Research on Search Performance of Crossover and Mutation in Real-Coded GA

The search performance of crossover operator and mutation operator of real-coded GA is studied. The gene level diversity of GA is defined firstly from mathematics angle; and referring to the concepts of "space" and "subspace" in linear algebra, the concepts of arithmetic crossover extended subspace of population, the optimization space and initial population space are defined, the extensibility of crossover and mutation in solution space is analyzed. Secondly, the idea of population "inward contraction" and "external expansion" is provided and the influence of crossover and mutation on population diversity is studied.