Pith. sign in

REVIEW 4 major objections 6 minor 55 references

Self-composing neural operators for high-frequency and multiscale PDE surrogates

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Repeatedly applying one fixed neural-operator backbone makes it a universal approximator, with error that shrinks as the number of repetitions grows.

desk verdict Fresh framework and strong USCT numbers, but the depth-scaling theorem is overstated as written; still worth a serious referee. read the letter →

arxiv 2508.20650 v2 pith:DENZZMRP submitted 2025-08-28 cs.LG

Juncai He , Xinliang Liu , Jinchao Xu This is my paper · ORCID
classification cs.LG MSC 68Txx65Mxx
keywords self-composingneuraloperatorlearningmultigridHelmholtzequationultrasoundcomputedtomographytrain-and-unrollspectralbiashigh-frequencyPDEsurrogates
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 a neural operator of the form $O(v)=P\circ (G\circ)^n\circ L(v)$ — one fixed backbone block $G$ applied $n$ times between a lifting map $L$ and a projection $P$ — is a universal approximator for every fixed $n$, and that for Lipschitz target operators the approximation error falls at rate $\epsilon+C_\epsilon/\log(n)$ as the composition count $n$ grows, with the backbone unchanged. If true, this turns depth into a free accuracy knob: PDE surrogates can be made more accurate by reusing the same parameter-efficient block instead of adding new parameters. That matters for high-frequency problems like ultrasound computed tomography, where standard Fourier-type operators struggle to represent sharp oscillations and large models strain memory. The paper proves both theorems, introduces a Train-and-Unroll curriculum that increases $n$ during training, and pairs the framework with a multigrid-inspired backbone with adaptive convolutions. On the USCT Helmholtz benchmark the resulting model reaches relative root-mean-square error 0.0028 at 300 kHz, roughly ten times lower than a Fourier neural operator and four times lower than the best compared Fourier variant.

What carries the argument

The central object is the self-composition operator $O(v)=P\circ (G\circ)^n\circ L(v)$, where $L$ lifts the coefficient field into a latent function space, $G$ is one neural operator of fixed network size applied $n$ times, and $P$ projects back to the solution space; this mirrors fixed-point iteration in PDE solvers by treating each application as one solver sweep. The depth-accuracy mechanism is Theorem 2.3, which relies on an external lemma stating that a fixed-size ReLU network's $n$-fold self-composition approximates Lipschitz functions to $O(n^{-1/k})$, giving the $1/\log(n)$ decay after the base error. For the Helmholtz application, the machinery is the multigrid V-cycle backbone with the adaptive-convolution mechanism $\operatorname{AdaConv}(k,x)=\bigl(\operatorname{MLP}(\operatorname{Filter}_k*k)\bigr)\odot\bigl(\operatorname{Filter}_x*x\bigr)$, which lets the learned smoother and PDE operator modulate their action by local wavenumber values.

What would settle it

Take a fixed Lipschitz target operator, such as the Helmholtz map on a simple two-layer medium, train the self-composed model at depths $n=1,2,4,\dots$ with identical backbone width and depth and identical data, and compare test error. If the error does not decay like $1/\log(n)$, or at least monotonically, as $n$ grows, or if the backbone size must grow with $n$ to see improvement, the central depth-accuracy claim is refuted. A cleaner check is to test the quoted fixed-size-composition theorem directly: a constant-size ReLU network's $n$-fold composition must reach $O(n^{-1/k})$ error on a known Lipschitz function such as a quadratic; if it plateaus, the proof mechanism fails.

Watch

Extended reading notes

Core claim

The core discovery is that self-composition alone is expressive: any continuous operator between Hilbert function spaces can be approximated to arbitrary accuracy by $P\circ(G\circ)^n\circ L$ even when the inner network $G$ is reused verbatim $n$ times (Theorem 2.1), and for Lipschitz operators (ones whose response changes by at most a constant factor times the change in input) one fixed $G$ of fixed size achieves error $\epsilon + C_\epsilon/\log(n)$ by increasing $n$ (Theorem 2.3). The proof reduces the operator to finitely many scalar functionals, approximates each scalar function by an $n$-fold self-composition of a fixed ReLU network using a quoted result that avoids growing the network, and then lifts that scalar approximation back to the function space using constant functions. The paper also identifies the practical mechanism: the Train-and-Unroll strategy starts with $n=1$, trains to convergence, then increments $n$ and warm-starts, so the deep self-composed model is reached by a curriculum rather than a single difficult optimization. With a multigrid V-cycle backbone and adaptive convolutions that multiply wave features by wavenumber-dependent masks, the model becomes MgNO-self and sets the reported low errors on the USCT task.

