and the mass of the object at its center can be inferred ( watson & wallin 1994 ) . in more recent observations , images of these masing features have been observed to move across the sky . for ngc4258 ,
the maser amplifications toward the center and at the sides are not at all similar .
emission toward the center is prominent , but only because the masing here is amplifying a strong continuum source at the center .
clearly , the clumpy appearance of the maser emission from ngc4258 indicates some form of irregularities in the medium of the disk . at the same time
, the medium must be smooth enough ( or the irregularities numerous enough ) that emission at the sides occurs mainly near the midlines where the lengths for coherence in velocity are greatest .
widely separated , but aligned " masers can be quite intense ( deguchi & watson 1989 ) . turbulence is prevalent in astrophysical environments of diffuse matter ( e.g. , supersonic turbulence in molecular clouds ) .
mhd turbulence is believed to be involved in the viscous dissipation mechanism for accretion disks .
it is thus natural to inquire as to whether the rather gentle irregularities in the optical depths that are caused by turbulence of a generic " form are sufficient to create spectra that are similar to those that are observed for ngc4258 .
we have thus performed computations that utilize standard methods to create representative turbulent velocity fields in the gas of a disk with dimensions and with keplerian velocities similar to those of the ngc4258 disk . for simplicity
, the maser opacity is taken as constant and maser saturation is ignored .
examples of spectra from our computations are given in figures 2 and 3 for rms turbulent velocities that are equal to the sound speed .
what is shown in these figures is the maser amplification due to the disk .
we have reasoned that the magnitudes of the amplification factors in these figures , when combined with best estimates for the intensities of the background continuum radiation that is being amplified , are approximately what is required to yield the observed fluxes .
these spectra can be compared with observations in , e.g. , myoshi et al .
( 1995).rms turbulent velocities that are approximetely equal to the sound speed in the masing gas seem to be necessary to obtain spectra that are similar to the observed spectra .
deguchi ( 1982 ) originally demonstrated that representative turbulent velocity fields obtained in a somewhat analogous way lead to images for circumstellar masers that are similar to the observed images ( also , sobolev , wallin , & watson 1998 for spectra and images of methanol masers in a turbulent medium ) .
the possibility for a quite different origin of the masing features at the sides of the disk has been recognized by maoz & mckee ( 1998 ) .
the reasoning in this interpretation is that the mass of the disk may be within the range where the toomre stability parameter @xmath8 indicates that spiral density waves will form .
the features at the sides are then proposed to occur where the lines of sight are tangent to these spirals .
shocks form at the edges of these waves where gas flows into the wave , and provide the conditions for the water masers . because the pattern of the waves is rotating , the locations where the tangents occur drift outward in radius to lower keplerian rotational velocities .
an acceleration of the doppler velocities of the masing features is thus predicted .
the absence of such accelerations in refined observations ( bragg et al .
1999 ) is in conflict with the proposed interpretation at least for what seem to be the most likely choices for the relevant parameters . in a somewhat related idea but directed toward the presumed masing disk of ngc1068 , kumar ( 1999 ) describes how gravitational instabilities in the disk can create the masing clumps when the mass of the disk is larger than that for which spiral density waves are expected .
other discussions of the nature of the masing disk include kartje , konigl , & elitzur ( 1999 ) , and babkovskaya & varshalovich ( 2000 ) .
novel aspects of radiative processes also have been examined ( deguchi 1994 ; gangadahara , deguchi , & lesch 1999 ) .
a pervasive issue in understanding and interpreting the observations of astrophysical masers is the degree of saturation . to proceed directly from the observed flux
, one must resort to highly uncertain estimates for the beaming angle ordinarily utilizing a cylindrical geometry for the masing region . for keplerian disks , the cylindrical approximation tends to overestimate the degree of saturation ( watson & wyld 2000 ) .
nevertheless , we and others have reasoned in this way that the masers in ngc4258 are likely to be no more than partially saturated ( see wallin et al.1998 ) .
an alternative approach is to recognize that ( in the absence of velocity gradients ) the spectral linebreadths of unsaturated masers are narrower than the thermal breadths , and that the spectral lines rebroaden to the full thermal breadths when the maser becomes saturated . despite assertions to the contrary , our own rigorous calculations ( emmering & watson 1994 ) demonstrate that the rebroadening does occur to the full maxwellian breadth for spherical as well as for linear masers and hence , almost surely this result is independent of geometry . at the gas kinetic temperature of 400k that is ordinarily adopted as the approximate minimum for the pumping of 22 ghz masers ,
the thermal linebreadth ( fwhm ) is 1 km s@xmath9 for water molecules .
i believe that it is far more likely that the 22 ghz line is actually a merger of the three strongest hyperfine components of the 6@xmath1 - 5@xmath2 transition .
they merge so well that the line profile can be undistorted . with the three merged hyperfine components ,
the minimum breadth for the fully rebroadened 22 ghz maser line is 1.5 km s@xmath9 ( nedoluha & watson 1991 ) .
not only are the linebreadths of some of the galactic water masers with the highest fluxes less than 1.5 km s@xmath9 , they are even less than 1 km s@xmath9 !
the prominent 1306 km s@xmath9 feature from the side of the disk of ngc4258 has a linewidth of 1.1 km s@xmath9 .
many other microfeatures have breadths less than 1.4 km s@xmath9 ( bragg , et al . 2000 ) , and some are even less than 1 km s@xmath9 .
note that velocity gradients , unresolved components , etc . can increase the linebreadth , but no valid way is known to reduce the breadth other than unsaturated amplification .
the relaxation of molecular velocities , mainly through the emission and absorption of trapped , infrared spectral line radiation ( goldreich & kwan 1974 ) is a way to understand how masers can become saturated while retaining the narrow linebreadth of unsaturated masers .
then , the rebroadening of the spectral line is postponed until the rate for stimulated emission exceeds the rate for velocity relaxation .
when velocity relaxation is important for h@xmath0o masers , the decay rate @xmath10 is reduced due to the trapping . however , the rate for velocity relaxation is approximately the value that ordinarily is adopted for @xmath10the einstein a - value for de - excitation of the molecular states by infrared emission ( approximately 2 s@xmath9 for the 22 ghz states ) .
thus , the numerical value of the rate for stimulated emission at which rebroadening occurs is essentially unchanged as a result of velocity relaxation .
though this has been established in detail only for the 22 ghz masing transition ( anderson & watson 1993 ) , i expect that it is a rather general result .
a fraction ( ordinarily taken as about five percent ) of the rest energy of the matter that falls onto a black hole emerges as the bolometric luminosity in a standard interpretation .
typically , some ten percent of the bolometric luminosity appears as x - rays . by assuming steady - state ,
the accretion rate through the disk in ngc4258 can be estimated as @xmath11 .
an outstanding issue for the underluminous " nuclei in mildly active galaxies such as ngc4258 is whether this estimate is valid and the accretion rate is lower than for the more luminous , active galactic nuclei .
alternatively , the accretion rates may be similar to those in the luminous nuclei , but much of the energy fails to appear in the bolometric luminosity because it is advected into the black hole ( advection dominated accretion flows , or adaf s ) .
the refined description of the masing disk in ngc4258 offers a possibility that valuable information can be obtained on this issue .
data at various wavelengths has been fit to an assumed adaf spectrum for ngc4258 and leads to @xmath12 ( lasota et al.1996 ; gammie , narayan , & blandford 1999 ; chary et al .
2000 ) . from the observations of the masing disk , maoz & mckee ( 1998 )
obtain @xmath13 in the spiral density wave interpretation whereas neufeld & maloney ( 1995 ) find @xmath14 based mainly on the premise that the masing extends outward in radius until the molecules disappear and the gas becomes atomic . here , @xmath15 is the coefficient from the shakura - sunyaev description of viscosity .
an approach that seems somewhat more direct involves determining the strength of the magnetic field in the disk ( or a useful upper limit ) by observing for the polarization of the maser radiation . assuming
a balance between the thermal and magnetic pressure within the disk , together with the shakura - sunyaev description of turbulence , then leads to a value for ( @xmath16 ) .
herrnstein et al .
( 1998 ) thus infer an upper limit @xmath17 from an upper limit of 300 mg for the magnetic field based on their failure to detect circular polarization at a level of 1.5 percent in the 1306 km s@xmath9 feature at the side of the disk . though this limit is not sensitive enough to discriminate between an adaf and a standard model , the prospects are good for observations of the central spectrum that are considerably more sensitive ( see also moran 2000 )
. however , the spectrum for the circular polarization of the central emission will be quite different from that of a two - level transition if the 22 ghz line is a merger of three hyperfine components ( watson & wiebe 2001 ) as i believe is most likely .
relating the magnetic field strength to the circular polarization probably is less difficult for isolated spectral lines ( see also nedoluha & watson 1992 ) . the mass in the disk can be related to @xmath18 utilizing the shakura - sunyaev prescription for the viscosity . in terms of the radius @xmath19 at the location of the central masers ,
the surface mass density at radius r is ( e.g. , desch et al . 1998 ) @xmath20 since the x - rays from the central source can penetrate to only about @xmath21 , they probably fail by a considerable factor in being able to maintain the 400k or so temperatures that would be required for masing by water throughout the disk even for the @xmath22 based on the standard " interpretation of flow through the accretion disk .
thus , a cold component of the disk probably exists that is more massive than the masing component that we detect .
the spectrum of ngc4258 often exhibits a narrow dip near the systemic velocity .
this may be due to absorption by the cold component . beyond the depths
to which the x - rays can penetrate and excluding isolated regions where there might be localized heating , the temperature probably is less than about 50k . with favorable assumptions about poorly understood processes , viscous heating within
the disk might be sufficient to heat a thin layer that could be the site of the observed masing ( desch et al .
it could not , however , provide sufficient energy to heat the entire thickness of the disk to such temperatures . note that heating by shocks in spiral density waves
is included in this assessment of the viscous heating since the origin of the energy for both is the change in gravitational potential energy as the matter drifts inward .
an estimate for the mass in the disk between the inner and outer radii at which masers are observed can readily be obtained by ignoring the variation in temperature in the above expression for @xmath23 and integrating to obtain @xmath24 the central mass about which the masing disk is orbiting has a mass of @xmath25 .
i am not aware of a quantitative assessment for the maximum disk mass that is consistent with the absence of observed deviations from the keplerian rotation , but it seems unlikely to be larger than about 0.3 percent of the central mass . based on the above m@xmath26 , there is then little if any flexibility for @xmath22 to exceed the minimum value inferred from the standard description of accretion .
gravitational clumping and/or multiple thermodynamic phases may alter the simple analysis given above in essential ways .
certainly , the shakura - sunyaev prescription for viscosity is an extreme simplification .
for example , mhd simulations tend to indicate that the magnetic pressure ( and hence the likely dissipation of viscous heat ) decreases much less rapidly with distance away from the midplane of an accretion disk than does the gas kinetic pressure ( miller & stone 2000 ) , a viewpoint that has received support from other perspectives ( e.g. , stella & rosner 1984 ) . finally , by utilizing the shakura - sunyaev viscosity
once more , the toomre stability parameter @xmath8 can be expressed as @xmath27 a disk is stable for @xmath28 .
this marginal stability ( at best ) of the disk in ngc4258 underlies proposals for density waves and for the clumping of the mass .
we note that an entirely different , alternative approach exists for understanding accretion in disks ( blandford & payne 1982 ) , but does not lend itself to quantitative interpretations .
.2 in * appendix polarization of maser radiation * .1 in the contributions to issues of broad astrophysical importance from the observation of extragalactic masers , just as for galactic masers , often depend upon an understanding of the basic physics for the excitation and transport of astrophysical maser radiation .
this is especially true in efforts to infer information about the magnetic field from the observation of the polarization of the maser radiation .
confusion about the theory for maser polarization seems to be prevalent in the literature .
a summary at this time may thus be helpful .
i will focus here only on the limit in which the zeeman splitting @xmath29 is much smaller than the spectral linebreadth @xmath30 ( weak splitting `` ) .
this is the limit that is relevant for the circular polarization of extragalactic water masers , as well as for galactic h@xmath0o , sio , oh 1720 mhz , and ch@xmath31oh masers . for simplicity
, i will ignore any explicit consequences of cross - relaxation in this discussion .
cross - relaxation is analogous to velocity relaxation in that its main effect for polarization in the weak splitting regime is that the rate for cross - relaxation @xmath32 replaces the ' ' phenomenological decay rate " @xmath10 of the molecular populations in the discussion of this appendix ( goldreich , keeley , & kwan 1973b ) .
however , as in the case of velocity relaxation , @xmath10 ordinarily is reduced when cross - relaxation is important and @xmath32 has the approximate value that normally is associated with @xmath10approximately 2 s@xmath9 for the 22 ghz water masers
. a discussion of the theory of maser polarization naturally begins with the classic paper of goldreich , keeley , & kwan ( 1973a ; hereafter gkk ) .
it is therefore important to know exactly what idealized masing conditions were considered by gkk , and what answers they did and did not provide .
gkk considered the basic issue of the polarization of maser radiation that emerges from a linear maser when weak , continuum seed radiation is incident at the far end of a masing region in which there is a constant magnetic field .
they considered only an angular momentum j=1 - 0 masing transition and only equal pumping of the magnetic substates of j=1 .
i will refer to this scenario as the gkk idealization " .
_ the key restriction in gkk is that the solutions for the polarization given there are obtained only in limiting regimes for the maser intensity , as specified by the rate r for stimulated emission . _ with @xmath10 as the usual decay rate for the masing states , gkk find that there is linear polarization when ( a)@xmath33 or when ( b ) @xmath34 , and that the fractional linear polarization is zero when @xmath35 or @xmath36 . in regime ( a ) , the fractional linear polarization
( q / i ) is given by the widely quoted expressions ( q / i ) = -1 for angles @xmath37 between the magnetic field and the line - of - sight that are less than 35 degrees and for other angles between 35 and 90 degrees , @xmath38 .
in regime ( b ) , q / i is positive and can be as large as one - third .
positive q corresponds to a net linear polarization that is perpendicular to the magnetic field .
although it accomplished a lot , the gkk analysis was incomplete even for the gkk idealization because it did not tell us quantitatively how strong the inequalities that define the various regimes must be in order for the solutions given there for the linear polarization to be applicable .
more generally , the behavior of the polarization as a continuous function of maser intensity ( or degree of saturation ) was not provided .
several collaborative efforts at illinois have largely completed the picture for the gkk idealization .
i will now summarize the key results .
computations for the linear polarization as a continuous function of maser intensity were initially limited to intensities for which @xmath39 ( western & watson 1984 ) .
as expected , the computations agreed with the gkk solutions in the limit of high saturation @xmath40 . however , quite high saturations @xmath41 are necessary to attain the fractional linear polarizations of 70 percent for j=1 - 0 transitions which are sometimes observed for the sio masers . to attain the comparably high fractional polarization that is observed for masing of sio in the j=2 - 1 transition
, @xmath42 must be implausibly large .
it follows that linear polarization can not be entirely due to the gkk mechanism .
the next step was rephrasing the gkk density matrix equations in a way that is convenient for treating more generally the regime where @xmath43 is not satisfied .
an additional result in this investigation was the recognition that significant linear polarization is created in regime ( b ) [ where @xmath44 only for a j=1 - 0 transition ( deguchi & watson 1990 ) . by integrating the radiative transfer equations that are based on the density matrix formulation , nedoluha & watson ( 1990a ) finally obtained the full solution the fractional linear polarization for all @xmath45for the gkk idealization .
complete agreement with what is expected from gkk was obtained .
answers to the question of how strong must the inequalities be in order for the gkk solutions to be applicable , and how does the linear polarization behave when they are not valid , were obtained .
representative results are reproduced in figure 5 .
note that the direction of the linear polarization is neither parallel nor perpendicular to the projection of the magnetic field on the plane of the sky when @xmath46 ( i.e. , stokes u@xmath470 ) .
.2 in * b. additional considerations for the linear polarization * .1 in as noted above , the gkk mechanism alone of saturated masing in the presence of a magnetic field is unlikely to produce the highest fractional linear polarizations that are observed in the j=2 - 1 and higher transitions of the circumstellar sio masers .
actual masing environments in astronomy may be more complicated in essential ways than the idealized environment considered by gkk
. of these , anisotropic ( or @xmath48-dependent ) pumping of the magnetic substates is likely to be the most important .
the absorption and/or the escape of infrared radiation by the masing molecules ordinarily is essential in the pumping of astrophysical masers . if the angular distribution of this radiation is anisotropic when it is incident upon the molecules , or if the angular distribution with which it escapes is anisotropic , then the magnetic substates of a molecular energy level will be pumped unequally .
explicit calculations for the radiative transfer of anisotropic , infrared radiation that is involved in the pumping were performed in the context of the circumstellar sio masers to demonstrate that sufficient anisotropy can be present under conditions where the magnitude of the pumping is adequate to provide the maser power that is observed ( western & watson 1983 ) .
the circumstellar sio masers are natural candidates for anisotropic pumping since they are located within a few stellar radii of the star .
in addition , the medium in which they are located probably has strong velocity gradients which can create preferential directions for the escape of the infrared radiation . as a group ,
the circumstellar sio masers do exhibit the highest fractional linear polarizations . even though anisotropic pumping associated with infrared radiation may be the dominant cause for the linear polarization , the direction of the polarization vector is still likely to be either parallel or perpendicular to the direction of the magnetic field as projected onto the surface of the sky ( nedoluha & watson 1990a ) .
this will occur as long as @xmath29 is somewhat greater than @xmath45 and @xmath10 ( except possibly when @xmath45 and @xmath10 are comparable ) because the magnetic field is the dominant axis for symmetry .
an analogous effect occurs in the alignment of interstellar dust grains .
they are aligned either parallel or perpendicular to the magnetic field , regardless of the mechanism that causes the alignment , because the dust grains precess rapidly around the direction of the magnetic field .
faraday rotation within the maser can , however , reduce the magnitude and change the direction of the linear polarization just as it does for ordinary radiation ( see wallin & watson 1997 for an explicit calculation ) .
.2 in * c. circular polarization * .1 in gkk considered only the polarization characteristics of the radiation at the center of the spectral line . in the weak splitting limit ,
no circular polarization occurs at line center .
the issues can be discussed in two ways .
firstly , to what degree is stokes - v due to what we consider to be the standard zeeman effect altered by the saturation of a maser .
secondly , what are non - zeeman " effects that cause circular polarization .
the distinction between these may not always be clear , in principle . in practice
, it seems natural that alterations to the standard zeeman effect " should refer to changes that can be described in terms of the unequal populations of the magnetic substates that are created in saturated masers in the presence of a magnetic field when @xmath49 .
non - zeeman effects encompass everything else .
initial calculations of the influence of saturation on the standard zeeman effect focussed on the 22 ghz water transition ( nedoluha & watson 1992 ) . the calculations were limited to saturations @xmath50 .
they nevertheless demonstrated that the standard relationship @xmath51 for two - level transitions , which is applicable for thermal spectral lines and for masing lines that are unsaturated , is altered by the effects of saturation . at larger angles
@xmath52 , was found be greater by a factor of 2 or so for the f=7 - 6 hyperfine transition than would be expected from the standard relationship . since the 22 ghz transition probably is a result of the merger of three hyperfine components , the standard relationship between v and @xmath53 most likely is meaningless alternative relationships exist ( e.g. , watson & wiebe 2001 ) .
more extensive calculations have now been performed ( watson & wyld , in preparation ) for j=1 - 0 and j=2 - 1 masing transitions .
the quantity v/@xmath53 is increased in these calculations beyond that given by the standard relationship by the effects of saturation .
the magnitude of the increase is an increasing function of the angle @xmath37 until a maximum is reached at some large angle which depends somewhat on the degree of saturation . for @xmath54 _ increases , not decreases _ , with angle @xmath37 within this range of angles .
modifications to the standard zeeman effect , which have been discussed above , are likely to cause an uncertainty of only a factor of a few in the inferred magnetic fields .
in contrast , when they are effective , non - zeeman " mechanisms tend to create large circular polarization that is difficult to relate in a useful way to the strength of the magnetic field . non - zeeman circular polarization , as defined in the foregoing , occurs when the medium is anisotropic and the principal optical axes of the medium are not aligned with the direction of the linear polarization of radiation that is passing through the medium .
the effect is analogous to the creation of circularly polarized radiation when ordinary , linearly polarized light passes through a sheet of polaroid material ( or a quarter - wave plate " ) with axes that are not aligned with the direction of the polarization of the light . in astronomy , this effect is observed to occur for optical light when the direction of the magnetic field , and hence the direction of alignment of the polarizing dust grains , changes along the line of sight .
the two requirements are then firstly , the masing medium must be optically anisotropic and secondly , its optical axes and the direction of the linear polarization of the radiation must be misaligned along the path of the radiation
. the masing medium will be anisotropic when the populations of the magnetic substates are unequal
. they will be unequal when there is anisotropic pumping or when the linear polarization is created in the gkk mechanism .
more generally , when linearly polarized radiation is created within a medium , it follows that the medium is anisotropic .
the medium must be anisotropic to distinguish one direction from another which is required to establish a direction for the linear polarization of the radiation .
when linearly polarized radiation is observed , it then follows immediately that the first requirement for non - zeeman circular polarization is satisfied unless the masing medium is simply amplifying background radiation that is already linearly polarized to the same degree that is observed .
the masing medium is surely optically anisotropic in the spectral lines of the sio masers which tend to be have high linear polarization .
establishing when the second requirement for non - zeeman circular polarization is satisfied is more difficult .
when @xmath55 , the optical axes of the medium ordinarily will be aligned with the magnetic field .
thus , changes in the projected direction of the magnetic field on the plane of the sky within the masing gas will cause the optical axes and the direction of the linear polarization of the maser radiation to become misaligned . using a statistical description of turbulent velocity and magnetic fields
, we ( wiebe & watson 1998 ) have performed explicit computations to see what can be expected when the path lengths for masing and the changes in the direction of the magnetic field within these path lengths are determined by a plausible description of the medium .
the characteristics of the linear and circular polarization that emerge in this calculation are similar to what is observed for the circumstellar sio masers . even in the absence of any change in the magnetic field , the optical axes and
the direction of the linear polarization will become misaligned when @xmath56 .
this can easily be seen to occur , for example in figure 5 , where the direction of the linear polarization due to the gkk mechanism rotates by 90 degrees as @xmath45 approaches and significantly exceeds @xmath29 .
creation of non - zeeman circular polarization when @xmath56 has been investigated in considerable detail ( nedoluha & watson 1990b ; 1994 ) .
the conversion of linear to circular polarization in an anisotropic medium occurs because the phase velocity of a linearly polarized electromagnetic wave is different when its electric field is parallel to one , as opposed to the other , of the optical axes . within a spectral line , however , which of the axes has the larger phase velocity
is reversed at the center of the spectral line .
radiative transitions of molecules are analogous to classical resonances in making a contribution to both the real and imaginary parts of the dielectric constant of the medium ( e.g. , jackson 1999 ) .
the imaginary part of the dielectric constant describes the absorption and emission due to the radiative transition .
the contribution of the resonance to the real part of the dielectric constant determines the phase velocity of the medium at frequencies within the spectral line .
this contribution is antisymmetric about line center , and hence a reversal occurs at line center as to which phase velocity is the larger . as a result ,
an antisymmetric stokes - v is obtained when circularly polarized radiation is created by the non - zeeman mechanisms being described here .
the profile of this stokes - v is quite similar to the profile that is expected when the stokes - v is due to the standard zeeman effect at least for isolated , two - level transitions ( e.g. , wiebe & watson 1998 for representative spectra ) . finally , assertions have been made that `` fluctuations '' , `` instabilities '' and perhaps other ill defined , similar effects cause the polarization of masers to be quite different from the description in this appendix . in all cases where such effects have been examined in a meaningful way for astrophysical masers , they are unimportant ( gkk ; western 1983 ; watson 1994 ; wallin & watson 1995 ) .
non - local instabilities do exist in the radiative transfer equations for astrophysical masers , but only in a quite limited region of parameter space ( scappaticci & watson 1992 ) .
further , in alternatives that have been proposed to the gkk / illinois description of maser polarization , there is a fundamental failure to recognize that magnetic substates are no longer `` good '' or diagonal quantum states when @xmath45 approaches and exceeds @xmath29 .
density matrix methods are utilized for quantal calculations in this regime , and are employed by in the gkk / illinois calculations . | theoretical research on extragalactic masers ( and hence this review ) is almost entirely directed toward 22 ghz water masers in circumnuclear disks with the focus being the near ideal masing disk at the nucleus of the galaxy ngc4258 .
the discussion here is organized around ( 1 ) the excitation and conditions for the masers , ( 2 ) the spatial and spectral appearance of the maser emission , and ( 3 ) the mass accretion rate and structure of the disk . in addition , a summary is given of the basic physics ( spectral linebeadths and maser polarization ) that underlies certain interpretations of these and other astrophysical masers . # 1#2 |
the quantum xy spin chain and its extensions have been studied for a very long time and from many different perspectives .
it is due to a couple of reasons .
first of all , it is possible to obtain an exact solution ( for spin one - half case ) in the language of non - interacting fermions @xcite .
it is interesting for its own , and moreover it can be used in testing various techniques that are applied to a wide range of non - integrable systems @xcite .
another motivation is to use the model to describe the experimental data of quasi - one - dimensional systems @xcite . and
finally , in recent years the xy chain has been extensively examined in the context of quantum information theory , quantum entropy and entanglement @xcite @xcite .
the solution of the quantum spin one - half xy model is based on the jordan - wigner transformation . by means of this transformation
, the hamiltonian can be brought to a form that describes a system of non - interacting fermions .
this method has been used in seminal paper by lieb , schulz and mattis @xcite , where the anisotropic xy model without magnetic field was studied .
another important papers generalizing and extending the results of @xcite are : an exact solution of quantum ising chain in transverse magnetic field @xcite , the ground - state phase diagram of the anisotropic xy model in transverse magnetic field @xcite , the explicit expression for the ground - state energies @xcite , extensive computations of the correlation functions @xcite , a simple and short derivation of the formulae for numerous observables @xcite , the exact results for xyz model in magnetic field @xcite . the basic aim of our paper is to examine the _ differentiability properties _ of the expression for the ground state energy , which boils down to the study of the ground - state phase transitions order .
this formula , expressible in terms of elliptic integrals has been known for quite a long time @xcite , but some physical corollaries are , to our best knowledge , lacking .
let us next present a detailed study of these aspects of the ground state formula in comparison to the phase diagram of the xy model @xcite . the phase diagram is shown on fig .
[ phasedgm ] .
a good reference regarding the origin and nature of all phases ( especially the oscillatory phase ) is @xcite . ) , which splits into oscillatory and non oscillatory regions , and a paramagnetic phase ( @xmath0 ) .
parameters @xmath1 and @xmath2 are the anisotropy parameter and the magnetic field strength from equation ( [ ham1 ] ) . ]
the most interesting lines on the phase diagram are : _
i ) _ the vertical line @xmath3 , _ ii ) _ the isotropy line @xmath4 and _ iii ) _ the circle @xmath5 .
( here @xmath1 is magnetic field , and @xmath2 anisotropy parameter ; proper definitions are given in sec .
[ sec : hamanddiag ] . ) the first line _
i ) _ has been exhaustively examined ; it corresponds to the transition between the paramagnetic and ferromagnetic phases .
this transition is in the universality class of two - dimensional classical ising model @xcite .
the second line _ ii ) _ corresponds to the change of the direction of magnetization and is in the universality class of two decoupled ising models @xcite .
the third line corresponds to the transition between the ferromagnetic and oscillatory phases . in the literature ,
some rather vague assertions concerning the nature of this transition have been formulated : the thermodynamic functions do not exhibit singularities on this line @xcite , but no derivation has been given .
the expression for the ground state energy allows one to deduce that it is _ infinitely differentiable _ on this line . in this aspect ,
the transition between ferromagnetic and oscillatory phases resembles the kosterlitz - thouless one , but other aspects of these two transitions turned out to be quite different .
the paper is organized as follows . in the sec .
[ sec : hamanddiag ] we define the hamiltonian and briefly describe its diagonalization , which leads to an explicit expression for the ground - state energy .
its properties ( differentiability and asymptotic forms of solution in the neighbourhoods of phase transition lines ) are presented in the sec .
3 . the sec .
4 contains summary of results obtained , comparison with existing results and perspectives for future work .
we consider a spin-@xmath6 chain consisting of @xmath7 sites .
the spin sitting in a given site interacts only with its nearest neighbours .
the hamiltonian of such an interaction is given by @xmath8 , where @xmath9 are the nearest neighbours ; we denote it by writing @xmath10 .
moreover , the chain is placed in a transverse magnetic field @xmath11 .
therefore , the hamiltonian of the whole system is @xmath12 the spin operators , @xmath13 , @xmath14 and @xmath15 are proportional to the pauli matrices @xmath16 we rewrite expression ( [ ham0 ] ) introducing the anisotropy parameter @xmath17 $ ] , defined by : @xmath18 , where @xmath19 .
moreover , we denote @xmath20 as @xmath1 .
the hamiltonian ( [ ham0 ] ) , expressed in terms of those parameters is given by @xmath21-\alpha \sum_{i=1}^n s_i^z . \label{ham1}\ ] ] for @xmath4 hamiltonian ( [ ham1 ] ) describes fully isotropic interaction , whereas @xmath22 corresponds to the ising model .
we consider the ferromagnetic case , i.e. @xmath23 , @xmath24 .
the hamiltonian ( [ ham1 ] ) can be rewritten in the language of free fermions .
it is done by the so called _ jordan - wigner _ transformation @xcite , @xcite . for the free - fermion hamiltonian
one can explicitely calculate all eigenvalues and in particular obtain the expression for the ground - state energy . in thus section
we give the main points of the jordan - wigner transformation .
the details can be found for instance in @xcite , @xcite .
the first step is to write hamiltonian ( [ ham1 ] ) with @xmath25 in terms of spin - raising and spin - lowering operators @xmath26 the hamiltonian then reads @xmath27 the algebraic properties of operators @xmath28 are nonuniform , i.e. the raising and lowering operators from the same site anticommute , while operators from different sites commute .
the solution of this problem is the jordan - wigner transformation , which introduces the set of @xmath29 fermionic operators @xmath30 , @xmath31 given by @xmath32 the operators @xmath33 and @xmath34 satisfy purely fermionic commutation rules , i.e. @xmath35 the hamiltonian of the open chain then reads @xmath36-\alpha c_i^\dagger c_i\right ) % + \nonumber \\ + \frac{1}{2}\left(c_n^\dagger c_{1}+\gamma c_n^\dagger c_{1}^\dagger+h.c.\right ) % \left(\exp\left(i\pi \mathfrak{n}\right)+1\right ) .
\label{ham_open}\end{gathered}\ ] ] in calculating the ground state energy in the limit of large @xmath7 , one can impose boundary conditions that are more convenient from the computational point of view , and therefore obtain the _ c - cyclic chain _
@xcite , @xcite @xmath37-\alpha c_i^\dagger c_i\right),\,\,n+1\equiv1 . \label{ham_c}\ ] ] the hamiltonians ( [ ham_open ] ) , ( [ ham_c ] ) are quadratic forms in fermion creation and annihilation operators and can be diagonalised , i.e. written in the following form @xmath38 where @xmath39 are some fermionic operators .
assume , we have found the eigenvalues @xmath40 of hamiltonian ( [ ham_a ] ) . in order to find the ground state energy , @xmath41 ,
let us calculate the trace of hamiltonians ( [ ham_a ] ) and ( [ ham_diag ] ) @xmath42 the formula for the ground state energy per spin in the thermodynamic limit does not depend on the boundary conditions , because the difference between the c - cyclic and open chain tend to zero in the thermodynamic limit .
therefore , in all cases , the ground state energy can be calculated from equation ( [ trace ] ) , i.e. @xmath43 the eigenvalues for the cyclic boundary conditions are given by @xmath44 ^ 2+\gamma^2\sin^2(k),\ k=\frac{2\pi m}{n},\ m=0,1,\dots , n-1 . \label{eigenvs}\ ] ] this expression was derived for the first time by katsura in @xcite .
using fomulae ( [ eg_sum ] ) and ( [ eigenvs ] ) in the thermodynamic limit , one obtains the ground state energy in a form of the following integral @xmath45 the procedure of calculation of this integral is lengthy and tedious but straightforward . we skip it and present the final result @xcite : being not aware of paper @xcite .
after the first version of our manuscript was completed , prof .
j. stolze drew our attention to this paper . ]
the ground state energy @xmath46 is given by formulas @xmath47}\;,\;\;\;k=\sqrt{\frac{1-\alpha^2-\gamma^2}{1-\alpha^2 } } , & \mathrm{for}\ \alpha^2+\gamma^2<1 \\ -\frac{1-\alpha^2}{\pi\gamma}\bigg{[}\pi\big{(}\alpha^2;k\big{)}-k(k)+\frac{\gamma^2}{1-\alpha^2}e(k)\bigg{]}\;,\;\;\;k=\frac{\sqrt{\alpha^2+\gamma^2 - 1}}{\gamma } , & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha<1 \\ -\frac{\alpha^2 - 1}{\pi\sqrt{\alpha^2+\gamma^2 - 1}}\bigg{[}\pi\big{(}\frac{1}{\alpha^2};k\big{)}-k(k)+\frac{\alpha^2+\gamma^2 - 1}{\alpha^2 - 1}e(k)\bigg{]}\;,\;\;\ k=\frac{\gamma}{\sqrt{\alpha^2+\gamma^2 - 1 } } & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha>1 . \end{array } \right .
\label{eg_final}\ ] ] the above results are formulated in terms of complete elliptic integrals @xcite : @xmath48 which are called complete elliptic integrals of the first , second and the third kind , respectively . to make the expression more transparent
, we plot the function ( [ eg_final ] ) on the fig .
[ energy_map ] . .
one characteristic circle - shaped level set can be seen .
this is the circle @xmath5 , where @xmath49 . ]
the free energy per particle of the system described by a hamiltonian @xmath50 is of the form @xmath51 , where @xmath52 and @xmath53 is the inverse temperature .
when the system is in its ground state , one has to calculate the limit of @xmath54 as @xmath53 approaches infinity .
therefore @xmath55 then the ( transverse ) magnetization and the magnetic susceptibility are simply given by @xmath56 differentiating equations ( [ eg_final ] ) , one obtains @xmath57}\;,\;\;\;k=\sqrt{\frac{1-\alpha^2-\gamma^2}{1-\alpha^2 } } , & \mathrm{for}\ \alpha^2+\gamma^2<1 \\ \frac{1-\alpha^2}{\pi\alpha\gamma}\big{[}\pi\big{(}\alpha^2;k\big{)}-k(k)\big{]}\;,\;\;\ ; k=\frac{\sqrt{\alpha^2+\gamma^2 - 1}}{\gamma } , & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha<1 \\ \frac{\alpha^2 - 1}{\pi\alpha\sqrt{\alpha^2+\gamma^2 - 1}}\pi\big{(}\frac{1}{\alpha^2};k\big{)}\;,\;\;\ ; k=\frac{\gamma}{\sqrt{\alpha^2+\gamma^2 - 1 } } & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha>1 .
\end{array } \right .
\label{magnetization}\ ] ] and @xmath58}{\pi(\alpha^2+\gamma^2 - 1)}\;,\;\;\ ; k=\frac{\sqrt{\alpha^2+\gamma^2 - 1}}{\gamma } , & \mathrm{for}\ \alpha^2+\gamma^2>1,\
\alpha<1 \\ -\frac{k(k)-e(k)}{\pi\sqrt{\alpha^2+\gamma^2 - 1}}\;,\;\;\ ; k=\frac{\gamma}{\sqrt{\alpha^2+\gamma^2 - 1 } } & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha>1 .
\end{array } \right .
\label{susceptibility}\ ] ] the exemplary plots are shown on figure . [ magn_and_susc ] .47 .,title="fig : " ] .47 .,title="fig : " ] as one can see in figure [ susc ] , the magnetic susceptibility approaches infinity near @xmath3 . using the series expansions of the complete elliptic integrals near @xmath59 @xcite , one has that for @xmath60 small @xmath61+\frac{8 - 9\log2-\gamma^2 - 6\log\gamma}{4\pi\gamma^3}|1-\alpha|+ \\ ] ] @xmath62 another singularity arises when the second derivative of the energy with respect to @xmath1 is evaluated around @xmath4 ( line _ ii ) _ ) .
one may expect such a behaviour , because for @xmath63 the x - coupling constant , @xmath64 , is greater than the y - coupling constant , @xmath65 ( see eqs .
( [ ham0 ] ) and ( [ ham1 ] ) ) , while for @xmath66 we have @xmath67 .
this means that while crossing the line @xmath4 , the direction of magnetization changes from @xmath68 to @xmath69 direction .
it turns out that such a transition occurs as an ising type singularity , i.e. as a logarithmic divergence of the second derivative . using the expansions of the elliptic integrals around @xmath70 , @xmath71 or @xmath72 ,
depending on the value of @xmath73 for @xmath2 close to @xmath72 , one can see that this is the case for @xmath74 and @xmath75 small @xmath76+\frac{1}{4\pi\sqrt{1-\alpha^2}}\big(15 - 21\alpha^2+\ ] ] @xmath77\gamma^2 + \mathcal{o}(|\gamma|^3 ) .
\label{rozwiniecie2}\ ] ] for @xmath78 the ground state energy is smooth .
_ the transverse susceptibility diverges on both lines @xmath3 and @xmath4 in a logarithmic manner .
but the behaviour of the correlation functions is different .
the corresponding universality classes for these transitions are called ising model and two decoupled ising models , respectively .
they are governed by different correlation functions critical exponents @xcite .
these facts are well known . but using an exact solution , one can obtain the sub - leading exponents as well as amplitudes up to arbitrarily high order .
equations ( [ rozwiniecie1 ] ) and ( [ rozwiniecie2 ] ) present the first terms of the expansion . in this section
we will give the formula for all partial derivatives of function @xmath79 ( see equation ( [ eg_final ] ) ) with respect to parameter @xmath1 in the limit @xmath80 .
as shown in section [ section : magn_susc ] , the second derivatives of @xmath46 are given by @xmath81}{\pi(\alpha^2+\gamma^2 - 1)}\ ; , \;\;\ ; k=\frac{\sqrt{\alpha^2+\gamma^2 - 1}}{\gamma } , & \mathrm{for}\ \alpha^2+\gamma^2>1,\ \alpha<1 .
\label{2nd_long } \end{array } \right.\ ] ] we will next write the above formula in a more compact form , using identities @xmath82 which follow from equations 17.3.29 and 17.3.30 in @xcite . for @xmath83 ,
equations ( [ identities ] ) read @xmath84 where @xmath85 . inserting this result to formula ( [ 2nd_long ] ) ,
one obtains that @xmath86}{\pi(\alpha^2+\gamma^2 - 1)}=-\frac{1}{\pi\gamma}\frac{\left[e\left(\tilde k\right)-k\left(\tilde k\right)\right]}{\tilde k^2},\ \ \alpha<1.\ ] ] we will next use the identity @xmath87}{k^2},\ ] ] which can be derived using definitions of complete elliptic integrals ( equations ( [ elliptic ] ) ) simply by subtraction of the integrands on the right hand side and by differentiation under the integral sign on the left hand side .
the second derivative of the ground state energy now reads @xmath88 higher derivatives of the ground state energy can be therefore calculated as derivatives of the composite function @xmath89 . to this end
, we will use the following formula for the @xmath90th derivative of the composition of two functions @xcite @xmath91 where the sum runs through all positive integers @xmath92 such that @xmath93 and @xmath94 .
note that in the case at hand , the third derivarive of @xmath95 vanishes , hence equation ( [ nthderivative_gen ] ) simplifies .
namely , @xmath96 combining the above formula with equation ( [ gse_2nd_der ] ) , one obtains @xmath97 where the derivatives of function @xmath98 are known from its series expansion @xcite
@xmath99 ^ 2\frac{1}{1 - 2m}\ ] ] both limits of derivatives : left and right ones are equal up to an arbitrary order , so the ground - state energy @xmath46 is infinitely differentiable on the circle @xmath5 .
we have also checked the smoothness of the energy numerically , by plotting the derivatives of the ground state energy of order up to five along some sample curves passing transversally through the circle @xmath5 .
the results confirmed our calculations within numerical accuracy . because the character of correlations changes while crossing the line @xmath5 , one could expect that it should result in some kind of singularity .
however , as we have shown in previous paragraph , the ground state energy is smooth on the aforementioned circle .
such a behaviour resembles the _ kosterlitz - thouless transition _ , where certain thermodynamic quantities are smooth .
so one can ask whether this transition is present in the considered xy spin chain .
let us shortly review the main properties of the k - t transition , necessary to verify this hypothesis .
the kosterlitz - thouless transition occurs in many two - dimensional classical models : coulomb gas , sine - gordon , xy and many others @xcite .
there is no spontaneous symmetry breaking and the free energy , although non - analytic , is infinitely differentiable during this transition .
the singular part of the free energy behaves as : @xmath100 near the critical temperature @xmath101 ( @xmath102 is a positive constant ) .
the kosterlitz - thouless transition occurs also in ground states of one - dimensional quantum chains under change of certain parameter @xmath103 ( for instance , an anisotropy parameter in the quantum xxz chain @xcite ; the @xmath104 quantum chain @xcite ; the ashkin - teller chain @xcite , @xcite ; spin-1 heisenberg models @xcite ) . in all cases ,
the ground - state energy is infinitely differentiable , and the energy gap @xmath105 over the ground state behaves near the critical value of parameter @xmath106 as @xmath107 ( @xmath102 is a positive constant ) .
the physical origin of the kosterlitz - thouless sometimes is clear @xcite , but sometimes is somewhat hidden @xcite , @xcite .
so we decided to check whether the k - t transition is present in the xy model . in the following paragraph
, we will investigate the aforementioned gap energy i.e. the difference between the energy of the first excited state and the energy of the ground state .
first , we will reproduce in an alternative way the well known result regarding the periodic chain .
then , we will show the results of the numerical calculations for an open chain , which reveal that the behaviour of the energy gap is qualitatively the same in both cases .
while the energy gap for the cyclic chain is well - studied , the open chain , to our best knowledge , has not yet been investigated .
the main reason for this is the lack of the translational symmetry , which makes the exact diagonalization far more difficult . from equation ( [ ham_diag ] )
, one can easily see that the gap energy is given by the minimal eigenvalue of the hamiltonian @xmath108 let us first reexamine the case of the @xmath109-cyclic boundary conditions . in order to find the minimal eigenvalue ( equation ( [ gap_general ] ) ) in the limit of @xmath110 ,
one has to solve the equation @xmath111 , i.e. @xmath112 the gap energy in then given by @xmath113 and it is clear that it does not describe the kosterlitz - thouless transition .
this result is consistent with the first - order of the expansion for the energy gap , which was derived in @xcite .
let us next switch our attention to the open boundary conditions .
the hamiltonian of an open chain after the jordan - wigner transformation is given by ( [ ham_open ] ) .
because of the lack of the translational symmetry , the exact diagonalization of the fermionic hamiltonian is more difficult , i.e. eigenvalues of one - particle problem are not analytically known for arbitrary @xmath7 .
however , one can do the numerical diagonalization .
the numerical calculations show that also in the case of an open chain , the gap energy behaves like @xmath114 for sufficiently long chains .
the exemplary plots depicting this phenomenon are shown in figure [ fig : gap_n ] , where the line @xmath115 was fitted .
.47 .47 the fitted values of @xmath116 are equal to zero within the accuracy of @xmath117 and @xmath118 respectively .
the values of parameter @xmath119 are also fitted very accurately - up to the magnitude of @xmath120 and @xmath121 respectively .
since @xmath116 is zero in the whole range of parameters , there is no kosterlitz - thouless transition .
however , the behaviour of @xmath122 in the limit or large @xmath7 , i.e. parameter @xmath119 from equation ( [ gap_fit ] ) , varies significantly on the choice of the boundary conditions .
however , in the strong - field region , @xmath78 , plots on figure ( [ fig : gap_n ] ) suggest that @xmath123 regardless on the choice of the boundary conditions .
it turns out that the behaviour of energy gap in the considered xy chain is _ not _ compatible with behaviour described by ( [ ktinqchains ] ) .
it is so due to the following reasons . *
consider the ordered phase , i.e. @xmath124 ( and @xmath74 ) . in this ordered phase
, the system exhibits the @xmath125 symmetry @xcite ( corresponding to plus and minus sign of magnetization ) , which implies the ( asymptotic ) degeneracy of two ground states .
so , the energy gap is zero in the ordered phase ( in the thermodynamic limit ) . *
consider the oscillating phase ( i.e. @xmath126 ) and take the line @xmath127const .
it turns out that along this line , the gap _
oscillates_. such a behaviour has been observed in @xcite ; we illustrate it on fig .
[ fig : gap_n ] .
moreover , for all values of @xmath1 , the gap tends to 0 in the thermodynamic limit .
.47 can be seen . moreover , with the raise of the chain length the gap tends to zero and the oscillations are faster . for a detailed description ,
see @xcite.,title="fig : " ] .47 can be seen .
moreover , with the raise of the chain length the gap tends to zero and the oscillations are faster . for a detailed description ,
see @xcite.,title="fig : " ] .47 can be seen .
moreover , with the raise of the chain length the gap tends to zero and the oscillations are faster . for a detailed description ,
see @xcite.,title="fig : " ] .47 can be seen .
moreover , with the raise of the chain length the gap tends to zero and the oscillations are faster . for a detailed description ,
see @xcite.,title="fig : " ] we have confirmed numerically that the value of energy gap , extrapolated to @xmath128 , _ is zero_. this observation precludes ( in a numerical manner ) the occurrence of the kosterlitz - thouless transition in the xy chain .
the distinguishing feature of k - t transition is the exponential closing of the gap together with the smoothness of ground - state energy at the transition point . in our case ,
energy gap is asymptotically zero in the neighbourhood of transition point , therefore the gap is not an appropriate object for testing analyticity of the ground - state energy . intuitively , one could conjecture that the non - analyticity combined with the smoothness of ground - state energy should be somehow related to the structure of energy levels , but we are not able to specify such a relation .
more precisely , we do nt know how to formulate asymptotic behaviour of energy levels in the neighbourhood of the circle @xmath5 , and leave this as an open problem .
we only conclude that the transition between oscillating and ferromagnetic phases is not of the kosterlitz - thouless type .
we have analysed the the formula for ground - state energy of the anisotropic xy model in transverse magnetic field .
we have examined differentiability properties of the expression .
it turned out that the second derivative of energy ( i.e. the transverse susceptibility ) is logarithmically divergent in the neighbourhood of the lines @xmath3 and @xmath4 .
this behaviour is characteristic for ising - type phase transition .
this fact has been known earlier ( see for instance @xcite ) , but here one can obtain sub - leading exponents and amplitudes up to arbitrary order . we have presented some of first terms of such an expansion . an interesting open problem ( as far as we know ) is the question of _ universality _ of ising - type critical behaviour near these lines .
the universality principle has been formulated more than 40 years ago , but rigorous proofs are very rare .
one of such results is the proof of universality for 2d ising model due to spencer @xcite
. it would be very interesting to adapt the technique used in @xcite to quantum xy chain , for lines @xmath3 and @xmath4 .
on the physical grounds , such universality is expected , but as far as we know the rigorous proof is lacking .
we have also established that the energy is differentiable up to arbitrary order on the line @xmath5 .
again , it is interesting to check whether this behaviour is stable ( universal ) against perturbations of hamiltonian .
another aspect of the smoothness of energy on the line @xmath5 , it is interesting to confront this result with paper @xcite , where differentiability property of _ renyi entropy _ has been analysed .
the result there is that derivative of renyi entropy is discontinuous on the line @xmath5 .
apparently , differentiability properties of energy and entropy are different .
it seems strange and it would be interesting to take a closer look at this question . * acknowledgments .
* we are grateful to prof .
j. stolze for drawing our attention to paper @xcite .
tomasz maciek is supported by polish ministry of science and higher education `` diamentowy grant '' no .
di2013 016543 and erc grant qolaps .
99 lieb , e. h. , schulz , t. , and mattis , d. c. : _ ann .
( n. y. ) _ * 16 * , 406 ( 1961 ) .
henkel , m. : _ conformal invariance and critical phenomena _ texts and monographs in physics .
springer verlag , 1999 .
bunder , j. e. and mckenzie , r. h. : _ phys .
rev . _ * b 60 * , 344 ( 1999 ) .
franchini , f. , its , a. r. , and korepin , v. e. : _ j. phys . a : math .
theor . _ * 41 * , 025302 ( 2008 ) ; arxiv 0707.2534 [ quant - phys ] franchini , f. , its , a. r. , jin , b .- q . , and korepin , v. e. : _ j. phys . a : math .
_ * 40 * , 8467 ( 2007 ) batle , j. and casas , m. : _ phys
_ * a 82 * , 062101 ( 2010 ) franchini , f. , its , a. r. , korepin , v. e. , and takhtajan , l. a. : _ quantum inf . process _ * 10 * , 325 ( 2011 ) eisler , v. , karevski , d. , platini , t. and peschel , i. : _ j. stat
_ p01023 ( 2008 ) arxiv:0711.0289 wendenbaum , p. , platini , t. , karevski , d. : _ phys .
rev . _ * a 91 * , 040303 ( 2015 ) .
pfeuty , p. : _ ann . phys .
( n. y. ) _ * 57 * , 79 ( 1970 ) .
barouch , e. , mccoy , b. m. and dresden , m. : _ phys .
rev . _ * a 2 * , 1075 ( 1970)i barouch , e. and mccoy , b. m. : _ phys .
_ * a 3 * , 786 ( 1971 ) taylor , j. h. and mller , g. : _ physica _ * 130 a * , 1 ( 1985 ) .
vaidya , h. g. and tracy , c. a. : _ physica _ * 92 a * , 1 ( 1978 ) .
jimbo , m. , miwa , t. , mori , y. , and sato , m. : _ physica _ * 1 d * , 80 ( 1980 ) .
karevski , d. : _ j. phys .
_ * a33 * , l313 ( 2000 ) arxiv 0009038 [ cond - mat ] kurmann , j. , thomas , h. and mller , g. : _ physica _ * 112 a * , 235 ( 1982 ) hoeger , c. , von gehlen , g. and rittenberg , v. : _ j. phys .
_ * a 18 * , 1813 ( 1985 ) den nijs , m. : _ the domain wall theory of two - dimensional commensurate incommensurate phase transitions .
_ in : _ phase transitions and critical phenomena _ vol .
by c. domb and j. l. lebowitz , academic press , 1988 .
mattis , d. c. : _ the theory of magnetism ii _ , springer series in solid state sciences v. 55 , springer verlag , berlin , heidelberg 1985 .
katsura , s. : _ phys .
rev . _ * 127 * , 1508 ( 1962 ) .
fichtenholz , g. m. : _ rachunek rniczkowy i cakowy ( differential and integral calculus ) _ , vol .
2 . wydawnictwo naukowe pwn , warszawa 2013 abramowitz , m. and stegun , i. a. ( eds . ) .
_ handbook of mathematical functions with formulas , graphs , and mathematical tables _ , 9th printing .
new york : dover , 1972 whittaker , e. t. and watson , g. n. _ a course in modern analysis _
cambridge , england : cambridge university press , 1990 h .- n .
huang , s. a. m. marcantognini and n. j. young _ chain rules for higher derivatives _
, 2005 http://www1.maths.leeds.ac.uk/ nicholas / abstracts / faadibruno3.pdf kosterlitz j. m. and thouless , d. j. : _ j. phys .
_ * c 6 * , 1181 ( 1973 ) amaj , l. : _ acta phys .
slovaca _ * 60 * , 155 ( 2010 ) allton , c. r. and hamer , c. j. :
_ j. phys .
_ * a 21 * , 2417 ( 1988 ) kohmoto , m. , den nijs , m. , and kadanoff , l.p . : _ phys . rev . _
* b 24 * , 5229 ( 1981 ) baake , m. , von gehlen , g. and rittenberg , v. : _ j. phys .
_ * a 20 * , l479 and l487 ( 1987 ) neirotti , j. , p. , de oliveira , m. , j. , _ phys
. rev . _ * b 59 * , 5 ( 1998 ) henkel , m. : _ j. phys . _ * a 20 * , 995 ( 1987 ) spencer , t. : _ physica _ * 279 a * , 250 ( 2000 ) | we investigate an explicite formula for ground state energy of the anisotropic xy chain in transverse magnetic field .
in particular , we examine the smoothness properties of this expression .
we explicitly demonstrate that the ground - state energy is infinitely differentiable on the boundary between ferromagnetic and oscillatory phases .
we also confirm known 2d - ising type behaviour in the neighbourhood of certain lines of phase diagram and give more detailed information there , calculating a few next - to - leading exponents as well as the corresponding amplitudes . |
recent observations on the origin of the uv upturn phenomenon of elliptical galaxies showed that hot hb and post - hb stars are the dominant uv sources in these systems ( ferguson et al .
1991 ; oconnell et al . 1992 ; king et al . 1992 ; bertola et al . 1995 ; brown et al 1997 ) . as demonstrated in our recent investigation ( yi et al .
1999 ) , it is still not clear , however , whether the dominant uv sources are metal poor ( metal - poor hb model " ; lee 1994 ; park & lee 1997 ) or metal rich ( metal - rich hb model " ; bressan et al .
1994 ; dorman et al 1995 ; yi et al 1998 ) .
the metal - poor hb model " suggests that the dominant uv sources are very old , hot metal - poor hb stars and their post - hb progeny , although metal - rich pagb stars also contribute some uv flux . in this picture ,
nearby giant ellipticals are about 3 gyr older than the milky way galaxy , which would imply non - zero cosmological constant from time scale test . on the other hand , the metal - rich hb model
" suggests that the dominant uv sources are super metal - rich hot hb stars that experienced enhanced mass - loss and helium enrichment .
if so , nearby giant ellipticals are similar in age to the milky way galaxy , and there is no conflict with zero @xmath2 cosmology from time scale test . clearly , it is of considerable importance to understand the origin of the uv upturn .
considering this situation , we have recently investigated the evolution of uv flux with look - back time to see if two models predict substantially different evolutionary patterns .
we found that this would indeed provide some strong observational test on the origin of the uv upturn phenomenon ( yi et al .
the major uncertainty in these model calculations is the mass of pagb stars , since it is only poorly constrained from direct observations . in this paper , we report our progress in investigating the effect of pagb mass in the modeling of uv evolution of elliptical galaxies .
our model calculations show that hot hb and post - hb stars contribute most of the uv light in nearby giant ellipticals ( fig .
1 ) . as lookback time increses , however , pagb stars are getting more dominat uv sources since the mean masses of helium burning stars are too high to be hot enough hb stars at younger ages .
( at very large look - back times , hot main sequence stars eventually become the major uv source . ) since the contribution from pagb stars overwhelms the uv spectrum at higher redshifts , the model sed is strongly affected by the current uncertainty of the pagb mass as look - back time increases .
we present , in figure 2 , the 1500 - v color evolution of giant ellipticals predicted by both the metal - poor " and metal - rich " hb models under two assumptions regarding the pagb mass .
it is clear from figure 2 that the 1500 - v is not strongly affected by the uncertainty of the pagb mass for the nearby giant ellipticals since the pagb contribution to the total uv flux is less than 10% .
pagb treatment becomes significant , however , for the ellipticals at @xmath3 .
fortunately , both metal - poor " and metal - rich " models predict more or less the same 1500 - v colors at @xmath0 , and thus far - uv observations at these redshifts ( cf .
brown et al . 1998 ) would provide crucial constraint on the pagb mass .
our models in figure 2 suggest that the observations at lower redshifts ( @xmath1 ) would still provide strong constraints on two models on the origin of the uv upturn .
future observations of ellipticals from the _ stis / hst _ and planned _ galex _ ( martin et al 1998 ) space uv facility will provide crucial database required for more concrete calibration of our uv dating techniques for old stellar systems .
bertola , f. , et al .
1995 , , 438 , 680 bressan , a. , chiosi , c. , & fagotto , f. 1994 , , 94 , 63 brown , t. m. , ferguson , h. c. , davidsen , a. f. , & dorman , b. 1997 , , 482 , 685 brown , t. m. , ferguson , h. c. , deharveng , j
.- m . , & jedrzejewski , r. 1998 , , 508 , l139 ferguson , h. c. , et al . 1991 , , 382 , l69 king , i. r. , et al .
1992 , , 397 , l35 lee , y .- w .
1994 , , 430 , l113 martin , c. , et al .
1998 , , 29 , 1309 oconnell , r. w. , et al . 1992 . , 395 , l45 park , j .- h .
, & lee , y .- w .
1997 , , 476 , 28 yi , s. , demarque , p. , & oemler , a. 1998 , , 492 , 480 yi , s. , lee , y .- w . ,
woo , j .- h . , park , j .- h . ,
demarque , p. , & oemler , a. 1999 , , 513 , 128 | the uv upturn phenomenon observed in elliptical galaxies is attractive for its potential value as an age indicator of old stellar systems .
we present our most recent population models for the uv evolution of elliptical galaxies .
we confirm that the dominant uv sources are either metal - poor or metal - rich hot horizontal - branch ( hb ) stars in local giant ellipticals , but we also note that the contribution from post - asymptotic - giant - branch ( pagb ) stars overwhelms the uv spectrum at higher redshifts ( look - back times ) .
the model uv spectral energy distribution ( sed ) is therefore strongly affected by the current uncertainty of the mean mass of pagb stars at higher redshifts .
fortunately , our models suggest that the far - uv observations at @xmath0 could produce strong constraint on the pagb mass , while observations at lower redshifts ( @xmath1 ) would still provide constraints on the models on the origin of the uv upturn .
future observations of ellipticals from the _ stis / hst _ and planned _ galex _ space uv facility will provide crucial database required for more concrete calibration of our uv dating techniques for old stellar systems . |
t dwarfs are brown dwarfs that exhibit methane absorption bands at 1.6 and 2.2 @xmath1 @xcite , and thus have effective temperatures t@xmath3 @xmath4 1200 - 1300 k @xcite .
the prototype for this class , gl 229b @xcite , was identified as a cool companion to the nearby m1v star gl 229a .
recently , seven field objects @xcite and another companion object @xcite have also been identified as t dwarfs .
the rapid discovery of these cool brown dwarfs has been driven by new sky surveys , such as the two micron all sky survey ( * ? ? ? * hereafter 2mass ) and the sloan digital sky survey @xcite ; and deep near - infrared surveys , such as the eso new technology telescope deep field @xcite .
the t dwarfs identified to date are remarkably similar to gl 229b , with colors in the range @xmath50.2 @xmath4 j - k@xmath6 @xmath4 0.2 .
near - infrared spectra are correspondingly similar @xcite , likely due to the saturation of h@xmath7o and ch@xmath2 bands that dominate this wavelength regime .
subtle differences in the magnitudes and shapes of h- and k - band flux peaks are observed , due to increased ch@xmath2 , h@xmath7o , and h@xmath7 collision - induced absorption ( cia ) toward cooler effective temperatures @xcite , and variations in the depths of near - infrared h@xmath7o and ch@xmath2 bands are discerned when compared to gl 229b @xcite .
nonetheless , the similarity of the near - infrared spectra suggests that either the objects thus far identified are very similar in temperature , around 1000 k @xcite , or that near - infrared features are fairly insensitive to temperature , making the definition of a t dwarf spectral sequence in this wavelength regime a difficult proposition , at least at low resolution .
we report the discovery of a t dwarf by 2mass which is unique among its counterparts , as it shows significant differences in its near - infrared features while retaining defining ch@xmath2 bands .
this object , 2massi j0559191 - 140448 ( hereafter 2mass j0559 - 14 ) , is also 0.4 mag brighter than gl 229b and more than 1 mag brighter than the field t dwarfs discovered thus far .
it is the first brown dwarf to be identified by the newly commissioned cornell massachusetts slit spectrograph ( * ? ? ?
* hereafter cormass ) , mounted on the palomar 60-inch telescope . in @xmath82
we discuss the selection of 2mass j0559 - 14 from 2mass data and its spectral identification by cormass . in @xmath83
we discuss the observed spectral features and argue that 2mass j0559 - 14 is a warm t dwarf , possibly close to the transition temperature between l and t spectral classes .
we discuss the brightness of this object and its role in future spectroscopic investigation of the t dwarf class in @xmath84 .
2mass j0559 - 14 was initially selected as a t dwarf candidate from the 2mass point - source working database .
details on the selection criteria for 2mass t dwarfs are discussed in @xcite . using the palomar 60-inch infrared camera @xcite
, we successfully re - imaged this candidate on 23 september 1999 ( ut ) at j - band , confirming it as a bona - fide candidate ( i.e. , not a minor planet or artifact ) .
2mass j0559 - 14 is the brightest 2mass t dwarf candidate confirmed to date , with j = [email protected] ( table 1 ) ; it is also the reddest candidate confirmed , with j - k@xmath6 = [email protected] .
optical serc - ej @xcite and j - band 2mass images of the 2mass j0559 - 14 field are shown in figure 1 ; no optical counterpart is seen to b@xmath9 @xmath10 23 .
low - resolution near - infrared spectral data of 2mass j0559 - 14 were obtained on 24 october 1999 ( ut ) using cormass .
this newly commissioned instrument is an r @xmath10 300 prism cross - dispersed near - infrared spectrograph , with a 256 @xmath11 256 nicmos3 array and a 40 lines mm@xmath12 grating , blazed at 4.8 @xmath1 .
cormass was designed primarily for the spectral classification of candidate low - mass objects color - selected from the 2mass database . the instrument s echelle format provides simultaneous coverage of the _
z_jhk bands , for 0.8 @xmath4 @xmath13 @xmath4 2.5 @xmath1 .
the fixed slit has a width of 2@xmath14 and a length of 15@xmath14 .
further details on this instrument can be found in @xcite .
conditions during the observations were not photometric , and estimated seeing was @xmath10 1@xmath155 , with variable thin cirrus throughout the night .
total on - source time was 2400 seconds divided into sets of 300-second integrations , nodding @xmath10 5@xmath14 along the slit between exposures .
spectra were reduced using standard iraf routines .
after correction for bad pixels , flat field and flux calibration images were corrected for nicmos3 reset - decay bias ( also known as shading ) as follows : a quadratic was fit to a row - by - row clipped median of the top quarter of the array , the portion unused by the spectrograph .
the fit was extrapolated to 128 rows for the top two quadrants , duplicated for the bottom two quadrants , and subtracted row - by - row from all pixels .
all images were flat - fielded with a pixel responsivity solution from the apflatten task using dome flats from all five nights of the run summed together
. nodded image pairs for the science object were subtracted against each other to remove sky background and reset - decay bias .
the apall task was used for spectral extraction .
wavelength calibration was accomplished using spectral observations of the planetary nebula ngc 7027 obtained on 25 october 1999 ( ut ) .
flux calibration was done by dividing by the reduced spectrum of the a2v standard hd 77281 @xcite , which was observed at similar airmass as 2mass j0559 - 14 and hand corrected for stellar h paschen and brackett recombination lines .
the ratio was then multiplied by a 8810 k @xcite blackbody to complete the flux calibration .
finally , all 300-second observations for each order were weighted by the spectra mean and combined with the scombine task using a trimmed average .
orders were stitched together with scombine after hand deletion of noisy data at the ends of the orders .
the spectrum of 2mass j0559 - 14 is shown in figure 2 , along with optical ( 0.87 - 1.0 @xmath1 ; burgasser et al .
2000b ) and near - infrared ( 1.0 - 2.35 @xmath1 ; strauss et al .
1999 ) data for sdssp j162414.37 + 002915.6 ( hereafter sdss 1624 + 00 ) .
the combined sdss 1624 + 00 spectrum was smoothed to the resolution of cormass .
spectra for both objects are normalized to one at 1.27 @xmath1 ( j - band peak ) .
close up views of 0.88 - 1.01 and 1.15 - 1.345 @xmath1 are shown in figure 3 .
table 2 summarizes the spectral features detected in 2mass j0559 - 14 , with identifications from @xcite , @xcite , @xcite , and @xcite
. only major absorption bands of h@xmath7o and ch@xmath2 are tabulated .
the characteristic ch@xmath2 bands at 1.6 and 2.2 @xmath1 are present , as are bands at 1.1 and 1.3 @xmath1 identified from laboratory data @xcite which are blended with h@xmath7o bands at the same wavelengths .
an feh feature is seen at 0.9896 @xmath1 ( 0 - 0 band of a@xmath16@xmath17-x@xmath16@xmath17 ) which has also been identified in sdss 1624 + 00 @xcite .
we do not detect the higher order 0 - 1 feh band at 1.191 @xmath1 , which is seen to weaken in the latest l dwarfs @xcite .
two sets of k i doublets are noted at 1.1690 & 1.1773 @xmath1 ( 4p @xmath18p@xmath19 @xmath5 3d @xmath18d ) and 1.2432 & 1.2522 @xmath1 ( 4p @xmath18p@xmath19 @xmath5 5s @xmath18s ) , as is cs i at 0.8943 @xmath1 ( 6s @xmath18s@xmath20 @xmath5 6p @xmath18p@xmath20 ) .
we do not detect the 2 - 0 x@xmath21@xmath22@xmath23-x@xmath21@xmath22@xmath23 band of co at 2.3 @xmath1 .
comparison between 2mass j0559 - 14 and sdss 1624 + 00 reveals significant differences in spectral morphology . in figure 2
, it is apparent that the slope between 0.9 and 1.05 @xmath1 is shallower in 2mass j0559 - 14 , likely due to decreased absorption by the pressure - broadened k i doublet at 0.7665 and 0.7699 @xmath1 @xcite .
ch@xmath2 and h@xmath7o features are generally weaker in 2mass j0559 - 14 , as noted by the significantly weakened 1.1 - 1.2 @xmath1 trough between @xmath24- and j - band peaks .
the decreased ch@xmath2 opacity noticeably affects the shape of the j - band peak near 1.27 @xmath1 , as the weak ch@xmath2 wings at 1.24 and 1.30 @xmath1 carve out less flux on either side of the peak . in figure 2 ,
a significant flux offset is readily apparent at the base of the 1.6 @xmath1 ch@xmath2 band in 2mass j0559 - 14 , and relative enhancement of flux at both h- and k - bands in this object is almost certainly due to decreased h@xmath7 cia ( 1 - 0 quadrupole ) , h@xmath7o , and ch@xmath2 opacity .
the weak ch@xmath2 bands seen in 2mass j0559 - 14 are unique among the current sample of t dwarfs , and the reduced opacity can be most readily explained if this object is warmer than other known t dwarfs . at higher effective temperatures ,
the dominant carbon - bearing species changes from co to ch@xmath2 at small optical depth , so that the ch@xmath2 column density will be less than that of cooler t dwarfs , and observed band strengths correspondingly weaker .
increased thermal flux will also be seen at the base of these bands , particularly at 1.6 @xmath1 , which is unaffected by h@xmath7o absorption .
the lower ch@xmath2 column density directly affects the h@xmath7o column density , via the reaction co + 3 h@xmath7 @xmath25 ch@xmath2 + h@xmath7o , leading to shallower bands at 1.1 and 1.45 @xmath1 .
water can also be heated and dissociated by dust layers deep in the photosphere @xcite .
finally , decreased h@xmath7 cia opacity at k - band , congruous with reduced ch@xmath2 and h@xmath7o opacity , will result in redder j - k@xmath6 colors with warmer t@xmath3 .
these features are observed in 2mass j0559 - 14 , and its warm temperature is independently supported by the detection of the 0.9896 @xmath1 feh band , which is present but weakening in the latest l dwarfs @xcite .
a similar argument has been made for sdss 1624 + 00 by @xcite , which was one of only three t dwarfs in that paper to show this feature .
sdss 1624 + 00 has been shown to be a warm object via optical continuum measurements between broadened na i and k i features @xcite , while @xcite argue that this object is both warmer and dustier than gl 229b based on its shallower ch@xmath2 and h@xmath7o bands . by analogy , 2mass j0559 - 14 should be warmer still .
the detection of excited k i lines at j - band , which are seen to weaken in the latest l dwarfs @xcite , is further evidence of the warmth of this object .
the spectral features in 2mass j0559 - 14 suggest that it is close to the l / t transition temperature .
the lack of co detection at 2.3 @xmath1 is not necessarily contradictory to this hypothesis , as overlying ch@xmath2 absorption beyond 2.2 @xmath1 may mask this weaker feature .
naturally , metallicity , dust , and gravity could also play roles in the band strengths seen in 2mass j0559 - 14 ; however , temperature is likely to be the dominant determinant given the concordance of spectral features as discussed above .
we can make a conservative t@xmath3 constraint for this object based on the temperature of gl 229b , which is clearly cooler , and a temperature estimate of the l8v companion dwarf gl 584c @xcite ; this translates into a range of 1000 @xmath4 t@xmath3 @xmath4 1300 k. parallax and bolometric luminosity measurements of this object would allow a direct determination of temperature .
the relative brightness of 2mass j0559 - 14 as compared to other t dwarfs suggests that it may be a nearby brown dwarf .
we can estimate its distance using simple scaling arguments based on gl 229b = [email protected] pc @xcite , t@xmath3 = 960@xmath070 k @xcite , r @xmath26 r@xmath27 = 7.1 @xmath11 10@xmath28 cm @xcite , j = [email protected] , l = ( [email protected] ) @xmath11
10@xmath29 l@xmath30 , and bc@xmath9 = [email protected] @xcite . ] .
first , if we assume this object has the same intrinsic luminosity as gl 229b , we estimate its distance to be 4.6 pc .
however , 2mass j0559 - 14 is probably a warmer object than gl 229b , so it should be more distant .
an alternate estimate can be made if we adopt a radius r @xmath26 r@xmath31 and bounding t@xmath3 @xmath4 1300 k , so that l @xmath4 2.2 @xmath11 10@xmath32 l@xmath30 . using a gl 229b bolometric correction
, we obtain m@xmath9 @xmath33 14.2 and thus @xmath34 @xmath4 8.4 pc .
note , however , that an l8v has m@xmath9 = 15.0 @xcite , implying that our bc@xmath9 value may be too high ; indeed , bc@xmath9 may be only 1.3 for an l8v @xcite . adopting a smaller bc@xmath9 yields a smaller distance .
therefore , barring multiplicity , 2mass j0559 - 14 is probably 5 - 8 pc distant , and likely falls within the 8 pc nearby star sample defined by @xcite .
the brightness of this object allows substantial follow - up over a broad wavelength range , particularly for @xmath13 @xmath35 2.5 @xmath1 , where the fundamental absorption bands of h@xmath7o ( 2.7 @xmath1 ) , nh@xmath36 ( 3.0 @xmath1 ) , ch@xmath2 ( 3.3 @xmath1 ) , h@xmath7s ( 3.8 @xmath1 ) , and co ( 4.7 @xmath1 ) are found .
the latter 1 - 0 x@xmath21@xmath22@xmath23-x@xmath21@xmath22@xmath23 band of co can aid in constraining its atmospheric abundance @xcite , essential in testing hypotheses of mixing @xcite and transparency @xcite in t dwarf atmospheres .
deuterated molecules , such as ch@xmath36d , are of interest for study in the 3 - 5 @xmath1 range , as they yield information on the deuterium burning history of these brown dwarfs .
mid - infrared wavelengths ( 8 - 14 @xmath1 ) are sensitive to vibration - rotation bands of alkali chlorides and sulfides , which are more abundant than their atomic alkali counterparts at low temperatures and high pressures @xcite and may serve as excellent temperature discriminants .
this region also contains the strong fundamental @xmath37@xmath7 band of nh@xmath36 ( 10.5 @xmath1 ) and various silicate features .
additionally , optical spectra shortward of 0.9 @xmath1 are critical in determining the behavior of the alkalis in cool brown dwarfs , particularly pressure - broadened na i ( 0.5890 & 0.5896 @xmath1 ) and k i ( 0.7665 & 0.7699 @xmath1 ) doublets , and cs i lines at 0.8521 and 0.8943 @xmath1 , which may also be used as temperature discriminants @xcite .
while these features have been seen in t dwarfs such as sdss 1624 + 00 @xcite , detailed investigation of li i ( 0.6708 @xmath1 ) and rb i ( 0.7800 & 0.7948 @xmath1 ) lines have been hampered by the lack of detectable flux at these wavelengths .
investigation in this spectral regime is currently underway .
we thank m. strauss for the kind use of the sdss 1624 + 00 near - infrared spectrum and acknowledge useful discussions with c. griffith and m. marley .
we thank our anonymous referee for helpful comments and suggestions .
we also recognize the efforts of the 2mass staff and scientists in creating a truly incredible astronomical resource , and the expert assistance of the palomar observatory staff during imaging and spectroscopic observations .
a. j. b. , j. d. k. , and j. e. g. acknowledge the support of the jet propulsion laboratory , california institute of technology , which is operated under contract with the national aeronautics and space administration .
j. c. w. acknowledges support by nasa grant nag5 - 4376 .
observations from palomar observatory were made as part of a continuing collaboration between the california institute of technology and cornell university .
the digitized sky survey was produced at the space telescope science institute under us government grant nag w-2166 .
serc - ej data were scanned by dss from photographic data obtained using the uk schmidt telescope , operated by the royal observatory edinburgh , with funding from the uk science and engineering research council .
dss images were obtained from the canadian astronomy data centre , which is operated by the herzberg institute of astrophysics , national research council of canada .
iraf is distributed by the national optical astronomy observatories , which are operated by the association of universities for research in astronomy , inc . , under cooperative agreement with the national science foundation .
this publication makes use of data from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center , funded by the national aeronautics and space administration and the national science foundation .
arnouts , s. , dodorico , s. , cristiani , s. , zaggia , s. , fontana , a. , & giallongo , e. , , 341 , 641 auman , j. , jr .
