🤖 AI Summary
This paper addresses the computational tractability of correlated equilibria in max-polymatrix games—multi-player games where each player’s utility is the maximum over marginal payoffs from pairwise interactions. For this long-standing open problem, we establish, for the first time, the polynomial expectation property for such games. Leveraging this property, we design a polynomial-time algorithm that explicitly constructs a correlated equilibrium. Theoretical analysis proves that a correlated equilibrium with polynomially bounded support always exists and can be computed in polynomial time. This result resolves the fundamental questions of existence and efficient computability of correlated equilibria in max-polymatrix games. Moreover, it yields the first equilibrium computation framework for high-dimensional interactive decision-making settings with rigorous polynomial-time guarantees.
📝 Abstract
We address an open question which addresses the computability of correlated equilibria in a variant of polymatrix where each player's utility is the maximum of their edge payoffs. We demonstrate that this max-variant game has the polynomial expectation property, and conclude that there indeed exists a polynomial correlated equilibrium scheme.