1: % ****** Start of file template.aps ****** %

2: %%

3: %%

4: %%   This file is part of the APS files in the REvTeX 4 distribution.

5: %%   version 4.0 of REvTeX, August 2001

6: %%

7: %%

8: %%   Copyright (c) 2001 The American Physical Society.

9: %%

10: %%   See the REvTeX 4 README file for restrictions and more information.

11: %%

12: %

13: % This is a template for producing manuscripts for use with REvTEX 4.0

14: % Copy this file to another name and then work on that file.

15: % That way, you always have this original template file to use.

16: %

17: % Group addresses by affiliation; use superscriptaddress for long

18: % author lists, or if there are many overlapping affiliations.

19: % For Phys. Rev. appearance, change preprint to twocolumn.

20: % Choose pra, prb, prc, prd, pre, prl, prstab, or rmp for journal

21: %  Add 'draft' option to mark overfull boxes with black boxes

22: %  Add 'showpacs' option to make PACS codes appear

23: %  Add 'showkeys' option to make keywords appear

24: %\documentclass[aps,prl,preprint,groupedaddress]{revtex4}

25: %\documentclass[aps,prl,preprint,superscriptaddress]{revtex4}

26:     \documentclass[aps,pre,groupedaddress,amsmath,eqsecnum,showpacs]{revtex4}

27:     %\documentclass[aps,prb,groupedaddress,amsmath,preprint,eqsecnum,showkeys]{revtex4}

28:     \usepackage{bm}

29:     \usepackage{graphicx}

30:    %\usepackage{showkeys}

31: % You should use BibTeX and apsrev.bst for references

32: % Choosing a journal automatically selects the correct APS

33: % BibTeX style file (bst file), so only uncomment the line

34: % below if necessary.

35: %\bibliographystyle{apsrev}

36: \newcommand\beq{\begin{equation}}

37: \newcommand\eeq{\end{equation}}

38: \newcommand\beqa{\begin{eqnarray}}

39: \newcommand\eeqa{\end{eqnarray}}

40: \newcommand{\dd}{\text{d}}

41: \newcommand{\nn}{\nonumber\\}

42: 

43: \newcommand{\qq}{r}

44: 

45: \newcommand{\al}{\alpha}

46: \newcommand{\at}{\frac{1+\alpha}{2}}

47: \newcommand{\atp}{\left(\frac{1+\alpha}{2}\right)}

48: \newcommand{\sg}{(\widehat{\bm{\sigma}}\cdot\mathbf{g})}

49: \newcommand{\si}{\widehat{\sigma}}

50: \newcommand{\nutz}{\nu_{2|0}}

51: \newcommand{\nuzt}{\nu_{0|2}}

52: \newcommand{\nuto}{\nu_{2|1}}

53: \newcommand{\nuzth}{\nu_{0|3}}

54: \newcommand{\nufz}{\nu_{4|0}}

55: \newcommand{\nutt}{\nu_{2|2}}

56: \newcommand{\nuzf}{\nu_{0|4}}

57: \newcommand{\omegatz}{\omega_{2|0}}

58: \newcommand{\omegazt}{\omega_{0|2}}

59: \newcommand{\omegato}{\omega_{2|1}}

60: \newcommand{\omegazth}{\omega_{0|3}}

61: \newcommand{\omegafz}{\omega_{4|0}}

62: \newcommand{\omegatt}{\omega_{2|2}}

63: \newcommand{\omegazf}{\omega_{0|4}}

64: \newcommand{\nubar}{\omega_{0|2}}

65: \newcommand{\omeganu}{\nu}

66: 

67: 

68: 

69: 

70: 

71: \begin{document}

72: % Use the \preprint command to place your local institutional report

73: % number in the upper righthand corner of the title page in preprint mode.

74: % Multiple \preprint commands are allowed.

75: % Use the 'preprintnumbers' class option to override journal defaults

76: % to display numbers if necessary

77: %\preprint{}

78: %Title of paper

79: \title{Third and fourth degree collisional moments for inelastic Maxwell models}

80: % repeat the \author .. \affiliation  etc. as needed

81: % \email, \thanks, \homepage, \altaffiliation all apply to the current

82: % author. Explanatory text should go in the []'s, actual e-mail

83: % address or url should go in the {}'s for \email and \homepage.

84: % Please use the appropriate macro foreach each type of information

85: % \affiliation command applies to all authors since the last

86: % \affiliation command. The \affiliation command should follow the

87: % other information

88: % \affiliation can be followed by \email, \homepage, \thanks as well.

89: \author{Vicente Garz\'o}

90: \email{vicenteg@unex.es}

91: \homepage{http://www.unex.es/fisteor/vicente/}

92: \author{Andr\'es Santos}

93: \email{andres@unex.es} \homepage{http://www.unex.es/fisteor/andres/}

94: \affiliation{Departamento de F\'{\i}sica, Universidad de

95: Extremadura, E-06071 Badajoz, Spain}

96: %\homepage[]{Your web page}

97: %\thanks{}

98: %\altaffiliation{}

99: %Collaboration name if desired (requires use of superscriptaddress

100: %option in \documentclass). \noaffiliation is required (may also be

101: %used with the \author command).

102: %\collaboration can be followed by \email, \homepage, \thanks as well.

103: %\collaboration{}

104: %\noaffiliation

105: \date{\today}

106: \begin{abstract}

107: The third and fourth degree collisional moments for $d$-dimensional

108: inelastic Maxwell models are exactly evaluated in terms of the

109: velocity moments, with explicit expressions for the associated

110: eigenvalues and cross coefficients as functions of the coefficient

111: of normal restitution. The results are applied to the analysis of

112: the time evolution of the moments (scaled with the thermal speed) in

113: the free cooling problem. It is observed that the characteristic

114: relaxation time toward the homogeneous cooling state decreases as

115: the anisotropy of the corresponding moment increases. In particular,

116: in contrast to what happens in the one-dimensional case, all the

117: anisotropic moments of degree equal to or less than four vanish in

118: the homogeneous cooling state for $d\geq 2$.

119: \end{abstract}

120: % insert suggested PACS numbers in braces on next line

121: \pacs{05.20.Dd, 45.70.-n, 51.10.+y}

122: % insert suggested keywords - APS authors don't need to do this

123:  %\keywords{Inelastic Maxwell models; Boltzmann equation; Collisional moments; Homogeneous cooling state}

124: 

125: %\maketitle must follow title, authors, abstract, \pacs, and \keywords

126: \maketitle

127: % body of paper here - Use proper section commands

128: % References should be done using the \cite, \ref, and \label commands

129: %\section{Introduction\label{sec1}}

130: %\section{Introduction\label{sec1}}

131: % Put \label in argument of \section for cross-referencing

132: %\section{\label{}}

133: %\subsection{}

134: %\subsubsection{}

135: 

136: \maketitle

137: 

138: 

139: \section{introduction}

140: A realistic model capturing the influence of dissipation on the dynamic properties of granular systems consists

141: of a gas of inelastic hard spheres (IHS) with a constant coefficient of normal restitution $\alpha \leq 1$

142: \cite{BP04}. For  sufficiently low densities, the Boltzmann equation for IHS provides the adequate framework to

143: describe the time evolution of the one-particle velocity distribution function $f(\mathbf{r},\mathbf{v};t)$

144: \cite{BDS97}. However, the intricacy of the Boltzmann collision operator for IHS makes it difficult to obtain

145: exact results. For instance, the fourth cumulant $a_2$ of the velocity distribution in the so-called homogeneous

146: cooling state (HCS) is not exactly known, although good estimates of it have been proposed

147: \cite{NE98,MS00,BP06}. For inhomogeneous situations, explicit expressions for the Navier--Stokes (NS) transport

148: coefficients are approximately obtained by considering the leading terms in a Sonine polynomial expansion

149: \cite{BDKS98,BC01,GD02,GSM07}.

150: 

151: As in the elastic case, part of the above difficulties can be overcome by considering the so-called Maxwell

152: models, i.e., models for which the collision rate is independent of the relative velocity of the two colliding

153: particles. Inelastic Maxwell models (IMM) have attracted the attention of physicists and mathematicians since

154: the beginning of the century

155: \cite{BCG00,CCG00,NK00,C01,EB02a,EB02,EB02bis,BMP02,KN02,KN02bis,NK02,BC02,MP02,MP02bis,BC03,BCT03,SE03,S03,

156: NK03,G03,BE04,GA05,BG06,ETB06,S07}. The structure of the Boltzmann collision operator for IMM has the advantage

157: of allowing for the derivation of a number of exact properties, such as the high-velocity tails

158: \cite{EB02a,EB02,EB02bis,BMP02,KN02bis,KN02} and the cumulants \cite{BMP02,EB02,MP02,MP02bis,S03,GA05} in

159: homogeneous situations, the NS transport coefficients \cite{S03,GA05}, and the rheology under simple shear flow

160: \cite{C01,G03}. As a consequence, it is possible to explore the influence of inelasticity on the dynamic

161: properties in a clean way, without the need of introducing additional, and sometimes uncontrolled,

162: approximations. Apart from their academic interest, it turns out that the IMM  reliably describe the properties

163: of IHS in some situations, as happens in the simple shear flow problem \cite{G03} and for the NS transport

164: coefficients associated with the mass flux \cite{GA05}. Furthermore, it is interesting to remark that recent

165: experiments \cite{KSSAON05} for magnetic grains with dipolar interactions are well described by IMM.

166: 

167: The aim of this paper is to contribute to the advancement in the knowledge of exact properties of IMM by

168: evaluating all the third and fourth degree moments of the Boltzmann  collision operator for an arbitrary number

169: of dimensions $d$. The knowledge of those collisional moments, along with that of the second degree collisional

170: moments \cite{S03,G03}, opens up a number of interesting applications. For instance, one can investigate the

171: temporal relaxation toward the HCS, starting from arbitrary initial conditions (not necessarily isotropic), as

172: measured by the lowest degree moments (namely, the fourth degree moments) which signal the non-Gaussian

173: character of the asymptotic velocity distribution function. This issue will be covered in this paper.

174: 

175: This paper is organized as follows. In Section \ref{sec2} the Boltzmann equation for IMM is presented. Next, the

176: Ikenberry polynomials \cite{TM80} $Y_{2r|i_1i_2\ldots i_s}(\mathbf{V})$ of degree $k=2r+s$ are introduced and

177: their associated collisional moments $J_{2r|i_1i_2\ldots i_s}$ for $k=3$ and 4 are evaluated, the technicalities

178: being relegated to an Appendix. The results are applied to the relaxation problem of the (scaled) moments toward

179: their asymptotic values in the HCS in Section \ref{sec3}. The paper is closed in Section \ref{sec4} with a brief

180: discussion of the results obtained here.

