Average optimality inequality for continuous-time Markov decision processes in Polish spaces (Q2472191): Difference between revisions

This paper is concerned with the average cost optimality criterion for continuous-time jump Markov decision processes. Assuming some regularity conditions and exponential uniform ergodicity, the author establishes the optimality inequality by employing a vanishing discount factor approach, Fatou's lemma and Tauberian theorem. An optimal policy is obtained as a measurable selector from this optimality inequality. A similar result for discrete-time Markov decision processes was obtained by \textit{M. Schäl} [Math. Oper. Res. 18, No. 1, 163--172 (1993; Zbl 0777.90079)].