A note on strong 1-optimal policies in Markov decision chains with unbounded costs

Abstract.

We extend a result by Cavazos-Cadena and Lasserre on the existence of strong 1-optimal stationary policies in Markov decision chains with countable state spaces, uniformly ergodic transition probabilities and bounded costs to a larger class of models with unbounded costs and the so-called V-uniform ergodic transition structure.