1967 , , 14 , 171 basri , g. , mohanty , s. , allard , f. , hauschildt , p. h. , delfosse , x. , martn , e. l. , forveille , t. , & goldman , b. 2000 , , in press burgasser , a. j. , et al . 2000a , , 531 , l57 burgasser , a. j. , kirkpatrick , j. d. , reid , i. n. , liebert , j. , gizis , j. e. , & brown , m. e. 2000b , , 120 , in press burgasser , a. j. , et al . 1999 , , 522 , l65 burrows , a. , marley , m. s. , & sharp , c. m. 2000 , , 531 , 438 burrows , a. , & sharp , c. m. 1999 , , 512 , 843 burrows , a. , & liebert , j. 1993 , rev.mod
65 , 301 cuby , j. g. , saracco , p. , moorwood , a. f. m. , dodorico , s. , lidman , c. , comern , f. , & spyromilio , j. 1999 , , 349 , l41 danielson , r. e. 1966 , , 143 , 949 elias , j. h. , frogel , j. a. , matthews , k. , & neugebauer , g. 1982 , , 87 , 1029 fegley , b. , jr . , & lodders , k. 1996 , , 472 , l37 fink , u. , & larson , h. p. 1979
, , 233 , 1021 griffith , c. a. , & yelle , r. v. 1999 , , 519 , l85 gunn , j. e. , & weinberg , d. h. 1995 , in wide - field spectroscopy and the distant universe , ed .
s. maddox & a. aragn - salamanca ( singapore : world scientific ) , p. 3
kirkpatrick , j. d. , reid , i. n. , liebert , j. , gizis , j. e. , burgasser , a. j. , monet , d. g. , dahn , c. c. , nelson , b. , & williams , r. j. 2000 , , 120 , in press kirkpatrick , j. d. , et al .
1999 , , 519 , 802 leggett , s. k. , toomey , d. w. , geballe , t. r. , & brown , r. h. 1999 , , 517 , l139 leggett , s. k. , allard , f. , & hauschildt , p. h. 1998 , , 509 , 836 liebert , j. , reid , i. n. , burrows , a. , burgasser , a. j. , kirkpatrick , j. d. , & gizis , j. e. 2000 , , 533 , in press lodders , k. 1999 , , 519 , 793 marley , m. s. , saumon , d. , guillot , t. , freedman , r. s. , hubbard , w. b. , burrows , a. , & lunine , j. i. 1996 , science , 272 , 1919 mclean , i. , et al .
2000 , , 533 , l45 morgan , d. h. , tritton , s. b. , savage , a. , hartley , m. , & cannon , r. d. 1992 , in digitised optical sky surveys , ed .
h. t. macgillivray & e. b. thomson ( dordrecht : boston ) , p. 11
murphy , d. c. , persson , s. e. , pahre , m. a. , sivaramakrishnan , a. , & djorgovski , s. g. 1995 , , 107 , 1234 nakajima , t. , tsuji , t. , maihara , t. , iwamuro , f. , motohara , k. , taguchi , t. , hata , r. , tamura , m. , & yamashita , t. 2000 , , 52 , 87 nakajima , t. , oppenheimer , b. r. , kulkarni , s. r. , golimowski , d. a. , matthews , k. , & durrance , s. t. 1995 , , 378 , 463 noll , k. s. , geballe , t. r. , & marley , m. s. 1997 , , 489 , l87 oppenheimer , b. r. 1999 , ph.d . thesis ( pasadena : caltech ) oppenheimer , b. r. , kulkarni , s. r. , matthews , k. , van kerkwijk , m. h. 1998 , , 502 , 932 pearse , r. w. b. , & gaydon , a. g. 1963 , the identification of molecular spectra ( 3d ed . ;
new york : john wiley & sons ) perryman , m. a. c. , et al .
1997 , , 323 , l49 phillips , j. g. , davis , s. p. , lindgren , b. , & balfour , w. j. 1987 , , 65 , 721 reid , i. n. , kirkpatrick , j. d. , gizis , j. e. , dahn , c. c. , monet , d. g. , williams , r. j. , liebert , j. , & burgasser , a. j. 2000 , , 119 , 369 reid , i. n. , et al . 1999 , , 521 , 613 reid , i. n. , & gizis , j. e. 1997 , , 113 , 2246 skrutskie , m. f. , et al .
1997 , in the impact of large - scale near - ir sky surveys , ed .
f. garzon ( dordrecht : kluwer ) , p. 25
strauss , m. a. , et al .
1999 , , 522 , l61 tokunaga , a. t. 2000 , in allen s astrophysical quantities , fourth edition , ed .
a. n. cox ( new york : springer - verlag ) , p. 151 tsvetanov , z. i. , et al .
2000 , , 531 , l61 wiese , w. l. , smith , m. w. , & glennon , b. m. 1966 , atomic transition probabilities , vol . 1 .
( washington , d. c. : gpo ) wilson , j. c. , et al .
2000 , , in preparation cs i & 0.8943 & 6s @xmath18s@xmath20 @xmath5 6p @xmath18p@xmath20 & @xcite + k i & broadened up to @xmath10 0.9 & 4s @xmath18s@xmath20 @xmath5 4p @xmath18p@xmath38 & @xcite + h@xmath7o & 0.925 - 0.95 & @xmath37@xmath7 = 0 , @xmath37@xmath39 + @xmath37@xmath36 = 3 & @xcite + h@xmath7o & 0.95 - 0.985 & @xmath37@xmath7 = 2 , @xmath37@xmath39 + @xmath37@xmath36 = 2 & @xcite + feh & 0.9896 & 0 - 0 band of a@xmath16@xmath17-x@xmath16@xmath17 & @xcite + h@xmath7o & 1.07 - 1.11 & @xmath37@xmath7 = -1 , @xmath37@xmath39 + @xmath37@xmath36 = 3 & @xcite + ch@xmath2 & 1.1 - 1.24 & 3@xmath37@xmath36 & @xcite + h@xmath7o & 1.11 - 1.16 & @xmath37@xmath7 = 1 , @xmath37@xmath39 + @xmath37@xmath36 = 2 & @xcite + h@xmath7o & 1.16 - 1.23 & @xmath37@xmath7 = 3 , @xmath37@xmath39 + @xmath37@xmath36 = 1 & @xcite + k i & 1.1690 & 4p @xmath18p@xmath19 @xmath5 3d @xmath18d & @xcite + k i & 1.1773 & 4p @xmath18p@xmath19 @xmath5 3d @xmath18d & @xcite + k i & 1.2432 & 4p @xmath18p@xmath19 @xmath5
5s @xmath18s & @xcite + k i & 1.2522 & 4p @xmath18p@xmath19 @xmath5 5s @xmath18s & @xcite + ch@xmath2 & 1.3 - 1.5 & @xmath37@xmath7 + 2@xmath37@xmath36 & @xcite + h@xmath7o & 1.33 - 1.43 & @xmath37@xmath7 = 0 , @xmath37@xmath39 + @xmath37@xmath36 = 2 & @xcite + h@xmath7o & 1.43 - 1.52 & @xmath37@xmath7 = 2 , @xmath37@xmath39 + @xmath37@xmath36 = 1 & @xcite + ch@xmath2 & 1.6 - 1.8 & 2@xmath37@xmath36 & @xcite + h@xmath7o & 1.71 - 1.80 & @xmath37@xmath7 = -1 , @xmath37@xmath39 + @xmath37@xmath36 = 2 & @xcite + h@xmath7o & 1.80 - 2.08 & @xmath37@xmath7 = 1 , @xmath37@xmath39 + @xmath37@xmath36 = 1 & @xcite + ch@xmath2 & 2.2 - 2.6 & @xmath37@xmath7 + @xmath37@xmath36 & @xcite + h@xmath7 & centered at 2.4 & 1 - 0 quadrupole ( cia ) & @xcite + | we report the discovery of a bright ( j = [email protected] ) methane brown dwarf , or t dwarf , by the two micron all sky survey .
this object , 2massi j0559191 - 140448 , is the first brown dwarf identified by the newly commissioned cormass instrument mounted on the palomar 60-inch telescope .
near - infrared spectra from 0.9 - 2.35 @xmath1 show characteristic ch@xmath2 bands at 1.1 , 1.3 , 1.6 , and 2.2 @xmath1 , which are significantly shallower than those seen in other t dwarfs discovered to date . coupled with the detection of an feh band at 0.9896 and two sets of k i doublets at j - band
, we propose that 2mass j0559 - 14 is a warm t dwarf , close to the transition between l and t spectral classes .
the brightness of this object makes it a good candidate for detailed investigation over a broad wavelength regime and at higher resolution . |
there is a growing interest in the nonextensive generalization of the conventional boltzmann -gibbs statistical mechanics .
tsallis thermostatistics ( tt ) is one of these generalization attempts [ 1 ] and in this study , we generalize the maier - saupe theory ( mst ) [ 2 ] , a mean field theory especially used in nematic liquid crystals , and the variation of the order parameter of paa ( p - azoxyanisole , a nematic liquid crystals ) versus temperature is investigated within tt formalism .
mst fails to explain the experimental data for some of the nematics and paa is one of these nematics .
it is well known that boltzmann - gibbs statistics ( bg ) seems to fail for the systems , ( i ) with long range microscopic interactions , ( ii ) with long range memory effects and ( iii ) which evolves in a multifractal space - time .
therefore there is a tendency for nonextensive formalisms .
it is seen in the literature that from 1988 up to present days , tt not only has been applied to various concepts of thermostatistics [ 3 - 20 ] , but also achieved to solve some physical systems where bg statistics is known to be inadequate . amongst them , stellar polytropes [ 21 ] , levy - like anomalous diffusions [ 22 ] , two - dimensional euler turbulence [ 23 ] , solar neutrino problem [ 24 ] , velocity distributions of galaxy clusters [ 25 ] , fully developed turbulence [ 26 ] , electron - positron and other high - energy collisions [ 27 ] and anomalous diffusion of hydra viridissima [ 28 ] could be mentioned .
the axioms of tt are the following 1 ) the entropy of the physical system is defined by @xmath0 where k is a constant , p@xmath1 is the probability of the system in the i th microstate , w is the total number of configurations and q is the entropic index ( q@xmath2 ) which is a measure of the nonextensivity of the system .
it is important to note that if q@xmath3 , then s@xmath4 , which is the well known shannon entropy .
\2 ) q - expectation value of an observable o is given by @xmath5 which again recovers the conventional expectation value when q=1 .
an approach that has proved to be useful in developing a theory of long range orientational order and the where v is the molar volume , a is a constant independent of pressure , volume and temperature , @xmath6 is the angle between the long molecular axis and the preferred axis , s is the order parameter and given by @xmath7 the internal energy per mole is @xmath8 where b = a/(ktv@xmath9 ) , ( a=13@xmath10 erg.@xmath11 , v=225 @xmath12forpaa ) , n is the avogadro number , t is temperature .
the partition function for a single molecule is given by @xmath13 so that the entropy is @xmath14 .\ ] ] then the helmholtz free energy is @xmath15 .\end{aligned}\ ] ] the condition for equilibrium is given by @xmath16 or @xmath17 this equation is satisfied when @xmath18 the theoretical curve for s versus t could be determined from ( 8) namely the equilibrium condition .
in the frame of tt , the generalized partition function is written as @xmath19 ^{\frac{1}{1-q}}\ ] ] where @xmath20 .
the average value of cos@xmath21 is @xmath22 for the sake of simplicity , we replace cos@xmath6 with x@xmath1 . in eq.(10 ) , let us write @xmath23 ^{\frac{1}{1-q } } \nonumber \\ & = & \left [ 1+(1-q)\frac{bs}{2}(3x_{i}^{2}-1)\right ] ^{\frac{1}{1-q}}.\end{aligned}\ ] ] substituting eq.(11 ) into eq.(10 ) , we obtain @xmath24 ^{\frac{1}{1-q}}dx_{i}}{\int_{0}^{1}% \left [ 1+(1-q)\frac{bs}{2}(3x_{i}^{2}-1)\right ] ^{\frac{1}{1-q}}dx_{i}}.\ ] ] to integrate the terms in eq.(12 ) , the series expansions of exp@xmath25 is written in terms of q and then the leading two terms are taken : @xmath26 \nonumber \\ & & -\frac{1}{8}\exp \left [ \frac{1}{2}bs(3x_{i}^{2}-1)\right ] b^{2}s^{2}(3x_{i}^{2}-1)^{2}(1-q).\end{aligned}\ ] ] then , we have solved equation ( 12 ) and plotted s = f(t ) curve for various values of q. in figure 1 , the order parameter versus temperature for various values of q are presented . as seen from fig.1 ,
gmst is in good aggreement with the experimental data for q=0.988 .
it seems that with a small departure from the standard theory one could be able to explain some of the experimental data available in the study of nematics .
gmst recovers mst when q=1 .
near the nematic - isotropic phase transition temperature , the aggreement between gmst and the experimental data is broken . in table 1 , the variations of the density of paa with respect to temperature are given .
in addition , in the recent studies , the entropic index is related to the number of particles of the system [ 29,30 ] .
then a better aggreement with the experimental data could be obtained , which will be the subject of a forthcoming study . from gmst ,
the value of the order parameter s at t@xmath27 is determined as 0.4 which is in good aggreement with the experimental values [ 31 - 35 ] , whereas 0.443 is given for s in mst .
mst is the most widely used molecular field theory in the nematic liquid crystals .
although this theory is successful in the study of some of the nematics , it fails for some others , one of which is paa , p - azoxyanisole . in this study , it is observed that , after the generalization of mst , the variation of the order parameter versus the temperature shows a better aggremeent with the experimental data .
fig.1 exhibits the effect of nonextensivity .
it is remarkable to note that the value of s at t@xmath27 is 0.4 which is given in refs.[31,32,33 ] .
another important point is the variation of the density of paa with respect to the temperature[2 ] , presented in table 1 . as temperature increases ,
the density of paa decreases .
moreover , as mentioned earlier , in recent studies , the entropic index is related to the number of particles of the system [ 29,30 ] .
therefore it is expected that q , the entropic index , could be related to the density , which will be the subject of another study in near future and the variation of s near t@xmath27 , nematic - isotropic phase transition temperature , could be explained hopefully by gmst . following this study , a tendency to investigate liquid crystals within nonextensive formalisms
could be expected .
one of us ( o.k . ) would like to thank u. tirnakli for valuable comments and suggestions .
[ 1 ] c. tsallis , j. stat phys . 52 , 479 ( 1988 ) ; e.m.f . curado , c. tsallis , j. phys .
a24 , l69 ( 1991 ) ; corrigenda : 24 , 3187 ( 1991 ) ; 25 , 1019 ( 1992 ) .
[ 2 ] maier , w. and saupe , a. , a15 , 287 ( 1960 ) . [ 3 ] n. ito , c. tsallis , nuovo cimento d11 , 907 ( 1989 ) . [ 4 ] r.f.s .
andrade , physica a175 , 285 ( 1991 ) ; a203 , 486 ( 1994 ) ; u. tirnakli , d. demirhan , f. bykkili , acta .
phys . pol .
a91 , 1035 ( 1997 ) .
[ 5 ] a.m. mariz , phys .
a165 , 409 ( 1992 ) ; j.d .
ramshaw , phys .
a175 , 169 and 171 ( 1993 ) .
[ 6 ] f. bykkili , d. demirhan , phys .
a181 , 24 ( 1993 ) ; f.bykkili , d.demirhan , a. gle , phys .
a197 , 209 ( 1995 ) .
[ 7 ] a. plastino , a.r .
plastino , phys .
a177 , 177 ( 1993 ) .
[ 8 ] a.r .
plastino , a. plastino , physica a202 , 438 ( 1994 ) .
[ 9 ] e.p .
da silva , c. tsallis , e.m.f .
curado , physica a199 , 137 ( 1993 ) .
[ 10 ] a. plastino , c. tsallis , j. phys .
a26 , l893 ( 1993 ) .
[ 11 ] d.a .
stariolo , phys .
a185 , 262 ( 1994 ) .
[ 12 ] a.r .
plastino , a. plastino , c. tsallis , j. phys .
a27 , 5707 ( 1994 ) .
[ 13 ] f. bykkili , d. demirhan , z. phys .
b99 , 137 ( 1995 ) .
[ 14 ] s. curilef , c. tsallis , physica a215 , 542 ( 1995 ) .
[ 15 ] f. bykkili , d. demirhan , u. tirnakli , physica a238 , 285 ( 1997 ) .
[ 16 ] c. tsallis , f.c .
sa barreto , e.d .
loh , phys .
e52 , 1447 ( 1995 ) .
[ 17 ] u. tirnakli , f. buyukkili , d. demirhan , physica a240 , 657 ( 1997 ) .
[ 18 ] a.k .
rajagopal , phys .
74 , 1048 ( 1995 ) ; physica b212 , 309 ( 1995 ) .
[ 19 ] a.k .
rajagopal , phys .
76 , 3469 ( 1996 ) .
[ 20 ] u. tirnakli , s.f .
zeren , f. bykkili , d. demirhan , z. phys .
b104 , 341 ( 1997 ) .
[ 21 ] a.r .
plastino , a. plastino , phys .
a174 , 384 ( 1993 ) .
[ 22 ] p.a .
alemany , d.h .
zanette , phys .
e49 , r956 ( 1994 ) ; c. tsallis , s.v.f .
levy , a.m.c .
souza , r. maynard , phys .
75 , 3589 ( 1995 ) ; d.h .
zanette , p.a .
alemany , phys .
75 , 366 ( 1995 ) ; m.o .
careres , c.e .
budde , phys .
77 , 2589 ( 1996 ) ; d.h .
zanette , p.a .
alemany , phys .
77 , 2590 ( 1996 ) .
[ 23 ] b.m .
boghosian , phys .
e53 , 4754 ( 1996 ) .
[ 24 ] g. kaniadakis , a. lavagno , p. quarati , phys .
b369 , 308 ( 1996 ) .
[ 25 ] a. lavagno , g. kaniadakis , m.r .
monteiro , p. quarati , c. tsallis , astrophys .
lett . and comm .
35 , 449 ( 1998 ) .
[ 26 ] t. arimitsu and n. arimitsu , phys .
e 61 , 3237 ( 2000 ) ; t. arimitsu and n. arimitsu , j. phys .
a 33 , l235 ( 2000 ) ; c. beck , physica a 277 , 115 ( 2000 ) ; c. beck , g.s .
lewis and h.l .
swinney , phys .
e 63 , 035303 ( 2001 ) .
[ 27 ] d.b .
walton and j. rafelski , phys .
84 , 31 ( 2000 ) ; g. wilk and z. wlodarcsyk , nucl .
b ( proc . suppl . ) 75a , 191 ( 1999 ) ; g. wilk and z. wlodarczyk , phys
84 , 2770 ( 2000 ) ; m.l.d . ion and d.b .
ion , phys .
b 482 , 57 ( 2000 ) ; i. bediaga , e.m.f . curado and j. miranda , physica a 286 , 156 ( 2000 ) ; c. beck , physica a 286 , 164 ( 2000 ) . [ 28 ] a. upadhyaya , j .-
rieu , j.a .
glazier and y. sawada , physica a 293 , 549 ( 2001 ) .
[ 29 ] s. abe , s. martinez , f. pennini , a. plastino , phys .
a278 , 249 ( 2001 ) .
[ 30 ] c. tsallis , r.s .
mendes , a.r .
plastino , physica a261 , 534 ( 1998 ) .
[ 31 ] v. zwetkoff , acta physicochim .
urss 16 , 132 ( 1942 ) .
[ 32 ] w. maier , a. saupe , z. naturforschg . 13a , 564 ( 1958 ) .
[ 33 ] w. maier , g. englert , z. electrochem .
( 1960 ) .
[ 34 ] p. chatelain , bull .
78 , 262 ( 1955 ) .
[ 35 ] h. lippmann , ann .
( 6 ) 20 , 265 ( 1957 ) . [ 36 ] chandrasekhar s. , madhusudana n.v . ,
journal phys .
radium 30 , c4 - 24 ( 1969 ) . [ 37 ] p. chatelain and m. germain , c.r.hebd .
seances acad .
259 , 127 ( 1964 ) .
* table 1 : * the variation of the density of paa versus temperature [ 2 ] . + * figure 1 : * plot of the order parameter versus reduced temperature t / t@xmath28 continuous line : maier - saupe approximations ( q@xmath29 ) .
dashed lines : gmst approximation for various values of q. filled circles : optical data [ 36 ] . filled triangles : from refractive index measurements [ 37 ] .
[ cols="^,^,^,^",options="header " , ] | in this study , one of the mean - field theories in nematics , the maier - saupe theory ( mst ) , is generalized within tsallis thermostatistics ( tt ) .
the variation of the order parameter versus temperature has been investigated and compared with experimental data for paa ( p - azoxyanisole ) . so far we believe that this is the first attempt of the application of tt in liquid crystals .
it is well known that mst fails to explain the experimental data for some of the nematics , one of which is paa .
however generalized mst ( gmst ) is able to account for the experimental data for a long range of temperatures .
also in this study , the effect of nonextensivity is shown for various values of the entropic index . _
pacs number(s ) : 05.20.-y , 05.70.-a , 61.30.-v , 61.30.cz , 61.30.gd_ |
the idea that our universe may be a hypermembrane in a five dimensional spacetime has received some attention in the last few years after it was realized that gravity could be localized on a three - brane embedded in an anti - de sitter spacetime @xcite . since then
, much work has been done in a cosmological context @xcite and there is hope that a consistent ( i.e. , mathematically self - contained and observationally satisfying ) high dimensional model might soon be formulated .
for instance , it has been proposed that a model @xcite based on such ideas could present itself as an alternative to the inflationary paradigm , although for the time being the controversy as to whether or not such a model might have anything in common with our universe is still going on @xcite .
the idea is not new however , but has evolved from the standard kaluza - klein approach to that of particle localization on a higher dimensional defect @xcite . in particular , it has been shown that massless bulk scalars and gravitons share the property to have a zero mode localized on the brane @xcite in the randall - sundrum model .
various mechanisms @xcite have been invoked according to which it would be possible to confine massless gauge bosons on a brane , so that there is hope to achieve a reasonable model including all the known interactions in a purely four dimensional effective model .
a mechanism permitting localization of massless fermions on a domain wall was described in refs .
however , although appealing this mechanism might be , it should be emphasized that actual fermions , as seen on an everyday basis in whatever particle physics experiment , are massive , so that a realistic fermionic matter model on the brane must accommodate for such a mass .
the question of localization of massive fermions on the brane thus arises naturally , and it is the purpose of this work to provide the transverse brane and fermionic structure that leads to this localization . up to now , fermions have been confined under the restricting hypothesis that the brane self gravity was negligible @xcite , or that it was embedded in a minkowski spacetime with one @xcite or two @xcite transverse dimensions ( see also @xcite for the localization of fermion on a string - like defect in five dimension ) . our goal is to transpose the original work of ref .
@xcite to the brane context . for that purpose
, we realize the brane as a domain wall . such domain wall configurations in anti - de sitter space
have already been studied @xcite .
we will assume that five dimensional fermions are yukawa - coupled to the domain wall forming higgs field , as in the usual case of cosmic strings .
in this respect , our work somehow extends ref .
@xcite , where the mass term was put by hand , and ref .
@xcite where the gravity of the wall was neglected .
we start , in the following section , by recalling the domain wall configuration of a higgs field in a five dimensional anti - de sitter spacetime and discuss briefly its properties . in section [ sec_dirac ] , we describe the dynamical equations of fermions coupled to this domain wall in order to show that they obey a schrdinger - like equation with an effective potential which can trap massive modes on the wall .
the asymptotic structure , i.e. , deep in the bulk ( far from the brane ) , is not minkowski space , so that the effective potential felt by the fermions possesses a local minimum at the brane location , but no global minimum , as first pointed out in ref .
@xcite . as a consequence ,
the bound states are metastable and fermions can tunnel to the bulk .
we then provide an analytical approximation of the effective potential , thanks to which we compute analytically , in section [ sec_analy ] , the mass spectrum of the fermions trapped on the brane .
we obtain the mass of the heaviest fermion that can live on the brane and estimate its tunneling rate .
this result is compared to a full numerical integration , performed in section [ sec_num ] . in a last section ,
we investigate the parameter space and , after having compared our results to previous ones , we conclude that there exists a wide region in the parameter space for which the fermion masses can be made arbitrary low , i.e. , comparable to the observed small values ( with respect to the brane characteristic energy scale ) , while their confinement life - time can be made much larger than the age of the universe .
such models can therefore be made viable as describing realistic matter on the brane .
we consider the action for a real scalar field @xmath1 coupled to gravity in a five dimensional spacetime @xmath2\sqrt{g}\,{{\rm d}}^5 x \equiv \int \sqrt{g}\,{{\rm d}}^5 x [ { \cal l}_{_{\rm grav}}+{\cal l}_\lambda+{\cal l}_\phi ] \ ] ] where @xmath3 is the five dimensional metric with signature @xmath4 , @xmath5 its ricci scalar , @xmath6 the five dimensional cosmological constant and @xmath7 , @xmath8 being the five dimensional gravity constant .
capital latin indices @xmath9 run from 0 to 4 .
the potential of the scalar field @xmath1 is chosen to allow for topological membrane ( domain wall like ) configurations , @xmath10 where @xmath11 is a coupling constant and @xmath12 , @xmath13=m^{3/2}$ ] , @xmath14=m^2 $ ] , @xmath15=m^{-1}$ ] , @xmath16=m^{3/2}$ ] and @xmath17=m^{-3/2}$ ] ( @xmath18 being a unit of mass ) . ] . motivated by the brane picture
, we choose the metric of the bulk spacetime to be of the warped static form @xmath19 where @xmath20 is the four dimensional minkowski metric of signature ( @xmath21 ) , and @xmath22 the coordinate along the extra - dimension .
greek indices @xmath23 run from 0 to 3 . with this metric ansatz ,
the einstein tensor components reduce to @xmath24 where a prime denotes differentiation with respect to @xmath22 .
the non - vanishing components of the matter stress - energy tensor @xmath25 are given by @xmath26 it follows that the five dimensional einstein equations @xmath27 can be cast in the form @xmath28 while the klein gordon equation takes the form @xmath29 eqs .
( [ einstein1]-[kg ] ) is a set of three differential equations for two independent variables ( @xmath1 and @xmath30 ) . indeed , as can easily be checked , the klein - gordon equation stems from the einstein equations provided @xmath31 . to study the domain wall configuration , we choose to solve the first einstein equation ( [ einstein1 ] ) together with the klein - gordon equation ( [ kg ] ) .
this system of equations must be supplemented with boundary conditions . by definition of the topological defect like configuration , we require that the higgs field vanishes on the membrane itself , i.e. , @xmath32 for @xmath33 , while it recovers its vev in the bulk , so that @xmath34 .
note that the sign choice made here is arbitrary and corresponds to the so - called kink solution ; the opposite choice ( i.e. @xmath35 ) would lead to an anti - kink whose physical properties , as far as we are concerned , are exactly equivalent . as for the metric function @xmath30 , it stems from the requirement that one wants to recover anti - de sitter asymptotically , so that one demands that @xmath36 tends to a constant for @xmath37 .
this constant can be determined using eq .
( [ einstein2 ] ) , so that @xmath38 .
note that as @xmath22 changes sign at the brane location , there is no choice for the sign of the function @xmath30 in this case .
note also that , as is well known , the static hypothesis implies that the bulk cosmological constant @xmath6 must be negative , and therefore the five dimensional spacetime to be anti - de sitter . with the convenient dimensionless rescaled variables @xmath39 the dynamical equations read @xmath40 @xmath41 where a dot refers to a derivative with respect to @xmath42 and the two dimensionless ( positive ) parameters @xmath43 and
@xmath44 are defined by @xmath45 these parameters are not independent since , for an arbitrary value of @xmath44 say , there is only one value of @xmath43 for which the boundary condition @xmath46 , or equivalently @xmath47 , is satisfied .
this stems from the fact that eq .
( [ eq1 ] ) is a first order equation in @xmath48 , so that only one boundary condition is freely adjustable , and we choose it to be at @xmath49 .
once this choice is made , the value of @xmath48 on the brane is completely determined , and unless the parameters are given the correct values , it does not vanish .
as the solution must be symmetric with respect to the extra dimension coordinate @xmath22 , one must tune the parameters in order to have a meaningfull solution ( i.e. for which the metric and its first derivative are continuous at @xmath50 ) .
this is reminiscent of the relation that should hold between the brane and bulk cosmological constants @xcite .
( [ eq1 ] -[eq2 ] ) have been solved numerically with the relevant boundary conditions .
the field profiles are depicted on fig .
[ fig3 ] for two arbitrary values of the parameter @xmath44 .
the relation between the parameters @xmath43 and @xmath44 such that the metric is regular at the brane location is depicted on fig .
[ rpu2 ] . as can be seen on the figure , it consists essentially in two power laws . for small values of @xmath44 ,
one finds roughly @xmath51 , which becomes exact in the limit @xmath52 , while for large values of @xmath44 , one gets @xmath53 , which is , again , exact in the limit @xmath54 .
we were able to find the best fit @xmath55 , \label{relab}\ ] ] which , as can be seen on the figure , is almost exact everywhere .
this translates into the relation @xmath56 between the 5 dimensional cosmological constant and the microscopic parameters , which corresponds to the usual relations between brane and bulk cosmological constants .
this section is devoted to the description of the dirac equation in five dimensions with the domain wall configuration obtained in the previous section . the minimal representation of spinors in five dimensions can be chosen to be four dimensional @xcite .
the five dimensional clifford algebra can then be constructed from the usual four dimensional one by adding the @xmath57 matrix to close the algebra . if @xmath58 design the usual four dimensional dirac matrices in minkowski space in the chiral representation , the dirac matrices in five dimensional minkowski space , @xmath59 , are @xmath60 and they satisfied the usual clifford algebra @xmath61 where @xmath62 stands for the five dimensional minkowski metric . since in this representation
the dirac matrices satisfy @xmath63 the five dimensional spinors have neither weyl nor majorana represention .
it follows that the dirac lagrangian in five dimension for fermions coupled to the higgs domain wall is necessary of the form @xmath64\psi\ ] ] where the lorentz covariant derivative with spin connection is @xcite @xmath65 we emphasize that the sign of the coupling @xmath66 of the spinor @xmath67 to the higgs field is arbitrary and represents a coupling either to kink or to anti - kink domain wall . for definiteness , we shall consider in what follows only the case of a kink coupling , and thus assume without lack of generality that @xmath68=m^2 $ ] and @xmath69= m^{-1/2}$ ] . ] .
the variation of the lagrangian ( [ lagrangien ] ) leads to the equation of motion of the spinor field , namely the dirac equation in five dimensional anti - de sitter space for a fermionic field coupled to a higgs field , @xmath70 \psi=0.\ ] ] this equation involves the matrix @xmath57 ( through @xmath71 ) and it is thus convenient to split the four dimensional right- and left - handed components of the five dimensional spinor and to separate the variables as @xmath72 \psi(x^\mu),\ ] ] where @xmath73 is a four dimensional dirac spinor , while @xmath74 and @xmath75 are yet - underdetermined functions of @xmath22 . in what follows ,
we want the five dimensional dirac equation to yield an effective four dimensional massive dirac equation , with an effective mass @xmath76 ( energy eigenvalue of the bound state ) .
such a requirement implies that @xmath77 or , equivalently , in terms of the right- and left- handed components @xmath78 where the right- and left - handed components of the four dimensional spinor are defined as @xmath79 contrary to the case studied in ref .
@xcite , the mass @xmath76 is not an arbitrary parameter and will be determined later .
choosing @xmath80 and @xmath81 as the independent variables instead of @xmath67 and @xmath82 , and inserting equation ( [ diracmass ] ) into the equation of motion ( [ fermmvt ] ) while using the splitting ansatz ( [ sepansatz ] ) yields the differential system for the two functions @xmath83 , @xmath84{{\mathcal{u}}}_{\rm r}(y)&= & -m{{\rm e}}^{\sigma(y)}{{\mathcal{u}}}_{\rm l}(y ) , \label{systurul1 } \\
\left[\partial_y-2 \sigma'(y)+g_{_{\rm f } } \phi\right]{{\mathcal{u}}}_{\rm l}(y)&= & m{{\rm e}}^{\sigma(y)}{{\mathcal{u}}}_{\rm r}(y ) .
\label{systurul2}\end{aligned}\ ] ] to simplify the notations , it is convenient to introduce the dimensionless rescaled bulk components of the fermions @xmath85 in terms of which the system ( [ systurul1]-[systurul2 ] ) takes the form @xmath86\right)\ , { \widetilde{\mathcal{u}}}_{\rm r } & = & - { \mu}\ , { { \rm e}}^{\sigma}{\widetilde{\mathcal{u}}}_{\rm l } , \label{syst1}\\ \left(\partial_\varrho + \left[{\gamma_{_{\rm f}}}h - \frac{1}{2 } s\right]\right)\ , { \widetilde{\mathcal{u}}}_{\rm l } & = & { \mu}\ , { { \rm e}}^{\sigma}{\widetilde{\mathcal{u}}}_{\rm r } , \label{syst2}\end{aligned}\ ] ] with the dimensionless rescaled mass and coupling constant @xmath87 let us first concentrate on the special case @xmath88 .
the system ( [ syst1]-[syst2 ] ) then consists in two decoupled differential equations and the zero mode states @xcite are recovered .
asymptotically , these functions behave as @xmath89 thus , only the left - handed solution @xmath90 may remain bounded @xcite , and yet provided @xmath91 indeed , the right - handed zero modes could have been obtained by considering the coupling of fermions to the anti - kink higgs profile would have yield right - handed solution with the constraint @xmath92 .
we thus recover the well - known fact that massless fermions _ must _ be single - handed in a brane model , contrary to the ordinary four dimensional field theory in which they simply _
can_. let us now focus on the more interesting massive case for which @xmath93 . then the system ( [ syst1]-[syst2 ] )
can be decoupled by eliminating @xmath94 say . for that purpose ,
we differentiate ( [ syst2 ] ) with respect to @xmath42 and express @xmath95 using equation ( [ syst1 ] ) and @xmath94 using equation ( [ syst2 ] ) again to get @xmath96 { \widetilde{\mathcal{u}}}_{\rm l}= 0,\\ \label{orderzerour } & & { \widetilde{\mathcal{u}}}_{\rm r } = \frac{{{\rm e}}^{-\sigma}}{{\mu}}\left[\partial_\varrho + \left({\gamma_{_{\rm f}}}h - \frac{1}{2 } s \right)\right]{\widetilde{\mathcal{u}}}_{\rm l}.\end{aligned}\ ] ] this system is strictly equivalent to the initial system ( [ syst1]-[syst2 ] ) since differentiating eq .
( [ orderzerour ] ) and then using eq .
( [ ordertwoul ] ) gives back eq .
( [ syst1 ] ) .
it is thus important to keep both equations .
note that the integration of the first equation ( [ ordertwoul ] ) will require two initial conditions but that @xmath94 will then be completely determined and thus requires no extra constant of integration . as a consequence ,
it is sufficient to solve the second order equation ( [ ordertwoul ] ) for @xmath90 in order to fully determine the left- and right - handed bulk fermion profiles .
. ( [ ordertwoul ] ) can be recast into a schrdinger - like second order differential equation @xmath97 where the function @xmath98 is defined by @xmath99 with the new function @xmath100 our aim will now be to find the zero modes of this new equation ; as previously discussed , they will be equivalent to the massive bound states we are looking for on the brane . in order for the fermions to be confined on the brane
, the minimum of @xmath101 needs to be negative to imply an exponential decrease of @xmath102 in the bulk .
this is essentially equivalent to the condition ( [ cond_bounded ] ) that was obtained for the case of zero modes .
we shall assume henceforth that this condition also holds for massive modes , i.e. , that the value of @xmath103 necessary to bind massive fermions on the brane is at least that to bind massless ones .
indeed , eq . ( [ omega2 ] ) shows that the minimum of @xmath101 can only be negative for large values of the parameter @xmath103 .
however , since the first term of ( [ omega2 ] ) increases exponentially at large distance from the brane , then , if @xmath93 , @xmath101 will necessarily become positive . this will yield asymptotic radiative behaviors of the spinor bulk components .
physically , it can be interpreted as a tunneling of the fermions from the brane to the bulk @xcite . on the other hand , on the brane , the higgs field @xmath104 and the derivative of the warp factor @xmath48 vanish , so that @xmath105 is positive . as a result , the fermions can freely propagate in a tiny region around the brane , but certainly only for particular values of @xmath101 ( and thus of @xmath106 ) satisfying the boundary conditions with the surrounding exponential decreasing regions .
the effective potential @xmath107 , depicted in fig .
[ figveff ] , exhibits a local minimum on the brane and minima at infinity .
the modes trapped on the brane are thus expected to have discrete masses @xmath106 on the brane and non - zero probability of tunneling into the bulk .
since the higgs and warp factor profiles are not known analytically , it is a priori impossible to solve eq .
( [ schrodlike ] ) analytically .
nevertheless , since the fermions are expected to be trapped in a neighborhood of the brane , we can look for series solutions in @xmath42 . in [ subsec_approx ]
, we give an approximation of the effective potential @xmath108 which will then be used to determine the bound states and the mass spectrum in [ subsec_bound ] .
we end this section by determining in [ subsec_tunel ] the tunneling rate in this approximation .
the validity of this approach is difficult to assess and it will be justified _ a posteriori _ on the ground of a full numerical integration of the system in the following section . in the whole section , it is assumed that @xmath109 . in a neighborhood of the brane
we can expand the higgs and warp factor profiles as @xmath110 where both the constant and quadratic terms vanish for symmetry reasons . to simplify the analysis ,
we shall make use of the equations of motion in the form @xmath111 and @xmath112 where the function @xmath113 is defined as @xmath114 ( recall that @xmath115 ) . plugging the expansions ( [ fieldexpand1 ] ) and ( [ fieldexpand2 ] ) into eqs .
( [ dots ] ) and ( [ doth ] ) , it follows that the function @xmath113 can be expanded up to third order as @xmath116 inserting this expression back into the equations of motion yields the three coefficients @xmath117,\label{coeffs2}\\ h_3 & = & \frac{2}{3 } h_1 \left[\frac{\alpha}{3 } h_1 ^ 2 - \beta \right]\label{coeffs3}\end{aligned}\ ] ] in terms of the coefficient @xmath118 @xmath119 then , the frequency @xmath101 can be expanded as an harmonic oscillator potential @xmath120 where @xmath121 and @xmath122 are given by @xmath123 .
\label{omega2shape}\end{aligned}\ ] ] the function @xmath101 is well approximated by @xmath124 only near the brane and the expansion ( [ omega2near ] ) is no longer valid at large distance where the exponential term dominates [ see eq .
( [ omega2 ] ) ] .
once the fixed asymptotic values of the higgs and warp factor are reached , the frequency ( [ omega2 ] ) behaves as @xmath125 the analytical estimate of the function @xmath101 is thus obtained by matching the two limiting asymptotic behaviors ( [ omega2near ] ) and ( [ omega2infty ] ) , respectively @xmath124 closes to the brane and @xmath126 far from it , as @xmath127 the dimensionless matching distance @xmath128 has to be solution of @xmath129 in order to get a continuous function . note that
, because of the symmetry on both sides of the wall , we can assume @xmath130 without lack of generality .
for large values of @xmath103 , and using eq .
( [ omega2shape ] ) , we get @xmath131 leading to @xmath132 as it turns out , the faster the asymptotic solution is reached , the better the approximation works .
this is the case in particular for @xmath133 , the exact ( numerically integrated ) effective potential and its approximation are compared in fig .
[ figapprox ] .
the global shapes are effectively the same , and in spite of uncertainties at intermediate regions due to this crude approximation , it is reasonable to expect the same fermion physical behaviors in both potentials . note that for ( cosmologically favored ) higher value of @xmath134 , the higgs field and the warp factor reach more rapidly their asymptotic values leading thus to a better agreement between the two potentials , as can be seen on fig .
[ figapprox ] . given the approximate frequency ( [ omega2estim ] ) , the equation of motion for the left - handed bulk spinor component @xmath102 reduces to @xmath135{\widehat{\mathcal{u}}}_{\rm l}=0,\label{approxfar}\\ \varrho<\varrho_{_{\rm m } } : & \qquad & \left[\partial_\varrho^2 + \omega_0 ^ 2-\omega \varrho^2\right]{\widehat{\mathcal{u}}}_{\rm l}=0,\label{approxnear}\end{aligned}\ ] ] with the requirement that @xmath102 and its derivative are continuous at @xmath128 .
note also that we consider only the case @xmath136 , physics on both sides of the brane being completely symmetric under the transformation @xmath137 .
let us consider the solutions in each region separately .
* @xmath138 : we introduce the new variable @xmath139 in terms of which eq .
( [ approxfar ] ) reduces to a standard bessel differential equation @xmath140\ , { \widehat{\mathcal{u}}}_{\rm l } = 0,\ ] ] the order of which being given by @xmath141 since @xmath101 is positive at infinity , the asymptotic form of the solution is necessary radiative , as was already pointed out in the previous section .
the most general solution of the bessel equation ( [ bessel ] ) is a linear superposition of hankel functions .
since we are interested only in ingoing waves in order to study a tunneling process , the most general solution takes the form @xmath142 where @xmath143 is the hankel function of the first kind , propagating towards the brane at infinity @xcite @xmath144 and @xmath145 is an arbitrary complex constant . *
@xmath146 : performing operations similar to those of the previous case , we cast eq .
( [ approxnear ] ) on the form @xmath147 { \widehat{\mathcal{u}}}_{\rm l}=0,\ ] ] in which we have introduced the new variable and parameter @xmath148 the general solutions of eq .
( [ pcf ] ) are the parabolic cylinder functions , namely @xmath149 and @xmath150 , of which @xmath151 can be expressed as linear superpositions . in the limit @xmath152 , these solutions scale as @xcite @xmath153 since we are interested in confined fermion states on the brane , only the exponentially decreasing function is relevant , so that the general solution near the brane reads @xmath154 where @xmath155 is a complex integration constant .
the general solution @xmath156 for all @xmath42 is obtained by matching the two different solutions at @xmath157 . since @xmath128 corresponds to the maximum positive value of the effective potential [ see fig .
[ figapprox ] ] , it is reasonable to consider that the hankel function at that point can be expanded around small values of their argument with respect to their order @xcite , i.e. , @xmath158 while the parabolic cylinder functions can be taken in their large argument asymptotic limit ( [ asymppcf ] ) .
this is the same kind of approximation as that made to derive the effective potential .
physically , the initial conditions on the brane , i.e. , @xmath159 and @xmath160 , are chosen in such a way that the asymptotic exponentially growing function @xmath150 contribution is everywhere negligible .
once these initial conditions are fixed , they fully determine the solution on the other side of the brane , i.e. , for @xmath161 .
the asymptotic expansion ( [ asymppcf ] ) can be analytically extended to @xmath162 and yields @xcite @xmath163 thus , once @xmath159 and @xmath160 are fixed , the matchings between @xmath164 and @xmath165 on one side , and @xmath166 on the other side fully determine the bulk component @xmath102 for all @xmath42 .
the last constraint comes from eq .
( [ orderzerour ] ) determining the right - handed spinor bulk function .
it is well defined if and only if both @xmath102 and @xmath167 are not singular . in fact , the derivative of the parabolic cylinder function @xmath149 is generally discontinuous at @xmath168 . with the help of the wronskian of @xmath149 and @xmath169 @xcite @xmath170
we can construct the derivative discontinuity at @xmath168 .
this is @xmath171 where we have used the particular value @xcite @xmath172 imposing that the derivative of @xmath102 is continuous at @xmath168 results in imposing that the jump ( [ jump ] ) vanishes .
this is the case if and only if @xmath173 is solution of @xmath174 since @xmath175 is singular for negative integer arguments , this condition is satisfied only for @xmath176 where @xmath177 is a positive integer .
note that @xmath178 can not be odd since then the numerator of the wronskian ( [ nulwronsk ] ) will also be singular resulting in a finite derivative jump at @xmath168 .
the condition ( [ quantization ] ) shows that the trapped fermions on the brane have necessarily discrete masses @xmath179 which read , using the values of the parameters ( [ omega2brane ] ) , ( [ omega2shape ] ) and ( [ defa ] ) , @xmath180}{36{\gamma_{_{\rm f}}}^2 } } \right .
\left(1+\frac{\alpha h_1}{6{\gamma_{_{\rm f } } } } \left[(4n + 1)^2 + 1 \right ] \right ) \right].\end{aligned}\ ] ] this mass spectrum is valid for @xmath181 since our derivation assumed that @xmath109 . in the limiting case where @xmath182
, it reduces to the much simpler form for the lowest masses @xmath183 on the other hand , @xmath184 can not reach very large values since it is necessary to have a potential barrier in order to have bound states . from the expression of the effective potential ( [ omega2infty ] ) , the barrier is found to disappear when @xmath185 again in the limit where @xmath182 , using the value ( [ rhomatch ] ) of @xmath128 , one gets the maximum accessible reduced mass @xmath186 for @xmath106 as @xmath187 the maximum number of distinct massive states trapped on the brane can thus be estimated to be @xmath188.\ ] ] for the parameters chosen in fig . [ figveff ] ,
one obtains @xmath189 and there are @xmath190 massive modes trapped on the brane .
since the effective potential becomes negative at infinity , the massive modes trapped on the brane are subject only to a finite potential barrier .
they are in a metastable state and can tunnel from the brane to the bulk . in this section
we use our previous analytic solution to estimate the tunneling rate .
would this rate be too high , one would observe an effective violation of energy - momentum conservation on the brane , i.e. , in four dimension , thereby contradicting observation .
our starting point is the analytic solution for the left - handed bulk function that was derived in the previous section @xmath191 the transmission factor can easily be derived from the matching conditions of the left - handed bulk fermion component at @xmath157 .
first of all , @xmath102 has to be continuous . using the expansions ( [ asymppcf ] ) and ( [ hankelmatch ] ) and the value ( [ quantization ] ) of the parameter @xmath173 , permits to find the relation @xmath192 between the coefficients @xmath155 and @xmath145 . making use of the expression ( [ rhomatch ] ) for @xmath128 yields @xmath193}.\ ] ] assuming , as above , that @xmath182 ( so that @xmath194 ) we can expand @xmath195 as @xmath196 so that , using expressions ( [ shapeapp ] ) , ( [ rhomatch ] ) and ( [ defmut ] ) respectively of @xmath122 , @xmath128 , and @xmath197 , we get that @xmath198 is approximated by @xmath199}.\ ] ] the transmission coefficient from the brane to the bulk associated with @xmath102
can thus be defined by @xmath200 with @xmath201 , is evaluated at the turning point @xmath202 where the spinor bulk component begins to propagate freely . using the behaviour ( [ uzero ] ) of the function at the origin ,
the ratio ( [ rappamp ] ) and the properties @xcite of the hankel function @xmath203 the transmission coefficient ( [ th ] ) reduces to @xmath204}.\end{aligned}\ ] ] it follows , using the definition ( [ defuh ] ) , that the probability for a trapped particle on the brane to tunnel to the bulk is given by @xmath205 the characteristic time for a fermionic mode trapped on the brane can be roughly estimated by @xmath206 where @xmath207 represents the typical length , in the fifth dimension , felt by a particle on the brane . as can be seen on fig .
[ figbulkradiate ] , the spinor bulk components are exponentially damped as soon as the effective potential becomes positive .
thus @xmath208 can be estimated by the solution of @xmath209 so that , keeping in mind that @xmath210 , @xmath211 the life - time @xmath212 of a fermionic bound state on the brane labelled by @xmath177 @xmath213 can be estimated by @xmath214\right)}.\end{aligned}\ ] ] we recall that , due to the approximations performed in the previous derivation , this estimate is valid only for @xmath215 .
nevertheless , the argument in the exponential amplifies the transition from bound states to tunneling ones for masses @xmath216 , as intuitively expected .
an order of magnitude of the minimal coupling constant @xmath103 leading to stable bound states can thus be estimated by requiring that the lowest massive state does not tunnel , @xmath217 using the two values ( [ appspectrum ] ) and ( [ defmtmax ] ) , this implies that @xmath218 as a numerical application , for the higgs and gravity parameters used in fig .
[ figveff ] , we get @xmath219 .
numerically , it is simpler and more convenient to solve the first order differential system ( [ syst1]-[syst2 ] ) .
a runge - kutta integration method was used , on both sides of the brane . in order to suppress the exponential growth ,
we integrate from the turning point , @xmath202 , where the solution begins to propagate freely , toward the brane . in this way
, we get only @xmath149 near the brane . the radiative solution for @xmath220 ,
is simply obtained by integrating from the turning point toward infinity , with initial conditions determined by the matching with the exponential decreasing solution near the wall .
the same method is used on the other side of the brane , but this time , by means of the last free parameter , we impose the continuity of one bulk spinor component on the brane ( @xmath90 say ) .
generally , the other bulk spinor component will be discontinuous at @xmath50 , as expected from the analytical study since @xmath221 is generally discontinuous .
the mass spectrum is thus obtained by requiring the continuity of @xmath94 on the brane .
the bulk spinor components computed this scheme have been plotted for the first massive modes trapped on the brane in fig .
[ figbulkcomp ] , for @xmath222 . the lowest mass is numerically found to be @xmath223 and was estimated analytically , from ( [ appspectrum ] ) , to be @xmath224 leading to a precision of 0.5% for the analytical estimate .
the second mass is numerically found to be @xmath225 , which has to be compared to its analytical estimate @xmath226 .
again , the precision of this estimate is of about 1% .
as predicted from ( [ defnmax ] ) , there are @xmath190 massive bound states the lightest masses of which are summed up in table [ table ] . on fig .
[ figbulkradiate ] , we plot the last @xmath227 trapped mode ; it has a tiny radiative component , as expected for a tunneling mode .
.comparison of the numerical values and analytical estimates of the first six bound states reduced mass @xmath106 , together with the heaviest mode , computed for @xmath222 . [
cols="^,^,^,^,^,^,^,^,^ " , ] in conclusion , the numerics confirm that the approximations of the previous section and our estimates are accurate up to 1%-10% ( see table [ table ] ) .
we shall now discuss the cosmological constraints existing on the kind of model we have been considering here .
most of these constraints come from brane models in which the wall structure is replaced by an infinitely thin four dimensional layer . as discussed in the previous sections ,
such an approximation is equivalent , within our framework , to asking that the combination @xmath228 be much larger than unity . in this limit ,
equivalent to the large @xmath44 limit since , from eq .
( [ relab ] ) , @xmath229 , we can replace the stress - energy tensor ( [ tmunu ] ) by the effective four dimensional surface distribution @xmath230 whose isotropic tension @xmath231 is obtained by integration in the transverse direction to yield @xmath232 \equiv \sqrt{|\lambda| } \eta^2 \xi ( \beta)\ ] ] where the function @xmath233 can be expressed as @xmath234.\ ] ] fig .
[ tinf ] shows the variation of @xmath235 as a function of @xmath44 . in the limit @xmath236 ,
it is clear that @xmath237 , so that the brane tension behaves as @xmath238 which will be used to derive the relevant cosmological constraints .
note also that the discrepancy between the above formula and the actual value of @xmath231 becomes important ( more than 100 % error say ) for @xmath239 , which is already rather far from the thin brane limit usually considered .
the model described in this article depends of five parameters , four describing the spacetime and scalar field dynamics @xmath240 and one concerning the fermions @xmath241 ) .
with the domain wall structure assumed , only four of these parameters are independent [ see eq .
( [ rellambda ] ) ] .
it is convenient to replace this set of parameters by the three mass scales @xmath242 and the dimensionless parameter @xmath243 .
these parameters are subject to a number of constraints , namely 1 .
the four dimensional gravitational constant must agree with its observed value @xmath244 with @xmath245gev . using the expression of the four dimensional planck mass in terms of the five dimensional analog and of the brane tension gives @xcite @xmath246 2 .
the brane cosmological constant @xcite @xmath247 must also agree with the standard observational bound @xmath248 .
this implies @xmath249 note that in the limit @xmath236 , this relation is equivalent to eq .
( [ tension ] ) .
this means that this condition is readily satisfied in the thin brane limit . at this point , it is worth emphasizing that this is precisely the limit in which the analytic approximation for fermion masses are the most accurate .
3 . there must not be any deviation of the law of gravity on the brane with respect to the inverse square newton law above 1 millimeter @xcite .
this implies @xcite @xmath250 4 .
finally , we require the fermion stress - energy tensor to be negligible with the brane stress - energy , so that we impose that the mass of the heaviest fermion is smaller than the brane mass scale . by means of eq .
( [ defmtmax ] ) , this condition reads @xmath251 where @xmath118 ends up being function of @xmath44 only by means of eqs ( [ relab ] ) and ( [ def_h1 ] ) , @xmath252 in the limit @xmath236 , eqs .
( [ mmax ] ) and ( [ h1beta ] ) combine to give the constraint on the coupling constant @xmath253 it follows from the relations ( [ rel4 ] ) and ( [ rel2 ] ) that @xmath254 so that the three mass scales must satisfy @xmath255 which , together with eq .
( [ gammaf ] ) and ( [ cond_bounded ] ) yields @xmath256 where , as discussed below eq .
( [ defuh ] ) , the lower bound is very conservative .
now , let us examine the stability of the fermion confinement on the brane and the restriction on their life - time imposed by the previous conditions .
we require , at least , one massive bound state to have a life - time longer than the age of the universe , i.e. , @xmath257 using eq .
( [ lifetime ] ) , one roughly gets @xmath258\right)}.\ ] ] the condition ( [ t_u ] ) together with the former constraint ( [ numval ] ) can be written as @xmath259 on fig .
[ figlifetime ] , we present the contour plot of the dimensionless function @xmath260 , for @xmath261 and @xmath262 , which correspond respectively to a particle life - time of the order of the age of the universe , and of the proton life - time lower limit .
for @xmath263 , there are in principle two allowed regions , corresponding to strong and weak coupling limits , i.e. @xmath182 and @xmath264
. however , the lower bound on @xmath103 , which comes from the requirement that fermions are actually trapped on the brane , pushes the weak coupling allowed region to very high values of @xmath44 , in practice @xmath265 for @xmath266 . note also that this already rather extreme value is based on the conservative estimate given by eq .
( [ cond_bounded ] ) . for @xmath267 , the weak coupling region completely disappears , while the strong coupling allowed region shrinks rapidely : for @xmath268
, the life - time can not exceed the age of the universe because @xmath269 .
considering @xmath270 therefore turns out to be the relevant limit if one wishes to have fermionic bound states living on the brane .
in this article , we have considered fermions coupled to a higgs field with a domain wall structure in a five dimensional anti - de sitter spacetime .
this domain wall can be thought of as a realization of a brane universe . after , studying the domain wall configuration , we solved the dirac equation and showed that there exists massive fermionic bound states trapped on the wall .
we develop both analytic approximation to compute the mass spectrum and the tunneling time .
this was compared to a full numerical integration of the dynamical equations that revealed the accuracy of our approximation scheme .
we recover the fact that massive fermions tunnel to the bulk @xcite .
investigation in the parameter space shows that , for models satisfying the cosmological constraints , the relevant confinement life - time can be much greater that either the age of the universe or the proton life - time .
this was made possible by the derivation of the analytic estimate .
one of our central result is the derivation of an analytic mass spectrum for fermions trapped on a brane - like four dimensional spacetime . in particular , as could have been anticipated @xcite , it was shown that the allowed masses are quantized , with a spectrum varrying , in the strong coupling limit , as @xmath271 .
such a spectrum is indeed in contradiction with experimental measurements of particle masses @xcite , which is not surprising given the simplicity of the model .
it however opens the possibility to build more realistic theories in which mass quantization would stem naturally from extra dimensions .
we thank jrme martin for numerous enlightning discussion , and in particular for pointing to us ref .
@xcite on parabolic cylinder functions .
jpu thanks linstitut dastrophysique de paris for hospitality while this work was carried out .
p. bintruy , c. deffayet , d. langlois , nucl .
b * 565 * ( 2000 ) 269 ; c. cski , m. graesser , c. kolda , j. terning , phys . lett .
b * 462 * ( 1999 ) 34 ; j. m. cline , c. grojean , g. servant , ( 1999 ) 4245 ; p. bintruy , c. deffayet , u. ellwanger , d. langlois , phys .
b * 477 * ( 2000 ) 285 ; p. kraus , jhep * 9912 * ( 1999 ) 11 ; t. shiromizu , k. maeda , m. sasaki , ( 2000 ) 024012 ; e. flanagan , s. tye , i. wasserman , ( 2000 ) 044039 ; r. maartens , d. wands , b. bassett , i. heard , ( 2000 ) 041301 .
see also v. a. rubakov , _ large an infinte extra dimensions _
, hep - ph/0104152 and references therein .
j. khoury , b. a. ovrut , p. j. steinhardt , n. turok , _ the ekpyrotic universe : colliding branes and the origin of the hot big bang _ , hep - th/0103239 ; j. khoury , b. a. ovrut , p. j. steinhardt , n. turok , _ a brief comment on ` the pyrotechnic universe ' _ , hep - th/0105212 ; r. y. donagi , j. khoury , b. a. ovrut , p. j. steinhardt , n. turok , _ visible branes with negative tension in heterotic m - theory _ , hep - th/0105199 ; g. dvali , m. shifman , phys .
b * 396 * ( 1997 ) 64 ; ibid . * 407 * ( 1997 ) 452 ; g. dvali , g. gabadadz , m. shifman , phys . lett .
b * 497 * ( 2001 ) 271 ; p. dimopoulos , k. farakos , a. kehagias , g. koutsoumbas , _ lattice evidence for gauge field localization on a brane _ , hep - th/0007079 ; m. j. duff , j. t. liu , w. a. sabra , _ localization of supergravity on the brane _ , hep - th/0009212 ; i. oda , _ a new mechanism for trapping of photon _ , hep - th/0103052 , k. ghoroku , a. nakamura , _ massive vector trapping as a gauge boson on a brane _ , hep - th/0106145 ; e. kh .
akhmedov , _ dynamical localization of gauge fields on a brane _ , hep - th/0107223 .
j. c. long , a. b. churnside , j. c. price , _ gravitational experiment below 1 millimeter and comment on shielded casimir backgrounds for experiments in the micron regime _ , proceedings of the ixth marcel grossmann meeting ( rome , 28 july 2000 ) . | we construct an explicit model to describe fermions confined on a four dimensional brane embedded in a five dimensional anti - de sitter spacetime .
we extend previous works to accommodate massive bound states on the brane and exhibit the transverse structure of the fermionic fields .
we estimate analytically and calculate numerically the fermion mass spectrum on the brane , which we show to be discrete .
the confinement life - time of the bound states is evaluated , and it is shown that existing constraints can be made compatible with the existence of massive fermions trapped on the brane for durations much longer than the age of the universe
. 0.25 cm |
quantum phase transitions between a fermi liquid and magnetic phases have been a subject of experimental and theoretical investigations for several decades @xcite .
the transition to a spin - density - wave ( sdw ) in particular is relevant to problems of current interest . in the cuprates ,
daou _ et al . _
@xcite argued that the fermi surface change associated with this transition is a key to understanding anomalous normal state properties .
recent studies in the pnictides @xcite , heavy fermion materials @xcite and organic superconductors @xcite focus on the relation between the sdw and superconductivity .
in fact , strong experimental similarities between quantum critical behavior in the organics , pnictides and cuprates have been pointed out recently @xcite .
the electron - doped cuprates @xcite have provided an early example where quantum critical behavior has been inferred from the temperature dependence of resistivity at low temperature .
it was measured to be linear @xcite from 35mk to 10k in pr@xmath11ce@xmath12cuo@xmath13 ( pcco ) at doping @xmath14 .
more recent transport @xcite and thermopower measurements @xcite also suggest the presence of a quantum critical point at a similar doping .
the precise nature of the quantum critical point remains however unclear , as thoroughly discussed in ref . .
for example , it has also been suggested that the quantum critical point coincides with the onset of superconductivity in the overdoped regime @xcite . .
the arrows gives examples of pseudo - nesting conditions , namely of points such that @xmath15 is equal to the antiferromagnetic wave vector .
the @xmath0 ordering wave - vector is defined with respect to the @xmath16 coordinate system . in the field theory approach introduced later , we work in the rotated ( @xmath17 ) coordinate system . ] here we study quantum critical behavior associated with the transition between a sdw phase and a fermi liquid when the wave vector @xmath0 of the sdw connects hot spots with parallel fermi velocities .
the two fermi surfaces connected by @xmath0 in this case are tangent to each other , as shown in fig .
[ bzf ] . on a spherical fermi surface
, the fermi wave vector would satisfy the condition @xmath18 .
we call this pseudo - nesting @xcite
. we will explain why the qcp can be located at , or close to , the pseudo - nesting filling @xmath19 .
this occurs naturally in the one - band hubbard model for the electron - doped cuprates and we will perform some of our calculations specifically for this case , although the frequency and temperature dependencies that we find are valid more generally .
the methods that we describe below can be applied to electron - doped cuprates because these materials are described by a hubbard model in an intermediate coupling regime where one can neglect effects induced by mott physics @xcite .
the theory of hertz @xcite and millis@xcite has formed the basis for much of the work on quantum critical phenomena . in this approach ,
fermions are integrated out and an effective bosonic theory for the collective modes is studied using standard renormalization group methods that can be taken to high order @xcite .
it has been pointed out by abanov and chubukov @xcite that for a commensurate sdw at the upper critical dimension , namely @xmath20 for @xmath4 , all the coefficients of the bosonic theory are singular so that one must treat simultaneously the bosonic collective modes and the fermions .
metlitski and sachdev @xcite have reexamined this problem and obtained the non - fermi liquid behavior at the hot spots , and shown that the bosonic sdw spectrum does not obey dynamic scaling with @xmath4 but instead that a super - power - law form is obtained .
they have also thoroughly discussed the failure of the @xmath21 expansion at higher order , leading to a strong - coupling problem .
more recently , it has been argued @xcite that non - fermi liquid corrections are also important away from the hot spots .
an alternate approach is the self - consistent - renormalized theory of moriya @xcite .
it is in the universality class of the spherical model and as such its critical behavior will not be exactly that expected for the o(3 ) model .
nevertheless , it can be accurate away from the critical point and provide leading order estimates for the exponents .
the two - particle - self - consistent ( tpsc ) theory @xcite is a related approach that has critical behavior similar to that of moriya , including logarithmic corrections @xcite .
it has the advantage that although it is an approximate solution to the hubbard model , it is quantitatively very close to benchmark quantum monte carlo results @xcite .
it is non - perturbative , does not include any phenomenological parameters , has internal consistency checks , and satisfies a number of exact results .
previous theoretical studies have mostly been done for the case where the quantum critical point occurs when the fermi velocities at the hot spots that are connected by the sdw are not parallel , unlike the case of parallel velocities we consider here ( see fig .
[ bzf ] ) .
such a case of parallel fermi velocities is generic in one dimension , but at first sight appears as an accident in two dimensions , because upon translation by the @xmath0 sdw wave vector , the fermi surfaces touch at only one point .
if the surfaces were flat , we would recover the case of perfect nesting encountered in one dimension .
the curvature here provides a cutoff , and so we refer to the situation with parallel fermi velocities as `` pseudo - nesting '' @xcite .
such fermi surfaces have also been studied in three dimensions @xcite , but in the presence of this pseudo - nesting the spin susceptibility is singular in two dimensions @xcite and the analysis must be redone .
altshuler , ioffe and millis @xcite first looked at the case where the instability is at @xmath22 , hence connects parallel segments of the fermi surface , but the sdw wavelength is not commensurate with the lattice .
they found that the transition is weakly first order with an intermediate scaling regime when the sdw wavelength is close to @xmath0 .
the scaling regime was obtained in a systematic expansion in a number proportional to the inverse number of fermion flavors .
krotkov and chubukov @xcite found different results for the self - energy .
here we consider only the commensurate case .
some of our results differ from those of previous authors because they overlooked the significance of the umklapp process shown by the top double arrow in fig .
[ bzf ] .
we use two different approaches .
we obtain finite temperature results appropriate for the sdw quantum critical point of electron - doped cuprates using tpsc .
then a field - theory for the spin - fermion model allows us to find the finite - frequency zero temperature results and some finite temperature results .
the results of both approaches are consistent .
we do not , however , consider the possibility of a first order transition @xcite .
the rest of this paper is organized as follows . in sec .
[ sec : model ] we present the model along with general arguments suggesting why one should expect the antiferromagnetic quantum critical point to be located close to pseudo - nesting , namely at a filling where the @xmath0 wave vector connects parts of the fermi surface that are tangent , or equivalently with parallel fermi velocities .
[ sec : tfinite ] contains the finite temperature results .
they are obtained with tpsc , which is described in sec .
[ sub : tpsc ] . analytical results for the behavior at the qcp are illustrated with numerical examples appropriate for electron - doped cuprates in the subsections of sec .
[ sub : analytical ] .
the critical behavior is obtained in sec .
[ sub : criticaltpsc ] .
zero temperature finite - frequency results are treated in sec .
[ sec : fieldtheory ] with field theoretical methods .
the lagrangian appears in sec .
[ sub : lagrangian ] followed by sections on the polarization bubble ( spin susceptibility ) [ sub : polarizationft ] , on the electron self - energy [ sub : selfft ] and on the irrelevance of the quartic term [ sub : scaling ] .
an appendix on vertex corrections [ sec : vertex ] appears after the summary in sec.[sec : summary ] .
consistency between tpsc and field theory results are pointed out in the field theory sec .
[ sec : fieldtheory ] .
in this section , we introduce the model and give generic arguments why we expect the quantum critical point to often be located close to the filling where translation by the antiferromagnetic wave vector leads to fermi surfaces that are tangent to each other , as illustrated in fig . [ bzf ] .
while the frequency and temperature dependencies that we find do not depend on details of the model , specific numerical examples at finite temperature calculations will be performed on the two - dimensional @xmath23 hubbard model on the square lattice at weak to intermediate coupling .
the model is given by @xmath24 where @xmath25 are the hopping integrals , @xmath26 are the site index , @xmath27 is the spin label , @xmath28 and @xmath29 are the particle creation and annihilation operators .
doubly occupied sites cost an energy @xmath30 and @xmath31 .
units are such that @xmath32 , @xmath33 and lattice spacing is unity .
the kinetic energy of a single - particle excitation in momentum space is obtained from@xmath34 with the sum over @xmath35 running over all neighbors of any of the sites @xmath36 .
the chemical potential @xmath37 is chosen so that we have the required density .
one can explain on general grounds the filling where the qcp is likely to occur .
figure [ fig : suscepatqcp ] displays the lindhard function @xmath38 along the @xmath39 direction for different fillings . its maximum is at @xmath0 so it is symmetric in @xmath40 and @xmath39 .
there are two remarkable features .
first , below a certain doping @xmath19 , the maximum value is almost independent of filling @xcite , and second it falls rapidly as soon as the filling exceeds @xmath19 .
the filling @xmath19 corresponds to the point where the fermi surfaces joined by @xmath0 touch instead of intersecting .
a @xmath41 change in filling leads to almost @xmath42 drop in value of the susceptibility .
if we consider a simple stoner criterion for the transition , we would conclude that if @xmath30 takes the value @xmath43 , then the qcp would be close to this filling @xmath19 .
this does not require fine tuning because the value of @xmath30 that should enter the stoner criterion is the value renormalized by kanamori - brckner screening @xcite .
this renormalized value becomes essentially @xmath30 independent when @xmath30 becomes of the order of the bandwidth because the two - body wave function creates a cusp to minimize double - occupancy @xcite and the renormalized interaction can not become larger .
this maximum renormalized value in tpsc , @xmath44 , takes a value @xcite near @xmath45 .
in addition , in tpsc the value of @xmath44 self - consistently adjusts itself to the value necessary to prevent a finite temperature phase transition on the sdw side of the qcp .
although , at sufficiently low temperature , details will start to matter and one needs to start to tune the value of @xmath30 to find the qcp precisely at @xmath19 , there is an intermediate temperature scale that can be quite broad where fine tuning is unnecessary .
point as a function of doping for @xmath46 , @xmath47 and @xmath48 , values that are appropriate for electron - doped cuprates .
the rapid fall with filling larger than @xmath49 , close to the critical filling , is apparent.,title="fig : " ] +
in this section , we use the non - perturbative two - particle self - consistent ( tpsc ) approach @xcite .
this approach respects the pauli principle , the mermin - wagner theorem and conversation laws .
it also contains quantum fluctuations in crossed channels that lead to kanamori - brckner screening .
@xcite it is valid in the weak to intermediate coupling regime @xmath50 and not too deep in the renormalized classical regime where a pseudogap is observed .
it has been benchmarked on quantum monte carlo calculations on the hubbard model . @xcite .
tpsc has been shown to be in the @xmath51 universality class of the @xmath52 model @xcite .
it has the same critical behavior as moriya theory and hence has the same logarithmic corrections @xcite .
these logarithms have the same functional form as those of the renormalization group asymptotically close to the quantum critical point , but in tpsc and in moriya theory the mode - mode coupling term does not flow , hence the corrections may differ in the details from the renormalization group @xcite
. quantum critical behavior of the susceptibility and of the self - energy in the closely related spin - fermion model has been discussed by abanov et al .
@xcite .
it has been argued from detailed comparisons of numerical calculations with experiment @xcite that strong - coupling physics is not important for electron - doped cuprates , at least not too close to half - filling .
hence , tpsc is appropriate to study these compounds .
it gives a satisfactory description of arpes data @xcite , and the temperature @xmath53 , where the pseudogap seen in arpes opens up experimentally , corresponds to that where the antiferromagnetic correlation length coincides with the thermal de broglie wavelength @xcite , as predicted for two - dimensional precursors of three - dimensional long - range order @xcite .
hence , all the numerical results are presented in units where @xmath54 , @xmath55 , @xmath32 ( with @xmath56 component of spin defined by @xmath57 ) for values of the hubbard model hopping parameters appropriate for electron - doped cuprates , namely second and third nearest - neighbor hopping @xmath47 and @xmath48 @xcite .
interaction strengths @xmath46 and @xmath58 , again in the range appropriate for electron - doped cuprates , @xcite will be considered .
we first present the formalism and then give analytical and numerical results for the qcp .
given the hubbard model parameters , tpsc has no adjustable parameter .
irreducible vertices are obtained self - consistently and in such a way that the pauli principle and conservation laws are obeyed . the formal derivation is given in refs . .
here , we simply present the equations that are solved . in tpsc , the retarded spin @xmath59 and charge
@xmath60 susceptibilities are written as@xmath61 where @xmath62 is the non - interacting retarded lindhard function at wave vector @xmath63 and angular frequency @xmath64@xmath65 here , @xmath66 is the fermi function @xmath67 , @xmath68 is the temperature and @xmath69 is the total number of sites .
the effective spin interaction @xmath44 is evaluated without adjustable parameter using the _ _ ansatz _ _ @xcite@xmath70 with the local - moment sum rule that follows from the fluctuation - dissipation theorem@xmath71 where @xmath72 and @xmath73 is the double occupancy .
we dropped the site index using translational invariance and we used the pauli principle to write @xmath74 similarly the irreducible vertex @xmath75 entering @xmath76 is found using a sum - rule that is the analog of eqn.([sum_rule ] ) for spin : @xmath77 the crossing symmetric self - energy is obtained from @xmath78 g_{\sigma } ^{\left ( 1\right ) } ( k+q).\label{self}%\ ] ] the superscript @xmath79 reminds us that we are at the second level of approximation .
