Optimal $C^{\frac{1}{2}}$ regularity of the Boltzmann equation in non-convex domains

This paper, titled "Optimal $C^{1/2}$ Regularity of the Boltzmann Equation in Non-Convex Domains" by Gayoung An and Donghyun Lee, addresses a long-standing challenge in kinetic theory regarding the regularity of solutions to the Boltzmann equation under physical boundary conditions in non-convex geometries.

1. The Problem

The authors study the Boltzmann equation with specular reflection boundary conditions in an exterior domain $\Omega$ of a uniformly convex object.

The central difficulty lies in the geometry of the boundary. In non-convex domains (such as the exterior of a convex obstacle), billiard trajectories can undergo "grazing collisions"—where a particle hits the boundary at a nearly tangential angle. Near these grazing sets, the trajectory map (the mapping of initial position and velocity to the state after a bounce) is not differentiable; specifically, it exhibits a square-root singularity ($C^{1/2}$ regularity).

Previous research (notably Kim and Lee, 2023) established $C^{\beta}$ regularity for $\beta < 1/2$ by averaging the specular singularity. However, reaching the optimal $C^{1/2}$ regularity was hindered because, as a particle approaches the boundary, the geometric effects of the domain's curvature vanish, making the boundary appear "flat" and causing the singularity integration to fail at the critical exponent $\beta = 1/2$.

2. Methodology: Dynamical Singular Regime Integration

To overcome the limitations of previous "static" averaging methods, the authors introduce a novel methodology termed "Dynamical Singular Regime Integration."

• Shifted Coordinates: The authors introduce "shifted" positions ($\tilde{x}$) and velocities ($\tilde{v}$) to decompose trajectories into singular and non-singular parts. This allows for a precise measurement of the "Specular Singularity"—the mathematical term describing the loss of regularity during a grazing collision.
• Dynamical vs. Static Integration: Instead of fixing a spatial location $x$ and integrating over velocity fluctuations (the static approach), the authors follow the characteristic trajectory $X(s; t, x, v)$ through time. By integrating the singularity along the path of the particle, they can exploit the fact that a particle only spends a limited amount of time in the "singular regime" (near the boundary and near grazing angles).
• Comparison Principle: To handle general uniformly convex objects, the authors use a comparison principle, bounding the complex geometry of the domain by a circumscribed sphere. This allows them to use explicit calculations from spherical geometry to prove results for general convex shapes.
• Seminorm Construction: They define new weighted seminorms ($X$ and $V$) that incorporate both velocity and time integrations. This structure is specifically designed to capture and control the singular behavior encountered during the Duhamel integration of the collision operator.

3. Key Contributions

• New Integration Scheme: The development of the dynamical singular regime integration is the primary mathematical innovation. It allows the researchers to treat the singularity as a time-integrable phenomenon rather than a spatial/velocity-based obstruction.
• Handling the Small-Velocity Singularity: The authors identify that while $C^{1/2}$ regularity is achievable, a singularity persists when the velocity $|v| \to 0$ near the boundary. They successfully incorporate this into a weighted H"older norm using a weight function $W$.
• Geometric Analysis: They provide rigorous estimates for the backward exit time ($t_b$) and the sensitivity of billiard trajectories in non-convex domains, proving that the "grazing" effect is integrable.

4. Results

The main result is Theorem 1.4, which establishes the optimal $C^{1/2}_{x,v}$ H"older regularity for the solution $f(t, x, v)$ of the Boltzmann equation.

Specifically, if the initial data $f_0$ is in the weighted H"older space $A^{1/2}(f_0)$, then for a sufficiently small time $T$, the solution $f$ satisfies:
$$\sup_{0 \leq t \leq T} \sup_{|(x,v)-(\bar{x},\bar{v})| \leq 1} \frac{W^{-1} \langle v \rangle^{-1} e^{-\varpi \langle v \rangle^2 t} |f(t, x, v) - f(t, \bar{x}, \bar{v})|}{|(x, v) - (\bar{x}, \bar{v})|^{1/2}} \leq C$$
where $W$ is a weight function that accounts for the singularity near the boundary and at zero velocity. This result confirms that the regularity of the Boltzmann equation is limited by the $C^{1/2}$ nature of the billiard trajectories.

5. Significance

This paper is significant for several reasons:

1. Mathematical Completeness: It closes a gap in the regularity theory for the Boltzmann equation in non-convex domains, proving that the regularity is exactly as good as the underlying billiard dynamics allow.
2. Theoretical Toolset: The "dynamical singular regime integration" provides a new framework that could potentially be applied to other kinetic equations (like the Vlasov equation) or other boundary conditions where grazing singularities occur.
3. Physical Insight: It provides a rigorous mathematical foundation for understanding how the geometry of a container influences the smoothness of a rarefied gas distribution, particularly how the "flatness" of a boundary at a microscopic scale interacts with the macroscopic curvature.