Pith. sign in

REVIEW 3 major objections 5 minor 44 references

Graph Neural Network Surrogates for Contacting Deformable Bodies with Necessary and Sufficient Contact Detection

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read By adding a contact-loss term computed from a continuous collision detection algorithm that applies both necessary and sufficient conditions, this paper shows that graph neural network surrogates for soft deformable bodies generalize…

desk verdict A useful proof-of-concept whose central claim—that sufficient-condition CCD drives the regularization—is not isolated by the experiments; the end-step penalty and missing necessary-only baseline undercut the attribution. read the letter →

arxiv 2507.13459 v1 pith:23DAVYSW submitted 2025-07-17 cs.CE cs.AIcs.LGcs.NAmath.NA

Vijay K. Dubey (1) , Collin E. Haese (1) , Osman Gültekin (1) , David Dalton (2) , Manuel K. Rausch (1) , Jan N. Fuhg (1) ((1) The University of Texas at Austin , (2) University of Glasgow) This is my paper
classification cs.CEcs.AIcs.LGcs.NAmath.NA MSC 74M1568T0774S05
keywords graphneuralnetworkscontactmechanicscontinuouscollisiondetectiondeformablebodiessurrogatemodelingsofttissueregularization
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 tries to establish that a graph neural network can learn the contact of two soft deformable bodies, not just rigid or rigid-soft contact, when training includes a contact loss computed by a continuous collision detection algorithm that applies both necessary and sufficient conditions. On two benchmarks, inflating membranes with varying reference geometry and a closing bioprosthetic aortic valve, the authors find that adding this contact term regularizes training and improves generalization to unseen cases. They report up to a thousand-fold inference speedup over the finite-element solver, while acknowledging a major increase in training cost. The practical claim is that precise contact awareness in the loss, rather than larger networks alone, is what lets the surrogate respect contact on data it has not seen.

What carries the argument

The load-bearing object is the continuous collision detection routine turned into a differentiable loss. For each time step, axis-aligned bounding boxes cull candidate triangle pairs, neighboring triangles are removed, and vertex-face and edge-edge coplanarity filters are applied; these are the necessary conditions. A cubic polynomial is then formed for each candidate pair and its roots are found within the time step, followed by linear-inequality tests for vertex-face pairs and segment-intersection tests for edge-edge pairs, which are the sufficient conditions. The maximum of the six vertex-face and nine edge-edge responses for a confirmed collision gives the contact response, normalized by the length scale, and its mean absolute value over the batch is the contact loss added to the dynamic mean-squared-error acceleration loss.

What would settle it

Construct a synthetic one-step trajectory of two triangles that is known to interpenetrate and then separate completely, with the crossing occurring strictly between the start and end times. If the continuous collision detection reports no collision and zero contact response, then the claimed sufficient condition is not sufficient, and the regularization benefit attributed to the contact loss would be called into question for exactly the tunneling cases the method says it handles.

Watch

Extended reading notes

Core claim

The central claim is that a graph neural network trained with an additional contact loss, derived from a continuous collision detection routine with both necessary and sufficient conditions, generalizes better for deformable soft-body contact than the same network trained only on acceleration errors. The collision detection identifies triangle pairs that become coplanar during a time step using bounding-box culling and vertex-face and edge-edge filters as necessary conditions, then confirms actual collision by finding the roots of cubic polynomials and checking vertex-in-face and edge-segment intersection conditions as sufficient conditions. The contact response is the maximum orthogonal distance or edge midpoint distance at the end of the step for confirmed colliding pairs, normalized by a problem length scale and used as an unsupervised penalty. Across two benchmarks, the contact-loss-trained network shifts the distribution of contact violations downward on validation and test sets, and the effect appears for both simple contact at a known plane and complex contact at varying planes and element angles.

Load-bearing premise

The penalty for penetration is estimated from distances measured at the end of the time step for pairs found by assuming each vertex moves along a straight line at a constant speed; if two elements pass through each other and separate inside one step in a way this linear approximation misses, the contact loss will not represent the real contact violation.