181: 

182: \section{Collisional moments for IMM\label{sec2}}

183: 

184: In the absence of external forces, the inelastic Boltzmann equation for a granular gas reads \cite{BP04}

185:  \beq

186: \left(\partial_t +\mathbf{v}\cdot\nabla\right)f(\mathbf{r},\mathbf{v};t)=J[\mathbf{v}|f,f], \label{2.1}

187: \eeq

188: where $J[\mathbf{v}|f,f]$ is the Boltzmann collision operator. The form of the operator $J$ for IMM can be

189: obtained from the form for IHS by replacing the IHS collision rate (which is proportional to the relative

190: velocity of the two colliding particles) by an effective velocity-independent collision rate. With this

191: simplification, the form of $J$ becomes \cite{NK03}

192: \begin{equation}

193: J\left[{\bf v}_{1}|f,f\right] =\frac{\omeganu}{n\Omega_d} \int

194: \dd{\bf v}_{2}\int \dd\widehat{\boldsymbol{\sigma}} \left[

195: \alpha^{-1}f({\bf v}_{1}')f({\bf v}_{2}')-f({\bf v}_{1})f({\bf

196: v}_{2})\right] .

197: \label{1}

198: \end{equation}

199: Here,

200: \beq

201: n=\int\dd \mathbf{v} f(\mathbf{v})

202: \eeq

203: is the number density, $\omeganu$ is an effective collision

204: frequency (to be chosen later), $\Omega_d=2\pi^{d/2}/\Gamma(d/2)$ is

205: the total solid angle in $d$ dimensions, and $\alpha\leq 1$ refers

206: to the constant coefficient of restitution. In addition, the primes

207: on the velocities denote the initial values $\{{\bf v}_{1}^{\prime},

208: {\bf v}_{2}^{\prime}\}$ that lead to $\{{\bf v}_{1},{\bf v}_{2}\}$

209: following a binary collision:

210: \begin{equation}

211: \label{3}

212: {\bf v}_{1}^{\prime}={\bf v}_{1}-\frac{1}{2}\left( 1+\alpha

213: ^{-1}\right)(\widehat{\boldsymbol{\sigma}}\cdot {\bf

214: g})\widehat{\boldsymbol {\sigma}}, \quad {\bf v}_{2}^{\prime}={\bf

215: v}_{2}+\frac{1}{2}\left( 1+\alpha^{-1}\right)

216: (\widehat{\boldsymbol{\sigma}}\cdot {\bf

217: g})\widehat{\boldsymbol{\sigma}}\;,

218: \end{equation}

219: where ${\bf g}={\bf v}_1-{\bf v}_2$ is the relative velocity of the colliding pair and

220: $\widehat{\boldsymbol{\sigma}}$ is a unit vector directed along the centers of the two colliding spheres. The

221: collision frequency $\omeganu$ can be seen as a free parameter in the model. Its dependence on the coefficient

222: of restitution $\alpha$ can be chosen to optimize the agreement with  the results obtained from the Boltzmann

223: equation  for IHS. In particular, to get the same expression for the cooling rate as the one found for IHS

224: (evaluated in the local equilibrium approximation) one takes the choice \cite{S03}

225: \begin{equation}

226: \label{4}

227: \omeganu=\frac{d+2}{2}\nu_0, \quad

228: \nu_0=\frac{4\Omega_d}{\sqrt{\pi}(d+2)}n\sigma^{d-1}\sqrt{\frac{T}{m}},

229: \end{equation}

230: where $\sigma$ is the diameter of the spheres. Note that, in any

231: case, the results derived in this paper will be independent of the

232: specific choice of $\nu_0$.

233: 

234: 

235: A useful identity for an arbitrary function $h({\bf v})$ is given by

236: \begin{equation}

237: \label{5}

238: \mathcal{J}[h]\equiv \int \dd{\bf v}_1 h({\bf v}_1)J[{\bf

239: v}_1|f,f]=\frac{\omeganu}{n\Omega_d} \int \dd{\bf v}_{1}\,\int

240: \dd{\bf v}_{2}f({\bf v}_{1})f({\bf v}_{2}) \int

241: \dd\widehat{\boldsymbol{\sigma}}\,\left[h({\bf v}_1'')-h({\bf

242: v}_1)\right],

243: \end{equation}

244: where

245: \begin{equation}

246: \label{6}

247: {\bf v}_{1}''={\bf v}_{1}-\frac{1}{2}(1+\alpha)(

248: \widehat{\boldsymbol{\sigma }}\cdot {\bf

249: g})\widehat{\boldsymbol{\sigma}}

250: \end{equation}

251: denotes the post-collisional velocity. If $h(\mathbf{v})$ is a

252: polynomial, then

253: \beq

254: \mathcal{M}[h]\equiv \int \dd\mathbf{v} h(\mathbf{v}) f(\mathbf{v})

255: \label{2.2}

256: \eeq

257: is its associated velocity moment and $\mathcal{J}[h]$ is the

258: corresponding collisional moment.

259: 

260: In the case of Maxwell models (both elastic and inelastic), it is convenient to introduce the Ikenberry

261: polynomials \cite{TM80} $Y_{2r|i_1i_2\ldots i_s}(\mathbf{V})$ of degree $k=2r+s$, where

262: $\mathbf{V}=\mathbf{v}-\mathbf{u}(\mathbf{r})$ is the peculiar velocity, $\mathbf{u}(\mathbf{r})$ being the mean

263: flow velocity defined as

264: \beq \mathbf{u}=\frac{1}{n}\int\dd \mathbf{v} \mathbf{v}f(\mathbf{v}). \eeq

265: The

266: Ikenberry polynomials are defined as $Y_{2r|i_1i_2\ldots i_s}(\mathbf{V})=V^{2r}Y_{i_1i_2\ldots

267: i_s}(\mathbf{V})$, where $Y_{i_1i_2\ldots i_s}(\mathbf{V})$ is obtained by subtracting from

268: $V_{i_1}V_{i_2}\ldots V_{i_s}$ that homogeneous symmetric polynomial of degree $s$ in the components of

269: $\mathbf{V}$ such as to annul the result of contracting the components of $Y_{i_1i_2\ldots i_s}(\mathbf{V})$ on

270: any pair of indices. The polynomials functions $Y_{2r|i_1i_2\ldots i_s}(\mathbf{V})$ of degree smaller than or

271: equal to four are \beq Y_{0|0}(\mathbf{V})=1,\quad Y_{0|i}(\mathbf{V})=V_i, \label{X0} \eeq \beq

272: Y_{2|0}(\mathbf{V})=V^2,\quad Y_{0|ij}(\mathbf{V})=V_iV_j-\frac{1}{d}V^2\delta_{ij}, \label{X1} \eeq \beq

273: Y_{2|i}(\mathbf{V})=V^2 V_i,\quad

274: Y_{0|ijk}(\mathbf{V})=V_iV_jV_k-\frac{1}{d+2}V^2\left(V_i\delta_{jk}+V_j\delta_{ik}+V_k\delta_{ij}\right),

275: \label{X2} \eeq \beq Y_{4|0}(\mathbf{V})=V^4,\quad

276: Y_{2|ij}(\mathbf{V})=V^2\left(V_iV_j-\frac{1}{d}V^2\delta_{ij}\right), \label{X3} \eeq \beqa

277: Y_{0|ijk\ell}(\mathbf{V})&=&V_iV_jV_kV_\ell-\frac{1}{d+4}V^2

278: \left(V_iV_j\delta_{k\ell}+V_iV_k\delta_{j\ell}+V_iV_\ell\delta_{jk}

279: +V_jV_k\delta_{i\ell}+V_jV_\ell\delta_{ik}+V_kV_\ell\delta_{ij}\right)\nn

280: &&+\frac{1}{(d+2)(d+4)}V^4\left(\delta_{ij}\delta_{k\ell}+\delta_{ik}\delta_{j\ell}+\delta_{i\ell}\delta_{jk}\right)\nn

281: &=&V_iV_jV_kV_\ell-\frac{1}{d+4}\left[Y_{2|ij}(\mathbf{V})\delta_{k\ell}+Y_{2|ik}(\mathbf{V})\delta_{j\ell}

282: +Y_{2|i\ell}(\mathbf{V})\delta_{jk} +Y_{2|jk}(\mathbf{V})\delta_{i\ell}\right.\nn &&\left.

283: +Y_{2|j\ell}(\mathbf{V})\delta_{ik}

284: +Y_{2|k\ell}(\mathbf{V})\delta_{ij}\right]-\frac{1}{d(d+2)}V^4\left(\delta_{ij}\delta_{k\ell}+

285: \delta_{ik}\delta_{j\ell}+\delta_{i\ell}\delta_{jk}\right). \label{X4} \eeqa

286: 

287: 

288: Here we will use the notation $M_{2r|i_1i_2\ldots

289: i_s}=\mathcal{M}[Y_{2r|i_1i_2\ldots i_s}]$ and $J_{2r|i_1i_2\ldots

290: i_s}=\mathcal{J}[Y_{2r|i_1i_2\ldots i_s}]$ for the associated

291: moments and collisional moments, respectively. Note that

292: $M_{0|0}=n$, $J_{0|0}=0$ (conservation of mass), $M_{0|i}=0$ (by

293: definition of the peculiar velocity), $J_{0|i}=0$ (conservation of

294: momentum), and $M_{2|0}=pd/m$, where $p=nT$ is the hydrostatic

295: pressure, $T$ being the granular temperature. Moreover,

296: $M_{0|ij}=(P_{ij}-p\delta_{ij})/m$, where $P_{ij}$ is the pressure

297: tensor and $M_{2|i}=2q_i/m$, where $\mathbf{q}$ is the heat flux

298: vector. The moment $M_{2|0}$,  the number density $n$, and the flow

299: velocity $\mathbf{u}$ are the hydrodynamic fields, while the moments

300: $M_{0|ij}$ and $M_{2|i}$ constitute the momentum and energy fluxes,

301: respectively. The remaining third degree moments $M_{0|ijk}$ and the

302: moments of a degree $k\geq 4$ are not directly related to the

303: hydrodynamic description, but they are useful to provide information

304: about the velocity distribution function. In particular, the moment

305: $M_{4|0}$ is related to the fourth cumulant $a_2$ as

306: \beq

307: a_2=\frac{m^2 }{d(d+2)nT^2}M_{4|0}-1,

308: \label{2.3}

309: \eeq

310: while the moments $M_{0|ijk}$, $M_{0|ijk\ell}$, and $M_{2|ij}$

311: measure the degree of anisotropy of the velocity distribution.

