General Methodology for Converting Sequential Evolutionary Algorithms into Parallel Algorithms with OpenMP as Applied to Combinatorial Optimization Problems

This paper presents a general methodology for the conversion of sequential evolutionary algorithms into parallel evolutionary algorithms using OpenMP, independent of the problem being approached. Because of the diversity of existing problems and the variation between each one, it is difficult to find a standard or a method to make problems parallel. Analyses that are specific to problems being approached exist, but there is not a general methodology. This article shows its methodology by applying it to a combinatorial optimization problem of Vehicle Routing with time windows, using an evolutionary heuristic with intelligent mutation and k-means clustering as a solution method. This paper shows the analysis of the sequential algorithm and the task assignment to each processor in the parallel algorithm. The justification of the elements used for the conversion of sequential and parallel versions is also shown.