$L_p$-estimates of the conormal derivative problem for parabolic equations with time measurable coefficients and $A_p$-weights

Basic Information

• Paper ID: 2510.21139
• Title: $L_p$-estimates of the conormal derivative problem for parabolic equations with time measurable coefficients and $A_p$-weights
• Authors: Hongjie Dong (Brown University), Pilgyu Jung (Yonsei University), Doyoon Kim (Korea University)
• Classification: math.AP (Analysis of PDEs)
• Submission Date: October 24, 2025
• Paper Link: https://arxiv.org/abs/2510.21139v1

Abstract

This paper investigates weighted mixed-norm estimates for divergence-form parabolic equations with conormal derivative boundary conditions on Reifenberg-flat domains. The leading coefficients are assumed to be merely measurable in the time variable and possess small mean oscillation in the spatial variables. The boundary estimates are derived by overcoming regularity issues through the use of half-time derivative estimates.

Research Background and Motivation

Core Problem

This paper studies the $L_p$ theory for the following second-order divergence-form parabolic equation: $u_t - D_i(a_{ij}D_ju) + \lambda u = D_t^{1/2}h + D_ig_i + f \quad \text{in } \mathbb{R} \times \Omega$ equipped with conormal derivative boundary conditions, where $\Omega$ is a Reifenberg-flat domain.

Research Significance

1. Theoretical Importance: The $L_p$ theory for elliptic and parabolic equations has been extensively studied over the past decades, but the theory remains largely open for Neumann boundary value problems with time-irregular coefficients and irregular boundaries.
2. Technical Challenges: When the time irregularity of coefficients is combined with geometric irregularity of the boundary, the problem becomes genuinely non-trivial: one must treat $u_t$ as a distribution rather than as a function in $L_p$ spaces.
3. Limitations of Existing Methods:
   • For Dirichlet problems, extensive results on interior and boundary estimates exist
   • For conormal derivative problems for elliptic equations, results are available (e.g., Byun-Wang 2005)
   • However, for parabolic conormal derivative problems with time-irregular coefficients, no prior results exist

Motivation for Innovation

The obstacle encountered by traditional methods: when applying existing techniques on Reifenberg-flat domains, a term $u_t I_{\Omega^*}$ appears on the right-hand side (where $\Omega^*$ is the part of the half-space outside the domain), but solutions to divergence-form parabolic equations typically do not satisfy $u_t \in L_p$. If the coefficients were independent of $t$, one could exploit the fact that $u_t$ itself is also a solution; however, when coefficients depend on $t$ without regularity, this approach no longer applies.

Core Contributions

1. Introduction of Half-Time Derivative Technique: The first systematic use of half-time derivatives $D_t^{1/2}u$ in conormal derivative problems for parabolic equations, which is the key innovation for overcoming time irregularity.
2. Establishment of New Equation Form: Study of non-traditional parabolic equations containing half-time derivative terms $D_t^{1/2}h$, which is crucial for handling equations with time-irregular coefficients on Reifenberg-flat domains.
3. Weighted Mixed-Norm Estimates: Establishment of estimates in $L_{p,q,\omega}$ spaces where the weight function $\omega(t,x) = \omega_1(x)\omega_2(t)$, with $\omega_1 \in A_p(\mathbb{R}^d)$ and $\omega_2 \in A_q(\mathbb{R})$.
4. Complete Existence and Uniqueness Theory: For sufficiently large $\lambda$, existence and uniqueness of solutions are established.
5. Application to Standard Equations: Results are applied to traditional parabolic equations without half-time derivative terms on $(0,T) \times \Omega$.

Detailed Methodology

Problem Formulation

Main Equation (Equation 2.2): $u_t - D_i(a_{ij}D_ju) + \lambda u = D_t^{1/2}h + D_ig_i + f \quad \text{in } Q = \mathbb{R} \times \Omega$ equipped with conormal derivative boundary conditions.

