Non-linear parabolic PDEs with rough coefficients and critical data: existence, uniqueness and regularity of weak solutions

1. Problem Statement and Motivation

The paper addresses the well-posedness (existence, uniqueness, and regularity) of weak solutions to non-linear parabolic partial differential equations (PDEs) of the form:
$$\partial_t u - \text{div}(A(t, x, u)\nabla u) = F(u, \nabla u)$$
Specifically, the authors focus on the reaction-diffusion equation:
$$\partial_t u - \text{div}(A \nabla u) = \phi(u), \quad u(0) = u_0$$
where:

• Rough Coefficients: The diffusion matrix $A: \mathbb{R}^n \to \mathbb{R}^{n \times n}$ is merely bounded, measurable, and elliptic. It lacks any continuity or smoothness assumptions (e.g., it may represent diffusivity in heterogeneous media like brain tissue).
• Critical Data: The initial data $u_0$ belongs to homogeneous Besov spaces $\dot{B}^\alpha_{p,p}$ with negative regularity ($\alpha < 0$). These are "scaling-critical" spaces, meaning the norm is invariant under the natural scaling of the equation.
• Non-linearity: The reaction term $\phi(u)$ exhibits polynomial growth of order $1+\rho$ (where $\rho > 2/n$) and satisfies a local Lipschitz condition.

The Challenge: Existing methods (such as maximal regularity in $L^p_t L^q_x$ spaces) fail in this setting. When coefficients are rough, standard parabolic Sobolev embeddings do not provide sufficient "hypercontractivity" (the ability of the solution operator to map lower integrability spaces to higher ones) to handle the super-linear growth of the non-linearity. Furthermore, standard weak solution theories often require initial data in $L^2$ or positive regularity spaces, which excludes the critical negative regularity regime.

2. Methodology and Framework

The authors develop a novel framework combining harmonic analysis and PDE theory to overcome these obstacles.

A. Weighted Z-Spaces as Solution Spaces

Instead of standard $L^p_t L^q_x$ spaces, the authors utilize weighted Z-spaces ($Z^{r,q}_\beta$) as the primary solution spaces.

• Definition: These spaces measure local integrability via averages over parabolic Whitney boxes (scaled balls in space-time) and global integrability via $L^r$ norms of these averages.
• Advantage: They allow for different local ($q$) and global ($r$) integrability exponents, providing the flexibility needed to handle rough coefficients and critical data.
• Connection to Initial Data: The authors leverage a known characterization where the free evolution of Besov data $u_0 \in \dot{B}^\alpha_{p,p}$ naturally lives in a specific weighted Z-space.

B. Hypercontractive Singular Integral Operators (SIOs)

A central tool is the development of a theory for hypercontractive SIOs on weighted Z-spaces.

• Concept: The solution operator for the linear parabolic equation (the Duhamel operator) acts as a singular integral operator that improves integrability (maps $L^q$ to $L^r$ with $r > q$).
• Novelty: Unlike classical Calderón-Zygmund theory, these operators do not necessarily possess pointwise kernel representations (due to rough coefficients). Instead, they satisfy "off-diagonal estimates" and decay properties.
• Result: The authors prove that these operators map $Z^{p,q}_\beta \to Z^{p,r}_{\beta+\kappa}$, effectively compensating for the loss of integrability caused by the super-linear non-linearity $\phi(u)$.

C. Mild-to-Weak Transference

The proof strategy involves a two-step approach:

1. Mild Solutions: First, establish existence and uniqueness of "mild solutions" (fixed points of the integral equation $u = E_L(u_0) + L(\phi(u))$) using Banach's fixed-point theorem and the hypercontractivity of the Duhamel operator.
2. Weak Solutions: Transfer these results to "weak solutions" (satisfying the PDE in the distributional sense with $\nabla u \in L^2_{loc}$).
   • Mild $\to$ Weak: Use bootstrapping regularity to show that mild solutions have gradients in the required weighted Z-spaces.
   • Weak $\to$ Mild: This is the harder direction. It requires proving that a weak solution satisfies the integral equation. This is achieved by establishing super-linear Reverse Hölder (RH) inequalities for weak solutions and proving a self-improving property for these inequalities. This allows the authors to upgrade the local integrability of the weak solution sufficiently to identify it as a mild solution.

3. Key Contributions

1. New Function Space Framework: The systematic use of weighted Z-spaces for non-linear parabolic problems with rough coefficients, allowing for critical (negative regularity) initial data.
2. Hypercontractive SIO Theory: A general theory of singular integral operators on Z-spaces that extends beyond the classical kernel-based setting, specifically tailored for operators generated by elliptic operators with rough coefficients.
3. Self-Improving Super-Linear RH Inequalities: A new principle establishing that weak solutions to reaction-diffusion equations satisfy Reverse Hölder inequalities that "self-improve" the local integrability exponent. This is crucial for bridging the gap between weak and mild solutions.
4. Maximal Well-Posedness: Proving the existence of a maximal weak solution without requiring smallness conditions on the initial data or the non-linearity.

4. Main Results

Theorem 7.1 (Well-posedness of the Reaction-Diffusion Equation):
For a growth parameter $\rho > 2/n$ and initial data $u_0 \in \dot{B}^\alpha_{p,p}$ (where the parameters satisfy the scaling relation $2/\rho = n/p - \alpha$):

• Existence: There exists a local weak solution $(u, \tau)$ with $\tau > 0$.
• Regularity: The solution satisfies $u \in Z^{r,q}_{n/2r - 1/\rho}$ and $\nabla u \in Z^{p,2}_{n/2p - 1/\rho - 1/2}$ for a wide range of admissible parameters $r, q$.
• Uniqueness: The solution is unique within the class of weak solutions satisfying the specified regularity conditions.
• Maximality: The solution is maximal; it cannot be extended beyond its lifespan $\tau$ while maintaining the required regularity.

Specific Technical Bounds:

• The range of admissible global integrability $r$ is $(RD_-(n, \rho), RD_+(n, \rho))$, where $RD_-(n, \rho) = (n+2)\rho/2$ and $RD_+(n, \rho) = n(1+\rho)\rho/2$.
• The theory handles the case where the diffusion matrix $A$ is only measurable, covering physical scenarios like tumor growth in heterogeneous brain tissue where diffusivity jumps between gray and white matter.

5. Significance and Impact

• Overcoming Rough Coefficients: This work breaks the barrier of requiring continuous or smooth coefficients for critical data problems. It extends the scope of non-linear parabolic PDE theory to highly irregular media.
• Critical Data Handling: By working with negative regularity Besov spaces, the paper addresses the most singular initial data allowed by scaling, which is essential for understanding blow-up phenomena and the behavior of solutions near singularities.
• Generalizability: The authors note that their toolkit (hypercontractive SIOs and self-improving RH inequalities) is robust enough to be applied to other models, including:
  • Burgers-type equations.
  • Quasi-linear problems.
  • Generalized harmonic map heat flows.
  • Systems of reaction-diffusion equations (e.g., Fitzhugh-Nagumo).
• Theoretical Advancement: The introduction of "hypercontractive SIOs" and the "mild-weak transference principle" provides a new blueprint for analyzing non-linear evolution equations where standard energy methods or maximal regularity fail.

In summary, this paper establishes a comprehensive well-posedness theory for non-linear parabolic PDEs with rough coefficients and critical initial data, utilizing a sophisticated interplay between harmonic analysis (Z-spaces, SIOs) and PDE regularity theory (Reverse Hölder inequalities, bootstrapping).