Random iterates of monotone maps (Q2268902)

Random dynamical systems generated by monotone maps of closed subsets of \(\mathbb{R}^k\) (monotone under the partial order \(x\leq y\) if \(x_j\leq y_j\) for all \(1\leq j\leq k\)) into itself are considered. Invariant distributions for the corresponding Markov processes \(X_{n+1}=\alpha_n(X_n)\), where \(\alpha_n\) is a sequence of i.i.d. monotone maps with certain distribution \(Q\) and \(X_0\) is independent of \(\alpha_n, n=1,2,\dots\), are studied. The main result shows that there exists a unique invariant probability \(\pi\) for the Markov process \(X_n\) which is stable. This result extends a result of \textit{L. E. Dubins} and \textit{D. A. Freedman} [Ann. Math. Stat. 37, 837--868 (1966; Zbl 0147.16404)] to multi-dimensional spaces and improves a result of \textit{R. Bhattacharya} and \textit{M. Majumdar} [J. Theor. Probab. 12, No. 4, 1067--1087 (1999; Zbl 0961.60064)].