Solving Imperfect Recall Games

Imperfect recall games remain an unexplored part of the game theory, even though recent results in abstraction algorithms show that the imperfect recall might be the key to solving the immense games found in the real world efficiently. Our research objective is to develop the first algorithm capable of outperforming the state-of-the-art solvers applied directly to perfect recall games, by creating and solving its imperfect recall abstraction. To achieve this, we need to answer a set of open questions considering the imperfect recall games. Namely (1) what properties allow the existence of Nash equilibrium in behavioral strategies in imperfect recall games and (2) what information are the players allowed to forget to guarantee that the resulting imperfect recall game can be used to solve the original game with a bounded error.