$L^{\infty}$-error estimates for Neumann boundary value problems on graded meshes

Thomas Apel, Johannes Pfefferer, Sergejs Rogovs, Max Winkler

Abstract

This paper investigates a priori pointwise error estimates for finite element solutions of boundary value problems with Neumann boundary conditions in polygonal domains. Due to the presence of domain corners, the convergence rate of numerical solutions is typically lower than in smooth domains. To address this, the authors investigate the use of local mesh refinement (graded meshes) near corners. By exploiting regularity results in weighted Sobolev spaces, this work provides the first sharp convergence order of $h^2 |\ln h|$ for piecewise linear finite element approximations in the Neumann case. Furthermore, the results are extended to semilinear boundary value problems using a supercloseness argument between the Ritz projection and the finite element solution.

Introduction

In the field of partial differential equations (PDEs), the Finite Element Method (FEM) is a cornerstone for numerical simulation. While standard error analysis often assumes smooth domain boundaries, real-world engineering applications frequently involve polygonal geometries with sharp corners. These corners introduce singularities where the solution's derivatives can become unbounded, significantly degrading the accuracy and convergence rate of standard numerical methods.

While the "Dirichlet" case (fixed boundary values) has been extensively studied, the "Neumann" case (fixed flux on the boundary) presents unique technical challenges. Specifically, there has been a lack of sharp pointwise ($L^\infty$) error estimates for Neumann problems on graded meshes. Establishing these estimates is critical for applications where the maximum error must be strictly controlled, such as in safety-critical engineering or optimal control problems.

Methodology

To achieve sharp error estimates, the authors utilize a combination of mesh refinement strategies and advanced functional analysis:

Graded Meshes

Rather than employing a uniform grid, the authors use "graded meshes" that become systematically denser as they approach the singular corners of the domain. This local refinement compensates for the reduced regularity of the solution near the corners, allowing the global error to maintain a high order of convergence.

Weighted Sobolev Spaces and Dyadic Decomposition

The analysis is conducted within weighted Sobolev spaces, where derivatives are weighted by their distance from the domain corners. To prove the $L^\infty$ estimates, the authors employ a dyadic decomposition technique, partitioning the domain into concentric shells surrounding the corners. Local estimates are derived for each shell and then aggregated to form a global bound.

Supercloseness in Semilinear Problems

A significant contribution of this work is the extension of linear results to semilinear (nonlinear) problems. The authors demonstrate that the finite element solution is "superclose" to the Ritz projection of the continuous solution. This relationship allows the sharp convergence rates established for linear equations to be transferred to the nonlinear regime.

Key Insights

  • Sharp Logarithmic Terms: Previous research often resulted in sub-optimal logarithmic terms, such as $|\ln h|^{3/2}$. This paper proves that the exponent is exactly one ($|\ln h|$), providing a tighter and more accurate bound.
  • Neumann-Specific Challenges: The proof for Neumann conditions is more complex than the Dirichlet counterpart because the boundary data requires more sophisticated handling in weighted spaces, and standard tools like the Poincaré inequality cannot be applied directly.
  • Linear to Nonlinear Transferability: The use of supercloseness provides a powerful framework for extending high-quality linear estimates to a broader range of physical phenomena, including heat radiation and chemical reaction models.

Numerical Validation

The theoretical findings were verified using a benchmark numerical example involving a circular sector with an interior angle $\omega$. The study compared a convex domain ($\omega = 3\pi/4$) and a non-convex L-shaped domain ($\omega = 3\pi/2$).

In the L-shaped domain test, the results were striking:

  • Uniform Mesh: The convergence rate was only 0.67, far below the ideal rate for linear elements.
  • Graded Mesh: By applying a grading parameter ($\mu = 0.3$), the convergence rate improved to 1.96, closely matching the theoretical "Gold Standard" of 2.0.

Conclusion

This paper establishes the first sharp $L^\infty$ error estimates for Neumann problems on polygonal domains using graded meshes. By precisely specifying the required regularity of input data and proving the $h^2 |\ln h|$ convergence rate, the authors provide a rigorous foundation for practitioners. While the current work focuses on 2D domains, it sets the stage for future research into 3D polyhedral singularities and the convergence of optimal control problems involving Neumann boundary conditions.

Summary updated

Abstract

This paper deals with a priori pointwise error estimates for the finite element solution of boundary value problems with Neumann boundary conditions in polygonal domains. Due to the corners of the domain, the convergence rate of the numerical solutions can be lower than in case of smooth domains. As a remedy the use of local mesh refinement near the corners is investigated. In order to prove quasi-optimal a priori error estimates regularity results in weighted Sobolev spaces are exploited. This is the first work on the Neumann boundary value problem where both the regularity of the data is exactly specified and the sharp convergence order $h^2 \lvert \ln h \rvert $ in the case of piecewise linear finite element approximations is obtained. As an extension we show the same rate for the approximate solution of a semilinear boundary value problem. The proof relies in this case on the supercloseness between the Ritz projection to the continuous solution and the finite element solution.
BibTeX
@inproceedings{apel2018l,
  author = {Thomas Apel and Johannes Pfefferer and Sergejs Rogovs and Max Winkler},
  title = {{$L^\{\textbackslash\{\}infty\}$-error estimates for Neumann boundary value problems on graded meshes}},
  booktitle = {IMA},
  year = {2018},
  doi = {10.1093/imanum/dry076},
}
Figure