Approximate invariant subspaces and quasi-Newton optimization methods

Serge Gratton, P.L. Toint

Research output: Contribution to journalArticle

69 Downloads (Pure)

Abstract

New approximate secant equations are shown to result from the knowledge of (problem dependent) invariant subspace information, which in turn suggests improvements in quasi-Newton methods for unconstrained minimization. A new limited-memory Broyden-Fletcher-Goldfarb-Shanno using approximate secant equations is then derived and its encouraging behaviour illustrated on a small collection of multilevel optimization examples. The smoothing properties of this algorithm are considered next, and automatic generation of approximate eigenvalue information demonstrated. The use of this information for improving algorithmic performance is finally investigated on the same multilevel examples. © 2010 Taylor & Francis.
Original languageEnglish
Pages (from-to)507-529
Number of pages23
JournalOptimization Methods and Software
Volume25
Issue number4
DOIs
Publication statusPublished - 1 Aug 2010

Fingerprint

Quasi-Newton Method
Invariant Subspace
Optimization Methods
Chord or secant line
Newton-Raphson method
Unconstrained Minimization
Data storage equipment
Smoothing
Eigenvalue
Optimization
Dependent

Keywords

  • secant equation
  • multigrid techniques
  • quasi-Newton methods
  • Unconstrained optimization

Cite this

@article{340c413070934797b13f0217dc420791,
title = "Approximate invariant subspaces and quasi-Newton optimization methods",
abstract = "New approximate secant equations are shown to result from the knowledge of (problem dependent) invariant subspace information, which in turn suggests improvements in quasi-Newton methods for unconstrained minimization. A new limited-memory Broyden-Fletcher-Goldfarb-Shanno using approximate secant equations is then derived and its encouraging behaviour illustrated on a small collection of multilevel optimization examples. The smoothing properties of this algorithm are considered next, and automatic generation of approximate eigenvalue information demonstrated. The use of this information for improving algorithmic performance is finally investigated on the same multilevel examples. {\circledC} 2010 Taylor & Francis.",
keywords = "secant equation, multigrid techniques, quasi-Newton methods, Unconstrained optimization",
author = "Serge Gratton and P.L. Toint",
note = "Publication code : FP SB010/2007/11 ; QA 0002.2/001/07/11",
year = "2010",
month = "8",
day = "1",
doi = "10.1080/10556780902992746",
language = "English",
volume = "25",
pages = "507--529",
journal = "Optimization Methods and Software",
issn = "1055-6788",
publisher = "Taylor & Francis",
number = "4",

}

Approximate invariant subspaces and quasi-Newton optimization methods. / Gratton, Serge; Toint, P.L.

In: Optimization Methods and Software, Vol. 25, No. 4, 01.08.2010, p. 507-529.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Approximate invariant subspaces and quasi-Newton optimization methods

AU - Gratton, Serge

AU - Toint, P.L.

N1 - Publication code : FP SB010/2007/11 ; QA 0002.2/001/07/11

PY - 2010/8/1

Y1 - 2010/8/1

N2 - New approximate secant equations are shown to result from the knowledge of (problem dependent) invariant subspace information, which in turn suggests improvements in quasi-Newton methods for unconstrained minimization. A new limited-memory Broyden-Fletcher-Goldfarb-Shanno using approximate secant equations is then derived and its encouraging behaviour illustrated on a small collection of multilevel optimization examples. The smoothing properties of this algorithm are considered next, and automatic generation of approximate eigenvalue information demonstrated. The use of this information for improving algorithmic performance is finally investigated on the same multilevel examples. © 2010 Taylor & Francis.

AB - New approximate secant equations are shown to result from the knowledge of (problem dependent) invariant subspace information, which in turn suggests improvements in quasi-Newton methods for unconstrained minimization. A new limited-memory Broyden-Fletcher-Goldfarb-Shanno using approximate secant equations is then derived and its encouraging behaviour illustrated on a small collection of multilevel optimization examples. The smoothing properties of this algorithm are considered next, and automatic generation of approximate eigenvalue information demonstrated. The use of this information for improving algorithmic performance is finally investigated on the same multilevel examples. © 2010 Taylor & Francis.

KW - secant equation

KW - multigrid techniques

KW - quasi-Newton methods

KW - Unconstrained optimization

UR - http://www.scopus.com/inward/record.url?scp=77951197373&partnerID=8YFLogxK

U2 - 10.1080/10556780902992746

DO - 10.1080/10556780902992746

M3 - Article

VL - 25

SP - 507

EP - 529

JO - Optimization Methods and Software

JF - Optimization Methods and Software

SN - 1055-6788

IS - 4

ER -