Result Directed Computing in a Functional Logic Language

H. Kuchen, F.J. Lopez Fraguas

	   In a functional logic language, especially in a constraint
	   functional logic language, often the topmost constructor is
	   known which an expression should be evaluated to. E.g.~in
	   a guarded expression $b \rightarrow e$, $b$ should only
	   be evaluated to {\it true}.	Evaluating $b$ to {\it false}
	   would only waste resources and risk to run into an infinite
	   computation. Hence, some kind of result directed computation
	   can be used to reduce the search space by cutting uninteresting
	   computations. One possibility to achieve this result direction
	   is to integrate it into the abstract machine which is the basis
	   for the runtime system. Here, we show how to achieve result
	   directed computation by means of a program transformation. A
	   runtime system without special facilities for result directed
	   computing can be used to run the transformed programs. Thus,
	   runtime overhead for result directed computing is avoided.