Pith. sign in

REVIEW 4 major objections 5 minor 42 references

The Neural Approximated Virtual Element Method for Elasticity Problems

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Neural-network-predicted basis functions replace stabilization in a polygonal method for elasticity, matching expected convergence orders.

desk verdict Solid extension of NA VEM to elasticity with honest but in-distribution numerical evidence; the generalization claim is plausible but unproven, and the error-constant comparison is muddied by different error metrics. read the letter →

arxiv 2507.05786 v1 pith:VQEGTTYS submitted 2025-07-08 math.NA cs.NA

Stefano Berrone , Moreno Pintore , Gioana Teora This is my paper · ORCID
classification math.NAcs.NA MSC 65N3065N1274S0568T07
keywords NAVEMvirtualelementmethodneuralnetworkapproximationelasticitypolygonalmeshesstabilization-freenonlinearbasisfunction
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper extends the Neural Approximated Virtual Element Method (NA VEM) to linear and nonlinear elasticity. It claims that virtual basis functions can be approximated element-wise by neural networks trained only on boundary data, so that no stabilization or projection operators are needed. On distorted quadrilateral and Voronoi meshes, the method reaches the expected convergence orders, $O(h^2)$ in the $L^2$ error and $O(h)$ in the $H^1$ seminorm, with lower error constants than the standard virtual element method. On a large-displacement nonlinear benchmark, it stays stable and oscillation-free even when the standard method requires an incremental loading strategy to maintain accuracy.

What carries the argument

The local approximation space is $H^{\mathrm{NN}}_{j,E} = \mathrm{span}\{\tilde p_\beta\}_{\beta=1}^{2\ell_{\mathrm{NN}}+1} \cup \{\Phi^{j-1}_{j,E}, \Phi^j_{j,E}, \Phi^{j+1}_{j,E}\}$, where the polynomials are harmonic polynomials up to degree $\ell_{\mathrm{NN}}$ and the three auxiliary functions are built from a universal harmonic-rational function $\Phi(z)$ with poles $z_\alpha = 1 + 2\exp(-4(\sqrt{N_1}-\sqrt{\alpha}))$. Two neural networks map an encoding of the pair (vertex, element) to coefficients: $c^\varphi$ for the basis function via loss (18) and $c^q$ for the gradient via loss (22). The load-bearing transfer principle, proven in [22], is that small boundary losses suffice for accurate interior approximation of both the function and its gradient. Training uses an RMS loss and a two-stage optimization with ADAM followed by a self-scaled BFGS optimizer, which the paper reports lowers the training losses compared with [22]. The gradient network is deliberately not the derivative of the basis-function network; this introduces a consistency error but suppresses gradient oscillations and makes the method more stable.

What would settle it

Evaluate the trained networks on polygons with extreme aspect ratios or non-convex shapes outside the RDQM and VM training distributions; the central claim is falsified if the $L^2$ or $H^1$ errors stop decreasing at the expected rates or the convergence curves level off.

Watch

Extended reading notes

Core claim

The authors construct two small neural networks per polygon-vertex class: one predicts the coefficients of a basis-function approximation in a space spanned by harmonic polynomials plus three element-dependent auxiliary harmonic functions, and a second predicts coefficients for the gradient of that basis function. Both networks are trained by minimizing boundary trace losses, namely the $H^{1/2}(\partial E)$ loss (18) for the function and the tangential $L^2(\partial E)$ loss (22) for the gradient, using the fact that virtual basis functions are known in closed form on element boundaries. The paper relies on the result from [22] that small boundary losses guarantee accurate interior approximation of both the basis functions and their gradients. The assembled discrete problem then follows the standard finite element structure, with no projector or stabilization term. Numerical tests on linear elasticity, on the nonlinear benchmark from [10], and on a Neo-Hookean hyperelastic problem show the expected convergence rates and lower error constants relative to VEM, and the large-deformation test converges without the incremental force strategy that VEM requires.

Load-bearing premise

The load-bearing premise is that minimizing the boundary trace losses (18) and (22) is enough to guarantee accurate interior approximation of both the basis functions and their gradients on any polygon of a given vertex count, and that networks trained on the RDQM and VM datasets generalize to unseen polygons of the same vertex count.

