A relative backward perturbation theorem for the eigenvalue problem
From MaRDI portal
If \(| \cdot |\) denotes absolute values taken componentwise the following statement is formulated: If \(| r| =| A\hat x-{\hat \lambda}\hat x| \leq \eta | A| | \hat x|\) holds for an approximate eigenpair (\({\hat \lambda}\),\^x) of A, then there exists a perturbation matrix E satisfying \(| E| \leq \eta | A|\) such that (\({\hat \lambda}\),\^x) is an exact eigenpair of the matrix \(A+E\).
Recommendations
- Relative perturbation results for eigenvalues and eigenvectors of diagonalisable matrices
- Relative perturbation theory for a class of diagonalizable Hermitian matrix pairs
- Relative perturbation bounds of matrix eigenvalues
- A note on the backward error analysis for eigenvalue problems
- Relative Perturbation Theory: I. Eigenvalue and Singular Value Variations
Cites work
Cited in
(3)
This page was built for publication: A relative backward perturbation theorem for the eigenvalue problem
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1262696)