Autonomous stochastic perturbations of dynamical systems (Q1415865)

White-noise-type random perturbations of a dynamical system \(\dot x _{t} = b(x_{t})\) in \(\mathbb R^{n}\) have been studied thoroughly since the sixties of the last century. In this paper, a different type of random perturbations is considered, namely, the system \(\dot X^\varepsilon_{t} = b(X^\varepsilon_{t}) + \varepsilon\beta(X^\varepsilon_{t})\) in \(\mathbb R^{n}\), \(X^\varepsilon_0 = x\), \(\varepsilon>0\) a small parameter, is treated, where \(\beta = (\beta(x),\, x\in\mathbb R^{n})\) is a random field with smooth enough paths. Under suitable hypotheses on \(b\) and \(\beta\), the weak limit of \(X^\varepsilon\) as \(\varepsilon\downarrow 0\) is identified in two particular cases: for the one-dimensional Hamiltonian system \(\ddot q^\varepsilon_{t} + f(q^\varepsilon _{t}) = -\varepsilon\beta(q^\varepsilon_{t})\dot q^ \varepsilon_{t}\) (with \(\beta>0\)) and for a certain Hamiltonian system with many degrees of freedom. The proofs are only sketched.