Pith. sign in

REVIEW 4 major objections 6 minor 52 references

Learning Mappings in Mesh-based Simulations

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read An 85%-information grid encoding lets a UNet learn mesh mappings better than Fourier and transformer baselines.

desk verdict A solid, honest empirical study built on a standard bilinear-scatter encoding; the information-theoretic claim sits on a hand-set constant and the 'consistent gains' wording overreaches. read the letter →

arxiv 2506.12652 v1 pith:6EZQOKNY submitted 2025-06-14 cs.LG

classification cs.LG
keywords pointcloudencodingmesh-basedsimulationtopologyconvolutionalneuralnetworkUNetoperatorgridrepresentationscientificmachinelearning
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 claims that the main bottleneck for learning mappings from mesh-based simulation data is the encoding, not the network. It introduces a parameter-free encoder that spreads each mesh node's footprint onto the vertices of a surrounding regular grid, producing a structured image-like representation that standard convolutions can consume directly. The authors quantify that this footprint encoding retains about 85% of the available positional information at their reference setting, versus about 35% for binary and count-based encodings. They then pair the encoder with a UNet variant (E-UNet) and report that it matches or beats Fourier-neural-operator and transformer baselines across five 2D problems and one 3D problem, with better data efficiency, noise robustness, and lower training cost, while also enabling recovery of full responses from partial point observations.

What carries the argument

The load-bearing object is the footprint-based topology encoding: each point scatters four weights $w_{LT}, w_{LB}, w_{RT}, w_{RB}$ to the vertices of its enclosing grid cell, with the weights equal to the areas of the sub-rectangles defined by the point's position, so the encoded grid $G_o$ accumulates geometry without trainable parameters. The companion normalization $H = G_u/(G_o+10^{-6})$ turns accumulated response values into weighted averages, making reconstruction a single bilinear interpolation with a provable $O(h^\beta)$ error that degrades gracefully as field smoothness decreases. E-UNet then learns the grid-to-grid map $G_o \mapsto \tilde{G}_u$ using a U-shaped convolutional network whose final output is multiplied by $G_o$ as a mask, with an edge-weighted relative-$L_2$ loss that concentrates training near sharp gradients.

What would settle it

Measure the actual uncertainty in recovering point positions from $G_o$ at $r=128$ with $N=2000$, infer the true $\delta_\mathrm{eff}$, and recompute $H_G$ from Equation (11); if the inferred $\eta$ exceeds roughly 0.1, the 85% figure drops below the about-35% value of binary and count-based encodings at the paper's reference setting.

Watch

Extended reading notes

Core claim

The central claim is that an analytic, parameter-free grid encoding can carry enough of the mesh topology for convolutional networks to learn high-accuracy mappings from point clouds to response fields. Each point contributes four weights to the vertices of its enclosing cell, with weights equal to the areas of the sub-rectangles the point's position cuts out of that cell; summing these contributions builds an encoded topology grid $G_o$ that preserves the total point count. The encoded response grid $G_u$ is normalized by $G_o$ to form weighted averages, which are then inverted by bilinear interpolation with an $O(h^\beta)$ error bound that depends on the smoothness of the underlying field. The paper reports that this encoding keeps roughly 85% of the available information at $N=2000$, $r=128$ (versus about 35% for binary and count-based encodings), and that E-UNet trained on these representations outperforms FNO-based, GeoFNO, GNOT, and GeomDeepONet baselines in most benchmarks, with the notable exception of the Circles problem where GNOT is more accurate but far more expensive.

Load-bearing premise

The 85% information figure rests on a hand-picked and unmeasured blur constant inside the formula for how much the grid smears point positions; if the true blur is larger, the claimed advantage over binary and count-based encodings shrinks or disappears.

Editorial extensions

If this is right

  • Any point cloud that fits in a regular coordinate box can be fed to standard CNNs or FFT-based models without learned preprocessing, because the encoder has no trainable parameters.
  • Because the encoded grid preserves total point count and is invertible up to a controlled interpolation error, predictions made on the grid can be mapped back to the original scattered node locations, so the method is usable as a surrogate for mesh-based solvers.
  • The information-content comparison predicts that the advantage of the footprint encoding over binary and count-based encodings grows with the number of points and remains at high resolution, with the 85% figure as the headline instance.
  • Chaining a small grid-to-grid network that reconstructs the full encoded topology from a partial encoded response lets the pipeline recover responses at missing sensor locations within the trained dropout range, something pointwise transformers such as GNOT are not designed to do.
  • Training time scales as $O(N + r^2(1 + (c_\mathrm{in}+c_\mathrm{out})c + (1+2L)c^2))$, so the encoding itself contributes only linearly in point count, making grid resolution $r$ the main cost lever.

