CONTROL-DRIVEN CONSTRAINT PROPAGATION

This article is concerned with two styles of programming: constraint programming and the one employed in coordination languages. A generic framework for constraint propagation is presented using coordination languages. More precisely, concern is with the realization of a control-driven coordination-based version of the generic iteration algorithm for compound domains of K. R. Apt. The goal is to describe constraint propagation as the coordination of cooperative agents, and thus, to provide a flexible, scalable, and generic framework for constraint propagation. Our framework overcomes inherent problems of the parallel and distributed algorithms of E. Monfroy and J.-H. Réty. Other benefits of this coordination based framework is that it does not require special modeling of CSPs and opens up ways for new constraint propagation strategies.