312: 

313: 

314: 

315: As in the elastic case, the mathematical structure of the collision

316: operator \eqref{1} implies that a collisional moment of degree $k$

317: can be expressed in terms of velocity moments of a degree less than

318: or equal to $k$. More specifically, the choice of the polynomials

319: $Y_{2r|\bar{s}}(\mathbf{V})$, where we have introduced the

320: short-hand notation $\bar{s}\equiv i_1i_2\ldots i_s$, yields the

321: following structural form for the collisional moments

322: $J_{2r|\bar{s}}$:

323: \beq

324: J_{2r|\bar{s}}=-\nu_{2r|s}

325: M_{2r|\bar{s}}+{\sum_{r',r'',\bar{s}',\bar{s}''}}^\dagger\lambda_{r'r''|\bar{s}'\bar{s}''\bar{s}}M_{2r'|\bar{s}'}M_{2r''|\bar{s}''},

326: \label{2.4}

327: \eeq

328: where the dagger in the summation denotes the constraints

329: $2(r'+r'')+s'+s''=2r+s$, $2r'+s'\geq 2$, and $2r''+s''\geq 2$. Since

330: the first term on the right-hand side of  Eq.\ \eqref{2.4} is

331: linear, then $\nu_{2r|s}$ represents the eigenvalue of the

332: linearized collision operator corresponding to the eigenfunction

333: $Y_{2r|\bar{s}}(\mathbf{V})$.

334: 

335: 

336: 

337: Let us now display the explicit expressions for the collisional

338: moments $J_{2r|i_1i_2\ldots i_s}$ for $k=2r+s\leq 4$. We start with

339: the second degree moments.

340: 

341: \subsection{Second degree collisional moments}

342: The second degree collisional moments were already evaluated in

343: Ref.\ \onlinecite{S03}. They are given by

344: \beq

345: J_{2|0}=-\nutz M_{2|0},\quad J_{0|ij}=-\nuzt M_{0|ij},

346: \label{X5}

347: \eeq

348: where the expressions for the eigenvalues $\nutz$ and $\nuzt$ are

349: \beq

350: \nutz=\frac{d+2}{4d}\left(1-\al^2\right)\nu_0,

351: \label{X6a}

352: \eeq

353: \beq

354: \nuzt=\frac{(1+\al)(d+1-\al)}{2d}\nu_0=\nutz+\frac{(1+\alpha)^2}{4}\nu_0.

355: \label{X6}

356: \eeq

357:  The quantity $\nutz$ is not but

358: the cooling rate, i.e., the rate of change of the granular

359: temperature due to the inelasticity of the collisions. The

360: eigenvalue   $\nuzt$ is the collision frequency associated with the

361: NS shear viscosity and  reduces to $\nu_0$ in the elastic limit. The

362: second equality in Eq.\ \eqref{X6} decomposes $\nuzt$ into the part

363: inherent to the collisional cooling plus the genuine part of the

364: momentum collisional transfer. As shown below, a similar

365: decomposition can be carried out for the eigenvalues $\nu_{2r|s}$.

366: 

367: 

368: \subsection{Third degree collisional moments}

369: The evaluation of the  third degree collisional moments $J_{2|i}$

370: and $J_{0|ijk}$ is performed in the Appendix. The results are

371: \beq J_{2|i}=-\nuto M_{2|i}, \quad J_{0|ijk}=-\nuzth M_{0|ijk},

372: \label{X7}

373: \eeq

374: where

375: \beq

376: \nuto=\frac{(1+\al)\left[5d+4-\al(d+8)\right]}{8d}\nu_0=\frac{3}{2}\nutz+\frac{(1+\al)^2(d-1)}{4d}\nu_0,

377: \label{X7b}

378: \eeq

379: \beq

380:  \nuzth=\frac{3}{2}\nuzt.

381: \label{X8}

382: \eeq

383: Equation \eqref{X7b} was first obtained in Ref.\ \onlinecite{S03}.

384: The eigenvalue $\nuto$ is the collision frequency associated with

385: the NS thermal conductivity. It is interesting to note that

386: $(\nuto-\frac{3}{2}\nutz)/(\nuzt-\nutz)=(d-1)/d$, which generalizes

387: the simple relationship, holding for elastic Maxwell models, between

388: the collision frequencies associated with the thermal conductivity

389: and the shear viscosity. An even simpler extension is provided by

390: Eq.\ \eqref{X8}.

391: 

392: 

393: 

394: \subsection{Fourth degree collisional moments}

395: The fourth degree collisional moments are also worked out  in the

396: Appendix. They can be written as

397: \beq

398: J_{4|0}=-\nufz M_{4|0}+\lambda_1 n^{-1}M_{2|0}^2-\lambda_2

399: n^{-1}M_{0|ij}M_{0|ji},

400: \label{X9}

401: \eeq

402: \beq

403: J_{2|ij}=-\nutt M_{2|ij}+\lambda_3 n^{-1}M_{2|0}M_{0|ij}-\lambda_4

404: n^{-1}\left(M_{0|ik}M_{0|kj}-\frac{1}{d}M_{0|k\ell}M_{0|\ell

405: k}\delta_{ij}\right),

406: \label{X11}

407: \eeq

408: \beqa

409: J_{0|ijk\ell}&=&-\nu_{0|4}M_{0|ijk\ell}+\lambda_5

410: n^{-1}\left[M_{0|ij}M_{0|k\ell}+

411: M_{0|ik}M_{0|j\ell}+M_{0|i\ell}M_{0|jk}-\frac{2}{d+4}\left(

412: M_{0|ip}M_{0|pj}\delta_{k\ell}\right.\right. \nn

413: &&\left.+M_{0|ip}M_{0|pk} \delta_{j\ell}

414: +M_{0|ip}M_{0|p\ell}\delta_{jk}+M_{0|jp}M_{0|pk}\delta_{i\ell}

415: +M_{0|jp}M_{0|p\ell}\delta_{ik}+M_{0|kp}M_{0|p\ell}\delta_{ij}\right)\nn

416: &&\left.

417: +\frac{2}{(d+2)(d+4)}M_{0|pq}M_{0|qp}\left(\delta_{ij}\delta_{k\ell}+

418: \delta_{ik}\delta_{j\ell}+\delta_{i\ell}\delta_{jk}\right) \right].

419: \label{X12}

420: \eeqa

421: In Eqs.\ (\ref{X9})--(\ref{X12}),  the usual summation convention

422: over repeated indices is assumed. The collision frequencies (or

423: eigenvalues) $\nu_{2r|{s}}$ and the cross coefficients $\lambda_i$

424: are given by

425: \begin{figure}

426: \includegraphics[width=0.95\columnwidth]{fig1.eps}

427: \caption{(Color online) Plot of the (reduced) fourth degree

428: eigenvalues $\nuzf^*$, $\nutt^*$, and $\nufz^*$ (top panels) and of

429: the shifted eigenvalues $\omegazf$, $\omegatt$, and $\omegafz$

430: (bottom panels) as functions of the coefficient of restitution. The

431: left and right panels correspond to $d=2$ and $d=3$, respectively.

432: \label{fig1}}

433: \end{figure}

434: \beqa

435: \nufz&=&\frac{(1+\al)\left[12d+9-\alpha(4d+17)+3\alpha^2-3\alpha^3\right]}{16d}\nu_0\nn

436: &=&2\nutz+\frac{(1+\al)^2\left(4d-7+6\alpha-3\alpha^2\right)}{16d}\nu_0,

437: \label{X10}

438: \eeqa

439: \beqa

440: \nutt&=&\frac{(1+\al)\left[7d^2+31d+18-\al(d^2+14d+34)

441: +3\al^2(d+2)-6\al^3\right]}{8d(d+4)}\nu_0\nn &=&

442: 2\nutz+\frac{(1+\al)^2\left[3d^2+7d-14+3\al(d+4)-6\al^2\right]}{8d(d+4)}\nu_0,

443: \label{X15}

444: \eeqa

445: \beqa

446: \nuzf&=&\frac{(1+\al)\left[2d^3+21d^2+61d+

447: 39-3\al(d+3)(d+5)+3\al^2(d+3)-3\al^3\right]}{2d(d+4)(d+6)}\nu_0\nn

448: &=&2\nutz+\frac{(1+\al)^2\left[d^3+9d^2+17d-9+3\al(d+4)-3\al^2\right]}{2d(d+4)(d+6)}\nu_0,

449: \label{X18}

450: \eeqa

451: \beq

452: \lambda_1=\frac{(1+\al)^2(d+2)\left(4d-1-6\al+3\al^2\right)}{16d^2}\nu_0,

453: \label{X14}

454: \eeq

455: \beq

456: \lambda_2=\frac{(1+\al)^2\left(1+6\al-3\al^2\right)}{8d}\nu_0,

457: \label{X13}

458: \eeq

459: \beq

460: \lambda_3=\frac{(1+\al)^2\left[d^2+5d-2-3\al(d+4)+6\al^2\right]}{8d^2}\nu_0,

461: \label{X17}

462: \eeq

463: \beq

464: \lambda_4=\frac{(1+\al)^2\left[2-d+3\al(d+4)-6\al^2\right]}{4d(d+4)}\nu_0,

465: \label{X16}

466: \eeq

467: \beq

468: \lambda_5=\frac{(1+\al)^2\left[d^2+7d+9-3\al(d+4)+3\al^2\right]}{2d(d+4)(d+6)}\nu_0.

469: \label{X19}

470: \eeq

471: Equations \eqref{X10} and \eqref{X14} coincide with the results of

472: Ref.\ \onlinecite{S03}.

473: 

474: 

475: We have checked that, in the elastic case ($\alpha=1$) and for three-dimensional systems ($d=3$), all the

476: expressions reported in this Section reduce to known results \cite{TM80,GS03}. In the one-dimensional elastic

477: case ($d=1$, $\alpha=1$), the gas behaves as an ideal gas because a collision is equivalent to exchanging the

478: labels of both colliding particles. As a consequence, $J_{2r|\bar{s}}=0$. It is easy to check that Eqs.\

479: \eqref{X5}--\eqref{X19} are consistent with this property since the coefficients affecting the non-vanishing

480: moments are zero, i.e., $\nuto=\nufz=\lambda_1=0$.

