Steplength selection in interior-point methods for quadratic programming

Frank Curtis; Jorge Nocedal

doi:10.1016/j.aml.2006.05.020

Steplength selection in interior-point methods for quadratic programming

Frank Curtis, Jorge Nocedal^*

^*Corresponding author for this work

Industrial Engineering and Management Sciences

Research output: Contribution to journal › Article › peer-review

5 Scopus citations

Abstract

We present a new strategy for choosing primal and dual steplengths in a primal-dual interior-point algorithm for convex quadratic programming. Current implementations often scale steps equally to avoid increases in dual infeasibility between iterations. We propose that this method can be too conservative, while safeguarding an unequally-scaled steplength approach will often require fewer steps toward a solution. Computational results are given.

Original language	English (US)
Pages (from-to)	516-523
Number of pages	8
Journal	Applied Mathematics Letters
Volume	20
Issue number	5
DOIs	https://doi.org/10.1016/j.aml.2006.05.020
State	Published - May 2007

Keywords

Barrier method
Interior-point method
Nonlinear optimization
Quadratic programming

ASJC Scopus subject areas

Applied Mathematics

Access to Document

10.1016/j.aml.2006.05.020

Cite this

@article{6008ddc4eada465e9e202558be14862c,

title = "Steplength selection in interior-point methods for quadratic programming",

abstract = "We present a new strategy for choosing primal and dual steplengths in a primal-dual interior-point algorithm for convex quadratic programming. Current implementations often scale steps equally to avoid increases in dual infeasibility between iterations. We propose that this method can be too conservative, while safeguarding an unequally-scaled steplength approach will often require fewer steps toward a solution. Computational results are given.",

keywords = "Barrier method, Interior-point method, Nonlinear optimization, Quadratic programming",

author = "Frank Curtis and Jorge Nocedal",

note = "Funding Information: This work was supported in part by a grant from the Intel Corporation. ",

year = "2007",

month = may,

doi = "10.1016/j.aml.2006.05.020",

language = "English (US)",

volume = "20",

pages = "516--523",

journal = "Applied Mathematics Letters",

issn = "0893-9659",

publisher = "Elsevier Limited",

number = "5",

}

TY - JOUR

T1 - Steplength selection in interior-point methods for quadratic programming

AU - Curtis, Frank

AU - Nocedal, Jorge

N1 - Funding Information: This work was supported in part by a grant from the Intel Corporation.

PY - 2007/5

Y1 - 2007/5

N2 - We present a new strategy for choosing primal and dual steplengths in a primal-dual interior-point algorithm for convex quadratic programming. Current implementations often scale steps equally to avoid increases in dual infeasibility between iterations. We propose that this method can be too conservative, while safeguarding an unequally-scaled steplength approach will often require fewer steps toward a solution. Computational results are given.

AB - We present a new strategy for choosing primal and dual steplengths in a primal-dual interior-point algorithm for convex quadratic programming. Current implementations often scale steps equally to avoid increases in dual infeasibility between iterations. We propose that this method can be too conservative, while safeguarding an unequally-scaled steplength approach will often require fewer steps toward a solution. Computational results are given.

KW - Barrier method

KW - Interior-point method

KW - Nonlinear optimization

KW - Quadratic programming

UR - http://www.scopus.com/inward/record.url?scp=33845284269&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33845284269&partnerID=8YFLogxK

U2 - 10.1016/j.aml.2006.05.020

DO - 10.1016/j.aml.2006.05.020

M3 - Article

AN - SCOPUS:33845284269

SN - 0893-9659

VL - 20

SP - 516

EP - 523

JO - Applied Mathematics Letters

JF - Applied Mathematics Letters

IS - 5

ER -

Steplength selection in interior-point methods for quadratic programming

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this