Genetic algorithm belongs to evolutionary algorithm category. It is a good algorithm prototype for handling real parameter optimization, and its efficiency can be enhanced by executing generations in parallel. Based on a three-parent crossover and a diversity operator, this paper investigates the running time of executing genetic algorithm in parallel. Specifically, parallel execution is realized based on multicore central processing unit of computer. Extensive experiments are conducted on a set of mathematical test functions. The running times of genetic algorithm with and without parallel execution are compared based on the types of optimization problem. Moreover, the results are presented from one core to eight cores. A time increase curve is fitted based on polynomial model, which could assist users to conduct parallel genetic algorithm to solve problems.
展开▼
