Making Adaptive an Interval Constraint Propagation Algorithm Exploiting Monotonicity

Ignacio Araya (1) , Gilles Trombettoni (2) , Bertrand Neveu (3)

1 Departamento de Informatica [Valparaíso, Chile]
2 Laboratoire d'Informatique, Signaux, et Systèmes de Sophia-Antipolis (I3S) / Equipe CEP
3 LIGM - Laboratoire d'Informatique Gaspard-Monge

A new interval constraint propagation algorithm, called MOnotonic Hull Consistency (Mohc), has recently been proposed. Mohc exploits monotonicity of functions to better filter variable domains. Embedded in an interval-based solver, Mohc shows very high performance for solving systems of numerical constraints (equations or inequalities) over the reals. However, the main drawback is that its revise procedure depends on two user-defined parameters. This paper reports a rigourous empirical study resulting in a variant of Mohc that avoids a manual tuning of the parameters. In particular, we propose a policy to adjust in an auto-adaptive way, during the search, the parameter sensitive to the monotonicity of the revised function.

Domaines: Intelligence artificielle [cs.AI]

Ignacio Araya, Gilles Trombettoni, Bertrand Neveu. Making Adaptive an Interval Constraint Propagation Algorithm Exploiting Monotonicity. CP 2010, 16th International Conference on Principles and Practice of Constraint Programming, Sep 2010, St Andrews, Scotland, United Kingdom. pp.61-68, ⟨10.1007/978-3-642-15396-9_8⟩. ⟨hal-00654414⟩