Pith. sign in

REVIEW 4 major objections 3 minor 47 references

Training physics-informed networks on a shared latent geometry, with PDE residuals pulled back through the deformation gradient, yields large accuracy gains on unseen rotated geometries when training data is scarce.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 05:26 UTC pith:JJ7OPHBJ

load-bearing objection The pullback idea is standard, but this is a careful, honestly reported test of it on a hard nonlinear PDE with limited data; the big wins on rotations are real but rest on single runs, and the advertised shape-gradient mechanism is not what was actually implemented. the 4 major comments →

arxiv 2607.22215 v1 pith:JJ7OPHBJ submitted 2026-07-24 cs.LG physics.comp-ph

Latent PDE mapping for efficient physics-informed learning across geometries with limited data

Ingvild Askim Adde , Mary M. Maleckar , Gabriel Balaban This is my paper
classification cs.LG physics.comp-ph
keywords latent PDE mappingphysics-informed neural networksgeometric generalizationAliev-Panfilovdeformation gradientlimited datacardiac electrophysiologyshape gradients
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper claims that training physics-informed networks on a predefined latent geometry—pulling PDE residuals and boundary conditions back through the deformation gradient—recovers shape-gradient information that ordinary physics-informed losses discard. On the nonlinear Aliev-Panfilov cardiac PDE with only fifteen training geometries, the mapped models achieve roughly four to six times lower mean relative L2 error than latent-geometry-only baselines on rotation-dominated shape families, in both 2D and 3D, and with either explicit deformation parameters or PCA descriptors. Gains are modest for affine and shear families, indicating the technique helps most when geometric variation substantially changes the solution.

Core claim

Latent PDE mapping turns a geometry-dependent training problem into a parametric one: the PDE residual and boundary conditions are rewritten on a fixed reference domain with the deformation gradient F and Jacobian J encoding each geometry. The authors show that conventional physics losses implicitly drop a boundary-motion term when taking shape derivatives, whereas the mapped formulation lets the network's geometry sensitivity flow through the solution on a fixed domain. On the Aliev-Panfilov model, this yields error reductions of 4.3x in 2D rotation families and 3.9-5.1x in 3D rotations about the y- and z-axes, where rotations alter the fiber orientation and therefore the solution most.

What carries the argument

The deformation gradient F(X,s)=∇_X Φ^{-1}(X,s) and its determinant J, used to pull back the strong-form PDE, boundary conditions, and data from each physical geometry to a preselected latent domain. In the implementation, F and J are precomputed per geometry and treated as constant coefficient fields during training, so the shape gradient actually computed is the solution sensitivity ∂V/∂s and ∂W/∂s, not the deformation-map sensitivity ∂F/∂s and ∂J/∂s.

Load-bearing premise

The method assumes an exact deformation gradient from each physical geometry to the latent one, and the implemented training treats F and J as frozen constants, so if real geometries lack known maps or the map's own sensitivity matters, the advertised shape-gradient mechanism and practical applicability are unsupported.

