Performance Evaluation of Reproduction Operators in Genetic Algorithm

The performance of a GA largely depends on its parameters: crossover, mutation and selection. There exist many crossover and mutation operators are proposed. The primary interest of this paper is to investigate the effectiveness of the various reproduction operators. The conceptual characteristics of the combination of reproduction operators in the context of Travelling Salesman Problem (TSP) are discussed. Extensive experiments are conducted to compare the performance of 3-crossovers and 3-mutation operators. The computational experiments are performed and the results are collected. Statistical tests are conducted that demonstrate the superiority of 2-point cut crossover and swap mutation operators combination.