Antiprenexing for WSkS: A Little Goes a Long Way

We study light-weight techniques for preprocessing of WSkS formulae in an automata-based decision procedure as implemented, e.g., in Mona. The techniques we use are based on antiprenexing, i.e., pushing quantifiers deeper into a formula. Intuitively, this tries to alleviate the explosion in the size of the constructed automata by making it happen sooner on smaller automata (and have the automata minimization reduce the output). The formula transformations that we use to implement antiprenexing may, however, be applied in different ways and extent and, if used in an unsuitable way, may also cause an explosion in the size of the formula and the automata built while deciding it. Therefore, our approach uses informed rules that use an estimation of the cost of constructing automata for WSkS formulae. The estimation is based on a model learnt from runs of the decision algorithm on various formulae. An experimental evaluation of our technique shows that antiprenexing can significantly boost the performance of the base WSkS decision procedure, sometimes allowing one to decide formulae that could not be decided before.