@xmath80 is the same green s function as that used to compute the susceptibilities @xmath81 .
charge fluctuations @xmath76 are included in numerical calculations but they are neglected in the analytical results because they are small .
since the self - energy is constant at the first level of approximation , this means that @xmath80 is the non - interacting green s function with the chemical potential that gives the correct filling .
this chemical potential @xmath37 is slightly different from the one that we must use in @xmath82 to obtain the same density @xcite .
unless otherwise specified , all the numerical results below are obtained using the matsubara frequency version of equations ( [ chisp ] ) to ( [ self ] ) without any approximation , hence they are valid at arbitrary distance from the quantum critical point .
we begin below with the ornstein - zernicke form of the spin susceptibility that is usually valid when the correlation length is large .
the case where there is perfect nesting leads us naturally to the pseudo - nesting condition relevant for this paper .
the situation , however , is not as simple as usual since the ornstein - zernicke form for the spin susceptibility is incorrect in our case , as we will explain .
the self - energy is treated at the end of this section .
[ [ general - case ] ] general case + + + + + + + + + + + + when the correlation length is large , one usually assumes that the denominator of the spin susceptibility can be expanded around the wave vectors @xmath83 , where the maxima in @xmath84 occur in @xmath85dimensions .
one then obtains @xmath86 where @xmath63 is measured with respect to the wave vector @xmath87 where the spin suseptibility is maximum ( @xmath0 in our case ) . defining @xmath88 as the value of the interaction at the mean - field sdw transition , the other quantities in the previous expression are @xmath89 in the expression for the spin susceptibility ,
the denominators are expanded around the @xmath90 wave vector . to obtain analytical results for the imaginary part of the self - energy @xmath91 in eqn.([self ] )
we use the spectral representation for the susceptibilities and for the green s function , perform the sum of the internal matsubara frequency and then the analytical continuation neglecting the charge fluctuations , to obtain @xmath92 \chi_{sp}^{\prime\prime } \left ( \mathbf{q}_{\vert},q_{\perp}\left ( \mathbf{k}_{f}+\mathbf{q}% _ { d},\omega,\omega^{\prime}\right ) ; \omega^{\prime}\right ) \label{imsigmareel}%\]]where @xmath93 , the component of @xmath63 parallel to the fermi velocity @xmath94 , is obtained from the solution of the equation @xmath95 for all fermi wave vectors , where @xmath96 the above equation reduces to @xmath97 where @xmath98 is the fermi velocity in the hot region , i.e. where @xmath99 in the asymptotic form of the spin susceptibility eqn.([chi_as ] ) , the wave vector appears only in the form @xmath100 so that keeping this general form in the equation for the self - energy eqn.([imsigmareel ] ) , we obtain @xmath101 \chi_{sp}^{\prime\prime } \left ( \mathbf{q}_{\vert},(\omega+\omega')/v_f ; \omega^{\prime},t\right ) . \label{selfwithchipp}\ ] ] normally , one expects @xmath102 to be a temperature independent constant of the order of the lattice spacing and @xmath103 to be a constant of the order of the fermi energy . in the case of perfect nesting , or of pseudo - nesting , this is not the case .
[ [ perfect - nesting ] ] perfect nesting + + + + + + + + + + + + + + + although the case we are interested in does not correspond to perfect nesting , understanding that case first will facilitate our task later .
there is perfect nesting when the equality @xmath104 is satisfied for all wave vectors , with @xmath83 the nesting wave vector .
this case was treated by virosztek and ruvalds@xcite .
the quantities @xmath105 and @xmath106 that are usually assumed temperature independent , here become temperature dependent .
we show this below . for perfect nesting , the lindhard function becomes@xmath107 so that changing to an energy integral we have@xmath108 where @xmath109 is the density of states . the real part at zero frequency on the other hand is given by@xmath110 in two dimensions , there is a well known logarithmic divergence of the density of states @xmath111 at the van hove singularity .
neglecting this logarithmic divergence that appears only for a special filling in the hole - doped case , we take @xmath111 as a constant . in that case , integrating by part and replacing the upper bound by infinity in the convergent integral we are left with @xmath112 where @xmath113 is a temperature independent constant .
we also have @xmath114 these results suggest that the quantity @xmath106 defined by eqn.([gamma_0 ] ) scales as @xmath115 following ref .
@xmath116 we move on to demonstrate analytically that @xmath105 in eqn.([xsi02 ] ) scales as @xmath117 .
the @xmath118 dependence , fundamentally comes from the second derivative of @xmath119 in eqn.([xhi1log ] ) .
we shall now make this argument more rigorous . keeping for a while a general notation where @xmath36 is some direction in the brillouin zone , and @xmath63
is measured with respect to the center of the zone , one can write @xmath120
where @xmath121 measuring @xmath63 with respect to @xmath83 we evaluate the above second derivative at @xmath122 as before , for perfect nesting we have @xmath123 the last equation shows that @xmath124 scales as @xmath125 times a function of @xmath126 in the integrals , the derivatives of the type @xmath127 will not introduce singular terms in temperature .
hence , replacing them by some average value in the brillouin zone , we can change the integration variable to energy and the most singular terms in temperature will come from @xmath128 neglecting the energy dependence of the density of states , we are left with@xmath129 using the definition of @xmath105 , eqn.([xsi02 ] ) , and the result for @xmath106 eqn.([gamma0avecxsi0 ] ) above , we have that @xmath130 [ [ pseudo - nesting ] ] pseudo - nesting + + + + + + + + + + + + + + touches the antiferromagnetic brillouin zone . the important integration region is over a rectangle of thickness @xmath68 and width @xmath131 . the critical chemical potential @xmath132 where the antiferromagnetic zone boundary touches @xmath133 on the fermi surface is the solution of @xmath134 .
this corresponds to a filling @xmath135 for @xmath136 and @xmath137 . for @xmath46 , the critical filling that we find , @xmath138 ,
is slightly larger .
, title="fig : " ] + in this subsection we show that , for the pseudo - nesting case , the main contribution to the lindhard function at @xmath139 has the same form as in the perfect nesting case except for a temperature dependant prefactor .
the previous calculation illustrates that the main contribution to the quantities of interest , @xmath105 and @xmath140 come from nested regions of the fermi surface . in the pseudo - nesting case @xcite
illustrated in fig .
[ fig : integrationdomain ] , the fermi surface displaced by @xmath139 just touches the original fermi surface , with the fermi velocities of the two surfaces that are parallel at the touching point . as in the perfect nesting case , the most important contribution to the integral for the lindhard function around @xmath139 comes from the regions in @xmath141-space connected by @xmath139 , with a width around the fermi surface that corresponds to an energy range @xmath142 .
now , imagine that we divide the integral over @xmath143 near one of those points of the fermi surface , for example @xmath144 , into two components , @xmath145 parallel to the fermi velocity , i.e. perpendicular to the fermi surface , and @xmath146 parallel to the fermi surface at @xmath147 . for @xmath143 near that region of the fermi surface
, we can write in two dimensions @xmath148 where we have measured wave vectors with respect to @xmath147 and used the fact that @xmath149 the quantity @xmath150 measures the curvature of the fermi surface . from that approximation
we have @xmath151 and thus @xmath152 since the terms in @xmath153 cancel out in that expression , this approximation is valid if the next term in the series of @xmath154 is negligible compared with the first one , namely if @xmath155 and since @xmath156 , we have the following upper bound for the temperature , @xmath157 now , for the fermi function , over a region around @xmath147 , we have @xmath158 since the derivative of the fermi function is even in @xmath159 , @xmath160 the region where this is valid is given by the condition @xmath161 which is satisfied if @xmath162 .
therefore , from and , the lindhard function takes the form @xmath163 where @xmath164 is a domain such that @xmath165 while @xmath166 is the domain such that @xmath167 @xmath168 the integration over @xmath166 thus gives a factor proportional to @xmath169 . the integral over @xmath170 can be transformed into an integral over energy in the same way as the perfect nesting case , with a constant density of states determined by the fermi velocity .
the domain delimited by @xmath164 and @xmath166 is depicted in fig .
[ fig : integrationdomain ] .
overall then , the final result will be that @xmath171 where the @xmath172 prefactor comes from the @xmath173 integration .
a similar reasoning leads to@xmath174 which means that the regular temperature independent term represented here by @xmath175 dominates at low temperature . repeating the same analogous arguments for @xmath176 we find that @xmath177 which implies from the definition of @xmath178 eqn.([gamma_0 ] ) and eqs.([chi(1)0]-[chi(1)0 ] ) that @xmath179the same result for @xmath106 , within logarithmic corrections , as if we had perfect nesting . in three dimensions ,
the correction compared to perfect nesting is determined by the area spanned by @xmath180 , proportional to @xmath181 , hence we would have had @xmath182 and again @xmath183 results of numerical calculations shown in fig .
[ fig : xsi0_and_gamma_0 ] confirm the power law temperature dependencies found above . at the actual filling @xmath19 where the fermi surface is tangent to the antiferromagnetic zone boundary ,
the power laws extend to low temperature .
and @xmath178 evaluated at @xmath135 obtained from the condition that the fermi surface is tangent to the antiferromagnetic zone boundary , as explained in the caption of fig .
[ fig : integrationdomain ] . ] and @xmath178 evaluated at @xmath135 obtained from the condition that the fermi surface is tangent to the antiferromagnetic zone boundary , as explained in the caption of fig .
[ fig : integrationdomain ] . ] given that @xmath102 and @xmath106 are now temperature dependent , we should check whether the small @xmath63 and small @xmath64 expansion of the denominator that lead to eqn.([chi_as ] ) is still valid .
normally , the expansion is of the form @xmath184 with @xmath185 a constant . since the function falls on a scale @xmath186
the higher order term @xmath187 can be neglected . however , our case is different .
the coefficients of the expansion in powers of @xmath63 are singular at @xmath188 .
for example we have @xmath189 where the @xmath190 comes from counting the powers of @xmath68 associated with derivatives in eqs.([d2chidq2initial ] ) to ( [ d2chidq2 ] ) and the @xmath172 from the restriction to the @xmath191 integral as usual . knowing the scaling of @xmath105
, we can rewrite @xmath192 so that we are left with@xmath193 where @xmath194 is a constant .
the susceptibility will preserve a scaling form as a function of @xmath195 and @xmath196 if the scaling exponent is @xmath3 .
indeed , in that case @xmath197 and since @xmath198 @xmath199 , the susceptibility becomes @xmath200 with @xmath201 constants .
each higher power of @xmath202 has an additional power of @xmath203 coming from the additional derivatives of the non - interacting susceptibility and the scaling form is preserved to all orders . for frequency , there are also higher order terms , @xmath204 etc .
hence we have the general scaling form @xmath205 we check in the following section that this is consistent with the tpsc self - consistency condition .
the quantum critical behavior has been thoroughly studied in ref . for the case where the ornstein - zernicke form is valid .
this analysis does not apply here because of the more general form of the spin susceptibility obtained in the previous section .
nevertheless , it is interesting to note that with the ornstein - zernicke form , one obtains @xmath206 for both perfect and pseudo nesting .
hence , simply taking into account the temperature dependence of @xmath178 , we recover @xmath3 scaling , namely@xmath207 note that with a temperature independent value for @xmath178 we recover the more usual result @xcite @xmath4 .
the physics of the result for the correlation length is however quite different from the calculation with the ornstein - zernicke form .
indeed , in the latter case , it is the self - consistency relation eqn.([sum_rule ] ) that determines the temperature dependence of the correlation length . in the present case
, we found that temperature dependence in the previous section without invoking the self - consistency relation .
we will show in sec.([sub : scaling ] ) that indeed in our case , the self - consistency relation leads to irrelevant corrections to the temperature dependence of the correlation length .
the scaling of the correlation length can be obtained from numerical calculations by plotting , for example , the inverse of the width of the real part of the spin susceptibility at zero frequency measured at various fractions of the maximum value . for @xmath58
, the critical doping corresponds to @xmath135 where the fermi surface is tangent to the antiferromagnetic zone boundary . for that case ,
[ fig : chi_sp](a ) shows that whether we measure the width at half - maximum or at some other fraction of the maximum , that width scales essentially as @xmath2 , with small deviations probably coming from the fact that we have not reached the asymptotic regime .
we also show on this figure the correlation length @xmath208 obtained from the ornstein - zernicke form eqn.([xsi2 ] ) using @xmath44 from the self - consistency relation eqn.([sum_rule ] ) .
deviations from the @xmath2 power law occur if we measure the width of the spin susceptibility too far in the tails , i.e. for a small fraction of the maximum ( not shown ) .
as demonstrated in fig .
[ fig : chi_sp](b ) , deviations from @xmath2 also occur at low temperature for a value of @xmath30 ( @xmath209 in the present example ) where the critical point does not occur precisely when the fermi surface is tangent to the antiferromagnetic zone boundary . at various fractions of the maximum height . on ( a ) , for @xmath58 , the critical doping corresponds to @xmath135 , where the fermi surface is tangent to the antiferromagnetic zone boundary .
the temperature scale is too small to detect possible logarithmic corrections .
the results are consistent with @xmath3 . on ( b )
, deviations from @xmath2 occur at low temperature because , for the chosen value @xmath46 , the calculation is at the critical doping @xmath210 , slightly away from the point where the fermi surface is tangent to the antiferromagnetic zone boundary .
also shown , @xmath208 obtained from the ornstein - zernicke form eqn.([xsi2 ] ) using @xmath44 from the self - consistency relation eqn.([sum_rule ] ) ] at various fractions of the maximum height . on ( a ) , for @xmath58 , the critical doping corresponds to @xmath135 , where the fermi surface is tangent to the antiferromagnetic zone boundary .
the temperature scale is too small to detect possible logarithmic corrections .
the results are consistent with @xmath3 . on ( b )
, deviations from @xmath2 occur at low temperature because , for the chosen value @xmath46 , the calculation is at the critical doping @xmath210 , slightly away from the point where the fermi surface is tangent to the antiferromagnetic zone boundary .
also shown , @xmath208 obtained from the ornstein - zernicke form eqn.([xsi2 ] ) using @xmath44 from the self - consistency relation eqn.([sum_rule ] ) ] to conclude this section , we show that one can easily obtain the temperature scaling for two more quantities .
first , from the general form for the susceptibility eqn.([suscep_simple ] ) used above , the static susceptibility at @xmath0 scales as @xmath211 which can be checked directly numerically , or more simply deduced from the temperature dependent results for @xmath212 , eqn.([xsi_qcp ] ) , and @xmath213 , eqn.([xsi_0 ] ) .
behave as @xmath7 over the accessible temperature range .
calculations are done with @xmath58 , @xmath47 and @xmath48 at the quantum critical filling @xmath135 .
the lower figures are the corresponding results for @xmath46 , @xmath47 and @xmath48 at the quantum critical point @xmath214 . since in that case
the fermi surface is not exactly tangent to the antiferromagnetic zone boundary , the @xmath7 behavior near @xmath215 is recovered only if the temperature is high enough that details of the fermi surface can not be resolved .
the black lines on ( b ) and ( d ) are the curves defined by @xmath216 where @xmath217 is the component of @xmath218 parallel to @xmath219 at a given angle @xmath220 . ]
behave as @xmath7 over the accessible temperature range .
calculations are done with @xmath58 , @xmath47 and @xmath48 at the quantum critical filling @xmath135 .
the lower figures are the corresponding results for @xmath46 , @xmath47 and @xmath48 at the quantum critical point @xmath214 .
since in that case the fermi surface is not exactly tangent to the antiferromagnetic zone boundary , the @xmath7 behavior near @xmath215 is recovered only if the temperature is high enough that details of the fermi surface can not be resolved .
the black lines on ( b ) and ( d ) are the curves defined by @xmath216 where @xmath217 is the component of @xmath218 parallel to @xmath219 at a given angle @xmath220 . ]
behave as @xmath7 over the accessible temperature range .
calculations are done with @xmath58 , @xmath47 and @xmath48 at the quantum critical filling @xmath135 .
the lower figures are the corresponding results for @xmath46 , @xmath47 and @xmath48 at the quantum critical point @xmath214 .
since in that case the fermi surface is not exactly tangent to the antiferromagnetic zone boundary , the @xmath7 behavior near @xmath215 is recovered only if the temperature is high enough that details of the fermi surface can not be resolved .
the black lines on ( b ) and ( d ) are the curves defined by @xmath216 where @xmath217 is the component of @xmath218 parallel to @xmath219 at a given angle @xmath220 . ]
behave as @xmath7 over the accessible temperature range .
calculations are done with @xmath58 , @xmath47 and @xmath48 at the quantum critical filling @xmath135 .
the lower figures are the corresponding results for @xmath46 , @xmath47 and @xmath48 at the quantum critical point @xmath214 .
since in that case the fermi surface is not exactly tangent to the antiferromagnetic zone boundary , the @xmath7 behavior near @xmath215 is recovered only if the temperature is high enough that details of the fermi surface can not be resolved .
the black lines on ( b ) and ( d ) are the curves defined by @xmath216 where @xmath217 is the component of @xmath218 parallel to @xmath219 at a given angle @xmath220 . ] finally , the nuclear magnetic resonance ( nmr ) relaxation rate @xmath221 can be obtained from the two - dimensional version of the moriya formula @xmath222 where @xmath223 is proportional to the hyperfine matrix element . taking this as a constant and using the general scaling form eqn.([chispscaling ] ) , a simple change of integration variable shows that @xmath224 however , the integral over momenta @xmath225 also contains contributions far from the peak in the susceptibility . there the susceptibility is essentially temperature independent .
there is thus a korringa contribution @xmath226 that is dominant at low temperature . to find the scaling of the self - energy
, we use the scaling form of the susceptibility eqn.([chispscaling ] ) to rewrite the self - energy eqn.([selfwithchipp ] ) in the form@xmath92 \frac{1}{\sqrt{t}}g\left(\frac{q_{\vert}}{t},\frac{(\omega+\omega')/v_f}{t};\frac{\omega}{t}\right ) % \ ] ] specializing to two dimensions and remembering the scaling of the bose and fermi functions with frequency and temperature , we change integration variables to @xmath227 and @xmath228 and we are left with @xmath229 where @xmath230 is a scaling function . at various angles along the fermi surface .
the result of a power law fit is shown on ( b ) and ( d ) . the dashed lines on ( a ) and ( c ) correspond to the fitted power laws . at the hot spot located at @xmath215 ,
@xmath231 scales as @xmath232 .
the frequency range is small because of the low temperature saturation shown on the next figure .
we have verified that the crossover from @xmath232 to the fermi liquid regime @xmath233 occurs on a broader angular scale when the temperature is higher , as expected from the results of fig .
[ fig : self_t_tpsc ] .
calculations are done at @xmath234 , @xmath47 and @xmath48 at @xmath58 and @xmath235 for ( a ) and ( b ) and @xmath46 , @xmath214 for ( c ) and ( d ) . ] at various angles along the fermi surface .
the result of a power law fit is shown on ( b ) and ( d ) .
the dashed lines on ( a ) and ( c ) correspond to the fitted power laws . at the hot spot located at @xmath215 ,
@xmath231 scales as @xmath232 .
the frequency range is small because of the low temperature saturation shown on the next figure .
we have verified that the crossover from @xmath232 to the fermi liquid regime @xmath233 occurs on a broader angular scale when the temperature is higher , as expected from the results of fig .
[ fig : self_t_tpsc ] .
calculations are done at @xmath234 , @xmath47 and @xmath48 at @xmath58 and @xmath235 for ( a ) and ( b ) and @xmath46 , @xmath214 for ( c ) and ( d ) . ] at various angles along the fermi surface .
the result of a power law fit is shown on ( b ) and ( d ) .
the dashed lines on ( a ) and ( c ) correspond to the fitted power laws . at the hot spot located at @xmath215 ,
@xmath231 scales as @xmath232 .
the frequency range is small because of the low temperature saturation shown on the next figure .
we have verified that the crossover from @xmath232 to the fermi liquid regime @xmath233 occurs on a broader angular scale when the temperature is higher , as expected from the results of fig .
[ fig : self_t_tpsc ] .
calculations are done at @xmath234 , @xmath47 and @xmath48 at @xmath58 and @xmath235 for ( a ) and ( b ) and @xmath46 , @xmath214 for ( c ) and ( d ) . ] at various angles along the fermi surface .
the result of a power law fit is shown on ( b ) and ( d ) .
the dashed lines on ( a ) and ( c ) correspond to the fitted power laws . at the hot spot located at @xmath215 ,
@xmath231 scales as @xmath232 .
the frequency range is small because of the low temperature saturation shown on the next figure .
we have verified that the crossover from @xmath232 to the fermi liquid regime @xmath233 occurs on a broader angular scale when the temperature is higher , as expected from the results of fig .
[ fig : self_t_tpsc ] .
calculations are done at @xmath234 , @xmath47 and @xmath48 at @xmath58 and @xmath235 for ( a ) and ( b ) and @xmath46 , @xmath214 for ( c ) and ( d ) . ]
the latter result can be checked numerically at @xmath236 where we expect @xmath237 .
[ fig : self_t_tpsc](a ) and [ fig : self_t_tpsc](c ) displays the temperature dependent scattering rate for various angles @xmath220 along the fermi surface for @xmath58 and @xmath235 and @xmath46 and @xmath214 , respectively . the line @xmath238 is horizontal in the brillouin zone appearing in the inset . in fig .
[ fig : self_t_tpsc](a ) , at the hot spot located at @xmath239 , we recover the predicted result , @xmath7 .
this is best illustrated in fig .
[ fig : self_t_tpsc](b ) by a plot of the local slope of the preceding log - log plot .
as we move away from the hot spot , fermi liquid behavior appears to be recovered .
there are well known logarithmic corrections in two dimensions @xcite that may explain why we seem to deviate from exactly @xmath240 .
one also notices that the @xmath7 behavior occurs over a wider range of angles when the temperature is high .
this is easily understood from figs .
[ fig : self_t_tpsc](b ) and [ fig : self_t_tpsc](d ) that illustrates how temperature affects the domain where the pseudo - nesting occurs in the spin susceptibility . the solid black line in figs .
[ fig : self_t_tpsc](b ) and [ fig : self_t_tpsc](d ) is defined by @xmath216 where @xmath217 is the component of @xmath218 parallel to @xmath219 at a given angle @xmath220 .
[ fig : self_t_tpsc](c ) and figs .
[ fig : self_t_tpsc](d ) are for @xmath241 .
since at this quantum critical point the fermi surface is not tangent to the antiferromagnetic zone boundary , the @xmath7 behavior occurs near @xmath239 only at high enough temperature . at low temperature ,
deviations become apparent . when @xmath242 , the scaling form for the self - energy eqn.([self - scaling ] ) predicts @xmath243 . however , at zero temperature , or when @xmath242 , the analytical approach taken above fails because the expansion of the spin susceptibility in @xmath196 and @xmath195 is no longer justified and we can not expect the latter result to be correct .
nevertheless , tpsc can be solved numerically . to set the stage for the next section where
calculations are performed analytically directly at zero temperature , we show in fig . [
fig : self_w_tpsc ] the result of the numerical calculations for @xmath242 for two values of the interaction strength at a doping near @xmath19 . on figs .
[ fig : self_w_tpsc](a ) and [ fig : self_w_tpsc](b ) , @xmath58 while @xmath46 on figs .
[ fig : self_w_tpsc](c ) and [ fig : self_w_tpsc](d ) . at the hot spot for @xmath58 ,
the scaling of the imaginary part of the self energy is very close to the expected result @xmath232 . for @xmath46
, there is a larger discrepancy with the predicted scaling because at @xmath214 the fermi surface does not touch @xmath133 and thus the present theory does not apply anymore at low temperature . away from the hot spot ,
fermi liquid behavior is recovered .
again , logarithmic corrections are inaccessible from the numerical solution of the full tpsc equations because of the limited range of available frequencies : scaling is no - longer valid at frequency of the order of the fermi energy , while at low frequency there is a saturation arising from the finite temperature .
this saturation is illustrated in fig .
[ fig : self_saturation_tpsc ] .
we discuss analytically the @xmath188 regime in the following section where logarithmic corrections are found .
the @xmath6 temperature dependence of the static @xmath0 spin susceptibility obtained above will also be recovered .
at the hot spot for two different temperatures .
the saturation at low frequency occurs at higher frequency when the temperature is higher .
calculations are done with @xmath46 , @xmath47 and @xmath48 at the quantum critical filling @xmath244 appropriate for electron - doped cuprates.,title="fig : " ] +
in this section we study the properties of fermionic excitations close to the hot spots within the field - theoretic framework of a spin - fermion model .
this effective low - energy theory describes fermions with a parabolic dispersion ( represented by fields , @xmath245 ) interacting with sdw fluctuations ( represented by a o(3 ) vector field , @xmath246 ) . as shown in fig .
[ bzf ] there are four hot - spots on the fermi surface which are connected by the sdw wave - vector @xmath247 .
earlier studies@xcite of the spin - fermion model in the present context did not include the umklapp processes properly and we show in the following that a correct treatment of these terms modifies the results drastically .
we start with the two patch ( denoted by @xmath248 ) model @xcite in the rotated @xmath249 coordinate system .
the umklapp contributions will be discussed later . in order to simplify the notation
, we have rescaled our coordinates to get rid of the fermi velocity and curvature of the fermi - surface .
the corresponding lagrangian takes the form @xmath250 here we have promoted each fermion field to have @xmath69 flavors ( the flavor index is suppressed ) .
the yukawa - coupling , @xmath251 , is chosen to be @xmath252 . as a result of this ,
the couplings of all the bosonic terms in the last line above are scaled by a factor of @xmath69 as they will appear naturally upon integrating out the fermion fields .
we do nt include the kinetic energy of the boson , @xmath253 , as this is an irrelevant term @xcite .
the bare fermion propagator is given by , @xmath254 the fermi surfaces are located at @xmath255 and @xmath256 for patch @xmath257 and @xmath258 respectively . from fig .
[ bzf ] we immediately observe that @xmath259 and also @xmath260 connect two more points in the bz .
however , it is convenient to fold back the points within the bz , which effectively gives rise to two more patches .
these can be described by rotating the original patches by @xmath261 .
let us denote @xmath262 by @xmath263 .
then , the equations of these two additional fermi surfaces are given by @xmath264 .
physically , these two scattering processes are very different since in the former case , the @xmath265fluctuation scatters fermions that disperse strongly in the direction transverse to the fermi surface while in the latter case , they disperse strongly in the tangential direction .
this will have interesting consequences in the behavior of the electron self energy as a function of the external frequency .
the rest of this section is organized as follows : in section [ rpa ] , we compute the rpa contributions to the sdw propagator including both direct as well as umklapp processes .
we then use the dressed bosonic propagator to evaluate the fermion self - energy in section [ fse ] at leading order in @xmath21 .
[ rpa ] at @xmath188 , the one loop polarization bubble ( fig . [ pol ] ) for the two - patch theory
is given by , @xmath266,\\ % \textnormal{i.e . } \\
\pi(q)&=&n[\pi_0(q_\tau,{\bf q})+\pi_0(q_\tau,-{\bf q})],\end{aligned}\ ] ] where we are working with imaginary frequencies @xmath267 and @xmath268 denote the three sdw - polarizations . after performing the integrals
, we obtain @xcite .
the internal solid lines in the loop correspond to the free fermion propagators ( different patches denoted by @xmath248 ) and the external wavy lines correspond to the boson @xmath269 . ]
@xmath270=\frac{\lambda^2}{2\pi } \sqrt{e_{{\bf q}}+\sqrt{e_{{\bf q}}^2 + |q_\tau|^2}}. \label{bubble}\ ] ] where , @xmath271 in the rpa propagator obtained after bubble summation , the @xmath272 contribution determines the location of the quantum critical point .
it is thus convenient to add and subtract this component to make the integrals convergent . from now on ,
we include @xmath272 in the definition of the bubble .
to make connection with results of the previous sections , we also quote the results for the bubble at finite temperature . in this case , @xmath273 where the contour @xmath124 has to be chosen appropriately .
therefore , the above integral simplifies to , @xmath274\ ] ] on integrating the above equation by parts , we obtain , @xmath275}\sqrt{\omega+\sqrt{\omega^2 + |q_\tau|^2 } } , \\
\pi_0(q)&=&\sqrt{t } f\bigg(\frac{|q_\tau|}{t},\frac{e_{\bf q}}{t}\bigg ) , \label{scalingpi0t}\end{aligned}\ ] ] where , @xmath276}\sqrt{y+\sqrt{y^2+\frac{|q_\tau|^2}{t^2}}}\ ] ] as found in ref . . the rpa propagator for the sdw fluctuations is then given by , @xmath277 } , % endmodified% \label{renprop}\ ] ] where we have included the rpa contribution arising from all four hot spots on the fermi surface and @xmath278
the terms ( @xmath279 ) are not equal to ( @xmath280 ) as was incorrectly assumed by the authors of ref . .
at the quantum critical filling @xmath281 , zero matsubara frequency @xmath282 but finite temperature , the @xmath283 term is negligible compared to the contribution of the bubbles @xmath284 . using the scaling form eqn.([scalingpi0 t ] ) and keeping only terms linear in @xmath225 in @xmath285 ( eqn .
[ eofq ] ) , one recovers the zero frequency limit of the previous result eqn.([chispscaling ] ) for the spin susceptibility . naively doing the analytical continuation in frequency
, the full scaling form would also follow .
for the rest of the computations , we consider @xmath188 and carefully take into account logarithmic corrections that were beyond the reach of the previous calculation . before we proceed to evaluate the fermionic self - energy ,
let us compute the forms of the real and imaginary parts of the retarded polarization bubble at @xmath188 , @xmath286 , where @xmath287 .
for the imaginary part , we obtain from eqn .
( [ bubble ] ) , @xmath288 , \\
\textnormal{im}\pi_r({\bf q},\omega)&=&\frac{-\lambda^2}{2\sqrt{2}\pi } \bigg[\theta ( \omega - e_{{\bf q}})\sqrt { \omega - e_{{\bf q}}}-\theta(- \omega - e_{{\bf q}})\sqrt{- \omega - e_{{\bf q}}}\bigg],\end{aligned}\ ] ] which is chosen in a way such that @xmath289 . the real part can also be obtained from eqn .
( [ bubble ] ) or from the kramers - kronig relation @xmath290.\end{aligned}\ ] ] these results agree with those of refs . .
the @xmath291 limit calculated in ref
. also agrees with the above .
[ fse ] we are interested in evaluating the electron self energy ( fig . [ se ] ) at @xmath188 , @xmath292 , which at leading order in @xmath21 is given by , @xmath293 at leading order in @xmath21 .
the @xmath265 propagator includes the one loop bubbles evaluated earlier . ] after analytic continuation @xmath294 we obtain the following expression for the imaginary part of the retarded self - energy @xmath295 \text{im } d_r(\q,\omega-\xi^\mp_{\k-\q}),\ ] ] where @xmath296 and @xmath297 . at @xmath188 and @xmath298
we are left with @xmath299 \text{im } d_r(\q,\omega-\xi^-_{-\q } ) \notag \\ & = & \frac{3\lambda^2}{4\pi^2 } \int dq_y \int_{-q_y^2}^{-q_y^2 + \omega } dq_x \ , \text{im } d_r(\q,\omega-\xi^-_{-\q } ) \notag \\ & \overset{\omega\rightarrow0}{\approx } & \frac{3\lambda^2 \omega}{4\pi^2 } \int dq_y \ , \text{im } d_r(\q,\omega-\xi^-_{-\q } ) \big|_{q_x \to -q_y^2 } \label{eq1}\end{aligned}\ ] ] note that @xmath300 for @xmath301 . in terms of @xmath302 , this can be rewritten as , @xmath303 where @xmath304 is the total retarded rpa bubble including direct and umklapp terms and we have ignored the @xmath305 term in the denominator . for @xmath306 , we get @xmath307 , @xmath308 , @xmath309 and @xmath310 .
therefore , re(im)@xmath311 are given by , @xmath312\nonumber\\ \text{im } \pi_r^{tot}(\q,\omega)=-\frac{\lambda^2}{4\pi } \bigg[\sqrt{2 \omega-3q_y^2 } \
, \theta(2 \omega-3q_y^2)&+&\sqrt{2 \omega+q_y^2 } \ , \theta(2 \omega+q_y^2)\nonumber \\ -\sqrt{-3q_y^2 - 2 \omega } \ , \theta(-3q_y^2 - 2 \omega)&-&\sqrt{q_y^2 - 2 \omega } \ , \theta(q_y^2 - 2 \omega ) \nonumber\\ + \sqrt{2 \omega+2q_y - q_y^4 } \ , \theta(2 \omega+2q_y - q_y^4)&+&\sqrt{2 \omega-2q_y - q_y^4 } \ , \theta(2 \omega-2q_y - q_y^4)\nonumber\\ -\sqrt{2q_y - q_y^4 - 2 \omega } \ , \theta(2q_y - q_y^4 - 2 \omega)&-&\sqrt{-2q_y - q_y^4 - 2 \omega } \ , \theta(-2q_y - q_y^4 - 2 \omega ) \bigg]\nonumber\\\end{aligned}\ ] ] as a starting point , we can drop the @xmath313 terms in the limit of small @xmath64 and retain only the @xmath39 terms in the @xmath314 contributions that we take into account .
this is a consistent way of handling these terms , since if typical @xmath315 , then @xmath313 is smaller than @xmath39 , so that it is justified to drop these terms .
since the integrand is an even function of @xmath39 , we integrate only over @xmath316 .
then , the expression for self energy reduces to , @xmath317\nonumber\\ a&=&\frac{\sqrt{2 \omega+2q_y}+\sqrt{2 \omega-2q_y}+\sqrt{2 \omega-3q_y^2}+\sqrt{2 \omega+q_y^2}}{\bigg[\frac{1}{\lambda^2 } q_y^2+\frac{\sqrt{2q_y+2 \omega}+\sqrt{2 \omega-2q_y}+\sqrt{3q_y^2 + 2 \omega}+\sqrt{2 \omega - q_y^2}}{4\pi}\bigg]^2+\frac{[\sqrt{2 \omega+2q_y}+\sqrt{2 \omega-2q_y}+\sqrt{2 \omega-3q_y^2}+\sqrt{2 \omega+q_y^2}]^2}{16\pi^2 } } \nonumber\\ b&=&\frac{\sqrt{2 \omega+2q_y}-\sqrt{2q_y-2 \omega}+\sqrt{2 \omega-3q_y^2}+\sqrt{2 \omega+q_y^2}}{\bigg[\frac{1}{\lambda^2 } q_y^2+\frac{\sqrt{2q_y+2 \omega}+\sqrt{2q_y-2 \omega}+\sqrt{3q_y^2 + 2 \omega}+\sqrt{2 \omega - q_y^2}}{4\pi}\bigg]^2+\frac{[\sqrt{2 \omega+2q_y}-\sqrt{2q_y-2 \omega}+\sqrt{2 \omega-3q_y^2}+\sqrt{2 \omega+q_y^2}]^2}{16\pi^2 } } \nonumber\\ c&=&\frac{\sqrt{2 \omega+2q_y}-\sqrt{2q_y-2 \omega}+\sqrt{2 \omega+q_y^2}}{\bigg[\frac{1}{\lambda^2 } q_y^2+\frac{\sqrt{2q_y+2 \omega}+\sqrt{2q_y-2 \omega}+\sqrt{3q_y^2 + 2 \omega}+\sqrt{3q_y^2 - 2 \omega}+\sqrt{2 \omega - q_y^2}}{4\pi}\bigg]^2+\frac{[\sqrt{2 \omega+2q_y}-\sqrt{2q_y-2 \omega}+\sqrt{2 \omega+q_y^2}]^2}{16\pi^2 } } \nonumber\\ d&=&\frac{\sqrt{2q_y+2 \omega}-\sqrt{2q_y-2 \omega}-\sqrt{q_y^2 - 2 \omega}+\sqrt { q_y^2 + 2 \omega}}{\bigg[\frac{1}{\lambda^2 } q_y^2+\frac{\sqrt{2q_y+2 \omega}+\sqrt{2q_y-2 \omega}+\sqrt{3q_y^2 + 2 \omega}+\sqrt{3q_y^2 - 2 \omega}}{4\pi}\bigg]^2+\frac{[\sqrt{2q_y+2 \omega}-\sqrt{2q_y-2 \omega}-\sqrt{q_y^2 - 2 \omega}+\sqrt{q_y^2 + 2 \omega}]^2}{16\pi^2}}\nonumber\\ \label{reg}\end{aligned}\ ] ] at small frequencies the dominant contribution to the imaginary part of the self - energy comes from the second integral between @xmath318 and @xmath319 , which scales as @xmath320
. the contributions from the other regions scale as @xmath321 and thus are negligible at low frequencies .
the correct prefactor can be obtained by expanding the numerator and the denominator of integrand @xmath113 for small frequencies @xmath64 and retaining only the largest terms , which gives @xmath322 one then integrates over @xmath39 to obtain , @xmath323 we note here that the self - energy is less singular compared to earlier works where umklapp scattering was not taken into account , in which case the self - energy scales as @xmath324.@xcite . ,
calculated numerically from the second line of eqn . , shown as a red solid line .
the black dashed line is the asymptotic result eqn . .
for comparison , the blue dash - dotted line indicates @xmath232 .
right : different contributions to @xmath325 .
the red line shows the dominant contribution , denoted by @xmath113 in the main text .
black dashed line : asymptotic result eqn . .
the other three curves show the sub - leading contributions @xmath326 and @xmath327 , which scale as @xmath321 . ]
so far we havent addressed an important issue : what happens if we include the renormalization of the boson - fermion vertex ?
in the two - patch theory originally considered by altschuler et al .
@xcite , the one - loop correction to this vertex was found to be logarithmically singular .
however in appendix [ appver ] , we show that the full four - patch theory does not have this singularity .
based on our discussion so far , we see that the additional umklapp terms considered in our work play a very crucial role at the critical point . in the absence of these contributions ,
the self - energy was more singular than what we have found here .
moreover , the vertex correction was also found to be singular .
however , here we have shown that the singular behavior is washed out when we include the additional scattering contributions . in the tpsc approach we imposed two - particle self - consistency in the form of a sum - rule that is similar to the spherical model . in the present field - theory approach , this amounts to imposing @xmath328 where the expectation value is taken with respect to the fermions and bosons .
we have argued that the @xmath3 scaling does not come from the self - consistency condition . to confirm this result ,
in this subsection we obtain the scaling of the quartic term in the boson lagrangian . integrating out the fermions
, the polarization operator @xmath329 appears in the quadratic term of the boson lagrangian . emphasizing the scaling
only , this term is symbolically written as @xmath330 this is the most relevant quadratic term . integrating out the large wave number modes for @xmath331 and rescaling @xmath225 and @xmath64 such that @xmath332 , @xmath333 with @xmath334 , returns the new cutoff @xmath335 to its original value @xmath336 .
invariance of the quadratic term written in terms of the prime variables then imposes that @xmath337 .
the effect of this tree level scaling on the quartic term is that @xmath338 this in turn means that @xmath339 scales to zero and is thus irrelevant .
we have argued that for bare interaction strengths @xmath30 in the intermediate coupling range , commensurate sdw fluctuations at @xmath0 and band parameters similar to those of electron - doped cuprates , the antiferromagnetic quantum critical point naturally occurs close to the filling where the fermi surface points joined by @xmath0 are nearly tangent to each other .
as long as the temperature or frequency are not too low , the limiting case of tangent fermi surfaces describes the physics . in this pseudo - nesting situation ,
the fermi liquid behavior breaks down .
quasiparticles still exist but the self - energy and spin susceptibility , for example , are different from those predicted by fermi liquid theory .
we considered this problem at zero temperature , or for frequencies larger than temperature , using a field - theoretical model of gapless collective bosonic modes ( sdw fluctuations ) interacting with fermions .
the imaginary part of the retarded fermionic self - energy close to the hot spots scales as @xmath9 .
this is less singular than earlier predictions of the form @xmath10 .
the difference arises from the effects of umklapp terms that were not included in previous studies . at finite temperature ,
we have used tpsc to study this problem and have obtained numerical results for the one - band hubbard model with band parameters and interaction strength appropriate for electron - doped cuprates . neglecting logarithmic corrections , we found analytically and numerically that the correlation length @xmath1 scales like @xmath2 , namely @xmath3 instead of the naive @xmath4 .
the static spin susceptibility @xmath5 scales like @xmath6 , and the correction @xmath8 to the korringa nmr relaxation rate is subdominant .
nmr experiments are difficult in electron - doped cuprates .
we also found that the imaginary part of the self - energy at the hot spot scales like @xmath7 .
the latter result and the @xmath9 frequency dependence of the self - energy should be experimentally verifiable with angle - resolved photoemission spectroscopy ( arpes ) in electron - doped cuprates .
recent transport measurements in these compounds @xcite have found a @xmath7 behavior of the resistivity above the quantum critical point at the end of the overdoped side of the superconducting dome .
while there may be a relation with the above result if antiferromagnetic fluctuations disappear at the same time , one must also be careful not to equate scattering rate with resistivity because in a simple picture it is the inverse of the scattering rates that are averaged over the fermi surface .
this suggests that in the resistivity , fermi liquid behavior of the cold spots should short - circuit the non - fermi liquid behavior of the hot spots .
@xcite we thank erez berg , andrey chubukov , lev ioffe and max metlitski for useful discussions .
this research was supported by the national science foundation under grant dmr-1103860 ( d.c .
, s.s . ) , by the austrian science fund ( fwf)-erwin schrdinger fellowship j 3077-n16 ( m.p . ) , by a muri grant from afosr ( s.s . ) and by nserc , the tier i canada research chair program ( a .- m .
m.s.t is grateful to the harvard physics department , cifar , and the center for materials theory at rutgers university for support during the writing of this work .
partial support was also provided by the mit - harvard center for ultracold atoms .
computer intensive calculations were performed on computers provided by cfi , mels , calcul qubec and compute canada .
[ appver ] in this appendix , we compute the 1-loop correction to the boson - fermion vertex , which is defined as , @xmath340 where we are working again with imaginary frequencies . the expression for the diagram in fig . [ bf ] can be written as , @xmath341 we now use the identity @xmath342 twice to simplify the above expression .
then on defining @xmath343 , we have , @xmath344\nonumber\\ \bigg[\frac{1}{-i ( l_\tau+p_\tau)-(l_x+p_x)+(l_y+p_y)^2}\bigg]\bigg[\frac{1}{{\bf l}^2+r+[\pi_0({\bf l})+\pi_0(-{\bf l})+\pi_0({\bf \tilde{l}})+\pi_0(-{\bf \tilde{l}})]}\bigg].\end{aligned}\ ] ] let us now evaluate this for zero external momenta @xmath345 at the critical point and check for singularities .
the expression reduces to , @xmath346\bigg[\frac{1}{-i l_\tau - l_x+l_y^2}\bigg]\nonumber\\\bigg[\frac{1}{{\bf l}^2+\pi_0({\bf l},l_\tau)+\pi_0(-{\bf l},l_\tau)+\pi_0({\bf \tilde{l}},l_\tau)+\pi_0(-{\bf \tilde{l}},l_\tau)}\bigg ] .
\label{renver}\end{aligned}\ ] ] the above integrand has a very complex structure .
let us therefore analyze the ( non-)singular nature of this diagram by power counting .
one needs to be careful as the @xmath347propagator has many combinations of powers of the internal momenta .
we begin by rescaling the variables as , @xmath348 eqn .
[ renver ] then takes the form , @xmath349\bigg[\frac{1}{-i l_\tau'-l_x'+1 } \bigg ] \nonumber \\ & & \times \frac{1}{l^2+[lf(l_x',l_\tau')+lf(-l_x',l_\tau')+\sqrt{l}g_+(l , l_x',l_\tau')+\sqrt{l}g_-(l , l_x',l_\tau')]/2\pi } , \\ f(l_x',l_\tau')&=&\sqrt{(1/2-l_x')+\sqrt{(1/2-l_x')^2 + l_\tau'^2 } } , \\
g_\pm((l , l_x',l_\tau')&=&\sqrt{(l^3l_x'^2/2\pm1)+\sqrt{(l^3l_x'^2/2\pm1)^2 + l^2l_\tau'^2}}\end{aligned}\ ] ] all we need to do now is to check whether this expression ( which is so far exact ) is singular in the ir and uv . in the ir
, we can ignore the momentum dependence of the fermionic green s functions compared to @xmath350 in the denominator .
moreover , @xmath351 and @xmath352 in this small momentum limit .
therefore , the above expression reduces to , @xmath353 where @xmath354 is a small cutoff . but the above expression is convergent , so that there are no ir singularities .
let us now check for uv singularities .
we proceed by introducing a characteristic lower cutoff @xmath336 which is large , but finite , such that the integration runs from @xmath336 to @xmath355 .
then , in the limit of these large momenta , we have , @xmath356\bigg[\frac{1}{-i l_\tau'-l_x ' } \bigg]\frac{1}{l^2+[l\sqrt{\sqrt{l_x'^2 + l_\tau'^2}+l_x'}+2l^2l_x']/2\pi } , \nonumber \\\end{aligned}\ ] ] where we have ignored the contribution from @xmath357 compared to @xmath358 and @xmath359 compared to @xmath360 .
we first want to do the integral over @xmath361 and @xmath362 .
it is more convenient to change to polar coordinates , @xmath363 .
however , we estimate @xmath364 and eliminate the @xmath220 dependence , which does not give rise to any singularities . then , @xmath365/2\pi } .\ ] ] we also notice that @xmath366 in the uv . therefore , ignoring the @xmath367 term , we obtain , @xmath368 both the @xmath369 and @xmath370 integrals can be performed easily to verify that @xmath371 is convergent in the uv
. therefore , the 1-loop vertex correction is non - singular both in the uv and in the ir . | many correlated materials display a quantum critical point between a paramagnetic and a spin - density wave ( sdw ) state .
the sdw wave vector connects points , so - called hot spots , on opposite sides of the fermi surface .
the fermi velocities at these pairs of points are in general not parallel . here
we consider the case where pairs of hot spots coalesce , and the wave vector @xmath0 of the sdw connects hot spots with parallel fermi velocities . using the specific example of electron - doped cuprates , we first show that kanamori screening and generic features of the lindhard function make this case experimentally relevant .
the temperature dependence of the correlation length , the spin susceptibility and the self - energy at the hot spots are found using the two - particle - self - consistent theory and specific numerical examples worked out for band and interaction parameters characteristic of the electron - doped cuprates .
while the curvature of the fermi surface at the hot spots leads to deviations from perfect nesting , the pseudo - nesting conditions lead to drastic modifications of the temperature dependence of these physical observables : neglecting logarithmic corrections , the correlation length @xmath1 scales like @xmath2 , namely @xmath3 instead of the naive @xmath4 , the @xmath0 static spin susceptibility @xmath5 like @xmath6 , and the imaginary part of the self - energy at the hot spots like @xmath7 . the correction @xmath8 to the korringa nmr relaxation rate is subdominant .
we also consider this problem at zero temperature , or for frequencies larger than temperature , using a field - theoretical model of gapless collective bosonic modes ( sdw fluctuations ) interacting with fermions .
the imaginary part of the retarded fermionic self - energy close to the hot spots scales as @xmath9 .
this is less singular than earlier predictions of the form @xmath10 .
the difference arises from the effects of umklapp terms that were not included in previous studies . |
for any integer @xmath3 we note @xmath4 the symmetric group on @xmath3 elements and @xmath5 an integer partition of @xmath6 with @xmath7 parts sorted in decreasing order .
if @xmath8 is the number of parts of @xmath9 that are equal to @xmath10 , then we may write @xmath9 as @xmath11 $ ] and define @xmath12 and @xmath13 .
a partition @xmath14 is usually represented as a young diagram of @xmath15 boxes arranged in @xmath16 lines so that the @xmath10-th line contains @xmath17 boxes . given a box @xmath18 in the diagram of @xmath14 , let @xmath19 be the number of boxes to the north , south , east , west of @xmath18 respectively .
these statistics are called * co - leglength , leglength , armlength , co - armlength * respectively .
we note for some parameter @xmath0 : @xmath20 moreover , @xmath21 is the conjugate of partition @xmath9 , for two integer partitions @xmath9 and @xmath22 , we denote @xmath23 if for all @xmath24 , @xmath25 and we denote @xmath26 the quantity @xmath27 let @xmath28 be the ring of symmetric functions
. denote @xmath29 the monomial symmetric function indexed by @xmath14 on indeterminate @xmath30 , @xmath31 and @xmath32 the power sum and schur symmetric functions respectively . whenever the indeterminate is not relevant we shall simply write @xmath33 , @xmath34 and @xmath35
. let @xmath36 be the scalar product on @xmath28 such that the power sum symmetric functions verify @xmath37 where @xmath38 is the kronecker delta .
+ the schur symmetric functions @xmath35 are characterized by the fact that they form an orthogonal basis of @xmath28 for @xmath36 ( they form even an orthonormal basis ) and the transition matrix between schur and monomial symmetric functions is upper unitriangular . as shown by schur , the @xmath35 are of critical interest in the representation theory of the symmetric group .
+ the * zonal polynomials * @xmath39 constitute another important basis of @xmath28 directly linked with the theory of the * zonal spherical functions*. zonal polynomials verify the same properties as the @xmath35 if the scalar product is replaced by @xmath40 with @xmath41 .
+ more generaly , using an additional parameter @xmath0 , henry jack @xcite introduced the basis of * jack symmetric functions * @xmath42 , that can be characterized as the set of symmetric functions verifying : * the @xmath42 are orthogonal for the alternative scalar product @xmath43 that verifies : @xmath44 * the transition matrix between the @xmath42 and the monomial symmetric functions is upper triangular and the coefficient in @xmath33 of the expansion of @xmath42 in the monomial basis is equal to @xmath45 .
formally it means that the @xmath42 may be expressed with the help of some scalar coefficients @xmath46 as : @xmath47 according to the above definition , @xmath48 is the normalized schur symmetric function @xmath49 and @xmath50 is the zonal polynomial @xmath39 .
in @xcite goulden and jackson studied the series @xmath51 more precisely , they investigate the * connection series * i.e. the coefficients @xmath52 in the power sum expansion of @xmath53 defined by @xmath54 they conjecture that the @xmath52 are polynomials in @xmath55 with positive integer coefficients and of degree at most @xmath56 .
they also conjecture that the coefficients @xmath52 can be written in the form @xmath57 where the summation is over a subset of matchings ( fixed point free involutions in @xmath58 ) and @xmath59 is a combinatorial statistic correlated with the `` nonbipartiteness '' of matchings .
they show the conjecture in the cases @xmath60 $ ] and @xmath61 $ ] . +
as shown in section [ sec : mot ] , aside their importance in the study of jack symmetric functions ( and therefore schur and zonal polynomials ) , there are several motivations to study this kind of coefficients .
they provide a general approach to the study of classical connection coefficients in the group algebra of the symmetric group .
these coefficients are also strongly related to the theory of jack characters that generalize both the irreducible characters of the symmetric group and the zonal spherical functions .
finally ( see e.g. @xcite ) , they are of notable interest to statisticians as they arise in the study of asymmetric random walks on the symmetric group .
+ in this paper we develop further this unified framework to study connection coefficients .
we define the numbers @xmath62 for integer @xmath18 greater or equal to @xmath63 and @xmath64 for @xmath65 .
@xmath66\sum_{\beta \vdash n}\frac{1}{\langle j^{{\alpha}}_\beta , j^{{\alpha}}_\beta\rangle_{{\alpha}}}\prod_{i}j_\beta^\alpha(x^i)\ ] ] we use the terminology of * jack connection coefficients * for these numbers .
notably we focus on the case when most of the @xmath67 are equal to @xmath68 $ ] .
we denote @xmath69 ( with @xmath70 appearing @xmath71 times , @xmath72 ) , i.e. : @xmath73 \sum_{\beta \vdash n}\frac{1}{\langle j^{{\alpha}}_\beta , j^{{\alpha}}_\beta\rangle_{{\alpha}}}\prod_{i}j_\beta^\alpha(x^i)\ ] ] the coefficients @xmath74 are of particular interest . as shown by hanlon in @xcite
, they give the probability that a random walk of length @xmath71 on @xmath4 which begins at the identity permutation ends at a permutation of type @xmath14 .
the transition probabilities for this random walk are such that only moves from a permutation @xmath75 to @xmath76 where @xmath77 is a transposition are possible .
the probability of moving from a permutation of cycle length @xmath78 to a permutation of cycle length @xmath79 is @xmath0 times greater than moving to a permutation of cycle length @xmath80 .
additionally , as shown in section [ sec : jcclbo ] , the @xmath74 give the coefficients in the power sum expansion of the symmetric function obtained by applying @xmath71 times the laplace beltrami operator to @xmath81 .
the motivations for choosing the definition of equation ( [ eq : df ] ) for jack connection coefficients are detailed in section [ sec : cccadca ] , equations ( [ eq : lemsymjc ] ) and ( [ eq : lemdcajc ] ) .
denote @xmath82 . in section [ sec : pt
] , we show the following main results .
[ thm : zero ] let @xmath74 be defined as above .
then for any integer partition @xmath9 we have @xmath83 for @xmath84 .
[ thm : pol ] for @xmath85 , @xmath74 can be written in the form @xmath86 where the @xmath87 are integers that verify @xmath88 as a consequence of theorem [ thm : pol ] we have the following corollary : let @xmath9 be an integer partition of @xmath3 and @xmath71 an integer greater than @xmath89 , then @xmath90 is a polynomial in @xmath91 with integer coefficients of degree at most @xmath92 .
we also show : [ thm : polpos ] let @xmath9 be an integer partition of @xmath3 and @xmath71 an integer greater than @xmath89 , then @xmath90 is a polynomial in @xmath93 with * non negative * integer coefficients of degree at most @xmath92 .
[ thm : denes ] let @xmath74 be defined as above . in the limit case
@xmath94 , we have @xmath95 in the specific case @xmath96 , equation [ eq : thm ] reads @xmath97 we view this later formula as a generalization of the classical formula attributed to dnes _ @xcite _ for the number of minimal factorizations of a long cycle in the symmetric group into a product of transpositions .
the paper is organized as follows . in section [ sec : mot ] we focus on the motivations for studying jack connection coefficients as defined in equation ( [ eq : df ] ) .
we also provide some background results about classical connections coefficients .
section [ sec : pt ] provides the proofs of theorems [ thm : zero ] , [ thm : pol ] , [ thm : polpos ] and [ thm : denes ] .
section [ sec : fr ] gives additional results based on the development in the previous section .
for any integer partition @xmath9 of @xmath3 let @xmath98 be the * conjugacy class * of @xmath4 containing the permutations of cycle type @xmath14 .
the cardinality of the conjugacy classes is given by @xmath99 .
additionally , @xmath100 is the * hyperoctahedral group * ( i.e. the centralizer of @xmath101 in @xmath58 ) . as shown in e.g. (
* vii.2 ) the * double cosets * of @xmath100 in @xmath58 are also indexed by integer partitions of @xmath3 .
we denote by @xmath102 the double coset indexed by @xmath103 consisting of all the permutations @xmath104 of @xmath58 such that @xmath105 has cycle type @xmath106 .
we have @xmath107 and @xmath108 .
the link between schur ( resp .
zonal ) polynomials and irreducible characters of the symmetric group ( resp .
zonal spherical functions ) is given by the decomposition of the @xmath35 ( resp .
@xmath39 ) in the power sum basis @xmath109 where @xmath110 is the value of the irreducible character of the symmetric group @xmath111 indexed by integer partition @xmath9 at any element of @xmath112 and @xmath113 .
the value of the zonal spherical function indexed by @xmath14 of the gelfand pair @xmath114 at the elements of the double coset @xmath115 is given by @xmath116 .
using the fact that both schur and zonal symmetric functions are special cases of jack symmetric functions , it is natural to focus on a more general form of equations ( [ eq : characters ] ) and ( [ eq : spherical ] ) .
formally , let @xmath117 denote the coefficient of @xmath118 in the expansion of @xmath42 in the power sum basis : @xmath119 according to equations ( [ eq : characters ] ) and ( [ eq : spherical ] ) , up to a normalization factor , the @xmath117 s coincide with the irreducible characters of the symmetric group and the zonal spherical functions in the cases @xmath120 and @xmath2 . in the general case doga and
fray @xcite named them * jack characters*. + the coefficients in the power sum expansion of jack symmetric functions have received significant attention over the past decades . as an example , hanlon conjectured a first combinatorial interpretation for them in @xcite in terms of digraphs .
stanley proved various results for these coefficients in @xcite .
for instance for simple values of @xmath22 we have the following formulas : @xmath121}^{\lambda}(\alpha ) = 1,\\ & \theta_{[1^{n-2}2 ^ 1]}^{\lambda}(\alpha ) = \alpha n({\lambda } ' ) - n({\lambda}),\\ & \theta_{(n)}^{\lambda}(\alpha ) = \prod_{s \in { \lambda}\setminus\{(1,1)\}}({\alpha}a'(s)-l'(s)).\end{aligned}\ ] ] more recent works by lasalle ( see e.g. @xcite and @xcite ) reconsidered these coefficients as generalizations of irreducible characters of the symmetric group .
lasalle conjectured various polynomial properties in @xmath91 for the @xmath117 .
doga and fray ( @xcite ) proved them partially and showed that jack characters are polynomials in @xmath91 with rational coefficients . together with p. niady @xcite they conjectured an expression involving a measure of `` non - orientability '' of locally orientable hypermaps ( an expression also introduced in @xcite ) .
doga and fray also proved in @xcite that the @xmath122 introduced by goulden and jackson are polynomials in @xmath55 with rational coefficients . by abuse of notation
let @xmath98 ( resp .
@xmath102 ) also represent the formal sum of its elements in the group algebra @xmath123 ( resp .
@xmath124 ) . then @xmath125 ( resp .
@xmath126 ) forms a basis of the * class algebra * ( resp . * double coset algebra * ,
i.e. the commutative subalgebra of @xmath124 identified as the hecke algebra of the gelfand pair @xmath127 ) . for @xmath128 ( @xmath129 )
, we define the * connection coefficients of the class algebra * @xmath130 and the * connection coefficients of the double coset algebra * @xmath131 by @xmath132\prod_{i \geq 2}c_{\lambda^i } , \;\;\;\;\ ; b^{{\lambda}^1}_{\lambda^2,\ldots,{\lambda}^s } = [ k_{{\lambda}^1}]\prod_{i \geq 2}k_{\lambda^i}.\ ] ] from a combinatorial point of view @xmath133 is the number of ways to write a given permutation @xmath134 of @xmath135 as the ordered product of @xmath136 permutations @xmath137 where @xmath138 is in @xmath139 . similarly , @xmath131 counts the number of ordered factorizations of a given element in @xmath140 into @xmath136 permutations of @xmath141 .
connections coefficients are also strongly related to characters and zonal spherical functions .
[ lem : sym ] the connection coefficients of the symmetric group and the double coset algebra are linked to the irreducible characters of the symmetric group and the zonal spherical functions through the equations @xmath142 using the fact that the elements @xmath143 form a complete set of orthogonal indempotents for the center of @xmath123 , it is easy to show ( see @xcite for the detailed computation ) that @xmath144 as a result we get @xmath132\prod_{i \geq 2}c_{\lambda^i } = \frac{1}{|s_n|}\sum_{\beta \vdash n}\frac{\prod_{i \geq 2}|c_{\lambda^i}|\chi^\beta_{{\lambda}^i}}{deg(\chi^\beta)^{s-2}}\chi^\beta_{{\lambda}^1}.\ ] ] but @xmath145 and @xmath99
. this provides the desired result . using similar techniques as in the proof of equation ( [ eq : lemsym ] )
, one can show ( see @xcite ) that @xmath146 extracting the coefficient in @xmath140 gives the desired formula . using the formalism of jack characters equations ( [ eq : lemsym ] ) and ( [ eq : lemdca ] ) read : @xmath147 equations ( [ eq : lemsymjc ] ) and ( [ eq : lemdcajc ] ) are very similar and the question of a more general approach to connection coefficients for any parameter @xmath0 appears to be very natural .
jack connection coefficients provide the desired approach . for the sake of simplicity ,
further in this paper we use the following formulation for jack connection coefficients : @xmath148 @xmath149}(\alpha)\right)^r.\ ] ] this formulation is justified as stanley showed in @xcite that the scalar product of the @xmath150 may be written as : @xmath151 except for special cases no closed formulas are known for the coefficients @xmath133 and @xmath131 . using an inductive argument bdard and
goupil @xcite first found a formula for @xmath152 in the case @xmath153 , which was later reproved by goulden and jackson @xcite via a bijection with a set of ordered rooted bicolored trees .
later , using characters of the symmetric group and a combinatorial development , goupil and schaeffer @xcite derived an expression for connection coefficients of arbitrary genus as a sum of positive terms ( see biane @xcite for a succinct algebraic derivation ; and poulalhon and schaeffer @xcite , and irving @xcite for further generalizations ) .
closed form formulas of the expansion of the generating series for the @xmath154 ( for general @xmath18 ) and @xmath155 in the monomial basis were provided by morales and vassilieva and vassilieva in @xcite , @xcite , @xcite and @xcite .
papers @xcite and @xcite use the topological interpretation of @xmath154 and @xmath155 in terms of unicellular locally orientable hypermaps and constellations of given vertex degree distribution .
jackson ( @xcite ) computed a general expression for the generating series of the @xmath156 in terms of some explicit polynomials .
this expression allows him to compute the following formula when @xmath157 and @xmath158 $ ] : @xmath159,\ldots,[1^{n-2}2 ^ 1 ] } = \frac{(s-1)!}{n!}n^{s-1}2^{n - s}[x^{s-1}]sh^{n-1}x.\ ] ] when @xmath160 , @xmath161,\ldots,[1^{n-2}2 ^ 1]}$ ] is equal to @xmath162 .
this is the famous result classically attributed to dnes ( @xcite ) .
shapiro , shapiro and vainhstein @xcite reproved the elegant closed form generating series in equation ( [ eq : j ] ) . + additionally we define @xmath163\log\left(1+\sum_{\rho , k}\frac{t^{|\rho|}u^k}{z_\rho k!}c^{\rho}_{\underbrace{[1^{n-2}2 ^ 1],\ldots,[1^{n-2}2 ^ 1]}_{k\ ; factors}}p_\rho\right)\ ] ] where the sums over @xmath70 is over all the non empty partitions .
goulden and jackson proved the following result ( see @xcite ) : @xmath164 this is the number of minimal transitive factorizations of a permutation in @xmath165 into a product of transpositions .
in this section we show various formulas for jack connection coefficients that allow us to prove theorems [ thm : zero ] , [ thm : pol ] , [ thm : polpos ] and [ thm : denes ] . as a first result
we show the relation between the values of jack connection coefficients in parameters @xmath0 and @xmath166 .
we have the following theorem : [ thm : inv ] let @xmath64 for @xmath65 and @xmath167 .
the jack connection coefficients for parameters @xmath91 et @xmath166 are linked through the relation @xmath168 let @xmath169 be the automorphism on @xmath28 defined by @xmath170 for any integer @xmath71 .
we have ( see @xcite ) @xmath171 as a consequence @xmath172 furthermore it is clear from the definition that @xmath173 finally , @xmath174 and the result follows . for indeterminate @xmath175 define the * laplace beltrami operator * by @xmath176 in this section we show that the coefficients in the power sum expansion of @xmath177 are jack connection coefficients .
we have the theorem : [ thm : ar ] let @xmath74 be the jack connection coefficients defined by equation ( [ eq : ar ] ) .
we have the following equality : @xmath178{d({\alpha})}^r(p_1^n),\ ] ] where @xmath179{d({\alpha})}^r(p_1^n)$ ] denotes the coefficient of @xmath34 in the power sum expansion of @xmath180 . to prove theorem [ thm : ar ] , we use two classical properties of jack symmetric functions .
first , stanley in @xcite showed that the scalar product of the @xmath181 may be written as @xmath182 as a consequence , we have the following identity for jack characters : @xmath183 equivalently , @xmath184 as a result we have the following lemma : the following relation between jack and power sum symmetric functions holds : [ lem : jp ] @xmath185 it is a direct consequence from equation ( [ eq : orth ] ) . summing over @xmath186 on both sides reads @xmath187 the formula is proved by setting @xmath188 $ ] . in this case @xmath189}^{\lambda}({\alpha } ) = 1 $ ] and the result follows .
secondly , it is well known that the jack symmetric functions are eigenfunctions of @xmath190 : @xmath191}^{\lambda}({\alpha})j_{\lambda}^{\alpha}.\ ] ] theorem [ thm : ar ] is a direct consequence of lemma [ lem : jp ] and equation ( [ eq : eigen ] ) : @xmath192}}(\alpha)\right)^r\theta^\beta_{{\lambda}}(\alpha)p_{\lambda}\\ & = \sum_{\beta \vdash n}\frac{1}{h_\beta(\alpha)h'_\beta(\alpha)}\left(\theta^\beta_{{[1^{n-2}2 ^ 1]}}(\alpha)\right)^rj_\beta^{\alpha}\\ & = \sum_{\beta \vdash n}\frac{1}{h_\beta(\alpha)h'_\beta(\alpha)}{d({\alpha})}^r(j_\beta^{\alpha})\\ & = { d({\alpha})}^r \left ( \sum_{\beta \vdash n}\frac{1}{h_\beta(\alpha)h'_\beta(\alpha)}j_\beta^{\alpha}\right)\\ & = { d({\alpha})}^r\left ( \frac{p_1^n}{{\alpha}^nn!}\right).\end{aligned}\ ] ] the second equality comes from the definition of @xmath193 as the coefficients in the power sum expansion of the jack symmetric functions , the third one is @xmath71 times the application of equation ( [ eq : eigen ] ) and the last one is the application of lemma [ lem : jp ] .
successive applications of the laplace beltrami operator to@xmath194 give the values of the @xmath74 s . for @xmath195 or
@xmath196 and @xmath103 we get @xmath197},\\ \nonumber { d({\alpha})}^2\left ( \frac{p_1^n}{{\alpha}^nn!}\right ) & = \frac{1}{2{\alpha}^{n-1}(n-2)!}p_{[1^n]}+ \frac{({\alpha}-1)}{{\alpha}^{n-1}(n-2)!}p_{[1^{n-2}2 ^ 1]}\\ & \phantom{lalal}+\frac{1}{4{\alpha}^{n-2}(n-4)!}p_{[1^{n-4}2 ^ 2]}+\frac{1}{{\alpha}^{n-2}(n-3)!}p_{[1^{n-3}3 ^ 1]}.\end{aligned}\ ] ] in order to show theorems [ thm : pol ] and [ thm : polpos ] we need an additional classical lemma : [ lem : dp ] operator @xmath190 can be expressed as @xmath198 where for @xmath103 : @xmath199 the following property holds : for any @xmath103 , the coefficients in the power sum expansion of @xmath200 , @xmath201 and @xmath202 are non negative integers . : * from equation ( [ eq : n ] ) , it is obvious that the quantity @xmath203 is a non negative integer so @xmath179n(p_{\lambda})$ ] is a non negative integer . *
the expression of @xmath201 is symmetric in @xmath10 and @xmath204 so the coefficient @xmath205 vanishes by adding the terms for indices @xmath206 and @xmath207 when @xmath208 .
when @xmath209 , assume @xmath210 .
then @xmath211 as @xmath212 is a non negative integer , the coefficients in the power sum expansion of @xmath201 are always non negative integers . * regarding operator @xmath213
the property is proved by noticing that for @xmath209 , @xmath214 is a non negative integer .
using theorem [ thm : ar ] we may rewrite jack connection coefficients @xmath215 as : @xmath216 as a result , @xmath217 is a polynomial in @xmath91 of degree at most @xmath71 with integer coefficients . for @xmath218
, there exist integers @xmath219 such that @xmath220 but from theorem [ thm : inv ] @xmath221 recall the definition @xmath222 . as a consequence to the previous equation @xmath223 equating the coefficients in the powers of @xmath0 in both sides of the equation finishes the proof of theorem [ thm : pol ] .
we get @xmath224 using the previous equation , as the summation parameter @xmath10 is less or equal to @xmath71 : @xmath225 finally the following expressions hold @xmath226 equations ( [ eq : aaa ] ) and ( [ eq : bbb ] ) prove theorem [ thm : pol ] . + the proof of theorem [ thm : polpos ] is a direct consequence of @xmath227 and the previous remarks in this section regarding the polynomial properties of @xmath215 in @xmath91 .
we study further the combinatorial properties of the laplace - beltrami operator to prove theorems [ thm : zero ] an [ thm : denes ] .
+ the power sum expansions of the actions of operators @xmath228 , @xmath229 and @xmath213 on a given @xmath34 have the following properties : * @xmath230n(p_{\lambda } ) = 0 $ ] * @xmath231u(p_{\lambda } ) = 0 $ ] * @xmath232s(p_{\lambda } ) = 0 $ ] as a direct consequence , we have the lemma : [ lem : u ] for any integer partition @xmath9 of @xmath3 with @xmath233 parts , the coefficient of @xmath34 of @xmath180 verifies @xmath234{d({\alpha})}^r(p_1^n ) = { \alpha}^r[p_{\lambda}]u^r(p_1^n)\ ] ] furthermore , if @xmath235 , then @xmath179{d({\alpha})}^r(p_1^n ) = 0 $ ] .
the final remark of lemma [ lem : u ] proves theorem [ thm : zero ] . in order to prove theorem [ thm : denes ] ,
it is sufficient to study operator @xmath229 . looking more precisely at the action of @xmath229 on a given power sum symmetric function @xmath34
, one can see that the terms @xmath236u(p_{\lambda})$ ] are non zero only when @xmath22 is of the form @xmath237 , i.e. when @xmath22 is obtained by combining ( adding ) two parts of @xmath9
. the contributions of the combinations of @xmath17 and @xmath238 to @xmath236u(p_{\lambda})$ ] are additive and equal to @xmath239 .
the sum of all the contributions @xmath240 for all the possible @xmath22 combining two parts of @xmath9 gives @xmath201 .
as an example , there are @xmath241 ways of combining two parts of the partition @xmath242 $ ] .
any of these combinations yields the partition @xmath243 $ ] and in this particular case @xmath244 .
we have @xmath245}.\ ] ] further , there are @xmath246 ways of combining two @xmath63 s in @xmath243 $ ] to get the partition @xmath247 $ ] ( @xmath248 ) and @xmath249 ways of combining one @xmath63 and the part @xmath196 to get the partition @xmath250 $ ] ( @xmath251 ) .
it follows that @xmath252 } ) = \frac{(n-2)(n-3)}{2}p_{[1^{n-4}2 ^ 2 ] } + 2(n-2)p_{[1^{n-3}3 ^ 1]}.\ ] ] iterating @xmath253 times operator @xmath229 on @xmath81 yields a non zero coefficient only in @xmath254 ( equal to @xmath255}$ ] ) .
this coefficient is the sum of the contributions of all the possible successive combinations of the parts of @xmath242 $ ] .
figure [ fig : comb ] shows how these various ways of combination contribute to the final coefficient in @xmath254 for @xmath256 ( @xmath257 } ) = 3000p_{[5 ^ 1]}$ ] ) .
