Pith. sign in

REVIEW 5 major objections 5 minor 45 references

Enforced Interface Constraints for Domain Decomposition Method of Discrete Physics-Informed Neural Networks

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read Interface continuity in decomposed PINNs can be enforced by interpolation instead of by loss penalties, keeping energy-based training accurate and parallel-friendly.

desk verdict The EIC mechanism looks inert in the equations as written: overridden interface node values never enter the Gauss-point strain energy, so the central continuity claim is unsupported. read the letter →

arxiv 2505.10925 v1 pith:QIUO2ZD2 submitted 2025-05-16 cs.CE

classification cs.CE
keywords discretephysics-informedneuralnetworksdomaindecompositionenforcedinterfaceconstraintsnon-conformingmeshesenergy-basedlearninglinearelasticityparalleltrainingweakspatialconstraint
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 argues that domain decomposition does not need an interface penalty term in a physics-informed neural network if the interface condition is imposed geometrically. In the proposed discrete PINN, each subdomain is meshed independently and the system energy is evaluated by Gaussian quadrature over the elements; at non-matching interfaces, the displacement of a slave node is replaced by the shape-function interpolation from the nearest master element. The authors claim this enforced interface constraint preserves displacement continuity exactly, removes the loss-balancing problems of penalty or soft constraints, and decouples the weak spatial constraint that can smear genuine gaps. On two- and three-dimensional linear-elastic benchmarks, they report maximum relative errors of 0.78% to 3.42% against non-conforming finite elements, with multi-GPU training speeding up large cases by roughly a third.

What carries the argument

The load-bearing object is the node-element coupling defined by Eq. (11): $u(\mathbf{x}_o)=\sum_i N_i^e(\xi,\eta)\mathbf{u}_i^e$, where the displacement of an interface node is expressed through the shape functions of a neighboring subdomain's element evaluated at the node's inverse-mapped reference coordinates. This mapping is built once in preprocessing by solving $R(\xi)=\sum_i N_i(\xi)\mathbf{x}_i - \mathbf{x}_o = 0$ via Newton-Raphson; at each training epoch the mapped value replaces the network output at that node before the Gaussian-quadrature energy of Eq. (6) is assembled. It converts interface continuity from a soft objective into an exact kinematic relation, and it is what allows subdomains to be meshed independently and trained in parallel.

What would settle it

Run a two-subdomain linear-elastic problem with a non-conforming interface under a steep stress gradient, compare to a refined finite element solution, and measure the normal traction jump across the interface: if the jump stays large as meshes and epochs increase while displacement errors are small, then the pointwise displacement constraint is too weak to deliver the true interfacial equilibrium.

Watch

Extended reading notes

Core claim

The central claim is that continuity across non-conforming subdomain interfaces in an energy-based discrete PINN can be enforced as a hard constraint rather than as a soft loss term. The mechanism replaces the neural network's prediction for an interface node in one subdomain with displacements interpolated from the nearest element of the neighboring subdomain, after mapping the node into the element's reference coordinates. Because the replacement happens before energy evaluation, the total potential energy remains a single functional, with no interface residual, no auxiliary collocation points, and no weighting coefficients to tune. The paper then shows, in 2D and 3D linear elasticity, that this hard constraint preserves displacement continuity, avoids the weak spatial constraint artifact around narrow gaps, retains accuracy (maximum relative errors of 0.78% to 3.42% against non-conforming finite element solutions), and allows multi-GPU training with error curves identical to single-GPU training.

Load-bearing premise

The load-bearing premise is that pointwise interpolation of a slave interface node from the nearest master element gives enough kinematic coupling that the energy-minimizing solution is the correct physical one, even though no traction equilibrium or weak interface condition is imposed.

Editorial extensions

If this is right

  • Subdomains can be meshed independently, so preprocessing for complex geometries is simpler and non-matching interfaces need no special coupling elements.
  • Because the loss function is a single energy term, no interface weights or auxiliary collocation points have to be tuned, reducing loss-balance pathologies.
  • Separate subnetworks per subdomain decouple the weak spatial constraint, so narrow physical gaps produce true displacement discontinuities instead of artificial smearing.
  • Multi-GPU parallel training is accuracy-preserving: the paper reports identical error curves to single-GPU training while cutting large-scale training time by about 32%.
  • Accuracy remains in the reported 0.78% to 3.42% maximum relative-error band across 2D and 3D problems, with error rising mildly at larger scales but decreasing with more training epochs.

