1: \documentclass[runningheads]{cl2emult}

2: \usepackage{psfig}

3: \usepackage{psfrag}   % replace labels in figures

4: \usepackage{epsf}

5: \usepackage{makeidx}  % allows index generation

6: \usepackage{graphicx} % standard LaTeX graphics tool

7: \usepackage{subeqnar} % subnumbers individual equations

8: \usepackage{multicol} % used for the two-column index

9: \usepackage{citesort}

10: \usepackage{amssymb}

11: \usepackage{amsmath}

12: \usepackage{latexsym}

13: \usepackage{cropmark} % cropmarks for pages without

14: \usepackage{lnp}      % placeholder for figures

15: \makeindex            % used for the subject index

16: 

17: \newcommand{\euler}[1]{{\usefont{U}{eur}{m}{n}#1}}

18: \newcommand{\eulerbold}[1]{{\usefont{U}{eur}{b}{n}#1}}

19: \newcommand{\umu}{\mbox{\euler{\char22}}}

20: \newcommand{\umub}{\mbox{\eulerbold{\char22}}}

21: \begin{document}

22: 

23: \frontmatter

24: 

25: %\input{Book/Title}

26: %\cleardoublepage

27: 

28: %\markboth{Preface}{Preface}

29: %\input{Book/Preface}

30: %\clearpage

31: 

32: %\markboth{Contents}{Contents}

33: %\tableofcontents

34: \mainmatter

35: \setcounter{page}{100}

36: 

37: %\addcontentsline{toc}{part}{Kinetic Theory and Hydrodynamics}

38: %\part*{Kinetic Theory\\[0.3cm] and\\[0.3cm] Hydrodynamics}

39: %\include{Ernst/Ernst}                    %FINISHED

40: %\include{Zippelius/Zippelius}            %FINISHED (check \stackrel)

41: %\include{Brey/Brey}                      %FINISHED

42: %\include{Goldhirsch/Goldhirsch}          %FINISHED

43: %\include{Brilliantov/PP2}                 %FINISHED

44: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

45: %  \begin{document}

46: 

47: \title*{{\small In: T. P\"oschel and S. Luding (eds.), {\em Granular Gases}, Lecture Notes in Physics Vol. 564, Springer (Berlin, 2000), p. 100}\vspace*{0.5cm}\\ Granular Gases with Impact-velocity Dependent Restitution Coefficient}

48: \label{BPpage}

49: 

50: \toctitle{Granular Gases with impact-velocity dependent restitution coefficient}

51: 

52: \titlerunning{Granular Gases with impact-velocity dependent restitution coefficient}

53: 

54: \author{Nikolai~V.~Brilliantov\inst{1}$^,$\inst{2} \and Thorsten P\"oschel\inst{2}}

55: 

56: \authorrunning{N.\,V.~Brilliantov \and T.~P\"oschel}

57: \tocauthor{N.\,V.~Brilliantov, T.~P\"oschel}

58: 

59: \institute{Moscow State University, Physics Department, Moscow 119899, Russia.\\ e-mail:~nbrillia@physik.hu-berlin.de

60:   \and Humboldt-Universit\"at, Institut f\"ur Physik,

61:   Invalidenstr. 110, D-10115 Berlin, Germany. e-mail:~thorsten@physik.hu-berlin.de, http://summa.physik.hu-berlin.de/$\sim$thorsten}

62: 

63:   \maketitle              % typesets the title of the contribution

64: 

65: \begin{abstract}

66:    We consider collisional models for granular particles and analyze the 

67:    conditions under which the restitution coefficient might be a constant.

68:    We show that these conditions are not consistent with known collision 

69:    laws. From the generalization of the Hertz contact  law for 

70:    viscoelastic particles we obtain the coefficient of normal restitution 

71:    $\epsilon$ as a function of the normal component of the impact velocity

72:    $v_{\rm imp}$. Using $\epsilon(v_{\rm imp})$ we describe the 

73:    time evolution of temperature and of the velocity distribution function of 

74:    a granular gas in the homogeneous cooling regime, 

75:    where the particles collide according to the viscoelastic law. We show that for the studied

76:    systems the simple scaling hypothesis for the velocity distribution 

77:    function is violated, i.e. that its evolution is not determined only by 

78:    the time dependence of the thermal velocity. We observe, that the deviation 

79:    from the Maxwellian distribution, which we quantify by the value of the second 

80:    coefficient of the Sonine polynomial expansion of the velocity distribution 

81:    function,  does not depend on time 

82:    monotonously.  At first stage of the evolution it increases on the mean-collision 

83:    time-scale up to a maximum value and then decays to zero at the second stage,

84:    on the time scale corresponding to the evolution of the granular gas 

85:    temperature. For granular gas in the homogeneous cooling regime we also 

86:    evaluate the time-dependent self-diffusion coefficient of 

87:    granular particles. We analyze the time dependence of the mean-square displacement 

88:    and discuss its impact on clustering. Finally, we discuss 

89:    the problem of the relevant internal time for the systems of interest.  

90: \end{abstract}

91: 

92: \section{Introduction}

93: 

94: Granular gases, i.e. systems of inelastically colliding particles, are 

95: widely spread in nature. They may be exemplified by industrial dust or 

96: interterrestrial dust; the behavior of matter in planetary rings is also 

97: described in terms of the granular gas dynamics. As compared with common 

98: molecular gases, the steady removal of kinetic energy  in these systems 

99: due  to dissipative collisions causes a variety of nonequilibrium phenomena, 

100: which have been very intensively studied 

101: %experimentally (e.g. \cite{BSHPexper}), numerically 

102: (e.g.~\cite{BSHPnumer,McNamara96,GZ93,NoijeErnst:97,EsipovPoeschel:97,NoijeErnstRing,GoldshteinShapiro95,EBEuro,NEBO97,BrilliantovPoeschel:1998d,HuthmannZippelius:97}). In most of these studies the coefficient of 

103: restitution, which characterizes the energy lost in the collisions, was assumed 

104: to be constant. This approximation, although providing a considerable 

105: simplification, and allowing to understand the main effects in granular gas dynamics, is not always justified (see also the paper by Thornton in this book~\cite{ThorntonHere}). Moreover, sometimes it 

106: occurs to be too crude to describe even qualitatively the  features of  

107: granular gases. Here we discuss the properties of granular gases 

108: consisting of viscoelastically colliding particles which implies an impact-velocity dependent restitution coefficient. The results are compared with results for gases consisting of particles which interact via a constant restitution coefficient and we see that the natural assumption of viscoelasticity leads to {\em qualitative} modifications of the gas properties.

109: 

110: The following problems will be addressed:

111: \begin{enumerate}

112: \item[$\bullet$]Why does the restitution coefficient $\epsilon$ depend on the

113: impact velocity $v_{\rm imp}$?

114: 

115: \item[$\bullet$]How does it depend on the impact velocity?

116: 

117: \item[$\bullet$]What are the consequences of the dependence of $\epsilon$

118: on $v_{\rm imp}$ on the collective behavior of particles in granular

119: gases? In particular how does $\epsilon=\epsilon\left(v_{\rm imp}\right)$ influence: 

120: \begin{itemize}

121: \item the evolution of temperature with time?

122: \item the evolution of the velocity distribution function with time? 

123: \item the self-diffusion in granular gases?

124: \end{itemize}

125: \end{enumerate}

126: 

127: In what follows we will show that the dependence of the restitution 

128: coefficient on the impact velocity is a very basic property of  

129: dissipative particle collisions, whereas the assumption of a constant restitution coefficient 

130: for the collision of three-dimensional spheres may lead to a physically incorrect dependence of the dissipative force on 

131: the compression rate of the colliding particles. From the Hertz 

132: collision law and the general relation between the elastic and dissipative

133: forces we deduce the dependence of the restitution coefficient on the 

134: impact velocity, which follows purely from scaling considerations. 

135: We also give the corresponding relation obtained from rigorous theory. 

136: Using the dependence $\epsilon(v_{\rm imp})$ we derive the time 

137: dependence of the temperature, the time-evolution 

138: of the velocity distribution function and describe self-diffusion in granular gases in the homogeneous cooling regime. 

139: 

140: \section{Dependence of the restitution coefficient on the impact velocity}

141: 

142: The collision of two particles may be characterized by the compression 

143: $\xi$ and by the compression rate $\dot{\xi}$, as shown on Fig.~\ref{fig:sketch}. The 

144: compression gives rise to the elastic force $F_{\rm el}(\xi)$, while the 

145: dissipative  force $F_{\rm diss}(\xi,\dot{\xi})$ appears due to the compression 

146: rate. 

147: \begin{figure}[htbp]

148: \centerline{\psfig{figure=coll.eps,width=5.5cm}}

149:   \caption{Sketches of two colliding spheres. The compression $\xi$ is equal to $2R-|\vec{r}_1-\vec{r}_2|$, with $\vec{r}_{1/2}$ being the particles positions.

150: The compression rate is $\dot{\xi}=v_1-v_2$. For simplicity the head on 

151: collision of identical spheres is shown.}

152:   \label{fig:sketch}

153: \end{figure}

154: 

155: If the compression and the compression rate are not very large, one can 

156: assume the dependence of the elastic and dissipative 

157: force on $\xi$ and $\dot{\xi}$

158: \begin{eqnarray}

159: &&F_{\rm el}(\xi) \sim \xi^{\alpha} \\

160: &&F_{\rm diss}(\xi,\dot{\xi}) \sim \dot{\xi}^{\beta } \xi^{\gamma }\,.

161: \end{eqnarray}

162: The dimension analysis yields the following functional 

163: form for the dependence of the restitution coefficient on the impact velocity

164: \cite{rospap}: 

165: 

166: \begin{equation}

167: \epsilon(v_{\rm imp})= 

168: \epsilon \left( v_{\rm imp}^{\frac{ 2 (\gamma -\alpha)}{1+\alpha}+\beta} \right)

169: \end{equation}

170: Therefore, the condition for a constant restitution coefficient imposes the 

171: relation between the exponents $\alpha$, $\beta$ and $\gamma$~\cite{Taguchi93,Luding98}:

172: 

173: \begin{equation}

174: 2\left(\gamma -\alpha\right)+\beta\,\left ( 1+\alpha\right ) = 0\,.

175: \label{condition}

176: \end{equation}

177: For compressions which do not exceed the plasticity 

178: threshold, the particle's material behaves as a viscoelastic medium. Then 

179: it may be generally shown~\cite{BSHP1,BSHP2,remark2} that the  

180: relation 

181: 

182: \begin{equation}

183: \label{eldiss}

184: F_{\rm diss}= A\, \dot\xi \frac{\partial}{\partial{\xi}} F_{\rm el}(\xi) \, 

185: \end{equation}

186: between the elastic and dissipative force holds, independently 

187: on the shape of the bodies in contact, provided three conditions

188: are met~\cite{BrilPoshprep}: 

189: 

190: \begin{enumerate}

191: \label{item:condistions}

192: \item[(i)] The elastic components of the stress tensor 

193:   $\sigma_{\rm el}^{ik}$ depend linearly on

194:   the components of the deformation tensor $u_{ik}$~\cite{LandauLifshits}.

195: \item [(ii)] The dissipative components of the stress tensor 

196:   $\sigma_{\rm diss}^{ik}$ depend

197:   linearly on the components of the deformation rate tensor $\dot{u}_{ik}$~\cite{LandauLifshits}.

198: \item[(iii)] The conditions of quasistatic motion are provided, i.e.

199:   $\dot{\xi} \ll  c$, $\tau_{\rm vis} \ll \tau_c$ \cite{BSHP1,BSHP2} 

200:   (here $c$ is the

201:   speed of sound in the material of particles and $\tau_{\rm vis}$ is the 

202:   relaxation time of viscous processes in its bulk).

203: 

204: \end{enumerate}

205: The constant $A$ in Eq.~(\ref{eldiss}) reads~\cite{BSHP1,BSHP2}

206: %??? check Spahn

207: \begin{equation}

208: \label{A}

209: A=\frac13\, \frac{(3\eta_2-\eta_1)^2}{(3\eta_2+2\eta_1)}

210: \left[\frac{(1-\nu^2)(1-2\nu)}{Y \nu^2}\right]\,.

211: \end{equation} 

212: where $Y$ and $\nu$ are respectively the Young modulus and the

213: Poisson ratio of the particle material and the viscous constants 

214: $\eta_1$, $\eta_2$ relate (linearly) the dissipative stress

215: tensor $\sigma_{\rm diss}^{ik}$ to the deformation rate tensor 

216: $\dot{u}_{ik}$ \cite{BSHP1,BSHP2,LandauLifshits}.

