Parallel strategies for GRASP with path-relinking

Abstract

A Greedy Randomized Adaptive Search Procedure (GRASP) is a metaheuristic for combinatorial optimization. It usually consists of a construction procedure based on a greedy randomized algorithm and a local search. Path-relinking is an intensification strategy that explores trajectories that connect high quality solutions. We analyze two parallel strategies for GRASP with path-relinking and propose a criterion to predict parallel speedup based on experiments with a sequential implementation of the algorithm. Independent and cooperative parallel strategies are described and implemented for the 3-index assignment problem and the job-shop scheduling problem. The computational results for independent parallel strategies are shown to qualitatively behave as predicted by the criterion.