A nonequilibrium distribution for stochastic thermodynamics
Jean-Luc Garden
TL;DR
This work introduces an extended Gibbs canonical distribution by incorporating an internal nonequilibrium variable $\xi(t)$, enabling a microscopic description of work and uncompensated heat within stochastic thermodynamics. By allowing energy levels $E_i(\lambda,\xi)$ to depend on both a control parameter $\lambda$ and the internal state $\xi$, and defining the stochastic work $\delta W_i=(\partial E_i/\partial \lambda)_{\xi} d\lambda$ and the stochastic uncompensated heat $\delta Q'_i=A_i d\xi$ with $A_i=- (\partial E_i/\partial \xi)_{\lambda}$, the framework recovers macroscopic laws upon averaging and yields new nonequilibrium identities. It derives a Jarzynski-type nonequilibrium work relation, $\overline{e^{-\beta W_i}}=e^{-\beta\Delta F^{eq}}$, and a corresponding nonequilibrium heat relation, $\overline{e^{-\beta Q'_i}}=1$, showing that the distributions of work, heat, and uncompensated heat are interdependent through entropy fluctuations. The approach provides microscopic insight into entropy production and connects stochastic fluctuations to classical thermodynamic quantities, with potential experimental relevance for probing energetic and entropic changes in driven small systems.
Abstract
The canonical distribution of Gibbs is extended to the case of systems outside equilibrium. The distribution of probabilities of a discrete energy levels system is used to provide a microscopic definition of work, along with a microscopic definition of the uncompensated heat of Clausius involved in nonequilibrium processes. The later is related to the presence of non-conservatives forces with regards to the variation of the external parameters. This new framework is used to investigate the nonequilibrium relations in stochastic thermodynamics. A new relation is derived for the random quantity of heat associated to the nonequilibrium work protocol. We finally show that the distributions of probabilities of work, heat and uncompensated heat are non-independent each other during a nonequilibrium process.
