🔢 mathematics

Robust near-diagonal Green function estimates

This paper establishes sharp, robust near-diagonal pointwise bounds for the Green function of nonlocal fractional operators as the order approaches 2, achieving these results independently of the Dirichlet heat kernel to cover cases where the heat kernel fails to satisfy isotropic bounds.

Original authors: Moritz Kassmann, Minhyun Kim, Ki-Ahm Lee

Published 2026-08-28

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Moritz Kassmann, Minhyun Kim, Ki-Ahm Lee

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the vast landscape of mathematics, there is a branch dedicated to understanding how things spread, settle, and interact across space. This field, known as potential analysis, often deals with the invisible forces that govern everything from the flow of electricity to the movement of heat. At the heart of this study lies a powerful mathematical tool called the Green function. You can think of this function as a map that reveals how a single point of influence, like a drop of ink in water or a single spark in a dark room, affects its entire surroundings. For centuries, mathematicians have known how to draw these maps for simple, smooth systems, but the world is often messy and complex. In recent decades, scientists have turned their attention to "nonlocal" systems, where a point does not just interact with its immediate neighbors but can reach out and influence points far away, much like a radio signal that skips over obstacles to hit a distant receiver.

For a specific type of these long-range interactions, defined by a parameter that controls how far the influence reaches, mathematicians have long sought a precise description of the Green function. The challenge has been that the standard methods used to create these maps rely on a different, related tool called the heat kernel, which describes how a system evolves over time. While this approach works well for fixed settings, it breaks down when the system approaches a critical threshold where the rules of interaction change from long-range jumps to short-range steps. At this tipping point, the old tools become unreliable, and the constants in the equations blow up, making it impossible to see the true behavior of the system. This gap has left a blind spot in our understanding of how these nonlocal systems behave as they transition into the familiar, smooth world we experience every day.

A team of researchers has now filled this gap by proving that the Green function for these nonlocal systems follows a very specific, predictable pattern near the point of influence, even as the system approaches that critical threshold. Their work demonstrates that the strength of the influence at a distance follows a clear rule: the effect gets weaker as you move away, but it does so in a way that remains consistent and stable, regardless of how close the system gets to the transition point. Crucially, they achieved this without using the heat kernel, the very tool that had previously caused the equations to fail. By developing a new approach that bypasses the heat kernel entirely, they were able to show that the Green function remains well-behaved and robust, providing a reliable map even in situations where the time-evolution map is chaotic or undefined.

The significance of this finding lies in its ability to unify two different worlds. In the past, mathematicians could describe the behavior of these systems when the interaction range was fixed, but they could not guarantee that the description would hold up as the system changed. The new results prove that the bounds on the Green function stay uniform and do not degenerate as the system approaches the limit. This means that the complex, long-range interactions can be understood with the same clarity as the smooth, short-range interactions that govern classical physics. The researchers showed that this stability holds true even for systems that are not perfectly symmetric, meaning the influence can be stronger in some directions than others, a scenario where previous methods failed completely.

To reach this conclusion, the team had to construct a new framework for analyzing these functions. They started by proving that a unique Green function exists for a wide variety of these nonlocal operators, establishing a solid foundation before attempting to measure its properties. They then developed a series of estimates, which are mathematical bounds that limit how large or small the function can be. By carefully tracking the constants in these bounds, they demonstrated that the values do not explode or vanish as the system's parameters change. This required them to ignore the traditional heat kernel method and instead rely on direct properties of the interaction rules themselves. They showed that even when the underlying rules for interaction are irregular or anisotropic, the resulting Green function still adheres to a strict, isotropic pattern near the source, behaving as if the system were perfectly balanced.

The researchers also addressed the question of symmetry, proving that the influence of point A on point B is exactly the same as the influence of point B on point A, provided the interaction rules are symmetric. This might seem obvious, but in the complex world of nonlocal operators, proving such a fundamental property requires rigorous demonstration. Their proof relies on the stability of the function they established earlier, showing that the limit of their approximations preserves this symmetry. This result is essential for ensuring that the mathematical model is consistent with physical reality, where the laws of interaction usually do not depend on the direction of observation.

One of the most striking aspects of their work is the discovery that the Green function can be well-behaved even when the heat kernel is not. In some complex scenarios, the way a system evolves over time is so erratic that it cannot be bounded by simple rules, yet the final, steady-state map of influence remains smooth and predictable. This separation of behaviors was previously unknown and suggests that the Green function contains more robust information about the system than the time-dependent evolution does. The researchers provided specific examples where the heat kernel fails to show rotational symmetry, meaning the spread of influence looks different depending on the direction, yet the Green function maintains a consistent, rotationally symmetric bound. This finding challenges the assumption that the two tools must always behave in tandem and opens the door to studying systems that were previously considered too irregular to analyze.

The paper concludes by confirming that these results are not just theoretical curiosities but apply to a broad class of operators that model real-world phenomena. The conditions required for the proof are mild enough to cover many practical cases, including those with rough or irregular coefficients. By establishing these sharp, near-diagonal bounds, the authors have provided a reliable tool for future research, allowing mathematicians to study nonlocal systems with the same confidence they have in classical systems. The work stands as a testament to the power of finding new paths through old problems, showing that by stepping away from established methods, one can uncover deeper truths that were previously hidden in the noise of degenerating constants.

Technical Summary: Robust Near-Diagonal Green Function Estimates

Problem Statement and Motivation
The paper addresses the derivation of near-diagonal pointwise bounds for the Green function $G_\Omega(x, y)$ associated with nonlocal operators of fractional order $\alpha \in (0, 2)$ . While explicit formulas and sharp estimates exist for specific cases (e.g., the fractional Laplacian on a ball), existing literature often relies on the representation of the Green function as the time-integral of the Dirichlet heat kernel, $G^{(\alpha)}(x, y) = \int_0^\infty p^{(\alpha)}_\Omega(t; x, y) dt$ .

