Long time behavior of discrete velocity kinetic equations
Gayrat Toshpulatov · 2025
Read Long time behavior of discrete velocity kinetic equations by Gayrat Toshpulatov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-03646.txt.
Introduction
Gayrat Toshpulatov studies the long-time behavior of discrete-velocity kinetic equations, in this 2025 paper. Such equations track gas particles that move at a fixed set of speeds. The work covers one and three dimensions with wraparound boundaries.
Solutions are shown to decay toward the global balance state at an exponential rate in the natural energy measure. The result holds for a wide class of collision rules, including two classic named models. The decay rates are explicit and constructive, not just claimed to exist. The proof builds Lyapunov functions by reshaping the classic entropy of Boltzmann.
Catalog record
| CREATOR | Gayrat Toshpulatov |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 76 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-03646
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Long time behavior of discrete velocity kinetic equations Gayrat Toshpulatov Abstract We study long time behavior of some nonlinear discrete velocity kinetic equations in the one and three dimensions with periodic boundary conditions. We prove the exponential time decay of solutions towards the global equilibrium in the L 2 L^{2} space. Our result holds for a wide class of interaction rates including the Goldstein-Taylor and Carleman equations, and the estimates on the rate of convergence are explicit and constructive. The technique is based on the construction of suitable Lyapunov functionals by modifying Boltzmann’s entropy. Keywords: Kinetic theory of gases, discrete velocity models, the Goldstein-Taylor equation, the Carlemann equation, convergence to equilibrium, hypocoercivity. 2020 Mathematics Subject Classification: 35Q82, 35B40, 82C40. 1 Introduction We consider the initial value problem \begin{cases}\partial_{t}u+c\partial_{x}u=k(u,v,x)(v-u),\,\,\,\,&x\in\mathbb{T},\,\,t>0,\\ \partial_{t}v-c\partial_{x}v=k(u,v,x)(u-v),\,\,\,\,\,&x\in\mathbb{T},\,\,t>0,\\ u_{|t=0}=u_{0},\,\,v_{|t=0}=v_{0},&x\in\mathbb{T}.\end{cases} (1) This equation describes the evolution of the velocity distribution of a gas composed of two species that move at a constant speed c>0 in the x- direction. The variables t\geq 0 and x\in\mathbb{T}=\mathbb{R}/\mathbb{Z} stand for time and position, respectively. The first unknown u=u(t,x) denotes the density function of particles moving at speed c in the positive x- direction, while the second unknown v=v(t,x) is the density function of particles moving at speed c in the negative x- direction. The function k(u,v,x) is non-negative and describes the interaction between gas particles. For example, the case k(u,v,x)=1 is known as the Goldstein-Taylor model [ 14 ] , [ 27 ] , which can be converted to the telegraph equation. The case k(u,v,x)=u+v was introduced by Carlemann in [ 7 ] to describe the interaction of two type of particles such that an interaction of two particles of the first type results into two particles of the second type and vice versa. The macroscopic variables for ( 1 ) are the mass density \rho\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}u+v and the momentum j\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}u-v. With this notations, ( 1 ) is equivalent to \begin{cases}\partial_{t}\rho+c\partial_{x}j=0,\\ \partial_{t}j+c\partial_{x}\rho=-2kj.\end{cases} (2) Equation ( 1 ) can be extended to the three dimensions. In this case the gas particles can move in directions parallel to one of the axes x_{1},x_{2},x_{3} either in the positive direction or the negative one at a constant speed c>0 . Let u_{i}=u_{i}(t,x_{1},x_{2},x_{3}) (respectively v_{i}=v_{i}(t,x_{1},x_{2},x_{3}) ) denote the density of particles moving at speed c in the positive (respectively negative) x_{i}- direction for i\in\{1,2,3\} . Then we have \begin{cases}\partial_{t}u_{i}+c\partial_{x_{i}}u_{i}=k(u_{1},u_{2},u_{3},v_{1},v_{2},v_{3},x)(\rho-6u_{i})\,\,\,\,&x\in\mathbb{T}^{3},\,\,t>0,\\ \partial_{t}v_{i}-c\partial_{x_{i}}v_{i}=k(u_{1},u_{2},u_{3},v_{1},v_{2},v_{3},x)(\rho-6v_{i}),\,\,\,\,&x\in\mathbb{T}^{3},\,\,t>0,\\ {u_{i}}_{|t=0}=u_{i,0},\,\,{v_{i}}_{|t=0}=v_{i,0},\,\,\,\,&x\in\mathbb{T}^{3},\end{cases} (3) where \rho\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}\sum_{i=1}^{3}(u_{i}+v_{i}) denotes mass density function as in the one dimensional case. We denote the mass densities in the x_{i}- directions as \rho_{i}\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}u_{i}+v_{i},\,\,\,i\in\{1,2,3\}. The momentum is defined as j\mathrel{\mathchoice{\vbox{\hbox{$\displaystyle:$}}}{\vbox{\hbox{$\textstyle:$}}}{\vbox{\hbox{$\scriptstyle:$}}}{\vbox{\hbox{$\scriptscriptstyle:$}}}{=}}\begin{pmatrix}u_{1}-v_{1}\\ u_{2}-v_{2}\\ u_{3}-v_{3}\end{pmatrix}\in\mathbb{R}^{3}. With this notations, the following macroscopic system can be derived from ( 3 ) \begin{cases}\partial_{t}\rho+c\mathrm{div}_{x}j=0,\\ \partial_{t}j+c\begin{pmatrix}\partial_{x_{1}}\rho_{1}\\ \partial_{x_{2}}\rho_{2}\\ \partial_{x_{3}}\rho_{3}\end{pmatrix}=-6kj.\end{cases} (4) Whenever u,v is a classical solution to ( 1 ), by integrating the first equation in ( 2 ) we obtain conservation of mass \int_{\mathbb{T}}(u(t,x)+v(t,x))dx=\int_{\mathbb{T}}(u_{0}(x)+v_{0}(x))dx,\,\,\,\,\forall\,t\geq 0. Another interesting property of ( 1 ) is that a family of functionals decay under the time evolution of the classical solution u,v . Let \varphi be a C^{1} convex function defined on [0,\infty) (e.g., \varphi(u)=u^{p},\,p\geq 1 ). Multiplying the first and second equations in ( 1 ) by \varphi^{\prime}(u) and \varphi^{\prime}(v), respectively, and integrating over \mathbb{T}, we obtain \frac{d}{dt}\int_{\mathbb{T}}(\varphi(u)+\varphi(v))dx=-\int_{\mathbb{T}}k(u,v,x)(u-v)(\varphi^{\prime}(u)-\varphi^{\prime}(v))dx. (5) Since \varphi is convex, the right hand side of this equation is non-positive. Hence, the integral \int_{\mathbb{T}}\big[\varphi(u)+\varphi(v)\big]dx is a non-increasing function of t\geq 0. The right hand side of ( 5 ) vanishes if u(t,x)=v(t,x) for all t>0 and x\in\mathbb{T}. We call such functions as a local equilibrium . One can check that a local equilibrium u,v solves ( 1 ) if u and v are equal to a positive constant m_{\infty} for all t\geq 0 and x\in\mathbb{T}. We call such functions as a global equilibrium . The three dimensional equation ( 3 ) has the same properties as the one dimensional one ( 1 ). The initial mass is conserved \int_{\mathbb{T}^{3}}\sum_{i=1}^{3}(u_{i}(t,x)+v_{i}(t,x))dx=\int_{\mathbb{T}^{3}}\sum_{i=1}^{3}(u_{0,i}(x)+v_{0,i}(x))dx,\,\,\,\,\,\forall\,t\geq 0. As in the one dimensional case, for any convex function \varphi:[0,\infty)\to\mathbb{R}, one can check that \displaystyle\frac{d}{dt}\int_{\mathbb{T}^{3}}\sum_{i=1}^{3}(\varphi(u_{i})+\varphi(v_{i}))dx= \displaystyle-\frac{1}{2}\int_{\mathbb{T}^{3}}k\sum_{\begin{subarray}{c}i,j=1\\ i\neq j\end{subarray}}^{3}(\varphi^{\prime}(u_{i})-\varphi^{\prime}(u_{j}))(u_{i}-u_{j})dx \displaystyle-\frac{1}{2}\int_{\mathbb{T}^{3}}k\sum_{\begin{subarray}{c}i,j=1\\ i\neq j\end{subarray}}^{3}(\varphi^{\prime}(v_{i})-\varphi^{\prime}(v_{j}))(v_{i}-v_{j})dx \displaystyle-\frac{1}{2}\int_{\mathbb{T}^{3}}k\sum_{i,j=1}^{3}(\varphi^{\prime}(u_{i})-\varphi^{\prime}(v_{j}))(u_{i}-v_{j})dx\leq 0. (6) Hence, the integral \int_{\mathbb{T}^{3}}\sum_{i=1}^{3}(\varphi(u_{i})+\varphi(v_{i}))dx is a non-increasing function of t\geq 0. From ( 1 ), we also conclude that u_{i},v_{i} is a local equilibrium (i.e., the right hand side of ( 1 ) vanishes) if u_{i}(t,x)=v_{i}(t,x)={\rho(t,x)}/{6} for all i\in\{1,2,3\}, t\geq 0, and x\in\mathbb{T}^{3}, where we remind \rho=\sum_{i=1}^{3}(u_{i}+v_{i}). u_{i},v_{i} is a global equilibrium if there exists m_{\infty}>0 and u_{i}(t,x)=v_{i}(t,x)=m_{\infty} for all i\in\{1,2,3\}, t\geq 0, and x\in\mathbb{T}^{3}. Let u(t,x),v(t,x) be the solution of the one dimensional equation ( 1 ). On the basis of the decay of the integral \int_{\mathbb{T}}(\varphi(u)+\varphi(v))dx , one can expect that \int_{\mathbb{T}}(\varphi(u)+\varphi(v))dx decays its minimum as t\to\infty. From ( 5 ) we see that u,v solves ( 1 ) and minimizes \int_{\mathbb{T}}(\varphi(u)+\varphi(v))dx at the same time if u=v=m_{\infty} for some constant m_{\infty} and for all t\geq 0,\,x\in\mathbb{T}. Hence, one can conjecture that u(t,x) and v(t,x) converges to a constant m_{\infty} as t\to\infty. Since u(t,x)+v(t,x) has the same as u_{0}(x)+v_{0}(x) , we expect that m_{\infty}=\frac{1}{2}\int_{\mathbb{T}}(u_{0}(x)+v_{0}(x))dx. The same conjecture can be formulated for the three dimensional equation ( 3 ), i.e., u_{i}(t,x) and v_{i}(t,x) converge to m_{\infty}=\frac{1}{6}\int_{\mathbb{T}^{3}}\sum_{i=1}^{3}(u_{i,0}(x)+v_{i,0}(x))dx as t\to\infty for all i\in\{1,2,3\}. Thus, it is an interesting problem to prove (or disprove) these convergences and to estimate the convergence rate. Our goal is to prove these convergences for the equations ( 1 ) and ( 3 ) with explicit and constructive convergence rates. There are many works concerning the equations ( 1 ) and ( 3 ). We refer [ 13 , 5 , 23 , 19 ] for a general introduction to the discrete velocity kinetic equations. Existence and uniqueness of weak solutions to ( 1 ) and ( 3 ) was proven by Lions and Toscani in [ 19 ] ; see Theorem 2.2 and 2.5 below. The diffusion approximation of ( 1 ) has been studied mostly for the case k(u,v,x)=(u+v)^{\alpha} with \alpha\leq 1. In this case, the asymptotics leads to the fast diffusion equation if \alpha\in[0,1], the slow diffusion equation if \alpha\in[-1,0), and the porous media equation if \alpha<-1; see [ 24 , 19 , 25 , 15 ] . Concerning the large time behavior, there are studies only for the Goldstein-Taylor equation (i.e., k(u,v,x)=1 ) and the Carleman equation (i.e., k(u,v,x)=u+v ). In [ 8 ] , algebraic decay of solutions to its equilibrium in L^{2}(\mathbb{T}) was proved for the Goldstein-Taylor equation with a spatially dependent interaction rate k=k(x)\in H^{1}(\mathbb{T}) and for initial data u_{0},v_{0}\in H^{2}(\mathbb{T}). The authors used the entropy method [ 9 ] which consists of obtaining differential inequalities for the entropy and its dissipation. Bernard and Salvarani in [ 3 ] proved exponential convergence to equilibrium for the Goldstein-Taylor equation with initial data u_{0},v_{0}\in H^{1}(\mathbb{T}) by using its connection to the telegraph equation. Their decay rate is optimal and the interaction rate k=k(x)\geq 0 does not have to be bounded from below by a positive constant. Similar results were obtained in [ 1 ] for the one and three dimensional Goldstein-Taylor equations by constructing suitable Lypunov functionals. The long time behavior of C^{1} solutions to the one dimensional Carleman equation was studied by Illner and Reed in the whole domain [ 17 ] and in the torus [ 18 ] . The authors proved that