Multi-Objective Propagation in Constraint Programming

Bounding constraints are used to bound the tolerance of solutions under certain undesirable features. Standard solvers propagate them one by one. Often times, it is easy to satisfy them independently, but difficult to satisfy them simultaneously. Therefore, the standard propagation methods fail. In this paper we propose a novel approach inspired in multi-objective optimization. We compute a multi-objective lower bound set that, if large enough, can be used to detect the inconsistency of the problem. Our experiments on two domains inspired in real-world problems show that propagation of additive bounding constraints using our approach is clearly superior than previous approaches.