Average Drift Analysis and Population Scalability

This paper aims to study how the population size affects the computation time of evolutionary algorithms (EAs) in a rigorous way. The computation time of EAs can be measured by either the number of generations (hitting time) or the number of fitness evaluations (running time) to find an optimal solution. Population scalability is the ratio of the expected hitting time between a benchmark algorithm and an algorithm using a larger population size. Average drift analysis is introduced to compare the expected hitting time of two algorithms and to estimate lower and upper bounds on the population scalability. Several intuitive beliefs are rigorously analyzed. It is proven that: 1) using a population sometimes increases rather than decreases the expected hitting time; 2) using a population cannot shorten the expected running time of any elitist EA on any unimodal function on the time-fitness landscape, however, this statement is not true in terms of the distance-based fitness landscape; and 3) using a population cannot always reduce the expected running time on deceptive functions, which depends on whether the benchmark algorithm uses elitist selection or random selection.

[1]  Pietro Simone Oliveto,et al.  Improved runtime analysis of the simple genetic algorithm , 2013, GECCO '13.

[2]  Jens Jägersküpper,et al.  Combining Markov-Chain Analysis and Drift Analysis , 2011, Algorithmica.

[3]  John N. Tsitsiklis,et al.  Introduction to Probability , 2002 .

[4]  John R. Koza,et al.  A Parallel Implementation of Genetic Programming that Achieves Super-Linear Performance , 1998, Inf. Sci..

[5]  Marvin Künnemann,et al.  Royal road functions and the (1 + λ) evolutionary algorithm: Almost no speed-up from larger offspring populations , 2013, 2013 IEEE Congress on Evolutionary Computation.

[6]  Lishan Kang,et al.  Convergence of Genetic Evolution Algorithms for Optimization , 1995, Parallel Algorithms Appl..

[7]  Enrique Alba,et al.  Parallelism and evolutionary algorithms , 2002, IEEE Trans. Evol. Comput..

[8]  Adam Prügel-Bennett,et al.  Benefits of a Population: Five Mechanisms That Advantage Population-Based Algorithms , 2010, IEEE Transactions on Evolutionary Computation.

[9]  Xin Yao,et al.  From an individual to a population: an analysis of the first hitting time of population-based evolutionary algorithms , 2002, IEEE Trans. Evol. Comput..

[10]  Yuren Zhou,et al.  A Runtime Analysis of Evolutionary Algorithms for Constrained Optimization Problems , 2007, IEEE Transactions on Evolutionary Computation.

[11]  Dirk Sudholt,et al.  Analysis of different MMAS ACO algorithms on unimodal functions and plateaus , 2009, Swarm Intelligence.

[12]  Heinz Mühlenbein,et al.  The Science of Breeding and Its Application to the Breeder Genetic Algorithm (BGA) , 1993, Evolutionary Computation.

[13]  Jun He.  A note on the first hitting time of (1 + λ) evolutionary algorithm for linear functions with boolean inputs , 2010, IEEE Congress on Evolutionary Computation.

[14]  Rajeev Motwani,et al.  Randomized Algorithms , 1995, SIGA.

[15]  Zbigniew Michalewicz,et al.  GAVaPS-a genetic algorithm with varying population size , 1994, Proceedings of the First IEEE Conference on Evolutionary Computation. IEEE World Congress on Computational Intelligence.

[16]  Xin Yao,et al.  A New Approach for Analyzing Average Time Complexity of Population-Based Evolutionary Algorithms on Unimodal Problems , 2009, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[17]  Xin Yao,et al.  Analysis of Scalable Parallel Evolutionary Algorithms , 2006, 2006 IEEE International Conference on Evolutionary Computation.

[18]  Xin Yao,et al.  Towards an analytic framework for analysing the computation time of evolutionary algorithms , 2003, Artif. Intell..

[19]  Jonathan E. Rowe,et al.  Theoretical analysis of local search strategies to optimize network communication subject to preserving the total number of links , 2009, Int. J. Intell. Comput. Cybern..

[20]  Xin Yao,et al.  On the Easiest and Hardest Fitness Functions , 2012, IEEE Transactions on Evolutionary Computation.

[21]  Yang Yu,et al.  A new approach to estimating the expected first hitting time of evolutionary algorithms , 2006, Artif. Intell..

[22]  Dirk Sudholt,et al.  Adaptive population models for offspring populations and parallel evolutionary algorithms , 2011, FOGA '11.

