Section: Bilateral Contracts and Grants with Industry

Google

Participants : Jean-Guillaume Fages, Xavier Lorca, Jean-Charles Régin, Louis Martin Rousseau.

Within the context of a Google grant involving Jean-Charles Régin from Nice University , Louis-Martin Rousseau from Polytechnique Montreal and Xavier Lorca and Jean-Guillaume Fages from TASC the following work on graph constraints has been undertaken.

In constraint programming, specific syntax for expressing unweighted circuit constraints in a graph have been proposed already since the first CP systems were developed. Most current CP systems contain a constraint to model unweighted circuits, although the associated filtering algorithm may be quite different for each system. Weighted circuit constraints are less common in CP systems, as the weights and the circuit are typically handled separately. However, several filtering algorithms have been proposed in the literature that can be applied to the weighted circuit constraint. Currently, no professional CP system offers any of these algorithms. Thus, solving problems that contain weighted circuit constraints remains a challenge for constraint programming. One of the main contributions of our work is to expand the reach of constraint programming solvers to complex routing problems. We will propose more effective filtering algorithms for the weighted circuit constraint that are guided by the resolution of real world instances and not only by benchmarking.