🔢 mathematics

Hypocoercivity for a thermostatted Boltzmann equation

This paper establishes global well-posedness and exponential convergence to a unique non-Maxwellian stationary state for a thermostatted Boltzmann equation on a 3D torus with a spatially varying temperature reservoir, demonstrating that while self-collisions can initially increase relative entropy, the system ultimately relaxes exponentially under uniform Gaussian bounds.

Original author: Josephine Evans, Alexander Pilakoutas

Published 2026-11-06

📖 5 min read🧠 Deep dive

CC BY 4.0

Original author: Josephine Evans, Alexander Pilakoutas

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 author. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a gas trapped inside a box, a collection of countless tiny particles bouncing off one another and the walls. In the quiet, undisturbed state of equilibrium, these particles settle into a predictable pattern where their speeds and positions follow a specific statistical rule, much like how a crowd of people might spread out evenly in a room. This is the standard picture of a gas at rest. But nature is rarely so still. Often, gases are subjected to external forces or temperature differences that keep them in a constant state of flux, pushing energy from one place to another. This is the realm of non-equilibrium, where a steady flow of heat or matter is maintained not by a temporary disturbance, but by a continuous, unchanging setup. Understanding how such systems behave is crucial for everything from designing efficient engines to modeling the atmosphere, yet the mathematics required to describe them is notoriously difficult because the usual tools for analyzing gases break down when the system is not in perfect balance.

In a new study, researchers Josephine Evans and Alexander Pilakoutas have tackled a specific, challenging version of this problem: a gas on a three-dimensional torus, which can be thought of as a space that loops back on itself in all directions, like a video game world where you exit one side and reappear on the other. In this model, the gas particles collide with each other, but they also interact with a thermal reservoir, a kind of invisible thermostat that heats or cools the gas depending on where the particles are located. Crucially, this thermostat is not uniform; it is hotter in some regions and colder in others. This spatial variation drives a constant flow of energy through the gas, creating a non-equilibrium steady state. The researchers wanted to know if such a state actually exists, whether it is unique, and how the gas behaves as it settles into this state.

The team proved that for a wide range of conditions, a unique steady state does indeed exist. They showed that if the gas starts with a certain distribution of particles, it will eventually settle into a specific, unchanging pattern, and it will do so at an exponential rate, meaning the distance to the final state shrinks rapidly over time. However, this steady state is not the simple, predictable pattern seen in equilibrium. Because the temperature of the reservoir changes from place to place, the final distribution of the gas cannot be a simple local Maxwellian, which is the standard term for the equilibrium distribution of a gas at a single temperature. Instead, the gas develops a complex structure that reflects the varying temperature of its environment.

One of the most striking findings concerns the behavior of entropy, a measure of disorder or randomness in the system. In many physical processes, entropy tends to increase or stay the same as a system evolves, a principle known as the second law of thermodynamics. However, the researchers discovered that in this specific setup, the relative entropy—a measure of how far the current state is from the final steady state—can actually increase for a short time, even if the gas starts very close to the final state. This happens because the particles are colliding with each other while also interacting with the varying temperature reservoir. These self-collisions can initially push the system slightly further away from its destination before the overall forces of the system pull it back in. This temporary increase in disorder is a subtle but important feature of non-equilibrium systems that had not been fully characterized in this context.

The study also clarifies the conditions under which these results hold. The researchers demonstrated that the gas must be sufficiently dominated by collisions with the thermal reservoir compared to collisions between the gas particles themselves. If the particles collide with each other too frequently, the complex behavior driven by the temperature gradient might be overwhelmed. By establishing strict mathematical bounds on the initial state of the gas and the strength of these interactions, the authors were able to prove that the system is well-behaved and will always converge to the unique steady state. Their work provides a rigorous foundation for understanding how heat flows through gases in complex environments, offering a clearer picture of the mechanisms that drive energy transport in systems far from equilibrium.

The implications of this work extend beyond pure theory. By proving the existence and stability of these non-equilibrium states, the researchers have opened the door to more accurate models of heat conduction in gases. In the future, these models could be scaled up to connect with macroscopic laws, such as Fourier's law of heat conduction, which describes how heat moves through materials in everyday life. The ability to predict how a gas will behave when subjected to varying temperatures is a fundamental step toward understanding and controlling thermal processes in engineering and nature. The study confirms that even in a system driven by complex, spatially varying forces, there is an underlying order and predictability, provided the interactions are understood correctly.

Technical Summary: Hypocoercivity for a Thermostatted Boltzmann Equation

Problem Statement
This paper investigates the existence, uniqueness, and stability of nonequilibrium stationary states (NESS) for a gas modeled by the Boltzmann equation on the three-dimensional torus ( $T^3$ ). The system consists of particles undergoing self-collisions and collisions with a spatially varying thermal reservoir (thermostat). The reservoir temperature $\tau(x)$ is non-constant, heating the gas in some regions and cooling it in others, thereby driving a nonequilibrium energy current.

