A new condition for the existence of optimal stationary policies in average cost Markov decision processes (Q1076617): Difference between revisions

The author considers a discrete time, countable state Markov decision processes with finite decision sets and bounded costs. He obtains conditions under which (possibly) unbounded solution to the average cost equation for the optimal value exists and yields an optimal stationary policy. In the special case in which every stationary policy induces an ergodic Markov chain, he obtains a new form for the optimality equation and gives a sufficient condition for the existence of an optimal stationary policy. The results are illustrated by some examples.