Searching for i-Good Lemmas to Accelerate Safety Model Checking

Abstract / and its variants have been the prominent approaches to safety model checking in recent years. Compared previous model-checking algorithms like (Bounded Model Checking) (Interpolation Checking), is attractive due completeness (vs. ) scalability ). maintains an over-approximate state sequence for proving correctness. Although refinement methodology known be crucial performance, literature lacks a systematic analysis of problem. We propose approach based on definition i - good lemmas , introduction two kinds heuristics, i.e., steer search towards construction $$i$$ i -good lemmas. The applicable variant (Complementary Approximate Reachability), it very easy integrate within existing systems. implemented heuristics into open-source checkers, as well mature platform, carried out extensive experimental evaluation HWMCC benchmarks. results show that proposed can effectively compute more lemmas, thus improve performance all above checkers.