What would settle it

Compute the two deformation-map sensitivity terms ∂R/∂F · ∂F/∂s and ∂R/∂J · ∂J/∂s for the rotation families and check whether their magnitudes are negligible against the solution-sensitivity terms; alternatively, run latent PDE mapping on geometries whose deformation gradient is approximated by noisy or estimated maps and see whether the 4-6x advantage over baselines disappears.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • In geometric families where boundary motion is large and the PDE solution changes substantially, latent PDE mapping gives roughly 4-6x lower mean relative L2 error than latent-geometry-only baselines, across two network architectures and two geometric descriptors.
  • Because the PDE is evaluated on a fixed latent domain, collocation points can be shared across geometries, and the added cost is modest during training and negligible at inference since F and J are computed once during meshing.
  • The framework is architecture-agnostic and formulated for diffeomorphic geometries, so the same pull-back construction can be attached to graph networks, neural operators, or other physics-informed backbones.
  • The paper's numerical quantification shows that the boundary-motion term neglected by conventional physics losses is larger than the retained shape gradient in every 2D family tested, giving a concrete diagnostic for when latent PDE mapping should help.
  • Effectiveness persists when geometries are described by PCA scores rather than explicit deformation parameters, which is the more realistic setting for anatomical populations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The theoretical motivation advertises the full shape gradient including ∂F/∂s and ∂J/∂s, but the experiments freeze F and J; if those deformation-map sensitivity terms are non-negligible, the empirical gains may come less from 'accurate shape gradients' than from simply evaluating residuals on a fixed reference domain.
  • A testable screening rule follows from the paper's own measurements: latent PDE mapping should help most when the boundary-motion term ∂x/∂s·n is large and the residual on the boundary is non-negligible, which is consistent with the small gains on expansion and shear families.
  • For real-world anatomies without closed-form deformation maps, the method would need numerically estimated F and J; performance likely degrades with map-error, and the paper leaves this unquantified.
  • Static deformations are assumed throughout; time-dependent or moving-boundary problems would require reinstating the dropped ∂F/∂s and ∂J/∂s terms, so the practical extension to morphing geometries is an open question.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 3 minor

Summary. This paper proposes latent PDE mapping (LPM), a method for training physics-informed neural networks (PINNs and PI-DONs) on families of geometries. The PDE residual and boundary terms are pulled back from physical domains Ω(s) to a fixed latent domain Ω0 using the deformation gradient F and Jacobian J, and the solution is represented as a function of latent coordinates and a shape descriptor s. The method is instantiated on the anisotropic Aliev–Panfilov cardiac electrophysiology model in 2D and 3D, with training on 10 geometries per family and evaluation on held-out internal and external parameter ranges. The reported mean relative L2 errors show consistent but modest gains on expansion/shear families and large gains on rotation families, with a ~4–6× reduction claimed on external rotations. Ablations include LG-PINN/DON (latent geometry but no residual pullback), PA-PINN, and Basic-PINN, with either deformation parameters or PCA scores as shape descriptors.

Significance. The underlying pullback construction is mathematically standard and clearly presented; if the empirical claims hold, LPM would be a useful, architecture-agnostic ingredient for geometry-generalizable physics-informed models. The paper deserves credit for a detailed derivation, public synthetic data, and a broader comparison than many PINN papers. However, the headline quantitative claims hinge on experimental details that are not yet established: the external rotation families may not actually be outside the training distribution under 180° symmetry, all results are single-seed, and the implemented loss does not exercise the full advertised shape gradient. The significance of the contribution therefore remains conditional.

