Computational experiments with scaled initial Hessian approximation for the Broyden family methods

M. Al-Baali, D. Conforti, R. Musmanno

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In this paper we consider two alternative choices for the factor used to scale the initial Hessian approximation, before updating by a member of the Broyden family of updates for quasi-Newton optimization methods. By extensive computational experiments carried out on a set of standard test problems from the CUTE collection, using efficient implementations of the quasi-Newton method, we show that the proposed new scaling factors are better, in terms of efficiency achieved (number of iterations, number of function and gradient evaluations), than the standard choice proposed in the literature.

Original languageEnglish
Pages (from-to)375-389
Number of pages15
JournalOptimization
Volume48
Issue number3
Publication statusPublished - 2000

Fingerprint

Quasi-Newton Method
Computational Experiments
Scaling Factor
Newton-Raphson method
Approximation
Efficient Implementation
Test Problems
Updating
Optimization Methods
Update
Experiments
Gradient
Iteration
Alternatives
Evaluation
Standards
Family
Experiment
Factors
Scaling

Keywords

  • Broyden's class
  • Initial self-scaling
  • Quasi-Newton methods
  • Unconstrained optimization

ASJC Scopus subject areas

  • Management Science and Operations Research
  • Applied Mathematics
  • Control and Optimization

Cite this

Computational experiments with scaled initial Hessian approximation for the Broyden family methods. / Al-Baali, M.; Conforti, D.; Musmanno, R.

In: Optimization, Vol. 48, No. 3, 2000, p. 375-389.

Research output: Contribution to journalArticle

@article{4f68416b4972497a9b46b08c229ea7a8,
title = "Computational experiments with scaled initial Hessian approximation for the Broyden family methods",
abstract = "In this paper we consider two alternative choices for the factor used to scale the initial Hessian approximation, before updating by a member of the Broyden family of updates for quasi-Newton optimization methods. By extensive computational experiments carried out on a set of standard test problems from the CUTE collection, using efficient implementations of the quasi-Newton method, we show that the proposed new scaling factors are better, in terms of efficiency achieved (number of iterations, number of function and gradient evaluations), than the standard choice proposed in the literature.",
keywords = "Broyden's class, Initial self-scaling, Quasi-Newton methods, Unconstrained optimization",
author = "M. Al-Baali and D. Conforti and R. Musmanno",
year = "2000",
language = "English",
volume = "48",
pages = "375--389",
journal = "Optimization",
issn = "0233-1934",
publisher = "Taylor and Francis Ltd.",
number = "3",

}

TY - JOUR

T1 - Computational experiments with scaled initial Hessian approximation for the Broyden family methods

AU - Al-Baali, M.

AU - Conforti, D.

AU - Musmanno, R.

PY - 2000

Y1 - 2000

N2 - In this paper we consider two alternative choices for the factor used to scale the initial Hessian approximation, before updating by a member of the Broyden family of updates for quasi-Newton optimization methods. By extensive computational experiments carried out on a set of standard test problems from the CUTE collection, using efficient implementations of the quasi-Newton method, we show that the proposed new scaling factors are better, in terms of efficiency achieved (number of iterations, number of function and gradient evaluations), than the standard choice proposed in the literature.

AB - In this paper we consider two alternative choices for the factor used to scale the initial Hessian approximation, before updating by a member of the Broyden family of updates for quasi-Newton optimization methods. By extensive computational experiments carried out on a set of standard test problems from the CUTE collection, using efficient implementations of the quasi-Newton method, we show that the proposed new scaling factors are better, in terms of efficiency achieved (number of iterations, number of function and gradient evaluations), than the standard choice proposed in the literature.

KW - Broyden's class

KW - Initial self-scaling

KW - Quasi-Newton methods

KW - Unconstrained optimization

UR - http://www.scopus.com/inward/record.url?scp=26444533599&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=26444533599&partnerID=8YFLogxK

M3 - Article

VL - 48

SP - 375

EP - 389

JO - Optimization

JF - Optimization

SN - 0233-1934

IS - 3

ER -