Given a two-player zero-sum discounted-payoff stochastic game, we introduce three classes of continuous-time best-response dynamics, stopping-time best-response dynamics, closed-loop best-response dynamics, and open-loop best-response dynamics. We show the global convergence of the first two classes to the set of minimax strategy profiles, and the convergence of the last class when the players ...

Subjective games of incomplete information are formulated where some of the key assumptions of Bayesian games of incomplete information are relaxed. The issues arising because of the new formulation are studied in the context of a class of nonzero-sum, two-person games, where each player has a different model of the game. The static game is investigated in this note. It is shown that the proper...

We introduce a method of decomposing a p-player normal form game into 2p simultaneously-played component games, each distinguished by the set of “active” players whose choices influence payoffs. We then prove that a normal form game is a potential game if and only if in each of the component games, all active players have identical payoff functions, and that in this case, the sum of these share...

We consider a game in which “meta-players” choose finite automata to play a repeated stage game. Meta-players’ utilities are lexicographic, first increasing in the (limit-of-the-means) payoffs of the repeated game and second decreasing in the number of states in their automaton. We examine the outcomes in this game which satisfy a version of evolutionary stability that has been modified to perm...

A coupling game is a new model for partly cooperative and partly non-cooperative games. This paper describes benefits that result when coupling rational cooperation, under some situations, is increased. We prove that if a coupling game is strictly rational and partly cooperative, the higher the coupling factors, the larger the social payoff, which is the sum of the payoffs of all players. Other...

We study a zero sum differential game under strong assumptions of convexity — the cost is convex for one player, and concave for the other. An explicit necessary and sufficient condition for a saddle point of the game is given in terms of convex analysis subgradients of the conjugate of the cost function. A generalized Hamiltonian equation is shown to describe saddle trajectories of the game.

We prove that for a class of zero-sum differential games with incomplete information on both sides, the value admits a probabilistic representation as the value of a zero-sum stochastic differential game with complete information, where both players control a continuous martingale. A similar representation as a control problem over discontinuous martingales was known for games with incomplete i...

We present a game-theoretic account of irrational agent behavior and define conditions under which irrational behavior may be considered quasi-rational. To do so, we use a 2-player, zero-sum strategic game, parameterize the reward structure and study how the value of the game changes with this parameter. We argue that for any “underdog” agent, there is a point at which the asymmetry of the game...

This paper presents a novel coevolutionary model which is used to create strong game (The Virus Game) playing strategies. We use two approaches to coevolve Artificial Neural Networks (ANN) which evaluate board positions of a two player zero-sum game (The Virus Game). The first approach uses the coevolution with initial population of random ANN and second approach is a novel coevolutionary model...