On the Convergence of NSGA-II Algorithm to a Single Pareto Optimal Solution with a Continuous and Quadratic Multi-objective Optimization Program

Abstract

In this paper, we test a fast elitist Non-Dominant Sorting Genetic Algorithm (NSGA-II) on a continuous quadratic multiobjective optimization problem. We use a big size of multiobjective programming for the Gross Domestic Product (GDP) quarterly disaggregation and we test different values of the NSGA-II parameters. We come to the conclusion that if the parameters are judiciously chosen, the NSGA-II algorithm converges to a single Pareto optimal solution in a regular and bounded set for the quadratic multiobjective optimization problem we used. It should be noted that elitism in the NSGA-II algorithm contributes to accelerating the rate of convergence and the overall performance of the genetic algorithm incorporated in it.