Reading between the lines

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

  • Beyond the paper, the same hard-constraint idea should transfer to any variational problem with a potential energy functional, such as steady heat conduction, incompressible elasticity, or incremental plasticity, where slave-node values can be eliminated before energy assembly.
  • A testable extension would be to check whether master/slave asymmetry matters: with unidirectional mapping, swapping the roles of the two subdomains should ideally leave the energy and errors nearly unchanged, and a large difference would expose a directional bias.
  • The reported error levels are measured against a non-conforming finite element solution whose convergence is not demonstrated; a mesh-refinement study on the reference FEM would clarify whether the 0.78% to 3.42% numbers bound method error or include reference error.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes a discrete physics-informed neural network (dPINN) framework combined with domain decomposition (DDM) and an enforced interface constraint (EIC) mechanism. The method computes the network loss from element-wise Gaussian quadrature of strain energy and external work, rather than from strong-form PDE residuals. For interface continuity, EIC precomputes node–element pairs across subdomain interfaces and, at each training epoch, replaces the predicted displacement of slave interface nodes with shape-function interpolations from the adjacent master element. Two- and three-dimensional linear elastic problems are tested against non-conforming FEM solutions, reporting maximum relative errors of 0.78%–3.42%. Multi-GPU parallel training speedups are also reported. The central claims are that EIC enforces interface continuity without auxiliary loss penalties, eliminates the 'weak spatial constraint' (WSC) effect, and improves training stability and efficiency.

Significance. If the EIC mechanism were correctly and completely formulated, the contribution would be practically valuable: it would give a discrete energy-based PINN with hard interface constraints, no penalty-weight tuning, support for nonconforming subdomain meshes, and natural parallelization. The energy-minimization principle is sound, and the FEM validation strategy is appropriate. The paper ships no code, but the numerical experiments are extensive in scale (up to 3.8 million elements) and consistently compare against nonconforming FEM references. However, the significance as currently presented is undermined by a load-bearing gap in the derivation: as written, the EIC override does not demonstrably affect the minimized loss. The 'strict continuity' claim also overstates the mechanism, which enforces pointwise slave-node displacement only and imposes no traction or weak equilibrium at the interface. The WSC effect is introduced as an ad hoc concept without controlled isolation, and accuracy is reported without variance or a mesh-refinement study of the FEM reference.

