Geometric Stochastic Fluctuation Relations

• Geometric stochastic fluctuation relations are a framework linking trajectory geometry with stochastic thermodynamics and information theory through Riemannian metrics and symmetry transformations.
• The approach employs the stochastic Fisher information metric to quantify entropy-space distances along trajectories, thereby extending traditional fluctuation theorems into geometric domains.
• Key results include refined uncertainty relations and entropy production bounds, with demonstrated applications in Brownian motion, quantum dots, and driven mesoscopic systems.

Geometric stochastic fluctuation relations constitute a unifying framework connecting the geometry of trajectory space, stochastic thermodynamics, and information theory. They provide fluctuation theorems and uncertainty principles that are structurally governed by geometric objects, such as Riemannian metrics on trajectory or configuration manifolds, and symmetry-induced transformations in path spaces. These relations give rise to refined constraints on fluctuations, dissipation, and irreversibility in both classical and quantum stochastic systems, accounting for geometry beyond the standard role of time-reversal.

1. Stochastic Metrics, Fluctuation Geometry, and Stochastic Fisher Information

A foundational concept in geometrizing fluctuation relations is the association of a Riemannian metric to the space of probability distributions or to stochastic trajectories. In classical settings, the Fisher information metric provides a distinguished choice, encoding the sensitivity of probability measures to parameter variations and defining an infinitesimal statistical distance. For a time-dependent Markovian process with density $P(x,t)$, the instantaneous (time-parametrized) Fisher information $\mathcal{I}_F(t)$ can be extended at the level of individual trajectories to the stochastic Fisher information (SFI): $$\iota_F(x,t) = [\partial_t \ln P(x,t)]^2 = [\partial_t s_{\rm sys}(x,t)]^2,$$ where $s_{\rm sys}(x,t) = -\ln P(x,t)$ is the stochastic system entropy. The SFI equips each trajectory with a fluctuating local metric,

$d\ell^2(x,t) = \iota_F(x,t) dt^2,$

and the total stochastic length,

$$\ell[x(\cdot)] = \int_0^\tau dt\, |\partial_t s_{\rm sys}(x(t),t)|,$$

interprets as the accumulated "distance" in entropy space traversed by the realization $x(\cdot)$ (Melo et al., 2024).

The general program of fluctuation geometry constructs a Riemannian manifold $\mathcal{M}$ from a parametric family of distributions $dp(x|\theta)$, with metric $g_{ij}(x|\theta)$ defined via the covariant Hessian $-D_iD_j \log \rho(x|\theta)$ and further endowed with the Levi-Civita connection and curvature tensor $R_{ijkl}$. Non-vanishing curvature reflects irreducible statistical correlations; the scalar curvature $R(x|\theta)$ quantifies departures from Gaussianity (Velazquez, 2013, Velazquez, 2011).

2. Geometric Fluctuation Relations: Trajectory-level Theorems

Fluctuation relations incorporating geometric metrics generalize nonequilibrium fluctuation theorems by recognizing the role of path-dependent geometric quantities. For the SFI metric, two primary forms arise:

• Detailed Geometric Fluctuation Relation:

$$\frac{P_F(\iota_F)}{P_B(\hat\iota_F)} = \exp\{\beta q[x(\cdot)] + \ell[x(\cdot)]\},$$

where $P_F$ and $P_B$ are the forward and backward SFI sequence distributions along $x(\cdot)$ and its time-reversal, $q[x(\cdot)]$ is the stochastic heat, and $\ell[x(\cdot)]$ the entropy-space length (Melo et al., 2024). This extends Crooks’ theorem by incorporating the trajectory-specific metric.

• Integral Geometric Fluctuation Relation:

$\langle \exp\{\beta q + \ell\} \rangle_F = 1,$

yielding, by Jensen's inequality, the upper bound $$\langle \ell \rangle \le -\langle \beta q \rangle = -\Delta S_{\rm bath}$$. The average entropic distance cannot exceed the entropy delivered to the heat bath, effectively recasting the second law in geometric terms (Melo et al., 2024).

Analogous geometric extensions have been developed in the context of spatial symmetries or non-involutive bijections acting on trajectory space. For any invertible trajectory transformation $\mathcal{R} : \Omega \to \Omega$, the ratio of path probabilities satisfies

$$\frac{P[\gamma]}{P[\mathcal{R}\gamma]} = \exp(\Sigma[\gamma]),$$

with $\Sigma[\gamma]$ the geometric entropic functional, which becomes linear in the presence of a covariant observable $A_T$ and matrix symmetry $R$, $\Sigma[\gamma] = (w, A_T[\gamma])$ (Chetrite et al., 20 Mar 2025, Marcantoni et al., 2020).

3. Symmetry-Induced and Isometric Fluctuation Relations

Beyond time-reversal, dynamical symmetries (e.g., spatial translations, rotations, or permutations) induce geometric fluctuation relations for suitably defined observables. For a continuous-time Markov process invariant under a configuration-space bijection $R: X \to X$, and observable $K(\omega_t)$ covariant under $K(\mathcal{R}\omega_t) = U K(\omega_t)$, one obtains finite- and long-time fluctuation relations at the level of generating functions and rate functions (Marcantoni et al., 2020, Chetrite et al., 20 Mar 2025).

Isometric fluctuation relations generalize further to rotations in observable space: $$\frac{P^{\text{cano},s}[K_T = a]}{P^{\text{cano},s}[K_T = Ra]} = \exp[(I - R^T)s, R^{-1}a],$$ with $R$ an isometry and $s$ the bias field (Chetrite et al., 20 Mar 2025). These geometric FRs capture the transformation behavior of vector-valued path observables under group actions, and hold under broad conditions for both finite-state Markov chains and diffusions.

Geometric fluctuation relations also arise in non-stationary, cyclically driven systems where geometric phases (Berry–Sinitsyn–Nemenman curvature) dominate current fluctuations, producing cyclic FRs with non-Gaussian signatures (Hino et al., 2019, Watanabe et al., 2017). In quantum and interacting settings, geometric FRs reflect coupling between charge and energy statistics, with interaction effects (e.g., Coulomb blockade) directly quantifiable through geometric corrections in the FR (Riwar et al., 2020).

4. Trajectory-Dependent Uncertainty Relations and Non-Gaussian Effects

The stochastic geometric approach enables trajectory-level uncertainty relations and bounds on entropy production rates. For an individual realization, the Cauchy–Schwarz inequality gives

$$j[x(\cdot)] = \tau \int_0^\tau dt\, \iota_F(x(t),t) \ge [\ell[x(\cdot)]]^2,$$

with $j$ the trajectory action and $\ell$ the stochastic length. The normalized variance, $$\sigma^2[x(\cdot)] = j[x(\cdot)] - \ell[x(\cdot)]^2$$, quantifies the dispersion of the information rate; ensemble averaging yields a lower bound on information rate fluctuations reminiscent of thermodynamic uncertainty relations for single-trajectory metrics (Melo et al., 2024).

In periodically driven pumps and nonequilibrium quantum systems, geometric fluctuation relations produce high-order corrections (e.g., cubic terms in $J$), resulting in non-Gaussian tails in the current distribution, directly reflecting geometric phases in parameter space (Hino et al., 2019, Watanabe et al., 2017).

5. Applications and Model Systems

Geometric stochastic fluctuation relations are illustrated and validated in a broad array of model systems:

Model/System	Geometric Quantity	Key Fluctuation Relation/Effect
Overdamped Brownian motion	SFI, entropy-space length $\ell$	$\langle e^{\beta q+\ell}\rangle = 1$ (Melo et al., 2024)
Harmonic traps in fast motion	SFI, breakdown of integral FR in non-quasistatic regime	$\langle e^{\beta q+\ell}\rangle \neq 1$ away from $v\to 0$
Spin–boson quantum pumps	Geometric Berry-Sinitsyn-Nemenman curvature	Extended FT with geometric, non-Gaussian corrections (Watanabe et al., 2017)
Interacting quantum dot	Geometric gauge term in full counting statistics	Joint FR for charge/energy with interaction-induced geometric correction (Riwar et al., 2020)
Markov chains with spatial symmetries	Isometric and translation FRs; vector observables	FRs for swapping even/odd bonds; spatial rotation covariance (Chetrite et al., 20 Mar 2025)

Experimental implications encompass single-electron charge pumps, mesoscopic heat engines, cold-atom transport, and optical-tweezer setups with time-resolved density measurements, where geometric corrections to standard fluctuation and dissipation relations can be directly accessed (Hino et al., 2019, Watanabe et al., 2017, Melo et al., 2024).

6. Conceptual and Mathematical Implications

Geometric stochastic fluctuation relations recast the structure of nonequilibrium statistical mechanics, exposing an intrinsic connection between fluctuating geometric quantities (metrics, curvature) and thermodynamic irreversibility. Scalar and tensorial curvatures in fluctuation geometry diagnose irreducible correlations and control the Gaussianity of fluctuations (Velazquez, 2013). Geometric FRs illuminate the role of symmetry, topology, and parameter-space structure in dictating the constraints and trade-offs of stochastic flows.

A plausible implication is a geometric classification of fluctuation phenomena—distinct types of curvatures and symmetries encode separate families of fluctuation relations, bounds, and uncertainty principles. The scope of geometric stochastic fluctuation relations encompasses classical and quantum systems, covering both Markovian and non-Markovian dynamics, and extends naturally to systems with time-dependent driving, vectorial observables, and non-involutive transformations of trajectory space.