A critical limitation of this classical approach is that the constants in the resulting estimates degenerate as $\alpha \to 2^-$ . This occurs because the decay properties of the heat kernel shift from polynomial to exponential at $\alpha = 2$ , making it difficult to track constants uniformly. Furthermore, this method fails in scenarios where the Green function satisfies isotropic bounds but the corresponding heat kernel does not (e.g., operators with anisotropic or cone-restricted kernels). The authors aim to establish near-diagonal bounds where the constants remain uniform as $\alpha \to 2^-$ , thereby recovering the classical second-order bounds in the limit without relying on heat kernel estimates. The authors explicitly note that while their results are "much weaker in many respects" compared to existing sharp estimates for fixed $\alpha$ , they provide the crucial advantage of robustness.

Methodology and Framework
The authors develop a variational approach that avoids the heat kernel entirely. The operator $L$ is defined via a symmetric measurable kernel $k(x, y)$ on $\mathbb{R}^d \times \mathbb{R}^d \setminus \text{diag}$ , satisfying a specific integrability condition (1.1). The associated bilinear form is $E_\Omega(u, v) = \iint (u(y)-u(x))(v(y)-v(x))k(x, y) dy dx$ .

The analysis relies on a set of structural assumptions on the kernel $k$ :

(E $\ge$ ): A lower bound condition ensuring the energy form dominates the fractional Sobolev energy $E_\alpha$ .
(U1) & (U2): Upper bound conditions on the kernel, specifically $k(x, y) \le c_K(2-\alpha)|y-x|^{-d-\alpha}$ , which control the singularity and the "tail" behavior.
(H): An annulus Harnack inequality for the Green function.
(UJS): A condition related to the uniform jump size, which implies the Harnack inequality in conjunction with (E $\ge$ ) and (U2).

The proof strategy proceeds in several stages:

Existence and Uniqueness: Regularized Green functions $G_\rho$ are constructed using the Lax–Milgram theorem on bounded domains. Uniform estimates in weak $L^p$ spaces and fractional Sobolev spaces $W^{\beta/2, q}$ are established independently of the regularization parameter $\rho$ and the order $\alpha$ (away from zero).
Local Boundedness with Tail: The authors prove a local boundedness result (Theorem 4.2) for subsolutions involving a nonlocal tail term. This is achieved via Moser's iteration, adapted to handle the non-integrability of the Green function near the singularity by utilizing the uniform weak $L^p$ estimates derived earlier.
Upper Bounds: By combining the local boundedness result with the uniform weak $L^p$ estimates, the authors derive the upper bound $G(x, y) \le C|x-y|^{\alpha-d}$ .
Lower Bounds: The lower bound is established by modifying the classical proof for second-order operators (specifically the localization technique in [GW82]). The authors introduce cut-off functions to isolate the singularity and control nonlocal quantities (integrals over annuli and exterior domains) using the Harnack inequality and the structural assumptions on $k$ .
Symmetry: The symmetry of the Green function is proven using the convergence of regularized sequences and the continuity of the solutions.

Key Contributions and Results
The paper establishes the following main theorems:

Theorem 1.3 (Existence and Uniqueness): Under mild assumptions (E $\ge$ ), a Green function exists. If (U1) also holds, it is unique. The function satisfies uniform membership in $W^{\beta/2, q}$ and weak $L^{d/(d-\alpha)}$ spaces, with constants independent of $\alpha$ as $\alpha \to 2^-$ .
Theorem 1.4 (Upper Bound): Assuming (E $\ge$ ) and (U2), the Green function satisfies $G(x, y) \le C|x-y|^{\alpha-d}$ for all $x, y \in \Omega$ . The constant $C$ depends on $d, c_*, c_K, \alpha_0$ but is uniform in $\alpha$ and $\Omega$ .
Theorem 1.5 (Lower Bound): Assuming (E $\ge$ ), (U1), and the Harnack inequality (H), the Green function satisfies $G(x, y) \ge C|x-y|^{\alpha-d}$ for $|x-y| \le \text{dist}(y, \partial\Omega)/2$ . The constant is again robust as $\alpha \to 2^-$ .
Theorem 1.7 (Symmetry): Under (E $\ge$ ) and (U1), the Green function is symmetric, i.e., $G(x, y) = G(y, x)$ .

Significance and Claims
The primary novelty claimed by the authors is the robustness of the estimates. Unlike previous works where constants degenerate as $\alpha \to 2^-$ , the constants in these bounds remain uniform. This allows the recovery of near-diagonal bounds for second-order operators by taking the limit $\alpha \to 2^-$ .

A second significant contribution is the ability to handle cases where the heat kernel does not satisfy rotational bounds, yet the Green function does. The authors provide examples (Example 1.8) involving kernels supported on double cones or specific annular regions where the heat kernel lacks isotropic bounds, but the Green function satisfies the standard isotropic estimates derived in this paper. This demonstrates that the heat kernel approach is not necessary for establishing Green function bounds in these anisotropic settings.

The paper also corrects a serious mistake in a previous work ([KS02], Proposition 3.3) regarding the application of a theorem to functions that do not vanish on the complement of a specific domain.

Scope and Limitations
The results are established for dimensions $d \ge 3$ . The authors explicitly note that they do not discuss extensions to $d=2$ or $d=1$ with specific $\alpha$ ranges. The bounds are "near-diagonal," meaning the lower bound is valid only when $x$ is sufficiently close to $y$ relative to the boundary distance. The paper focuses on pointwise bounds and does not claim to provide global estimates up to the boundary for general non-translation-invariant kernels, noting this as an interesting task for future work.

More like this