Enhancing the Constraint-Solving Power of CLP(FD) by Means of Path-Consistency Methods

We consider methods for handling constraints over finite domains based on the notion of path consistency. We show that these methods can be easily integrated in the glass-box constraint-solver of the clp(FD) system. Besides the clp(FD) indexical constraints encoding (partial) arc-consistency methods, we introduce new primitive constraints, that we call m-constraints, which encode path-consistency methods. These new constraints increase the constraint-solving power and the extensibility of the clp(FD) language.