cond-mat0201175/cm.tex
1: %revised version: 11/07/01
2: \documentstyle[preprint,aps,pre]{revtex}
3: \begin{document}
4: \tighten
5: \title{Rheological properties in a low-density granular mixture}
6: \author{Jos\'e Mar\'{\i}a Montanero\footnote[1]{Electronic address:
7: jmm@unex.es}}
8: \address{Departamento de Electr\'onica e Ingenier\'{\i}a Electromec\'anica,\\
9: Universidad de Extremadura, E-06071 Badajoz, Spain}
10: \author{Vicente Garz\'{o}\footnote[2]{Electronic address: vicenteg@unex.es}}
11: \address{Departamento de F\'{\i}sica, Universidad de Extremadura, E-06071 \\
12: Badajoz, Spain}
13: 
14: \date{\today}
15: \maketitle
16: 
17: \begin{abstract}
18: 
19: Steady simple shear flow of a low-density binary mixture of inelastic smooth
20: hard spheres is studied in the context of the Boltzmann equation. This equation 
21: is solved by using two different and complementary approaches: a Sonine 
22: polynomial expansion and the direct simulation Monte Carlo method. The 
23: dependence of the shear and normal stresses as well as of the steady 
24: granular temperature on both the dissipation and the parameters of the 
25: mixture (ratios of masses, concentration, and sizes) is analyzed. In 
26: contrast to previous studies, the theory predicts and the simulation 
27: confirms that the partial temperatures of each species are different, even 
28: in the weak dissipation limit. In addition, the simulation shows that the 
29: theory reproduces fairly well the values of the shear stress and the 
30: phenomenon of normal stress differences. On the other hand, here
31: we are mainly interested in analyzing transport in the homogeneous
32: shear flow so that, the possible formation of particle clusters is ignored in 
33: our description.
34: 
35: Keywords: Granular mixture of gases; Simple shear flow; Kinetic theory;
36: Direct Simulation Monte Carlo Method. 
37:  
38: \end{abstract}
39: \draft
40: \vspace{1cm}
41: \pacs{PACS number(s): 45.70.Mg, 05.20.Dd, 51.10.+y, 47.50.+d}
42: 
43: \bigskip \narrowtext
44: 
45: 
46: \section{Introduction}
47: \label{sec1}
48: 
49: 
50: 
51: Many features associated with dissipation in rapid granular flows can be 
52: well represented by a fluid of hard spheres with inelastic collisions. 
53: In the simplest model the grains are taken to be smooth so that the 
54: inelasticity is characterized by means of a constant coefficient of normal 
55: restitution. The essential difference with respect to normal fluids is 
56: the absence of energy conservation, which leads to modifications 
57: of the usual hydrodynamic equations.  In recent years, the Boltzmann and 
58: Enskog equations have been generalized to account for inelastic binary  
59: collisions. These equations have been solved by means of an expansion akin 
60: to the Chapman-Enskog method up to the Navier-Stokes order and detailed 
61: expressions for the corresponding transport coefficients have been obtained 
62: \cite{BDKS98,RET99}. These expressions are not restricted to the 
63: low-dissipation limit and comparison with Monte Carlo simulations indicate 
64: that the results are very accurate, even for strong 
65: dissipation \cite{BMC99}. In the context of multicomponent granular gases, 
66: most of the existing work appears to be based on weak dissipation 
67: approximations \cite{JM89,Z95,AW98,WA99}. Given that the inelasticity is
68: small, an usual assumption in these studies is to consider
69: a single temperature variable characterizing the entire mixture.
70: However, as one of the authors pointed out\cite{Z95}, the equipartition of
71: energy is not completely justified beyond the low-dissipation limit and it
72: is necessary to offer theories involving mixtures of granular materials in
73: which the kinetic temperatures of species $T_i$ are different from the
74: mixture temperature $T$. As a matter of fact, recent experiments
75: \cite{FM02} and simulations \cite{CH02} on {\em driven} granular mixtures
76: show that the two types of grains do not attain the same granular
77: temperature. In terms of the mean square velocities of species, this
78: implies a violation of the classical equipartition theorem.
79: A related finding for a binary mixture undergoing homogeneous
80: cooling (i.e., an unforced system) has been reported by
81: Garz\'o and Dufty \cite{GD99,GD01} from a kinetic theory analysis. 
82: 
83: 
84: 
85: %However, as one of the authors pointed out
86: %\cite{Z95}, the assumption of equal temperatures is not completely justified 
87: %and it is necessary to offer theories involving mixtures of granular 
88: %materials in which the kinetic temperatures of species $T_i$ are different 
89: %from the mixture temperature $T$. Recently, a description of 
90: %hydrodynamics in a low-density binary granular mixture with a comparable 
91: %accuracy to that for the single gas (valid for arbitrary degree of 
92: %dissipation) has been done \cite{GD99,GD01}.
93: %{\tt In contrast with the common
94: %granular temperature, these theories show that in general, the mixture has 
95: %two (different) kinetic temperatures.}
96: %Within the context of the 
97: %Boltzmann equation, the corresponding expressions for the mass, heat, and 
98: %momentum fluxes are exact through the Navier-Stokes regime. 
99:  
100: 
101: All the above works refer to near equilibrium situations. Very little is 
102: known about far from equilibrium states. This is true for both molecular 
103: and granular fluids due to the intricacy of the Boltzmann and 
104: Enskog collision operators. Nevertheless, the difficulties are even harder 
105: for granular gases since gradients in the system can be controlled by 
106: dissipation in collisions and not only by the boundary and initial 
107: conditions. Thus, for instance, a granular system with uniform boundaries 
108: at constant temperature develops spatial inhomogeneities \cite{BC98}. 
109: 
110: 
111: One of the simplest far from equilibrium physical situations 
112: corresponds to the simple shear flow. Macroscopically, it is characterized 
113: by uniform density and temperature and a constant velocity profile. In the 
114: case of molecular fluids, this state is not stationary since the 
115: temperature increases monotonically in time due to viscous heating. However, 
116: for granular fluids a steady state is possible when the viscous heating is 
117: exactly balanced by the inelastic cooling. As a consequence, for a 
118: given shear rate, the temperature is a function of the restitution 
119: coefficient in the steady state. This steady state is precisely what we want 
120: to analyze here. 
121: 
122: In the case of a one-component system, the simple shear flow has been 
123: extensively studied. Thus, Lun {\em et al.}\cite{LSJC84} obtained the 
124: rheological properties of a dense gas for small inelasticity, while Jenkins 
125: and Richman\cite{JR88} used a maximum-entropy approximation to solve the 
126: Enskog equation. An extension of the Jenkins and Richman work 
127: \cite{JR88} to highly inelastic spheres has been recently 
128: made\cite{CR98,C01}. For low-density granular gases, Sela {\em et 
129: al.}\cite{SGN96} have been able to get a perturbation solution of the 
130: Boltzmann equation to third order in the shear rate, finding normal stress 
131: differences at this level of approximation. On the other hand, some 
132: progresses have been done by using model kinetic equations in the 
133: low-density limit \cite{BMM97} as well as for dense gases \cite{MGSB99}. In 
134: both works, comparison with Monte Carlo simulations shows an excellent 
135: agreement even for strong dissipation.
136: Similar studies for multicomponent systems are more scarce.
137: Most of them are based on a Navier-Stokes description of the hydrodynamic
138: fields\cite{JM89,Z95,AW98,WA99} and, therefore, they are restricted to small 
139: shear rates, which for the steady shear flow is equivalent to the 
140: low-dissipation limit. As said before, although these studies permit
141: different temperatures for the two species, they lead to equal partial
142: granular temperatures $T_i$ in the quasielastic limit.
143: A primary attempt to include temperature differences was made by
144: Jenkins and Mancini \cite{JM87}, although applications of this theory which
145: appear in the literature incorporate the assumption of equipartition of
146: energy\cite{note}. Since this assumption is not
147: completely justified \cite{GD99,MP99,BRGD99,HWRL00,SD01}
148: for highly inelastic spheres, the problem of describing
149: the simple shear flow from a {\em multi-temperature} theory is still open. 
150:  
151:  
152:  
153: The aim of this paper is to get the rheological properties of a binary granular 
154: mixture subjected to the simple shear flow in the framework of the Boltzmann 
155: equation. Two complementary routes are followed. First, the set of 
156: coupled Boltzmann equations are solved by using a first-Sonine 
157: polynomial approximation with a Gaussian measure. The main 
158: characteristic of our solution is that the reference Gaussian 
159: distributions are
160: defined in terms of the kinetic temperatures $T_i$ instead of the 
161: mixture temperature $T$. Consequently,
162: we do not assume {\em a priori} the equality of the 
163: three temperatures and the temperature ratio $T_1/T_2$ is 
164: consistently determined from the solution to the Boltzmann equations.
165: It is found that the partial temperatures of each species
166: are clearly different and so, the energy
167: is not equally distributed between both species. The consequences of this 
168: effect on the rheological properties are significant, as shown below. Once 
169: the temperature ratio is known, we get explicit expressions for the elements 
170: of the pressure tensor. The results are general and no limited 
171: to weak inelasticity or specific values of the parameters of the mixture. 
172: As a second alternative and to test the reliability of the theoretical 
173: predictions, we have used the Direct Simulation Monte Carlo (DSMC) method 
174: \cite{B94} to numerically solve the Boltzmann equation in the simple shear 
175: flow. Although the DSMC method was originally devised for molecular fluids, 
176: its extension to dealwith inelastic collisions is easy \cite{F00,MG01}.
177: For the elements of
178: the pressure tensor the agreement between theory and simulation turns out to 
179: be very good over a wide range of values of the restitution coefficients, 
180: mass ratios, concentration ratios and size ratios. It must be 
181: noted that in this paper we are interested in analyzing transport 
182: properties in the {\em uniform} shear flow. As several authors have shown, 
183: \cite{varios} the simple shear flow is unstable to long enough
184: wavelength perturbations so that clusters of particles are spontaneously
185: developed throughout the system. Here, we will restrict ourselves to the
186: {\em uniform} case, assuming that the system has reached such a state, and
187: without paying attention to the possible formation of particle clusters
188: (microstructure).
189:  
190: 
191: 
192: The plan of the paper is as follows. In Sec.\ \ref{sec2}, the coupled
193: Boltzmann equations and the corresponding hydrodynamic equations are 
194: recalled. The steady shear flow problem is also introduced in Sec.\ 
195: \ref{sec2}, while the Sonine approximation is discussed in Sec.\ \ref{sec3}. 
196: Section \ref{sec4} deals with the Monte Carlo simulation of the Boltzmann 
197: equation particularized for steady simple shear flow. The 
198: comparison between theory and simulation is presented in Sec.\ \ref{sec5} 
199: and we close the paper in Sec.\ \ref{sec6} with a short discussion. 
200:  
201:  
202:  
203: \section{The Boltzmann equation and the simple shear flow}
204: \label{sec2}
205: 
206: We consider a binary mixture of smooth hard spheres of masses $m_{1}$ and 
207: $m_{2}$ and diameters $\sigma _{1}$ and $\sigma _{2}$. The 
208: inelasticity of collisions are characterized by three independent constant 
209: coefficients of normal restitution $\alpha_{11}$, $\alpha_{22}$, and 
210: $\alpha_{12}=\alpha_{21}$, where $\alpha_{ij}$ is the restitution 
211: coefficient for collisions between particles of species $i$ with $j$. In the
212: low-density regime, the distribution functions $f_{i}({\bf r},{\bf v};t)$ $
213: (i=1,2)$ for the two species verify the set of nonlinear Boltzmann equations 
214: \cite{GD99}  
215: \begin{equation}
216: \left( \partial _{t}+{\bf v}_{1}\cdot \nabla \right) f_{i}({\bf r},{\bf v}
217: _{1},t)=\sum_{j}J_{ij}\left[ {\bf v}_{1}|f_{i}(t),f_{j}(t)\right] \;.