Weak Solution Definition (Definition 2.1): For $h, g_i, f \in L_{p,q,\omega}(Q)$, $u \in H_p^{1/2,1}(Q)$ is a solution if for any $\phi \in H_{p',q',\omega'}^{1/2,1}(Q)$: $\int_Q \left(-\mathcal{H}(D_t^{1/2}u)D_t^{1/2}\phi + a_{ij}D_juD_i\phi + b_iD_iu\phi + cu\phi\right) dX = \int_Q \left(hD_t^{1/2}\phi - g_iD_i\phi + f\phi\right) dX$

where $\mathcal{H}$ is the Hilbert transform.

Core Technical Innovations

1. Definition and Properties of Half-Time Derivatives

For $\phi \in C_0^\infty(Q)$: $D_t^{1/2}\phi(t,x) = \frac{1}{\sqrt{8\pi}} \int_{\mathbb{R}} \frac{\phi(t+\ell, x) - \phi(t,x)}{|\ell|^{3/2}} d\ell$

For general functions, definition is given through weak formulation (Equation 2.1).

Key Identity (Remark 2.4): $\int_Q \mathcal{H}(D_t^{1/2}u)D_t^{1/2}\phi dX = \int_Q u\phi_t dX$

This connects half-time derivatives with ordinary time derivatives.

2. Key Technique for Handling Boundary Irregularity

Core Idea of Lemma 4.1: For Reifenberg-flat domain $\Omega_R(x_0)$, through coordinate transformation the local boundary is approximated by a plane. Let $\Omega_R^*(x_0) = \mathbb{R}^d_- \cap \Omega_R(x_0)$ (the part of the half-space outside the domain).

Key Transformation: Rewrite $u_t I_{\Omega^*}$ as: $u_t I_{\Omega^*} = -D_t^{1/2}(\mathcal{H}(D_t^{1/2}u)I_{\Omega^*})$

Thus, near a subdomain with flat boundary, the equation becomes: $u_t - D_i(a_{ij}(t)D_ju) + \lambda u \approx D_t^{1/2}(\mathcal{H}(D_t^{1/2}u)I_{\Omega^*}) + D_ig_i + f$

Even when $h=0$, the right-hand side contains a half-time derivative term.

3. Hierarchical Estimation Strategy

First Level: Basic $L^2$ Solvability (Theorem 3.1): Using the Lax-Milgram theorem, for arbitrary domains (without regularity requirements) and coefficients (satisfying only ellipticity), establish $L^2$ estimates: $\|D_t^{1/2}u\|_2 + \|Du\|_2 + \sqrt{\lambda}\|u\|_2 \leq N(\|h\|_2 + \|g_i\|_2 + \lambda^{-1/2}\|f\|_2)$

Second Level: $L^p$ Theory for Special Coefficients (Proposition 3.3): For coefficients satisfying Assumption 3.2 (depending on $t$, $x_1$, or $(t,x_1)$), establish $L^p$ estimates on $\mathbb{R} \times \mathbb{R}^d$.

Third Level: Local Estimates (Lemma 3.6): Local estimates on cylindrical domains $\Pi_R = (0,R) \times \mathbb{R}^{d-1}$: $\left(2^{-1}R^{-2}\int_{-R^2}^{R^2}\int_{\Pi_R} |D_t^{1/2}u|^p + |Du|^p + \lambda^{p/2}|u|^p dx dt\right)^{1/p} \leq N\sum_{j=0}^\infty 2^{-j/4}(\cdots)$

Fourth Level: Decomposition on Reifenberg-flat Domains (Proposition 4.2): For $U = (Du, \sqrt{\lambda}u)$, there exists a decomposition $U = W + V$ where:

• $W$: controllable part satisfying $L^p$ estimates
• $V$: regular part satisfying $L^{p_1}$ estimates ($p_1 > p$)

