Invited Talk by S. K. Manikandan at Chalmers Condensed Matter Seminar Series

EquiNET uses a graph neural network to infer entropy production from non-equilibrium molecular trajectories and uses this information to estimate equilibrium free-energy differences. (Image by S. K. Manikandan.)

Nonequilibrium Fluctuations as Probes of Thermodynamics
Date: 6 October 2026
Time: 11:00
Place: PJ seminar room, Fysik Origo

The first and second laws of thermodynamics provide the basic principles for describing energy transformations and the emergence of the arrow of time in physical systems. Heat dissipation and entropy production, central quantities in these laws, are often difficult to quantify precisely in experiments because the associated energy exchanges are distributed over a vast number of typically inaccessible degrees of freedom in the environment. This is especially true at microscopic scales, where the relevant energy scales are often of the order of kB T, comparable to thermal fluctuations in the environment, and can lie several orders of magnitude below the sensitivity limits of state-of-the-art room-temperature calorimetry.

Recent advances in nonequilibrium statistical physics and data-driven inference provide new approaches for overcoming these challenges. In this talk, I will first introduce our theoretical results showing how the time dependence of dynamical fluctuations of a system (due to interactions with its environment) can be exploited to quantify dissipation at scales inaccessible to calorimetry. I will then demonstrate how these measurements can be used to address a long-standing challenge related to inferring equilibrium free-energy differences from nonequilibrium measurements. The results will be illustrated using experiments and simulations of biophysical systems, demonstrating both the practical applicability of the method and the challenges that remain.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.