一句话:将语义推导的可行插值定理应用于随机归结,并由此得到团-着色公式随机归结反驳的下界。
随机归结是 Buss、Kołodziejczyk 和 Thapen 提出的一种可靠命题证明系统:它在归结基础上允许把随机选取的子句(在特定技术条件下)加入初始子句集。本文研究如何为该系统建立可行插值,即把 Krajíček 关于语义推导的一般可行插值定理应用到随机归结。作为推论,作者证明了团-着色公式不存在短的随机归结反驳,给出了该类公式随机归结复杂度的下界。
Random resolution, defined by Buss, Kolodziejczyk and Thapen (JSL, 2014), is a sound propositional proof system that extends the resolution proof system by the possibility to augment any set of initial clauses by a set of randomly chosen clauses (modulo a technical condition). We show how to apply the general feasible interpolation theorem for semantic derivations of Krajicek (JSL, 1997) to random resolution. As a consequence we get a lower bound for random resolution refutations of the clique-coloring formulas.Comment: Preprint April 2016, revised September and October 2016