borg logo

an algorithm portfolio toolkit and research project at the University of Texas at Austin


Current release

Solver collections

Training data

Competition entries

Borg software and data sets

The project source repository is public and hosted on GitHub. Installation and usage instructions are provided in the associated documentation, which is incomplete but under active expansion. Feel free to contact us with any questions that the documentation doesn't answer.

Recent entries of the borg portfolio into solver competitions are also available for download. These entries are of historical interest only, and contain known bugs! If you are interested in benchmarking borg, or applying it to some problem, you should use the most recent release.

The most significant practical obstacle to algorithm portfolio research is the computational cost of acquiring solver behavior data. As a service to the algorithm portfolio research community, the borg project also makes solver behavior data publicly available, along with tools for easily integrating these data into portfolio experiments. These data sets represent decades of processor time, and were collected on the 1500-node Mastodon cluster at the University of Texas at Austin. These data sets will be revised over time, and new sets will be released.

The project also packages suites of modern solvers for several problem domains, including satisfiability (SAT), maximum satisfiability and optimization (MAX-SAT), and pseudo-Boolean satisfiability and optimization (PB). These suites are useful as portfolio subsolvers. See the CREDITS file in each collection for citations to each solver's respective author(s).