Using Implications for Optimizing State Set Representations of Linear Hybrid Systems

In this paper we present an approach to exploit pre-calculated implication knowledge in the construction of LinAIGs which represent sets of states of Linear Hybrid Systems. Our method computes implications between linear constraints and uses this information to strengthen SAT-based equivalence checks which occur during the construction of the LinAIGs. The approach is evaluated on several hybrid model checking benchmarks where LinAIGs are used as the core data-structure of the model checker. The results show that the use of implications can significantly reduce the number of applications of expensive SAT-modulo-Theories (SMT) methods, and thus can accelerate LinAIG compaction methods which use equivalence checks.