Some Remarks on Optimality Conditions for Markovian Decision Process

Abstract

This paper is concerned with a discrete time parameter Markovian decision problem. The expected total returns for infinite horizon are considered as the power series of discount factor. Some optimality criterions, for example, β-optimal, 1-optimal and so on, are discussed from the view point of the theory of infinite series. And a new optimality criterion is introduced. This criterion is valuable to construct an intuitive optimal policy theoretically.