Parallel Multi-Objective Evolutionary Algorithm with Multi-Front Equitable Distribution

In multi-objective context, the evolutionary approach offers specific mechanisms such as Pareto selection, elitism and diversification. These techniques are proved to be efficient to characterize the Pareto Front. However, their high computing time constitutes a major handicap for their expansion. The parallelization of Multi-Objective Evolutionary Algorithms (MOEAs) may be an efficient way to overcome this problem. This parallelization aims not only to achieve time saving by distributing the computational effort but also to get benefit from the algorithmic aspect by the cooperation between different populations and evolutionary schemes. In this paper we propose a new parallel multi-objective evolutionary algorithm with multi-front equitable distribution which is based on an elitist technique. Every population evolves differently on a processor and cooperates with the others to preserve genetic diversity and to obtain a set of diversified non dominated solutions.