[23]  Jens Jägersküpper,et al.  A Blend of Markov-Chain and Drift Analysis , 2008, PPSN.

[24]  Pietro Simone Oliveto,et al.  On the runtime analysis of the Simple Genetic Algorithm , 2014, Theor. Comput. Sci..

[25]  Per Kristian Lehre,et al.  Negative Drift in Populations , 2010, PPSN.

[26]  Günter Rudolph,et al.  Finite Markov Chain Results in Evolutionary Computation: A Tour d'Horizon , 1998, Fundam. Informaticae.

[27]  David E. Goldberg,et al.  The Gambler's Ruin Problem, Genetic Algorithms, and the Sizing of Populations , 1999, Evolutionary Computation.

[28]  Carsten Witt,et al.  Population size versus runtime of a simple evolutionary algorithm , 2008, Theor. Comput. Sci..

[29]  Xin Yao,et al.  Drift analysis and average time complexity of evolutionary algorithms , 2001, Artif. Intell..

[30]  Pietro Simone Oliveto,et al.  Analysis of Diversity-Preserving Mechanisms for Global Exploration , 2009, Evolutionary Computation.

[31]  Tobias Storch,et al.  On the Choice of the Parent Population Size , 2008, Evolutionary Computation.

[32]  Pietro Simone Oliveto,et al.  Analysis of population-based evolutionary algorithms for the vertex cover problem , 2008, 2008 IEEE Congress on Evolutionary Computation (IEEE World Congress on Computational Intelligence).

[33]  Kalyanmoy Deb,et al.  Genetic Algorithms, Noise, and the Sizing of Populations , 1992, Complex Syst..

[34]  Daniel Johannsen,et al.  Random combinatorial structures and randomized search heuristics , 2010 .

[35]  Günter Rudolph,et al.  Convergence of evolutionary algorithms on the n-dimensional continuous space , 2013, IEEE Transactions on Cybernetics.

[36]  Ingo Wegener,et al.  On the Optimization of Monotone Polynomials by Simple Randomized Search Heuristics , 2005, Combinatorics, Probability and Computing.

[37]  Kenneth A. De Jong,et al.  Design and Management of Complex Technical Processes and Systems by Means of Computational Intelligence Methods on the Choice of the Offspring Population Size in Evolutionary Algorithms on the Choice of the Offspring Population Size in Evolutionary Algorithms , 2004 .

[38]  Zbigniew Michalewicz,et al.  Parameter Control in Evolutionary Algorithms , 2007, Parameter Setting in Evolutionary Algorithms.

[39]  Benjamin Doerr,et al.  Multiplicative drift analysis , 2010, GECCO.

[40]  Xin Yao,et al.  A study of drift analysis for estimating computation time of evolutionary algorithms , 2004, Natural Computing.

[41]  Jens Jägersküpper,et al.  Rigorous runtime analysis of a (μ+1)ES for the sphere function , 2005, GECCO '05.

[42]  Marvin Künnemann,et al.  How the (1+λ) evolutionary algorithm optimizes linear functions , 2013, GECCO '13.

[43]  B. Hajek.  Hitting-time and occupation-time bounds implied by drift analysis with applications , 1982, Advances in Applied Probability.

[44]  Xin Yao,et al.  Erratum to: Drift analysis and average time complexity of evolutionary algorithms [Artificial Intelligence 127 (2001) 57-85] , 2002, Artif. Intell..

[45]  Carsten Witt,et al.  Population Size vs. Mutation Strength for the (1+λ) EA on OneMax , 2015, GECCO.

[46]  Thomas Jansen,et al.  On the analysis of the (1+1) evolutionary algorithm , 2002, Theor. Comput. Sci..

[47]  Carsten Witt,et al.  Optimizing Linear Functions with Randomized Search Heuristics - The Robustness of Mutation , 2012, STACS.

[48]  Charles M. Grinstead,et al.  Introduction to probability , 1999, Statistics for the Behavioural Sciences.

[49]  Enrique Alba,et al.  Parallel evolutionary algorithms can achieve super-linear performance , 2002, Inf. Process. Lett..

[50]  Dirk Sudholt,et al.  The choice of the offspring population size in the (1,λ) EA , 2012, GECCO '12.

[51]  Leslie Ann Goldberg,et al.  Adaptive Drift Analysis , 2010, PPSN.

[52]  E. Cantu-Paz,et al.  The Gambler's Ruin Problem, Genetic Algorithms, and the Sizing of Populations , 1997, Evolutionary Computation.