Performance Study of a Distributed Genetic Algorithm with Parallel Cooperative-Competitive Genetic Operators
Hernan E. Aguirre, Kiyoshi Tanaka, Shinjiro Oshita · 2002
ly simple and any gain we might expect reducing theo verall time to co1Wq/6I1 co1W seemmino6 Ho wever, the pro cessing time viewpow taloM misses the dynamics that can arise fro o peratoS withcoh1I6q/ tary roW/ acting in parallel. The balance betweencroWR ver and mutatio is crucial to the perfo/q71O o f GAs. One way to pursue better balances, andtherefo1 better perfo7 mance, isto co bine cro6R ver with higher mutatio rates. Higher mutatioI parallel to croIq ver can give an e#cient framewo rkto wards thisgo1W in which the received December 18, 2001. Manuscript revised April 2, 2002. Final manuscript received May 24, 2002. + The aut8C= are wit t he Facult y ofEngineering, Shinshu Universit y, Nagano-shi, 380-8553 Japan. a) E-mail: ktil: [email protected] strengthso f the individualo perato0 can be kept withoh interferingof with theo1/0 [5]. Rather than as a hardware accelerato6 themo/ significant gainsfro the parallel applicatio o fo perato7 within parallel GAs coSS co0 fro explo