🔬 condensed matter

Exact Fluctuation-Response Relations for Underdamped Langevin Dynamics

This paper establishes an exact finite-time fluctuation-response equality for underdamped Langevin dynamics that reveals response, rather than mean current, as the fundamental link between fluctuations and dissipation, enabling a variational characterization of entropy production and overcoming the limitations of conventional thermodynamic uncertainty relations.

Original authors: Tan Van Vu, Van Tuan Vo, Ruicheng Bao, Keiji Saito

Published 2026-08-21

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Tan Van Vu, Van Tuan Vo, Ruicheng Bao, Keiji Saito

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the chaotic world of heat and motion, where tiny particles jitter and drift, there is a fundamental rule that links how much energy is wasted to how precisely a system can perform a task. This rule, known as the thermodynamic uncertainty relation, suggests that to make a current of particles flow with perfect steadiness, a system must pay a steep price in the form of wasted heat. For decades, scientists have used this principle to understand everything from the motors inside living cells to the flow of electricity in circuits. The logic seemed straightforward: if you want less randomness in the flow, you must generate more entropy, or disorder. However, this rule was built on assumptions that worked well for slow, sluggish systems but began to crack when applied to objects that have mass and momentum, like a ball rolling down a hill or a nanoparticle suspended in air. In these faster, underdamped systems, the old rules sometimes failed, leaving a gap in our understanding of how energy, motion, and randomness truly connect.

A team of researchers has now filled this gap by deriving a new, exact mathematical framework that works for these faster systems at any moment in time, not just after a long wait. They discovered that the key to understanding the cost of precision in these moving systems is not the average speed of the flow itself, but rather how sensitive that flow is to changes in friction. By treating the system's response to a slight tweak in its resistance as the central quantity, they found a relationship that holds true perfectly, even when the system is being pushed and pulled by external forces. This new equality acts as a precise ruler, allowing scientists to measure the total energy wasted by a system simply by observing how its motion changes when the friction is adjusted, a method that works even for very short observation times where previous methods failed.

The researchers arrived at this insight by looking at the mathematics of probability in a new way. They focused on a specific type of motion where a particle's velocity is directly affected by random kicks from its environment, but its position changes based on that velocity. In such systems, the randomness of the particle's path and the average direction it travels often probe different aspects of the physics, making the old rules unreliable. To solve this, the team developed a method that tracks the expected future contribution of a particle's journey from any given moment. They showed that the total variation, or jitter, in a particle's path can be split into two parts: the uncertainty inherited from where the particle started, and the new uncertainty generated by the noise it encounters as it moves. Crucially, they proved that this newly generated jitter is exactly equal to the square of how much the average flow changes when the friction is slightly altered.

This finding leads to a powerful new tool for measuring energy loss. The researchers demonstrated that if you choose to measure how the system responds to a change in friction, you get a bound on the energy wasted that is always true and can be made perfectly tight. In other words, there is always a specific way to measure the flow such that the relationship between the jitter and the energy cost becomes an exact equality. This allows for a precise calculation of the total entropy produced, which is the measure of wasted energy, simply by analyzing the statistics of the particle's motion and its sensitivity to friction. This is a significant improvement over older methods, which often provided only loose estimates or failed entirely when the observation time was short.

Perhaps the most striking result of this work is what happens when the researchers tested it on a simple, driven system where a particle diffuses freely under a constant force. They found that for certain types of measurements that pay attention to the particle's velocity, the old rule of thumb breaks down completely. In these cases, the uncertainty factor—the ratio of jitter to the square of the average flow—can drop exponentially as the energy cost increases, suggesting that the system becomes incredibly precise with very little waste. This seemed to violate the fundamental cost-precision trade-off. However, when the researchers applied their new friction-based rule to the same system, the violation disappeared. The new rule held firm, showing that while the average flow might become a poor indicator of precision, the system's sensitivity to friction remains a reliable and universal guide.

The implications of this work extend beyond theoretical physics. The researchers showed that their method is not just a mathematical curiosity but is experimentally accessible. In real-world setups, such as a tiny particle trapped by light in a gas or an electrical circuit with thermal noise, the friction and the random noise are linked by a fundamental law. By slightly changing the pressure of the gas or the resistance in the circuit, scientists can alter the friction and the noise together. Measuring how the particle's motion responds to this change allows for a direct and exact calculation of the energy dissipated, even in systems that are far from equilibrium. This provides a new, robust way to infer the thermodynamic cost of processes in nanoscale devices and biological motors, where measuring heat directly is often impossible.

Ultimately, this study reshapes our understanding of the relationship between fluctuations and dissipation in moving systems. It establishes that in the realm of underdamped dynamics, the quantity that directly links the randomness of motion to the cost of energy is not the mean current itself, but the response of that current to changes in friction. By shifting the focus from the average flow to the system's sensitivity, the researchers have uncovered a structure that is exact, universal, and valid at any moment in time. This framework unifies the concepts of fluctuation, response, and thermodynamic inference, offering a clear path forward for understanding the fundamental limits of precision in the physical world.

