Pith. sign in

REVIEW 3 major objections 4 minor 51 references

Benchmarking Multi-fidelity Neural Operators on Complex PDE Problems with Non-trivial Fidelity Differences

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read Transfer learning is the only multi-fidelity strategy that consistently beats the high-fidelity baseline across four PDE benchmarks; direct low-fidelity-input injection degrades as the fidelity gap grows.

desk verdict Useful benchmark comparison with a solid internal ranking, but the 'model-form fidelity' claim rests on a POD-projection proxy that needs to be fixed or reframed before the paper is publishable. read the letter →

arxiv 2608.04708 v1 pith:4S4TVOWC submitted 2026-08-05 cs.CE

classification cs.CE
keywords neuraloperatormulti-fidelitylearningtransferFourierproperorthogonaldecompositionDarcyflowsmokeinflowautoregressiveerroraccumulation
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

Multi-fidelity neural operators try to compensate for scarce high-fidelity (HF) simulation data by exploiting abundant low-fidelity (LF) data. This paper systematically compares four ways to do that — two-step, residual, intermediate, and transfer learning — on four PDE test cases, including two newly designed to create large LF-HF gaps: a Darcy flow whose LF solution is obtained by keeping only the two dominant proper orthogonal decomposition modes, and an unsteady smoke-inflow problem whose LF-HF discrepancy grows over time. Its central finding is that only transfer learning, which initialises the HF neural operator with weights pretrained on LF data, consistently outperforms a model trained on HF data alone. Methods that feed LF predictions into the HF model as inputs tend to amplify errors, most severely in autoregressive time-stepping (using earlier predictions as inputs to later steps), when the LF and HF solutions disagree substantially. The intended upshot is that practitioners should prefer transfer learning for multi-fidelity PDE surrogate modelling when fidelity levels differ by more than grid resolution.

What carries the argument

The machinery is a controlled comparison of four multi-fidelity integration schemes on one baseline architecture (Fourier neural operator, with a wavelet operator in the Darcy cases), plus two purpose-built benchmarks that separate fidelity-gap size from grid effects. The modified Darcy case creates its low-fidelity data by rank-2 proper orthogonal decomposition, $A_2=U\Sigma_2 V^T$ (Eqs. 19–20), raising the mean absolute LF-HF discrepancy from 0.02647 to 0.151; the smoke-inflow case pairs a 96×120 low-fidelity grid with a 256×320 high-fidelity grid in a time-dependent smoke-advection problem, where discrepancy grows with time and single-step rollout during training is followed by autoregressive evaluation. These benchmarks are what let the paper attribute performance differences to the LF-HF mismatch mechanism rather than to trivial resolution similarity.

What would settle it

Build a version of the Darcy or smoke-inflow benchmark where the low-fidelity data comes from a solver with a genuinely different governing equation or closure (for example, an actual turbulence model rather than a finer or coarser grid), rerun all four strategies, and check whether transfer learning still wins; if two-step or residual matches it there, the central claim is limited to the paper's projection-style fidelity gap.

Watch

Extended reading notes

Core claim

The paper's central claim is that the way low-fidelity information enters a neural operator determines whether multi-fidelity training helps or hurts. On all four test cases — 1D stochastic Poisson, 2D Darcy flow, a modified Darcy flow with a large fidelity gap, and unsteady smoke inflow — the transfer-learning strategy is the only one that consistently beats the HF-only baseline. The two-step and residual strategies, which form $\hat{y}_{HF}=F(x,\hat{y}_{LF})$ or learn the residual $r=y_{HF}-y_{LF}$, and the intermediate architecture, which shares a trunk and emits LF from a hidden layer, all lose accuracy when the LF-HF discrepancy is large; in the unsteady case the gap widens over time and autoregressive rollout compounds the error. The paper interprets this as evidence that injecting LF outputs as input features confuses the HF network, whereas LF-pretrained weights supply a structural prior that HF fine-tuning can refine.

Load-bearing premise

The paper's third test case assumes that filtering the high-fidelity solutions down to their two dominant patterns mimics the gap between two genuinely different physics models, even though the low-fidelity data is derived from the very high-fidelity solutions it is compared against, so the transfer-learning conclusion may not hold for true model-form discrepancies.