Reading between the lines

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

  • Inference: the information-content ratio is sensitive to the hand-set constant $\eta$ in $\delta_\mathrm{eff}^2 \approx \delta^2 + \eta h^2$; calibrating $\eta$ from actual reconstruction errors on a few meshes would turn the 85% claim into a falsifiable, dataset-specific number rather than an illustrative one.
  • Inference: the same footprint construction could be extended to higher-order interpolation footprints, such as quadratic or spline weights, to improve the $O(h^\beta)$ bound on non-smooth fields at the cost of slightly larger stencils; the paper's own error analysis indicates where those gains would appear.
  • Inference: the partial-observation recovery experiment suggests a general design pattern for sensor-to-field reconstruction: learn the missing topology first, then apply a pretrained field mapper, rather than trying to inpaint the field directly.
  • Inference: a multi-resolution variant that allocates fine grids only near sharp features would directly address the paper's stated limitation of cubic cost in 3D, because the encoder is local and resolution-independent.
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 / 6 minor

Summary. The paper introduces a grid-based encoding scheme for mesh-based point clouds: each point scatters bilinear (2D) or trilinear (3D) weights to the vertices of its enclosing grid cell, producing a topology grid Go and a response grid Gu. The authors pair this encoding with a UNet (E-UNet) that maps Go (plus auxiliary inputs) to an approximation of Gu, then reconstructs point-wise responses via interpolation. They provide an information-theoretic comparison of Go against binary and count-based encodings, an error analysis of the reconstruction, a time-complexity analysis, and experiments on five 2D benchmarks and one 3D benchmark, including data-efficiency, noise-robustness, and partial-observation recovery studies. Public code and datasets are promised.

Significance. If the main information-content claim were quantitatively substantiated, this would be a valuable practical contribution: a simple, parameter-free encoding that turns irregular meshes into standard grid tensors, usable with efficient CNN/FFT backbones, and backed by an open implementation. The empirical study is careful in several respects: five independent training runs, median reporting, error statistics in Appendix A, and comparisons against FNO, GeoFNO, GNOT, and GeomDeepONet. The principal weakness is that the headline quantitative claim—85% retained information versus about 35% for binary/count encodings—rests on an unmeasured constant η, and the abstract's 'consistently outperforms' wording is stronger than what the reported tables show.

major comments (4)
  1. [Section 2.1.2, Eqs. (10)-(11)] The central information-content claim depends on the unmeasured constant η in δ_eff² ≈ δ² + ηh². The authors state 'If the second derivatives of the topology field are O(0.1), we have η ≈ 0.01,' but they do not derive η from the actual point distributions, fit it to reconstruction error, or validate it on any of the five benchmarks. The sensitivity is material: for N=2000, r=128, δ=10^-3, the retained information is about 21.2 kbits (≈85%) at η=0.01, but drops to about 15.4 kbits (≈62%) at η=0.1 and to about 8.9 kbits (≈36%) at η=1, the last being comparable to the binary/count values. Because the '85% vs 35%' statement is the paper's theoretical motivation, this constant must either be measured, derived, or the claim must be rephrased as conditional on an unverified smoothness assumption.
  2. [Section 3.2.1, Table 1 and Section 3.2.2, Table 2] The abstract and Section 1 state that E-UNet 'consistently outperforms' Fourier- and transformer-based baselines. The reported medians do not support the unqualified version of that claim: in Table 1, GNOT achieves a lower median Lre2 on Circles (1.18E-02 vs E-UNet's 1.38E-02), and in Table 2, GNOT achieves a lower median Lre2 on the 3D Solid example (3.67E-02 vs E-UNet's 4.48E-02). The body text acknowledges these cases, but the high-level summary should be qualified to match the data.
  3. [Section 3.2.1, Table 1 footnote and Figure 6] The reported E-UNet results are selected as 'the resolution with the most accurate predictions' on the test set. This test-based model selection can inflate the reported performance relative to baselines that are trained at a single resolution, and it makes the comparison harder to interpret. The authors should either use a validation-based resolution selection or report the performance at each resolution without test-set selection.
  4. [Section 2.3, Eqs. (24)-(26) and Section 4] The O(h²) error bound is derived under the assumption that the normalized response H is in C² with bounded second derivatives. The authors themselves note in Section 4 that this smoothness condition is violated near the NACA shock and at the Maze walls, and Figure 7 shows that all models indeed have the largest errors in those regions. The bound therefore does not apply exactly in the regimes where the encoding is most stressed; this limitation should be stated at the point where the bound is introduced, and ideally supplemented by empirical convergence rates for the non-smooth cases.
minor comments (6)
  1. [Section 1] There is a typo in 'V oronoi tessellation-based interpolation'; it should read 'Voronoi tessellation-based interpolation'.
  2. [Section 2.3] The word 'Simliarly' should be 'Similarly'.
  3. [Section 2.4.1] The phrase 'neighboring gird points' should be 'neighboring grid points'.
  4. [Section 3.2.1] The word 'numnber' should be 'number'.
  5. [Table A1] The numeric formatting in Appendix A contains stray spaces (e.g., '1 .122E −02') and at least one malformed entry ('1.284E − 0 2'). Please reformat the table for consistency and readability.
  6. [Reference [48]] The reference title contains 'operater'; it should be 'operator'.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular dependency: benchmark claims are external and the only self-citation is non-load-bearing; hand-set eta affects the information estimate but is not fitted to the results.

