Modifying Bfgs Method

Abstract

In this paper, we propose a modiied BFGS method and study the global and superlinear convergence properties of this method. We show that under certain circumstances this modiied BFGS method corrects the eigenvalues better than the BFGS does. Our numerical results support this claim and also indicate that the modiied BFGS method may be competitive with the BFGS method in general. This modiied method can also be regarded as an implementation technique of the standard BFGS method.

Showing 1-10 of 13 references

Superlinear convergence of a class of -bounded rank-one update methods

  • A Liao
  • 1991

A tool for the analysis of quasi-Newton methods with application to unconstrained minimization

  • R H Byrd, J Nocedal
  • 1989

Quasi-Newton updates in abstract spaces

  • M J Todd
  • 1984

Collinear scaling and sequential estimation in sparse optimization algorithms

  • D C Sorensen
  • 1982

Mor e. Quasi-Newton methods, motivation and theory

  • J E Dennis
  • 1977