Improving abstract interpretations by combining domains

Abstract

In this paper we consider static analyses based on abstract interpretation of logic programs over combined domains. It is known that analyses over combined domains potentially provide more information than obtainable by performing the independent abstract interpretations. However, the construction of a combined analysis often requires redefining the basic operations for the combined domain. We demonstrate for logic programs that in practice it is possible to obtain precision in a combined analysis without redefining the basic operations. We also propose a way of performing the combination which can be more precise than the straightforward application of the classical “reduced product” approach, while keeping the original components of the basic operations. The advantage of the approach is that proofs of correctness for the new domains are not required and implementations can be reused. We illustrate our results by showing that a combined sharing analysis—constructed from “old” proposals—compares well with other “new” proposals suggested in recent literature both from the point of view of efficiency and accuracy.