full rationale

The derivation chain is self-contained rather than circular. The topology and response encodings are defined analytically in Section 2.1 by deterministic footprint aggregation (Eqs. 1-13), and the reconstruction is the corresponding bilinear interpolation (Eq. 14); the encodings are not defined in terms of the predicted outputs or the benchmark errors. The comparative results in Section 3 are evaluated against external datasets from prior work (FNO, GeoFNO, GNOT, GeomDeepONet), and no constant is fitted to the test errors; the network is trained to match encoded responses, not to match the claimed information percentages. The 85%-vs-35% information-content claim does rest on the hand-set value eta about 0.01 in Eq. (11), justified only by the assertion that the topology field's second derivatives are O(0.1), but this is an unmeasured input to a theoretical estimate, not a parameter fitted to the benchmark outcomes, so it is a correctness risk rather than circularity. Similarly, the O(h^2) error bound in Section 2.3 explicitly assumes twice-differentiable fields, and the authors acknowledge violations near NACA shocks and Maze walls (Section 4, Figure 7), which is a stated limitation, not a circular step. The one self-citation, reference [26], supports a peripheral claim about spectral bias of low-mode FNOs and is not load-bearing for the encoding's construction, the information estimate, or the benchmark results. Overall, the paper's central claims do not reduce to their inputs by construction.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claims rest on three unproven premises: a locally affine field assumption for the bilinear weights, a smoothness assumption for the O(h^2) error bound, and the ad hoc effective quantization model with η=0.01. The grid resolution r is selected per problem using test performance, which acts as a free choice that can inflate reported accuracy. No new physical entities are introduced.

free parameters (2)
  • η (eta) in effective quantization model = 0.01
    Hand-picked constant in δ_eff^2=δ^2+ηh^2 (Eq. 10-11); directly controls the claimed 85% information retention. No derivation or measurement is given.
  • Grid resolution r per problem = various (e.g., 128, 256, 64)
    Selected per benchmark and reported for the best-performing r (Table 1), which is a form of test-set hyperparameter selection.
assumptions (3)
  • domain assumption Within each grid cell, the underlying topology field varies locally affinely
    Used in Section 2.1.1 to justify the bilinear weights; standard interpolation assumption but not strictly true for arbitrary point clouds.
  • domain assumption The response field H is twice continuously differentiable with bounded second derivatives for the O(h^2) error bound
    Assumed in Section 2.3, case (1); violated near shocks and interfaces in NACA and Maze, as the authors acknowledge.
  • ad hoc to paper Effective spatial quantization follows δ_eff^2 ≈ δ^2 + η h^2
    Introduced in Eq. (10) with no derivation; underpins the information-content comparison.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Mappings in Mesh-based Simulations." pith.science (2026). https://pith.science/paper/6EZQOKNY

@misc{pith2026250612652,
  author       = {Pith},
  title        = {Pith review of: Learning Mappings in Mesh-based Simulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6EZQOKNY}},
  note         = {Machine review of arXiv:2506.12652}
}
read the original abstract