218: \label{2.1}
219: \end{equation}
220: The Boltzmann collision operator $J_{ij}\left[ {\bf v}_{1}|f_{i},f_{j}\right]
221: $ describing the scattering of pairs of particles is 
222: \begin{eqnarray}
223: J_{ij}\left[{\bf v}_{1}|f_{i},f_{j}\right]  &=&\sigma _{ij}^{2}\int d{\bf v}
224: _{2}\int d\widehat{\bbox {\sigma }}\,\Theta (\widehat{\bbox {\sigma}}\cdot 
225: {\bf g}_{12})(\widehat{\bbox {\sigma }}\cdot {\bf g}_{12})  \nonumber \\
226: &&\times \left[ \alpha_{ij}^{-2}f_{i}({\bf r},{\bf v}_{1}^{\prime},t)f_{j}(
227: {\bf r},{\bf v}_{2}^{\prime },t)-f_{i}({\bf r},{\bf v}_{1},t)f_{j}({\bf r},
228: {\bf v}_{2},t)\right] \;,  \label{2.2}
229: \end{eqnarray}
230: where $\sigma_{ij}=\left( \sigma_{i}+\sigma_{j}\right)/2$, $\widehat{
231: \bbox {\sigma}}$ is a unit vector along their line of centers, $\Theta $ is
232: the Heaviside step function, and ${\bf g}_{12}={\bf v}_{1}-{\bf v}_{2}$. The
233: primes on the velocities denote the initial values $\{{\bf v}_{1}^{\prime},
234: {\bf v}_{2}^{\prime}\}$ that lead to $\{{\bf v}_{1},{\bf v}_{2}\}$
235: following a binary collision: 
236: \begin{equation}
237: {\bf v}_{1}^{\prime }={\bf v}_{1}-\mu _{ji}\left( 1+\alpha_{ij}    
238: ^{-1}\right)(\widehat{\bbox {\sigma}}\cdot {\bf g}_{12})\widehat{\bbox 
239: {\sigma}},
240: \quad {\bf v}_{2}^{\prime}={\bf v}_{2}+\mu_{ij}\left( 
241: 1+\alpha_{ij}^{-1}\right) (\widehat{\bbox {\sigma}}\cdot {\bf 
242: g}_{12})\widehat{\bbox{\sigma}}\;,  \label{2.3}
243: \end{equation}
244: where $\mu_{ij}=m_{i}/\left( m_{i}+m_{j}\right) $. The relevant
245: hydrodynamic fields are the number densities $n_{i}$, the flow velocity $
246: {\bf u}$, and the temperature $T$. They are defined in terms of moments of
247: the distributions $f_{i}$ as 
248: \begin{equation}
249: n_{i}=\int d{\bf v}_{1}f_{i}({\bf v}_{1})\;,\quad \rho {\bf u}
250: =\sum_{i}\rho_i{\bf u}_i=\sum_{i}\int d{\bf v}_{1}m_{i}{\bf v}_{1}f_{i}({\bf 
251: v}_{1})\;, 
252:  \label{2.4}
253: \end{equation}
254: \begin{equation}
255: nT=\sum_in_iT_i=\sum_{i}\int d{\bf v}_{1}\frac{m_{i}}{3}
256: V_1^{2}f_{i}({\bf v}_{1})\;,  \label{2.5}
257: \end{equation}
258: where $n=n_{1}+n_{2}$ is the total number density, $\rho 
259: =\rho_1+\rho_2=m_{1}n_{1}+m_{2}n_{2}$ is the total mass density, and ${\bf 
260: V}_1={\bf v}_1-{\bf u}$ is the peculiar velocity. Equations (\ref{2.4}) and 
261: (\ref{2.5}) also define the flow velocity ${\bf u}_i$ and the partial 
262: temperature $T_i$ of species $i$.  
263: 
264: The collision operators conserve the number of particles of each species 
265: and the total momentum, but the total energy is not conserved: 
266: \begin{equation}
267: \int d{\bf v}_{1}J_{ij}[{\bf v}_{1}|f_{i},f_{j}]=0\;,  \label{2.6}
268: \end{equation}
269: \begin{equation}
270: \sum_{i,j}\int d{\bf v}_{1}m_{i}{\bf v}_{1}J_{ij}[{\bf v}_{1}|f_{i},f_{j}]=0
271: \;,  \label{2.7}
272: \end{equation}
273: \begin{equation}
274: \sum_{i,j}\int d{\bf v}_{1}\case{1}{2}m_{i}V_{1}^{2}J_{ij}[{\bf v}
275: _{1}|f_{i},f_{j}]=-\case{3}{2}nT\zeta \;,  
276: \label{2.8}
277: \end{equation}
278: where $\zeta$ is identified as the ``cooling rate'' due to inelastic
279: collisions among all species. At a kinetic level, it is also convenient to 
280: discuss energy transfer in terms of the ``cooling rates'' $\zeta_i$ for the 
281: partial temperatures $T_i$. They are defined as 
282: \begin{equation}
283: \label{2.8.1}
284: \zeta_i=-\frac{2}{3n_iT_i}\sum_j\int 
285: d{\bf v}_{1}\case{1}{2}m_{i}V_{1}^{2}J_{ij}[{\bf 
286: v}_{1}|f_{i},f_{j}]\;.
287: \end{equation}
288: The total cooling rate $\zeta$ can be written as 
289: \begin{equation} 
290: \label{2.8.2}
291: \zeta=T^{-1}\sum_i\ x_iT_i\zeta_i\;,
292: \end{equation}
293: $x_i=n_i/n$ being the molar fraction of species $i$.  
294: 
295: 
296: 
297: {}From Eqs.\ (\ref{2.4})--(\ref{2.8}), the macroscopic balance equations for 
298: the mixture can be obtained. They are given by 
299: \begin{equation}
300: D_{t}n_{i}+n_{i}\nabla \cdot {\bf u}+\frac
301: {\nabla \cdot {\bf j}_{i}}{m_{i}}=0\;,  \label{2.9}
302: \end{equation}
303: \begin{equation}
304: D_{t}{\bf u}+\rho ^{-1}\nabla \cdot {\sf P}=0\;,  \label{2.10}
305: \end{equation}
306: \begin{equation}
307: D_{t}T-\frac{T}{n}\sum_{i}\frac{\nabla \cdot {\bf j}_{i}}{m_{i}}+\frac{2}{
308: 3n}\left( \nabla \cdot {\bf q}+{\sf P}:\nabla {\bf u}\right) =-\zeta
309: \,T\;.  \label{2.11}
310: \end{equation}
311: In the above equations, $D_{t}=\partial _{t}+{\bf u}\cdot \nabla $ is the
312: material derivative, 
313: \begin{equation}
314: {\bf j}_{i}=m_{i}\int d{\bf v}_1\,{\bf V}_1\,f_{i}({\bf V}_1)
315: \label{2.11bb}
316: \end{equation}
317: is the mass flux for species $i$ relative to the local flow, 
318: \begin{equation}
319: {\sf P}=\sum_i {\sf P}_i=\sum_{i}\,\int d{\bf v}_1\,m_{i}{\bf V}_1{\bf 
320: V}_1\,f_{i}({\bf V}_1) 
321: \label{2.12}
322: \end{equation}
323: is the total pressure tensor, and  
324: \begin{equation}
325: {\bf q}=\sum_i {\bf q}_i=\sum_{i}\,\int d{\bf 
326: v}_1\,\case{1}{2}m_{i}V_1^{2}{\bf V}_1\,f_{i}({\bf V}_1)
327: \label{2.13}
328: \end{equation}
329: is the total heat flux. The partial contributions to the pressure tensor, 
330: ${\sf P}_i$, and the heat flux, ${\bf q}_i$, coming from species $i$ can be 
331: identified from Eqs.\ (\ref{2.12}) and (\ref{2.13}). 
332: 
333: As said in the Introduction, here we are interested in evaluating the 
334: rheological properties of a granular binary mixture subjected to the simple 
335: shear flow. From a macroscopic point of view, this state is characterized by 
336: a constant linear velocity profile ${\bf u}={\bf u}_i={\sf a}\cdot {\bf r}$, 
337: where the elements of the tensor ${\sf a}$ are 
338: $a_{k\ell}=a\delta_{kx}\delta_{\ell y}$, $a$ being the constant shear rate. 
339: In addition, the partial densities $n_i$ and the granular temperature $T$
340: are uniform, while the mass and heat fluxes vanish by symmetry reasons. 
341: Thus, the (uniform) pressure tensor is the only nonzero flux in the problem. 
342: On the other hand, the temporal variation of the granular temperature arises 
343: from the balance of two opposite effects: viscous heating and dissipation in 
344: collisions. In the steady state both mechanisms cancel each other and the 
345: temperature remains constant. In that case, according to the balance energy 
346: equation (\ref{2.11}), the shear stress $P_{xy}$ and the cooling rate 
347: $\zeta$ are related by
348: \begin{equation}
349: \label{2.14}
350: aP_{xy}=-\frac{3}{2}\zeta p,
351: \end{equation}
352: where $p=nT$ is the pressure. Our aim is to analyze this 
353: steady state by means of an (approximate) analytical method as well as by  
354: performing Monte Carlo simulations of the Boltzmann equation. 
355: %Before doing 
356: %it, it is worth mentioning that sheared granular media has different steady 
357: %states, depending on the initial and boundary conditions. 
358: %However, in this paper we will restrict ourselves to the simple shear flow 
359: %described before and no attention will be paid to its stability.
360: 
361: 
362: The simple shear flow becomes spatially uniform when one refers the 
363: velocities of the particles to a frame moving with the flow velocity ${\bf 
364: u}$: $f_i\left({\bf r},{\bf v}_1\right)\rightarrow f_i({\bf V}_1)$. 
365: Consequently, the corresponding Boltzmann equations (\ref{2.1}) read  
366: \begin{equation}
367: \label{2.15}
368: -a V_{1,y}\frac{\partial}{\partial 
369: V_{1,x}}f_1({\bf V}_1)=J_{11}[{\bf V}_1|f_1,f_1]+J_{12}[{\bf V}_1|f_1,f_2]\;,
370: \end{equation}
371: \begin{equation}
372: \label{2.16}
373: -a V_{1,y}\frac{\partial}{\partial 
374: V_{1,x}}f_2({\bf V}_1)=J_{22}[{\bf V}_1|f_2,f_2]+J_{21}[{\bf V}_1|f_2,f_1]\;.
375: \end{equation}
376: The elements of the partial pressure tensors ${\sf P}_i$ ($i=1,2$) can be 
377: obtained by multiplying the Boltzmann equations by $V_{1,k}V_{1,\ell}$ and 
378: integrating over ${\bf V}_1$. The result is
379: \begin{equation}
380: \label{2.17}
381: a_{k m}P_{1,\ell m}+a_{\ell m}P_{1,k m}=A_{k \ell}^{11}+A_{k\ell}^{12}
382: \quad (1\leftrightarrow2)\;,
383: \end{equation}
384: where 
385: \begin{equation}
386: \label{2.18}
387: A_{k\ell}^{ij}=m_i\int d{\bf V}_1 V_{1,k}V_{1,\ell} J_{ij}[{\bf 
388: V}_1|f_i,f_j]\;.
