GRASP—a new search algorithm for satisfiability
Proceedings of the 1996 IEEE/ACM international conference on Computer-aided design
A machine program for theorem-proving
Communications of the ACM
Efficient conflict driven learning in a boolean satisfiability solver
Proceedings of the 2001 IEEE/ACM international conference on Computer-aided design
Survey propagation: An algorithm for satisfiability
Random Structures & Algorithms
Google's PageRank and Beyond: The Science of Search Engine Rankings
Google's PageRank and Beyond: The Science of Search Engine Rankings
IJCAI'99 Proceedings of the 16th international joint conference on Artificial intelligence - Volume 2
The community structure of SAT formulas
SAT'12 Proceedings of the 15th international conference on Theory and Applications of Satisfiability Testing
Hi-index | 0.00 |
Despite the success of modern SAT solvers on industrial instances, most of the progress relies on intensive experimental testing of improvements or new ideas. In most cases, the behavior of CDCL solvers cannot be predicted and even small changes may have a dramatic positive or negative effect. In this paper, we do not try to improve the performance of SAT solvers, but rather try to improve our understanding of their behavior. More precisely, we identify an essential structural property of industrial instances, based on the Eigenvector centrality of a graphical representation of the formula. We show how this static value, computed only once over the initial formula casts new light on the behavior of CDCL solvers. We also advocate for a better partitionning of industrial problems. Our experiments clearly suggest deep discrepancies among the families of benchmarks used in the last SAT competitions.