Hölder regularity of parabolic equations with Dirichlet boundary conditions and application to reaction-diffusion and reaction-cross-diffusion systems

Basic Information

• Paper ID: 2511.18872
• Title: Hölder regularity of parabolic equations with Dirichlet boundary conditions and application to reaction-diffusion and reaction-cross-diffusion systems
• Authors: Hector Bouton, Laurent Desvillettes, Helge Dietert (Université Paris Cité and Sorbonne Université, CNRS, IMJ-PRG)
• Classification: math.AP (Analysis of Partial Differential Equations)
• Submission Date: November 24, 2025
• Paper Link: https://arxiv.org/abs/2511.18872v1

Abstract

This paper extends the authors' recent work on Neumann boundary conditions to the Dirichlet boundary condition setting. The core object of study is the parabolic equation $a\partial_t w - \Delta w = f$ with rough coefficients $a$, equipped with homogeneous Dirichlet boundary conditions and the special assumption $\partial_t w \geq 0$. Based on this result, the authors prove the existence of global strong solutions for the three-dimensional triangular Shigesada-Kawasaki-Teramoto (SKT) cross-diffusion system with Lotka-Volterra reaction terms, as well as estimates for reversible chemical reaction-diffusion systems, all under Dirichlet boundary conditions.

Research Background and Motivation

Problem Background

1. Core Problem: Study Hölder regularity of parabolic equations with non-constant rough coefficients under Dirichlet boundary conditions. Specifically, consider the equation: $a(t,x)\partial_t w(t,x) - \Delta w(t,x) = f(t,x), \quad \partial_t w \geq 0$ where $a$ satisfies the ellipticity condition $0 < a_0 \leq a \leq c_0 a_0 < \infty$.
2. Significance:
   • Hölder regularity is a fundamental problem in PDE theory, crucial for understanding solution properties
   • Results apply to nonlinear systems in biomathematics, chemical reaction kinetics, and other fields
   • Provides key technical tools for establishing global existence for cross-diffusion and reaction-diffusion systems
3. Limitations of Existing Methods:
   • The classical Krylov-Safonov (1981) result, though covering more general cases, has non-explicit constants and requires stronger spatial assumptions
   • Previous work by the authors boutondesvillettes2025 only handles Neumann boundary conditions
   • Under Dirichlet boundary conditions, solution behavior near the boundary is more complex, requiring new techniques
4. Research Motivation:
   • Utilize the monotonicity assumption $\partial_t w \geq 0$ to simplify proofs and obtain explicit constants
   • Require only critical Lebesgue space control on $f$
   • Provide complete treatment of boundary conditions with clear theoretical support for applications

Core Contributions

1. Main Theoretical Result (Theorem 1): Under the condition $\gamma := 2 - \frac{2}{p} - \frac{d}{q} > 0$, the non-negative solution $w$ of equation (1) belongs to $C^{0,\alpha}([0,T]\times\Omega)$ with the estimate: $\|w\|_{C^{0,\alpha}([0,T]\times\Omega)} \leq C_*\left(\|f\|_{L^p((0,T];L^q(\Omega))} + \|w^{in}\|_{C^1(\Omega)}\right)$
2. Application to Reversible Chemical Systems (Proposition 1): Proves global existence of strong solutions for the reversible chemical reaction system $\partial_t u_i - d_i\Delta u_i = (-1)^i(u_1u_3 - u_2u_4), \quad i=1,\ldots,4$ under Dirichlet boundary conditions in dimensions up to four.
3. Application to SKT Cross-Diffusion System (Proposition 2): Proves global existence of strong solutions for the triangular SKT system\partial_t u - \Delta[(d_1+\sigma v)u] = f_u(u,v)\\ \partial_t v - d_2\Delta v = f_v(u,v) \end{cases}$$ under Dirichlet boundary conditions in dimensions up to four.
4. Technical Innovation: Develops oscillation decay techniques specifically adapted to Dirichlet boundary conditions, which differ essentially from the Neumann case.

