On the existence of good stationary strategies for nonleavable stochastic games (Q1972563)

The main result of this paper asserts that in any nonleavable game with finite state space, there is a uniformly \(\varepsilon\)-optimal stationary strategy available to player II. The method also yields that in nonleavable games with countably infinite state space, if the value function \(V\) of the nonleavable game is greater than or equal to the utility function \(u\), there exists a uniformly optimal strategy for player II. If \(V(x)< u(x)\) for some \(x\), there need not be an optimal strategy for player II even when the state space of the game is finite.