On the Convergence of Rolling Horizon Procedure and the Average Criterion
Abstract
We deal with the average criterion on Markov Decision Processes (MDP) to evaluate the performance of the policies. In IEEE A.Control, 35(10), Hernández-Lerma and Lasserre give some sufficient conditions for the convergence of the rolling horizon procedure. We present more general conditions based on properties of Markov chains and value iteration procedure.