We describe a method for computing a minimum-state automaton to act as an intermediate assertion in assume-guarantee reasoning, using a sampling approach and a Boolean satisfiability solver. For a set of synthetic benchmarks intended to mimic common situations in hardware verification, this is shown to be significantly more effective than earlier approximate methods based on Angluin's L* algorithm. For many of these benchmarks, this method also outperforms BDD-based model checking and

more »

... ion-based model checking. We also demonstrate how domain knowledge can be incorporated into our algorithm to improve its performance. 1. Identifying an appropriate decomposition of the system. 2. Identifying a simple assumption.