Editorial extensions

If this is right

  • The same trained networks can be reused for different materials, constitutive laws, body forces, and meshes whose polygons have the same vertex count, because training uses only geometric information.
  • Nonlinear problems become simpler: no stabilization parameter needs tuning, and the large-displacement benchmark converges with plain Newton-Raphson, without the incremental force strategy that VEM needs.
  • Point-wise evaluation of basis functions and gradients is available everywhere, so quantities such as the deformation determinant can be computed directly rather than through a projector.
  • The method retains the expected finite element convergence rates on polygonal meshes, with lower error constants than VEM in the reported tests.
  • The per-element neural network evaluation cost becomes negligible on fine meshes, and is further amortized in vector-valued elasticity because scalar basis functions are reused across components.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial extension: if the boundary-to-interior transfer holds beyond the tested mesh families, the method could be applied to non-convex or badly shaped elements, provided the training set adequately covers such geometries.
  • Editorial extension: the generalization of the trained networks to polygons outside the RDQM and VM training distributions is untested; a natural stress test is to evaluate on extreme-aspect-ratio or non-convex elements, where boundary trace losses may not control interior gradients.
  • Editorial extension: the independence of the basis-function and gradient networks creates an inconsistency that may influence stress accuracy; the paper reports stability benefits but does not analyze the resulting consistency error.
  • Editorial extension: because the approximation is element-wise and vertex-local, the method could combine naturally with adaptive refinement or hanging-node strategies, as the authors note in their conclusions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The manuscript introduces the Neural Approximated Virtual Element Method (NA VEM) for linear and nonlinear elasticity in two dimensions. The key idea is to replace the implicitly defined VEM basis functions with explicit neural-network approximations that are trained once, in an offline phase, to match the known boundary traces of the lowest-order VEM basis functions on reference polygons. In the online phase the networks return closed-form basis functions and their gradients, so the discrete equations are assembled as in a standard finite element method, without stabilization or projection operators. The paper presents the discrete formulation, details the network architecture and training losses, and reports numerical tests for linear elasticity, a nonlinear benchmark with analytical solutions, and a Neo-Hookean hyperelasticity problem. The central claims are that NA VEM attains the expected O(h^2) L2 and O(h) H1 convergence rates and achieves lower error constants than VEM, while avoiding stabilization and being simpler in nonlinear regimes.

Significance. If the claims are substantiated, NA VEM would be a practically attractive stabilization-free polygonal method for nonlinear elasticity, with the possible offline-online advantage that the same trained networks can be reused across different meshes, materials, and constitutive laws. The paper has several genuine strengths: the benchmarks use external analytical solutions, the nonlinear tests are substantially more demanding than a single linear Poisson problem, comparisons are made against independent VEM and FEM results, and no free parameter is fit to the PDE solution in the training procedure, so the experiments are not circular. The improved training losses and optimizer are also documented quantitatively in Table 1. However, the two load-bearing numerical claims—lower error constants than VEM and reusability on general meshes—are not fully supported by the evidence as presented: the error metrics used for NA VEM and VEM are different, and the test meshes are drawn from the same distributions used for training. These issues are fixable but require additional experiments and clarification.

