Abstract (EN):
In this paper, we address the Capacitated Single Allocation ¿-Hub Location Problem (CSA¿HLP) in which the capacities of the hubs limit the flows in the network and every non-hub node must be allocated to only one hub. The objective is to choose a fixed number of ¿ nodes to be established as hubs that minimizes the costs of allocating all the non-hub nodes to the chosen hubs. We propose a simple Relaxation Adaptive Memory Programming (RAMP) approach that uses Lagrangean Relaxation with subgradient optimization to explore the dual side, a projection method to project dual solutions into the primal solutions space and an improvement method to guide the search in the primal side. The computational results obtained on a classical set of benchmark problems showed that our algorithm achieved the best results in the literature, demonstrating the advantages of exploring primal-dual relationships.
Idioma:
Inglês
Tipo (Avaliação Docente):
Científica