J4

• 计算机科学 • Previous Articles     Next Articles

Variety and Convergence of Genetic Algorithm

SUN Qi-long, CHEN Luo-wu, LIU Yan-liang, CHEN Jie, HU Xiao-min, ZHANG Jun    

  1. Department of Computer Science and Technology, Sun Yat-Sen University, Guangzhou 510275, China
  • Received:2006-05-20 Revised:1900-01-01 Online:2006-08-26 Published:2006-08-26
  • Contact: ZHANG Jun 

Abstract: This paper deals with the influence of variety and convergence of Genetic Algorithm (GA) by the selection operator on the final solution. Several selection operators were introduced such as roulette selection operator which keeps population variety well, and tournament selection operator which keeps population convergence well. Based on the synergism of variety and convergence, the standard tournament selection operator is improved so as to get two new selection operators: keeping old population tournament selection operator and new tournament selection operator. After the experiment, we have come to a conclusion: If the population variety is increased, the speed of popula tion convergence will decrease; accelerating the speed of convergence will greatly destroy the variety. The former can impact the chance to find the optimal solution, and the later can enhance the speed of the genetic algorithm. In order to find a better solution, we should consider their synergism.

Key words: genetic algorithm, selection operator, selection press ure, variety, convergence, compromise

CLC Number: 

  • TP311.133