Please use this identifier to cite or link to this item: http://dx.doi.org/10.14279/depositonce-6545
Main Title: Structured pseudospectra and the condition of a nonderogatory eigenvalue
Author(s): Karow, Michael
Type: Article
Language Code: en
Abstract: Let $\lambda$ be a nonderogatory eigenvalue of $A\in\mathbb{C}^{n\times n}$ of algebraic multiplicity m. The sensitivity of $\lambda$ with respect to matrix perturbations of the form $A\leadsto A+\Delta$, $\Delta\in\boldsymbol{\Delta}$, is measured by the structured condition number $\kappa_{\boldsymbol{\Delta}}(A,\lambda)$. Here $\boldsymbol{\Delta}$ denotes the set of admissible perturbations. However, if $\boldsymbol{\Delta}$ is not a vector space over $\mathbb{C}$, then $\kappa_{\boldsymbol{\Delta}}(A,\lambda)$ provides only incomplete information about the mobility of $\lambda$ under small perturbations from $\boldsymbol{\Delta}$. The full information is then given by the set $K_{\boldsymbol{\Delta}}(x,y)=\{y^*\Delta x;$ $\Delta\in\boldsymbol{\Delta},$ $\|\Delta\|\leq1\}\subset\mathbb{C}$ that depends on $\boldsymbol{\Delta}$, a pair of normalized right and left eigenvectors $x,y$, and the norm $\|\cdot\|$ that measures the size of the perturbations. We always have $\kappa_{\boldsymbol{\Delta}}(A,\lambda)=\max\{|z|^{1/m};$ $z\in K_{\boldsymbol{\Delta}}(x,y)\}$. Furthermore, $K_{\boldsymbol{\Delta}}(x,y)$ determines the shape and growth of the $\boldsymbol{\Delta}$-structured pseudospectrum in a neighborhood of $\lambda$. In this paper we study the sets $K_{\boldsymbol{\Delta}}(x,y)$ and obtain methods for computing them. In doing so we obtain explicit formulae for structured eigenvalue condition numbers with respect to many important perturbation classes.
URI: https://depositonce.tu-berlin.de//handle/11303/7272
http://dx.doi.org/10.14279/depositonce-6545
Issue Date: 30-Nov-2010
Date Available: 14-Dec-2017
DDC Class: 518 Numerische Analysis
512 Algebra
Subject(s): eigenvalues
structured perturbations
pseudospectra
condition numbers
License: http://rightsstatements.org/vocab/InC/1.0/
Journal Title: SIAM Journal on Matrix Analysis and Applications
Publisher: Society for Industrial and Applied Mathematics
Publisher Place: Philadelphia, Pa.
Volume: 31
Issue: 5
Publisher DOI: 10.1137/070695836
Page Start: 2860
Page End: 2881
EISSN: 1095-7162
ISSN: 0895-4798
Appears in Collections:FG Numerische Lineare Algebra » Publications

Files in This Item:
File Description SizeFormat 
2010_karow.pdf297.56 kBAdobe PDFThumbnail
View/Open


Items in DepositOnce are protected by copyright, with all rights reserved, unless otherwise indicated.