Abstract
This paper presents a method to simultaneously produce multiple solutions to unconstrained multi-objective optimization problems. The proposed methodology uses populations of sets instead of populations of individuals and iterative calls to a Genetic Algorithm (IGA) to obtain a set of solutions spread across the Pareto set in the objective space. The superiority of such an approach to single run, conventional population Pareto GAs is shown. The various difficulties of the algorithm and the methods used to overcome them are detailed. Finally, the paper expands upon how this method can be used with or without user inputs, and shows an analysis of its performance by applying it to a succession of increasingly difficult problems, identifying its range of application.