major comments (4)
  1. [Section 2.2 / Table 1 / Tables 3–4] The claimed out-of-distribution gains on rotation families are undermined by a symmetry degeneracy. Under the natural centered-coordinate interpretation of the latent square/cube, rotation by θ+180° maps Ω0 to itself, and since the diffusion tensor D is invariant under R180 (D = R180 D R180^T), the physical problem for θ+180° is identical to that for θ. Because G*_rot / H*_rot are defined only by θ ∉ [-90,90], every external angle is equivalent to an internal one (e.g., 120° ≡ -60°). Thus the ~4.3–5.1× improvements on G*_rot, Hy*_rot, Hz*_rot in Tables 3–4 do not demonstrate generalization to unseen geometries. Please either use a latent geometry without 180° symmetry or choose an external range within one symmetry period (e.g., internal [-45,45], external [45,135] with endpoints excluded).
  2. [Section 4.1 / Tables 3–4] No multi-seed evaluation is reported. The ± values in Tables 3 and 4 are standard deviations across test geometries for a single training run; the paired Wilcoxon tests treat each geometry as an independent observation from that one run and cannot quantify stochastic variability. As the authors acknowledge in Section 4.1, the influence of random initialization is unquantified. The central claim of 'generalizable models from limited data' requires at least 3–5 seeds per configuration with the resulting error distributions reported, especially for the headline rotation-family comparisons.
  3. [Section 2.1.2 / Eq. (12) / Appendix A.1, Eq. (34)] The implemented shape gradient drops the deformation-map sensitivity terms. The text claims Eq. (12) computes the shape gradient directly and that 'the integrand does not vary with s', but R_LPM in Eq. (6) depends on s through F and J; the full chain rule in Eq. (33) includes ∂R/∂F ∂F/∂s and ∂R/∂J ∂J/∂s. In training these terms are absent because F and J are precomputed and treated as constant coefficient fields (Appendix A.1). Consequently, the advertised mechanism—accurate shape-gradient calculation—is not what is exercised in the experiments; only the solution-sensitivity part (34) remains. Please either implement the full gradient or revise the motivation/claims to state that the benefit of LPM in these experiments comes from the residual pullback itself, not from deformation-map gradients.
  4. [Section 2.2] The PCA descriptors are constructed using all 85 geometries, including the 35 external test geometries ('The sampled geometries covered the full parameter ranges of the internal and external geometric families... n=85'). The test points therefore influence the PCA basis and the scores used as network inputs. This is a form of test leakage for the PCA experiments in Tables 3 and 4 and weakens the claim that LPM works with low-dimensional descriptors on unseen geometries. PCA should be fit on the training geometries only, or the experiment should be repeated in a fully nested fashion.
minor comments (3)
  1. [Appendix A, Eq. (30)] The boundary-condition integral should be over ∂Ω0 with dS, not over Ω0 with dΩ0; the current equation is dimensionally inconsistent.
  2. [Section 2.1.2] The sentence 'the integrand does not vary with s' is inaccurate: R_LPM depends on s through F, J, and u. Consider rephrasing to 'the integration domain does not vary with s'.
  3. [Tables 3–4] The text refers to 'best and second-best performing models'; please clarify explicitly that comparisons are within each architecture (PINN vs PINN, DON vs DON), as a reader may otherwise attempt cross-architecture comparisons that the paper says are out of scope.

Circularity Check

0 steps flagged

No circularity: latent PDE mapping is standard pullback calculus with predictions scored against independent FEM solutions; acknowledged limitations are robustness issues, not circular reasoning.

full rationale

The central derivation (Eq. 5-8; Appendix A) is a textbook change-of-variables: the residual on Ω(s) is pulled back to Ω0 via F = ∇X Φ^{-1} and J = det F, giving Eq. 8. This is not defined in terms of the quantity being predicted: the prediction target is the FEM transmembrane potential from openCARP, and accuracy is measured with the independent relative L2 error in Eq. 20. No fitted constant is renamed as a prediction, no self-citation carries the argument (the authors do not cite their own prior work as load-bearing), and no uniqueness theorem is imported to force the choice of mapping. The paper's stated limitations are real but not circular: Section 4.1 acknowledges 'reported experiments were conducted without a systematic evaluation across multiple random seeds' (statistical robustness); Appendix A.1 states F and J 'were not part of the computational graph' so the implemented shape gradient is Eq. 34, not the full Eq. 33 (theory-implementation gap); and the PCA descriptors in Section 2.2 are computed over the combined internal/external sample, a possible leakage concern. Each of these affects how strongly the empirical claims can be trusted, but none makes the derivation equivalent to its inputs. The pullback is a genuine transformation of the PDE residual, and the claimed improvements are evaluated against an external FEM benchmark, not against the method's own outputs.

