Abstract
Recent advances in drift analysis have given us better and better tools for understanding random processes, including the run time of randomized search heuristics. In the setting of multiplicative drift we do not only have excellent bounds on the expected run time, but also more general results showing the strong concentration of the run time. In this paper we investigate the setting of additive drift under the assumption of strong concentration of the “step size” of the process. Under sufficiently strong drift towards the goal we show a strong concentration of the hitting time. In contrast to this, we show that in the presence of small drift a Gambler’s-Ruin-like behavior of the process overrides the influence of the drift, leading to a maximal movement of about \(\sqrt{t}\) steps within t iterations. Finally, in the presence of sufficiently strong negative drift the hitting time is superpolynomial with high probability; this corresponds to the well-known Negative Drift Theorem.
Similar content being viewed by others
Notes
Note that iterating this idea leads to an example where, under arbitrary additive drift, the expected number of iterations until n is reached is 2, seemingly contradicting the Additive Drift Theorem; however, this iterated example requires an unbounded search space, which is ruled out by the requirements of the Additive Drift Theorem.
In fact, the term sub-Gaussian originally entailed \(\delta = \infty \), while the version with finite \(\delta \) is called locally sub-Gaussian [3]. Furthermore, this terminology is usually applied to single random variables, not to martingales.
Note that Eq. (2) is typically required to also hold for negative \(z \ge -\delta \), in which case even stronger statements can be made. However, we want to keep the scope of these definitions wide.
References
Azuma, K.: Weighted sums of certain dependent random variables. Tohoku Math. J. 19, 357–367 (1967)
Buldygin, V.V., Kozachenko, Y.V.: Sub-Gaussian random variables. Ukr. Math. J. 32, 483–489 (1980)
Chareka, P., Chareka, O., Kennedy, S.: Locally sub-Gaussian random variables and the strong law of large numbers. Atl. Electron. J. Math. 1, 75–81 (2006)
Doerr, B., Goldberg, L.A.: Adaptive drift analysis. Algorithmica 65, 224–250 (2013)
Doerr, B., Johannsen, D., Winzen, C.: Multiplicative drift analysis. Algorithmica 64, 673–697 (2012)
Doerr, B., Künnemann, M.: How the (1+\(\lambda \)) evolutionary algorithm optimizes linear functions. In: Proceedings of GECCO (Genetic and Evolutionary Computation Conference), pp. 1589–1596 (2013)
Dubhashi, D.P., Panconesi, A.: Concentration of Measure for the Analysis of Randomized Algorithms. Cambridge University Press, Cambridge (2009)
Fan, X., Grama, I., Liu, Q.: Hoeffding’s inequality for supermartingales. Stoch. Process. Appl. 122, 3545–3559 (2012)
Fan, X., Grama, I., Liu, Q.: Exponential inequalities for martingales with applications. Electron. J. Probab. 20, 1–22 (2015)
Hajek, B.: Hitting-time and occupation-time bounds implied by drift analysis with applications. Adv. Appl. Probab. 13, 502–525 (1982)
He, J., Yao, X.: A study of drift analysis for estimating computation time of evolutionary algorithms. Nat. Comput. 3, 21–35 (2004)
Johannsen, D.: Random Combinatorial Structures and Randomized Search Heuristics. Ph.D. thesis, Universität des Saarlandes. Available online at http://scidok.sulb.uni-saarland.de/volltexte/2011/3529/pdf/Dissertation_3166_Joha_Dani_2010 (2010)
Kötzing, T.: Concentration of first hitting times under additive drift. In: Proceedings of GECCO (Genetic and Evolutionary Computation Conference)
Lehre, P.K., Witt, C.: Concentrated hitting times of randomized search heuristics with variable drift. In: Proceedings of ISAAC (International Symposium on Algorithms and Computation), pp. 686–697 (2014)
McDiarmid, C.: Concentration. In: Habib, M., McDiarmid, C., Ramirez-Alfonsin, J., Reed, B. (eds.) Probabilistic Methods for Algorithmic Discrete Mathematics. Algorithms and Combinatorics, Vol. 16, pp. 195–248. Springer, Berlin, Heidelberg (1998)
Mitavskiy, B., Rowe, J.E., Cannings, C.: Preliminary theoretical analysis of a local search algorithm to optimize network communication subject to preserving the total number of links. In: Proceedings of CEC (Congress on Evolutionary Computation), pp. 1484–1491 (2008)
Oliveto, P.S., Witt, C.: Simplified drift analysis for proving lower bounds in evolutionary computation. Algorithmica 59, 369–386 (2011)
Oliveto, P.S., Witt, C.: Erratum: simplified drift analysis for proving lower bounds in evolutionary computation. 1211.7184 (2012)
Oliveto, P.S., Witt, C.: On the runtime analysis of the simple genetic algorithm. Theor. Comput. Sci. 545, 2–19 (2014)
Rivasplata, O.: Subgaussian random variables: an expository note. http://www.math.ualberta.ca/~orivasplata/publications/subgaussians (2012)
Rowe, J.E., Sudholt, D.: The choice of the offspring population size in the (1, \(\lambda \)) EA. In: Proceedings of GECCO (Genetic and Evolutionary Computation Conference), pp. 1349–1356 (2012)
Wormald, N.C.: The differential equation method for random graph processes and greedy algorithms. In: Lectures on Approximation and Randomized Algorithms, pp. 73–155 (1999)
Acknowledgments
I would like to thank Tobias Friedrich, Benjamin Doerr and Anton Krohmer for many useful discussions on the topic of this paper; all of them also provided valuable pointers to the literature. A further source of pointers to the literature as well as many helpful comments came from the reviewers of the conference version this paper, who have proved to be very knowledgeable in this subject. Most importantly, Carsten Witt pointed me to possible extensions of the Azuma–Hoeffding Inequality, as well as to some computations in his own publications, which helped to extend this paper to cases of non-bounded variables, making the contribution significantly more valuable; this paper would have a much more restricted scope without his advice. Finally, I would like to thank the reviewers of the journal version of this paper; they pointed out places for correction and improvement, as well as more useful literature.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Kötzing, T. Concentration of First Hitting Times Under Additive Drift. Algorithmica 75, 490–506 (2016). https://doi.org/10.1007/s00453-015-0048-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00453-015-0048-0