Optimal average value convergence in nonhomogeneous Markov decision processes (Q1323097)

This paper deals with an infinite state nonhomogeneous Markov decision process with average reward criterion. The authors proved the following two structural results: (1) Under the Doeblin condition, the problem is equivalent to a discounted problem. (2) Under the same condition, the optimal finite horizon average values converge to the infinite horizon optimal one.