Positive and negative combination effects of crossover and mutation operators in sequencing problems

When a genetic algorithm is applied to sequencing problems, various crossover and mutation operators are applicable. Because the performance of genetic algorithms depends on the choice of such operators, we have to carefully select appropriate operators for constructing high performance genetic algorithms. Moreover the performance also depends on the specifications of the crossover and mutation probabilities. We discuss the selection of genetic operators and their probability specifications for sequencing problems. First we examine the performance of various crossover and mutation operators by computer simulations on flowshop scheduling problems. Then we point out that the combination of high performance crossover and mutation operators does not always lead to a high performance genetic algorithm because of the negative combination effect of those two operators. Finally we explain how a high performance genetic algorithm can be constructed by utilizing the positive combination effect of crossover and mutation