Many real-world physics and engineering problems arise in geometrically complex domains discretized by meshes for numerical simulations. The nodes of these potentially irregular meshes naturally form point clouds whose limited tractability poses significant challenges for learning mappings via machine learning models. To address this, we introduce a novel and parameter-free encoding scheme that aggregates footprints of points onto grid vertices and yields information-rich grid representations of the topology. Such structured representations are well-suited for standard convolution and FFT (Fast Fourier Transform) operations and enable efficient learning of mappings between encoded input-output pairs using Convolutional Neural Networks (CNNs). Specifically, we integrate our encoder with a uniquely designed UNet (E-UNet) and benchmark its performance against Fourier- and transformer-based models across diverse 2D and 3D problems where we analyze the performance in terms of predictive accuracy, data efficiency, and noise robustness. Furthermore, we highlight the versatility of our encoding scheme in various mapping tasks including recovering full point cloud responses from partial observations. Our proposed framework offers a practical alternative to both primitive and computationally intensive encoding schemes; supporting broad adoption in computational science applications involving mesh-based simulations.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

52 extracted references · 39 canonical work pages

  1. [1]

    Convolution Optimization, pages 85–126

    Albert Chun-Chen Liu and Oscar Ming Kin Law. Convolution Optimization, pages 85–126. IEEE, 2021

  2. [2]

    Convolutional neural networks for steady flow approxi- mation

    Xiaoxiao Guo, Wei Li, and Francesco Iorio. Convolutional neural networks for steady flow approxi- mation. In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pages 481–490, 2016

  3. [3]

    Deep learning for real-time aerodynamic evaluations of arbitrary vehicle shapes

    Sam Jacob Jacob, Markus Mrosek, Carsten Othmer, and Harald K ¨ostler. Deep learning for real-time aerodynamic evaluations of arbitrary vehicle shapes. arXiv preprint arXiv:2108.05798, 2021

  4. [4]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18 , pages 234–241. Springer, 2015

  5. [5]

    Surfnet: Super-resolution of turbulent flows with transfer learning using small datasets

    Octavi Obiols-Sales, Abhinav Vishnu, Nicholas P Malaya, and Aparna Chandramowlishwaran. Surfnet: Super-resolution of turbulent flows with transfer learning using small datasets. In 2021 30th International Conference on Parallel Architectures and Compilation Techniques (PACT), pages 331–

  6. [6]

    Computa- tionally effective estimation of supersonic flow field around airfoils using sparse convolutional neural network

    Ming-Yu Wu, Jiang-Zhou Peng, Zhi-Ming Qiu, Zhi-Hua Chen, Yu-Bai Li, and Wei-Tao Wu. Computa- tionally effective estimation of supersonic flow field around airfoils using sparse convolutional neural network. Fluid Dynamics Research, 55(3):035504, 2023

  7. [7]

    Sparse convo- lutional neural networks

    Baoyuan Liu, Min Wang, Hassan Foroosh, Marshall Tappen, and Marianna Pensky. Sparse convo- lutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 806–814, 2015

  8. [8]

    UCNN: A Convolutional Strategy on Unstructured Mesh

    Mengfei Xu, Shufang Song, Xuxiang Sun, and Weiwei Zhang. Ucnn: A convolutional strategy on unstructured mesh. arXiv preprint arXiv:2101.05207, 2021

Show all 52 references
  1. [9]

    Deep parametric continuous convolutional neural networks

    Shenlong Wang, Simon Suo, Wei-Chiu Ma, Andrei Pokrovsky, and Raquel Urtasun. Deep parametric continuous convolutional neural networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 2589–2597, 2018

  2. [10]

    U-net-based surrogate model for evaluation of microfluidic channels

    Tuyen Quang Le, Pao-Hsiung Chiu, and Chinchun Ooi. U-net-based surrogate model for evaluation of microfluidic channels. International Journal of Computational Methods, 19(07):2141018, 2022

  3. [11]

    A finite element-convolutional neural network model (fe- cnn) for stress field analysis around arbitrary inclusions

    Mohammad Rezasefat and James D Hogan. A finite element-convolutional neural network model (fe- cnn) for stress field analysis around arbitrary inclusions. Machine Learning: Science and Technology, 4(4):045052, 2023

  4. [12]

    Geometric compression for interactive transmission

    Olivier Devillers and P-M Gandoin. Geometric compression for interactive transmission. In Proceed- ings Visualization 2000. VIS 2000 (Cat. No. 00CH37145), pages 319–326. IEEE, 2000

