Weak Galerkin methods for second order elliptic interface problems

Guo-Wei Wei, Lin Mu, Junping Wang, Guowei Wei 0001, Xiu Ye, Shan ZhaoJournal of Computational Physics 2013 · DOI

TL;DR

This paper introduces a novel Weak Galerkin Finite Element Method (WG-FEM) for solving second-order elliptic partial differential equations with discontinuous coefficients and interfaces. The method demonstrates remarkable accuracy, achieving higher-order convergence rates even for solutions with low regularity due to nonsmooth interfaces, outperforming existing methods.

What Problem Does This Paper Address?

Many real-world phenomena, from fluid mechanics to electromagnetics and biological processes, are modeled by elliptic interface problems. These problems are characterized by equations with discontinuous coefficients and the need to satisfy specific conditions at interfaces between different materials or regions. A significant challenge arises when these interfaces are nonsmooth (e.g., sharp edges, cusps), leading to solutions with low regularity (i.e., the solution or its derivatives are not smooth). Existing numerical methods often struggle to achieve high-order accuracy in such scenarios, with many limited to second-order convergence. This paper aims to develop a robust and accurate numerical method that can effectively handle these challenging interface problems, particularly those with nonsmooth interfaces and low-regularity solutions.

Methods

The core of this paper is the development and analysis of a Weak Galerkin Finite Element Method (WG-FEM). Unlike traditional finite element methods that enforce continuity of solution components strongly, WG-FEM utilizes discontinuous functions within elements and enforces continuity only weakly through carefully designed discrete differential operators. This approach offers significant flexibility in handling complex geometries and interface conditions.

Key aspects of the WG-FEM presented include:

  • Discontinuous Element Functions: Trial and test functions are allowed to be discontinuous across element boundaries.
  • Weak Continuity Enforcement: Continuity is enforced through specific terms in the weak formulation, avoiding the need for explicit continuity constraints on the discrete solution.
  • Body-Fitted Meshes: The method uses finite element meshes that align with the interface, which is crucial for accurately representing the geometry.
  • Raviart-Thomas Elements: The paper utilizes Raviart-Thomas elements for the discrete gradient, which are well-suited for mixed finite element methods and have desirable properties for handling divergence-free spaces.
  • Theoretical Framework: The paper provides a rigorous convergence analysis for the proposed WG-FEM, proving its uniqueness and deriving error estimates.

Datasets

Not discussed in the paper. The paper focuses on developing and validating a numerical method using known analytical solutions for various benchmark problems.

Key Insights and Ideas

  • Robustness with Nonsmooth Interfaces: The WG-FEM demonstrates remarkable robustness in handling nonsmooth interfaces, which are a major bottleneck for many existing numerical methods. It achieves high-order accuracy even when the interface has geometric singularities.
  • Handling Low-Regularity Solutions: The method's ability to maintain high-order convergence for solutions with low regularity (due to function singularities or geometric features) is a significant breakthrough. This is particularly important for accurately capturing the behavior of physical phenomena in complex scenarios.
  • Flexibility of Weak Formulation: The weak enforcement of continuity provides a powerful tool for dealing with interface conditions, allowing for more straightforward implementation and analysis compared to methods that require strict continuity.
  • Improved Accuracy over Existing Methods: The numerical results consistently show that the WG-FEM achieves better accuracy, particularly in the $L_\infty$ norm, compared to what is typically reported for other methods on similar problems, especially those with low-regularity solutions.

Key Results

The paper presents extensive numerical results for various 2D elliptic interface problems, including those with smooth and nonsmooth interfaces, discontinuous coefficients, and low-regularity solutions.

  • Second-Order Convergence for Smooth Interfaces: For problems with smooth interfaces and well-behaved solutions, the WG-FEM achieves second-order convergence in the $L_\infty$ norm for the solution and first-order convergence for the gradient, aligning with theoretical predictions.
  • High-Order Convergence for Nonsmooth Interfaces: For Lipschitz continuous interfaces and solutions with $C^1$ continuity, the WG-FEM achieves convergence rates close to second order for the solution and first order for the gradient.
  • Significant Improvement for Low-Regularity Solutions: For solutions with $H^2$ continuity (indicating singularities), the WG-FEM achieves convergence rates of approximately 1.75 for the solution and 1 for the gradient in the $L_\infty$ norm. This is a substantial improvement over existing methods, which are often limited to first-order convergence in such cases.
  • Robustness Across Examples: The method consistently demonstrates good performance across a range of challenging examples, including those with high contrast in coefficients and complex interface geometries.

Limitations and Future Work

  • Focus on 2D: The current study is limited to two-dimensional problems. While the principles are expected to extend to 3D, further investigation is needed.
  • Mesh Quality: The paper notes that mesh quality can affect the convergence rates, particularly for less regular meshes. While the WG-FEM is robust, careful mesh generation, especially around complex interface features, can further enhance accuracy.
  • Higher-Order WG-FEM: The paper primarily uses piecewise constant finite elements ($P_0$). The authors suggest that exploring higher-order WG-FEMs could lead to even better accuracy and is a direction for future research.
  • Analytical Solutions: The numerical validation relies on problems with known analytical solutions. Extending the validation to problems where analytical solutions are not available would be valuable.
  • 3D Extension: Extending the WG-FEM to three-dimensional interface problems is a natural and important next step.
  • Other PDE Types: Investigating the application of WG-FEM to other types of PDEs, such as parabolic or hyperbolic equations with interfaces, would broaden its applicability.

Summary updated

BibTeX
@article{wei2013weak,
  author = {Guo-Wei Wei and Lin Mu and Junping Wang and Guowei Wei and Xiu Ye and Shan Zhao},
  title = {{Weak Galerkin methods for second order elliptic interface problems}},
  journal = {Journal of Computational Physics},
  year = {2013},
  doi = {10.1016/j.jcp.2013.04.042},
}
Figure