Even the fastest SMT solvers have performance problems with regular expressions from real programs. Because these performance issues often arise from the problem representation (e.g. non-deterministic finite automata get determinized and regular expressions get unrolled), we revisit Boolean finite automata, which allow for the direct and natural representation of any Boolean combination of regular languages. By applying the IC3 model checking algorithm to Boolean finite automata, not only can we efficiently answer emptiness and universality problems, but through an extension, we can decide satisfiability of multiple variable string membership problems. We demonstrate the resulting system's effectiveness on a number of popular benchmarks and regular expressions.

Abstract Several new algorithms for deciding emptiness of Boolean combinations of regular languages and of languages of alternating automata have been proposed recently, especially in the context of analysing regular expressions and in string constraint solving. The new algorithms demonstrated a significant potential, but they have never been systematically compared, neither among each other nor with the state-of-the art implementations of existing (non)deterministic automata-based methods. In this paper, we provide such comparison as well as an overview of the existing algorithms and their implementations. We collect a diverse benchmark mostly originating in or related to practical problems from string constraint solving, analysing LTL properties, and regular model checking, and evaluate collected implementations on it. The results reveal the best tools and hint on what the best algorithms and implementation techniques are. Roughly, although some advanced algorithms are fast, such as antichain algorithms and reductions to IC3/PDR, they are not as overwhelmingly dominant as sometimes presented and there is no clear winner. The simplest NFA-based technology may sometimes be a better choice, depending on the problem source and the implementation style. We believe that our findings are relevant for development of automata techniques as well as for related fields such as string constraint solving.