This package implements a number of tools that operate on DIMACS CNF files.
Here is a short summary of the tools and their usage:
- cnf-cat [FILE] ...
Concatenate the clauses of one or more CNF files.
- cnf-clause [[LITERAL] ... 0] ...
Generate a CNF file containing the clauses specified on the command
line. Example:
$ cnf-clause 1 2 0 -1 3 0
p cnf 3 2
1 2 0
-1 3 0
- cnf-grep VARIABLE [FILE]
Read the CNF file specified by FILE (or standard input if omitted)
and output the clauses containing the variable VARIABLE. Example:
$ cnf-clause 1 2 0 -1 3 0 | cnf-grep 2
p cnf 3 1
1 2 0
- cnf-renumber [FILE]
Read the CNF file specified by FILE (or standard input if omitted)
and rename the variables so that they start at 1 and go to N, where
N is the number of distinct variables in the input. Example:
$ cnf-clause 100 101 0 | cnf-pack
p cnf 2 1
1 2 0
- cnf-propagate [FILE]
Read the CNF file specified by FILE (or standard input if omitted)
and carry out unit propagation recursively until no further
propagations can be made. Note that no clauses are removed; both
unit clauses and tautologies (e.g. 1 -1 0) will remain in the output.
Example:
$ cnf-clause 1 2 0 -1 3 0 -3 0 | cnf-propagate
p cnf 3 3
2 0
-1 0
-3 0
- cnf-shuffle-clauses [FILE]
Shuffle the order of the clauses, but leave the order of literals
within each clause intact.
- cnf-shuffle-literals [FILE]
Shuffle the order of the literals within each clause, but leave the
order of the clauses intact.
- cnf-shuffle-variables [FILE]
rENAME the variables randomly using a one-to-one mapping. The
resulting CNF output is equisatisfiable with the CNF input.
Example:
$ cnf-clause 1 2 3 0 | cnf-shuffle-variables
p cnf 3 1
2 3 1 0
- cnf-sort-clauses [FILE]
Order the clauses in ascending order. Example:
$ cnf-clause 3 2 1 0 -3 -2 -1 0 | cnf-sort-clauses
p cnf 3 2
-3 -2 -1 0
3 2 1 0
- cnf-sort-literals [FILE]
Order the literals within each clause in ascending order. Example:
$ cnf-clause 3 2 1 0 -3 -2 -1 0 | cnf-sort-literals
p cnf 3 2
1 2 3 0
-3 -2 -1 0
- cnf-stat [FILE]
Print some simple statistics about the CNF input. Example:
$ cnf-stat 20-r-16-160-r-00-shuffle-1.cnf
Literals: 121157 (61952 negative, 59205 positive)
Clause length histogram:
1: 1
2: 15608
3: 29980
- cnf-fuzz-nossum
Generate a random problem. By Vegard Nossum, with additions by
Mate Soos
- cnf-fuzz-biere
CNF fuzzer by Armin Biere (April 2010). See Armin Biere's paper
in SAT'10 proceedings for details
- cnf-fuzz-brummayer.py
Fuzzing Tool for CNF, written by Robert Daniel Brummayer, 2009
- compare.pl
Compares two solvers side-by-side, by Vegard Nossum.
Usage:
perl compare.pl program1 program2 args
The output from program1 will be printed in yellow/brown, while the
output from program2 will be printed in green. Output lines are
prefixed with a timestamp which is basically the number of seconds
(not CPU time) since the program was started. The arguments will be
supplied to both programs.
- sgen4
The program generates small, difficult satisfiability benchmarks in
the DIMACS cnf format. The benchmark is written to standard output
and the available options are:
-n number-of-variables This specifies the minimum number of variables - there may be slightly more.
-sat Generate a satisfiable instance
-unsat Generate an unsatisfiable instance
-[no]reorder Randomly reorder literals and clauses in the generated instance
-s seed Specify a seed for random number generation
-m model-filename If a satisfiable instance is being generated, write a satisfying model
-min-variables The normal mode of operation is to be as difficult as possible for a small number of literals. This options means use as few variables as possible.
Typical usage:
sgen4 -n 60 -unsat > unsat-60.cnf
sgen4 -n 150 -sat > sat-60.cnf
sgen4 -n 150 -sat -m sat-60.model > sat-60.cnf
See file sgen4descr.pdf for further details