Theorem proving in propositional logic on vector computers using a generalized Davis-Putman procedure

Abstract

The Davis-Putman procedure (DPP) is an efficient method for solving the theorem proving problem in propositional logic. The authors present an effective technique for vectorizing the DPP. To speed up the execution of DPP, the rules used by the procedure are first generalized by considering more than one literal at a time. Then vectorized algorithms based on the generalized rules are proposed. Experiments are conducted on vector computers. The results show that the vectorized version of the Davis-Putnam procedure is effective in solving a variety of instances of the theorem proving problem in propositional logic. The vectorized version of the DPP was effective in cutting down the search space and thus substantially improving the search efficiency. >