Axiom & Free-Parameter Ledger

4 free parameters · 6 axioms · 0 invented entities

The central method uses standard calculus; no fitted physical constants or invented entities. The main ad hoc choices are experimental design choices (latent geometry, loss weights, PCA dimension), plus one significant disclosed approximation (frozen F and J).

free parameters (4)
  • latent geometry choice = 10×10 mm square (2D); 10×10×10 mm cube (3D)
    Chosen by hand; no sensitivity study. The paper admits the influence of the latent geometry is unexplored.
  • loss weights = equal weights (1) on data/phys/bc/ic terms
    Chosen without tuning; could affect the relative benefit of LPM vs ablations.
  • PCA descriptor dimension = 2 principal components
    Arbitrarily chosen as a low-dimensional geometry descriptor; not swept.
  • train/validation/test split sizes = 10 train / 5 validation / 35 internal test / 35 external test per family
    Defines the limited-data regime; a different split could change conclusions.
axioms (6)
  • domain assumption A diffeomorphic map Φ(s): x(s)→X exists between each physical geometry and the latent geometry
    Used to define the deformation gradient in Eq. (5); all tested geometries are generated by explicit deformations, so it holds.
  • domain assumption F is invertible
    Required for F^{-1} in Eq. (28); the rotation/affine/quadratic parameter ranges avoid singularities.
  • domain assumption The mapping is time-independent
    Needed for ∂(JV)/∂τ = J ∂V/∂τ in Eqs. (31)–(32); time-dependent deformations are declared out of scope.
  • ad hoc to paper F and J can be treated as constant coefficient fields during training
    Eq. (34) drops ∂F/∂s and ∂J/∂s; the paper acknowledges this as an approximation and a limitation.
  • domain assumption FEM solutions from openCARP are accurate enough as ground truth
    Relies on the Niederer et al. benchmark for mesh and time-step levels; not independently verified here.
  • standard math Neural network training reaches a low enough loss for generalization comparisons to be meaningful
    Standard ML assumption; no formal convergence guarantees are provided.

pith-pipeline@v1.3.0-alltime-deepseek · 21811 in / 14648 out tokens · 133093 ms · 2026-08-01T05:26:14.599702+00:00 · methodology

0 comments
read the original abstract

In this study, we introduce latent PDE mapping, a broadly applicable physics-informed learning technique designed to enable efficient geometric generalization with sparse training data. Latent PDE mapping pulls back geometry-specific PDE residuals and boundary conditions to a predefined latent geometry via the deformation gradient, thereby enabling the automated calculation of geometry-consistent shape gradients that are missing in conventional physics-informed machine learning formulations. We demonstrate the utility of latent PDE mapping in solving the anisotropic Aliev-Panfilov PDE of cardiac electrophysiology using both physics-informed neural networks and physics-informed deep operator networks. The Aliev-Panfilov PDE serves as a challenging exemplar: a nonlinear, time-dependent PDE benchmark with sharp gradients that are expensive to capture using traditional numerical solvers. To represent the limited data regime, we train the networks using just fifteen geometric samples drawn from parameterized distributions in two and three spatial dimensions. While modest improvements appear for geometries parameterized by affine and shear deformations, latent PDE mapping demonstrates significant benefits on select geometric families, achieving a factor ~4-6 reduction in mean relative L2 error. Furthermore, our results show that the computational cost of applying latent PDE mapping was modest during network training, and negligible at inference. Taken together, our study highlights how latent PDE mapping facilitates the creation of generalizable physics-informed machine learning models from limited sets of training geometries.

Figures

Figures reproduced from arXiv: 2607.22215 by Gabriel Balaban, Ingvild Askim Adde, Mary M. Maleckar.

Figure 1
Figure 1. Figure 1: Latent PDE mapping (LPM) enables accurate predictions on novel, unseen geometries. (a) Training phase: Spatiotemporal data from each physical geometry are mapped to a predefined latent geometry via a diffeomorphic mapping (ΦN ), and passed to a neural network (N N θ) alongside shape parameters (s) encoding each geometry. FEM data, PDE residuals, and BC residuals defined on the physical geometries are pulle… view at source ↗
Figure 2
Figure 2. Figure 2: The figure shows an example geometry from [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Schematic of the neural network architectures employed in our study. (a) The [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Snapshots of predicted transmembrane voltages ( [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Quantification of the mean shape gradient and mean missing boundary shape gradient, together with their [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Snapshots of predicted transmembrane voltages ( [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Boxplots of the average training time per epoch in 2D and 3D (top row) and the average inference time per [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

