Burnetas, A.N., 1995. Computing Optimal Policies for Markovian Decision Processes Using Simulation. Probability in the Engineering and Informational Sciences, 9, pp.525-537.