The primary mathematical challenges addressed are:

Existence and Uniqueness: Proving the existence of a unique stationary state $F_\alpha$ for the nonlinear equation when self-collisions are present ( $\alpha > 0$ ) and the reservoir temperature varies.
Stability: Establishing global well-posedness and exponential convergence to this stationary state for initial data satisfying specific Gaussian bounds, even when the initial data is not close to the steady state.
Entropy Dynamics: Analyzing the behavior of relative entropy. While the linear case ( $\alpha=0$ ) satisfies a data-processing inequality (non-increasing entropy), the paper investigates whether the addition of self-collisions allows relative entropy to increase initially, despite eventual exponential decay.

Methodology
The authors employ a perturbative approach, treating the nonlinear equation as a small perturbation of the linear Boltzmann equation with a thermostat ( $\alpha = 0$ ).

Linear Analysis ( $\alpha=0$ ): The linear evolution forms a Markov semigroup $P^\tau_t$ . The authors establish exponential decay in a weighted $L^1$ space ( $L^1(W)$ ) using Harris's theorem. This requires verifying a drift condition (moment bounds) and a minorization condition (lower bounds on transition probabilities).
Norm Transfer: To handle the nonlinear term, control in the weighted $L^\infty$ space ( $L^\infty(W)$ ) is necessary. The authors utilize an operator splitting argument (following Mischler) and a double expansion of Duhamel's formula. This technique allows them to transfer the $L^1(W)$ decay to $L^\infty(W)$ by changing variables from intermediate velocities to earlier collision positions, effectively bounding the nonlinear collision term.
Nonlinear Well-Posedness: Using the linear decay estimates, the authors construct global mild solutions via a contraction mapping argument. They define a class of probability densities with uniform Gaussian pointwise and moment bounds. For sufficiently small self-collision strength $\alpha$ , the solution map at a specific time $t_c$ acts as a strict contraction on this class, guaranteeing a unique fixed point (the stationary state).
Entropy Analysis: The derivative of relative entropy is decomposed into dissipation terms from reservoir collisions, dissipation from self-collisions, and a "cross term" arising from self-collisions. The authors analyze the sign of this cross term to determine if entropy can increase.

Key Contributions and Results

Global Well-Posedness and Exponential Convergence (Theorem 1.1):
- For isotropic cutoff kernels (Maxwell molecules or hard potentials) and reservoir temperatures bounded between fixed positive values ( $0 < \tau_- \le \tau(x) \le \tau_+ < \infty$ ), the authors prove the existence of a unique stationary state $F_\alpha$ within a class of densities bounded by a Gaussian $e^{-a|v|^2}$ (where $a < 1/(2\tau_+)$ ).
- Solutions starting from initial data in this class converge exponentially to $F_\alpha$ in the sum of weighted $L^1$ and $L^\infty$ norms.
- The result holds for initial data far from the stationary state, provided the self-collision parameter $\alpha$ is sufficiently small relative to the Gaussian bounds of the initial data.
Non-Maxwellian Nature of Stationary States (Lemma 5.5):
- If the reservoir temperature $\tau(x)$ is not constant, the stationary state $F_\alpha$ cannot be a local Maxwellian. This is proven by showing that if $F$ were a local Maxwellian, the reservoir collision term would force $\tau$ to be constant.
Initial Growth of Relative Entropy (Theorem 1.2 and Theorem 6.3):
- In the linear case ( $\alpha=0$ ), relative entropy to the stationary state is non-increasing.
- The paper demonstrates that for $\alpha > 0$ and non-constant $\tau$ , arbitrarily small perturbations of the stationary state can cause the relative entropy to increase initially.
- This occurs because the "cross term" in the entropy production rate (arising from the interaction between the density perturbation and the velocity distribution) can be positive and dominate the dissipative terms for small times.
- Despite this initial growth, the solutions still converge exponentially to the stationary state for large times, as guaranteed by Theorem 1.1.
Entropy Production and Energy Current:
- The authors derive a stationary entropy balance equation showing strictly positive total entropy production when $\tau$ is non-constant.
- They establish balance laws for mass, momentum, and energy, showing that the energy current is driven by the temperature gradient and the difference between the gas temperature and the reservoir temperature.

Significance and Claims
The paper establishes global well-posedness and exponential relaxation to a unique NESS for the nonlinear thermostatted Boltzmann equation with spatially varying temperatures, without assuming the solution is close to equilibrium initially. This work resolves a previously open problem regarding nonlinear stability for this Boltzmann model, which had been left open in prior BGK studies [27, Section 1.3.2].

The significance of the entropy growth result lies in challenging the monotonicity of relative entropy in nonlinear nonequilibrium systems. The authors note that while the classical H-theorem ensures entropy production is non-negative for the total system (gas plus reservoir), the relative entropy of the gas with respect to a non-Maxwellian stationary state is not necessarily monotonic. The paper explicitly constructs perturbations where this quantity increases, highlighting the complex interplay between transport and nonlinear collisions in driving the system toward a non-trivial steady state.

The work bridges the gap between linear kinetic theory (where hypocoercivity and Harris's theorem are well-established) and nonlinear problems with spatial inhomogeneities, offering a framework for understanding heat conduction and energy currents in kinetic models beyond the hydrodynamic limit.

More like this