shilch / mustool

MUST

MUST is a tool for online enumeration of minimal unsatisfiable subsets (MUSes) of a given unsatisfiable set of constraints. The tool currently implements three online MUS enumeration algorithms: MARCO [9], TOME [2], and ReMUS [3], and supports MUS enumeration in three constraint domains: SAT, SMT, and LTL. A paper [1] describing MUST will be presented at TACAS 2020.

We distribute this source code under the MIT licence. See ./LICENSE for mode details.

Installation

To be able to deal with the three constraint domains, our tool employs several external libraries. In particular, we use:

• miniSAT [4] and zlib library for dealing with the SAT domain
• z3 [6] for dealing with the SMT domain
• and SPOT [7] for dealing with the LTL domain.

MiniSAT is packed directly with the source code of our tool; you do not install it separately. Zlib has to be installed. Finally, z3 and/or SPOT need to be installed only if you want to use our tool in the SMT and/or LTL domains. Note that installation of z3 and SPOT might take several hours.

Installation of zlib

Zlib is a part of the package zlib1g-dev, you should be able to install it with:

sudo apt install zlib1g-dev

Installation of z3

z3 can be downloaded at https://github.com/Z3Prover/z3 Please follow its README file for installation instructions. Basically, the following should do the trick:

python scripts/mk_make.py
cd build; make
sudo make install

Installation of SPOT

SPOT can be downloaded from https://spot.lrde.epita.fr/ Again, follow installation instructions that are provided by its authors. Basically, the following should do the trick:

./configure --disable-python
make
sudo make install
sudo ldconfig

Note the "sudo ldconfig", that is important. Do not forget to run it, otherwise SPOT might not be visible for our tool.

Building our tool

Building our tool requires at least zlib to be installed. To build our tool, go to its main directory (the one where is this README.md file) and run:

make

This will make our tool only with support for the SAT domain. If you want to support also the other domains, you have to install z3 and/or SPOT first. Then run one of the following:

make USESMT=YES
make USELTL=YES
make USESMT=YES USELTL=YES

If you run make repeatedly, e.g. if you have decided to allow another domain, run first "make clean" and then appropriate make.

Running our tool

in the main directory, run "./must file" where file is an input file of constraints. You can use one of our examples, e.g.:

./must examples/test.cnf
./must examples/test.smt2
./must examples/test.ltl
./must examples/bf1355-228.cnf

To run the tool with a time limit (always recommended), use e.g.:

timeout 20 examples/bf1355-228.cnf

To save the identified MUSes into a file, run:

timeout 10 ./must -o output_file examples/test.cnf

To see all the available parameters, run:

./must -h

Other Third-Party Tools

Besides the above mentioned tools, we also use two single MUS extraction algorithms: muser2 [5] and mcsmus [8]. You do not have to install these. Muser2 is presented in our repo in a binary form, and mcsmus is compiled with our code.

References

• [1] Jaroslav Bendík and Ivana Černá: MUST: Minimal Unsatisfiable Subsets Enumeration Tool. TACAS 2020.
• [2] Jaroslav Bendík, Nikola Beneš, Ivana Černá, Jiří Barnat: Tunable Online MUS/MSS Enumeration. FSTTCS 2016.
• [3] Jaroslav Bendík, Ivana Černá, Nikola Beneš: Recursive Online Enumeration of All Minimal Unsatisfiable Subsets. ATVA 2018.
• [4] http://minisat.se/
• [5] https://bitbucket.org/anton_belov/muser2
• [6] https://github.com/Z3Prover/z3
• [7] https://spot.lrde.epita.fr/
• [8] https://bitbucket.org/gkatsi/mcsmus
• [9] Mark H. Liffiton, Alessandro Previti, Ammar Malik, João Marques-Silva: Fast, flexible MUS enumeration. Constraints 21(2), 2016.

Citing

If you use our tool (MUST) in your research, please cite our paper that presented MUST:

@inproceedings{DBLP:conf/tacas/BendikC20,
  author    = {Jaroslav Bend{\'{\i}}k and
               Ivana Cern{\'{a}}},
  title     = {{MUST:} Minimal Unsatisfiable Subsets Enumeration Tool},
  booktitle = {{TACAS} {(1)}},
  series    = {Lecture Notes in Computer Science},
  volume    = {12078},
  pages     = {135--152},
  publisher = {Springer},
  year      = {2020}
}

Contact

In case of any troubles, do not hesitate to contact me, Jaroslav Bendik, the developer of the tool, at xbendik=at=fi.muni.cz.