  5. [13]

    Learning collision situation to convolutional neural network using collision grid map based on probability scheme

    Jun Hyeong Jo and Chang-bae Moon. Learning collision situation to convolutional neural network using collision grid map based on probability scheme. Applied Sciences, 10(2):617, 2020

  6. [14]

    Subdle: Identification of substructures in cosmological simulations with deep learning-an image segmentation approach to substructure finding

    Michela Esposito, Stefano Borgani, and Giuseppe Murante. Subdle: Identification of substructures in cosmological simulations with deep learning-an image segmentation approach to substructure finding. Astronomy & Astrophysics, 689:A33, 2024. 30

  7. [15]

    Factorized implicit global convolution for automotive computational fluid dynamics prediction

    Chris Choy, Alexey Kamenev, Jean Kossaifi, Max Rietmann, Jan Kautz, and Kamyar Azizzadenesheli. Factorized implicit global convolution for automotive computational fluid dynamics prediction. arXiv preprint arXiv:2502.04317, 2025

  8. [16]

    Dgpolarnet: dynamic graph convolution network for lidar point cloud semantic segmentation on polar bev

    Wei Song, Zhen Liu, Ying Guo, Su Sun, Guidong Zu, and Maozhen Li. Dgpolarnet: dynamic graph convolution network for lidar point cloud semantic segmentation on polar bev. Remote Sensing , 14(15):3825, 2022

  9. [17]

    Global field reconstruction from sparse sensors with voronoi tessellation-assisted deep learning

    Kai Fukami, Romit Maulik, Nesar Ramachandra, Koji Fukagata, and Kunihiko Taira. Global field reconstruction from sparse sensors with voronoi tessellation-assisted deep learning. Nature Machine Intelligence, 3(11):945–951, 2021

  10. [18]

    Efficient deep data assimilation with sparse observations and time-varying sensors

    Sibo Cheng, Che Liu, Yike Guo, and Rossella Arcucci. Efficient deep data assimilation with sparse observations and time-varying sensors. Journal of Computational Physics, 496:112581, 2024

  11. [19]

    Applying convolutional neural networks to data on unstructured meshes with space-filling curves

    Claire E Heaney, Yuling Li, Omar K Matar, and Christopher C Pain. Applying convolutional neural networks to data on unstructured meshes with space-filling curves. Neural Networks, 175:106198, 2024

  12. [20]

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

    Han Gao, Luning Sun, and Jian-Xun Wang. Phygeonet: Physics-informed geometry-adaptive convo- lutional neural networks for solving parameterized steady-state pdes on irregular domain. Journal of Computational Physics, 428:110079, 2021

  13. [21]

    Mmgp: a mesh morphing gaussian process-based machine learning method for regression of physical problems under nonparametrized geometrical vari- ability

    Fabien Casenave, Brian Staber, and Xavier Roynard. Mmgp: a mesh morphing gaussian process-based machine learning method for regression of physical problems under nonparametrized geometrical vari- ability. Advances in Neural Information Processing Systems, 36:43972–43999, 2023

  14. [22]

    Fourier neural operator for parametric partial differential equations

    Zongyi Li, Nikola Kovachki, Kamyar Azizzadenesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier neural operator for parametric partial differential equations. arXiv preprint arXiv:2010.08895, 2020

  15. [23]

    Fourier neural operator with learned deformations for pdes on general geometries

    Zongyi Li, Daniel Zhengyu Huang, Burigede Liu, and Anima Anandkumar. Fourier neural operator with learned deformations for pdes on general geometries. Journal of Machine Learning Research , 24(388):1–26, 2023

  16. [24]

    Beyond regular grids: Fourier-based neural operators on arbitrary domains

    Levi Lingsch, Mike Y Michelis, Emmanuel De B´ezenac, Sirani M Perera, Robert K Katzschmann, and Siddhartha Mishra. Beyond regular grids: Fourier-based neural operators on arbitrary domains. arXiv preprint arXiv:2305.19663, 2023

  17. [25]

    Mitigating spectral bias in neural operators via high-frequency scaling for physical systems

    Siavash Khodakarami, Vivek Oommen, Aniruddha Bora, and George Em Karniadakis. Mitigating spectral bias in neural operators via high-frequency scaling for physical systems. arXiv preprint arXiv:2503.13695, 2025

  18. [26]

    Parametric encoding with attention and convolution mitigate spectral bias of neural partial differential equation solvers

