Stabilization-Free ๐ โก ( ๐๐ฎ๐ซ๐ฅ ) \mathbf{H}\left(\operatorname{\mathbf{curl}}\right) and ๐ โก ( div ) \mathbf{H}\left(\operatorname{div}\right) -Conforming Virtual Element Method
Yuxuan Liao, Xue Feng, Yidong Huang ยท 2025
Read Stabilization-Free ๐ โก ( ๐๐ฎ๐ซ๐ฅ ) \mathbf{H}\left(\operatorname{\mathbf{curl}}\right) and ๐ โก ( div ) \mathbf{H}\left(\operatorname{div}\right) -Conforming Virtual Element Method by Yuxuan Liao, Xue Feng, Yidong Huang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-15168.txt.
Introduction
Yuxuan Liao, Xue Feng, and Yidong Huang remove a crutch from a flexible numerical method for field problems, in a 2025 paper. The virtual element method solves equations on oddly shaped meshes, handy for electromagnetics, but it leans on stabilization terms that warp its accuracy and cost. The authors build a version that needs no stabilization for the two key space families used in field problems, those respecting curl and those respecting divergence.
Their trick is a leaner projection step, using so-called serendipity spaces trimmed to the fewest possible unknowns. The construction covers the full chain of related spaces in three dimensions while keeping the properties engineers rely on, such as continuity across cell borders and clean hand-offs between spaces. They prove the method keeps its accuracy and test it on Maxwell eigenvalue problems, where the computed vibration modes of electromagnetic cavities matched expectations.
Catalog record
| CREATOR | Yuxuan Liao, Xue Feng, Yidong Huang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 112 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-15168
Browse the full library ยท About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Stabilization-Free ๐ โก ( ๐๐ฎ๐ซ๐ฅ ) \mathbf{H}\left(\operatorname{\mathbf{curl}}\right) and ๐ โก ( div ) \mathbf{H}\left(\operatorname{div}\right) -Conforming Virtual Element Method Yuxuan Liao, Xue Feng, Yidong Huang Abstract Standard Virtual Element Method (VEM) requires stabilization terms that significantly affect the numerical computation performance. In this work, we propose a stabilization-free VEM for general order ๐ โก ( ๐๐ฎ๐ซ๐ฅ ) \mathbf{H}(\operatorname{\mathbf{curl}}) and ๐ โก ( div ) \mathbf{H}(\operatorname{div}) -conforming spaces by constructing novel serendipity projectors and corresponding serendipity spaces with minimum number of DoFs. Our approach handles the full De Rham complex chain in โ 3 \mathbb{R}^{3} while preserving essential properties including boundary continuity and commutativity. Since the number of DoFs are minimized, computational overhead is greatly reduced. The optimal approximation properties are rigorously proven and validated through Maxwell eigenvalue problems with numerical experiments. keywords Virtual Element Method, serendipity space, stabilization, H(curl) and H(div)-conforming, error analysis โ โ runningheads: Stabilization-Free \mathbf{H}\left(\operatorname{\mathbf{curl}}\right) and \mathbf{H}\left(\operatorname{div}\right) -Conforming VEM / Yuxuan Liao et al. MSC 65N12, 65N22, 65N30 1 Introduction The Virtual Element Method (VEM) [ 8 ] , introduced as a natural extension of the finite element method (FEM), has gained considerable attention for its exceptional capability to handle general star-shaped polytopes. Such flexibility offers significant advantages over conventional finite element methods, particularly for complex geometries and adaptive mesh refinement. For comprehensive reviews of VEMโs advantages and applications, we refer to [ 2 ] . However, its geometric flexibility comes at a cost. Unlike traditional FEM, the virtual element function space \mathbf{V}(K) is no longer purely polynomial, necessitating projections onto polynomial spaces for numerical computation. Specifically, to solve