A Memetic Algorithm with Parallel Local Search for Flowshop Scheduling Problems.

We propose a new memetic algorithm to minimize the makespan for the flowshop scheduling problem in two variants: the classic permutation setting and the no-wait statement. Our algorithm hybridizes the local search technique into the framework of a steady-state genetic algorithm. A local search heuristic is applied on each iteration and explores the insertion neighborhood. The execution of the local search is parallelized using the CUDA framework for Graphics Processing Units. The initial population is constructed by means of greedy constructive heuristics. A computational experiment on the benchmark instances shows that the proposed algorithm yields results competitive to those of well-known algorithms for the flowshop problem.