Time and Size Limited Harvesting Models of Genetic Algorithm

  • Subbiah Baskaran
  • David Noever
Conference paper


In this paper we formulate and investigate a novel model of a Genetic Algorithm (GA) in which the genetic population is allowed to grow with a delay in selection. And during selection, the excess growth over a preset constant size is harvested. Two possible delay modes result in two harvesting schemes called time and size limited harvesting. The two schemes generalize the standard genetic algorithm in the direction of treating population size as a stochastic parameter. If the delay threshold is one, then both schemes reduce to the standard genetic algorithm. The retention of low fitness members for extended period in the evolving population promotes preservation of schema pathways which enable escape from local optima and also help alleviate premature convergence. The extended model is successfully applied to a difficult two-dimensional non-stationary problem for tracking time-varying optima in real time.


Genetic Algorithm Fitness Landscape Excess Growth Standard Genetic Algorithm Delay Mode 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Holland, J.H: Adaptation in Natural and Artificial Systems, Ann Arbor, University of Michigan Press, 1975.Google Scholar
  2. [2]
    Watson, J.D et. al.,: Recombinant DNA, Second Edition, W.H. Freeman and Company, NY, 1992.Google Scholar
  3. [3]
    Noever, D., Baskaran, S., Schuster, P.: Understanding Genetic Algorithm Dynamics using harvesting Strategies, Physica D, Vol. 72, pp. 456–435, 1995.Google Scholar
  4. [4]
    Smith, R.E., Goldberg, D.E.: Diploidy and dominance in artificial genetic search, Complex Systems, Vol. 6, pp. 21–28, 1992.MathSciNetGoogle Scholar
  5. [5]
    Gobb, H.G., Grefenstette, J.J.: Genetic Algorithms for Tracking Changing Environments, Proc. of 5th ICGA, Vol. 5, pp. 523–529, 1993.Google Scholar
  6. [6]
    Mori, K., Tsukiyama, M., Fukuda, T.: Immune Algorithm with Searching Diversity and its Application to Resource Allocation Problem, Trans. IEE of Japan, Vol. 113-C, No. 10, pp. 872–878, 1993.Google Scholar
  7. [7]
    Mori, N., Kita, H., Nishikawa, Y.: Application to a changing environment by means of the Thermodynamical Genetic Algorithm, Proc. of. 4th PPSN (PPSN′96), Vol. 4, pp. 513–522, 1996.Google Scholar
  8. [8]
    Dasgupda, D., Mcgregor, D.P.: Non-stationary Function Optimization using the Structured Genetic Algorithm, Proc. of. 2nd PPSN (PPSN′92), Vol. 2, pp. 145–154, 1992.Google Scholar
  9. [9]
    Whitley, D.: The genitor algorithm and selection pressure: why rank based allocation of reproductive trials is best, In Proceedings of the Third Int’l. Conference on Genetic Algorithms and their Applications, 116–121, J.D. SchafFer, (ed.), Morgan Kaufmann, June, 1989.Google Scholar
  10. [10]
    Cerf, R.: The dynamics of mutation selection algorithms with large population sizes, Ann. Inst. H. Poincaré Probab. Statist., Vol. 32, pp. 455–508, 1996.MathSciNetMATHGoogle Scholar
  11. [11]
    Leung, Y., Gao, Y., Xu, Z.B.: Degree of population diversity — A perspective on premature convergence in genetic algorithms and its Markov chain analysis, IEEE Trans. Neural Networks, Vol. 8, pp. 1165–1175, 1997.CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Wien 1999

Authors and Affiliations

  • Subbiah Baskaran
    • 1
    • 2
    • 3
  • David Noever
    • 3
  1. 1.Institut fuer Theoretische ChemieWienAustria
  2. 2.Raytheon ITSSUSA
  3. 3.Biophysics Branch ES76 National Aeronautics and Space AdministrationGeorge C. Marshall Space Flight CenterHuntsvilleUSA

Personalised recommendations