Kirchweger, M., Peitl, T., & Szeider, S. (2023). Co-Certificate Learning with SAT Modulo Symmetries. In E. Elkind (Ed.), Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI-23) (pp. 1944–1953). International Joint Conferences on Artificial Intelligence. https://doi.org/10.24963/ijcai.2023/216
E192-01 - Forschungsbereich Algorithms and Complexity
-
Erschienen in:
Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI-23)
-
ISBN:
978-1-956792-03-4
-
Datum (veröffentlicht):
2023
-
Veranstaltungsname:
Thirty-Second International Joint Conference on Artificial Intelligence
en
Veranstaltungszeitraum:
19-Aug-2023 - 25-Aug-2023
-
Veranstaltungsort:
Macao, China
-
Umfang:
10
-
Verlag:
International Joint Conferences on Artificial Intelligence
-
Peer Reviewed:
Ja
-
Keywords:
Constraint Satisfaction; Optimization; Satisfiabilty; Solvers and tools
en
Abstract:
We present a new SAT-based method for generating all graphs up to isomorphism that satisfy a given co-NP property. Our method extends the SAT Modulo Symmetry (SMS) framework with a technique that we call co-certificate learning. If SMS generates a candidate graph that violates the given co-NP property, we obtain a certificate for this violation, i.e., `co-certificate' for the co-NP property. The co-certificate gives rise to a clause that the SAT solver, serving as SMS's backend, learns as part of its CDCL procedure. We demonstrate that SMS plus co-certificate learning is a powerful method that allows us to improve the best-known lower bound on the size of Kochen-Specker vector systems, a problem that is central to the foundations of quantum mechanics and has been studied for over half a century. Our approach is orders of magnitude faster and scales significantly better than a recently proposed SAT-based method.