Three types of differential evolution applied to the facility location problem

Abstract

Facility location problem is a combinatorial optimization problem with many variants and a number of real-world applications in operations research and industry. A number of exact and approximate methods has been developed to tackle this complex task. Populational nature-inspired metaheuristic optimization methods have been investigated in context of the facility location problem as well. They are able to find excellent problem solutions, but often rely on domain-specific local search and employ heuristic steps. In this work, we study the application of three different flavours of the differential evolution algorithm to a hard variant of the facility location problem. The methods are used as pure metaheuristics without any domain specific knowledge and the efficiency of different optimization strategies they represent is evaluated and compared on a test data set.