217: 

218: \ From Eq.~(\ref{eldiss}) follows that 

219: \begin{equation}

220: \beta=1 \qquad \gamma = \alpha -1 \,.

221: \label{1conditionres}

222: \end{equation}

223: 

224: 

225: Consider now a collision of three-dimensional spherical particles of 

226: radii $R_1$ and $R_2$. The Hertz contact contact law gives for the 

227: elastic force~\cite{Hertz.Brill}

228: \begin{equation}

229: \label{Hertzlaw}

230: F_{\rm el} =\rho \, \xi^{3/2} \,,

231: \qquad \rho \equiv \frac{2Y}{3(1-\nu^2)}\sqrt{R^{\rm eff}}\,,

232: \end{equation}

233: where $R^{\rm eff} \equiv R_1R_2/(R_1+R_2)$. With the set of exponents, $\alpha =3/2$, $\beta=1$ and $\gamma = 1/2$, 

234: which generally follows from the basic laws of the viscoelastic collision, 

235: the condition for the constant restitution coefficient, Eq.~(\ref{condition}), is 

236: obviously not satisfied. For spherical particles the restitution coefficient could be constant 

237: only for $\gamma = 1/4$; this, however, is not consistent with the collision laws. 

238: Instead one obtains the functional dependence

239: %\vspace*{-0.3cm}

240: \begin{equation}

241: \epsilon=\epsilon \left( v_{\rm imp}^{1/5} \right) \,.

242: \label{eps15}

243: \end{equation}

244: Note that this conclusion comes from the general analysis of viscoelastic 

245: collisions with no other assumptions needed. Therefore, the dependence of the restitution coefficient on the impact velocity, Eq.~(\ref{eps15}), is a natural property, provided the assumption on viscoelasticity holds true which is the case in a wide range of impact velocities (see discussion on page \pageref{item:condistions}). We want to mention that the functional dependence Eq.~(\ref{eps15}) was already given in \cite{Kuwabara} using heuristic arguments. 

246: 

247: 

248: Rigorous calculations~\cite{TomThor} yield for the dependence of the 

249: restitution coefficient on the impact velocity:

250: \begin{equation}

251: \label{epsC1C2}

252: \epsilon=1-C_1 A \kappa^{2/5}v_{\rm imp}^{1/5}

253: +C_2A^2 \kappa^{4/5}v_{\rm imp}^{2/5}  \mp \cdots

254: \end{equation}

255: with 

256: \begin{equation}

257: \kappa = \left( \frac32 \right)^{3/2}  

258: \frac{Y\sqrt{R^{\rm eff}}}{m^{\rm eff} (1-\nu^2)}

259: \end{equation}

260: where $m^{\rm eff}=m_1m_2/(m_1+m_2)$ ($m_{1/2}$ are the masses of the colliding 

261: particles). Numerical values for the constants $C_1$ and 

262: $C_2$ obtained in~\cite{TomThor} may be also written in a more convenient 

263: form~\cite{rospap}:

264: 

265: \begin{equation}

266: \label{C1C2}

267: C_1= \frac{ \Gamma(3/5)\sqrt{\pi}}{2^{1/5}5^{2/5} \Gamma(21/10)} =

268: 1.15344, \qquad C_2=\frac{3}{5}C_1^2\, .

269: \end{equation}

270: Although the next-order coefficients of the above expansion $C_3=-0.483582$, 

271: $C_4=0.285279$, are now available~\cite{rospap}, we assume that 

272: the dissipative constant  $A$ is small enough to ignore these high-order 

273: terms. (For large $A$ a very accurate Pad\'e approximation for 

274: $\epsilon(v_{\rm imp})$ has been proposed recently~\cite{rospap}). 

275: 

276: \section{Time-evolution of temperature and of the velocity distribution function}

277: 

278: We consider a granular gas composed of $N$ identical particles 

279: confined in a volume $\Omega$. The particles are assumed to be smooth 

280: spheres, so that the collision properties are determined by the normal component of the relative 

281: motion only. The gas is supposed to be dilute enough so that one can 

282: assume binary collisions (i.e. neglect multiple collisions) and ignore the collision duration as compared with the mean free time in between successive collisions. 

283: We assume that the initial velocities of the 

284: particles (more precisely the temperature, which we define below) are not 

285: very large to assure viscoelastic properties of the collisions, i.e. to avoid plastic deformations and fragmentation. 

286: The final velocities are assumed not to be very small which allows to neglect surface forces as adhesion and others. Under these restrictions one can apply the 

287: viscoelastic collision model. Furthermore, we assume that dissipation is not 

288: large, so that the second-order expansion  (\ref{epsC1C2}) for 

289: $\epsilon(v_{\rm imp})$ describes the collisions accurately. We analyze the 

290: granular gas in the regime of homogeneous cooling, i.e. in the 

291: pre-clustering regime, when the gas is homogeneously distributed in space. 

292: 

293: The impact velocity $v_{\rm imp}$ of colliding smooth spheres, 

294: which determines the value of the restitution coefficient according to 

295: Eq.~(\ref{epsC1C2}),  is given by the normal component of the relative velocity

296: \begin{equation}

297: v_{\rm imp}=\left|\vec{v}_{12} \cdot \vec{e} \right|~~~~\mbox{with}~~~~ \vec{v}_{12}=\vec{v}_1-\vec{v}_2\,.

298: \end{equation}

299: The unit vector $\vec{e} = \vec{r}_{12}/ |\vec{r}_{12}|$ gives the 

300: direction of the intercenter vector $\vec{r}_{12}=\vec{r}_{1}-\vec{r}_{2}$ at 

301: the instant of the collision. 

302: 

303: The evolution of the granular gas proceeds by elementary collision events in 

304: which the pre-collisional velocities of colliding particles $\vec{v}_1$,  

305: $\vec{v}_2$, are converted into after-collisional ones, $\vec{v}_1^*$, $\vec{v}_2^*$, 

306: according to the rules

307: \begin{equation}

308: \label{directcoll}

309: \begin{split}

310: %\label{v1v2v1*v2*}

311: \vec{v}_1^*&=\vec{v}_1-

312: \frac12 \left[1+\epsilon(|\vec{v}_{12} \cdot \vec{e} | )\right] 

313: (\vec{v}_{12} \cdot \vec{e})\vec{e} \\

314: \vec{v}_2^*&=\vec{v}_2+ \frac12  

315: \left[1+\epsilon(|\vec{v}_{12} \cdot \vec{e} | )\right] 

316: (\vec{v}_{12} \cdot \vec{e})\vec{e} \,,

317: \end{split}

318: \end{equation}

319: where $\epsilon$ depends on the impact velocity 

320: $v_{\rm imp}=|\vec{v}_{12} \cdot \vec{e} |$. Due to the {\em direct} collision 

321: (\ref{directcoll}) the population in the velocity phase-space near the points 

322: $\vec{v}_1$,  $\vec{v}_2$ decreases, while near the points $\vec{v}_1^*$, 

323: $\vec{v}_2^*$ it increases. The decrease  of the population near 

324: $\vec{v}_1$,  $\vec{v}_2$ caused by the direct collision, is (partly) 

325: counterbalanced by its increase in the {\em inverse} collision, where 

326: the after-collisional velocities are $\vec{v}_1$,  $\vec{v}_2$ with 

327: the pre-collisional ones $\vec{v}_1^{**}$, $\vec{v}_2^{**}$. The 

328: rules for the inverse collision read

329: \begin{equation}

330: \label{inversecoll}

331: \begin{split}

332: %\label{v1v2v1*v2*}

333: \vec{v}_1&=\vec{v}_1^{**}-

334: \frac12 \left[ 1+\epsilon(|\vec{v}_{12}^{**} \cdot \vec{e} |) \right]

335: \left( \vec{v}_{12}^{**} \cdot \vec{e} \right)\vec{e} \\

336: \vec{v}_2&=\vec{v}_2^{**}+

337: \frac12 \left[ 1+\epsilon(|\vec{v}_{12}^{**} \cdot \vec{e} |) \right]

338: \left( \vec{v}_{12}^{**} \cdot \vec{e} \right)\vec{e}\,.

339: \end{split}

340: \end{equation}

341: Note that in contrast to the case of $\epsilon ={\rm const}$, the 

342: restitution coefficients in the inverse and in the direct collisions

343: are different.  

344: 

345: The Enskog-Boltzmann equation describes the evolution of the population of particles in the phase space on the mean-field level.

346: The evolution is characterized by the distribution 

347: function $f(\vec{r},\vec{v},t)$, which for the force-free case does not 

348: depend on $\vec{r}$ and obeys the equation~\cite{NoijeErnst:97,resibua}

349: \begin{multline}

350: \label{collint} 

351: \frac{\partial}{\partial t}f\left(\vec{v}_1,t\right) =

352: g_2(\sigma)\sigma^2 \int d \vec{v}_2 \int d\vec{e}

353: \Theta(-\vec{v}_{12} \cdot \vec{e}) |\vec{v}_{12} \cdot \vec{e}|\\

354: \times \left\{\chi f(\vec{v}_1^{**},t)f(\vec{v}_2^{**},t)-

355: f(\vec{v}_1,t)f(\vec{v}_2,t) \right\}  \equiv g_2(\sigma)I(f,f)

356: \end{multline}

357: where $\sigma$ is the diameter of the particles.

358: The contact value of the pair distribution function~\cite{CarnahanStarling}

359: \begin{equation}

360: g_2(\sigma)=(2-\eta)/2(1-\eta)^3\,,

361: \end{equation}

362: accounts on the mean-field level for the increasing frequency of collisions due to 

363: excluded volume effects with $\eta=\frac16\, \pi n \sigma^3$ being the 

364: volume fraction. 

365: 

366: The first term in the curled brackets in the right-hand side of 

367: Eq.~(\ref{collint}) refers to the ``gain'' term for the population in the 

368: phase-space near  the point $\vec{v}_1$, while the second one is the 

369: ``loss'' term. The Heaviside function $\Theta(-\vec{v}_{12} \cdot \vec{e})$ discriminates approaching particles (which do collide) from separating particles (which do not collide), and

370: $|\vec{v}_{12} \cdot \vec{e}|$ gives the length of the collision cylinder.

371: Integration in Eq.~(\ref{collint}) is performed over all 

372: velocities $\vec{v}_2$ and interparticle vectors  $\vec{e}$ in the direct 

373: collision. Equation~(\ref{collint}) accounts also 

374: for the inverse collisions via the factor $\chi$, which appears due to 

375: the Jacobian of the transformation 

376: $\vec{v}_1^{**},\vec{v}_2^{**} \to  \vec{v}_1,\vec{v}_2$, and due to the 

377: difference between the lengths of the collision cylinders of  the 

378: direct and the inverse collision:

379: \begin{equation}

380: \chi=

381: \frac{{\cal D}(\vec{v}_1^{**},\vec{v}_2^{**} )}{{\cal D}(\vec{v}_1,\vec{v}_2)}

382: \, 

383: \frac{|\vec{v}_{12}^{**} \cdot \vec{e}|}{|\vec{v}_{12} \cdot \vec{e}|}\,.

384: \end{equation}

385: 

386: For constant restitution coefficient the factor $\chi$ is a 

387: constant

388: \begin{equation}

389: \chi = \frac{1}{\epsilon^2} = {\rm const} \, , 

390: \end{equation}

391: while for $\epsilon=\epsilon\left(v_{\rm imp}\right)$, as given in Eq.~(\ref{epsC1C2}), it reads~\cite{BPMaxw}

392: \begin{equation}

393: \label{chi}

394: \chi = 1 + \frac{11}{5}C_1 A \kappa^{2/5} |\vec{v}_{12} \cdot \vec{e}|^{1/5}

395: +\frac{66}{25}C_1^2 A^2 \kappa^{4/5} |\vec{v}_{12} \cdot \vec{e}|^{2/5} 

396: +\cdots

397: \end{equation}

398: \ From Eq.~(\ref{chi}) it follows that $\chi =\chi(|\vec{v}_{12} \cdot \vec{e}|)$. 

399: Since the average velocity in granular gases changes with time, such a

400: dependence of $\chi$ means, as we will show below, that $\chi$ and, therefore, the velocity distribution function itself depend explicitly on time.

401: The time dependence of $\chi$ changes drastically the properties of the collision integral 

402: and destroys the simple scaling form of the velocity distribution function, 

403: which holds for the case of the constant restitution coefficient 

404: (e.g.~\cite{EsipovPoeschel:97,NoijeErnst:97}). 

