National Conference on Artificial Intelligence · 1998 · 93 citations · 2 references
Mathematical ProgrammingEngineeringConstrained OptimizationComputational ComplexityDiscrete OptimizationAlldiff ConstraintConstraint ProgrammingOperations ResearchConstraint SolvingData ScienceDiscrete MathematicsCombinatorial OptimizationComputational GeometryApproximation TheoryComputer ScienceAlldiff ConstraintsAlgorithmic Information TheoryConstraint SatisfactionFiltering AlgorithmFormal MethodsBound Consistency
Some n-ary constraints such as the alldiff constraints arise naturally in real life constraint satisfaction problems (CSP). General purpose filtering algorithms could be applied to such constraints. By taking the semantics of the constraint into account, it is possible to design more efficient filtering algorithms. When the domains of the variables are totally ordered (e.g. all values are integers). then filtering based on bound consistency may be very useful. We present in this paper a filtering algorithm for the alldiff constraint based on bound consistency whose running time complexity is very low. More precisely, for a constraint involving n variables, the time complexity of the algorithm is O(nlog(n)) which improyes preyionsly published results. The implementation of this algorithm is cliscussed. and we give some experimental results that prove its practical utility.
2