Detailed Methodology

Task Definition

Input:

• Bounded $C^2$ domain $\Omega \subset \mathbb{R}^d$
• Coefficients $a(t,x)$ satisfying ellipticity conditions
• Forcing term $f \in L^p((0,T];L^q(\Omega))$ where $\gamma = 2-\frac{2}{p}-\frac{d}{q} > 0$
• Initial value $w^{in} \in C^1(\Omega)$ satisfying $w^{in}|_{\partial\Omega} = 0$

Output:

• Hölder continuity of solution: $w \in C^{0,\alpha}([0,T]\times\Omega)$
• Explicit Hölder norm estimate

Constraints:

• Homogeneous Dirichlet boundary condition: $w = 0$ on $(0,T]\times\partial\Omega$
• Monotonicity: $\partial_t w \geq 0$
• Non-negativity: $w \geq 0$

Core Technical Framework

1. Dirichlet Heat Kernel Estimates (Lemmas 1-2)

Lemma 2 (Key Lower Bound Estimate): For the Dirichlet heat kernel $\Gamma_{B(0,R),D}$ in the ball $B(0,R)$, taking $T_R := a_0\frac{9R^2}{32d}$, we have $\inf_{x,y\in B(0,R/4)} \inf_{t\in[\frac{T_R}{2c_0a_0}, \frac{T_R}{a_0}]} \Gamma_{B(0,R),D}(t,x,y) \gtrsim_{c_0,d} R^{-d}$

Proof Strategy:

• Construct comparison function $\Psi(t,x,y) := p(t,|x-y|) - \sup_{0\leq s\leq t}p(s,3R/4)$
• Apply maximum principle to obtain $\Gamma_{B(0,R),D}(t,x,y) > \Psi(t,x,y)$
• For $x,y\in B(0,R/4)$ and appropriate time ranges, compute precisely to obtain positive lower bound

2. Oscillation Decay (Proposition 3)

Core Idea: Prove geometric decay of oscillation in regions away from the boundary.

Proposition 3: For $R>0$ satisfying $\beta R^2 \leq T$ (where $\beta = \frac{9a_0}{32d}$), the solution $w:[−\beta R^2, 0]\times B(0,R)\to[0,1]$ satisfies $\text{osc}_{(-\beta R^2/16,0]\times B(0,R/4)} w \leq 1-\delta + C_f R^\gamma \|f\|_{L^p((−\beta R^2,0);L^q(B(0,R)))}$

Proof Strategy:

• Consider two cases: $|\{w(-T_R,\cdot)\geq\frac{1}{2}\}\cap B(0,R/4)| \geq \frac{1}{2}|B(0,R/4)|$ or its complement
• Construct comparison function $v$ satisfying $(a_0c_0\partial_t - \Delta)v = f$
• Apply comparison principle and Lemma 2's lower bound to obtain $w(t,x) > \delta - C_2R^\gamma\|f\|$
• Iterate to achieve geometric decay rate

3. Boundary Neighborhood Estimates (Lemma 3)

Key Lemma 3: For $z\in\Omega$, let $d_z := d(z,\partial\Omega)$ denote distance to boundary, then $w(t,x) \leq C(\|w^{in}\|_{\text{Lip}(\Omega)} + \|f\|_{L^p([0,T],L^q(\Omega))}) d_x^{\tilde{\gamma}}$ where $\tilde{\gamma} := \min(\gamma, 1-2\varepsilon)$

Proof Key Points:

• Case 1 (Small time $t^{1/2}\leq d_x$): Use integral representation of heat kernel and Lemma 1
• Case 2 (Large time $t > d_x^2$): Use Gaussian estimates $\Gamma_{\Omega,D}(t,x,y) \lesssim \frac{d_x d_y}{t^{\frac{d+2}{2}}}e^{-c\frac{|x-y|^2}{t}}$ Decompose integral into three parts for careful estimation

4. Small Time Estimates (Lemma 4)

Lemma 4: $|w(t,x) - w(0,x)| \leq C(\|w^{in}\|_{C^1(\Omega)} + \|f\|_{L^p([0,T],L^q(\Omega))}) t^{\frac{\min(1,\gamma)}{2}}$

Proof Method:

• Apply comparison principle: $v_{a_0c_0} \leq w \leq v_{a_0}$
• Decompose $v_{a_0}$ as $v_{a_0,f} + \tilde{v}_{a_0}$
• Use Schauder estimates on $\tilde{v}_{a_0}$ to obtain parabolic Hölder continuity
• Apply Lemma 1 to $v_{a_0,f}$

Complete Proof Architecture of Theorem 1

Unified Treatment of Three Cases:

1. Near Boundary ($\min(d_x,d_y) \leq 2|x-y|$):
   • Directly apply Lemma 3
   • Obtain $|w(t,x)-w(t',y)| \lesssim |x-y|^{\tilde{\gamma}}$
2. Small Time ($t'^{1/2}, t^{1/2} \leq |x-y|$):
   • Apply Lemma 4 and Lipschitz property of initial value
   • Obtain $|w(t,x)-w(t',y)| \lesssim |x-y|^{\tilde{\gamma}}$
3. Interior and Large Time ($|x-y| \leq \min(t'^{1/2}, t^{1/2}, \frac{\min(d_x,d_y)}{2})$):
   • Take scale $R := \min(t^{1/2}, \min(d_x,d_y))$
   • If $\min(d_x,d_y) \leq t^{1/2}$: use Lemma 3
   • If $\min(d_x,d_y) > t^{1/2}$: use Lemma 4
   • Apply Proposition 3's oscillation decay, iterate to obtain: $o_k \lesssim R^{\tilde{\gamma}}(\|f\| + \|w^{in}\|)\Lambda^k$ where $\Lambda > \max(\frac{1}{4^\gamma}, 1-\delta)$
   • Choose $k_0$ such that $R4^{-k_0} \leq \max(|x-y|, (\frac{|t-t'|}{\beta})^{1/2})$
   • Obtain $|w(t,x)-w(t',y)| \lesssim \max(|x-y|, |t-t'|^{1/2})^{\alpha_\Lambda}$

Detailed Application Systems

Application 1: Reversible Chemical System (Proposition 1)

System Model: $\partial_t u_i - d_i\Delta u_i = (-1)^i(u_1u_3 - u_2u_4), \quad i=1,2,3,4$ representing the reversible reaction $A_1 + A_3 \rightleftharpoons A_2 + A_4$

Proof Strategy:

1. Standard A Priori Estimates (Equation 12): $\sum_{i=1}^4 \int_\Omega \frac{u_i^{p+1}}{p+1}(T) + \frac{4p}{(p+1)^2}\sum_{i=1}^4 d_i\int_0^T\int_\Omega |\nabla(u_i^{\frac{p+1}{2}})|^2 \leq \text{Initial} + C_p\sum_{i=1}^4\int_0^T\int_\Omega u_i^{p+2}$
2. Define Auxiliary Function: $w := \int_0^t(\sum_{i=1}^4 d_i u_i)$
   • Verify: $w\geq 0$, $\partial_t w \geq 0$, $w|_{\partial\Omega}=0$
   • Compute: $\Delta w = \sum_{i=1}^4 u_i - \sum_{i=1}^4 u_i^{in}$
   • Define $a := \frac{\sum_{i=1}^4 u_i}{\sum_{i=1}^4 d_i u_i}$ satisfying $\frac{1}{\max d_i} \leq a \leq \frac{1}{\min d_i}$
   • Obtain: $a\partial_t w - \Delta w = \sum_{i=1}^4 u_i^{in}$
3. Apply Theorem 1: Obtain $\sum_{i=1}^4 u_i \in C^{0,\alpha}([0,T]\times\Omega)$
4. Interpolation Inequality (Equation 13): $\left\|\sum_{i=1}^4 u_i - \sum_{i=1}^4 u_i^{in}\right\|_{L^{\frac{2(3-\alpha)}{2-\alpha}}(\Omega)}^3 \leq C\|w\|_{C^{0,\alpha}(\Omega)}^{\frac{3}{3-\alpha}} \left\|\nabla\left[\sum_{i=1}^4 u_i - \sum_{i=1}^4 u_i^{in}\right]\right\|_{L^2(\Omega)}^{\frac{3(2-\alpha)}{3-\alpha}}$
5. Bootstrap Argument: Combine estimates (12) and (13) to obtain $u_i \in L^{3+\delta}([0,T]\times\Omega)$, then all $L^p$ estimates follow.

Application 2: SKT Cross-Diffusion System (Proposition 2)

System Model: