This archive provides a demonstration version of a proof-generating
SAT solver based on BDDs. The solver generates proofs in LRAT format.
A checker for this format is included.
The program also generates and tests two benchmark problems: the
mutilated chessboard problem and the pigeonhole problem.
1. System Requirements (Tested with TACAS virtual environment)
* Python interpreter. By default, the program runs python3.
You can change by editing the "INTERP" definition in the
top-level Makefile
* C compiler. The LRAT checker is written in C. Just about
any compiler should work. The default is gcc.
2. Installation and Running Demonstration
All code and benchmark data can be downloaded in two different ways:
A. Using GitHub
git clone https://github.com/rebryant/pgbdd-artifact
B. Getting the Zip File
wget http://www.cs.cmu.edu/~bryant/download/pgbdd-artifact.zip
unzip pgbdd-artifact.zip
Once downloaded, the two demonstrations can be run as:
cd pgbdd-artifact
make chess
make pigeon
For both sets of benchmarks, a lot of stuff gets printed, but the
final summary information is saved in files chess-results.txt and
pigeon-results.txt in the top-level directory. See the included file
"artifact-documentation.pdf" for more information about the benchmarks
and how to interpret the results.
3. Makefile options:
install:
Compiles the LRAT checker. No other installation steps are required.
chess:
Generate and test versions of the N x N mutilated chessboard up to
N=40. This demonstrates the power of the BDD-based approach plus a
novel method of organizing the sequencing of conjunction and
qexistential quantification operations inspired by symbolic model
checking.
A lot of stuff gets printed, but the final summary information is
saved in a file "chess-results.txt" in the top-level directory. See
the included file "artifact-documentation.pdf" for more information
about the benchmarks and how to interpret the results.
pigeon:
Generate and test versions of the pigeonhole problem with N holes
and N+1 pigeons up to N=40. This demonstrates the power of the
BDD-based approach plus a novel method of organizing the sequencing
of conjunction and existential quantification operations inspired by
symbolic model checking.
A lot of stuff gets printed, but the final summary information is
saved in a file "pigeon-results.txt" in the top-level directory.
See the included file "artifact-documentation.pdf" for more
information about the benchmarks and how to interpret the results.
test:
Generate and run simple examples of the two benchmarks. A lot of
stuff gets printed out, but you should see the statement "VERIFIED"
for each benchmark. All runtime data are saved in files named as
follows:
benchmark/chess-data/chess-NNN-MMMMMM.data
NNN: Size of chess board
MMMMMM: Method used. linear/bucket/column/noquant
Of these, column is the most interesting, generating proofs
with growth O(n^2.7). The files contains lots of information
about the execution. The word VERIFIED indicates success.
benchmark/pigeon-data/pigeon-tseitin-NNN-MMMMMM.data
NNN: Number of holes
MMMMMM: Method used. linear/bucket/column
Of these, column is the most interesting, generating proofs
with growth O(n^3). The files contains lots of information
about the execution. The word VERIFIED indicates success.
regress:
Generate and run an extensive set of examples of the two benchmarks
using different strategies for scheduling operations. These have
been scaled down to use less than 30 seconds each. A lot of stuff
gets printed out, but you should see the statement "VERIFIED" for
each benchmark. The data files are located and named as described
above.
clean:
Delete all generated files, except for any benchmark data
superclean:
Delete all generated files, including the benchmark data
4. Using PGBDD as a standalone SAT solver
The PGBDD SAT solver is available as the program pgbdd.py in the
directory pgbdd-artifact/solver. Its use is documented in the file
pgbdd-artifact/solver/README.txt.