Infinite products of random matrices and repeated interaction dynamics (Q985329)

The paper considers products of independent, identically distributed random matrices, which are contractions for some suitable norm, and which have a deterministic common eigenvector with associated eigenvalue~\(1\). The product can then be written as the sum of a fluctuating and of a decaying process. While the decaying process converges to zero exponentially fast, the Cesàro averages of the fluctuating process converge to a limit which can be characterized by the common eigenvector and by the spectral data of the expectation of the matrices. The results are applied to inhomogeneous Markov chains with random transition matrices and to random repeated interaction open quantum systems.