GENERIC CRSs INFINITE - DOMAIN PROBLEMS FINITE - DOMAIN PROBLEMS

The lack of a uni ed framework for constraint problems has resulted in a diversity of constraint solution techniques and methods for describing constraint problems. A uni ed framework would allow the construction of a kernel constraint engine, capable of solving constraint problems composed of models from di erent domains, such as engineering and geometric problems. As a rst step towards such a uni ed framework, a uni ed constraint representation scheme is necessary. This report presents reductions, a method of comparing constraint representation schemes in terms of expressiveness. A hierarchy of current constraint representation schemes can then be constructed. The notion of a generic constraint representation scheme is also introduced and a generic scheme is a potential candidate to be used as a uni ed constraint representation scheme. Reductions are used to prove that certain currently used schemes are, in fact, generic.