Load-bearing premise

The load-bearing premise is a quoted external result: one fixed-size network, applied $n$ times to itself, can approximate any function whose output changes by at most a constant factor times the input change, with error shrinking like $n^{-1/k}$, without growing the network. If that result fails in the form used, the paper's guarantee that deeper self-composition reduces error collapses.

Editorial extensions

If this is right

  • Increasing the composition count $n$ lowers the guaranteed approximation error to $\epsilon + C_\epsilon/\log(n)$ with no new parameters, so depth becomes a controlled accuracy knob for a fixed backbone.
  • Because a practitioner can stop training at any intermediate depth, Train-and-Unroll outputs a whole family of surrogates and lets compute be traded against accuracy at deployment time.
  • For the USCT Helmholtz problem, the self-composed multigrid model reduces relative error to 0.0028 at 300 kHz, 0.0036 at 400 kHz, and 0.0049 at 500 kHz, outperforming FNO, AFNO, and BFNO on both RRMSE and maximum error.
  • On Darcy and multiscale benchmarks, self-composition cuts parameter counts by about a factor of three (0.17M versus 0.57M for the MgNO backbone) while keeping accuracy essentially on par with the non-shared model.

Reading between the lines

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

  • A testable extension the paper leaves open is swapping the multigrid backbone for a different local update inside the same $P\circ(G\circ)^n\circ L$ shell; if the depth-scaling benefit persists, self-composition is a general accuracy lever rather than a multigrid artifact.
  • The paper's own error-versus-depth curves decay faster than the proven $1/\log(n)$ rate; an editorial inference is that the bound is loose, and the practical depth scaling may be closer to $1/n$ for smoother operators.
  • Because the same weights are reused at every stage, self-composition imposes a strong weight-tying prior: the same solver sweep is assumed useful everywhere. An ablation that unshares the per-stage weights would isolate how much of the gain is due to sharing itself.
  • In the large-$n$ limit the composed block approaches a fixed point of $G$, suggesting self-composition could serve as a warm start or regularizer for implicit-style surrogates; the paper does not explore that connection.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a self-composing neural operator (SC-NO) framework of the form O(v)=P∘(G∘)^n∘L(v), in which a single backbone operator G is applied n times to a lifted latent state, motivated by iterative PDE solvers such as multigrid. The authors prove a universal approximation result for any fixed n (Theorem 2.1) and a quantitative depth-accuracy bound of ϵ + C_ϵ/log(n) for Lipschitz target operators (Theorem 2.3), introduce a Train-and-Unroll strategy that grows composition depth during training, and report experiments on Darcy flow benchmarks and a 300--500 kHz USCT Helmholtz task where the MgNO-self model achieves substantially lower RRMSE than FNO, AFNO, BFNO, and UNet.

Significance. If the theoretical claims were fully established, the paper would make a useful contribution to parameter-efficient neural operator design: the self-composition architecture shares parameters across depth, the Train-and-Unroll curriculum is a practical training scheme, and the USCT results at high frequencies address a genuinely hard benchmark. The paper is also honest in its 'Discussion on the convergence rate,' acknowledging that the proved O(1/log n) rate is slow and that practical gains may come from parameter sharing and initialization rather than asymptotic depth scaling. However, as written, the central depth-accuracy theorem is not demonstrated because of its quantifier structure and its dependence on an unstated external approximation theorem, and the empirical isolation of the self-composition mechanism is incomplete. No code, data split details, or error bars are provided, so the empirical claims are currently not reproducible.

