Experimental study of the population parameters settings in cooperative multi-agent system solving instances of the VRP

Abstract

Usually a lot of experiments are often required in order to tune population-based algorithms, designed for solving difficult optimization problems. Individial features of a particular problem, different parameters of population of individuals, or structure of the algorithm may influence results produced by the system. The paper aims at evaluating experimentally to what extent (if any) different values of the population parameters controlled by the user in a multi-agent system solving instances of Vehicle Routing Problem influence computational results. The reported experiment involved several methods of creating an initial population of solutions and several cooperating agents representing improvement heuristics working in parallel. © 2013 Springer-Verlag.