Advertisement

Generalized descent for global optimization

  • A. O. Griewank
Contributed Papers

Abstract

This paper introduces a new method for the global unconstrained minimization of a differentiable objective function. The method is based on search trajectories, which are defined by a differential equation and exhibit certain similarities to the trajectories of steepest descent. The trajectories depend explicitly on the value of the objective function and aim at attaining a given target level, while rejecting all larger local minima. Convergence to the gloal minimum can be proven for a certain class of functions and appropriate setting of two parameters.

Key Words

Global optimization generalized descent search trajectories target level 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Schubert, B. O.,Sequential Optimization of Multimodal Discrete Function with Bounded Rate of Change, Management Science, Vol. 18, pp. 687–693, 1972.Google Scholar
  2. 2.
    Aird, T. J., andRice, J. R.,Systematic Search in High Dimensional Sets, SIAM Journal of Numerical Analysis, Vol. 14, pp. 296–312, 1977.Google Scholar
  3. 3.
    Dickson, L. C. W., Gomulka, J., andSzegö, G. P.,Toward a Global Optimization Technique, Toward Global Optimization, Edited by L. Dickson and G. Szegö, North-Holland Publishing Company, Amsterdam, Holland, 1975.Google Scholar
  4. 4.
    Branin, F. H.,Widely Convergent Method for Finding Multiple Solutions of Nonlinear Equations, IBM Journal of Research and Development, Vol. 16, pp. 504–522, 1972.Google Scholar
  5. 5.
    Hirsch, M.,Differential Topology, Springer-Verlag, New York, New York, 1976.Google Scholar
  6. 6.
    Smale, S.,A Convergent Process of Price Adjustment and Global Newton Methods, Journal of Mathematical Economics, Vol. 3, pp. 107–220, 1976.Google Scholar
  7. 7.
    Keller, H. B.,Global Homotopics and Newton Methods, Recent Advances in Numerical Analysis, Edited by C. DeBoor and G. H. Golub, Academic Press, New York, New York, 1978.Google Scholar
  8. 8.
    John, F.,Ueber die Vollstaendigkeit der Relationen von Morse fuer die Anzahlen Kritischer Punkte, Mathematische Annalen, Vol. 109, pp. 381–394, 1934.Google Scholar
  9. 9.
    Griewank, A.,A Generalized Descent Method for Global Optimization, Australian National University, MS Thesis, 1977.Google Scholar
  10. 10.
    Willmore, T. J.,An Introduction to Differential Geometry, Clarendon Press, Oxford, England, 1959.Google Scholar
  11. 11.
    Inomata, S., andCumada, M.,On the Golf Method, Denshi Gijutso Sogo Kenkyujo, Tokyo, Japan, Bulletin of the Electrotechnical Laboratory, Vol. 25, pp. 495–512, 1964.Google Scholar
  12. 12.
    Zidkov, N. P., andSiedrin, B. M.,A Certain Method of Search for the Minimum of a Function of Several Variables, Computing Methods and Programming, Izdat Moscow University, Vol. 10, pp. 203–210, 1968.Google Scholar
  13. 13.
    Incerti, S., Parisi, V., andZirilli, F.,A New Method for Solving Nonlinear Simultaneous Equations, SIAM Journal of Numerical Analysis, Vol. 16, pp. 779–789, 1979.Google Scholar
  14. 14.
    Griewank, A.,Analysis and Modification of Newton's Method at Singularities, Australian National University, PhD Thesis, 1980.Google Scholar
  15. 15.
    Powell, M. J. D.,A New Algorithm for Unconstrained Optimization, Nonlinear Programming, Edited by J. B. Rosen, O. L. Mangasarian, and K. Ritter, Academic Press, New York, New York, 1970.Google Scholar
  16. 16.
    Gear, C. W.,Numerical Initial-Value Problems in Ordinary Differential Equations, Prentice-Hall, Englewood Cliffs, New Jersey, 1971.Google Scholar
  17. 17.
    Anonymous, International Mathematical and Statistical Libraries, Library 2, Edition 6, Vol. 1, Houston, Texas, 1977.Google Scholar

Copyright information

© Plenum Publishing Corporation 1981

Authors and Affiliations

  • A. O. Griewank
    • 1
  1. 1.Department of Applied Mathematics and Theoretical PhysicsCambridgeUnited Kingdom

Personalised recommendations