🔢 mathematics

A Nonlocal $p$ -Laplacian Interface Model with Sharp Interface

This paper proposes an energy-based nonlocal $p$ -Laplacian interface model that preserves a sharp interface and incorporates various boundary conditions, proving its convergence to local counterparts via $\Gamma$ -convergence and validating the results through numerical experiments.

Original authors: Kehan Shi, Zuoqiang Shi, Tangjun Wang

Published 2026-06-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kehan Shi, Zuoqiang Shi, Tangjun Wang

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

Imagine you are trying to model how heat spreads through a room, or how water flows through a sponge. Usually, we use math to describe how things change from one tiny point to the next right beside it. This is like looking at a picture through a microscope: you see the immediate neighbors.

But sometimes, things interact over a distance. Maybe a drop of ink in water doesn't just affect the water touching it, but spreads its influence a little further out. This is called a "nonlocal" interaction.

This paper introduces a new mathematical tool to handle these "long-distance" interactions, specifically when there is a sharp boundary (an interface) separating two different materials.

Here is the breakdown of what the authors did, using everyday analogies:

1. The Problem: The "Wall" Between Two Worlds

Imagine a room divided by a thin, invisible wall. On the left side, the floor is made of wood; on the right, it's made of tile.

The Local Way: Traditional math says, "If you are on the wood, you only talk to the wood next to you. If you are on the tile, you only talk to the tile." At the wall, the rules get tricky. You have to manually tell the math, "Hey, the temperature must be the same on both sides of the wall, but the heat flow might jump."
The Nonlocal Way: The authors wanted to create a model where the wood and tile "feel" each other across a small distance, not just at the exact edge. The challenge? Standard nonlocal math usually blurs the wall, turning that sharp line into a fuzzy, thick zone. The authors wanted to keep the wall sharp (crisp and clear) while still allowing for long-distance interactions.

2. The Solution: A "Penalty" System

The authors built a new mathematical "scorecard" (called an Energy Functional) to find the best solution. Think of this scorecard as a game where you want to minimize the "cost" of the system.

The Diffusion Cost: This part measures how much the material "stretches" or changes between points. If points far apart are very different, the cost goes up.
The Boundary Penalty: This is the clever part. To keep the wall sharp, they added a special rule: If the value on the wood side and the value on the tile side are different right at the wall, you get a huge penalty.
- This forces the solution to be continuous (smooth) across the wall, just like in the real world, without needing to blur the wall into a thick zone.
The "Ghost" Interaction: They also used a "smoothing" technique (averaging) to handle the specific rules about how heat or flow jumps across the wall. It's like having a referee who looks at the average of the wood side and the average of the tile side to decide if the rules are being followed.

3. The "p-Laplacian" Twist

The paper also shows this method works for more complex, "non-linear" situations.

Imagine the floor isn't just wood or tile, but a material that gets stiffer the harder you push it. This is the p-Laplacian problem.
The authors showed their "sharp wall" method works here too, even when the material behaves in a tricky, non-linear way. They proved that their new model is flexible enough to handle these complex "membrane" conditions where the flow depends on the difference between the two sides.

4. The Proof: Zooming Out

The most important claim of the paper is about convergence.

Imagine you have a nonlocal model with a "horizon" (a distance over which points can talk to each other). Let's call this distance $\delta$ (delta).
The authors proved mathematically that as you shrink this distance $\delta$ down to zero (making the interaction purely local), their new model perfectly transforms into the traditional, standard model we already know and trust.
It's like taking a high-resolution photo and slowly zooming out until it looks exactly like a standard painting. The "fuzzy" nonlocal effects disappear, and you are left with the sharp, classic interface problem.

5. The Computer Test

Finally, they didn't just do the math on paper; they built a computer program to test it.

They created a virtual room with a sharp boundary.
They ran the simulation with different "horizon" sizes.
The Result: As they made the horizon smaller, the computer's answer got closer and closer to the known "true" answer. The error dropped at a predictable, steady rate (first-order convergence), proving their method is accurate and efficient.

Summary

In short, the authors created a new way to model problems with sharp boundaries (like a wall between two materials) that allows for long-distance interactions.

Old way: Blur the wall to make math easier, or struggle to force the wall to stay sharp.
Their way: Keep the wall sharp, use a "penalty" to enforce the rules, and prove that as the "long-distance" effect shrinks away, the model naturally becomes the standard one we already use.

They verified this with rigorous math and computer simulations, showing that their method is a robust, flexible tool for solving complex interface problems.

