Efficient local search for several combinatorial optimization problems. (Recherche locale performante pour la résolution de plusieurs problèmes combinatoires)
This Ph.D. thesis concerns algorithms for Combinatorial Optimization Problems. In Combinatorial Optimization Problems the set of feasible solutions is discrete or can be reduced to a discrete one, and the goal is to find the best possible solution. Specifically, in this research we consider three different problems in the field of Combinatorial Optimization including One-dimensional Bin Packing (and two similar problems), Machine Reassignment Problem and Rolling Stock Problem. The first one is a classical and well known optimization problem, while the other two are real world and very large scale problems arising in industry and have been recently proposed by Google and French Railways (SNCF) respectively. For each problem we propose a local search based heuristic algorithm and we compare our results with the best known results in the literature. Additionally, as an introduction to local search methods, two metaheuristic approaches, GRASP and Tabu Search are explained through a computational study on Set Covering Problem.