Computing Equilibria of Two-Person Games from the Extensive Form

The Lemke-Howson algorithm for computing equilibria of finite 2-person non-cooperative games in normal form is modified to restrict the computations to the ordinarily small portion corresponding to the strategies actually used by the players, and further it is shown that in games with perfect recall these strategies can be generated as needed from an auxiliary analysis of the players' decision trees derived from the extensive form of the game.

• 저널Management Science · 18(7) · 448–460
• 토픽Artificial Intelligence in Games · Artificial Intelligence
• DOI10.1287/mnsc.18.7.448
• 저자Robert Wilson