405: 

406: Nevertheless, some important properties of the collision integral are preserved. Namely, it may be shown that the relation 

407: \begin{eqnarray}

408: \label{deraver} 

409: &&\!\!\!\!\!\!\frac{d}{dt} \left< \psi(t) \right> =\int d\vec{v}_1 \psi (\vec{v}_1)

410: \frac{\partial}{\partial t} f(\vec{v}_1, t) = g_2(\sigma)

411: \int d\vec{v}_1 \psi (\vec{v}_1) I(f,f) =\\

412: &&\!\!\!\!\!\!\frac{g_2(\sigma)\sigma^2}{2} \int d\vec{v}_1d\vec{v}_2 \int d\vec{e}\,

413: \Theta(-\vec{v}_{12} \cdot \vec{e}) |\vec{v}_{12} \cdot \vec{e}|

414: f(\vec{v}_1, t)f(\vec{v}_2, t) \Delta \left[ \psi( \vec{v}_1)+

415: \psi( \vec{v}_2) \right] \nonumber 

416: \end{eqnarray}

417: holds true, where 

418: \begin{equation}

419: \left< \psi(t) \right> \equiv \int d\vec{v} \psi (\vec{v}) f(\vec{v}, t)

420: \end{equation}

421: is the average of some function $\psi (\vec{v})$, and 

422: \begin{equation}

423: \Delta \psi(\vec{v}_i) \equiv \left[\psi(\vec{v}_i^*)-\psi(\vec{v}_i) \right]

424: \end{equation}

425: denotes the change of $\psi( \vec{v}_i)$ in a direct collision. 

426: 

427: Now we introduce the temperature of the three-dimensional granular gas,

428: \begin{equation}

429: \label{deftemp1}

430: \frac{3}{2} n T(t)=\int d \vec{v} \frac{m v^2}{2} f(\vec{v},t)\,,

431: \end{equation}

432: where $n$ is the number density of granular particles 

433: ($n=N/ \Omega$), and the characteristic velocity $v_0^2(t)$ is related to temperature via

434: \begin{equation}

435: \label{deftemp}

436: T(t)=\frac12 mv_0^2(t)\,.

437: \end{equation}

438: First we try the scaling ansatz

439: \begin{equation}

440: \label{veldis}

441: f(\vec{v}, t)=\frac{n}{v_0^3(t)} \tilde{f}(\vec{c})

442: \end{equation}

443: where 

444: $\vec{c} \equiv \vec{v}/v_0(t)$ and following 

445: \cite{NoijeErnst:97,GoldshteinShapiro95} assume that deviations from the 

446: Maxwellian distribution are not large, so that  $\tilde{f}(\vec{c})$ may 

447: be expanded into a convergent series with the leading term being the 

448: Maxwellian distribution $\phi(c) \equiv \pi^{-3/2} \exp(-c^2)$.  

449: It is convenient to use the Sonine polynomials expansion 

450: \cite{NoijeErnst:97,GoldshteinShapiro95}

451: \begin{equation}

452: \label{Soninexp}

453: \tilde{f}(\vec{c})=\phi(c) \left\{1 + \sum_{p=1}^{\infty} a_p S_p\left(c^2\right) \right\}

454: \, .

455: \end{equation}

456: These polynomials are orthogonal, i.e.

457: \begin{equation}

458: \label{Soninortog}

459: \int d \vec{c}\, \phi (c) S_p(c^2)S_{p^{\prime}}\!\left(c^2\right) 

460: = \delta_{pp^{\prime}}{\cal N}_p\,,

461: \end{equation}

462: where $\delta_{pp^{\prime}}$ is the Kronecker delta and  ${\cal N}_p$ is the 

463: normalization constant. The first few polynomials  read

464: \begin{equation}

465: \label{Soninfewfirst}

466: \begin{split}

467: S_0(x)&=1 \\

468: S_1(x)&=-x^2 +\frac32 \\

469: S_2(x)&=\frac{x^2}{2}-\frac{5x}{2}+\frac{15}{8}\,.

470: \end{split}

471: \end{equation}

472: Writing the Enskog-Boltzmann equation in terms of the scaling variable

473: $\vec{c}_1$, one observes that the factor $\chi$ may not be expressed 

474: only in terms of the scaling variable, but it depends also on the 

475: characteristic velocity $v_0(t)$, and thus depends on time. Therefore, the 

476: collision integral also occurs to be time-dependent. As a result, it is not 

477: possible to reduce 

478: the Enskog-Boltzmann equation to a pair of equations, 

479: one for the time evolution of the temperature and another for the 

480: time-independent scaling function, whereas for $\epsilon=$const. the Boltzmann-Enskog equation is separable, e.g.~\cite{NoijeErnst:97,GoldshteinShapiro95,EsipovPoeschel:97}. Formally adopting the approach of Refs.~\cite{NoijeErnst:97,GoldshteinShapiro95} for $\epsilon=$const., one would obtain  

481: time-dependent coefficients $a_p$ of the Sonine polynomials expansion. 

482: This means that the simple scaling ansatz (\ref{veldis}) is violated 

483: for the case of the impact-velocity dependent restitution coefficient. 

484: 

485: Thus, it seems natural to write the distribution function in the following 

486: general form  

487: 

488: \begin{equation}

489: \label{genscal}

490: f(\vec{v}, t)=\frac{n}{v_0^3(t)}\tilde{f}(\vec{c}, t)

491: \end{equation}

492: with 

493: \begin{equation}

494: \label{genSoninexp}

495: \tilde{f}(\vec{c})

496: =\phi(c) \left\{1 + \sum_{p=1}^{\infty} a_p(t) S_p(c^2) \right\}

497: \end{equation}

498: and find then equations for the {\em time-dependent} coefficients $a_p(t)$. 

499: Substituting (\ref{genscal}) into the Boltzmann equation (\ref{collint})

500: we obtain

501: \begin{equation}

502: \label{geneqveldis}

503: \frac{\mu_2}{3} 

504: \left(3 + c_1 \frac{\partial}{\partial c_1} \right) \tilde{f}(\vec{c}, t) +

505: B^{-1} \frac{\partial}{\partial t} \tilde{f}(\vec{c}, t) =

506: \tilde{I}\left( \tilde{f}, \tilde{f} \right)

507: \end{equation}

508: with

509: \begin{equation}

510: B=B(t) \equiv v_0(t) g_2(\sigma) \sigma^2 n\,.

511: \end{equation}

512: We define the 

513: dimensionless collision integral:

514: \begin{multline}

515: \label{dimlcolint}

516: \tilde{I}\left( \tilde{f}, \tilde{f} \right)=\\

517: \int d \vec{c}_2 \int d\vec{e}

518: \Theta(-\vec{c}_{12} \cdot \vec{e}) |\vec{c}_{12} \cdot \vec{e}|

519: \left\{\tilde{\chi} \tilde{f}(\vec{c}_1^{**},t) \tilde{f}(\vec{c}_2^{**},t)-

520: \tilde{f}(\vec{c}_1,t)\tilde{f}(\vec{c}_2,t) \right\} 

521: \end{multline}

522: with the reduced factor $\tilde{\chi}$

523: \begin{equation}

524: \label{chiscal}

525: \tilde{\chi} = 1 + 

526: \frac{11}{5}C_1 \delta^{\prime} |\vec{c}_{12} \cdot \vec{e}|^{1/5}

527: +\frac{66}{25}C_1^2 \delta^{\prime \, 2} |\vec{c}_{12} \cdot \vec{e}|^{2/5}

528: +\cdots

529: \end{equation}

530: which depends now on time via a quantity

531: \begin{equation}

532: \label{deltaprime}

533: \delta^{\, \prime} (t) \equiv A \kappa^{2/5} \left[2T(t)\right]^{1/10} 

534: \equiv \delta \left[2T(t)/T_0 \right]^{1/10}\,. 

535: \end{equation}

536: Here $\delta \equiv A \kappa^{2/5}[T_0]^{1/10}$, $T_0$ is the initial 

537: temperature, and for simplicity we assume the unit mass, $m=1$. 

538: We also define the moments of the dimensionless collision integral

539: \begin{equation}

540: \label{mup}

541: \mu_p \equiv - \int d \vec{c}_1 c_1^p 

542: \tilde{I}\left( \tilde{f}, \tilde{f} \right)\ , 

543: \end{equation}

544: so that the second moment describes the rate of the temperature change:

545: \begin{equation}

546: \label{dTdt}

547: \frac{dT}{dt} =-\frac23 BT\mu_2\,.

548: \end{equation} 

549: Equation~(\ref{dTdt}) follows from the definitions of the temperature and of the 

550: moment $\mu_2$. Note that these moments depend on time, in contrast to the 

551: case of the constant restitution coefficient, where these moments are time-independent~\cite{NoijeErnst:97}. 

552: 

553: Multiplying both sides of Eq.~(\ref{geneqveldis}) with $c_1^p$ and integrating 

554: over $d \vec{c}_1$, we obtain

555: \begin{equation}

556: \label{momeq}

557: \frac{\mu_2}{3} p \left< c^p \right> -B^{-1}\sum_{k=1}^{\infty} 

558: \dot{a}_k \nu_{kp} = \mu_p

559: \end{equation}

560: where integration by parts has been performed and we define

561: \begin{eqnarray}

562: \label{nukp}

563: \nu_{kp} &\equiv& \int \phi(c) c^p S_k(c^2) d\vec{c} \\

564: \left< c^p \right>  &\equiv& \int c^p \tilde{f}(\vec{c}, t)  d\vec{c}\,.

565: \label{cpdef}

566: \end{eqnarray}

567: The calculation of $\nu_{kp}$ is straightforward; the first few of these  read:

568: $\nu_{22}=0$, $\nu_{24}=\frac{15}{4}$. The odd moments 

569: $\left< c^{2n+1} \right> $ vanish, while the even ones 

570: $\left< c^{2n} \right> $ may be expressed in terms of $a_k$ with 

571: $0 \leq k  \leq n$, namely, 

572: $\left< c^2 \right>  =\frac32 - \frac32 a_1$. On the other hand, from the

573: definition of temperature and of the thermal velocity in Eqs.~(\ref{deftemp}) and (\ref{deftemp1})

574: follows that $\left< c^2 \right> = \frac32$ and thus, $a_1=0$. 

575: Similar considerations yield 

576: $\left< c^4 \right> = \frac{15}{4}\left( 1 + a_2 \right)$. 

577: The moments $\mu_p$ may be expressed in terms of coefficients 

578: $a_2, a_3, \cdots$ too; therefore, the system Eq.~(\ref{momeq}) is an infinite (but closed) set of equations for these 

579: coefficients. 

580: 

581: It is not possible to get a general solution of the problem. However, since

582: the dissipative parameter $\delta$ is supposed to be small, the deviations from the 

583: Maxwellian distribution are presumably small too. Thus, we assume that one can neglect all  

584: high-order terms with $p>2$ in the expansion (\ref{genSoninexp}). Then Eq.~(\ref{momeq}) 

585: is an equation for the coefficient $a_2$. For $p=2$ Eq.~(\ref{momeq}) converts into an identity,

586: since $\left< c^2 \right>=\frac32$, $a_1=0$, $\nu_{22}=0$ and 

587: $\nu_{24}=\frac{15}{4}$. For $p=4$ we obtain

588: \begin{equation}

589: \label{eqa2}

590: \dot{a}_2-\frac43\, B\mu_2 \left(1+a_2 \right)+\frac{4}{15}B\mu_4 =0\,.

591: \end{equation} 

592: In Eq.~(\ref{eqa2}) $B$ 

593: depends on time as 

594: \begin{equation}

595: B(t)=(8 \pi)^{-1/2} \tau_c(0)^{-1}[T(t)/T_0]^{1/2}\,,

596: \end{equation}

597: where $\tau_c(0)$ is related to the initial mean-collision time, 

598: \begin{equation}

599: \tau_c(0)^{-1}=4 \pi^{1/2}g_2(\sigma) \sigma^2 n T_0^{1/2}\,.

600: \end{equation}

601: The time evolution of the temperature is determined by Eq.~(\ref{dTdt}), i.e. by the 

602: time dependence of $\mu_2$. 

603: 

604: The time-dependent coefficients $\mu_p(t)$ may be expressed in terms of $a_2$ owing to 

605: their definition Eq.~(\ref{mup}) and the approximation $\tilde{f}= \phi (c) [ 1+a_2(t)S_2(c^2)]$. 

