A Reinforcement Learning Algorithm Based on Policy Iteration for Average Reward Empirical Results with Yield Management and Convergence Analysis. di Gosavi, Abhijit