Performance of Neural and Polynomial Operator Surrogates
Josephine Westermann, Benno Huber, Thomas O’Leary-Roseberry, Jakob Zech · 2026
Read Performance of Neural and Polynomial Operator Surrogates by Josephine Westermann, Benno Huber, Thomas O’Leary-Roseberry, Jakob Zech (2026) 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-2604-00689.txt.
Description
Westermann and colleagues compare neural and polynomial stand-in models that replace costly repeated runs of equations with changing inputs. They test two neural versions, one built on a reduced basis and one using Fourier methods, against two polynomial versions, one sparse-grid and one tensor-train, on heat-spread and stretchy-material problems. No method wins everywhere. Polynomials use data far more efficiently for smooth inputs, while the Fourier method converges fastest for rough inputs.
Catalog record
| CREATOR | Josephine Westermann, Benno Huber, Thomas O’Leary-Roseberry, Jakob Zech |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 111382 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.00689