606: We finally obtain:

607: \begin{multline}

608: \label{mupa2}

609: \mu_p=-\frac12 \int d\vec{c}_1\int d\vec{c}_2 \int d\vec{e} 

610: \Theta(-\vec{c}_{12} \cdot \vec{e}) |\vec{c}_{12} \cdot \vec{e}| \phi(c_1) \phi(c_2)\\ \times 

611: \left\{1+a_2\left[S_2(c_1^2)+S_2(c_2^2) \right] + a_2^2\,S_2(c_1^2)S_2(c_2^2) \right\}

612: \Delta (c_1^p+c_2^p) 

613: \end{multline}

614: with the definition of $\Delta (c_1^p+c_2^p)$ given above. Calculations 

615: performed up to the second order in terms of the dissipative parameter $\delta$ 

616: yield~\cite{BPMaxw}:

617: \begin{equation}

618: \label{mu2A}

619: \mu_2= \sum_{k=0}^{2} \sum_{n=0}^2 {\cal A}_{kn} \delta^{\, \prime \,k } a_2^n

620: \end{equation} 

621: where

622: \begin{align}

623: \label{A1A6}

624: {\cal A}_{00}&=0; \quad &{\cal A}_{01}&=0; \quad &{\cal A}_{02}&=0 \nonumber\\

625: {\cal A}_{10}&=\omega_0; \quad &{\cal A}_{11}&=\frac{6}{25}\omega_0; \quad &{\cal A}_{12}&=\frac{21}{2500} \omega_0 \\

626: {\cal A}_{20}&=-\omega_1; \quad & {\cal A}_{21}&=-\frac{119}{400}\omega_1; \, \quad &{\cal A}_{22}&=-\frac{4641}{640000}\omega_1\nonumber

627: \end{align}

628: with 

629: \begin{eqnarray}

630: \omega_0 &\equiv& 2 \sqrt{2 \pi} 2^{1/10} \Gamma \left (\frac{21}{10} \right)C_1=6.48562 \ldots\\

631: \omega_1 &\equiv& \sqrt{2 \pi} 2^{1/5} \Gamma \left (\frac{16}{5} \right)C_1^2=9.28569 \ldots

632: \end{eqnarray}

633: Similarly  

634: \begin{equation}

635: \label{mu4A}

636: \mu_4= \sum_{k=0}^{2} \sum_{n=0}^2 {\cal B}_{kn} \delta^{\, \prime \,k } a_2^n

637: \end{equation} 

638: with

639: 

640: \begin{align}

641: %\label{B1B3}

642: {\cal B}_{00}&=0; \quad &{\cal B}_{01}&=4\sqrt{2 \pi}; \quad &{\cal B}_{02}&=\frac{1}{8}\sqrt{2 \pi} \nonumber \\

643: \label{B4B6}

644: {\cal B}_{10}&= \frac{56}{10}\omega_0; \quad  &{\cal B}_{11}&=\frac{1806}{250}\omega_0; \quad &{\cal B}_{12}&=\frac{567}{12500}\omega_0 \\

645: %\label{B7B9}

646: {\cal B}_{20}&= -\frac{77}{10}\omega_1;\quad  &{\cal B}_{21}&=-\frac{149054}{13750}\omega_1;\quad &{\cal B}_{22}&=-\frac{348424}{5500000}\omega_1\nonumber

647: \end{align}

648: 

649: Thus, Eqs.~(\ref{dTdt}) and (\ref{eqa2}), together with Eqs.~(\ref{mu2A}) and  

650: (\ref{mu4A}) form a  closed set to find the time evolution of the temperature and 

651: the coefficient $a_2$. We want to stress an important difference for the time evolution 

652: of temperature for the case of the impact-velocity dependent restitution coefficient, 

653: as compared to that of a constant restitution coefficient. 

654: In the former case it is coupled to the time evolution of the coefficient $a_2$, while  

655: in the latter case there is no such coupling since $a_2={\rm const}$. 

656: This coupling may lead in to a rather peculiar time-dependence of 

657: the temperature. 

658: 

659: Introducing  the reduced temperature $u(t) \equiv T(t)/T_0$ we recast the set of equations

660: (\ref{dTdt}) and (\ref{eqa2}) into the following form:

661: \begin{multline}

662: \label{genseteq1}

663: \dot{u}+\tau_0^{-1}u^{8/5}\left( \frac53 +\frac25 a_2+\frac{7}{500}a_2^2 \right)\\

664: -\tau_0^{-1}q_1 \delta \, u^{17/10}

665: \left( \frac53 +\frac{119}{240}a_2 +\frac{1547}{128000}a_2^2 \right) =0

666: \end{multline}

667: \begin{equation}

668: \label{genseteq2}

669: \dot{a}_2-r_0u^{1/2}\mu_2 \left(1 +a_2 \right) + \frac15 r_0u^{1/2}\mu_4=0\,,

670: \end{equation}

671: where we introduce the characteristic time  

672: \begin{equation}

673: \label{tau0}

674: \tau_0^{-1}=\frac{16}{5} q_0 \delta \cdot \tau_c(0)^{-1}

675: \end{equation} 

676: with

677: \begin{eqnarray}

678: q_0&=&2^{1/5}\Gamma(21/10)C_1/8=5^{-2/5}\sqrt{\pi}\Gamma(3/5)/8=0.173318\ldots \\

679: r_0 &\equiv& \frac{2}{3 \sqrt{2 \pi}} \tau_c(0)^{-1}\\

680: q_1 &\equiv& 2^{1/10} (\omega_1/\omega_0) =1.53445 \ldots

681: \end{eqnarray}

682: As shown below the characteristic time $\tau_0$ describes the time evolution of the temperature.

683: To obtain these equations we use the 

684: expressions for $\mu_2(t)$, $B(t)$, and for the coefficients ${\cal A}_{nk}$. Note that the 

685: characteristic time $\tau_0$ is $\delta^{-1} \gg 1$ times larger than the collision 

686: time $ \sim \tau_c(0)$. 

687: 

688: We will find the  solution to these equations as expansions in terms of the small 

689: dissipative parameter $\delta$ ($\delta^{\, \prime}(t) = \delta \cdot 2^{1/10}u^{1/10}(t)$):

690: \begin{eqnarray}

691: \label{expudel}

692: &&u=u_0+ \delta \cdot u_1 +\delta^2 \cdot u_2 +\cdots \\

693: \label{expadel}

694: &&a_2=a_{20}+\delta \cdot a_{21}+\delta^2 \cdot a_{22} +\cdots

695: \end{eqnarray} 

696: Substituting Eqs.~(\ref{mu2A},\ref{mu4A},\ref{expudel},\ref{expadel}) into 

697: Eqs.~(\ref{genseteq1},\ref{genseteq2}), one can solve these equations perturbatively, 

698: for each order of $\delta$. The solution of the order of ${\cal O}(1)$ reads for

699: the coefficient $a_2(t)$ \cite{BPMaxw}:

700: \begin{equation}

701: \label{a20tsmall}

702: a_{20}(t) \approx a_{20}(0)e^{-4t\left/\left(5 \tau_{\rm E}(0)\right)\right.}\,,

703: \end{equation} 

704: where $\tau_{\rm E} =\frac32 \tau_c$ is the Enskog relaxation time, so that 

705: $a_{20}(t)$ vanishes for $t \sim \tau_0$. This refers to the relaxation of an initially non-Maxwellian velocity distribution to the Maxwellian distribution. Note that the relaxation occurs within few collisions per particle, similarly to the relaxation of common molecular gases. 

706: 

707: We now assume that the initial distribution is Maxwellian, i.e., that $a_{20}(0)=0$ for 

708: $t=0$. Then the deviation from the Maxwellian distribution originates from the 

709: inelasticity of the interparticle collisions. For the case of  $a_{20}(0)=0$ the 

710: solution of the order of ${\cal O}(1)$ for the reduced temperature reads

711: \begin{equation}

712: \label{T(t)del1}

713: \frac{T(t)}{T_0}=u_0(t)= \left(1 + \frac{t}{\tau_0} \right)^{-5/3}\,,

714: \end{equation} 

715: which coincides with the time-dependence of the temperature obtained previously 

716: using scaling arguments~\cite{TomThor} (up to a constant $\tau_0$ which may 

717: not be determined by scaling arguments). 

718: 

719: The solution for $a_2(t)$ in linear approximation with respect to $\delta$ reads

720: \begin{equation}

721: \label{a21gensolLi}

722: a_2(t)=\delta \cdot a_{21}(t)=-\frac{12}{5}w(t)^{-1}

723: \left\{ {\rm Li} \left[ w(t) \right]-{\rm Li} \left[ w(0) \right] \right\}

724: \end{equation} 

725: where 

726: \begin{equation}

727: \label{w(t)}

728: w(t) \equiv \exp \left[ \left(q_0 \delta \right)^{-1} \left(1+t/\tau_0 \right)^{1/6} \right]\,.

729: \end{equation} 

730: and with the logarithmic Integral

731: \begin{equation}

732: {\rm Li}(x)=\int\limits_0^x\frac{1}{\ln(t)} dt\,.

733: \end{equation}

734: For $t \ll \tau_0$ the coefficient $a_2(t)$ 

735: (\ref{a21gensolLi}) reduces to 

736: \begin{equation}

737: \label{a21tsmallsol}

738: a_{2}(t)=

739: - \delta \cdot h \left( 1- e^{-4t\left/\left(5\tau_{\rm E}(0)\right)\right.} \right)

740: \end{equation} 

741: where 

742: \begin{equation}

743: h \equiv 2^{1/10} \left( {\cal B}_{10}-5{\cal A}_{10}\right)/16 \pi 

744: =(3/10)\Gamma(21/10)2^{1/5}C_1=0.415964\,.

745: \end{equation}

746: As it follows from Eq.~(\ref{a21tsmallsol}), after a transient 

747: time of the order of few collisions per particle, i.e. for 

748: $\tau_{\rm E}(0) < t \ll \tau_0$, $a_{2}(t)$ saturates at the 

749: ``steady-state''- value $-h\, \delta=-0.415964\,\delta$, i.e. it changes only slowly on the time-scale $\sim \tau_c(0)$. On the other hand, for $t \gg \tau_0$ one obtains

750: \begin{equation}

751: \label{a21tlargesol}

752: a_{2}(t) \simeq

753: - \delta \cdot h \left( t/\tau_0  \right)^{-1/6}

754: \end{equation} 

755: so that $a_{2}(t)$ decays to zero on a 

756: time-scale $\sim \tau_0$, i.e. slowly in the collisional time-scale 

757: $\sim \tau_c(0) \ll \tau_0$. The velocity distribution thus tends asymptotically 

758: to the Maxwellian distribution. For this regime the first-order correction 

759: for the reduced temperature, $u_1(t)$, reads~\cite{BPMaxw}:

760: \begin{equation}

761: \label{u1asympsol}

762: u_1(t)=\left( \frac{12}{25}h+2q_1 \right)(t/\tau_0)^{-11/6}=

763: 3.26856\,(t/\tau_0)^{-11/6}\,,

764: \end{equation}

765: where we used the above results for the constants $h$ and $q_1$. From the last 

766: equation one can see  how the coupling between the temperature evolution and the 

767: evolution of the velocity distribution influences the evolution of temperature. Indeed, if there 

768: were  no such coupling, there would be no coupling term in Eq.~(\ref{genseteq1}), and 

769: thus no contribution from $\frac{12}{25}h$ to the prefactor of $u_1(t)$ in Eq.~(\ref{u1asympsol}). This would noticeably change  the time behavior of $u_1(t)$. 

770: On the other hand, the leading term in the time dependence of temperature, 

771: $u_0(t)$, is not affected by this kind of coupling. 

772: 

773: In Fig.~\ref{fig:a2} and Fig.~\ref{fig:T} we show the time dependence of the coefficient 

774: $a_2(t)$ of the Sonine polynomial expansion and of the temperature of the 

775: granular gas. The analytical findings are compared with the numerical 

776: solution of the system (\ref{genseteq1},\ref{genseteq2}). As one can see from the 

777: figures the analytical theory reproduces fairly well the 

778: numerical solution for the case of small $\delta$. 