389: \end{equation}
390: {}From Eq.\ (\ref{2.17}), in particular, one obtains
391: \begin{equation}
392: \label{2.19}
393: a P_{1,xy}=-\case{3}{2}p_1\zeta_1\;,
394: \end{equation}
395: \begin{equation}
396: \label{2.20}
397: a P_{1,yy}=A_{xy}^{11}+A_{xy}^{12}\;,
398: \end{equation}
399: \begin{equation}
400: \label{2.21}
401: 0=A_{yy}^{11}+A_{yy}^{12}=A_{zz}^{11}+A_{zz}^{12}\;.
402: \end{equation}
403: Here, $p_1=n_1T_1=(P_{1,xx}+P_{1,yy}+P_{1,zz})/3$ is the partial pressure of 
404: species $1$ and upon writing Eq.\ (\ref{2.19}) we have considered the 
405: relation (\ref{2.8}). The corresponding equations for ${\sf P}_2$ can be 
406: easily written just by interchanging the indices $1$ and $2$. Thus, the 
407: determination of the elements of the partial pressure tensors ${\sf P}_i$ is 
408: a closed problem once the cooling rates $\zeta_i$ and the collisional 
409: moments $A_{k\ell}^{ij}$ are known. This requires the explicit knowledge of 
410: the velocity distribution functions $f_i$. 
411: 
412: \section{Approximate solution}
413: \label{sec3}   
414: 
415: 
416: Unfortunately, solving the Boltzmann equations (\ref{2.15}) and 
417: (\ref{2.16}) is a formidable task and it does not seem possible to get the 
418: exact forms of the distributions $f_i$, even in the one-component case. A 
419: possible way to overcome such a problem is to expand $f_i$ in a complete set 
420: of polynomials with a Gaussian measure and then truncate the series. In 
421: practice, Sonine polynomials are used. This approach is similar to the usual 
422: moment method for solving kinetic equations in the elastic case. In the 
423: context of granular gases, this strategy has been widely applied in the past 
424: few years in the one-component case as well as for multicomponent systems 
425: and excellent approximations have been obtained by retaining only the first 
426: two terms. Therefore, one can expect to get a reasonable estimate for 
427: $\zeta_i$ and $A_{k\ell}^{ij}$ by using the following approximation for 
428: $f_i$: 
429: \begin{equation}
430: \label{2.22}
431: f_i({\bf V}_1)\to 
432: f_{i,M}({\bf V}_1)\left[1+\frac{m_i}{2T_i}C_{i,k\ell}\left(V_{1,k}V_{1,\ell}
433: -\frac{1}{3}V_1^2\delta_{k\ell}\right)\right],
434: \end{equation}
435: where $f_{i,M}$ is a Maxwellian distribution at the temperature of the 
436: species $i$, i.e.,
437: \begin{equation}
438: \label{2.23} 
439: f_{i,M}({\bf V}_1)=n_i 
440: \left(\frac{m_i}{2\pi 
441: T_i}\right)^{3/2}\exp\left(-\frac{m_iV^2}{2T_i}\right). 
442: \end{equation} 
443: As we will show later, in general the three temperatures 
444: $T$, $T_1$, and $T_2$ are different in the inelastic case. For this reason
445: we choose the parameters in the Maxwellians so that it is  
446: normalized to $n_i$ and provides the exact second moment of $f_i$.
447: The Maxwellians $f_{i,M}$ for the two species
448: can be quite different due to the temperature differences. This aspect is
449: essential in our two-temperature theory and has not been taken into account
450: in all previous studies. The coefficient ${\sf C}_i$ can be identified by requiring the moments 
451: with respect to $V_{1,k}V_{1,\ell}$ of the trial function (\ref{2.22}) to be the same as those for the exact 
452: distribution $f_i$. This leads to 
453: \begin{equation}
454: \label{2.24}
455: C_{i,k\ell}=\frac{P_{i,k\ell}}{p_i}-\delta_{k\ell}.
456: \end{equation}
457: With this approximation, the integrals appearing in the expressions of 
458: $\zeta_i$ and $A_{k\ell}^{ij}$ can be evaluated and the details are given in 
459: Appendices \ref{appA} and \ref{appB}.  
460:  
461:  
462: In order to express the solution of the system of equations for the pressure 
463: tensor, it is convenient to introduce dimensionless quantities. Thus, we 
464: introduce the reduced cooling rates $\zeta_i^*=\zeta_i/\nu$, 
465: the reduced temperature $T^*=\nu^2/a^2$, and the reduced 
466: pressure tensors ${\sf P}_i^*={\sf P}_i/x_i p$. The (reduced) total pressure 
467: tensor ${\sf P}^*={\sf P}/p=x_1{\sf P}_{1}^*+x_2{\sf P}_2^*$.  Here, 
468: $\nu=\sqrt{\pi}n\sigma_{12}^2v_0$ is a characteristic collision frequency 
469: and $v_0=\sqrt{2T(m_1+m_2)/m_1m_2}$ is a thermal velocity defined in terms 
470: of the temperature of the mixture $T$. Notice that, for given values of the 
471: parameters of the mixture, $T^{*}$ and ${\sf P}^*$ are  
472: functions of the restitution coefficients $\alpha_{ij}$ only. 
473: 
474: According to the symmetry of the problem, $P_{i,xz}=P_{i,yz}=0$, so that the 
475: nonzero elements are $P_{i,xx}$, $P_{i,yy}$, $P_{i,zz}$, and 
476: $P_{i,xy}=P_{i,yx}$. The three normal elements are not independent since 
477: $P_{i,xx}^*+P_{i,yy}^*+P_{i,zz}^*=3\gamma_i$, where the temperature ratios 
478: $\gamma_i=T_i/T$ are given by 
479: \begin{equation}
480: \label{2.25}
481: \gamma_1=\frac{\gamma}{1+x_1(\gamma-1)},\quad 
482: \gamma_2=\frac{1}{1+x_1(\gamma-1)}\;,
483: \end{equation}
484: with $\gamma=T_1/T_2$. The temperature ratio $\gamma$ provides information 
485: about how the kinetic energy is distributed between both species. In 
486: addition, according to Eq.\ (\ref{2.21}), $P_{i,yy}^*=P_{i,zz}^*$. 
487: Consequently, the partial pressure tensors have four relevant elements, say 
488: for instance: ${\sf {\cal P}}\equiv \{P_{1,yy}^*, P_{2,yy}^*, P_{1,xy}^*, 
489: P_{2,xy}^{*}\}$. Taking into account the results derived in the Appendices, 
490: Eqs.\ (\ref{2.20}) and (\ref{2.21}) (plus their corresponding counterparts 
491: for species $2$) can be written as
492: \begin{equation}
493: \label{2.26} 
494: {\sf {\cal L}}{\sf{\cal P}}={\sf {\cal Q}},
495: \end{equation}
496: where ${\sf {\cal L}}$ is the $4\times 4$ matrix 
497: \begin{equation}
498: \label{2.26.1}
499: {\sf {\cal L}}=\left(
500: \begin{array}{cccc}
501: 1&0&-(G_{11}+G_{12})\nu/a&-H_{12}\nu/a\\
502: 0&1&-H_{21}\nu/a&-(G_{22}+G_{21})\nu/a\\
503: -(G_{11}+G_{12})&-H_{12}&0&0\\
504: -H_{21}&-(G_{22}+G_{21})&0&0
505: \end{array}
506: \right),
507: \end{equation}
508: and 
509: \begin{equation}
510: \label{2.26.2}
511: {\sf {\cal Q}}=\left(
512: \begin{array}{c}
513: 0\\
514: 0\\
515: F_{11}+F_{12}\\
516: F_{22}+F_{21}
517: \end{array}
518: \right).
519: \end{equation}
520: Here, the functions $F_{ij}$, $G_{ij}$, and $H_{ij}$ are defined 
521: in the Appendix \ref{appB}. The solution to Eq.\ (\ref{2.26}) is
522: \begin{equation}
523: \label{2.27}
524: {\sf {\cal P}}={\sf {\cal L}}^{-1}{\sf {\cal Q}}\;.
525: \end{equation}
526: Equation (\ref{2.27}) gives the nonzero elements of the pressure tensors 
527: ${\sf P}_i^*$ in terms of the reduced temperature $T^*$ (or the 
528: reduced shear rate $a/\nu$), the temperature ratio $\gamma$, the 
529: restitution coefficients and the parameters of the mixture. The dependence 
530: of $T^*$ on the coefficients $\alpha_{ij}$ can be obtained from the energy 
531: balance equation (\ref{2.14})
532: \begin{equation}
533: \label{2.28}
534: T^{*-1/2}=-\frac{3}{2}\frac{\zeta^*}{P_{xy}^*}=
535: -\frac{3}{2}\frac{x_1\gamma_1\zeta_1^*+x_2\gamma_2\zeta_2^*}{x_1P_{1,xy}^*+
536: x_2P_{2,xy}^*}.
537: \end{equation}
538: Finally, when Eqs.\ (\ref{2.27}) and (\ref{2.28}) are used in Eq.\ 
539: (\ref{2.19}) (or its counterpart for the species 2), one gets a ${\em 
540: closed}$ equation for the temperature ratio $\gamma$, that can be solved 
541: numerically. In reduced units, this equation can be written as 
542: \begin{equation}
543: \label{2.29}
544: \gamma=
545: \frac{\zeta_2^*}{\zeta_1^*}\frac{P_{1,xy}^*}{P_{2,xy}^*}.
546: \end{equation}
547: 
548: 
549: In the elastic limit ($\alpha_{ij}=1$), we recover previous results 
550: derived for molecular gases\cite{MGS95}. A simple and interesting case 
551: corresponds to the case of mechanically equivalent particles ($m_1=m_2$, 
552: $\alpha_{11}=\alpha_{22}=\alpha_{12}\equiv\alpha$, 
553: $\sigma_{11}=\sigma_{22}$). In this limit, Eqs.\ (\ref{2.27})--(\ref{2.29}) 
554: leads to $\gamma=1$, ${\sf P}^*={\sf P}_{1}^*={\sf P}_{2}^*$, with   
555: \begin{equation}
556: \label{2.30}
557: P_{yy}^*=\frac{2}{3}\frac{2+\alpha}{3-\alpha}\;,
558: \end{equation}
559: \begin{equation}
560: \label{2.31}
561: P_{xy}^*=-\frac{5}{3}\frac{2+\alpha}{(1+\alpha)(3-\alpha)^2}\frac{a}{\nu}\;,
562: \end{equation}
563: \begin{equation}
564: \label{2.32}
565: P_{xx}^*=3-2P_{yy}^*,
566: \end{equation}
567: and 
568: \begin{equation}
569: \label{2.33}
570: T^{*-1}\equiv\frac{a^2}{\nu^2}=\frac{3}{5}\frac{(1+\alpha)(3-\alpha)^2}
571: {2+\alpha}(1-\alpha^2)\;.
572: \end{equation}
573: These expressions differ from the results derived in Ref.\ 
574: \onlinecite{BMM97} by using a model kinetic equation. However, for practical 
575: purposes, the discrepancies between both approximations are quite small, 
576: even for moderate values of the restitution coefficient. It is also 
577: interesting to consider the limit of weak dissipation
578: ($1-\alpha_{ij}\ll 1$), in which case it is
579: possible to get analytical results. For the sake of simplicity, let us 
580: assume that all the particles have the same coefficient of restitution, 
581: namely, $\alpha_{11}=\alpha_{22}=\alpha_{12}\equiv \alpha$. To get the first 
582: order corrections in the quasielastic limit, we introduce the perturbation 
583: parameter $\epsilon\equiv (1-\alpha^2)^{1/2}$ and perform a series expansion 
584: around $\epsilon=0$. The details of the calculation are presented in 
585: Appendix \ref{appC} and here we only quote the final results. First, the 
586: leading term of the reduced shear rate $a/\nu$ (which is a measure of the 
587: steady granular temperature) is
588: \begin{equation}
589: \label{2.34}
590: \frac{a}{\nu}=a_0 (1-\alpha^2)^{1/2}+\cdots.
