From a theoretical viewpoint, the (tree-)decomposition methods offer a good approach for solving Constraint Satisfaction Problems (CSPs) when their (tree)-width is small. In this case, they have often shown their practical interest. So, the literature (coming from Mathematics or AI) has concentrated its efforts on the minimization of a single parameter, the tree-width. Nevertheless, experimental studies have shown that this parameter is not always the most relevant to consider for solving CSPs. In this paper, we experimentally show that the decomposition algorithms of the state of the art produce clusters (a tree-decomposition is a tree of clusters) having several connected components. Then we highlight that such clusters create a real problem for the efficiency of solving methods. To avoid this kind of problem, we consider here a new kind of graph decomposition called Bag-Connected Tree-Decomposition, which considers only tree-decompositions such that each cluster is connected. We propose a first polynomial time algorithm to find such decompositions. Finally, we show experimentally that using these bag-connected tree-decompositions improves significantly the solving of CSPs by decomposition methods.
展开▼
