Multiplicity in genetic algorithms to face multicriteria optimization

When establishing the Pareto-optimal front, effective multicriteria optimization involves simultaneous parallel search for multiple members of a genetic algorithm population. In one of these approaches due to Eiben and Lis (1997) rather than conducting multiple independent single objective searches, all the individuals in the population, “speciated” by criterion, explore the problem space expecting that the increased parallel processing of schemata improves effectiveness to find more solutions in the Pareto-optimal range. The present paper investigates the problem of using a genetic algorithm on a set of test functions which allows a multisexual population, multiple parents and multiple crossovers per mating, attempting to build a Pareto optimal set of larger size. Also, while creating a new population a selection process for replacement favours those new created solutions that are inclined to appertain to the Pareto front. As a result, the performance of the method produce an evenly distributed and larger set of efficient points