major comments (4)
  1. [Section 4, Eqs. (29)-(30)] The claimed lower error constants of NA VEM relative to VEM are not established because the two methods are measured with different error functionals. Equation (29) compares the NA VEM solution directly pointwise (u - u_NN), while Eq. (30) compares VEM through the projected quantities u - Π1 u_VEM and ∇u - Π0 ∇u_VEM. Projection is a smoothing operation, so a lower value of the VEM error functional does not necessarily correspond to a more accurate VEM solution; conversely, the comparison between the two functionals is not apples-to-apples. To support the 'lower error constants' claim, the authors should compare both methods on the same quantity, for example by applying the same projection operators to the NA VEM solution, or by using an enhanced VEM solution that can be evaluated pointwise. The paper should also report computed convergence rates (EOC values) rather than relying only on visual inspection of log-log plots.
  2. [Section 4.1, paragraph beginning 'We also note that the elements...'] The numerical evidence for the central claim of a reusable, general polygonal method is in-distribution only. The training datasets RDQM and VM used in Section 3.3 are the same distributions from which the test meshes in Section 4.1 and 4.2 are drawn, and the manuscript explicitly states that 'the elements in this family of meshes are well represented by the training set. Thus, the network accuracy does not influence these results.' This is an admission that the reported convergence rates do not test generalization to genuinely new polygon shapes. Since the introduction and Section 3.3 present the offline-online reuse across 'various meshes' as a key advantage, the authors should include out-of-distribution experiments, for example polygonal meshes with strong aspect-ratio distortion, non-convex polygons, or vertex counts outside the trained classes, and report the resulting errors. Without such experiments, the claim that the same networks can be applied to arbitrary polygons of a given vertex count is unsupported.
  3. [Section 3.4, Eqs. (24)-(25)] The NA VEM discrete space is nonconforming: the manuscript acknowledges that the neural-network basis functions are not exactly continuous across elements, yet the degrees of freedom are not doubled. No consistency estimate or patch test is provided for this nonconforming approximation. Since the method is presented as a finite-element-like discretization, the observed convergence needs to be supported by either a standard patch test or a discussion of the consistency error introduced by the trace mismatch. As it stands, the convergence rates in Section 4 could in principle be masking a small but uncontrolled variational crime, especially on meshes that are less well represented by the training set.
  4. [Section 4, Figures 2, 4, 5] The convergence-rate claims O(h^2) and O(h) are supported only by visual inspection of log-log plots; no computed experimental orders of convergence are tabulated. Moreover, in Test 2 the L2 errors are said to behave similarly but are not reported. The authors should provide a table or listing of computed EOC values for each mesh family, error norm, and test case. This is essential for verifying that the convergence is actually of the expected order and for comparing the two methods fairly.
minor comments (5)
  1. [Section 3.1, Eq. (13)] In Eq. (13), the phrase 'with = 1, . . . , Nv_E' is missing the index variable; it should read 'with j = 1, . . . , Nv_E'.
  2. [Section 3.2, paragraph after Eq. (18)] The text says 'his gradient can be approximated'; this should be 'this gradient can be approximated'.
  3. [Section 4.2, caption of Figure 5] The caption of Figure 5 says 'Figure 4 reports the behaviour...' but should refer to Figure 5 itself.
  4. [Section 3.2 and 3.3] The input encoding strategy and the affine map from a general polygon to the reference square are described only by reference to [22]. Since the encoding is essential for reproducing the method, a brief self-contained description or an appendix entry would improve the paper.
  5. [Section 4.3, Figures 7 and 8] The mesh sizes for the mesh in Figure 7 and the FEM reference mesh in Figure 8 are not stated, which makes the visual comparison of oscillations difficult to interpret quantitatively.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: neural basis functions are trained on exact VEM boundary traces and benchmarked against external analytical elasticity solutions, with no PDE outputs used in training.

full rationale

The derivation chain is self-contained and non-circular. The neural networks are trained to minimize the boundary-trace losses (18) and (22), whose targets phi_j,E and q_j,E dot t are known in closed form on the boundary of E; no solution data from the target elasticity problems enter the training. The transfer from small boundary losses to accurate interior basis functions is justified by a cited theorem in the authors' prior work [22], but that theorem is a mathematical stability result for harmonic functions, not an assumption of the elasticity convergence being tested. All convergence benchmarks in Section 4 are compared against analytical solutions (Test 1: equation (32); Test 2: the two closed-form displacements; Test 3: a FEM reference on a fine triangular mesh), so the reported rates are not fitted outputs. The paper's admission in Section 4.1 that the distorted-square elements are 'well represented by the training set' is a limitation on the generality of the generalization evidence, not a circular reduction: the network is still evaluated on unseen elements, and no fitted parameter is re-labelled as a prediction. The self-citations to [21,22] are disclosed and concern the method's own construction; they do not smuggle in the elasticity results, since the cited works treat general elliptic problems and the current benchmarks are external. Hence the central claim has independent content and no step reduces by construction to its inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim depends on standard VEM theory, the boundary-trace computability of virtual basis functions, a transfer theorem from [22] connecting boundary losses to interior accuracy, and an empirical generalization assumption for the trained networks. No new physical entities are introduced. The hyperparameters are chosen by hand, not fitted to the target elasticity solutions.

