SCIENCE CHINA Information Sciences, Volume 62, Issue 10: 202101(2019) https://doi.org/10.1007/s11432-018-9852-8

Wave models and dynamical analysis of evolutionary algorithms

More info
  • ReceivedOct 18, 2018
  • AcceptedMar 29, 2019
  • PublishedSep 3, 2019


By drawing an analogy between the population of an evolutionary algorithm and a gas system (which we call a particle system), we first build wave models of evolutionary algorithms based on aerodynamics theory. Then, we solve the models' linear and quasi-linear hyperbolic equations analytically, yielding wave solutions. These describe the propagation of the particle density wave, which is composed of leftward and rightward waves. We demonstrate the convergence of evolutionary algorithms by analyzing the mechanism underlying the leftward wave, and investigate population diversity by analyzing the rightward wave. To confirm these theoretical results, we conduct experiments that apply three typical evolutionary algorithms to common benchmark problems, showing that the experimental and theoretical results agree. These theoretical and experimental analyses also provide several new clues and ideas that may assist in the design and improvement of evolutionary algorithms.


This work was supported by National Natural Science Foundation of China (Grant No. 61672391).


[1] Back T. Evolutionary algorithms in theory and practice: evolution strategies, evolutionary programming, genetic algorithms. Oxford: Oxford University Press, 1996. Google Scholar

[2] Wang F, Zhang H, Li K. A hybrid particle swarm optimization algorithm using adaptive learning strategy. Inf Sci, 2018, 436-437: 162-177 CrossRef Google Scholar

[3] Guo S M, Yang C C. Enhancing Differential Evolution Utilizing Eigenvector-Based Crossover Operator. IEEE Trans Evol Computat, 2015, 19: 31-49 CrossRef Google Scholar

[4] Wegener I. Methods for the analysis of evolutionary algorithms on pseudo-boolean functions. Evolutionary optimization. Boston: Springer, 2003: 349-369 DOI: 10.17877/DE290R-15272. Google Scholar

[5] Beyer H G. Convergence analysis of evolutionary algorithms that are based on the paradigm of information geometry.. Evolary Computation, 2014, 22: 679-709 CrossRef PubMed Google Scholar

[6] Derrac J, García S, Hui S. Analyzing convergence performance of evolutionary algorithms: A statistical approach. Inf Sci, 2014, 289: 41-58 CrossRef Google Scholar

[7] Tan C J, Neoh S C, Lim C P. Application of an evolutionary algorithm-based ensemble model to job-shop scheduling. J Intell Manuf, 2019, 30: 879-890 CrossRef Google Scholar

[8] Wu H, Kuang L, Wang F. A multiobjective box-covering algorithm for fractal modularity on complex networks. Appl Soft Computing, 2017, 61: 294-313 CrossRef Google Scholar

[9] Goldberg D E, Segrest P. Finite Markov chain analysis of genetic algorithms. In: Proceedings of the 2nd International Conference on Genetic Algorithms, Cambridge, 1987. 1: 1. Google Scholar

[10] Rudolph G. Finite Markov chain results in evolutionary computation: A tour d'horizon. Fund Inform, 1998, 35: 67--89. Google Scholar

[11] He J, Yao X. Drift analysis and average time complexity of evolutionary algorithms. Artificial Intelligence, 2001, 127: 57-85 CrossRef Google Scholar

[12] Sudholt D. A New Method for Lower Bounds on the Running Time of Evolutionary Algorithms. IEEE Trans Evol Computat, 2013, 17: 418-435 CrossRef Google Scholar

[13] Yu Y, Qian C, Zhou Z H. Switch Analysis for Running Time Analysis of Evolutionary Algorithms. IEEE Trans Evol Computat, 2015, 19: 777-792 CrossRef Google Scholar

[14] Bian C, Qian C, Tang K. A general approach to running time analysis of multi-objective evolutionary algorithms. In: Proceedings of 27th International Joint Conference on Artificial Intelligence (IJCAI), Stockholm, 2018. 1405--1411. Google Scholar

[15] Mori N, Yoshida J, Tamaki H, et al. A thermodynamical selection rule for the genetic algorithm. In: Proceedings of IEEE International Conference on Evolutionary Computation, Perth, 1995. 1: 188. Google Scholar

[16] Cornforth T W, Lipson H. A hybrid evolutionary algorithm for the symbolic modeling of multiple-time-scale dynamical systems. Evol Intel, 2015, 8: 149-164 CrossRef Google Scholar

