English

Achieving High Accuracy with PINNs via Energy Natural Gradients

Machine Learning 2023-08-16 v2 Numerical Analysis Numerical Analysis

Abstract

We propose energy natural gradient descent, a natural gradient method with respect to a Hessian-induced Riemannian metric as an optimization algorithm for physics-informed neural networks (PINNs) and the deep Ritz method. As a main motivation we show that the update direction in function space resulting from the energy natural gradient corresponds to the Newton direction modulo an orthogonal projection onto the model's tangent space. We demonstrate experimentally that energy natural gradient descent yields highly accurate solutions with errors several orders of magnitude smaller than what is obtained when training PINNs with standard optimizers like gradient descent or Adam, even when those are allowed significantly more computation time.

Keywords

Cite

@article{arxiv.2302.13163,
  title  = {Achieving High Accuracy with PINNs via Energy Natural Gradients},
  author = {Johannes Müller and Marius Zeinhofer},
  journal= {arXiv preprint arXiv:2302.13163},
  year   = {2023}
}

Comments

Published version