lrAA: Low-Rank Anderson Acceleration
Daniel Appelö, Yingda Cheng · 2025
Read lrAA: Low-Rank Anderson Acceleration by Daniel Appelö, Yingda Cheng (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-03909.txt.
Description
Appelo and Cheng accelerate low-rank solutions of nonlinear matrix equations with a new Anderson method, in a paper from twenty twenty-five. Such equations arise when nonlinear partial differential equations are discretized in space. Full solutions are huge, so low-rank forms are essential. Their method, lrAA, performs every step in low-rank form.
An adaptive tolerance schedule trims rank as residuals guide the process. A new cross approximation handles nonlinear terms with error control. It warm-starts each round from the last approximation for speed. Tests span Laplace, Bratu, Monge-Ampere, and Allen-Cahn problems. The method beats rivals on speed and accuracy across the set. It controls both rank growth and iteration counts well.
Catalog record
| CREATOR | Daniel Appelö, Yingda Cheng |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 56 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-03909