Non-stationary Function Optimization using Polygenic Inheritance

Non-stationary function optimization has proved to be a di cult subject for genetic algorithms (GAs). Standard haploid GAs nd it di cult to track a moving target and tend to converge to a local optimum that appears early in a run. We have surveyed a number of diploid GAs which use classic dominant recessive expressions, and outline some possible reasons why they have failed to gain wide acceptance [3]. A new haploid system, Shades, is then described which uses polygenic inheritance, in which several genes contribute to each phenotypic trait. The rst instance of this in natural biology was discovered in 1909 by Nilsson-Ehle when he showed that the kernel color in wheat, an additive trait, was in fact managed by two pairs of genes. Inheritance of genes of this type is known as polygenic inheritance, and the more loci involved in the calculation of a trait, the smoother the phenotypic space.