Technical Summary: Exact Fluctuation-Response Relations for Underdamped Langevin Dynamics

Problem Statement
Thermodynamic uncertainty relations (TURs) establish a fundamental trade-off between the precision of a current and the thermodynamic cost (dissipation) required to sustain it. While these relations have been extensively developed for overdamped systems and Markov jump processes, their application to underdamped Langevin dynamics presents significant challenges. In underdamped systems, thermal noise acts directly only on velocity degrees of freedom, while experimentally relevant transport currents often accumulate through position. This mismatch causes conventional mean-current TURs to fail or require modification, as fluctuations and currents probe different sectors of phase space. Furthermore, physically relevant perturbations in underdamped dynamics (such as varying the friction coefficient) simultaneously alter both the drift and the diffusion terms, rendering standard likelihood-ratio approaches—which typically assume fixed noise strength—insufficient. The central question addressed is whether fluctuations, responses, and dissipation can be unified within a single exact framework for underdamped dynamics that remains valid at arbitrary finite times.

Methodology
The authors develop a framework based on the Doob martingale associated with general finite-time additive observables. They consider a $d$ -dimensional underdamped Langevin system driven by time-dependent forces and subject to additive noise. The key steps in their methodology include:

Martingale Representation: For a general additive observable $J_\tau = \int_0^\tau [\phi_t(z_t)dt + w_t(z_t) \circ dv_t]$ , the authors define a conditional expectation $G_t(z)$ representing the future contribution to the observable. They construct the Doob martingale $M_t = J_t + G_t(z_t)$ and apply Itô's formula to derive an exact differential representation of the martingale in terms of the Wiener process.
Exact Variance Identity: By integrating the martingale representation and applying the Itô isometry, they derive an exact decomposition of the observable's variance. This decomposition separates the variance into a contribution inherited from the initial ensemble ( $Var_0$ ) and a dynamically generated contribution expressed as a quadratic form of a local response kernel.
Fluctuation-Response Connection: The authors connect the martingale amplitude to the linear response of the observable average under an infinitesimal source injected into the velocity-space probability current. This allows them to express the dynamically generated variance exactly as the squared local susceptibility measured in the diffusion metric.
Variational Characterization: By optimizing over perturbation fields, they establish a dual variational representation for the dynamically generated variance.
Specific Perturbation Analysis: They specifically analyze the perturbation corresponding to the irreversible probability flow (friction response), where the perturbation cost equals the total entropy production.

Key Contributions and Results

Exact Finite-Time Fluctuation-Response Equality: The paper derives an exact identity (Eq. 10) linking the dynamically generated variance of an arbitrary additive observable to the squared local response kernel. This equality holds for arbitrary initial ensembles, time-dependent driving, and without imposing time-reversal parity constraints.
Sharp Response Bounds and Variational Principle: The equality yields sharp lower bounds on the dynamically generated variance for specific perturbation directions. Crucially, the authors show that the bound is tight and can be saturated at any finite observation time by choosing an appropriate observable. This leads to an exact variational principle for the total entropy production (Eq. 17).
Friction-Response Thermodynamic Uncertainty Relation: By selecting the irreversible probability flow as the perturbation, the authors derive a friction-response TUR (Eq. 16). Unlike conventional TURs, which relate variance to the square of the mean current, this relation relates the dynamically generated variance to the square of the friction susceptibility ( $\gamma \partial_\gamma \langle J_\tau \rangle$ ). This bound is valid for any additive observable and is saturable at arbitrary finite times.
Exponential Violation of Conventional TUR: Using the exact long-time representation, the authors demonstrate that for driven free diffusion, conventional TURs can be violated arbitrarily strongly. For specific velocity-resolved antisymmetric currents, the conventional uncertainty factor decays exponentially with the thermodynamic cost (dissipation), even in linear Gaussian dynamics. This violation arises from the nonlinear velocity-resolved readout, not from nonlinearity in the underlying dynamics.
Robustness of Friction-Response TUR: In contrast to the conventional TUR, the friction-response TUR retains its universal lower bound (Eq. 21) even in regimes where the conventional factor vanishes. This confirms that the friction susceptibility, rather than the mean current, is the quantity that correctly links fluctuations and dissipation in underdamped systems.

Significance
The paper establishes that response, rather than the mean current itself, is the fundamental quantity linking fluctuations and dissipation in underdamped dynamics. The derived framework unifies fluctuation-response relations, precision bounds, and thermodynamic inference for underdamped systems. By providing an exact variational principle for entropy production that is saturable at finite times, the work offers a rigorous operational tool for thermodynamic inference. The results clarify the limitations of conventional TURs in underdamped regimes and provide a corrected, universally valid alternative based on friction susceptibility. The findings are applicable to experimental platforms where friction and thermal noise can be varied together, such as optically levitated nanoparticles in gases or electrical RLC circuits.

More like this