Deep learning of thermodynamic laws from microscopic dynamics

TL;DR

This paper demonstrates that deep neural networks can learn and encode thermodynamic laws, such as entropy and adiabatic accessibility, directly from microscopic molecular dynamics data without prior physical assumptions.

Contribution

It shows that a DNN can discover emergent thermodynamic principles from microscopic simulations, bridging microscopic dynamics and macroscopic laws.

Findings

DNN learns an order relation consistent with adiabatic processes

Internal DNN representations act as entropy measures

Machine learning can uncover macroscopic physical laws from microscopic data

Abstract

We numerically show that a deep neural network (DNN) can learn macroscopic thermodynamic laws purely from microscopic data. Using molecular dynamics simulations, we generate the data of snapshot images of gas particles undergoing adiabatic processes. We train a DNN to determine the temporal order of input image pairs. We observe that the trained network induces an order relation between states consistent with adiabatic accessibility, satisfying the axioms of thermodynamics. Furthermore, the internal representation learned by the DNN act as an entropy. These results suggest that machine learning can discover emergent physical laws that are valid at scales far larger than those of the underlying constituents -- opening a pathway to data-driven discovery of macroscopic physics.