[ lem : recu ] let @xmath10 and @xmath204 be two integers greater or equal to 1 .
the following recursion holds for the coefficients in the power sum expansion of @xmath258 : @xmath259u^{i+j-2}&(p_{[1^{i+j}]})=\\ & \frac{1}{2^{\delta_{ij}}}\binom{i+j}{i}\binom{i+j-2}{i-1}\left ( [ p_{i}]u^{i-1}(p_{[1^{i}]})\right ) \left ( [ p_{j}]u^{j-1}(p_{[1^{j}]})\right ) \end{aligned}\ ] ] as noticed in the previous paragraph , the application of operator @xmath260 on @xmath261}$ ] provides non zero coefficients in the @xmath34 s where partition @xmath9 is obtained by @xmath71 successive combinations of parts starting with @xmath262 $ ] . at each iteration
two parts are combined and the resulting coefficient depends only on the chosen parts ( and not the whole partition at this step ) . as a result
the coefficient obtained by @xmath263 combinations of parts to get partition @xmath206 from @xmath262 $ ] is equal to the product of the coefficients obtained by @xmath264 combinations of parts to get @xmath265 from @xmath266 $ ] for @xmath267 multiplied by : * the number of ways of selecting the initial @xmath63-parts that will be combined to get @xmath268 ( there are @xmath269 such ways ) * the number of ways of selecting the @xmath270 ( among @xmath263 ) iterations used to combine parts in order to get @xmath268 ( there are @xmath271 such ways , all yielding obviously the same contribution ) . when @xmath209 this total number is divided by @xmath196 not to count twice identical scenarios .
combining the formula of lemma [ lem : recu ] with equations ( [ eq : ard ] ) and ( [ eq : defu ] ) implies the following recursive formula on the number @xmath272 : @xmath273 which simplifies as @xmath274 denote @xmath275 . as shown in @xcite , one can solve the above recursion by considering the generating function @xmath276 using equation ( [ eq : reca ] ) , one gets @xmath277 which gives @xmath278 using the initial conditions @xmath279 and @xmath280 , one obtains the functional equation @xmath281 and necessarily @xmath282 the proof of theorem [ thm : denes ] is completed by noticing that lemma [ lem : recu ] can be easily generalized as @xmath283u^{|{\lambda}|-\ell({\lambda})}(p_{[1^{|{\lambda}| } ] } ) = \binom{|{\lambda}|}{{\lambda}}\binom{|{\lambda}|-\ell({\lambda})}{{\lambda}-1}\prod_{i}\left ( [ p_{{\lambda}_i}]u^{{\lambda}_i-1}(p_{[1^{{\lambda}_i}]})\right ) \ ] ] where @xmath284 is the partition @xmath285 of @xmath286
we have : @xmath287u^{|{\lambda}|-\ell({\lambda})}\left ( \frac{p_{[1^{|{\lambda}|}]}}{{\alpha}^{|{\lambda}|}|{\lambda}|!}\right ) & = \frac{(|{\lambda}|-\ell({\lambda}))!}{\prod_i{({\lambda}_i-1)!}}\prod_{i}\left ( [ p_{{\lambda}_i}]u^{{\lambda}_i-1}\left ( \frac{p_{[1^{{\lambda}_i}]}}{{\alpha}^{{\lambda}_i}{\lambda}_i!}\right)\right ) \\ \nonumber aut_{\lambda}a^{|{\lambda}|-\ell({\lambda})}_{{\lambda}}({\alpha } ) & = \frac{(|{\lambda}|-\ell({\lambda}))!}{\prod_i{({\lambda}_i-1)!}}\prod_{i}a_{({\lambda}_i)}^{{\lambda}_i-1}({\alpha})\\ aut_{\lambda}a^{|{\lambda}|-\ell({\lambda})}_{{\lambda}}({\alpha } ) & = \frac{(|{\lambda}|-\ell({\lambda}))!}{\prod_i{({\lambda}_i-1)!}}\prod_{i}\frac{1}{{\alpha}{\lambda}_i}{\lambda}_i^{{\lambda}_i-2}\end{aligned}\ ] ]
in this section we look at a less refined definition of jack connection coefficients and focus only on the number of parts of the @xmath67 .
we define @xmath288 stanley @xcite evaluated jack symmetric functions at @xmath289 ( @xmath290 @xmath63 s ) and showed the classical formula @xmath291 but as @xmath292 we have the following formula : @xmath293 being true for any integer @xmath290 , this polynomial identity is actually true for any scalar @xmath294 : @xmath295 as a result we get the following explicit general formulation : [ thm : genjackson ] let @xmath296 be the coefficients defined in equation ( [ eq : pi ] ) and @xmath297 ( @xmath65 ) @xmath18 scalar indeterminate .
we have the following formula for any integer @xmath298 : @xmath299 in the case @xmath300 jackson used this formula to compute explicitly @xmath154 with @xmath301,[1^{n-2},2 ^ 1]\}$ ] for @xmath302 . in this particular case @xmath303}^{\lambda}(1 ) & = \delta_{{\lambda } , [ 1^{n - t},t^1]}(-1)^{n - t-1}(n - t-1)!(t-1)!,\\ \theta_{[1^{n-2}2 ^ 1]}^{[1^{n - t},t^1 ] } & = -\frac{n(n-2t-1)}{2},\\ h_{[1^{n - t},t^1]}(1 ) & = ( n-1)(t-1)!(n - t-1)!.\end{aligned}\ ] ] these very simple expressions allow a computation of the close form formula ( [ eq : j ] ) . when @xmath304 the formula for the jack connection coefficients does not simplify .
consider the exponential generating function @xmath305 using theorem [ thm : ar ] , one immediately gets @xmath306(x ) = \left[\exp\left({td({\alpha})}\right)\left ( \frac{p_1^n}{{\alpha}^nn ! }
\right ) \right](x)\end{aligned}\ ] ] as a consequence we have the following differential equation : @xmath307 this equation is a general form of the differential equations studied by goulden and jackson for the generating series of the number of minimal transitive factorizations of a permutation into transpositions in @xcite . using a similar development as in section [ sec : pt ]
one can compute the coefficient @xmath308}({\alpha})$ ] .
we have @xmath309}({\alpha } ) = \frac{1}{{\alpha}^{\ell({\lambda})}z_{\lambda}}[p_\mu ] d({\alpha})(p_{\lambda}).\ ] ] as a consequence @xmath310}({\alpha } ) = ( { \alpha}-1){\alpha}^{-\ell({\lambda})}z_{\lambda}^{-1}n({\lambda}').\ ] ] if @xmath237 then @xmath309}({\alpha } ) = \frac{{\lambda}_i{\lambda}_j}{2^{\delta_{i , j}}}{\alpha}^{-\ell({\lambda})+1}z_{\lambda}^{-1}.\ ] ] if @xmath311 then @xmath309}({\alpha } ) = \frac{{\lambda}_i}{2^{\delta_{{\lambda}_i,2k}}}{\alpha}^{-\ell({\lambda})}z_{\lambda}^{-1}.\ ] ] unicellular hypermaps embedded in orientable ( resp .
locally orientable ) surfaces are counted by @xmath312 ( resp .
@xmath313 ) . using the equation above and lemma [ lem : sym ] ,
we get : @xmath314 } = 0 , \;\ ;
\frac{1}{|b_n|}b^{(n)}_{(n),[1^{n-2}2 ^ 1 ] } = \binom{n}{2},\\ & c^{(n)}_{(n - i , i),[1^{n-2}2 ^ 1 ] } = n/2^{\delta_{n,2i}},\;\;\frac{1}{|b_n|}b^{(n)}_{(n - i , i),[1^{n-2}2 ^ 1 ] } = n/2^{\delta_{n,2i}}\end{aligned}\ ] ] 1 f. bdard and a. goupil . the poset of conjugacy classes and decomposition of products in the symmetric group , , 35(2):152160 , 1992 .
+ p. biane .
nombre de factorisations dun grand cycle , , 51:4 , 2004 .
+ g. chapuy and c. stump .
counting factorizations of coxeter elements into products of reflections , , 2012 .
+ j. dnes .
the representation of a permutation as the product of a minimal number of transpositions and its connection with the theory of graphs , , 4:6370 , 1959 .
+ m. doga and v. fray .
on kerov polynomials for jack characters , as:539550 , 2013 .
+ m. doga and v. fray .
gaussian fluctuations of young diagrams and structure constants of jack characters , , 2014 .
+ m. doga , v. fray and p. niady .
jack polynomials and orientability generating series of maps , , 2013 .
+ i. p. goulden and d. m. jackson . the combinatorial relationship between trees , cacti and certain connection coefficients for the symmetric group , , 13:357365 , 1992
. + i. p. goulden and d. m. jackson .
connection coefficients , matchings , maps and combinatorial conjectures for jack symmetric functions , , 348(3):873892 , 1996 .
+ i. p. goulden and d. m. jackson .
transitive factorizations into transpositions and holomorphic mappings on the sphere , , 125(1):5160 , 1997 .
+ a. goupil and g. schaeffer .
factoring n - cycles and counting maps of given genus , , 19:819834(16 ) , 1998 .
+ j. irving . on the number of factorizations of a full cycle , , 113(7):15491554 , 2006 .
+ p. j. hanlon .
jack symmetric functions and some combinatorial properties of young symmetrizers , , 47:3770 , 1988 .
+ p. j. hanlon . a markov chain on the symmetric group and jack symmetric functions , , 99:123140 , 1992 .
+ p. j. hanlon , r. p. stanley and j. r. stembridge .
some combinatorial aspects of the spectra of normally distributed random matrices , , 138:151174 , 1992
. + h. jack . a class of symmetric polynomials with a parameter , , 69:118 , 1970 .
+ d. m. jackson .
some combinatorial problems associated with products of conjugacy classes of the symmetric group , , 49(2):363369 , 1988 . + m. lassalle . a positivity conjecture for jack polynomials , , 15(4):661681 , 2008 .
+ m. lassalle .
jack polynomials and free cumulants , , 222(6):22272269 , 2009 .
+ i. macdonald .
symmetric functions and hall polynomials , , second edition , 1999 . + a. h. morales and e. a. vassilieva .
bijective enumeration of bicolored maps of given vertex degree distribution , , ak:661672 , 2009 .
+ a. h. morales and e. a. vassilieva .
bijective evaluation of the connection coefficients of the double coset algebra , , ao:681692 , 2011 .
+ d. poulalhon and g. schaeffer .
, , 254:433458 , 2000 . + g. schaeffer and e. a. vassilieva . a bijective proof of jackson s formula for the number of factorizations of a cycle , , 115(6):903924 , 2008 .
+ b. shapiro , m. shapiro and a. vainshtein . ramified covering of @xmath315 with one degenerate branching point and edge - ordered graphs , , 180:219227 , 1997 .
+ a. b. shukla . a short proof of cayley s tree formula , , 2009 .
+ r. p. stanley
. some combinatorial properties of jack symmetric functions , , 77:76115 , 1989 .
+ e. a. vassilieva .
explicit generating series for connection coefficients , ar:123134 , 2012 .
+ e. a. vassilieva .
long cycle factorizations : bijective computation in the general case , as:10771088 , 2013 | this article is devoted to the computation of jack connection coefficients , a generalization of the connection coefficients of two classical commutative subalgebras of the group algebra of the symmetric group : the class algebra and the double coset algebra .
the connection coefficients of these two algebraic structures are of significant interest in the study of schur and zonal polynomials as well as the irreducible characters of the symmetric group and the zonal spherical functions .
furthermore they play an important role in combinatorics as they give the number of factorizations of a permutation into a product of permutations with given cyclic properties and , in some cases , the number of hypermaps embedded in orientable and locally orientable surfaces with specified vertex degree distribution .
+ usually studied separately , these two families of coefficients share strong similar properties . first ( partially ) introduced by goulden and jackson in 1996 , jack connection coefficients provide a natural unified approach closely related to the theory of jack polynomials , a family of bases in the ring of symmetric functions indexed by a parameter @xmath0 that generalizes both schur ( case @xmath1 ) and zonal polynomials ( case @xmath2 ) .
jack connection coefficients are also directly linked to jack characters , a general view of the characters of the symmetric group and the zonal spherical functions .
goulden and jackson conjectured that these coefficients are polynomials in @xmath0 with nice combinatorial properties , the so - called _ matchings - jack conjecture_. + in this paper , we use the theory of jack symmetric functions and the laplace beltrami operator to show the polynomial properties of jack connection coefficients in some important cases .
we also provide explicit formulations including notably a generalization of a classical formula of denes for the number of minimal factorizations of a permutation into transpositions . |
an indirect exciton in a coupled quantum well structure ( cqw ) is a bound pair of an electron and a hole confined to spatially separated qws .
the small overlap of the electron and the hole wavefunctions increases the lifetime of indirect excitons by orders of magnitude compared to the lifetime of regular direct excitons .
in addition , indirect excitons are oriented dipoles with a built - in dipole moment @xmath0 , where @xmath1 is close to the separation between the qw centers .
the repulsive dipole - dipole interaction between indirect excitons allows them to screen in - plane disorder in the structure @xcite . as a result
, indirect excitons can travel over large distances before optically decaying @xcite .
the long lifetimes of indirect excitons also allow them to cool down to low temperatures , well below the temperature of quantum degeneracy @xmath2 @xcite ( in the studied cqw with the exciton mass @xmath3 and the spin degeneracy @xmath4 , @xmath5k for the density per spin state @xmath6 @xmath7 ) .
furthermore , the built - in dipole moment of an indirect exciton allows control of exciton transport by voltage @xcite and light @xcite .
the long lifetime , large transport distances , efficient cooling , and control of transport make indirect excitons a model system for studying transport of cold bosons in condensed matter materials .
studies of indirect excitons have resulted in observations of patterns in the exciton emission , including the inner ring @xcite , external ring @xcite , localized bright spots @xcite , and macroscopically ordered exciton state @xcite ( ring shapes are specific to point excitation geometry ) . the inner ring has been explained in terms of exciton transport and cooling : laser excitation heats the exciton gas , excitons cool towards the lattice temperature as they travel away from the excitation spot , the cooling results in an increase in the occupation of the low - energy optically active exciton states @xcite and , in turn , the appearance of an emission ring around the excitation spot @xcite . in this paper
, we report on the study of the excitation energy dependence of the inner ring in the exciton emission pattern .
we find that by adjusting the excitation energy to the direct exciton resonance , we can reduce heating due to laser excitation , facilitating the realization of a cold and dense exciton gas .
the cqw structure used in this experiment contains two 8 nm gaas qws separated by a 4 nm al@xmath8ga@xmath9as barrier . the cqw diagram is given in fig .
the sample was grown by molecular - beam epitaxy .
the effective spacing between the electron and hole layers is given by @xmath10 nm @xcite , close to the nominal distance between the qw centers .
details on the cqw structure can be found in ref .
@xcite .
indirect excitons were photogenerated by a tunable ti : sapphire laser focused to a 2.8 @xmath11 m spot on the sample .
excitation energy @xmath12 was varied between 1.55 and 1.85 ev .
for this excitation below the al@xmath8ga@xmath9as barrier energy , no excess free carriers are photogenerated , and both the external ring and the localized bright spots are absent @xcite .
exciton density was controlled by the laser excitation power @xmath13 .
spatial resolution was 1.4 @xmath11 m .
the spectra were measured with a liquid nitrogen cooled charge - coupled device ( ccd ) placed after a double - grating spectrometer .
the measurements were performed at an applied gate voltage of 1.2 v , at which the indirect excitons are approximately 20 mev lower in energy than direct excitons in the structure .
the cryostat bath temperature was 1.6 k. figure 1b shows an intensity profile @xmath14 of the inner ring in the emission of indirect excitons .
the major characteristics of the inner ring are indicated : @xmath15 is the half width at half maximum ( hwhm ) , @xmath16 is the average maximum intensity , and @xmath17 is the depth of the inner ring .
the contrast @xmath18 of the inner ring gives a measure of heating of the exciton gas by laser excitation as detailed below .
figure 1c shows @xmath19-energy images of the exciton emission for varying @xmath13 .
a higher emission energy corresponds to a higher exciton density because of the repulsive dipole - dipole interaction between indirect excitons @xcite . for the considered densities ,
the emission linewidth is @xmath20 mev .
it corresponds to the exciton emission linewidth , which is determined by the homogeneous and inhomogeneous broadening and can be small @xcite . in comparison ,
the linewidth characteristic for the emission of an electron - hole plasma is close to the sum of the electron and hole fermi energies in the plasma and is considerably larger @xcite . the narrow emission line , which is characteristic for the inner ring reported in refs .
@xcite and in the present paper , indicates that in all these experiments the inner ring forms in a system of excitons rather than in electron - hole plasma @xcite .
is the half width at half maximum ( hwhm ) ; @xmath16 is the average maximum intensity ; @xmath17 is the depth of the ring .
( c ) @xmath19-energy images of the exciton emission for different excitation powers @xmath13
. excitation energy @xmath21 ev .
the position of the laser excitation spot is @xmath19 = 0.,width=321 ] figure 2a presents the spatially integrated emission intensity of indirect excitons @xmath22 as a function of excitation energy @xmath12 .
@xmath23 follows the absorption curve of the cqw , with resonances at the excitation energy equal to the heavy hole direct exciton ( hh dx ) energy @xmath24 ev and light hole direct exciton ( lh dx ) energy @xmath25 ev .
figures 2b - f present the excitation power dependence of the inner ring at these values of @xmath12 along with a higher nonresonant excitation energy @xmath26 ev . both the ring extension @xmath15 ( fig .
2b ) and contrast @xmath18 ( fig .
2c ) increase with increasing excitation density @xmath13 .
this behavior is understood as follows : at low exciton densities , indirect excitons are localized by the in - plane disorder potential in the structure and do not travel away from the excitation spot .
the size of the exciton cloud is close to the size of the excitation spot .
no inner ring is observed in this regime ( figs .
2d - f , blue curves ) . at high densities ,
excitons screen the disorder potential .
delocalized excitons travel away from the excitation spot and cool toward the lattice temperature during their travel .
this results in the enhancement of the occupation of low - energy optically active exciton states @xcite and , in turn , the appearance of the ring in the exciton emission pattern @xcite ( figs .
2d - f , orange and black curves ) . for @xmath13 = 50 @xmath11w .
( b ) hwhm @xmath15 and ( c ) contrast @xmath18 of the inner ring as a function of excitation power @xmath13 for @xmath27 1.570 ev ( black @xmath28 ) , 1.588 ev ( red @xmath29 ) and 1.676 ev ( green @xmath30 ) , indicated in ( a ) .
( d - f ) @xmath13 dependence of the indirect exciton emission profile for the above values of @xmath12 .
the laser excitation profile is shown by the dashed line in ( d).,width=321 ] the detailed excitation energy dependence of the integrated emission intensity @xmath22 , ring extension @xmath15 , and contrast @xmath18 is presented in figure 3 . increased absorption at the hh dx and lh dx resonances leads to a higher exciton population at @xmath31 and 1.588 ev , conveyed by the peaks in @xmath23 ( fig .
a higher exciton population leads to a broader emission pattern and corresponding peaks in @xmath32 ( fig .
3b ) . figure 3c shows the ring contrast @xmath18 as a function of excitation energy
. a strong reduction of the ring contrast is observed at low @xmath12 , indicating a strong reduction of the excitation - induced heating of the exciton gas .
the highest exciton density ( fig . 3a ) , the largest ring extension ( fig .
3b ) , and the smallest ring contrast ( fig .
3c ) are observed for the excitation tuned to the hh dx resonance . at @xmath12 below the hh dx resonance ,
the inner ring vanishes as a result of reduced absorption ( fig .
3a - c ) . , ( b ) extension @xmath15 , and ( c ) contrast @xmath18 of the inner ring as a function of excitation energy @xmath33 for excitation power @xmath34w ( red ) and @xmath35w ( black )
. the data for 10 and @xmath35w in ( a ) are normalized at @xmath36 ev.,width=321 ]
following the approach developed in @xcite , we solved a set of coupled equations to find the spatial profiles of the indirect exciton density , temperature and emission intensity . in the steady - state scenario relevant to the experiment , the indirect exciton density @xmath37 satisfies the following transport equation : @xmath38 - \gamma_{\rm opt } n_{\rm x } + \lambda = 0 . \label{eq1}\ ] ] the two terms in the square brackets of equation ( [ eq1 ] ) are due to the diffusion and drift currents of excitons , respectively .
the disorder - dependent diffusion coefficient @xmath39 $ ] accounts for the ability of dipole - orientated indirect excitons to screen the disorder potential . here , @xmath40 is the dipole - dipole interaction potential and @xmath41 mev is the amplitude of the disorder potential .
the exciton mobility is given by the generalized einstein relation , @xmath42 @xcite .
the optical decay rate @xmath43 incorporates the fact that only excitons in the lowest energy states may decay optically @xcite .
the generation rate of excitons , @xmath44 follows the gaussian profile of the laser where @xmath45 is the distance from the center of the excitation spot .
the transport equation ( [ eq1 ] ) is solved in 2d geometry with radial symmetry of the density assumed . for @xmath46 = 0.5 mev ( black ) and 30 mev ( red ) .
( d ) contrast of the inner ring @xmath18 as a function of @xmath47 calculated from the simulated exciton emission profiles.,width=321 ] the exciton temperature @xmath48 , which stands in the expressions for the diffusion coefficient , mobility , and decay rate is found by solving the following heat balance equation , @xmath49 here , @xmath50 is the rate of cooling of indirect excitons by a bath of bulk longitudinal acoustic phonons , which act to restore the exciton temperature to the lattice temperature . within the excitation spot ,
heating occurs due to the injection of high energy excitons , which then thermalize within the timescales much shorter than the cooling time associated with acoustic phonons @xcite .
the heating rate @xmath51 is characterized by the pump rate @xmath52 and the excess energy acquired by photoexcited excitons @xmath47 , which increases with photon energy .
expressions for @xmath50 , @xmath51 and @xmath53 and all other parameters are given in @xcite .
in previous studies , the heating / cooling rate which arises from optical decay of indirect excitons was found to be small in comparison to @xmath50 and @xmath51 @xcite .
it is therefore neglected in these calculations . from the numerical solution to equations ( [ eq1],[eq2 ] ) , one can extract the emission intensity @xmath54 , density @xmath55 , and temperature @xmath48 of indirect excitons , shown as functions of @xmath19 in figs .
4a , 4b , and 4c , respectively , for two different excess energies @xmath47 .
the contrast of the inner ring obtained from these calculations for various values of @xmath47 is presented in fig .
the model illustrates the sensitivity of the exciton temperature to the tuning of the laser .
exciton cooling via scattering with acoustic phonons becomes less effective against the heating supplied by excitation with increasing excitation energy .
since the intensity reduction in the center of the inner ring is due to a higher exciton temperature in the region of laser excitation , the contrast of the inner ring @xmath18 gives a measure of the exciton heating .
figure 3 shows that tuning the laser excitation to the hh dx resonance @xmath12 = 1.570 ev both maximizes the light absorption and minimizes the laser - induced heating of excitons , thus facilitating the realization of a cold and dense exciton gas .
there is a qualitative agreement between the observed and simulated shapes of emission profiles ( see fig .
2d - f and fig .
the simulated ring contrast @xmath18 increases with excess energy @xmath56 ( fig .
4d ) , and therefore with excitation energy , due to increasing exciton temperature in the excitation spot ( fig .
an increase of the ring contrast with excitation energy is in qualitative agreement with experimental data ( see fig . 3c and fig .
note however that the simulations are based on a simplified qualitative model , which includes only exciton relaxation via emission and absorption of acoustic phonons and only one energy band - the band of hh indirect excitons .
such model qualitatively explains the main observed characteristics of the inner ring , including the enhancement of the ring contrast with excitation energy , reported in this paper .
a quantitative comparison with the experiment requires a theory which takes into account exciton relaxation via emission and absorption of optical phonons as well as hh and lh bands of direct and indirect excitons .
the development of such theory forms the subject for a future work .
we studied the excitation energy dependence of the inner ring in the indirect exciton emission pattern .
such ring forms due to exciton transport and cooling .
we found that the excitation - induced heating of the exciton gas is suppressed at lower excitation energies , reaching a minimum at the excitation energy tuned to the heavy hole direct exciton resonance , below which the inner ring disappears due to the diminished light absorption in the structure .
we found that tuning the excitation laser to the heavy hole direct exciton energy both increases the density of indirect excitons and effectively suppresses the laser - induced heating of indirect excitons , thus facilitating the realization of a cold and dense exciton gas .
we thank nikolay gippius , aaron hammack , and leonidas mouchliadis for discussions .
support of this work by nsf and epsrc is gratefully acknowledged . | we report on the excitation energy dependence of the inner ring in the exciton emission pattern .
the contrast of the inner ring is found to decrease with lowering excitation energy .
excitation by light tuned to the direct exciton resonance is found to effectively suppress excitation - induced heating of indirect excitons and facilitate the realization of a cold and dense exciton gas .
the excitation energy dependence of the inner ring is explained in terms of exciton transport and cooling . |
among the various approximations existing in the literature to describe a diluted bose condensed gas at finite temperature , the generalized random phase approximation ( grpa ) has been the subject of several studies @xcite .
this approximation has attracted a special attention since it is the only one in the literature with two important properties : 1 ) in agreement with the hugenholtz - pines theorem @xcite , it predicts the observed gapless and phonon - like excitations ; 2 ) the mass , momentum and energy conservation laws are fulfilled in the gas dynamical description . an approximation that satisfies these properties is said to be _ gapless _ and _ conserving _ @xcite .
besides these unique features , the grpa predicts also other phenomena , namely a second branch of excitations and the dynamical screening of the interaction potential .
these phenomena appear also in the case of a gas of charged particles or plasma .
the possibility of a second kind of excitation has been explained quite extensively in @xcite .
there is a distinction between the single particle excitations and the collective excitations . in the case of a plasma
, the first corresponds to the electrically charged excitations and its dispersion relation is obtained from the pole of the one particle green function .
the second corresponds to the plasmon which is a chargeless excitation whose the dispersion relation is obtained from the pole of the susceptibility function .
the plasmon mediates the interaction between two charged excitations .
more precisely , during the interaction , one charged excitation emits a virtual plasmon which is subsequently reabsorbed by another charged excitation ( see fig.1 ) . remarkably , such a description holds also for a bose gas with single atom excitations carrying one unit of atom number and with gapless collective excitations with no atom number .
the poles of the green functions have a similar structure above the critical point . but
below this critical point , the existence of a macroscopic condensed fraction
_ hybridizes _ the collective and single particle excitations so that the poles of the one particle green function and the susceptibility function mix to form common branches of collective excitations @xcite .
thus , at the difference of a plasma , the presence of a condensed fraction prevents the direct observation of the atom - like excitation through the one particle green function .
the dynamical screening effect predicted in the grpa appears much more spectacular in a bose gas .
the screening effect of the coulombian interaction is well known to explain the dissociation of salt diluted in water into its ions ( see fig.2a ) .
but it also provides an explanation to the superfluidity phenomenon i.e. the possibility of a metastable motion without any friction .
most of the literature on superfluidity is usually devoted to the study of metastable motion in a toroidal geometry like , for example , an annular region between two concentric cylinders possibly in rotation @xcite . in this simply connected geometry
, the angular momentum about the axis of the cylinder of the superfluid is quantized in unit of @xmath2 .
the metastability of the motion is explained by the impossibility to go continuously from one quantized state to another due to the difficulty to surmount an enormous free - energy barrier .
this is not the situation we want to address in this paper .
we are rather focusing on the explanation of the superfluid ability to flow without any apparent friction with its surrounding .
the landau criterion is a necessary but not sufficient condition for superfluidity .
it tells about the kinematic conditions under which an external object can move relatively to a superfluid without damping its relative velocity by emitting a phonon - like collective excitation . for a dilute bose gas at low temperature , it amounts to saying that this relative velocity must be lower than the sound velocity @xcite .
the external object is assumed to be macroscopic and can be an impurity @xcite , an obstacle like a lattice @xcite or even the normal fluid @xcite . in particular , this criterion does not taken into account the fact that the normal fluid is microscopically composed of thermal excitations . in a bose condensed gas , even though their relative velocity is on average lower than the critical one , many of these excitations are very energetic with a relative velocity high enough to allow the phonon emission . in the grpa where these excitations correspond to the thermal atoms and under the condition of the landau criterion ,
such a process is forbidden as shown qualitatively from fig.2b .
the effect of an external perturbation of the condensed atoms caused for example by the thermal atoms is attenuated by the dynamical screening .
this screening is total in the sense that no effective mutual binary interaction allows a collision process which would be essential for a dissipative relaxation of the superfluid motion .
the purpose of this paper is to show that these peculiar phenomena could in principle be observed in a raman scattering process .
this process induces a transition for a given frequency @xmath0 and a wavevector @xmath1 determined from the difference of the frequencies and the wavevectors of two laser beams @xcite . for each wavevector corresponding to the transferred momentum , one can arbitrarily tune the frequency in order to reach the resonance energy associated to the excitation .
unlike the bragg scattering which allows the observation of the bogoliubov phonon - like collective excitation @xcite , the raman scattering is more selective .
not only the gas is probed with a selected energy transition and transferred momentum , but the atoms are scattered into a selected second internal hyperfine level . through a zeeman splitter , they can be subsequently analyzed separately from unscattered atoms . according to the grpa ,
the scattered thermal atoms become distinguishable from the unscattered ones and thus release the gap energy due to the exchange interaction . in a previous study @xcite , we showed that this gap appears as a resonance in the frequency spectrum of the atom transition rate at @xmath3 .
the possibility of momentum transfer allows to analyze the influence of the screening of the external perturbation induced by the raman light beams .
the paper is divided as follows . in section 2 ,
we review the time - dependant hartree - fock ( tdhf ) equations for a spinor condensate and study the linear response function to an external potential which gives results equivalent to the grpa .
sections 3 and 4 are devoted to the bragg and raman scatterings respectively .
section 5 ends up with the conclusions and the perspectives .
we start from the time - dependant hartree - fock equations for describing two component spinor bose gas @xcite labeled by @xmath4 .
the atoms have a mass @xmath5 , feel the external potential @xmath6 and the hartree and fock mean field interaction potential characterized by the coupling constants @xmath7 expressed in terms of the scattering lengths @xmath8 between components @xmath9 and @xmath10 ( @xmath11 ) .
note that no fock mean field ( or exchange ) interaction energy appears between condensed atoms .
these equations describe the time evolution of a set of spinor wave function @xmath12 describing @xmath13 atoms labeled by @xmath14 and depending on the position @xmath15 and on the time @xmath16 . for the condensed mode ( @xmath17 ) , these are : @xmath18 for a non condensed mode ( @xmath19 ) , these are @xmath20 the non condensed spinors remain orthogonal during their time evolution in the thermodynamic limit . in general
, the spinor associated to the condensed mode does not remain orthogonal with the others . but
according to @xcite , the non orthogonality is not important in the thermodynamic limit for smooth external potential .
another way of justifying the non orthogonality is to start from an ansatz where the condensed spinor mode is described in terms of a coherent state and the non condensed ones in terms of a complete set of orthogonal fock states i.e. @xmath21 where @xmath22 is the atom creation operator in the mode @xmath14 and @xmath23 .
the theory remains _ conserving _ because the conservation laws are preserved on average but becomes non _ number conserving _ since the quantum state is not an eigenstate of the total particle number operator .
this procedure is justified in the thermodynamic limit since the total particle number fluctuations are relatively small during the time evolution .
in contrast , instead of using spinor wavefunctions , the alternative method based on the use of excitation operators is number conserving @xcite .
the atom number @xmath13 for each mode is supposed time - independent in the tdhf . strictly speaking
, a collision term must be added in order to allow population transfers between the various modes .
these equations are valid in the collisionless regime i.e. on a time scale shorter than the average time between two collisions @xmath24 where @xmath25 is the average velocity and @xmath26 is the scattering cross section . in these conditions ,
the resulting frequency spectrum has a resolution limited by @xmath27 .
the magnitude order of resolution of interest is given by the @xmath28 s so we require @xmath29 which is generally the case when @xmath30 .
these conditions are fulfilled for the parameter values considered in this work . in the following
, we will restrict our analysis to a bulk gas embedded in a volume @xmath31 . at @xmath32 , we assume all atoms in thermodynamic equilibrium in the level @xmath33 and that @xmath34 except for @xmath35 which is constant and fixes the energy shift between the two sub - levels . in that case , the solutions of the tdhf are orthogonal plane waves with @xmath14 corresponding to the momentum @xmath36 : @xmath37}{\sqrt{v } } \left(\begin{array}{c}1 \\ 0 \end{array } \right)\end{aligned}\ ] ] where we define the hartree - fock energy for atoms with momentum @xmath36 : @xmath38 where @xmath39 and where the condensed and total particle densities are @xmath40 and @xmath41 .
eq.([drs ] ) corresponds to the dispersion relation of the single particle excitation . at equilibrium , @xmath42 - 1 ) \ ] ] is the bose - einstein distribution . below the condensation point
, the chemical potential becomes @xmath43 and the macroscopic occupation @xmath44 is fixed to satisfy the total number conservation .
for @xmath45 , we apply an external potential . for the bragg and raman scatterings ,
these are respectively : @xmath46\end{aligned}\ ] ] we solve the system through a perturbative expansion : @xmath47 the equations of motion for the first order corrections are for the case of bragg and raman scatterings respectively : @xmath48 \psi^{(1)}_{1,{\mathbf{k } } } = \nonumber \\ \left[v_{11}+\!\sum_{\mathbf{k ' } } g_{11 } ( 2-\delta_{{\mathbf{k'}},{\mathbf{0}}}\delta_{{\mathbf{k}},{\mathbf{0 } } } ) ( { \psi^{(0)*}_{1,{\mathbf{k ' } } } } \psi_{1,{\mathbf{k'}}}^{(1 ) } + c.c . )
n_{\mathbf{k ' } } \right]\ ! \psi^{(0)}_{1,{\mathbf{k } } } \\ \label{p21 } \left[i{\partial_t } + \frac{\nabla^2_{\mathbf{r}}}{2 m } - \omega_0 - g_{12}(n-\delta_{{\mathbf{k}},{\mathbf{0}}}n_{{\mathbf{0 } } } ) \right ] \psi^{(1)}_{2,{\mathbf{k } } } = \left[v_{12 } + g_{12}\sum_{\mathbf{k ' } } n_{\mathbf{k ' } } { \psi^{(0)*}_{1,{\mathbf{k ' } } } } \psi^{(1)}_{2,{\mathbf{k'}}}\right ] \psi^{(0)}_{1,{\mathbf{k}}}\end{aligned}\ ] ] these two set of integral equations can be solved exactly using the methods developed in @xcite .
defining the fourier transforms : @xmath49 } v_{ab}({\mathbf{r}},t)\end{aligned}\ ] ] one obtains in the level 1 for the condensed mode : @xmath50 for the non condensed modes ( @xmath51 ) : @xmath52 and in the level 2 for all modes : @xmath53 these formulae resemble the one obtained from the non interacting bose gas excepted for the mean field term in ( [ psir ] ) and the extra factors representing the screening effect . for the bragg scattering
, these factors can be written as @xcite : @xmath54 where @xmath55 -8g_{11}\chi_0({\mathbf{q}},\omega)g_{11 } n_{{\mathbf{0}}}\epsilon_{\mathbf{q}}\end{aligned}\ ] ] is the propagator for the collective excitations , @xmath56 is the bogoliubov excitation energy , @xmath57 is the sound velocity and @xmath58 is the susceptibility function describing the normal atoms . for the raman scattering , it is @xmath59 where @xmath60 knowing the fourier transform of the potential @xmath61 and @xmath62 , eqs.([psib0],[psib],[psir ] ) are calculated using the contour integration method over @xmath63 by analytic continuation in the lower half plane . as a consequence ,
the poles of the integrand tell about the excitation frequencies induced by the external perturbation .
the pole of the propagator containing @xmath36 corresponds to atom excitation involving one mode only while the poles coming from the screening factors correspond to the excitations involving all modes @xmath36 collectively .
thus , the tdhf approach predicts both single atom and collective excitations . note that the single mode excitation is not possible for the condensed atoms since the corresponding pole is compensated by a zero coming from the screening factor .
the expressions ( [ psib0],[psib],[psir ] ) have an interpretation shown in fig.3 .
an atom of momentum @xmath36 is scattered into a state of momentum @xmath64 by means of an external interaction mediated by a virtual collective excitation of momentum @xmath65 . [ vc ]
let us first review the bragg scattering process . up to the second order in the bragg potential , the atoms number for any mode @xmath36
can be decomposed into an unscattered part : @xmath66\end{aligned}\ ] ] and a scattered part : @xmath67 is determined through the conservation relation @xmath68 . generally speaking within the sublevel 1 , the scattered atoms can not be distinguished from the unscattered ones .
but in order to understand the underlying physics , we assume that distinction is possible . within the second order perturbation theory ,
the quantity of interest is the scattered atom rate per unit of time and is expected to reach a stationary value after a certain transition time . in the following , we shall analyze these transition rates for time long enough that transient effects disappear . in these conditions ,
a perturbative approach is still valid for very large time provided that the scattered atom number remains low compared to unscattered ones .
this last requirement is always satisfied with a sufficiently weak external perturbation . at zero temperature ,
only the condensed wave function is modified and eq.([psib0 ] ) becomes after contour integration over @xmath63 : @xmath69\end{aligned}\ ] ] the response function is only resonant at the bogoliubov energy @xmath70 .
also no transient response appears at zero temperature . using ( [ psib0 ] ) and ( [ psib ] ) , the total number of scattered atom can be obtained by determining the total momentum : @xmath71 through @xcite : @xmath72 using eq.([t0 ] )
, we recover that : @xmath73 where @xmath74 is the static structure factor .
the delta function comes from the relation @xmath75 .
the result ( [ suscb ] ) obtained in the grpa is identical to the one obtained from the bogoliubov approach where @xmath76 can be calculated equivalently from the four points correlation function @xcite .
but in any case the generated phonon like excitation is still a part of the macroscopic wave function @xmath77 . at temperatures different from zero ,
the poles become imaginary which means that any bogoliubov excitation is absorbed by a thermal atom excitation @xcite .
this phenomenon is known as the landau damping .
so for long time , only the residues of ( [ psib0 ] ) with poles touching the real axis contribute whereas the others give rise to transient terms negligible for long time .
thus the perturbative part becomes : @xmath78 using the property @xmath79 , the total number in the condensed mode reaches a constant value @xmath80 and the scattered thermal atom rate is given by : @xmath81 from ( [ p2 ] ) , we deduce for the imaginary susceptibility : @xmath82 the basic interpretation of these formulae is the following . at finite temperature ,
the collective excitation modes created by the external perturbation are damped over a time given by the inverse of the landau damping .
so the number of collectively excited condensed atom reaches the constant value ( [ n0scat ] ) when the produced collective excitations rate compensates their absorption rate by thermal atoms .
this constant value is higher for a transition frequency and a transferred momentum close to the resonance @xmath83 .
the formula ( [ nscatt ] ) is a generalization of the fermi - golden rule when the screening effect is taken into account .
the external potential perturbs the thermal atoms of momentum @xmath36 in two channels by transferring a momentum @xmath84 and a transition energy @xmath85 such that the resulting single atom excitation has a momentum @xmath86 and a kinetic energy @xmath87 .
the presence of the screening factor amplifies or reduces the scattering rate .
amplification ( or anti - screening ) occurs for a frequency close to the resonance energy @xmath88 of the collective excitations . on the contrary ,
dynamical screening occurs for a frequency close to the pole of the screening factor and is total for transition involving condensed atom at @xmath89 .
thus , in grpa , attempt to generate incoherence through single condensed atom scattering is forbidden at finite temperature .
only collective excitations affect the condensed mode but they are damped and therefore can not contribute to effectively transfer condensed atoms to a different mode @xcite .
it is taught in standard textbooks @xcite that , in the impulse approximation used for large @xmath1 , the response of the system is sensitive to the momentum distribution of the gas , since the atoms behave like independent particle . in particular ,
a delta peak is expected to account for the presence of a condensate fraction .
the difficulty of the observation of this peak could be explained by this impossibility of a single condensed atom excitation at finite temperature . for completeness , let us mention that interaction with thermal atoms can be also totally screened and inspection of the formulae ( [ k ] ) shows that this happens for @xmath90 @xcite .
4 shows these features in the frequency spectrum for the total momentum rate ( [ chit ] ) at fixed @xmath1 .
we choose the typical density observed experimentally for @xmath91 at the trap center @xcite .
these results can be put in direct relation with the analysis of impurity scattering @xcite . indeed , the dynamic response function is related to the dynamic structure factor through the fluctuation - dissipation theorem : @xmath92 .
the dynamic structure factor is directly connected to the transition probability rate @xmath93 that an external particle or impurity changes its initial momentum @xmath94 and energy @xmath95 into @xmath96 and @xmath97 respectively : @xmath98 where @xmath99 is the fourier transform of the interaction potential between the impurity and the atom gas .
the total rate of scattering @xmath100 results from a virtual process involving emission and absorption of the collective excitations : @xmath101 as a consequence , the impurity scattering is possible provided that the energy and momentum are conserved in a effective collision with a thermal atom of momentum @xmath36 mediated by a virtual collective excitation . note that total screening prevents impurity scattering involving ongoing and outgoing condensed atoms .
in contrast , for temperature close to zero , the landau damping approaches zero since @xmath102 so that the application of eq.([suscb ] ) to ( [ gam ] ) leads to an on - energy shell process of absorption and emission of a collective excitation . we obtain : @xmath103 where @xmath104 .
this limit case leads to the apparent interpretation of an impurity interacting with a thermal bath of phonon - like quasi - particle .
this situation has been considered in @xcite in the study of the impurity dynamics . instead , eq.([gam2 ] )
provides a generalization for higher temperature emphasizing that any external particle can excite a single thermal atom alone but not a condensed one .
the conclusions so far obtained in the bragg process can be extended straightforwardly to the case of raman scattering with the difference that only one channel of scattering is possible . for the purpose of simplicity ,
we choose the case @xmath105 .
also this channel is easier to access experimentally . defining the detuning @xmath106 , explicit calculations of the spinor component ( [ psir ] ) in the second sublevel
give : @xmath107 so we obtain for the atom number in the mode @xmath36 : @xmath108 by summing over all the modes , we obtain the density rate transferred in level 2 : @xmath109 where we define the imaginary part @xmath110 of the intercomponent susceptibility function : @xmath111 this last formulae is also the one obtained in the grpa @xcite .
again we find a similar structure as the intracomponent case . in this process
, thermal atoms with an initial momentum @xmath36 and energy @xmath112 are transferred into a second level with momentum @xmath113 and energy @xmath114 provided @xmath115 . in absence of screening , a resonance appears at the detuning @xmath116 .
the first term corresponds to the usual recoil energy while the second is the gap energy @xmath117 that results from the exchange interaction . during the raman transition
, the transferred atoms become distinguishable from the others and release this gap energy .
the scattering rate is determined through the imaginary part of the susceptibility eq.([chirpa ] ) versus the transition frequency @xmath0 and at fixed @xmath1 .
[ fig:1 ] and [ fig:2 ] show the corresponding resonance around this gap in absence of screening .
the screening effect strongly reduces the raman scattering and , in particular , forbids it for atoms with momentum @xmath36 such that @xmath118 .
this case corresponds to @xmath119 and includes also the condensed atoms ( @xmath120 ) .
the graphs illustrate well the effect of the macroscopic wave function that deforms its shape in order to attenuate locally the external potential displayed by the raman light beams and to prevent incoherent scattering of the condensed atom .
the experimental observation of this result would explain some of the reasons for which a superfluid condensate moves coherently without any friction with its surrounding .
anti - screening occurs in the region close to the resonance frequency @xmath88 of the collective mode . at zero temperature
, we recover @xmath121 @xcite while for non zero temperature the collective modes become damped for @xmath122 @xcite .
these results can be compared to the one obtained from the bogoliubov non conserving approximation developed in @xcite and valid only for a weakly depleted condensate .
this approach implicitly assumes that the only elementary excitations are the collective ones and form a basis of quantum orthogonal states that describe the thermal part of the gas .
consequently , this formalism predicts no gap and no screening .
instead , the intercomponent susceptibility describes transitions involving the two collective excitation modes of phonon @xmath123 and of rotation in spinor space @xmath124 : @xmath125 where @xmath126^{1/2}$ ] .
this function does not preserve the f - sum rule associated to the @xmath127 symmetry .
in contrast to the grpa , a delta peak describes a spinor rotation transition of the condensed fraction , and two other transitions involve the excitation transfer from a phonon mode into a rotation mode and the excitation creation in the two modes simultaneously . for small @xmath1 ,
these processes remain dispersive since the frequency transition depends on the momentum @xmath36 . as a consequence , the resulting spectrum shown in figs .
5 and 6 is broader . in particular
, the process of creation in the two modes favors transition with positive frequency .
note also the maximum of the curve separating the region involving a transition atom - atom like ( high @xmath36 ) and the one involving a transition phonon - atom like ( low @xmath36 ) .
all these features established so far for the bulk case allow a clear comparison between the grpa and the bogoliubov approaches . in the real case of a parabolic trap , the inhomogeneity induces a supplementary broadening of the spectrum that prevents the direct observation of the screening .
this effect as well as the finite time resolution and the difference between the scattering lengths will be discussed in a subsequent work .
we have analyzed the many body properties that can be extracted from the raman scattering in the framework the grpa . the calculated spectrum allows to show the existence of a second branch of excitation but also the screening effect which prevents the excitation of the condensed mode alone .
the observation of phenomena like the gap and the dynamical screening could have significant repercussions on our microscopic understanding of a finite temperature bose condensed gas and its superfluidity mechanism . on the contrary , the non - observation of these phenomena would imply that the _ gapless _ and _ conserving _ grpa is not valid . in that case
, a different approximation has to be developed in order to explain what will be observed . as an alternative ,
the idea to use the bogoliubov approach has been also discussed .
but unfortunately , the violation of the f - sum rule is a serious concern regarding this _ non conserving _
approach @xcite .
all these aspects emphasize the importance of the experimental study of the raman scattering at finite temperature .
pn thanks the referees for usefull comments and acknowledges support from the belgian fwo project g.0115.06 , from the junior fellowship f/05/011 of the kul research council , and from the german avh foundation .
p. szpfalusy and i. kondor , ann . phys . *
82 * , 1 - 53 ( 1974 ) ; a. minguzzi , m.p .
tozi , j. phys .
matter * 9 * 10211 ( 1997 ) ; m. fliesser , j. reidl , p. szpfalusy , r. graham , phys .
a * 64 * , 013609 ( 2001 ) ; j. reidl , a. csordas , r. graham , p. szpfalusy , phys . rev .
a * 61 * , 043606 ( 2000 ) .
zhang and h.a .
fertig , phys .
a * 74 * , 023613 ( 2006 ) .
zhang and h.a .
fertig , phys .
a * 75 * , 013601 ( 2007 ) .
p. navez , j. low .
phys . * 138 * , 705 - 710 ( 2005 ) ; p. navez , physica a * 356 * , 241 - 278 ( 2005 ) ; p. navez and r. graham , phys .
a * 73 * , 043612 ( 2006 ) .
oktel and l.s .
levitov , phys .
lett . * 83 * , 1 ( 1999 ) ; m. .
oktel and l.s .
levitov , phys .
a * 65 * , 063604 ( 2002 ) ; m. .
oktel , t.c .
killian , d. kleppner and l.s .
levitov , phys .
a * 65 * , 033617 ( 2002 ) ; m. .
oktel and l.s .
levitov , phys .
lett . * 88 * , 230403 ( 2002 ) .
p. navez , physica a * 387 * , 4070 ( 2008 ) .
a. griffin , phys .
b * 53 * , 9341 ( 1996 ) ; p.c .
hohenberg and p.c .
martin , ann .
( n.y . ) * 34 * , 291 ( 1965 ) ; a. griffin , excitations in a bose - condensed liquid ( university press , cambridge , 1993 ) .
n.m . hugenholtz and d. pines , phys . rev . *
116 * , 489 ( 1959 )
stamper - kurn , a.p .
chikkatur , a. grlitz , s. inouye , s. gupta , d.e .
pritchard , and w. ketterle , phys .
lett . * 83 * , 2876 ( 1999 ) .
f. zambelli , l. pitaevskii , d.m .
stamper - kurn and s. stringari , phys .
rev . a * 61 * , 063608 ( 2000 ) ; r. ozeri , n. katz , j. steinhauer , and n. davidson , rev .
phys . * 77 * , 187 ( 2005 ) .
a. j. leggett , rev .
phys . * 73 * , 307 ( 2001 ) ; l. pitaevskii and s. stringari , _ bose - einstein condensation _ ,
( clarendon press , 2003 ) ; c. j. pethick and h. smith , _ bose - einstein condensation in dilute gases _ , ( cambridge university press , 2001 )
. a. j. leggett , _ superfluidity _ , rev .
phys . * 71 * , s318 ( 1999 ) .
a. p. chikkatur , a. grlitz , d. m. stamper - kurn , s. inouye , s. gupta , and w. ketterle , phys .
85 * , 483 ( 2000 ) .
f. s. cataliotti , s. burger , c. fort , p. maddaloni , f. minardi , a. trombettoni , a. smerzi , and m. inguscio , science * 293 * , 843 ( 2001 ) .
d. a. huse and e.d .
siggia , j. low .
* 46 * , 137 ( 1981 ) .
d. pines and p. nozires , the theory of quantum liquids ( addison - wesley , readin , ma , 1989 ) , vol .
montina , phys .
a * 67 * , 053614 ( 2003 ) ; a. montina , phys . rev .
a * 66 * , 023609 ( 2002 ) w. b. colson and a. l. fetter , j. low temp
. phys . * 33 * , 231 ( 1978 ) ; w. zhang , d. l. zhou , m .- s .
chang , m. s. chapman , and l. you , phys .
* 95 * , 180403 ( 2005 ) . | we explore the many body physics of a bose condensed atom gas at finite temperature through the raman transition between two hyperfine levels . unlike the bragg scattering where the phonon - like nature of the collective excitations has been observed ,
a different branch of thermal atom excitation is found theoretically in the raman scattering .
this excitation is predicted in the generalized random phase approximation ( grpa ) and has a gapped and parabolic dispersion relation .
the gap energy results from the exchange interaction and is released during the raman transition .
the scattering rate is determined versus the transition frequency @xmath0 and the transferred momentum @xmath1 and shows the corresponding resonance around this gap .
nevertheless , the raman scattering process is attenuated by the superfluid part of the gas .
the macroscopic wave function of the condensate deforms its shape in order to screen locally the external potential displayed by the raman light beams
. this screening is total for a condensed atom transition in order to prevent the condensate from incoherent scattering .
the experimental observation of this result would explain some of the reasons why a superfluid condensate moves coherently without any friction with its surrounding . |
coxeter transformations play an important role in the development of lie group and lie algebra theory . for a compact lie group , a coxeter transformation or a coxeter element is defined to be the product of all the reflections in the root system initially . h. s. coxeter studies these elements and their eigenvalues in @xcite .
he also observes that the number of roots in the corresponding root system is equal to the product of the order of the coxeter transformation ( called the coxeter number ) and the number of eigenvalues of the coxeter transformation , which is proved by a. j. coleman in @xcite .
+ in @xcite , v. f. subbotin and r. b. stekolshchik consider the eigenvalues and the jordan canonical forms of the coxeter transformations for symmetrizable cartan matrices .
they prove that the jordan canonical form of the coxeter transformation is diagonal if and only if the tits form is non - degenerate .
for an extended dynkin diagram , the jordan canonical form of the coxeter transformation contains only one @xmath3 jordan block and all eigenvalues have norm @xmath4 .
they also prove the following : if the tits form is indefinite and degenerate , then the number of two - dimensional jordan blocks coincides with dimension of the kernel of the tits form , and the remaining jordan blocks are @xmath5 .
c. m. ringel generalizes their results to generalized cartan matrices in @xcite , he shows that the spectral radius @xmath6 of the coxeter transformation has norm greater than @xmath4 , and @xmath6 is also an eigenvalue with multiplicity one .
a. j. coleman computes characteristic polynomials for the coxeter transformations of all extended dynkin diagrams , including the case with cycles @xmath7 in @xcite .
there is a nice survey @xcite on this subject and recent development + the coxeter transformation is also important in the study of representations of algebras , quivers , partially ordered sets ( posets ) and lattices , see @xcite . its distinguished role in these areas is related to the construction of the coxeter functors @xmath8 ( see @xcite ) and given by i. n. bernstein , i. m. gelfand and v. a. ponomarev in @xcite . given any path algebra @xmath9 with finite global dimension , let @xmath10 be the cartan matrix .
it is well - known that the coxeter transformation coincides with @xmath11 .
one interesting result is that the spectral radius @xmath6 of the coxeter transformation is related to the representation type of the hereditary algebra .
concretely , let @xmath9 be a hereditary algebra .
then @xmath12 @xmath9 is representation - finite if @xmath13 and @xmath4 is not a root of the coxeter polynomial @xmath14 .
@xmath12 @xmath9 is tame if @xmath15 .
@xmath12 @xmath9 is wild if @xmath16 .
+ when we consider the bounded derived category of a finite dimensional algebra @xmath9 and its grothendieck group , the coxeter transformation is just the linear transformation on the grothendieck group induced by the functor @xmath17 $ ] ( see , e.g. , @xcite ) , where @xmath18 is the serre functor ( see @xcite ) and @xmath19 $ ] is the inverse of the shift functor which is denoted by @xmath20 $ ] .
so in general , we can define the coxeter transformation of a triangulated category with serre functor @xmath21 to be the linear transformation induced by the functor @xmath22 $ ] on grothendieck group .
it is an invariant under triangulated equivalences . about the derived equivalences of rings and algebras , j. rickard get interesting results in @xcite .
+ besides the bounded derived categories of finite dimensional algebras ( see , e.g. , @xcite ) , another important class of triangulated categories is the bounded derived categories of coherent sheaves on ( smooth projective ) varieties , see , e.g. , @xcite .
an interesting question is when the derived category of a variety is equivalent to the one of a finite dimensional algebra . in 1970 s ,
a. a. beilinson constructs an algebra derived equivalent to @xmath23 , in this case , tilting sheaves play an important role . in @xcite , a. i. bondal and m. m. kapranov
give an useful method to find a tilting sheaf : ( strongly ) exceptional sequences of sheaves ( in particular line bundles ) .
after that , many smooth complete varieties are found to admit a strongly exceptional sequences of line bundles , and most of them are smooth complete toric varieties , @xcite .
a. king conjectures that all smooth toric varieties admit a strongly exceptional sequence of line bundles in @xcite , but this is proved false by l. hille and m. perling in @xcite .
+ by serre duality @xcite , we know that the derived categories of smooth complete varieties admit serre functors , so we can define the coxeter transformations for them .
the main goal of this paper is to investigate the coxeter transformation of the derived category of coherent sheaves on smooth complete varieties .
we compute their characteristic polynomials and jordan canonical forms .
we also find relations between these invariants and the geometric properties of the varieties .
these invariants give necessary conditions for the bounded derived category of a smooth variety @xmath24 to be triangulated equivalent to the bounded derived category of a finite dimensional algebra @xmath9 .
+ the paper is organized as follows . in section 2 ,
we recall some basic properties of smooth varieties and facts needed for our main results . in section 3 , we prove that if the rank of the grothendieck group of a smooth complete variety of dimension @xmath2 is finite , say @xmath0 , then the characteristic polynomial of the coxeter transformation is @xmath1 , see theorem [ main theorem 1 ] .
+ in section 4 , we study the jordan canonical forms of coxeter transformations .
we first prove that for a smooth complete toric variety with ample canonical or anticanonical sheaf , its betti numbers are uniquely determined by and also determine the number of the jordan blocks in the jordan canonical form of the coxeter transformation and their dimensions , see theorem [ main theorem 3 ] .
we then prove that for a rational surface , the number of the jordan blocks is equal to the rank of the picard group ; furthermore if the rank of the picard group is not @xmath25 , then the jordan canonical form of the coxeter transformation contains only one jordan block of dimension 3 , and the others are of dimension one ; otherwise , it has two jordan block of dimension @xmath26 , and the others are of dimension @xmath4 , see theorem [ main theorem 2 ] . as a corollary ,
if a rational surface or a smooth complete toric variety @xmath24 with ample canonical or anticanonical divisor is derived equivalent to a finite dimensional algebra @xmath9 , then the betti numbers of @xmath24 are determined by the matrix @xmath11 , see corollary [ corollary derived equivalent ] . + in the final section 5 , we first prove that for two smooth toric varieties @xmath24 and
@xmath27 , if the jordan canonical form of the coxeter transformations of @xmath24 and @xmath27 are @xmath28 and @xmath29 respectively , then the sizes and number of jordan blocks in the jordan canonical form of the coxeter transformation of @xmath30 are the same as that of @xmath31 , see proposition [ theorem jordan block ] .
we then give a formula to compute the jordan canonical form for tensor product of invertible matrices , see proposition [ proposition jordan block ] .
this formula generalizes a formula for the tensor product of two matrices in @xcite and it is useful to construct graded local frobenius algebras .
the authors deeply thank their supervisors professor bin zhang and professor liangang peng for helpful discussions .
throughout this paper , we assume that the ground field @xmath32 is an algebraically closed field of characteristic @xmath33 . + let @xmath24 be a variety over @xmath32 .
the category of coherent sheaves on @xmath24 is denoted by @xmath34 .
the locally free coherent sheaves on @xmath24 are called _ vector bundles _ , and they form a full subcategory of @xmath34 which is denoted by @xmath35 .
the _ grothendieck group of coherent sheaves _ on a variety @xmath24 , denoted by @xmath36 , is defined to be generated by @xmath37 $ ] , the isomorphism classes of coherent sheaves on @xmath24 , subject to the relations @xmath37=[{{\mathcal f}}']+[{{\mathcal f}}'']$ ] for every exact sequence @xmath38 similarly , we have the following definition .
the _ grothendieck group of vector bundles _ on a variety @xmath24 , denoted by @xmath39 , is defined to be generated by @xmath40 $ ] , the isomorphism classes of vector bundles on @xmath24 , subject to the relations @xmath40=[\mathbb{e}']+\mathbb{e}''$ ] for every exact sequence @xmath41 in the affine case , every short exact sequence of vector bundles splits . therefore , @xmath42 for an affine variety @xmath43 , where @xmath44 is the grothendieck group of projective modules over @xmath9 , see @xcite .
the embedding of the category of vector bundles on @xmath24 into the category of coherent sheaves induces the _ cartan homomorphism _ @xmath45
this homomorphism is in general neither injective nor surjective , but for a smooth variety @xmath24 , it is an isomorphism . [ theorem cartan hom ] ( see ,
e.g. , @xcite ) for a smooth variety , the cartan homomorphism is an isomorphism .
therefore , for a smooth variety @xmath24 , we do not distinguish between @xmath36 and @xmath39 , and just use @xmath39 to denote it .
+ we denote the bounded derived category of coherent sheaves on @xmath24 by @xmath46 or just @xmath47 . it is well known that @xmath47 is a triangulated category with shift functor @xmath20 $ ] .
the grothendieck group of @xmath47 is defined to be generated by the isomorphism class of complexes of coherent sheaves , subject to the distinguished triangles .
the following lemma is well - known .
let @xmath24 be a variety over @xmath32 .
then the grothendieck group of @xmath47 is isomorphic to @xmath36 . in particular ,
if @xmath24 is smooth , then the grothendieck group of @xmath47 is isomorphic to @xmath39 .
in fact , for any abelian category @xmath48 , we have that the grothendieck group of @xmath49 is isomorphic to @xmath50 . for this reason
, we do not distinguish the grothendieck group of @xmath49 and @xmath50 in the following .
+ the tensor product of vector bundles induces a commutative ring structure on @xmath39 , we call it the _ grothendieck ring _ of @xmath24 . in general , @xmath36 is a @xmath39-module .
+ ( @xcite ) even though @xmath24 is a smooth complete variety of dimension @xmath2 , we know that the grothendieck group @xmath39 is possibly huge , i.e. , with infinite rank . in the following
, we recall basic facts of _ chow groups _ of varieties , see @xcite .
let @xmath51 denote the group of @xmath32-dimensional algebraic cycles on a variety @xmath24 .
the corresponding quotient group of @xmath51 modulo the rational equivalences is called the _ chow group _ of @xmath32-cycles , denoted by @xmath52 .
let @xmath53 for a general ( not necessarily smooth ) complete @xmath2-dimensional variety @xmath24 , the group of cycles of codimension @xmath32 modulo rational equivalence is denoted by @xmath54 .
if in addition @xmath24 is smooth , then there is a product @xmath55 where @xmath56 is the diagonal embedding .
it coincides with the geometric intersection of cycles in the case of transverse intersections .
we denote this product by @xmath57 for @xmath58 and @xmath59 .
this product corresponds to cup product in cohomology , and there is a ring homomorphism @xmath60 but this morphism is neither monomorphic nor epimorphic in general , @xcite .
+ let @xmath24 be a variety of dimension @xmath2 .
the _ chern character
_ @xmath61 of a vector bundle @xmath62 of rank @xmath63 is given by @xmath64 where @xmath65 in particular , for a line bundle @xmath66 and its corresponding cartier divisor @xmath67 on @xmath24 , we have @xmath68 @xmath69 + for an abelian group @xmath70 , we use the notation @xmath71 .
( @xcite)[lemma chern map ] let @xmath24 be a smooth complete variety .
there is a ring isomorphism @xmath72\mapsto \mathrm{ch}_x(\e).\ ] ] for a normal variety @xmath24 , the sheaf of zariski @xmath73-forms is defined to be @xmath74 where @xmath75 is the inclusion of the smooth locus of @xmath24 .
( see , e.g. , @xcite ) the _ canonical sheaf _ of a normal variety @xmath24 is @xmath76 , where @xmath2 is the dimension of @xmath24 .
this is a reflexive sheaf of rank 1 , so that @xmath77 for some weil divisor @xmath67 on @xmath24 .
this divisor is called the _ canonical divisor _ of @xmath24 , denoted by @xmath78 .
notice that for a smooth variety @xmath24 , the canonical divisor is a cartier divisor .
( see , e.g. , @xcite ) [ theorem serre duality ] let @xmath79 be the canonical sheaf of a complete normal cohen - macaulay variety @xmath24 of dimension @xmath2 . then for every coherent sheaf @xmath80 on @xmath24 , there is a natural isomorphism @xmath81 @xmath82 so for any @xmath83 on @xmath24 , there is a natural isomorphism @xmath84 .
this motivates the following definition .
( @xcite ) let @xmath48 be a @xmath32-linear hom - finite category .
an autoequivalent functor @xmath18 is called a _ serre functor _ of @xmath48 , if there exists a natural isomorphism @xmath85 for any object @xmath86 , where @xmath87 .
in particular , a serre functor is unique up to a natural isomorphism .
+ when we consider the bounded derived category @xmath47 for a smooth complete variety @xmath24 of dimension @xmath2 , by theorem [ theorem serre duality ] , we know that @xmath47 has a serre functor @xmath88 $ ] .
the importance of the existence of serre functor in a krull - schmidt triangulated category is that it is equivalent to the existence of auslander - reiten triangles in it .
the auslander - reiten functor is equivalent to @xmath89 $ ] , so the functor @xmath90 $ ] is the auslander - reiten functor of @xmath47 . in the case of finite dimensional algebras ,
the coxeter transformations are defined to be induced by the auslander - reiten functors on grothendieck groups .
so we have the following definition similar to that of finite dimensional algebras .
let @xmath24 be a smooth complete variety of dimension @xmath2 and @xmath79 be its canonical sheaf .
then the transformation induced by @xmath90 $ ] on @xmath91 is called the _ coxeter transformation _ of @xmath24 .
furthermore , if the rational grothendieck group @xmath91 is finite dimensional , then the characteristic polynomial of the coxeter transformation is called the _
coxeter polynomial _ of @xmath24 .
if two triangulated categories with serre functors are triangulated equivalent , then their grothendieck groups are isomorphic and equivalent functors commute with serre functors .
it is easy to see that the coxeter polynomial is an invariant of triangulated equivalences .
in section 2 , we define the coxeter polynomials for a smooth complete variety . in this section , we compute it explicitly . [ main theorem 1 ] let @xmath24 be a smooth complete variety of dimension @xmath2 and the rank @xmath0 of the grothendieck group @xmath39 be finite .
then the coxeter polynomial of @xmath24 is @xmath1 . in particular , all eigenvalues of the coxeter transformation are @xmath92 . because @xmath24 is a smooth complete variety , we have the following ring isomorphism @xmath93\mapsto { \operatorname{ch}\nolimits}_x(\mathbb{e}).\ ] ] the coxeter polynomial is the characteristic polynomial of the induced linear transformation @xmath94 on grothendieck group by the auslander - reiten functor @xmath90 $ ] . for any object @xmath95
, we have @xmath96=(-1)^{n-1}[w^{*}]$ ] , so we only need compute the induced linear transformation @xmath97 on grothendieck group by the functor @xmath98 .
the eigenvalues of @xmath94 are the ones of @xmath97 multiplied by @xmath92 .
+ let @xmath99 be the induced transformation on @xmath100 by the intersection product with @xmath101 . for any vector bundle @xmath102 on @xmath24 , we have @xmath103)&=({\operatorname{ch}\nolimits}_x)_\q ( [ w\otimes\omega_x])\quad\quad\quad\quad\\ & = ( { \operatorname{ch}\nolimits}_x)_\q([w])\cdot({\operatorname{ch}\nolimits}_x)_\q([\omega_x])\\ & = \psi({\operatorname{ch}\nolimits}_x)_\q([w]).\quad\quad\quad\quad\quad\,\,\,\end{array}\ ] ] because @xmath39 is generated by vector bundles , we have the following commutative diagram .
@xmath104^{({\operatorname{ch}\nolimits}_x)_\q } \ar[dd]^{\phi ' } & & { \operatorname{ch}\nolimits}^\ast(x)_\q \ar[dd]^{\psi}\\ \\ k_0(x)_\q\ar[rr]^{({\operatorname{ch}\nolimits}_x)_\q } & & { \operatorname{ch}\nolimits}^\ast(x)_\q}\ ] ] since @xmath105 is an isomorphism , it is easy to see that the matrix representing the linear transformation @xmath97 is similar to the matrix representing @xmath99 , so we only need compute the eigenvalues of @xmath99 . +
for any element @xmath106\in { \operatorname{ch}\nolimits}^i(x)_\q$ ] , @xmath107\in{\operatorname{ch}\nolimits}^j(x)_\q$ ] , we have @xmath108\cdot [ f_j]\in { \operatorname{ch}\nolimits}^{i+j}(x)_\q \mbox { and } \psi([e_i])=[e_i\cdot ( { \operatorname{ch}\nolimits}_x(\omega_x))].\ ] ] since @xmath109 , @xmath110 let @xmath111 be the rank of @xmath112 . we choose classes @xmath113\in { \operatorname{ch}\nolimits}^i(x)$ ] for @xmath114 , @xmath115 , such that @xmath116|j=1,\dots , b_i\}$ ] form a basis of @xmath117 . for any element
@xmath113 $ ] , @xmath118)=[e_{ij}\cdot ( { \operatorname{ch}\nolimits}_x(\omega_x))]=[e_{ij}]+[e_{ij}\cdot k_x]+\cdots+ \frac{[e_{ij}\cdot k_x^{n - i}]}{(n - i)!}.\ ] ] using this basis , we can write @xmath119)$ ] as a linear combination of @xmath120 for @xmath121\in { \operatorname{ch}\nolimits}^{i+k}(x)$ ] , it is a linear combination of @xmath122 $ ] since @xmath123| 1\leq l\leq b_{i+k } \}\ ] ] is a basis of @xmath124 .
so the matrix representing @xmath99 under the basis @xmath113 $ ] is an @xmath125 upper triangular matrix with @xmath4 in the diagonal .
so all the eigenvalues of @xmath99 are @xmath4 .
furthermore , we know that the coxeter polynomial is @xmath1 .
the proof is complete .
the following definition is practically important to find an algebra derived equivalent to a smooth projective variety .
( @xcite ) a coherent sheaf @xmath126 on a smooth projective variety @xmath24 is called a _ tilting sheaf _
if \(1 ) it has no higher self - extensions , i.e. @xmath127 for all @xmath128 , \(2 ) the endomorphism algebra of @xmath126 , @xmath129 , has finite global homological dimension , \(3 ) the direct summands of @xmath126 generate the bounded derived category @xmath47 .
the importance of tilting sheaf is that it induces an equivalence between @xmath47 and @xmath130 ( the bounded derived category of finite generated right @xmath9-modules ) .
if @xmath126 is a vector bundle , we call it a _ tilting bundle_. if @xmath47 admits a tilting sheaf , then its grothendieck group is free . for a finite dimensional algebra with finite global dimension ,
it is well - known that @xmath131 admits auslander - reiten triangles such that the auslander - reiten translation @xmath132 is an equivalence , see @xcite . on the level of the grothendieck group
, @xmath132 induces the coxeter transformation @xmath133 .
let @xmath134 be a complete system of pairwise non - isomorphic simple @xmath9-modules , @xmath135 be the corresponding projective covers and @xmath136 be the injective envelopes .
the integers @xmath137 yield an integral @xmath138 matrix @xmath139 with the property that @xmath140 represents the coxeter transformation @xmath133 under the basis of @xmath44 formed by the classes @xmath141,\dots , [ p_n]$ ] . + from theorem [ main theorem 1 ] , we have the following corollary directly .
let @xmath9 be a finite dimensional algebra over @xmath32 with finite global dimension .
if @xmath130 is triangulated equivalent to a bounded derived category of a smooth complete variety , then all the eigenvalues of the coxeter transformation of @xmath9 are equal , either @xmath4 or @xmath142 .
we only need the fact that the grothendieck group of a finite dimensional algebra is free and its rank is finite .
[ example projective space ] let @xmath143 be the following quiver : @xmath144^{r_{11 } } \ar@<1ex>[rr]^{r_{12}}\ar@<-3.5ex>[rr]^{\vdots}_{r_{1n } } & & \circ^2\ar@<4ex>[rr]^{r_{21 } } \ar@<1ex>[rr]^{r_{22}}\ar@<-3.5ex>[rr]^{\vdots}_{r_{2n}}&&\circ^3 & \cdots&\circ^{n-1 } \ar@<4ex>[rr]^{r_{(n-1)1 } } \ar@<1ex>[rr]^{r_{(n-1)2}}\ar@<-3.5ex>[rr]^{\vdots}_{r_{(n-1)n } } & & \circ^n}.\ ] ] for any path @xmath145 , where @xmath146 and @xmath147 , we associate a monomial @xmath148 to @xmath149 in @xmath150 $ ] .
let @xmath151 , where @xmath152\rangle$ ] .
a. a. beilinson proves that @xmath130 is equivalent to @xmath153 in @xcite . by theorem [ main theorem 1 ] , we know that the coxeter polynomial of @xmath9 is @xmath154 . for a fixed variety
, it is very difficult to find a finite dimensional algebra which is derived equivalent to it .
l. costa and r. m. mir - roig proved that the following toric varieties have tilting bundles whose summands are line bundles , see @xcite .
\(i ) any @xmath155-dimensional smooth complete toric variety with splitting fan .
\(ii ) any @xmath155-dimensional smooth complete toric variety with picard number @xmath156 .
\(iii ) del pezzo toric surfaces . l. hille and m. perling prove that any smooth complete rational surface which can be obtained by blowing up a hirzebruch surface @xmath157 two - times in several points , also has a tilting bundle whose summands are line bundles .
see @xcite .
[ example rational surface ] let @xmath24 be a variety which can be obtained from @xmath158 by blowing up one - time in @xmath159 points .
let @xmath143 be the following quiver : @xmath160\ar@<-0.25ex>[ddrr ] & & & \\ & & \circ^{2 ^ 2 } \ar@<0.25ex>[drr]\ar@<-0.25ex>[drr ] & & & \\
\circ^1 \ar[uurr]\ar[urr]\ar[drr]\ar[ddrr]&&\vdots & & \circ^{3}\ar@<-1ex>[r]\ar@<1ex>[r]\ar[r ] & \circ^4 .
\\ & & \circ^{2^{t-1 } } \ar@<0.25ex>[urr]\ar@<-0.25ex>[urr ] & & & \\ & & \circ^{2^t } \ar@<0.25ex>[uurr]\ar@<-0.25ex>[uurr ] & & & \\ } \ ] ] when @xmath161 , there are just three arrows from vertex @xmath4 to vertex @xmath162 . when @xmath163 , there is additionally one arrow @xmath164 from vertex @xmath4 to vertex @xmath162 , at this time , the quiver is as the following diagram shows .