47 extracted references · 2 canonical work pages

  1. [1]

    Meshless physics-informed deep learning method for three-dimensional solid mechanics

    Abueidda, D.W., Lu, Q., Koric, S., 2021. Meshless physics-informed deep learning method for three-dimensional solid mechanics. International Journal for Numerical Methods in Engineering 122, 7182–7201

    2021

  2. [2]

    Machine learning in fluid dynamics—physics- informed neural networks (pinns) using sparse data: A review

    Ali, M., Miron, P., Mönnigmann, M., Nikolay, B., et al., 2025. Machine learning in fluid dynamics—physics- informed neural networks (pinns) using sparse data: A review. Fluids 10, 226

    2025

  3. [3]

    A simple two-variable model of cardiac excitation

    Aliev, R.R., Panfilov, A.V ., 1996. A simple two-variable model of cardiac excitation. Chaos, Solitons & Fractals 7, 293–301

    1996

  4. [4]

    Uncovering near-wall blood flow from sparse data with physics- informed neural networks

    Arzani, A., Wang, J.X., D’Souza, R.M., 2021. Uncovering near-wall blood flow from sparse data with physics- informed neural networks. Physics of Fluids 33

    2021

  5. [5]

    Physics-informed neural networks for transformed geometries and manifolds

    Burbulla, S., 2023. Physics-informed neural networks for transformed geometries and manifolds. arXiv preprint arXiv:2311.15940

    Pith/arXiv arXiv 2023

  6. [6]

    Solved in unit domain: Jacobinet for differentiable coordinate-transformed pinns

    Chen, X., Yang, J., Zhang, J., Yang, R., Liu, X., Wang, H., Zheng, T., Ren, Z., Hu, W., 2026. Solved in unit domain: Jacobinet for differentiable coordinate-transformed pinns. Journal of Computational Physics 563, 115074

    2026

  7. [7]

    Engineering Applications of Artificial Intelligence 127, 107324

    Costabal, F.S., Pezzuto, S., Perdikaris, P., 2024.δ-pinns: physics-informed neural networks on complex geometries. Engineering Applications of Artificial Intelligence 127, 107324

    2024

  8. [8]

    Scientific machine learning through physics–informed neural networks: Where we are and what’s next

    Cuomo, S., Di Cola, V .S., Giampaolo, F., Rozza, G., Raissi, M., Piccialli, F., 2022. Scientific machine learning through physics–informed neural networks: Where we are and what’s next. Journal of Scientific Computing 92, 88

    2022

  9. [9]

    Physics-informed graph neural network emulation of soft-tissue mechanics

    Dalton, D., Husmeier, D., Gao, H., 2023. Physics-informed graph neural network emulation of soft-tissue mechanics. Computer Methods in Applied Mechanics and Engineering 417, 116351

    2023

  10. [10]

    Pinning cerebral blood flow: analysis of perfusion mri in infants using physics-informed neural networks

    Galazis, C., Chiu, C.E., Arichi, T., Bharath, A.A., Varela, M., 2025. Pinning cerebral blood flow: analysis of perfusion mri in infants using physics-informed neural networks. Frontiers in network physiology 5, 1488349

    2025

  11. [11]

    Phygeonet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state pdes on irregular domain

    Gao, H., Sun, L., Wang, J.X., 2021. Phygeonet: Physics-informed geometry-adaptive convolutional neural networks for solving parameterized steady-state pdes on irregular domain. Journal of Computational Physics 428, 110079

    2021

  12. [12]

    Physics-informed graph neural galerkin networks: A unified framework for solving pde-governed forward and inverse problems

    Gao, H., Zahr, M.J., Wang, J.X., 2022. Physics-informed graph neural galerkin networks: A unified framework for solving pde-governed forward and inverse problems. Computer Methods in Applied Mechanics and Engineering 390, 114502

    2022

  13. [13]

    Physics-informed deep neural operator networks, in: Machine learning in modeling and simulation: methods and applications

