Training thermodynamic computers by gradient descent
Proceedings of the National Academy of Sciences · 07.04.2026 · 123 (14)
DOI: 10.1073/pnas.2528413123 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 03.04.2026. Verfasst von Stephen Whitelam. Beteiligte Einrichtung: Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, CA 94720. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: DOE | SC | Basic Energy Sciences (BES).
Abstract (Originalsprache)
We show how to adjust the parameters of a thermodynamic computer by gradient descent in order to perform a desired computation at a specified observation time. Within a digital simulation of a thermodynamic computer, training proceeds by maximizing the probability with which the computer would generate an idealized dynamical trajectory. The idealized trajectory is designed to reproduce the activations of a neural network trained to perform the desired computation. This teacher-student scheme results in a thermodynamic computer whose finite-time dynamics enacts a computation analogous to that of the neural network. The parameters identified in this way can be implemented in the hardware realization of the thermodynamic computer, which will perform the desired computation automatically, driven by thermal noise. We demonstrate the method on a standard image-classification task, and estimate the thermodynamic advantage—the ratio of energy costs of the digital and thermodynamic implementations—to exceed seven orders of magnitude. Our results establish gradient descent as a viable training method for thermodynamic computing, enabling application of the core methodology of machine learning to this emerging field.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Molecular Foundry, Lawrence Berkeley National Laboratory, Berkeley, CA 94720.
ORCID-Profile: Stephen Whitelam
Förderung
- DOE | SC | Basic Energy Sciences (BES)
Zitieren
APA
Whitelam, S. (2026). Training thermodynamic computers by gradient descent. Proceedings of the National Academy of Sciences, 123(14). https://doi.org/10.1073/pnas.2528413123
Vancouver
Whitelam S. Training thermodynamic computers by gradient descent. Proceedings of the National Academy of Sciences. 2026;123(14). doi:10.1073/pnas.2528413123
BibTeX
@article{Whitelam2026,
author = {Whitelam, Stephen},
title = {Training thermodynamic computers by gradient descent},
journal = {Proceedings of the National Academy of Sciences},
year = {2026},
volume = {123},
number = {14},
doi = {10.1073/pnas.2528413123}
}
EUVIMED