A Two-Levels Local Search Algorithm for Random SAT Instances with Long Clauses
Abramé André, Djamal Habet, Donia Toumi · Frontiers in artificial intelligence and applications · 2014
We present a local search algorithm exploiting two efficient solvers for SAT. The first one is based on the configuration checking strategy and the second one on an algorithm of the Walksat family. This new solver is dedicated to solve random k-SAT instances, such that k ≥ 4. We have carried tests on the instances of the SAT Challenge 2012. The obtained results confirm the relevance of our approach.