Nash Convergence of Gradient Dynamics in General-Sum Games

Multi-agent games are becoming an increasingly prevalent formalism for the study of electronic commerceand auctions. The speed at which transactions can take place and the growing complexity of electronic marketplaces makes the study of computationally simple agents an appealing direction. In this work, we analyze the behavior of agents that incrementally adapt their strategy through gradient ascent on expected payo , in the simple setting of two-player, two-action, iterated general-sum games, and present a surprisingresult. Weshowthat either the agents will converge to a Nash equilibrium, or if the strategies themselves do not converge, then their average payo s will nevertheless converge to the payo s of a Nash equilibrium.

• Type: Conference
• Year: 2000
• Where: UAI
• Authors: Satinder P. Singh, Michael J. Kearns, Yishay Mansour
• Keywords: Electronic Commerceand Auctions, Nash Equilibrium, Prevalent Formalism, UAI 2000, UAI 2008