Dirk Sudholt
ABSTRACT
We re-investigate a fundamental question: how effective is crossover in combining building blocks? Although this has been discussed controversially for decades, we are still lacking a rigorous and intuitive answer. We provide such answers for royal road functions and OneMax, where every bit is a building block. For the latter we prove that a simple GA with uniform crossover is twice as fast as the fastest EA using only standard bit mutation, up to small-order terms. The reason is that crossover effectively turns neutral mutations into improvements by combining the right building blocks at a later stage. Compared to mutation-based EAs, this makes multi-bit mutations more useful. Introducing crossover changes the optimal mutation rate on OneMax from 1/n to (1+5)/2 Å 1/n H 1.618/n. Similar results are proved for k-point crossover. Experiments and statistical tests confirm that our findings apply to a broad class of building-block functions.
- M. Abramowitz and I. A. Stegun. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Dover, New York, ninth Dover printing, tenth GPO printing edition, 1964.Google Scholar
- M. Dietzfelbinger, B. Naudts, C. Van Hoyweghen, and I. Wegener. The analysis of a recombinative hill-climber on H-IFF. IEEE Transactions on Evolutionary Computation, 7(5):417--423, 2003. Google ScholarDigital Library
- B. Doerr, E. Happ, and C. Klein. Crossover can provably be useful in evolutionary computation. Theoretical Computer Science, 425(0):17--33, 2012. Google ScholarDigital Library
- S. Droste, T. Jansen, and I. Wegener. On the analysis of the (1+1) evolutionary algorithm. Theoretical Computer Science, 276:51--81, 2002. Google ScholarDigital Library
- S. Fischer and I. Wegener. The one-dimensional Ising model: Mutation versus recombination. Theoretical Computer Science, 344(2--3):208--225, 2005. Google ScholarDigital Library
- S. Forrest and M. Mitchell. Relative building block fitness and the building block hypotheses. In Proc. of FOGA 2, pages 198--226. Morgan Kaufmann, 1993.Google Scholar
- T. Jansen, K. A. De Jong, and I. Wegener. On the choice of the offspring population size in evolutionary algorithms. Evolutionary Computation, 13:413--440, 2005. Google ScholarDigital Library
- T. Jansen and I. Wegener. On the choice of the mutation probability for the (1+1) EA. In PPSN '00, volume 1917, pages 89--98. Springer, 2000. Google ScholarDigital Library
- T. Jansen and I. Wegener. On the analysis of evolutionary algorithms--a proof that crossover really can help. Algorithmica, 34(1):47--66, 2002.Google ScholarDigital Library
- T. Jansen and I. Wegener. Real royal road functions--where crossover provably is essential. Discrete Applied Mathematics, 149:111--125, 2005. Google ScholarDigital Library
- T. Jansen and C. Zarges. Analysis of evolutionary algorithms: from computational complexity analysis to algorithm engineering. In Proc. of FOGA '11, pages 1--14. ACM, 2011. Google ScholarDigital Library
- T. Kötzing, D. Sudholt, and M. Theile. How crossover helps in pseudo-Boolean optimization. In Proc. of GECCO '11, pages 989--996. ACM Press, 2011. Google ScholarDigital Library
- J. Lassig and D. Sudholt. General scheme for analyzing running times of parallel evolutionary algorithms. In PPSN '10, pages 234--243. Springer, 2010. Google ScholarDigital Library
- M. Mitchell, S. Forrest, and J. H. Holland. The royal road function for genetic algorithms: fitness landscapes and GA performance. In Proc. of the 1st European Conference on Artificial Life, pages 245--254. MIT Press, 1992.Google Scholar
- M. Mitchell, J. H. Holland, and S. Forrest. When will a genetic algorithm outperform hill climbing? In Advances in Neural Information Processing Systems, pages 51--58. Morgan Kaufmann, 1994.Google Scholar
- F. Neumann, P. S. Oliveto, and C. Witt. Theoretical analysis of fitness-proportional selection: landscapes and efficiency. In Proc. of GECCO '09, pages 835--842. ACM Press, 2009. Google ScholarDigital Library
- F. Neumann and M. Theile. How crossover speeds up evolutionary algorithms for the multi-criteria all-pairs-shortest-path problem. In Proc. of PPSN '10, pages 667--676. Springer, 2010. Google ScholarDigital Library
- T. Storch and I. Wegener. Real royal road functions for constant population size. Theoretical Computer Science, 320:123--134, 2004. Google ScholarDigital Library
- D. Sudholt. Crossover is provably essential for the Ising model on trees. In Proc. of GECCO '05, pages 1161--1167. ACM Press, 2005. Google ScholarDigital Library
- D. Sudholt. General lower bounds for the running time of evolutionary algorithms. In PPSN '10, pages 124--133. Springer, 2010. Google ScholarDigital Library
- D. Sudholt. A new method for lower bounds on the running time of evolutionary algorithms. ArXiv e-prints, 2011. Available from http://arxiv.org/abs/1109.1504.Google Scholar
- D. Sudholt and C. Thyssen. Running time analysis of ant colony optimization for shortest path problems. Journal of Discrete Algorithms, 2012. To appear. Google ScholarDigital Library
- R. A. Watson and T. Jansen. A building-block royal road where crossover is provably essential. In Proc. of GECCO '07, pages 1452--1459. ACM, 2007. Google ScholarDigital Library
- I. Wegener and C. Witt. On the optimization of monotone polynomials by simple randomized search heuristics. Combinatorics, Probability and Computing, 14:225--247, 2005. Google ScholarDigital Library
- C. Witt. Runtime analysis of the (¼+1) EA on simple pseudo-Boolean functions. Evolutionary Computation, 14(1):65--86, 2006. Google ScholarDigital Library
- C. Witt. Optimizing Linear Functions with Randomized Search Heuristics - The Robustness of Mutation. In Proc. of STACS '12, pages 420--431, 2012.Google Scholar
Index Terms
- Crossover speeds up building-block assembly
Recommendations
How crossover helps in pseudo-boolean optimization
GECCO '11: Proceedings of the 13th annual conference on Genetic and evolutionary computationUnderstanding the impact of crossover on performance is a major problem in the theory of genetic algorithms (GAs). We present new insight on working principles of crossover by analyzing the performance of crossover-based GAs on the simple functions ...
How crossover speeds up building block assembly in genetic algorithms
We reinvestigate a fundamental question: How effective is crossover in genetic algorithms in combining building blocks of good solutions? Although this has been discussed controversially for decades, we are still lacking a rigorous and intuitive answer. ...
Fast mutation in crossover-based algorithms
GECCO '20: Proceedings of the 2020 Genetic and Evolutionary Computation ConferenceThe heavy-tailed mutation operator proposed in Doerr et al. (GECCO 2017), called fast mutation to agree with the previously used language, so far was successfully used only in purely mutation-based algorithms. There, it can relieve the algorithm ...
Comments