Its performance is tested over three sets of instances for the inventory location routing, location-routing and inventory-routing problems.
Results of state-of-the-art metaheuristics are also compared on standard sets of instances for the classical LRP, the two-echelon LRP and the truck and trailer problem.
(All new best results are available at http://prolog.univie.ac.at/research/DARP/.) In a next step, we apply the proposed hybrid LNS to two sets of instances for which optimal solutions are known.
> In our computational study, two groups of instances were considered: Randomly generated graphs from [43]: For this set of instances the parameters for the generation were set as follows: ˈ S ˈ ∈ { 20, 50, 100 }, ˈ R ˈ ∈ { 20, 50, 100 }.
Combine it with the base set of instances forming a new data set for the next over_sampling process.
Increasing only the core costs did not significantly change the behavior of the GRASP algorithm for this set of instances.
Running time and the quality of solutions of Ljubic' VNS heuristic and DLS are compared for the set of instances introduced in [33].
On the publicly available set of benchmark instances for barge container routing, we demonstrate that this model provides very tight dual bounds and significantly outperforms the existing approaches from the literature for splittable demands.
Here, we use an L1-regularized L2-loss SVM, which solves the following optimization problem for a set of instance-label pairs x → i, y i, x → i ∈ R n, y i ∈ { -1, + 1}, i = 1, …, l: (1) min w → | | w → | | 1 + C ∑ i = 1 l max 0, 1 − y i w → T x → i 2, where C ≥ 0 is a penalty parameter.
