Constraint updating

Abstract

In the areas of arti® cial intelligence and computer-aided design there are many fundamental problems that can be formulated and solved by constraint techniques. Th e problems are often of a dynamic nature in the sense that the original problem changes in the course of time ; furthermore, the granularity of the changes is often located on a tuple level. So, working on that level ensures maximal¯exibility and problem adequacy. Here, both the modi® cations and the recomputations of the aOE ected constraints are working on that level. W ith the NP-complete complexity of global constraint satisfaction in mind, it is desirable to decrease the average complexity of the consistency algorithm. This objective may be achieved by applying the changes incrementally to only the aOE ected parts of the network avoiding total recomputations wherever possible. Especially in applications with large domains and many solution tuples, the reduction of the computational complexity is enormous.

Fine-grained conflict resolution in constraint satisfaction problems

“…Freuder and W allace (1996) have re-presented a republication of a wellknown paper in the area of constraint relaxation and acknowledged further related work (see, for instance, p. 64 of this work). Hower and Jacobi (1998) have dealt with CSP modi® cations on the ® ne-grained tuple level, to permit more or fewer solution tuples. Finally, Jampel (1996) has compiled additional material.…”