On limiting behavior of stationary measures for stochastic evolution systems with small noise intensity (Q2193949)

This paper establishes a close connection between deterministic dynamical systems and their stochastic perturbations by considering the limiting behavior of stationary measures for stochastic evolution systems with small random perturbations. The general study is given considering \((\Omega, {\mathcal F}, P)\) a probability space, \((M,\rho)\) a Polish space, and \({\mathcal P}(M)\) the set of all probability measures on \(M\). The main result of the paper is obtained under a probability convergence hypothesis. Let \(\mu^\varepsilon\) be a stationary measure for a stochastic process \(X^\varepsilon\) with small \(\varepsilon\) and \(X^0\) be a semiflow on the Polish space \(M\). If \(\{\mu^\varepsilon : 0<\varepsilon\le\varepsilon_0\}\) is tight, then all their limits in weak sense are \(X^0\)-invariant and their supports are contained in the Birkhoff center of \(X^0\) as \(\varepsilon\) tends to zero. This result, compared with other existing results, which mostly focus on stochastic ordinary differential equations with non-degenerate noise, gives much more precise positions for limiting measures to be concentrated on. Utilizing the obtained results, Bernoulli's lemniscate with non-degenerate noise such that stationary measures converge weakly to a delta measure at a saddle of the drift vector field is constructed. A lot of other examples are given and also applications to stochastic reaction-diffusion equations, stochastic \(2D\) Navier-Stokes equations and stochastic Burgers type equations driven by Brownian motion or Lévy processes are presented.