RE-EXAMINATION OF MARKOV POLICIES FOR ADDITIVE DECISION PROCESS
Toshiharu Fujita · Bulletin of informatics and cybernetics · 1997
The purpose of this paper is to ensure that Markov policy is enough for the additive decision process. We show that there exists an optimal policy which is Markov for both stochastic additive decision process and deterministic one. We also verify through a multi-stage stochastic decision tree method that, among the class of general policies, there exist an optimal Markov policy. This fact is of course obtained by solving the regular recursive equation.