CNF2OBDD or BDD_MINISAT_ALL
Takahisa Toda
The University of Electro-Communications
Graduate School of Informatics and Engineering
Associate Professor
A CNF to OBDD compiler or a formula-BDD caching AllSAT solver, implemented on top of MiniSat-C v1.14.1, is presented.
This software has two names because of a double meaning.
It may be called cnf2obdd to put more emphasis on a CNF to OBDD compiler aspect, which supports rich operations for postprocessing, and it also may be called bdd_minisat_all on an AllSAT solver aspect, which simply generates all solutions, a familiar format to wider users.
We surveyed major techniques of AllSAT solvers and conducted comprehensive experiments, which is published as an open access paper here in ACM Journal of Experimental Algorithmics.
Boolean formula should be in DIMACS CNF format.
For details of DIMACS CNF format and benchmark problems, see SATLIB.
If no option is given, standard mode is selected.
$ tar zxvf bdd_minisat_all-1.0.0.tar.gz
$ cd bdd_minisat_all-1.0.0
$ make [options]
list of options
s standard: debug information used by debugger is generated at compilation time, and detailed solver status is reported at runtime.
p profile: in addition to standard setting, profile information used by gprof is generated at compilation time and several tests are performed at runtime.
d debug: in addition to standard setting, several tests are performed at runtime and no optimization is applied.
r release: release version, compiled with dynamic link
rs static: release version, compiled with static link
clean executable files, object files, etc are removed.
If an output file is specified, all satisfying assignments to a CNF are generated in DIMACS CNF format without problem line.
Notice: there may be as many number of assignments as can not be stored in a disk space.
If you want to use refresh option, define REFRESH in Makefile
.
Usage: ./bdd_minisat_all [options] input-file [output-file]
-n<int> maximum number of obdd nodes: if exceeded, obdd is refreshed
Since variable orderings significantly affect solver's performance, we recommend that users in advance execute the software MINCE to determine a good variable ordering.
Program behavior can be controlled by defining or not defining the following macros in
Makefile
.
-
CUTSETCACHE: Cutset is selected as formula-BDD caching. If this is not defined, separator is selected.
-
LAZY: Cache operations are reduced (Recommended).
Select one of the two base solver types: a blocking solver or a non-blocking solver.
-
NONBLOCKING non-blocking procedure is selected as a base solver. If this is not defined, blocking procedure is selected. For options of non-blocking solvers, see nbc_minisat_all.
-
REFRESH: refresh option in command line is enabled. If the number of BDD nodes exceeds a specified threshold, all solutions are dumpted to a file (if output file is specified in command line), all caches are refreshed, and search is continued.
Other options are as follows.
-
TRIE_RECURSION: Recursive version of trie is used. If this is not defined, iterative version is used.
-
REDUCTION: Reduction of compiled OBDD into fully reduced one is performed using CUDD library (Optional). To do this, obtain CUDD library and modify Makefile.
-
GMP: GNU MP bignum library is used to count solutions. To do this, obtain GMP library and modify Makefile.
bdd_minisat_all is implemented by modifying MiniSat-C_v1.14.1. Please confirm the license file included in this software.
A huge number of assignments may be generated and disk space may be exhausted. To avoid disk overflow, take measure such as using ulimit command.
Moreover, bdd_minisat_all requires much memory. In case of memory shortage, the solver halts with the following error message.
error:"file name":"line number":"function name":memory allocation failed
In this case, define the REFRESH macro in Makefile, compile the code again, execute with -n option.
-
Takahisa Toda, Takehide Soh, Implementing Efficient All Solutions SAT Solvers, Retrived October 5th, 2015 from: arXiv:1510.00523.
-
N. E\E9n, N. S\F6rensson : MiniSat Page: MiniSat-C_v1.14.1, accessed on 15 Dec. 2014.
-
N. E\E9n, N. S\F6rensson : An Extensible SAT-solver, In Proceedings of the 6th international conference of Theory and Applications of Satisfiability Testing, pages 502--518, 2004.
-
Grumberg, Orna and Schuster, Assaf and Yadgar, Avi: Memory Efficient All-Solutions SAT Solver and Its Application for Reachability Analysis, Formal Methods in Computer-Aided Design, LNCS Vol.3312, pp.275-289, 2004.
-
Gebser, Martin and Kaufmann, Benjamin and Neumann, Andr{\'e} and Schaub, Torsten: Conflict-driven Answer Set Enumeration, in Proceedings of the 9th International Conference on Logic Programming and Nonmonotonic Reasoning, pp.136--148, 2007.
-
Morgado, A. and Marques-Silva, J.: Good learning and implicit model enumeration, In Proceedings of 17th IEEE International Conference on Tools with Artificial Intelligence, 2005. ICTAI 05.
-
Yinlei Yu and Subramanyan, P. and Tsiskaridze, N. and Malik, S.: All-SAT Using Minimal Blocking Clauses, In Proceedings of 27th International Conference on VLSI Design and 2014 13th International Conference on Embedded Systems, 2014, pp.86-91.
-
McMillan, KenL.: Applying SAT Methods in Unbounded Symbolic Model Checking, Computer Aided Verification, LNCS Vol.2404, pp.250-264. 2002.
-
Weinan Zhao and Weimin Wu: ASIG: An all-solution SAT solver for CNF formulas, In Proceedings of 11th IEEE International Conference on Computer-Aided Design and Computer Graphics, 2009. CAD/Graphics '09, pp. 508-513, 2009.
-
T. Toda, K. Tsuda: BDD Construction for All Solutions SAT and Efficient Caching Mechanism, Track on Constraint Solving and Programming and Knowledge Representation and Reasoning (CSP-KR) part of the 30th Annual ACM Symposium on Applied Computing, 2015.
-
J. Huang, A. Darwiche: Using DPLL for efficient OBDD construction, In Proceedings of the 7th international conference on Theory and Applications of Satisfiability Testing, pages 157--172, 2005.
-
Y, Crama, and P. Hammer: Boolean Functions: Theory, Algorithms, and Applications. Encyclopedia of Mathematics and its Applications. Cambridge University Press (2011).
-
R.E. Bryant: Graph-Based algorithm for Boolean function manipulation, IEEE Trans. Comput., Vol.35, pp.677-691 (1986)
-
D.E. Knuth: The Art of Computer Programming Volume 4a, Addison-Wesley Professional, New Jersey, USA (2011)
-
F. Somenzi: CU Decision Diagram Package: Release 2.5.0., accessed on 4 September 2012.