Performance of first- and second-order methods for ℓ1-regularized least squares problems

Kimon Fountoulakis; Jacek Gondzio

doi:10.1007/s10589-016-9853-x

Performance of first- and second-order methods for ℓ1-regularized least squares problems

Kimon Fountoulakis, Jacek Gondzio

School of Mathematics

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

Abstract

We study the performance of first- and second-order optimization methods for ℓ1-regularized sparse least-squares problems as the conditioning of the problem changes and the dimensions of the problem increase up to one trillion. A rigorously defined generator is presented which allows control of the dimensions, the conditioning and the sparsity of the problem. The generator has very low memory requirements and scales well with the dimensions of the problem.

Original language	English
Pages (from-to)	605-635
Number of pages	32
Journal	Computational optimization and applications
Volume	65
Issue number	3
Early online date	14 Jun 2016
DOIs	https://doi.org/10.1007/s10589-016-9853-x
Publication status	Published - Dec 2016

Access to Document

10.1007/s10589-016-9853-x

trillion14dec15Accepted author manuscript, 1.18 MB
art%3A10.1007%2Fs10589-016-9853-xFinal published version, 1.6 MB

Cite this

@article{29db6680bb304f61954a48f8433846f2,

title = "Performance of first- and second-order methods for ℓ1-regularized least squares problems",

abstract = "We study the performance of first- and second-order optimization methods for ℓ1-regularized sparse least-squares problems as the conditioning of the problem changes and the dimensions of the problem increase up to one trillion. A rigorously defined generator is presented which allows control of the dimensions, the conditioning and the sparsity of the problem. The generator has very low memory requirements and scales well with the dimensions of the problem.",

author = "Kimon Fountoulakis and Jacek Gondzio",

year = "2016",

month = dec,

doi = "10.1007/s10589-016-9853-x",

language = "English",

volume = "65",

pages = "605--635",

journal = "Computational optimization and applications",

issn = "0926-6003",

publisher = "Springer",

number = "3",

}

TY - JOUR

T1 - Performance of first- and second-order methods for ℓ1-regularized least squares problems

AU - Fountoulakis, Kimon

AU - Gondzio, Jacek

PY - 2016/12

Y1 - 2016/12

N2 - We study the performance of first- and second-order optimization methods for ℓ1-regularized sparse least-squares problems as the conditioning of the problem changes and the dimensions of the problem increase up to one trillion. A rigorously defined generator is presented which allows control of the dimensions, the conditioning and the sparsity of the problem. The generator has very low memory requirements and scales well with the dimensions of the problem.

AB - We study the performance of first- and second-order optimization methods for ℓ1-regularized sparse least-squares problems as the conditioning of the problem changes and the dimensions of the problem increase up to one trillion. A rigorously defined generator is presented which allows control of the dimensions, the conditioning and the sparsity of the problem. The generator has very low memory requirements and scales well with the dimensions of the problem.

U2 - 10.1007/s10589-016-9853-x

DO - 10.1007/s10589-016-9853-x

M3 - Article

SN - 0926-6003

VL - 65

SP - 605

EP - 635

JO - Computational optimization and applications

JF - Computational optimization and applications

IS - 3

ER -

Performance of first- and second-order methods for ℓ1-regularized least squares problems

Abstract

Access to Document

Fingerprint

Cite this