A method for separating nearly multiple eigenvalues for Hermitian matrix. (English) Zbl 1112.65032
The following problem is discussed: Given a symmetric matrix and assume that by one of the methods of verified computation a small interval has been determined which encloses \(k\) of its eigenvalues. Verify that they are not a k-fold eigenvalue. An algorithm using verified computation, e.g. using INTLAB, is given, which can decide this question. Two examples with \(k=2\) (the infamous Wilkinson matrix) and \(k=3\) illustrate the method.
Reviewer: Ludwig Elsner (Bielefeld)
MSC:
65F15 | Numerical computation of eigenvalues and eigenvectors of matrices |
65G20 | Algorithms with automatic result verification |
15A42 | Inequalities involving eigenvalues and eigenvectors |
Keywords:
multiple eigenvalue; symmetric matrix; verified computation; numerical examples; eigenvalue enclosure; algorithm; Wilkinson matrixSoftware:
INTLABReferences:
[1] | Rump, S. M., INTLAB-INTerval LABoratory, (Cesndes, Tibor, Developments in Reliable Computing (1999), Kluwer Academic Publishers: Kluwer Academic Publishers The Netherlands), 77-104 · Zbl 0949.65046 |
[2] | Rump, S. M., Computational error bounds for multiple or nearly multiple eigenvalues, Linear Algebra Appl., 324, 209-226 (2001) · Zbl 0986.65031 |
[3] | Wilkinson, J. H., The Algebric Eigenvalue Problem (1965), Oxford University Press: Oxford University Press London · Zbl 0258.65037 |
[4] | Yamamoto, N., A simple method for error bounds of eigenvalues of symmetric matrices, Linear Algebra Appl., 324, 227-234 (2001) · Zbl 0981.65042 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.