Using Problem Generators to Explore the
Effects of Epistasis

Kenneth A. De Jong
Mitchell A. Potter
William M. Spears

 

In this paper we develop an empirical methodology for studying the behavior of evolutionary algorithms based on problem generators. We then describe three generators that can be used to study the effects of epistasis on the performance of EAs. Finally, we illustrate the use of these ideas in a preliminary exploration of the effects of epistasis on simple GAs.