Strategies and heuristics for exploiting tree-decompositions of constraint networks

This paper deals with methods exploiting treedecomposition approaches for solving constraint networks. We consider here the practical efficiency of these approaches, by defining three classes of variable orders more and more dynamic which preserve the time complexity bound and an extension of this theoretical time complexity bound to increase the dynamic aspect of these orders. We propose heuristics in each class to improve the runtime of the methods. Then, we study empirically the practical interest of the proposed heuristics in order to point up the most interesting ones. Finally, the proposed theoretical extension of the time complexity bound is assessed from a practical viewpoint.