Key estimate (boundary case, Equations 4.7-4.8): $(|W|^p)^{1/p}_{C_R(X_0)} \leq N\gamma_0^{1/\nu'}\sum_{j=0}^\infty 2^{-j/4}(|U|^\nu)^{1/\nu}_{C_{2^{j/2}R,R}(X_0)} + \cdots$

4. Level Set Method

Core Inequality (Equation 5.9): For $s > s_0$, $\omega(A(\kappa s)) \leq N\gamma^\beta \omega(B(s))$

where:

• $A(s) = \{|U| > s\}$
• $B(s)$ involves maximal functions $M(I_Q|U|^{p_0})$, $SM(I_Q|U|^\nu)$, etc.
• $\gamma$ is a selectable small parameter, $\gamma_0$ comes from coefficient assumptions

Through the "crawling of ink spots" lemma and the Hardy-Littlewood maximal function theorem: $\|U\|_{L_p^p} = p\int_0^\infty \omega(A(s))s^{p-1}ds \leq I_1 + I_2$

where $I_1$ involves estimates on small cylinders and $I_2$ is controlled via the Hardy-Littlewood maximal function theorem.

By choosing $\gamma$ and $\gamma_0$ sufficiently small, the left-hand side terms are absorbed.

Technical Assumptions

Assumption 2.6 ($\gamma_0$):

1. Small BMO condition for coefficients (Equation 2.6): $\frac{1}{|Q_r|}\int_{Q_r(X)} |a_{ij}(s,y) - \frac{1}{|B_r|}\int_{B_r(x)} a_{ij}(s,z)dz| dy ds \leq \gamma_0$
2. Reifenberg-flat condition: For any $(t,x) \in \mathbb{R} \times \partial\Omega$ and $r \in (0, R_0]$, there exists a coordinate system such that: $\{y_1 > 0\} \cap B_r(x) \subset \Omega_r(x) \subset \{y_1 > -2\gamma_0 r\} \cap B_r(x)$

Main Results

Theorem 2.8 (Main Theorem)

Let $p, q \in (1,\infty)$, $\omega_1 \in A_p(\mathbb{R}^d)$, $\omega_2 \in A_q(\mathbb{R})$, with $[\omega_1]_{A_p} + [\omega_2]_{A_q} \leq K$. There exist $\gamma_0 = \gamma_0(d,\delta,p,q,K)$ and $N = N(d,\delta,p,q,K)$ such that under Assumption 2.6 ($\gamma_0$), for $u \in H_{p,q,\omega}^{1/2,1}(Q)$ satisfying Equation (2.7) and conormal derivative boundary conditions:

$\|D_t^{1/2}u\|_{L_{p,q,\omega}(Q)} + \|Du\|_{L_{p,q,\omega}(Q)} + \sqrt{\lambda}\|u\|_{L_{p,q,\omega}(Q)}$$\leq N(\|h\|_{L_{p,q,\omega}(Q)} + \|g_i\|_{L_{p,q,\omega}(Q)} + \lambda^{-1/2}\|f\|_{L_{p,q,\omega}(Q)})$

holds for $\lambda \geq \lambda_0$. For $\lambda > \lambda_0$, the solution exists and is unique.

Theorem 2.9 (Standard Equations)

For standard equations without half-time derivatives: $u_t - D_i(a_{ij}D_ju) + \lambda u = D_ig_i + f \quad \text{in } \Omega_T = (-\infty, T) \times \Omega$

Under the same assumptions: $\|Du\|_{L_{p,q,\omega}(\Omega_T)} + \sqrt{\lambda}\|u\|_{L_{p,q,\omega}(\Omega_T)} \leq N(\|g_i\|_{L_{p,q,\omega}(\Omega_T)} + \lambda^{-1/2}\|f\|_{L_{p,q,\omega}(\Omega_T)})$

Corollary 2.10 (Cauchy Problem)