    Mehdi Shishehbor, Shirin Hosseinmardi, and Ramin Bostanabad. Parametric encoding with attention and convolution mitigate spectral bias of neural partial differential equation solvers. Structural and Multidisciplinary Optimization, 67(7):128, 2024

  19. [27]

    Bubbleml: A multiphase multiphysics dataset and benchmarks for machine learning

    Sheikh Md Shakeel Hassan, Arthur Feeney, Akash Dhruv, Jihoon Kim, Youngjoon Suh, Jaiyoung Ryu, Yoonjin Won, and Aparna Chandramowlishwaran. Bubbleml: A multiphase multiphysics dataset and benchmarks for machine learning. Advances in Neural Information Processing Systems, 36:41...

  20. [28]

    Inductive representation learning on large graphs

    Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. Advances in neural information processing systems, 30, 2017

  21. [29]

    Learning mesh-based simulation with graph networks

    Tobias Pfaff, Meire Fortunato, Alvaro Sanchez-Gonzalez, and Peter Battaglia. Learning mesh-based simulation with graph networks. In International conference on learning representations, 2020

  22. [30]

    Convolution, aggrega- tion and attention based deep neural networks for accelerating simulations in mechanics

    Saurabh Deshpande, Ra ´ul I Sosa, St´ephane PA Bordas, and Jakub Lengiewicz. Convolution, aggrega- tion and attention based deep neural networks for accelerating simulations in mechanics. Frontiers in Materials, 10:1128954, 2023

  23. [31]

    Graph neural networks and implicit neural representation for near- optimal topology prediction over irregular design domains

    Minsik Seo and Seungjae Min. Graph neural networks and implicit neural representation for near- optimal topology prediction over irregular design domains. Engineering Applications of Artificial Intelligence, 123:106284, 2023

  24. [32]

    Graph neural network-based surrogate models for finite element analysis

    Meduri Venkata Shivaditya, Jos ´e Alves, Francesca Bugiotti, and Frederic Magoules. Graph neural network-based surrogate models for finite element analysis. In 2022 21st International Symposium on Distributed Computing and Applications for Business Engineering and Science (DCA...

  25. [33]

    A graph-based probabilistic geometric deep learning framework with online enforcement of physical constraints to predict the criticality of defects in porous materials

    Vasilis Krokos, St ´ephane PA Bordas, and Pierre Kerfriden. A graph-based probabilistic geometric deep learning framework with online enforcement of physical constraints to predict the criticality of defects in porous materials. International Journal of Solids and Structures, ...

  26. [34]

    Understanding gnn computational graph: A coordinated computation, io, and memory perspective

    Hengrui Zhang, Zhongming Yu, Guohao Dai, Guyue Huang, Yufei Ding, Yuan Xie, and Yu Wang. Understanding gnn computational graph: A coordinated computation, io, and memory perspective. Proceedings of Machine Learning and Systems, 4:467–484, 2022

  27. [35]

    Pointnet: Deep learning on point sets for 3d classification and segmentation

    Charles R Qi, Hao Su, Kaichun Mo, and Leonidas J Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 652–660, 2017

  28. [36]

    Pointnet++: Deep hierarchical feature learning on point sets in a metric space

    Charles Ruizhongtai Qi, Li Yi, Hao Su, and Leonidas J Guibas. Pointnet++: Deep hierarchical feature learning on point sets in a metric space. Advances in neural information processing systems, 30, 2017

  29. [37]

    A point-cloud deep learning framework for predic- tion of fluid flow fields on irregular geometries

    Ali Kashefi, Davis Rempe, and Leonidas J Guibas. A point-cloud deep learning framework for predic- tion of fluid flow fields on irregular geometries. Physics of Fluids, 33(2), 2021

  30. [38]

    Kolmogorov–arnold pointnet: Deep learning for prediction of fluid fields on irregular geometries

    Ali Kashefi. Kolmogorov–arnold pointnet: Deep learning for prediction of fluid fields on irregular geometries. Computer Methods in Applied Mechanics and Engineering, 439:117888, 2025

  31. [39]

    Physics-informed pointnet: On how many irregu- lar geometries can it solve an inverse problem simultaneously? application to linear elasticity

    Ali Kashefi, Leonidas J Guibas, and Tapan Mukerji. Physics-informed pointnet: On how many irregu- lar geometries can it solve an inverse problem simultaneously? application to linear elasticity. Journal of Machine Learning for Modeling and Computing, 4(4), 2023

  32. [40]

    Tree species classification using ground-based lidar data by various point cloud deep learning methods

    Bingjie Liu, Huaguo Huang, Yong Su, Shuxin Chen, Zengyuan Li, Erxue Chen, and Xin Tian. Tree species classification using ground-based lidar data by various point cloud deep learning methods. Remote Sensing, 14(22):5733, 2022

  33. [41]

    Gnot: A general neural operator transformer for operator learning

    Zhongkai Hao, Zhengyi Wang, Hang Su, Chengyang Ying, Yinpeng Dong, Songming Liu, Ze Cheng, Jian Song, and Jun Zhu. Gnot: A general neural operator transformer for operator learning. In Inter- national Conference on Machine Learning, pages 12556–12569. PMLR, 2023. 32

  34. [42]

    Pinnsformer: A transformer-based framework for physics-informed neural networks

    Zhiyuan Zhao, Xueying Ding, and B Aditya Prakash. Pinnsformer: A transformer-based framework for physics-informed neural networks. arXiv preprint arXiv:2307.11833, 2023

  35. [43]

    Transformer for partial differential equations’ operator learning

    Zijie Li, Kazem Meidani, and Amir Barati Farimani. Transformer for partial differential equations’ operator learning. arXiv preprint arXiv:2205.13671, 2022

  36. [44]

    Machine learning for modelling unstructured grid data in computational physics: a review

    Sibo Cheng, Marc Bocquet, Weiping Ding, Tobias Sebastian Finn, Rui Fu, Jinlong Fu, Yike Guo, Eleda Johnson, Siyi Li, Che Liu, et al. Machine learning for modelling unstructured grid data in computational physics: a review. arXiv preprint arXiv:2502.09346, 2025

  37. [45]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. Attention is all you need. In I. Guyon, U. V on Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett, editors, Advances in Neural...

  38. [46]

    Text2pde: Latent diffusion models for accessible physics simulation

    Anthony Zhou, Zijie Li, Michael Schneier, John R Buchanan Jr, and Amir Barati Farimani. Text2pde: Latent diffusion models for accessible physics simulation. arXiv preprint arXiv:2410.01153, 2024

  39. [47]

    Real-time single image and video super-resolution using an efficient sub- pixel convolutional neural network

    Wenzhe Shi, Jose Caballero, Ferenc Husz ´ar, Johannes Totz, Andrew P Aitken, Rob Bishop, Daniel Rueckert, and Zehan Wang. Real-time single image and video super-resolution using an efficient sub- pixel convolutional neural network. In Proceedings of the IEEE conference on comp...

  40. [48]

    An isotropic 3 × 3 image gradient operater

    Irwin Sobel. An isotropic 3 × 3 image gradient operater. Machine vision for three-dimensional scenes, pages 376–379, 1990

  41. [49]

    Junyan He, Seid Koric, Diab Abueidda, Ali Najafi, and Iwona Jasiuk. Geom-deeponet: A point-cloud- based deep operator network for field predictions on 3d parameterized geometries.Computer Methods in Applied Mechanics and Engineering, 429:117130, 2024

  42. [50]

    The well: a large-scale collection of diverse physics simulations for machine learning

    Ruben Ohana, Michael McCabe, Lucas Meyer, Rudy Morel, Fruzsina Agocs, Miguel Beneitez, Marsha Berger, Blakesly Burkhart, Stuart Dalziel, Drummond Fielding, et al. The well: a large-scale collection of diverse physics simulations for machine learning. Advances in Neural Informa...

  43. [51]

    Clawpack: building an open source ecosystem for solving hyperbolic pdes

    Kyle T Mandli, Aron J Ahmadia, Marsha Berger, Donna Calhoun, David L George, Yiannis Had- jimichael, David I Ketcheson, Grady I Lemoine, and Randall J LeVeque. Clawpack: building an open source ecosystem for solving hyperbolic pdes. PeerJ Computer Science, 2:e68, 2016

  44. [52]

    Real-time inference and extrapolation via a diffusion-inspired temporal transformer operator (ditto)

    Oded Ovadia, Vivek Oommen, Adar Kahana, Ahmad Peyvan, Eli Turkel, and George Em Karniadakis. Real-time inference and extrapolation via a diffusion-inspired temporal transformer operator (ditto). arXiv preprint arXiv:2307.09072, 2023. 33

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.