major comments (5)
  1. [§2.2, Eq. (11); §2.1, Eq. (6)] The central EIC mechanism is not connected to the loss that is minimized. The loss in Eq. (6) is evaluated as a Gaussian quadrature of the strain energy, where the integrand σ(x_g)ᵀ ε(x_g) is computed from derivatives of the neural-network displacement field at Gauss points. Eq. (11) overrides point values of slave interface nodes, but those point values do not appear in the Gauss-point strain integrand as the equations are stated. Consequently, unless the element displacement field is explicitly constructed as a shape-function interpolation of neural-network nodal values, the override cannot change the strain energy of either subdomain, and the two subnetworks remain essentially uncoupled in the objective. The manuscript never states such an FE-style interpolation in Section 2.1 or 2.2. This must be corrected by either deriving the element displacement field from nodal values and shape functions (so that the override affects the quadrature integrand) or by explaining precisely how the override enters the energy functional.
  2. [§3, all numerical experiments] No comparison is made against soft-constraint DDM-PINNs, such as penalty-based or flux-based interface losses, even though the abstract and introduction claim that EIC eliminates the need for auxiliary loss terms and avoids loss imbalance. The reported accuracy is only relative to non-conforming FEM. A direct baseline comparison (e.g., a penalty-interface dPINN and a flux-interface variant) is needed to substantiate the central claim that EIC improves interface continuity and training stability over traditional soft-constraint approaches.
  3. [§3.1 and §3.2] Accuracy is reported without variance or statistical characterization for the main displacement-error results; the ten independent runs mentioned in §3.2.1 are used only for training-time boxplots. Also, the non-conforming FEM reference is assumed converged, but no mesh-independence or h-refinement study is provided. Without these, statements such as 'the overall accuracy remains within acceptable bounds' are not quantitatively supported to the standard one would expect for a method claiming to deliver controlled accuracy.
  4. [§2.2 and §4] The claim of 'strict' or 'hard' continuity across subdomain interfaces is overstated. Eq. (11) enforces displacement equality only at slave interface nodes, via one-sided interpolation from the master element; it does not enforce displacement continuity on the master side (the master nodes are not constrained by the slave field), nor does it impose traction continuity or a weak equilibrium condition across the interface. The resulting field is continuous in a pointwise sense only at the selected slave nodes, not in the strong sense claimed in the abstract and conclusion.
  5. [§3.1, 'Weak spatial constraint (WSC)'] The WSC effect is presented as a key phenomenon that the method must decouple, but it is never defined mathematically or isolated experimentally. The experiments show that a single network fails to represent a gap discontinuity; this is a known limitation of continuous coordinate-based networks and does not by itself validate the existence of a special 'weak spatial constraint' mechanism. A controlled study varying only the interface treatment while holding the network capacity fixed, or an analytic demonstration of how the input-coordinate proximity induces the effect, is needed to support the paper's conceptual framing.
minor comments (5)
  1. [Section 2 (organization paragraph)] The paragraph at the end of the Introduction contains 'In Section 2, xx. In Section 3, xx. Finally, xx in Section 4.' These placeholders should be filled with actual content.
  2. [General] There are typographical errors and garbled text in several equations and figure captions; for example, 'σ x ε' in Eq. (6) appears without the intended superscripts, and the text contains 'tthe' in Section 2.1. The manuscript needs a careful proofreading pass.
  3. [Reference [5]] Reference [5] is a book citation that uses a third-party redirect URL (sc.panda985.com) and no author list or publication year; this should be replaced with a proper bibliographic entry.
  4. [§3.1] The load magnitude is written as '43.6 10×' and later '36.4 10×' (and in §3.3 as '36.4 10×'); these are likely 3.6×10^4 kN and should be formatted consistently.
  5. [§2.3, Eq. (15)] The notation in Eq. (15) is ambiguous: the conditions use '∈ Γ_D' and '> Γ_D' (the latter likely '∉ Γ_D'), and the definitions of g(x) and l(x) are not fully specified. Clarify the symbols and the distinction.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the dPINN loss is an independent energy functional, FEM is used only as a benchmark, and EIC is a hard constraint rather than a fitted parameter.

full rationale

The derivation chain is self-contained with respect to circularity. The dPINN loss in Eq. (6) is the potential energy computed by Gaussian quadrature; network weights are trained to minimize that physical functional, and the non-conforming FEM solution is used only as an external benchmark (Section 3), never as training data or as a fitted parameter. The EIC mechanism (Eq. (11)) is introduced as a hard replacement of slave-interface nodal displacements, and the claimed continuity at those nodes is a direct consequence of that definition, not a hidden reuse of the validation output. No self-citation is load-bearing: references such as [37] supply an architecture component (random Fourier features), but the central energy formulation and interface constraint do not rest on an author-imported uniqueness theorem or ansatz. One non-circular concern should be flagged: as written, Eq. (6) evaluates strain energy from network derivatives at Gauss points, while Eq. (11) overrides node values; because the paper does not state that element strain fields are reconstructed from the overridden nodal values, the mechanism by which EIC changes the minimized loss is not made explicit. This is a missing-support or correctness issue, not a circular step, so it does not raise the circularity score.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The method rests on standard linear elasticity, variational energy minimization, and standard numerical quadrature. No new physical entities are introduced; the WSC label is a conceptual description. The main hidden assumptions are the adequacy of pointwise slave-master interpolation for non-conforming interfaces and the convergence of the reference FEM solution.