591: \end{equation}
592: Next, the temperature ratio and the relevant elements of the (partial) 
593: pressure tensors can be written as
594: \begin{equation}
595: \label{2.35}
596: \gamma=1+\gamma_0 a_0^2 (1-\alpha^2)+\cdots,
597: \end{equation}
598: \begin{equation}
599: \label{2.36}
600: P_{i,yy}^*=1+P_{i,yy}^{(2)}a_0^2 (1-\alpha^2)+\cdots,
601: \end{equation}
602: \begin{equation}
603: \label{2.37}
604: P_{i,xx}^*=1+P_{i,xx}^{(2)} a_0^2 (1-\alpha^2)+\cdots,
605: \end{equation}
606: \begin{equation}
607: \label{2.38}
608: P_{i,xy}^*=P_{i,xy}^{(1)}a_0 (1-\alpha^2)^{1/2}+\cdots.
609: \end{equation}
610: In these equations, $a_0$, $\gamma_0$, and $P_{i,k\ell}^{(r)}$ are 
611: dimensionless coefficients that depend on the ratios of mass, concentrations 
612: and sizes. Their explicit expressions are given in Appendix \ref{appC}.
613: 
614: In summary, by using the Sonine approximation (\ref{2.22}), we have 
615: explicitly determined the rheological properties of the mixture as well as 
616: the reduced shear rate and the temperature ratio as functions of 
617: dissipation and mechanical parameters of the mixture. These 
618: theoretical predictions will be compared with those obtained from Monte 
619: Carlo simulations in Sec.\ \ref{sec5}. 
620:  
621:  
622: \section{Monte Carlo simulation}
623: \label{sec4}
624: 
625: {}From a numerical point of view, the Direct Simulation Monte Carlo (DSMC) 
626: method\cite{B94} is the most convenient algorithm to study non-equilibrium 
627: phenomena in the low-density regime. It was devised to mimic the dynamics 
628: involved in the Boltzmann collision term. 
629: The extension of the DSMC method to deal with inelastic collisions is 
630: straightforward\cite{MGSB99,F00,MG01}, and here we have used it to 
631: numerically solve the Boltzmann equation in the simple shear flow. In 
632: addition, since the simple shear flow is spatially homogeneous in the local 
633: Lagrangian frame, the simulation method becomes especially easy to 
634: implement. This is an important advantage with respect to molecular dynamics 
635: simulations. On the other hand, the restriction to this homogeneous state 
636: prevents us from analyzing the possible instability of simple shear flow or 
637: the formation of clusters or microstructures. 
638: 
639: The DSMC method as applied to the simple shear flow is as follows. 
640: The velocity distribution function of the species $i$ is represented by the 
641: peculiar velocities $\{{\bf V}_k\}$ of $N_i$ ``simulated" particles:
642: \begin{equation}
643: \label{4.1}
644: f_i({\bf V},t)\to n_i \frac{1}{N_i}\sum_{k=1}^{N_i} \delta({\bf V}-{\bf 
645: V}_k(t))\; .
646: \end{equation}
647: Note that the number of particles $N_i$ must be taken according to the 
648: relation $N_1/N_2=n_1/n_2$. At the initial state, one assigns velocities to 
649: the particles drawn from the Maxwell-Boltzmann probability distribution: 
650: \begin{equation}
651: \label{4.2}
652: f_i({\bf V},0)=n_i\ \pi^{-3/2}\ V_{0i}^{-3}(0)\ 
653: \exp\left(-V^2/V_{0i}^2(0)\right)\;,
654: \end{equation}
655: where $V_{0i}^2(0)=2T(0)/m_i$ and $T(0)$ is the initial temperature. To 
656: enforce a vanishing initial total momentum, the velocity of every particle 
657: is subsequently subtracted by the amount $N_i^{-1} \sum_k {\bf V}_k(0)$. In 
658: the DSMC method, the free motion and the collisions are uncoupled over a 
659: time step $\Delta t$ which is small compared with the mean free time and the 
660: inverse shear rate. In the local Lagrangian frame, particles of each species 
661: ($i=1,2$) are subjected to the action of a non-conservative inertial force 
662: ${\bf F}_i=-m_i\ {\sf a}\cdot{\bf V}$. This force is represented by the 
663: terms on the left-hand side of Eqs.\ (\ref{2.15}) and (\ref{2.16}). Thus, 
664: the free motion stage consists of making ${\bf V}_k\to {\bf V}_k-{\sf 
665: a}\cdot{\bf V}_k\Delta t$. In the collision stage, binary interactions 
666: between particles of species $i$ and $j$ must be considered. To simulate the 
667: collisions between particles of species $i$ with $j$ a sample of 
668: $\frac{1}{2} N_i \omega_{\text{max}}^{(ij)}\Delta t$ pairs is chosen at 
669: random with equiprobability. Here, $\omega_{\text{max}}^{(ij)}$ is an upper 
670: bound estimate of the probability that a particle of the species $i$ 
671: collides with a particle of the species $j$. 
672: Let us consider a pair $\{k,\ell\}$ belonging to this sample. Here, $k$ 
673: denotes a particle of species $i$ and $\ell$ a particle of species $j$. 
674: For each pair $\{k,\ell\}$ with velocities 
675: $\{{\bf V}_k,{\bf V}_{\ell}\}$, the following steps 
676: are taken: (1) a given direction $\widehat{\bbox \sigma}_{k\ell}$ is chosen 
677: at random with equiprobability; (2) the collision between particles $k$ and 
678: $\ell$ is accepted with a probability equal to $\Theta({\bf g}_{k\ell}\cdot 
679: \widehat{\bbox \sigma}_{k\ell})\omega_{k\ell}^{(ij)}/
680: \omega_{\text{max}}^{(ij)}$, where 
681: $\omega_{k\ell}^{(ij)}=4\pi \sigma_{ij}^2 n_j|{\bf g}_{k\ell}\cdot 
682: \widehat{\bbox \sigma}_{k\ell}|$ and ${\bf g}_{k\ell}={\bf V}_k-{\bf 
683: V}_{\ell}$; (3) if the collision is accepted, postcollisional velocities are 
684: assigned to both particles according to the scattering rules:
685: \begin{equation}
686: \label{4.3}
687: {\bf V}_{k}\to {\bf V}_{k}-\mu_{ji}
688: (1+\alpha_{ij})({\bf g}_{k\ell}\cdot \widehat{\bbox 
689: \sigma}_{k\ell})\widehat{\bbox \sigma}_{k\ell}\; ,
690: \end{equation}
691: \begin{equation}
692: \label{4.4}
693: {\bf V}_{\ell}\to {\bf V}_{\ell}+\mu_{ij}
694: (1+\alpha_{ij})({\bf g}_{k\ell}\cdot \widehat{\bbox 
695: \sigma}_{k\ell})\widehat{\bbox \sigma}_{k\ell}\;.
696: \end{equation}
697: In the case that in one of the collisions 
698: $\omega_{k\ell}^{(ij)}>\omega_{\text{max}}^{(ij)}$, the estimate of 
699: $\omega_{\text{max}}^{(ij)}$ is updated as 
700: $\omega_{\text{max}}^{(ij)}=\omega_{k\ell}^{(ij)}$. The procedure described 
701: above is performed for $i=1,2$ and $j=1,2$. 
702: 
703: 
704: In the course of the simulations, one evaluates the total pressure tensor 
705: and the partial temperatures. They are given as 
706: \begin{equation}
707: \label{4.5}
708: {\sf P}=\sum_{i=1}^{2} \frac{m_i n_i}{N_i}\sum_{k=1}^{N_i} {\bf V}_k {\bf 
709: V}_k\; ,
710: \end{equation}
711: \begin{equation}
712: T_i=\frac{m_i}{3N_i}\sum_{k=1}^{N_i}{\bf V}_k^2\; .
713: \end{equation}
714: To improve the statistics, the results are averaged over a number 
715: ${\cal N}$ of independent realizations or replicas. In our simulations we 
716: have typically taken a total number of particles $N=N_1+N_2=10^5$, a number 
717: of replicas ${\cal N}=10$, and a time step $\Delta t=3\times 10^{-3}
718: \lambda_{11}/V_{01}(0)$, where 
719: $\lambda_{11}=(\sqrt{2} \pi n_1 \sigma_{11}^2)^{-1}$ is the mean free path 
720: for collisions 1--1.
721: 
722: 
723: A complete presentation of the results is complex since there are many 
724: parameters involved: $\{\alpha_{ij}, m_1/m_2, n_1/n_2, 
725: \sigma_{11}/\sigma_{22}\}$. For the sake of concreteness, henceforth we will 
726: consider the case $\alpha_{11}=\alpha_{22}=\alpha_{12}\equiv\alpha$. 
727: In the steady state, the reduced quantities $T^*$, $\gamma_i$, and ${\sf 
728: P}^*$ are independent of the initial state for given 
729: values of the restitution coefficient and the ratios of mass, concentration 
730: and sizes. To illustrate it, in Fig.\ \ref{fig1} we present the time 
731: evolution of $T^*(t)$ for $\alpha=0.75$, 
732: $\sigma_{11}/\sigma_{22}=1$, $m_1/m_2=4$, $n_1/n_2=1/3$ and three different 
733: initial conditions. Time is measured in units of $\lambda_{11}/V_{01}(0)$. 
734: After an initial transient period, all curves converge to the same steady 
735: value, as predicted by the solution described in Sec.\ \ref{sec3}. The same 
736: qualitative behavior has been found for the temperature ratio and the 
737: elements of the reduced pressure tensor. Therefore, in the following we will 
738: focus on the dependence of the steady values of the reduced quantities on 
739: the restitution coefficient $\alpha$ and the parameters of the mixture, once 
740: we have checked they do not depend on the initial state. 
741:  
742: 
743: 
744: \section{Comparison between theory and Monte Carlo simulations}
745: \label{sec5} 
746:  
747: In this Section we compare the predictions of the Sonine approximation with 
748: the results obtained from the DSMC method. Our goal is to explore the 
749: dependence of $a^*$, $\gamma=T_1/T_2$ and the nonzero elements of ${\sf 
750: P}^*$ on $\alpha$, the mass ratio $\mu\equiv m_1/m_2$, the 
751: concentration ratio $\delta\equiv n_1/n_2$, and the ratio of sizes 
752: $w\equiv \sigma_{11}/\sigma_{22}$. 
753: 
754: First, we will investigate the dependence of the relevant quantities on 
755: $\alpha$ and $\mu$ for given values of $\delta$ and $w$. Recent molecular 
756: dynamics simulations for a dilute monocomponent system of smooth inelastic 
757: hard disks\cite{GT96} have supported an ``equation of state'' to a
758: sheared granular system in which the steady (reduced) temperature $T^*$ can 
759: be closely fitted by a linear function of $(1-\alpha^2)^{-1}$. Similar 
760: results have been obtained from kinetic models of the Boltzmann \cite{BMM97} 
761: and Enskog \cite{MGSB99} equations. An interesting question is whether this 
762: simple relationship can be extended to the case of multicomponent systems. 
763: The results obtained here (both from the simulations and from the kinetic 
764: theory analysis) for mixtures of different masses, concentrations or sizes 
765: show that $T^{*}$ is indeed a {\em quasi}-linear function of 
766: $(1-\alpha^2)^{-1}$. As an illustrative example, we consider the case $w=1$, 
767: $\delta=1$ (equimolar mixture), and three different values of the mass ratio 
768: $\mu=1$, $2$, and $10$. Figure \ref{fig2} shows $T^{*}$ versus 
769: $(1-\alpha^2)^{-1}$ as obtained from the simulations (symbols) and from the 
770: Sonine approximation (lines). It is evident that the kinetic theory has an 
771: excellent agreement with the simulation results and also that $T^{*}$ is 
772: practically linear in $(1-\alpha^2)^{-1}$ whatever the mass ratio considered 
773: is. The slope of the straight lines increases as the disparity of masses 
774: increases. 