Editorial extensions

If this is right

  • With scarce high-fidelity data, initialising the HF neural operator with low-fidelity-pretrained weights is the reliable way to exploit LF data; it improves accuracy over HF-only training on all tested problems.
  • Two-step, residual, and intermediate architectures that inject LF predictions into the HF input should be avoided when the LF-HF gap is large, since they can underperform the HF baseline.
  • In time-dependent problems, direct LF injection is especially risky: the fidelity gap widens with time and autoregressive rollout turns small errors into large ones.
  • The benefit of LF data shrinks as HF data becomes plentiful, and non-transfer strategies can slightly degrade performance once 50 HF samples are available.
  • Benchmark design matters: close-to-HF low-fidelity data (grid-only differences) can mask differences between multi-fidelity strategies that appear only at larger fidelity gaps.

Reading between the lines

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

  • Editorial inference: the POD-truncated Darcy 'model-form' gap is not a true governing-equation difference — the LF data is a rank-2 projection of the HF snapshots, so it shares the HF mode structure and can only remove detail; the transfer-learning conclusion is therefore not established for genuinely different physics until a test with LF from a different solver is run.
  • Editorial inference: the discrepancy-error analysis suggests a practical diagnostic — measure the LF-HF gap before choosing a strategy, using transfer learning when the gap is large and any of the direct-injection methods when it is small.
  • Editorial inference: the mechanism at work, that weight-space transfer tolerates strong fidelity mismatch while input-space injection does not, is not specific to neural operators and could be tested in simpler multi-fidelity regression settings where the same failure mode should appear.
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 / 4 minor

Summary. The paper evaluates four multi-fidelity neural operator strategies—intermediate, two-step, residual, and transfer learning—against a high-fidelity-only baseline across four PDE test cases: a one-dimensional stochastic Poisson equation, a two-dimensional Darcy flow, a modified Darcy flow intended to emulate governing-equation discrepancies, and an unsteady smoke inflow problem. The main empirical claim is that only transfer learning consistently outperforms the high-fidelity baseline across all scenarios, while methods that feed low-fidelity predictions into the high-fidelity model are prone to error amplification under large LF-HF discrepancy, especially in autoregressive time-stepping. The paper also introduces two new test cases: a POD-truncated Darcy flow and a time-dependent smoke inflow benchmark.

Significance. If the central empirical claims hold, the paper is a useful systematic comparison for the multi-fidelity neural operator community. Its strengths include a consistent evaluation protocol with repeated random train/test splits, a clear comparison against a high-fidelity-only baseline, and a new unsteady smoke inflow dataset that addresses an underrepresented temporal setting. The observation that transfer learning is more robust than direct LF-input injection under large discrepancies is practically relevant and falsifiable. However, the modified Darcy test case, which is the key large-discrepancy case supporting the transfer-learning conclusion, rests on a construction that is not a faithful emulation of model-form discrepancies, and the text does not rule out test information leakage through the SVD. These issues are load-bearing for the paper's main claim and require careful resolution.

major comments (3)
  1. [Section 3.3, Eqs. (19)-(20)] The paper does not state whether the SVD in Eq. (19) is computed on the full snapshot matrix A, including the test split, or only on training snapshots. If the full matrix is used, the low-fidelity representation of each test input is constructed from the dominant modes of the high-fidelity test solutions themselves, which leaks information from the test set into the inputs. This would invalidate the third test case as a ranking experiment. The manuscript must specify the split used and, if the SVD included test data, rerun the experiment with a training-only SVD.
  2. [Section 3.3, Eqs. (19)-(20) and Fig. 10] Even with a correct train/test split, a rank-2 POD projection of the HF solution does not emulate a governing-equation discrepancy such as RANS versus LES. The LF and HF fields share the same POD basis by construction, so the LF field is a projection of the HF field that only removes spatial detail. This is an information-loss gap, not a model-form gap, and it cannot produce the physically different solutions that a genuinely different model would yield. Consequently, the conclusion that transfer learning is robust under 'substantial model discrepancies' is not established for true model-form mismatches; it is established only for projection-type discrepancies. The authors should either construct the LF data from an independent solver or model, or substantially weaken the claim that this test case resembles governing-equation differences.
  3. [Section 3.4 and Table 6] The text states that the training rollout is restricted to a single step ('we restrict the rollout to a single step', and 'the HF model is trained with a rollout number of one'), but Table 6 lists Rollout number 3 for both LF and HF models. This contradiction matters because the paper attributes the two-step model's failure to error accumulation in an autoregressive setting. If the HF model was actually trained with rollout 3, the stated training protocol is wrong; if it was trained with rollout 1, the table is wrong and the claims about autoregressive error accumulation should be reconsidered or clarified.
minor comments (4)
  1. [Table 2] The first numeric cell in the table reads '504.43×10−2', which appears to be a typographical error; the intended value is likely '4.43×10−2'. The row labels for the first two columns are also difficult to parse.
  2. [Section 3.1] The text says 'the FNO yields a better performance compared to other kernels, such as the Wavelet neural operator (WNO), in this test case' but the comparison is not shown. This claim should either be supported with a displayed result or removed.
  3. [Section 3.2 and Abstract] Section 3.2 states that both transfer learning and the two-step architecture are the only approaches that consistently surpass the HF baseline across all training sample configurations in that test case, while the abstract and conclusion state that only transfer learning consistently outperforms the baseline across all scenarios. The scope of the 'only transfer learning' claim should be clarified, since two-step also achieves this in the 2D Darcy case.
  4. [Fig. 12] The claim that transfer learning is 'the least sensitive' to increasing discrepancy is based on only two discrepancy levels per model. The figure shows two points per model, so the slope comparison is not statistically supported; this should be acknowledged or supplemented with more discrepancy levels.