free parameters (4)
  • learning rate = 1e-3
    Adam initial learning rate with cosine annealing; standard hyperparameter, not fitted to physical data.
  • network width = 112 (single-network) and 56 (per subdomain)
    Chosen to keep parameter counts comparable between architectures; hand-set.
  • training epochs = 20,000 (default), 25,000 (extended)
    Hand-set; longer training improves accuracy.
  • random Fourier feature scale = not specified
    Random Fourier features are used, but the scale or bound is not reported, making reproduction difficult.
assumptions (5)
  • domain assumption Linear elastic constitutive model with small strains
    Eq. (2) assumes Cauchy stress and the strain-displacement relation in Eq. (4); experiments fix E=3.0 GPa and Poisson's ratio 0.3.
  • domain assumption Principle of minimum potential energy holds for the equilibrium problem
    The loss in Eq. (6) is the total potential energy; the method assumes its minimizer is the unique physical solution.
  • standard math Gaussian quadrature with the chosen number of points accurately integrates element energies
    Standard FEM practice; the exact quadrature order is not specified in the paper.
  • domain assumption Newton-Raphson inverse mapping converges to the correct reference coordinates for every interface node
    The preprocessing relies on this for all node-element pairs, but no failure cases are reported.
  • domain assumption Neural networks have sufficient capacity to represent the subdomain solutions
    Accuracy is demonstrated empirically; no universal approximation guarantee is proven.
invented entities (1)
  • Weak spatial constraint (WSC) effect
    purpose: To describe the tendency of a single coordinate-based network to smooth over spatial gaps
    This is a descriptive label for a known smoothness bias of neural networks, not a new physical entity. No independent falsifiable handle is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Enforced Interface Constraints for Domain Decomposition Method of Discrete Physics-Informed Neural Networks." pith.science (2026). https://pith.science/paper/QIUO2ZD2

@misc{pith2026250510925,
  author       = {Pith},
  title        = {Pith review of: Enforced Interface Constraints for Domain Decomposition Method of Discrete Physics-Informed Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QIUO2ZD2}},
  note         = {Machine review of arXiv:2505.10925}
}
read the original abstract

This study presents a discrete physics-informed neural network (dPINN) framework, enhanced with enforced interface constraints (EIC), for modeling physical systems using the domain decomposition method (DDM). Built upon finite element-style mesh discretization, the dPINN accurately evaluates system energy through Gaussian quadrature-based element-wise integration. To ensure physical field continuity across subdomain interfaces, the EIC mechanism enforces interfacial displacement constraints without requiring auxiliary sampling or loss penalties.This formulation supports independent meshing in each subdomain, simplifying preprocessing and improving computational flexibility. Additionally, by eliminating the influence of weak spatial constraints (WSC) commonly observed in traditional PINNs, the EIC-dPINN delivers more stable and physically consistent predictions.Extensive two- and three-dimensional numerical experiments validate the proposed framework's accuracy and demonstrate the computational efficiency gains achieved through parallel training. The results highlight the framework's scalability, robustness, and potential for solving large-scale, geometrically complex problems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 16 canonical work pages

  1. [1]

    Introduction In the rapidly evolving field of the scientific and engineering computation, Physics-Informed Neural Networks (PINNs) [1,2] have emerged as a powerful computational approach, demonstrating unique advantages and significant potential. By integrating fundamental physical laws into the le arning process of neural networks, PINNs not only enable ...

  2. [2]

    To ensure a well-posed problem, the boundary conditions ()k  are imposed on the domain 5 boundary k , with the prescribed boundary values ()kg x

    The methodologies 2.1 Discrete physics-informed neural networks For standard PINNs, the primary objective is to solve partial differential equations (PDEs), which can generally be written in the following form [8]: ( ( ); ) = ( ), ( ( )) = ( ), d kkk f g       ux λ xx ux x x   , (1) where () denotes the differential operator, ()ux represen...

  3. [3]

    Numerical experiments and discussion This section evaluates the predic tive performance of the propos ed discrete physics-informed neural network (dPINN) framework with enforced interface constraints (EIC) under the domain decomposition method (DDM). The primary objective is to assess its capacity to decouple the wea k spatial constraint (WSC) effect with...

  4. [4]

    Conclusion This study presents a discrete physics-informed neural network (dPINN) framework grounded in finite element discretization as the core modeling approach. In contrast to conventional collocation-based PINNs that approximate system energy via Monte Carlo integration, the proposed dPINN u tilizes Gaussian quadrature for element-wise energy computa...