481:  Moreover, for  one-dimensional inelastic gases ($d=1$, $\alpha<1$),

482: our expressions for $\nutz$, $\nuto$, $\nufz$, and $\lambda_1$ agree with the results derived by Ben-Naim and

483: Krapivsky \cite{NK00}, who obtained the exact expressions for all the collisional moments, namely $J_{2r|0}$ and

484: $J_{2r|x}$.

485: 

486: 

487: 

488: 

489: While the $\alpha$ dependence of the second and third degree

490: eigenvalues, Eqs.\ \eqref{X6a}, \eqref{X6}, \eqref{X7b}, and

491: \eqref{X8}, is relatively simple, that of the fourth degree

492: eigenvalues  \eqref{X10}--\eqref{X18} and the cross coefficients

493: \eqref{X14}--\eqref{X19} is more involved. Figure \ref{fig1} shows

494: the $\alpha$ dependence of the (reduced) eigenvalues $\nufz^*$,

495: $\nutt^*$, and $\nuzf^*$, where

496: $\nu_{2r|s}^*\equiv\nu_{2r|s}/\nu_0$, and the shifted eigenvalues

497: $\omegafz$, $\omegatt$, and $\omegazf$, where we have called

498: $\omega_{2r|s}\equiv \nu_{2r|s}^*-(r+s/2)\nutz^*$, for $d=2$ and

499: $d=3$. While $\nuzf^*$ decays monotonically as the inelasticity

500: increases, the other two eigenvalues $\nutt^*$ and $\nufz^*$ start

501: growing, reach a maximum, and then decay. The maximum  value of

502: $\nutt^*$  occurs at $\alpha\simeq 0.40$ for $d=2$ and at

503: $\alpha\simeq 0.67$ for $d=3$. In the case of $\nufz^*$, the maximum

504: occurs at $\alpha\simeq 0.18$ and $\alpha\simeq 0.30$ for $d=2$ and

505: $d=3$, respectively. However, when the part associated with the

506: cooling rate is subtracted from the bare eigenvalues, the resulting

507: shifted quantities $\omegafz$, $\omegatt$, and $\omegazf$ exhibit a

508: monotonic behavior. As shown in Section \ref{sec3}, these shifted

509: eigenvalues are the relevant ones in the time relaxation  of the

510: \emph{scaled} moments in the free cooling problem. Therefore, the

511: decrease of $\omegafz$, $\omegatt$, and $\omegazf$ implies that the

512: characteristic relaxation times of the (scaled) fourth degree

513: moments toward their asymptotic values increase with dissipation.

514: 

515: 

516: \begin{figure}

517: \includegraphics[width=0.95\columnwidth]{fig2.eps}

518: \caption{(Color online) Plot of the (reduced) cross coefficients

519: $\lambda_i^*$,

520:  as functions

521: of the coefficient of restitution. The left and right panels

522: correspond to $d=2$ and $d=3$, respectively.

523: \label{fig2}}

524: \end{figure}

525: 

526: It is instructive to compare the fourth degree eigenvalues with the

527: second and third degree eigenvalues. In the elastic case, one has

528: $\omegafz=\omegato<\omegazt=\omegatt<\omegazth<\omegazf$ for $d=2$

529: and $\omegafz=\omegato<\omegazt<\omegatt<\omegazth<\omegazf$ for

530: $d=3$. We have observed that inelasticity breaks the degeneracy

531: $\omegafz=\omegato$ for both dimensionalities (yielding

532: $\omegafz<\omegato$) and the degeneracy $\omegazt=\omegatt$ for

533: $d=2$ (yielding $\omegatt<\omegazt$). The inelasticity also affects

534: the ordering of the eigenvalues: for $d=2$ one has

535: $\omegafz<\omegato<\omegatt<\omegazt<\omegazth<\omegazf$  if

536: $0.17<\alpha<1$ and

537: $\omegafz<\omegato<\omegatt<\omegazt<\omegazf<\omegazth$  if

538: $0<\alpha<0.17$; for $d=3$ the ordering is

539: $\omegafz<\omegato<\omegazt<\omegatt<\omegazth<\omegazf$ if

540: $0.43<\alpha<1$ and

541: $\omegafz<\omegato<\omegatt<\omegazt<\omegazth<\omegazf$ if

542: $0<\alpha<0.43$. Since, except $\omegafz$, these quantities are

543: related to moments which vanish in isotropic states, the fact that

544: $\omegafz$ is the smallest one implies that (as expected on physical

545: grounds) the characteristic time needed to achieve an isotropic

546: state is shorter than the one needed to reach the asymptotic state.

547: 

548: 

549: 

550: 

551: Let us consider now the (reduced) cross coefficients

552: $\lambda_i^*\equiv \lambda_i/\nu_0$ ($i=1,\ldots,5$), which measure

553: the coupling of the second degree moments to the evolution of the

554: fourth degree moments. Their dependence on dissipation is shown in

555: Fig.\ \ref{fig2}. It is apparent that the effect of inelasticity on

556: $\lambda_i^*$ is more pronounced than on the fourth degree

557: eigenvalues. For elastic collisions,

558: $\lambda_3^*=0<\lambda_5^*<\lambda_4^*=\lambda_2^*=\lambda_1^*$ for

559: $d=2$ and

560: $\lambda_5^*<\lambda_3^*<\lambda_4^*=\lambda_2^*<\lambda_1^*$ for

561: $d=3$. This ordering changes with inelasticity. Moreover,

562: $\lambda_1^*$, $\lambda_2^*$, and $\lambda_4^*$ significantly

563: decrease with increasing dissipation, $\lambda_3^*$ has a

564: non-monotonic behavior, and $\lambda_5^*$ is nearly constant. Note

565: that the coefficient $\lambda_4$ does not actually play any role in

566: $d=2$ since the combination

567: $M_{0|ik}M_{0|kj}-\frac{1}{d}M_{0|k\ell}M_{0|\ell k}\delta_{ij}$

568: appearing in the collisional moment $J_{2|i}$, cf.\ Eq.\

569: \eqref{X11}, vanishes in the two-dimensional case.

570: 

571: 

572: 

573: 

574: 

575: \section{Relaxation to the homogeneous cooling state\label{sec3}}

576: The results derived in the preceding Section can be applied to several interesting situations. Here we will

577: consider the most basic problem, namely the  time evolution of the moments of degree less than or equal to four

578: (both isotropic and anisotropic) in the \textit{homogeneous} free cooling state \cite{BP04}. In that case, the

579: Boltzmann equation \eqref{2.1} becomes

580: \beq

581: \partial_t f(\mathbf{v},t)=J[\mathbf{v}|f,f],

582: \label{Z1} \eeq which must be complemented with a  given initial condition $f(\mathbf{v},0)$. Since the

583: collisions are inelastic, the granular temperature $T(t)$ monotonically decays in time and so a steady state

584: does not exist. In the context of IMM, it has been proven \cite{BC03,BCT03} that, provided that

585: $f(\mathbf{v},0)$ has a finite moment of some degree higher than two, $f(\mathbf{v},t)$ asymptotically tends

586: toward a self-similar solution of the form \beq f(\mathbf{v},t)\to n [v_0(t)]^{-d} \phi(V/v_0(t)), \label{3.1}

587: \eeq where $v_0(t)\equiv \sqrt{2T(t)/m}$ is the thermal speed and $\phi(c)$ is an  isotropic distribution that

588: is only known in the one-dimensional case \cite{BMP02}. According to Eq.\ \eqref{3.1}, the \emph{scaled} moments

589: \beq

590: M_{2r|\bar{s}}^*(t)\equiv n^{-1}[v_0(t)]^{-(2r+s)}M_{2r|\bar{s}}(t) \label{3.1a}

591: \eeq

592: must tend asymptotically to

593: \beq M_{2r|\bar{s}}^*(t)\to \mu_{2r|\bar{s}}\equiv \int\dd\mathbf{c}\, Y_{2r|\bar{s}}(\mathbf{c})\phi(c).

594: \label{3.2}

595: \eeq

596: Due to the isotropy of $\phi(c)$, then $\mu_{2r|\bar{s}}=0$ unless $s=0$. Moreover, it is known that the scaled

597: distribution $\phi(c)$ exhibits an algebraic high velocity tail \cite{EB02a,EB02,KN02,KN02bis} of the form

598: $\phi(c)\sim c^{-d-\gamma(\alpha)}$, so that the moments $\mu_{2r|0}$ diverge if $2r\geq \gamma(\alpha)$. The

599: quantity $\gamma(\alpha)$ obeys a transcendental equation whose solution is always $\gamma(\alpha)>4$, except

600: for $d=1$. Consequently, for any value of $\alpha$ and $d\geq 2$, the scaled moment $M_{4|0}^*(t)$ goes to a

601: well defined value $\mu_{4|0}$, while the remaining scaled moments of degree equal to or less than four are

602: anisotropic (except of course $M_{0|0}^*=1$ and $M_{2|0}^*=d/2$) and so they tend to zero. The main goal of this

603: Section is to analyze in detail the relaxation of the second, third, and fourth degree moments (both isotropic

604: and anisotropic) toward their asymptotic values.

605: 

606: 

607: 

608: Taking velocity moments in both sides of Eq.\ \eqref{Z1} one has

609: \beq

610: \partial_t M_{2r|\bar{s}}=J_{2r|\bar{s}}.

611: \label{Z2}

612: \eeq

613: In particular,

614: \beq

615: \partial_t M_{2|0}=-\nutz M_{2|0}.

616: \label{Z3} \eeq

617: Since $M_{2|0}=dnT/m$, Eq.\ \eqref{Z3}  is the equation for the time evolution of the granular

618: temperature and $\nutz$ is the cooling rate. The solution of Eq.\ \eqref{Z3} is \beq

619: T(t)=\frac{T(0)}{\left[1+\nutz(0)t/2\right]^2}, \label{Z3b} \eeq where $T(0)$ is the initial temperature and

620: $\nutz(0)\propto T^{1/2}(0)$ is the initial cooling rate. Equation \eqref{Z3b} is not but Haff's law

621: \cite{BP04}.

622: 

623: 

624: 

625: 

626: 

627: Let us consider now the scaled moments \eqref{3.1a}. In that case,

628: from Eqs.\ (\ref{Z2}) and (\ref{Z3}) one simply gets

629: \beq

630: \partial_\tau M_{2r|\bar{s}}^*=J_{2r|\bar{s}}^*+\frac{2r+s}{2}\nutz^* M_{2r|\bar{s}}^*,

631: \label{Z5}

632: \eeq

633: where  $J_{2r|\bar{s}}^*\equiv J_{2r|\bar{s}}/\nu_0nv_0^{2r+s}$ and

634: \beq

635: \tau=\int_0^t\dd t'\, \nu_0(t')

636: \label{Z6}

637: \eeq

638: measures time as the number of (effective) collisions per particle.

