Spectral Normalization and Voigt-Reuss net: A universal approach to microstructure-property forecasting with physical guarantees
Sanath Keshav, Julius Herb, Felix Fritzen*, S. Keshav, J. Herb, F. Fritzen · 2025
Read Spectral Normalization and Voigt-Reuss net: A universal approach to microstructure-property forecasting with physical guarantees by Sanath Keshav, Julius Herb, Felix Fritzen*, S. Keshav, J. Herb, F. Fritzen (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-2504-00712.txt.
Description
Keshav, Herb, and Fritzen predict material properties with networks that respect physical bounds in twenty twenty-five. Mixed materials need fast predictions of stiffness, conductivity, and similar traits. Simulations are accurate but costly and hard to get gradients from. Machine learning surrogates are fast yet can predict values that cannot occur.
Classical theory bounds the response of any mix between Voigt and Reuss limits. Their spectral normalisation forces predictions inside those bounds by design. The trick works with any features and any model of that kind. It applies to elasticity, heat conduction, and permeability tensors alike. A test network beat standard neural networks on large data sets. Accuracy and robustness rose, and the gains did not hinge on which features were used. The authors offer physical guarantees rare in machine learning.
Catalog record
| CREATOR | Sanath Keshav, Julius Herb, Felix Fritzen*, S. Keshav, J. Herb, F. Fritzen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 81 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-00712