  5. [5]

    Raissi, P

    M. Raissi, P. Perdikaris, G.E. Karniadakis, Physics-informe d neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partia l differential equations, Journal of Computational Physics 378 (2019) 686–707. https://doi.org/10.1016/j.jcp.2018.10.045

  6. [6]

    Cuomo, V .S

    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, J Sci Comput 92 (2022) 88. https://doi.org/10.1007/s10915-022-01939-z

  7. [7]

    H. Hu, L. Qi, X. Chao, Physics-informed Neural Networks (PI NN) for computational solid mechanics: Numerical frameworks and applicat ions, Thin-Walled Structures 2 05 (2024) 112495. https://doi.org/10.1016/j.tws.2024.112495

  8. [8]

    S. Cai, Z. Mao, Z. Wang, M. Yin, G.E. Karniadakis, Physics- informed neural networks (PINNs) for fluid mechanics: a review, Acta Mech. Sin. 37 (2021) 1727–1738. https://doi.org/10.1007/s10409-021-01148-1

Show all 45 references
  1. [9]

    https://sc.panda985.com/extdomains/books.google.com/books/about/Domain_Decomposition_Methods_Algorithm s.html?hl=zh-CN&id=tpSPx68R3KwC (accessed April 6, 2025)

    Domain Decomposition Methods - Algorithms and Theory, n.d. https://sc.panda985.com/extdomains/books.google.com/books/about/Domain_Decomposition_Methods_Algorithm s.html?hl=zh-CN&id=tpSPx68R3KwC (accessed April 6, 2025)

  2. [10]

    Smith, Domain Decomposition Methods for Partial Differ ential Equations, in: D.E

    B.F. Smith, Domain Decomposition Methods for Partial Differ ential Equations, in: D.E. Keyes, A. Sameh, V . Venkatakrishnan (Eds.), Parallel Numerical Algorithms, Springer Netherlands, Dordrecht, 1997: pp. 225–243. https://doi.org/10.1007/978-94-011-5412-3_8

  3. [11]

    Graham, P.O

    I.G. Graham, P.O. Lechner, R. Scheichl, Domain decompositio n for multiscale PDEs, Numer. Math. 106 (2007) 589–626. https://doi.org/10.1007/s00211-007-0074-1

  4. [12]

    Jagtap, E

    A.D. Jagtap, E. Kharazmi, G.E. Karniadakis, Conservative ph ysics-informed neural networks on discrete domains for conservation laws: Applications to forward and inve rse problems, Computer Methods in Applied Mechanics and Engineering 365 (2020) 113028. https://doi.org/10.1016...

  5. [13]

    Jagtap, G.E

    A.D. Jagtap, G.E. Karniadakis, Extended Physics-Informed Ne ural Networks (XPINNs): A Generalized Space- Time Domain Decomposition Based Deep Learning Framework for Non linear Partial Differential Equations, Communications in Computational Physics 28 (2020). https://doi.org/10...

  6. [14]

    Shukla, A.D

    K. Shukla, A.D. Jagtap, G.E. Karniadakis, Parallel physics-informed neural networks via domain decomposition, Journal of Computational Physics 447 (2021) 110683. https://doi.org/10.1016/j.jcp.2021.110683

  7. [15]

    Moseley, A

    B. Moseley, A. Markham, T. Nissen-Meyer, Finite basis phys ics-informed neural networks (FBPINNs): a scalable domain decomposition approach for solving differential equations, Adv Comput Math 49 (2023) 62. https://doi.org/10.1007/s10444-023-10065-9

  8. [16]

    X. Meng, Z. Li, D. Zhang, G.E. Karniadakis, PPINN: Pararea l physics-informed neural network for time- dependent PDEs, Computer Methods in Applied Mechanics and Engin eering 370 (2020) 113250. https://doi.org/10.1016/j.cma.2020.113250

  9. [17]

    Kharazmi, Z

    E. Kharazmi, Z. Zhang, G.E.M. Karniadakis, hp-VPINNs: Variational physics-informed neural networks with domain decomposition, Computer Methods in Applied Mechanics and Engineering 374 (2021) 113547. https://doi.org/10.1016/j.cma.2020.113547

  10. [18]