major comments (4)
  1. [Section 2.3, Theorem 2.3 (Eq. 2.28)] The statement of Theorem 2.3 has the quantifier structure 'for any n there exist L_n, P_n, G_n', so it bounds the optimal error of the n-fold composition class, not the accuracy of a single fixed backbone as n grows. The conclusion after the proof that 'the deeper, the larger n, the better' is not a logical consequence of the theorem. Since the paper's central motivation is that repeatedly applying the same backbone G improves accuracy, either prove a fixed-G version (G independent of n) or explicitly restate the claim as a property of the class and soften the interpretation accordingly. The paper's own 'Discussion on the convergence rate' concedes that the O(1/log n) rate is slow and that practical gains may stem from parameter-sharing regularization and Train-and-Unroll initialization; this concession should be reconciled with the surrounding narrative.
  2. [Section 2.3, Lemma 2.2 (Eqs. 2.9-2.10)] The proof of Lemma 2.2 is not self-contained or verifiable as written. It invokes Theorem 1.3 of [52] without stating its hypotheses or conclusion, and the rescaling step 'take n = q n ≥ max_i n_i' followed by 'g_i = (g_i∘)^q' is not a valid derivation unless Theorem 1.3 supplies a single fixed-size network whose n-fold composition approximates each F_i for every n. If the external theorem only supplies an n_i-dependent network, then increasing the composition count by q does not preserve the approximation error, and the argument collapses. This lemma is load-bearing for both Theorem 2.1 and Theorem 2.3, so the authors must either state and prove the adaptation of the external theorem in full or provide a self-contained proof. As printed, the notation also mixes n_i and n inconsistently, which further obscures the argument.
  3. [Section 5, Table 2 and Section 2.4, Table 1] The empirical evidence for the benefit of self-composition is incomplete. Table 2 compares MgNO-self with UNet, FNO, AFNO, and BFNO but includes no non-self MgNO baseline, so the large gains on the USCT Helmholtz task cannot be attributed to self-composition rather than to the multigrid-inspired AdaConv backbone. On the Darcy benchmarks in Table 1, MgNO-self is consistently less accurate than MgNO (e.g., L2 error 0.187 vs 0.153 on smooth, 0.371 vs 0.339 on rough, 0.800 vs 0.715 on multiscale), which further weakens the claim that repeated composition improves accuracy. The authors should add a non-self MgNO baseline on USCT and report controlled ablation experiments that isolate composition depth from backbone design.
  4. [Section 5, numerical experiments] The reported SOTA results are single point estimates with no error bars, no number of seeds, no data split details, and no training hyperparameters; no code is provided. For a methods paper whose main empirical claim is a large margin over baselines, this is not sufficient to assess robustness, especially given the small baseline set. Please report means and standard deviations over multiple seeds, describe the data split and preprocessing for OpenBreastUS, and release code or provide detailed training configurations.
minor comments (6)
  1. [Title] The title in the paper header differs from the arXiv title; please align the submitted title with the published version.
  2. [Section 2.3, Eq. (2.44)] The arithmetic in Eq. (2.44) is inconsistent with the preceding bounds: the first term should be ϵ/2 if it is to match Eq. (2.29), not ϵ/3. Also, Theorem 2.3 should state n ≥ 2, since log(1) = 0.
  3. [Lemma 2.2] The proof of Lemma 2.2 uses F_i and eF_i interchangeably in Eq. (2.10); please clarify the notation throughout the proof.
  4. [Throughout] There are numerous typos, e.g., 'self-comprising' in the Theorem 2.1 preamble, 'naural motivation', 'campactness', and 'comprehensivly benched'; a careful proofread is needed.
  5. [Figure 2] Figure 2 is referenced but the visual is not present in the submitted text; ensure the figure and its caption are included and that the axes and labels are described.
  6. [Table 2] The note 'MgNO-self mode is named after MgNO' is unclear and should be expanded or removed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central depth-scaling theorem is imported from an external fixed-size ReLU composition result, and the empirical claims are evaluated on external benchmarks and baselines.