Editorial extensions

If this is right

  • Adding the contact loss reduces penetration on unseen test cases: contact-loss distributions shift to lower values for the contact-loss-trained network than for dynamic-only networks on both benchmarks.
  • The benefits appear for symmetric contact at a known plane (aortic valve) and asymmetric contact at varying planes and element normal angles (inflating membranes with varying geometry).
  • The framework handles varying reference geometries even when the shape parameters are not provided as input features.
  • Inference is about six times faster than the finite-element solver on CPU and more than a thousand times faster on GPU, with a per-epoch training cost roughly 30 to 40 times higher when the contact loss is active.

Reading between the lines

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

  • The paper does not test this, but the contact loss is unsupervised and could be applied to self-contact or multi-body contact scenarios where labeled contact data is unavailable, since it is computed purely from predicted mesh positions.
  • One untested route to reducing the training penalty is to pretrain a dynamic-only network and fine-tune with the contact loss for a small number of epochs; the paper's two-stage schedule already hints this might preserve most of the generalization gain.
  • The observed failure mode of accumulated rollout error suggests a testable combination of contact loss with noise-injected training to stabilize autoregressive inference.
  • The comparison with a larger dynamic-only network hints that contact loss may substitute for model capacity in some regimes; a proper ablation varying both capacity and contact-loss weight would quantify that trade-off.
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

3 major / 5 minor

Summary. The manuscript proposes a graph neural network surrogate for predicting the deformation of contacting soft bodies. The architecture is an encode-process-decode GNN with mesh-space and world-space edges, and training augments the standard acceleration mean-squared-error loss with an unsupervised contact loss computed from a continuous collision detection (CCD) algorithm. The CCD pipeline applies bounding-box culling, necessary vertex-face and edge-edge coplanarity filters, and sufficient-condition polynomial-root checks; the contact response is taken as the maximum end-of-step distance among detected pairs. The method is tested on two FEBio-generated benchmarks: inflating membranes with varying reference geometry and a bioprosthetic aortic valve model, comparing small and large dynamic-only networks with a small network trained with the extra contact loss. The paper reports reduced contact loss on held-out data, marginal position-error improvements, large training-cost increases, and up to roughly two-thousand-fold inference speedups, and it concludes that the contact loss has a regularizing effect that improves generalization.

Significance. If the regularization claim holds, the paper makes a useful contribution: it is among the first to integrate a sufficient-condition continuous collision detection algorithm into GNN training for soft-body contact, going beyond the necessary-only filters used in prior work such as Zhu et al. The paper is also commendably transparent about computational costs, hardware dependence, and the trade-off between training time and generalization, and it promises code and data release. However, the empirical support for the central claim is currently thinner than the abstract suggests: the test sets are very small, the comparisons are single runs with no seed variation, and no ablation isolates the sufficient-condition component from a generic contact penalty. The contribution is plausible and worth publishing after the evidence is strengthened or the claims are appropriately tempered.