weak formulations of the form a(u,v)=b(v) , the local discrete bilinear form must be reformulated as a_{K}(\bm{\mathit{\Pi}}_{k}u,\bm{\mathit{\Pi}}_{k}v) , where \bm{\mathit{\Pi}}_{k}:\mathbf{V}(K)\to\mathbb{P}_{k}(K) denotes the polynomial projection. The fundamental challenge in standard VEM arises from the instability of this projection: since typically \dim\mathbb{P}_{k}^{d}(K)<\dim\mathbf{V}(K) , the inequality \|\bm{\mathit{\Pi}}_{k}v\|_{K}\gtrsim\|v\|_{K} fails to hold. Consequently, stabilization terms of the form a_{K}(v,v)\lesssim S_{K}(v,v)\lesssim a_{K}(v,v)\ \forall v\in\ker\bm{\mathit{\Pi}}_{k} are essential to maintain the coercivity. These stabilization terms present several critical drawbacks: they significantly impact numerical stability and condition numbers [ 7 ] , require careful problem-specific design (especially for nonlinear problems [ 22 ] ), and exhibit isotropic behavior that proves problematic for anisotropic coefficients or meshes [ 9 ] . Existing Stabilization-Free Approaches. Recent efforts to eliminate stabilization terms have pursued two distinct strategies. The first is to modify the virtual element space, so that a stable \mathrm{L}^{2} -polynomial projection can be computed. The first work for \mathrm{H}^{1} -conforming virtual element spaces is [ 10 ] . It utilizes the lowest order \mathrm{H}^{1} -seminorm projector \bm{\mathit{\Pi}}^{\nabla}_{1} , and construct the virtual element space with functions equivalent under this projector: \begin{split}\mathbf{V}_{1,l}\left(F\right):=\left\{v\in\mathrm{H}^{1}\mid\left(\bm{\mathit{\Pi}}^{\nabla}_{1}v-v,p\right)_{F}=0\ \forall p\in\mathbb{P}_{l+1}(F),\Delta v\in\mathbb{P}_{l+1}(F),\right.\\ \left.v_{\mid E}\in\mathbb{P}_{1}(E)\ \forall E\in\partial F,v\in C^{0}(\partial F)\right\}.\end{split} (1.1) Then the \mathrm{L}^{2} -projection of \nabla v to \mathbb{P}^{2}_{l}\left(F\right) is computable, so \left(\nabla u,\nabla v\right)_{F}\ \forall u,v\in\mathbf{V}_{1,l}\left(F\right) is computed as \left(\bm{\mathit{\Pi}}_{l}\nabla u,\bm{\mathit{\Pi}}_{l}\nabla v\right)_{F} . This discrete bilinear form requires no stabilization since \left\|\nabla v\right\|_{F}\approx\left\|\bm{\mathit{\Pi}}_{l}\nabla v\right\|_{F} . This technique, while can be extended to general order spaces, inevitably requires the internal DoFs for k>1 . Since \dim\mathbb{P}_{l}^{2}(F)\geq\dim\mathbf{V}(F) is necessary for \bm{\mathit{\Pi}}_{l} to be stable, these internal DoFs, though can be eventually eliminated by static condensation, still require l much larger than it truly needs to be, which would greatly increase the computational cost. While in [ 12 ] this technique is further extended to general order spaces, by replacing \bm{\mathit{\Pi}}^{\nabla}_{1} with the serendipity projector \bm{\mathit{\Pi}}^{\mathrm{S}}_{k} . Later in [ 23 , 24 ] , the same technique is applied to nonlinear two-dimensional (2D) and three-dimensional (3D) finite elasticity problems, which greatly exhibit the advantages of stabilization-free VEM. Since the serendipity projector can be computed using only the boundary DoFs, it overcomes the internal DoFs issue. However, the trivially chosen serendipity projector, do not preserve many crucial properties required by other weak formulations. For example, the lack of commutativity with differential operators, makes it unapplicable to mixed formulations. The other strategy is to modify the projection space. In [ 13 ] , an \mathbf{H}\left(\operatorname{div}\right) -conforming macro finite element \mathbf{V}^{\operatorname{div}}_{k-1}\left(K\right) is constructed, and the \mathrm{L}^{2} -projection from the conventional virtual element space \mathbf{V}_{k}\left(K\right) to \mathbf{V}^{\operatorname{div}}_{k-1}\left(K\right) is stable and computable. But this method would dilute