English

Mastering high-dimensional dynamics with Hamiltonian neural networks

Neural and Evolutionary Computing 2020-08-11 v1 Chaotic Dynamics

Abstract

We detail how incorporating physics into neural network design can significantly improve the learning and forecasting of dynamical systems, even nonlinear systems of many dimensions. A map building perspective elucidates the superiority of Hamiltonian neural networks over conventional neural networks. The results clarify the critical relation between data, dimension, and neural network learning performance.

Keywords

Cite

@article{arxiv.2008.04214,
  title  = {Mastering high-dimensional dynamics with Hamiltonian neural networks},
  author = {Scott T. Miller and John F. Lindner and Anshul Choudhary and Sudeshna Sinha and William L. Ditto},
  journal= {arXiv preprint arXiv:2008.04214},
  year   = {2020}
}

Comments

7 pages, 9 figures

R2 v1 2026-06-23T17:45:17.363Z