Article ID: | iaor20043709 |
Country: | Netherlands |
Volume: | 24 |
Issue: | 5 |
Start Page Number: | 223 |
End Page Number: | 234 |
Publication Date: | Jun 1999 |
Journal: | Operations Research Letters |
Authors: | Federgruen Awi, Aviv Yossi |
Keywords: | markov processes |
This paper deals with Markov decision processes with a countable state space. We demonstrate that a single, relatively simple condition suffices to guarantee that the value-iteration method converges and that an optimal policy can be computed via this method, once the existence of a solution to the average cost optimality equation has been established via any of the many available sets of existence conditions.