论文信息 - Linkage equilibrium and genetic algorithms

Linkage equilibrium and genetic algorithms

The mathematical analysis of recombination/crossover in population genetics and genetic algorithms is very difficult. We investigate a result from H. Geiringer (1944), which states that by repeated applications of the recombination operator alone (with no selective pressure) the gene frequencies of diploid organisms converge to linkage equilibrium. The result is valid for any recombination/crossover scheme. It gives strong support to the conjecture that recombination is not able to explore nonlinear gene interactions. This fact explains the difficulties genetic algorithms have with optimization of fitness functions with nonlinear gene interactions.

H. Muhlenbein | U.K. Chakraborty | Uday K. Chakraborty | Heinz Muehlenbein

[1] Heinz Mühlenbein,et al. On the Mean Convergence Time of Evolutionary Algorithms without Selection and Mutation , 1994, PPSN.

[2] John H. Holland,et al. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence , 1992 .

[3] H. Tietze,et al. Über das Schicksal gemischter Populationen nach den Mendelschen Vererbungsgesetzen , 2022 .

[4] Gilbert Syswerda,et al. Uniform Crossover in Genetic Algorithms , 1989, ICGA.

[5] H. Mühlenbein,et al. Gene Pool Recombination in Genetic Algorithms , 1996 .

[6] T. Nagylaki. Introduction to Theoretical Population Genetics , 1992 .

[7] R. B. Robbins. Some Applications of Mathematics to Breeding Problems III. , 1917, Genetics.

[8] M Pagano. The mean. , 1992, Nutrition.

[9] R. B. Robbins. Some Applications of Mathematics to Breeding Problems. , 1917, Genetics.

[10] H. Geiringer. On the Probability Theory of Linkage in Mendelian Heredity , 1944 .