A characterization of the perfect equilibria of infinite horizon games (Q1072945): Difference between revisions

The paper extends the ideas of \textit{D. Fudenberg} and \textit{D. Levine} [ibid. 31, 251-268 (1983; Zbl 0521.90106)] who have made the following observation: under some circumstances, in an infinite extensive game of n players \(\Gamma\), a sequence of players' strategies is a perfect equilibrium of \(\Gamma\) if and only if it is a limit of a sequence of \(\epsilon_ n\)-perfect equilibria of truncated games \(\Gamma_ n\) \((\Gamma_ n\) arises from \(\Gamma\) by truncating it after n steps), with \(\epsilon_ n\to 0\) while \(n\to \infty.\) The original concept of convergence w.r.t. a metric has been replaced in the paper under review by the convergence w.r.t. a ''weak'' and a ''strong'' topology in the set of all sequences of players' strategies, and w.r.t. any topology in between.