Abstract

In this paper we study the problem of yielding robust performances from current state-of-the-art solvers for quantified Boolean formulas (QBFs). Analyzing the results of the 2006 QBF solvers competition (QBFEVAL’06), we find out that a set of inexpensive syntactic features is sufficient to determine the best solver to run on each formula with reasonable accuracy. Building on top of existing QBF solvers, we implement a new multi-engine solver which can learn its solver selection strategy from available datasets of solvers competitions such as QBFEVAL’06. Experimental results confirm that our solver is always more robust than each single engine, and that it is stable with respect to various perturbations of the dataset used for learning. As further experiments show, these results are in great part explained by a handful of features playing a crucial role in our solver.

I documenti depositati in UnissResearch sono protetti dalle leggi che regolano il diritto d'autore