Kinetic Fokker-Planck equations with Maxwell boundary conditions

Technical Summary: Kinetic Fokker-Planck Equations with Maxwell Boundary Conditions

Problem Statement
This work addresses the boundary regularity theory for linear kinetic Fokker-Planck equations in a bounded domain $\Omega \subset \mathbb{R}^n$. The equation under consideration is:
$$\partial_t f + v \cdot \nabla_x f - \text{div}_v(A\nabla_v f) = B \cdot \nabla_v f + F$$
subject to Maxwell boundary conditions. These conditions model the reflection of particles at the spatial boundary $\partial \Omega$ by interpolating between two extreme physical scenarios: diffuse reflection (perfect thermalization) and specular reflection (perfectly smooth elastic bounce). The interpolation is governed by an accommodation coefficient $\alpha \in [0, 1]$.

Prior to this work, rigorous regularity results were restricted to the endpoint cases: $\alpha = 0$ (diffuse) and $\alpha = 1$ (specular). The intermediate regime $\alpha \in (0, 1)$ remained an open problem, particularly regarding the optimal Hölder regularity of solutions near the grazing set $\gamma_0 = \{(x, v) \in \partial \Omega \times \mathbb{R}^n : n_x \cdot v = 0\}$. The authors aim to resolve the regularity behavior for the entire intermediate regime and extend these findings to broader classes of reflection-type boundary conditions, including super-elastic collisions.

Methodology
The authors develop a unified approach that overcomes the limitations of existing techniques, which fail when both specular and diffuse components are present simultaneously. The methodology relies on several novel components:

1. Unified Boundary Framework: The study generalizes the Maxwell condition to a form $f = \alpha R_b f + (1-\alpha)g$ on the incoming boundary $\gamma_-$. Here, $R_b$ represents a generalized reflection operator (specular when $b=1$, super-elastic when $b \in (0, 1)$), and $g$ represents an in-flow datum. This framework allows the treatment of both the standard Maxwell condition (where $g$ is a nonlocal operator $Nf$) and super-elastic collisions.

2. De Giorgi-Nash-Moser Theory at the Boundary:

   • Local Boundedness: The authors establish local boundedness for weak solutions with merely measurable coefficients. A key difficulty is that standard extension techniques (mirror reflection for specular, constant extension for diffuse) fail for $\alpha \in (0, 1)$. To overcome this, they introduce a mechanism to prove a gain of integrability in all variables up to the boundary. This involves extending the solution to the whole space, identifying the boundary trace as a measure supported on the grazing set, and deriving a kinetic Caccioppoli-type inequality. Crucially, they utilize the dissipativity of the boundary flux, which holds when $\alpha/b^2 \leq 1$.
   • Hölder Regularity: Once boundedness is established, they employ an oscillation decay scheme. The proof treats the problem as a kinetic equation with an in-flow condition, leveraging the fact that $\alpha < 1$ ensures the boundary data oscillation decays sufficiently fast.

3. Liouville Theorems and Blow-Up Analysis:

   • To determine the optimal regularity exponent, the authors utilize a blow-up technique. They classify the blow-up limits of solutions near the grazing set.
   • They prove a 1D Liouville theorem for the limiting equation $v \partial_x h - \partial_{vv} h = 0$ subject to the reflection condition. This requires establishing a boundary Harnack-type principle and a maximum principle for solutions in the half-space.
   • A critical step involves constructing an explicit homogeneous solution $\phi_{\alpha, b}$ and proving that any global solution with suitable growth is a multiple of this function. This classification relies on a novel maximum principle and a modified mirror-type reflection argument adapted to the operator $R_b$.

4. Transfer of Regularity and Schauder Estimates:

   • The authors establish transfer of regularity results from the velocity variable $v$ to the spatial variable $x$ up to the boundary, even for equations with source terms in dual fractional Sobolev spaces.
   • For smooth coefficients, they derive Schauder-type estimates away from the grazing set and optimal Hölder estimates up to the grazing set using a perturbation argument and difference quotient techniques.

Key Contributions and Results

1. Optimal Regularity for Maxwell Conditions ($\alpha \in (0, 1)$):
   The paper establishes that solutions are Hölder continuous with an optimal exponent $\lambda_\alpha$. Specifically, for smooth coefficients, the solution $f$ belongs to $C^{\lambda_\alpha}$ up to the grazing set, where:
   $$\lambda_\alpha = \frac{3}{\pi} \arccos\left(\frac{\alpha}{2}\right) - 1$$
   This exponent is shown to be sharp; the authors construct explicit solutions demonstrating that $f \notin C^{\lambda_\alpha + \epsilon}$ for any $\epsilon > 0$.

   • As $\alpha \to 0$, $\lambda_\alpha \to 1/2$, recovering the known $C^{1/2}$ regularity for diffuse reflection.
   • As $\alpha \to 1$, $\lambda_\alpha \to 0$, indicating that the high regularity ($C^{4,1}$) observed in the specular case is a rigid feature that disappears immediately for any $\alpha < 1$.

2. Regularity for General Reflection Conditions:
   The results extend to the generalized condition $f = \alpha R_b f + (1-\alpha)g$ with a super-elastic parameter $b \in (0, 1]$. The optimal exponent is given by $\lambda_{\alpha, b}$, defined implicitly by:
   $$2 \cos\left(\pi \left(\frac{\lambda}{3} + \frac{1}{3}\right)\right) = \alpha b^\lambda$$
   This provides the first systematic regularity theory for kinetic equations with super-elastic boundary conditions.

3. Regularity with Measurable Coefficients:
   The authors prove that weak solutions remain Hölder continuous ($C^\beta$) even when the diffusion matrix $A$ is merely measurable (uniformly elliptic) and coefficients $B, F$ are bounded. This extends the De Giorgi-Nash-Moser theory to the boundary for this class of kinetic equations.

4. Extensions to Generalized Models:
   The framework is shown to accommodate:

   • Flux-dependent accommodation coefficients $\alpha(t, x)$.
   • Nonlocal reflection operators defined by general kernels $K(v, w)$ instead of the Maxwellian kernel.

Significance and Claims
The paper claims to provide the first regularity results for kinetic equations with Maxwell boundary conditions in the intermediate regime $\alpha \in (0, 1)$. It resolves the open question of how optimal regularity interpolates between the diffuse ($C^{1/2}$) and specular ($C^{4,1}$) limits, revealing a fundamental discontinuity in regularity behavior as soon as the specular symmetry is broken ($\alpha < 1$).

The authors emphasize that their unified approach is not limited to the specific Maxwell condition but extends to a broad class of reflection-type boundary conditions, including super-elastic collisions relevant to granular gases. The explicit identification of the optimal regularity exponent $\lambda_{\alpha, b}$ and the proof of its sharpness represent a significant advancement in the boundary regularity theory of kinetic equations. The results are intended to serve as a foundation for establishing conditional regularity and continuation criteria for nonlinear kinetic models, such as the Landau and non-cutoff Boltzmann equations, in bounded domains.