major comments (3)
  1. [§2.2 and §2.3, Eqs. (10)-(16), Fig. 3] The contact response in Eqs. (10)-(13) is evaluated only at the final time t+δt, even though the continuous collision detection finds roots inside the interval (t, t+δt). For a pair of elements that interpenetrate and separate within the same time step, the penalty is the final distance rather than the maximum penetration, and this final distance can be near zero despite a large tunneling violation. This is in tension with the introduction's claim that the CCD handles elements that 'completely go through each other during a time step' and with the description of r_Ii as representative of the extent of the contact violation. Please either compute a maximum-over-time or time-integrated contact response, or provide evidence that the end-of-step response is sufficient to produce the claimed regularization effect.
  2. [§4.2 and §4.3] The central novelty is the sufficient-condition CCD, but the comparison between S-D and S-DC can only attribute the observed improvement to the presence of a contact term, not to sufficiency. The manuscript itself states in §4.2 that ablation studies could not be performed, so the reader cannot distinguish the proposed sufficient-condition contribution from, for example, an end-of-step penalty or a necessary-only filter. To support the abstract and §1 claims, add at least one baseline with a necessary-only collision filter or an end-of-step distance penalty; if the computational cost makes this infeasible, the claims in the abstract and conclusions should be weakened accordingly.
  3. [§3.2, §4.2, Figs. 6-8] The generalization claims rest on 25 simulations per benchmark split as 8:1:1, leaving only two or three held-out simulations in the validation and test sets, and on a single training run per network configuration. The reported S-DC benefits, particularly the small position-error differences in Figs. 6 and 7, are therefore fragile. Please provide seed-averaged results with confidence intervals, or a cross-validation scheme, before asserting that the contact loss leads to better generalization.
minor comments (5)
  1. [§4.2] The phrase 'i.e., (w_d,w_c),(0,0) after the thousandth epoch' appears to contain a typo; it should presumably read (w_d,w_c)=(1,0) before epoch 1000 and (w_d,w_c)=(1,w_c) afterward.
  2. [Appendix A] The sentence about options to improve numerical stability in the Adam optimizer is duplicated; please remove the repeated sentence.
  3. [Table A.5] The caption uses both D(·) and D(·) for dimension and depth, which is easy to confuse; please use distinct symbols for these two quantities.
  4. [§3.1.1, Eq. (18)] The notation z[u/l] and the undefined variables x and y make the undulation formula difficult to parse; please define the in-plane coordinates and the domain explicitly.
  5. [Fig. 6] The x-axis tick labels in Fig. 6 do not match the epoch values discussed in the text (e.g., 100 and 500); please make the axis labels consistent.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the contact-loss regularization claim is evaluated against held-out FEM data and does not reduce to the training loss by construction.

full rationale

The paper's central comparison (S-D vs. S-DC) is a controlled comparison with identical initialization, and the held-out test set is not used to select the contact-loss weight w_c; Appendix A states that 'the weight (w_c) is chosen using the information in the training set.' The contact loss L_c is computed from the GNN-predicted trajectory only (Eqs. 10-16), so it is a physics-based regularizer rather than a relabeled output of the FEM target; the supervised learning signal remains the dynamic loss of Eq. (14) against FEBio ground truth. The reported evaluation metrics include the independent displacement error L_p (Eq. 21), so the claimed generalization benefit is not a tautology of minimizing L_c on the training set. The inference speedups in Table 3 are wall-clock comparisons against an external FEM solver, not a quantity derived from the network's own predictions. The self-citations [3] and [27] support the surrogate-model framing and the encode-process-decode/skew-symmetric message-passing architecture; these are methodological scaffolding, not the load-bearing premise that contact loss improves generalization. The absence of a necessary-only ablation and the fact that Eqs. (10)-(11) evaluate penetration depth only at the end of the time step are genuine scientific limitations that weaken attribution of observed improvements to the sufficient-condition continuous collision detection, but they are not instances of a prediction reducing to its inputs by construction, a fitted parameter being renamed as a prediction, or an argument that depends on a self-citation chain.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central empirical claim depends on standard ML hyperparameters, the FEM ground truth, and the correctness of the contact detection and contact-response formulation. No new physical entities are introduced; the main adjustable quantities are training hyperparameters and problem-dependent geometric scales such as collision radius and contact-loss weight.

free parameters (4)
  • Contact loss weight w_c = not reported numerically
    Chosen to bring L_c to the scale of L_d at epoch 1000 and used in Eq. (17) for all -DC networks.
  • Collision radius R = 1.45e-1 and 6.32e-4 for the two problems
    Used to build world-space edges (Eq. 2) and determines which triangle pairs can be checked; values listed in Table 2.
  • Network architecture and training hyperparameters = see Table A.5 and Figure A.10
    Widths, depths, message-passing rounds, learning-rate schedule, weight decay, and batch sizes are chosen by hand or by GPU memory constraints and affect the reported comparisons.
  • Number of training epochs = 3000 for varying geometry, 9400 for aortic valve
    Chosen as the last common checkpoint to compare networks fairly and limited by computational cost (Section 4.2, Appendix A).
assumptions (5)
  • domain assumption Vertex trajectories between t and t+δt are linearly interpolated using uniform average velocity.
    Used in Section 2.2 to build continuous collision detection; questionable for high acceleration or large nonlinear deformations.
  • domain assumption Provot's sufficient-condition algorithm [35], with the modified edge-edge segment intersection check, correctly identifies all penetrating triangle pairs under the linearized trajectory.
    Relied on for the 'sufficient' part of contact detection; not formally verified in this paper.
  • ad hoc to paper The contact response defined by max orthogonal distance at t+δt (Eqs. 10-13) is an adequate penalty for collision severity.
    The contact loss is built on this definition; collisions that resolve within the time step may yield zero or misleading response.
  • domain assumption FEBio FEM solutions are treated as ground truth for training and evaluation.
    Dataset generation uses FEBio (Section 3), and all errors are measured against it.
  • domain assumption Quadrilateral shell meshes can be converted to triangular shell meshes without changing the contact physics being learned.
    Appendix B states the conversion is done to use the continuous collision detection algorithm.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph Neural Network Surrogates for Contacting Deformable Bodies with Necessary and Sufficient Contact Detection." pith.science (2026). https://pith.science/paper/23DAVYSW

@misc{pith2026250713459,
  author       = {Pith},
  title        = {Pith review of: Graph Neural Network Surrogates for Contacting Deformable Bodies with Necessary and Sufficient Contact Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/23DAVYSW}},
  note         = {Machine review of arXiv:2507.13459}
}
read the original abstract

Surrogate models for the rapid inference of nonlinear boundary value problems in mechanics are helpful in a broad range of engineering applications. However, effective surrogate modeling of applications involving the contact of deformable bodies, especially in the context of varying geometries, is still an open issue. In particular, existing methods are confined to rigid body contact or, at best, contact between rigid and soft objects with well-defined contact planes. Furthermore, they employ contact or collision detection filters that serve as a rapid test but use only the necessary and not sufficient conditions for detection. In this work, we present a graph neural network architecture that utilizes continuous collision detection and, for the first time, incorporates sufficient conditions designed for contact between soft deformable bodies. We test its performance on two benchmarks, including a problem in soft tissue mechanics of predicting the closed state of a bioprosthetic aortic valve. We find a regularizing effect on adding additional contact terms to the loss function, leading to better generalization of the network. These benefits hold for simple contact at similar planes and element normal angles, and complex contact at differing planes and element normal angles. We also demonstrate that the framework can handle varying reference geometries. However, such benefits come with high computational costs during training, resulting in a trade-off that may not always be favorable. We quantify the training cost and the resulting inference speedups on various hardware architectures. Importantly, our graph neural network implementation results in up to a thousand-fold speedup for our benchmark problems at inference.

Figures

Figures reproduced from arXiv: 2507.13459 by the authors.

Figure 1
Figure 1. Schematic of the framework used in this work. [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Changes due to modification of edge-edge su [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Computed contact response (Equation (13)) using the implemented contact algorithm. [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Problems used for the numerical experiments. [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Total (L) over training set for (a) varying geometry and (b) aortic valve problems. Note that the losses are average across all batches at the respective epoch. 4.3. Impact of Contact Loss Inclusion on Learning 4.3.1. During the Training Process During the training pro…
Figure 6
Figure 6. Figure 6: Quartile boundaries (box and whisker) and mean values (circles) of position loss, contact loss, and position error when inference is done [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Quartile boundaries (box and whisker) and mean values (circles) of position loss, contact loss, and position error when inference is done [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Average position error (Equation (23)) and average accumulated position error (Equation (25)), both averaged across nodes and graphs [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Position error (Equation (22)) for a randomly selected simulation from the test set for each of the problems at the penultimate time step. [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