[17] Li Y X, Zou X F, Kang L S. A new dynamical evolutionary algorithm based on statistical mechanics. J Comput Sci Technol, 2003, 18: 361-368 CrossRef Google Scholar

[18] Li Y X, Xiang Z L, Xia J N. Dynamical system models and convergence analysis for simulated annealing algorithm. Chinese Journal of Computers, 2018, 1-13. http://kns.cnki.net/kcms/detail/11.1826.TP.20181114.1046.002.html. Google Scholar

[19] Li Y X, Xiang Z L, Zhang W Y. A relaxation model and time complexity analysis for simulated annealing algorithm. Chinese Journal of Computers, 2018. to be published. Google Scholar

[20] Zhou Y L. One-Dimensional Unsteady Hydrodynamics. Beijing: Science China Press, 1998. Google Scholar

[21] Lamb H. Hydrodynamics. Cambridge: Cambridge University Press, 1993. Google Scholar

[22] Gu C H, Li D Q. Mathematical Physics Equations. Beijing: People's Education Press, 1982. Google Scholar

[23] Zhang Y. Expansion Waves and Shock Waves. Beijing: Peking University Press, 1983. Google Scholar

[24] Shi Y, Eberhart R C. Empirical study of particle swarm optimization. In: Proceedings of IEEE International Conference on Evolutionary Computation, Washington, 1999. 3: 1945--1950. Google Scholar

[25] Liang J J, Qu B Y, Suganthan P N. Problem Definitions and Evaluation Criteria for the CEC 2014 Special Session and Competition on Single Objective Real-Parameter Numerical Optimization. Zhengzhou University and Nanyang Technological University, Technical Report. 2013. Google Scholar

[26] ?repin?ek M, Liu S H, Mernik M. Exploration and exploitation in evolutionary algorithms. ACM Comput Surv, 2013, 45: 1-33 CrossRef Google Scholar

[27] Liu S H, Mernik M, Bryant B R. To explore or to exploit: An entropy-driven approach for evolutionary algorithms. Int J Knowledge-based Intell Eng Syst, 2009, 13: 185-206 CrossRef Google Scholar

[28] Tang K, Yang P, Yao X. Negatively Correlated Search. IEEE J Sel Areas Commun, 2016, 34: 542-550 CrossRef Google Scholar

[29] Ursem R K. Diversity-guided evolutionary algorithms. In: Proceedings of International Conference on Parallel Problem Solving from Nature, Berlin, 2002. 462--471. Google Scholar

  • Figure 1

    (Color online) Illustration showing (a) the rightward linear wave and (b) clusters of characteristic lines.

  • Figure 2

    (Color online) Wave propagation for $f(x)$, showing the (a) particle density distribution, (b) volume elasticity coefficient $\kappa$, and (c) wave propagation speed $a$ over time.

  • Figure 3

    (Color online) (a) ${\rm~TSP}(n=50)$; (b) ${\rm~TSP}(n=100)$.

  • Figure 4

    (Color online) PSO applied to (a) $f_1$ and (b) $f_2$.

  • Figure 5

    (Color online) DE applied to (a) $f_1$ and (b) $f_2$.

  • Figure 6

    (Color online) PSO vs. DE, applied to (a) $f_1$ and (b) $f_2$.

  • Figure 7

    (Color online) $a$ vs. $a_{sw}$, for (a) $f_1$ and (b) $f_2$.

  • Table 1   Test functions used for PSO and DE
    Test function $D$ Search range $f_{\rm~min}$ Name
    $f_1(y)=\sum_{i=1}^{D-1}(100(y_{i}^2-y_{i+1}^2)+(y_i-1)^2)+F_{4}^*$,30 $[-100,100]^{D}$400Shifted and Rotated
    where $y=M(\frac{2.048z}{100})+1$, $M$ is a rotation matrix,Rosenbrock [25]
    $z=x-o$, and $o=[o_1,o_2,\ldots,o_n]$ is the shifted
    global optimum.
    $f_2(y)=\sum_{i=1}^D\frac{y_{i}^2}{4000}-\prod_{i=1}^D{\rm~cos}(\frac{y_i}{\sqrt~i})+1+F_{7}^*$,30$[-100,100]^{D}$700 Shifted and Rotated
    where $y=M(\frac{600z}{100})$, $M$ is a rotation matrix,Griewank [25]
    $z=x-o$, and $o=[o_1,o_2,\ldots,o_n]$ is the shifted
    global optimum.

Copyright 2020 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有

京ICP备18024590号-1       京公网安备11010102003388号