New average optimality conditions for semi-Markov decision processes in Borel spaces (Q438786)

This paper deals with semi-Markov decision processes on a Borel space with the so-called ratio-average expected cost criterion. The objective is to provide a set of conditions under which there exists an optimal stationary policy. The key assumption is that the so-called relative difference of the optimal discounted cost value function is bounded by integrable two functions. Further, the authors make use of the well-known \textit{P. J. Schweitzer's} data transformation [J. Math. Anal. Appl. 34, 495--501 (1971; Zbl 0218.90070)] and analyze two optimality inequalities. The optimal policy is obtained as a selector of the minima in one of the inequalities. Similar techniques of semi-Markov decision processes with unbounded cost functions were used by different authors, see for example [\textit{F. Luque-Vásquez} and \textit{O. Hernández-Lerma}, Appl. Math. 26, No. 3, 315--331 (1999; Zbl 1050.90566)]; [\textit{A. Jaśkiewicz}, Math. Methods Oper. Res. 54, No. 1, 1--19 (2001; Zbl 1031.90062)]; [\textit{A. Federgruen}, \textit{P. J. Schweitzer} and \textit{H. C. Tijms}, Math. Oper. Res. 8, 298--313 (1983; Zbl 0513.90085)].