What is a Fluctuation Theorem?

This book provides a modern review of Fluctuation Relations and Fluctuation Theorems in nonequilibrium statistical mechanics. It focuses on the pioneering perspectives of Gallavotti and Cohen, according to which a fluctuation theorem describes the statistics of the deviations of entropy production from its expected value. For time-reversal invariant systems, these fluctuations obey a universal (i.e., model-independent) symmetry called the fluctuation relation. The probabilistic framework introduced in the first part of the book allows for a very general formulation of Fluctuation Relations and Theorems for both deterministic and stochastic dynamical systems. The authors further explore models of physical interest, illustrating this framework by concrete applications. The second part of the book focuses on chaotic dynamics. The formulation of two general Fluctuation Theorems, followed by the detailed study of a concrete example, provides the reader with an understanding of both the theoretical and practical aspects of the subject.

💡 Research Summary

The monograph offers a comprehensive and mathematically rigorous treatment of fluctuation relations (FR) and fluctuation theorems (FT) in nonequilibrium statistical mechanics. It begins with a historical overview, tracing the development from early fluctuation‑dissipation ideas (Einstein‑Smoluchowski, Johnson‑Nyquist, Onsager) through the pioneering works of Bochkov‑Kuzovlev, Evans‑Cohen‑Morris, and finally the Gallavotti‑Cohen chaotic‑hypothesis framework. The authors emphasize that FRs describe a universal symmetry of the probability distribution of a time‑reversal‑odd observable—most commonly the entropy production rate—while FT refers to the precise mathematical statement of this symmetry.

The core of the book is built on large deviation theory. After introducing the Large Deviation Principle (LDP) and the associated rate function (I(s)), the authors show that the transient (finite‑time) FR follows directly from the Radon‑Nikodym derivative between the path measure and its time‑reversed counterpart, yielding \