Constraint Satisfaction - Algorithms and Complexity Analysis

The constraint satisfaction problem (CSP) comprises n variables with associated nite domains (with a maximal cardinality d), and some combinations of value assignments (\constraints") to the variables; then, in order to get the globally consistent solution, we need to compute the set of all n-tuples consistent with the given constraints. There are a lot of papers which exploit speciic constraint network structures; others focus on local constraint propagation. However, nally, the global solution is wanted, and in this regard, I would like to introduce here a general framework which really yields this globally consistent solution. It is well-known that local consistency does not imply global consistency. Anyway, to start directly with a higher-level mechanism may not be a bad idea. The algorithms presented here may get consulted when nothing can be exploited regarding network topology or other features. Remark This report may supersede #14/90. The current issue (#17/94) is a slight update of the following work: On global constraint satisfaction.