Average optimality in dynamic programming on Borel spaces — unbounded costs and controls

Abstract

We show the existence of average-cost optimal stationary policies in a class of discrete-time Markov control processes with Borel state space, unbounded costs, and non-compact control sets.