Title :
Guiding CNF-SAT search via efficient constraint partitioning
Author :
Durairaj, Vijay ; Kalla, Priyank
Author_Institution :
Dept. of Electr. & Comput. Eng., Utah Univ., Salt Lake City, UT, USA
Abstract :
Contemporary techniques to identify a good variable order for SAT rely on identifying minimum tree-width decompositions. However, the problem of finding a minimal width tree decomposition for an arbitrary graph is NP complete. The available tools and methods are impractical, as they cannot handle large and hard-to-solve CNF-SAT instances. This work proposes a hypergraph partitioning based constraint decomposition technique as an alternative to contemporary methods. We model the CNF-SAT problem on a hypergraph and apply min-cut based bi-partitioning. Clause-variable statistics across the partitions are analyzed to further decompose the problem, iteratively. The resulting tree-like decomposition provides a variable order for guiding CNF-SAT search. Experiments demonstrate that our partitioning procedure is fast, scalable and the derived variable order results in significant increase in performance of the SAT engine.
Keywords :
computability; computational complexity; constraint theory; logic design; optimisation; search problems; trees (mathematics); CNF-SAT search; NP complete; arbitrary graph; clause-variable statistics; constraint decomposition technique; efficient constraint partitioning; hypergraph partitioning; min-cut based bi-partitioning; minimum tree-width decompositions; Circuits; Cities and towns; Computational complexity; Engines; Information analysis; Statistical analysis; Tree data structures; Tree graphs;
Conference_Titel :
Computer Aided Design, 2004. ICCAD-2004. IEEE/ACM International Conference on
Print_ISBN :
0-7803-8702-3
DOI :
10.1109/ICCAD.2004.1382629