Abstract
We consider polynomial eigenvalue problems P(A,α,β)x=0 in which the matrix polynomial is homogeneous in the eigenvalue (α,β) ∈ℂ 2. In this framework infinite eigenvalues are on the same footing as finite eigenvalues. We view the problem in projective spaces to avoid normalization of the eigenpairs. We show that a polynomial eigenvalue problem is well-posed when its eigenvalues are simple. We define the condition numbers of a simple eigenvalue (α,β) and a corresponding eigenvector x and show that the distance to the nearest ill-posed problem is equal to the reciprocal of the condition number of the eigenvector x. We describe a bihomogeneous Newton method for the solution of the homogeneous polynomial eigenvalue problem (homogeneous PEP). © 2002 Elsevier Science Inc.
| Original language | English |
|---|---|
| Pages (from-to) | 71-94 |
| Number of pages | 23 |
| Journal | Linear Algebra and its Applications |
| Volume | 358 |
| Issue number | 1-3 |
| DOIs | |
| Publication status | Published - 1 Jan 2003 |
Keywords
- Condition number
- Matrix polynomial
- Polynomial eigenvalue problem
- Quadratic eigenvalue problem
Fingerprint
Dive into the research topics of 'Perturbation theory for homogeneous polynomial eigenvalue problems'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver