Computing Better Approximate Pure Nash Equilibria in Payoff-maximization Potential Games
TL;DR
It is shown that standard approaches based on unilateral improvement moves can fail to provide any finite approximation guarantee even for simple extensions of Pd--Flip games, so an algorithmic framework based on coordinated moves by small groups of players is proposed, whose approximation guarantee and number of moves are controlled by two natural game parameters, the stretch and the spread.
Abstract
Potential games are a fundamental class of games in which pure Nash equilibria are guaranteed to exist, yet computing such equilibria is computationally intractable for several subclasses. This has led to extensive research on computing approximate pure Nash equilibria. In this paper, we study payoff-maximization potential games. For these games, strong approximation guarantees are known only for restricted subclasses, most notably Pd--Flip games. We show that standard approaches based on unilateral improvement moves can fail to provide any finite approximation guarantee even for simple extensions of Pd--Flip games. To overcome this limitation, we propose an algorithmic framework based on coordinated moves by small groups of players, whose approximation guarantee and number of moves are controlled by two natural game parameters, the stretch and the spread, which are bounded for broad classes of games of interest. In the special case of Pd--Flip games, our framework can be configured to recover the existing algorithm, matching its approximation guarantee and the number of moves performed.