A Genetic Model Based On Simulated Crossover Of Quaternary Genes For Quadratic Fitness

We present a genetic model based on simulated recombination of fixed sequences of quaternary genes, (assuming four distinct forms, or alleles). States and dynamics of the infinite population genetic system, represented by the model, are derived for quadratic fitness functions. The bivariate marginal distribution genetic algorithm, devised to simulate the system, is experimentally compared with the univariate marginal distribution genetic algorithm with bit-based simulated crossover (in the case of infinite populations) for the MAX - CUT problem.