Virtual Error-bin Methodology

• Virtual Error-bin Methodology is a framework that decomposes discretization errors into distinct bins using mesh regularity, inverse estimates, and norm equivalences.
• The approach employs canonical VEM interpolants and detailed interpolation error analysis to support both a priori and a posteriori error control.
• It underpins adaptive mesh refinement algorithms that achieve optimal convergence rates in both isotropic and anisotropic, conforming and nonconforming mesh settings.

The Virtual Error-bin Methodology is a rigorous framework for the derivation, analysis, and implementation of error estimates and adaptive strategies in the Virtual Element Method (VEM) for elliptic partial differential equations on general polygonal and polyhedral meshes. The methodology enables the decomposition of the total discretization error into conceptually distinct "bins," each admitting quantitative analysis rooted in mesh and solution regularity assumptions, inverse estimates, norm equivalences, and interpolation error bounds. This approach supports both a priori and a posteriori error control for conforming and nonconforming VEM schemes, in isotropic and anisotropic mesh environments, and underpins effective adaptive mesh refinement algorithms (Chen et al., 2017, Cangiani et al., 2016, Cao et al., 2018, Antonietti et al., 2020).

1. Mesh Assumptions and Auxiliary Structures

In the Virtual Error-bin Methodology, the mesh $\mathcal{T}_h$ for the computational domain $\Omega \subset \mathbb{R}^d$ (typically $d=2,3$) consists of general polygons (or polyhedra) satisfying explicit regularity hypotheses. Two critical conditions are frequently imposed:

• Simple polygonality: Each element $K \in \mathcal{T}_h$ is an open, simply-connected, non-self-intersecting polygon (or polyhedron).
• Virtual quasi-uniform triangulation: Each $K$ admits a subdivision (triangulation) $\mathcal{T}_K$ that is shape-regular and quasi-uniform, such that each polygonal edge of $K$ is the edge of some triangle in $\mathcal{T}_K$, the mesh-size $h_T \simeq h_K$ for all $T \in \mathcal{T}_K$, and triangulation constants (shape-regularity, quasi-uniformity, bounded number of triangles) are uniform in $K$.

For nonconforming or anisotropic mesh settings, further geometric conditions generalize these hypotheses to include bounds on the number of faces, height and hourglass conditions, extended-patch properties, and finite overlaps of convex hulls to permit elements with high aspect ratios and nontrivial geometry (Chen et al., 2017, Cao et al., 2018).

2. Inverse Inequalities and Norm Equivalence

Key to the methodology are robust local inverse estimates and norm-equivalence results between degrees of freedom (DoFs) and function norms. For the local conforming VEM space

$$V_{K,1}(K) = \{ v \in H^1(K) : v|_{\partial K} \text{ is piecewise } P^k,\, \Delta v \in P^l(K) \}$$

with $l \geq k-2$, the following properties hold (constants are uniform in $K$):

• Inverse estimate: For all $v \in V_{K,1}(K)$,

$\|\nabla v\|_{0,K} \leq C h_K^{-1} \|v\|_{0,K}.$

• Norm equivalence between DoFs and $L^2$-norm: Let $X(v)$ denote the scaled DoF vector,

$$c_1 h_K \|X(v)\|_{\ell^2} \leq \|v\|_{0,K} \leq c_2 h_K \|X(v)\|_{\ell^2}.$$

These form the algebraic and analytic backbone for both the stability of commonly used stabilization bilinear forms and the underlying discretization (Chen et al., 2017). Analogous norm equivalence properties are essential in the reliability proofs for residual-type a posteriori estimators (Cangiani et al., 2016).

3. Interpolation Error Decomposition

Central to error binning is the construction and analysis of canonical VEM interpolants $I_K$, defined to match prescribed degrees of freedom (vertex values, edge/face moments, internal moments) of a smooth function $v \in H^{k+1}(K)$. The interpolation error satisfies

$$\|v - I_K v\|_{0,K} + h_K \|\nabla(v - I_Kv)\|_{0,K} \leq C h_K^{k+1} |v|_{k+1,K}.$$

The proof leverages harmonic and moment-preserving interpolants, Poincaré inequalities, and auxiliary finite element constructions on $\mathcal{T}_K$ (Chen et al., 2017). These interpolation estimates, together with quasi-interpolants (Clement-type) in adaptive settings, provide the quantitative tools needed for both a priori and a posteriori analysis (Cangiani et al., 2016, Antonietti et al., 2020).

