Convergence of some time inhomogeneous Markov chains via spectral techniques (Q2372459)

The authors are concerned with finite state inhomogeneous Markov chains whose transition matrices admit the same invariant distribution. Even if this is a very special assumption, a large class of such chains is provided by some time inhomogeneous random walks on finite groups, where the invariant distribution is the uniform one. Rates of convergence to the invariant distribution are provided in terms of the second dominant eigenvalues of certain linear operators associated with the transition matrices. The results obtained allow to give sharp bounds for procedures such as semi-random transpositions and semi-random insertions in a suffling process, that can be considered as inhomogeneous random walks on the symmetric group.