Adaptive artificial chemistry for Nash equilibria approximation

Abstract

A simple artificial chemistry model designed for computing Nash equilibria of continuous games, called adaptive artificial chemistry for Nash equilibria is presented. This method mimics elementary chemical reactions between molecules representing strategy profiles of a non-cooperative game while using a generative relation for Nash equilibria in order to direct the search towards the equilibrium. Experimental results - indicating the potential of the proposed method - are performed on Cournot oligopolies for up to 1,000 players.