44 extracted references · 23 canonical work pages

  1. [1]

    Zienkiewicz, R

    O. Zienkiewicz, R. Taylor, D. Fox, General problems in solid mechanics and nonlinearity, in: The Finite Element Method for Solid and Structural Mechanics, Elsevier, 2014, pp. 1–20.doi:10.1016/B978-1-85617-634-7.00001-6. 17 Varying Geometrya. Epochs Epochs Epochs Epochs Epochs Epochs Case S-D Case L-DCase S-DC Total Training Loss Dynamic Training Loss Cont...

    work page doi:10.1016/b978-1-85617-634-7.00001-6 2014

  2. [2]

    Wriggers, Computational Contact Mechanics, Springer Berlin Heidelberg, Berlin, Heidelberg, 2006.doi:10.1007/ 978-3-540-32609-0

    P. Wriggers, Computational Contact Mechanics, Springer Berlin Heidelberg, Berlin, Heidelberg, 2006.doi:10.1007/ 978-3-540-32609-0. URLhttps://doi.org/10.1007/978-3-540-32609-0

    doi:10.1007/978-3-540-32609-0 2006

  3. [3]

    J. N. Fuhg, A. Fau, U. Nackenhorst, State-of-the-art and comparative review of adaptive sampling methods for kriging, Archives of Compu- tational Methods in Engineering 28 (2021) 2689–2747.doi:10.1007/s11831-020-09474-6

    work page doi:10.1007/s11831-020-09474-6 2021

  4. [4]

    Kumar, D

    S. Kumar, D. M. Kochmann, What machine learning can do for computational solid mechanics, in: F. Aldakheel, B. Hudobivnik, M. Soleimani, H. Wessels, C. Weißenfels, M. Marino (Eds.), Current Trends and Open Problems in Computational Mechanics, Springer International Publishing, Cham, 2022, pp. 275–285.doi:10.1007/978-3-030-87312-7\_27. URLhttps://doi.org/1...

    work page doi:10.1007/978-3-030-87312-7 2022

  5. [5]

    Goodbrake, S

    C. Goodbrake, S. Motiwale, M. S. Sacks, A neural network finite element method for contact mechanics, Computer Methods in Applied Mechanics and Engineering 419 (2024) 116671.doi:10.1016/j.cma.2023.116671

    work page arXiv 2024

  6. [6]

    Padula, M

    G. Padula, M. Girfoglio, G. Rozza, A brief review of reduced order models using intrusive and non-intrusive techniques, PAMM 24 (4) (2024) e202400210.doi:10.1002/pamm.202400210. 18 Training Set Validation Set Test Set Aortic Valve Varying Geometry Figure B.12: Pairwise scatter plot of all parameters used to generate the dataset labelled by the set they be...

    doi:10.1002/pamm.202400210 2024

  7. [7]

    Ghattas, K

    O. Ghattas, K. Willcox, Learning physics-based models from data: perspectives from inverse problems and model reduction, Acta Numerica 30 (2021) 445–554

    work page 2021

  8. [8]

    Amsallem, M

    D. Amsallem, M. J. Zahr, C. Farhat, Nonlinear model order reduction based on local reduced-order bases, International Journal for Numerical Methods in Engineering 92 (10) (2012) 891–916.doi:10.1002/nme.4371

    doi:10.1002/nme.4371 2012

Show all 44 references
  1. [9]

    Krenciszek, R

    J. Krenciszek, R. Pinnau, Model reduction of contact problems in elasticity: proper orthogonal decomposition for variational inequalities, in: Progress in Industrial Mathematics at ECMI 2012, Springer, 2014, pp. 277–284

    2012

  2. [10]

    Balajewicz, D

    M. Balajewicz, D. Amsallem, C. Farhat, Projection-based model reduction for contact problems, International Journal for Numerical Methods in Engineering 106 (8) (2016) 644–663, _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/nme.5135.doi:10.1002/nme.5135. URLhttps://o...

