Information Propagation in Partial

The aim of this work is to incorporate constraints into the partial evaluation procedure in order to obtain further specialisation through greater (constraint-based) information propagation. We incorporate constraints into the algorithm for the partial evaluation of the Escher language, a functional logic language based on rewriting. The preliminary implementation of the partial evaluator uses positive and negative information propagation by constraints; using this additional information, the partial evaluator optimises a naive pattern matcher to one equivalent to the KMP algorithm. This is the rst known implementation of the propagation of negative information in the partial evaluation of functional logic languages. Future work includes an extension to other constraint domains, including arithmetic, boolean, and nite domain.