Evolutionary Dynamics of Genetic Algorithms in the Fitness Landscape Including Neutral Networks

Abstract

Neutral networks, which are landscapes containing neighboring points of equal fitness, have attracted much research interest in recent years. In this paper, we conduct a series of computer simulations to investigate the effect of an error threshold on the moving speed of a population as well as a variable mutation rate strategy against ruggedness. Two kinds of GA are adopted. One is the simple GA where the mutation rate is constant, and the other is the operon-GA whose effective mutation rate is changing at each locus independently according to the history of the genetic search. The results demonstrate that the moving speed of a population is correlated with the selection pressure as well as the mutation rate. The variable mutation rate strategy is beneficial in the cases of the simplest test function and complex test functions. This tendency becomes clearer with the increase of ruggedness in the test functions.