@xmath165\ar@<-0.25ex>[drr ] & & & \\ \circ^1\ar[urr]\ar[rrrr]^b & & & & \circ^{3}\ar@<-1ex>[r]\ar@<1ex>[r]\ar[r ] & \circ^4}\ ] ] from @xcite , we know that @xmath47 is equivalent to @xmath130 , where @xmath151 , @xmath166 is described in section @xmath167 in @xcite . by theorem [ main theorem 1 ] , we know that the coxeter polynomial of @xmath9 is @xmath168 since the rank of the grothendieck group of @xmath9 is @xmath169 .
we prove in the previous section that for a smooth complete variety @xmath24 , the coxeter transformation of @xmath24 is a linear transformation with eigenvalues @xmath4 or @xmath142 . in this section ,
we compute its jordan canonical form .
it is also an invariant under derived equivalences .
+ for an integer @xmath170 and an element @xmath171 or @xmath32 , let @xmath172 which is the _
jordan block _ of dimension @xmath173 with eigenvalues @xmath149 .
first , we recall some definitions and facts about fano varieties and toric varieties .
( see , e.g. , @xcite ) a complete normal variety @xmath24 is said to be a _
gorenstein fano variety _ if the anticanonical divisor @xmath174 is cartier and ample .
thus gorenstein fano varieties are projective .
+ in this subsection , we always assume that @xmath175 is a lattice of rank @xmath2 .
( @xcite ) a _ toric variety _ is an irreducible normal variety @xmath24 containing a torus @xmath176 as a zariski open subset such that the multiplication of @xmath177 on itself extends to an algebraic action of @xmath177 on @xmath24 . by the theory of toric varieties , such a toric variety @xmath24
is characterized by a fan @xmath178 of strongly convex polyhedral cones in @xmath179 if @xmath24 is normally separated .
( @xcite ) a _ convex polyhedral cone _ in @xmath179 is a set of the form @xmath180 where @xmath181 is finite .
a convex polyhedral cone @xmath182 is called _ strongly convex _ if @xmath182 contains no linear subspace
. it is called rational if @xmath183 . in this paper
, we always consider strongly convex rational polyhedral cones and call it cone for simplicity .
a _ fan _ in @xmath179 is a finite collection of cones @xmath184 such that : @xmath185 every @xmath186 is a strongly convex rational polyhedral cone .
@xmath187 for all @xmath186 , each face of @xmath182 is also in @xmath188 .
@xmath189 for all @xmath190 , the intersection @xmath191 is a face of each .
let @xmath192 be the dual lattice .
for any cone @xmath182 , denote @xmath193 the dual cone in @xmath194 by @xmath195 @xmath193 is also strongly convex if @xmath182 is strongly convex .
@xmath196 gives a character @xmath197 defined by @xmath198 in this way , @xmath199 is the character lattice of @xmath177 .
so for any semigroup @xmath200 , its _ semigroup algebra _ is @xmath201=\{\sum_{m\in s}c_m\chi^m\,|\,c_m\in \c\mbox { and } c_m=0\mbox { for all but finitely many } m\},\ ] ] with multiplication induced by @xmath202 a face @xmath132 of @xmath182 is given by @xmath203 , where @xmath204 and @xmath205 is the hyperplane defined by @xmath0 .
let @xmath206 be the semigroup .
then @xmath207 and @xmath208 $ ] is the localization @xmath209_{\chi^m}$ ] . to every cone @xmath186
, there is an affine variety @xmath210)$ ] .
if @xmath132 is a face of @xmath182 , then we have a natural embedding @xmath211 .
now consider the collection of affine toric varieties @xmath212)$ ] , where @xmath182 runs over all cones in a fan @xmath188 .
let @xmath213 and @xmath214 be any two of these cones and let @xmath215 .
we have an isomorphism @xmath216 which is the identity on @xmath217 .
define @xmath218 to be the variety glued by the affine varieties @xmath219 along the subvarieties @xmath220 , see ( @xcite ) .
+ the variety @xmath218 is a normally separated toric variety for a fan @xmath188 in @xmath179 .
if @xmath188 is complete , then @xmath218 is compact in the classical topology .
let @xmath221 for any cone @xmath182 in @xmath179 , points of @xmath219 are in bijective correspondence with semigroup homomorphisms @xmath222 .
for each cone @xmath182 , we have a point of @xmath219 defined by @xmath223 we denote this point by @xmath224 and call it the _ distinguished point _ corresponding to @xmath182 .
now , by the @xmath177-action on @xmath218 , we see that each cone @xmath186 corresponds to a torus orbit @xmath225 it is known that @xmath226 , where @xmath227 and @xmath228 is the sublattice of @xmath175 spanned by the points in @xmath229 , ( @xcite ) . for a fan @xmath188 in @xmath179 ,
we denote by @xmath230 the set of the cones of dimension @xmath231 in @xmath188 .
so for any ray @xmath232 , we get a prime @xmath177-invariant divisor @xmath233 on @xmath218 .
@xmath234 consider the fan @xmath188 in @xmath235 in _ figure_-@xmath4 , where @xmath236 has standard basis @xmath237 .
then @xmath238 and the divisor @xmath239 which is also the hyperplane divisor @xmath240 .
@xmath241 consider the fan @xmath188 in @xmath235 in _ figure_-@xmath26 .
then @xmath242 which is the hirzbruch surface if @xmath243 ; in particular , @xmath244 if @xmath245 .
we also get that the divisors @xmath246 , @xmath247 , so @xmath248 is a basis of the picard group . in the following ,
we denote @xmath249 and @xmath250 by @xmath251 and @xmath252 respectively . if @xmath218 is complete and simplicial , then @xmath253 as rings , @xcite .
let @xmath111 be the rank of @xmath254 .
then @xmath111 is the @xmath255-th betti number of @xmath218 .
denote the set of the maximal cones in a fan @xmath188 by @xmath256 .
( @xcite)[lemma toric variety ] let @xmath218 be a smooth complete toric variety
. then the groups @xmath257 and @xmath258 are free abelian groups of rank @xmath259 .
( @xcite ) let @xmath24 be a normal projective variety of dimension @xmath2 and @xmath260 be the class of an ample divisor on @xmath24 .
then we have the following : @xmath234 for all @xmath114 , multiplication by @xmath261 defines an isomorphism @xmath262 @xmath241 for @xmath263 , multiplication by @xmath264 @xmath265 is injective .
it is easy to see that ( 2 ) follows from ( 1 ) .
generally , for any variety @xmath24 , an element in @xmath266 which satisfies condition ( 1 ) of the hard lefchetz theorem is called a _ lefschetz element_. + we have a simple observation .
[ lemma nilpotent matrix ] let @xmath9 be an @xmath125 nilpotent matrix over @xmath267 .
then the number and the dimensions of the jordan blocks in the jordan canonical form of @xmath268 are the same as @xmath9 , where @xmath269 .
@xmath9 is a nilpotent matrix , then there exists a @xmath125 invertible matrix @xmath251 such that @xmath270 , where @xmath271 s are the jordan blocks of @xmath9 , and the eigenvalues of @xmath9 are zeros .
@xmath272 , where @xmath273 for @xmath274 .
so we only need prove that @xmath275 has only one jordan block .
it is easy to see that @xmath276 where @xmath277 is the dimension of @xmath275 .
all the eigenvalues of @xmath275 are @xmath4 , and it is easy to see that @xmath278 has rank @xmath279 , where @xmath280 is the identity matrix of dimension @xmath277 .
so @xmath275 has only one eigenvector @xmath281 , the proof of this lemma is complete .
now we can prove our main result .
[ main theorem 3 ] let @xmath218 be an n - dimensional smooth complete toric variety with ample canonical divisor or anticanonical divisor .
let @xmath111 be the rank of @xmath282 for @xmath114 and @xmath283 .
then we have the following : @xmath185 the number of jordan blocks of the jordan canonical form of the coxeter transformation is @xmath284}$ ] .
@xmath187 the maximal dimension of the jordan blocks in the jordan canonical form of the coxeter transformation is @xmath285 .
@xmath189 the jordan canonical form of the coxeter transformation has @xmath286 jordan blocks of dimension @xmath287 for any @xmath288 $ ] , and these are all its jordan blocks .
@xmath289 if the jordan canonical form of the coxeter transformation has @xmath290 @xmath291 jordan blocks of dimension @xmath292 for @xmath274 and @xmath293 , then @xmath234 @xmath294 , @xmath295 .
@xmath241 @xmath296}=t_1+\cdots+t_s\quad \mbox{if } i = s .
\end{split}\right.\ ] ] for simplicity , we denote @xmath297 by @xmath78 , which is the canonical divisor of @xmath218 .
we only need consider the case that the canonical divisor @xmath78 is ample , since in the case that the anticanonical divisor is ample , we can consider the inverse transformation of the coxeter transformation and the jordan normal form of a transformation with eigenvalues @xmath4 is the same as its inverse . since @xmath218 is complete and simplicial
, we get that @xmath298 and @xmath299 for @xmath114 .
so by hard lefschetz theorem , we know that @xmath300 is isomorphic for @xmath301 $ ] .
+ denote by @xmath302 the linear transformation induced by the intersection product with @xmath78 .
then @xmath303 is a nilpotent transformation and the coxeter transformation is @xmath304 . by lemma [ lemma nilpotent matrix ] , we only need prove the results for @xmath303 . + we know that @xmath305 is one - dimensional .
let @xmath306 be a generator of @xmath305 .
then the subspace spanned by @xmath307 is an invariant subspace of @xmath303 , its dimension is @xmath285 .
since @xmath308 and @xmath309 for any @xmath310 , it is easy to see that @xmath311 is an eigenvector of @xmath303 and @xmath307 is the longest chain which contains @xmath311 , so the jordan block corresponding to @xmath311 is of dimension @xmath285 .
+ from hard lefschetz theorem , we know that @xmath312 is surjective , so @xmath313 . since @xmath314 , we can choose a basis @xmath315 of @xmath316 , they are also eigenvectors of @xmath303 . since @xmath317 is an isomorphism , there exist unique @xmath318 in @xmath319 such that @xmath320 for @xmath321 .
thus for any @xmath321 , @xmath322 spans an invariant subspace of @xmath303 , its dimension is @xmath323 since @xmath324 are linear independent for fixed @xmath231 .
it is easy to see that @xmath322 is also the longest chain which contains @xmath325 .
so we get that the canonical form of @xmath303 has @xmath326 jordan blocks of dimension @xmath323 .
do it inductively , one get that the canonical form of @xmath303 has @xmath286 jordan blocks of dimension @xmath287 for any @xmath288 $ ] . to prove that these are all its jordan blocks
, we only need check the following equality : @xmath327}(b_i - b_{i-1})(n-2i+1)=\sum_{i=0}^n b_i.\ ] ] + notice that @xmath328 from the poincare duality or hard lefschetz theorem .
case 1 : @xmath2 is odd .
the left hand side is equal to @xmath329}2b_i$ ] .
the right hand side is equal to @xmath330 + 1)b_{[\frac{n}{2}]}+\sum_{i=0}^{[\frac{n}{2}]-1}((n-2i+1)-(n-2(i-1)+1))b_i= 2b_{[\frac{n}{2}]}+\sum_{i=0}^{[\frac{n}{2}]-1}2b_i\ ] ] since @xmath283 .
the equality is valid .
case 2 : @xmath2 is even .
this case is similar to case 1 .
the left hand side is equal to @xmath329 - 1}2b_i+b_{[\frac{n}{2}]}$ ] .
the right hand side is equal to @xmath330 + 1)b_{[\frac{n}{2}]}+\sum_{i=0}^{[\frac{n}{2}]-1}((n-2i+1)-(n-2(i-1)+1))b_i= b_{[\frac{n}{2}]}+\sum_{i=0}^{[\frac{n}{2}]-1}2b_i\ ] ] since @xmath283 .
the equality is valid .
so we get ( c ) .
now , ( a ) and ( b ) are trivial .
+ for ( d ) , if the jordan canonical form of the coxeter transformation has @xmath290 jordan blocks of dimension @xmath292 for @xmath274 and @xmath331 , then from ( a ) , ( b ) and ( c ) , it is easy to see that @xmath294 , @xmath295 . for the last statement , by ( c ) , we know that @xmath332 ( @xmath274 ) for some integer @xmath333 .
the formula in ( 2 ) make sense since @xmath334 is even for any @xmath274 .
if the betti numbers @xmath111 are as ( 2 ) shows , then we can check directly that the jordan canonical form of the coxeter transformation has @xmath290 @xmath291 jordan blocks of dimension @xmath292 .
so we only need prove that the betti numbers are unique .
if there exist @xmath335 such that they also satisfy the requirement , then @xmath336 . we denote @xmath337\},\ ] ] where @xmath295 .
so @xmath338 . if @xmath339 $ ] , then either @xmath340 or @xmath341 . for simplicity
, we assume that @xmath340
. then the jordan canonical form has @xmath342 jordan blocks of dimension @xmath343 . by ( c )
, we must also get that @xmath344 .
but @xmath345 , so @xmath346 .
contradict to our assumption .
so the betti numbers @xmath111 are unique .
the proof is complete .
[ remark lefschetz elment ] from the proof , we know that the above theorem is also valid if the canonical bundle of a smooth complete toric variety is a lefschetz element .
so an interesting question is when this condition holds .
the following example shows that there exist smooth complete varieties of dimension greater than @xmath26 with neither ample anonical nor ample anticanonical divisor also have the same properties as theorem [ main theorem 2 ] says .
let @xmath24 be the blow - up of @xmath347 at two fixed points under the action of the torus .
it is easy to check that neither the canonical nor the anticanonical divisor of @xmath24 is ample .
notice that @xmath24 is also a toric variety with fan @xmath188 generated by the columns of the following matrix @xmath348 denote the cone generated by the @xmath231-th column by @xmath349 .
then the maximal cones in @xmath188 are @xmath350 , @xmath351 , @xmath352 , @xmath353 , @xmath354 , @xmath355 , @xmath356 and @xmath357 .
let @xmath358 be the divisor corresponding to the one - dimensional cone generated by the @xmath359 column for @xmath360 .
then the chow group @xmath100 is generated by @xmath361 with the relations @xmath362 , @xmath363 , @xmath364 , @xmath365 , @xmath366 .
so @xmath367 is a basis of the chow group .
it is easy to get that @xmath368 .
one can check that the jordan canonical form of the coxeter transformation is @xmath369 .
( @xcite)[lemma betti number ] let @xmath188 be a complete simplicial fan in @xmath370
. then the betti numbers of @xmath218 are given by @xmath371 where @xmath372 is the @xmath373-th betti number .
then we have the following corollary of theorem [ main theorem 3 ] .
let @xmath188 be a complete smooth fan in @xmath374 .
if @xmath218 is a variety with ample canonical divisor or anticanonical divisor , then the jordan canonical form of the coxeter transformation has exactly @xmath375 many jordan blocks of dimension @xmath376 for any @xmath377 $ ] . by theorem [ main theorem 3 ] and lemma [ lemma betti number ] , we know that the jordan canonical form of the coxeter transformation has exactly @xmath378-\left[\sum_{i = k-1}^n(-1)^{i - k+1}\left(\begin{array}{cc}i\\k-1 \end{array } \right)|\sigma(n - i)|\right ] \quad(4\mbox{-}2)\ ] ] jordan blocks of dimension @xmath376 for any @xmath377 $ ] .
then ( 4 - 2 ) is equal to @xmath379-\left[\sum_{i = k}^n(-1)^{i - k+1}\left(\begin{array}{cc}i\\k-1 \end{array } \right)|\sigma(n - i)|\right]\hspace{2.5cm}\\ & -|\sigma(n - k+1)| \hspace{11.6cm}\\ = & \left\{\sum_{i = k}^n(-1)^{i - k}\left[\left(\begin{array}{cc}i\\k \end{array } \right)+ \left(\begin{array}{cc}i\\k-1 \end{array } \right ) \right]|\sigma(n - i)|\right\}-|\sigma(n - k+1)|\hspace{3.9cm}\\ = & \left [ \sum_{i = k}^n(-1)^{i - k}\left(\begin{array}{cc}i+1\\k \end{array } \right)|\sigma(n - i)| \right ] -|\sigma(n - k+1)|\hspace{5.9cm}\\ = & \sum_{i = k-1}^n(-1)^{i - k}\left(\begin{array}{cc}i+1\\k \end{array } \right)|\sigma(n - i)| . \hspace{8.3cm}\end{aligned}\ ] ] in the following , we compute the jordan canonical form of the coxeter transformation for lower dimensional toric fano varieties . from @xcite , we know that there are five toric fano surface , also called toric del pezzo surfaces .
we give a list of the jordan blocks of the coxeter transformations in the following .
@xmath380 from @xcite and @xcite , we know that there are 18 toric fano three - fold .
we give a list of the jordan blocks of the coxeter transformations in the following .
@xmath381 from @xcite , we know that there are 124 toric fano 4-folds .
+ now , we consider rational surfaces . by the classification of rational surfaces , for any given rational surface @xmath24
, there exists a sequence of blow - downs @xmath382 where @xmath383 is either @xmath158 or some hirzebruch surface @xmath157 .
we fix such a sequence .
denote @xmath384 the classes of the total transformations on @xmath24 of the exceptional divisors of the blow - ups @xmath111 .
if @xmath385 , we get @xmath386 as a basis of @xmath387 with the following intersection relations : @xmath388 if @xmath389 , we get @xmath390 as a basis of @xmath387 with @xmath391 @xmath392 see , e.g. , @xcite .
the following lemma characterizes the canonical divisor and the picard group of the blow - up .
( @xcite ) [ lemma blow up ] @xmath234
let @xmath393 be a dominating morphism between smooth varieties such that @xmath394 .
then @xmath395 where @xmath396 is the ramification divisor of @xmath397 .
@xmath241 let @xmath398 be a blow - up of @xmath24 at a smooth subvariety @xmath27 of codimension @xmath399 , and let @xmath400 be the exceptional divisor of the blow up .
then @xmath401 and @xmath402 [ lemma rational surface ] let @xmath24 be a rational surface .
then the canonical divisor is a lefschetz element if and only if the rank of the picard group is not @xmath25 .
there exists a sequence of blow - downs @xmath403 case 1 @xmath404 . by lemma [ lemma blow up ] , we know @xmath405 since the canonical divisor of @xmath158 is @xmath406 . from the intersection relations ( 4 - 3 )
, we get that @xmath407.\ ] ] but @xmath408 for @xmath409 , it follows that @xmath78 is a lefschetz element if and only if @xmath410 , which is equivalent to @xmath411 .
furthermore , @xmath386 is a basis of @xmath387 , so the rank of the picard group is @xmath412 and then the result follows .
case 2 @xmath413 .
similarly , since the canonical divisor of @xmath157 is @xmath414 , we know @xmath415 . from the intersection relations ( 4 - 4 ) , we get that @xmath416.\ ] ] it follows that @xmath78 is a lefschetz element if and only if @xmath417 .
furthermore , @xmath418 is a basis of @xmath387 , so the rank of the picard group is @xmath419 and then the result follows .
[ main theorem 2 ] let @xmath24 be a rational surface .
@xmath234 the number of the jordan blocks in jordan canonical form of the coxeter transformation of @xmath24 is equal to @xmath420 .
@xmath241 if @xmath421 , then the jordan canonical form of the coxeter transformation has a jordan block of dimension @xmath162 , the others are of dimension @xmath4 ; otherwise , the jordan canonical form of the coxeter transformation has two jordan blocks of dimension @xmath26 , and eight jordan blocks of dimension @xmath4 . by lemma [ lemma rational surface ] and remark [ remark lefschetz elment ] ,
we get ( 1 ) and the first statement of ( 2 ) . for the second statement of ( 2 ) , the matrix which represents the coxeter transformation is @xmath422 .
if @xmath24 is a blow - up of @xmath158 , by lemma [ lemma rational surface ] , we know @xmath423 and @xmath424 has a basis @xmath425 since @xmath426 .
it is easy to see that @xmath427 and @xmath428 are one - dimensional and generated by @xmath429 $ ] and @xmath430 $ ] respectively .
so @xmath429 , h , r_1,\dots , r_9,[pt]$ ] is a basis of @xmath100 .
we denote by @xmath99 the linear transformation induced by @xmath431 on the @xmath100 .
it is easy to see the jordan blocks of @xmath99 has the same dimension as the jordan blocks of @xmath94 . by lemma [ lemma blow up ] , we know @xmath432 since the canonical divisor of @xmath158 is @xmath406 . from the intersection relations
, we can get @xmath433 the matrix of @xmath99 under the above basis is as follows @xmath434 it is easy to check that this matrix has two jordan blocks of dimension @xmath26 , and eight jordan blocks of dimension @xmath4 .
if @xmath24 is a blow - up of @xmath157 , similarly , we get that @xmath435 and @xmath436 $ ] is a basis of @xmath100 . also by lemma
[ lemma blow up ] , since the canonical divisor of @xmath157 is @xmath414 , we know @xmath437 . from the intersection relations , we get @xmath438 the matrix of @xmath99 under the above basis is as follows @xmath439 it is easy to check that this matrix also has two jordan blocks of dimension @xmath26 , and eight jordan blocks of dimension @xmath4 .
so ( 2 ) is valid .
[ corollary derived equivalent ] if a rational surface or a smooth complete toric variety @xmath24 with ample canonical or anticanonical divisor is derived equivalent to a finite dimensional algebra @xmath9 , then the betti numbers of @xmath24 are determined by the matrix @xmath11 . since @xmath24 is derived equivalent to @xmath9 , we know that the jordan canonical form of the coxeter transformation @xmath440 is the same as that of @xmath11 . from theorem [ main theorem 3 ] and theorem [ main theorem 2 ] ,
we know that the rank of @xmath112 is determined by the matrix @xmath11 . for @xmath24 is a rational surface or a smooth complete toric variety , we know that @xmath441 is the @xmath255-th betti number .
so the betti numbers of @xmath24 are determined by the matrix @xmath11 . at the end of this section
, we continue to consider the examples in section 3 .
[ example projective space2 ] from theorem [ main theorem 3 ] , the jordan canonical form of the coxeter transformation of @xmath23 has only one jordan block .
let @xmath9 be the algebra in example [ example projective space ] .
then the jordan canonical form of the coxeter transformation @xmath133 has only one jordan block .
let @xmath9 be the algebra in example [ example rational surface ] .
then the jordan canonical form of the coxeter transformation @xmath133 has @xmath412 jordan blocks with only one of dimension @xmath162 , and the others are of dimension @xmath4 if @xmath411 .
otherwise , @xmath133 has @xmath412 jordan blocks with two of dimension @xmath26 , and the others are of dimension @xmath4 if @xmath423 .
to construct graded local frobenius algebras over an algebraically closed field @xmath32 , it is important to find out jordan canonical forms of tensors product of square matrices , see @xcite .
in fact , it is known that any graded local frobenius algebra is of the form of @xmath442 , where @xmath443 is a finite dimensional @xmath32-vector space , @xmath444 , and @xmath445 a @xmath32-linear map satisfying several conditions .
further , if we decompose @xmath443 as @xmath446 , where @xmath447 are invariant subspaces of @xmath448 and @xmath449 , then the conditions of @xmath450 can be described in terms of each @xmath451
. then we have to consider a jordan canonical form of @xmath452 as an element in @xmath453 .
+ from tensor product of matrices , we can define the operator @xmath454 as follows @xmath455 ( @xcite)[lemma jordan block ] the following holds for integers @xmath456 and @xmath457 : @xmath458 ) & \mbox { if } \alpha=0=\beta\\ j(0,s)^{\oplus t } & \mbox { if } \alpha=0\neq\beta\\ j(0,t)^{\oplus s}&\mbox { if } \alpha\neq0=\beta\\ \bigoplus^s_{i=1}j(\alpha\beta , s+t+1 - 2i)&\mbox { if } \alpha\neq0\neq\beta \end{array}\right . .\ ] ] because @xmath24 and @xmath27 are two smooth complete toric varieties , from lemma [ lemma toric variety ] and @xmath465 , @xmath466 , we know @xmath467 and @xmath468 . from the knneth formula , we have @xmath469 but @xmath30 is a smooth complete toric variety , so @xmath470 .
thus ( 1 ) is valid .
( b)if the jordan canonical form of the coxeter transformations of @xmath24 and @xmath27 are @xmath28 and @xmath29 respectively , then the sizes and number of jordan blocks in the jordan canonical form of the coxeter transformation of @xmath30 are the same as that of @xmath31 .
denote @xmath474 and @xmath475 .
let @xmath79 and @xmath476 be the canonical bundles of @xmath24 and @xmath27 respectively . by lemma [ lemma variety product ] , we know @xmath477 let @xmath440 and @xmath478 be the coxeter transformations of @xmath24 and @xmath27 respectively
. then @xmath440 and @xmath478 are induced by the functors @xmath479 $ ] and @xmath480 $ ] respectively .
furthermore , @xmath481 is the coxeter transformation of @xmath30 induced by @xmath482 $ ] . also let @xmath483 be the transformation induced by @xmath484 , and the others are similar .
for any smooth complete toric variety @xmath102 , we know @xmath485 as rings , so for any @xmath486 , we also use @xmath149 to denote the corresponding element in @xmath487 .
we also use @xmath488 and @xmath489 to denote the induced transformation on @xmath487 . from the knneth formula
, we have an isomorphism of rings @xmath490 which maps @xmath491 to the cochain cross product @xmath492 .
the ring structure on @xmath493 is defined to be @xmath494 if @xmath495 and @xmath496 .
notice that @xmath497 acts on @xmath498 as @xmath499 and @xmath500 .
so the induced transformation @xmath497 acts on @xmath493 as @xmath501 . for any @xmath502
, we get the induced transformation @xmath497 on @xmath493 as follows : @xmath503 the third equality holds since for any smooth complete toric variety @xmath102 , if @xmath504 , then @xmath231 is even .
therefore @xmath505 is the coxeter transformation of @xmath30 under the isomorphisms @xmath506 .
but @xmath507 has the same jordan canonical form as @xmath508 except the eigenvalues .
the proof is complete .
[ proposition jordan block ] let @xmath510 and @xmath511 be @xmath159 integers
. then @xmath512 } ( j(\alpha , n+1 - 2k)^{\oplus ( c_k - c_{k-1})}),\ ] ] where @xmath513 and @xmath514 s are given in formula @xmath515 and @xmath516 . by lemma [ lemma jordan block ] , do it inductively , we can see that @xmath517 is a finite direct sum of some jordan blocks and their eigenvalues are equal to @xmath149 .
since the jordan canonical form of the tensor product of these jordan blocks does not depend on @xmath518 if they are all non - zero , we only need do it for @xmath519 or @xmath142 for @xmath520 . by example
[ example projective space2 ] , we know that the jordan canonical form of the coxeter transformation of @xmath521 has only one jordan block of dimension @xmath522 with eigenvalues @xmath523 . from proposition [ theorem jordan block ]
, we also know that the jordan canonical form of the coxeter transformation of @xmath524 is the same as @xmath525 except the eigenvalues . from the knneth formula ,
we know that the @xmath373-th betti number of @xmath24 is @xmath514 where @xmath514 is as the formula ( 5 - 2 ) shows . by theorem [ main theorem 3 ] , we get that @xmath517 has @xmath526 jordan blocks of dimension @xmath527 for @xmath377 $ ] since @xmath528 is the dimension of @xmath24 .
let @xmath9 and @xmath70 be two finite dimensional algebras with finite global diemension over algebraically closed field @xmath32 .
then the relation between the jordan canonical form of the coxeter transformations of @xmath9 , @xmath70 and @xmath529 is the same as proposition [ theorem jordan block ] describes . in particular , @xmath530 [ proposition deireved equivalent ] let @xmath24 , @xmath27 be two smooth projective varieties over @xmath32 and @xmath9 , @xmath70 be two finite dimensional algebras with finite global dimensions also over @xmath32 . if @xmath531 and @xmath532 induced by tilting bundles , then @xmath533 .
let @xmath400 and @xmath534 be the tilting bundles in @xmath47 and @xmath535 respectively which induce @xmath531 and @xmath532
. then @xmath536 and @xmath537 for all @xmath128 , and @xmath538 , @xmath539 . by lemma 3.4.1 in @xcite
, we get that @xmath540 generates @xmath541 . applying the knneth formula for locally free sheaves on algebraic varieties , we get @xmath542 since @xmath9 and @xmath70 has finite global dimensions , so is @xmath543 . from @xcite , we know that @xmath547 is triangulated equivalent to @xmath130 , where @xmath548 and @xmath549_{a_2 } \ar@<0.5ex>[r]^{a_1 } & \circ}$ ] .
so @xmath550 is triangulated equivalent to @xmath551 , where @xmath552 and @xmath553 is the following quiver and @xmath554 .
@xmath555_{c_2 } \ar@<0.5ex>[rr]^{c_1 } & & \circ \\ \\ \circ \ar@<-0.5ex>[uu]_{a_2 } \ar@<0.5ex>[uu]^{a_1 } \ar@<-0.5ex>[rr]_{b_1 } \ar@<0.5ex>[rr]^{b_2 } & & \circ \ar@<-0.5ex>[uu]_{d_1 } \ar@<0.5ex>[uu]^{d_2}}\ ] ] aaa m. auslander , i. reiten and s. o. smal , _ representation theory of artin algebras _ , cambridge studies in advanced math . *
36 * , cambridge univ . press , 1995 .
v. v. batyrev , toroidal fano 3-folds , _ math .
ussr - izv .
_ * 19 * ( 1982 ) : 1325 . v. v. batyrev , on the classification of toric fano 4-folds , _ j. math .
( new york ) _ * 94 * , no . 1 ( 1999 ) : 10211050 .
a. a. beilinson , coherent sheaves on @xmath23 and problems of linear algebra , _ funkt .
* 12 * ( 1979 ) : 214216 .
a. a. beilinson , j. bernstein and p. deligne , faisceaux pervers , in analysis and topology on singular spaces , i(luminy,1981 ) , _ astrisque * 100 * soc .
france , paris _ , ( 1982 ) : 5171
. i. n. bernstein , i. m. gelfand and v. a. ponomarev , coxeter functors , and gabriel s theorem , _ uspehi mat .
nauk _ 28 , no . 2 ( 1973 )
english translation : _ russian math . surveys _ * 28 * , no . 2 ( 1973 ) : 1732
. a. i. bondal , representation of associative algebras and coherent sheaves , _ math .
ussr izvestiya _ * 34 * ( 1990 ) : 2342 .
a. i. bondal and m. m. kapranov , representable functors , serre functors , and mutations , _ izv . akad .
nauk sssr ser
_ * 53 * ( 1989 ) : 11831205 , russian ; english translation : _ math .
ussr izv . _
* 35 * ( 1990 ) : 519541 .
a. i. bondal and d. o. orlov , reconstruction of a variety from the derived category and groups of autoquivalences , _ compositio math .
_ * 125 * ( 2001 ) : 327344 .
a. i. bondal and m. van den bergh , generators and representability of functors in commutative and noncommutative geometry , _ moscow math
. j. _ * 3 * ( 2003 ) : 136 .
t. bridgeland , derived categories of coherent sheaves , in _ proceedings of the international congress of matematicians(icm ) , madrid , spain , august 22 - 30 , 2006 .
volume ii : invited lectures _ : 563582 .
zrich : european mathematical society ( ems ) , 2006 .
w. bruns and j. gubeladze , _ polytopes , rings and k - theory _ , new york : springer , 2009 .
a. j. coleman , the betti numbers of the simple lie groups , _ canad . j. math .
_ * 10 * ( 1958 ) : 349356 .
a. j. coleman , killing and the coxeter transformation of kac - moody algebras , _ invent .
_ * 95 * , no . 3 ( 1989 ) : 447477 .
l. costa and r. m. mir - roig , tilting sheaves on toric varieties , _ math .
z. _ * 248 * ( 2004 ) : 849865 .
d. cox , j. little and h. schenck .
_ toric varieties .
_ volume * 124 * of graduate studies in mathematics .
american mathematical society , providence , ri , 2011 .
h. s. coxeter , the product of the generators of a finite group generated by reflections , _ duke math . j. _ * 18 * ( 1951 ) : 765782 .
v. dlab and c. m. ringel , indecomposable representations of graphs and algebras , _ mem .
* 6 * , no . 173 ( 1976 ) .
v. dlab and c. m. ringel , eigenvalues of coxeter transformations and the gelfand - kirillov dimension of the preprojective algebras , _ proc .
soc . _ * 83 * ( 1981 ) : 228232 .
g. ewald , on the classification of toric varieties , _
discrete comput.geom._ * 3 * ( 1988 ) : 4954
. w. fulton , introduction to toric varieties , _ ann . of math .
studies , princeton _ * 131 * ( 1993 ) .
w. fulton , _ intersection theory .
_ second ed .
berlin : springer , 1998 . m. goresky and r. macpherson , intersection homology theory ii , _ invent . math .
_ * 72 * ( 1983 ) : 77129 .
d. happel , _ triangulated categories in representation theory of finite dimensional algebras , _ london math . soc .
lecture notes ser .
* 119 * , cambridege univ . press , 1988 .
d. happel , auslander - reiten triangles in derived categories of finite - dimensional algebras , _ proc .
* 112 * ( 1991 ) : 641648 .
r. hartshorne , _ algebraic geometry .
_ number * 52 * in graduate texts in mathematics , new york : springer , 1977 .
l. hille and m. perling , a counterexample to king s conjecture , _ comp . math _
* 142 * ( 2006 ) : 15071521 .
l. hille and m. perling , exceptional sequences of invertible sheaves on rational surfaces , _ compos .
_ * 147 * ( 2011 ) : 12301280 .
a. king , tilting bundles on some rational surfaces , unpublished manuscript , 1997 .
h. lenzing and j. a. de la pea , spectral analysis of finite dimensional algebras and singularities , _ in trends in representation theory of algebras and related topics . _
( ems ser .
congr . rep .
zrich , 2008 ) : 541588 .
a. martsinkovsky and a. vlassov , the representation rings of @xmath556 $ ] , preprint , http://www.math.neu.edu/martsinkovsky/mathindex.html .
t. oda , _ convex bodies and algebraic geometry . _
volume * 15 * of ergebnisse der mathematik und ihrer grenzgebiete ( 3 ) .
berlin : springer , 1988 . translated from the japanese .
j. a. de la pea , coxeter transformations and the representation theory of algebras , _ in finite dimensional algebras and related topics . _
( nato asi series c. math .
sci . * 424 * , kluwer academic publishers , dordrecht , 1994 ) : 223253 . j. a. de la pea and m. takane , spectral properties of coxeter transformations and applications , _ archiv math .
( basel ) _ * 55 * ( 1990 ) : 120134 .
m. perling , some quiver describing the derived category on the toric del pezzo surfaces , preprint , http://www.mathematik.uni-kl.de/ perling / quivers.pdf .
m. perling , examples for exceptional sequences of invertible sheaves on rational surfaces , _ geometric methods in representation theory ii , seminaire et congres _ * 25 * ( 2010 ) : 369389 .
j. rickard , morita theory for derived categories , _
j. london math .
* 39 * ( 1989 ) : 436456 .
j. rickard , derived equivalences as derived functors , _ j. london math .
* 43 * ( 1991 ) : 3748 . c. m. ringel , the spectral radius of the coxeter transformations for a generalized cartan matrix , _ math .
ann . _ * 300 * , no . 2 ( 1994 ) : 331339 .
h. sato , toward the classification of higher - dimensional toric fano varieties , _ tohoku math .
j. _ * 52 * , no.3 ( 2000 ) : 383413 .
r. b. stekolshchik , study of coxeter transformations and regularity problems associated with representation theory of graphs , _
ph.d.thesis , kiev , _ 95p , 1985 ,
r. b. stekolshchik , _ notes on coxeter transformations and the mckay correspondence . _
berlin : springer monographs in mathematics , 2008 .
v. f. subbotin and r. b. stekolshchik , the jordan form of the coxeter transformation , and applications to representations of finite graphs , _ funkcional .
i priloen . _
* 12 * , no . 1 ( 1978 ) : 8485 .
english translation : _ functional anal .
* 12 * , no . 1 ( 1978 ) : 6768 .
t. wakamatsu , on graded frobenius algebras , _ j. algebra . _ * 267 * ( 2003 ) : 377395 .
k. watanabe and m. watanabe , the classification of fano 3-folds with torus embeddings , _
tokyo j. math .
_ * 5 * ( 1982 ) : 3748 . | in this paper , we study the coxeter transformations of the derived categories of coherent sheaves on smooth complete varieties .
we first obtain that if the rank of the grothendieck group is finite , say @xmath0 , then its characteristic polynomials is @xmath1 , where @xmath2 is dimension of the variety .
we then study the jordan canonical forms of the coxeter transformations for rational surfaces , smooth complete toric varieties with ample canonical or anticanonical bundles , and prove that their jordan canonical forms can determine and be determined by the betti numbers of these varieties . as an application , we compute the jordan canonical forms of tensor products of matrices .
xy [ section ] [ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]case [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]construction [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary p corresponding author ] |
the anisotropy of stellar orbits in elliptical galaxies can be estimated from the temperature of the hot interstellar gas through which they move .
these two quite different galactic attributes are intimately related by the jeans equation for the stars and the condition for hydrostatic equilibrium in the gas .
this simple relationship is valuable since both the anisotropy and gas temperature are difficult to extract from optical and x - ray observations , respectively .
many massive elliptical galaxies are nearly spherical ( merritt & trembly 1996 ) and slowly rotating , but their stellar velocity ellipsoids are not in general isotropic .
the non - spherical nature of the stellar velocity dispersion is represented by the parameter @xmath6 where @xmath7 is the radial stellar velocity dispersion and @xmath8 is the dispersion in a transverse direction , i.e. @xmath9 . if the orbits are predominantly radial , @xmath10 , the line of sight velocity profile becomes more strongly peaked than a gaussian profile with increasing projected radius @xmath11 ; if the orbits are mostly tangential , @xmath12 , the profile is more flat - topped and becomes broader with increasing radius .
both @xmath0 and the galactic potential @xmath13 can be determined from optical observations of the line of sight velocity dispersion @xmath14 , the optical surface brightness distribution and the deviation of the stellar line profiles from a gaussian as expressed by the line - symmetric coefficient @xmath15 in a gauss - hermite expansion ( e , g , van der marel & franx 1993 ) .
stellar line profiles observed in most massive e galaxies indicate a preference for radial orbits with @xmath16 ( bender , saglia & gerhard 1994 ; gerhard et al . 1998 ; saglia et al . 2000 ) .
accurate @xmath0 profiles from optical data require high quality data and considerable care in reduction and analysis .
the radially anisotropic nature of stellar orbits in luminous elliptical galaxies provides important and otherwise unavailable information about the merger history of these galaxies ( naab , burkert & hernquist 1999 ) , so an improved or independent determination of @xmath0 would be desirable .
similarly , the radial variation of the hot gas temperature @xmath17 k in elliptical galaxies depends on the spatial and spectral resolutions of x - ray detectors and the accuracy of the three - dimensional decomposition that converts @xmath18 as a function of projected radius @xmath11 to physical radius @xmath19 .
in addition , it is often unclear whether the hot gas at any radius has a single temperature or a multitude of temperatures as might be expected if the gas were cooling .
it is unclear if the gas cools at all .
for example , high resolution x - ray spectra with xmm - newton of the large e galaxy ngc 4636 fail to show emission lines expected from gas at intermediate temperatures , such as the 0.574 kev ovii line ( xu et al .
2002 ) , suggesting that the gas is not cooling below @xmath20 as in classical cooling flows .
however , bregman , miller & irwin ( 2001 ) detected the ovi 1032,1038 doublet in ngc 4636 emitted from gas at @xmath21 k , implying that cooling near the expected rate may occur after all .
as an additional source of confusion , the observed x - ray spectra can often be significantly improved by assuming two quite different discrete temperatures at each galactic radius ( e.g. buote 2002 for ngc 1399 ; buote et al .
2002a and tamura et al .
2003 for ngc 5044 ) .
while the origin and physical nature of gas with only two ( or a limited continuum of ) temperature phases may be conceptually problematical , it may be related to certain heating mechanisms .
nevertheless , there are good reasons for suspecting the accuracy or interpretation of gas temperature profiles @xmath1 in elliptical galaxies . in the following discussion
we illustrate the close relationship between the radial variations of @xmath0 and @xmath1 using data for the well - observed giant elliptical ngc 4472 . while currently available data for this galaxy are adequate to illustrate this relationship , more accurate optical and x - ray data will be necessary to verify @xmath0 or @xmath1 with confidence .
we recognize that ngc 4472 is not perfectly spherical ; its optical image has been variously classified as e1 or e2 , however , optical determinations of @xmath0 for ngc 4472 and other nearly - spherical elliptical galaxies have been made under the assumption of spherical symmetry ( e.g. kronawitter et al .
indeed the very concept of the @xmath22 anisotropy parameter is only valid in spherical geometry .
similarly , the gas temperature @xmath1 determined from x - ray observations of ngc 4472 and other nearly spherical elliptical galaxies are based on the assumption of spherical symmetry .
one justification for this is that the stellar gravitational potential that confines the stars and hot gas within the half - light radius is always more spherical than the underlying stellar luminosity distribution ( e.g. brighenti & mathews 1996 ) .
our discussion of the close relationship between stellar anisotropy and gas temperature is a thinly - veiled request for observers to concentrate on those luminous e galaxies for which the most accurate x - ray and optical data can be acquired .
the jeans equation for the radial stellar velocity dispersion @xmath7 is @xmath23 where @xmath3 is the stellar luminosity density corresponding to the stellar velocity dispersions @xmath7 and @xmath8 .
the equation for hydrostatic equilibrium in the hot interstellar gas is @xmath24 here @xmath25 is the isothermal sound speed , @xmath26 is the proton mass and @xmath27 is the molecular weight .
the uniformity of the stellar mass to light ratio within the effective radius of ngc 4472 is excellent evidence that the hot gas is in hydrostatic equilibrium ( brighenti & mathews 1997 ) .
both equations contain the same gravitational term @xmath28 . eliminating this term results in an equation explicitly independent of the confining mass : @xmath29 = { d c^2 \over d r } + { c^2 \over r } \left [ { d \log \rho \over d \log r } \right].\ ] ]
this equation has been discussed previously , particularly in the context of the anisotropic orbits of galaxies in rich clusters that contain hot gas ( e.g. fabricant et al .
in addition to hydrostatic equilibrium , the validity of this equation depends on the assumption that the gas pressure exceeds that of any other interstellar pressure such as turbulence , cosmic rays , etc .
if @xmath0 and @xmath1 ( as well as @xmath30 and @xmath31 ) are known from observations , then equation ( 3 ) can be solved for @xmath32 and the line of sight velocity dispersion is then found from @xmath33 \ell_*(r ) ( r^2 - r^2)^{-1/2 } r dr \over \int_r^{r_t } \ell_*(r ) ( r^2 - r^2)^{-1/2 } r dr } \ ] ] ( e.g. binney & mamon 1982 ) . in this equation @xmath11
is the projected radius and @xmath34 is some radius significantly beyond the limit of the available data ( we assume @xmath35 ) .
if @xmath0 is known , equation ( 3 ) can be solved for @xmath32 for various parameterized temperature profiles @xmath1 until @xmath14 from equation ( 4 ) agrees with observed velocity dispersions . by this
means it is possible to derive the gas temperature profile @xmath1 from a knowledge of @xmath0 .
conversely , if the gas temperature @xmath1 is securely known from x - ray observations , then @xmath0 can be determined by requiring that solutions @xmath36 of equations ( 3 ) and ( 4 ) pass through the observations .
this close relationship between @xmath0 and @xmath1 is possible and valuable because @xmath31 , @xmath30 and @xmath36 are much easier to determine from observations than either @xmath0 or @xmath1 . except for the very central region ,
the stellar brightness in ngc 4472 is well - fit with a de vaucouleurs profile .
the stellar b - band luminosity density @xmath37 is determined by the total luminosity @xmath38 @xmath39 and the effective radius : @xmath40 kpc , assuming a distance @xmath41 mpc . however , within a small `` break '' radius , @xmath42 pc , the stellar luminosity flattens to @xmath43 ( gebhardt et al . 1996 ; faber et al .
1997 ) . in figure 1
we compare the ( un - normalized ) stellar luminosity density @xmath31 in ngc 4472 with the hot gas density profile @xmath30 determined from x - ray observations .
the remarkable proportionally @xmath44 was first discovered by trinchieri , fabbiano & canizares ( 1986 ) .
x - ray observations of the hot gas temperature in the central region of ngc 4472 are shown in figure 2 .
only three observations of the gas temperature are available in the region within 10 kpc where @xmath22 has been determined from optical observations .
the filled circles show rosat deprojected gas temperature observations from buote ( 2000 ) . the linear temperature profile defined by these two temperature observations @xmath45 is shown in figure 2 with a solid line . from chandra data
soldatenkov , vikhlinin & pavlinsky ( 2003 ) have recently provided an additional temperature measurement of @xmath46 at radius 1 = 0.0824 kpc that is shown with a filled square in figure 2 .
the linear fit to the buote data passes within the error bars of the soldatenkov et al .
observation , so all three observations are consistent with a linear temperature profile within the effective radius @xmath47 kpc .
although the x - ray data is sparse in the region of interest , it is in perfect qualitative agreement with linear gas temperature profiles observed in other bright elliptical galaxies in @xmath48 ( fig . 1 of brighenti & mathews 1997 ) .
the open circles in figure 3 show observations of the line of sight stellar velocity dispersion in ngc 4472 ( from bender et al .
1994 ) that were used by kronawitter et al .
( 2000 ) to determine @xmath0 from optical observations alone .
the filled circles in figure 3 are the data of fried & illingworth ( 1994 ) .
both sets of observations lie entirely within the effective ( half - light ) radius , @xmath49 kpc .
for some reason the normalization is systematically different in the two sets of optical observations .
the fried - illingworth data have smaller error bars , less overall scatter and resemble in overall form the velocity dispersion profiles @xmath14 of other well - observed ellipticals [ e.g. gerhard et al . ( 1998 ) for ngc 6703 ] .
the observational uncertainties apparent in the data sets shown in figure 3 make it difficult to determine with confidence any radial dependence of the stellar anisotropy .
consequently , we assume that the orbital anisotropy @xmath22 in equation ( 3 ) is constant over the range @xmath50 kpc . in figure 3
we compare solutions of equations ( 3 ) and ( 4 ) with the observed line of sight stellar velocity dispersion in ngc 4472 .
each solution curve @xmath14 is based on the linear gas temperature profile in figure 2 and a particular value of @xmath22 .
more information about the solutions shown in figure 3 can be found in table 1 .
of the three solid lines designed to fit the fried - illingworth data , solution 2 with initial @xmath51 km s@xmath52 and @xmath53 fits the data best .
solutions 1 and 3 show additional marginally acceptable fits although they fall respectively below and above the fried - illingworth data at smaller projected radii @xmath11 .
evidently @xmath54 is the most likely anisotropy based on the gas temperature observations . in some of our solutions of equation ( 3 ) @xmath55 becomes unphysically negative at a radius @xmath56 less than the outer limit of our integration in equation ( 4 ) , @xmath57 ; for these solutions we replace @xmath34 with @xmath56 . in solutions 4 - 6 , shown with dashed lines in figure 3 , we find that the anisotropy parameter must have values @xmath58 to be in approximate agreement with the data of bender et al .
once @xmath59 is selected for each of these dashed line solutions in figure 3 , the vertical position of the solution is extremely sensitive to @xmath22 , i.e. , @xmath22 must be specified to one part in 1000 to pass through the center of the bender et al . data .
one may doubt if nature can be so discriminating
. nevertheless , for either set of optical data , the corresponding values of @xmath22 are about twice as large as the @xmath60 found by kronawitter et al .
( 2000 ) from the bender et al .
data . the preferred orbital anisotropy in ngc 4472 determined by kronawitter et al .
( 2000 ) from the bender et al .
observations can be accurately represented with @xmath61 which peaks at @xmath62 kpc where @xmath63 . in this section
we assume that @xmath64 is correct and solve equation ( 3 ) for @xmath65 and @xmath1 . as before we consider only linear temperature variations defined by two temperatures @xmath66 and @xmath67 evaluated respectively at @xmath68 kpc and @xmath69 kpc . we seek values of @xmath66 and @xmath67 for which solutions @xmath14 of equation ( 4 ) pass through or near the bender et al . optical data .
we begin with a set of solutions for which @xmath70 and the temperature profile is constrained to agree with the observed linear @xmath1 in figure 2 .
when @xmath31 and @xmath30 are taken from figure 1 as before , the only remaining undetermined parameter required to solve equations ( 3 ) and ( 4 ) is @xmath59 . the four short dashed lines in figure 4 show the line of sight velocity dispersion profile @xmath14 for @xmath71 km s@xmath52 , corresponding to solutions 7 - 10 in table 1 . since we are using the kronawitter et al .
@xmath64 , it is appropriate to compare these solutions only with the data of bender et al . from which @xmath64 was derived .
of these , the best - fitting solution is 9 , but the overall flat or positive slope [ @xmath72 at larger @xmath11 does not resemble the @xmath14 profiles of most well - observed e galaxies . since none of these solutions is a particularly good fit to the bender et al .
data , it appears that @xmath64 found from optical data alone is somewhat inconsistent with the observed hot gas temperature profile . to explore alternate x - ray temperature profiles that may agree better with the @xmath14 of bender et al . , we consider solution 11 in which the hot gas is isothermal at the mean value of the two buote temperatures in figure 2 , @xmath73 k. this solution , shown with a dotted line in figure 4 , decreases slightly with @xmath11 , in better agreement with data in @xmath74 kpc
. solutions with negative @xmath75 may fit even better , but all known x - ray observations for e galaxies have positive @xmath75 for @xmath76 .
finally , we consider solution 12 based on a temperature profile having the same slope as the observations in figure 2 , but with uniformly higher temperatures , @xmath77 k and @xmath78 k. in solution 12 ( solid line in figure 4 ) the radially decreasing line of sight dispersion @xmath14 follows the bender et al .
data better than any other curve in figure 4 , possibly suggesting that the gas temperature should be higher than the observations in figure 2 . therefore ,
either the gas temperatures from x - ray observations in figure 2 are too low or @xmath64 and the @xmath14 data from which it was derived are inaccurate .
we have shown that equations ( 3 ) and ( 4 ) establish a useful correspondence between the anisotropy of stellar orbits in elliptical galaxies @xmath0 and the temperature of the hot interstellar gas @xmath1 . in practice ,
the @xmath79 relation is based on the assumption of hydrostatic equilibrium for the hot gas and requires reliable observations of the stellar and gas densities and the stellar line of sight velocity dispersions .
this connection between @xmath22 and @xmath18 is possible only if the galaxies are nearly spherical , as emphasized by maggorian & ballantyne ( 2002 ) .
if the gas temperature profile @xmath1 in an e galaxy is securely known from x - ray observations , @xmath0 can be determined from the observed stellar velocity dispersion using equations ( 3 ) and ( 4 ) . in principle , this method of determining @xmath0 could be done at galactic radii much larger than @xmath80 where stellar observations of @xmath15 may be difficult but where observations of @xmath14 and x - ray gas temperatures are generally reliable .
conversely , if @xmath0 can be observed with high precision from optical observations , the hot gas temperature @xmath1 can be found from equations ( 3 ) and ( 4 ) by using the gas density from x - ray observations .
this means of determining @xmath1 may be particularly valuable in regions of the hot gas atmospheres where x - ray spectra indicate a two - temperature or limited multi - temperature thermal structure ( e.g. buote 2002b ) that may complicate the determination of the average local gas temperature . as a representative example , we have used the @xmath22@xmath18 relation to determine @xmath0 in the bright elliptical galaxy ngc 4472 from the hot gas temperature profile @xmath1 found from x - ray observations .
the resulting anisotropy , @xmath81 , indicates that the stellar orbits are considerably more radially biased than implied by the smaller @xmath82 determined from optical observations alone ( kronawitter et al . 2000 ) .
these large @xmath22 for ngc 4472 are based on the fried - illingworth stellar velocity dispersions @xmath14 .
somewhat lower anisotropies , @xmath58 are consistent with the velocity dispersion observations of bender et al . , but these @xmath22 still exceed the @xmath82 found optically by kronawitter et al .
( 2000 ) . alternatively ,
if the kronawitter et al . @xmath70 and the bender et al .
( 1994 ) data are correct , then the gas temperature in ngc 4472 may be higher than previously thought from x - ray observations and possibly have a different slope @xmath75 .
these conclusions apply only to the limited region @xmath76 for which @xmath14 data currently exist for ngc 4472 .
it is clear , at least for ngc 4472 , that currently available observations are not sufficiently accurate to establish either @xmath0 or @xmath1 with complete confidence .
accurate optical observations of @xmath14 and stellar line profiles can be observed with 8 - 10 meter telescopes to several @xmath80 and @xmath1 is also often known at these large radii .
in addition , it may be possible to determine gas temperature profiles with more accurate or prolonged x - ray observations than those in figure 2 .
the @xmath0 determined from x - ray emission can serve to calibrate the accuracy of purely optical determinations or to provide a test for hydrostatic equilibrium . to accomplish this
, it would be desirable to obtain high quality stellar dispersion and gas temperature observations of those galaxies such as ngc 4472 , ngc 4649 , ngc 1399 , m87 , etc .
that are luminous at both optical and x - ray frequencies .
.4 in studies of the evolution of hot gas in elliptical galaxies at uc santa cruz are supported by nasa grants nag 5 - 8409 nag 5 - 13275 and nsf grants ast-9802994 and ast-0098351 for which we are very grateful .
bender , r. , saglia , r. p. , & gerhard , o. 1994 , mnras 269 , 785 binney , j. j. & mamon , g. a. 1982 , mnras , 200 , 361 bregman , j. n. , miller , e. d. & irwin , j. a. 2001 , apj 553 , l125 brighenti , f. & mathews , w. g. 1996 , apj , 470 , 747 brighenti , f. & mathews , w. g. 1997 , apj , 486 , l83 buote , d. a. 2000 , apj , 539 , 172 buote , d. a. 2002 , apj 574 , l135 buote , d. a. 2002b , apj ( in press ) ( astro - ph/0206422 ) buote , d. a. et al . 2002a , apj ( submitted ) ( astro - ph/0205362 ) buote , d. a. , jeltema , t. e. , canizares , c. r. , garmire , g. p. 2002b
, apj , 577 , 183 .
buote , d. a. 2000 , apj 539 , 172 faber , s. m. et al .
1997 , aj , 114 , 1771 fabricant , d. g. , kent , s. m. & kurtz , m. j. 1989 , apj 336 , 77 fried , j. w. & illingworth , g. d. 1994 , aj 107 , 992 gebhardt , k. et al .
1996 , aj , 112 , 105 gerhard , o. , jeske , g. , saglia , r. p. & bender , r. 1998 , mnras , 295 , 197 irwin , j. a. & sarazin , c. l. 1996 , apj 471 , 663 kronawitter , a. , saglia , r. p. , gerhard , o. & bender , r. 2000 , a&as 144 , 53 loewenstein , m. , mushotzky , r. f. , angelini , l. , arnaud , k. a. & quataert , e. 2001 , apj 555 , l21 magorrian j. & ballantyne , d. 2002 , mnras 322 , 702 merritt d. & tremblay b. 1996 , aj 111 , 2243 naab , t. , brudert , a. & hernquist , l. 1999 , apj 523 , l133 saglia , r. p. , kronawitter , a. , gerhard , o. & bender , r. 2000 , aj , 119 , 153 a&a 343 , 420 soldatenkov , d. a. , vikhlinin , a. a. & pavlinsky , m. n. 2003 , astronomy letters , 2003 , 29 , 298 tamura , t. , kaastra , j. s. , makishima , k. & takahashi , i. 2003 , a&a ( submitted ) ( astro - ph/0211603 ) trinchieri , g. , fabbiano , g. , & canizares , c. r. 1986 , apj , 310 , 637 van der marel , r. p. & franx , m. 1998 , apj , 407 , 525 xu , h. , kahn , s. m. , peterson , j. r. , behar , e. , paerels , f. b. s. , mushotzky , r. f. , jernigan , j. g. , brinkman , a. c. & makishima , k. 2002 , apj , 579 , 600 rlllccccc @xmath83 & & & & & & & & 1 & 0.6333 & .906 & 1.368 & 390 . & 249 . & 25.6 & 282 . & 172 . 2 & 0.7125 & .906 & 1.368 & 430 . & 253 . & 23.7 & 293 . & 162 . 3 & 0.7775 & .906 & 1.368 & 470 .
@xmath84 & 306 . & 171 . 4 & 0.4766 & .906 & 1.368 & 330 . & 276 .
& @xmath84 & 306 .
& 295 . 5 & 0.63 & .906 & 1.368 & 390 . & 290 . & @xmath84 & 318 .
& 270 . 6 & 0.773 & .906 & 1.368 & 470 . & 326 . &
@xmath84 & 334 . & 245 .
& & & & & & & & @xmath85 & & & & & & & & 7 & @xmath86 & .906 & 1.368 & 269.10 & 205 . &
@xmath84 & 247 . & 210 .
8 & @xmath86 & .906 & 1.368 & 269.15 & 219 .
& @xmath84 & 272 . & 281 .
9 & @xmath86 & .906 & 1.368 & 269.20 & 232 . &
@xmath84 & 295 . & 337 .
10 & @xmath86 & .906 & 1.368 & 269.25 & 244 .
& @xmath84 & 317 . & 385 .
11 & @xmath86&1.137 & 1.137 & 330.38 & 254 . &
@xmath84 & 315 . & 303 .
12 & @xmath86&1.306 & 1.768 & 332.9 & 258 . &
@xmath84 & 316 . & 295 . | we draw attention to the close relationship between the anisotropy parameter @xmath0 for stellar orbits in elliptical galaxies and the temperature @xmath1 of the hot interstellar gas . for nearly spherical galaxies
, the gas density @xmath2 can be accurately determined from x - ray observations and the stellar luminosity density @xmath3 can be accurately found from the optical surface brightness .
the jeans equation and hydrostatic equilibrium establish a connection between @xmath0 and @xmath1 that must be consistent with the observed stellar velocity dispersion .
optical observations of the bright elliptical galaxy ngc 4472 indicate @xmath4 within the effective radius .
however , the x - ray gas temperature profile @xmath1 for ngc 4472 requires significantly larger anisotropy , @xmath5 , about twice as large as the optical value .
this strong preference for radial stellar orbits must be understood in terms of the formation history of massive elliptical galaxies .
conversely , if the smaller , optically determined @xmath0 is indeed correct , we are led to the important conclusion that the temperature profile @xmath1 of the hot interstellar gas in ngc 4472 must differ from that indicated by x - ray observations , or that the hot gas is not in hydrostatic equilibrium .
.2 in |
the problem of obtaining additional physical effects only due to multiple repetitions of the same measurement or interaction has been discussed both analytically @xcite and experimentally @xcite .
these phenomena , in which one may preserves in time an initially prepared state or even `` guide '' its time evolution to another final predetermined state @xcite in contrast to the known rules of quantum mechanics by which the result of measurement can not be known beforehand , are collectively termed quantum zeno effect @xcite .
they were discussed exclusively at the level of either the schroedinger equation @xcite , or by using the density matrix @xcite .
we show in this work , using specific examples , that these phenomena may be found also in the context of quantum field theory .
moreover , it has been shown @xcite , using the spin example , that these repetitions not only preserves or guide to some predetermined state but also may result in entirely new effects as will be explained .
we show , using quantum field theory , that this is indeed the case and not only in the spin case .
we show this for the two most discussed cases in relation to the many body problem in quantum field theory @xcite : 1 ) the many body system in which the constituent particles are not interacting with one another , but are submitted to an external potential @xmath1 , and 2 ) the many body system in which the constituent particles are interacting with one another . in both cases the single particle propagator can be represented by an infinite series from which we can get the energies and the lifetime of the relevant system @xcite . in the expression `` single particle propagator '' we mean especially the specific green function @xmath2 which is the probability amplitude that if at the time @xmath3 we add a particle in state @xmath4 to the system in its ground state , then at the time @xmath5 the system will be found in its ground state with an added particle in the state @xmath6 @xcite . the propagator @xmath2 is termed the `` dressed '' or `` clothed '' propagator to differentiate it from the free ( bare ) propagator @xmath7 which has the same meaning of a probability amplitude as that of @xmath2 , but with no perturbing interaction resulting from either an external potential or from some interaction among the particles composing the system .
we remark that the `` clothed '' propagator is conventionally estimated @xcite by summing to an infinite order over some selective series which is always characterized by the same basic diagram ( from a very large number of possible diagrams ) repeated to all orders . from the sum over this series one
derives physical results like the ground and excited energy states of the system @xcite .
that is , the physical phenomena appear after summing to infinite order over this set of series of repetitions of the same diagram .
there exists a large number of examples corroborating this .
the known hartree @xcite and hartree - fock @xcite quantum field realizations of physical phenomena are the results of summing to an infinite order over only the same repeated diagram .
that is , over only the bubble terms @xcite in the first case , and over only the bubble and open oyster terms in the second case @xcite .
likewise , the random phase approximation method ( rpa ) is based upon summing over only the terms called the ring terms @xcite .
the basic phonon relations are derived @xcite from summing to an infinite order over only the same repeated ( to all orders ) process which represents the einstein constant frequency phonon .
the plasmon characteristics have been derived by summing over only the `` pair bubbles '' terms @xcite .
even the two particle propagator is handled by summing over only what is termed the ladder terms @xcite .
for all the above and many other cases this summing over the same repeated process results in a new particle , the quasi particle @xcite , with a characteristic energy , an effective mass , and a finite lifetime .
these infinite repetitions over the same process dress the initial `` bare '' particle and transform it to another one with different energy , mass , and lifetime .
we will show in section 3 that if we have no repetitions then we have also no quasiparticles and no excited energy states .
thus , according to the previous discussion , the starting point will not be the general series which is not summable @xcite , but a selective series which is generally a series of only one process ( from actually a very large number of possible processes ) and all its different orders .
here , in order to emphasize this element of repetition and its essential role in the formation of the zeno effect @xcite we discuss a special version of the last series in which the terms of these series are not all the orders of the once performed relevant interaction , but _ all the orders of the @xmath0 times repetitions of it _ , as will be explained in the following sections . also ,
using the bubble and open - oyster examples we illustrate the aharonov - vardi conclusion @xcite , with respect to spin rotations , that even when the physical mechanisms ( potentials and interactions ) , that cause the time evolutions of the physical systems , are absent , nevertheless , the large number of repetitions of the `` measurement '' of the corresponding observables induces this type of time evolution . in our case
we obtain , by these repetitions , an induced continuous spectrum of excited state energies in a finite interval . in section 2 use
is made of the vacuum amplitude @xmath8 @xcite and the unique nature of the zeno effect @xcite to show this effect for the bubble process @xcite , and for the general unlinked diagram with @xmath0 identical links @xcite . in section 3 the zeno effect
is shown also for the case in which the initial and final states of the system are different .
this is demonstrated for the specific open - oyster process @xcite , and for the general case of different initial and final states of the system in which the amplitude has a value greater than unity .
the vacuum amplitude , as defined in the literature ( see , for example , @xcite ) , takes into account all the various processes that lead from the ground state , back to the same state . here , in order to discuss the zeno effect @xcite which is characterized by a large number of repetitions of the same process , we adopt a restricted vacuum amplitude formalism that involves repetitions of only one particular process . as we have pointed out , the hartree and hartree - fock procedures , for example ,
belong to this category .
as mentioned , our basic diagram is the @xmath0 times repetitions of the process that begins and ends at the same state , where in the limit of dense measurement @xmath0 tends to be a very large number .
that is , this basic diagram is , actually , composed of @xmath0 identical parts .
thus , the terms of the infinite series representing the vacuum amplitude must signify the different orders of this basic @xmath0-times - repeated interaction .
the first term of this infinite series is the free term when no interaction occurs in the time interval @xmath9 ( we specify the initial time by @xmath10 ) .
the value of this first term of the vacuum amplitude is unity @xcite , since it expresses the fact that in the unperturbed case the probability amplitude for the quantum system to stay in its ground state is unity .
the second term denotes the basic diagram , just described .
the third term denotes the probablity when this @xmath0-times - repeated interaction is performed twice in the time interval @xmath9 etc . as an example for this process we take the bubble interaction @xcite in which an external potential lifts the system at the time @xmath11 out of its initial state @xmath12 creating a hole , and instantaneously puts it back in , destroying the hole . in the energy - time representation the probability amplitude for the occurence of the bubble process
is given by @xcite @xmath13 where @xmath14 is the external potential that transmits the system from the state @xmath12 back again to the same state @xmath12 .
@xmath14 does not depends on @xmath11 so it can be moved out of the integral sign in eq ( [ e1 ] ) .
the point correlation function @xmath15 is the probability amplitude that at the time @xmath3 a hole in state @xmath12 has been added and instantaneously removed ( destroyed ) from the system in its ground state @xcite .
the value of @xmath15 is -1 ( see @xcite ) .
the minus sign in eq ( [ e1 ] ) is for the fermion loop @xcite of the bubble process .
the integration time from @xmath10 to @xmath11 is the time it takes this process to occur .
if this bubble interaction is repeated @xmath0 times over the same total finite time @xmath9 , we obtain for the probability amplitude to find the system at time @xmath11 to have the same state it has at time @xmath10 @xcite @xmath16v^n_{ll } \nonumber \end{aligned}\ ] ] where @xmath17 is the dyson time ordered product operator @xcite .
the division by @xmath18 is because we take into account all the possible orders of the times @xmath19 .
here each @xmath20 have the same constant value ( of @xmath21 as we have remarked ) , so we obtain from the equation ( [ e2 ] ) @xmath22 the last equation is the probability amplitude to find the system at the time @xmath11 , after it has been interacted upon @xmath0 times by the same bubble interaction , to have the same state it has at the time @xmath10 . now ,
as we have mentioned we must take into account all the possible orders of this @xmath0 times repeated interaction .
if , for example , this @xmath0-th order interaction is repeated two , three , and four times over the same finite total time @xmath9 , we obtain for the relevant probability amplitudes @xmath23 , @xmath24 , and @xmath25 respectively .
the divisions by @xmath26 , @xmath27 , and @xmath28 take into account the possible time orders among these @xmath0-th order interactions ( repeated two , three , and four times ) besides the extra @xmath18 times permutations for each such @xmath0 times repeated interaction .
we note that since , as we have remarked , each such @xmath0-th order interaction is treated as _ the basic interaction _
its @xmath0 parts are not time permuted with the @xmath0 parts of any other identical basic interaction . repeating this @xmath0th order bubble process @xmath0 times , and taking the former equations into account we obtain for the probability amplitude ( denoted by @xmath29 ) to find the system in the time @xmath11 to be in the same state it was in the initial time @xmath10 .
@xmath30 we are interested in showing the existence of the zeno effect in the limit of dense measurement , that is , of a very large @xmath0 .
we obtain @xmath31 that is , the probability to remain with the initial state after all these interactions is unity which is the zeno effect @xcite .
we can generalize from the specific bubble interaction to a general one . the only condition this general interaction has to fulfil is to start and end at the same state , so that when it is repeated @xmath0 times , the resulting @xmath0-th order diagram is composed of @xmath0 unlinked identical links .
now , it is known @xcite that the value of an unlinked diagram with @xmath0 unlinked links @xmath32 is @xmath33 , no matter what is the character of @xmath32 .
thus , denoting our fundamental generalized interaction by @xmath32 , and repeating the same process , as we have done for the bubble interaction , we obtain the following vacuum probability amplitude @xmath34 ( to start and end at the same state ) in the zeno limit @xmath35 that is , the quantum zeno effect may occur in the framework of quantum field theory .
this derivation is general in that we do not have to specify the fundamental repeated interaction @xmath32 .
the same conclusion can also be obtained by considering the ground state energy of the perturbed system which is obtained by using the vacuum amplitude from eq ( [ e6 ] ) .
this ground state energy is obtained from the following relation , known as the linked cluster theorem @xcite @xmath36 where @xmath37 is the ground state energy of the unperturbed hamiltonian corresponding to the unperturbed ground state @xmath38 which is assumed to be the initial state of the system , and @xmath39 is a positive infinitesimal such that @xmath40 , and @xmath41 for any finite @xmath42 .
@xmath8 , in our case , is the @xmath43 from eq ( [ e6 ] ) .
one sees from the general linked cluster expansion given , for example , by mattuck ( in @xcite p. 110 ) that the expansion ( [ e6 ] ) results from including only the bubble contribution .
thus , substituting in eq ( [ e7 ] ) for @xmath8 ( @xmath43 from eq ( [ e6 ] ) ) we obtain @xcite @xmath44 thus , we see that in the zeno limit the initial energy ( the initial state ) is preserved .
this is true for any general process @xmath32 , such that when repeated @xmath0 times the value of its @xmath0 unlinked parts diagram ( we are restricted here to the vacuum amplitude case ) is @xmath33 .
all we have to do is to use the general @xmath43 from eq ( [ e6 ] ) , and eq ( [ e7 ] ) .
the result we obtain is identical to eq ( [ e8 ] ) .
all our discussion thus far of the bubble zeno effect uses the vacuum amplitude , and so is restricted to the case where the initial and final states of the system were the ground state .
we generalize now to any other state and take into account explicitly the unperturbed propagators which connect neighbouring interactions . here also our basic unit is , because of the zeno effect , the @xmath0-times - repeated bubble interaction .
this general bubble process is now more natural than the former , since each bubble interaction is naturally related to the former and to the following identical ones by connecting paths which are the free propagators @xmath45 defined as the free propagation of the system from the time @xmath3 to @xmath5 without any disturbance whatever .
thus , in order to accomodate to this situation we have to multiply each fundamental bubble process given by eq ( [ e1 ] ) by the free propagators @xmath46 and @xmath47 , the first leads from the initial time @xmath10 to the time of the interaction @xmath3 and the second from @xmath3 to the time after the interaction @xmath5 , so that eq ( [ e1 ] ) would be written as @xmath48 where @xmath49 is the initial and final state of each such fundamental bubble process .
the interaction is denoted now by @xmath50 that signifies that our system begins and ends at the same state @xmath49 , creating and destroying a hole in state @xmath12 ( if the system interacts only with an external potential then this interaction is denoted by @xmath51 as is done for the vacuum amplitude case ) .
@xmath50 is a probability amplitude that does not depend on time and is given by @xcite @xmath52 and @xmath53 has the same meaning as in the former case .
the free propagator @xmath47 has the following value @xcite @xmath54 with @xmath55 substituting from eq ( [ e10 ] ) into eq ( [ e9 ] ) we obtain @xmath56 now , since we deal with identical repetitions of the same interaction all the @xmath50 s are the same . also all the @xmath57 s are , for the same reason , identical to each other
moreover , we can take also the time differences @xmath58 , especially for large @xmath0 , to be the same .
thus , taking these considerations into account , we write the relevant modified form of eq ( [ e2 ] ) as follows @xmath59\int_{t_0}^t\int_{t_0}^{t_1 } \ldots \int_{t_0}^{t_{n-1 } } \cdot \nonumber \\ & & \cdot e^{-i\epsilon_k(t_n - t_0 ) } dt_1dt_2 \ldots dt_n= ( v_{klkl})^n\int_{t_0}^t\int_{t_0}^{t_1 } \ldots \int_{t_0}^{t_{n-2}}(\frac{e^{-i\epsilon_k(t_{n-1}-t_0)}}{-i\epsilon_k}- \frac{1}{(-i\epsilon_k)})dt_1dt_2\ldots dt_{n-1}= \nonumber \\ & & = ( v_{klkl})^n ( \frac{e^{-i\epsilon_k(t - t_0)}}{(-i\epsilon_k)^{n-1}}-\frac{1}{(-i\epsilon_k)^{n-1 } } -\frac{(t - t_0)}{(-i\epsilon_k)^{n-2}}- \frac{(t - t_0)^2}{(-i\epsilon_k)^{n-3}}-\ldots)= \label{e12 } \\ & & = ( v_{klkl})^n(\frac{e^{-i\epsilon_k(t - t_0)}}{(-i\epsilon_k)^{n-1}}- \sum_{m=0}^{n-1}\frac{(t - t_0)^m}{m!(-i\epsilon_k)^{n-1-m } } ) \nonumber \end{aligned}\ ] ] here , we have taken @xcite @xmath60 . expanding the exponent @xmath61 in a taylor series
we obtain from the last equation @xmath62 the left hand side of figure 1 shows the @xmath0 times repetitions of the bubble process which is represented as a circle .
these unconnected repetitions conform to eq ( [ e2 ] ) .
the right hand side of the figure shows these @xmath0 times repetitions connected by leading paths , and so they conform to eq ( [ e13 ] ) .
we note that since what interests us in this work is the limit of very large @xmath0 of these @xmath0-times repeated interactions , represented by equations ( [ e12])-([e13 ] ) in this section and eq ( [ e28 ] ) in the following one , these @xmath0 multiple interactions are to be regarded as one connected unseparated process ( see the discussion before eq ( [ e4 ] ) ) and not as repetitions over improper self energy parts @xcite , so we can use the following dyson s equation @xcite as we have done in equations ( [ e18 ] ) , ( [ e29 ] ) and ( [ e34 ] ) .