free parameters (5)
  • Harmonic polynomial order l_NN = 20
    Chosen by hand (Section 3.3) to define the local approximation space H_NN; the text says fine-tuning is left to future work, so the reported accuracy depends on this choice.
  • Reference scaling R_NN = 3
    Hand-chosen scaling of the reference square S_NN in Section 3.1, affecting the conditioning of the harmonic basis.
  • Network architecture (5 layers, 50 neurons, tanh) = 5x50, tanh
    Fixed in Section 3.3 for all experiments; no sensitivity analysis is provided.
  • Regularization coefficient = 1e-8
    L2 weight-norm regularization added to both losses (Section 3.3), chosen without tuning study.
  • Training schedule = 5000 ADAM + 5000 self-scaled BFGS epochs
    Section 3.3; this schedule was selected empirically and contributes to the reported training losses.
assumptions (5)
  • standard math Virtual element space V_{h,1}(E) with componentwise harmonic basis functions and vertex degrees of freedom
    Section 2.1; standard VEM theory [7,9], used as target for neural approximation.
  • domain assumption Boundary trace continuity of virtual basis functions: phi_{j,E} is known piecewise linear on partial E
    Section 3.1-3.2; this makes the training losses (18) and (22) computable.
  • domain assumption The harmonic polynomials plus auxiliary functions Phi span a space rich enough to approximate VEM basis functions accurately (theorem from [22] connecting boundary losses to interior errors)
    Section 3.1; the paper cites [22] for this property, does not reprove it here.
  • domain assumption A suitable affine map exists from each element E to a reference element inside S_NN, and the input encoding of (v_j, E) retains enough geometric information
    Section 3.1; the paper refers to [22] and [32] for the map and encoding, assuming they preserve accuracy across arbitrary polygons.
  • ad hoc to paper Trained networks generalize from training polygons (random quadrilaterals and Voronoi cells) to test polygons with the same vertex count
    Section 3.3 and Section 4.1; the test meshes are chosen to be well represented by the training set, so this is an untested empirical assumption for new geometries.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Neural Approximated Virtual Element Method for Elasticity Problems." pith.science (2026). https://pith.science/paper/VQEGTTYS

@misc{pith2026250705786,
  author       = {Pith},
  title        = {Pith review of: The Neural Approximated Virtual Element Method for Elasticity Problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VQEGTTYS}},
  note         = {Machine review of arXiv:2507.05786}
}
read the original abstract

We present the Neural Approximated Virtual Element Method to numerically solve elasticity problems. This hybrid technique combines classical concepts from the Finite Element Method and the Virtual Element Method with recent advances in deep neural networks. Specifically, it is a polygonal method in which the virtual basis functions are element-wise approximated by a neural network, eliminating the need for stabilization or projection operators typical of the standard virtual element method. We present the discrete formulation of the problem and provide numerical tests on both linear and non-linear elasticity problems, demonstrating the advantages of having a simple discretization, particularly in handling non-linearities.

Figures

Figures reproduced from arXiv: 2507.05786 by the authors.

Figure 1
Figure 1. The second one (steps S.4–S.5) depends only on the problem and can be handled as in the standard Finite Element Method [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Test 1: Left: Last refinement of the Distorted-Square family of meshes. Right: Behaviour of errors [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 3
Figure 3. Test 2: Last refinements related to the two families of meshes. [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: reports the behaviour of the H1 -errors in (29) and (30), as the mesh parameter h decreases, for the test Case 1. Concerning this test case, we observe that the expected order of convergence is attained for both the methods with and without employing the incremental st…
Figure 5
Figure 5. Figure 5: Test 2, Case 2: Behaviour of the H1 -errors (29)–(30) as the mesh parameter h decreases. “(U)” means that the incremental force method is adopted. Left: Cartesian grids family. Right: Distorted-Voronoi grids family. figure, we draw a slice of the first component of the…
Figure 6
Figure 6. Figure 6: Test 2: Slice of the component u1 related to Case 2 and the last refinement of the Distorted-Voronoi family. Red: VEM. Blue: NAVEM. Black: exact displacement. We observe that the blue curve overlaps the black one. Finally, we observe that the robustness that characteri…
Figure 7
Figure 7. Figure 7: Test 3: Left: Mesh used in the simulation. Center: Displacement in the [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Test 3: Left: Mesh used to compute the FEM reference solution. Right: Displacement in the [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 21 canonical work pages

  1. [1]

    Canuto, D

    C. Canuto, D. Fassino, Higher-order adaptive virtual element methods with contraction properties, Mathematics in Engineering 5 (6) (2023) 1–33. doi:10.3934/mine.2023101. URL https://www.aimspress.com/article/doi/10.3934/mine.2023101

    work page doi:10.3934/mine.2023101 2023

  2. [2]

