Skip to content

NTU-ALComLab/ssat-benchmarks

Repository files navigation

Collection of Stochastic SAT Instances

Purpose

This collection of stochastic SAT (SSAT) instances is intended to be a benchmarking suite for evaluating the efficiency of decision procedures for SSAT solving.

Format

The SSAT instances in this collection are encoded with the sdimacs format, which adapts the qdimacs format for quantified Boolean formulas.

A randomly quantified variable x (with probability p) is written as r p x 0. For example, the SSAT query:

exist x1, exist x2, random p=0.5 x3, random p=0.5 x4. (x1 or x3) and (x2 != x4).

is encoded as follows:

p cnf 4 3
e 1 0
e 2 0
r 0.5 3 0
r 0.5 4 0
1 3 0
2 4 0
-2 -4 0

Collected SSAT Families

The collection consists of the following families of SSAT instances:

  • re-random-k-CNF/: RE-SSAT instances converted from random k-CNF formulas
  • re-strategic-company/: RE-SSAT instances converted from strategic-company QBFs
  • re-PEC/: RE-SSAT instances encoding the PEC problem
  • er-random-k-CNF/: ER-SSAT instances converted from random k-CNF formulas
  • ere-sand-castle/: ERE-SSAT instances encoding the sand-castle problem
  • ere-ToiletA/: ERE-SSAT instances converted from ToiletA QBFs
  • ere-conformant/: ERE-SSAT instances converted from conformant-planning QBFs
  • ere-MaxCount/: ERE-SSAT instances converted from maximum-model-counting formulas

Contribution

This repository is open for submission of new SSAT instances. Please fork the repository, commit your changes, and send a pull request.

License

The SSAT instances are under open-source license Apache 2.0.

Contact

You are welcome to create an issue for discussion.

About

Collection of SSAT instances

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published