Circularity Check

1 steps flagged · score 2.0 of 10

No significant circularity in the empirical comparison; the only constructional self-referentiality is in the modified-Darcy fidelity gap, which limits external validity without forcing the transfer-learning ranking.

  1. self definitional [Section 3.3, Eqs. (19)-(20)]
    "we decompose the solution using a proper orthogonal decomposition (POD) method. The snapshot matrix A of the solution is decomposed through a lower rank approximation ... In this case, we only retain the two most dominant modes, resulting in a reduced-order snapshot matrix A2. A2 = UΣ2VT ... With this approach, we preserve the overall structure of the solution while selectively removing finer details in the lower-fidelity data, mimicking the multi-fidelity scenario based on governing equations, as seen in fluid dynamics problems."

    By Eq. (20), the low-fidelity snapshot matrix A2 is defined as the high-fidelity snapshot matrix A truncated to its two dominant POD modes, so the LF and HF fields share the same POD basis by construction. The paper uses this construction to claim it 'mimics' RANS-vs-LES-type governing-equation discrepancies, but such discrepancies are model-form differences that can change the solution structure, whereas POD truncation can only remove detail from the very HF solution used as ground truth. Any conclusion that transfer learning is robust to 'substantial model discrepancies' drawn from this test case is therefore a statement about robustness to self-generated projection error rather than to independent model-form mismatch.

full rationale

The central claim is an empirical benchmark result: Tables 2, 4, 5 and Fig. 16 compare RMSE and MAE of transfer learning, two-step, residual, and intermediate architectures against an HF-only baseline. The transfer-learning ranking is not derived from an equation in which low-fidelity data is a fitted input; it is measured, and the paper even reports cases where multi-fidelity does not help and where residual regression degrades sharply (Table 5). There is no fitted parameter renamed as a prediction and no uniqueness theorem imported from the same authors. The only constructional circularity is in the modified Darcy test case (Eq. 20): LF data is a rank-2 POD projection of the HF snapshots, so the 'governing-equation-like' discrepancy is not an independent model-form gap. The text does not state whether the POD is computed on training snapshots only; if the test split is included, the LF test inputs would inherit HF test modes, which would bias the comparison. That is a genuine methodological caveat but not a demonstrated reduction of the concluding ranking. The smoke-inflow data is self-cited via [48], whose authors overlap with this paper, but it is a released Zenodo dataset with fixed simulation outputs, so the self-citation is not load-bearing in a way that makes the results circular. The paper itself flags a related limitation: 'the current smoke inflow test case does not reflect fidelity differences arising from variations in governing equations.' Overall, the main ranking has independent empirical content and should be judged a low-circularity benchmark; the modified-Darcy projection is a constructional weakness that caps the external-validity claim, not a derivation that forces the conclusion.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper contributes an empirical comparison, so the ledger is light: no fitted parameters and no invented entities. Hand-chosen quantities that affect the results are the intermediate loss weight alpha in Eq. 14 (value never reported), the rollout count for the smoke inflow models (3 per Table 6 but 1 per Section 3.4), and the POD rank k=2 that sets the modified-Darcy discrepancy level. The load-bearing assumptions are that the chosen architectures can represent the target solutions, that prior datasets are used faithfully, and, most importantly, that grid coarsening and POD truncation stand in for real model-form fidelity differences; the last is the paper's own construction and is asserted without validation.