    A 3D-1D Virtual Element Method for Modeling Root Water Uptake

    S. Berrone, S. Ferraris, D. Grappein, G. Teora, F. Vicini, A 3d-1d virtual element method for modeling root water uptake (2024). arXiv:2412.12884. URL https://arxiv.org/abs/2412.12884

    work page Pith review arXiv 2024

  3. [3]

    Berrone, F

    S. Berrone, F. Vicini, Effective polygonal mesh generation and refinement for vem, Mathematics and Computers in Simulation 231 (2025) 239–258. doi:https://doi.org/10.1016/j.matcom.2024.12.007. URL https://www.sciencedirect.com/science/article/pii/S0378475424004786

    doi:10.1016/j.matcom.2024.12.007 2025

  4. [4]

    S. E. Leon, D. W. Spring, G. H. Paulino, Reduction in mesh bias for dynamic fracture using adaptive splitting of polygonal finite elements, International Journal for Numerical Methods in Engineering 100 (8) (2014) 555–576. arXiv:https://onlinelibrary.wiley.com/doi/pdf/10.1002/nme.4744, doi:https://doi.org/10. 1002/nme.4744. URL https://onlinelibrary.wiley...

    work page doi:10.1002/nme.4744 2014

  5. [5]

    S. O. R. Biabanaki, A. R. Khoei, P. Wriggers, Polygonal finite element methods for contact-impact problems on non-conformal meshes, Computer Methods in Applied Mechanics and Engineering 269 (2014) 198–221. URL https://api.semanticscholar.org/CorpusID:119486202

    work page 2014

  6. [6]

    A. L. Gain, G. H. Paulino, L. S. Duarte, I. F. Menezes, Topology optimization using polytopes, Computer Methods in Applied Mechanics and Engineering 293 (2015) 411–430. doi:https://doi.org/10.1016/j.cma.2015. 05.007. URL https://www.sciencedirect.com/science/article/pii/S0045782515001735

    doi:10.1016/j.cma.2015 2015

  7. [7]

    Beirão da Veiga, F

    L. Beirão da Veiga, F. Brezzi, A. Cangiani, G. Manzini, A. Russo, Basic principles of Virtual Element Methods, Mathematical Models and Methods in Applied Sciences 23 (01) (2013) 199–214

    work page 2013

  8. [8]

    Beirão da Veiga, F

    L. Beirão da Veiga, F. Brezzi, L. D. Marini, A. Russo, Virtual element method for general second order elliptic problems on polygonal meshes, Mathematical Models and Methods in Applied Sciences 26 (04) (2016) 729–750. arXiv:https://doi.org/10.1142/S0218202516500160, doi:10.1142/S0218202516500160. URL https://doi.org/10.1142/S0218202516500160

    doi:10.1142/s0218202516500160 2016

Show all 42 references
  1. [9]

    Beirão da Veiga, F

    L. Beirão da Veiga, F. Brezzi, L. D. Marini, Virtual elements for linear elasticity problems, SIAM Journal on Numerical Analysis 51 (2) (2013) 794–812. arXiv:https://doi.org/10.1137/120874746, doi:10.1137/ 120874746. URL https://doi.org/10.1137/120874746

    2013 doi

  2. [10]

    Beirão da Veiga, C

    L. Beirão da Veiga, C. Lovadina, D. Mora, A virtual element method for elastic and inelastic problems on polytope meshes, Computer Methods in Applied Mechanics and Engineering 295 (2015) 327–346. doi:https: //doi.org/10.1016/j.cma.2015.07.013. URL https://www.sciencedirect.com...

    2015 doi

  3. [11]

    S. C. Brenner, L. R. Scott, The Mathematical Theory of Finite Element Methods, V ol. 15 of Texts in Applied Mathematics, Springer, 2008. doi:10.1007/978-0-387-75934-0 . URL http://dx.doi.org/10.1007/978-0-387-75934-0

    2008 doi

  4. [12]

    Ciarlet, The Finite Element Method for Elliptic Problems, V ol

    Ph. Ciarlet, The Finite Element Method for Elliptic Problems, V ol. 40 of Classics in Applied Mathematics, Society for Industrial and Applied Mathematics (SIAM), Philadephia, 2002. arXiv:https://epubs.siam.org/doi/ pdf/10.1137/1.9780898719208, doi:10.1137/1.9780898719208. URL ...

