The paper [1] by Liu, Madhavan, and Tegmark sought to use machine learning methods to elicit known conservation laws for several systems. However, in their example of a damped 1D harmonic oscillator they made seven serious errors, causing both their method and result to be incorrect. In this Comment, those errors are reviewed.
Cite
@article{arxiv.2404.02896,
title = {Comment on "Machine learning conservation laws from differential equations"},
author = {Michael F. Zimmer},
journal= {arXiv preprint arXiv:2404.02896},
year = {2025}
}
Comments
3 pages, 1 figure; comment on https://doi.org/10.1103/PhysRevE.106.045307. This update now includes: Error#7, Fig1, Eqn5, a new paragraph in Afterword