775: 
776: The temperature ratio and the non-zero elements of the pressure tensor are 
777: plotted in Figs.\ \ref{fig3} and \ref{fig4} $(a-d)$, respectively, as a 
778: function of the dissipation parameter $\alpha$ for the same cases as those 
779: considered in Fig.\ \ref{fig2}. The curves corresponding to $\mu<1$ can be 
780: easily inferred from them. Figure \ref{fig3} clearly shows that, except 
781: for mechanically equivalent particles, the partial temperatures are 
782: different, even for moderate dissipation (say $\alpha\simeq 0.9$). This 
783: means that the traditional assumption of equipartition of fluctuation energy
784: begins to fail.
785: This effect is generic of multicomponent dissipative systems and is
786: consistent with results recently derived in the homogeneous cooling 
787: state\cite{GD99,SD01} as well as in driven systems\cite{FM02,CH02}. 
788: The extent of the equipartition violation depends
789: on the concentrations and the mechanical differences of the particles (e.g.,
790: masses, sizes, restitution coefficients), and is greater when the differences
791: are large. The agreement between theory and simulation is again excellent.
792: %The results obtained for the steady reduced shear rate $a^{*2}$ are shown in 
793: %Fig.\ \ref{fig3}. 
794: With respect to the pressure tensor, Fig.\ \ref{fig4} $(a-d)$, we see that 
795: the theory captures well the main trends observed for the rheological 
796: properties. At a quantitative level, the agreement is better in the case of 
797: the shear stress $P_{xy}^*$ and the normal element $P_{xx}^*$, while the 
798: discrepancies for the normal stresses $P_{yy}^*$ and $P_{zz}^*$ are larger 
799: than in the case of the temperature ratio, especially as the restitution 
800: coefficient decreases. On the other hand, the theory only predicts normal 
801: stress differences in the plane of shear flow ($P_{xx}^*\neq 
802: P_{yy}^*=P_{zz}^*$) while the simulation also shows that there is anisotropy 
803: in the plane perpendicular to the flow velocity, $P_{zz}^*> P_{yy}^*$. This 
804: kind of anisotropy has been also observed in molecular dynamics simulations 
805: of shear flows \cite{HS92}. Nevertheless, these relative normal stress 
806: differences in this plane are very small and decrease as $\alpha$ increases. 
807:  
808:  
809: The influence of the concentration ratio $\delta$ on the temperature ratio 
810: and the rheological properties is shown in Figs.\ \ref{fig5} and \ref{fig6} 
811: $(a-d)$, respectively, for $w=1$, $\mu=4$, and two values of $\delta$: 
812: $\delta=1/3$ and $\delta=3$. We observe again a strong dependence of the 
813: temperature ratio $\gamma$ on dissipation. For a given value of $\alpha$, 
814: the temperature ratio increases as the molar fraction of the heavy species 
815: decreases. Concerning shear stresses, we see that they are practically 
816: independent of the concentration ratio since all the curves collapse in a 
817: common curve. A more significant influence is observed for the normal 
818: stresses. In general, the agreement with the theory is good although the 
819: discrepancies are more important in the case of $\delta=1/3$. 
820: Finally, the influence of the size of the particles on the rheological 
821: properties is illustrated in Fig.\ \ref{fig7} $(a-d)$. We consider 
822: an equimolar mixture ($\delta=1$) of particles of equal mass ($\mu=1$) for 
823: two different values of the size ratio: $w=1$, and $w=2$. We see that 
824: similar conclusions to those previously found in Figs.\ 
825: \ref{fig3}--\ref{fig6} are obtained when one considers mixtures of particles 
826: of different sizes. 
827:  
828:  
829: 
830: \section{Discussion}
831: \label{sec6}
832: 
833: In this paper we have addressed the problem of a low-density granular 
834: mixture constituted by smooth inelastic hard spheres and subjected to a 
835: linear shear flow $u_x=ay$. We are interested in the steady state where the 
836: effect of viscosity is compensated for by the dissipation in collisions. Our 
837: description applies for arbitrary values of the shear rate $a$ or the 
838: inelasticity of the system and not restriction on the values of masses, 
839: concentrations and sizes are imposed in the system. The study has 
840: been made by using two different and complementary routes. On the one hand, 
841: the set of coupled Boltzmann equations are solved by means a Sonine 
842: polynomial approximation  and, on the other hand, Monte Carlo simulations 
843: are performed to numerically solve the Boltzmann equations.
844: Given that the partial temperatures $T_i$ of each species can be different,
845: the reference Maxwellians in the Sonine expansion are defined at the
846: temperature for that species. This is one of the new features of our
847: expansion. On the other hand, to put this work
848: in a proper context, it must be noticed that we have restricted our 
849: considerations to states in which the only gradient is the one associated 
850: with the shear rate so that density and velocity fluctuations are not 
851: allowed in the numerical simulation. 
852: 
853: We have focused on the analysis of the dependence of the steady 
854: (reduced) temperature $T^*$ and the (reduced) pressure tensor ${\sf P}^*$ on 
855: the coefficients of restitution $\alpha_{ij}$ and the parameters of the 
856: mixture, namely the mass ratio $\mu$, the concentration ratio $\delta$ and 
857: the size ratio $w$. The results clearly indicate that the deviation of the 
858: above quantities from their functional forms for elastic collisions is quite 
859: important, even for moderate dissipation. In particular, the temperature 
860: ratio, which measures the distribution of kinetic energy between both 
861: species, is different from unity and presents a complex dependence on the 
862: parameters of the problem. This result contrasts with previous results 
863: derived for granular mixtures\cite{JM89,Z95,AW98,WA99} where it was 
864: consistently assumed the equality of the partial temperatures in the small
865: inelasticity limit. In the same way
866: as in the homogeneous cooling state problem\cite{GD99,MG01},
867: the deviations from
868: the energy equipartition can be weak or strong depending on the mechanical 
869: differences between the species and the degree of dissipation. On the other
870: hand, the simulation as well as the theoretical results also show that 
871: the steady total temperature $T^*$ can be fitted 
872: by a linear function of $(1-\alpha^2)^{-1}$ with independence of the values 
873: of the parameters of the mixture. This extends previous results derived in 
874: the context of simple granular gases by using molecular dynamics \cite{GT96} 
875: or Monte Carlo simulations\cite{BMM97,MGSB99}. With respect to the 
876: rheological properties, comparison between theory and simulation shows a 
877: good quantitative agreement, especially for the shear stress $P_{xy}^*$, 
878: which is the most relevant element of the pressure tensor in a shearing
879: problem.  Although the kinetic theory also predicts normal stresses, the 
880: discrepancies between theory and simulation are larger than those found for 
881: the temperature ratio or the shear stress.
882: 
883: 
884: It is illustrative to make some comparison between the predictions
885: made from our two-temperature theory with those obtained if the
886: differences in the partial temperatures were neglected ($T_1=T_2=T$). For
887: instance, let us consider the mixture $\sigma_{11}=\sigma_{22}$, $n_1=n_2$,
888: and $m_1=10m_2$ with $\alpha=0.75$. In this case, for the $xy$ and $yy$ elements
889: of the pressure tensor, the simulation results are $-P_{xy}^*=0.498$ and
890: $P_{yy}^*=0.723$. Our two-temperature theory predicts $-P_{xy}^*=0.498$ and
891: $P_{yy}^*=0.743$ while the single-temperature theory (assumption made in
892: previous works) gives $-P_{xy}^*=0.456$ and $P_{yy}^*=0.815$. Clearly,
893: inclusion of the two-temperature effects improves the theoretical
894: estimations and makes a significant (quantitative) difference
895: with respect to the predictions of the single-temperature theory.
896:  
897:  
898: As a final comment, let us mention that the study made here can in principle be
899: extended in both aspects, kinetic theory and simulations, to the revised 
900: Enskog equation in order to assess the influence of finite density on the 
901: rheological properties of the mixture. Work along this line is in progress.  
902:  
903: \acknowledgments
904: 
905: Partial support from the Ministerio de Ciencia y Tecnolog\'{\i}a (Spain)
906: through Grant No. BFM 2001-0718 is acknowledged.  
907: 
908: 
909: 
910: \appendix
911: \section{Evaluation of the cooling rates}
912: \label{appA}
913: 
914: In this Appendix the (reduced) cooling rates $\zeta_i^*$ are evaluated by 
915: using the first Sonine approximation (\ref{2.22}). The cooling rate is given 
916: by
917: \begin{equation}
918: \label{a1}
919: \zeta_i^*=-\frac{2}{3}\pi^{-1/2}\theta_i\sum_j\int d{\bf 
920: V}_1^*V_1^{*2}J_{ij}^*[{\bf V}_1^*|f_i^*,f_j^*],
921: \end{equation}
922: where $\theta_i=1/(\gamma_i\mu_{ji})$, ${\bf V}_1^*={\bf V}_1/v_0$, 
923: $J_{ij}^*=(v_0^2/n_in\sigma_{12}^2)J_{ij}$, and $f_i^*=(v_0^3/n_i)f_i$. 
924: Henceforth, it will be understood that dimensionless quantities will be used 
925: and the asterisks will be deleted to simplify the notation. A useful 
926: identity for an arbitrary function $h({\bf V}_{1})$ is given by 
927: \begin{eqnarray}
928: \int d{\bf V}_{1}h({\bf V}_{1})J_{ij}\left[ {\bf V}_{1}|f_{i},f_{j}\right]
929: &=&x_j\left(\frac{\sigma _{ij}}{\sigma_{12}}\right)^{2}\int \,d{\bf 
930: V}_{1}\,\int \,d{\bf V}_{2}f_{i}(
931: {\bf V}_{1})f_{j}({\bf V}_{2})  \nonumber \\
932: & &\times \int d\widehat{\bbox {\sigma}}\,\Theta (\widehat{\bbox {\sigma}}
933: \cdot {\bf g}_{12})(\widehat{\bbox {\sigma }}\cdot {\bf g}_{12})\,\left[ h(
934: {\bf V}_{1}^{^{\prime \prime }})-h({\bf V}_{1})\right] \;,  \label{a2}
935: \end{eqnarray}
936: with 
937: \begin{equation}
938: {\bf V}_{1}^{^{\prime \prime }}={\bf V}_{1}-\mu _{ji}(1+\alpha_{ij})(
939: \widehat{\bbox {\sigma }}\cdot {\bf g}_{12})\widehat{\bbox {\sigma}}\;.
