LGApr 3, 2024
Comment on "Machine learning conservation laws from differential equations"
arXiv:2404.02896v32 citationsh-index: 1
Originality Synthesis-oriented
AI Analysis
This is an incremental correction for researchers in computational physics and machine learning, addressing methodological flaws in prior work.
The paper critiques a previous study for seven serious errors in applying machine learning to find conservation laws, specifically in the case of a damped harmonic oscillator, leading to incorrect methods and results.
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.