the advantages of VEM to some extent, since it requires simplical tessellation of polyhedra and the construction of corresponding broken polynomial spaces, while VEM with the original form, is designed to handle more general polytopes. Our Contribution. In this work, we introduce a fundamentally different stabilization-free VEM that addresses the limitations of existing approaches while preserving all essential mathematical and computational properties. Our method is built upon four key innovations: Complete De Rham Complex Treatment: Unlike previous works that handle \mathrm{H}^{1} space, we construct commutative serendipity projectors for the entire De Rham complex chain in \mathbb{R}^{3} . This unified treatment not only ensures seamless integration across different differential operators, but also can be systematically extended to higher-dimensional De Rham complexes by recursively applying the same technique to boundaries of each dimension. Preservation of Essential Mathematical Properties: Our serendipity spaces rigorously maintain boundary continuity, operator conformity, exact commutativity with differential operators, and B-compatibility [ 11 ] . Computational Efficiency: Our approach achieves stabilization-free computation using only serendipity projections computed with minimum number of DoFs. This reduces the stable projector order significantly, for general order virtual element spaces. Notably, even for stabilized treatment, the reduction in DoFs is also substantial as it is applicable to eigenvalue problems where internal DoFs cannot be eliminated by static condensation. The resulting stabilization-free VEM preserves all advantages of the original method while eliminating stabilization-related complications, making it particularly valuable for mixed formulations, nonlinear problems, and applications requiring robust numerical behavior across diverse mesh configurations. 2 Preliminaries We utilize standard Sobolev spaces whose detailed definitions can be found in [ 19 ] . For a Lipschitz domain D\subseteq\mathbb{R}^{d} with d=1,2,3 : โข \mathrm{L}^{2}(D) : space of square-integrable functions โข \mathrm{H}^{s}(D) : space of functions with square-integrable derivatives up to order s\geq 0 We use bold upright typeface for vector-valued spaces: โข \mathbf{H}^{s}(\operatorname{\mathbf{curl}},D) : functions whose curl belongs to \mathbf{H}^{s}(D) โข \mathbf{H}^{s}(\operatorname{div},D) : functions whose divergence belongs to \mathrm{H}^{s}(D) For any function space X , we denote: โข \|\cdot\|_{X} and |\cdot|_{X} : norms and seminorms equipped to X โข \|\cdot\|_{s,D} and |\cdot|_{s,D} : norms and seminorms for \mathrm{H}^{s}(D) or \mathbf{H}^{s}(D) โข (\cdot,\cdot)_{D} : \mathrm{L}^{2} -inner product in domain D When s=0 , the subscript s is omitted. When the domain is clear from context, the subscript D is omitted. The subscript 0 denotes the space of functions with zero trace, and the subscript \operatorname{div}0 denotes the space of functions with zero divergence. For linear spaces X\subseteq Y , Y/X denotes the subspace of Y naturally isomorphic to the quotient space in conventional notation. For polynomial spaces, we adopt the following notation: โข \mathbb{P}^{q}_{k,d}(D) : polynomials from \mathbb{R}^{d} to \mathbb{R}^{q} of degree at most k โข \mathbb{P}^{q}_{k_{1}+1:k_{2},d}(D) : \mathbb{P}^{q}_{k_{2},d}(D)/\mathbb{P}^{q}_{k_{1},d}(D) Following standard convention: q is omitted when q=1 , d is omitted when clear from context, and k is omitted for arbitrarily large polynomial degree. We reserve specific notation for these entities: โข \mathbf{V} : finite element spaces โข \bm{\mathit{\Pi}} : general projectors โข \bm{\mathit{\Pi}}^{q}_{k,d} : \mathrm{L}^{2} -projector onto \mathbb{P}^{q}_{k,d} The subscripts q,k,d follow the same