Skip to content

Latest commit

 

History

History
59 lines (38 loc) · 3.09 KB

README.md

File metadata and controls

59 lines (38 loc) · 3.09 KB
Raw

SPUR - Satisfying Perfectly Uniform Random Sampler

docs docs

SPUR - Published at the SAT 2018 Conference
Winner of Best Student Paper

Title: Fast Sampling of Perfectly Uniform Satisfying Assignments
Authors: Dimitris Achlioptas, Zayd Hammoudeh, and Panos Theodoropoulos

SPUR is a fast uniform SAT assignment sampler. It was presented at SAT-2018.

Building and Running the Binary

For user convenience, we have included in the root of this repository a build script, build.sh. Running this script will build SPUR from source.

After building the binary files, SPUR can be run by calling from the command line:

build\Release\spur -s <#Samples> -cnf <FormulaFile>

where <#Samples> is the number of samples to collect (>=1) and <FormulaFile> is a path to Boolean CNF formula in DIMACS file format.

The full set of command line options are:

  • -cnf [cnf_file] Path to the CNF file
  • -s [s] Number of models "s" to sample uniformly at random
  • -tp Forces two-pass sampling. (Only applicable when s=1)
  • -out [out_file] Path to write the specified samples
  • -no-sample-write Disable writing the final samples to a file.
  • -count-only Perform only model counting. Disable sampling.
  • -q Quiet mode
  • -v Verbose and trace mode
  • -d Debug mode
  • -t [s] Set time bound to s seconds
  • -cs [n] Set max cache size to n MB

Dependencies

  • Operating System: This software is only tested on Ubuntu Linux and Mac OS High Sierra. Additional operating systems, but functionality is not guaranteed nor implied.
  • C++ Version: C++11
  • Libraries:

Output Format

As explained in our paper, the generated witnesses are represented as tuples. The first item in the tuple is an integer between 1 and the number of samples (inclusive); this number represents the number of witnesses entailed by this tuple.

The second item in the tuple is a variable assignment. Variables are ordered from 1 to N, where N is the total number of variables. Each variable is assigned to either "0", "1", or "*". Observe that the "*" corresponds to an unconstrained variable that can be either "0" or "1".

Example: The tuple "3,10*0*" means that three witnesses are entailed by the assignment where variable one is 0b1, variables two and four are 0b0, while variables three and five can take either value.

Python Libraries Available

Python libraries for running the SPUR binary are available upon request. Email zayd.hammoudeh@gmail.com to request access. Please include in the email request your name, email address, and institution.

Special Thanks

The SPUR sampler is built on top of sharpSAT, which was developed by Marc Thurley.