English

Preconditioning and Numerical Stability in Neural Network Training for Parametric PDEs

Numerical Analysis 2026-02-02 v1 Numerical Analysis

Abstract

In the context of training neural network-based approximations of solutions of parameter-dependent PDEs, we investigate the effect of preconditioning via well-conditioned frame representations of operators and demonstrate a significant improvement on the performance of standard training methods. We also observe that standard representations of preconditioned matrices are insufficient for obtaining numerical stability and propose a generally applicable form of stable representations that enables computations with single- and half-precision floating point numbers without loss of precision.

Keywords

Cite

@article{arxiv.2601.23185,
  title  = {Preconditioning and Numerical Stability in Neural Network Training for Parametric PDEs},
  author = {Markus Bachmayr and Wolfgang Dahmen and Chenguang Duan and Mathias Oster},
  journal= {arXiv preprint arXiv:2601.23185},
  year   = {2026}
}
R2 v1 2026-07-01T09:28:05.418Z