Violation of thermodynamic speed limits as a criterion for physical model selection
Open Access
- 28 May 2026
- journal article
- Published by IOP Publishing in Machine Learning: Science and Technology
- Vol. 7 (3) , 035035
Abstract
Machine learning algorithms are now being used to automate the discovery of physical equations from measurement data. However, under or overfitting the dynamics can lead to model misspecification and unreliable predictions. Here we show that thermodynamic speed limits–fundamental bounds on the rates of heat, entropy, and other observables–provide a diagnostic: if essential dynamics are omitted, observable speeds can appear to exceed the bound. These apparent violations signal missing physics and a structurally incomplete description of the dynamics. Using an underdamped Langevin oscillator and an active Brownian particle as benchmarks, we prove that neglecting inertia or activity leads to observable rates that exceed the limit, in some cases indefinitely. Because true violations are impossible, speed-limit violations are a criterion for rejecting physically inconsistent representations of nonequilibrium processes.Keywords
Funding Information
- National Science Foundation (2231469)
- U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Funding for Accelerated, Inclusive Research (DE-SC-0024305)
This publication has 50 references indexed in Scilit:
- Non-Equilibrium Dynamics and Non-Gaussian Fluctuations of an Optical Matter System Manifesting PseudorotationACS Nano, 2025
- Optimal design of experiments for optimization-based model calibration using Fisher information matrixReliability Engineering & System Safety, 2021
- Variational Approach for Learning Markov Processes from Time Series DataJournal of Nonlinear Science, 2019
- Author Correction: VAMPnets for deep learning of molecular kineticsNature Communications, 2018
- Acoustic trapping of active matterNature Communications, 2016
- Brownian motion at short time scalesAnnalen der Physik, 2013
- Damping of Acoustic Vibrations of Single Gold Nanoparticles Optically Trapped in WaterNano Letters, 2012
- Resonances arising from hydrodynamic memory in Brownian motionNature, 2011
- Direct observation of the full transition from ballistic to diffusive Brownian motion in a liquidNature Physics, 2011
- Dynamic mode decomposition of numerical and experimental dataJournal of Fluid Mechanics, 2010