full rationale

I find no circular derivation chain. The operator approximation theorems (Theorem 2.1 and Theorem 2.3) reduce Lipschitz operator approximation to scalar Lipschitz approximation on [0,1]^k via projection, lifting by constant functions, and the pointwise-constant construction in equations (2.18)-(2.22). The depth-accuracy bound is imported from Zhang, Lu, and Zhao [52], an external ICML theorem on compositions of fixed-size ReLU networks; it is not a result of the present authors, and its assumptions do not include the target operator-approximation claim, so citation [52] is independent support rather than self-citation. Self-citations to MgNet [22,23] and MgNO [21] appear in the motivation and architectural context (e.g., Figure 3 caption) but are not load-bearing for universality or for the error-rate theorem. The paper's own 'Discussion on the convergence rate' explicitly concedes that the O(1/log n) rate is slow and that practical gains may come from parameter-sharing regularization and the Train-and-Unroll strategy; this is a soundness caveat, not a circularity. The skeptic's concern about Theorem 2.3 — that G is allowed to depend on n, so the theorem does not literally prove that a single backbone becomes more accurate as n grows — is a quantifier-scope and interpretation issue in the proof, not a self-referential reduction: the stated bound is for the class of n-fold compositions and follows from [52] if that theorem is used as stated. Lemma 2.2 is compressed and contains a notational slip (ni versus n in equation (2.10)), but that is a technical soundness risk, not circularity. The empirical USCT comparisons use the external OpenBreastUS benchmark [51] and external baselines; no fitted parameter is relabeled as a prediction. Overall, the central claims are not equivalent to their inputs by construction.

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

No fitted constants are hidden in the theory; the constants C_ϵ are existential and independent of n. The main load-bearing inputs are the borrowed approximation theorem, the Lipschitz assumption, and the function-space embedding. The adaptive convolution mechanism is an architectural component, not an invented physical entity.

assumptions (3)
  • standard math Theorem 1.3 in Zhang, Lu, Zhao (2023): fixed-size ReLU networks can approximate Lipschitz functions with error O(n^{-1/k}) under n-fold self-composition.
    Borrowed external theorem used in Lemma 2.2 and Theorem 2.3; not proven in this paper.
  • domain assumption Target operator O* is Lipschitz continuous.
    Assumed in Theorem 2.3 to convert the [52] rate into the stated 1/log(n) bound.
  • domain assumption The function space Z contains the constant function 1(x) and the domain is bounded.
    Required for the lifting and projection construction in the proofs of Theorems 2.1 and 2.3.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Self-composing neural operators for high-frequency and multiscale PDE surrogates." pith.science (2026). https://pith.science/paper/DENZZMRP

@misc{pith2026250820650,
  author       = {Pith},
  title        = {Pith review of: Self-composing neural operators for high-frequency and multiscale PDE surrogates},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DENZZMRP}},
  note         = {Machine review of arXiv:2508.20650}
}
read the original abstract

Addressing the computational challenges of high-frequency and multiscale partial differential equations (PDEs), this work introduces a self-composing neural operator (SC-NO) framework. Inspired by classical fixed-point iterative solvers (e.g., multigrid, domain decomposition), the proposed architecture constructs a deep operator by repeatedly applying a single, parameter-efficient backbone block. This design mimics the update step of a numerical solver, allowing the model to progressively resolve complex solution features without increasing the parameter count. For practical training, we develop an adaptive ``Train-and-Unroll'' strategy that grows the composition depth during training, acting as a curriculum from shallow to deep self-composed models. We demonstrate the efficacy of this framework on the Helmholtz equation for ultrasound computed tomography (USCT), a problem characterized by high-frequency wave propagation in highly heterogeneous media. By instantiating the backbone with a multigrid-inspired architecture, the SC-NO effectively mitigates the spectral bias often observed in standard operator learning baselines. Numerical experiments show that our method reduces the prediction error significantly compared to Fourier Neural Operators (FNO) and their variants in the 300--500 kHz regime. Furthermore, we provide theoretical analysis linking the self-composition depth to approximation accuracy.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 43 canonical work pages

  1. [52]

    Zhang, J

    S. Zhang, J. Lu, and H. Zhao , On enhancing expressive power via compositions of single fixed-size relu network , in International Conference on Machine Learning, PMLR, 2023, pp. 41452–41487

    work page 2023

  2. [1]

    S. Bai, J. Z. Kolter, and V. Koltun , Deep equilibrium models, Advances in neural informa- tion processing systems, 32 (2019)

    work page 2019

  3. [2]

    J. A. L. Benitez, T. Furuya, F. F aucher, A. Kratsios, X. Tricoche, and M. V. de Hoop , Out-of-distributional risk bounds for neural operators with applications to the helmholtz equation, Journal of Computational Physics, 513 (2024), p. 113168

    work page 2024

  4. [3]

