Limit theorems for random walks with dynamical random transitions (Q1343605): Difference between revisions

The authors consider the following model (called ``dynamical random transitions'') of introducing additional randomness into transition probabilities of a random walk on a group \(G\). Let \(\theta\) be an ergodic conservative measure type preserving transformation of a \(\sigma\)-finite measure space \((\Omega,m)\), and \(\omega\mapsto \mu_ \omega\) be a measurable map from \(\Omega\) to the space of probability measures on \(G\). Then any given \(\omega \in \Omega\) gives rise to an (inhomogeneous in time) Markov chain on \(G\) whose transition probabilities at time \(n\) are determined by the measure \(\mu_{\theta^ n \omega}\). Using general methods of the theory of group representations in Banach spaces the authors find conditions when the resulting Markov chain on \(G\) is mixing for almost every \(\omega\) (i.e., its tail or asymptotic \(\sigma\)-algebra is trivial) in the case when \(G\) is either Abelian or compact. For compact groups such results were earlier obtained by \textit{D. S. Mindlin} and \textit{B. A. Rubshtejn} [Theory Probab. Appl. 33, No. 2, 355-357 (1988); translation from Teor. Veroyatn. Primen. 33, No. 2, 376-379 (1988; Zbl 0651.60011) and Ann. Inst. Henri Poincaré, Probab. Stat. 30, No. 2, 213-233 (1994; Zbl 0808.60012)] assuming stationarity, i.e., when the transformation \(\theta\) is measure preserving and the measure \(m\) is finite.