@xmath63,\ ] ] where @xmath17 is the dyson s time ordered product .
the right hand side of eq ( [ e14 ] ) is generally used because the @xmath64 s do not commute . here
the @xmath64 s take numerical values ( see equations ( [ e12 ] ) , ( [ e13 ] ) , and ( [ e28 ] ) ) , and so we do not have here any commutation problems .
thus , the @xmath65 from eq ( [ e11 ] ) , for example , could have been written and substituted in eq ( [ e12 ] ) as @xmath66 now , we have to take into account all the possible orders of the @xmath0 times repeated interaction process given by eq ( [ e12 ] ) .
for example , the second order process , is @xmath67 and the @xmath0th order process @xmath68 where the expression @xmath69 contains @xmath0 terms .
we want to demonstrate the zeno effect in the dense measurement limit , that is , for very large @xmath0 .
so , repeating this @xmath0th order bubble interaction to all orders , taking the former equations into account , adding and subtracting 1 , and using the dyson s equation we obtain for the probability amplitude to find the system at time @xmath11 in the same state it was at the initial time @xmath10 ( compare with eq ( [ e5 ] ) ) @xmath70 the last outcome is obtained by using the last results of equations ( [ e12 ] ) and ( [ e13 ] ) from which we obtain @xmath71 .
@xmath72 is the probability amplitude to begin and end at the same state without any interaction .
this no - interaction process , like the basic bubble interaction discussed here , is an @xmath0-times - repeated process .
that is , @xmath72 is the @xmath0 times repetitions of the free propagator given by eq ( [ e10 ] ) , so that the time allocated for each is @xmath73 .
thus , @xmath72 , with the help of eq ( [ e10 ] ) and in the zeno limit where @xmath74 , is @xmath75 from equations ( [ e18])-([e19 ] ) we obtain for the zeno limit of the probability of the bubble process @xmath76 that is , in the limit of the zeno effect we obtain for the bubble process , when it is represented by either eq ( [ e1 ] ) ( in the vacuum amplitude case ) or by the more general eq ( [ e9 ] ) , a probability of unity to begin and end in the same state .
we must again note that taking into account only the bubble process , from the large number of possible different processes , is the earlier hartree method @xcite of dealing with the interacting many body system .
but unlike this hartree point of view in which the bubble interaction is taken once to all orders , here in order to emphasize the important role of these identical repetitions to the zeno effect this bubble interaction is taken @xmath0 times to all orders where @xmath77 .
now , we discuss the other ( excited ) states of the system .
the conventional procedure that yields the excited state energies is to find the poles of the propagator @xmath78 @xcite which is the fourier transform of the propagator @xmath79 .
the last propagator is the probability amplitude to find the system at the time @xmath11 , _ after interaction _ , in the same state it has started from at the time @xmath10 , and it is , for the zeno process , no other than the @xmath80 we found in eq ( [ e18 ] ) .
thus , we must transform this equation from the @xmath81 representation to the @xmath82 one .
we do this by finding the @xmath82 representation of @xmath72 from eq ( [ e19 ] ) using the fourier transform method @xmath83 the @xmath84 in the exponent comes from multiplying by @xmath85 , where @xmath84 is an infinitesimal satisfying @xmath86 , and @xmath87 , ( @xmath88 is a constant ) @xcite .
we do this in order to remain with a finite result for this exponent when @xmath89 .
the @xmath90 is the @xmath0 times repetitions of the free propagator @xmath91 which is the @xmath82 representation of @xmath47 from eq ( [ e10 ] ) .
we are interested in the limit of very large @xmath0 , and as seen from eq ( [ e21 ] ) when @xmath92 we , actually , have a pole for each value of @xmath93 that satisfies @xmath94 , that is , @xmath95 .
there are no excited energies outside this range .
we note that in the many body interaction picture the excited energy @xmath57 is equal to @xcite the difference between the excited state energy of the interacting @xmath96-particle system and the ground state of the interacting @xmath97-particle system .
thus , if the bubble process is performed once and the selective series of this once performed process is summed to all orders , as in the hartree method , one obtains excited state energies at the value given by eq ( [ e23 ] ) .
but when this bubble process is repeated @xmath0 times and the selective series of this @xmath0-times repeated process is summed to all orders , as we have just done in equation ( [ e12])-([e18 ] ) , we obtain from eq ( [ e21 ] ) excited state energies for all values of @xmath93 that satisfy @xmath94 .
that is , we obtain a large number ( continuum ) of extra excited energies that has been added _ only because of these identical repetitions of the same bubble process_. this mechanism of obtaining physical results as a consequence of just repeating the same process which by itself , without these repetitions , does not yield these results has already been noted in @xcite in connection with rotations that occur only because of a large number of repetitions of the same measurement . speaking in terms of quasi - particles
@xcite we can write the @xmath82 representation of @xmath98 from eq ( [ e18 ] ) , using eq ( [ e21 ] ) , as @xmath99 @xmath100 is the lifetime of the quasi - particle , and since @xmath84 is small @xmath100 is very large , so these quasi - particles with the extra excited energies just mentioned have a very large lifetime .
we must note that the relevant excited state energies @xmath101 obtained when the bubble process is performed once and the selective series of this once performed process is summed to all orders is just the hartree @xmath101 @xcite .
@xmath102 when the bubble process is repeated @xmath0 times , then as can be seen from equations ( [ e13]),([e18])-([e19 ] ) , and ( [ e21])-([e22 ] ) the @xmath101 s obtained do not depend on any potential @xmath1 .
this , as we have remarked , is in accord with the aharonov - vardi conclusion @xcite that the physical mechanisms that trigger the time evolutions of the system does not play an essential role , since the mere large number of repetitions of the same measurement is the cause of this time evolution .
we note that aharonov and vardi show this for the spin @xmath103 particle example , but it is obvious from their representation that this conclusion is a general one . we have shown this for the bubble process for which a large number of repetitions results in excited energies that do not depend upon any potential .
we show in the next section that if we have no repetitions then we do not have any excited energies . in summary , we find that when the interaction involved does not end at the same state it has began from and if it is not repeated then no excited state results from such an interaction ( see eqs ( [ e31])-([e32 ] ) in the next section and the discussion that follows ) . if this interaction begins and ends at the same state as in the hartree model then a single pole @xmath104 is found ( see eq ( [ e23 ] ) ) . and when this interaction is repeated @xmath97 times then in the limit of @xmath105 one find a continuum of poles ( cut ) for all values of @xmath106 that satisfy @xmath107 , where @xmath108 is the energy by which the involved system propagates during the interaction .
that is , as has been remarked in @xcite the large number of repetitions produces new stable physical effects ( see also eq ( [ e22 ] ) and the discussion that follows it ) that do not appear in the absence of them . and the more larger the number of these repetitions on the same time interval , as in the discussion here in which the repeated interaction is not taken by itself but by its @xmath97 time repetitions where @xmath105 , the more larger is the new stable physical effect as the cut found here ( see eqs ( [ e21])-([e22 ] ) and the relevant discussion there ) instead of the single pole of the hartree model .
we note that the quasi - particles related to these poles have a very long lifetime so that once they are formed they do not decay fastly .
we , now , show that we can apply the zeno effect @xcite also for the general case , where the system ends at the time @xmath11 in some specific state which is not identical to the initial one from which it has started at the time @xmath10 . in this context
we do not use the standard zeno effect at a state ( where the system returns to the same state it has started from ) , as discussed in the previous section , but apply a zeno effect along some definite feynman path of possible states in the sense of aharonov and vardi @xcite .
that is , if we do dense measurement along any definite feynman path of states then we make it actual in the sense that its probability amplitude is unity .
here we begin at some predetermined initial state and end at another predetermined final one .
this aspect of the quantum zeno effect in which the evolution of the relevant quantum system is _ guided _
, by means of dense measurement , to the corresponding prefixed final state is termed in @xcite the dynamical quantum zeno effect , in contrast to the usual quantum zeno effect ( in which the system starts and ends at the same state ) which is termed in @xcite the static quantum zeno effect .
the propagator in this general case is the probability amplitude that if the system begins at the initial time @xmath10 in a specific state , then it will be found at another specific one at the later time @xmath11 . as in the former section , in order to emphasize the important role of repetitions for the zeno effect , the basic diagram is the @xmath0 times repetitions of this interaction , where in the limit of dense measurement @xmath0 becomes very large number .
thus , the terms of the infinite series representing the propagator signify the different orders of this @xmath0-repeated - interaction . in this case
the repetitions is along some definite path connecting the initial and final states , and not local repetition as in the bubble example .
we choose , as in the bubble case , some example that may be described from two points of view .
one is the situation when the interaction is triggered by an external potential that acts @xmath0 , @xmath109 , @xmath110 times etc .
the other , more natural , interaction is that caused by the correlations between different particles that comprise the system .
unlike the bubble case , in both points of view there must be a connecting path between any two neighbouring interactions since they are not identical to each other , as will be explained in detail later . here
the initial state of each such interaction is not identical to the initial state of the former one , _ but to its final state_. the only difference between the external potential situation and the correlation - between - particles one is in the character of the interaction which in the former case is denoted by @xmath111 , that is , a particle that begins at state @xmath49 is interacted upon by an external potential that changes its state to that of @xmath12 ( compare with the external potential situation of the bubble case in which a particle begins and ends at the same state , and therefore the external potential is denoted by @xmath51 ) . in the correlation - between - particles situation
this interaction is denoted by @xmath112 @xcite ( compare with the @xmath50 of the correlation - between - particles situation of the bubble case @xcite ) a fundamental interaction in which the system ends at the time @xmath11 in a state different from the one with which it has started from at the initial time @xmath10 is , for example , what is termed the open - oyster diagram @xcite .
we must remark that this interaction is calculated to be @xcite as one in which the particle that left the interaction site at the later time @xmath11 is in the same state @xmath49 with which another particle enters the interaction site at the initial time @xmath10 .
nevertheless , we discuss here another version of this interaction in which the particle that leaves the interaction site at the time @xmath11 is in the state @xmath113 , and not in the initial one @xmath49 .
we also call this interaction open - oyster . in the external potential version of this interaction
an incoming particle at state @xmath49 enters the potential region at the time @xmath10 .
then at time @xmath11 the potential knocks another particle out of the state @xmath114 into state @xmath12 , thus creating a particle in state @xmath12 , and a hole in state @xmath114 . at the same time
@xmath11 the particle in @xmath49 is knocked into the hole in @xmath114 , and thus annihilated with it .
the particle in @xmath12 continues propagating out of the potential region .
this process is referred to as an exchange scattering @xcite , compared to the forward scattering of the bubble process in which the particle emerges in the same direction ( i.e , momentum state ) as it has entered .
on the right hand side of figure 2 we see this open - oyster interaction , and on the left hand side of it we see @xmath0 times repetitions of this process over the same time interval @xmath9 . in the energy - time representation
the probability amplitude for the occurence of the open - oyster process is given by @xcite : @xmath115 the difference between the bubble process that may represent the static zeno effect @xcite ( when repeated a large number of times ) , and the open - oyster process , that may be regarded as an example of the dynamic zeno effect @xcite ( when performed many times ) , can be understood in the following way @xcite : suppose we have a family of states denoted as @xmath116 , where @xmath117 , such that @xmath118 , where @xmath119 is the initial state of the quantum system .
we assume that successive states differ infinitesimally from one another , so that we have @xmath120 . denoting , as before , the total finite time of the @xmath0 repeated interactions by @xmath9 , and the time it takes to perform each such interaction by @xmath121 we have @xmath122 .
now , the open - oyster interaction may be regarded as , actually , projecting the evolving wave function at the time @xmath123 on the state @xmath116 .
so when @xmath0 becomes very large in the limit of the zeno effect we obtain actually @xmath124 .
this is the dynamic zeno effect of @xcite .
the static zeno effect is the special case when @xmath125 for all @xmath49 .
if we describe this process in terms of the correlation between the different particles of the system then in this interaction an incoming particle in state @xmath49 performs in a simultaneous manner several tasks ; 1 ) it strikes another particle from state @xmath114 to state @xmath12 , 2 ) creates a hole in @xmath114 , 3 ) is annihilated with the hole in @xmath114 , and the particle in @xmath12 leaves the system .
the open - oyster interaction is written now as @xmath126 now , since the last two equations ( [ e24 ] ) and ( [ e25 ] ) are identical to each other , except for the subscripts of the potential @xmath1 , we concentrate our attention on eq ( [ e25 ] ) with the understanding that what we say about it holds also for eq ( [ e24 ] ) .
@xmath112 denotes the interaction just described , and the @xmath127 s are the free propagators given by eq ( [ e10 ] ) .
we must note again that the successive repetitions of the open - oyster interaction , required for the discussion of the dynamic zeno effect , are not characterized as being identical to each other , as in the bubble process , but that each such fundamental interaction begins from the point ( state ) in which the former interaction ends .
thus , we have to take into account the path that connects each two such neighbouring interactions .
this connecting path is , of course , the free propagator @xmath128 .
substituting now from eq ( [ e10 ] ) into eq ( [ e25 ] ) , and assuming that @xmath112 does not depend on @xmath11 we obtain @xmath129 where we have used the value of 1 for @xmath130 . using eq ( [ e25 ] ) we write for the @xmath0-th order open - oyster process @xmath131 where we have assumed that for large @xmath0 all the potentials that transfer the system between two neighbouring states are equal to each other , that is , @xmath132 . carrying out the @xmath0 integrals of the last equation we obtain an expression with @xmath133 terms , each of which is a fraction with a numerator that is a difference of exponentials in the energies @xmath134 s multiplied by the times @xmath135 , and the denominator is a multiplication of @xmath0 different factors .
this @xmath133 terms expression can be grouped into @xmath0 different groups in which the number of terms are arranged as @xmath136 .
all the terms of the same group have the same numerator up to a sign , but a different denominator , so we can reduce the number of all the terms of each group to 1 by taking the common denominator of all the terms that belong to the same group . in such a way the total number of terms of the original expression is reduced from @xmath133 to @xmath0 .
thus , we obtain @xmath137 it can be seen that all the @xmath0 numerators of the last equation are differences of sines and cosines , whereas each one of the corresponding @xmath0 denominators is a product of @xmath0 factors that are differences of energies .
when @xmath0 is very large , which we always assume in this work , we have @xmath138 ( since neighbouring states differ infinitesimally ) , so in this limit we have at least two factors in each denominator that tend to zero .
thus , although all the @xmath0 terms of equation ( [ e28 ] ) are multiplied by the factor @xmath139 ( @xmath1 is a probability amplitude that satisfies @xmath140 ) we obviously have @xmath141 .
we are interested , as in the bubble case , in the repetitions to all orders of @xmath142 from eq ( [ e28 ] ) .
beginning from this equation it is not hard to obtain the various orders of @xmath142 .
so , if we take the infinite series ( that denotes the various orders of the @xmath0 repetitions process @xmath142 ) , adding and subtracting 1 , and taking the relation @xmath141 into account we obtain , using the dyson s equation , for the general probability amplitude in the zeno limit @xmath143 @xmath144 is the amplitude for our system to begin in some specific initial state @xmath116 at the time @xmath10 , and end in another different state @xmath145 at the time @xmath11 without any interaction whatever on our system .
this no - interaction process is obviously zero if the final state is different from the initial one ( see , for example , @xcite ) , so we obtain for the _ probability _ of the open - oyster process in the zeno limit @xmath146 thus , we see that in this limit we obtain for the open - oyster process a probability of unity to end at a specific prescribed state different from another specific initial one .
we now show that we have no excited state energies for the open - oyster process in the zeno limit . for this purpose we must find , in this limit , the poles of the propagator @xmath147 which is the fourier transform of the propagator @xmath148 given by eq ( [ e29 ] ) .
thus , using the fourier transform procedure , multiplying by @xmath149 @xcite , and using @xmath150 we obtain @xmath151 where the @xmath84 is , as in eq ( [ e21 ] ) ( see the discussion after eq ( [ e21 ] ) ) , an infinitesimal quantity that satisfies @xmath152 , and @xmath87 , where @xmath88 is some finite number .
this @xmath84 has been introduced in order to have a finite result for the exponent of eq ( [ e31 ] ) in the limit @xmath89 ( see appendix i in @xcite ) . from the last equation
we obtain that the poles of @xmath153 , which are the excited energy states of the physical system are @xmath154 that is , there exists no excited energy states in the zeno limit of the open - oyster process .
the reason , as we have remarked , is the absence of local repetitions in the version we have adopted here for the open - oyster process .
that is , we discuss here a process in which the state of the particle that leaves the system is different from the state of the one that enters . and
when this process is repeated the initial state of the entering particle in the repeated process is the final state of the leaving particle in the former one .
thus , this process is not locally repeated , and this absence of repetitions entails the absence of excited states for the system .
that is , all the energies of the @xmath96-particle system are equal , in the zeno limit , to each other and to the ground state energy of the @xmath97-particle system ( see @xcite , p. 41 ) .
in contrast to this situation , when we have local repetitions of some process , then we have excited states of the physical system .
that is , if the selective series of this process is composed of repeated to all order terms like the hartree selective series of the bubble process , then excited states are obtained ( see eq ( [ e23 ] ) ) .
many more additional excited states are obtained when this summation to all orders is over the @xmath0-times repetitions of this process as we have obtained for the bubble process in the former section ( see eq ( [ e21 ] ) ) .
now , if we discuss this open - oyster process from the conventional point of view @xcite where the energy of the leaving particle is the same as that of the entering one , and the summation to all orders is over the once - performed open - oyster process and not over the @xmath0th times repetitions of it , then we obtain for the @xmath101 @xcite @xmath155 where @xmath112 is the physical interaction that generates this open - oyster interaction .
that is , the excited state energies of the system are determined by these repetitions , as has been remarked in @xcite ( see the discussion after eq ( [ e21 ] ) ) we must note that the result of eq ( [ e30 ] ) is obtained not only for the open - oyster case , but also for any other arbitrary interaction for which the amplitude @xmath156 to ends in a specific state different from the initial one satisfies @xmath157 .
if we denote the propagator ( the full propagator , not the free one ) of such interaction by @xmath158 , its free propagator by @xmath159 , and adding and subtracting 1 the propagator takes the following form @xmath160 in obtaining the result of eq ( [ e34 ] ) we made use of the facts that @xmath161 , and @xmath157 so that @xmath162 .
we see , therefore , that also for the general case , where the system reaches at the time @xmath11 a different state from that in which it started , we get a probability of 1 in the dense measurement limit .
thus , we see that the zeno effect @xcite may be effective in the framework of quantum field theory .
we show that the zeno effect may be discussed also in the context of quantum field theory .
we have used in section 2 the dyson s equation and the bubble example to demonstrate the static zeno effect , in which the initial and final states of the system are the same . in section 3
we have used the open - oyster example and the dyson s equation to demonstrate the dynamic zeno effect , in which the initial and final states of the system are different . in this work
the dyson s equation has been used to infinitely sum to all orders over the @xmath0 times repetitions of these two processes .
it has been shown in sections 2 and 3 that the probability amplitudes to find the final state of the system identical to the initial one in the bubble case , or different in the open oyster case tend both to unity as the number of repetitions @xmath0 becomes large .
we have found in section 2 that repeating the bubble process a large number of times in a finite total time results in obtaining a large number ( cut ) of additional excited energy states that emerge only because of these repetitions . by this
we have corroborated the same conclusion arrived to by aharonov and vardi with respect to spin rotation .
we have found , accordingly , in section 3 for the open - oyster process that the absence of any repetition results in the absence of excited state energies .
i wish to thank l.p.horwitz for discussions on this subject , and for his review of the manuscript 99 1 in b. misra and e. c. sudarshan , j. math .
phys , * 18 * , 756 ( 1977 ) ; `` decoherence and the appearance of a classical world in quantum theory '' , d. giulini , e. joos , c. kiefer , j. kusch , i. o. stamatescu and h. d. zeh , springer - verlag , ( 1996 ) ; saverio pascazio and mikio namiki , phys .
rev a * 50 * , 6 , 4582 , ( 1994 ) ; a. peres , phys .
rev d * 39 * , 10 , 2943 , ( 1989 ) , a. peres and amiram ron , phys .
rev a * 42 * , 9 , 5720 ( 1990 ) r. a. harris and l. stodolsky , j. chem .
phys , * 74 * , 4 , 2145 ( 1981 ) ; mordechai bixon , chem .
phys , * 70 * , 199 - 206 , ( 1982 ) ; marcus simonius , phys . rev .
lett , * 40 * , 15 , 980 ( 1978 ) r. j. cook , physica scripta t * 21 * , 49 - 51 ( 1988 ) ; w. m. itano , d. j. heinzen , j. j. bollinger , and d. j. wineland , phys .
rev a , * 41 * , 2295 - 2300 , ( 1990 ) ; a.g.kofman and g.kurizki , _ phys.rev a _ * 54 * , 3750 - 3753 ( 1996 ) ; g.kurizki , a.g.kofman and v.yudson , _ phys.rev a _ * 53 * r35 ( 1995 ) ; s.r.wilkinson , c.f.bharucha , m.c.madison , p.r.morrow , q.niu , b.sundaram , and m.g.raizen , _ nature _ * 387 * , 575 - 577 ( 1997 ) y. aharonov and m. vardi , phys.rev d * 21 * , 2235 , ( 1980 ) p. facchi , a. g. klein , s. pascazio and l. schulman , phys.lett a * 257 * , 232 - 240 , ( 1999 ) , richard .
d. mattuck , @xmath163 edition , mcgraw - hill international book company , ( 1976 ) , @xmath163 edition , g. mahan , plenum press new york ( 1993 ) ; `` a course on many body theory applied to solid state physics '' , c. enz , world scientific , ( 1992 ) , by richard .
d. mattuck , annals of physics _ 27 _ , 216 - 226 , ( 1964 ) richard .
p. feynman , rev .
phys,*20 * , 2 , 367 ( 1948 ) ; `` quantum mechanics and path integrals '' , richard .
p. feynman and a. r. hibbs , mcgraw - hill book company ( 1965 ) , h. haken , north - holland publishing company , ( 1981 ) the diagrams that represent repetitions over improper self energy parts can be separated into unconnected self energy parts by removing the connecting particles lines . generally , it is known @xcite that such diagrams can not be summed over by using dyson s equation as we do below ( see , for example , equations ( [ e18 ] ) and ( [ e34 ] ) ) , since these improper diagrams would have to be counted more than once @xcite . , gert roepstorff , springer - verlag , ( 1994 ) , @xmath163 edition , e. merzbacher , john wiley and sons , ( 1961 ) i. m. gelfand and a. m. yaglom , j. math .
phys * 1*. 1 , 48 - 69 , ( 1960 ) , volume 4 , i. m. gelfand and n. ya .
vilenkin , academic press , 1964 | _ we show , using quantum field theory , that performing a large number of identical repetitions of the same measurement does not only preserve the initial state of the wave function ( the zeno effect ) , but also produces additional physical effects .
we first demonstrate that a zeno type effect can emerges also in the framework of quantum field theory , that is , as a quantum field phenomenon .
we also derive a zeno type effect from quantum field theory for the general case in which the initial and final states are different .
the basic physical entities dealt with in this work are not the conventional once - perfomed physical processes , but their @xmath0 times repetition where @xmath0 tends to infinity .
we show that the presence of these repetitions entails the presence of additional excited state energies , and the absence of them entails the absence of these excited energies . _ |
macroscopic measurements of brain activity , such as the electroencephalogram ( eeg ) , magnetoencephalogram or local field potentials ( lfp ) , display approximately @xmath0 frequency scaling in their power spectra @xcite .
the origin of such @xmath0 `` noise '' is at present unclear .
@xmath0 spectra can result from self - organized critical phenomena @xcite , suggesting that neuronal activity may be working according to such states @xcite .
alternatively , the @xmath0 scaling may be due to filtering properties of the currents through extracellular media @xcite .
the latter hypothesis , however , was resting on indirect evidence , and still needs to be examined theoretically , which is one of the motivations of the present paper . a _ continuum model _ @xcite of lfps incorporated the inhomogeneities of the extracellular medium into continuous spatial variations of conductivity ( @xmath1 ) and permittivity ( @xmath2 ) parameters .
this model reproduced a form of low - pass frequency filtering in some conditions , while considering the extracellular medium as locally neutral with @xmath1 and @xmath2 parameters independent of frequency .
this model was not entirely correct , however , because macroscopic measurements in cortex revealed a frequency dependence of electrical parameters @xcite .
we will show here that it is possible to keep the same model structure by including plausible causes for the frequency dependence . in a _ polarization model _
@xcite of lfps , the variations of conductivity and permittivity were considered by explicitly taking into account the presence of various cellular processes in the extracellular space around the current source . in particular
, it was found that the phenomenon of surface polarization was fundamental to explain the frequency dependence of lfps .
the continuum model @xcite incorporated this effect phenomenologically through continuous variations of @xmath1 and @xmath2 . in the polarization model ,
the extracellular medium is `` reactive '' in the sense that it reacts to the electric field by polarization effects .
it is also locally non - neutral , which enables one to take into account the non - instantaneous character of polarization , which is at the origin of frequency dependence according to this model @xcite . in the present paper , we propose a `` diffusion - polarization '' model that synthesizes these previous approaches and which takes into account both microscopic and macroscopic measurements .
this model includes ionic diffusion , which we will show has a determinant influence on frequency filtering properties .
the model also includes electric polarization , which also influences the frequency - dependent electric properties of the tissue .
we show that taking into account ionic diffusion and electric polarization allows us to quantitatively account for the macroscopic measurements of electric conductivity in cortical tissue according to the experiments of gabriel et al .
@xcite . however , recent measurements of logothetis et al .
@xcite evidenced that the frequency dependence of cortical tissue was negligible , therefore in contradiction with the measurements of gabriel et al .
we show here that the `` diffusion - polarization '' model can be consistent with both types of experiments , and thus may reconcile this contradiction
. we will also examine whether this model can also explain the @xmath0 frequency scaling observed in lfp or eeg power spectra .
finally , we consider possible ways for experimental test of the predictions of this model .
the final goal of this approach is to obtain a model of local field potentials which is consistent with both macroscopic measurements of conductivity and permittivity , and the microscopic features of the structure of the extracellular space around the current sources .
the numerical simulations presented in section [ sec - sim ] describe the impedance of the extracellular medium based on the following equation @xmath3 this equation gives the @xmath4-frequency component of the impedance at point @xmath5 in extracellular space , in spherical symmetry ( see ref .
@xcite and eq .
[ eqgeneralm ] for details ) .
to evaluate this equation , we use matlab , which computes the riemann sum : @xmath6 where @xmath7 is the integration step ( @xmath8 ) and @xmath9 is determined for a `` slice '' of 1 mm .
we also use the parametric model of gabriel et al .
@xcite to simulate the frequency dependence of electrical parameters @xmath1 and @xmath2 of the extracellular fluid from gray matter ( at a temperature of @xmath10 ) .
this model is valid for frequencies included in the range of 10 hz to @xmath11 hz @xcite . according to this model
, the absolute complex and macroscopic permittivity and conductivity ( measured between 10 and @xmath12 hz ) in cortical gray matter is given by the cole - cole parametric model @xcite : @xmath13 where the sum runs over 4 polarization modes @xmath14 , @xmath15 is vacuum permittivity , @xmath16 is the permittivity relative to @xmath17 , @xmath18 is the static conductivity at @xmath19 according to the chosen parametric model .
the parameters under the sum of eq .
[ eq4 ] are given in table 1 .
+ table 1 here
we start by outlining a macroscopic model with frequency - dependent electrical parameters ( section [ sec - theor ] ) , and we discuss the main physical causes for this frequency dependence ( section [ sec - freqdep ] ) .
we then constrain the model to macroscopic measurements of electrical parameters , and provide numerical simulations to test the model and reproduce the experimental observations ( section [ sec - sim ] ) .
finally , we propose a possible way to test the model experimentally ( section [ sec - test ] ) . in this section ,
we derive the equations governing the time evolution of the extracellular potential .
we follow a formalism similar to the one developed previously @xcite , except that we reformulate the model macroscopically , to allow the electrical parameters ( the conductivity @xmath1 and permittivity @xmath2 ) to depend on frequency , as demonstrated by macroscopic measurements @xcite .
the physical causes of this macroscopic frequency dependence will be examined in section [ sec - freqdep ] .
we begin by deriving a general equation for the electrical potential when the electrical parameters are frequency dependent .
we start from maxwell equations , taking the 1@xmath20 and the divergence of the 4@xmath21 maxwell equation in a medium with constant magnetic permeability , giving : @xmath22 where @xmath23 , @xmath24 and @xmath25 are respectively the electric displacement , current density and charge density in the medium surrounding the sources .
moreover , in a linear medium the equations linking the electric field @xmath26 with electric displacement @xmath23 , and with current density @xmath24 , gives : @xmath27 and @xmath28 the fourier transforms of these equations are respectively @xmath29 and @xmath30 , where we allow @xmath1 and @xmath2 to depend on frequency .
given the limited precision of measurements , we can consider @xmath31 for frequencies smaller than 1000 hz .
thus , we can assume that @xmath32 such that the complex fourier transform of eqs .
[ eqtemps ] can be written as : @xmath33 consequently , we have @xmath34 compared to previous derivations ( see eq .
49 in ref .
@xcite ) , this equation is a more general form in which the electrical parameters can be dependent on frequency . in principle , it is sufficient to solve eq .
[ eqgeneral ] in the extracellular medium to obtain the frequency dependence of lfps .
however , in practice , this equation can not be solved because the structure of the medium is too complex to properly define the limit conditions .
the associated values of electric parameters must be specified for every point of space and for each frequency , which represents a considerable difficulty .
one way to solve this problem is to consider a macroscopic or mean - field approach .
this approach is justified here by the fact that the values measured experimentally are averaged values , which precision depends on the measurement technique . because our goal is to simulate those measured values , we will use a macroscopic model , in which we take spatial averages of eq .
[ eqgeneral ] , and make a continuous approximation for the spatial variations of these average values .
this type of approximation can be found in the classic theory of electromagnetism ( see for example chapters 9 and 10 in maxwell s treaty for electricity and magnetism @xcite ) . to this end
, we define macroscopic electric parameters , @xmath35 and @xmath36 , as follows : @xmath37 and @xmath38 where @xmath39 is the volume over which the spatial average is taken .
we assume that @xmath39 is of the order of @xmath40 , and is thus much smaller than the cortical volume , so that the mean values will be dependent of the position in cortex .
because the average values of electric parameters are statistically independent of the mean value of the electric field , we have : @xmath41 where the first term in the righthand represents the `` dissipative '' contribution , and the second term represents the `` reactive '' contribution ( reaction from the medium ) . here , all physical effects , such as diffusion , resistive and capacitive phenomena , are integrated into the frequency dependence of @xmath36 and @xmath35 .
we will examine this frequency dependence more quantitatively in section [ sec - freqdep ] .
the complex fourier transform of @xmath42 then becomes : @xmath43 where @xmath44 is the complex conductivity .
we can also assume @xmath45 such that @xmath46 because @xmath47 and @xmath48 , the expressions above ( eqs .
[ ff ] ) can also be written in the form : @xmath49 we note that starting from the continuum model @xcite , where only spatial variations were considered , and generalizing this model by including frequency - dependent electric parameters , gives the same mathematical form as the original model ( compare with eq 49 in ref .
this form invariance will allow us to introduce in section [ sec - freqdep ] the surface polarization phenomena by including an _ ad hoc _ frequency dependence in @xmath50 and @xmath51 .
the physical causes of this macroscopic frequency dependence is that the cortical medium is microscopically non - neutral ( although the cortical tissue is macroscopically neutral ) .
such a local non - neutrality was already postulated in a previous model of surface polarization @xcite .
this situation can not be accounted by eq .
[ eqgeneral ] if @xmath50 and @xmath51 are frequency independent ( in which case @xmath52 when @xmath53 ) .
thus , including the frequency dependence of these parameters enables the model to capture a much broader range of physical phenomena .
finally , a fundamental point is that the frequency dependences of the electrical parameters @xmath50 and @xmath51 can not take arbitrary values , but are related to each - other by the kramers - kronig relations @xcite : @xmath54 and @xmath55 where principal value integrals are used .
these equations are valid for any linear medium ( i.e. , when eqs . [ link1 ] and [ link2 ] are linear ) .
these relations will turn out to be critical to relate the model to experiments , as we will see below .
note that , contrary to frequency dependence , the spatial dependences of @xmath50 and @xmath51 are independent of each other , because these dependences are related to the spatial distribution of elements within the extracellular medium . to obtain an expression for the extracellular potential ,
we need to solve eq .
[ eqgeneralm ] , which is possible analytically only if we consider a simplified geometry of the source and surrounding medium .
the first simplification is to consider the source as monopolar
. the choice of a monopolar source does not intrinsically reduce the validity of the results because multipolar configurations can be composed from the arrangement of a finite number of monopoles @xcite .
in particular , if the physical nature of the extracellular medium determines a frequency dependence for a monopolar source , it will also do so for multipolar configurations .
a second simplification will be to consider that the current source is spherical and that the potential is uniform on its surface .
this simplification will enable us to calculate exact expressions for the extracellular potential and should not affect the results on frequency dependence .
a third simplification is to consider the extracellular medium as isotropic .
this assumption is certainly valid within a macroscopic approach , and justified by the fact that the neuropil of cerebral cortex is made of a quasi - random arrangement of cellular processes of very diverse size @xcite .
this simplified geometry will allow us to determine how the physical nature of the extracellular medium can determine a frequency dependence of the lfps , independently of other factors ( such as more realistic geometry , propagating potentials along dendrites , etc ) .
thus , considering a spherical source embedded in an isotropic medium with frequency - dependent electrical parameters , combining with eq .
[ eqgeneralm ] , we have : @xmath56 integrating this equation gives the following relation between two points @xmath5 and @xmath57 in the extracellular space , @xmath58 = r_2^{2 } \
\frac{d < v_\omega>_{|_v}}{dr}(r_2 ) \ \left [ \sigma_{\omega}(r_2 ) + i \omega \epsilon_{\omega}(r_2 ) \right ] ~ .\ ] ] assuming that the extracellular potential vanishes at large distances ( @xmath59 ) , we find @xmath60 this equation is analogous to a similar expression derived previously ( eq . 25 in ref .
@xcite ) , but more general .
the two formalisms are related by : @xmath61 instead of @xmath62 ( see eq . 4 in ref .
this difference is due to the fact that the conductivity here depends on frequency . in the following
, we will use the simplified notations @xmath63,@xmath64 and @xmath65 instead of @xmath66,@xmath67 and @xmath68 , respectively . using the relation @xmath69 , the impedance @xmath70 is given by : @xmath71 } \label{impedance}\ ] ] where @xmath72 and @xmath73 is the distance between the center of the source and the position defined by @xmath74 . in the following ,
we successively consider two different cases of extracellular medium : first , `` non - reactive '' media , in which the current passively flows into the medium ; second , `` reactive '' media , in which some properties ( such as charge distribution ) may change following current flow . for each medium , we will consider two types of physical phenomena , the current produced by the electric field , and the current produced by ionic diffusion , as schematized in fig .
[ polariz ] ) .
`` non reactive '' media ( @xmath75 and @xmath76 ) are equivalent to `` resistive '' media , in which the resistance ( or equivalently , the conductivity ) does not change following the flow of current .
the simplest type of such configuration consists of a resistive medium ( such as a homogeneous conductive fluid ) in which current sources solely interact via their electric field . applying eq.[impedance ] to this configuration is equivalent to model the extracellular potential by coulomb s law : @xmath77 where @xmath78 is the extracellular potential at a position defined by @xmath74 in extracellular space , @xmath73 is the absolute distance between @xmath74 and the center of the current source . here
, the conductivity ( @xmath79 ) is independent of space and frequency , and thus , this model is not compatible with macroscopic measurements of frequency dependence @xcite .
it is , however , the most frequently used model to calculate extracellular field potentials @xcite .
this model will be referred to as `` model n '' in the following . because current sources are ionic currents , there is flow of ions inside or outside of the membrane , and another physical phenomena underlying current flow is ionic diffusion .
let us consider a resistive medium such as a homogeneous extracellular conductive fluid with electric parameters @xmath80 and in which the ionic diffusion coefficient is @xmath81 .
when the extracellular current is exclusively due to ionic diffusion , the current density depends on frequency as @xmath82 ( see appendix [ secscal ] ) .
a resistive medium behaves as if it had a resistivity equal to @xmath83 , where @xmath84 is complex .
the parameter @xmath85 is the resistivity for very high frequencies , and reflects the fact that the effect of ionic diffusion becomes negligible compared to calorific dissipation ( ohm s law ) for very high frequencies . when ionic diffusion is dominant compared to electric field effects , the real part of @xmath84 is much larger than @xmath85
. the frequency dependence of conductivity will be given by : @xmath86 applying eq .
[ impedance ] to this configuration gives the following expression for the electric potential as a function of distance : @xmath87 this expression shows that , in a non - reactive medium , when the extracellular current is dominated by ionic diffusion ( compared to that directly produced by the electric field ) , then the conductivity will be frequency dependent and will scale as @xmath82 .
this model will be referred to as `` model d '' in the following .
note that , if the electric field primes over ionic diffusion , then we have the situation described by model n above . in reality ,
extracellular media contain different charge densities , for example due to the fact that cells have a non - zero membrane potential by maintaining differences of ionic concentrations between the inside and outside of the cell .
such charge densities will necessarily be influenced by the electric field or by ionic diffusion .
as above , we first consider the case with only electric field effects and will consider next the influence of diffusion and the two phenomena taken together .
electric polarization is a prominent type of `` reaction '' of the extracellular medium to the electric field . in particular
, the ionic charges accumulated over the surface of cells will migrate and polarize the cell under the action of the electric field .
it was shown previously in a theoretical study that this `` surface polarization '' phenomena can have important effects on the propagation of local field potentials @xcite .
if a charged membrane is placed inside an electric field @xmath88 , there is production of a secondary electric field @xmath89 given by ( see eq . 31 in ref .
@xcite ) : @xmath90 this expression is the frequency - domain representation of the effect of the inertia of charge movement associated with surface polarization , reflecting the fact that the polarization does not occur instantaneously but requires a certain time to setup .
this frequency dependence of the secondary electric field was derived in ref .
@xcite for a situation where the current was exclusively produced by electric field .
the parameter @xmath91 is the characteristic time for charge movement ( maxwell - wagner time ) and equals @xmath92 , where @xmath93 and @xmath94 are respectively the absolute ( tangential ) permittivity and conductivity of the membrane surface , respectively , and are in general very different from the permittivity and the conductivity of the extracellular fluid .
let us now determine for zero - frequency the amplitude of the secondary field @xmath88 produced between two cells embedded in a given electric field .
first , we assume that it is always possible to trace a continuous path which links two arbitrary points in the extracellular fluid ( see fig . [ modelesimple]b ) .
consequently , the domain defined by extracellular fluid is said to be _
linearly connex_. in this case , the electric potential arising from a current source is necessarily continuous in the extracellular fluid .
second , in a first approximation , we can consider that the cellular processes surrounding sources are arranged randomly ( by opposition to being regularly structured ) and their distribution is therefore approximately isotropic .
consequently , the field produced by a given source in such a medium will also be approximately isotropic . also consequent to this quasi - random arrangement
, the equipotential surfaces around a spherical source will necessarily cut the cellular processes around the source ( fig .
[ modelesimple]b ) .
suppose that at time @xmath95 , an excess of charge appears at a given point in extracellular space , then a static electric field is immediately produced . at this time , currents begins to flow in extracellular fluid , as well as inside the different cellular processes surrounding the source .
these cells begin to polarize , with @xmath96 as the characteristic polarization time .
asymptotically , the system will reach an equilibrium where the polarization will neutralize the electric field , such that there is no electric field inside the cells ( zero current ) .
now suppose that , in the asymptotic regime , there would still be a current flowing in between cells ( in the extracellular fluid ) , then we have two possibilities .
first , the equipotential surfaces are discontinuous , or they cut the membrane surfaces ( as illustrated in fig .
[ modelesimple]b ) .
the first possibility is impossible because it would imply an infinite electric field .
the second possibility is also impossible , because cells are isopotential due to polarization .
therefore , we can say that , asymptotically , there is no current flowing in extracellular fluid at @xmath97 , and necessarily this is equivalent to a dielectric medium . in other words ,
_ a passive inhomogeneous medium with randomly distributed passive cells is a perfect dielectric at zero frequency_. in this case , the conductivity must tend to 0 when frequency tends to 0 .
thus , in the following , we assume that @xmath98 where @xmath99 is the field produced by the source .
+ figure [ modelesimple ] here
it follows that the expression for the current density in extracellular space as a function of the electric field is given by : @xmath100 in addition , for cerebral cortical tissue , we have @xmath101 for frequencies @xmath102 and @xmath103 ( see ref
. @xcite .
thus , an excellent approximation of the conductivity can be written as : @xmath104 applying eq .
[ impedance ] gives : @xmath105 this model describes the effect of polarization in reaction to the source electric field , and will be referred to as
model p in the following .
+ figure [ polariz ] here
the propagation of current in the medium is dominated by ionic diffusion currents or by currents produced by the electric field , according to the values of @xmath106 and @xmath107 with respect to @xmath108 .
the values of @xmath106 and @xmath107 are respectively inversely proportional to the square root of the global ionic diffusion coefficient in the extracellular fluid , and of membrane surface ( see appendix [ secscal ] ) .
we apply the reasoning based on the connex topology of the cortical medium ( see above ) to deduce the order of magnitude of the induced field for zero frequency @xmath88 @xmath109 where @xmath110 because the `` tangential '' conductivity on membrane surface is given by @xmath111 when the current is dominated by either electric field or ionic diffusion ( see eq . [ diffus ] ) .
it follows that the expression for the current density in extracellular space as a function of the electric field is given by @xmath112 because @xmath113 in cortical tissue for frequencies @xmath114 and @xmath103 ( see ref .
@xcite ) .
thus , we have the following expression for the complex conductivity of the extracellular medium : @xmath115 where @xmath116 .
thus , we have obtained a unique expression ( eq . [ apparent ] ) for the apparent conductivity in extracellular space outside of the source , and its frequency dependence due to differential ohm s law , electric polarization phenomena and ionic diffusion .
these phenomena are responsible for an apparent frequency - dependence of the electric parameters , which will be compared to the frequency dependence observed in macroscopic measurements of conductivity ( section [ sec - sim ] ) . finally , eqs .
[ impedance ] and [ apparent ] imply that the macroscopic impedance of a homogeneous spherical shell of width @xmath117 is given by : @xmath118 in the following , this model will be referred as the `` diffusion - polarization '' model , or `` dp '' model , and we will use the above expressions ( eqs .
[ apparent ] and [ impedance1 ] ) to simulate experimental measurements .
we first consider the experiments of gabriel et al .
@xcite who measured the frequency dependence of electrical parameters for a large number of biological tissues . in these experiments
, the biological tissue was placed in between two capacitor plates , which were used to measure the capacitance and leak current using the relation @xmath119 , imposing the same current amplitude at all frequencies . because the admittance value is proportional to @xmath120 ,
measuring the admittance provides direct information about @xmath121 and @xmath122 . to stay coherent with the formalism developed above
, we will assume that the capacitor has a spherical geometry .
the exact geometry of the capacitor should in principle have no influence on the frequency dependence of the admittance , because the geometry will only affect the proportionality constant between @xmath123 and @xmath124 .
in the case of a spherical capacitor , by applying eq.[impedance1 ] , we obtain : @xmath125=4\pi\frac{r_1r_2}{r_2-r_1}\sigma_z^m\ ] ] we also take into account the fact that the resistive part is always greater than the reactive part for low frequencies ( @xmath126 1000 hz ) , which is expressed by @xmath127 this relation can be verified for example from the gabriel et al.measurement @xcite , where it is true for the whole frequency band investigated experimentally ( between 10 and 10@xmath128 hz ) .
the real part of @xmath129 then takes the form @xmath130 where @xmath131 by substituting this value of @xmath132 , the inverse of the conductivity ( the resistivity ) is given by : @xmath133 \label{sigmam}\ ] ] finally , to reproduce gabriel et al .
experiments , we assume @xmath134 . by developing in series the last term ( in parenthesis ) of eq . [ sigmam ] , we obtain : @xmath135 + figure [ puis1 ] here
[ gab ] corresponds to the conductivity @xmath36 , as measured in the experimental conditions of gabriel et al.experiments ( the permittivity @xmath35 is obtained by applying kramers - kronig relations ) .
figure [ puis1 ] shows that this expression for the conductivity can explains the measurements in the frequency range of 10 to 1000 hz , which are relevant for lfps . to obtain this agreement , we had to assume in eq .
[ apparent ] a relatively low maxwell - wagner time of the order of @xmath136 ( @xmath137 between 1 hz and 10 hz ) , @xmath138 ( for frequencies smaller than 100 hz ) .
this value of maxwell - wagner time is necessary to explain gabriel s experiments , and may seem very large at first sight .
however , the maxwell - wagner time is not limited by physical constraints , because we have by definition @xmath139 . in principle , the value of @xmath140 can be very small , approaching zero , while @xmath141 can take very large values .
for example , taking gabriel et al .
measurements in aqueous solutions of nacl and in gray matter @xcite , gives values of @xmath96 comprised between 1 ms and 100 ms for frequencies around 10 hz .
thus , the model predicts that in the gabriel et al .
experiments , the transformation of electric current carried by electrons to ionic current in the biological medium necessarily implies an accumulation of ions at the plates of the capacitor .
this ion accumulation will in general depend on frequency , because the conductivity and permittivity of the biological medium are frequency dependent .
this will create a concentration gradient across the biological medium , which will cause a ionic diffusion current opposite to the electric current .
this ionic current will allow a greater resulting current because surface polarization is opposite to the electric field .
figure [ puis1 ] shows that such conditions give frequency - dependent macroscopic parameters consistent with gabriel et al
. measurements .
the parameter choices to obtain this agreement can be justified qualitatively because the ionic diffusion constant on cellular surfaces is probably much smaller than in the extracellular fluid , such that @xmath142 .
this implies the existence of a frequency band @xmath143 for which @xmath82 is negligible with respect to @xmath107 , but not with respect to @xmath106 because these constants are inversely proportional to the square root of their respective diffusion coefficients .
thus , the approximation that we suggest here is that this band @xmath143 finishes around 100 hz in the gabriel et al
. experimental conditions .
it is important to note that this parameter choice is entirely dependent on the ratio between ionic diffusion current and the current produced by the electric field , and thus will be depend on the particular experimental conditions .
it is interesting to note that the present model and the phenomenological cole - cole model @xcite predict different behaviors of the conductivity for low frequencies ( @xmath12610 hz ) . in the present model , the conductivity tends to zero when frequency tends to zero , while in the cole - cole extrapolation , it tends to a constant value @xcite .
the main difference between these models is that the cole - cole model is phenomenological and has never been deduced from physical principles for low frequencies , unlike the present model which is entirely deduced from well - defined physical phenomena .
we next consider the experiments of logothetis et al .
@xcite , which reported a resistive medium , in contrast with gabriel et al . experiments . in these experiments ,
4 electrodes were aligned and spaced by 3 mm in monkey cortex .
the first and last electrodes were used to inject current , while the two intermediate electrodes were used to measure the extracellular voltage .
the voltage was measured at different frequencies and current intensities .
one important point in this experimental setup is that the intensity of the current was such that the voltage at the extreme ( injecting ) electrodes saturates .
one of the consequences of this saturation was to limit ionic diffusion effects , as discussed by the authors @xcite .
this voltage saturation will diminish the concentration difference near the source ( we would have an amplification if this was not the case ) .
it follows that , in the logothetis et al .
experiments , the ratio between diffusion current and electric field current is very small .
thus , in this case we use values of parameters @xmath144 very different from those assumed above to reproduce gabriel et al . experiments , in particular @xmath145 . as we will see in the discussion section
, this situation may be different from the physiological conditions .
nevertheless , the large distance between electrodes suggests that the relation between current and voltage is linear because the current density is roughly proportional to the inverse of squared distance to the source .
consequently , we can suppose that in the logothetis et al .
experiments , the ionic gradient is negligible , which prevents ionic diffusion currents .
thus , in this experiment , most if not all of the extracellular current is due to electric - field effects . in this situation , the conductivity ( eq . [ apparent ] ) becomes : @xmath146 which is similar to the `` model p '' above .
moreover , taking the same maxwell - wagner time @xmath91 as above for gabriel et al .
experiments ( which corresponds to a cutoff frequency of 1 hz ) , we have for frequencies above 10 hz : @xmath147 similar to a resistive medium . thus , in the logothetis et al
. experiments , the saturation phenomenon entrains current propagation in the biological medium as if the medium was quasi - resistive for frequencies larger than 10 hz .
this constitutes a possible explanation of the contrasting results in logothetis et al . and gabriel et al
. measurements . the third type of experimental observation is the fact that the power spectral density ( psd ) of lfps or eeg signals displays 1/f frequency scaling @xcite . to examine
if this 1/f scaling can be accounted for by the present formalism , we consider a spherical current source embedded in a continuous macroscopic medium .
we also assume that the psd of the current source is a lorentzian , which could derive for example from randomly occurring exponentially decaying postsynaptic currents @xcite ( see fig .
[ dsp ] ) . to simulate this situation we used the `` diffusion - polarization '' model with ionic diffusion and electric field effects in a reactive medium .
we have estimated above that surface polarization phenomena have a cutoff frequency of the order of 1 hz , and will not play a role above that frequency .
so , if we focus on the psd of extracellular potentials in the frequency range larger than 1 hz , we can consider only the effect of ionic diffusion ( in agreement with the gabriel et al .
experiments see above ) .
thus , we can approximate the conductivity as ( see eq . [ apparent ] ) : @xmath148 where @xmath85 is a constant .
it follows that the extracellular voltage around a spherical current source is given by ( see eq . [ impedance ] ) : @xmath149 where @xmath150 is the radius of the source .
+ figure [ dsp ] here in other words , we can say that the extracellular potential is given by the current source @xmath151 convolved with a filter in 1/@xmath152 , which is essentially due to ionic diffusion ( warburg impedance ; see refs .
a white noise current source will thus result in a psd scaling as 1/f , and can explain the experimental observations , as shown in fig .
experimentally recorded lfps in cat parietal cortex display lfps with frequency scaling as 1/f for low frequencies , and 1/f@xmath153 for high frequencies
[ dsp]a - b ) .
following the same procedure as in ref .
@xcite , we reconstructed the synaptic current source from experimentally recorded spike trains ( fig .
[ dsp]c - d ) .
the psd of the current source scales as a lorentzian ( fig .
[ dsp]e ) as expected from the exponential nature of synaptic currents . calculating the lfp around the source and taking into account ionic diffusion ,
gives a psd with two frequency bands , scaling in 1/f for low frequencies , and 1/f@xmath153 for high frequencies
[ dsp]f ) .
this is the frequency scaling observed experimentally for lfps in awake cat cortex @xcite .
we conclude that ionic diffusion is a plausible physical cause of the @xmath0 structure of lfps for low frequencies .
two important points must be noted .
first , the diffusion - plolarization model does not automatically predict @xmath0 scaling at low frequencies , but rather implements an @xmath0 filter , which may result in frequency scaling with larger slopes .
second , the same experimental situation may result in different frequency scaling , and this is also consistent with the diffusion - polarization model .
these two points are illustrated in fig .
[ dsp2 ] which shows a similar analysis as fig .
[ dsp ] but during slow - wave sleep in the same experiment .
the lfp is dominated by slow - wave activity ( fig .
[ dsp2]a ) , and the different units display firing patterns characterized by concerted `` pauses '' ( gray lines in fig .
[ dsp2]b ) , characteristic of slow - wave sleep and which are also visible in the reconstructed synaptic current ( fig .
[ dsp2]c ) .
the psd shows a similar scaling as @xmath154 as for wakefulness , but the scaling at low frequencies is different ( slope around -2 at low frequencies ; see fig .
[ dsp2]d ) .
the psd reconstructed using the diffusion - polarization model displays similar features ( compare with fig .
[ dsp2]e ) .
this analysis shows that the diffusion - polarization model qualitatively accounts for different regions of frequency scaling found experimentally in different frequency bands and network states .
+ figure [ dsp2 ] here in this final section , we examine a possible way to test the model experimentally .
the main prediction of the model is that , in natural conditions , the extracellular current perpendicular to the source is dominated by ionic diffusion .
the experiments realized so far @xcite used macroscopic currents that did not necessarily respect the correct current flow in the tissue .
we suggest to create more naturalistic current sources by creating ionic currents using a micropipette placed in the extracellular medium . by using periodic current injection during very short time
@xmath155 compared to the period ( small duty cycle ) , we can measure using the same electrode the extracellular voltage @xmath156 ( using a fixed reference far away from the source ) .
if the period of the source is shorter than the relaxation time of the system , the voltage will integrate , which is due to charge accumulation . because @xmath156 is directly proportional to the amount of charge emitted as a function of time during @xmath155 ( capacitive effect of the extracellular medium ) , the time variation of @xmath156 is directly proportional to the ionic diffusion current .
in such conditions , if the extracellular medium is purely resistive , as predicted by logothetis et al .
experiments , the relaxation time should be very small , of the order of @xmath157 @xcite , which would prevent any integration phenomena and charge accumulation for frequencies below @xmath158 hz .
if the medium has a slower relaxation , due to polarization and ionic diffusion , then we should observe voltage integration and charge accumulation for physiological frequencies ( @xmath126 1000 hz ) . to illustrate the difference between these two situations , we consider the simplest case of a non - reactive medium ( as in model d above ) , in which the current can be produced by ionic diffusion or by the electric field , or by both . to calculate the time variations of ionic concentration and extracellular voltage
, we consider the current density : @xmath159 + \sigma \vec{e}\ ] ] according to the differential law for charge conservation and poisson law , we have : @xmath160 when ionic diffusion is negligible compared to ohm s law , we have : @xmath161 it follows that the charge density is given by : @xmath162 on the other hand , if ionic diffusion is the primary cause of current propagation in the extracellular medium , then the relaxation time should be much larger and thus , integration should be observed .
when ionic diffusion is dominant , we have : @xmath163 instead of eq . [ pe ] .
the general solution is : @xmath164 the difference between the expressions above ( eqs .
[ sol1 ] and [ sol2 ] ) shows that the time variation of charge density is different according to which current dominates , electric field current or ionic diffusion current .
the same applies to the electric potential between the electrode and a given reference , because this potential is linked to charge density through poisson s law .
therefore , this experiment would be crucial to clearly show which of the two current primes for currents perpendicular to the source ( this would not apply to longitudinal currents , like axial currents in dendrites ) . in the hypothetical case of dominant ionic diffusion ,
the cortex would be similar to a warburg impedance and one can estimate the macroscopic diffusion coefficient using eq .
[ sol2 ] .
thus , using a micropipette injecting periodic current pulses , it should be possible to test the capacity of the medium to create charge accumulation for physiological frequencies .
if this is the case , this would constitute evidence that ionic currents are non - negligible in the physiological situation .
in the present paper , we have proposed a framework to model local field potentials , and which synthesizes previous measurements and models .
this framework integrates microscopic measurements of electric parameters ( conductivity @xmath1 and permittivity @xmath2 ) of extracellular fluids , with macroscopic measurements of those parameters ( @xmath50 , @xmath51 ) in cortical tissue @xcite .
it also integrates previous models of lfps , such as the _ continuum model _
@xcite , which was based on a continuum hypothesis of electric parameters variations in extracellular space , or the _ polarization model _
@xcite , which explicitly considered different media ( fluid and membranes ) and their polarization by the current sources .
the present model is more general and also integrates ionic diffusion , which is predicted as a major determinant of the frequency dependence of lfps .
this `` diffusion - polarization '' model also accounts for observations of @xmath0 frequency scaling of lfp power spectra , which is due here to ionic diffusion , and is therefore predicted to be a consequences of the genesis of the lfp signal , rather than being solely due to neuronal activity ( see ref .
finally , this work suggests that ephatic interactions between neurons can occur not only through electric fields but that ionic diffusion should also be considered in such interactions .
as discussed in section [ simplif ] , the present model rests on several approximations , which were necessary to obtain the analytic expressions used here .
these approximations were that current sources were considered as monopolar entities ( longitudinal currents such as axial currents in dendrites were not taken into account ) , the current source was spherical and the extracellular medium was considered isotropic .
because multipolar effects can be reconstructed from the superposition of monopoles @xcite , the monopolar configuration should not affect the results on frequency dependence , as long as the extracellular current perpendicular to the source is considered .
similarly , the exact geometry of the current source should have no influence on the frequency dependence far away from the sources .
however , in the immediate vicinity of the sources , the geometric nature and the synchrony of synaptic currents can have influences on the power spectrum @xcite .
another assumption is that the extracellular medium is isotropic , which was justified within the macroscopic framework followed here .
these factors , however , will influence the exact shape of the lfps .
more quantitative models including a more sophisticated geometry of current sources and the presence of membrane excitability and action potentials should be considered ( e.g. , see refs .
@xcite ) .
the main prediction of the present model is that ionic diffusion is an essential physical cause for the frequency dependence of lfps .
we have shown that the presence of ionic diffusion allows the model to account quantitatively for the macroscopic measurements of the frequency dependence of electric parameters in cortical tissue @xcite .
ionic diffusion is responsible for a frequency dependence of the impedance as @xmath165 for low frequencies ( @xmath126 1000 hz ) , which directly accounts for the observed @xmath0 frequency scaling of lfp and eeg power spectra during wakefulness @xcite ( see fig .
[ dsp ] ) .
note that the eeg is more complex because it depends on the diffusion of electric signals across fluids , dura matter , skull , muscles and skin .
however , this `` filtering '' is of low - pass type , and may not affect the low - frequency band , so there is a possibility that the 1/f scaling of eeg and lfps have a common origin . the present model is consistent with the view that this apparent `` @xmath0 noise '' in brain signals is not generated by self - organized features of brain activity , but is rather a consequence of the genesis of the signal and its propagation through extracellular space @xcite .
it is important to note that the fact that ionic diffusion may be responsible for @xmath0 frequency scaling of lfps is not inconsistent with other factors which may also influence frequency scaling .
for example , the statistics of network activity and more generally network state can affect frequency scaling .
this is apparent when comparing awake and slow - wave sleep lfp recordings in the same experiment , showing that the @xmath0 scaling is only seen in wakefulness but @xmath166 scaling is rather seen during sleep @xcite ( see fig .
[ dsp2 ] ) . in agreement with this , recent results indicate that the correlation structure of synaptic activity may influence frequency scaling at the level of the membrane potential , and that correlated network states scale with larger ( more negative ) exponents @xcite .
we also investigated ways to explain the measurements of logothetis et al .
@xcite , who reported that the extracellular medium was resistive and therefore did not display frequency dependence , in contradiction with gabriel et al .
@xcite measurements .
we summarize and discuss our conclusions below . in gabriel
experiments @xcite , one measures permittivity and conductivity in the medium in between two metal plates .
this forms a capacitor , which ( macroscopic ) complex impedance is measured .
this measure actually consists of two independent measurements , the real and imaginary part of the impedance .
these values are used to deduce the macroscopic permittivity and macroscopic conductivity of the medium . however , at the interface between the medium and the metal plates , the flow of electrons in the metal corresponds to a flow of charges in the tissue , and a variety of phenomena can occur , which can interfere the measurement .
the accumulation of charges that occurs at the interface between the electrode and the extracellular fluid implies a polarization impedance , which depends on the interaction between ions and the metal plate .
because this accumulation of charge implies a variation of concentration , the flow of ions may involve an important component of ionic diffusion . in logothetis et al .
experiments , a system of 4 electrodes is used , the two extreme electrodes inject current in the medium , while the two electrodes in the middle are exclusively used to measure the voltage .
this system is supposed to be more accurate than gabriel s , because the electrodes that measure voltage are not subject to charge accumulation . however , the drawback of this method are nonlinear effects .
the magnitude of the injected current is such that the voltage at the extreme electrodes saturates .
this voltage saturation also implies saturation of concentration ( capacitive effect between electrodes ) , which limits ionic diffusion currents .
thus , the ratio between ionic diffusion currents and the currents due to the electric field is greatly diminished relative to gabriel et al . experiments .
we think that natural current sources are closer to gabriel et al .
situation for several reasons .
first , the magnitude of the currents produced by biological sources is far too low for saturation effects .
second , the flow of charges across ion channels will produce perturbations of ionic concentration , which will be re - equilibrated by diffusion .
the effects may not be as strong as the perturbations of concentrations induced by gabriel et al .
type of experiments , but ionic diffusion should play a role in both cases .
this is precisely one of the aspects that should be evaluated in further experiments .
the logothetis et al .
experiments were done using a 4-electrode setup which neutralizes the influence of electrode impedance on voltage measurements @xcite .
this system was used to perform high - precision impedance measurements , also avoiding ionic diffusion effects @xcite . indeed , these experimental conditions , and the apparent resistive medium , could be reproduced by the present model if ionic diffusion was neglected .
the present model therefore formulates the strong prediction that ionic diffusion is important , and that any measurement technique should allow ionic diffusion to reveal the correct frequency - dependent properties of impedance and electric parameters in biological tissue .
the critical question that remains to be solved is whether , in physiological conditions , ionic diffusion plays a role as important as suggested here .
we propose a simple method to test this hypothesis .
the frequency dependence could be evaluated by using an extracellular electrode injecting current in conditions as close as possible to physiological conditions ( a micropipette would be appropriate ) . by measuring the integration of the extracellular voltage following periodic current injection
, one could estimate the `` relaxation time '' of the medium with respect to charge accumulation .
if this relaxation time occurs at time scales relevant to neuronal currents ( milliseconds ) rather than the fast relaxation predicted by a purely resistive medium ( picoseconds ) , then ionic diffusion will necessarily occur in physiological conditions , which would provide evidence in favor of the present mechanism .
in this appendix , we evaluate the ratio between ionic diffusion currents and electric field currents in the extracellular space directly adjacent to the source .
this ratio measures if the ionic diffusion current perpendicular to the membrane is greater than the electric field current .
we will design this ratio by the term @xmath167 .
we have in general @xmath168 + \sigma \nabla v ~ , \ ] ] where the first term in right hand is the electric current density produced by ionic diffusion , while the second term is that produced by differential ohm s law . for a displacement @xmath169 in the direction across the membrane ( from inside to outside )
, we have approximatively : @xmath170_{\delta r = 10 nm}+\sigma_{\omega } \delta v_{\delta r = 10 nm}\ ] ] suppose that we have a spherical cell of @xmath171 radius , embedded in sea water and at resting potential .
the resting membrane potential is a dynamic equilibrium between inflow and outflow of charges , in which these two fluxes are equal on ( temporal ) average .
fluctuations of current around this average in the extracellular medium around the membrane have all characteristics of thermal noise @xcite because the shot noise ( see refs .
@xcite and @xcite ) is zero when the current is zero on average , such that the net charge on the external side of the membrane varies around a mean value with the same characteristics as white noise ( thermal noise ) .
these fluctuations will therefore be present also at the level of the membrane potential . in this appendix
, we evaluate the order of magnitude of the electric current caused by ionic diffusion , relative to the electric field for this situation of dynamic equilibrium .
first , the ratio between the membrane voltage noise and the variation of total charge concentration is given by @xmath172 because the cell s capacitance is given by @xmath173 ( @xmath174 ) second , mass conservation imposes : @xmath175_{tot } ~ , \ ] ] where @xmath176 and @xmath177 is the volume of the spherical shell containing the charges . because the charges are not uniformly distributed inside the cell , but rather distributed within a thin spherical shell adjacent to the membrane , because the electric field developed across the membrane is very intense ( of the order of @xmath178 ) ,
thus , the width of the shell is of the order of @xmath179 , where the volume of the spherical shell is approximately equal to @xmath180 . in this case
, we have for monovalent ions ( @xmath181 ) @xmath182_{tot } \simeq 2.2\times10^{-38}\delta[c]_{tot}=2.2\times10^{-38}\delta[c]_{\delta r = 10 nm}\ ] ] if we assume that the variation of concentration on the adjacent border of the exterior surface of the cell is over a width of @xmath183 . in this case
, we have @xmath184_{\delta r = 10 nm}}{\delta v_{membrane } } = \frac{k_1}{k_2 } \simeq 10^{27}~ c / m^3v\ ] ] third , the potential difference between the cell surface and @xmath183 away of it , is given by @xmath185 consequently , the ratio between ionic diffusion current and electric diffusion current caused by thermal noise in sea water obeys @xmath186_{\delta r = 10 nm}\delta t}{\sigma_{\omega}^{sea}\delta v_{\delta r = 10 nm } } = \frac{e d_{sea}k_1}{\sigma_{\omega}^{sea } k_2}\simeq \frac{10^{2}}{\sigma_{\omega}^{sea } } ~ , \ ] ] where the diffusion constant of @xmath187 or @xmath188 in sea water is of the order of @xmath189 .
this implies that the ratio @xmath73 is much larger than 1 for frequencies @xmath1261000 hz because @xmath190 of sea water is necessarily @xmath191 .
because tortuosity is given by @xmath192 , and is comprised between 1.6 and 2.2 ( for small and large molecules , respectively ) in cerebral cortex @xcite , the macroscopic diffusion constant in cortex is certainly larger than @xmath193 .
thus , we have @xmath194 where @xmath195 ( see ref .
@xcite ) .
this evaluation shows that the phenomenon of ionic diffusion is essential to determine the current field in the cortex .
finally , we note that we did not need to evaluate the absolute magnitude of @xmath196 in our evaluation .
this evaluation is valid for a physical situation where we have a permanent white noise over a distance of 10 nm , independently of the intensity of this noise ( which in practice will depend on many factors , such as the size of the cell , the number of ion channels , etc ) .
in this appendix , we calculate the frequency dependence of ionic diffusion current outside of a spherical current source .
we consider a constant variation of ionic concentration , @xmath197 , on the surface of the source , and a null variation at an infinite distance ( warburg conditions ) .
the diffusion equation for a given ionic species is @xmath198 where @xmath197 is the perturbation of concentration @xmath199 of ion @xmath200 around the steady - state value , and @xmath201 is the associated diffusion coefficient .
this diffusion coefficient depends of the ionic species considered and the structure of the medium .
because the geometry of the problem and the limit conditions respect spherical symmetry , we use spherical coordinates . in this coordinate system
, we have @xmath202\ ] ] because @xmath197 does not depend on @xmath203 and of @xmath204 ( spheric symmetry ) .
the fourier transform of @xmath197 with respect to time gives : @xmath205 which general solution is given by : @xmath206 for a variation of concentration at the source border which is independent of frequency and which satisfies the warburg hypothesis ( the variation of concentration tends to zero at an infinite distance @xcite ) , we have : @xmath207 where @xmath73 is the distance between the center of the source and @xmath150 is the radius of the source . thus , the electric current density produced by ionic diffusion is given by : @xmath208 where @xmath209 is the charge of ions @xmath200 . because we can consider that the source and extracellular medium form a spherical capacitor , the voltage difference between the surface of the source and infinite distance is given by @xmath210 where @xmath211 is the capacitance value .
thus , the electric impedance of the medium is given by : @xmath212 for a source of radius @xmath213 and a macroscopic ionic diffusion coefficient of the order of @xmath214 , and for frequencies @xmath215 1 hz , we can approximate the impedance by : @xmath216 the same expression for the impedance is also obtained in cylindric coordinates or planar cartesian coordinates ( not shown ) .
note that if several ionic species are present , then the superposition principle applies ( fick equations are linear ) and therefore the contribution of each ion will add - up .
the diffusion constants for different ions are of the same order of magnitude ( for na@xmath217 , k@xmath217 , cl@xmath218 , ca@xmath219 ) , so no particular ion would be expected to dominate .
we thank christoph kayser , nikos logothetis , axel oeltermann and anonymous reviewers for useful comments on the manuscript . research supported by cnrs , anr and the european community ( facets project ) .
99[bibliographie - debut ] bhattacharya , j. , and h. petsche .
universality in the brain while listening to music .. * 268 * : 2423 - 2433 .
bdard , c. , h. krger and a. destexhe . 2004 . modeling extracellular field potentials and the frequency - filtering properties of extracellular space .
_ biophys .
j. _ * 86 * : 1829 - 1842 .
bdard , c. , h. krger , and a. destexhe . 2006 .
model of low - pass filtering of local field potentials in brain tissue .
_ physical review e_. * 73 * : 051911 .
bdard , c. , h. krger , and a. destexhe . 2006 .
does the 1/f frequency scaling of brain signals reflect self - organized critical states ? .
_ physical review lett . _ * 97 * : 118102 .
bdard , c. and a. destexhe . 2008 .
a modified cable formalism for modeling neuronal membranes at high frequencies .
_ biophys .
j. _ * 94 * : 1133 - 1143 .
beggs , j. , and d. plenz .
neuronal avalanches in neocortical circuits .
_ j. neurosci._. * 23 * : 11167 - 11177 .
berg , h.c .. 1983 .
_ random walks in biology . _ princeton university press , princeton .
braitenberg , v. and a. shutz .
cortex : statistics and geometry of neuronal connectivity . _
( 2nd ed . ) , springer - verlag , berlin , germany .
buckingham , m.j.j .
_ noise in electronic devices and systems .
_ john wiley & sons , new york .
cole , k.s . and r.h .
1941 . dispersion and absorption in dielectrics i. alternating current characteristics .
_ j. chem . phys._. * 9 * : 341 - 351 .
destexhe , a. , d. contreras and m. steriade .
spatiotemporal analysis of local field potentials and unit discharges in cat cerebral cortex during natural wake and sleep states . _
j. neurosci._. * 19 * : 4595 - 4608 .
diard , j - p . , b. le gorrec , c. montella .
linear diffusion impedance .
general expression and applications .
_ j. electroanalytical chem . _ . *
471 * : 126 - 131 .
foster , kr .
, hp . schwan .
dielectric properties of tissues and biological materials : a critical review .
reviews biomed . engineering_. * 17 * : 25 - 104 .
gabriel , s. , r.w .
lau , and c. gabriel .
the dielectric properties of biological tissues : i. literature survey .
_ phys . med .
biol._. * 41 * : 2231 - 2249 .
gabriel , s. , r.w .
lau , and c. gabriel .
the dielectric properties of biological tissues : ii .
measurements in the frequency range 10 hz to 20 ghz .
_ phys . med .
biol._. * 41 * : 2251 - 2269 .
gabriel , s. , r.w .
lau , and c. gabriel .
the dielectric properties of biological tissues : iii .
parametric models for the dielectric spectrum tissues .
_ phys . med .
biol._. * 41 * : 2271 - 2293 .
geddes , l.a .. 1997 .
historical evolution of circuit models for the electrode - electrolyte interface .
eng._. * 25 * : 1 - 14 .
gold , c. , d.a .
henze , c. koch , and g. buzsaki .
2006 . on the origin of the extracellular action potential waveform : a modeling study .
_ j. neurophysiol._. * 95 * : 3113 - 3128 .
hooge , f.n .
p.a . bobbert .
on the correlation function of 1/f noise . _ physica b _ * 239 * : 223 - 230 .
hooge , f.n .
1/f noise is no surface effect .
_ physics letters _ * 29a * : 139 - 140 .
jensen , h.j .. 1998 .
_ self - organized criticality : emergent complex behavior in physical and biological systems . _ cambridge university press , cambridge , uk .
kronig , r.d.l .. 1926 . on the theory of dispersion of x - rays .
soc . am._.
* 12 * : 547 .
logothetis , nk , c. kayser , a. oeltermann .
2007 . in vivo measurement of cortical impedance spectrum in monkeys : implications for signal propagation .
_ neuron _ * 55 * : 809 - 823 .
landau , l.d . and e.m .
_ electrodynamics of continuous media . _ pergamon press , moscow , russia .
mcadams , e.t . and
j. jossinet . 1992 . a physical interpretation of schwan s limit current of linearity .
engineering . _ * 20 * : 307 - 319 .
novikov , e. , a. novikov , d. shannahoff - khalsa , b. schwartz , and j. wright .
scale - similar activity in the brain .
