Enhanced Integration of QBF Solving Techniques
Sven Reimer, Florian Pigorsch, Christoph Scholl, Bernd Becker · 2012
In this paper we present a novel QBF solving technique which is based on the integration of a search based (DPLL) and a rewriting based approach: While traversing the search space in a DPLL manner, we iteratively generate many sub-problems, which are handed over to the rewriting method one by one. Instead of just communicating back satisfiability results of the individual sub-problems, we collect as many constraints derived by the rewriting based solver as possible, and transfer them back to the search based solver. This allows not only to prune the current branch, but also to avoid the unnecessary traversal of search paths in different regions of the search tree. We also discuss heuristics to determine suitable switching points between these two methods. We present first promising results that underline the potential of our approach. 1.