639: The effect of the second term on the right-hand side of Eq.\

640: \eqref{Z5} is to shift the eigenvalues $\nu_{2r|s}^*$ to

641: $\omega_{2r|s}=\nu_{2r|s}^*-(r+s/2)\nu_{2|0}^*$. For instance,

642: \beq

643: \partial_\tau M_{4|0}^*=-\omegafz M_{4|0}^*+\lambda_1^*

644: \frac{d^2}{4}-\lambda_2^* M_{0|ij}^*M_{0|ji}^*.

645: \label{Z8}

646: \eeq

647: 

648: \subsection{The one-dimensional case}

649: In the one-dimensional case, the only scaled moments of degree equal

650: to or less than four are (apart from $M_{0|0}^*=1$ and

651: $M_{2|0}^*=\frac{1}{2}$) $M_{2|x}^*$ and $M_{4|0}^*$. Their

652: evolution equations are

653: \beq

654: \partial_\tau M_{2|x}^*=0,

655: \label{Z17}

656: \eeq

657: \beq

658: \partial_\tau

659: M_{4|0}^*=\frac{3}{16}(1-\alpha^2)^2\left(M_{4|0}^*+\frac{3}{4}\right). \label{Z16} \eeq The solution of Eq.\

660: \eqref{Z16} is \beq M_{4|0}^*(\tau)=M_{4|0}^*(0)

661: e^{\frac{3}{16}(1-\alpha^2)^2\tau}+\frac{3}{4}\left[e^{\frac{3}{16}(1-\alpha^2)^2\tau}-1\right]. \label{Z18.2}

662: \eeq

663: This solution shows that the (scaled) fourth degree moment monotonically increases with time, i.e.,

664: $\mu_{4|0}=\infty$. This is consistent with the exact HCS solution found by Baldassarri \emph{et

665: al}.\cite{BMP02}, namely

666: \beq \phi(c)=\frac{2^{3/2}}{\pi}\frac{1}{(1+2c^2)^2}. \label{Z18}

667: \eeq

668: 

669: On the other hand, Eq.\ \eqref{Z17} shows that

670: $M_{2|x}^*(\tau)=M_{2|x}^*(0)$, i.e., if the initial state is

671: anisotropic with $M_{2|x}(0)\neq 0$ then one has

672: $M_{2|x}(t)=M_{2|x}(0)[T(t)/T(0)]^{3/2}$. The constancy of

673: $M_{2|x}^*$ implies that any initial anisotropy does not vanish in

674: the scaled velocity distribution function for long times. As a

675: consequence, while the distribution \eqref{Z18} represents the

676: asymptotic form $\phi(c)$ for a wide class of \emph{isotropic}

677: initial conditions, it cannot be reached, strictly speaking, for any

678: \emph{anisotropic} initial state. Whether or not there exists a

679: generalization of \eqref{Z18} for anisotropic states is, to the best

680: of our knowledge, an open problem. Since the symmetry of the

681: distribution \eqref{Z18} implies that $M_{2|x}^*\equiv\langle

682: c_x^3\rangle =0$ but the average $\langle |c|^3\rangle$ diverges,  a

683: small correction to the form \eqref{Z18} could accommodate a finite

684: value of $ \langle c_x^3\rangle$.

685: 

686: 

687: \subsection{The two-dimensional case}

688: As is known, one-dimensional systems are generally not very

689: realistic and so they can exhibit peculiar properties. However,

690: two-dimensional systems are usually considered as representative of

691: the features present in real systems.

692: 

693: For the sake of simplicity, in the remainder of this Section we will consider the two-dimensional case. The set

694: of independent moments of second, third, and fourth degree will be taken as

695: \beq \{M_{0|xx}^*,M_{0|xy}^*\},

696: \eeq \beq \{M_{2|x}^*,M_{2|y}^*,M_{0|xxx}^*,M_{0|xxy}^*\},

697: \eeq

698: \beq

699: \{M_{4|0}^*,M_{2|xx}^*,M_{2|xy}^*,M_{0|xxxx}^*,M_{0|xxxy}^*\}, \label{4th}

700: \eeq

701: respectively. The remaining moments are simply related to the above ones as $M_{0|yy}^*=-M_{0|xx}^*$,

702: $M_{0|xyy}^*=-M_{0|xxx}^*$, $M_{0|yyy}^*=-M_{0|xxy}^*$,  $M_{2|yy}^*=-M_{2|xx}^*$,

703: $M_{0|yyyy}^*=-M_{0|xxyy}^*=M_{0|xxxx}^*$, $M_{0|xyyy}^*=-M_{0|xxxy}^*$. {}From Eq.\ \eqref{Z5}, it is easy to

704: obtain the time dependence of the (scaled) second and third degree moments:

705: \beq

706: M_{0|ij}^*(\tau)=M_{0|ij}^*(0)e^{-\omegazt \tau}, \label{Z13}

707: \eeq

708: \beq M_{2|i}^*(\tau)=M_{2|i}^*(0)e^{-\omegato

709: \tau}, \quad M_{0|ijk}^*(\tau)=M_{0|ijk}^*(0)e^{-\omegazth \tau}.

710: \eeq

711: In the case of the fourth degree moments, one has to deal with inhomogeneous linear differential equations

712: involving the second degree moments. The solutions are

713: \beqa M_{4|0}^*(\tau)&=&M_{4|0}^*(0)e^{-\omegafz

714: \tau}+\frac{\lambda_1^*}{\omegafz}\left(1-e^{-\omegafz \tau}\right)\nn

715: &&-\frac{2\lambda_2^*}{2\omegazt-\omegafz}\left[M_{0|xx}^{*2}(0)+M_{0|xy}^{*2}(0)\right]\left(e^{-\omegafz

716: \tau}-e^{-2\omegazt \tau}\right), \label{Z9}

717: \eeqa

718: \beq M_{2|ij}^*(\tau)=M_{2|ij}^*(0)e^{-\omegatt

719: \tau}+\frac{\lambda_3^*}{\omegazt-\omegatt}M_{0|ij}^*(0)\left(e^{-\omegatt \tau}-e^{-\omegazt \tau}\right),

720: \label{Z10} \eeq \beq M_{0|xxxx}^*(\tau)=M_{0|xxxx}^*(0)e^{-\omegazf

721: \tau}+\frac{3}{2}\frac{\lambda_5^*}{2\omegazt-\omegazf}\left[M_{0|xx}^{*2}(0)-M_{0|xy}^{*2}(0)\right]\left(e^{-\omegazf

722: \tau}-e^{-2\omegazt \tau}\right), \label{Z11}

723: \eeq

724: \beq M_{0|xxxy}^*(\tau)=M_{0|xxxy}^*(0)e^{-\omegazf

725: \tau}+\frac{3\lambda_5^*}{2\omegazt-\omegazf}M_{0|xx}^{*}(0)M_{0|xy}^{*}(0)\left(e^{-\omegazf

726: \tau}-e^{-2\omegazt \tau}\right). \label{Z12}

727: \eeq

728: Equations \eqref{Z13}--\eqref{Z12} show that all the moments,

729: except $M_{4|0}^*$, tend to zero for sufficiently long times. The asymptotic expression of $M_{4|0}^*$ is

730: \beq

731: M_{4|0}^*\to\mu_{4|0}=\frac{\lambda_1^*}{\omegafz}=2\frac{7-6\alpha+3\alpha^2}{1+6\alpha-3\alpha^2}, \label{Z14}

732: \eeq

733: which agrees with previous results \cite{S03}.

734: 

735: 

736: As an illustration, let us analyze the time evolution of the scaled

737: fourth degree moments \eqref{4th} for the following initial

738: anisotropic distribution:

739: \beq

740: f(\mathbf{v},0)=\frac{n}{3}\left[\delta(\mathbf{V}-\mathbf{V}_1)+\delta(\mathbf{V}-\mathbf{V}_2)+\delta(\mathbf{V}-\mathbf{V}_3)\right],

741: \label{Z15}

742: \eeq

743: where $\mathbf{V}_1=v_0(0)\widehat{\mathbf{x}}$,

744: $\mathbf{V}_2=(v_0(0)/\sqrt{2})\widehat{\mathbf{y}}$, and

745: $\mathbf{V}_3=-\mathbf{V}_1-\mathbf{V}_2$. Here,

746: $v_0(0)=\sqrt{2T(0)/m}$ is the initial thermal speed, where the

747: initial temperature $T(0)$ is arbitrary. Figure \ref{fig3} shows the

748: evolution of the moments \eqref{4th} for two values of the

749: coefficient of restitution: $\alpha=1$ (elastic system) and

750: $\alpha=0.5$ (strongly inelastic system). It is quite apparent that

751: the number of collisions needed to reach the HCS values increases

752: with the inelasticity, as expected from Fig.\ \ref{fig1}. In the

753: particular case of $\alpha=0.5$, the relaxation times are about

754: twice the ones corresponding to $\alpha=1$. Moreover, since

755: $\omegazf>\omegatt>\omegafz$, we observe that the moments

756: $M_{0|ijk\ell}^*$ tend to zero more rapidly than the moments

757: $M_{2|ij}^*$, and that the isotropic moment $M_{4|0}^*$ reaches its

758: asymptotic value more slowly than the anisotropic moments.

759: \begin{figure}

760: \includegraphics[width=0.95\columnwidth]{fig3.eps}

761: \caption{(Color online) Time evolution of the scaled fourth degree

762: moments in the free cooling of a two-dimensional IMM with

763: $\alpha=0.5$ (solid lines). The time evolution to equilibrium of an

764: elastic system ($\alpha=1$) is also shown (dashed lines). In both

765: cases the initial state is described by Eq.\ \protect\eqref{Z15}.

766: The horizontal dotted lines in the bottom panel indicate the

767: corresponding asymptotic HCS values $\mu_{4|0}$.

768: \label{fig3}}

769: \end{figure}

770: 

771: \section{Discussion\label{sec4}}

772: As Maxwell already realized \cite{Maxwell}, scattering models where the collision rate of two particles

773: approaching each other with a relative velocity $\mathbf{g}$ is independent of the magnitude of $\mathbf{g}$

774: allows one to evaluate exactly the collisional moments without the explicit knowledge of the velocity

775: distribution function. In the conventional case of ordinary gases of particles colliding elastically, Maxwell

776: models are useful to find non-trivial exact solutions to the Boltzmann equation in far from equilibrium

777: situations \cite{GS03}. Needless to say, the introduction of inelasticity through a constant coefficient of

778: normal restitution $\alpha\leq 1$ opens up new perspectives for exact results, including the elastic case as a

779: special limit ($\alpha=1$). This justifies the growing interest in IMM by physicists and mathematicians alike in

780: the past few years.

781: 

782: The choice of the Ikenberry polynomials \cite{TM80} $Y_{2r|\bar{s}}$ of degree $2r+s$ allows one to express the

783: corresponding collisional moments $J_{2r|\bar{s}}$ in the form \eqref{2.4}: an eigenvalue $-\nu_{2r|s}$ times

784: the velocity moment $M_{2r|\bar{s}}$ plus a bilinear combination of moments of degree less than $2r+s$. In

785: particular, $\nu_{2|0}$ is the cooling rate of the gas. In this paper we have evaluated all the third and fourth

786: degree collisional moments of the IMM defined by Eq.\ \eqref{1}. In that context, the results are exact for

787: arbitrary values of $\alpha$ and apply to any dimensionality $d$. Known results are recovered for

788: three-dimensional elastic systems \cite{TM80,GS03} and for one-dimensional inelastic systems \cite{NK00}. We

789: have observed that some of the eigenvalues $\nu_{2r|s}$ do not have a monotonic dependence on $\alpha$, while

790: the shifted eigenvalues $\nu_{2r|s}-(r+s/2)\nu_{2|0}$ monotonically decrease with increasing inelasticity.

791: Moreover, given a value of $\alpha$ and a degree $2r+s$, the eigenvalues $\nu_{2r|s}$ increase with $s$. This

792: means that the larger the anisotropy of a moment $M_{2r|\bar{s}}$ the higher its collisional rate of change.

793: Although the above observations are based on the moments of degree $2r+s\leq 4$, we expect that they extend to

794: moments of higher degree.

795: 

796: As a simple application of the results derived in Section

797: \ref{sec2}, we have studied the time evolution of the  moments of

798: degree equal to or less than four in the free cooling state, in

799: which case the evolution of the moments scaled with the thermal

800: speed is governed by the shifted eigenvalues. An interesting feature

801: of the one-dimensional case is that  the heat flux

802: $q_x=(m/2)M_{2|x}$, when scaled with the thermal speed, does not

803: change in time, so that an initial anisotropic distribution cannot

804: evolve toward an asymptotic isotropic distribution. Thus, the exact

805: solution found by Baldassarri \emph{et al.}\cite{BMP02} does not

806: play a universal role, at least in a strict sense, unless the

807: initial distribution is isotropic. On the other hand, we have found

808: that all the anisotropic moments of degree equal to or less than

809: four vanish in the long time limit for $d\geq 2$. However, this does

810: not preclude the possibility that anisotropic moments of higher

811: degree diverge for $\alpha$ sufficiently small. We plan to explore

812: this possibility in the near future.

813: 

814: The explicit results provided in this paper can be useful for

815: studying different problems. An important application is the exact

816: derivation of the Burnett order constitutive equations for IMM, with

817: explicit expressions of the associated transport coefficients as

818: functions of $d$ and $\alpha$. This is possible because the

819: determination of the Burnett order pressure tensor and heat flux

820: requires the previous knowledge of the third and fourth degree

821:  collisional moments to Navier--Stokes order. Another

822: interesting problem is the so-called simple or uniform shear flow, which is an intrinsically non-Newtonian state

823: \cite{SGD04}. Apart from the rheological quantities, the results derived here allows one to analyze the time

824: evolution of the fourth degree velocity moments toward their steady state values \cite{SG07} and investigate

825: their possible divergence, in a similar way to the analysis carried out in the elastic case \cite{SGBD93}.

826: Finally, the generalized transport coefficients characterizing small perturbations around the simple shear flow

827: have been determined \cite{G07} and compared with those previously obtained for IHS \cite{G06} from a model

828: kinetic equation.

829: 

830: 

831: 

832: \acknowledgments

833: 

834: 

835: 

836: This research  has been supported by the Ministerio de Educaci\'on y Ciencia (Spain) through Grant No.\

837: FIS2007-60977, partially financed by FEDER funds.

838: 

839: 

840: 

841: 

842: \appendix*

843: \section{Explicit evaluation of the collisional moments\label{appA}}

844: In this Appendix we give the details of the derivation,   by using

845: the property \eqref{5}, of the collisional moments

846: $J_{2r|\bar{s}}=\mathcal{J}[Y_{2r|\bar{s}}]$ associated with the

847: Ikenberry polynomials of third and fourth degree. To carry out the

848: calculations we will need the angular integrals

849: \begin{equation}

850: \label{N9}

851: \int

852: \dd\widehat{\boldsymbol{\sigma}}\,(\widehat{\boldsymbol{\sigma}}\cdot

853: {\bf g})^{2\qq+1} \widehat{{\sigma}}_i=B_{\qq+1} g^{2\qq}{g_i},

854: \end{equation}

855: \begin{equation}

856: \label{N10}

857: \int

858: \dd\widehat{\boldsymbol{\sigma}}\,(\widehat{\boldsymbol{\sigma}}\cdot

859: {\bf g})^{2\qq} \widehat{{\sigma}}_i\widehat{{\sigma}}_j

860: =\frac{B_{\qq}}{2\qq+d} g^{2(\qq-1)}\left(2\qq{g_i}{g_j}+g^2

861: \delta_{ij}\right),

862: \end{equation}

863: \begin{equation}

864: \label{N21}

865: \int

866: \dd\widehat{\boldsymbol{\sigma}}\,(\widehat{\boldsymbol{\sigma}}\cdot

867: {\bf g})^{2\qq+1}

868: \widehat{\sigma}_i\widehat{\sigma}_j\widehat{\sigma}_k=\frac{B_{\qq+1}}{2(\qq+1)+d}

869: g^{2(\qq-1)}\left[ 2\qq

870: g_ig_jg_k+g^2\left(\delta_{ij}g_k+\delta_{ik}g_j+\delta_{jk}g_i\right)\right],

871: \end{equation}

872: \begin{eqnarray}

873: \label{N29}

874: \int

875: \dd\widehat{\boldsymbol{\sigma}}\,(\widehat{\boldsymbol{\sigma}}\cdot

876: {\bf g})^{2\qq}

877: \widehat{\sigma}_i\widehat{\sigma}_j\widehat{\sigma}_k

878: \widehat{\sigma}_{\ell}&=&\frac{B_{\qq}}{(2\qq+d)[d+2(\qq+1)]}

879: \left[4\qq(\qq-1)g^{2(\qq-2)}g_ig_jg_kg_{\ell}\right. \nonumber\\

880: & &

881: +g^{2(\qq-1)}2\qq\left(g_ig_j\delta_{k\ell}+g_ig_k\delta_{j\ell}+g_ig_{\ell}\delta_{jk}+

882: g_kg_j\delta_{i\ell}+g_{\ell}g_j\delta_{ik}+g_kg_{\ell}\delta_{ij}\right)\nonumber\\

883: & & \left.

884: +g^{2\qq}\left(\delta_{ij}\delta_{k\ell}+\delta_{ik}\delta_{j\ell}+\delta_{jk}\delta_{i\ell}\right)

885: \right].

886: \end{eqnarray}

887: Here, the coefficients $B_\qq$ are \cite{NE98}

888: \begin{equation}

889: \label{N11}

890: B_{\qq}=\int

891: \dd\widehat{\boldsymbol{\sigma}}\,(\widehat{\boldsymbol{\sigma}}\cdot

892: {\widehat{\bf

893: g}})^{2\qq}=\Omega_d\pi^{-1/2}\frac{\Gamma\left(\frac{d}{2}\right)

894: \Gamma\left(\qq+\frac{1}{2}\right)}{\Gamma\left(\qq+\frac{d}{2}\right)}.

895: \end{equation}

896: 

897: 

898: 

899: \subsection{Third degree moments\label{appAbis}}

900: We start by noting that the collision rule \eqref{6} implies that

901: \beqa

902: V_{1i}''V_{1j}''V_{1k}''&=&V_{1i}V_{1j}V_{1k}-

903: \at\sg\left(V_{1i}V_{1j}\si_k+V_{1i}V_{1k}\si_j+V_{1j}V_{1k}\si_i\right)\nn

904: &&+

905: \left(\at\right)^2\sg^2\left(V_{1i}\si_j\si_k+V_{1j}\si_i\si_k+V_{1k}\si_i\si_j\right)-

906: \left(\at\right)^3\sg^3\si_i\si_j\si_k.\nn &&

907: \label{A1}

908: \eeqa

909: Next, making use of Eqs.\ \eqref{N9}--\eqref{N21}, one gets

910: \beqa

911: \mathcal{J}[V_{1i}V_{1j}V_{1k}]&=&-\frac{n\nu_0}{2d}\at\left\{(d+2)\langle

912: g_iV_{1j}V_{1k}+g_jV_{1i}V_{1k}+g_kV_{1i}V_{1j}\rangle\right.\nn

913: &&\left.-\at\left[2\langle

914: g_ig_jV_{1k}+g_ig_kV_{1j}+g_jg_kV_{1i}\rangle+\langle

915: g^2\left(V_{1i}\delta_{jk}+V_{1j}\delta_{ik}+V_{1k}\delta_{ij}\right)\rangle\right]\right\},\nn

916: &&

917: \label{A2}

918: \eeqa

919: where the brackets are defined as

920: \begin{equation}

921: \label{N13}

922: \langle h({\bf V}_1,{\bf V}_2)\rangle \equiv \frac{1}{n^2}\int

923: \dd{\bf V}_1\int \dd{\bf V}_2 h({\bf V}_1,{\bf V}_2) f({\bf

924: V}_1)f({\bf V}_2).

925: \end{equation}

926:  and we have taken

927: into account that $\langle g_ig_jg_k\rangle=\langle

928: g^2g_{i}\rangle=0$. It is easy to get

929: \beq

930: n\langle g_iV_{1j}V_{1k}\rangle=n\langle

931: g_ig_{j}V_{1k}\rangle=M_{0|ijk}+\frac{1}{d+2}\left(M_{2|i}\delta_{jk}+M_{2|j}\delta_{ik}+M_{2|k}\delta_{ij}\right),

932: \label{A3}

933: \eeq

934: \beq

935: n\langle g^2V_{1i}\rangle=M_{2|i}.

936: \label{A4}

937: \eeq

938: Therefore,

939: \beq

940: \mathcal{J}[V_{1i}V_{1j}V_{1k}]=-\frac{\nu_0}{2d}\at\left[3(d+1-\alpha)M_{0|ijk}+\frac{5d+4-\alpha(d+8)}{2(d+2)}

941: \left(M_{2|i}\delta_{jk}+M_{2|j}\delta_{ik}+M_{2|k}\delta_{ij}\right)\right].

