Equivalence-Based Abstraction Refinement for \mu HORS Model Checking

Xin Li, Naoki Kobayashi

Abstract: Kobayashi and Igarashi proposed model checking of \(\mu \)HORS (recursively-typed higher-order recursion schemes), by which a wide range of programs such as object-oriented programs and multi-threaded programs can be precisely modeled and verified. In this work, we present a procedure for \(\mu \)HORS model checking that improves the procedure based on automata-based abstraction refinement proposed by Kobayashi and Li. The new procedure optimizes each step of the abstract-check-refine paradigm of the previous procedure. Specially, it combines the strengths of automata-based and type-based abstraction refinement as equivalence-based abstraction refinement. We have implemented the new procedure, and confirmed that it always outperformed the original automata-based procedure on runtime efficiency, and successfully verified all benchmarks which were previously impossible.