    2016 doi

  3. [11]

    Fauque, I

    J. Fauque, I. Ramière, D. Ryckelynck, Hybrid hyper-reduced modeling for contact mechanics problems, International Journal for Numerical Methods in Engineering 115 (1) (2018) 117–139, _eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1002/nme.5798.doi:10.1002/nme. 5798. URLht...

    2018 doi

  4. [12]

    Benaceur, A

    A. Benaceur, A. Ern, V . Ehrlacher, A reduced basis method for parametrized variational inequalities applied to contact mechanics, Interna- tional Journal for Numerical Methods in Engineering 121 (6) (2020) 1170–1197.doi:10.1002/nme.6261

    2020 doi

  5. [13]

    G. S. Pawar, S. S. Kulkarni, Efficient parametric model order reduction in contact mechanics, Computers & Structures 315 (2025) 107838. doi:10.1016/j.compstruc.2025.107838

    2025

  6. [14]

    K. S. Kollepara, J. M. Navarro-Jiménez, Y . Le Guennec, L. Silva, J. V . Aguado, On the limitations of low-rank approximations in contact mechanics problems, International Journal for Numerical Methods in Engineering 124 (1) (2023) 217–234.doi:10.1002/nme.7118

    2023 doi

  7. [15]

    Sahin, M

    T. Sahin, M. von Danwitz, A. Popp, Solving forward and inverse problems of contact mechanics using physics-informed neural networks, Advanced Modeling and Simulation in Engineering Sciences 11 (1) (2024) 11.doi:10.1186/s40323-024-00265-3

    2024 doi

  8. [16]

    Sahin, D

    T. Sahin, D. Wolff, A. Popp, Physics-informed neural networks for solving contact problems in three dimensions, arXiv preprint 19 arXiv:2412.09022 (2024).doi:10.48550/arXiv.2412.09022

    2024 doi

  9. [17]

    Eskinazi, B

    I. Eskinazi, B. J. Fregly, Surrogate modeling of deformable joint contact using artificial neural networks, Medical engineering & physics 37 (9) (2015) 885–891.doi:10.1016/j.medengphy.2015.06.006

    2015 doi

  10. [18]

    Kalliorinne, R

    K. Kalliorinne, R. Larsson, F. Pérez-Ràfols, M. Liwicki, A. Almqvist, Artificial neural network architecture for prediction of contact mechan- ical response, Frontiers in Mechanical Engineering 6 (2021) 579825.doi:10.3389/fmech.2020.579825

    2021

  11. [19]

    Houssein, An artificial neural network for predicting the roles of contacting bodies in computational contact mechanics (2025)

    H. Houssein, An artificial neural network for predicting the roles of contacting bodies in computational contact mechanics (2025)

    2025

  12. [20]

    Sahin, J

    T. Sahin, J. Bonari, S. Brandstaeter, A. Popp, Data-driven surrogate modeling techniques to predict the effective contact area of rough surface contact problems, arXiv preprint arXiv:2504.17354 (2025).doi:10.48550/arXiv.2504.17354

    2025 doi arXiv

  13. [21]

    W. L. Hamilton, Graph Representation Learning, Synthesis Lectures on Artificial Intelligence and Machine Learning, Springer International Publishing, Cham, 2020.doi:10.1007/978-3-031-01588-5. URLhttps://doi.org/10.1007/978-3-031-01588-5

    2020 doi

  14. [22]

    P. W. Battaglia, J. B. Hamrick, V . Bapst, A. Sanchez-Gonzalez, V . Zambaldi, M. Malinowski, A. Tacchetti, D. Raposo, A. Santoro, R. Faulkner, C. Gulcehre, F. Song, A. Ballard, J. Gilmer, G. Dahl, A. Vaswani, K. Allen, C. Nash, V . Langston, C. Dyer, N. Heess, D. Wierstra, P. ...