    S. Li, Y . Xia, Y . Liu, Q. Liao, A deep domain decomposition method based on Fourier features, Journal of Computational and Applied Mathematics 423 (2023) 114963. https://doi.org/10.1016/j.cam.2022.114963

  11. [19]

    Huang, Q.-Y

    G.-B. Huang, Q.-Y . Zhu, C.-K. Siew, Extreme learning machine: Theory and applications, Neurocomputing 70 (2006) 489–501. https://doi.org/10.1016/j.neucom.2005.12.126

  12. [20]

    Dwivedi, N

    V . Dwivedi, N. Parashar, B. Srinivasan, Distributed learning machines for solving forward and inverse problems in partial differential equations, Neurocomputing 420 (2021) 299–316. https://doi.org/10.1016/j.neucom.2020.09.006. 24

  13. [21]

    S. Dong, Z. Li, Local extreme learning machines and domain decomposition for solving linear and nonlinear partial differential equations, Computer Methods in Applied Mec hanics and Engineering 387 (2021) 114129. https://doi.org/10.1016/j.cma.2021.114129

  14. [22]

    Z. Hu, A.D. Jagtap, G.E. Karniadakis, K. Kawaguchi, Augmen ted Physics-Informed Neural Networks (APINNs): A gating network-based soft domain decomposition methodology, Engineering Applications of Artificial Intelligence 126 (2023) 107183. https://doi.org/10.1016/j.engappai.2023.107183

  15. [23]

    Droniou, R

    J. Droniou, R. Eymard, T. Gallouët, C. Guichard, R. Herbin , Non-conforming Finite Element Methods, in: J. Droniou, R. Eymard, T. Gallouët, C. Guichard, R. Herbin (Eds.), The Gradient Discretisation Method, Springer International Publishing, Cham, 2018: pp. 285–305. https://do...

  16. [24]

    Wohlmuth, A Mortar Finite Element Method Using Dual Spaces for the Lagrange Multiplier, SIAM Journal on Numerical Analysis (2006)

    B.I. Wohlmuth, A Mortar Finite Element Method Using Dual Spaces for the Lagrange Multiplier, SIAM Journal on Numerical Analysis (2006). https://doi.org/10.1137/S0036142999350929

  17. [25]

    Puso, T.A

    M.A. Puso, T.A. Laursen, A mortar segment-to-segment contact method for large deformation solid mechanics, Computer Methods in Applied Mechanics and Engineering 193 (2004 ) 601–629. https://doi.org/10.1016/j.cma.2003.10.010

  18. [26]

    Shi, A convergence condition for the quadrilateral Wils on element, Numer

    Z. Shi, A convergence condition for the quadrilateral Wils on element, Numer. Math. 44 (1984) 349–361. https://doi.org/10.1007/BF01405567

  19. [27]

    Hansbo, P

    A. Hansbo, P. Hansbo, An unfitted finite element method, b ased on Nitsche’s method, for elliptic interface problems, Computer Methods in Applied Mechanics and Engineering 191 (2002) 5537–5552. https://doi.org/10.1016/S0045-7825(02)00524-8

  20. [28]

    Stummel, The Generalized Patch Test, SIAM Journal on Nu merical Analysis (2006)

    F. Stummel, The Generalized Patch Test, SIAM Journal on Nu merical Analysis (2006). https://doi.org/10.1137/0716037

  21. [29]

    Ainsworth, Essential boundary conditions and multi-poin t constraints in finite element analysis, Computer Methods in Applied Mechanics and Engineering 190 (2001) 6323–63 39

    M. Ainsworth, Essential boundary conditions and multi-poin t constraints in finite element analysis, Computer Methods in Applied Mechanics and Engineering 190 (2001) 6323–63 39. https://doi.org/10.1016/S0045- 7825(01)00236-5

  22. [30]

    Nitsche, Über ein Variationsprinzip zur Lösung von Diri chlet-Problemen bei Verwendung von Teilräumen, die keinen Randbedingungen unterworfen sind, Abh.Math.Semin.Uni v.Hambg

    J. Nitsche, Über ein Variationsprinzip zur Lösung von Diri chlet-Problemen bei Verwendung von Teilräumen, die keinen Randbedingungen unterworfen sind, Abh.Math.Semin.Uni v.Hambg. 36 (1971) 9–15. https://doi.org/10.1007/BF02995904