free parameters (3)
  • alpha (intermediate loss weight) = not reported
    In Eq. 14, L = alpha MSE_HF + (1-alpha) MSE_LF + lambda ||theta||^2. The paper calls alpha a user-defined hyperparameter (Section 2.2.1) but never states its value, so the intermediate architecture's results cannot be reproduced or checked for fairness.
  • rollout number (smoke inflow) = 3 per Table 6, 1 per Section 3.4
    The rollout count controls how many autoregressive steps are used in training; Table 6 lists 3 for both LF and HF, while Section 3.4 states training used a single rollout step. The contradiction leaves the effective training protocol unspecified.
  • POD truncation rank k = 2
    The low-fidelity modified-Darcy data keeps only the two dominant modes (Eq. 20). The choice of k=2 sets the LF-HF discrepancy level (MAE 0.151) and is not justified or varied.
assumptions (4)
  • domain assumption Fourier and wavelet neural operators can approximate the PDE solution operators for the four test problems within the stated data budgets.
    The benchmark presumes the chosen FNO/WNO architectures are expressive enough for the 1D Poisson, Darcy, and smoke transport tasks; this is inherited from the operator-learning literature [20, 38] and not re-established here.
  • domain assumption LF-HF pairs in the test suite are representative of real multi-fidelity CFD scenarios, with grid coarsening and POD truncation standing in for RANS-versus-LES-type model discrepancies.
    Sections 1 and 3.3 argue grid-only fidelity differences are unrepresentative and that POD truncation mimics governing-equation variation; no independent evidence is offered that a rank-2 projection behaves like a different solver.
  • domain assumption The datasets from Tripura et al. [31] and Zakaria et al. [48] are used as-is and match the descriptions in those references.
    The 1D and 2D Darcy data are taken from [31] and the smoke inflow data from [48]; the paper does not regenerate or independently verify these datasets.
  • ad hoc to paper Rank-2 POD truncation of the HF snapshot matrix yields a valid low-fidelity solution whose fidelity gap is comparable to a change of governing equation.
    Eqs. (19)-(20) in Section 3.3 define the modified LF data as A2 = U Sigma2 V^T. This construction is specific to this paper and is asserted, not validated against an actual low-fidelity solver.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Benchmarking Multi-fidelity Neural Operators on Complex PDE Problems with Non-trivial Fidelity Differences." pith.science (2026). https://pith.science/paper/4S4TVOWC

@misc{pith2026260804708,
  author       = {Pith},
  title        = {Pith review of: Benchmarking Multi-fidelity Neural Operators on Complex PDE Problems with Non-trivial Fidelity Differences},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4S4TVOWC}},
  note         = {Machine review of arXiv:2608.04708}
}
read the original abstract

Solving PDE-governed physical problems is computationally expensive, limiting the availability of high-fidelity (HF) data for training neural operators, which typically require large datasets. Multi-fidelity learning addresses this by combining abundant low-fidelity (LF) data with limited HF samples. To evaluate several multi-fidelity neural operator strategies, including two-step, residual, intermediate, and transfer learning, we conduct a systematic assessment across four PDE test cases: two standard grid-based problems, a modified Darcy flow with LF-HF governing equation discrepancies, and an unsteady smoke inflow with temporal dynamics. Most prior studies focus on grid-based LF-HF differences, but real-world discrepancies, such as those between RANS and LES models in CFD, can be more complex. To capture these more challenging scenarios, we introduce two new test cases specifically designed to emulate nontrivial LF-HF mismatch beyond simple grid-resolution differences, enabling a more realistic assessment of multi-fidelity strategies under substantial model discrepancies. We find that methods directly feeding LF predictions into HF models are prone to error amplification under large LF-HF discrepancies, particularly in autoregressive settings. In contrast, transfer learning, which uses LF-trained weights to initialise the HF model, consistently improves accuracy by providing a robust prior refined with HF data. These results highlight the importance of establishing transfer learning as a reliable approach for efficient multi-fidelity PDE surrogate modelling.

Figures

Figures reproduced from arXiv: 2608.04708 by the authors.