    Cao, Choose a transformer: Fourier or galerkin, Advances in Neural Information Processing Systems, 34 (2021)

    S. Cao, Choose a transformer: Fourier or galerkin, Advances in Neural Information Processing Systems, 34 (2021)

    work page 2021

  5. [4]

    G. Chen, X. Liu, Q. Meng, L. Chen, C. Liu, and Y. Li , Learning neural operators on riemannian manifolds, arXiv preprint arXiv:2302.08166, (2023)

    arXiv 2023

  6. [5]

    J.-S. Chen, C. Pan, C.-T. Wu, and W. K. Liu , Reproducing kernel particle methods for large deformation analysis of non-linear structures , Computer Methods in Applied Mechanics and Engineering, 139 (1996), pp. 195–227

    work page 1996

  7. [6]

    R. T. Chen, Y. Rubanova, J. Bettencourt, and D. K. Duvenaud , Neural ordinary differ- ential equations, Advances in neural information processing systems, 31 (2018)

    2018

  8. [7]

    Y. Chen, X. Dai, M. Liu, D. Chen, L. Yuan, and Z. Liu , Dynamic convolution: Attention over convolution kernels , in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 11030–11039

    work page 2020

Show all 55 references
  1. [8]

    Y. Chen, B. Dong, and J. Xu , Meta-mgnet: Meta multigrid networks for solving parameter- ized partial differential equations, Journal of computational physics, 455 (2022), p. 110996

    2022

  2. [9]

    J. T. Connor, R. D. Martin, and L. E. Atlas , Recurrent neural networks and robust time series prediction, IEEE Transactions on Neural Networks, 5 (1994), pp. 240–254

    1994

  3. [10]

    X. Ding, X. Zhang, J. Han, and G. Ding , Scaling up your kernels to 31x31: Revisiting large 22 J. HE, X. LIU, J. XU kernel design in cnns , in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 11963–11975

    2022

  4. [11]

    E , A Proposal on Machine Learning via Dynamical Systems , Communications in Mathe- matics and Statistics, 5 (2017), pp

    W. E , A Proposal on Machine Learning via Dynamical Systems , Communications in Mathe- matics and Statistics, 5 (2017), pp. 1–11

    2017

  5. [12]

    S. W. Fung, H. Heaton, Q. Li, D. McKenzie, S. Osher, and W. Yin , Jfb: Jacobian-free backpropagation for implicit networks, in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 36, 2022, pp. 6648–6656

    2022

  6. [13]

    Geng, X.-Y

    Z. Geng, X.-Y. Zhang, S. Bai, Y. W ang, and Z. Lin , On training implicit models , Advances in Neural Information Processing Systems, 34 (2021), pp. 24247–24260

    2021

  7. [14]

    Gregor and Y

    K. Gregor and Y. LeCun , Learning fast approximations of sparse coding , in Proceedings of the 27th international conference on international conference on machine learning, 2010, pp. 399–406

    2010

  8. [15]

    Guibas, M

    J. Guibas, M. Mardani, Z. Li, A. Tao, A. Anandkumar, and B. Catanzaro , Adap- tive fourier neural operators: Efficient token mixers for transformers , arXiv preprint arXiv:2111.13587, (2021)

    2021 arXiv

  9. [16]

    Gupta, X

    G. Gupta, X. Xiao, and P. Bogdan , Multiwavelet-based operator learning for differential equations, Advances in Neural Information Processing Systems, 34 (2021), pp. 24048– 24062

    2021

  10. [17]

    Haber, L

    E. Haber, L. Ruthotto, E. Holtham, and S.-H. Jun , Learning across scales—multiscale methods for convolution neural networks , in Proceedings of the AAAI conference on arti- ficial intelligence, vol. 32, 2018

    2018

  11. [18]

    Hackbusch , Multi-grid methods and applications , vol

    W. Hackbusch , Multi-grid methods and applications , vol. 4, Springer Science & Business Media, 2013

    2013

  12. [19]

    W. Hao, X. Liu, and Y. Yang , Newton informed neural operator for solving nonlinear partial differential equations, in Advances in Neural Information Processing Systems, vol. 37, 2024

    2024

  13. [20]

