A simple algorithm for the inverse field of values problem

Author

Russell Carden

Affiliations

Department of Computational and Applied Mathematics, Rice University, Houston, TX 77005-1832, USA

E-mail

Russell.L.Carden@rice.edu

Journal

Inverse Problems Create an alert RSS this journal

Issue

Volume 25, Number 11

Citation

Russell Carden 2009 Inverse Problems 25 115019

doi: 10.1088/0266-5611/25/11/115019


 
Tag this article Full text PDF (362 KB)
Abstract

The field of values of a matrix is the closed convex subset of the complex plane comprising all Rayleigh quotients, a set of interest in the stability analysis of dynamical systems and convergence theory of matrix iterations, among other applications. Recently, Uhlig proposed the inverse field of values problem: given a point in the field of values, determine a vector for which this point is the corresponding Rayleigh quotient. Uhlig also devised a sophisticated algorithm involving random vectors and the boundaries of ellipses for solving the inverse field of values problem. We propose a simpler deterministic algorithm that must converge (in exact arithmetic) and for most points yields an exact result in only a few iterations. The algorithm builds upon the fact that the inverse field of values problem can be solved exactly in the two-dimensional case. We also resolve a conjecture posed by Uhlig concerning the number of linearly independent vectors that generate a point in the field of values and propose a more challenging inverse field of values problem that is of interest in eigenvalue computations.

 
PACS

02.30.Zz Inverse problems

02.60.-x Numerical approximation and analysis

02.10.Ud Linear algebra

MSC

15A29 Inverse problems

65F10 Iterative methods for linear systems (See also 65N22)

15A18 Eigenvalues, singular values, and eigenvectors

Subjects

Mathematical physics

Computational physics

Dates

Issue 11 ( 1 November 2009)

Received 7 June 2009 , in final form 29 September 2009

Published 5 November 2009



Users also read

What's this?
This innovative new feature generates a list of articles 'also read' by other users based on them reading the original article. Article abstracts citations and references are all considered and weighted accordingly. We hope that this will help you find relevant papers for your research.

  1. Greedy solution of ill-posed problems: error bounds and exact inversion
  2. Fine scale uncertainty in parameter estimation for elliptic equations

View by subject


Export