    2018 doi arXiv

  15. [23]

    P. W. Battaglia, R. Pascanu, M. Lai, D. Rezende, K. Kavukcuoglu, Interaction networks for learning about objects, relations and physics, arXiv:1612.00222 [cs] (Dec. 2016).doi:10.48550/arXiv.1612.00222. URLhttps://doi.org/10.48550/arXiv.1612.00222

    2016 doi arXiv

  16. [24]

    Sanchez-Gonzalez, J

    A. Sanchez-Gonzalez, J. Godwin, T. Pfaff, R. Ying, J. Leskovec, P. W. Battaglia, Learning to simulate complex physics with graph networks, version Number: 2 (2020).doi:10.48550/ARXIV.2002.09405. URLhttps://doi.org/10.48550/arXiv.2002.09405

    2020 doi arXiv

  17. [25]

    Mrowca, C

    D. Mrowca, C. Zhuang, E. Wang, N. Haber, L. Fei-Fei, J. B. Tenenbaum, D. L. K. Yamins, Flexible Neural Representation for Physics Prediction, arXiv:1806.08047 [cs] (Oct. 2018).doi:10.48550/arXiv.1806.08047. URLhttp://arxiv.org/abs/1806.08047

    2018 doi arXiv

  18. [26]

    F. d. A. Belbute-Peres, T. D. Economon, J. Z. Kolter, Combining Differentiable PDE Solvers and Graph Neural Networks for Fluid Flow Prediction, arXiv:2007.04439 [cs] (Aug. 2020).doi:10.48550/arXiv.2007.04439. URLhttp://arxiv.org/abs/2007.04439

    2007 doi arXiv

  19. [27]

    Dalton, H

    D. Dalton, H. Gao, D. Husmeier, Emulation of cardiac mechanics using graph neural networks, Computer Methods in Applied Mechanics and Engineering 401 (2022) 115645.doi:10.1016/j.cma.2022.115645. URLhttps://doi.org/10.1016/j.cma.2022.115645

    2022

  20. [28]

    Dalton, A

    D. Dalton, A. Lazarus, A. Rabbani, H. Gao, D. Husmeier, Graph neural network emulation of cardiac mechanics, 2021.doi:10.11159/ icsta21.127. URLhttps://avestia.com/ICSTA2021_Proceedings/files/paper/ICSTA_127.pdf

    2021

  21. [29]

    Pfaff, M

    T. Pfaff, M. Fortunato, A. Sanchez-Gonzalez, P. W. Battaglia, Learning mesh-based simulation with graph networks, CoRR abs/2010.03409 (2020).arXiv:2010.03409. URLhttps://doi.org/10.48550/arXiv.2010.03409

    2020 doi arXiv

  22. [30]

    K. R. Allen, Y . Rubanova, T. Lopez-Guevara, W. Whitney, A. Sanchez-Gonzalez, P. Battaglia, T. Pfaff, Learning rigid dynamics with face interaction graph networks, arXiv:2212.03574 [cs] (Dec. 2022).doi:10.48550/arXiv.2212.03574. URLhttp://arxiv.org/abs/2212.03574

    2022 doi arXiv

  23. [31]

    X. Zhu, Y . Qian, Q. Wang, Z. Feng, P.-A. Heng, Collision-aware interactive simulation using graph neural networks, Visual Computing for Industry, Biomedicine, and Art 5 (1) (2022) 15.doi:10.1186/s42492-022-00113-4. URLhttps://doi.org/10.1186/s42492-022-00113-4

    2022 doi

  24. [32]

    S. Lee, D. You, Data-driven prediction of unsteady flow over a circular cylinder using deep learning, Journal of Fluid Mechanics 879 (2019) 217–254.doi:10.1017/jfm.2019.700. URLhttps://www.cambridge.org/core/product/identifier/S0022112019007006/type/journal_article

    2019 doi

  25. [33]

    R. Wang, K. Kashinath, M. Mustafa, A. Albert, R. Yu, Towards Physics-informed Deep Learning for Turbulent Flow Prediction, in: Proceed- ings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, ACM, Virtual Event CA USA, 2020, pp. 1457–1466.doi...

