A generalization of Saad's theorem on Rayleigh-Ritz approximations
If \(A\) is a Hermitian matrix with eigenpair \((\lambda,x)\) and Ritz pair \((\mu,u)\) from the subspace \({\mathcal K}\), then \textit{Y. Saad} [Numerical methods for large eigenvalue problems: Theory and algorithm, Wiley, NY (1992; Zbl 0991.65039)] gives an estimate of the form \(\sin \angle (x,u)\leq C \sin\angle(x,{\mathcal K})\), with \(C\) depending of \(A\), \(\mathcal K\), \(x\), \(u\). This is generalized in this paper to eigenspaces of non-Hermitian matrices, i.e., the vector \(x\) can be replaced by an eigenspace \(\mathcal X\) and \(u\) by a Ritz space \(\mathcal U\). This bound is however not good enough to prove convergence when \(\mathcal K\) approaches \(\mathcal X\).
- Accuracy of singular vectors obtained by projection-based SVD methods
- The least squares and line search in extracting eigenpairs in Jacobi-Davidson method
- Two-sided Grassmann-Rayleigh quotient iteration
- Cluster robust error estimates for the Rayleigh-Ritz approximation. I: Estimates for invariant subspaces
- A generalization of Saad's bound on harmonic Ritz vectors of Hermitian matrices
- A generalisation of the Rayleigh-Ritz approximation
- The convergence of harmonic Ritz vectors and harmonic Ritz values, revisited
- Krylov-Schur-type restarts for the two-sided Arnoldi method
- An analogue of the results of Saad and Stewart for harmonic Ritz vectors
- Fast and accurate randomized algorithms for linear systems and eigenvalue problems
- A subspace method for large-scale trace ratio problems
This page was built for publication: A generalization of Saad's theorem on Rayleigh-Ritz approximations
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5935366)
