Tirage de solutions par ajout de contraintes tables aléatoires

Constraint solvers feature efficient algorithms to handle satisfaction and optimisation combinatorial problems. These features are not suited to new usages, such as the sampling of solutions. We propose here an algorithm to randomly sample solutions, based on the addition of randomly generated table constraints, without modifying the model of the problem. We implemented this method of resolution using an existing constraint solver. Our experiments show that this algorithm is an improvement over a random branching strategy in terms of quality of the randomness of the sampling.