940: \label{a3}
941: \end{equation}
942: This result applies for both $i=j$ and $i\neq j$. Use of this identity in 
943: Eq.\ (\ref{a1}) allows the angular integrals to be performed. The result is
944: \begin{eqnarray}
945: \label{a4}
946: \zeta_i&=&(1-\alpha_{ii}^2)\frac{1}{12}\sqrt{\pi} \theta_ix_i
947: \left(\frac{\sigma _{ii}}{\sigma_{12}}\right)^{2}\int \,d{\bf 
948: V}_{1}\,\int \,d{\bf V}_{2}\,g_{12}^3f_{i}({\bf V}_{1})f_{i}({\bf V}_{2}) 
949: \nonumber\\
950: & & +(1-\alpha_{ij}^2)\frac{1}{3}\sqrt{\pi} \theta_i\mu_{ji}^2x_j
951: \int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}\, g_{12}^3f_{i}({\bf 
952: V}_{1})f_{j}({\bf V}_{2}) \nonumber\\
953: & & +(1+\alpha_{ij})\frac{2}{3}\sqrt{\pi} \theta_i\mu_{ji}x_j
954: \int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}\,g_{12}
955: ({\bf g}_{12}\cdot {\bf G}_{ij})f_{i}({\bf V}_{1})f_{j}({\bf V}_{2}) ,
956: \end{eqnarray}
957: where ${\bf G}_{ij}=\mu_{ij}{\bf V}_1+\mu_{ji}{\bf V}_2$. Now we consider 
958: the Sonine approximation (\ref{2.22}) for the distributions $f_i$:
959: \begin{equation}
960: \label{a4.1}
961: f_{i}({\bf V}_1)\to \left(\frac{\theta_i}{\pi}\right)^{3/2}e^{-\theta_i 
962: V_1^2}
963: \left[1+\theta_i C_{i,k\ell}\left(V_{1,k}V_{1,\ell}-\frac{1}{3}
964: V_1^2\delta_{k\ell}\right)\right].
965: \end{equation}
966: Neglecting nonlinear terms in the tensor $C_{i,k\ell}$, the expression 
967: (\ref{a4}) can be written as
968: \begin{eqnarray}
969: \label{a5}
970: \zeta_i&=&(1-\alpha_{ii}^2)\frac{1}{12}\pi^{-5/2} \theta_i^{-1/2}x_i
971: \left(\frac{\sigma _{ii}}{\sigma_{12}}\right)^{2}\int \,d{\bf 
972: V}_{1}\,\int \,d{\bf V}_{2}\,g_{12}^3\ e^{-(V_1^2+V_2^2)}
973: \nonumber\\
974: & & +(1-\alpha_{ij}^2)\frac{1}{3}\pi^{-5/2} 
975: (\theta_i\theta_j)^{3/2}\mu_{ji}^2x_j \theta_i
976: \int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}\,g_{12}^3
977: \ e^{-(\theta_iV_1^2+\theta_jV_2^2)}
978: \nonumber\\
979: & & +(1+\alpha_{ij})\frac{2}{3}\pi^{-5/2} (\theta_i\theta_j)^{3/2}
980: \mu_{ji}x_j\theta_i\int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}\,g_{12}
981: ({\bf g}_{12}\cdot {\bf G}_{ij})\ e^{-(\theta_iV_1^2+\theta_jV_2^2)}.
982: \end{eqnarray}
983: Here, use has been made of the fact the scalar $\zeta_i^*$ cannot be coupled 
984: to the traceless tensor $C_{i,k\ell}$ so that the only nonzero contribution 
985: to the cooling rate comes from the Maxwellian term (first term of the right 
986: hand side of (\ref{a4.1})) of the distribution function. 
987: The first integral of Eq.\ (\ref{a5}) is straightforward and can be done
988: with a change of variables to relative and center of mass variables. The
989: next two integrals are somewhat more complicated and they can be performed 
990: by the change of variables 
991: \begin{equation}
992: \label{a6}
993: {\bf x}={\bf V}_1-{\bf V}_2, \quad {\bf y}=\theta_i{\bf V}_1+\theta_j
994: {\bf V}_2,
995: \end{equation}
996: with the Jacobian $(\theta_i+\theta_j)^{-3}$. The integrals can be now 
997: easily performed and the final result for $\zeta_1$ is 
998: \begin{eqnarray}
999: \label{a7}
1000: \zeta_1&=&\frac{2}{3}\sqrt{2}
1001: \left(\frac{\sigma 
1002: _{11}}{\sigma_{12}}\right)^{2}x_1\theta_1^{-1/2}(1-\alpha_{11}^2)\nonumber\\
1003: & & 
1004: +\frac{4}{3}x_2\mu_{21}\left(\frac{\theta_1+\theta_2}{\theta_1\theta_2}
1005: \right)^{1/2}(1+\alpha_{12}) 
1006: \left[2-\mu_{21}(1+\alpha_{12})\frac{\theta_1+\theta_2}{\theta_2}
1007: \right].
1008: \end{eqnarray} 
1009: The result for $\zeta_2$ is obtained from Eq.\ (\ref{a7}) by interchanging 
1010: $1$ and $2$.
1011: 
1012: 
1013: 
1014: \section{Evaluation of the collisional moments}
1015: \label{appB}  
1016: 
1017: In reduced units, the collisional moments $A_{k\ell}^{ij}$ are given by
1018: \begin{eqnarray}
1019: \label{b1}
1020: A_{k\ell}^{ij}&=&\frac{m_iv_0^2}{T}\pi^{-1/2}\,\int d{\bf V}_1 
1021: V_{1,k}V_{1,\ell}J_{ij}[{\bf V}_1|f_i,f_j]\nonumber\\
1022: &=& \frac{m_iv_0^2}{T}x_j 
1023: \pi^{-1/2}\left(\frac{\sigma_{ij}}{\sigma_{12}}\right)^{2}
1024: \int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}f_{i}(
1025: {\bf V}_{1})f_{j}({\bf V}_{2}) 
1026: \int d\widehat{\bbox {\sigma}}\,\Theta (\widehat{\bbox {\sigma}}
1027: \cdot {\bf g}_{12})\nonumber\\
1028: & & \times (\widehat{\bbox {\sigma }}\cdot {\bf g}_{12})
1029: \left(V_{1,k}^{''}V_{1,\ell}^{''}-V_{1,k}V_{1,\ell}\right),
1030: \end{eqnarray}
1031: where the identity (\ref{a2}) has been used. Substitution of (\ref{a3}) into 
1032: Eq.\ (\ref{b1}) allows the angular integral to be performed with the result
1033: \begin{eqnarray}
1034: \label{b2}
1035: A_{k\ell}^{ij}&=&-\frac{\sqrt{\pi}}{2} \frac{m_iv_0^2}{T} \mu_{ji}x_j 
1036: \left(\frac{\sigma _{ij}}{\sigma_{12}}\right)^{2}(1+\alpha_{ij})
1037: \int \,d{\bf V}_{1}\,\int \,d{\bf V}_{2}f_{i}(
1038: {\bf V}_{1})f_{j}({\bf V}_{2}) \nonumber\\
1039: & & \times 
1040: \left[g_{12}(G_{ij,k}g_{12,\ell}+G_{ij,\ell}g_{12,k})+\frac{\mu_{ji}}{2}
1041: (3-\alpha_{ij})g_{12}g_{12,k}g_{12,\ell}-\frac{\mu_{ji}}{6}
1042: (1+\alpha_{ij})g_{12}^3\delta_{k\ell}\right],
1043: \end{eqnarray}
1044: where $g_{12,k}=V_{1,k}-V_{2,k}$ and 
1045: $G_{ij,k}=\mu_{ij}V_{1,k}+\mu_{ji}V_{2,k}$. To perform the integral we use
1046: the Sonine approximation of $f_i$ and the change of variables 
1047: (\ref{a6}). When one neglects again nonlinear terms in the tensor ${\sf 
1048: C}_i$, the collisional moment $A_{k\ell}^{ij}$ becomes 
1049: \begin{eqnarray}
1050: \label{b4}
1051: A_{k\ell}^{ij}&=&-\frac{1}{2}\pi^{-5/2} \frac{m_iv_0^2}{T}\mu_{ji}x_j 
1052: \left(\frac{\sigma 
1053: _{ij}}{\sigma_{12}}\right)^{2}(1+\alpha_{ij})\frac{(\theta_i\theta_j)^{3/2}}
1054: {(\theta_i+\theta_j)^3}
1055: \int \,d{\bf x}\,\int \,d{\bf y}e^{-(b_{ij}x^2+d_{ij}y^2)}\nonumber\\
1056: & & \times 
1057: \left[1+\theta_i(\theta_i+\theta_j)^{-2}{\sf C}_i\cdot 
1058: ({\bf y}+\theta_j{\bf x})({\bf y}+
1059: \theta_j{\bf x})+\theta_j(\theta_i+\theta_j)^{-2}{\sf C}_{j}\cdot
1060: ({\bf y}-\theta_i{\bf x})({\bf y}-
1061: \theta_i{\bf x})\right]\nonumber\\
1062: & & \times
1063: \left[(\theta_i+\theta_j)^{-1}x(x_{k}y_{\ell}+x_{\ell}y_{k})
1064: +\lambda_{ij}xx_kx_{\ell}-\frac{\mu_{ji}}{6}
1065: (1+\alpha_{ij})x^3\delta_{k\ell}\right],
1066: \end{eqnarray}
1067: where 
1068: \begin{equation}
1069: \label{b4.1}
1070: b_{ij}=\theta_i\theta_j(\theta_i+\theta_j)^{-1},
1071: \end{equation}
1072: \begin{equation}
1073: \label{b4.2}
1074: d_{ij}=(\theta_i+\theta_j)^{-1},
1075: \end{equation}
1076: \begin{equation}
1077: \label{b5}
1078: \lambda_{ij}=2\frac{\mu_{ij}\theta_j-\mu_{ji}\theta_i}{\theta_i+\theta_j}+
1079: \frac{\mu_{ji}}{2}(3-\alpha_{ij}).
1080: \end{equation}
1081: The corresponding integrals can be now easily performed and, after some 
1082: algebra, the final result is
1083: \begin{eqnarray}
1084: \label{b5.1}
1085: A_{k\ell}^{ij}&=&\frac{2}{3}\frac{m_iv_0^2}{T}\mu_{ji}x_j 
1086: \left(\frac{\sigma 
1087: _{ij}}{\sigma_{12}}\right)^{2}(1+\alpha_{ij})
1088: \left(\frac{\theta_i+\theta_j}{\theta_i\theta_j}\right)^{3/2} \nonumber\\
1089: & & \times \left\{\left[\case{1}{5}\lambda_{ij}+\case{1}{2}\mu_{ji}
1090: (1+\alpha_{ij})\right]\delta_{k\ell}-2
1091: \frac{\theta_i\theta_j}{(\theta_i+\theta_j)^2}\left[\left(1+\case{3}{5}
1092: \lambda_{ij}\frac{\theta_i+\theta_j}{\theta_i}\right)\gamma_i^{-1}
1093: P_{i,k\ell}\right.\right.\nonumber\\
1094: & & \left.\left.-\left(1-\case{3}{5}
1095: \lambda_{ij}\frac{\theta_i+\theta_j}{\theta_j}\right)\gamma_j^{-1}
1096: P_{j,k\ell}\right]\right\}.