942: \label{A5}

943: \eeq

944: If one makes $j=k$ and sum over $j$ one gets the first equality of

945: Eq.\ (\ref{X7}) with $\nu_{2|1}$ given by  Eq.\ (\ref{X7b}). Also,

946: by subtracting

947: $\left(J_{2|i}\delta_{jk}+J_{2|j}\delta_{ik}+J_{2|k}\delta_{ij}\right)/(d+2)$

948: from both sides of Eq.\ (\ref{A5}) one gets the second equality of

949: Eq.\ (\ref{X7}) with $\nu_{0|3}$ given by  Eq.\ (\ref{X8}).

950: 

951: \subsection{Fourth degree moments\label{appB}}

952: Now the starting point is the collision rule

953: \beqa

954: V_{1i}''V_{1j}''V_{1k}''V_{1\ell}''&=&V_{1i}V_{1j}V_{1k}V_{1\ell}-

955: \at\sg\left(V_{1i}V_{1j}V_{1k}\si_\ell+V_{1i}V_{1j}V_{1\ell}\si_k+V_{1i}V_{1k}V_{1\ell}\si_j+V_{1j}V_{1k}V_{1\ell}\si_i\right)\nn

956: &&+

957: \left(\at\right)^2\sg^2\left(V_{1i}V_{1j}\si_k\si_\ell+V_{1i}V_{1k}\si_j\si_\ell+V_{1i}V_{1\ell}\si_j\si_k

958: +V_{1j}V_{1k}\si_i\si_\ell+V_{1j}V_{1\ell}\si_i\si_k\right.\nn

959: &&\left.+V_{1k}V_{1\ell}\si_i\si_j\right)-

960: \left(\at\right)^3\sg^3\left(V_{1i}\si_j\si_k\si_\ell+V_{1j}\si_i\si_k\si_\ell+V_{1k}\si_i\si_j\si_\ell+V_{1\ell}\si_i\si_j\si_k\right)\nn

961: &&+\atp^4\sg^4\si_i\si_j\si_k\si_\ell.

962: \label{B1}

963: \eeqa

964: After integrating over $\widehat{\bm{\sigma}}$,

965: \beqa

966: \mathcal{J}[V_{1i}V_{1j}V_{1k}V_{i\ell}]&=&-\frac{n\nu_0}{2d(d+4)(d+6)}\at\left\{(d+2)(d+4)(d+6)\langle

967: g_iV_{1j}V_{1k}V_{1\ell}+\stackrel{(3)}{\cdots}\rangle\right.\nn

968: &&-\at(d+4)(d+6)\left[2\langle

969: g_ig_jV_{1k}V_{1\ell}+\stackrel{(5)}{\cdots}\rangle+\langle

970: g^2(V_{1i}V_{ij}\delta_{k\ell}+\stackrel{(5)}{\cdots})\rangle\right]\nn

971: && +3(d+6)\atp^2 \left[2\langle

972: g_ig_jg_kV_{1\ell}+\stackrel{(3)}{\cdots}\rangle+\langle

973: g^2[(g_iV_{1j}+g_jV_{1i})\delta_{k\ell}+\stackrel{(5)}{\cdots}]\rangle\right]\nn

974: && \left. -3\atp^3\left[8\langle g_ig_jg_kg_\ell\rangle+4\langle

975: g^2(g_ig_j\delta_{k\ell}+\stackrel{(5)}{\cdots})\rangle+\langle

976: g^4\rangle

977: (\delta_{ij}\delta_{k\ell}+\stackrel{(2)}{\cdots})\right]\right\},\nn

978: &&

979: \label{B2}

980: \eeqa

981: where $\stackrel{(s)}{\cdots}$ denotes the $s$ terms obtained from

982: the canonical one by permutation of indices. Making $k=\ell$ and

983: summing over $k$ we obtain

984: \beqa

985: \mathcal{J}[V_1^2V_{1i}V_{1j}]&=&-\frac{n\nu_0}{2d(d+4)}\at\left\{(d+2)(d+4)\langle

986: 2 (\mathbf{g}\cdot\mathbf{V}_1)V_{1i}V_{1j}+V^2(

987: g_iV_{1j}+g_jV_{1i})\rangle\right.\nn &&-\at(d+4)\left[2\langle

988: V_1^2g_ig_j+2(\mathbf{g}\cdot\mathbf{V}_1)(g_iV_{1j}+g_j

989: V_{1i})\rangle +\langle

990: (d+6)g^2V_{1i}V_{1j}+g^2V_1^2\delta_{ij}\rangle\right]\nn &&

991: +3\atp^2 \left[(d+6)\langle g^2(g_iV_{1j}+g_j

992: V_{1i})\rangle+2\langle

993: (\mathbf{g}\cdot\mathbf{V}_1)(2g_ig_j+g^2\delta_{ij})\rangle\right]\nn

994: && \left. -3\atp^3\langle 4g^2g_ig_j+g^4\delta_{ij}\rangle\right\}.

995: \label{B2.2}

996: \eeqa

997:  Next, by

998: taking $i=j$ and summing over $i$, Eq.\ (\ref{B2.2}) yields

999: \beqa

1000: \mathcal{J}[V_1^4]&=&-\frac{n\nu_0}{2d}\at\left\{4(d+2)\langle

1001: (\mathbf{g}\cdot\mathbf{V}_1)V_1^2\rangle-2\at\left[(d+4)\langle

1002: g^2V_1^2\rangle +4\langle

1003: (\mathbf{g}\cdot\mathbf{V}_1)^2\rangle\right]\right.\nn  &&\left.

1004: +12\atp^2 \langle (\mathbf{g}\cdot\mathbf{V}_1)g^2\rangle

1005: -3\atp^3\langle g^4\rangle\right\}.

1006: \label{B2.3}

1007: \eeqa

1008: 

1009: 

1010: Now we express the averages in terms of the Ikenberry moments. Let

1011: us consider first the four-index averages:

1012: \beq

1013: n\langle

1014: g_iV_{1j}V_{1k}V_{1\ell}\rangle=M_{0|ijk\ell}+\frac{1}{d+4}\left(M_{2|ij}\delta_{k\ell}+\stackrel{(5)}{\cdots}\right)+\frac{1}{d(d+2)}

1015: M_{4|0}\left(\delta_{ij}\delta_{k\ell}+\stackrel{(2)}{\cdots}\right),

1016: \label{B3}

1017: \eeq

1018: \beqa

1019: n\langle

1020: g_ig_{j}V_{1k}V_{1\ell}\rangle&=&M_{0|ijk\ell}+\frac{1}{d+4}\left(M_{2|ij}\delta_{k\ell}+\stackrel{(5)}{\cdots}\right)+\frac{1}{d(d+2)}

1021: M_{4|0}\left(\delta_{ij}\delta_{k\ell}+\stackrel{(2)}{\cdots}\right)\nn

1022: &&

1023: +n^{-1}M_{0|ij}M_{0|k\ell}+\frac{n^{-1}}{d}M_{2|0}\left(M_{0|ij}\delta_{k\ell}+M_{0|k\ell}\delta_{ij}\right)

1024: +\frac{n^{-1}}{d^2}M_{2|0}^2\delta_{ij}\delta_{k\ell},\nn &&

1025: \label{B4}

1026: \eeqa

1027: \beqa

1028: n\langle

1029: g_ig_{j}g_{k}V_{1\ell}\rangle&=&M_{0|ijk\ell}+\frac{1}{d+4}\left(M_{2|ij}\delta_{k\ell}+\stackrel{(5)}{\cdots}\right)+\frac{1}{d(d+2)}

1030: M_{4|0}\left(\delta_{ij}\delta_{k\ell}+\stackrel{(2)}{\cdots}\right)\nn

1031: &&

1032: +n^{-1}\left(M_{0|ij}M_{0|k\ell}+\stackrel{(2)}{\cdots}\right)+\frac{n^{-1}}{d}M_{2|0}\left(M_{0|ij}\delta_{k\ell}+\stackrel{(5)}{\cdots}\right)

1033: +\frac{n^{-1}}{d^2}M_{2|0}^2\left(\delta_{ij}\delta_{k\ell}+\stackrel{(2)}{\cdots}\right),\nn

1034: &&

1035: \label{B5}

1036: \eeqa

1037: \beq

1038: \langle g_ig_{j}g_{k}g_{\ell}\rangle=2\langle

1039: g_ig_{j}g_{k}V_{1\ell}\rangle.

1040: \label{B6}

1041: \eeq

1042: Summing over two repeated indices we get the two-index averages:

1043: \beq

1044: n\langle g^2V_{1i}V_{1j}\rangle=n\langle

1045: V_1^2g_ig_j\rangle=M_{2|ij}+\frac{1}{d}M_{4|0}\delta_{ij}+n^{-1}M_{2|0}\left(M_{0|ij}+\frac{1}{d}M_{2|0}\delta_{ij}\right),

1046: \label{B7}

1047: \eeq

1048: \beqa

1049: n\langle g^2g_{i}V_{1j}\rangle&=&n\langle

1050: (\mathbf{g}\cdot\mathbf{V}_1)g_{i}g_{j}\rangle=M_{2|ij}+\frac{1}{d}M_{4|0}\delta_{ij}+2n^{-1}M_{0|ik}M_{0|kj}\nn

1051: &&+\frac{n^{-1}}{d}M_{2|0}\left[(d+4)M_{0|ij}

1052: +\frac{d+2}{d}M_{2|0}\delta_{ij}\right],

1053: \label{B8}

1054: \eeqa

1055: \beq

1056: \langle g^2g_{i}g_{j}\rangle=2\langle g^2g_{i}V_{1j}\rangle,

1057: \label{B9}

1058: \eeq

1059: \beq

1060: n\langle V_1^2 g_iV_{1j}\rangle=n\langle

1061: (\mathbf{g}\cdot\mathbf{V}_1)V_{1i}V_{1j}\rangle=M_{2|ij}+\frac{1}{d}M_{4|0}\delta_{ij},

1062: \label{B11}

1063: \eeq

1064: \beq

1065: n\langle

1066: (\mathbf{g}\cdot\mathbf{V}_1)g_{i}V_{1j}\rangle=M_{2|ij}+\frac{1}{d}M_{4|0}\delta_{ij}+n^{-1}M_{0|ik}M_{0|kj}+\frac{n^{-1}}{d}M_{2|0}\left[

1067: 2M_{0|ij}+\frac{1}{d}M_{2|0}\delta_{ij}\right].

