Any admissible cycle-convergence behavior is possible for restarted GMRES at its initial cycles.
From MaRDI portal
Publication:2889398
DOI10.1002/NLA.739zbMATH Open1249.65073arXiv0907.3573OpenAlexW2076892752MaRDI QIDQ2889398FDOQ2889398
Authors: Eugene Vecharynski, Julien Langou
Publication date: 7 June 2012
Published in: Numerical Linear Algebra with Applications (Search for Journal in Brave)
Abstract: Given a matrix order , a restart parameter (), a decreasing positive sequence , where , it is shown that there exits an -by- matrix and a vector with such that , , where is the residual at cycle of restarted GMRES with restart parameter applied to the linear system , with initial residual . Moreover, the matrix can be chosen to have any desired eigenvalues. We can also construct arbitrary cases of stagnation; namely, when for any . The restart parameter can be fixed or variable.
Full work available at URL: https://arxiv.org/abs/0907.3573
Recommendations
- On the residual norms, the Ritz values and the harmonic Ritz values that can be generated by restarted GMRES
- scientific article; zbMATH DE number 1300160
- The cycle-convergence of restarted GMRES for normal matrices is sublinear
- Discussion on convergence conditions for restarted GMRES
- Generalization of convergence conditions for a restarted GMRES
Cites Work
- Title not available (Why is that?)
- GMRES: A Generalized Minimal Residual Algorithm for Solving Nonsymmetric Linear Systems
- Krylov sequences of maximal length and convergence of GMRES
- Variational Iterative Methods for Nonsymmetric Systems of Linear Equations
- How Fast are Nonsymmetric Matrix Iterations?
- GMRES with Deflated Restarting
- Any Nonincreasing Convergence Curve is Possible for GMRES
- The superlinear convergence behaviour of GMRES
- Deflated and Augmented Krylov Subspace Techniques
- On the Occurrence of Superlinear Convergence of Exact and Inexact Krylov Subspace Methods
- Restarted GMRES preconditioned by deflation
- Adaptively Preconditioned GMRES Algorithms
- The Tortoise and the Hare Restart GMRES
- Analysis of acceleration strategies for restarted minimal residual methods
- Complementary cycles of restarted GMRES
- New conditions for non-stagnation of minimal residual methods
- The cycle-convergence of restarted GMRES for normal matrices is sublinear
- Some Remarks on the Elman Estimate for GMRES
- A simple strategy for varying the restart parameter in GMRES\((m)\)
- Generalization of convergence conditions for a restarted GMRES
- Some remarks on the restarted and augmented GMRES method
- Using spectral low rank preconditioners for large electromagnetic calculations
Cited In (14)
- Generalization of convergence conditions for a restarted GMRES
- On the Forsythe conjecture
- On the convergence of Q-OR and Q-MR Krylov methods for solving nonsymmetric linear systems
- Any finite convergence curve is possible in the initial iterations of restarted FOM
- On two generalized inverse eigenvalue problems for Hessenberg-upper triangular pencils and their application to the study of GMRES convergence
- Limited‐memory polynomial methods for large‐scale matrix functions
- Effective matrix-free preconditioning for the augmented immersed interface method
- Any Ritz value behavior is possible for Arnoldi and for GMRES
- Some observations on weighted GMRES
- On the residual norms, the Ritz values and the harmonic Ritz values that can be generated by restarted GMRES
- Complementary cycles of restarted GMRES
- The cycle-convergence of restarted GMRES for normal matrices is sublinear
- GMRES algorithms over 35 years
- Explanation of a phenomenon witnessed in pre-processed GMRES
This page was built for publication: Any admissible cycle-convergence behavior is possible for restarted GMRES at its initial cycles.
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2889398)