Hyperplane Elimination for Quickly Enumerating Local Optima

Part of the Lecture Notes in Computer Science book series (LNCS, volume 9595)

Abstract

Examining the properties of local optima is a common method for understanding combinatorial-problem landscapes. Unfortunately, exhaustive algorithms for finding local optima are limited to very small problem sizes. We propose a method for exploiting problem structure to skip hyperplanes that cannot contain local optima, allowing runtime to scale with the number of local optima instead of with the landscape size. We prove optimality for linear functions and Concatenated Traps, and we provide empirical evidence of optimality on NKq Landscapes and Ising Spin Glasses. We further refine this method to find solutions that cannot be improved by flipping r or fewer bits, which counterintuitively can reduce total runtime. While previous methods were limited to landscapes with at most \(2^{34}\) binary strings, hyperplane elimination can enumerate the same problems with \(2^{77}\) binary strings, and find all 4-bit local optima of problems with \(2^{200}\) binary strings.

Keywords

Landscape understanding Gray-Box Mk Landscapes 

References

  1. 1.
    Boese, K.D., Kahng, A.B., Muddu, S.: A new adaptive multi-start technique for combinatorial global optimizations. Oper. Res. Lett. 16(2), 101–113 (1994)MathSciNetCrossRefMATHGoogle Scholar
  2. 2.
    Chicano, F., Whitley, D., Sutton, A.M.: Efficient identification of improving moves in a ball for pseudo-boolean problems. In: Genetic and Evolutionary Computation Conference, Vancouver, BC, Canada, pp. 437–444. ACM, 12–16 July 2014Google Scholar
  3. 3.
    Deb, K., Goldberg, D.E.: Analyzing deception in trap functions. In: Foundations of Genetic Algorithms Conference, Vail, Colorado, 24–29 July 1992Google Scholar
  4. 4.
    Goldman, B.W., Punch, W.F.: Gray-box optimization using the parameter-less population pyramid. In: Genetic and Evolutionary Computation Conference, pp. 855–862. ACM (2015)Google Scholar
  5. 5.
    Goldman, B.W., Tauritz, D.R.: Linkage tree genetic algorithms: variants and analysis. In: Genetic and Evolutionary Computation Conference, Philadelphia, Pennsylvania, USA, pp. 625–632. ACM, 7–11 July 2012Google Scholar
  6. 6.
    Hsu, S.H., Yu, T.L.: Optimization by pairwise linkage detection, incremental linkage set, and restricted/back mixing: Dsmga-ii. In: Genetic and Evolutionary Computation Conference, pp. 519–526. ACM (2015)Google Scholar
  7. 7.
    Iclănzan, D., Daolio, F., Tomassini, M.: Learning inherent networks from stochastic search methods. In: Blum, C., Ochoa, G. (eds.) EvoCOP 2014. LNCS, vol. 8600, pp. 157–169. Springer, Heidelberg (2014)Google Scholar
  8. 8.
    Inoue, K., Hasegawa, T., Araki, Y., Mori, N., Matsumoto, K.: Adaptive control of parameter-less population pyramid on the local distribution of inferior individuals. In: Genetic and Evolutionary Computation Conference, pp. 863–870. ACM (2015)Google Scholar
  9. 9.
    Ochoa, G., Chicano, F., Tinós, R., Whitley, D.: Tunnelling crossover networks. In: Genetic and Evolutionary Computation Conference, pp. 449–456. ACM (2015)Google Scholar
  10. 10.
    Saul, L., Kardar, M.: The 2d\(\pm \)j ising spin glass: exact partition functions in polynomial time. Nucl. Phys. B 432(3), 641–667 (1994)CrossRefGoogle Scholar
  11. 11.
    Selman, B., Mitchell, D.G., Levesque, H.J.: Generating hard satisfiability problems. Artif. Intell. 81(1–2), 17–29 (1996)MathSciNetCrossRefGoogle Scholar
  12. 12.
    Tintos, R., Whitley, D., Chicano, F.: Partition crossover for pseudo-boolean optimization. In: Foundations of Genetic Algorithms Conference, pp. 137–149. ACM (2015)Google Scholar
  13. 13.
    Tomassini, M., Verel, S., Ochoa, G.: Complex-network analysis of combinatorial spaces: the n k landscape case. Phys. Rev. E 78(6), 066114 (2008)CrossRefGoogle Scholar
  14. 14.
    Verel, S., Ochoa, G., Tomassini, M.: Local optima networks of nk landscapes with neutrality. IEEE Trans. Evol. Comput. 15(6), 783–797 (2011)CrossRefGoogle Scholar
  15. 15.
    Whitley, D.: Mk landscapes, nk landscapes, max-ksat: a proof that the only challenging problems are deceptive. In: Genetic and Evolutionary Computation Conference, pp. 927–934. ACM (2015)Google Scholar
  16. 16.
    Whitley, D., Chen, W.: Constant time steepest descent local search with lookahead for nk-landscapes and max-ksat. In: Genetic and Evolutionary Computation Conference, pp. 1357–1364. ACM (2012)Google Scholar
  17. 17.
    Wright, A.H., Thompson, R.K., Zhang, J.: The computational complexity of N-K fitness functions. IEEE Trans. Evol. Comput. 4(4), 373–379 (2000)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  1. 1.BEACON Center for the Study of Evolution in ActionMichigan State UniversityEast LansingUSA

Personalised recommendations