Discovering a Local Fluctuation Theorem with Machine Learning
© The Physical Society of Japan
This article is on
Machine learning a timelocal fluctuation theorem for nonequilibrium steady states
(PTEP Editors' Choice)
Prog. Theor. Exp. Phys. 2023, 083A01 (2023).
A groundbreaking study reveals a new timelocal fluctuation theorem using machine learning, revolutionizing our understanding of deterministic nonequilibrium steady state systems.
The second law of thermodynamics states that for large systems, there is a direction associated with their evolution over large time scales. However, in the realm of nonequilibrium dynamics, especially where a system is driven far from equilibrium, understanding and quantifying thermodynamic relations is challenging.
In this regard, fluctuation theorems (FTs), which quantify the thermodynamic reversibility of a system in terms of a dissipation function, have emerged as an essential tool for characterizing the probability of observing events in small systems during short timescales that violate the second law of thermodynamics. However, these theorems have only been formulated for a limited class of systems.
Consequently, our understanding of deterministic nonequilibrium steady states—a system with unknown phase space distribution for which only asymptotic FTs hold— remains limited. This is because all attempts to evaluate their dissipation function require either the trajectory segment of interest to be relatively long or information about the entire trajectory around that segment.
In this study, we addressed this longstanding problem in statistical mechanics by utilizing a simple machine learning model. The model was trained to minimize the binary crossentropy loss function and discern whether a segment of a steady state trajectory progresses forward or backward in time. This yielded a function that satisfied an FT and relied only upon the information within the segment of interest.
Remarkably, this timelocal FT is applicable even to extremely short trajectory segments and holds true for systems far from equilibrium as well as for various nonequilibrium dynamics, surpassing all previously known theoretical approximations.
We further demonstrated that any model reaching the global minimum of the binary crossentropy loss function must be a wellcalibrated predictor of time’s arrow and satisfy an FT. This leads us to a new derivation for localtime FTs: an FT can be exactly satisfied for an arbitrarily small set of local information provided the correlation function between that information and the unknown, nonlocal information is known.
Since our theoretical treatment is applicable to feasible realworld measurements, which only capture information from a local region and for a finite duration, it could potentially have wideranging applications in computational modeling as well as realworld data collection.
Consequently, this study could eventually lead to an enhanced understanding of nonequilibrium steady state systems across diverse fields, including fluid dynamics, heat transfer, semiconductor operation, ion transport, biology, and climate modeling.
Machine learning a timelocal fluctuation theorem for nonequilibrium steady states
(PTEP Editors' Choice)
Prog. Theor. Exp. Phys. 2023, 083A01 (2023).
Share this topic
Fields
Related Articles

NetPattern of Muskmelon Rind Favors a Common Mathematical Law
Crossdisciplinary physics and related areas of science and technology
Statistical physics and thermodynamics
2023112
The surface of muskmelon is covered with a fine meshlike netpattern. The geometric features of the fine mesh appear unique for each individual, but hide an unexpected mathematical rule.

High Magnetic Field as a Tool for Discovery in Condensed Matter Physics
Magnetic properties in condensed matter
Dielectric, optical, and other properties in condensed matter
Electronic transport in condensed matter
Structure and mechanical and thermal properties in condensed matter
Electron states in condensed matter
Electronic structure and electrical properties of surfaces and nanostructures
Measurement, instrumentation, and techniques
20221213
The Journal of the Physical Society of Japan highlights in this special topic recent advances in modern physics that have been realized with the generation of pulsed high magnetic fields.

New Phases of Matter: Topological Physics in the 21^{st} Century
Theoretical Particle Physics
Fundamental Theory of Condensed Matter Physics, Statistical Mechanics, Fluid Dynamics,
Superconductivity
2022719
Researchers from Japan highlight in an editorial our current understanding of topological phases of matter, focusing on the classification of the newly discovered symmetry protected topological phases.

Tensor Networks Across Physics
Mathematical methods, classical and quantum physics, relativity, gravitation, numerical simulation, computational modeling
Statistical physics and thermodynamics
Magnetic properties in condensed matter
202267
Researchers from Japan provide the first comprehensive review of the historical development of tensor networks from a statistical mechanics viewpoint, with a focus on its theoretical background.

Violation of FluctuationDissipation Theorem Results in Robustness of Fluctuation Against Localization
Statistical physics and thermodynamics
Electronic transport in condensed matter
Electronic structure and electrical properties of surfaces and nanostructures
2022323
We study equilibrium current fluctuations in systems without timereversal symmetry, violating the fluctuationdissipation theorem. Notably, the offdiagonal fluctuation is insensitive to system imperfections in contrast to other fluctuations and conductivity.