AN ALGORITHM TO OBTAIN AN OPTIMAL STRATEGY FOR THE MARKOV DECISION PROCESSES, WITH PROBABILITY DISTRIBUTION FOR THE PLANNING HORIZON

Abstract

In this paper we formulate Markov Decision Processes with Random Horizon. We show the optimality equation for this problem, however there may not exist optimal stationary strategies. For the MDP (Markov–Decision–Process), with probability distribution for the planning horizon with infinite support, we show Turnpike Planning Horizon Theorem. We develop an algorithm obtaining an optimal first stage decision. We give some numerical examples.