Structured Backward Error and Condition of Generalized Eigenvalue Problems

From MaRDI portal





Backward errors describe stability of a numerical method while condition numbers describe sensitivity of the solution. The theory of backward error and conditioning is well developed for linear systems and least squares problems. The authors define and evaluate the backward error and the condition number for the generalized eigenvalue problem \(Ax=\lambda Bx\) where \(A, B\in \mathbb{C}^{n\times n}\). In both cases normwise analysis as well as componentwise analysis is done. A special attention is paid to the case of matrices with a fixed linear structure \[ A=\sum_{i=1}^ta_iU_i, \qquad B=\sum_{i=1}^tb_iV_i \] and perturbations preserving this structure \[ \Delta A=\sum_{i=1}^t\Delta a_iU_i, \qquad \Delta B=\sum_{i=1}^t\Delta b_iV_i. \] The analysis is illustrated by some numerical examples.




Cited in
(74)


Describes a project that uses

Uses Software






This page was built for publication: Structured Backward Error and Condition of Generalized Eigenvalue Problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4243516)