A note on the augmented Hessian when the reduced Hessian is semidefinite

Kurt M. Anstreicher, Margaret Wright

Research output: Contribution to journalArticle

Abstract

Certain matrix relationships play an important role in optimally conditions and algorithms for nonlinear and semidefinite programming. Let H be an n x n symmetric matrix. A an m x n matrix, and Z a basis for the null space of A. (In a typical optimization context H is the Hessian of a smooth function and A is the Jacobian of a set of constraints.) When the reduced Hessian ZTHZ is positive definite, augmented Lagrangian methods rely on the known existence of a finite ρ̄ > O such that, for all ρ > ρ̄, the augmented Hessian H + ρATA is positive definite. In this note we analyze the case when ZTHZ is positive semidefinite, i.e., singularity is allowed, and show that the situation is more complicated. In particular, we give a simple necessary and sufficient condition for the existence of a finite ρ̄ so that H + ρATA is positive semidefinite for ρ ≥ ρ̄. A corollary of our result is that if H is nonsingular and indefinite while ZTHZ is positive semidefinite and singular, no such ρ̄ exists.

Original languageEnglish (US)
Pages (from-to)243-253
Number of pages11
JournalSIAM Journal on Optimization
Volume11
Issue number1
DOIs
StatePublished - 2000

Fingerprint

Positive semidefinite
Positive definite
Augmented Lagrangian Method
Null Space
Semidefinite Programming
Symmetric matrix
Nonlinear Programming
Smooth function
Corollary
Singularity
Necessary Conditions
Optimization
Sufficient Conditions

Keywords

  • Augmented Hessian
  • Augmented Lagrangian methods
  • Inertia
  • Reduced Hessian

ASJC Scopus subject areas

  • Mathematics(all)
  • Applied Mathematics

Cite this

A note on the augmented Hessian when the reduced Hessian is semidefinite. / Anstreicher, Kurt M.; Wright, Margaret.

In: SIAM Journal on Optimization, Vol. 11, No. 1, 2000, p. 243-253.

Research output: Contribution to journalArticle

@article{f6b95db6a88647298c442c4f8b41429f,
title = "A note on the augmented Hessian when the reduced Hessian is semidefinite",
abstract = "Certain matrix relationships play an important role in optimally conditions and algorithms for nonlinear and semidefinite programming. Let H be an n x n symmetric matrix. A an m x n matrix, and Z a basis for the null space of A. (In a typical optimization context H is the Hessian of a smooth function and A is the Jacobian of a set of constraints.) When the reduced Hessian ZTHZ is positive definite, augmented Lagrangian methods rely on the known existence of a finite ρ̄ > O such that, for all ρ > ρ̄, the augmented Hessian H + ρATA is positive definite. In this note we analyze the case when ZTHZ is positive semidefinite, i.e., singularity is allowed, and show that the situation is more complicated. In particular, we give a simple necessary and sufficient condition for the existence of a finite ρ̄ so that H + ρATA is positive semidefinite for ρ ≥ ρ̄. A corollary of our result is that if H is nonsingular and indefinite while ZTHZ is positive semidefinite and singular, no such ρ̄ exists.",
keywords = "Augmented Hessian, Augmented Lagrangian methods, Inertia, Reduced Hessian",
author = "Anstreicher, {Kurt M.} and Margaret Wright",
year = "2000",
doi = "10.1137/S1052623499351791",
language = "English (US)",
volume = "11",
pages = "243--253",
journal = "SIAM Journal on Optimization",
issn = "1052-6234",
publisher = "Society for Industrial and Applied Mathematics Publications",
number = "1",

}

TY - JOUR

T1 - A note on the augmented Hessian when the reduced Hessian is semidefinite

AU - Anstreicher, Kurt M.

AU - Wright, Margaret

PY - 2000

Y1 - 2000

N2 - Certain matrix relationships play an important role in optimally conditions and algorithms for nonlinear and semidefinite programming. Let H be an n x n symmetric matrix. A an m x n matrix, and Z a basis for the null space of A. (In a typical optimization context H is the Hessian of a smooth function and A is the Jacobian of a set of constraints.) When the reduced Hessian ZTHZ is positive definite, augmented Lagrangian methods rely on the known existence of a finite ρ̄ > O such that, for all ρ > ρ̄, the augmented Hessian H + ρATA is positive definite. In this note we analyze the case when ZTHZ is positive semidefinite, i.e., singularity is allowed, and show that the situation is more complicated. In particular, we give a simple necessary and sufficient condition for the existence of a finite ρ̄ so that H + ρATA is positive semidefinite for ρ ≥ ρ̄. A corollary of our result is that if H is nonsingular and indefinite while ZTHZ is positive semidefinite and singular, no such ρ̄ exists.

AB - Certain matrix relationships play an important role in optimally conditions and algorithms for nonlinear and semidefinite programming. Let H be an n x n symmetric matrix. A an m x n matrix, and Z a basis for the null space of A. (In a typical optimization context H is the Hessian of a smooth function and A is the Jacobian of a set of constraints.) When the reduced Hessian ZTHZ is positive definite, augmented Lagrangian methods rely on the known existence of a finite ρ̄ > O such that, for all ρ > ρ̄, the augmented Hessian H + ρATA is positive definite. In this note we analyze the case when ZTHZ is positive semidefinite, i.e., singularity is allowed, and show that the situation is more complicated. In particular, we give a simple necessary and sufficient condition for the existence of a finite ρ̄ so that H + ρATA is positive semidefinite for ρ ≥ ρ̄. A corollary of our result is that if H is nonsingular and indefinite while ZTHZ is positive semidefinite and singular, no such ρ̄ exists.

KW - Augmented Hessian

KW - Augmented Lagrangian methods

KW - Inertia

KW - Reduced Hessian

UR - http://www.scopus.com/inward/record.url?scp=0034549140&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034549140&partnerID=8YFLogxK

U2 - 10.1137/S1052623499351791

DO - 10.1137/S1052623499351791

M3 - Article

VL - 11

SP - 243

EP - 253

JO - SIAM Journal on Optimization

JF - SIAM Journal on Optimization

SN - 1052-6234

IS - 1

ER -