Abstract
The idea to recombine two or more search points into a new solution is one of the main design principles of evolutionary computation (EC). Its usefulness in the combinatorial optimization context, however, is subject to a highly controversial discussion between EC practitioners and the broader Computer Science research community. While the former, naturally, report significant speedups procured by crossover, the belief that sexual reproduction cannot advance the search for high-quality solutions seems common, for example, amongst theoretical computer scientists. Examples that help understand the role of crossover in combinatorial optimization are needed to promote an intensified discussion on this subject.
We contribute with this work an example of a crossover-based genetic algorithm (GA) that provably outperforms any mutation-based black-box heuristic on a classic benchmark problem. The appeal of our examples lies in its simplicity: the proof of the result uses standard mathematical techniques and can be taught in a basic algorithms lecture.
Our theoretical result is complemented by an empirical evaluation, which demonstrates that the superiority of the GA holds already for quite small problem instances.
Keywords
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
See, for example, here: https://www.simonsfoundation.org/2010/05/18/why-sex/.
- 2.
See Sect. 3 for a discussion of the fact that sampling the parent without replacement improves the expected optimization time of this algorithm on OneMax. We do not apply this modified parent selection rule in the greedy \((\mu +1)\) GA\(_{\text {mod}}\) to highlight that the main improvement stems from the modified mutation step.
References
Pinto, E.C., Doerr, C.: Discussion of a more practice-aware runtime analysis for evolutionary algorithms. In: EA 2017, pp. 298–305 (2017)
http://www-desir.lip6.fr/~doerr/CarvalhoDoerr-PPSN18-Crossover.pdf
Corus, D., Oliveto, P.S.: Standard steady state genetic algorithms can hillclimb faster than mutation-only evolutionary algorithms. IEEE Trans. Evol. Comput. (2018, to appear)
Doerr, B., Doerr, C.: The impact of random initialization on the runtime of randomized search heuristics. Algorithmica 75, 529–553 (2016)
Doerr, B., Doerr, C.: Optimal static and self-adjusting parameter choices for the \((1+(\lambda,\lambda ))\) genetic algorithm. Algorithmica 80, 1658–1709 (2018)
Doerr, B., Doerr, C., Ebel, F.: From black-box complexity to designing new genetic algorithms. Theor. Comput. Sci. 567, 87–104 (2015)
Doerr, B., Doerr, C., Yang, J.: Optimal parameter choices via precise black-box analysis. In: GECCO 2016, pp. 1123–1130. ACM (2016)
Doerr, B., Happ, E., Klein, C.: Crossover can provably be useful in evolutionary computation. Theor. Comput. Sci. 425, 17–33 (2012)
Doerr, B., Johannsen, D., Kötzing, T., Lehre, P.K., Wagner, M., Winzen, C.: Faster black-box algorithms through higher arity operators. In: FOGA 2011, pp. 163–172. ACM (2011)
Goldman, B.W., Punch, W.F.: Fast and efficient black box optimization using the parameter-less population pyramid. Evol. Comput. 23, 451–479 (2015). https://github.com/brianwgoldman?tab=repositories
Jansen, T., Wegener, I.: The analysis of evolutionary algorithms - a proof that crossover really can help. Algorithmica 34, 47–66 (2002)
Jansen, T., Zarges, C.: Analysis of evolutionary algorithms: from computational complexity analysis to algorithm engineering. In: FOGA 2011, pp. 1–14. ACM (2011)
Lehre, P.K., Witt, C.: Black-box search by unbiased variation. Algorithmica 64, 623–642 (2012)
Sudholt, D.: Crossover speeds up building-block assembly. In: GECCO 2012, pp. 689–702. ACM (2012)
Sudholt, D.: A new method for lower bounds on the running time of evolutionary algorithms. IEEE Trans. Evol. Comput. 17, 418–435 (2013)
Sudholt, D.: How crossover speeds up building block assembly in genetic algorithms. Evol. Comput. 25, 237–274 (2017)
Witt, C.: Tight bounds on the optimization time of a randomized search heuristic on linear functions. Comb. Probab. Comput. 22, 294–318 (2013)
Acknowlegement
We thank the anonymous reviewers of this paper for their constructive feedback, which has helped us to improve the presentation of our main result. This research benefited from the support of the FMJH Program Gaspard Monge in optimization and operation research, and from the support to this program from EDF.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer Nature Switzerland AG
About this paper
Cite this paper
Pinto, E.C., Doerr, C. (2018). A Simple Proof for the Usefulness of Crossover in Black-Box Optimization. In: Auger, A., Fonseca, C., Lourenço, N., Machado, P., Paquete, L., Whitley, D. (eds) Parallel Problem Solving from Nature – PPSN XV. PPSN 2018. Lecture Notes in Computer Science(), vol 11102. Springer, Cham. https://doi.org/10.1007/978-3-319-99259-4_3
Download citation
DOI: https://doi.org/10.1007/978-3-319-99259-4_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-99258-7
Online ISBN: 978-3-319-99259-4
eBook Packages: Computer ScienceComputer Science (R0)