Efficient dataflow analysis of logic programs

Abstract

We investigate a framework for efficient flow analyses of logic programs. A major problem in this context is that unification can give rise to aliasing and dependencies between variables whose effects are difficult to predict, and which make sound flow analysis algorithms computationally expensive. We give a simple characterization of the class of flow analysis problems for which aliasing effects can be ignored without loss of soundness, and describe an efficient analysis procedure for this class of problems. The utility of our approach is illustrated by discussing its application to several analysis and optimization problems for logic programs. Our results are useful in the design of efficient flow analysis systems for logic programming languages.