    Z. Hao, Z. W ang, H. Su, C. Ying, Y. Dong, S. Liu, Z. Cheng, J. Song, and J. Zhu , Gnot: A general neural operator transformer for operator learning , in International Conference on Machine Learning, PMLR, 2023, pp. 12556–12569

    2023

  14. [21]

    J. He, X. Liu, and J. Xu , Mgno: Efficient parameterization of linear operators via multigrid , in International Conference on Learning Representations, 2024

    2024

  15. [22]

    He and J

    J. He and J. Xu , Mgnet: A unified framework of multigrid and convolutional neural network , Science China Mathematics, (2019), pp. 1–24

    2019

  16. [23]

    J. He, J. Xu, L. Zhang, and J. Zhu , An interpretive constrained linear model for resnet and mgnet, Neural Networks, 162 (2023), pp. 384–392

    2023

  17. [24]

    K. He, X. Zhang, S. Ren, and J. Sun , Deep residual learning for image recognition , in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770–778

    2016

  18. [25]

    K. He, X. Zhang, S. Ren, and J. Sun , Identity mappings in deep residual networks , in European Conference on Computer Vision, Springer, 2016, pp. 630–645

    2016

  19. [26]

    Hu and P

    J. Hu and P. Jin , A hybrid iterative method based on mionet for pdes: Theory and numerical examples, Mathematics of Computation, (2025)

    2025

  20. [27]

    X. Jia, B. De Brabandere, T. Tuytelaars, and L. V an Gool , Dynamic filter networks , in Advances in neural information processing systems, 2016, pp. 667–675

    2016

  21. [28]

    P. Jin, S. Meng, and L. Lu , Mionet: Learning multiple-input operators via tensor product , SIAM Journal on Scientific Computing, 44 (2022), pp. A3490–A3514

    2022

  22. [29]

    Kovachki, Z

    N. Kovachki, Z. Li, B. Liu, K. Azizzadenesheli, K. Bhattacharya, A. Stuart, and A. Anandkumar , Neural operator: Learning maps between function spaces , arXiv pre- print arXiv:2108.08481, (2021)

    2021 arXiv

  23. [30]

    Kovachki, Z

    N. Kovachki, Z. Li, B. Liu, K. Azizzadenesheli, K. Bhattacharya, A. Stuart, and A. Anandkumar , Neural operator: Learning maps between function spaces with appli- cations to pdes, Journal of Machine Learning Research, 24 (2023), pp. 1–97

    2023

  24. [31]

    Lanthaler, Z

    S. Lanthaler, Z. Li, and A. M. Stuart , The nonlocal neural operator: Universal approxi- mation, arXiv preprint arXiv:2304.13221, (2023)

    2023 arXiv

  25. [32]

    Z. Li, D. Z. Huang, B. Liu, and A. Anandkumar , Fourier neural operator with learned deformations for pdes on general geometries , arXiv preprint arXiv:2207.05209, (2022)

    2022 arXiv

  26. [33]

    Z. Li, N. B. Kovachki, K. Azizzadenesheli, K. Bhattacharya, A. Stuart, and A. Anand- kumar, Fourier NeuralOperator for Parametric Partial Differential Equations , in Inter- national Conference on Learning Representations, 2020

    2020

  27. [34]

    X. Liu, B. Xu, S. Cao, and L. Zhang , Mitigating spectral bias for the multiscale operator learning, Journal of Computational Physics, 506 (2024), p. 112944

    2024

  28. [35]

    L. Lu, P. Jin, G. Pang, Z. Zhang, and G. E. Karniadakis , Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators, Nature Machine SELF-COMPOSING NEURAL OPERATORS 23 Intelligence, 3 (2021), pp. 218–229

    2021

  29. [36]

    Y. Lu, A. Zhong, Q. Li, and B. Dong , Beyond finite layer neural networks: Bridging deep architectures and numerical differential equations, in International Conference on Machine Learning, PMLR, 2018, pp. 3276–3285

    2018

  30. [37]

    H. Luo, H. Wu, H. Zhou, L. Xing, Y. Di, J. W ang, and M. Long , Transolver++: An accu- rate neural solver for pdes on million-scale geometries , arXiv preprint arXiv:2502.02414, (2025)

    2025 arXiv

  31. [38]

    Marwah, A

    T. Marwah, A. Pokle, J. Z. Kolter, Z. Lipton, J. Lu, and A. Risteski , Deep equilibrium based neural operators for steady-state pdes , Advances in Neural Information Processing Systems, 36 (2023), pp. 15716–15737

    2023

  32. [39]

    Osnabrugge, S

    G. Osnabrugge, S. Leedumrongwatthanakun, and I. M. Vellekoop , A convergent born series for solving the inhomogeneous helmholtz equation in arbitrarily large media , Journal of computational physics, 322 (2016), pp. 113–124

    2016

  33. [40]

    Pathak, S

    J. Pathak, S. Subramanian, P. Harrington, S. Raja, A. Chattopadhyay, M. Mardani, T. Kurth, D. Hall, Z. Li, K. Azizzadenesheli, et al. , Fourcastnet: A global data- driven high-resolution weather model using adaptive fourier neural operators , arXiv pre- print arXiv:2202.11214, (2022)

    2022 arXiv

  34. [41]

    M. A. Rahman, Z. E. Ross, and K. Azizzadenesheli , U-no: U-shaped neural operators, arXiv e-prints, (2022), pp. arXiv–2204

    2022

  35. [42]

    Raonic, R

    B. Raonic, R. Molinaro, T. Rohner, S. Mishra, and E. de Bezenac , Convolutional neural operators, in ICLR 2023 Workshop on Physics for Machine Learning, 2023

    2023

  36. [43]

    Ronneberger, P

    O. Ronneberger, P. Fischer, and T. Brox , U-net: Convolutional networks for biomed- ical image segmentation , in International Conference on Medical image computing and computer-assisted intervention, Springer, 2015, pp. 234–241

    2015

  37. [44]

    Trottenberg, C

    U. Trottenberg, C. W. Oosterlee, and A. Schuller , Multigrid, Elsevier, 2000

    2000

  38. [45]

    V aswani, N

    A. V aswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin , Attention is all you need , Advances in neural information processing systems, 30 (2017)

    2017

  39. [46]

    H. Wu, T. Hu, H. Luo, J. W ang, and M. Long , Solving high-dimensional pdes with latent spectral models, arXiv preprint arXiv:2301.12664, (2023)

    2023 arXiv

  40. [47]

    Xu , Theory of multilevel methods , vol

    J. Xu , Theory of multilevel methods , vol. 8924558, Cornell University, 1989

    1989

  41. [48]

    B. Yang, G. Bender, Q. V. Le, and J. Ngiam , Condconv: Conditionally parameterized convolutions for efficient inference, in Advances in neural information processing systems, 2019, pp. 1305–1316

    2019

  42. [49]

    H. You, Q. Zhang, C. J. Ross, C.-H. Lee, M.-C. Hsu, and Y. Yu , A physics-guided neu- ral operator learning approach to model biological tissues from digital image correlation measurements, Journal of Biomechanical Engineering, accepted for publication, (2022), p. arXiv prepri...

    2022 arXiv

  43. [50]

    C. Zeng, Y. Zhang, J. Zhou, Y. W ang, Z. W ang, Y. Liu, L. Wu, and D. Z. Huang , Point cloud neural operator for parametric pdes on complex and variable geometries, arXiv preprint arXiv:2501.14475, (2025)

    2025 arXiv

  44. [51]

    Z. Zeng, Y. Zheng, H. Hu, Z. Dong, Y. Zheng, X. Liu, J. W ang, Z. Shi, L. Zhang, Y. Li, et al. , Openbreastus: Benchmarking neural operators for wave imaging using breast ultrasound computed tomography, arXiv preprint arXiv:2507.15035, (2025)

    2025 arXiv

  45. [53]

    Q. Zhao, Y. Ma, P. Boufounos, S. Nabi, and H. Mansour , Deep born operator learning for reflection tomographic imaging, in ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2023, pp. 1–5

    2023

  46. [54]

    J. Zhu, J. He, and Q. Huang , An enhanced v-cycle mgnet model for operator learning in numerical partial differential equations, Computational Geosciences, (2023), pp. 1–12

    2023

  47. [55]

    J. Zhu, J. He, L. Zhang, and J. Xu , Fv-mgnet: Fully connected v-cycle mgnet for interpretable time series forecasting, Journal of Computational Science, 69 (2023), p. 102005

    2023

Pith tools

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