Finding Connict Sets and Backtrack Points in Clp(<)

This paper presents a method for intelligent backtracking in CLP(<). Our method integrates a depth-rst intelligent backtracking algorithm developed for logic programming with an original constraint satisfaction algorithm which naturally generates sets of connicting constraints. We prove that if CLP(<) is assumed to cover strictly the domain of real numbers, then the constraint satisfaction algorithm provides minimal connict sets to be used as a basis for intelligent backtracking. We then extend the backtrack-ing method to cover a two-sorted domain, where variables can be bound to either structured terms or real numbers. We discuss a practical implementation of the algorithm using a generator-consumer approach to the recording of variable bindings, and we give an example of a CLP(<) program which beneets signiicantly from intelligent backtracking.