In this paper we take 4 different features of the SAT solver CaDiCaL, blocked clause elimination, vivification, on-the-fly self subsumption, and increasing the bound of variable elimination over the SAT Competitions benchmarks between 2009 and 2022. We study these features by both activating them one-by-one and deactivating them one-by-one. We have three hypothesis regarding the experiments: (i) disabling features is always harmful; (ii) the life span of the techniques is limited; and (iii) features simulate each other. Our experiments cannot confirm any of the hypothesis.

翻译：暂无翻译