Jarzynski Equality Counterexample?

TL;DR

This paper challenges the universal validity of the Jarzynski Equality by providing a counterexample involving a harmonic oscillator, demonstrating it only holds in the slow process limit and highlighting misapplications of the Feynman-Kac formula.

Contribution

The paper presents a counterexample showing the Jarzynski Equality fails for rapid processes and clarifies the incorrect application of the Feynman-Kac formula in previous studies.

Findings

JE holds only in the slow process limit

Counterexample with a harmonic oscillator demonstrates failure of JE in rapid processes

Misapplication of Feynman-Kac formula in prior work is identified

Abstract

Almost 25 years ago, Jarzynski published a paper in which it was asserted: the work done, W, in driving a system from state A to state B, characterized by the Helmholtz free energies FA and FB, satisfies an equality in which an average over an ensemble of measurements for W determines the difference in Free energy. Several features of this result require more detailed description, to be given in the text. The equality is significant and unexpected. So is the statement that the equality is independent of the rate of change from state A to state B. A few years ago, I had presented three papers in which the contraction of the description from full phase space to coordinate space only was made. This was motivated by the large difference in time scales for momenta relaxation and coordinate relaxation. The Jarzynski equality (JE) will be shown here to be correct only in the limit of extremely slow processes. The proposed counterexample is for a macromolecular harmonic oscillator with a Hooke's spring constant, k, that during the time interval linearly changes from k1 to k2. When performed very slowly, we obtain JE in a form in which the free energy is related to the ratio of the k's. However, when the transition is performed more rapidly the equality is lost. Jarzynski has suggested to me that Hummer and Szabo have shown that JE follows directly from the Feynman-Kac formula. We show here that F-K does not reproduce the direct calculation executed in this paper and that it has been misapplied in earlier papers.