779: \begin{figure}[htbp]

780: \centerline{\psfig{figure=linLi.eps,width=5.5cm}\psfig{figure=linLilong.eps,width=5.5cm}}

781: \centerline{\psfig{figure=a2scale.eps,width=5.5cm}}

782:   \caption{Time dependence of the second coefficient of the Sonine polynomial 

783: expansion $a_2(t)$. Time is given in units of the mean collisional time 

784: $\tau_c(0)$. (Left): $a_2 \times 1000$ (solid lines) for 

785: $\delta =0.001, 0.005, 0.01, 0.015$ 

786: %???? Please, check the last number, 0.015  !!!!!!!!!} 

787: (top to bottom) together with the linear approximation (dashed lines); (Right): the same as (left) but for larger times; (Middle): $-a_2(t)$ over time (log-scale) for $\delta =0.03, 0.01, 0.003, 0.001$ (top to bottom) together with the power-law asymptotics $\sim t^{-1/6}$.}

788:   \label{fig:a2}

789: \end{figure}

790: %Fig.2    %%%%% Yours Fig.1 -> Fig.2a, Fig.2 -> Fig.2b, Fig.9 -> Fig.2c %%%%%%

791: 

792: \begin{figure}[htbp]

793: \centerline{\psfig{figure=TdecayA.eps,width=5.5cm}\psfig{figure=TdecayB.eps,width=5.5cm}}

794: 

795: \centerline{\psfig{figure=TdecayAlog.eps,width=5.5cm}}  

796:   \caption{Time-evolution of the reduced temperature, $u(t)=T(t)/T_0$. 

797: The time is given in units of mean collisional time 

798: $\tau_c(0)$. Solid line: numerical solution, short-dashed:

799: $u_0(t)=(1+t/\tau_0)^{-5/3}$ (zero-order theory), 

800: %??? {\bf Please, check the power {-5/3} !!!!! in your simulations !!!!!!!!!}

801: long-dashed: $u(t)=u_0(t)+\delta \, u_1(t)$ (first-order theory). 

802: (Left): for $\delta=0.05, 0.1$ (top to bottom); (Right): $\delta=0.15, 0.25$ (top to bottom); 

803: (Middle): the same as (Left) but log-scale and larger ranges.}

804:   \label{fig:T}

805: \end{figure}

806: %Fig.3   %%%%%%%% Yours Fig.4 -> Fig.3a, Fig.5 -> Fig.3b, Fig.6 ->Fig.3c  %%%%%%%%%

807: %%%

808: 

809: As it follows from Fig.~\ref{fig:a2} (where the time is given  in collisional units), 

810: for small $\delta$ the following scenario of evolution of the velocity 

811: distribution takes place for a force-free granular gas. 

812: The initial Maxwellian distribution evolves to a non-Maxwellian distribution, with 

813: the discrepancy between these two characterized by the second coefficient of the 

814: Sonine polynomials expansion $a_2$. The deviation from the Maxwellian distribution 

815: (described by $a_2$) quickly grows, until it saturates after a few collisions per 

816: particle at a ``steady-state'' value. At this instant the deviation from the Maxwellian 

817: distribution is maximal, with the value $a_2 \approx - 0.4 \delta$ (Fig.~\ref{fig:a2}a). 

818: This refers to the first ``fast'' stage of the evolution, which takes place on a 

819: mean-collision time-scale $\sim \tau_c(0)$. After this maximal deviation is reached, 

820: the second ``slow'' stage of the evolution starts. At this stage $a_2$ decays to zero on 

821: the ``slow'' time scale $\tau_0 \sim \delta^{-1} \tau_c(0) \gg \tau_0(0)$, which 

822: corresponds to the time  scale of the temperature evolution (Fig.~\ref{fig:a2}b); the decay of the 

823: coefficient $a_2(t)$ in this regime occurs according to a power law $ \sim t^{-1/6}$ 

824: (Fig.~\ref{fig:a2}c). Asymptotically the 

825: Maxwellian distribution would be achieved, if the clustering process did not occur. 

826: 

827: Fig.~\ref{fig:T} illustrates the significance of the first-order correction $u_1(t)$ in the time-evolution 

828: of temperature. This becomes more important as the dissipation parameter $\delta$ 

829: grows (Figs.~\ref{fig:T}a,b). At large times the results of the first-order theory 

830: (with $u_1(t)$ included) practically coincide with the numerical results, while 

831: zero-order theory (without $u_1(t)$) demonstrates noticeable deviations (Fig.~\ref{fig:T}c). 

832: 

833: For larger values of $\delta$ the linear theory breaks down. Unfortunately, the 

834: equations obtained for the second order approximation ${\cal O}(\delta^2)$ are 

835: too complicated to be treated analytically. Hence, we studied them only numerically

836: (see Fig.~\ref{fig:a2time}). As compared to the case of small $\delta$, an additional intermediate regime in 

837: the time-evolution of the velocity distribution is observed. The first ``fast'' stage 

838: of evolution takes place, as before, on the time scale of few collisions per particle,

839: where maximal deviation from the Maxwellian distribution is achieved (Fig.~\ref{fig:a2time}). For 

840: $\delta \geq 0.15$ these maximal values of $a_2$ are  positive. Then, on the second 

841: stage (intermediate regime), which continues $10-100$ collisions, $a_2$ changes its 

842: sign and reaches a maximal negative deviation. Finally, on the third, slow stage, 

843: $a_2(t)$ relaxes to zero on the slow time-scale $\sim \tau_0$, just as for 

844: small $\delta$. In Fig.~\ref{fig:a2time} we show the first stage of the 

845: time evolution of $a_2(t)$ for systems  with large $\delta$. At a certain value of the dissipative parameter $\delta$ the behavior changes qualitatively, i.e. the system then reveals another time scale as discussed above. 

846: \begin{figure}[htbp]

847: \centerline{\psfig{figure=a2.01-02.eps,width=6cm}}  

848:   \caption{Time dependence of the second coefficient of the Sonine polynomial expansion $a_2(t) \times 100 $.

849: %??? {\bf Please, check the factor 100, my be this should be 10 as far as I can remember, that I saw.} 

850: Time is given in units of mean collisional time $\tau_c(0)$. $\delta=0.1, 0.11, 0.12, \ldots, 0.20$ (bottom to top).}

851:   \label{fig:a2time}

852: \end{figure}

853: %Fig.4   %%%%% Yours Fig.10 -> Fig.4 %%%%%%

854: %%

855: 

856: Figure~\ref{fig:a2evo} shows the numerical solution of Eqs.~(\ref{genseteq1}) and (\ref{genseteq2}) for the second Sonine coefficient $a_2(t)$ as a function of time. One can clearly distinguish the different stages of evolution of the velocity distribution function. A more detailed investigation of the evolution of the distribution function for larger dissipation is subject of present research~\cite{BPMaxw}.

857: 

858: \begin{figure}[htbp]

859: \centerline{\psfig{figure=a2.60.eps,width=5cm}\psfig{figure=a2.1200.eps,width=5cm}}

860: \centerline{\psfig{figure=a2.12000.eps,width=5cm}\psfig{figure=a2.1200000.eps,width=5cm}}

861:   \caption{The second Sonine coefficient $a_2$  for $\delta=0.16$ over time. The numerical solutions of Eqs.~(\ref{genseteq1}) and (\ref{genseteq2}) show all stages of evolution discussed in the text.}

862:   \label{fig:a2evo}

863: \end{figure}

864: 

865: The interesting property of the granular gases in the regime of homogeneous 

866: cooling is the overpopulation of the high-velocity tails in the velocity 

867: distribution~\cite{EsipovPoeschel:97}, which has been shown for granular gases consisting of particles which interact via a constant restitution coefficient,

868: $\epsilon ={\rm const}$. How does the velocity dependence of the 

869: restitution coefficient as it appears for viscoelastic spheres influence this effect?  We observe, that for the 

870: case of $\epsilon=\epsilon(v_{\rm imp})$ the functional form (i.e. the exponential 

871: overpopulation~\cite{EsipovPoeschel:97}) persists, but it decreases with time on the 

872: ``slow'' time-scale $\sim \tau_0$. Namely we obtain for the velocity distribution 

873: for $c \gg 1$ \cite{BPMaxw}:

874: \begin{equation}

875: \label{veldiscgg1}

876: \tilde{f}(\vec{c}, t) \sim 

877: \exp \left[ - \frac{b}{ \delta}\,c\left(1+\frac{t}{\tau_0} \right)^{1/6} \right]\, .

878: \end{equation}

879: where 

880: $b=\sqrt{\pi/2}\left(16q_0/5\right)^{-1}=2.25978\ldots$, which holds for 

881: $t \gg \tau_c(0)$.  

882: Again we see that the distribution tends asymptotically to the Maxwellian distribution, 

883: since the overpopulation vanishes as $t \to \infty$. 

884: 

885: Using the temperature and the velocity distribution of a granular gas as were derived in this section, one can 

886: calculate the kinetic coefficients. In the next section we consider the simplest one --

887: the self-diffusion coefficient. 

888: 

889: \section{Self-diffusion in granular gases of viscoelastic particles}

890: 

891: In the simplest case diffusion of particles occurs when there are 

892: density gradients in the system. 

893: The diffusion coefficient $D$ relates the flux of particles $\vec{J} $ to the 

894: density gradient $\nabla n$ according to a linear relation, provided the gradients are 

895: not too large:

896: \begin{equation}

897: \label{Jdiff}

898: \vec{J}= -D \nabla n\,.

899: \end{equation}

900: The coefficient $D$ also describes the statistical average of the  migration of a single particle. For {\em equilibrium} 

901: 3D-systems the mean-square displacement of a particle reads

902: \begin{equation}

903: \left\langle \left( \Delta r(t) \right)^2  \right\rangle_{\rm eq} 

904: = 6\,D\,t \,,

905: \label{Dgen}

906: \end{equation}

907: where $\left< \cdots \right>_{\rm eq}$ denotes the {\em equilibrium}

908: ensemble averaging. For {\em nonequilibrium} systems, such as granular gases, 

909: one should consider the time-dependent diffusion coefficient $D(t)$ and the corresponding 

910: generalization of Eq.~(\ref{Dgen}):

911: \begin{equation}

912: \left\langle \left( \Delta r(t) \right)^2  \right\rangle =

913: 6\, \int^t D(t^{\prime}) dt^{\prime}\,,

914: \label{Dgengen}

915: \end{equation}

916: where $\langle \cdots \rangle$ denotes averaging over the nonequilibrium ensemble. 

917: If the migration of a particle occurs in a uniform system composed of particles of the 

918: same kind, this process is called ``self-diffusion''. Correspondingly, the kinetic 

919: coefficient $D$ is called self-diffusion coefficient. 

920: 

921: To find  the mean-square  displacement, one writes 

922: \begin{equation}

923: \left\langle \left( \Delta r(t) \right)^2  \right\rangle =

924: \left< \int_0^{t}\vec{v}(t^{\prime})dt^{\prime}

925: \int_0^t  \vec{v}(t^{\prime \prime}) dt^{\prime \prime}  \right> \, 

926: \label{delR}

927: \end{equation}

928: and encounters then with the velocity autocorrelation function

929: 

930: $$

931: K_v(t^{\prime}, t) 

932: \equiv \left< \vec{v}(t^{\prime}) \vec{v}(t^{\prime \prime}) \right> 

933: $$

934: which should be evaluated in order to obtain the mean-square displacement and 

935: the self-diffusion coefficient. 

936: 

937: To calculate $K_v(t^{\prime}, t)$ we use the approximation of uncorrelated 

938: successive binary collision, which is valid for moderately dense systems, and 

939: an approach based on the formalism of the pseudo-Liouville operator 

940: ${\cal L}$~\cite{pseudo}. The pseudo-Liouville operator is defined as

941: \begin{equation}

942: i{\cal L}=\sum_j \vec{v}_j \cdot \frac{\partial}{\partial \vec{r}_j}

943: +\sum_{i<j}\, \hat{T}_{\mbox{\footnotesize\em ij}}\,.

944: \label{L}

945: \end{equation}

946: The first sum in (\ref{L}) refers to the free streaming of the

947: particles (the ideal part) while the second sum refers to the particle

948: interactions which are described by the binary collision

949: operators~\cite{Chandler}

950: \begin{equation}

951: \hat{T}_{\mbox{\footnotesize\em ij}}\!=\sigma^{2}\!\! \int\!\! d\vec{e}\, 

952: \Theta \left(- \vec{v}_{\mbox{\footnotesize\em ij}} \cdot \vec{e}\, \right)\!

953: |\vec{v}_{\mbox{\footnotesize\em ij}} \cdot \vec{e}\, | 

954: \delta\! \left( \vec{r}_{\mbox{\footnotesize\em ij}}- \sigma \vec{e} 

955: \right)\!\!