    2020

  26. [34]

    M. Tang, D. Manocha, R. Tong, Fast continuous collision detection using deforming non-penetration filters, in: Proceedings of the 2010 ACM SIGGRAPH symposium on Interactive 3D Graphics and Games, ACM, 2010, pp. 7–13.doi:10.1145/1730804.1730806. URLhttps://doi.org/10.1145/17308...

    2010

  27. [35]

    X. Provot, Collision and self-collision handling in cloth model dedicated to design garments, in: Computer Animation and Simulation’97: Proceedings of the Eurographics Workshop in Budapest, Hungary, September 2–3, 1997, Springer, 1997, pp. 177–189. URLhttps://doi.org/10.1007/9...

    1997 doi

  28. [36]

    Yuksel, A Fast & Robust Solution for Cubic & Higher-Order Polynomials, in: Special Interest Group on Computer Graphics and Interactive Techniques Conference Talks, ACM, 2022, pp

    C. Yuksel, A Fast & Robust Solution for Cubic & Higher-Order Polynomials, in: Special Interest Group on Computer Graphics and Interactive Techniques Conference Talks, ACM, 2022, pp. 1–2.doi:10.1145/3532836.3536266. URLhttps://doi.org/10.1145/3532836.3536266

    2022

  29. [37]

    Paszke, S

    A. Paszke, S. Gross, S. Chintala, G. Chanan, E. Yang, Z. DeVito, Z. Lin, A. Desmaison, L. Antiga, A. Lerer, Automatic differentiation in pytorch, in: NIPS-W, 2017

    2017

  30. [38]

    M. Fey, J. E. Lenssen, Fast graph representation learning with pytorch geometric, in: ICLR Workshop on Representation Learning on Graphs and Manifolds, V ol. abs/1903.02428, 2019.arXiv:1903.02428. URLhttps://doi.org/10.48550/arXiv.1903.02428

    1903 doi arXiv

  31. [39]

    D. P. Kingma, J. Ba, Adam: A Method for Stochastic Optimization, version Number: 9 (2014).doi:10.48550/ARXIV.1412.6980. 20 URLhttps://doi.org/10.48550/arXiv.1412.6980

    2014 doi arXiv

  32. [40]

    S. A. Maas, B. J. Ellis, G. A. Ateshian, J. A. Weiss, Febio: finite elements for biomechanics, Journal of Biomechanical Engineering 134 (1) (2012) 011005.doi:10.1115/1.4005694. URLhttps://doi.org/10.1115/1.4005694

    2012 doi

  33. [41]

    F. Xu, S. Morganti, R. Zakerzadeh, D. Kamensky, F. Auricchio, A. Reali, T. J. R. Hughes, M. S. Sacks, M. Hsu, A framework for designing patient-specific bioprosthetic heart valves using immersogeometric fluid–structure interaction analysis, International Journal for Numerical ...

    2018 doi

  34. [42]

    Thubrikar, The Aortic Valve, Routledge, 2017.doi:10.4324/9780203737163

    M. Thubrikar, The Aortic Valve, Routledge, 2017.doi:10.4324/9780203737163

    2017 doi

  35. [43]

    M. C. H. Wu, R. Zakerzadeh, D. Kamensky, J. Kiendl, M. S. Sacks, M.-C. Hsu, An anisotropic constitutive model for immersogeometric fluid–structure interaction analysis of bioprosthetic heart valves, Journal of Biomechanics 74 (2018) 23–31.doi:10.1016/j.jbiomech. 2018.04.012. U...

    2018 doi

  36. [44]

    Standring (Ed.), Gray’s anatomy: The Anatomical Basis of Clinical Practice, forty-first edition Edition, Elsevier Limited, 2016

    S. Standring (Ed.), Gray’s anatomy: The Anatomical Basis of Clinical Practice, forty-first edition Edition, Elsevier Limited, 2016. 21

    2016

Pith tools

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