1068: \label{B13}

1069: \eeq

1070: Summing again,

1071: \beq

1072: n\langle g^2V_1^2\rangle=M_{4|0}+n^{-1}M_{2|0}^2,

1073: \label{B12}

1074: \eeq

1075: \beq

1076: n\langle (\mathbf{g}\cdot\mathbf{V}_1)V_1^2\rangle=M_{4|0},

1077: \label{B17}

1078: \eeq

1079: \beq

1080: n\langle

1081: (\mathbf{g}\cdot\mathbf{V}_1)^2\rangle=M_{4|0}+n^{-1}M_{0|ij}M_{0|ji}+\frac{n^{-1}}{d}M_{2|0}^2,

1082: \label{B15}

1083: \eeq

1084: \beq

1085: n\langle

1086: (\mathbf{g}\cdot\mathbf{V}_1)g^2\rangle=M_{4|0}+2n^{-1}M_{0|ij}M_{0|ji}+n^{-1}\frac{d+2}{d}M_{2|0}^2,

1087: \label{B16}

1088: \eeq

1089: \beq

1090: \langle g^4\rangle=2\langle (\mathbf{g}\cdot\mathbf{V}_1)g^2\rangle.

1091: \label{B10}

1092: \eeq

1093: 

1094: Inserting Eqs.\ (\ref{B12})--(\ref{B10}) into Eq.\ (\ref{B2.3}) one

1095: gets Eq.\ (\ref{X9}). Analogously, from Eqs.\ (\ref{B2.2}) and

1096: (\ref{B2}) one gets, after some algebra, Eqs.\ (\ref{X11}) and

1097: (\ref{X12}), respectively.

1098: 

1099: 

1100: 

1101: 

1102: \begin{thebibliography}{00}

1103: 

1104: \bibitem{BP04}

1105: { Brilliantov N V and P\"oschel T} 2004 \emph{Kinetic Theory of

1106: Granular Gases} (Oxford: Oxford University Press)

1107: 

1108: \bibitem{BDS97} Brey J J,  Dufty J W and Santos A 1997 \emph{J. Stat. Phys.} \textbf{87}  1051

1109: 

1110: \bibitem{NE98}

1111:  van Noije T P C and  Ernst M H 1998 \emph{Gran. Matt.} \textbf{1}  57

1112: 

1113: \bibitem{MS00}

1114: Montanero J M and  Santos A 2000 \emph{Gran. Matt.} \textbf{2} 53

1115: 

1116: \bibitem{BP06}

1117: Brilliantov N and P\"oschel T 2006 \emph{Europhys. Lett.}

1118: \textbf{74}  424\\

1119: Erratum: 2006 \emph{Europhys. Lett.} \textbf{75} 75 188

1120: 

1121: \bibitem{BDKS98}

1122:  Brey J J, Dufty J W, Kim C S and Santos A 1998  \emph{Phys. Rev. E} \textbf{58} 4638

1123: 

1124: \bibitem{BC01}

1125: Brey J J and Cubero D 2001 \emph{Granular Gases} P\"oschel T and

1126: Luding S (eds) (Berlin: Springer) pp 59--78

1127: 

1128: 

1129: \bibitem{GD02}

1130: Garz\'o V and  Dufty J W 2002 \emph{Phys. Fluids} \textbf{14} 1476

1131: 

1132: \bibitem{GSM07}

1133:   Garz\'o V, Santos A and Montanero J M 2007 \emph{Physica A} \textbf{376}  94

1134: 

1135: 

1136: \bibitem{BCG00}

1137: Bobylev A V  Carrillo J A and  Gamba  I M 2000 \emph{J. Stat. Phys.}

1138: \textbf{98 } 743

1139: 

1140: \bibitem{CCG00}

1141:  Carrillo J A, Cercignani C and Gamba I M 2000 \emph{Phys. Rev. E} \textbf{62}  7700

1142: 

1143: \bibitem{NK00}

1144: Ben-Naim E and  Krapivsky P L 2000 \emph{Phys. Rev. E} \textbf{61}

1145: R5

1146: 

1147: \bibitem{C01}

1148: Cercignani C 2001 \emph{J. Stat. Phys.} \textbf{102}  1407

1149: 

1150: \bibitem{EB02a}

1151:  Ernst M H and Brito R 2002 \emph{Eurohys. Lett.} \textbf{58} 182

1152: 

1153: 

1154: \bibitem{EB02}

1155: Ernst M H and Brito R 2002 \emph{J. Stat. Phys.} \textbf{109} 407

1156: 

1157: \bibitem{EB02bis}

1158: Ernst M H and Brito R  2002 \emph{Phys. Rev. E} \textbf{65} 040301

1159: 

1160: \bibitem{BMP02}

1161: Baldasarri A, Marconi U M B and Puglisi A 2002 \emph{Europhys.

1162: Lett.} \textbf{58} 14

1163: 

1164: \bibitem{KN02}

1165: Ben-Naim E and  Krapivsky P L 2002 \emph{Phys. Rev. E} \textbf{66}

1166: 011309

1167: 

1168: \bibitem{KN02bis}

1169: Krapivsky P L and Ben-Naim E 2002  \emph{J. Phys. A: Math. Gen.}

1170: \textbf{35}  L147

1171: 

1172: \bibitem{NK02}

1173: Ben-Naim E and  Krapivsky P L 2002 \emph{Eur. Phys. J. E} \textbf{8}

1174: 507

1175: 

1176: \bibitem{BC02}

1177: Bobylev A V and Cercignani C 2002 \emph{J. Stat. Phys.} \textbf{106}

1178: 547

1179: 

1180: \bibitem{MP02}

1181:  Marconi U M B and Puglisi A 2002 \emph{Phys. Rev. E} \textbf{65}  051305

1182: 

1183: \bibitem{MP02bis}

1184: Marconi U M B and Puglisi A 2002 \emph{Phys. Rev. E} \textbf{66}

1185: 011301

1186: 

1187: 

1188: \bibitem{BC03}

1189: Bobylev A V and Cercignani C 2003 \emph{J. Stat. Phys.} \textbf{110}

1190: 333

1191: 

1192: \bibitem{BCT03}

1193: Bobylev A V, Cercignani C and Toscani G2003 \emph{J. Stat. Phys.}

1194: \textbf{111}  403

1195: 

1196: \bibitem{SE03}

1197: Santos A and Ernst M H 2003 \emph{Phys. Rev. E} \textbf{68} 011305

1198: 

1199: \bibitem{S03}

1200: Santos A 2003 \emph{Physica A} \textbf{321} 442

1201: 

1202: \bibitem{NK03}

1203: Ben-Naim E and  Krapivsky P L 2003 \emph{Granular Gases} P\"oschel T

1204: and Luding S (eds) (Berlin: Springer) pp 65--93

1205: 

1206: \bibitem{G03}

1207: Garz\'o V 2003 \emph{J. Stat. Phys.} \textbf{112}  657

1208: 

1209: \bibitem{BE04}

1210: Brito R and  Ernst M H 2004 \emph{Advances in Condensed Matter and

1211: Statistical Mechanics} Korutcheva E and  Cuerno R (eds (New York:

1212: Nova Science Publishers) pp 177--202

1213: 

1214: \bibitem{GA05}

1215: Garz\'o V and Astillero A 2005 \emph{J. Stat. Phys.} \textbf{118}

1216: 935

1217: 

1218: 

1219: \bibitem{BG06}

1220: Bobylev A V and  Gamba I M 2006 \emph{J. Stat. Phys.} \textbf{124}

1221: 497

1222: 

1223: \bibitem{ETB06}

1224: Ernst M H,  Trizac E and Barrat A 2006 \emph{J. Stat. Phys.}

1225: \textbf{124}  549

1226: 

1227: \bibitem{S07}

1228: Santos A 2007 \emph{Rarefied Gas Dynamics: 25th International

1229: Symposium on Rarefied Gas Dynamics} Rebrov A K and Ivanov MS (eds)

1230: (Novosibirsk: Publishing House of the Siberian Branch of the Russian

1231: Academy of Sciences) pp 143-148 and arXiv:cond-mat/0608627

1232: 

1233: 

1234: \bibitem{KSSAON05}

1235: Kohlstedt K,  Snezhko A,  Sapozhnikov M V,  Arnarsonn I S and

1236: Ben-Naim E 2005 \emph{Phys. Rev. Lett.} \textbf{95}  068001

1237: 

1238: \bibitem{TM80}

1239: Truesdell C and  Muncaster R G 1980 \emph{Fundamentals of Maxwell's

1240: Kinetic Theory of a Simple Monatomic Gas} (New York: Academic Press)

1241: 

1242: 

1243: \bibitem{GS03}

1244: Garz\'o V and Santos A 2003  \emph{Kinetic Theory of Gases in Shear

1245: Flows. Nonlinear Transport} (Dordrecht: Kluwer Academic Publishers)

1246: 

1247: 

1248: \bibitem{Maxwell}

1249: Maxwell J C 1867 \emph{Phil. Trans. Roy. Soc. (London)} \textbf{157}

1250:  49 reprinted in Brush S G 1966 \emph{Kinetic Theory, Vol 2,

1251: Irreversible Processes} (Oxford: Pergamon) pp 23--87

1252: 

1253: 

1254: 

1255: \bibitem{SGD04}

1256: Santos A,  Garz\'o V and Dufty J W 2004 \emph{Phys. Rev. E}

1257: \textbf{69}  061303

1258: 

1259: \bibitem{SG07}

1260: Santos A and Garz\'o V 2007 \textit{J. Stat. Mech.} P08021

1261: 

1262: \bibitem{SGBD93}

1263: Santos A, Garz\'o V,  Brey J J and Dufty J W 1993 \emph{Phys. Rev.

1264: Lett.} \textbf{71}  3971\\

1265: Erratum: 1994 \emph{Phys. Rev. Lett.} \textbf{72} 1392

1266: \\Santos A and  Garz\'o V 1995 \emph{Physica A} \textbf{213}  409

1267: 

1268: 

1269: \bibitem{G07}

1270: Garz\'o V 2007 \textit{J. Phys. A: Math. Theor.} 10729

1271: 

1272: 

1273: \bibitem{G06}

1274: Garz\'o V 2006 \emph{Phys. Rev. E} \textbf{73}  021304

1275: 

1276: 

1277: 

1278: 

1279: 

1280: 

1281: 

1282: 

1283: 

1284: 

1285: \end{thebibliography}

1286: \end{document}

1287: