On discounted approximations of undiscounted stochastic games and Markov decision processes with limited randomness

摘要

It is shown that the discount factor needed to solve an undiscounted mean payoff stochastic game to optimality is exponentially close to 1, even in one-player games with a single random node and polynomially bounded rewards and transition probabilities. For the class of the so-called irreducible games with perfect information and a constant number of random nodes, we obtain a pseudo-polynomial algorithm using discounts.