    Goswami, S., Bora, A., Yu, Y ., Karniadakis, G.E., 2023. Physics-informed deep neural operator networks, in: Machine learning in modeling and simulation: methods and applications. Springer, pp. 219–254

    2023

  14. [14]

    B-spline curve theory: An overview and applications in real life

    Hasan, M.S., Alam, M.N., Fayz-Al-Asad, M., Muhammad, N., Tunç, C., 2024. B-spline curve theory: An overview and applications in real life. Nonlinear Engineering 13, 20240054

    2024

  15. [15]

    Nonlinear Solid Mechanics: A Continuum Approach for Engineering

    Holzapfel, G.A., 2000. Nonlinear Solid Mechanics: A Continuum Approach for Engineering. Wiley

    2000

  16. [16]

    Principal component analysis: a review and recent developments

    Jolliffe, I.T., Cadima, J., 2016. Principal component analysis: a review and recent developments. Philosophical transactions of the royal society A: Mathematical, Physical and Engineering Sciences 374, 20150202

    2016

  17. [17]

    Physics-informed machine learning

    Karniadakis, G.E., Kevrekidis, I.G., Lu, L., Perdikaris, P., Wang, S., Yang, L., 2021. Physics-informed machine learning. Nature Reviews Physics 3, 422–440

    2021

  18. [18]

    Physics-informed latent neural operator for real-time predictions of time-dependent parametric pdes

    Karumuri, S., Graham-Brady, L., Goswami, S., 2026. Physics-informed latent neural operator for real-time predictions of time-dependent parametric pdes. Computer Methods in Applied Mechanics and Engineering 450, 118599. 23

    2026

  19. [19]

    Physics-informed pointnet: A deep learning solver for steady-state incompressible flows and thermal fields on multiple sets of irregular geometries

    Kashefi, A., Mukerji, T., 2022. Physics-informed pointnet: A deep learning solver for steady-state incompressible flows and thermal fields on multiple sets of irregular geometries. Journal of Computational Physics 468, 111510

    2022

  20. [20]

    Fourier neural operator with learned deformations for pdes on general geometries

    Li, Z., Huang, D.Z., Liu, B., Anandkumar, A., 2023a. Fourier neural operator with learned deformations for pdes on general geometries. Journal of Machine Learning Research 24, 1–26

  21. [21]

    Geometry-informed neural operator for large-scale 3d pdes

    Li, Z., Kovachki, N., Choy, C., Li, B., Kossaifi, J., Otta, S., Nabian, M.A., Stadler, M., Hundt, C., Azizzadenesheli, K., et al., 2023b. Geometry-informed neural operator for large-scale 3d pdes. Advances in Neural Information Processing Systems 36, 35836–35854

  22. [22]

    Deeponet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators

    Lu, L., Jin, P., Karniadakis, G.E., 2019. Deeponet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators. arXiv preprint arXiv:1910.03193

    Pith/arXiv arXiv 2019

  23. [23]

    Aerodynamic shape optimization investigations of the common research model wing benchmark

    Lyu, Z., Kenway, G.K., Martins, J.R., 2015. Aerodynamic shape optimization investigations of the common research model wing benchmark. AIAA journal 53, 968–985

    2015

  24. [24]

    Right ventricular shape and function: cardiovascular magnetic resonance reference morphology and biventricular risk factor morphometrics in uk biobank

    Mauger, C., Gilbert, K., Lee, A.M., Sanghvi, M.M., Aung, N., Fung, K., Carapella, V ., Piechnik, S.K., Neubauer, S., Petersen, S.E., et al., 2019. Right ventricular shape and function: cardiovascular magnetic resonance reference morphology and biventricular risk factor morphometrics in uk biobank. Journal of Cardiovascular Magnetic Resonance 21, 41

    2019

  25. [25]

    A framework for physics-informed deep learning over freeform domains

    Mezzadri, F., Gasick, J., Qian, X., 2023. A framework for physics-informed deep learning over freeform domains. Computer-Aided Design 160, 103520

    2023

  26. [26]

    A bi-atrial statistical shape model for large-scale in silico studies of human atria: model development and application to ecg simulations

    Nagel, C., Schuler, S., Dössel, O., Loewe, A., 2021. A bi-atrial statistical shape model for large-scale in silico studies of human atria: model development and application to ecg simulations. Medical Image Analysis 74, 102210

    2021

  27. [27]

    Verification of cardiac tissue electrophysiology simulators using an n-version benchmark

    Niederer, S.A., Kerfoot, E., Benson, A.P., Bernabeu, M.O., Bernus, O., Bradley, C., Cherry, E.M., Clayton, R., Fenton, F.H., Garny, A., et al., 2011. Verification of cardiac tissue electrophysiology simulators using an n-version benchmark. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 369, 4331–4351

    2011

  28. [28]

    Scaling digital twins from the artisanal to the industrial

    Niederer, S.A., Sacks, M.S., Girolami, M., Willcox, K., 2021. Scaling digital twins from the artisanal to the industrial. Nature Computational Science 1, 313–320

    2021

  29. [29]

    An active strain electromechanical model for cardiac tissue

    Nobile, F., Quarteroni, A., Ruiz-Baier, R., 2012. An active strain electromechanical model for cardiac tissue. International journal for numerical methods in biomedical engineering 28, 52–71

    2012

  30. [30]

    Geometry aware physics informed neural network surrogate for solving navier–stokes equation (gapinn)

    Oldenburg, J., Borowski, F., Öner, A., Schmitz, K.P., Stiehm, M., 2022. Geometry aware physics informed neural network surrogate for solving navier–stokes equation (gapinn). Advanced Modeling and Simulation in Engineering Sciences 9, 8

    2022

  31. [31]

    openCARP

    openCARP consortium, Augustin, C., Boyle, P.M., Loechner, V ., Colin, R., Huppé, A., Gsell, M., Houillon, M., Huang, Y .L.C., Hustad, K.G., Karabelas, E., Loewe, A., Myklebust, L., Neic, A., Nothstein, M., Plank, G., Prassl, A., Sánchez, J., Seemann, G., Stary, T., Thangamani, A., Tippmann, N., Trevisan Jost, T., Vigmond, E., Wülfers, E.M., Linder, M., 20...

    work page doi:10.35097/1979 2024

  32. [32]

    Exploration of continuous variability in collections of 3d shapes

    Ovsjanikov, M., Li, W., Guibas, L., Mitra, N.J., 2011. Exploration of continuous variability in collections of 3d shapes. ACM Transactions on Graphics (TOG) 30, 1–10

    2011

  33. [33]

    Physics-informed graph convolutional neural network for modeling geometry-adaptive steady-state natural convection

    Peng, J.Z., Aubry, N., Li, Y .B., Mei, M., Chen, Z.H., Wu, W.T., 2023. Physics-informed graph convolutional neural network for modeling geometry-adaptive steady-state natural convection. International Journal of Heat and Mass Transfer 216, 124593

    2023

  34. [34]

    The openCARP simulation environment for cardiac electrophysiology

    Plank*, G., Loewe*, A., Neic*, A., Augustin, C., Huang, Y .L.C., Gsell, M., Karabelas, E., Nothstein, M., Sánchez, J., Prassl, A., Seemann*, G., Vigmond*, E., 2021. The openCARP simulation environment for cardiac electrophysiology. Computer Methods and Programs in Biomedicine 208, 106223. doi: 10.1016/j.cmpb.2021. 106223

    work page doi:10.1016/j.cmpb.2021 2021

  35. [35]

    Physics-informed neural networks: A deep learning frame- work for solving forward and inverse problems involving nonlinear partial differential equations

    Raissi, M., Perdikaris, P., Karniadakis, G.E., 2019. Physics-informed neural networks: A deep learning frame- work for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational physics 378, 686–707

    2019

  36. [36]

