Exclusion Theorems and the Perturbation Analysis of the Generalized Eigenvalue Problem
King-wah Eric Chu
Abstract
King-wah Eric Chu
Abstract
A Bauer–Fike type theorem is proved for the eigenvalue problem $Ax = \lambda Bx$. A generalization is then applied to obtain perturbation bounds for clusters of eigenvalues. The conditioning of an eigenvalue (finite or infinite) is proved to be dependent on a Jordan type condition number, its eigenvector deficiency and the regularity of the matrix pencil $(A - \lambda B)$. An example illustrating the perturbation bounds is given.
OpenAlex reports 24 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
A Bauer–Fike type theorem is proved for the eigenvalue problem $Ax = \lambda Bx$. A generalization is then applied to obtain perturbation bounds for clusters of eigenvalues. The conditioning of an eigenvalue (finite or infinite) is proved to be dependent on a Jordan type condition number, its eigenvector deficiency and the regularity of the matrix pencil $(A - \lambda B)$. An example illustrating the perturbation bounds is given.
Key concepts: Eigenvalues and eigenvectors, Mathematics, Eigenvalue perturbation, Matrix pencil, Perturbation (astronomy), Lambda, Divide-and-conquer eigenvalue algorithm, Applied mathematics