Finite approximation to a zero-sum game with incomplete information (Q922970)

The authors investigate a scheme for approximating a two-person zero-sum game G of incomplete information by means of natural systems \(G_{mn}\) of finite subgames. They prove two theorem which say: Suppose that \(\alpha_{mn}\) is an optimal strategy for player I in \(G_{mn}\), and that \(\bar F_ m\) and \(G_ n\) are finite \(\sigma\)-fields. Then \[ \lim_{m\to \infty}\lim_{n\to \infty}Val_ G(\alpha_{mn})=V. \] Moreover, this convergence is uniform in the choices \(\alpha_{mn}\) is optimal strategies. The second result says that under an additional assumption first used by Milgran-Weber, if for \(m,n=1,2,...\), \(\alpha_{mn}\) is an optimal strategy for I in \(G_{mn}\), then \[ \lim_{m\to \infty,n\to \infty}Val_ G(\alpha_{mn})=V \] uniformly in the choices of optimal strategies.