A New Master Slave Model of Parallel Genetic Algorithms for Multi Objective Optimization Problems
Shinya Watanabe · 2000
In this paper, we propose a new parallel genetic algorithm model for multi objective optimization problems. That is one of a master slave model. In the proposed model, a master process chooses two individuals randomly and sends to a slave process. In slave process, all GA operations(crossover,mutaion,selection,evaluation,etc ... )are performed with two individuals. To clarify the characteristics and effectiveness of this model, the proposed model is applied to a knap- sack problem. Through the numerical examples, it becomes cleared that this model is suited to parallel computers and can keep the diversity of the solutions.