The stability of split-preconditioned FGMRES in four precisions
From MaRDI portal
Abstract: We consider the split-preconditioned FGMRES method in a mixed precision framework, in which four potentially different precisions can be used for computations with the coefficient matrix, application of the left preconditioner, and application of the right preconditioner, and the working precision. Our analysis is applicable to general preconditioners. We obtain bounds on the backward and forward errors in split-preconditioned FGMRES. Our analysis further provides insight into how the various precisions should be chosen; a suitable selection guarantees a backward error on the order of the working precision.
Recommendations
- Using FGMRES to obtain backward stability in mixed precision
- Mixed precision s‐step Lanczos and conjugate gradient algorithms
- Mixed-precision preconditionders in parallel domain decomposition solvers
- Mixed Precision Iterative Refinement with Sparse Approximate Inverse Preconditioning
- Five-Precision GMRES-Based Iterative Refinement
Cited in
(4)- Mixed precision sketching for least-squares problems and its application in GMRES-based iterative refinement
- Numerical study of mixed precision GMRES(m) preconditioned by deflation
- A modular framework for the backward error analysis of GMRES
- Mixed-Precision Computing for Scientific Discovery: Formats, Co-Design, and Responsible Approximation
This page was built for publication: The stability of split-preconditioned FGMRES in four precisions
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6430345)
