Abstract
Solving fourth-order partial differential equations involving the quad-curl operator is a significant challenge in computational electromagnetics and plasma physics. This paper presents a decoupled finite element method (FEM) that decomposes the complex quad-curl problem into a sequence of simpler Maxwell and Stokes-type sub-problems. By utilizing standard Nédélec edge elements and nonconforming $P_1-P_0$ elements, the proposed method bypasses the need for high-order continuity elements. We provide a rigorous error analysis, including both a priori and a posteriori estimates, and introduce an adaptive mesh refinement strategy based on a novel separate marking technique to handle solution singularities in non-convex domains.
Introduction
The quad-curl operator, denoted as $(\text{curl})^4$, is a fundamental component in mathematical models such as magnetohydrodynamics (MHD) and various electromagnetic wave propagation problems. Despite its importance, solving the quad-curl problem directly using the Finite Element Method (FEM) is notoriously difficult. As a fourth-order problem, standard "conforming" approaches require $H(\text{curl}^2)$-conforming elements, which necessitate high levels of continuity. In three dimensions, these element spaces are computationally expensive and mathematically intricate to implement.
While previous research has explored decoupling strategies—breaking the fourth-order operator into lower-order systems—these methods often struggle with accuracy at the boundaries or fail to efficiently handle singularities arising from complex geometries, such as re-entrant corners. This study addresses these limitations by developing a robust decoupled formulation that leverages mature, "off-the-shelf" finite element spaces while maintaining high-order accuracy through adaptive refinement.
The Decoupled Formulation
The core of the proposed approach is a three-step decoupling mechanism that transforms the quad-curl problem into a series of well-understood second-order problems:
- Step 1 (Maxwell-type): Solve a curl-curl equation for an auxiliary variable $w$.
- Step 2 (Stokes-type): Solve a Stokes-type equation for a second auxiliary variable $\phi$, using the result from the previous step.
- Step 3 (Maxwell-type): Solve a final Maxwell-type equation to obtain the primal solution $u$.
Element Selection
To ensure practical applicability, the authors utilize lower-order Nédélec edge elements for the Maxwell steps and nonconforming $P_1-P_0$ elements for the Stokes step. This selection is strategic: these elements are standard in most high-performance computing libraries, allowing researchers to plug the quad-curl problem into existing, highly optimized solvers (such as multigrid methods) rather than developing specialized code for fourth-order elements.
Adaptive Finite Element Method (AFEM)
A significant contribution of this work is the development of a reliable Adaptive Finite Element Method (AFEM) designed to handle singularities. In domains with sharp corners or complex boundaries, the solution may exhibit abrupt changes that standard uniform meshes cannot resolve efficiently.
Separate Marking Strategy
Traditional adaptive methods often use a single global error indicator. However, this paper introduces a separate marking strategy. By tracking the error contributions of the Maxwell and Stokes components independently, the algorithm can precisely identify which regions of the mesh require refinement for which specific sub-problem. This granular approach ensures that the mesh is only refined where necessary, maintaining computational efficiency without sacrificing accuracy.
Theoretical and Numerical Results
The paper provides a comprehensive mathematical foundation for the method, establishing both a priori error estimates (predicting accuracy based on mesh size) and a posteriori estimates (calculating error bounds from the computed solution).
Key Insights
- Regularity Lifting: The analysis reveals a "regularity lifting" effect where, due to the decoupling structure, the curl of the primal variable can be approximated with one order higher accuracy than the variable itself. This is particularly beneficial for applications where the magnetic or velocity fields (derived from the curl) are the primary variables of interest.
- Optimal Convergence: Numerical experiments on a smooth unit cube domain $\Omega = (0, \pi)^3$ confirmed that the method achieves quadratic order convergence ($O(h^2)$) for the curl of the solution and first-order convergence for the energy norm.
- Singularity Handling: In tests involving an L-shaped domain—a classic benchmark for singularities—the adaptive method successfully restored optimal convergence rates relative to the number of degrees of freedom (DoF). In contrast, standard uniform mesh refinement failed to maintain accuracy in the presence of the re-entrant corner singularity.
Conclusion
This decoupled finite element method offers a practical and mathematically rigorous path for solving quad-curl problems. By reducing the problem to a sequence of Maxwell and Stokes equations, the method allows for the use of mature numerical libraries and fast solvers. Furthermore, the introduction of a separate marking strategy for adaptive refinement ensures that the method remains robust even in the presence of geometric singularities. Future work may extend this analysis to more complex, non-homogeneous boundary conditions and explore the computational trade-offs in large-scale multiphysics simulations.
Summary updated