    2002 doi

  5. [13]

    Berrone, S

    S. Berrone, S. Scialò, G. Teora, The mixed virtual element discretization for highly-anisotropic problems: the role of the boundary degrees of freedom, Mathematics in Engineering 5 (6) (2023) 1–32. doi:10.3934/mine. 2023099. URL https://www.aimspress.com/article/doi/10.3934/mi...

    2023 doi

  6. [14]

    H. Chi, L. B. da Veiga, G. Paulino, Some basic formulations of the virtual element method (vem) for finite deformations, Computer Methods in Applied Mechanics and Engineering 318 (2017) 148–192. doi:https: //doi.org/10.1016/j.cma.2016.12.020. URL https://www.sciencedirect.com/...

    2017 doi

  7. [15]

    A. Chen, N. Sukumar, Stabilization-free serendipity virtual element method for plane elasticity, Computer Methods in Applied Mechanics and Engineering 404 (2023) 115784. doi:https://doi.org/10.1016/j.cma.2022. 115784. URL https://www.sciencedirect.com/science/article/pii/S0045...

    2023 doi

  8. [16]

    A. Chen, N. Sukumar, Stabilization-free virtual element method for plane elasticity, Computers & Mathematics with Applications 138 (2023) 88–105. doi:https://doi.org/10.1016/j.camwa.2023.03.002. URL https://www.sciencedirect.com/science/article/pii/S0898122123000913

    2023 doi

  9. [17]

    Credali, S

    F. Credali, S. Bertoluzza, D. Prada, Reduced basis stabilization and post-processing for the virtual element method, Computer Methods in Applied Mechanics and Engineering 420 (2024) 116693. doi:https://doi.org/10. 1016/j.cma.2023.116693. URL https://www.sciencedirect.com/scien...

    2024

  10. [18]

    Trezzi, U

    M. Trezzi, U. Zerbinati, When rational functions meet virtual elements: the lightning virtual element method, Calcolo 61 (3) (2024) 35. doi:10.1007/s10092-024-00585-1 . URL https://doi.org/10.1007/s10092-024-00585-1

    2024 doi

  11. [19]

    J. S. Hesthaven, G. Rozza, B. Stamm, et al., Certified reduced basis methods for parametrized partial differential equations, V ol. 590, Springer, 2016

    2016

  12. [20]

    Gopal, L

    A. Gopal, L. N. Trefethen, Solving laplace problems with corner singularities via rational functions, SIAM Journal on Numerical Analysis 57 (5) (2019) 2074–2094. arXiv:https://doi.org/10.1137/19M125947X, doi:10.1137/19M125947X. URL https://doi.org/10.1137/19M125947X

    2019 doi

  13. [21]

    Berrone, D

    S. Berrone, D. Oberto, M. Pintore, G. Teora, The lowest-order neural approximated virtual element method, in: Numerical Mathematics and Advanced Applications ENUMATH 2023, V ol. 1, Springer, 2025, pp. 129–138. URL https://doi.org/10.1007/978-3-031-86173-4_13

    2023 doi

  14. [22]

    Berrone, M

    S. Berrone, M. Pintore, G. Teora, The lowest-order neural approximated virtual element method on polygonal elements, Computers & Structures 314 (2025) 107753. URL https://doi.org/10.1016/j.compstruc.2025.107753

    2025

  15. [23]

    LeCun, Y

    Y . LeCun, Y . Bengio, G. Hinton, Deep learning, Nature 521 (7553) (2015) 436–444

    2015

  16. [24]

    Cuomo, V

    S. Cuomo, V . S. Di Cola, F. Giampaolo, G. Rozza, M. Raissi, F. Piccialli, Scientific machine learning through physics–informed neural networks: Where we are and what’s next, Journal of Scientific Computing 92 (3) (2022)

    2022

  17. [25]

    Abadi, et al., TensorFlow: Large-scale machine learning on heterogeneous systems, software available from tensorflow.org (2015)

    M. Abadi, et al., TensorFlow: Large-scale machine learning on heterogeneous systems, software available from tensorflow.org (2015). URL http://tensorflow.org/

    2015

  18. [26]

    Paszke, S

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Köpf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, S. Chintala, PyTorch: An imperative style, high-perform...