For the initial value problem on $(0,T) \times \Omega$ with $u(0,x) = 0$ and $\lambda = 0$: $\|Du\|_{L_{p,q,\omega}((0,T)\times\Omega)} + \|u\|_{L_{p,q,\omega}((0,T)\times\Omega)} \leq N(\|g_i\|_{L_{p,q,\omega}} + \|f\|_{L_{p,q,\omega}})$

Proof Framework

Step 1: Basic Solvability (Section 3)

• Theorem 3.1: $L^2$ solvability without regularity assumptions (Lax-Milgram)
• Proposition 3.3: $L^p$ solvability for special coefficients
• Lemma 3.6: Local estimates (iterative argument)
• Lemmas 3.7-3.12: Estimates on half-spaces and interiors with arbitrary integrability

Step 2: Estimates on Reifenberg-flat Domains (Section 4)

• Lemma 4.1: Key equation transformation introducing half-time derivatives
• Proposition 4.2: $W+V$ decomposition, special treatment for heat equations
• Proposition 4.5: Estimates involving $D_t^{1/2}u$

Step 3: Proof of Main Theorem (Section 5)

• Level set argument:
  • Define $A(s) = \{|U| > s\}$ and $B(s)$ (involving maximal functions)
  • Show that $CR(X_0) \subset B(s)$ implies $|QR(X_0) \cap A(\kappa s)| < \gamma|QR(X_0)|$
  • Apply "crawling of ink spots" lemma to obtain $\omega(A(\kappa s)) \leq N\gamma^\beta\omega(B(s))$
• Norm estimates: $\|U\|_{L_p^p} = p\int_0^\infty \omega(A(s))s^{p-1}ds \leq I_1 + I_2$ where $I_1$ involves estimates on small cylinders and $I_2$ is controlled via Hardy-Littlewood maximal function theorem.
• Extrapolation theorem: Generalize from $p=q$ case to general $p \neq q$ (using extrapolation theorem from 18)

Step 4: Standard Equations (Theorem 2.9 & Corollary 2.10)

• Transform problem on $(-\infty,T)$ to $\mathbb{R}$ through zero extension
• Apply results from Theorem 2.8
• Uniqueness: Lemma 5.1 via duality argument ($p \geq 2$) and energy methods

Technical Difficulties and Innovations

Difficulty 1: Time Irregularity

Problem: Coefficients $a_{ij}(t)$ are merely measurable, so direct differentiation with respect to $t$ is impossible.

Solution:

• Do not directly handle $u_t$
• Instead estimate $D_t^{1/2}u$ (half-order derivative)
• Use the identity $u_t I_{\Omega^*} = -D_t^{1/2}(\mathcal{H}(D_t^{1/2}u)I_{\Omega^*})$

Difficulty 2: Boundary Irregularity

Problem: Reifenberg-flat domains only satisfy $\epsilon$-approximation by planes condition.

Solution:

• Local coordinate transformation to approximate boundary by plane
• Control the measure of "bad set" $\Omega^*$ ($|\Omega^*| \leq N\gamma_0 R^d$)
• Absorb error terms through Hölder inequality and smallness of $\gamma_0$

Difficulty 3: Weighted Mixed Norms

Problem: $L_{p,q,\omega}$ spaces are more complex than $L_p$.

Solution:

• Use reverse Hölder inequality for $A_p$ weights (Equation 5.1)
• Control weights at different scales through doubling property
• Extrapolation theorem to generalize from $p=q$ to general $p \neq q$

Summary of Innovations

1. First systematic use of half-time derivatives: Novel application in conormal derivative problems
2. New equation form: Study equations containing $D_t^{1/2}h$ terms
3. Refined decomposition technique: $W+V$ decomposition with $W$ controllable and $V$ regular
4. Small parameter strategy: Achieve absorption by choosing $\gamma_0$ and $\gamma$ sufficiently small

Related Work

Elliptic Equations

• VMO coefficients: Chiarenza-Frasca-Longo 9,10, Caffarelli-Peral 8 using singular integral techniques
• Reifenberg-flat domains: Byun-Wang 3,5,6 using level set arguments and "crawling of ink spots" lemma
• Conormal derivative problems: Byun-Wang 6 (elliptic case), Dong-Kim 17 (higher-order systems)