_ phys . rev . e_. * 56 * : r2387-r2389 .
nunez , p.l .
, a. novikov and r. srinivasan .
_ electric fields of the brain .
_ ( 2nd ed ) oxford university press , oxford , uk .
nyquist , h .. 1928 .
thermal agitation of electric charge in conductors .
_ phys . rev._.
* 32 * : 110 - 113 .
pettersen , k.h . and g.t .
amplitude variability and extracellular low - pass filtering of neuronal spikes .
_ biophys j. _ * 94 * : 784 - 802
. pritchard , w.s .. 1992 .
the brain in fractal time : @xmath0-like power spectrum scaling of the human electroencephalogram .
j. neurosci . _
* 66 * : 119 - 129 .
purcell , e.m .. 1984 .
_ electricity and magnetism .
_ mcgraw hill , new york .
taylor , s.r . and e. gileadi
the physical interpretation of the warburg impedance .
_ corrosion _ * 51 * : 664 - 671
. vasilyev , a.m .. 1983 . _ an introduction to statistical physics .
_ mir editions , moscow , russia .
[ bibliographie - fin ]
+ figure [ polariz ] | the power spectrum of local field potentials ( lfps ) has been reported to scale as the inverse of the frequency , but the origin of this `` @xmath0 noise '' is at present unclear .
macroscopic measurements in cortical tissue demonstrated that electric conductivity ( as well as permittivity ) is frequency dependent , while other measurements failed to evidence any dependence on frequency . in the present paper , we propose a model of the genesis of lfps which accounts for the above data and contradictions .
starting from first principles ( maxwell equations ) , we introduce a macroscopic formalism in which macroscopic measurements are naturally incorporated , and also examine different physical causes for the frequency dependence .
we suggest that ionic diffusion primes over electric field effects , and is responsible for the frequency dependence .
this explains the contradictory observations , and also reproduces the @xmath0 power spectral structure of lfps , as well as more complex frequency scaling
. finally , we suggest a measurement method to reveal the frequency dependence of current propagation in biological tissue , and which could be used to directly test the predictions of the present formalism . |
the temperature - density phase diagram drastically changes with the range of the attractive part of the interparticle potential @xcite .
so that , when it is less than half of the particle s radius , the critical point locates below the freezing curve and the corresponding vapor - liquid phases are metastable @xcite .
this inhibits crystal formation . to overcome this obstacle ,
it was proposed to take advantage of natural fluctuations occurring in the vicinity of the critical point , which may produce the first stable seed to grow the crystal phase @xcite .
this pathway for crystallization was unsuccessfully applied to globular proteins solutions , which are known to follow the phase behavior described by short - range attractive model fluids @xcite . alternatively , temporal variations of protein concentration enhance crystallization , pointing out that fluctuations , though externally imposed , indeed promote nucleation and the crystal growth @xcite .
computer simulations of short - range interacting fluids easily access and remain in the metastable region of the phase space , which signals a free energy barrier between the vapor - liquid and vapor - solid states @xcite . at sufficiently low temperatures
, a liquid droplet evolves towards an imperfect solid made up of very small crystallites glued trough defective grain boundaries @xcite . on the other hand
, it has been shown that crystallization is enhanced for the patchy square well potential fluid . in that case
crystallization takes place after the formation of a high density droplet of self - assembled particles and well outside the metastable vapor - liquid region @xcite . in this work
we consider assisting the solid growth of an isotropic short - range interacting fluid by tempering its metastable vapor - liquid phases . in this way
we expect the system to approach to the vapor - solid ground state .
we also study the possibility of generating the solid seed at the vapor phase . to deal with these issues
, we employed a simulation algorithm that imposes sudden temperature changes over several system replicas .
monte carlo ( mc ) techniques are designed to sample form equilibrium .
however , standard mc in some cases can not drive the system to its ground state due to the existence of deep free - energy minima . in such cases ,
the replica exchange monte carlo ( remc ) method frequently improves the sampling @xcite . in this work
we combine the remc method with the slab technique @xcite , i.e. , a system initially containing a liquid slab surrounded by its vapor is replicated for different temperatures ( parallel tempering ) .
the general idea is to simulate @xmath0 replicas of the original system , being each replica at a different temperature , so that , the exchange of microstates among the cells is allowed ( swap moves ) .
these swap moves will make the replicas to follow some particular temperature routes , which can be then replicated by other simulation techniques .
thus , we use remc to expose the replicas in the metastable state to temperature fluctuations . by so doing ,
we expect some of these replicas to escape from metastability .
let us consider a model fluid made up of spherical particles at number density @xmath1 , interacting through the attractive square - well ( sw ) potential , given by @xmath2 for @xmath3 , @xmath4 for @xmath5 , and @xmath6 for @xmath7 ; being @xmath8 the particles diameter , @xmath9 and @xmath10 the well depth and width , respectively . for simplicity , we take the particles diameter as the unit length and @xmath9 is given in units of @xmath11 , being @xmath12 the boltzmann constant and @xmath13 the absolute temperature .
the corresponding reduced number density and temperature are given by @xmath14 and @xmath15 , respectively .
we employed @xmath16 parallelepipeds with sides @xmath17 and @xmath18 for simulating the vapor - liquid interface of the sw fluid .
thus , each simulation box is initially set with all particles randomly placed within the slab , in such a way that the particles concentration is @xmath19 0.7 and the slab is initially surrounded by vacuum .
this initial configuration is close to a metastable liquid - vapor state where liquid and vapor have similar volumes , which allows to search for seeding at both phases .
in addition , we also set the bulk dense phase having a thickness larger than @xmath20 .
this is , in our experience , enough to avoid a shift in the phase diagram due to the presence of the interfaces between the dense and the vapor phases .
periodic boundary conditions are set in the three directions .
the center of mass is placed at the box center .
verlet lists are implemented to improve performance .
the highest temperature is set near but below the critical temperature .
other temperatures are fixed by following a geometrically decreasing trend .
the swap acceptance rate is , in general , close to 20 % .
the replicas perform @xmath21 trials , while maximum displacements are varied to yield acceptance rates close to 30 % .
long displacement trials are also considered .
these displacements are important since they allow large jumps in the vapor phase with relatively large acceptance rates , while naturally carry out particle transference between both phases .
after that , maximum displacements turn fixed and the thermodynamic properties are calculated over additional @xmath21 configurations .
this study consists of computing the density profiles , radial distribution functions , average number of neighbors , and the order parameter @xmath22 , as a function of the temperature .
the average number of first neighbors , @xmath23 , is the average number of pairs having a center - center distance smaller than @xmath24 ( the vectors joining the centers of these pairs are named bonds ) .
the order parameter @xmath22 is defined as @xcite @xmath25 where @xmath26 is the average over all bonds and configurations of the spherical harmonics of the angles @xmath27 and @xmath28 ( these are the spherical angles of the bonds measured with respect to any fixed coordinate system , since @xmath22 is invariant ) .
from fig . [ profiles ] it is observed that , as temperature decreases , at @xmath29 the density profiles display an abrupt increase at the condensed phase while at the vapor phase a decrease occurs . the condensed phase of the fluid , with density @xmath30 ,
is located at the box central region whereas the vapor phase , with density @xmath31 , is contiguous to both sides of the condensed phase .
the profile corresponding to the highest temperature ( @xmath32 ) produces the lowest value of @xmath30 and the highest for @xmath33 , whereas the one corresponding to the lowest temperature ( @xmath34 ) yields the largest @xmath30 and smallest @xmath33 .
all the other density profiles are for intermediate temperatures .
thus , the density profiles appear divided in two classes : the high temperature class groups the density profiles with @xmath30 and @xmath33 in agreement with typical densities of liquid and vapor , respectively .
this kind of density profiles corresponds to the metastable vapor - liquid coexistence .
the low temperature class contains the higher values of @xmath30 and lower values of @xmath33 .
a distinction between the two classes of density profiles is the sharply defined interface for those at low temperature and the diffuse interface for those at high temperature .
the latter is typical for a vapor - liquid coexistence , where the interface width increases with temperature .
for the first simulation steps all density profiles behave like the high temperature class , since initial configurations are set in metastable states , i. e. , a liquid slab with density close to 0.7 and its corresponding vapor . at this stage
, all swap trials have acceptance rates over 20% . after a relatively large number of steps , driven by the temperature variations ,
a given replica escapes from its initial state towards its ground state , which give rise to the second kind of profiles .
the breaking away from the metastable state is accompanied by a strong decrease of the configuration energy , which makes the remc technique to preferable locate it at the lowest temperature .
a similar process occurs for other replicas until the remaining high temperature replicas unlikely access temperatures below @xmath35 . in parallel ,
a free energy gap among the replicas at the metastable region and those closer to their ground state is developed . due to this gap the swap moves between the high and low temperature replicas
turn practically forbidden .
consequently , the high temperature replicas do not access low temperatures states and so , they can not escape from the vapor - liquid metastable region . under this rationale ,
not all replicas break away towards the ground state , but just some of them , implying that the whole system of replicas does not reach equilibrium .
however , after approximately @xmath36 trials the whole system becomes stationary and thus , sampling is carried out .
the well defined condensed - phase and vapor regions directly lead to the vapor and condensed densities , @xmath33 and @xmath30 , by taking average at the different regions of the profiles .
this is done without the need of fitting a hyperbolic function , as it is commonly used .
[ phase - diagram ] shows the coexistence phase diagram obtained by plotting @xmath33 and @xmath30 as a function of temperature .
the horizontal line points out the discontinuity of the @xmath33 and @xmath30 branches at @xmath29 .
such a discontinuity separates the two density profile classes discussed in fig .
[ profiles ] . for the upper part of the diagram we employed the exponential fitting ( with @xmath37 ) to estimate the critical point @xcite .
we obtained @xmath38 and @xmath39 for the critical temperature and density , respectively .
as temperature decreases ( in fig . [ phase - diagram ] )
, there is an abrupt shift from vapor - liquid to a denser phase - vapor coexistence .
for @xmath40 we get the @xmath30 and @xmath33 results shown under the horizontal line .
these branches correspond to the low temperature class concentration profiles in fig .
[ profiles ] .
the density values of these branches indicate that the low temperature condensed phase is a solid .
this means that the imposed temperature drops on the replicas aid the system to avoid the free energy barrier between the metastable and the ground state .
that is , the crystal seeds are produced at low temperatures , which makes some of these microstates to appear at the other side of the free energy barrier when they are reset at high temperature in the swapping process . above the discontinuity
, our results well agree with those previously reported and carried out without interfaces @xcite .
the results of ref .
@xcite are shown as triangles in fig .
[ phase - diagram ] .
our results also agree with previous calculations using the slab technique@xcite . below the discontinuity , our data approach to the stable vapor - solid region described by liu _
( diamonds in fig .
[ phase - diagram ] ) .
actually , the vapor density branch falls over the interpolation of liu s data , which were obtained by imposing a vapor - solid interface @xcite . in our case the vapor - solid interface is spontaneously formed .
the observed agreement and the free energy decreasing confirms our assertion that above and below @xmath41 the data correspond to the metastable and stable coexistence diagrams , respectively . to characterize the high density phase , the first neighbor number , @xmath23 , and order parameter @xmath22 were followed as a function of temperature .
the results are shown in fig .
[ nn - q6 ] , where a vertical line is plotted at @xmath41 .
these quantities were obtained only for the condensed phase , and avoiding the interfaces . for @xmath42
, we obtained @xmath43 , and @xmath44 , which are the typical values for a liquid . at @xmath41 , @xmath23 and @xmath22 display a discontinuity : @xmath23 jumps from 7.95 to 11.20 , whereas @xmath22 from 0.064 to 0.400 .
it is known that the order parameter , @xmath22 , tends to @xmath45 for a completely random arrangement of particles , 0.5745 for a face cubic centered ( fcc ) crystal , 0.4848 for a hexagonal close packed ( hcp ) structure , and 0.3536 for a simple cubic geometry ( being @xmath46 the number of pairs considered for the calculation ) .
for @xmath47 we get @xmath48 and @xmath49 , which signatures the presence of a solid condensed phase constituted by fcc or / and hcp crystal domains .
we gain further insight of the condensed phase structure by building the radial distribution functions ( rdfs ) .
this is shown in fig .
[ gdrs ] for the lowest and highest computed temperatures .
the high temperature rdf shows the typical behavior for a sw liquid , i. e. , there are two discontinuities at @xmath50 and @xmath51 , followed by two broad peaks at @xmath52 and @xmath53 . at the location of the sw
the rdf has large values , being the largest at contact , @xmath50 , and the lowest at @xmath51 .
there is a nearly linear decreasing trend at the sw region . on the other hand
, the rdf for the lowest temperature shows several well defined peaks , pointing out a long range order . in this case
, the highest peak is at @xmath54 , and the following are at @xmath55 , 1.85 , 2.13 , and 2.40 , respectively , which are slightly shifted to the right with respect to those corresponding to a fcc crystal @xcite .
this is in common for all obtained solid phases .
this result contrasts with the rapid decay of the rdf for the extreme polycrystalline solid produced in the absence of temperature fluctuations , which has a density slightly above that of the liquid and a fractal structure @xcite . in practice ,
an equivalent procedure of the remc simulation technique could be reached by successively imposing increasing and decreasing variations of temperature on a colloidal suspension .
so that , in this way one could expect to produce a crystal seed at low temperature , which then can grow with the imposed fluctuating temperature .
the growing process incorporates particles and new grains at the boundary of the growing nucleus @xcite . at this point , a large temperature aids reorienting the adjacent grains and relaxing their boundaries , however , a large exposure may lead to its total melting . on the other hand ,
a low temperature fastens the growing processes with the consequent incorporation of defects . hence , tempering combines the reorienting , boundary relaxing , and growing processes of the nuclei , yielding a more coherent solid .
temperature fluctuations may also aid with subsequent grain coalescence processes .
finally , we included two representative snapshots of the vapor - liquid ( top ) , and vapor - solid ( bottom ) coexisting phases in the inset of fig .
[ gdrs ] . in these snapshots
many of the effects discussed throughout the paper are visualized . that is , the sharply defined vapor - solid interface ( bottom ) , the fcc solid structure ( bottom ) , the liquid structure ( top ) , and a more dense vapor phase coexisting with the liquid than with the solid ( top and bottom ) .
the crystalline structure explains the presence and absence of oscillations in the vapor - solid density profiles of fig .
[ profiles ] : if any of the crystal planes is parallel to the interface ( @xmath56-plane ) , oscillations appear , whereas they vanish in any other case .
this is why several low temperature profiles in fig .
[ profiles ] are flat even though they correspond to crystalline structures .
the crystal growth can not be controlled by the method , and so , the crystal planes may or may not be parallel to the interface .
the snapshots also show the aggregation of vapor particles at both temperatures .
however , no crystal nuclei are observed in the vapor phase in any case though the vapor is set relatively large in all cases .
we studied the response of the sw fluid metastable liquid - vapor coexistence to temperature fluctuations under the scheme of the replica exchange monte carlo algorithm .
even though we imposed a large vapor volume , the solid seeding was observed always inside the liquid phase .
this supports the hypothesis that the formation of a dense droplet is a necessary condition to produce the crystal @xcite .
the obtained solid is in all cases a nearly fcc monocrystal in opposition to the highly polycrystalline structure obtained by a simple quenching process @xcite .
our results suggest that following the seed formation at low temperatures , the crystal phase can grow and relax approaching to the ground state when temperature is increased and decreased successively .
the authors thank s. fraden for a fruitful discussion of some key ideas presented here . | the metastable vapor - liquid coexistence of short - range attractive fluids hinders the formation of crystal nuclei , which in turn makes difficult the progress of the system towards its vapor - solid ground state . in this letter
we show that crystal growth can be assisted by imposing temperature fluctuations . by
so doing the obtained solid is nearly a fcc monocrystal in contrast with the extreme polycrystalline structure obtained at low temperatures .
the study is carried out by combining the replica exchange monte carlo method and the standard slab technique .
the obtained results suggest a pathway for growing coherent crystals from the metastable liquid .
this is particularly relevant for the crystallization of globular proteins . |
complex hierarchical systems admit several levels of description . at a microscopic level ,
the interest lies on the dynamical behavior of the constituent elements .
thus , at this level of description , one needs to follow the time evolution of very many microscopic variables . on the other hand , at the macroscopic level
, one describes the system in terms of just a few collective variables characterizing it as a whole .
the laws describing the behavior of collective variables could , in principle , be deduced from the dynamical laws governing the elementary constituents . in practice
this is not an easy task as , in general , the dynamical equations for the collective variables can not be obtained from the microscopic dynamics without further approximations .
on the other hand , in general , when studying a complex systems , there are just the collective variables which are accessible to experimental measurements .
the detailed dynamics of the constituent elements has to be inferred from the information about the collective variables .
a fundamental property common to all complex systems made of a collection of interacting elements ( subsystems ) is the existence of feedback mechanisms of the whole system behavior into the dynamics of the subsystems @xcite .
the evolution of the system is determined by those of its elements in a statistical way and , at the same time , the feedback loops put the subsystems under the control of the global system .
such feedback loops , for example , give rise to the biological laws that characterize the phenomenon of life .
in fact , this mutual interlevel control introduces a self - regulation mechanism to the dynamics of the hierarchical system against external forces and boundary conditions . in this work ,
we have studied two very simple models describing a finite system composed of several subsystems .
each of them is characterized by a single stochastic variable @xmath0 .
the subsystems have an intrinsic nonlinear bistable dynamics and they are coupled among them either in a global or local form .
the whole system is characterized by a collective variable .
we will assume that the information about the system comes from the observation of the collective variable .
the randomness of the subsystem variables is due to a noise term in their dynamics .
we will take this noise to be a white gaussian noise characterized by a noise parameter @xmath1 .
this noise can originate for instance , from the weak interactions between the subsystems and a large thermal bath with a temperature @xmath2 .
a goal of this work is to study the form of the distribution function @xmath3 for the collective variable and to compare it with the distribution function of a single element @xmath4 for small finite systems . as we will see , in the absence of feedback , both distributions are unique , regardless of the initial preparation of the system .
this is a consequence of the fact that the joint probability distribution satisfies a linear fokker - planck equation .
if the observations of the collective variable are compatible with such a behavior , then we can accept the microscopic description .
but , if one observes that , depending upon the noise value and the strength of the subsystems coupling , the equilibrium distribution might not be unique and it might depend on the initial preparation of the system , then the linear fokker - planck equation ( or equivalently , the set of coupled langevin - like equations ) is not adequate . to deal with such a situation
, we propose stochastic dynamics leading to nonlinear fokker - plank equations .
the idea is simply an extension of the old weiss mean - field approach to the equilibrium properties of magnetic systems .
then , as the system parameters are varied , the dynamics leads us from just a single stationary solution to situations where two distributions are stable and the system goes to one or the other depending upon the initial conditions .
we should point out that such a non - uniqueness of a single variable distribution function was studied long ago by several authors @xcite .
they were able to derive a nonlinear fokker - planck equation for a single variable in the limit of an infinite system .
we emphasize that , in this work , we are interested in small , finite systems .
the intrinsic bistable nonlinearity of the subsystems dynamics forbids explicit analytical solutions , but it is essential for the behavior indicated above
. a closed , explicit , stochastic evolution equation for the collective variable can not be found from the subsystems dynamics .
we will then rely on numerical simulations of the stochastic evolution equations for the subsystems variables .
we consider a system formed by a set of @xmath5 subsystems each one described by a single stochastic variable @xmath6 , whose dynamics is given by the coupled langevin equations ( in dimensionless form ) @xmath7 where @xmath8 is the strength of the coupling and @xmath9 are gaussian white noises with @xmath10 this model was introduced some years ago by kometani and shimizu @xcite in a biophysical context and was later analyzed by desai and zwanzig @xcite and dawson @xcite from a more general statistical mechanical perspective .
an alternative formulation can be casted in terms of the linear fokker - planck equation for the joint probability distribution @xmath11 , @xmath12 where @xmath13 is the potential energy relief , @xmath14 with the single particle potential @xmath15 in the limit @xmath16 , one can write a nonlinear fokker - planck equation ( nlfpe ) for the one particle probability distribution , @xmath17 @xcite @xmath18p_{1}(x , t)\big\}+d\frac{\partial^{2}p_{1}(x , t)}{\partial x^{2 } } , \label{langinf}\ ] ] where @xmath19 in this limit , the system undergoes an order - disorder phase transition signaled by a change in the form of the equilibrium one - particle probability distribution @xmath4 and by the fact that for some values of the parameters @xmath8 and @xmath1 , the ergodicity is broken and there are more than one stable equilibrium distribution .
the system ends up in one of them depending on its initial preparation . in fig .
[ fig1 ] we sketch different regions in the parameter space @xmath20 and @xmath21 , separated by a transition line . below the transition line
, there is just one single distribution , regardless of the initial preparation of the system , leading to a single equilibrium @xmath4 with two maxima .
this is indicated in the figure by drawing a double well potential . on the other hand , above the transition line , two stable monomodal equilibrium distributions are possible . depending upon the initial condition the system relaxes to one or the other .
this feature is illustrated in the figure by drawing two asymmetrical single minima potentials .
let us define a collective variable @xmath22 then , the langevin equations ( [ eq001 ] ) can be casted in the equivalent form @xmath23 the nonlinearity of equations ( [ langfin ] ) prevent us from writing a closed langevin equation for @xmath24 .
note that there is a feedback of the collective dynamics in the langevin dynamics of each degree of freedom .
but , as we analyzed in a previous work @xcite , with this type of feedback for the finite model the single variable distribution is unique for any set of parameter values .
this is consistent with the fact that the joint probability distribution @xmath25 satisfies a linear fokker - plank equation with a unique distribution function for any values of the parameters .
we now propose to replace the model with @xmath26 here , @xmath27 represents a noise average of the collective variable defined in eq .
( [ eq007 ] ) . clearly , the langevin equation , eq .
( [ eq100 ] ) has to be solved concurrently with the evaluation of this average .
the proposal amounts to replace in the dynamical evolution of each individual degree of freedom the whole stochastic process @xmath24 by its noise average , in the spirit of weiss mean - field approximation to magnetism .
note that within this model , the collective variable is itself a fluctuating quantity , even though just its average value influences the subsystems dynamics .
this model is described by the ( dimensionless ) set of langevin equations @xmath28 with the periodic conditions @xmath29 and @xmath30 . the noises @xmath31 are white gaussian with the properties given in eq .
( [ eq002 ] ) . as in the global interaction model ,
the joint probability distribution function satisfies a linear fokker - planck equation similar to the one in eq .
( [ fpe ] ) but with the potential energy relief , @xmath32 then , also for this case , the joint probability distribution @xmath25 satisfies a linear fokker - plank equation with a unique distribution function for any values of the parameters . as in the global interaction case , we will now replace eq .
( [ nn1 ] ) with @xmath33 as said above , in both situations , even after using the weiss approach , the collective variable @xmath34 for @xmath5 finite is still a random process with a probability distribution defined by @xmath35 notice that within the mean field dynamics , the joint probability distribution @xmath36 factorizes as a product of single particle distributions , each of them satisfying a nlfpe similar to the one in ( [ langinf ] ) for the single particle distribution in the infinite size limit . + defining the characteristic function @xmath37 of a distribution @xmath38 as @xmath39 one can easily see that the generating function of the collective variable and that of a single particle variable are related by @xmath40 thus , the corresponding cumulant generating functions are related by @xmath41 .
then , the cumulants @xmath42 associated to the stochastic process @xmath24 , and those associated to the @xmath43 process , @xmath44 are related by @xmath45 in the limit of very large @xmath5 , @xmath46 becomes a very narrow peak around its first moment as indicated in @xcite .
we have solved numerically the corresponding langevin equations , eqs .
( [ langfin],[nn1 ] ) and eqs.~ ( [ eq100],[eq200 ] ) for systems with a small number of subsystems using very many noise realizations .
the numerical method used has been detailed elsewhere @xcite . after a convenient relaxation time so that the equilibrium probability distributions have been reached , we construct histograms that approximate the equilibrium probability distributions @xmath47 and @xmath4 .
notice that the generation of langevin trajectories for all the degrees of freedom for the mean field dynamics requires the knowledge of @xmath48 .
thus , at each time step during the evolution , this noise average is evaluated from the instantaneous values of each subsystem variable for each realization of the noise .
namely , @xmath49 where @xmath50 is the number of noise realizations considered .
let us first consider the case of global coupling with mean - field dynamics . in fig .
[ ps ] we depict the equilibrium distribution of the collective variable @xmath47 for several values of the noise strength and two system sizes , @xmath51 ( upper panel ) and @xmath52 ( lower panel ) . in all cases , we have used @xmath53 and initial conditions such that @xmath54 . as seen in the figure , the qualitative behavior is quite independent of @xmath5 , except for the narrowing of the distribution as @xmath5 increases . for the parameter values considered ,
@xmath47 is always monomodal , with a peak that is displaced from @xmath55 to higher values of @xmath56 as the noise is decreased .
note that the peak is located at positive values of @xmath56 because , at the initial time , we located the value of @xmath57 to be positive . had we started from the same initial values , but with a negative sign , the plots for @xmath47 would be as in fig .
[ ps ] , but with the peaks located at negative @xmath56 values . in fig .
[ px ] , we depict the equilibrium distribution for a single variable @xmath4 for several values of the noise strength and for two systems with @xmath51 ( upper panel ) and @xmath52 ( lower panel ) . in all cases ,
we have used @xmath53 .
again , the qualitative behavior is quite independent of @xmath5 .
by contrast with the behavior of @xmath47 , the single variable distribution @xmath4 changes its shape from a bimodal distribution centered at @xmath58 to a monomodal one , with a peak that is displaced to nonzero values of @xmath59 as the noise is decreased . in figs .
[ sm ] and [ xm ] we depict the behavior of the equilibrium first and second cumulant moments of the collective variable and that of an individual variable for a system with @xmath51 degrees of freedom and mean - field global dynamics .
the parameter values are indicated in the captions .
the average values have been obtained from the corresponding numerical integration involving the equilibrium histograms constructed from the langevin simulations . as expected from eq .
( [ cum ] ) , the first moment of the individual and the collective variables are the same , while the second cumulant for the collective variable is smaller than that of the individual one .
for the parameter values used in these figures , the probability distribution centered at @xmath55 ( or @xmath58 ) is unstable for @xmath1 smaller than a bifurcation value . in the figures
we have just depicted that branch of the stable first moments which are reached from the imposed initial condition @xmath54 .
the case of nearest neighbors interactions is qualitatively similar to the case of global interactions .
we will just depict as an example the contrast in the behavior of @xmath47 for the finite model in eq .
( [ nn1 ] ) and the corresponding one with the mean field dynamics eq .
( [ eq200 ] ) . in fig .
[ nnfin ] we plot the behavior of @xmath47 and @xmath4 for a system with @xmath51 and nearest neighbors coupling . in fig .
[ nnmf ] the same distributions are sketched but now with the systems described by the mean - field dynamics . in both dynamics
there is a change in the shape of the equilibrium distribution as the noise value is varied .
but for the finite case described by eq .
( [ nn1 ] ) the equilibrium distribution is always unique regardless of the initial preparation . on the other hand ,
the mean field dynamics leads to a bifurcation of the distribution as @xmath1 is varied . for large values of @xmath1
there is a single stable @xmath47 . as @xmath1
is decreased while keeping @xmath8 constant , the zero centered distribution becomes unstable and two stable distributions appear with peaks located at @xmath60 or @xmath61 depending on the initial preparation .
we have carried out numerical simulations to study some aspects of the behavior of complex stochastic systems formed by a finite number of subsystems with bistable intrinsic dynamics and mean field couplings .
we have contrasted the behavior of each individual degree of freedom with that of a collective variable characterizing the entire system .
the systems considered have finite sizes .
then , the collective variable is also a stochastic variable with a probability density with a finite width , in contrast with what occurs in the infinite size limit .
furthermore , the mean field dynamics is compatible with the possible existence of several probability distributions in some regions of the parameter space .
there is a transition from a single stable distribution to the several stable ones as the noise strength is varied for a fixed value of the coupling parameter . in the case of multiple distributions ,
the one that is observed depends upon the initial preparation .
we argue that if the stochastic collective variable is the accessible one , and it is such that its average value depends upon the initial conditions , the modelling of the subsystems dynamics requires the introduction of some sort of feedback of the average collective behavior on the individual dynamics .
an example of this feedback is the mean field dynamics considered in this paper .
it should be pointed out that in order to have the above mentioned transition between one or several distributions , the intrinsic bistable dynamics of each individual degree of freedom is essential .
99 k. kometani and h. shimizu , _ j. stat .
* 13 * , 473 ( 1975 ) .
r. c. desai and r. zwanzig , _ j. stat .
* 19 * , 1 ( 1978 ) .
d. a. dawson , _ j. stat .
phys . _ * 31 * , 29 ( 1983 ) .
m. shiino , _ phys .
a_*36*,2393 ( 1986 ) .
t. d. frank , nonlinear fokker - planck equations .
fundamentals and applications , springer , ( 2005 ) .
r. krsten , u. behn , _ phys .
rev . e_*94 * , 062135 ( 2016 ) .
j. gmez - ordez , j. m. casado , m. morillo , c. honisch , and r. friedrich , _ europhys .
* 88 * 400006 ( 2009 ) .
m. morillo , j .
gmez - ordez , and j. m. casado,_phys .
e _ * 52*,316 ( 1995 ) . | we consider the behavior of a collective variable in a complex system formed by a finite number of interacting subunits .
each of them is characterized by a degree of freedom with an intrinsic nonlinear bistable stochastic dynamics .
the lack of ergodicity of the collective variable requires the consideration of a feedback mechanism of the collective behavior on the individual dynamics .
we explore numerically this issue within the context of two simple finite models with a feedback mechanism of the weiss mean - field type : a global coupling model and another one with nearest neighbors coupling . |
geometry and inhomogeneity influence pattern formation in chemical and biological systems .
@xcite one example where these two factors play a crucial role is in the experimental observations of distinctive spiral wave dynamics on the surfaces of spherical beads , which are excitable inhomogeneous chemical media @xcite .
a biological example is the origin of abnormal cardiac rhythms in the human heart which depend on the anatomical substrate .
the heart possesses a complex nonplanar geometry with multiple chambers , with holes corresponding to valves and blood vessels .
some serious arrhythmias are associated with circulating spiral waves similar to those observed in chemical media @xcite . since
an abnormal anatomical substrate is a common finding in patients with some types of cardiac arrhythmias , and interventions that modify the anatomy are an accepted form of therapy @xcite , theoretical analyses of the interplay between geometry of the substrate and dynamics may help in the therapy of cardiac arrhythmias .
in this paper we study spiral wave dynamics on ( punctured ) spheres with spatially inhomogeneous excitability .
we show for punctured spheres that the sum of indices which characterize each spiral has to be zero .
moreover , we demonstrate that topological constraints imposed by the spherical geometry and inhomogeneity in excitability lead to the formation of pairs of spirals , with distinctive transient dynamics or as stable attractors , in which one spiral acts as a source and a second as a sink leading to a source - sink pair under a broad range of conditions .
our results explain the experimental observations of spirals on spherical beads @xcite . while we do not consider detailed models of cardiac wave propagation
, our results may apply to some generic aspects of atrial arrhythmias because the thin walls of the atria can be described as two - dimensional ( 2d ) inhomogeneous excitable media with specific geometrical features .
the mathematical description of spiral waves is based on the notion of phase which in turn allows one to characterize spiral waves by an index . from this description ,
a number of topological results placing restrictions on spiral wave dynamics can be derived .
@xcite with the exception of a finite number of singular points , with each point in an orientable and compact two - dimensional differentiable manifold @xmath0 we identify a unique phase lying on the unit circle , @xmath1 . the resulting phase map or phase field
is assumed to be continuously differentiable , except at the singular points .
the manifold can be triangulated @xcite ( subdividing it into a set of polygons ) , where none of the edges or vertices of the polygons pass through a singularity .
the index , @xmath2 ( sometimes also called the topological charge or winding number ) of a curve @xmath3 bounding a polygon is found by computing the line integral @xmath4 where the polygon is always traversed in a clockwise orientation . by continuity of @xmath5 , @xmath2 must be an integer .
the index of a singular point is uniquely defined as the index of any curve @xmath3 provided that @xmath3 encircles the point but no other singular points .
the index of a curve that does not enclose any singular points is obviously zero .
if the manifold @xmath0 has no boundaries , each edge of the triangulation is an edge of two polygons .
since the line integral adds up the change in phase along the various edges of the polygon , the sum of the indices of the singular points for a phase field in @xmath0 is @xmath6 where the sum is over all the singular points .
this follows since the contribution of the change in phase of each edge to the total integral is counted twice , but since the edge is traversed in opposite directions each time , the net contribution of each edge is zero .
this index theorem is applicable to tori , and other surfaces of genus different from 0 .
however , unlike the more familiar poincar index theorem ( see ref .
@xcite p. 74 for vector fields )
the sum of the indices of the singular points does not depend on the genus of the surface .
this index theorem for manifolds without boundaries can be extended to manifolds with boundaries . in the following
, we will consider the case of structures that arise from puncturing orientable and compact two - dimensional differentiable manifolds .
the index of a hole can be uniquely defined as the index of a curve @xmath3 provided that @xmath3 encircles the hole but no other holes or singular points and @xmath3 is positively oriented with respect to the domain which contains the hole and is bounded by @xmath3 . applying this definition and taking the summation in eq .
( [ indextheorem ] ) over the singularities and the hole , or the holes if there is more than one hole , the index theorem can be proven by the same line of arguments as for the case of manifolds without boundaries .
this extension is important in the heart , for example , where the atrium is punctured by valves and veins . in such cases one
is led to consider manifolds with holes ; for example , a sphere with a hole .
a sphere with a hole is topologically equivalent to a disk , and , indeed , the results for disks and for spheres with holes are consistent : for the disk @xmath7 , bounded by a curve @xmath3 , @xmath8 so that the sum of the indices of the singular points in the disk is equal to the index of the curve @xmath3 bounding the disk .
if there is a single singular point on the disk , with an index of + 1 , the index of the curve bounding the disk will also be + 1 .
imagine now the boundary of the disk to be brought together ( like a draw - string bag ) so that the boundary of the disk now defines a hole in the sphere .
in this geometry the curve @xmath3 will be traversed in an opposite orientation ( the hole is now inside @xmath3 ) from the direction it was traversed when it was the boundary of the disk .
now if there is a singular point with an index of + 1 on the sphere , the index of the hole is -1 , so that the sum of the indices is again zero .
since it is necessary to conserve the sum of the indices , singularities of index @xmath9 usually arise and are destroyed in pairs of opposite sign @xcite .
an exception occurs when singularities are destroyed by collision into a boundary , so that the index of the singular point and the index of the boundary both change simultaneously .
another exception occurs if there are singularities with index different from one . in such cases interactions between different singularities
can lead to destruction or creation of odd numbers of singularities @xcite .
the fitzhugh - nagumo ( fhn ) equation , @xcite @xmath10 where @xmath11 and @xmath12 are two scalar fields , @xmath13 is the ratio of the time scales associated with the two fields , and @xmath14 and @xmath15 are the constant diffusion coefficients , is a prototypical model for excitable media .
we choose @xmath16 and @xmath17 .
the real parameters @xmath18 and @xmath19 characterize the local dynamics and , hence , the local excitability . whenever @xmath20 , @xmath21 and @xmath22 , the fhn system is excitable . at @xmath23
a hopf bifurcation occurs such that for @xmath24 the system exhibits oscillations . in the following we take @xmath25 , @xmath26 and @xmath27 .
we consider a spherical shell whose outer and inner radii are @xmath28 and @xmath29 , respectively , and focus on thin spherical shells where @xmath30 , @xmath31 .
the radii are large enough to avoid a curvature - dependent loss of excitability @xcite , and the shell is sufficiently thin that the dynamics is effectively 2d corresponding to the dynamics on a sphere .
@xcite the initial condition is a domain of excited " state , adjacent to a domain of the refractory " state .
both domains have the forms of slices of the same size oriented from the north to the south pole @xcite and yield a pair of counter - rotating spirals . in order to apply the topological results to the fhn equation ,
it is necessary to first define the phase .
we define a phase , @xmath32 , based on the equation @xmath33
if @xmath34 and @xmath35 . thus , singular points at given @xmath36 are points @xmath37 in the medium for which @xmath38 and @xmath39 . for each @xmath36
, we obtain a continuously differentiable phase map @xmath40 that associates to each point in a well - defined domain @xmath41 a phase lying on the unit circle , @xmath1 . for our fhn medium , the domain is the surface of a ( punctured ) sphere reduced by a finite number of points where the phase is singular at fixed @xmath36 . rotating spiral waves in the fhn equations
are obviously associated with a singular point which is called the spiral core . in
what follows , we assume that there are only single - armed spirals so that a clockwise rotating source has an index of + 1 , and a counterclockwise rotating source has index -1 .
a clockwise rotating sink has an index of -1 , and a counterclockwise rotating sink has index + 1 . from eq .
( [ indextheorem ] ) , it is impossible to have a single rotating spiral wave on a sphere ; in addition , there must be at least another singular point or a hole with nonzero index . for excitable media ,
a non - zero index of a hole implies that wave fronts travel permanently around the hole such that the numbers of fronts travelling clockwise and counterclockwise are different .
this includes the particularly simple case of a single wave front travelling around the hole which can be considered as a spiral wave associated with the hole .
first consider homogeneous fhn media with a constant @xmath42 .
the wave fronts emanating from the spiral cores with opposite index annihilate " along the equator such that each spiral determines the dynamics on half of the sphere ( see fig .
[ exc_hom ] , left panel ) similar to what has been observed in ref .
@xcite for a different excitable system .
( color online ) left : spiral waves of excitation ( light fronts ) on the sphere for constant @xmath43 emanating from spiral cores close to the poles on an equator projection .
the white arrows show the direction of propagation .
one annihilation front along the equator can be identified .
right : sketch of the constant gradient in the inhomogeneous case .
the dashed lines are the equi-@xmath19 lines and we choose @xmath44 and @xmath45 .
the angle @xmath46 describes the orientation with respect to the axis from pole to pole .
the results described in the text do not depend qualitatively on the choice of @xmath46 or @xmath47 and @xmath48 as long as they yield stable spirals . ]
we have shown that this behavior is robust with respect to disorder in excitability with small amplitude and correlation length .
if random , uncorrelated spatial variations in @xmath19 exist on length scales much smaller than the diameter of the spiral core meander @xcite , the dynamics is able to average over such small - scale inhomogeneities .
the robustness explains why such states have been experimentally observed in some chemical reactions on spherical beads which are intrinsically inhomogeneous @xcite . applying a constant gradient in @xmath19 as sketched in the right panel of fig .
[ exc_hom ] leads to a different scenario .
the time evolution of the spiral pair may be partitioned into four distinct regimes as shown in figs . [ beta ] and [ d ] .
( color online ) the local excitability @xmath49 at the spiral cores versus time .
gradient - induced motion of the two spiral cores leads to a change in the local excitability at the cores with time .
the spiral period in the final state is @xmath50 .
four different regimes can be identified ( see text ) .
inset : the final bound pair of counter - rotating spirals in regime iii for @xmath51 is shown on an equator projection such that the point of lowest excitability lies on the central longitude .
the spiral closer to the equator has index @xmath52 while the other one has index @xmath53 . ]
( color online ) the distance between the two spiral cores @xmath54 versus @xmath36 .
gradient - induced motion of the two spiral cores leads to a change in the distance @xmath55 between the cores with time .
the lower and upper curves correspond to the distance in @xmath56 and @xmath57 , respectively .
the dashed lines are the respective upper bounds given by the size of the sphere . ] because of the gradient , the frequencies of the two spirals differ since a higher value of @xmath19 corresponds to lower excitability , which generally implies a lower spiral frequency @xcite . during a short transient t ,
the spiral with the higher frequency assumes control of the dynamics @xcite on the sphere .
the location on the sphere , where wave fronts emanating from the two spiral cores annihilate , moves toward the core of the low - frequency spiral . finally , the low - frequency spiral core is pushed farther from the high - frequency spiral core @xcite(see fig .
after this short transient , the wave fronts travel from pole to pole leading to the creation of a source - sink pair .
this ( intermediate ) state is shown in fig .
[ anti_series ] and corresponds to regime i in figs .
[ beta ] and [ d ] . viewed from the low excitability end of the sphere , the waves wind into a small region about the core , reminiscent of the structure of antispirals , i.e. , inward moving spirals seen in oscillatory media @xcite
. however , the origin of this inward spiral motion in oscillatory media differs and is distinct from that observed here . in oscillatory media , either spirals or antispirals are stable depending on system parameters and the wave length diverges on the border in the parameter space between these two regimes .
thus , antispirals exist independently of spirals .
this is not the case here because the generation of an inward moving spiral relies on the presence of a spiral source and spherical geometry . for example , consider the fhn system with a disc geometry and a radial gradient in excitability such that the maximum value of @xmath19 is located in the center of the disc . in this case , a source - sink pair can not occur because the high - frequency spiral , acting as a source , would push the other ( low - frequency ) spiral out of the system , excluding the presence of any strong random inhomogeneities in excitability which may pin the low - frequency spiral and prevent its motion ( see , e.g. , @xcite ) .
the lower panel of fig .
[ anti_series ] shows that remnants of the low frequency spiral persist in a small area around the core .
it has been speculated that antispiral waves might occur in cardiac tissue @xcite .
while excitable media can not support a regime of exclusive antispirals due to their excitable character , our results show that source - sink pairs with similar characteristics could form in the heart where the underlying dynamics is excitable , the medium is inhomogeneous and the topology is similar to a ( punctured ) sphere .
spiral dynamics of the type described above has been observed by maselko and showalter @xcite in experiments on the excitable belouzov - zhabotinsky reaction on spherical beads .
they attributed the generation of spiral source - sink pairs to inhomogeneities in the medium related to differing chemical environments .
this is consistent with our findings for systems with gradients and is further confirmed by the work reported in ref .
while the generation of source - sink pairs due to a gradient in the fhn medium investigated here is only an intermediate state ( but one that persists for approximately 240 spiral periods in our simulations ) , random spatial variations of the excitability with a correlation length comparable to the diameter of the spiral core meander or larger can lead to a final state consisting of such a source - sink pair @xcite .
this is due to the fact that the source can be trapped in a region of depressed local excitability .
( color online ) waves of excitation on the sphere in regime i of the gradient - induced dynamics shown in figs . [ beta ] and [ d ] .
a source - sink pair has formed . for random spatial variations of excitability with a correlation length comparable to the diameter of the spiral core
meander , the final state is very similar to the one shown here @xcite .
upper panel from left to right : view centered at the north pole , south pole and the equator .
the source at the south pole has index @xmath52 and the sink at the north pole index @xmath53 .
the black circles show possible choices of @xmath3 .
the white arrow shows the direction of wave propagation .
lower panel : dynamics at the north pole .
time increases from left to right with @xmath58 between snapshots . ]
not only does the gradient in the fhn medium change the local excitability but it also induces a drift of the spiral cores @xcite . for our model , the drift is rather slow compared to the transition to the source - sink pair which takes place during the transient regime t. this can be seen in fig .
( the fluctuations in @xmath49 are due to spiral meandering . ) in regime i , the dominating spiral drifts toward lower excitability and its wave fronts continuously push the other core in the opposite direction , thus , keeping the distance @xmath55 between the cores approximately constant ( see fig .
the fluctuations in @xmath54 are again due to meandering of the spiral cores . because of this drift
, the local excitabilities at the two spiral cores approach each other until they become equal . at this point , regime ii in figs .
[ beta ] and [ d ] is entered .
the dynamics change drastically : the slaved spiral reverses its drift direction and regains control over its own dynamics .
both spirals drift toward lower excitability . due to the geometric constraints imposed by the spherical geometry , the spirals approach each other until they form a bound pair ( at @xmath59 ) .
they finally reach a stable state ( at @xmath60 ) corresponding to regime iii in figs .
[ beta ] and [ d ] .
neither the _ average _ distance between the spirals nor the _ average _ local excitability changes further .
yet , on top of the persisting unsynchronized meandering of the two spiral cores , the bound pair slowly moves along a ( closed ) equi-@xmath19 curve on the surface of the sphere .
the direction of the motion depends on the initial condition , i.e. , whether the spiral with positive index was initially closer to the point of lowest excitability or to the point of highest excitability than the spiral with negative index .
the wave dynamics generated by this bound pair is shown in the inset of fig .
[ beta ] .
while kinematic theory applies only to spirals with large cores , it is instructive to note that this theory predicts that the direction of the drift due to gradients depends on the model system and its parameters @xcite .
although our simulations have shown the same direction of the drift for a range of parameters in the meander region of the fhn phase diagram , it is conceivable that , under different circumstances favoring a drift toward higher excitability , one spiral could act as a permanent source and a source - sink pair could be the final attractor .
such a scenario would also be consistent with the experimental findings in ref .
@xcite .
next , we consider a homogeneously excitable sphere with a single hole .
two scenarios can be observed depending on the location of the hole with respect to the spiral pair .
if a spiral wave is not permanently attached to the hole , the dynamics is very similar to the case without any hole .
if one of the spirals is permanently attached to the hole , the frequency of this spiral is lowered .
the size of the hole determines the frequency of the spiral because the wave front has to travel around the hole .
the transient dynamics is similar to that in regime t for the case with a gradient ; however , no drift of the spiral cores is induced and the final state is a spiral source - sink pair as shown in fig .
[ series_hole ] . not only is the net index conserved during the transition to a spiral source - sink pair but so is the index of the individual spirals : during the transition an outgoing counterclockwise ( clockwise ) oriented spiral is converted into an ingoing clockwise ( counterclockwise ) oriented spiral .
thus , the formation of spiral source - sink pairs conforms with the topological constraints .
if a gradient as well as a hole is present , the spiral drift discussed earlier also determines the final state , which depends on the hole s location in the gradient field .
for instance , if the point of lowest excitability in the medium is on the hole boundary , simulations show that the spiral which is not attached to the hole will stabilize close to this point . in this
final state odd numbers of wavefronts are attached to the hole , again conserving the topological charge of the hole .
( color online ) waves of excitation on the punctured sphere propagating towards the hole .
a source - sink spiral wave pair has formed in the final state .
upper panel from left to right : view centered at the north pole , south pole and the equator .
lower panel : dynamics at the south pole .
time increases from left to right with @xmath58 between snapshots . ]
inhomogeneities due to spatially - varying excitability on ( punctured ) spherical shells lead to complex spiral wave dynamics and the formation of source - sink spiral pairs in excitable media .
the results presented here are immediately applicable to excitable media in more complicated geometries such as tori or multi - holed tori and to situations in which multi - armed spirals are found .
this includes mathematical modelling of cardiac tissue .
the approach taken in this paper stresses constraints and aspects that apply to , and must be observed in , all realistic models of the heart satisfying certain criteria of continuity .
there are also implications for the treatment of cardiac arrhythmias . in cardiology
it is sometimes possible to develop maps showing the timing of the excitation over limited regions of heart @xcite . in this case
, a sink might be confused for a source ( of the arrhythmia ) , and this might have implications for the diagnosis of the mechanism and the choice of therapy . the current work shows how partial knowledge about what is happening in some regions that could be observed , might be helpful in establishing properties of dynamics that could not be observed . while the types of sinks we have described here have only been observed in chemical media @xcite so far , we certainly expect their existence in the cardiological domain . | we analyze the way topological constraints and inhomogeneity in the excitability influence the dynamics of spiral waves on spheres and punctured spheres of excitable media .
we generalize the definition of an index such that it characterizes not only each spiral but also each hole in punctured , oriented , compact , two - dimensional differentiable manifolds and show that the sum of the indices is conserved and zero . we also show that heterogeneity and geometry are responsible for the formation of various spiral wave attractors , in particular , pairs of spirals in which one spiral acts as a source and a second as a sink the latter similar to an antispiral .
the results provide a basis for the analysis of the propagation of waves in heterogeneous excitable media in physical and biological systems . |
an interesting and difficult problem in graph theory is the graph embedding problem .
the problem is as follows : given a collection of vertices in @xmath1 and the pairwise distances between some of them , find the positions of the vertices up to some arbitrary rigid motion .
the problem has shown to be np - hard @xcite ; however , solutions may be found using semidefinite programming @xcite .
unfortunately the embeddings given by semidefinite programming may be in some larger dimensional space rather than @xmath0-dimensional .
however , if the framework is _ universally rigid _ then solutions are automatically constrained to be @xmath0-dimensional . this paper seeks to address the question raised by gortler and thurston : `` for a graph @xmath4 which is @xmath0-ggr is there always a generic framework in the space of @xmath0-dimensional frameworks , @xmath5 that is universally rigid?''@xcite .
this question is answered in the affirmative for @xmath6 and @xmath7 , and for an infinite number of @xmath0-ggr graphs when @xmath8 .
the key tool to obtain this result is the hennenberg operation .
the hennenberg operation is a vertex and edge addition to a graph , which preserves important rigidity properties .
connelly proved that hennenberg operations preserve global rigidity @xcite .
this paper expands that result by proving that hennenberg operations preserve universal rigidity for suitable frameworks .
hennenberg operations and edge additions can be used to build @xmath0-ggr graphs from the complete graph @xmath9 .
all graphs built in this manner will possess a generic @xmath0-dimensional universally rigid ( @xmath0-gur ) framework .
every non - trivial 1 and 2-ggr graph is the result of hennenberg operations and edge additions to @xmath9 , allowing us to conclude the main result .
all 1 and 2-ggr graphs and an infinite number of @xmath0-ggr graphs for @xmath8 have a @xmath0-gur framework .
the complete answer to gortler and thurston s question is still unknown in higher dimensions but we offer the following conjecture .
all d - ggr graphs have a d - gur framework .
thanks to kiril ratmanski and andrew fanoe for their helpful comments and ideas .
thanks to timothy sun for showing that 1-ggr graphs can be characterized by the whitney synthesis theorem .
special thanks to dylan thurston for his enormous help throughout the entire process .
* graphs and frameworks : * a _ graph _ @xmath4 is a set @xmath10 of @xmath11 vertices , and a set @xmath12 of @xmath13 edges containing some collection of 2 member subsets of @xmath10 .
a _ framework _
@xmath14 , is an embedding of the vertices and edges in @xmath1 @xmath15 where each @xmath16 is the coordinate of a vertex in @xmath1 .
let @xmath5 denote the set of all frameworks of @xmath4 in dimension @xmath0 .
a framework is _ generic _ if the coordinates of its vertices do not satisfy any nontrivial polynomial with rational coefficients .
* local rigidity : * two different embeddings @xmath14 , and @xmath17 of a graph @xmath4 are said to be _ equivalent _ written @xmath14@xmath18@xmath17 if @xmath19 for all @xmath20 edges in @xmath12 .
a framework is said to be _ locally flexible _ if there exists a non - constant analytic path @xmath21\to { \mathbb{r}}^{vd}$ ] such that @xmath22 and @xmath23 $ ] and @xmath24 can not be extended to a euclidean motion of @xmath1 . if a framework is not locally flexible then it is _ locally rigid_. * infinitesimal rigidity : * the _ half edge length squared function _
@xmath25 , where @xmath26 @xmath27 is a map from frameworks to the space of edge lengths . if two frameworks are equivalent @xmath28 then by definition @xmath29 .
the jacobian @xmath30 of the edge - length squared function is called the _ rigidity matrix _ , and gives a simple way to determine local rigidity for generic frameworks .
the matrix is a @xmath31 matrix where each row represents an edge and each column represents the one of the @xmath0 coordinates of a vertex .
the rigidity matrix has an important connection to local rigidity , as the first derivatives of paths preserving edge lengths are in the kernel of the matrix .
thus every euclidean motion is in the kernel , and as long as the framework does not lie in some lower dimensional affine subspace of @xmath1 , the dimension of the kernel is at least @xmath32 .
a framework is _ infinitesimally rigid _ if the kernel of the rigidity matrix is minimal , or equivalently @xmath33 .
if a framework is infinitesimally rigid then it is locally rigid .
if a generic framework of a graph with d+1 or more vertices is locally rigid in @xmath1 , then it is infinitesimally rigid .
furthermore if there exists a single infinitesimally rigid framework , then every generic framework is infinitesimally rigid .
* stresses : * an _ equilibrium stress _ on a framework is an assignment of a real number @xmath34 to each edge of a framework satisfying @xmath35 .
the _ stress vector _ @xmath36 is a @xmath37 vector consisting of the ordered stresses on each edge .
it is not difficult to see that @xmath36 is in the kernel of the transpose of the rigidity matrix , and in fact the space of all stresses is precisely this kernel .
equilibrium stress matrix _ @xmath38 is a @xmath39 matrix satisfying :
\(1 ) @xmath40 \(2 ) @xmath41 if @xmath42 and @xmath43 \(3 ) @xmath44 if @xmath42 and @xmath45 \(4 ) @xmath46 is chosen so the row sum is zero .
the entries of the stress matrix are uniquely determined by @xmath36 so this paper will freely interchange between equilibrium stresses and stress matrices . since the row sum of the matrix is zero the vector of all ones @xmath47 is always in the kernel of the stress matrix .
additionally the @xmath0 coordinate projections of the vertices lie in the kernel of the stress matrix .
this condition is easy to check since @xmath35 must be satisfied . a generic framework with @xmath48 or more vertices can not lie in a lower dimensional affine subspace , so the @xmath0 coordinate projections are linearly independent .
thus for a generic framework with @xmath48 or more vertices , the nullity of the stress matrix is at least @xmath48 .
* redundant rigidity : * an edge in a locally rigid framework is _ redundant _ if removal of said edge preserves local rigidity .
a framework is _ redundantly rigid _ if every edge of the framework is redundant .
a framework is redundantly rigid if and only if there exists a stress which is nonzero on every edge .
* global rigidity : * two frameworks @xmath14 and @xmath17 are _ congruent _ written @xmath49 if @xmath19 for all @xmath50 . a @xmath0-dimensional framework @xmath14 is _ globally rigid in dimension _ @xmath0 if every framework in @xmath5 equivalent to @xmath14 is also congruent to @xmath14 . a graph is generically globally rigid in dimension @xmath0 ( @xmath0-ggr ) if and only if it has a generic framework with a stress matrix of minimal nullity d+1 , or it is the complete graph on d+1 or fewer vertices .
connelly showed that the stress condition was sufficient , and gortler , healy , and thurston showed that it was necessary , implying that global rigidity is a generic property . since global rigidity is a generic property , it can be thought of as a property of a graph , and such graphs are called _ generically globally rigid in dimension _
@xmath0 ( @xmath0-ggr ) .
the stress condition is not particularly intuitive , and other attempts have been made to characterize global rigidity combinatorially .
* vertex connectivity : * a graph is @xmath51__-vertex - connected _ _ if the deletion of any @xmath52 vertices leaves a connected graph .
if a graph is @xmath0-ggr then it is redundantly rigid and d+1-vertex - connected . in the same paper hendrickson
conjectured that these conditions would also be sufficient . in @xmath6
2-vertex - connectedness subsumes redundant rigidity , and it is easy to see that all 2-vertex - connected graphs are 1-ggr . in @xmath7 , connelly and
jackson and jordn proved that these conditions are sufficient for global rigidity ( see theorem 5.7 ) @xcite , @xcite .
unfortunately , in @xmath53 and above connelly showed that there are graphs which satisfy hendrickson s conditions , but fail to be globally rigid .
in fact there is not even a combinatorial characterization of local rigidity in dimensions 3 or higher @xcite .
the lack of a general combinatorial characterization of @xmath0-ggr graphs for @xmath8 makes conjecture 1.2 difficult to resolve .
* universal rigidity : * a @xmath0-dimensional framework , @xmath14 is _ universally rigid _ if every framework in @xmath54 equivalent to @xmath14 is also congruent to @xmath14 where @xmath55 . if the framework is also generic then it is a _
@xmath0-dimensional generic universally rigid framework _
( @xmath0-gur framework ) . unlike global rigidity
, universal rigidity is not a generic property .
a graph @xmath4 is _ @xmath0-semi - universally rigid _ ( @xmath0-sur ) if it has a @xmath0-gur framework , but has other generic frameworks which fail to be universally rigid in @xmath0-dimensions .
figure 1 has a very simple example of a graph with this behavior . in section 6
we construct a non - trivial family of @xmath0-sur graphs .
certain classes of graphs , like complete graphs , are generically universally rigid .
any framework of a complete graph is universally rigid . for frameworks of complete graphs equivalence and congruence
are the same condition .
if a framework is @xmath0-gur then it is a framework of a @xmath0-ggr graph .
if a framework is universally rigid then it must be globally rigid in dimension @xmath0 .
since the framework is generic all other generic @xmath0-dimensional frameworks are globally rigid so the graph is @xmath0-ggr .
* hennenberg operations : * a @xmath0-hennenberg operation is an edge and vertex addition to a graph that preserves important rigidity properties .
a @xmath0-hennenberg operation is performed on a graph by deleting an edge @xmath56 , adding a new vertex @xmath57 and edges @xmath58 and @xmath59 , and adding an additional @xmath60 edges to @xmath57 which are connected to any @xmath60 distinct vertices that are not @xmath61 or @xmath62 .
it is important to note that @xmath0-hennenberg operations may only be performed on graphs with at least @xmath48 vertices , since the operation requires @xmath48 distinct vertices .
connelly showed that @xmath0-hennenberg operations on a @xmath0-ggr graph resulted in another @xmath0-ggr graph @xcite .
this paper extends that result to show that a @xmath0-hennenberg operation takes a graph with a @xmath0-gur framework to another graph with a @xmath0-gur framework .
if a generic d - dimensional framework with d+2 or more vertices is universally rigid then it has a positive semi - definite ( psd ) stress matrix of nullity d+1 .
alfakih proved the theorem for a matrix of any rank @xcite , @xcite .
gortler and thurston strengthened the theorem by showing the stress matrix has minimal nullity .
* conics at infinity : * let @xmath63 be a symmetric @xmath64 non - zero matrix .
a conic at infinity @xmath65 is the set of points @xmath66 : @xmath67 .
the edge directions of a framework @xmath14 lie on a conic at infinity if there exists a non - zero symmetric matrix @xmath63 satisfying @xmath68 for all @xmath45 .
a d - dimensional framework with at least d+2 vertices is universally rigid if it has a psd stress matrix of nullity d+1 and its edge directions do not lie on a conic at infinity .
connelly proved this result across a number of papers @xcite and @xcite , however due to the centrality of the result to this paper a condensed and slightly modified proof is presented below .
let @xmath69 be the energy function @xmath70 .
@xmath71 where @xmath72 is the stress given by the fixed stress vector @xmath73 for the framework @xmath74 .
let @xmath75 be the @xmath76 vector of the @xmath77 coordinate entry of each of the @xmath11 vertices of @xmath14 , then an equivalent version of @xmath69 is given by @xmath78 where again @xmath79 is the fixed stress matrix determined by @xmath73 .
now let @xmath14 be a framework in dimension @xmath0 with a psd stress matrix @xmath80 , and whose edges do not lie on a conic at infinity .
let @xmath17 be some framework in dimension @xmath55 , with @xmath14@xmath18@xmath17 .
now consider @xmath81 , where @xmath82 is the stress vector determined by @xmath80 . since the coordinate projections of @xmath14 are in the kernel of @xmath80 , by equation ( 2 ) @xmath83 . by equation ( 1 )
@xmath81 is a function on the edge lengths .
thus if @xmath14@xmath18@xmath17 then @xmath84 .
@xmath80 is a psd matrix , so @xmath85 such that @xmath86 .
thus , @xmath87 @xmath88 the kernel of @xmath80 consists of the coordinate projections @xmath75 and the vector of all ones @xmath47 , thus each coordinant projection @xmath89 must be some linear combination of these vectors . therefore @xmath90 is obtained from @xmath91 by some affine transformation of @xmath1 to @xmath92 . in other words @xmath93 for some fixed @xmath94 matrix @xmath95 and a fixed @xmath96 vector @xmath97 for all verticies @xmath98 .
the transformation must preserve edge lengths so @xmath99 @xmath100 @xmath101 is clearly a symmetric matrix ; thus either it is the zero matrix or the edge directions lie on a conic at infinity . since the edge directions do not lie on a conic at infinity by hypothesis , @xmath102
let @xmath103,@xmath104 be vertices not necessarily sharing an edge .
then @xmath105 thus every framework @xmath17 in dimension @xmath2 equivalent to @xmath14 is also congruent to @xmath14 as desired .
let g(p ) be a generic d - dimensional framework of a finite graph g where each vertex has degree at least d. then the edge directions of g(p ) do not lie on a conic at infinity .
a generic d - dimensional framework with d+2 or more vertices is universally rigid if and only if it has a psd stress matrix of nullity d+1 .
theorem 3.1 already establishes one direction . to prove the other direction note the nullity of the stress matrix is @xmath48 , therefore the framework is globally rigid and thus @xmath106-connected .
since the graph is @xmath106-connected the degree of each vertex must be at least @xmath0 . by proposition 3.3
, the edge directions of the framework do not lie on a conic at infinity .
the framework then satisfies the hypothesis of theorem 3.2 .
in this section we derive a few simple results about stress matrices that we will need for our main argument .
the first four results deal with perturbations of eigenvalues and the last two results relate to kernels .
* normal matrices : * a real valued @xmath107 matrix @xmath95 is _ normal _ if @xmath108 .
it is easy to see that any symmetric matrix @xmath109 is normal since @xmath110 .
let a , f be @xmath107 matrices , with a , a+f normal .
let @xmath111 be the eigenvalues of a in some order and @xmath112 be the eigenvalues of a+f in some order .
then there exists a permutation @xmath113 of the integers @xmath114 such that @xmath115^{1/2 } \ ! \ ! \ ! \leq \lvert f\rvert_{2}\ ] ] let a and c be @xmath107 symmetric matrices .
let b be the @xmath107 matrix such that @xmath116 .
let @xmath117 be the smallest absolute value of an eigenvalue of a. if @xmath118 then c has at least as many positive and negative eigenvalues as a. @xmath95 and @xmath119 are symmetric , so they are normal and all their eigenvalues are real .
let @xmath111 be the eigenvalues of @xmath95 , and @xmath112 be the eigenvalues of @xmath119 . by the hoffman and wielandt theorem , @xmath120 @xmath121 thus @xmath122 . since @xmath123 if @xmath124 then @xmath125 and if @xmath126 then @xmath127 .
thus @xmath119 has at least as many positive and negative eigenvalues as @xmath95 .
let a be a @xmath39 stress matrix of nullity @xmath48 for some d - dimensional framework g(p ) .
let c be a @xmath39 stress matrix for some d - dimensional framework h(q ) , not necessarily different from g(p ) .
let b be the matrix such that @xmath116 .
let @xmath117 be the smallest absolute value of an eigenvalue of a. if @xmath118 then a and c have the same number of positive and negative eigenvalues .
@xmath95 and @xmath119 satisfy corollary 4.1 so @xmath119 has at least as many positive and negative eigenvalues as @xmath95 .
thus @xmath119 has nullity at most @xmath48 .
but @xmath48 is the minimal nullity for stress matrices so the rest of the of the eigenvalues of @xmath119 must be zero .
therefore @xmath95 and @xmath119 have the same number of positive and negative eigenvalues .
let a , b be @xmath39 stress matrices for some d - dimensional framework g(p ) , where a is of nullity d+1
. then @xmath128 such that @xmath95 and @xmath129 have the same number of positive and negative eigenvalues , and @xmath129 is a stress matrix for g(p ) .
it is easy to check that if @xmath95 and @xmath130 are stress matrices for a framework then @xmath131 is also a stress matrix for the same framework .
let @xmath132 , and let @xmath133 be the smallest absolute value of an eigenvalue of @xmath95 .
pick @xmath134 and apply corollary 4.2 .
now we turn to analyzing kernels of matrices .
first we define a way to measure distances between linear subspaces .
* hausdorff distance : * let ( @xmath135 , @xmath136 ) be a metric space and let @xmath95 , @xmath130 be subsets of @xmath135 .
the hausdorff distance between @xmath95 and @xmath130 , written @xmath137 is defined as @xmath138 the distance between linear subspaces @xmath139 and @xmath140 of @xmath1 is defined as the hausdorff distance between @xmath141 and @xmath142 , where @xmath143 is the unit sphere in @xmath144 .
@xmath143 is a bounded set so the distance will be well defined .
suppose that @xmath145 is a matrix whose entries are integral polynomial functions of the real variables @xmath146 .
let @xmath147 be the maximum rank of @xmath148 and suppose that @xmath149
. then @xmath150 such that for @xmath151 the hausdorff distance between the linear subspaces @xmath152 and @xmath153 is less than @xmath154 .
if a , b are psd matrices then @xmath155 .
let @xmath156 , @xmath157 be matrices such that @xmath158 and @xmath159 .
let @xmath160 .
@xmath161 therefore , @xmath162 and @xmath163 , which holds if and only if @xmath164 and @xmath165 .
thus @xmath160 if and only if @xmath166 and @xmath167 .
it follows that @xmath168 .
let h be obtained from g by a single edge addition .
if g has a d - gur framework then h has a d - gur framework .
let @xmath14 be the @xmath0-gur framework of @xmath4
. by theorem 3.1 @xmath14 has a psd stress matrix @xmath95 of nullity @xmath48 .
let @xmath169 be a @xmath0-dimensional framework of @xmath170 with the same vertex positions as @xmath14 .
@xmath14 was generic so @xmath169 is also generic .
define the stress on the new added edge to be @xmath171 then @xmath95 is a stress matrix for @xmath169 as well .
thus @xmath169 is a generic @xmath0-dimensional framework with a psd stress matrix of nullity @xmath48 .
therefore @xmath170 has a @xmath0-gur framework .
this proof is so trivial since adding an edge can not do anything but make a rigid framework more rigid .
let g be a graph with @xmath172 or more vertices .
let h be obtained from g by a single @xmath0-hennenberg operation .
if g has a d - gur framework then h has a d - gur framework .
let @xmath14 be the @xmath0-gur framework of @xmath4
. by theorem 3.1 @xmath14 has a stress matrix @xmath95 which is psd and of nullity @xmath48 .
@xmath14 must also be globally rigid , so by theorem 2.4 it is redundantly rigid .
@xmath14 is redundantly rigid , therefore it has a stress matrix @xmath130 for which the stress on every edge is non - zero .
pick @xmath173 so that every non - zero entry of @xmath95 is still non - zero in @xmath174 . by corollary 4.3 ,
pick @xmath175 so that @xmath176 is still psd and nullity @xmath48 .
pick @xmath177 and let @xmath178 .
@xmath80 is now an equilibrium stress matrix for @xmath14 which is psd of nullity @xmath48 and defines a non - zero stress @xmath179 on every edge .
arrange @xmath80 so that the two vertices whose edge was eliminated by the @xmath0-hennenberg operation , say @xmath61 and @xmath62 , correspond to the last two columns and rows of the matrix respectively .
let @xmath180 be a @xmath0-dimensional framework of @xmath170 , where @xmath181 for all @xmath182 and the extra vertex , @xmath183 , of @xmath170 is placed somewhere along the line going through the eliminated edge @xmath56 .
note that the configuration of @xmath180 is almost the same as the configuration of @xmath14 , and that @xmath180 is not a generic framework ( see figure 2 ) .
let @xmath184 be the directed length of the old edge , let @xmath185 be the directed length of edge @xmath58 , and let @xmath186 be the directed length of edge @xmath187 where we require @xmath188 .
note choosing a value for @xmath189 or @xmath190 determines the placement of the final vertex @xmath57 .
we define a stress vector @xmath191 on @xmath180 as follows : \(1 ) @xmath192 if @xmath20 is an edge of @xmath4 and @xmath170 , \(2 ) @xmath193 , \(3 ) @xmath194 , and \(4 ) @xmath195 if @xmath196 .
these conditions constitute an equilibrium stress on @xmath180 .
let @xmath197 be the @xmath198 stress matrix defined by this stress .
arrange @xmath197 so that the first @xmath11 rows and columns correspond to the same vertices as in @xmath80 , and the last row and column correspond to the new vertex @xmath57 .
let @xmath199 be a @xmath200 matrix , where the top left @xmath39 block has the same entries as @xmath80 and the rest of the entries are zero .
@xmath80 is psd of nullity @xmath48 , so @xmath199 is psd of nullity @xmath172 .
the last column and row of @xmath199 , which contains all zeros , corresponds to the vertex @xmath57 .
since the stress on each edge except for @xmath56 , @xmath58 , and @xmath187 is the same , @xmath199 and @xmath197 have the same entries except for a @xmath201 block in the bottom right corner .
let @xmath156 be the @xmath200 matrix such that @xmath202 .
let @xmath203 be the non - zero @xmath201 block of @xmath156 .
then @xmath204 since @xmath188 it follows that @xmath205 . if @xmath206 , pick @xmath207 .
then the diagonal entries of @xmath156 are positive , so @xmath156 is psd . if @xmath208 , pick @xmath209 .
then the diagonal entries of @xmath156 are positive , so @xmath156 is psd .
the sum of two psd matrices is psd so @xmath197 is psd . using lemma 4.5 , @xmath210 . because @xmath211 , the standard basis vector @xmath212 is not in the kernel of @xmath156 but is clearly in the kernel of @xmath199 .
thus , the dimension of the kernel of @xmath197 is strictly less than the dimension of the kernel of @xmath199 , so @xmath213 .
therefore , @xmath197 is psd and of nullity @xmath48 .
all that is left is to perturb @xmath180 to a generic configuration . to be able to do this @xmath180
must be infinitesimally rigid , and the following proposition proves that it is .
suppose g(p ) is a framework in @xmath1 and h is obtained from g by a hennenberg operation .
if h(q ) is a framework such that the @xmath60 additional edges and the subdivided edge of g do not lie in a @xmath60 affine subspace , then h(q ) is infinitesimally rigid . in order for the @xmath60 additional edges and the subdivided edge to lie in a @xmath214-dimensional affine subspace @xmath61 , @xmath62 , @xmath57 and the @xmath60 vertices connected to @xmath57 must all lie in a @xmath214-dimensional affine subspace
this means that for @xmath14 the vertices @xmath61 , @xmath62 and @xmath60 other vertices all lie in a @xmath214-dimensional affine subspace .
however such a configuration is not generic , contradicting our hypothesis on @xmath14 , so @xmath180 is infinitesimally rigid . by theorem 2.1 , we now know that every generic framework of @xmath170 in dimension @xmath0 will be infinitesimally rigid , so the kernel of the transpose rigidity matrix for all of these frameworks will be minimal .
non - generic points are measure zero in @xmath215 so @xmath216 there exists a generic framework , @xmath217 such that @xmath218 .
the entries of the rigidity matrix are polynomial functions of the vertex positions , so by lemma 4.4 @xmath219 @xmath220 such that the distance between @xmath221 and @xmath222 is less than @xmath154 .
let @xmath223 , then @xmath224 .
therefore @xmath225 such that @xmath226 .
define a stress @xmath227 on @xmath217 such that @xmath228 and let @xmath229 be the stress matrix defined by this stress vector .
the stress on each edge of @xmath217 differs from the stress on @xmath180 by at most @xmath230 .
let @xmath231 be the matrix satisfying @xmath232 .
the norm of the matrix @xmath233 for some constant @xmath234 .
finally , pick @xmath235 so that @xmath154 is so small that f satisfies the hypothesis of corollary 4.2 .
then @xmath229 is a psd stress matrix of nullity @xmath48 . by corollary 3.4
, @xmath217 is a @xmath0-gur framework .
this completes the proof of the theorem .
let g be a graph with @xmath172 or more vertices .
let h be obtained from g by any sequence of edge additions and @xmath0-hennenberg operations .
if g has a @xmath0-gur framework then h does as well .
for any d an infinite number of @xmath0-ggr graphs have a @xmath0-gur framework .
any graph built from @xmath0-hennenberg operations and edge additions to the complete graph @xmath9 will have a @xmath0-gur framework by corollary 5.4 and proposition 2.5 .
* path , path addition , and whitney synthesis : * let @xmath4 be a graph .
a _ path _ in @xmath4 is a sequence of vertices of @xmath4 such that each vertex in the sequence is connected to the next vertex in the sequence by an edge .
path addition _ to a graph @xmath4 is the addition to @xmath4 of a path between two existing vertices of @xmath4 , such that the edges and internal vertices of the path are not in @xmath4 . a _
whitney synthesis _ of a graph @xmath170 from a graph @xmath4 is a sequence of graphs @xmath236 where @xmath237 and @xmath238 , and @xmath239 is the result of a path addition to @xmath240 .
a graph g is 2-vertex connected if and only if g is a cycle or a whitney synthesis from a cycle .
every 1-ggr graph with 3 or more vertices can be built from a sequence of hennenberg operations and edge additions to @xmath241 . by theorem 2.4
every @xmath0-ggr graph with @xmath172 or more vertices must be @xmath48-vertex connected and redundantly rigid .
for a 1-dimensional framework , redundant rigidity is equivalent to the graph being 2-edge - connected .
2-edge - connectedness is a weaker condition than 2-vertex - connectedness , so only 2-vertex - connectivity needs to be considered . by the whitney synthesis theorem
every 2-vertex - connected graph is either a cycle or a path addition to a cycle .
a 1-hennenberg operation on a graph subdivides an edge by adding a vertex .
a cycle with @xmath11 vertices can easily be generated from @xmath241 by performing @xmath242 hennenberg operations on the graph .
a path addition is simply an edge addition to a graph , followed by any number of hennenberg operations on the added edge .
any sequence of path additions can be formed by successive edge additions and hennenberg operations .
if a graph @xmath4 with at least 4 vertices is redundantly rigid in @xmath243 and 3-vertex connected then @xmath4 can be obtained from @xmath244 by a sequence of hennenberg operations and edge additions .
redundant rigidity and 3-vertex connectivity are necessary and sufficient conditions for global rigidity in 2-dimensions , thus every 2-ggr graph on 4 or more vertices can be built from hennenburg operations and edge additions on @xmath244 every ggr graph in 1 and 2-dimensions has a universally rigid generic framework .
suppose @xmath4 is a 2-ggr graph ( 1-ggr graph ) .
if @xmath4 has 4 ( 3 ) or more vertices then @xmath4 satisfies theorem 5.7 ( proposition 5.6 ) .
thus corollary 5.4 applies to @xmath4 , so @xmath4 has a 2-gur framework ( 1-gur framework ) . if @xmath4 has 3 ( 2 ) or fewer vertices than @xmath4 is a complete graph , and therefore universally rigid
this completes the proof of theorem 1.1 .
we now turn to the construction of an infinite family of @xmath0-sur graphs .
recall that given some framework @xmath14 the space of stresses on @xmath14 is the kernel of the transpose rigidity matrix @xmath245 .
let @xmath236 be a sequence of graphs where @xmath246 is the result of a @xmath0-hennenberg operation on @xmath247 and @xmath248 has a @xmath0-gur framework . if @xmath249 and @xmath250 are generic frameworks then @xmath251 . by theorem 5.2 and proposition 2.6 ,
each @xmath246 is @xmath0-ggr .
therefore the generic frameworks of each @xmath246 are infinitesimally rigid so the rank of the rigidity matrix @xmath30 is maximal .
let @xmath252 and @xmath253 denote the number of vertices and edges of @xmath246 respectively .
for a generic framework @xmath250 @xmath254 . let @xmath255 where @xmath250 is generic . then @xmath256 .
now we proceed by induction .
the base case is trivial as @xmath257 .
now suppose that @xmath258 .
then @xmath259 thus @xmath260 .
let @xmath236 be a sequence of graphs where @xmath246 is the result of a @xmath0-hennenberg operation on @xmath247 and @xmath261 the complete graph on @xmath172 vertices .
if @xmath250 is generic then @xmath262 .
in particular @xmath250 has a one dimensional space of stresses .
@xmath261 the complete graph on @xmath172 vertices thus @xmath263 .
let @xmath249 be a generic framework .
@xmath248 is @xmath0-ggr so @xmath249 is infinitesimally rigid
. therefore @xmath264 by proposition 6.1 @xmath262 each @xmath246 has a @xmath0-gur framework .
thus we only need to show that for @xmath266 the graph @xmath246 has a generic @xmath0-dimensional framework which is not universally rigid . for
some @xmath267 let @xmath268 be a @xmath0-gur framework of @xmath269 . by corollary
@xmath268 has a @xmath270 stress matrix @xmath271 which is unique up to scaling .
@xmath268 is @xmath0-gur so @xmath271 is psd of nullity @xmath48 .
@xmath268 is also redundantly rigid so @xmath271 must also define a stress that is non - zero on every edge .
let @xmath272 and set @xmath273 .
we now proceed as in theorem 5.2 and we obtain @xmath274 stress matrices @xmath275 and @xmath276 .
the upper left @xmath277 block of @xmath275 has the same entries as @xmath271 and the entries of @xmath278 row and column of @xmath275 are all zero .
@xmath276 and @xmath275 have the same entries except for a @xmath201 block in the bottom right corner .
@xmath156 is the @xmath274 matrix such that @xmath279 .
let @xmath203 be the non - zero @xmath201 block of @xmath156 . recall that @xmath204 since @xmath188 it is not difficult to see that @xmath205 . if @xmath208 pick @xmath207 .
then the diagonal entries of @xmath156 are negative so @xmath156 is nsd . if @xmath206 pick @xmath280 @xmath281 .
then the diagonal entries of @xmath156 are negative so @xmath156 is nsd .
let @xmath282 be the last standard basis vector .
@xmath283 thus @xmath276 can not be psd .
however , @xmath284 is a psd @xmath274 matrix of nullity @xmath172 so @xmath285 thus it follows that @xmath286 .
the standard basis vector @xmath287 but @xmath288 so the two kernels are not identical .
the dimension of the kernel of @xmath156 is greater than or equal to the dimension of the kernel of @xmath284 .
therefore @xmath289 .
so @xmath290 such that @xmath291 . @xmath292
thus @xmath276 can not be nsd .
therefore @xmath276 is indefinite .
again we proceed as in theorem 5.2 and we obtain a generic framework @xmath235 such that @xmath293 and @xmath294 .
we then pick @xmath154 small enough that f satisfies the hypothesis of corollary 4.1 .
then @xmath295 has at least as many positive and negative eigenvalues as @xmath276 .
since @xmath276 is indefinite @xmath295 must also be indefinite . by corollary 6.2
@xmath295 is a unique stress matrix for @xmath296 up to scale .
therefore the stress matrices of @xmath296 are either indefinite or are the zero matrix .
the zero matrix has nullity greater than @xmath48 so @xmath296 has no psd stress matrix of nullity @xmath48 . finally by corollary 3.4 @xmath296
can not be universally rigid .
so @xmath297 is a @xmath0-sur graph . | a @xmath0-dimensional framework is an embedding of the vertices and edges of a graph in @xmath1 .
a @xmath0-dimensional framework is globally rigid if every other @xmath0-dimensional framework with the same edge lengths has the same pairwise distances between the vertices .
a graph is generically globally rigid in dimension @xmath0 ( @xmath0-ggr ) if every generic framework is globally rigid .
the @xmath0-dimensional framework of a @xmath0-ggr graph is universally rigid if for @xmath2 every @xmath3-dimensional framework with the same edge lengths has the same pairwise distances between the vertices .
we establish a strong connection between global and universal rigidity by showing that all 1 and 2-ggr graphs and an infinite number of higher dimensional @xmath0-ggr graphs have a generic universally rigid framework .
= 1 |
our experiment relies on the coherent manipulation of atomic quantum states using light pulses @xcite .
we use telecom based laser sources that provide high frequency stability and power in a compact and integrated setup @xcite . starting from a @xmath2 vapor , we load in @xmath5 a cloud of about @xmath6 atoms laser cooled down to 10 @xmath7k , and select the atoms in a magnetic field insensitive ( @xmath8 ) zeeman sublevel .
we then apply a velocity selective raman light pulse @xcite carrying two counterpropagating laser fields so as to keep @xmath9 atoms which enter the ai with a longitudinal velocity distribution corresponding to a temperature of 300 nk .
the raman laser beams are aligned along the plane wings direction ( y axis , figure 1 ) and are retroreflected by a mirror attached to the aircraft structure and following its motion .
the ai consists of a sequence of three successive raman light pulses to split , redirect and recombine the atomic wavepackets ( see figure 1d ) .
the acceleration measurement process can be pictured as marking successive positions of the free - falling atoms with the pair of raman lasers , and the resulting atomic phase shift @xmath10 is the difference between the phase of the two raman lasers at the atom s successive classical positions , with respect to the retroreflecting mirror @xcite .
as the raman beam phase simply relates to the distance between the atoms and the reference retroreflecting mirror , the ai provides a measurement of the relative mean acceleration @xmath11 of the mirror during the interferometer duration , along the raman beam axis .
the information at the output of the ai is a two - wave interference sinusoidal signal @xmath12 , where @xmath13 is the transition probability between the two @xmath2 ground states , and @xmath14 ( resp .
@xmath15 ) is the offset ( resp .
the amplitude ) of the interference fringes .
this signal is modulated by the atomic phase shift @xmath16 , where @xmath17 nm is the raman lasers effective wavevector and @xmath18 is the time between the light pulses ( see the first `` methods '' subsection for the calculation of the phase shift ) . in the aircraft ,
the acceleration along y ( figure 2a ) fluctuates over time by @xmath19 ( 1 standard deviation ) , and is at least three orders of magnitude greater than the typical signal variations recorded by laboratory - based matter wave inertial sensors .
for this reason , the signal recorded by the ai first appears as random , as shown in figure 2b . to quantify the information contained in the atomic measurements , we use mechanical accelerometers ( mas ) fixed on the retroreflecting mirror and
search for the correlation between the mas and the ai @xcite .
we use the signal @xmath20 continuously recorded by the mas to estimate the mean acceleration @xmath21 which is expected to be measured by the ai at time @xmath22 , with @xmath23 being the experimental cycle time ( see the first `` methods '' subsection ) . plotting the atomic measurements
@xmath24 versus @xmath21 reveals clear sinusoidal correlations between the mechanical sensors and the ai , both at 1-g ( figure 2c ) and in 0-g ( figure 2d ) .
this demonstrates that the ai truly holds information on the mirror acceleration @xmath11 .
we note that this result stands for the first demonstration of the operation of an atom accelerometer in an aircraft and in microgravity .
we now consider the application of our matter wave sensor to precise measurements of the plane acceleration , by operating the ai beyond its linear range .
for that purpose , we determine the ai acceleration response , defined by @xmath25 , independently from the mechanical devices .
we have developped a method ( see the `` methods '' section ) to estimate this response ( i.e. the parameters @xmath14 and @xmath15 ) , and the signal to noise ratio ( snr ) of the interferometer , which determines the acceleration noise of the sensor , @xmath26 .
the knowledge of @xmath14 and @xmath15 enables us to extract the acceleration @xmath27 from the atomic data @xmath24 by inverting the model @xmath28 . in this way
, the acceleration is known within the region where the interferometer model can be inverted unambiguously and corresponding to an acceleration interval of range @xmath29 . to obtain the total acceleration
, we need the information on the reciprocity region @xmath30 $ ] where the ai operates at measurement time @xmath31 , with @xmath32 being the interference fringe number where the measurement point is located ( see the `` methods '' section ) . to determine @xmath32
, we use the mas which have a reciprocal response over a wide acceleration range .
thus , our instrument consists in a hybrid sensor which is able to measure large accelerations thanks to the mechanical devices , and able to reach a high resolution thanks to the atom accelerometer .
figures 3a to 3c illustrate the measurement process that we use to measure the plane acceleration during successive 1-g and 0-g phases of the flight ( figure 3d ) , with high resolution . due to their limited performances , mainly their nonlinear response and intrinsic noise ( see the `` methods '' section ) ,
the mas provide a signal which is not perfectly proportional to the acceleration @xmath11 .
this leads to errors on the estimated acceleration @xmath33 which blur the mas - ai correlation function and might prevent from finding the fringe index @xmath32 where the ai operates .
these errors increase with the ai sensitivity and with the acceleration signal @xmath11 . for a given acceleration level ,
the good measurement strategy consists in increasing @xmath18 up to @xmath34 where the mas can still resolve the correlation fringes .
this will set the scale factor @xmath35 of the ai .
the sensitivity of the accelerometer is then determined by the snr of the matter wave sensor , which is estimated independently from the mas with our method .
for instance at 1-g , where @xmath36 , the mas enables us to increase the ai interrogation time up to @xmath37 and to resolve the correlation fringes .
future improvements will rely on the use of well characterized mas , in particular on a calibration of their scale factor at the @xmath38 level over the sensitivity bandwidth of the ai , to achieve interrogation times @xmath39 .
high frequency ( @xmath40 ) vibration damping could also be used to constrain the frequency range where the mas are needed , and therefore push forward a particular sensor technology .
second , parasitic inertial effects due to the rotation of the plane might be experienced by the matter wave sensor and not by the mas .
at 1-g , the atoms fall down before interacting with the raman beams ( aligned along the y axis ) , so that the interferometer has a physical area and is thus sensitive to the sagnac effect . the resulting coriolis acceleration measured by the ai and not by the mas
might impair the correlation and limit the performance of the hybrid sensor . for shot to shot fluctuations of the plane rotation of the order of @xmath41 , we estimate a limit to the sensitivity at the @xmath42 level at 1-g . this error source may be significantly reduced in the future with the use of additional sensors to measure the rotation of the plane and to take it into account in the calculation of the estimated acceleration . finally , the atomic snr limits the sensitivity of the inertial sensor . during the flight
, we measure at 1-g a snr of 3.1 for @xmath43 , which is in agreement with the value measured in our laboratory for the same interrogation time . with our experimental setup , the signal ( @xmath44 ) is essentially limited by the imperfections of the atomic beam splitters and mirror due to the temperature of the cloud and to the gaussian intensity profile of the raman beams , while the noise is mainly due to detection noise .
in these conditions , with @xmath23 , the acceleration noise of the ai equals @xmath45 . at this sensitivity level ,
the hybrid sensor is able to measure inertial effects more than 300 times weaker than the typical acceleration fluctuations of the aircraft .
we emphasize that reaching such a high resolution is possible thanks to the appropriate combination of mas ( see the `` methods '' section ) , the success of operating the ai in the plane , and the use of our method for the acceleration measurement . in the present configuration of the experiment
, the snr degrades at 1-g when @xmath46 increases above @xmath47 because the atoms fall down and escape the raman and detection beams @xcite .
this limitation could be overcome thanks to additional raman beam collimators and by changing the orientation of the detection lasers . in 0-g ,
the experiment falls with the atoms and the snr is not constrained by gravity any more .
the snr may also be improved significantly in the future thanks to a better detection system ( e.g. more stable detection lasers ) and the use of ultra - cold atoms .
we focus now on the microgravity operation of the matter wave inertial sensor and on its possible application to fundamental physics tests such as that of the uff .
such a test can be carried out with two ais measuring the acceleration of two different atoms with respect to the same mirror that retroreflects the raman lasers . in airborne or spaceborne experiments ,
the mirror constitutes an ill - defined inertial reference because of the craft s vibrations .
this might degrade the sensitivity of the test , unless the vibrations impact the two interferometers in the same way .
vibration noise rejection occurs when conducting differential measurements such as in the operation of gradiometers @xcite or gyroscopes @xcite , and is expected in uff tests based on atom interferometry @xcite .
the rejection efficiency depends on the two species used for the uff test and is maximum for simultaneous interrogation of the two atoms ( see supplementary information ) . in the case of a finite rejection
, the impact of the acceleration noise can further be reduced by measuring the vibrations of the mirror with mas to substract them from the differential phase measurement . in this way
, the mas are used to release the requirements on the vibration damping of the craft as they measure the accelerations not rejected in the differential operation of the two ais .
the efficiency of that technique is limited by the performances of the mas as their imperfections ( e.g. their non - linearities ) translate into residual vibration noise impairing the differential acceleration measurement . in the following ,
we investigate vibration noise rejection by operating our @xmath48rb sensor in a differential mode .
we use a sequence of 4 light pulses to build a two - loop ai @xcite which is equivalent to two successive one - loop interferometers head to tail ( figure 4b ) .
the 4-pulse ai provides a signal resulting from the coherent substraction of two spatially and temporally separated inertial measurements and is therefore expected to be less sensitive to the low frequency inertial effects . to illustrate the noise rejection , we operate the 3-pulse and 4-pulse interferometers with a same total interrogation time of 20 ms , and compare the mas - ai correlation for each geometry ( figure 4 ) . figures 4c and 4d show that the quality of the correlation ( ratio of the sinusoid contrast to the mean error bar ) is clearly improved in the differential geometry compared to the 3-pulse interferometer . to understand this difference
, we estimate the total noise of the correlation , defined as @xmath49 .
it results , in the linear range of the interferometer , from the quadratic sum of two independent contributions : the atomic phase noise @xmath50 , and the vibration noise @xmath51 not measured by the mas . as we can estimate @xmath52 independently from the correlation
( see the second `` methods '' subsection ) , the comparison of @xmath53 and @xmath52 indicates whether the sensor sensitivity is limited by the atomic noise ( @xmath54 ) or by the residual vibration noise ( @xmath55 ) . in figures 4e and 4f ,
we have represented @xmath53 ( errors bars ) and @xmath52 ( vertical spacing between the green lines ) for the one - loop and the two - loop interferometers , respectively .
figure 4e shows that the sensitivity of the 3-pulse sensor is limited by the vibration noise not measured by the mas , as @xmath55 . on the contrary , figure 4f reveals that the atomic phase noise is the main limit to the sensitivity of the 4-pulse sensor , as @xmath54 .
the snr in the 4-pulse interferometer is less than in its 3-pulse counterpart as the additional light pulse reduces the interference fringe contrast due to the raman beam intensity inhomogeneities and the transverse temperature of the cloud . in spite of greater atomic noise ,
the quality of the correlation 4f is better than that of correlation 4e , which shows that the 4-pulse sensor operates a vibration noise rejection . in the 4-pulse geometry ,
the two elementary interferometers operate one after another and share accelerations of frequency below @xmath56 . in a uff test , two ais of equal scale factor ( @xmath57 )
will interrogate two different atoms almost simultaneously , so that the noise rejection is expected to be much more efficient ( see @xcite and the supplementary information ) . in that case
, the precision of the inertial sensor might be limited by the atomic phase noise , i.e. by the snr of the interferometer . for the 3-pulse ai data in figure 4c and 4e ( @xmath58 )
, we estimate a snr of 2.1 , which corresponds to an acceleration sensitivity of the matter wave sensor of @xmath0 , in the context of the high vibration noise rejection expected for the uff test . that sensitivity level may be greatly improved in the future by using a highly collimated atomic source in microgravity @xcite .
we finally discuss possible improvements of our setup , both for inertial guidance and fundamental physics applications . in the former case , increasing the resolution of the accelerometer will be achieved by using well characterized mas to increase @xmath34 , and by improving the snr of the ai .
reaching an interrogation time @xmath59 in the plane ( where the acceleration fluctuations are @xmath60 rms at 1-g ) would require mas whose scale factor frequency response is determined with a relative accuracy of @xmath61 .
we believe that such precision can be achieved with state - of - the - art ma technology , e.g. capacitive mems sensors , and we now work at implementing these sensors in our setup . together with a shot noise limited ai with @xmath62 ( as in @xcite ) , the resolution of the hybrid sensor would be @xmath63 per shot , which would represent a major advance in inertial navigation as well as in airborne gravimetry .
we note that the present analysis does not report the in - flight bias of the matter wave sensor since no other airborne accelerometer of similar accuracy was available onboard to proceed to the comparison .
however , it has been demonstrated in laboratories that atom accelerometers can reach biases of the order of @xmath64 under appropriate conditions @xcite .
tests of fundamental physics would also require performant mas to remove the residual aircraft s or satellite s acceleration noise not rejected in the differential measurement .
for a uff test in the 0-g plane and a vibration rejection efficiency of 300 ( explained in @xcite and in the supplementary information ) , a differential acceleration sensitivity of @xmath65 per shot ( @xmath66 , @xmath67 ) could be achieved with mas of @xmath61 relative accuracy if the vibrations during the 0-g phase are damped to the @xmath68 level . in space
, high performance mas such as the sensors developped for the goce mission could be used to determine the residual accelerations of the satellite ( @xmath69 ) with a resolution @xcite of the order of @xmath70 .
the vibration rejection efficiency of 300 would thus limit the impact of the acceleration noise on the interferometric measurement to @xmath71 per shot , which would stand for a minor contribution in the error budget .
therefore , high precision test of the equivalence principle could be conducted in space without the strong drag - free constraints on the satellite that represent a major challenge in current space mission proposals . to conclude
, we have demonstrated the first airborne operation of a cold atom inertial sensor , both at 1-g and in microgravity .
we have shown how the matter wave sensor can measure the craft acceleration with high resolutions .
our approach proposes to use mechanical devices which probe the coarse inertial effects and allow us to enter the fine measurement regime provided by the atom accelerometer . in the future ,
instruments based on the combination of better characterized mechanical sensors and a shot noise limited ai could reach sensitivities of the order of few @xmath72 in one second aboard aircrafts . thus , our investigations indicate that sensors relying on cold atom technology may be able to detect inertial effects with resolutions unreached so far by instruments aboard moving crafts characterized by high acceleration levels .
cold atom sensors offer new perspectives in inertial navigation thanks to their long term stability as they could be used to correct the bias ( @xmath73 ) of the traditional sensors monitoring the crafts motion , below the @xmath72 level . in geophysics ,
airborne gravity surveys may also benefit from the accuracy of ais @xcite .
moreover , we have operated the first matter wave sensor in microgravity .
we have shown how differential interferometer geometries enable to reject vibration noise of the experimental platform where new types of fundamental physics tests will be carried out .
our result in 0-g suggests that the high sensitivity level of matter wave interferometers may be reached on such platforms , and support the promising future of ais to test fundamental physics laws aboard aircrafts , sounding rockets or satellites where long interrogation times can be achieved @xcite .
while many quantum gravity theories predict violations of the uff , ais may investigate its validity at the atomic scale , with accuracies comparable to those of ongoing or future experiments monitoring macroscopic or astronomical bodies @xcite .
for a time varying acceleration @xmath74 of the retroreflecting mirror , the phase of the interferometer at time @xmath22 is given by @xmath75 where @xmath76 is the acceleration response of the 3-pulse ai .
it is a triangle - like function @xcite that reads : @xmath77 \\ 2t-(t-\tilde{t_i } ) & \text{if } \
t \in [ \tilde{t_i}+t,\tilde{t_i}+2 t ] , \end{array } \right.\ ] ] with @xmath78 being the time of the first raman pulse at the @xmath79-th measurement .
the mean acceleration that we infer is defined as @xmath80 .
thanks to the mas , we estimate the phase @xmath81 which is expected to be measured by the interferometer by averaging in the time domain the signal @xmath20 by the ai response function : @xmath82 the mas - ai correlation function can be written as @xmath83 and expresses the probability to measure the atomic signal @xmath24 at time @xmath31 , given the acceleration signal @xmath20 recorded by the mas . the estimated acceleration used in figure 2
is defined by @xmath84 . for simplicity ,
we have neglected in equation ( 1 ) the raman pulse duration @xmath85 with respect to the interrogation time @xmath46 .
the exact formula can be found in @xcite and has been used in the data analysis to estimate the phase @xmath86 .
we calculate the probability density function ( pdf ) of the ai measurements @xmath24 and fit it with the pdf of a pure sine ( a `` twin - horned '' distribution ) convolved with a gaussian of standard deviation @xmath87 .
the fit function reads : @xmath88^{-1/2 } \times \frac{1}{\sigma_p\sqrt{2\pi}}\text{exp}\big({-\frac{(x - x^{\prime})^2}{2\sigma_p^2}}\big),\ ] ] where @xmath89 is a normalization factor . in this way
, we estimate the amplitude @xmath15 and the offset value @xmath14 of the interference fringes , i.e. we estimate the ai response @xmath90 .
the fitted parameters @xmath15 and @xmath87 provide an estimate of the in - flight signal to noise ratio of the interferometer , given by @xmath91 .
figure 5 illustrates the method for the data used in figure 2 and 3 , corresponding to the interrogation time @xmath92 ms .
the fitted parameters are @xmath93 , @xmath94 and snr@xmath95 .
this method is independent on the mas signal and informs both on the response of the ai , and on its one - shot acceleration sensitivity @xmath96 .
the characteristics of the interferometer are thus known without any additional calibration procedure . in particular
, we can estimate the noise level of the matter wave sensor , given by @xmath97 for a white atomic phase noise , and which equals @xmath98 for the data in figure 5 , where @xmath23 .
our analysis estimates the snr by taking into account only the atomic noise due to detection noise or fluctuations of the fringe offset and contrast .
it does not account for the laser phase noise that could impact the sensitivity of the acceleration measurement for long interrogation times @xcite .
however , we demonstrated the low phase noise of our laser system during previous parabolic flight campaigns @xcite , which is at least one order of magnitude below the estimated atomic phase noise for the interrogation times we consider in this work ( @xmath99 ) .
therefore , we have neglected in this communication the laser phase noise contribution when evaluating the sensitivity of the sensor .
thanks to the mas signal , we first determine the fringe number @xmath100 $ ] where the interferometer operates at time @xmath22 , with @xmath29 being the reciprocity interval of the matter wave sensor .
the values @xmath32 are represented by the black step - like curve in figure 3a .
second , we use the atomic measurements @xmath24 to deduce the acceleration @xmath101 measured by the ai in its reciprocity region @xmath102 $ ] , and given by : @xmath103 the total acceleration @xmath27 is finally computed as : @xmath104 the acceleration measurement process is illustrated in figure 6 .
the main limitation of the mas comes from the nonlinearities in their frequency response ( phase and gain ) , which means that the signal @xmath20 is not exactly proportional to the acceleration of the retroreflecting mirror .
this results in errors in the estimation of the phase @xmath86 that impair the mas - ai correlation .
the nonlinearities typically reach amplitudes @xmath105 within the ai bandwidth which equals @xmath106 500 hz . ( the acceleration frequency response @xmath107 of an ai has been measured in @xcite and corresponds to the response of a second order low pass filter of cutoff frequency @xmath56 ) . to reduce them ,
we combine two mechanical devices of relatively flat frequency response within two complementary frequency bands : a capacitive accelerometer ( sensorex sx46020 ) sensing the low frequency accelerations ( 0 to 1 hz ) , and a piezoelectric sensor ( imi 626a03 ) measuring the rapid fluctuations ( 1 to 500 hz ) . in this way
, we achieve @xmath108 . for @xmath37 ,
these nonlinearities correspond to errors on @xmath86 of about 0.5 rad rms .
further details on the errors in the estimation of the phase due to the mas nonlinearities are given in the supplementary information .
another limitation comes from the mas internal noise , especially that of the capacitive one whose noise level integrated in [ 0 - 1hz ] equals @xmath109 .
this noise is about one order of magnitude lower than this due to the non - linearities of the mas , and is independent of the acceleration level in the plane .
axis cross - talk of the mas of the order of @xmath110 is taken into account for the estimation of the phase , so that the errors on @xmath86 due to the mas axis coupling are negligible .
finally , the bias of the capacitive accelerometer is specified at the @xmath111 level , so that the mean estimated acceleration might drift during the flight ( 4 hours ) .
this results in a displacement of the dark fringe position in the mas - ai correlation plots .
this work is supported by cnes , dga , rtra - triangle de la physique , anr , esa , esf program euroquasar and fp7 program isense .
laboratoire charles fabry and lne - syrte are part of ifraf .
we are grateful to alain aspect and mark kasevich for discussions and careful reading of the manuscript , and to linda mondin for her productive implication in the i.c.e . project .
we thank andr guilbaud and the novespace staff for useful technical advice . | inertial sensors relying on atom interferometry offer a breakthrough advance in a variety of applications , such as inertial navigation , gravimetry or ground and space based tests of fundamental physics .
these instruments require a quiet environment to reach their performance and using them outside the laboratory remains a challenge . here , we report the first operation of an airborne matter wave accelerometer set up aboard a 0-g plane and operating during the standard gravity ( 1-g ) and microgravity ( 0-g ) phases of the flight . at 1-g ,
the sensor can detect inertial effects more than 300 times weaker than the typical acceleration fluctuations of the aircraft .
we describe the improvement of the interferometer sensitivity in 0-g , which reaches @xmath0 with our current setup .
we finally discuss the extension of our method to airborne and spaceborne tests of the universality of free fall with matter waves .
matter wave inertial sensing relies on the capability of controling the wave nature of matter to build an interferometer and accurately measure a phase difference @xcite . since the particle associated to the matter wave senses inertial or gravitational effects , the interferometer represents an accurate inertial probe .
in particular , atom interferometers ( ais ) have benefited from the outstanding developments of laser - cooling techniques and reached accuracies comparable to those of inertial sensors based on optical interferometry .
thanks to their long term stability , ais offer a breakthrough advance in accelerometry , gyroscopy and gravimetry , for applications to inertial guidance @xcite , geoid determinations @xcite , geophysics @xcite and metrology @xcite .
in addition , ais are excellent candidates for laboratory - based tests of general relativity ( gr ) that could compete with the current tests which consider astronomical or macroscopic bodies @xcite .
for example , ais may provide new answers to the question of whether the free fall acceleration of a particle is universal , i.e. independent of its internal composition and quantum properties .
although this principle known as the universality of free fall ( uff ) has been tested experimentally @xcite to a few parts in @xmath1 , various extensions to the current theoretical physics framework predict its violation @xcite .
it is thus important to test experimentally these theoretical models with different types of particles .
ais also open perspectives for further tests of gr such as the detection of gravitational waves @xcite .
all these fundamental tests may benefit from the long interrogation times accessible on microgravity platforms @xcite or in space @xcite .
because of its high sensitivity , running an ai has required until now low vibration and high thermal stability environments which can only be found in dedicated ground or underground platforms .
we report here the first operation of a matter wave inertial sensor in an aircraft , both at 1-g and in microgravity ( 0-g ) .
our matter wave interferometer uses @xmath2 atoms and operates aboard the novespace a300 - 0 g aircraft taking off from bordeaux airport , france @xcite .
this plane carries out parabolic flights during which 22 second ballistic trajectories ( 0-g ) are followed by two minutes of standard gravity flight ( 1-g ) .
the ai measures the local acceleration of the aircraft with respect to the inertial frame attached to the interrogated atoms which are in free fall . in the first part of this communication
, we describe the inertial measurements performed by our instrument and show how the matter wave sensor achieves a resolution level more than 300 times below the plane acceleration level .
we present the general method that is used to operate the ai over a wide acceleration range and to reach such a resolution . in the second part
, we demonstrate the first operation of a matter wave inertial sensor in 0-g .
microgravity offers unique experimental conditions to carry out tests of fundamental physics .
however , these experiments are conducted on platforms such as planes , sounding rockets or satellites which are not perfectly free - falling , so that the residual craft vibrations might strongly limit the sensitivity of the tests . overcoming this problem
generally requires the simultaneous operation of two sensors in order to benefit from a common mode vibration noise rejection .
for example , conducting a matter wave uff test implies the simultaneous interrogation of two different atomic species by two ais measuring their acceleration difference @xcite . in the present work ,
we investigate the 0-g operation of our one - species ai in a differential configuration to illustrate a vibration noise rejection .
our achievements ( 0-g operation and noise rejection ) constitute major steps towards a 0-g plane based test of the uff with matter waves at the 10@xmath3 level , and towards a space based test @xcite below @xmath4 .
such an experiment in space has been selected for the next medium - class mission in esa s cosmic vision 2020 - 22 in the frame of the ste - quest project @xcite .
this paper presents the first airborne and microgravity operation of a matter wave inertial sensor .
we introduce a new and original method that allows to use the full resolution of an atom interferometer in the presence of high levels of vibration .
we also show how high precision tests of the weak equivalence principle may be conducted with differential atom interferometry . |
according to the fundamental theory of the strong interactions , quantum chromo - dynamics ( qcd ) , normal nuclear matter is expected to undergo a phase transition to deconfined quarks and gluons when its density and/or temperature are high @xcite .
to produce such a quark - gluon plasma ( qgp ) in the laboratory , experiments involving collisions of nuclei at relativistic energies have been carried out at the cern sps @xcite .
possible evidences for the production of this new phase of matter have recently been announced @xcite .
further experiments at the relativistic heavy ion collider ( rhic ) , which allows collisions at much high energies than those available previously , are expected to provide a better opportunity to create the quark - gluon plasma and to study its properties . since
quarks and gluons can not be directly detected in experiments , many indirect observables have been proposed as possible signatures for the qgp .
these include : enhanced production of strange hadrons as a result of the short strangeness equilibration time in qgp @xcite , suppression of @xmath0 production due to color screening in qgp @xcite , quenching of high @xmath1 jets due to passage through the qgp @xcite , and enhancement of low mass dileptons as a result of medium modifications of hadron properties in high density matter @xcite .
although all these signals have been observed in heavy ion collisions at cern sps , alternative explanations without invoking the formation of the quark - gluon plasma have also been proposed . as the qgp
is expected to be produced at rhic , it is of interest to study these signatures in heavy ion collisions at such high energies . since the quark - gluon plasma has a finite size , exists for a finite time , and may not be in equilibrium , it is important to use a dynamical model to take into account these effects . using the parton cascade model @xcite , satz and srivastava @xcite
have recently studied the time evolution of the density profile of the parton system in these collisions and demonstrated the possibility of using a dynamical model to study the onset of deconfinement and its effect on quarkonium suppression . in this paper
, we shall investigate explicitly @xmath0 suppression in central nucleus - nucleus collisions at rhic energies using a multiphase transport model ( ampt ) @xcite .
since the ampt model includes both initial partonic and final hadronic interactions as well as the transition between these two phases of matter , it allows us to study the relative importance of the partonic and hadronic effects on @xmath0 suppression .
this paper is organized as follows . in section [ ampt ]
, we briefly describe the ampt model for ultrarelativistic heavy ion collisions and its extension to include @xmath0 production .
the mechanisms for @xmath0 suppression are given in section [ absorption ] .
results for au+au and s+s collisions at rhic are given in section [ results ] .
first , we show the time evolution of the particle number and energy densities in both the partonic and hadronic matters .
this is then followed by the time evolution of various @xmath0 suppression effects and the @xmath0 survival probability .
finally , a summary is given in section [ summary ] .
in the ampt model , the initial conditions are obtained from the minijets generated by the heavy ion jet interaction generator ( hijing ) @xcite by using a woods - saxon radial shape for the colliding nuclei and including the nuclear shadowing effect on partons via the gluon recombination mechanism by mueller - qiu @xcite .
after the passage of the colliding nuclei , the gyulassy - wang model @xcite is then used to generate the initial space - time information of partons .
the subsequent time evolution of the parton phase - space distribution is modeled by the zhang s parton cascade ( zpc ) @xcite , which at present includes only the gluon elastic scattering .
after partons stop interacting , the lund string fragmentation model @xcite is used to convert them into hadrons .
the dynamics of resulting hadronic matter is described by a relativistic transport model ( art ) @xcite .
details of the ampt model can be found in @xcite . to include @xmath0 in the ampt model , we use the perturbative approach @xcite as
its production probability is small in heavy ion collisions .
specifically , the reaction @xmath2 is selected from the pythia program whenever there is an inelastic scattering between projectile and target nucleons . instead of allowing the produced @xmath0 to undergo multiple interactions , fragmentation , and decay in the pythia
, its momentum is stored in a file and read into hijing .
the transverse position of the @xmath0 is determined by propagating it to the time of formation from the average transverse position of the colliding projectile and target nucleons . since the @xmath0 production probability is increased from @xmath3 to @xmath4 , each @xmath0 is given a probability of @xmath5 in order to have the correct number in an event , given by the total probability of all @xmath0 s .
the @xmath0 formation time is taken to be 0.5 fm/@xmath6 , which is suggested by the virtuality argument of kharzeev and satz @xcite and is also consistent with the uncertainty principle used for estimating the lifetime of an expanding @xmath7 @xcite . before this time , the @xmath8 pair is considered a precursor @xmath0 .
after a pair of @xmath8 is produced in the initial collisions , whether it can materialize as a @xmath0 depends on its interactions in the initial partonic and final hadronic matters . in the partonic matter , the @xmath8 pair , which are initially close in phase space and would normally form a @xmath0 after the formation time of 0.5 fm/@xmath6 , will move beyond the confinement distance of @xmath0 and become unbound if the plasma screening remains strong @xcite .
they will later combine with the more abundant light quarks to form instead the charm hadrons when the parton density is low .
the critical density for the plasma screening to be effective can be estimated using @xmath9 @xcite , where @xmath10 is the critical debye screening mass , @xmath11 is the qcd coupling constant , and @xmath12 is the slope parameter of the transverse momentum distribution . using @xmath13 gev , @xmath14 , and @xmath15 gev ,
one obtains a critical density of @xmath16 @xmath17 . following the method of ref .
@xcite , we first determine from the ampt model the time evolution of the radius of the volume in which the parton density is above the critical density .
a @xmath0 is thus not formed from the @xmath8 pair if its radial position after the formation time of 0.5 fm/@xmath6 is within the critical radius at that time . besides dissociation due to plasma screening , a @xmath0 may also be dissociated by collisions with gluons in the partonic matter @xcite .
the cross section for @xmath18 has been estimated in ref .
@xcite to be 3 mb .
similarly , a precursor @xmath0 can also be dissociated by gluon scattering .
the cross section is expected to be smaller as its size is smaller than a physical @xmath0 . in the present study , we ignore the difference between the scattering of a gluon with precursor and physical @xmath0 , and use the same cross section .
since this cross section is not well - known , we have also used a value of @xmath19 mb in the following study . in the hadronic matter following the parton stage ,
a @xmath0 can be further destroyed by collisions with hadrons .
the @xmath0 absorption cross section by baryons is taken to be 6 mb @xcite while that by meson is taken to be 3 mb @xcite above their respective thresholds .
although earlier studies based on the perturbative qcd @xcite and simple hadronic model @xcite give a much smaller @xmath0 absorption cross sections by hadrons , the above values are consistent with recent studies using the quark - interchange model @xcite and the more complete hadronic model @xcite . using these cross sections in the transport model , it has been found that the observed suppression of @xmath0 production in heavy ion collisions at the sps energies can be reasonably described except for the most central collisions @xcite . as the formation time of @xmath0 is comparable to the passing time ( @xmath20 fm/@xmath6 ) of two colliding nuclei at rhic energies , @xmath0 absorption by the projectile and target nucleons
is not expected to be important @xcite and will be ignored in the present study .
we note that as the @xmath0 is treated perturbatively in the simulation , the momenta of other hadrons are not affected by their interactions with @xmath0 @xcite .
in fig . [ fig : auauden5 ] , we show the time evolution of the particle number and energy densities of partons and hadrons in central au+au collisions at @xmath21 gev .
they are the densities in the central cell , which has a transverse radius of 1 fm for partons and 2 fm for hadrons .
the longitudinal dimension of the central cell is taken to be @xmath22 of the time @xmath23 , which is equivalent to taking the central space - time rapidity cell .
the results are not significantly changed if @xmath24 of the time is used for the longitudinal dimension .
we see that the initial parton density is about @xmath25 @xmath17 and is much higher than the @xmath0 dissociation critical density ( about @xmath26 @xmath17 ) given above .
the time evolution of these densities is seen to deviate from that based on the ideal bjorken boost invariant scenario @xcite , which would lead to a linear curve on the log - log plot .
this difference is mainly due to the more realistic treatment of initial collisions by using the the gyulassy - wang model for the formation time and the presence of radial flow as well as a gradual freeze - out @xcite at the later hadronic stage . from the ratio of the parton energy density to its number density
, one sees that the average energy of a parton is more than @xmath19 gev .
the parton stage lasts about @xmath27 fm/@xmath6 , while the hadron stage starts gradually at around @xmath28 fm/@xmath6 and lasts until about @xmath29 fm/@xmath6 .
the average energy of a hadron is less than @xmath19 gev , since the central rapidity is dominated by mesons instead of baryons in ultrarelativistic nuclear collisions . in fig .
[ fig : ssden9 ] , we show the results for central collisions of s+s at @xmath21 agev . due to the smaller size of the system comparing to that of au+au collisions , the plasma lifetime is shorter , i.e. , about @xmath30 fm/@xmath6 , and the hadron stage sets in at about @xmath27 fm/@xmath6 .
we note that even in the smaller parton system produced in s+s collisions , the initial density is much higher than the critical density for @xmath0 dissociation . in fig .
[ fig : auaurho4c ] and fig .
[ fig : ssrho4a6 ] , we show , respectively , the parton density in central au+au and s+s collisions at different times . as minijet gluons are produced from initial hard collisions , their densities at different radii reflect the number of initial binary nucleon - nucleon collisions . as seen from the figure , they are different from the initial nuclear density distribution given by a woods - saxon form .
as expected , both the size and lifetime of the partonic matter produced in s+s collisions are smaller than those in au+au collisions . from the time evolution of the parton density
, one can determine the time evolution of the critical radius for @xmath0 dissociation , and this is shown in fig .
[ fig : auaurhoc3 ] by full dots for au+au collisions and open dots for s+s collisions .
the solid lines are polynomial fits to the above results using the form , @xmath31 for @xmath32 , with @xmath33 for @xmath34 and @xmath35 for @xmath36 . for the partonic matter produced in au+au collisions , we have @xmath37 fm/@xmath6 , @xmath38 fm/@xmath6 , @xmath39 , @xmath40 , @xmath41 , and @xmath42 . for s+s collisions , they are @xmath43 fm/@xmath6 , @xmath44 fm/@xmath6 , @xmath45 , @xmath46 , @xmath47 , and @xmath48 . in au+au collisions ,
the critical radius for @xmath0 dissociation extracted from our model is about 6 fm at beginning of the parton cascade and vanishes after about 1.2 fm/@xmath6 .
the critical radius is reduced to about 2 fm initially and vanishes after only about 0.5 fm in s+s collisions .
since the duration of the partonic matter that is above the critical density for @xmath0 dissociation is shorter than the @xmath0 formation time , @xmath0 suppression due to plasma screening is therefore unimportant in central s+s collisions . including the above three mechanisms for @xmath0 suppression in the ampt model
, we have evaluated the @xmath0 survival probability in ultrarelativistic heavy ion collisions at rhic energies . in fig .
[ fig : prob2a ] , we show for central au+au collisions the time evolution of the @xmath0 formation probability @xmath49 , the @xmath0 absorption probability @xmath50 by gluon scattering , the @xmath0 dissociation probability @xmath51 by plasma screening , the @xmath0 absorption probability @xmath52 by hadrons , and the @xmath0 survival probability @xmath53 .
the results are taken for the @xmath0 produced in the central rapidity interval @xmath54 .
it is seen that the @xmath0 formation probability increases quickly with time and about @xmath55 of the @xmath0 are formed by @xmath56 fm/@xmath6 .
absorption by gluons starts very early in the process and ends at about @xmath19 fm/@xmath6 after dissociation due to plasma screening begins .
the latter also ends at about @xmath19 fm/@xmath6 .
both dissociation due to plasma screening and absorption by gluon collisions give comparable contributions , accounting for the suppression of about 90% of the @xmath0 , while absorption by hadrons contributes only about a few percent .
the final @xmath0 survivability is about 6% .
the results for the case without plasma screening is shown in fig .
[ fig : prob2b ] .
comparing to fig .
[ fig : prob2a ] , we see that the @xmath0 absorption probability by gluons is similar for time up to @xmath56 fm/@xmath6 , but it lasts longer until @xmath27 fm/@xmath6 .
this indicates that there is a competition between @xmath0 suppression due to gluon scattering and plasma screening , i.e. , some @xmath0 s that are destroyed by collisions with gluons would have been dissociated by plasma screening if they are not allowed to scatter with gluons .
[ fig : prob2c ] shows the results without @xmath0 absorption by gluon scattering . in this case , the plasma dissociation probability saturates quickly .
compared to that shown in fig .
[ fig : prob2b ] , the @xmath0 suppression probability is seen to start late but saturate early .
this difference in the survival probability may be seen by studying the azimuthal distribution of final survival @xmath0 in mid - central collisions .
it is interesting to note that @xmath0 suppression during the parton stage is similar whether when both plasma screening and gluon scattering are present or when only one of them is present .
the effect of reducing the @xmath0 scattering cross section by gluon is shown in fig .
[ fig : prob2d ] . compared with fig .
[ fig : prob2a ] , we see that the decrease in the contribution from gluon scattering is partly compensated by the plasma screening . in table
[ table : auauprob ] , we summarize the final absorption probabilities due to gluons ( @xmath57 ) , hadrons ( @xmath58 ) , plasma dissociation ( @xmath59 ) , and the survival probability ( @xmath60 ) in central au+au collisions .
it is seen that the @xmath0 survival probability after the parton stage , @xmath61 , is about 9% and 17% for @xmath62 3 mb and 1 mb , respectively .
the parton stage thus has a large effect on @xmath0 suppression in heavy ion collisions at rhic . in both cases ,
the @xmath0 suppression factor in the hadron stage is @xmath63 .
depending on the different assumptions on the mechanisms for @xmath0 suppression , the final survival probability can range from about @xmath64 to about @xmath65 .
this indicates that finite size effects may be important in @xmath0 suppression at rhic energies .
we have also studied @xmath0 suppression in s+s collisions at rhic energies , and the results are summarized in table [ table : ssprob ] . as expected from the discussion of fig . [
fig : auaurhoc3 ] , there is no contribution from plasma screening in collisions of such light system , although the initial density is above the critical density . as a result , the @xmath0 has an appreciable survival probability ( about 50% ) after absorption by gluons and hadrons .
since @xmath0 absorption by gluon scattering still contributes appreciably to the final @xmath0 suppression , it may provide an opportunity for studying this mechanism .
we note that the @xmath0 suppression factor in the hadron stage is about @xmath66 , which is smaller than in au+au collisions .
.final @xmath0 absorption probability @xmath57 by gluon scattering , @xmath0 dissociation probability @xmath59 by plasma screening , @xmath0 absorption probability @xmath58 by hadrons , and @xmath0 survival probability @xmath60 for au+au ( @xmath67 , @xmath21 agev ) .
@xmath68 indicates @xmath69 mb while @xmath70 indicates @xmath71 mb .
[ cols="<,^,^,^,^ " , ]
in summary , we have studied @xmath0 suppression in au+au and s+s collisions at rhic energies in a multiphase transport model .
it is found that finite size effects may be important in @xmath0 suppression at rhic energies even in central au+au collisions . both @xmath0 suppression due to plasma screening and gluon
scatterings are important in au+au collisions , leading to a final @xmath0 survival probability of only about @xmath64 . for s+s collisions , even though the initial energy density is above the critical density for @xmath0 dissociation , only @xmath0 absorption by gluon and hadron scatterings contribute to its suppression as a result of the short lifetime of the plasma .
@xmath0 suppression in heavy ion collisions between light nuclei thus provides a possible opportunity for studying the interactions between @xmath0 and gluon . in our study ,
we have ignored both @xmath0 formation from the recombination of @xmath8 pairs during hadronization and @xmath0 production from charm mesons in the final hadronic matter . the number of @xmath0 formed from charm quarks during hadronization can be roughly estimated from considerations of the phase space and the spin and color factors . at the end of parton stage ,
the volume of the partonic matter in the central unit of rapidity is approximately given by @xmath72 , where @xmath73 and @xmath74 are the transverse radius and longitudinal dimension , respectively .
for au+au collisions , they are @xmath75 and @xmath76 . the probability that an anticharm quark is found within a distance of the @xmath0 radius from a charm quark
is given by the ratio of the size of the @xmath0 to the size of the partonic matter . with a @xmath0 radius @xmath77 ,
one obtains the probability @xmath78 . in order for this charm pair to form a @xmath0
, their spins must be parallel and their total color should be neutral .
this reduces the probability by the factor @xmath79 . requiring that the charm pair has a relative momentum distribution within @xmath80 gev at @xmath81 mev further suppresss the probability by a factor of 0.29 .
since there are about 1.5 charm pairs per unit rapidity per central au+au collision at rhic , the number of @xmath0 formed from their recombination is about @xmath82 , which is much smaller than the expected number of @xmath0 , i.e. , @xmath83 . in the above ,
the factor 40 is the ratio of charm to @xmath0 production in initial hard collisions @xcite and the factor 0.061 is the final @xmath0 survival probability from table [ table : auauprob ] . as to @xmath0 production from charm mesons in hadronic matter
, it has been shown to be insignificant in heavy ion collisions at rhic , although it may not be negligible at lhc @xcite .
we have not included the effect due to @xmath84 and @xmath85 , which are also produced in initial collisions and can decay to @xmath0 . since these particles are less bound than @xmath0 , they are more likely to be dissociated and absorbed in both the quark - gluon plasma and the hadronic matter . as a result ,
inclusion of @xmath86 and @xmath85 is expected to lead to a smaller number of @xmath0 than predicted in the present study .
also , we have used the default gluon shadowing in hijing without flavor and scale dependence , which is different from the one suggested by eskola @xcite that includes such dependence .
if the latter is used , then we expect a reduction of the @xmath0 survival probability as the gluon density would be higher at rhic energies @xcite .
these effects will be studied in the future .
the authors thank m. gyulassy and j. nagle for helpful discussions .
they also thank the parallel distributed system facility at national energy research scientific computer center for providing computing resources .
this work was supported by the national science foundation under grant no .
phy-9870038 , the welch foundation under grant no .
a-1358 , and the texas advanced research program under grant nos .
fy97 - 010366 - 0068 and fy99 - 010366 - 0081 . the work of b.a .
li was supported in part by the nsf under grant no . 0088934 and the arkansas science and technology authority under grant no .
sa was partially supported by the natural science foundation and nuclear industry foundation of china .
quark matter99 , proceedings of the 14th international conference on ultra - relativistic nucleus - nucleus collisions , torino , italy , may 10 - 15 , 1999 , nucl . phys . * a661 * , 1c-766c ( 1999 ) , and references therein . | using a multiphase transport model , we study the relative importance of @xmath0 suppression mechanisms due to plasma screening , gluon scattering , and hadron absorption in heavy ion collisions at the relativistic heavy ion collider .
we find that for collisions between heavy nuclei such as au+au , both plasma screening and gluon scattering are important . as a result ,
the effect due to absorption by hadrons becomes relatively minor .
the final @xmath0 survival probability in these collisions is only a few percent . in the case of collisions between light nuclei such as s+s ,
the effect of plasma screening is , however , negligible in spite of the initial high parton density .
the final @xmath0 survival probability thus remains appreciable after comparable absorption effects due to gluons and hadrons . |
mixing by fluid flows is a ubiquitous natural phenomenon that plays a central role in many of the applied sciences and engineering .
a geophysical example is the mixing of aerosols ( e.g. , @xmath0 supplied by a volcano , say , or by human activity ) in the atmosphere .
aerosols are dispersed by molecular diffusion on the smallest scales but are more effectively spread globally by atmospheric flows .
the density and density fluctuations of some aerosols influence the albedo of the earth and thus have a significant environmental impact .
hence it is important to understand fundamental properties of dispersion , mixing , and the reduction of concentration fluctuations by stirring flow fields .
various aspects of mixing have been the focus of many review articles . at the most basic level
, the mixing of a passive scalar can be modeled by an advection - diffusion equation for the scalar concentration field with a specified stirring flow field . in this work
we will focus on problems where fluctuations in the scalar field are generated and sustained by temporally steady but spatially inhomogeneous sources .
the question of interest here is this : for a given source distribution , how well can a specified stirring flow mix the scalar field ? mixing effectiveness can be measured by the scalar variance over the domain .
a well - mixed scalar field will have a more uniform density with relatively `` small '' variance while increased fluctuations in the scalar density will be reflected in a `` large '' variance .
we put quotes around the quantifiers small and large because the variance is a dimensional quantity that needs an appropriate dimensional point of reference from which it is being measured .
several years ago thiffeault _ et al _ @xcite introduced a notion of `` mixing enhancement '' for a velocity field stirring a steadily sustained scalar by comparing the bulk ( space - time ) averaged density variance with and without advecting flow .
mixing is accomplished by molecular diffusion alone in the absence of stirring , which can be quite effective on small scales but is not generally so good at breaking up and dispersing large - scale fluctuations quickly . stirring can greatly enhance the transport of the scalar from regions of excess density to depleted regions , suppressing the variance far below its diffusion - only value .
the magnitude of this variance suppression by the stirring the ratio of the variance without stirring to the variance in the presence of stirring is a dimensionless quantity that provides a sensible gauge of the mixing effectiveness of the flow .
different advection fields will have different mixing efficiencies stirring scalars supplied by different sources .
it is then of obvious interest both to determine theoretical limits on mixing enhancements for various source configurations and to explore whether those limits may be approached or even perhaps achieved-for particular flows .
there have been many studies of stirring and mixing of a scalar with fluctuations sustained by spatially inhomogeneous sources and sinks .
some of the earliest are by townsend @xcite , who was concerned with the effect of turbulence and molecular diffusion on a heated filament .
he found that the spatial localization of the source enhanced the role of molecular diffusivity .
durbin @xcite and drummond @xcite introduced stochastic particle models to turbulence modeling , and these allowed more detailed studies of the effect of the source on diffusion .
sawford and hunt @xcite pointed out that small sources lead to a dependence of the variance on molecular diffusivity .
these models were further refined by @xcite . chertkov _
et al _ and balkovsky & fouxon @xcite addressed the case of a random , statistically - steady source .
in this paper we study the enhancement of mixing by an advection field using a particle - based computational scheme that is easy to implement and applicable to a variety of source distributions .
the idea is to develop a method that accurately simulates advection and diffusion of large numbers of particles supplied by a steady source , and to measure density fluctuations by `` binning '' the particles to produce an approximation of the hydrodynamic concentration field . unlike a numerical pde code
, a particle code does not prefer specific forms of the flow or the source ( pde methods generally work best with very smooth fields ) .
there is , however , no free lunch : the accuracy of the particle code is ultimately limited by the finite number of particles that can be tracked .
the limitation to finite numbers of particles inevitably introduces statistical errors due to discrete fluctuations in the local density and systematic errors in the variance measurements due to binning .
but these problems are tractable , and as we will show , the method proves to be quantitatively accurate and computationally efficient for some applications .
in this section we review basic facts about the mixing enhancement problem as formulated by thiffeault , doering & gibbon _ et al _ @xcite and developed by plasting & young @xcite , doering & thiffeault @xcite , shaw _
et al _ @xcite , and thiffeault and pavliotis @xcite . the dynamics is given by the advection - diffusion equation for the concentration of a passive scalar @xmath1 with time - independent but spatially inhomogeneous source field @xmath2 : @xmath3 where @xmath4 is the molecular diffusivity and @xmath5 is a specified advection field that satisfies ( at each instant of time ) the incompressibility condition @xmath6 for simplicity , the domain is the @xmath7-torus , i.e. , @xmath8^d$ ] with periodic boundary conditions .
we limit attention to stirring fields that satisfy the properties of statistical homogeneity and isotropy in space defined by @xmath9 where the overbar denotes time - averaging and @xmath10 is the root mean square speed of the velocity field , a natural indicator of the intensity of the stirring .
these are statistical properties of homogeneous isotropic turbulence on the torus , but they are also shared by many other kinds of flows .
we are interested in fluctuations in the concentration @xmath11 so the spatially averaged background density is irrelevant .
it is easy to see from that the spatial average of @xmath11 grows linearly with time at the rate given by the spatial average of @xmath12 .
hence we change variables to spatially mean - zero quantities @xmath13 and @xmath14 that satisfy @xmath15 ( we must also supply initial conditions for @xmath11 and/or @xmath16 but they play no role in the long - time steady statistics that we are interested in . ) the `` mixedness '' of the scalar may be characterized by , among other quantities , the long - time averaged variance of @xmath11 , proportional to the long - time averaged @xmath17 norm of @xmath16 , @xmath18 the smaller @xmath19 is , the more uniform the distribution .
the `` mixing enhancement '' of a stirring field is naturally measured by comparing the scalar variance to the variance with the same source but in the absence of stirring . to be precise ,
we compare @xmath19 to @xmath20 where @xmath21 is the solution to @xmath22 ( with , say , the same initial data although these will not affect the long - time averaged fluctuations ) .
the dimensionless _ mixing enhancement factor _ is then defined as @xmath23 this quantity carries the subscript @xmath24 because we can also define _ multiscale mixing enhancements _ @xcite by weighting large / small wavenumber components of the scalar fluctuations : @xmath25 as discussed in doering & thiffeault @xcite , shaw _
et al _ @xcite and shaw @xcite , @xmath26 provide a gauge of the mixing enhancement of the flow as measured by scalar fluctuations on relatively small and large length scales , respectively .
we refer to them as enhancement factors because if one were to define an effective , eddy , or equivalent diffusivity @xmath27 as the value of a molecular diffusion necessary to produce the same value of @xmath28 with stirring , then @xmath29 . in this paper , however , we will focus exclusively on @xmath30 , the mixing enhancement at `` moderate '' length scales .
there is a theoretical upper bound on @xmath30 valid for any statistically stationary homogeneous and isotropic stirring field @xcite : @xmath31 where @xmath32 are the fourier coefficients of the source and the pclet number , @xmath33 is a dimensionless measure of the intensity of the stirring .
generally , we anticipate that @xmath30 is an increasing function of pe and the estimate in guarantees that @xmath34 as @xmath35 , the `` classical '' scaling necessary if there is to be any residual variance suppression in the singular vanishing diffusion limit .
that is , if @xmath36 then @xmath37 has a nonzero limit as @xmath38 with all other parameters held fixed .
it is natural to refer to any of the possible sub - classical scalings as `` anomalous '' .
the upper limit to the mixing enhancement in depends on the stirring field only through @xmath10 via pe , but it depends on all the details of the source distribution . as studied in depth in references @xcite , the structure of the scalar source can have a profound effect on the high pe scaling of @xmath30 , notably for sources with small scales .
it is physically meaningful to consider measure - valued source - sink distributions , like delta - functions , with arbitrarily small scales .
it is precisely this source size dependence of @xmath39 that motivates the development of a computational method that can handle singular source distributions . in this study , for computational simplicity and efficiency
, we utilize the `` random sine flow '' as the stirring field . in the two - dimensional case
this is defined for all time by @xmath40 where @xmath41 is the period , @xmath42 , and @xmath43 and @xmath44 are random phases chosen independently and uniformly on @xmath45 in each half cycle , which assures the homogeneity of the flow field . in this case ,
@xmath46 . in the three - dimensional case ,
we employ @xmath47\hat{\textbf{\i}}\,,\ & nt < t \leq nt+ \tfrac13 t ; \nonumber \\
w\left[\sin \alpha \sin\left(2 \pi
z / l+\psi_{3}\right)+\cos \alpha \sin\left(2 \pi x / l+\phi_{1}\right ) \right]\hat{\textbf{\j}}\,,\ & nt+\tfrac13 t < t
\leq nt+\tfrac23 t ; \nonumber \\ w\left[\sin \alpha \sin\left(2 \pi x / l+\psi_{1}\right)+\cos \alpha \sin\left(2 \pi y / l+\psi_{2}\right ) \right]\hat{\textbf{k}}\,,\ & nt+\tfrac23 t < t \leq ( n+1)t , \nonumber \end{cases}\ ] ] where again @xmath46 , @xmath48 , and @xmath49 , @xmath50 and @xmath51 are uniform random numbers in @xmath52 chosen independently every @xmath53 .
the angle @xmath49 randomizes the shear direction to guarantee isotropy of the flow .
in a particle code for solving the advection - diffusion equation , the concentration field @xmath11 is represented by a distribution of particles .
particles are introduced by generating random locations using the properly normalized source @xmath2 as a probability distribution function , then they are transported by advection and diffusion .
the particle density , @xmath1 , is measured by covering the domain with bins counting the number of particles per bin .
a discrete particle method is employed because it can easily deal with small - scale sources such as @xmath54 functions .
it is also straightforward to implement with any advection field .
the downside of a particle method is that it necessarily involves two kinds of errors : the number density of particles calculated by dividing the domain into bins is only resolved down to the size of the bins , and
the measurement of @xmath11 always includes statistical errors due to the use of finite numbers of particles . at each time
step the system is evolved by advection , diffusion , the source , and sinks .
an advection - only equation would be solved by moving particles along characteristics , and a diffusion - only equation would be solved by adding independent gaussian noises to each coordinate of each particle . with both advection and diffusion
we need to solve a stochastic differential equation to determine the proper displacement of the particles during a time step .
the stochastic differential equation is @xmath55 where @xmath56 is a standard vector - valued wiener process . in order to solve
, we will consider cases where the displacement due to the noise in a subinterval of length @xmath57 ( where @xmath7 is the dimension ) is much smaller than the wavelength of the random sine flow .
this condition is realized better and better as pe increases .
then , during each subinterval , the drift field @xmath58 experienced by each particle can be approximated by a steady flow with a linear shear . in 2d , for the first half of the period for a particle starting at @xmath59 we approximate by @xmath60 and for the second half of the period , starting from @xmath61 , @xmath62 therefore , during the first half period we evolve the position of a particle through a time interval @xmath63 ( where @xmath64 need _ not _ be small ) by the map @xmath65 where @xmath66 and @xmath67 satisfy @xmath68 the variance - covariance matrix of @xmath66 and @xmath67 is @xmath69 which is realized by @xmath70 where @xmath71 and @xmath72 are independent @xmath73 random variables ( normally distributed with mean @xmath24 and standard deviation @xmath74 ) .
the matrix describes the evolution of a passive scalar field in a shear flow @xcite .
therefore the time evolution map during the first half period is @xmath75 a similar map is employed during the second half of the period .
these stochastic maps include the shear in the approximation that the shear remains constant for each particle during each half cycle that causes a `` distortion '' of a gaussian cloud of particles ; see fig .
[ fig : gauss ] .
the same calculations apply for the three - dimensional case : for the first subinterval , the time - evolution map is @xmath76\nonumber\\ & \phantom{\rightarrow x_0}+\sqrt{\tfrac{1}{6}(s_{2}^{\ 2}+s_{3}^{\ 2})\kappa ( \delta t)^{3}+2\kappa\delta t } \
, n_{1 } + s_{2}\sqrt{\tfrac12\kappa}(\delta t ) ^{\frac{3}{2 } } \ , n_{2 } + s_{3}\sqrt{\tfrac12\kappa}(\delta t ) ^{\frac{3}{2 } } \ , n_{3},\\ y_0 & \rightarrow y_0+\sqrt{2 \kappa \delta t } \ , n_{2},\\ z_0 & \rightarrow z_0+\sqrt{2 \kappa \delta t } \ , n_{3},\end{aligned}\ ] ] where @xmath77 , @xmath78 , and @xmath79 and @xmath80 are independent @xmath73 random variables .
the maps for the other subintervals can be obtained by cyclic permutation of the coordinates . the steady scalar source
is realized by introducing a new particle one by one using normalized @xmath2 as a probability distribution function .
numerically , such a probability distribution function can be realized by mapping uniform random numbers over @xmath81 $ ] with the inverse of the cumulative probability distribution function in question .
new particles are added constantly so the total number of particles continues to increase , which slows down the computation . to cope with increasing particles
, we implement a particle subtraction scheme .
particles eventually get well mixed and `` older '' particles do not contribute to the value of the hydrodynamic variance .
there is no added value in keeping track of particles that have been in the mix for a very long time , and we can simply remove them from the system after a sufficiently long time .
it is very important to keep track of the `` age '' of each particle , however , and to only remove sufficiently old well - mixed particles .
( for example , if a random fraction of particles is removed at regular time intervals , then the simulation becomes one of a system particles with a random finite lifetime , described by an advection - diffusion equation with an additional density decay term . ) in order to determine how old particles must be in order to safely remove them without affecting the hydrodynamic variance , prior to a full simulation run a test is performed as follows . starting from an initial set of @xmath82 particles located in space according to the source distribution , the flow and diffusion are allowed to act and the variance of the number of particles per bin , which decays with time , is monitored .
the number @xmath82 is of the order of the number of particles that are introduced in the full simulation during , say , an interval of length @xmath41 characteristic of the random sine flow .
the variance does not decay all the way to zero , however , but rather to the variance expected when @xmath82 particles are randomly distributed among the bins .
the time when the variance achieves this random - distribution variance , measured beforehand for a given flow and diffusion strength , is then the required `` aging '' time before particles can be safely removed in the full simulation with the steady source .
such a trial run is performed for each flow , diffusion strength , source distribution and particle number because this `` mixing time '' depends on all these factors .
futher details of the criteria for removing old particles and extensive tests and benchmark trials may be found in ref .
@xcite .
the variance @xmath83 is measured by monitoring the fluctuations in the number of particles per bin , and time - averaging . in @xmath7 dimensions
the domain is divided into @xmath84 bins and the code calculates @xmath85 , where @xmath86 is the number of particles in a bin . then @xmath83 is initially approximated by @xmath87 we say `` initially '' because the expression above includes both the hydrodynamic fluctuations of interest _ and _ discreteness fluctuations resulting solely from the fact that each bin contains a finite number of particles .
the subtraction scheme eliminates the `` well - mixed '' particles that do not contribute to the value of the hydrodynamic variance .
but even if the system were completely mixed so that theoretically , @xmath88 , the measured variance @xmath89 would be ( very close to , for small bins ) @xmath90 , which is on the order of @xmath91 , where @xmath92 is the total number of particles in the domain .
this follows from the fact that @xmath93 is represented in this particle method by only a finite number of particles in each finite size bin .
that is , @xmath19 as defined by ( [ measure ] ) is nonzero even when the particles are uniformly distributed : then the bulk variance includes fluctuations as if @xmath92 particles were randomly thrown in @xmath84 bins .
the helpful fact is that the bulk variance contribution from these _ background fluctuations _ due to finite numbers of particles in the bins
does not depend on ( i.e. , is uncorrelated with ) the hydrodynamic density variation from bin to bin .
the total contribution to the variance is the sum of the `` extra '' variance in each bin which is linear in the ( mean ) number of particles in each bin .
hence the sum of the variances is @xmath94 and the bulk variance contribution from the background fluctuations , @xmath95 , can simply be subtracted from the initial estimate for @xmath19 in ( [ measure ] ) .
the net result is our measured value of the hydrodynamic variance .
in addition to the inevitable fluctuations due to discreteness , density variations are observed only down to the length scales @xmath96 because of the binning density , which is another source of error in this procedure .
we use @xmath97 , which tests and benchmark studies indicate is sufficient for the examples studies here @xcite .
the variance is calculated once for each subinterval , and the instant when it is calculated is determined randomly in order to obtain an unbiased time average .
thus each subinterval is divided into two parts , before and after variance calculation , and the particle transport and source processes are appropriately adapted .
the final measured quantities are long time averages that are observed to be converged to within the error indicated on the plots below .
in order to investigate the effect of source - sink scales on maximal and actual mixing enhancements , we performed a series of simulations for square - shaped sources of various sizes @xmath98 as illustrated in fig .
[ fig : sqfig ] .
fig [ fig:2dbounds ] shows the upper bounds on @xmath30 for square sources and a @xmath54-function source in 2d computed from ( [ bound ] ) .
the upper bound for any finite - size source is asymptotically @xmath99 pe , but for the @xmath54-function source it is @xmath100 in the large pe limit . in 3d ,
the distinction between cubic sources and a @xmath54-function source is more apparent as shown in fig [ fig:3dbounds ] : the upper bound for a @xmath54-function source behaves @xmath101 in 3d .
we stress that these mixing enhancement bounds apply for _ any _ statistically homogeneous and isotropic flows stirring sources with these shapes .
simulation results for the random sine flow , shown in fig .
[ fig:2dmixing ] for 2d and fig .
[ fig:3dmixing ] for 3d qualitatively confirm the behavior of the enhancements suggested by the upper limits . as the source size shrinks , the measured mixing enhancement gets smaller in a way that is remarkably similar to the bounds . in these simulations
pe is varied by decreasing @xmath4 at a fixed values of @xmath102 and @xmath41 .
other values of @xmath41 and other ( shorter ) wavelengths of the stirring flow were also checked , producing similar plots .
these 2d simulation results have recently been confirmed quantitatively by a pde computation @xcite .
the simulations also show that the upper estimates can give the correct _ quantitative _ behavior of @xmath30 as a function of pe . indeed ,
in fig [ fig:3ddelta ] we plot the upper bound on @xmath30 for the @xmath54-function source in 3d and the measured enhancement from the simulations .
the upper bound , which scales anomalously @xmath101 , is an excellent predictor of the data . from this
we conclude that the random sine flow is an `` almost - optimal '' mixer ( among statistically homogeneous and isotropic flows ) for this source - sink distribution .
-function source .
the solid line is the upper bound for any shi flow and the data are mixing enhancements for the random sine flow measured in the discrete particle simulations.,height=226 ]
we have devised an accurate and computationally efficient particle method to study hydrodynamic variance suppression by a mixing flow .
rigorous upper bounds for the mixing enhancement @xmath30 , i.e. , the effective diffusion enhancement factor , were compared to measured enhancements for the simple random sine flow . a key prediction of the analysis in refs .
@xcite is that the source - sink shape is a determining factor in the mixing enhancement of any flow .
the simulation results reported here show that the upper estimates give the correct qualitative picture as regards the pe and source - shape dependence of @xmath30 .
future work should focus on investigating enhancements of other stirring flows .
no attempt has been made here to find a more efficient stirring flow ( or , indeed , the _ most _ efficient flow , if there is one ) whose enhancement approaches more closely ( or perhaps even saturates ) the upper bound .
it is remarkable that the simple random sine flow appears to saturate the upper bound scaling @xmath103 in 3d .
stirring with appropriate turbulent solutions to the navier stokes equation is also of significant interest .
the central question here is , is statistically homogeneous and isotropic turbulence generically an efficient mixer ?
the answer may depend on the source - sink distribution .
the authors thank jai sukhatme for helpful comments on the paper . this work was supported in part by us national science foundation through awards phy-0555324 and dms-0553487 , by the geophysical fluid dynamics program at woods hole oceanographic institution , and by the alexander von humboldt foundation .
d. j. thomson , `` a stochastic model for the motion of particle pairs in isotropic high - reynolds number turbulence , and its application to the problem of concentration variance , '' _ j. fluid mech . _ * 210 * , 113153 ( 1990 ) .
m. chertkov , g. falkovich , i. kolokolov , and v. lebedev , `` statistics of a passive scalar advected by a large - scale two - dimensional velocity field : analytic solution , '' _ phys .
e _ * 51 * ( 6 ) , 56095627 ( 1995a ) .
m. chertkov , g. falkovich , i. kolokolov , and v. lebedev , `` normal and anomalous scaling of the fourth - order correlation function of a randomly advected passive scalar , '' _ phys .
e _ * 52 * ( 5 ) , 49244941 ( 1995b ) .
e. balkovsky and a. fouxon , `` universal long - time properties of lagrangian statistics in the batchelor regime and their application to the passive scalar problem , '' _ phys .
e _ * 60 * ( 4 ) , 41644174 ( 1999 ) . t. a. shaw , `` bounds on multiscale mixing efficiencies , '' in _ proceedings of the 2005 summer program in geophysical fluid dynamics _ , 291 ( woods hole oceanographic institute , woods hole , ma , 2005 ) , ` http://www.whoi.edu/cms/files/shaw_21281.pdf ` .
t. okabe , `` a particle - simulation method to study mixing efficiencies , '' in _ proceedings of the 2006 summer program in geophysical fluid dynamics _ ( woods hole oceanographic institute , woods hole , ma , 2006 ) , ` http://www.whoi.edu/cms/files/taka_21241.pdf ` . | the mixing effectiveness , i.e. , the enhancement of molecular diffusion , of a flow can be quantified in terms of the suppression of concentration variance of a passive scalar sustained by steady sources and sinks . the mixing enhancement defined this way is the ratio of the rms fluctuations of the scalar mixed by molecular diffusion alone to the ( statistically steady - state ) rms fluctuations of the scalar density in the presence of stirring .
this measure of the effectiveness of the stirring is naturally related to the enhancement factor of the equivalent eddy diffusivity over molecular diffusion , and depends on the pclet number .
it was recently noted that the maximum possible mixing enhancement at a given pclet number depends as well on the structure of the sources and sinks .
that is , the mixing efficiency , the effective diffusivity , or the eddy diffusion of a flow generally depends on the sources and sinks of whatever is being stirred . here
we present the results of particle - based simulations quantitatively confirming the source - sink dependence of the mixing enhancement as a function of pclet number for a model flow . |
Subsets and Splits