Enlarged GMRES for reducing communication

Hussam Al Daas 1, * Laura Grigori 1, * Pascal Hénon 2 Philippe Ricoux 3
* Corresponding author
1 ALPINES - Algorithms and parallel tools for integrated numerical simulations
Inria de Paris, Institut National des Sciences Mathématiques et de leurs Interactions, LJLL - Laboratoire Jacques-Louis Lions
Abstract : We propose a variant of the GMRES method for solving linear systems of equations with one or multiple right-hand sides. Our method is based on the idea of the enlarged Krylov subspace to reduce communication. It can be interpreted as a block GMRES method. Hence, we are interested in detecting inexact breakdowns. We introduce a strategy to perform the test of detection. Furthermore, we propose an eigenvalues deflation technique aiming to have two benefits. The first advantage is to avoid the plateau of convergence after the end of a cycle in the restarted version. The second is to have a very fast convergence when solving the same system with different right-hand sides, each given at a different time (useful in the context of CPR preconditioner). With the same memory cost, we obtain a saving of up to 50 % in the number of iterations to reach convergence with respect to the original method.
Complete list of metadatas

Cited literature [30 references]  Display  Hide  Download

https://hal.inria.fr/hal-01497943
Contributor : Hussam Al Daas <>
Submitted on : Wednesday, March 29, 2017 - 3:39:00 PM
Last modification on : Tuesday, May 14, 2019 - 11:05:56 AM
Long-term archiving on : Friday, June 30, 2017 - 3:31:51 PM

File

RR-9049.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01497943, version 1

Citation

Hussam Al Daas, Laura Grigori, Pascal Hénon, Philippe Ricoux. Enlarged GMRES for reducing communication. [Research Report] RR-9049, Inria Paris. 2017. ⟨hal-01497943⟩

Share

Metrics

Record views

439

Files downloads

319