1097: \end{eqnarray}
1098: 
1099: 
1100: 
1101: {}From this general expression one can get the collisional moments 
1102: $A_{k\ell}^{11}$, $A_{k\ell}^{12}$, $A_{k\ell}^{22}$, and 
1103: $A_{k\ell}^{21}$. In particular, 
1104: \begin{equation}
1105: \label{b6}
1106: A_{k\ell}^{11}=F_{11}\delta_{k\ell}+G_{11} P_{1,k\ell},
1107: \end{equation}
1108: \begin{equation}
1109: \label{b7}
1110: A_{k\ell}^{12}=F_{12}\delta_{k\ell}+G_{12} P_{1,k\ell}+H_{12}P_{2,k\ell},
1111: \end{equation}
1112: where
1113: \begin{equation}
1114: \label{b8}
1115: F_{11}=\frac{4}{15}\sqrt{2}\mu_{21}^{-1}x_1\left(\frac{\sigma_{11}}
1116: {\sigma_{12}}\right)^2\theta_1^{-3/2}(1+\alpha_{11})(2+\alpha_{11}),
1117: \end{equation}
1118: \begin{equation}
1119: G_{11}=-\frac{2}{5}\sqrt{2}x_1\left(\frac{\sigma_{11}}{\sigma_{12}}
1120: \right)^2\theta_1^{-1/2}(1+\alpha_{11})(3-\alpha_{11}),
1121: \end{equation}
1122: \begin{equation}
1123: \label{b9}
1124: F_{12}=\frac{4}{3}x_2(1+\alpha_{12})
1125: \left(\frac{\theta_1+\theta_2}{\theta_1\theta_2}\right)^{3/2}\left[
1126: \case{1}{5}\lambda_{12}+\case{1}{2}\mu_{21}(1+\alpha_{12})\right],
1127: \end{equation}
1128: \begin{equation}
1129: \label{b10}
1130: G_{12}=-\frac{8}{3}x_2\mu_{21}(1+\alpha_{12})
1131: \left(\frac{\theta_1}{\theta_2(\theta_1+\theta_2)}\right)^{1/2}\left(
1132: 1+\frac{3}{5}\lambda_{12}
1133: \frac{\theta_1+\theta_2}{\theta_1}\right),
1134: \end{equation}
1135: \begin{equation}
1136: \label{b11}
1137: H_{12}=\frac{8}{3}x_2\mu_{12}(1+\alpha_{12})
1138: \left(\frac{\theta_2}{\theta_1(\theta_1+\theta_2)}\right)^{1/2}\left(
1139: 1-\frac{3}{5}\lambda_{12}
1140: \frac{\theta_1+\theta_2}{\theta_2}\right).
1141: \end{equation}
1142: The corresponding expressions for $F_{22}$, $G_{22}$, $F_{21}$, $G_{21}$, 
1143: and $H_{21}$ can be easily inferred from Eqs.\ (\ref{b8})--(\ref{b11}) by 
1144: just making the changes $1\leftrightarrow2$. From Eqs.\ 
1145: (\ref{b6})--(\ref{b11}) and (\ref{a7}), it is easy to prove the identity 
1146: \begin{equation}
1147: \label{b13}
1148: -\gamma_1\zeta_1=F_{11}+F_{12}+\left(G_{11}+G_{12}\right)\gamma_1+
1149: H_{12}\gamma_2,
1150: \end{equation}
1151: which is in fact required by the partial energy conservation 
1152: equation (\ref{2.19}) to support the solution found for the simple shear 
1153: flow problem. In the case of mechanically 
1154: equivalent particles, the expression of the collisional moments 
1155: $A_{k\ell}^{ij}$ are
1156: \begin{equation}
1157: \label{b12}
1158: A_{k\ell}^{11}+A_{k\ell}^{12}=\frac{4}{15}(1+\alpha)\left[(2+\alpha)
1159: \delta_{k\ell}-\frac{3}{2}(3-\alpha)P_{k\ell}\right].
1160: \end{equation} 
1161: When $\alpha=1$, Eq.\ (\ref{b12}) reduces to the results derived from the 
1162: Boltzmann equation in the first Sonine approximation. 
1163: 
1164: 
1165: 
1166: \section{Low dissipation limit}
1167: \label{appC}  
1168: 
1169: In this Appendix we derive the expressions for the main quantities of the 
1170: simple shear flow problem in the low-dissipation limit. The expansions in 
1171: powers of $\epsilon\equiv(1-\alpha^2)^{1/2}$ are given by Eqs.\ 
1172: (\ref{2.34})--(\ref{2.38}). For symmetry reasons, the expansion of 
1173: $P_{i,xy}^*$ has only odd powers, while those of the normal stresses (and 
1174: of the temperature ratio) have only even powers. However, from a practical 
1175: point of view, it is simpler to use $a^*\equiv a/\nu$ as a perturbation 
1176: parameter instead of $\epsilon$. Thus, the expansions 
1177: (\ref{2.34})--(\ref{2.38}) can be written in terms of $a^*$ as
1178: \begin{equation}
1179: \label{c1}
1180: \alpha=1+\alpha_0 a^{*2}+\cdots,
1181: \end{equation}
1182: \begin{equation}
1183: \label{c2}
1184: \gamma=1+\gamma_0 a^{*2}+\cdots,
1185: \end{equation}
1186: \begin{equation}
1187: \label{c3}
1188: P_{i,yy}^*=1+P_{i,yy}^{(2)}a^{*2}+\cdots,
1189: \end{equation}
1190: \begin{equation}
1191: \label{c4}
1192: P_{i,xx}^*=1+P_{i,xx}^{(2)} a^{*2}+\cdots,
1193: \end{equation}
1194: \begin{equation}
1195: \label{c5}
1196: P_{i,xy}^*=P_{i,xy}^{(1)}a^{*}+\cdots.
1197: \end{equation}
1198: Of course, both expansions are directly related, so that 
1199: \begin{equation}
1200: \label{c6}
1201: a_0=\frac{1}{\sqrt{-2\alpha_0}}.
1202: \end{equation}
1203: 
1204: 
1205: The coefficients $P_{i,k\ell}^{(r)}$ can be obtained from Eq.\ (\ref{2.27}) 
1206: by retaining terms up to second order in $a^*$. To do that, we need the 
1207: corresponding expansions of the quantities $F_{ij}$, $G_{ij}$, and $H_{ij}$. 
1208: They are given by
1209: \begin{eqnarray}
1210: \label{c7}
1211: F_{11}&=&\frac{8}{5}\sqrt{2\mu_{21}}x_1\left(\frac{\sigma_{11}}
1212: {\sigma_{12}}\right)^2\left[1+(\case{5}{6}\alpha_0-\case{3}{2}
1213: x_2\gamma_0)a^{*2}+\cdots \right]
1214: \nonumber\\
1215: &\equiv&F_{11}^{(0)}+F_{11}^{(2)}a^{*2}+\cdots,
1216: \end{eqnarray}
1217: \begin{eqnarray}
1218: \label{c8}
1219: G_{11}&=&-\frac{8}{5}\sqrt{2\mu_{21}}x_1\left(\frac{\sigma_{11}}
1220: {\sigma_{12}}\right)^2\left(1+\case{1}{2}\gamma_0x_2a^{*2}+\cdots 
1221: \right)
1222: \nonumber\\
1223: &\equiv&G_{11}^{(0)}+G_{11}^{(2)}a^{*2}+\cdots,
1224: \end{eqnarray} 
1225: \begin{eqnarray}
1226: \label{c9}
1227: F_{12}&=&\frac{16}{5}\mu_{21}x_2
1228: \left(1+\frac{5\alpha_0+\gamma_0(9x_2-7\mu_{12})}{6} a^{*2}+\cdots\right)
1229: \nonumber\\
1230: &\equiv&F_{12}^{(0)}+F_{12}^{(2)}a^{*2}+\cdots,
1231: \end{eqnarray}
1232: \begin{eqnarray}
1233: \label{c10}
1234: G_{12}&=&-\frac{16}{15}x_2\mu_{21}(3+2\mu_{12})
1235: \left\{1+\frac{5\alpha_0\mu_{12}
1236: -\gamma_0\left[4\mu_{12}^2+\mu_{12}(2x_1-1)+3(x_1-1)\right]}
1237: {2(3+2\mu_{12})}a^{*2}+\cdots\right\}
1238: \nonumber\\
1239: &\equiv&G_{12}^{(0)}+G_{12}^{(2)}a^{*2}+\cdots,
1240: \end{eqnarray}
1241: \begin{eqnarray}
1242: \label{c11}
1243: H_{12}&=&\frac{16}{15}x_2\mu_{12}
1244: \left[1+\frac{3}{2}\mu_{21}(\alpha_0-2\gamma_0\mu_{12})a^{*2}+\cdots\right]
1245: \nonumber\\
1246: &\equiv&H_{12}^{(0)}+H_{12}^{(2)}a^{*2}+\cdots.
1247: \end{eqnarray}
1248: The expressions of the coefficients $F_{ij}^{(r)}$, $G_{ij}^{(r)}$, and 
1249: $H_{ij}^{(r)}$ can be easily identified from the above equations. On the 
1250: other hand, the coefficients $F_{22}$, $G_{22}$, $F_{21}$, $G_{21}$, and 
1251: $H_{21}$ are obtained by just making the changes $1\leftrightarrow 2$, 
1252: $\alpha_0\to \alpha_0$ and $\gamma_0\to -\gamma_0$. Substitution of Eqs.\ 
1253: (\ref{c7})--(\ref{c11}) (and their counterparts for species 2) into Eq.\ 
1254: (\ref{2.27}) and considering only terms through second order in $a^*$ allows 
1255: one to get the coefficients $P_{i,k\ell}^{(r)}$ in terms of $\alpha_0$ and 
1256: $\gamma_0$. The final expressions of such coefficients will be omitted here 
1257: since they are very large and not very illuminating.   
1258: 
1259: 
1260: Once the pressure tensors are known, we are in conditions to get the 
1261: coefficients $\alpha_0$ and $\gamma_0$. First, the cooling rates behave as
1262: \begin{eqnarray}
1263: \label{c12}
1264: \zeta_1^*&=&-\frac{4}{3}\left[\left(\sqrt{2\mu_{21}}\left(\frac{\sigma_{11}}
1265: {\sigma_{12}}\right)^2x_1+2\mu_{21}x_2\right)\alpha_0-4\mu_{21}\mu_{12}x_2
1266: \gamma_0\right]a^{*2}+\cdots\nonumber\\
1267: &\equiv& 
1268: \left(A_1\alpha_0+B_1\gamma_0\right)a^{*2}+\cdots\;,
1269: \end{eqnarray}
1270: \begin{eqnarray}
1271: \label{c13}
1272: \zeta_2^*&=&-\frac{4}{3}\left[\left(\sqrt{2\mu_{12}}\left(\frac{\sigma_{22}}
1273: {\sigma_{12}}\right)^2x_2+2\mu_{12}x_1\right)\alpha_0+4\mu_{12}\mu_{21}x_1
1274: \gamma_0\right]a^{*2}+\cdots\nonumber\\
1275: &\equiv& 
1276: \left(A_2\alpha_0+B_2\gamma_0\right)a^{*2}+\cdots\;,
1277: \end{eqnarray}
1278: where again the coefficients $A_i$ and $B_i$ are easily identified. The 
1279: quantities $\alpha_0$ and $\gamma_0$ can be now easily obtained from the 
1280: partial balance of energy (\ref{2.19}), i.e.,
1281: \begin{equation}
1282: \label{c14}
1283: -\frac{2}{3}P_{i,xy}^{(1)}=A_i\alpha_0+B_i\gamma_0 \quad (i=1,2).
1284: \end{equation}
1285: The solution of this system of algebraic equations leads to 
1286: \begin{equation}
1287: \label{c15}
1288: \alpha_0=\frac{2}{3}\frac{B_1 P_{2,xy}^{(1)}-
1289: B_2 P_{1,xy}^{(1)}}{A_1B_2-A_2B_1},
1290: \end{equation}
1291: \begin{equation}
1292: \label{c16}
1293: \gamma_0=\frac{2}{3}\frac{A_2 P_{1,xy}^{(1)}-
1294: A_1P_{2,xy}^{(1)}}{A_1B_2-A_2B_1},
1295: \end{equation}
1296: where 
1297: \begin{equation}
1298: \label{c17}
1299: P_{1,xy}^{(1)}=\frac{H_{12}^{(0)}(F_{22}^{(0)}+F_{21}^{(0)})(G_{11}^{(0)}+
1300: G_{22}^{(0)}+G_{12}^{(0)}+G_{21}^{(0)})-(F_{11}^{(0)}+F_{12}^{(0)})\left[
1301: (G_{22}^{(0)}+G_{21}^{(0)})^2+H_{12}^{(0)}H_{21}^{(0)}\right]}{
1302: \left[(G_{11}^{(0)}+G_{12}^{(0)})(G_{22}^{(0)}+G_{21}^{(0)})-H_{12}^{(0)}
1303: H_{21}^{(0)}\right]^2},
1304: \end{equation}
1305: and $P_{2,xy}^{(1)}$ can be obtained from $P_{1,xy}^{(1)}$ by setting the 
1306: changes $1\leftrightarrow 2$. 