Figure 1
Figure 1. FNO schematics: (a) full architecture of FNO. (b) Fourier layers. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Intermediate multi-fidelity neural operator architecture [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Two-step multi-fidelity neural operator architecture [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (14 more)
Figure 4
Figure 4. Figure 4: Residual multi-fidelity neural operator architecture [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Transfer learning process schematics. 7 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: An example of the difference between LF and HF data in a 1-dimensional stochastic Poisson equation. We vary the number of training samples on the LF and HF models to investigate the effect of the number of training samples on the multi-fidelity model. Each experiment i…
Figure 7
Figure 7. Figure 7: Single realisation on 1-dimensional stochastic Poisson equation test data. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: An example of the difference between LF and HF data in the original 2-dimensional Darcy flow problem. To train the model, we use the model hyperparameters as given in [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Prediction and error plot of a single data point for each model on 2D Darcy flow in a triangular domain with a notch. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Comparison between the original high-fidelity data, original lower-fidelity data, and the reduced-order lower-fidelity data. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Prediction and error plot for each model on modified 2D Darcy flow in triangular domain with a notch. [PITH_FULL_IMAGE:figures/full_fig_p014_11.png]
Figure 12
Figure 12. Figure 12: Discrepancy vs mean absolute error (MAE) plot for the 2-dimensional Darcy flow problem. Blue circle indicates the original data points [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]
Figure 13
Figure 13. Figure 13: Training history for Darcy flow problem. Blue lines represent training histories using the original dataset described in Section 3.2, while [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 14
Figure 14. Figure 14: Low and high-fidelity smoke inflow data with di [PITH_FULL_IMAGE:figures/full_fig_p016_14.png]
Figure 15
Figure 15. Figure 15: HF training phase loss history for smoke inflow problem. [PITH_FULL_IMAGE:figures/full_fig_p017_15.png]
Figure 16
Figure 16. Figure 16: RMSE at each timestep for each multi-fidelity strategy. [PITH_FULL_IMAGE:figures/full_fig_p018_16.png]
Figure 17
Figure 17. Figure 17: Data discrepancy vs time for the smoke inflow data. [PITH_FULL_IMAGE:figures/full_fig_p018_17.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

51 extracted references · 26 canonical work pages

  1. [1]

    Huthwaite, Accelerated finite element elastodynamic simulations using the gpu, Journal of Computational Physics 257 (2014) 687–707

    P. Huthwaite, Accelerated finite element elastodynamic simulations using the gpu, Journal of Computational Physics 257 (2014) 687–707. doi:10.1016/j.jcp.2013.10.017. URLhttp://dx.doi.org/10.1016/j.jcp.2013.10.017

  2. [2]

    Bernardini, D

    M. Bernardini, D. Modesti, F. Salvadore, S. Pirozzoli, Streams: A high-fidelity accelerated solver for direct numerical simulation of com- pressible turbulent flows, Computer Physics Communications 263 (2021) 107906. doi:10.1016/j.cpc.2021.107906. URLhttp://dx.doi.org/10.1016/j.cpc.2021.107906

  3. [3]

    Drucker, C

    H. Drucker, C. J. C. Burges, L. Kaufman, A. Smola, V . Vapnik, Support vector regression machines, in: M. Mozer, M. Jordan, T. Petsche (Eds.), Advances in Neural Information Processing Systems, V ol. 9, MIT Press, 1996, p. 1

  4. [4]

    C. E. Rasmussen, C. K. I. Williams, Gaussian processes for machine learning, Adaptive Computation and Machine Learning series, MIT Press, London, England, 2005

  5. [5]

    R. K. Tripathy, I. Bilionis, Deep uq: Learning deep neural network surrogate models for high dimensional uncertainty quantification, Journal of computational physics 375 (2018) 565–588

  6. [6]

    G. Sun, S. Wang, A review of the artificial neural network surrogate modeling in aerodynamic design, Proceedings of the Institution of Mechanical Engineers, Part G: Journal of Aerospace Engineering 233 (16) (2019) 5863–5872

  7. [7]

    Chatterjee, An introduction to the proper orthogonal decomposition, Current Science 78 (7) (2000) 808–817

    A. Chatterjee, An introduction to the proper orthogonal decomposition, Current Science 78 (7) (2000) 808–817. URLhttp://www.jstor.org/stable/24103957

  8. [8]

    Pakravan, P

    S. Pakravan, P. A. Mistani, M. A. Aragon-Calvo, F. Gibou, Solving inverse-pde problems with physics-aware neural networks, Journal of Computational Physics 440 (2021) 110414. doi:10.1016/j.jcp.2021.110414. URLhttp://dx.doi.org/10.1016/j.jcp.2021.110414

Show all 51 references
  1. [9]

    Molinaro, Y

    R. Molinaro, Y . Yang, B. Engquist, S. Mishra, Neural inverse operators for solving pde inverse problems (2023). arXiv:2301.11167. URLhttps://arxiv.org/abs/2301.11167

  2. [10]

    J. Berg, K. Nystr ¨om, Neural networks as smooth priors for inverse problems for pdes, Journal of Computational Mathematics and Data Science 1 (2021) 100008. doi:10.1016/j.jcmds.2021.100008. URLhttp://dx.doi.org/10.1016/j.jcmds.2021.100008

  3. [11]

    Raissi, P

    M. Raissi, P. Perdikaris, G. E. Karniadakis, Multistep neural networks for data-driven discovery of nonlinear dynamical systems (2018). arXiv:1801.01236. URLhttps://arxiv.org/abs/1801.01236

  4. [12]

    S. Pan, K. Duraisamy, Data-driven discovery of closure models, SIAM Journal on Applied Dynamical Systems 17 (4) (2018) 2381–2413. doi:10.1137/18m1177263. URLhttp://dx.doi.org/10.1137/18M1177263

  5. [13]

    M. M. Rai, N. K. Madavan, Aerodynamic design using neural networks, AIAA Journal 38 (1) (2000) 173–182. doi:10.2514/2.938. URLhttp://dx.doi.org/10.2514/2.938

  6. [14]

    X. Du, P. He, J. R. Martins, Rapid airfoil design optimization via neural networks-based parameterization and surrogate modeling, Aerospace Science and Technology 113 (2021) 106701. doi:10.1016/j.ast.2021.106701. URLhttp://dx.doi.org/10.1016/j.ast.2021.106701

  7. [15]

    Murtagh, Multilayer perceptrons for classification and regression, Neurocomputing 2 (5–6) (1991) 183–197

    F. Murtagh, Multilayer perceptrons for classification and regression, Neurocomputing 2 (5–6) (1991) 183–197. doi:10.1016/0925- 2312(91)90023-5. URLhttp://dx.doi.org/10.1016/0925-2312(91)90023-5 18

  8. [16]

    Popescu, V

    M.-C. Popescu, V . E. Balas, L. Perescu-Popescu, N. Mastorakis, Multilayer perceptron and neural networks, WSEAS Trans. Cir. and Sys. 8 (7) (2009) 579–588

  9. [17]

    Lecun, L

    Y . Lecun, L. Bottou, Y . Bengio, P. Haffner, Gradient-based learning applied to document recognition, Proceedings of the IEEE 86 (11) (1998) 2278–2324. doi:10.1109/5.726791. URLhttp://dx.doi.org/10.1109/5.726791

  10. [18]

    Raissi, P

    M. Raissi, P. Perdikaris, G. Karniadakis, Physics-informed neural networks: A deep learning framework for solving forward and inverse prob- lems involving nonlinear partial differential equations, Journal of Computational Physics 378 (2019) 686–707. doi:10.1016/j.jcp.2018.10.0...

  11. [19]

    L. Lu, P. Jin, G. Pang, Z. Zhang, G. E. Karniadakis, Learning nonlinear operators via deeponet based on the universal approximation theorem of operators, Nature Machine Intelligence 3 (3) (2021) 218–229. doi:10.1038/s42256-021-00302-5. URLhttp://dx.doi.org/10.1038/s42256-021-00302-5

  12. [20]

    Kovachki, Z

    N. Kovachki, Z. Li, B. Liu, K. Azizzadenesheli, K. Bhattacharya, A. Stuart, A. Anandkumar, Neural operator: Learning maps between function spaces (2021). doi:10.48550/ARXIV .2108.08481. URLhttps://arxiv.org/abs/2108.08481

  13. [21]

    D. E. Myers, Matrix formulation of co-kriging, Journal of the International Association for Mathematical Geology 14 (3) (1982) 249–257. doi:10.1007/bf01032887. URLhttp://dx.doi.org/10.1007/BF01032887

  14. [22]

    A. I. Forrester, A. S ´obester, A. J. Keane, Multi-fidelity optimization via surrogate modelling, Proceedings of the Royal Society A: Mathemat- ical, Physical and Engineering Sciences 463 (2088) (2007) 3251–3269. doi:10.1098/rspa.2007.1900. URLhttp://dx.doi.org/10.1098/rspa.2007.1900

  15. [23]

    P. S. Palar, T. Tsuchiya, G. T. Parks, Multi-fidelity non-intrusive polynomial chaos based on regression, Computer Methods in Applied Mechanics and Engineering 305 (2016) 579–606. doi:10.1016/j.cma.2016.03.022. URLhttp://dx.doi.org/10.1016/j.cma.2016.03.022

  16. [24]

