An improved algorithm for solving communicating average reward Markov decision processes (1991)

by M Haviv, M L Puterman