In this paper we develop a discretized version of the dynamic programming algorithm and derive error bounds for the approximate value and policy functions. We show that under the proposed scheme the computed value function converges quadratically to the true value function and the computed...