Singularly perturbed stochastic systems (Q4940419)

Stochastic models of systems are determined by two processes, namely, a switched process which describes the evolution of the system, and a switching process which describes the changes of a random medium. It is assumed that the evolution of a system process has a semigroup property, and the random medium has an ergodic property. The mentioned properties of stochastic systems select the class of systems represented by random evolution regarded as an operator-valued stochastic process in a Banach space. Efficient mathematical tools of analysis are based on the problems of singular perturbation for reducible invertible operators and on the martingale characterization of Markov processes. The purpose of this paper is to order the main schemes of asymptotic analysis of stochastic systems with two scales of time. Stochastic systems are considered in the series scheme with a small series parameter \(\varepsilon >0\) and with two scales of time: real time for a system and rapid time for a switching process. Problems of singular perturbation of the reducible invertible operators are classified and their applications to the analysis of stochastic Markov systems represented by random evolutions are considered. The phase merging, averaging and diffusion approximation schemes are discussed for dynamical systems with rapid Markov switchings.