4. Error-bin Framework and Error Decomposition

The methodology formalizes error control by decomposing the total discretization error $u-u_h$ (where $u$ is the exact and $u_h$ the VEM approximation) into discrete bins: $$u-u_h = (u - I_h u) \quad\text{(approximation bin)} + \quad (I_h u - u_h)\quad\text{(discrete bin)}$$ Each term admits quantitative estimation:

• Approximation bin: Bounded by the interpolation error estimates.
• Consistency bin: Difference $a(I_h u, v_h) - a_h(I_h u, v_h)$, non-zero only on the stabilization part, estimated via norm equivalences and stability.
• Discrete bin: Estimated using Céa-type arguments, ultimately controlled by stability and consistency constants.
• Data oscillation bin: Included if the right-hand side $f$ lacks exact polynomial representation, resulting in an $O(h^k)$ term.

The energy norm error obeys

$$\|u - u_h\|_{1,\Omega} \leq C h^{s} |u|_{s+1,\Omega}, \qquad s = \min(k, \,\text{regularity} - 1),$$

with mesh-independent constants depending only on auxiliary triangulation regularity, reproducing optimal $h^k$ convergence for sufficiently regular $u$ (Chen et al., 2017).

For a posteriori error estimation, residual-equation-based estimators of the form

$$\eta^2 = \sum_E \big( \eta_E^2 + \operatorname{osc}_E^2 + S_E + I_E \big)$$

are fully computable from DoFs and polynomial projections, yielding guaranteed upper (reliability) and lower (efficiency) bounds for the VEM error (Cangiani et al., 2016).

5. Generalizations: Anisotropy, Nonconformity, and Advanced Error Control

The methodology extends to nonconforming VEM, especially in the presence of mesh anisotropy. For linear nonconforming VEM, mesh patches $\omega_K$ enclosing possibly highly anisotropic elements $K$ are introduced to compensate for degenerate geometric properties. The corresponding error equations involve patch-elliptic projectors and stabilization terms weighted by global patch diameter, providing robustness and optimal $h$-convergence in the energy norm regardless of local element aspect ratios (Cao et al., 2018). For higher-order VEM and mixed/H(div)-conforming approximations, the bin decomposition framework and patch-based stabilizations remain applicable.

A posteriori error analysis for anisotropic meshes is enabled by patchwise quasi-interpolants and anisotropy-weighted error indicators. In this context, indicators incorporate the covariance structure of the element (or patch), gradient recovery tensors, and mesh-size ratios, affording reliability bounds with constants independent of element anisotropy. Adaptive refinement employs cut-based element splitting directed by the anisotropy of both the cell and the estimated error, delivering significant savings in degrees of freedom and preserving optimal convergence rates observed in numerical experiments (Antonietti et al., 2020).

6. Adaptive Algorithms and Practical Implementation

Adaptive mesh refinement under the Virtual Error-bin Methodology follows the sequence:

1. Solve the discrete VEM problem on the current mesh.
2. Estimate local error indicators ($\eta_E$, $\operatorname{osc}_E$, stabilization and virtual inconsistency).
3. Mark elements using a selection criterion (e.g., Dörfler marking to capture a prescribed fraction of the total error).
4. Refine marked elements (polygonal bisection, barycentric subdivision for isotropic or anisotropic strategies).
5. Project/Interpolate the current solution onto the new mesh via the canonical VEM interpolant.
6. Iterate until convergence criteria are reached, e.g., estimator below tolerance or maximal degrees of freedom.

The estimator is readily assembled from local DoFs, L² and energy projections, and local geometric data, with explicit handling of hanging nodes and coplanar subdivisions (Cangiani et al., 2016, Antonietti et al., 2020). Numerical results demonstrate the robustness and optimality of adaptive error control for smooth, singular, internal-layer, and coefficient-jump problems.

7. Extensions, Open Questions, and Impact

The Virtual Error-bin Methodology underpins much of the rigorous error analysis and adaptive control fundamental to modern VEM literature. Notable extensions include recovery-based and goal-oriented error estimators, preconditioners adapted to rapidly evolving polygonal meshes, extension to three spatial dimensions, and anisotropic and mixed conforming/nonconforming VEM. Outstanding questions relate to the explicit characterization of trace/interpolation constants in three dimensions and efficient coarsening strategies in unsteady problems (Cangiani et al., 2016).

The methodology provides a unified, flexible framework accommodating arbitrary polytopal meshes, supports provable error control, and scales across conforming, nonconforming, isotropic, and anisotropic settings. Its principles continue to inform new developments in VEM discretizations and adaptive methodologies (Chen et al., 2017, Cao et al., 2018, Cangiani et al., 2016, Antonietti et al., 2020).