956: \left(\hat{b}_{\mbox{\footnotesize\em ij}}^{\vec{e}}-1 \right)  \,,

957: \label{Tij}

958: \end{equation}

959: where $\Theta(x)$ is the Heaviside function.  The operator

960: $\hat{b}_{\mbox{\footnotesize\em ij}}^{{\vec{e}}}$ is defined as

961: \begin{equation}

962: \hat{b}_{\mbox{\footnotesize\em ij}}^{\vec{e}} f \left (\vec{r}_{i},

963:   \vec{r}_{j}, \vec{v}_{i},\vec{v}_{j} \cdots \right)=f \left

964:   (\vec{r}_{i}, \vec{r}_{j},

965:   \vec{v}^{*}_{i},\vec{v}^{*}_{j} \cdots \right) \, , 

966: \end{equation}

967: where $f$ is some function of the dynamical variables and $\vec{v}^{*}_{i}$ and $\vec{v}^{*}_{j}$ are the postcollisional velocities from Eq.~(\ref{directcoll}). 

968: The pseudo-Liouville operator gives the time derivative of any

969: dynamical variable $B$ (e.g.~\cite{resibua}):

970: \begin{equation}

971: \frac{d}{dt} B\left( \left\{ \vec{r}_i, \vec{v}_i  \right\}, t \right)= 

972: i{\cal L}\, B\left( \left\{     \vec{r}_i, \vec{v}_i \right\}, t \right)\,.

973: \label{derA}

974: \end{equation}

975: Therefore, the  time evolution of $B$

976: reads ($t>t^{\prime}$)

977: \begin{equation}

978: B\left( \{ \vec{r}_i, \vec{v}_i  \}, t \right)=

979: e^{i{\cal L} (t-t^{\,\prime} )} 

980: B\left( \{     \vec{r}_i, \vec{v}_i \}, t^{\,\prime} \right)\,.

981: \label{evolA}

982: \end{equation}

983: With Eq.~(\ref{evolA}) 

984: the time-correlation function reads

985: \begin{equation}

986: \left< B(t^{\prime})B(t) \right >=

987: \int d\Gamma \rho(t^{\prime}) B(t^{\prime}) e^{i{\cal L} (t-t^{\prime})}

988: B(t^{\prime})\,,

989: \label{evolAA}

990: \end{equation}

991: where $\int d\Gamma$ denotes integration over all degrees of freedom

992: and $\rho(t^{\prime})$ depends on temperature $T$, density $n$, etc.,

993: which change on a time-scale $t \gg \tau_c$.

994: 

995: Now we assume that

996: \begin{enumerate}

997: \item[(i)] the coordinate part and the velocity part of the distribution 

998:    function $\rho(t)$ factorize, and 

999: \item [(ii)] the molecular chaos hypothesis is valid.

1000: \end{enumerate}

1001: This suggests the following form of the distribution function:

1002: \begin{equation}

1003: \rho(t)=\rho(\vec{r}_1, \ldots, \vec{r}_N)\cdot f(\vec{v}_1,t) \ldots f(\vec{v}_N,t)\,.

1004: \label{molchaos}

1005: \end{equation}

1006: In accordance with the

1007: molecular chaos assumption  the sequence of the

1008: successive collisions occurs without correlations. If the variable $B$

1009: does not depend on the positions of the particles, its

1010: time-correlation function decays exponentially~\cite{Chandler1}:

1011: 

1012: \begin{equation}

1013: \left\langle B(t^{\prime})B(t) \right \rangle =

1014: \left< B^2 \right>_{t^{\prime}} e^{-\left.\left|t-t^\prime\right|\right/\tau_B(t^\prime)}

1015: ~~~~\left(t > t^{\prime}\right) \,.

1016: \label{AAexp}

1017: \end{equation}

1018: where $\langle \cdots \rangle_{t^\prime}$ denotes the averaging with

1019: the distribution function taken at time $t^{\prime}$.  The relaxation

1020: time $\tau_B$ is inverse to the initial slope of the autocorrelation

1021: function~\cite{Chandler1}, as it may be found from the time derivative

1022: of $\left\langle B(t^{\prime} )B(t) \right \rangle$ taken at

1023: $t=t^{\prime}$. Equations~(\ref{evolAA}) and (\ref{AAexp}) then yield

1024: \begin{equation}

1025: -\tau_B^{-1}(t^{\prime})=\int\!\!d\Gamma \rho(t^{\prime}) B {i{\cal L} }B /

1026: \left\langle B^2 \right \rangle_{t^{\prime}}= 

1027: \frac{\left\langle B i{\cal L}  B \right 

1028: \rangle_{t^{\prime}}}{\left\langle B^2 \right \rangle_{t^{\prime}}}\,.

1029: \label{ALA}

1030: \end{equation}

1031: The relaxation time $\tau_B^{-1}(t^{\prime})$ depends on time via

1032: the distribution function $\rho(t^{\prime})$ and varies on the

1033: time-scale $t \gg \tau_c$.

1034: 

1035: Let $B(t)$ be the velocity of some particle, say $\vec{v}_1(t)$.  Then

1036: with $3T(t)=\left\langle v^2 \right \rangle_t$, 

1037: Eqs.~(\ref{AAexp}) and (\ref{ALA}) (with Eqs.~(\ref{L}) and (\ref{Tij})) read~\cite{BrilliantovPoeschel:1998d}

1038: \begin{equation}

1039: \left\langle \vec{v}_1 (t^{\prime})\cdot  \vec{v}_1(t)\right\rangle =

1040: 3T(t^{\prime}) e^{-|t-t^{\prime}|/\tau_v(t^{\prime})}

1041: \label{vvexp}

1042: \end{equation} 

1043: \begin{equation}

1044: -\tau_v^{-1}(t^\prime)=

1045: (N-1) \frac{\left< \vec{v}_1 \cdot \hat{T}_{12} 

1046: \vec{v}_1\right>_{t^{\prime}}}{\left< \vec{v}_1  \cdot \vec{v}_1 \right>_{t^{\prime}}}\,.

1047: \label{vTv}

1048: \end{equation} 

1049: To obtain Eq.~(\ref{vTv}) we take into account that ${\cal L}_0

1050: \vec{v}_1=0$, $\hat{T}_{\mbox{\footnotesize\em ij}}\,\vec{v}_1=0$ (for $i\neq 1$) and the identity of the particles. 

1051: 

1052: Straightforward calculation yields for the case of a constant 

1053: restitution coefficient:

1054: \begin{equation}

1055: \tau_v^{-1}(t)=\frac{\epsilon +1}{2}\frac83 n \sigma^2 g_2(\sigma) 

1056: \sqrt{\pi T(t)} =\frac{\epsilon +1}{2}  \tau_E^{-1}(t)\,, 

1057: \label{tEt}

1058: \end{equation}

1059: where $\tau_E(t)=\frac32\,\tau_c(t)$ is the Enskog relaxation time

1060: \cite{resibua}. Note that according to Eq.~(\ref{tEt}),   

1061: $\tau_v =\frac{2}{1+\epsilon} \tau_E > \tau_E$, i.e., 

1062: the velocity correlation time for inelastic collisions exceeds that of

1063: elastic collisions. This follows from partial suppression of the

1064: backscattering of particles due to inelastic losses in their normal

1065: relative motion, which, thus, leads to more stretched particle trajectories, 

1066: as compared to the elastic case.

1067: 

1068: Similar (although somewhat more complicated) computations may be 

1069: performed for the system of viscoelastic 

1070: particles yielding

1071: \begin{equation}

1072: \tau_v^{-1}(t)= 

1073: \tau_E^{-1}(t) \left[1+\frac{3}{16}a_2(t)-4q_0 \, \delta \,u^{1/10}(t) \right]\,,

1074: \label{tauva}

1075: \end{equation} 

1076: where $q_0=0.173318$ has been already introduced and $a_2(t)$, $u(t)$ are the 

1077: same as defined above.

1078: To obtain Eq.~(\ref{tauva}) we neglect terms of the order 

1079: of ${\cal O}(a_2^2)$, 

1080: ${\cal O}(\delta^2)$ and ${\cal O}(a_2\, \delta)$. 

1081: 

1082: Using the velocity correlation function one writes

1083: \begin{equation}

1084: \left\langle \left( \Delta r(t) \right)^2  \right\rangle 

1085: =2 \int_0^t dt^\prime 3T (t^\prime) \int_{t^\prime}^t 

1086: dt^{\prime\prime} e^{-|t^{\prime\prime}-t^\prime|/\tau_v(t^\prime)}\,.

1087: \label{Difvel}

1088: \end{equation}

1089: On the short-time scale $t \sim \tau_c$, $T (t^\prime)$ and

1090: $\tau_v(t^\prime)$ may be considered as constants. Integrating in

1091: (\ref{Difvel}) over $t^{\prime\prime}$ and equating with

1092: (\ref{Dgengen}) yields for $t \gg \tau_c \sim \tau_v$ the

1093: time-dependent self-diffusion coefficient

1094: \begin{equation}

1095: D(t)= T(t) \tau_v(t)\,.

1096: \label{Difviatau}

1097: \end{equation}

1098: 

1099: Substituting the dependencies for $u(t)=T(t)/T_0$ and $a_2(t)$ as functions of time, which has been derived in 

1100: the previous section, we obtain the time dependence of the coefficient of 

1101: self-diffusion $D(t)$. For $t \gg \tau_0$ this may be given in an explicit form:

1102: \begin{equation}

1103: \frac{D(t)}{D_0} \simeq \left(\frac{t}{\tau_0} \right)^{-5/6} \!\!

1104: +\delta \left(4q_0+q_1+\frac{21}{400}h \right) \left(\frac{t}{\tau_0} \right)^{-1}\,,

1105: \label{Diffinal}

1106: \end{equation} 

1107: where the constants $q_0$, $q_1$ and $h$ are given above. Hence,

1108: the prefactor in the term proportional to $\delta$ reads 

1109: $\left(4q_0+q_1+\frac{21}{400}h \right)=2.24956$, and $D_0$ is the initial Enskog value 

1110: of the self-diffusion coefficient

1111: \begin{equation}

1112: D_0^{-1} =  \frac83 \, \pi^{1/2} n g_2(\sigma) \sigma^2 T_0^{-1/2}\,.

1113: \end{equation}  

1114: Correspondingly, the mean-square displacement reads

1115: asymptotically for $t \gg \tau_0 $:

1116: \begin{equation}

1117: \left\langle \left( \Delta r(t) \right)^2  \right\rangle  \sim 

1118: t^{1/6}+ b \, \delta\, \log t + \ldots\,,

1119: \label{dRasym}

1120: \end{equation}

1121: where $b$ is some constant.

1122: This dependence holds true for times 

1123: \begin{equation}

1124: \tau_c(0)\,\delta^{-1} \ll t \ll \tau_c(0)\, \delta^{-11/5}\,.

1125: \label{ineq}

1126: \end{equation}

1127: The

1128: first inequality in Eq.~(\ref{ineq}) follows from the condition $\tau_0 \ll t$, while the

1129: second one follows from the condition  $\tau_c(t) \ll

1130: \tau_0$, which means that temperature changes are slow on the collisional time-scale. For the constant restitution coefficient one obtains

1131: \begin{equation}

1132: T(t)/T_0=\left[ 1+\gamma_0 t/\tau_c(0) \right]^{-2}\,,  

1133: \end{equation}

1134: where $\gamma_0 \equiv \left(1-\epsilon^2\right)/6$ \cite{GZ93,NEBO97}. Thus, using 

1135: Eqs.~(\ref{tEt}) and  (\ref{Difviatau}) one obtains for the

1136: mean-square displacement in this case

1137: \begin{equation}

1138: \label{drcons}

1139: \left< \left( \Delta r(t) \right)^2

1140: \right>  \sim \log t \, .

1141: \end{equation}

1142: As it follows from Eqs.~(\ref{dRasym}) and (\ref{drcons}) the impact-velocity 

1143: dependent restitution coefficient, Eq.~(\ref{epsC1C2}), leads to a significant 

1144: change of the long-time behavior of the mean-square displacement of 

1145: particles in the laboratory-time. Compared to its logarithmically 

1146: weak dependence for the constant restitution coefficient, the 

1147: impact-velocity dependence of the restitution coefficient

1148: gives rise to a considerably faster increase of this quantity with time, 

1149: according to a power law. 

1150: 

1151: One can also compare the dynamics of the system in its 

1152: inherent-time scale. First we consider the average cumulative number of 

1153: collisions per particle ${\cal N}(t)$ as an inherent measure for time 

1154: (e.g.~\cite{McNamara96,EBEuro}). 

