Revisiting Bremermann's genetic algorithm. I. Simultaneous mutation of all parameters

Abstract

Hans Bremermann was one of the pioneers of evolutionary computation. Many of his early suggestions for designing evolutionary algorithms anticipated future inventions, including scaling mutations to be inversely proportional to the number of parameters in the problem, as well as many forms of recombination. This paper explores the gain in performance that occurs when Bremermann's original evolutionary algorithm (H.J. Bremermann et al., 1966) is extended to include the simultaneous mutation of every component in a candidate solution. Bremermann's original perspective was entirely "genetic", where each component corresponded to a gene, and therefore multiple simultaneous changes were viewed as occurring with geometrically decreasing probability. Experiments indicate that a change in perspective to a "phenotypic" view, where all the components change at once, can lead to more rapid optimization on linear systems of equations.