    2019

  19. [27]

    Bradbury, R

    J. Bradbury, R. Frostig, P. Hawkins, M. J. Johnson, C. Leary, D. Maclaurin, G. Necula, A. Paszke, J. Van- derPlas, S. Wanderman-Milne, Q. Zhang, JAX: composable transformations of Python+NumPy programs, http://github.com/google/jax (2018)

    2018

  20. [28]

    Beirão da Veiga, F

    L. Beirão da Veiga, F. Brezzi, L. D. Marini, A. Russo, The virtual element method, Acta Numerica 32 (2023) 123–202. doi:10.1017/S0962492922000095

    2023 doi

  21. [29]

    Kim, Introduction to Nonlinear Finite Element Analysis, Springer US, 2014

    N. Kim, Introduction to Nonlinear Finite Element Analysis, Springer US, 2014. URL https://books.google.it/books?id=PDeSBQAAQBAJ

    2014

  22. [30]

    Wriggers, Nonlinear Finite Element Methods, Springer Berlin Heidelberg, 2008

    P. Wriggers, Nonlinear Finite Element Methods, Springer Berlin Heidelberg, 2008. URL https://books.google.it/books?id=IaV0wgG2jacC 17 The Neural Approximated Virtual Element Method for Elasticity Problems A PREPRINT

    2008

  23. [31]

    Wriggers, F

    P. Wriggers, F. Aldakheel, B. Hudobivnik, Virtual Element Methods in Engineering Sciences, Springer Interna- tional Publishing, 2023. URL https://books.google.it/books?id=KdnfEAAAQBAJ

    2023

  24. [32]

    Berrone, G

    S. Berrone, G. Teora, F. Vicini, Improving high-order vem stability on badly-shaped elements, Mathematics and Computers in Simulation 216 (2024) 367–385. doi:https://doi.org/10.1016/j.matcom.2023.10.003. URL https://www.sciencedirect.com/science/article/pii/S0378475423004287

    2024 doi

  25. [33]

    J. B. Perot, C. Chartrand, A mimetic method for polygons, Journal of Computational Physics 424 (2021) 109853. doi:https://doi.org/10.1016/j.jcp.2020.109853. URL https://www.sciencedirect.com/science/article/pii/S0021999120306276

    2021

  26. [34]

    Goodfellow, Y

    I. Goodfellow, Y . Bengio, A. Courville, Y . Bengio, Deep learning, V ol. 1, MIT press Cambridge, 2016

    2016

  27. [35]

    Kingma, J

    D. Kingma, J. Ba, Adam: a method for stochastic optimization, International Conference on Learning Representa- tions (2014). doi:https://doi.org/10.48550/arXiv.1412.6980

    2014 doi arXiv

  28. [36]

    Wright, J

    S. Wright, J. Nocedal, et al., Numerical Optimization, V ol. 35, Springer, 1999

    1999

  29. [37]

    Kiyani, K

    E. Kiyani, K. Shukla, J. F. Urbán, J. Darbon, G. E. Karniadakis, Which optimizer works best for physics-informed neural networks and kolmogorov-arnold networks?, arXiv preprint arXiv:2501.16371 (2025)

    2025

  30. [38]

    Yu, mvem: A matlab software package for the virtual element methods (2022)

    Y . Yu, mvem: A matlab software package for the virtual element methods (2022). arXiv:2204.01339. URL https://arxiv.org/abs/2204.01339

    2022 arXiv

  31. [39]

    Ahmad, A

    B. Ahmad, A. Alsaedi, F. Brezzi, L. Marini, A. Russo, Equivalent Projectors for Virtual Element Methods, Comput Math Appl 66 (3) (2013) 376–391

    2013

  32. [40]

    Abbas, A

    M. Abbas, A. Ern, N. Pignet, Hybrid high-order methods for finite deformations of hyperelastic materials, Computational Mechanics 62 (4) (2018) 909–928. doi:10.1007/s00466-018-1538-0 . URL https://doi.org/10.1007/s00466-018-1538-0

    2018 doi

  33. [41]

    Lamperti, M

    A. Lamperti, M. Cremonesi, U. Perego, A. Russo, C. Lovadina, A hu–washizu variational approach to self- stabilized virtual elements: 2d linear elastostatics, Computational Mechanics 71 (2023) 935–955. doi:10.1007/ s00466-023-02282-2 . URL https://doi.org/10.1007/s00466-023-02282-2 18

    2023 doi

  34. [88]

    URL https://doi.org/10.1007/s10915-022-01939-z

    doi:10.1007/s10915-022-01939-z . URL https://doi.org/10.1007/s10915-022-01939-z

    doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.