  23. [31]

    Wilson, T

    P. Wilson, T. Teschemacher, P. Bucher, R. Wüchner, Non-conforming FEM-FEM coupling approaches and their application to dynamic structural analysis, Engineering Structu res 241 (2021) 112342. https://doi.org/10.1016/j.engstruct.2021.112342

  24. [32]

    D. Shi, L. Wang, X. Liao, A new nonconforming mixed finite element scheme for second order eigenvalue problem, Applied Mathematics and Computation 273 (2016) 842–855. https://doi.org/10.1016/j.amc.2015.10.064

  25. [33]

    Zhao, W.J

    J. Zhao, W.J. Chen, S.H. Lo, A refined nonconforming quadr ilateral element for couple stress/strain gradient elasticity, International Journal for Numerical Methods in Engi neering 85 (2011) 269–288. https://doi.org/10.1002/nme.2962

  26. [34]

    Rathore, W

    P. Rathore, W. Lei, Z. Frangella, L. Lu, M. Udell, Challenges in Training PINNs: A Loss Landscape Perspective, (2024). https://doi.org/10.48550/arXiv.2402.01868

  27. [35]

    S. Wang, Y . Teng, P. Perdikaris, Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks, SIAM J. Sci. Comput. 43 (2021) A3055–A3081. https://doi.org/10.1137/20M1318043

  28. [36]

    Szabó, Finite Element Analysis: Method, Verification and Validation, 2nd ed, John Wiley & Sons, Incorporated, Newark, 2021

    B.A. Szabó, Finite Element Analysis: Method, Verification and Validation, 2nd ed, John Wiley & Sons, Incorporated, Newark, 2021

  29. [37]

    J. Bai, T. Rabczuk, A. Gupta, L. Alzubaidi, Y . Gu, A physics-informed neural network technique based on a modified loss function for computational 2D and 3D solid mechan ics, Comput Mech 71 (2023) 543–562. https://doi.org/10.1007/s00466-022-02252-0. 25

  30. [38]

    Fried, Numerical integration in the finite element meth od, Computers & Structures 4 (1974) 921–932

    I. Fried, Numerical integration in the finite element meth od, Computers & Structures 4 (1974) 921–932. https://doi.org/10.1016/0045-7949(74)90015-7

  31. [39]

    Wang, Y .L

    J. Wang, Y .L. Mo, B. Izzuddin, C.-W. Kim, Exact Dirichlet boundary Physics-informed Neural Network EPINN for solid mechanics, Computer Methods in Applied Mechanics and Engineering 414 (2023) 116184. https://doi.org/10.1016/j.cma.2023.116184

  32. [40]

    Nguyen, P

    V .P. Nguyen, P. Kerfriden, M. Brino, S.P.A. Bordas, E. Bonisoli, Nitsche’s method for two and three dimensional NURBS patch coupling, Comput Mech 53 (2014) 1163–1182. https://doi.org/10.1007/s00466-013-0955-3

  33. [41]

    S. Wang, H. Wang, P. Perdikaris, On the eigenvector bias of Fourier feature networks: From regression to solving multi-scale PDEs with physics-informed neural networks, Computer Methods in Applied Mechanics and Engineering 384 (2021) 113938. https://doi.org/10.1016/j.cma.2021.113938

  34. [42]

    J.L. Ba, J.R. Kiros, G.E. Hinton, Layer Normalization, (2016). https://doi.org/10.48550/arXiv.1607.06450

  35. [43]

    L. Lu, R. Pestourie, W. Yao, Z. Wang, F. Verdugo, S.G. Johnson, Physics-Informed Neural Networks with Hard Constraints for Inverse Design, SIAM J. Sci. Comput. 43 (2021) B1105–B1132. https://doi.org/10.1137/21M1397908

  36. [44]

    Paszke, S

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. C hanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Köpf, E. Y ang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, S. Chintala, PyTorch: An Imperati ve Style, High-Perf...

  37. [45]

    Kingma, J

    D.P. Kingma, J. Ba, Adam: A Method for Stochastic Optimiza tion, (2017). https://doi.org/10.48550/arXiv.1412.6980

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.