1155:  It may be found by integrating 

1156: $d{\cal N}=\tau_c(t)^{-1}dt$ \cite{NEBO97}. For a constant restitution 

1157: coefficient $\epsilon$ one obtains ${\cal N}(t) \sim \log t$, while for the impact-velocity 

1158: dependent $\epsilon\left( v_{\rm imp} \right)$ one has ${\cal N}(t) \sim t^{1/6}$. Therefore, the temperature and the mean-square

1159: displacement behave in these cases as\\

1160: 

1161: \begin{center}

1162: %\begin{table}

1163: %\centering

1164: %\caption{xx}

1165: \renewcommand{\arraystretch}{1.4}

1166: \setlength\tabcolsep{5pt}

1167: \begin{tabular}{l|l}

1168: \hline

1169: $\epsilon={\rm const}$ & $\epsilon=\epsilon\left( v_{\rm imp} \right)$\\ 

1170: \hline\noalign{\smallskip}                   

1171: $T({\cal N}) \sim e^{-2(1-\epsilon^2){\cal N}}$ & $T({\cal N}) \sim {\cal N}^{-10} $ \\ 

1172: \noalign{\smallskip}

1173: \hline

1174: \noalign{\smallskip}                

1175:  $\left< \left( \Delta r({\cal N}) \right)^2\right>  \sim {\cal N}$ & $\left< \left( \Delta r({\cal N}) \right)^2\right>  \sim {\cal N}$ \\

1176: \noalign{\smallskip}                

1177: \hline

1178: \end{tabular}

1179: %\label{Tab1a}

1180: %\end{table}

1181: \end{center}

1182: \vspace{0.1cm}

1183: 

1184:  If the number of collisions per particle ${\cal N}(t)$ would be the relevant

1185: quantity specifying the stage of the granular gas evolution, one would

1186:  conjecture that the dynamical behavior of a granular gas with a

1187: constant $\epsilon$ and velocity-dependent $\epsilon$ are identical,

1188: provided an ${\cal N}$-based time-scale is used.  Whereas in equilibrium systems the number of collisions is certainly an appropriate measure of time, in nonequilibrium systems this value has to be treated with more care. As a trivial example may serve a particle bouncing back and forth between two walls, each time it hits a wall it loses part of its energy: If one describes this system using a ${\cal N}$-based time, one would come to the conclusion that the system conserves its energy, which is certainly not the proper description of the system. According to our

1189: understanding, therefore, the number of collision is not an appropriate time scale to describe physical reality. Sometimes, it may be even misleading.

1190: 

1191: Indeed, as it was shown in Ref.~\cite{EBEuro}, the value of

1192: ${\cal N}_c$, corresponding to a crossover from the linear regime of

1193: evolution (which refers to the homogeneous cooling state) to the nonlinear

1194: regime (when clustering starts) may differ by orders of magnitude,

1195: depending on the restitution coefficient and on the density of the granular

1196: gas.  Therefore, to analyze the behavior of a granular gas, one can try

1197: an alternative inherent time-scale, ${\cal T}^{-1} \equiv T(t)/T_0$

1198: which is based on the gas temperature. Given two

1199: systems of granular particles at the same density and the same initial

1200: temperature $T_0$, consisting of particles colliding with constant and

1201: velocity-dependent restitution coefficient, respectively, the time

1202: ${\cal T}$ allows to compare directly their evolution. A strong

1203: argument to use a temperature-based time has been given by Goldhirsch

1204: and Zanetti~\cite{GZ93} who have shown that as the temperature decays, the 

1205: evolution of the system changes from a linear regime to a nonlinear one. 

1206: Recent numerical results of Ref.~\cite{EBEuro}

1207: also support our assumption: It was shown that while ${\cal N}_c$

1208: differs by more than a factor of three for two different systems, 

1209: the values of ${\cal T}_c$, (defined, as ${\cal T}_c=T({\cal N}_c)/T_0$) are very close~\cite{EBEuro}.  These arguments show that one could consider ${\cal T}$ as a 

1210: relevant time-scale to analyze the granular gas evolution.

1211: 

1212: With the temperature decay $T({\cal N})/T_0 \sim e^{-2 \gamma_0{\cal N}}$ for a constant 

1213: restitution coefficient and $T({\cal N})/T_0 \sim {\cal N}^{-10}$ for the 

1214: impact-velocity dependent one, we obtain the following dependencies:\\

1215: \begin{center}

1216: %\begin{table}

1217: %\centering

1218: %\caption{xx}

1219: \renewcommand{\arraystretch}{1.4}

1220: \setlength\tabcolsep{5pt}

1221: \begin{tabular}{l|l}

1222: \hline\noalign{\smallskip}

1223:  $\epsilon={\rm const}$ & $\epsilon=\epsilon\left( v_{\rm imp} \right)$\\ 

1224: \noalign{\smallskip}

1225: \hline

1226: \noalign{\smallskip}                   

1227:  $T \sim \frac{1}{ {\cal T} }$ & $T \sim \frac{1}{ {\cal T} } $ \\ 

1228: \noalign{\smallskip}

1229:  $\left< \left( \Delta r({\cal T}) \right)^2\right>  \sim \log {\cal T}$ & $\left< \left( \Delta r({\cal T}) \right)^2 \right>  \sim {\cal T}^{1/10}$ \\

1230: \noalign{\smallskip}

1231: \hline

1232: \end{tabular}

1233: %\label{Tab2a}

1234: %\end{table}

1235: \end{center}

1236: \vspace{0.1cm}

1237: 

1238: This shows that in the temperature-based time-scale, in which the  cooling 

1239: of both systems is synchronized, the mean-square displacement grows logarithmically 

1240: slow for the case of constant restitution coefficient and much faster, as a power 

1241: law, for the system of viscoelastic particles with  $\epsilon = \epsilon(v_{\rm imp})$.

1242: Thus, we conclude that clustering may be retarded for the latter system. 

1243: 

1244: \section{Conclusion}

1245: In conclusion, we considered kinetic properties of granular gases composed of 

1246: viscoelastic particles, which implies the 

1247: impact-velocity dependence of the restitution coefficient. We found that 

1248: such dependence gives rise to some new effects in granular gas dynamics:

1249: (i) complicated, non-monotonous time-dependence of the coefficient $a_2$ of 

1250: the Sonine polynomial expansion, which describes the deviation of the 

1251: velocity distribution from the Maxwellian and (ii) enhanced spreading of 

1252: particles, which depends  on time as a power law, compared to a logarithmically 

1253: weak dependence  for the systems with a constant $\epsilon$. 

1254: 

1255: The Table below compares the properties of granular gases consisting of particles interacting via a constant coefficient of restitution $\epsilon ={\rm const}$ and consisting of viscoelastic particles where the collisions are described using an impact velocity dependent restitution coefficient $\epsilon = \epsilon(v_{\rm imp})$:\\

1256: \begin{center}

1257: %\begin{table}

1258: %\centering

1259: %\caption{xx}

1260: \renewcommand{\arraystretch}{1.4}

1261: \setlength\tabcolsep{5pt}

1262: \begin{tabular}{l|l}

1263: \hline\noalign{\smallskip}

1264:  $\epsilon={\rm const}$ & $\epsilon=\epsilon\left( v_{\rm imp} \right)$\\ 

1265: \noalign{\smallskip}

1266: \hline

1267: \noalign{\smallskip}                   

1268: $\epsilon $  is a model parameter & $\epsilon=1-C_1A \kappa^{2/5}v_{\rm imp}^{1/5}+\cdots $ \\ 

1269:  &$C_1=1.15396$, $C_2=\frac35 C_1^2$, $\ldots$ \\

1270:  &$\kappa=\kappa(Y, \nu, m, R)$ \\

1271:  &$A=A(\eta_1, \eta_2, Y, \nu)$ \\

1272:  &all quantities are defined via parameters \\[-0.15cm]

1273:  &of the particle material $Y$, $\nu$, $\eta_{1/2}$ \\[-0.15cm]

1274:  &and their mass and radius. \\

1275: \noalign{\smallskip}

1276: \hline

1277: \noalign{\smallskip}

1278: \multicolumn{2}{c}{\bf Small parameter} \\

1279:  $1-\epsilon^2$ -- does not depend & $\delta =A\kappa^{2/5}T_0^{1/10}$ -- depends \\[-0.15cm]

1280: on the state of the system & on the initial temperature $T_0$.  \\

1281: \noalign{\smallskip}

1282: \hline

1283: \noalign{\smallskip}

1284: \multicolumn{2}{c}{\bf Temperature}\\

1285:  $T=T_0 \left( 1+t/\tau_0^{\prime}\right)^{-2}$ & $T=T_0\left( 1+t/\tau_0\right)^{-5/3}$ \\

1286: \noalign{\smallskip}

1287: \hline

1288: \noalign{\smallskip}

1289: \multicolumn{2}{c}{\bf Velocity distribution}\\

1290:  $f(\vec{v}, t)=\frac{n}{v_0^3(t)} \tilde{f}(\vec{c})$

1291: & $f(\vec{v}, t)=\frac{n}{v_0^3(t)}\tilde{f}(\vec{c}, t) $ \\ 

1292: $\tilde{f}(\vec{c},t)=\phi(c) \left\{1 + \sum_{p=1}^{\infty} a_p S_p(c^2) \right\}$ &

1293: $\tilde{f}(\vec{c})=\phi(c) \left\{1 + \sum_{p=1}^{\infty} a_p(t) S_p(c^2) \right\}$ \\

1294:  $a_2 ={\rm const.} $ & $a_2=a_2(t)$ -- is a (complicated) \\[-0.15cm]

1295: & function of time. \\

1296: \noalign{\smallskip}

1297: \hline

1298: \noalign{\smallskip}

1299: \multicolumn{2}{c}{\bf Self-diffusion}\\

1300:  $\left< \left( \Delta r(t) \right)^2\right>  \sim \log t $ & $\left< \left( \Delta r(t) \right)^2 \right>  \sim t^{1/6} $ \\

1301: \noalign{\smallskip}

1302: \hline

1303:   \end{tabular}

1304: %\label{Tab3a}

1305: %\end{table}

1306: \end{center}

1307: \vspace{0.1cm}

1308: 

1309: 

1310: \section*{Acknowledgements}

1311: 

1312: We thank M. H. Ernst and I. Goldhirsch for valuable discussions.

1313: 

1314: \begin{thebibliography}{99}

1315: \addcontentsline{toc}{section}{References}

1316: \bibitem{BSHPnumer}

1317: %\bibitem{BSHPtheor}

1318: %\bibitem{theo1} 

1319: %J.~T.~Jenkins, {\em Rapid flow of granular materials},

1320: %in: R.~J.~Knops and A.~A.~Lacey (eds.), {\em Non--classical

1321: %continuum mechanics}, Cambridge University Press, (Cambridge, 1987);

1322: Y. Du, H. Li, and L. P. Kadanoff, Phys. Rev. Lett. {\bf 74}, 1268 (1995);

1323: T. Zhou and L. P. Kadanoff, Phys. Rev. E {\bf 54}, 623 (1996);

1324: %H. M. Jaeger, S. R. Nagel, and R. P. Behringer, Rev. Mod. Phys {\bf 68}, 1259 (1996);

1325: %S. Succi, J. Wang, and Y-H. Qian, J. Mod. Phys. C {\bf 8}, 999 (1997);

1326: %\bibitem{Kadanoff2}  

1327: %F. Bridges, K. D. Supulver, D. N. C. Lin, R. Knight, and M. Zafra, ICARUS {\bf 123}, 422 (1996);

1328: A. Goldshtein, M. Shapiro, and C. Gutfinger, J. Fluid Mech. {\bf 316}, 29 (1996);

1329: A. Goldshtein, V. N. Poturaev, and I. A. Shulyak, Izvestiya Akademii Nauk SSSR, Mechanika Zhidkosti i Gaza {\bf 2}, 166 (1990);

1330: J. T. Jenkins and M. W. Richman, Arch. Part. Mech. Materials {\bf 87}, 355 (1985);

1331: V. Buchholtz and T. P\"oschel, Granular Matter {\bf 1}, 33 (1998);

1332: E. L. Grossman, T. Zhou, and E. Ben-Naim, Phys. Rev. E {\bf 55}, 4200 (1997);

1333: %\bibitem{Ernst1} 

1334: F. Spahn, U. Schwarz, and J. Kurths, Phys. Rev. Lett. {\bf 78}, 1596 (1997);

1335: S. Luding and H. J. Herrmann, Chaos {\bf 9}, 673 (1999);

1336: S. Luding and S. McNamara, Gran. Matter {\bf 1}, 113 (1998);

1337: T. P. C. van Noije, M. H. Ernst, and R. Brito, Phys. Rev. E {\bf 57}, R4891 (1998);

1338: %\bibitem{Ernst2} 

1339: J. A. C. Orza, R. Brito, T. P. C. van Noije, and M. H. Ernst,  

1340: Int. J. Mod. Phys. C {\bf 8}, 953 (1997);

1341: %\bibitem{SelaGoldhirsh:98}

1342: N. Sela and I. Goldhirsch, J. Fluid. Mech., {\bf 361}, 41 (1998);

1343: N. Sela, I. Goldhirsch, and H. Noskowicz, Phys. Fluids {\bf 8}, 2337 (1996);

1344: I. Goldhirsch, M.-L. Tan, and G. Zanetti, J. Sci. Comp. {\bf 8}, 1 (1993);

1345: %\bibitem{Brey}

1346: T. Aspelmeier, G. Giese, and A. Zippelius, Phys. Rev. E {\bf 57}, 857 (1997);

1347: J. J. Brey and D. Cubero, Phys. Rev. E {\bf 57}, 2019 (1998);

1348: J. J. Brey, J. W. Dufty, C. S. Kim, and A. Santos, Phys. Rev. E {\bf 58}, 4648 (1998);

1349: J. J. Brey, M. J. Ruiz-Montero, and F. Moreno, Phys. Rev. E {\bf 55}, 2846 (1997);

1350: J. J. Brey, D. Cubero, and M. J. Ruiz-Montero, Phys. Rev. E {\bf 59}, 1256 (1999);

1351: %\bibitem{Kumaran99}

1352: V. Kumaran,

1353: Phys.Rev.E, {\bf 59}, 4188, (1999);

1354: %\bibitem{GarzoDufty99}

1355:   V. Garzo and J. W. Dufty, Phys. Rev. E, {\bf 59}, 5895, (1999).

