Quasistable gradient and Hamiltonian systems with a pairwise interaction randomly perturbed by Wiener processes. (English) Zbl 1027.60061

Skorokhod, Anatoli V.

J. Appl. Math. Stochastic Anal. 16, No. 1, 45-67 (2003).

Let \(w_k\) be a sequence of independent standard Wiener processes in \(R^d\), \(x_k\) be a finite or infinite sequence of \(R^d\)-valued stochastic processes that describe the evolution of a set of randomly perturbed pairwise interacting particles. Gradient and Hamiltonian systems of the form \[ dx_k(t)=F_k(t) dt+\sigma dw_k(t) \quad\text{and}\quad d\dot{x}_k(t)=F_k(t) dt+ \sigma dw_k(t), \] respectively, are considered, where \(F_k=-\sum_{j\neq k}U_k(x_k(t)-x_j(t))\). The assumptions on the potential \(U\) and on initial conditions, sufficient for the quasistability of both the systems, are established. Here quasistability means that any joint distributions of corresponding finite \(N\)-particle systems have a limit as \(N\rightarrow\infty\). The main assumptions include the existence of gradient (Hamiltonian) free particle system, the relation \(\lim_{s\downarrow 0}s^{\alpha+1}u''(s)=a\) for \(\alpha<1\), and some estimates for the initial configuration.

Reviewer: Yuliya S.Mishura (Kyïv)

MSC:

60H10	Stochastic ordinary differential equations (aspects of stochastic analysis)
60H30	Applications of stochastic analysis (to PDEs, etc.)
34F05	Ordinary differential equations and systems with randomness
37Jxx	Dynamical aspects of finite-dimensional Hamiltonian and Lagrangian systems
60G46	Martingales and classical analysis
60K35	Interacting random processes; statistical mechanics type models; percolation theory

Keywords:

gradient and Hamiltonian systems; Wiener process; quasistability