Parabolic Equations

• Singular integral methods: Bramanti-Cerutti 2
• Time measurable coefficients: Krylov 26 first studied this, using sharp functions and Fefferman-Stein theorem
• Dirichlet problems: Dong-Kim 14,16 established interior and boundary estimates
• Weighted theory: Dong-Kim 18, Choi-Kim 12

Position of This Work

• First result: Time measurable coefficients + Reifenberg-flat domains + conormal derivative boundary conditions
• Method innovation: Half-time derivative technique is the key breakthrough
• Theory completion: Fills important gap in $L_p$ theory for parabolic equations

Conclusions and Discussion

Main Conclusions

1. Establish complete $L_p$ theory for conormal derivative problems with time-measurable coefficients on Reifenberg-flat domains
2. Half-time derivatives are effective tools for handling time irregularity
3. Weighted mixed-norm estimates are optimal (consistent with known elliptic results)

Theoretical Significance

• Perfects the $L_p$ theory system for parabolic equations
• Provides new methods for more general boundary value problems
• Half-time derivative technique may have broader applications

Limitations

1. Parameter restrictions: Requires $\lambda \geq \lambda_0$ (possibly nonzero); resolved for Cauchy problems via exponential transformation $w = e^{-\lambda t}u$
2. Coefficient conditions: Requires small BMO in spatial variables ($\gamma_0$ small); cannot handle arbitrary VMO
3. Domain restrictions: Reifenberg-flat condition; does not include more general Lipschitz domains
4. Dimension dependence: Constants depend on dimension $d$

Future Directions

1. Extension to systems: Remark 2.12 mentions possible generalization but not detailed
2. Time weights: Consider more general time weights $\omega_2 \in A_q((0,T))$ (Remark 2.11)
3. Nonlinear problems: Can methods apply to quasilinear equations?
4. Optimal $\gamma_0$: Determine precise dependence of $\gamma_0$
5. $\lambda_0 = 0$: Can the lower bound on $\lambda$ be removed?

In-Depth Evaluation

Strengths

1. Methodological Innovation ★★★★★

• Breakthrough technique: Application of half-time derivatives in conormal derivative problems is truly innovative, elegantly solving a long-standing open problem
• Technical depth: Combines multiple advanced techniques (level sets, maximal functions, weighted theory, fractional calculus)
• Systematicity: Clear hierarchy from $L^2$ to $L_p$, from unweighted to weighted

2. Theoretical Completeness ★★★★★

• Existence and uniqueness: Not only estimates but complete solvability theory
• Multiple boundary conditions: Covers Dirichlet, Neumann, and mixed boundary conditions
• Generalizability: Derives results for standard equations from equations with half-time derivatives

3. Technical Rigor ★★★★★

• Detailed proofs: 42 pages of detailed proofs with every technical detail explained
• Complete lemmas: Numerous auxiliary lemmas support main results
• Boundary cases: Careful treatment of various boundary situations (interior, boundary, mixed)

4. Writing Quality ★★★★☆

• Clear structure: Introduction → main results → auxiliary results → boundary estimates → proof of main theorem
• Clear motivation: Section 1 clearly explains why half-time derivatives are needed
• Standard notation: Complete notation definitions (Section 2)

Weaknesses

1. Technical Complexity

• Readability: Difficult for non-specialists; requires deep PDE background
• Tedious calculations: Many estimates involve complex inequality chains (e.g., derivation of (4.11))
• Parameter selection: Logic for choosing multiple small parameters $\gamma_0, \gamma, p_0, p_1, \nu$ requires careful understanding

2. Result Limitations

• $\lambda$ restriction: Requires $\lambda \geq \lambda_0 > 0$ (though Cauchy problem resolved via transformation)
• Coefficient conditions: Small BMO is stronger than VMO
• Constant dependence: Dependence of constants on parameters not explicitly given

