New schemes for simplifying binary constraint satisfaction problems

Wady Naanaa (1)

1 UM - Université de Monastir - University of Monastir

Finding a solution to a Constraint Satisfaction Problem (CSP) is known to be an NP-hard task. This has motivated the multitude of works that have been devoted to developing techniques that simplify CSP instances before or during their resolution. The present work proposes rigidly enforced schemes for simplifying binary CSPs that allow the narrowing of value domains, either via value merging or via value suppression. The proposed schemes can be viewed as parametrized generalizations of two widely studied CSP simplification techniques, namely, value merging and neighbourhood substitutability. Besides, we show that both schemes may be strengthened in order to allow variable elimination, which may result in more significant simplifications. This work contributes also to the theory of tractable CSPs by identifying a new tractable class of binary CSP.

Mots clés: Constraint satisfaction problems; value merging; variable elimination; tractable CSP

Domaines: Intelligence artificielle [cs.AI]; Complexité [cs.CC]

Wady Naanaa. New schemes for simplifying binary constraint satisfaction problems. Discrete Mathematics and Theoretical Computer Science, 2020, vol. 22 no. 1, ⟨10.23638/DMTCS-22-1-10⟩. ⟨hal-01731250v4⟩