No label defined (Q4540411)

The authors give sufficient conditions for the stability and boundedness in the \(p\)th mean of stochastic dynamical systems with an \(n\)-dimensional state \(x_t\) which apply equally to discrete-time, continuous-time and hybrid systems. (For instance, they guarantee that the \(p\)th mean of the distance from an invariant set does not exceed a prescribed threshold, if the system starts close enough to the invariant set.) The conditions are stated in terms of certain drift conditions [in the sense of \textit{S. P. Meyn} and \textit{R. L. Tweedie}, Markov chains and stochastic stability (Springer, Berlin) (1993; Zbl 0925.60001)] that is with the help of a discrete-time version of a Lyapunov function \(V(x,t)\) and the means of its increments \( \Delta V(x_{t_k},t_k) = V(x_{t_{k+1}},t_{k+1}) - V(x_{t_k},t_k)\), \(t_k \rightarrow \infty\) as \( k \rightarrow \infty\). As an application they show that the trivial solution (state, control)\(= (0,0)\) of a discrete-time feedback control system that is linear in the state and perturbed by multiplicative (or additive) noise is uniformly asymptotically stable in the \(p\)th mean (or uniformly ultimately bounded in the \(p\)th mean, respectively), given certain algebraic conditions for the coefficient matrices and sufficiently small noise intensity.