Abstract
A refined a priori error analysis of the lowest-order (linear) nonconforming virtual element method (VEM) for approximating a model Poisson problem is developed in both 2D and 3D. A set of new geometric assumptions is proposed on the shape regularity of polytopal meshes. A new error equation for the lowest-order (linear) nonconforming VEM is derived for any choice of stabilization, and a new stabilization using a projection on an extended element patch is introduced for the error analysis on anisotropic elements.
Introduction
The Virtual Element Method (VEM) has emerged as a powerful tool for solving partial differential equations on general polytopal meshes. Unlike traditional Finite Element Methods (FEM), which typically require "shape-regular" elements—meaning triangles or tetrahedra should not be overly thin or flat—VEM is designed to handle more complex geometries. However, while VEM often performs surprisingly well on anisotropic elements, such as long, thin "slivers," theoretical guarantees for such cases have been lacking, particularly in 3D. This paper provides a rigorous mathematical foundation for the robustness of the linear nonconforming VEM on anisotropic meshes.
Methodology and Geometric Assumptions
The authors address the challenges of anisotropic elements through two primary innovations. First, they introduce a set of relaxed geometric assumptions (Assumptions A–E). Instead of the standard "chunkiness" requirements, they propose "Height" and "Hourglass" conditions. These conditions ensure that functions can be properly extended from a face into the element's interior, even when the element is highly stretched.
Second, the paper introduces a novel extended patch stabilization. In standard VEM, a stabilization term (a "penalty") is used to ensure the bilinear form is well-posed. On anisotropic elements, standard stabilization parameters can blow up. To counter this, the authors propose "borrowing" stability from a larger, better-shaped neighborhood, known as an extended element patch ($\omega_K$). This approach is conceptually similar to "ghost penalty" methods used in other unfitted mesh techniques.
Key Insights
A critical advantage of VEM in this context is its use of gradient projections rather than exact gradients. This "smoothing" effect reduces sensitivity to local element distortions. Furthermore, by focusing on the nonconforming space—where degrees of freedom (DoFs) are defined by face averages—the interpolation error is naturally zero in certain norms. This avoids the complex "edge-pairing" techniques often required for conforming methods on anisotropic meshes.
The analysis also bypasses traditional norm equivalence arguments, which fail on poorly shaped elements. Instead, the authors derive a new error equation and utilize a mesh-dependent norm $|\|\cdot\||$ naturally induced by the bilinear form.
Results and Convergence
The primary result of this work is the proof of optimal error estimates (Theorems 4.8 and 4.12). The authors demonstrate that the error in the mesh-dependent norm converges at an optimal rate of $O(h)$, where $h$ is the mesh size:
$$|\|u - u_h\|| \lesssim h|u|_{2,\Omega}$$
This confirms that accuracy is maintained even when elements are extremely thin or stretched. Crucially, this analysis is successfully extended to 3D polyhedra, providing a path for more robust 3D simulation software.
Conclusion and Future Work
This research provides the necessary theoretical backing for using nonconforming VEM on highly irregular meshes. While the extended patch stabilization introduces some additional computational cost—as the stencil involves neighbors of neighbors, making the stiffness matrix slightly denser—the gain in robustness is significant. Future work may extend these geometric conditions to higher-order VEM and a broader range of elliptic problems beyond the Poisson equation. Additionally, optimizing the implementation of the extended patch search for large-scale meshes remains an area for further development.
Summary updated