New effective neighborhoods for the permutation flow shop problem

We propose an extension of the Taillard's implementation, which allows to remove efficiently the less well inserted jobs in a permutation. We describe then six new neighborhoods for the permutation flow shop problem. Computational results show clearly that at least three of them are better than the insertion move. Their application into a simple metaheuristic is very effective, since a new upper bound has been found for a hard Taillard's instance.