Local thermodynamic equilibrium for some stochastic models of Hamiltonian origin (Q2385203): Difference between revisions

The authors study a class of one-dimensional models introduced in \textit{J.-P. Eckmann} and \textit{L.-S. Young} [Europhys. Lett. 68, 790--796 (2004; Zbl 1136.82368)] that are stochastic realisations of Hamiltonian models of heat conduction. They prove that in the infinite volume limit a local equilibrium is attained with a linear energy profile. The model can be briefly described as follows. Let \(\xi_i(t)\) be the energy stored at site \(i\in\{1,\dots,n\}\) at time \(t\). The sites exchange the energy by means of a tracer particle whose position and energy are denoted by \(\alpha (t)\) and \(\eta (t)\). Given a state \((\xi_i(t),\alpha (t),\sigma (t))\) at the time \(t\), the system waits an exponentially distributed time with the mean \(\sqrt {\eta (t)}\) and then jumps to a new state: the tracer particle exchanges its energy with the site \(\alpha (t)\), according some rules, and then jumps to one of the neighbours of \(\alpha (t)\) with the same probability, all the other energies are left unchanged. If the tracer jumps out of \(\{1,\dots,n\}\), it is replaced instantly by a new tracer (started at \(1\) or \(n\)) whose energy is exponentially distributed with mean \(T_L\) or \(T_R\), denoting the temperatures of reservoirs on the left, resp. right, side of the system. The authors show that for any \(T_L\) and \(T_R\) there is an equilibrium probability distribution, for which they give an exact formula if \(T_L=T_R\). They further study the projection of the equilibrium distribution on the energies of a finite number of sites \(\{xN-l,\dots,xN+l\}\), \(x\in (0,1)\). They show that this projection converges as \(N\to \infty\) to the equilibrium measure with \(T=x T_R + (1-x) T_L\) projected to \(2l +1 \) sites.