IRKA is a Riemannian Gradient Descent Method

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mlinarić, Petar, Beattie, Christopher A., Drmač, Zlatko, Gugercin, Serkan
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911949887897600
author Mlinarić, Petar
Beattie, Christopher A.
Drmač, Zlatko
Gugercin, Serkan
author_facet Mlinarić, Petar
Beattie, Christopher A.
Drmač, Zlatko
Gugercin, Serkan
contents The iterative rational Krylov algorithm (IRKA) is a commonly used fixed-point iteration developed to minimize the $\mathcal{H}_2$ model order reduction error. In this work, IRKA is recast as a Riemannian gradient descent method with a fixed step size over the manifold of rational functions having fixed degree. This interpretation motivates the development of a Riemannian gradient descent method utilizing as a natural extension variable step size and line search. Comparisons made between IRKA and this extension on a few examples demonstrate significant benefits.
format Preprint
id arxiv_https___arxiv_org_abs_2311_02031
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle IRKA is a Riemannian Gradient Descent Method
Mlinarić, Petar
Beattie, Christopher A.
Drmač, Zlatko
Gugercin, Serkan
Numerical Analysis
Systems and Control
Optimization and Control
The iterative rational Krylov algorithm (IRKA) is a commonly used fixed-point iteration developed to minimize the $\mathcal{H}_2$ model order reduction error. In this work, IRKA is recast as a Riemannian gradient descent method with a fixed step size over the manifold of rational functions having fixed degree. This interpretation motivates the development of a Riemannian gradient descent method utilizing as a natural extension variable step size and line search. Comparisons made between IRKA and this extension on a few examples demonstrate significant benefits.
title IRKA is a Riemannian Gradient Descent Method
topic Numerical Analysis
Systems and Control
Optimization and Control
url https://arxiv.org/abs/2311.02031