    Koziel, A

    S. Koziel, A. Bekasiewicz, I. Couckuyt, T. Dhaene, Efficient multi-objective simulation-driven antenna design using co-kriging, IEEE Trans- actions on Antennas and Propagation 62 (11) (2014) 5900–5905. doi:10.1109/tap.2014.2354673. URLhttp://dx.doi.org/10.1109/TAP.2014.2354673

  17. [25]

    X. Liu, W. Zhao, D. Wan, Multi-fidelity co-kriging surrogate model for ship hull form optimization, Ocean Engineering 243 (2022) 110239. doi:10.1016/j.oceaneng.2021.110239. URLhttp://dx.doi.org/10.1016/j.oceaneng.2021.110239

  18. [26]

    X. Meng, G. E. Karniadakis, A composite neural network that learns from multi-fidelity data: Application to function approximation and inverse pde problems, Journal of Computational Physics 401 (2020) 109020. doi:10.1016/j.jcp.2019.109020. URLhttp://dx.doi.org/10.1016/j.jcp.20...

  19. [27]

    M. Guo, A. Manzoni, M. Amendt, P. Conti, J. S. Hesthaven, Multi-fidelity regression using artificial neural networks: Efficient ap- proximation of parameter-dependent output quantities, Computer Methods in Applied Mechanics and Engineering 389 (2022) 114378. doi:10.1016/j.cma....

  20. [28]

    Chakraborty, Transfer learning based multi-fidelity physics informed deep neural network, Journal of Computational Physics 426 (2021) 109942

    S. Chakraborty, Transfer learning based multi-fidelity physics informed deep neural network, Journal of Computational Physics 426 (2021) 109942. doi:10.1016/j.jcp.2020.109942. URLhttp://dx.doi.org/10.1016/j.jcp.2020.109942

  21. [29]

    Y . Lyu, X. Zhao, Z. Gong, X. Kang, W. Yao, Multi-fidelity prediction of fluid flow based on transfer learning using fourier neural operator, Physics of Fluids 35 (7) (jul 2023). doi:10.1063/5.0155555. URLhttp://dx.doi.org/10.1063/5.0155555

  22. [30]

    H. Tang, Q. Kong, J. P. Morris, Multi-fidelity fourier neural operator for fast modeling of large-scale geological carbon storage (2024). arXiv:2308.09113. URLhttps://arxiv.org/abs/2308.09113

  23. [31]

    Tripura, A

    T. Tripura, A. Thakur, S. Chakraborty, Multi-fidelity wavelet neural operator surrogate model for time-independent and time-dependent reliability analysis, Probabilistic Engineering Mechanics 77 (2024) 103672. doi:10.1016/j.probengmech.2024.103672. URLhttp://dx.doi.org/10.1016...

  24. [32]

    A. A. Howard, M. Perego, G. E. Karniadakis, P. Stinis, Multifidelity deep operator networks for data-driven and physics-informed problems, Journal of Computational Physics 493 (2023) 112462. doi:10.1016/j.jcp.2023.112462. URLhttp://dx.doi.org/10.1016/j.jcp.2023.112462

  25. [33]

    Renard, S

    N. Renard, S. Deck, ZONAL DETACHED EDDY SIMULATION OF UNSTEADY AIRFOIL AERODYNAMICS, in: AERO 2020+1 - 55th 3AF International Conference on Applied Conference, Poitiers (virtuel), France, 2021, p. 1. URLhttps://hal.science/hal-03206330

  26. [34]

    Hendrycks, K

    D. Hendrycks, K. Gimpel, Gaussian error linear units (gelus) (2023). arXiv:1606.08415. URLhttps://arxiv.org/abs/1606.08415

  27. [35]

    Elfwing, E

    S. Elfwing, E. Uchibe, K. Doya, Sigmoid-weighted linear units for neural network function approximation in reinforcement learning (2017). arXiv:1702.03118. URLhttps://arxiv.org/abs/1702.03118

  28. [36]

    A. F. Agarap, Deep learning using rectified linear units (relu) (2019). arXiv:1803.08375. URLhttps://arxiv.org/abs/1803.08375

  29. [37]

    Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, A. Anandkumar, Neural operator: Graph kernel network for partial differential equations (2020). arXiv:2003.03485. URLhttps://arxiv.org/abs/2003.03485

  30. [38]

    Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, A. Anandkumar, Fourier neural operator for parametric partial 19 differential equations (2021). arXiv:2010.08895. URLhttps://arxiv.org/abs/2010.08895

  31. [39]

    Eldred, A

    M. Eldred, A. Giunta, S. Collis, Second-Order Corrections for Surrogate-Based Optimization with Model Hierarchies, American Institute of Aeronautics and Astronautics, 2004, Ch. 1, p. 1. arXiv:https://arc.aiaa.org/doi/pdf/10.2514/6.2004-4457, doi:10.2514/6.2004-4457. URLhttps:/...

  32. [40]

    T. D. Robinson, M. S. Eldred, K. E. Willcox, R. Haimes, Surrogate-based optimization using multifidelity models with variable parameteri- zation and corrected space mapping, AIAA Journal 46 (11) (2008) 2814–2822. doi:10.2514/1.36043. URLhttp://dx.doi.org/10.2514/1.36043

  33. [41]

    Leifsson, S

    L. Leifsson, S. Koziel, Multi-fidelity design optimization of transonic airfoils using physics-based surrogate modeling and shape-preserving response prediction, Journal of Computational Science 1 (2) (2010) 98–106. doi:10.1016/j.jocs.2010.03.007. URLhttp://dx.doi.org/10.1016/...

  34. [42]

    Zhang, N

    Y . Zhang, N. H. Kim, C. Park, R. T. Haftka, Multifidelity surrogate based on single linear regression, AIAA Journal 56 (12) (2018) 4944–4952. doi:10.2514/1.j057299. URLhttp://dx.doi.org/10.2514/1.J057299

  35. [43]

    M. G. Fern ´andez-Godino, S. Dubreuil, N. Bartoli, C. Gogu, S. Balachandar, R. T. Haftka, Linear regression-based multifidelity sur- rogate for disturbance amplification in multiphase explosion, Structural and Multidisciplinary Optimization 60 (6) (2019) 2205–2220. doi:10.1007...

  36. [44]

    Giselle Fern ´andez-Godino, Review of multi-fidelity models, Advances in Computational Science and Engineering 1 (4) (2023) 351–400

    M. Giselle Fern ´andez-Godino, Review of multi-fidelity models, Advances in Computational Science and Engineering 1 (4) (2023) 351–400. doi:10.3934/acse.2023015. URLhttp://dx.doi.org/10.3934/acse.2023015

  37. [45]

    Babaee, P

    H. Babaee, P. Perdikaris, C. Chryssostomidis, G. E. Karniadakis, Multi-fidelity modelling of mixed convection based on experimental corre- lations and numerical simulations, Journal of Fluid Mechanics 809 (2016) 895–917. doi:10.1017/jfm.2016.718. URLhttp://dx.doi.org/10.1017/j...

  38. [46]

    Perdikaris, M

    P. Perdikaris, M. Raissi, A. Damianou, N. D. Lawrence, G. E. Karniadakis, Nonlinear information fusion algorithms for data-efficient multi- fidelity modelling, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 473 (2198) (2017) 20160751. doi:1...

  39. [47]

    L. Lu, X. Meng, S. Cai, Z. Mao, S. Goswami, Z. Zhang, G. E. Karniadakis, A comprehensive and fair comparison of two neural op- erators (with practical extensions) based on fair data, Computer Methods in Applied Mechanics and Engineering 393 (2022) 114778. doi:10.1016/j.cma.202...

  40. [48]

    Zakaria, A

    K. Zakaria, A. N. Sinisuka, P. Palar, L. Zuhal, Prediction of smoke-inflow using recursive fourier neural operator dataset (2025). doi:10.5281/ZENODO.15629890. URLhttps://zenodo.org/doi/10.5281/zenodo.15629890

  41. [49]

    P. Holl, N. Thuerey,Φ flow (PhiFlow): Differentiable simulations for pytorch, tensorflow and jax, in: International Conference on Machine Learning, PMLR, 2024

  42. [50]

    Lippe, B

    P. Lippe, B. S. Veeling, P. Perdikaris, R. E. Turner, J. Brandstetter, PDE-Refiner: Achieving Accurate Long Rollouts with Temporal Neural PDE Solvers, in: Thirty-seventh Conference on Neural Information Processing Systems, 2023, p. 1. URLhttps://openreview.net/forum?id=Qv6468llWS

  43. [51]

    Kohl, L.-W

    G. Kohl, L.-W. Chen, N. Thuerey, Benchmarking autoregressive conditional diffusion models for turbulent flow simulation (2024). arXiv:2309.01745. URLhttps://arxiv.org/abs/2309.01745 20

Pith tools

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