Finding Regions of Local Repair in Hierarchical Constraint Satisfaction

Harald Meyer auf'm Hofe

DFKI DFKI Research Reports (RR) 97-05 1997.


Algorithms for solving constraint satisfaction problems (CSP) have been successfully applied to several fields including scheduling, design, and planning. Latest extensions of the standard CSP to constraint optimization problems (COP) additionally provided new opportunities for solving several problems of combinatorial optimization more efficiently. Basically, two classes of algorithms have been used for searching constraint satisfaction problems (CSP): local search methods and systematic tree search extended by the classical constraint-processing techniques like e.g. forward checking and backmarking. Both classes exhibit characteristic advantages and drawbacks. This report presents a novel approach for solving constraint optimization problems that combines the advantages of local search and tree search algorithms which have been extended by constraint-processing techniques. This method proved applicability in a commercial nurse scheduling system as well as on randomly generated problems.

RR-97-05.pdf (pdf, 528 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence