Heuristics for Selecting Predicates for Partial Predicate Abstraction

TL;DR

This paper introduces two heuristics for selecting effective predicates in partial predicate abstraction, improving the analysis of infinite-state systems modeled with linear integer arithmetic.

Contribution

It proposes novel heuristics that guide predicate selection using problem instance analysis and previous verification results, enhancing abstraction precision.

Findings

Heuristics improve predicate selection for partial abstraction.

Experimental results demonstrate effectiveness in CTL model checking.

Analysis discusses advantages and disadvantages of each heuristic.

Abstract

In this paper we consider the problem of configuring partial predicate abstraction that combines two techniques that have been effective in analyzing infinite-state systems: predicate abstraction and fixpoint approximations. A fundamental problem in partial predicate abstraction is deciding the variables to be abstracted and the predicates to be used. In this paper, we consider systems modeled using linear integer arithmetic and investigate an alternative approach to counter-example guided abstraction refinement. We devise two heuristics that search for predicates that are likely to be precise. The first heuristic performs the search on the problem instance to be verified. The other heuristic leverages verification results on the smaller instances of the problem. We report experimental results for CTL model checking and discuss advantages and disadvantages of each approach.