1307: 
1308: In summary, Eqs.\ (\ref{c15}) and (\ref{c16}) give $\alpha_0$ and 
1309: $\gamma_0$, respectively, while $a_0$ is given by (\ref{c6}) 
1310: and the coefficients $P_{i,k\ell}^{(r)}$ are obtained from Eq.\ (\ref{2.27}) 
1311: when only terms through second order in the dissipation parameter 
1312: $\epsilon$ are retained. 
1313:  
1314:  
1315:  
1316: 
1317: 
1318: 
1319: \begin{references}
1320: 
1321: \bibitem{BDKS98}J. J. Brey, J. W. Dufty, C.-S. Kim, A. Santos,
1322: Phys. Rev. E 58 (1998) 4638.
1323: 
1324: \bibitem{RET99}V. Garz\'o, J. W. Dufty, Phys. Rev. E 59 (1999) 5895.
1325: 
1326: \bibitem{BMC99}J. J. Brey, M. J. Ruiz-Montero, D. Cubero,
1327: Europhys. Lett. 48 (1999) 359.
1328: 
1329: \bibitem{JM89}J. T. Jenkins, F. Mancini, Phys. Fluids A 1 (1989) 2050. 
1330:  
1331: \bibitem{Z95}P. Zamankhan, Phys. Rev. E 52 (1995) 4877.
1332: 
1333: \bibitem{AW98}B. Arnarson, J. T. Willits, Phys. Fluids 10 (1998) 1324.
1334: 
1335: \bibitem{WA99}J. T. Willits, B. Arnarson, Phys. Fluids 11 (1999) 3116.
1336: 
1337: \bibitem{FM02}K. Feitosa, N. Menon, cond-mat/0111391. 
1338: 
1339: \bibitem{CH02}M. Alam, R. Clelland, C. M. Hrenya, in: Materials
1340: Research Society Symposium Proceedings 627, San Francisco, 2000; 
1341: R. Clelland, C. M. Hrenya, Phys. Rev. E (2002) (to be published). 
1342: 
1343: \bibitem{GD99}V. Garz\'{o}, J. W. Dufty, Phys. Rev. E 60 (1999) 5706.
1344: 
1345: \bibitem{GD01}V. Garz\'o, J. W. Dufty, Phys. Fluids (2002) (to be published)
1346: and cond-mat/0105395.
1347: 
1348: \bibitem{BC98}J. J. Brey, D. Cubero, Phys. Rev. E 57 (1998) 2019.
1349: 
1350: \bibitem{LSJC84} C. K. K. Lun, S. B. Savage, D. J. Jeffrey, N. 
1351: Chepurniy, J. Fluid Mech. 140 (1984) 223.
1352: 
1353: \bibitem{JR88}J. T. Jenkins, M. W. Richman, J. Fluid Mech. 192 (1988) 313.
1354: 
1355: \bibitem{SGN96}N. Sela, I. Goldhirsch, S. H. Noskowicz,
1356: Phys. Fluids 8 (1996) 2337.
1357: 
1358: \bibitem{CR98}C.-S. Chou, M. W. Richman, Physica A 259 (1998) 430.
1359: 
1360: \bibitem{C01}C.-S. Chou, Physica A 290 (2001) 341. 
1361: 
1362: \bibitem{BMM97}J. J. Brey, M. J. Ruiz-Montero, F. Moreno,
1363: Phys. Rev. E 55 (1997) 2846. 
1364:  
1365: \bibitem{MGSB99}J. M. Montanero, V. Garz\'o, A. Santos, J. J. Brey, 
1366: J. Fluid Mech. 389 (1999) 391.  
1367: 
1368: \bibitem{JM87}J. T. Jenkins, F. Mancini, J. Applied Mech. 54 (1987) 27.
1369: 
1370: \bibitem{note}As a matter of fact, the temperature differences found in Ref.
1371: \onlinecite{JM87} are due to effects of finite density (effects of
1372: excluded area) and so vanish in the low-density limit. 
1373: 
1374: 
1375: 
1376: \bibitem{MP99}P. A. Martin, J. Piasecki, Europhys. Lett. 46 (1999) 613.
1377: 
1378: \bibitem{BRGD99}J. J. Brey, M. J. Ru\'{\i}z-Montero, R. Garc\'{\i}a-Rojo,
1379: J. W. Dufty, Phys. Rev. E 60 (1999) 7174.
1380: 
1381: \bibitem{HWRL00}L. Huilin, L. Wenti, B. Rushan, Y. Lidan, D. Gidaspow,
1382: Physica A 284 (2000) 265.
1383: 
1384: \bibitem{SD01}A. Santos, J. W. Dufty, Phys. Rev. Lett. 86 (2001) 4823.
1385: 
1386: \bibitem{B94}G. Bird, Molecular Gas Dynamics and the Direct Simulation
1387: of Gas Flows, Clarendon, Oxford, 1994.
1388: 
1389: \bibitem{F00}A. Frezzotti, Physica A 278 (2000) 161. 
1390: 
1391: \bibitem{MG01}J. M. Montanero, V. Garz\'o, Gran. Matt. (2002)
1392: (to be published).
1393: 
1394: \bibitem{varios}See for instance, M. A. Hopkins, M. Y. Louge, Phys. Fluids A
1395: 3 (1991) 47; S. B. Savage, J. Fluid Mech. 241 (1992) 109; M. Babic,
1396: J. Fluid Mech. 254 (1993) 127; I. Goldhirsch, M.-L. Tan, G. Zanetti,
1397: J. Sci. Comp. 8 (1993) 1; P. J. Schmid, H. K. Ky”tomaa,
1398: J. Fluid Mech. 264 (1994) 255 (1994). 
1399: 
1400: 
1401: \bibitem{MGS95}C. Mar\'{\i}n, V. Garz\'o, A. Santos, Phys. Rev. E 52 
1402: (1995) 4942; C. Mar\'{\i}n, V. Garz\'o, Phys. Fluids 8 (1996) 2756.  
1403: 
1404: \bibitem{GT96}I. Goldhirsch, M. L. Tan, Phys. Fluids 8 (1996) 1753.
1405: 
1406: 
1407: \bibitem{HS92}M. A. Hopkins, H. H. Shen, J. Fluid Mech. 244 (1992) 477.
1408: 
1409: 
1410: 
1411: \end{references}
1412: 
1413: 
1414: \begin{figure}
1415: \caption{Time evolution of the reduced granular temperature 
1416: $T^*(t)=\nu^2(t)/a^2$ as obtained from Monte Carlo simulation of the 
1417: Boltzmann equation for $\alpha=0.75$, $\protect\sigma_{11}= 
1418: \protect\sigma_{22}$, $n_1/n_2=1/3$, $m_1/m_2=4$, and starting from three 
1419: different initial conditions. Time is measured in units of 
1420: $\lambda_{11}/V_{01}(0)$.}
1421: \label{fig1}
1422: \end{figure}
1423: 
1424: \begin{figure}
1425: \caption{Plot of the reduced granular temperature $T^*=\nu^2/a^2$ versus 
1426: the parameter $(1-\alpha^2)^{-1}$ as obtained from simulation (symbols) and 
1427: the Sonine approximation (lines), for $w=\sigma_{11}/\sigma_{22}=1$, 
1428: $\delta=n_1/n_2=1$ and three different values of the mass ratio 
1429: $\mu=m_1/m_2$: $\mu=10$ (solid line and circles); $\mu=2$ (dashed line and 
1430: squares), and $\mu=1$ (dotted line and triangles).}
1431: \label{fig2}
1432: \end{figure}
1433: 
1434: \begin{figure} 
1435: \caption{Plot of the temperature ratio $\gamma=T_1/T_2$ as a function 
1436: of the restitution coefficient $\alpha$ as obtained from simulation 
1437: (symbols) and the Sonine approximation (lines). We have considered  
1438: $w=\sigma_{11}/\sigma_{22}=1$, $\delta=n_1/n_2=1$ and three different values 
1439: of the mass ratio $\mu=m_1/m_2$: $\mu=10$ (solid line and circles); $\mu=2$ 
1440: (dashed line and squares), and $\mu=1$ (dotted line and triangles).}
1441: \label{fig3}
1442: \end{figure}
1443: 
1444: 
1445: \begin{figure} 
1446: \caption{Plot of the reduced elements of the pressure tensor (a) 
1447: $-P_{xy}^*=-P_{xy}/p$, (b) $P_{xx}^*=P_{xx}/p$, (c) $P_{yy}^*=P_{yy}/p$ and 
1448: (d) $P_{zz}^*=P_{zz}/p$ versus the restitution coefficient $\alpha$ as 
1449: obtained from simulation (symbols) and the Sonine approximation (lines). We 
1450: have considered  $w=\sigma_{11}/\sigma_{22}=1$, $\delta=n_1/n_2=1$ and three 
1451: different values of the mass ratio $\mu=m_1/m_2$: $\mu=10$ (solid line and 
1452: circles); $\mu=2$ (dashed line and squares), and $\mu=1$ (dotted line and 
1453: triangles).}
1454: \label{fig4}
1455: \end{figure}
1456: 
1457: \begin{figure}
1458: \caption{Plot of the temperature ratio $\gamma=T_1/T_2$ as a function of the 
1459: restitution coefficient $\alpha$ as obtained from simulation (symbols) and 
1460: the Sonine approximation (lines). We have considered  
1461: $w=\sigma_{11}/\sigma_{22}=1$, $\mu=m_1/m_2=4$ and two different values of 
1462: the concentration ratio $\delta=n_1/n_2$: $\delta=1/3$ (solid line and 
1463: circles), and $\delta=3$ (dashed line and squares).}
1464: \label{fig5}
1465: \end{figure}
1466: 
1467: 
1468: \begin{figure}
1469: \caption{Plot of the reduced elements of the pressure tensor (a) 
1470: $-P_{xy}^*=-P_{xy}/p$, (b) $P_{xx}^*=P_{xx}/p$, (c) $P_{yy}^*=P_{yy}/p$ and 
1471: (d) $P_{zz}^*=P_{zz}/p$ versus the restitution coefficient $\alpha$ as 
1472: obtained from simulation (symbols) and the Sonine approximation (lines). We 
1473: have considered  $w=\sigma_{11}/\sigma_{22}=1$, $\mu=m_1/m_2=4$ and two 
1474: different values of the concentration ratio $\delta=n_1/n_2$: $\delta=1/3$ 
1475: (solid line and circles), and $\delta=3$ (dashed line and squares).}
1476: \label{fig6}
1477: \end{figure}
1478: 
1479: \begin{figure}
1480: \caption{Plot of the reduced elements of the pressure tensor (a) 
1481: $-P_{xy}^*=-P_{xy}/p$, (b) $P_{xx}^*=P_{xx}/p$, (c) $P_{yy}^*=P_{yy}/p$ and 
1482: (d) $P_{zz}^*=P_{zz}/p$ versus the restitution coefficient $\alpha$ as 
1483: obtained from simulation (symbols) and the Sonine approximation (lines). We 
1484: have considered  $\mu=m_1/m_2=1$, $\delta=n_1/n_2=1$, and two 
1485: different values of the size ratio $w=\sigma_{11}/\sigma_{22}$: $w=2$ 
1486: (solid line and circles), and $w=1$ (dashed line and triangles).}
1487: \label{fig7}
1488: \end{figure}
1489: 
1490: 
1491: \end{document}
1492: 
1493: 
1494: