Understanding the Relative Strength of QBF CDCL Solvers and QBF Resolution

TL;DR

This paper models QCDCL QBF solvers as proof systems, analyzing their strength, limitations, and introducing new policies that can outperform traditional methods on certain formulas.

Contribution

It formally models QCDCL as proof systems, compares their strength to Q-resolution, and introduces new heuristics and proof systems with improved performance.

Findings

QCDCL proof system is incomparable to Q-resolution with exponential separations.

Lower bounds for QCDCL, including exponential bounds for random QBFs.

A new QCDCL proof system p-equivalent to Q-resolution with adapted heuristics.

Abstract

QBF solvers implementing the QCDCL paradigm are powerful algorithms that successfully tackle many computationally complex applications. However, our theoretical understanding of the strength and limitations of these QCDCL solvers is very limited. In this paper we suggest to formally model QCDCL solvers as proof systems. We define different policies that can be used for decision heuristics and unit propagation and give rise to a number of sound and complete QBF proof systems (and hence new QCDCL algorithms). With respect to the standard policies used in practical QCDCL solving, we show that the corresponding QCDCL proof system is incomparable (via exponential separations) to Q-resolution, the classical QBF resolution system used in the literature. This is in stark contrast to the propositional setting where CDCL and resolution are known to be p-equivalent. This raises the question what formulas are hard for standard QCDCL, since Q-resolution lower bounds do not necessarily apply to QCDCL as we show here. In answer to this question we prove several lower bounds for QCDCL, including exponential lower bounds for a large class of random QBFs. We also introduce a strengthening of the decision heuristic used in classical QCDCL, which does not necessarily decide variables in order of the prefix, but still allows to learn asserting clauses. We show that with this decision policy, QCDCL can be exponentially faster on some formulas. We further exhibit a QCDCL proof system that is p-equivalent to Q-resolution. In comparison to classical QCDCL, this new QCDCL version adapts both decision and unit propagation policies.