3. Application Guidance

• Concrete examples: Lack specific examples showing how to verify Assumption 2.6
• Numerical aspects: No numerical experiments or algorithms
• Practical scenarios: Physical application background not discussed

4. Comparison with Existing Results

• Quantitative comparison: No quantitative comparison with known results (e.g., time-continuous coefficient case)
• Optimality: Sharpness of estimates not discussed

Impact Assessment

Contribution to Field ★★★★★

• Fills gap: Solves important open problem in $L_p$ theory for parabolic equations
• Methodology: Half-time derivative technique may inspire research on other problems
• Citation potential: Expected to become important reference in this direction

Practical Value ★★★☆☆

• Theory-focused: Primarily theoretical contribution; direct application value limited
• Potential applications: May apply to irregular boundary problems in fluid mechanics, heat conduction
• Numerical methods: Can provide theoretical support for numerical methods

Reproducibility ★★★★☆

• Complete proofs: All proof steps detailed
• Clear citations: Known results clearly cited
• Technical threshold: Requires specialized knowledge to verify and apply

Applicable Scenarios

1. Theoretical Research:
   • Study more general parabolic boundary value problems
   • Develop fractional-order time derivative theory
   • Extend to nonlinear or higher-order equations
2. Numerical Analysis:
   • Provide error estimates for numerical methods on irregular domains
   • Guide adaptive mesh algorithms
3. Application Fields:
   • Heat conduction in non-homogeneous media (irregular coefficients)
   • Fluid problems with complex geometric boundaries
   • Option pricing in mathematical finance (irregular boundary conditions)

Recommendations for Future Research

1. Short-term:
   • Provide concrete examples of Reifenberg-flat domains
   • Study precise value of $\lambda_0$
   • Extend to parabolic systems
2. Medium-term:
   • Develop numerical methods
   • Apply to specific physical models
   • Study nonlinear generalizations
3. Long-term:
   • Develop more general fractional-order time derivative theory
   • Extend to more irregular domains and coefficients
   • Develop corresponding stochastic PDE theory

Overall Assessment

This is an outstanding theoretical work that solves an important open problem in $L_p$ theory for parabolic equations. The introduction of half-time derivative technique is a genuine innovation that elegantly overcomes technical obstacles arising from the combination of time irregularity and boundary irregularity. The proofs are rigorous and complete, with a well-developed theoretical system.

Recommendation: ★★★★★ (Strongly Recommended)

Target Readers:

• Theoretical PDE researchers
• Harmonic and real analysis experts
• Numerical analysis researchers (theoretical foundation)

Reading Suggestions:

1. First understand Section 1 motivation (why half-time derivatives are needed)
2. Carefully study Lemma 4.1 (core technique)
3. Follow level set argument in Section 5
4. For beginners, start with $L^2$ case (Theorem 3.1) and unweighted case

Key References

6 S.-S. Byun and L. Wang, The conormal derivative problem for elliptic equations with BMO coefficients on Reifenberg flat domains, Proc. London Math. Soc. (2005) - Pioneering work for elliptic case

16 H. Dong and D. Kim, On the $L_p$-solvability of higher order parabolic and elliptic systems with BMO coefficients, Arch. Ration. Mech. Anal. (2011) - Important reference for Dirichlet problems

18 H. Dong and D. Kim, On $L_p$-estimates for elliptic and parabolic equations with $A_p$ weights, Trans. Amer. Math. Soc. (2018) - Foundation for weighted theory

24 P. Jung and D. Kim, $L_p$-estimates for parabolic equations in divergence form with a half-time derivative, J. Differential Equations (2025) - Direct predecessor of half-time derivative technique

26 N. V. Krylov, Parabolic and elliptic equations with VMO coefficients, Comm. Partial Differential Equations (2007) - Pioneering work on time measurable coefficients