Advertisement

Mathematical Programming

, Volume 93, Issue 3, pp 495–515 | Cite as

A decomposition procedure based on approximate Newton directions

  • A. J. Conejo
  • F. J. Nogales
  • F. J. Prieto

Abstract.

 The efficient solution of large-scale linear and nonlinear optimization problems may require exploiting any special structure in them in an efficient manner. We describe and analyze some cases in which this special structure can be used with very little cost to obtain search directions from decomposed subproblems. We also study how to correct these directions using (decomposable) preconditioned conjugate gradient methods to ensure local convergence in all cases. The choice of appropriate preconditioners results in a natural manner from the structure in the problem. Finally, we conduct computational experiments to compare the resulting procedures with direct methods.

Keywords

Conjugate Gradient Gradient Method Nonlinear Optimization Special Structure Computational Experiment 
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.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2002

Authors and Affiliations

  • A. J. Conejo
    • 1
  • F. J. Nogales
    • 2
  • F. J. Prieto
    • 3
  1. 1.E.T.S. de Ingenieros Industriales, Univ. de Castilla-La Mancha, Ciudad Real, Spain, e-mail: aconejo@ind-cr.uclm.esES
  2. 2.E.T.S. de Ingenieros Industriales, Univ. de Castilla-La Mancha, Ciudad Real, Spain, e-mail: fjnogale@ind-cr.uclm.esES
  3. 3.Dept. of Statistics and Econometrics, Univ. Carlos III de Madrid, Spain, e-mail: fjp@est-econ. uc3m.esES

Personalised recommendations