Programming Constraint Inference Engines

Author: Christian Schulte

Editor: Gert Smolka

Existing constraint programming systems offer a fixed set of inference engines implementing search strategies such as single, all, and best solution search. This is unfortunate, since new engines cannot be integrated by the user. The paper presents first-class computation spaces as abstractions with which the user can program inference engines at a high level. Using computation spaces, the paper covers several inference engines ranging from standard search strategies to techniques new to constraint programming, including limited discrepancy search, visual search, and saturation. Saturation is an inference method for tautology-checking used in industrial practice. Computation spaces have shown their practicability in the constraint programming system Oz.