Avoiding Dynamic Delays in Functional Logic Programs

. In several functional logic programming languages, functional expressions must be evaluated before unification with another term, but this can only be done if the functional expression is ground. If the functional expression is non-ground, then unification must be delayed until all the arguments have become instantiated to ground terms. If the delaying mechanism uses dynamic tests, the program will be unnecessarily inefficient. We present an analysis method for statically determining at what point in the program a given functional expression can be evaluated. This analysis is then used for transforming a functional logic program into an equivalent program, in which most dynamic delays are avoided. We show that this transformation can lead to considerable improvements in performance, compared to a dynamic delaying strategy. 1 Introduction During the last decade, an immense amount of research has been done on combining logic programming with functional and equational languages. The d..