Technical Summary: A Nonlocal p-Laplacian Interface Model with Sharp Interface

Problem Statement
The paper addresses the challenge of formulating interface problems within a nonlocal framework. Classical interface problems involve partial differential equations (PDEs) defined on a domain $\Omega$ divided by an interface $\Gamma$ , where the solution or its derivatives exhibit discontinuities. Specifically, the authors consider the extension of the Poisson interface problem and the $p$ -Laplacian interface problem with membrane conditions to nonlocal settings. A primary difficulty in nonlocal formulations is that classical local interface conditions (such as continuity of the solution and prescribed jumps in normal flux) cannot be directly imposed. Existing approaches often approximate the sharp interface $\Gamma$ with a volumetric layer of thickness $\delta$ , which can complicate the modeling of independent diffusion in subdomains and the handling of boundary data.

Methodology
The authors propose a novel energy-based nonlocal approximation that retains a sharp interface $\Gamma$ . Instead of modifying the domain geometry, the model utilizes specific energy functionals where the interface conditions are encoded implicitly.

Energy Formulation:
- For the linear case ( $p=2$ ), the local energy functional $E(u)$ is approximated by a nonlocal functional $E_\delta(u)$ . The nonlocal diffusion term is constructed using a kernel $\Lambda(x,y)$ that restricts interactions to within subdomains $\Omega_1$ and $\Omega_2$ , ensuring independent nonlocal diffusion in each region.
- The Dirichlet interface condition (jump in flux) is handled via a penalty term involving nonlocal averages ( $K_{\delta,i} * u$ ) along the interface $\Gamma$ .
- The Neumann boundary condition is naturally incorporated through the variational structure.
- The framework is extended to the nonlinear $p$ -Laplacian case with membrane conditions, where the jump in the solution across the interface is penalized. The resulting nonlocal functional $F_\delta(u)$ shares the same structural form as $E_\delta(u)$ .
Theoretical Analysis:
- Existence and Uniqueness: The authors establish the existence and uniqueness of minimizers for the nonlocal functionals $E_\delta$ and $F_\delta$ using the direct method of the calculus of variations. This relies on a newly derived nonlocal Poincaré inequality that accounts for the interface jump penalty.
- $\Gamma$ -Convergence: The core theoretical contribution is the proof of $\Gamma$ -convergence. The authors demonstrate that as the nonlocal horizon $\delta$ vanishes ( $\delta \to 0$ ), the nonlocal functionals $\Gamma$ -converge to their local counterparts ( $E$ and $F$ ).
- Compactness: Combined with compactness results for nonlocal $p$ -Laplacian operators, the $\Gamma$ -convergence implies that the minimizers of the nonlocal problems converge to the minimizers of the local interface problems in the $L^p$ norm.
Numerical Implementation:
- The authors employ an efficient finite element method based on the tensor-product structure of Gaussian kernels and multi-cubic polynomial bases.
- This approach decomposes high-dimensional integrals into products of one-dimensional integrals, significantly reducing computational cost.
- Special quadrature rules (Gauss-Legendre) are tailored to handle the boundary layers arising from the cubic basis functions near cell interfaces.

Key Results

Sharp Interface Preservation: Unlike previous volume-constraint approaches, the proposed model maintains a sharp interface, allowing for independent nonlocal diffusion in disjoint subdomains without extending the interface data into the bulk.
Convergence Proof: Rigorous proofs confirm that the minimizers of the nonlocal functionals converge to the local minimizers as $\delta \to 0$ . The convergence rate is observed to be first-order.
Numerical Validation: Numerical experiments in one and two dimensions verify the theoretical convergence. The $L^2$ error between the nonlocal solution and the ground-truth local solution decays linearly with $\delta$ , and the results are robust across different mesh resolutions.

Significance and Claims
The paper claims that the proposed sharp-interface nonlocal model offers a more natural and flexible framework for interface problems compared to existing volumetric approximations. By formulating the problem via energy functionals, the difficult-to-handle normal flux interface conditions are implicitly satisfied. The authors highlight that this approach:

Facilitates the extension to more complex interface problems, as demonstrated by the successful application to the nonlinear $p$ -Laplacian with membrane conditions.
Avoids the need to extend interface data (e.g., $g$ ) from $L^2(\Gamma)$ to $H^1(\Omega)$ .
Provides a mathematically rigorous link between nonlocal and local interface models through $\Gamma$ -convergence, ensuring that the nonlocal model is a valid approximation of the classical physical problem as the horizon vanishes.

The work does not propose new physical applications but rather provides a robust mathematical and numerical tool for modeling existing interface phenomena in fields such as fluid dynamics and materials science where nonlocal effects are relevant.