1356: %%%%%%%%%%%%%%%%%%

1357: \bibitem{McNamara96}

1358:   S. McNamara and W. R. Young, Phys. Rev. E {\bf 53}, 5089  (1996).   

1359: 

1360: \bibitem{GZ93} I.~Goldhirsch and G.~Zanetti, {\em Phys.~Rev.~Lett.}, {\bf 70}, 1619 (1993).

1361: 

1362: \bibitem{NoijeErnst:97} 

1363:   T. P. C. van Noije and M. H. Ernst, 

1364:   Granular Matter, {\bf 1}, 57 (1998).

1365: %%%%%%%%%%%%%%%%%%%%%%%%%

1366: \bibitem{EsipovPoeschel:97} 

1367:   S. E. Esipov and T. P\"{o}schel, J. Stat.

1368:   Phys., {\bf 86}, 1385 (1997).

1369: \bibitem{NoijeErnstRing} 

1370:   T. P. C. van Noije, M. H. Ernst and R.Brito, 

1371:   Physica A {\bf 251}, 266 (1998).

1372: %%%%%%%%%%%%%%%%%%%%%

1373: \bibitem{GoldshteinShapiro95}

1374:   A. Goldshtein and M. Shapiro, 

1375:   J. Fluid. Mech., {\bf 282}, 75 (1995). 

1376: %%%%%%%%%%%%%%%%%%%%

1377: \bibitem{EBEuro}R. Brito and M. H. Ernst,  

1378:   Europhys. Lett. {\bf 43}, 497 (1998).

1379: \bibitem{NEBO97}T. P. C. van Noije and M. H.

1380:   Ernst, R. Brito and J. A. G. Orza, Phys. Rev. Lett. {\bf 79}, 411

1381:   (1997).  

1382: \bibitem{BrilliantovPoeschel:1998d} 

1383:   N. V. Brilliantov and T.

1384:   P\"oschel, Phys. Rev. E {\bf 61}, 1716 (2000).

1385: % Diffusion paper

1386: \bibitem{HuthmannZippelius:97} 

1387:     M. Huthmann and A. Zippelius, { Phys.

1388:     Rev. E.} {\bf 56}, R6275 (1997); S. Luding, M. Huthmann, 

1389:     S. McNamara  and A. Zippelius, { Phys.

1390:     Rev. E.} {\bf 58}, 3416 (1998). 

1391: 

1392: \bibitem{ThorntonHere} C. Thornton, {\em Contact mechanics and coefficients of restitution}, (in this volume, page 184.

1393: %\pageref{Thorntonpage}).

1394: 

1395: %\bibitem{SalazarBrenig99}

1396: %  J.M.Salazar and L.Brenig, 

1397: %  Phys.Rev.E, {\bf 59}, 2093 (1999).

1398: 

1399: 

1400: \bibitem{rospap}

1401:   R. Ram\'{\i}rez, T. P\"oschel, N. V. Brilliantov,

1402:   and T. Schwager, Phys. Rev. E {\bf 60}, 4465 (1999).

1403: 

1404: \bibitem{Taguchi93} Y. Taguchi, Europhys. Lett. {\bf 24}, 203 (1993).

1405: 

1406: \bibitem{Luding98} S. Luding, {\em Collisions and Contacts between two particles}, in: H. ~J. Herrmann, J. -P. Hovi, and S. Luding (eds.), {\em Physics of dry granular media - NATO ASI Series E350}, Kluwer (Dordrecht, 1998), p. 285. 

1407: 

1408: \bibitem{BSHP1}

1409:   N. V. Brilliantov, F. Spahn, J.-M. Hertzsch, and T.

1410:   P\"oschel, Phys. Rev. E {\bf 53}, 5382 (1996).

1411: \bibitem{BSHP2}

1412:   J.-M. Hertzsch,  F. Spahn, and N. V. Brilliantov, J. Phys. II (France), 

1413:   {\bf 5}, 1725 (1995).

1414: \bibitem{remark2}

1415:   Derivation of the dissipative force given in~\cite{BSHP1,BSHP2} for colliding 

1416:   spheres may be straightforwardly generalized to obtain the relation 

1417:   (\ref{eldiss}) (or (A17) in~\cite{BSHP1,BSHP2}) for colliding bodies of any 

1418:   shape, provided that displacement field in the bulk of the material 

1419:   of bodies in contact is a one-valued function of the compression 

1420:   (see also~\cite{BrilPoshprep}).

1421: \bibitem{BrilPoshprep}

1422:   N. V. Brilliantov and T. P\"oschel, in preparation. 

1423:   

1424: \bibitem{LandauLifshits}

1425:     L. D. Landau and E. M. Lifschitz, {\it Theory of Elasticity},

1426:     Oxford University Press (Oxford, 1965). 

1427: \bibitem{Hertz.Brill}

1428:   H. Hertz, J. f. reine u. angewandte Math. {\bf 92}, 156 (1882).

1429: 

1430: \bibitem{Kuwabara} G. Kuwabara and K. Kono, Jpn. J. Appl. Phys. {\bf 26}, 1230 (1987).

1431: 

1432: \bibitem{TomThor}

1433:   T. Schwager and T. P\"oschel, Phys. Rev. E

1434:   {\bf 57}, 650 (1998).

1435: 

1436: \bibitem{resibua} P. Resibois and M. de Leener, {\em Classical Kinetic

1437:     Theory of Fluids} (Wiley, New York, 1977).

1438: \bibitem{CarnahanStarling}

1439:   N. F. Carnahan, and K. E. Starling, J. Chem. Phys., {\bf 51}, 635 (1969). 

1440:   

1441: \bibitem{BPMaxw}

1442:  N. V. Brilliantov and T. P\"oschel, Phys. Rev. E, {\bf 61}, 5573 (2000)

1443: 

1444: \bibitem{pseudo} 

1445:   The term ``pseudo'' was initially referred to the

1446:   dynamics of systems with singular hard-core potential

1447:   \cite{Ernst:69,resibua}.

1448: \bibitem{Chandler} For application to ``ordinary'' fluids, see

1449:   \cite{Chandler1} and to granular systems

1450:   \cite{NoijeErnstRing,HuthmannZippelius:97}.  A rigorous

1451:   definition of ${\cal L}$ includes a pre-factor, preventing

1452:   successive collisions of the same pair of particles

1453:   \cite{Ernst:69,resibua} which, however, does not affect the present

1454:   analysis.

1455: 

1456: \bibitem{Chandler1} 

1457:   D. Chandler, {J. Chem. Phys.} {\bf 60}, 3500,

1458:   3508 (1974); B. J. Berne, {\em ibid} {\bf 66}, 2821, (1977); 

1459:   N. V. Brilliantov and O. P. Revokatov, 

1460:   Chem. Phys. Lett. {\bf 104}, 444 (1984).

1461: 

1462: \bibitem{Ernst:69} 

1463:   M. H. Ernst, J. R. Dorfman, W. R. Hoegy and J. M.

1464:   J. van Leeuwen, Physica A {\bf 45}, 127 (1969).

1465: 

1466: \end{thebibliography}

1467: 

1468: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1469: %%%%%%%%%%%%%    FIGURE CAPTINOS  %%%%%%%%%%%%%%%%%%%%%%%%%%%%

1470: 

1471: 

1472: 

1473: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1474: %%%%%%%%%%%%%%%%%%%%%%  END OF FIGURE CAPTINOS   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%                       

1475: 

1476: %INDEX%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1477: %\clearpage

1478: %\addcontentsline{toc}{section}{Index}

1479: %\flushbottom

1480: %\printindex

1481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1482: 

1483: %\end{document}

1484: 

1485: 

1486: 

1487: 

1488: 

1489: 

1490: 

1491: 

1492: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1493: %\include{Jenkins/Jenkins}                %FINISHED

1494: %\include{Henrique/Henrique}              %FINISHED

1495: 

1496: %\addcontentsline{toc}{part}{Collisions and One-dimensional Models}

1497: %\part*{Collisions\\[0.3cm] and\\[0.3cm] One-dimensional Models}

1498: %\include{Bridges/Bridges}                 %FINISHED

1499: %\include{Thornton/Thornton}               %FINISHED

1500: %\include{Ramirez/Ramirez}

1501: %\include{Poeschel/PoeschelBrilliantov}   %FINISHED

1502: %\addtocontents{toc}{\vfill\newpage}

1503: 

1504: %\addcontentsline{toc}{part}{Vibrated Granular Media}

1505: %\addtocontents{toc}{\vspace*{-0.07cm}}

1506: %\part*{Vibrated Granular Media}

1507: %\include{Huntley/Huntley}

1508: %\addtocontents{toc}{\vspace*{-0.07cm}}

1509: %\include{Clement/Clement}                %FINISHED

1510: %\addtocontents{toc}{\vspace*{-0.07cm}}

1511: %\include{Falcon/Falcon}                  %FINISHED

1512: %\addtocontents{toc}{\vspace*{-0.07cm}}

1513: %\include{Farkas/Farkas}                  %FINISHED

1514: %\addtocontents{toc}{\vspace*{-0.07cm}}

1515: %\include{Goldshtein/goldshtein}          %FINISHED

1516: %\addtocontents{toc}{\vspace*{-0.07cm}}

1517: 

1518: %\addcontentsline{toc}{part}{Granular Astrophysics}

1519: %\part*{Granular Astrophysics}

1520: %\addtocontents{toc}{\vspace*{-0.07cm}}

1521: %\include{Brahic/Brahic}

1522: %\addtocontents{toc}{\vspace*{-0.07cm}}

1523: %\include{Salo/salo}                          %FINISHED (confirm e-mail)

1524: %\addtocontents{toc}{\vspace*{-0.07cm}}

1525: %\include{Hanninen/Hanninen}                  %FINISHED

1526: %\addtocontents{toc}{\vspace*{-0.07cm}}

1527: %\include{Spahn/Spahn2}                       %FINISHED

1528: %\addtocontents{toc}{\vspace*{-0.07cm}}

1529: 

1530: %\addcontentsline{toc}{part}{Towards Dense Granular Systems}

1531: %\addtocontents{toc}{\vspace*{-0.07cm}}

1532: %\part*{Towards Dense Granular Systems}

1533: %\include{Luding/Luding}                      %FINISHED

1534: %\addtocontents{toc}{\vspace*{-0.07cm}}

1535: %\include{Urbach/Urbach}                      %FINISHED

1536: %\addtocontents{toc}{\vspace*{-0.07cm}}

1537: %\include{Hong/hong_new}                      %FINISHED

1538: %\addtocontents{toc}{\vspace*{-0.07cm}}

1539: %\include{Herrmann/honnef99}                  %FINISHED

1540: 

1541: %\include{Book/author_index}

1542: 

1543: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1544: %\clearpage

1545: %\addcontentsline{toc}{chapter}{Index}

1546: %\flushbottom

1547: %\printindex

1548: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1549: 

1550: \end{document}

1551: 

1552: