Combining Solvers in a Meta Constraint Logic Programming Architecture

Abstract

We present a general technique for the combination and the integration of different Constraint Logic Programming (CLP) solvers. The main idea behind the work concerns the possibility of building meta CLP architectures by adding CLP solvers in a natural and effective manner. In the meta architecture, levels are constraint solvers each reasoning on constraints of the underlying system. The architecture presented starts from a meta Constraint Logic Programming general scheme. A distinguishing feature of the architectural scheme concerns its operational semantics which can be seen as a general combination method for data and control of two constraint solvers. A set of linking rules define how systems exchange data, while a set of transition rules define how systems combine their control flow. We propose a specialization of a meta CLP architecture on finite domains. The specialization concerns the possibility of combining qualitative and quantitative reasoning in a CLP framework. This combination can be useful, for example, in the field of temporal reasoning.