Modeling Variation in Cooperative Coevolution Using Evolutionary Game Theory R. Paul Wiegand Department of Computer Science George Mason University Fairfax, VA 22030 paul@tesseract.org William C. Liles Department of Computer Science George Mason University Fairfax, VA 22030 wliles@gmu.edu Kenneth A. De Jong Department of Computer Science George Mason University Fairfax, VA 22030 kdejong@gmu.edu Abstract Though coevolutionary algorithms are currently used for optimization purposes, practi- tioners are often plagued with difficulties due to the fact that such systems frequently behave in counter intuitive ways that are not well understood. This paper seeks to extend work which uses evolutionary game theory (EGT) as a form of dynamical systems mod- eling of coevolutionary algorithms in order to begin to answer questions regarding how these systems work. It does this by concentrating on a particular subclass of cooperative coevolutionary algorithms, for which multi-population symmetric evolutionary game the- oretic models are known to apply. We examine dynamical behaviors of this model in the context of static function optimization, by both formal analysis, as well as model valida- tion study. Finally, we begin looking at the effects of variation by extending traditional EGT, offering some introductory analysis, as well as model validation. In the course of this study, we investigate the effects of parameterized uniform crossover and bit-flip mu- tation.