Download PDFOpen PDF in browser

Unsatisfiability Proofs for Parallel SAT Solver Portfolios with Clause Sharing and Inprocessing

15 pagesPublished: September 29, 2016

Abstract

State-of-the-art SAT solvers are highly tuned systematic-search procedures augmented with formula simplification techniques. They emit unsatisfiability proofs in the DRAT format to guarantee correctness of their answers. However, the DRAT format is inadequate to model some parallel SAT solvers such as the award-winning system \plingeling. In \plingeling, each solver in the portfolio applies clause addition and elimination techniques. Clause sharing is restricted to clauses that do not contain melted literals. In this paper, we develop a transition system that models the computation of such parallel portfolio solvers. The transition system allows us to formally reason about portfolio solvers, and we show that the formalism is sound and complete. Based on the formalism, we derive a new proof format, called parallel DRAT, which can be used to certify UNSAT answers.

Keyphrases: DRAT, formal model, parallel portfolio, SAT, unsatisfiability proofs

In: Christoph Benzmüller, Geoff Sutcliffe and Raul Rojas (editors). GCAI 2016. 2nd Global Conference on Artificial Intelligence, vol 41, pages 24--38

Links:
BibTeX entry
@inproceedings{GCAI2016:Unsatisfiability_Proofs_for_Parallel,
  author    = {Tobias Philipp},
  title     = {Unsatisfiability Proofs for Parallel SAT Solver Portfolios with Clause Sharing and Inprocessing},
  booktitle = {GCAI 2016. 2nd Global Conference on Artificial Intelligence},
  editor    = {Christoph Benzm\textbackslash{}"uller and Geoff Sutcliffe and Raul Rojas},
  series    = {EPiC Series in Computing},
  volume    = {41},
  pages     = {24--38},
  year      = {2016},
  publisher = {EasyChair},
  bibsource = {EasyChair, https://easychair.org},
  issn      = {2398-7340},
  url       = {https://easychair.org/publications/paper/vlFf},
  doi       = {10.29007/68qz}}
Download PDFOpen PDF in browser