MParCO: A Minimalist Parallel Framework for Combinatorial Optimization Applications

Abstract

This paper addresses the exploration of the potential of distributed parallelism in combinatorial optimization applications through a minimalist approach. In contrast to several parallel frameworks proposed in the last decades, specialized in parallelizing classical methods such as Branch-and-Bound and well-known metaheuristics, we propose a framework, so-called MParCO, that focuses essentially on assuring asynchronism, an essential ingredient for the performance of enumerative methods for combinatorial optimization problems.