Captain Jack: New Variable Selection Heuristics in Local Search for SAT

  • Dave A. D. Tompkins
  • Adrian Balint
  • Holger H. Hoos
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6695)


Stochastic local search (SLS) methods are well known for their ability to find models of randomly generated instances of the propositional satisfiability problem (SAT) very effectively. Two well-known SLS-based SAT solvers are Sparrow, one of the best-performing solvers for random 3-SAT instances, and VE-Sampler, which achieved significant performance improvements over previous SLS solvers on SAT-encoded software verification problems. Here, we introduce a new highly parametric algorithm, Captain Jack, which extends the parameter space of Sparrow to incorporate elements from VE-Sampler and introduces new variable selection heuristics. Captain Jack has a rich design space and can be configured automatically to perform well on various types of SAT instances. We demonstrate that the design space of Captain Jack is easy to interpret and thus facilitates valuable insight into the configurations automatically optimized for different instance sets. We provide evidence that Captain Jack can outperform well-known SLS-based SAT solvers on uniform random k-SAT and ‘industrial-like’ random instances.


Search Step Stochastic Local Search Promising Step Promising Variable Stochastic Local Search Algorithm 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Ansótegui, C., Bonet, M.L., Levy, J.: Towards industrial-like random SAT instances. In: IJCAI 2009, pp. 387–392 (2009)Google Scholar
  2. 2.
    Balint, A., Diepold, D., Gall, D., Gerber, S., Kapler, G., Retz, R.: EDACC - an advanced platform for the experiment design, administration and analysis of empirical algorithms. In: LION-2011 (to appear)Google Scholar
  3. 3.
    Balint, A., Fröhlich, A.: Improving stochastic local search for SAT with a new probability distribution. In: Strichman, O., Szeider, S. (eds.) SAT 2010. LNCS, vol. 6175, pp. 10–15. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  4. 4.
    Biere, A.: PicoSAT essentials. JSAT 4, 75–97 (2008)zbMATHGoogle Scholar
  5. 5.
    bwGRiD: Member of the German D-Grid initiative, funded by the Ministry of Education and Research and the Ministry for Science, Research and Arts Baden-WürttembergGoogle Scholar
  6. 6.
    Chvátal, V., Szemerédi, E.: Many hard examples for resolution. Journal of the ACM 35(4), 759–768 (1988)CrossRefzbMATHGoogle Scholar
  7. 7.
    Hoos, H.H.: Computer-aided design of high-performance algorithms. Tech. Rep. TR-2008-16, University of British Columbia (2008)Google Scholar
  8. 8.
    Hutter, F., Hoos, H.H., Leyton-Brown, K., Stützle, T.: ParamILS: An automatic algorithm configuration framework. Journal of Artificial Intelligence Research 36, 267–306 (2009)zbMATHGoogle Scholar
  9. 9.
    Hutter, F., Hoos, H.H., Stützle, T.: Automatic algorithm configuration based on local search. In: AAAI 2007, pp. 1152–1157 (2007)Google Scholar
  10. 10.
    KhudaBukhsh, A.R., Xu, L., Hoos, H.H., Leyton-Brown, K.: SATenstein: Automatically building local search SAT solvers from components. In: IJCAI 2009, pp. 517–524 (2009)Google Scholar
  11. 11.
    Li, C.M., Huang, W.Q.: Diversification and determinism in local search for satisfiability. In: Bacchus, F., Walsh, T. (eds.) SAT 2005. LNCS, vol. 3569, pp. 158–172. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  12. 12.
    Mertens, S., Mézard, M., Zecchina, R.: Threshold values of random k-SAT from the cavity method. Random Structures & Algorithms 28, 340–373 (2006)CrossRefzbMATHGoogle Scholar
  13. 13.
    Pham, D.N., Thornton, J., Gretton, C., Sattar, A.: Combining adaptive and dynamic local search for satisfiability. JSAT 4, 149–172 (2008)zbMATHGoogle Scholar
  14. 14.
    Prestwich, S.: Random walk with continuously smoothed variable weights. In: Bacchus, F., Walsh, T. (eds.) SAT 2005. LNCS, vol. 3569, pp. 203–215. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  15. 15.
    Thornton, J., Pham, D.N., Bain, S., Ferreira Jr., V.: Using cost distributions to guide weight decay in local search for SAT. In: Ho, T.-B., Zhou, Z.-H. (eds.) PRICAI 2008. LNCS (LNAI), vol. 5351, pp. 405–416. Springer, Heidelberg (2008)CrossRefGoogle Scholar
  16. 16.
    Tompkins, D.A.D.: Dynamic Local Search for SAT: Design, Insights and Analysis. Ph.D. thesis, University of British Columbia (2010)Google Scholar
  17. 17.
    Tompkins, D.A.D., Hoos, H.H.: UBCSAT: An implementation and experimentation environment for SLS algorithms for SAT and MAX-SAT. In: H. Hoos, H., Mitchell, D.G. (eds.) SAT 2004. LNCS, vol. 3542, pp. 306–320. Springer, Heidelberg (2005)CrossRefGoogle Scholar
  18. 18.
    Tompkins, D.A.D., Hoos, H.H.: Dynamic scoring functions with variable expressions: New SLS methods for solving SAT. In: Strichman, O., Szeider, S. (eds.) SAT 2010. LNCS, vol. 6175, pp. 278–292. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  19. 19.
    Xu, L., Hutter, F., Hoos, H.H., Leyton-Brown, K.: SATzilla: Portfolio-based algorithm selection for SAT. Journal of Artificial Intelligence Research 32, 565–606 (2008)zbMATHGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Dave A. D. Tompkins
    • 1
  • Adrian Balint
    • 2
  • Holger H. Hoos
    • 1
  1. 1.Department of Computer ScienceUniversity of British ColumbiaCanada
  2. 2.Institute of Theoretical Computer ScienceUlm UniversityGermany

Personalised recommendations