    Universal solution manifold networks (usm-nets): non-intrusive mesh-free surrogate models for problems in variable domains

    Regazzoni, F., Pagani, S., Quarteroni, A., 2022. Universal solution manifold networks (usm-nets): non-intrusive mesh-free surrogate models for problems in variable domains. Journal of Biomechanical Engineering 144, 121004

    2022

  37. [37]

    Deeponet-enhanced bayesian inference for biphasic tumor growth modeling

    Sotiropoulos, G., Litvinov, S., Kaltenbach, S., Harkos, C., Atzarakis, K., Stylianopoulos, T., Koumoutsakos, P., Papadopoulos, V ., 2026. Deeponet-enhanced bayesian inference for biphasic tumor growth modeling. Computer Methods in Applied Mechanics and Engineering 456, 118919. 24

    2026

  38. [38]

    Physics-informed deep learning for simultaneous surrogate modeling and pde-constrained optimization of an airfoil geometry

    Sun, Y ., Sengupta, U., Juniper, M., 2023. Physics-informed deep learning for simultaneous surrogate modeling and pde-constrained optimization of an airfoil geometry. Computer Methods in Applied Mechanics and Engineering 411, 116042

    2023

  39. [39]

    Latentpinns: Generative physics-informed neural networks via a latent representation learning

    Taufik, M.H., Alkhalifah, T., 2025. Latentpinns: Generative physics-informed neural networks via a latent representation learning. Artificial Intelligence in Geosciences , 100115

    2025

  40. [40]

    Artificial intelligence in cardiac electrophysiology: enhancing mapping and ablation precision

    Tran, H.H.V ., Thu, A., Twayana, A.R., Fuertes, A., Gonzalez, M., Harrison, J.L., Mehta, K.A., James, M., Basta, M., Frishman, W.H., et al., 2025. Artificial intelligence in cardiac electrophysiology: enhancing mapping and ablation precision. Cardiology in Review , 10–1097

    2025

  41. [41]

    Exact dirichlet boundary physics-informed neural network epinn for solid mechanics

    Wang, J., Mo, Y ., Izzuddin, B., Kim, C.W., 2023a. Exact dirichlet boundary physics-informed neural network epinn for solid mechanics. Computer Methods in Applied Mechanics and Engineering 414, 116184

  42. [42]

    An expert’s guide to training physics-informed neural networks

    Wang, S., Sankaran, S., Wang, H., Perdikaris, P., 2023b. An expert’s guide to training physics-informed neural networks. arXiv preprint arXiv:2308.08468

    Pith/arXiv arXiv

  43. [43]

    Learning the solution operator of parametric partial differential equations with physics-informed deeponets

    Wang, S., Wang, H., Perdikaris, P., 2021. Learning the solution operator of parametric partial differential equations with physics-informed deeponets. Science advances 7, eabi8605

    2021

  44. [44]

    Individual comparisons by ranking methods

    Wilcoxon, F., 1945. Individual comparisons by ranking methods. Biometrics bulletin 1, 80–83

    1945

  45. [45]

    Physics-informed meshgraphnets (pi- mgns): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes

    Würth, T., Freymuth, N., Zimmerling, C., Neumann, G., Kärger, L., 2024. Physics-informed meshgraphnets (pi- mgns): Neural finite element solvers for non-stationary and nonlinear simulations on arbitrary meshes. Computer Methods in Applied Mechanics and Engineering 429, 117102

    2024

  46. [46]

    A scalable framework for learning the geometry-dependent solution operators of partial differential equations

    Yin, M., Charon, N., Brody, R., Lu, L., Trayanova, N., Maggioni, M., 2024. A scalable framework for learning the geometry-dependent solution operators of partial differential equations. Nature Computational Science 4, 928–940

    2024

  47. [47]

    Physics-informed geometry-aware neural operator

    Zhong, W., Meidani, H., 2025. Physics-informed geometry-aware neural operator. Computer Methods in Applied Mechanics and Engineering 434, 117540. 25

    2025