Learning Hidden Physics and System Parameters with Deep Operator Networks

Birupaksha Pal; Dibakar Roy Sarkar; Somdatta Goswami; Vijay Kag

arxiv: 2412.05133 · v3 · submitted 2024-12-06 · 💻 cs.LG

Learning Hidden Physics and System Parameters with Deep Operator Networks

Dibakar Roy Sarkar , Vijay Kag , Birupaksha Pal , Somdatta Goswami This is my paper

Pith reviewed 2026-05-23 07:41 UTC · model grok-4.3

classification 💻 cs.LG

keywords deep operator networkshidden physicsparameter identificationinverse modelingpartial differential equationsphysics discoverysparse observations

0 comments

The pith

Deep operator networks can discover unknown PDE terms and identify governing parameters from sparse noisy observations across multiple equation families.

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

The paper introduces two frameworks that extend operator learning to recover hidden physics and system parameters without the retraining demands of prior data-driven approaches. One framework learns mappings of unknown physical operators to handle diverse PDE families, while the other combines a pretrained operator with inverse modeling to extract parameters directly from sensor data. This addresses limitations in existing methods that struggle with noise, sparsity, or lack of generalization. If correct, the work shows that operator-based models can support physics discovery and parameter estimation in settings where full equation knowledge or dense measurements are unavailable.

Core claim

The central claim is that the Deep Hidden Physics Operator identifies mappings of unknown physical operators to discover hidden PDE terms across families of equations, while a second framework that pairs pretrained DeepONet with physics-informed inverse modeling infers system parameters from sparse sensor data; both achieve relative solution errors on the order of 10 to the minus 2 and parameter errors on the order of 10 to the minus 3 on benchmarks including the Reaction-Diffusion system, Burgers' equation, the 2D Heat equation, and the 2D Helmholtz equation, even with limited and noisy observations.

What carries the argument

The Deep Hidden Physics Operator (DHPO), which maps inputs to unknown physical operators to discover hidden terms in PDEs without case-by-case retraining.

If this is right

The same operator can recover unknown terms in multiple distinct PDE families without retraining for each new equation.
Parameter values can be extracted directly from sparse sensor readings once an operator is pretrained.
Accuracy holds when data contain noise levels typical of real measurements.
The approach supplies a single framework that performs both physics discovery and parameter identification.

Where Pith is reading between the lines

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

The method could reduce the data volume needed for inverse problems in engineering systems where only partial sensor coverage is feasible.
It might extend naturally to time-evolving systems by treating the operator as a time-stepping map.
A practical test would involve feeding the learned operator experimental data from a physical device rather than simulated benchmarks.

Load-bearing premise

The learned operator mappings stay accurate and generalize across different PDE families even when observations are sparse and noisy.

What would settle it

Apply the DHPO framework to a PDE family outside the reported benchmarks using only 10 percent of the usual sensor points and check whether solution errors remain below order 10 to the minus 2.

Figures

Figures reproduced from arXiv: 2412.05133 by Birupaksha Pal, Dibakar Roy Sarkar, Somdatta Goswami, Vijay Kag.

**Figure 2.** Figure 2: Schematic representation of the proposed architecture for unknown system parameters identifi [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Performance analysis for Reaction diffusion equation: (a) Mean test error over varying [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗

**Figure 4.** Figure 4: Performance analysis for GRFs and Modified GRFs input function spaces: (a) GRFs basis [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗

**Figure 5.** Figure 5: Reaction diffusion equation: Comparison of reference and predicted results for two representative [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗

**Figure 6.** Figure 6: Reaction diffusion equation: Comparison of reference and predicted results for two representative [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗

**Figure 7.** Figure 7: Reaction diffusion equation: Comparison of reference and predicted results for two representative [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗

**Figure 8.** Figure 8: Reaction Diffusion Equation: Comparison of the reference solution and the predicted solution. [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗

**Figure 9.** Figure 9: Reaction Diffusion Equation: Distribution of absolute error of [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗

**Figure 10.** Figure 10: Performance analysis for Burger’s equation: (a) Mean test error over varying [PITH_FULL_IMAGE:figures/full_fig_p018_10.png] view at source ↗

**Figure 11.** Figure 11: Burgers’ equation: Comparison of reference and predicted results for two representative test [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗

**Figure 12.** Figure 12: Burgers’ Equation: Comparison of the reference solution and the predicted solution. The black [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗

**Figure 13.** Figure 13: Burgers’ Equation: Distribution of absolute error of [PITH_FULL_IMAGE:figures/full_fig_p020_13.png] view at source ↗

read the original abstract

Discovering hidden physical laws and identifying governing system parameters from sparse observations are central challenges in computational science and engineering. Existing data-driven methods, such as physics-informed neural networks (PINNs) and sparse regression, are limited by their need for extensive retraining, sensitivity to noise, or inability to generalize across families of partial differential equations (PDEs). In this work, we introduce two complementary frameworks based on deep operator networks (DeepONet) to address these limitations. The first, termed the Deep Hidden Physics Operator (DHPO), extends hidden-physics modeling into the operator-learning paradigm, enabling the discovery of unknown PDE terms across diverse equation families by identifying the mapping of unknown physical operators. The second is a parameter identification framework that combines pretrained DeepONet with physics-informed inverse modeling to infer system parameters directly from sparse sensor data. We demonstrate the effectiveness of these approaches on benchmark problems, including the Reaction-Diffusion system, Burgers' equation, the 2D Heat equation, and 2D Helmholtz equation. Across all cases, the proposed methods achieve high accuracy, with relative solution errors on the order of O(10^-2) and parameter estimation errors on the order of O(10^-3), even under limited and noisy observations. By uniting operator learning with physics-informed modeling, this work offers a unified and data-efficient framework for physics discovery and parameter identification, paving the way for robust inverse modeling in complex dynamical systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper adapts DeepONet into DHPO for hidden operator discovery and a pretrained inverse pipeline for parameter ID, reporting O(10^-2) solution and O(10^-3) parameter errors on four PDE benchmarks, but supplies almost no experimental controls or comparisons.

read the letter

The colleague should know two things up front. First, the work takes the DeepONet framework and applies it to two inverse tasks: learning unknown PDE terms via an operator called DHPO, and recovering system parameters from sparse noisy sensors using a pretrained DeepONet plus physics-informed loss. Second, the only concrete numbers given are relative errors on the order of 0.01 for the solution fields and 0.001 for the parameters across Reaction-Diffusion, Burgers, 2D heat, and 2D Helmholtz cases. Those are the headline claims; everything else is framing around existing limitations of PINNs and sparse regression. What is actually new is the concrete DHPO formulation that treats the unknown physics as an operator to be learned, plus the two-stage pretrained-DeepONet pipeline for parameter recovery. The paper does a clean job of stating the problem in operator-learning language and of picking standard test problems that let them show results under limited data. The approach follows the usual DeepONet training pattern without obvious internal contradictions or circular fitting. The soft spots sit in the evidence. The abstract states the error orders but gives no baseline numbers against PINNs or other operator methods, no ablation on data density or noise level, no error bars, and no discussion of whether each PDE required its own retraining. Without those controls it is impossible to tell whether the reported accuracies are competitive or simply the result of per-problem tuning. The claim that the method works across PDE families without case-specific retraining is also left unsupported by the material shown. The citation pattern is standard and appropriate for the subfield. This paper is for people already working on operator networks or scientific machine learning who want to see one more combination tried on inverse problems. A reader who needs a new benchmark or a ready-to-adapt code pipeline might extract something useful; anyone looking for a rigorous head-to-head comparison will come away wanting more. I would send it to peer review so the full methods, training details, and experimental tables can be examined.

Referee Report

1 major / 0 minor

Summary. The paper introduces two complementary DeepONet-based frameworks for discovering hidden physical laws and identifying system parameters from sparse and noisy observations. The first is the Deep Hidden Physics Operator (DHPO), which extends hidden-physics modeling into the operator-learning setting to identify mappings of unknown physical operators across PDE families. The second combines a pretrained DeepONet with physics-informed inverse modeling to infer parameters directly from sensor data. Effectiveness is demonstrated on the Reaction-Diffusion system, Burgers' equation, the 2D Heat equation, and the 2D Helmholtz equation, with reported relative solution errors of O(10^{-2}) and parameter estimation errors of O(10^{-3}).

Significance. If the numerical results are supported by proper validation, the work would provide a data-efficient alternative to PINNs and sparse regression for inverse problems by uniting operator learning with physics-informed modeling, potentially improving generalization across PDE families without per-case retraining.

major comments (1)

Abstract: The abstract states that the proposed methods achieve relative solution errors on the order of O(10^{-2}) and parameter estimation errors on the order of O(10^{-3}) across four benchmarks, but supplies no derivation details, baseline comparisons, error-bar analysis, or ablation studies. This absence makes it impossible to verify whether the reported errors support the central claims or result from post-hoc tuning.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive review. We address the single major comment below and propose revisions to strengthen the presentation.

read point-by-point responses

Referee: Abstract: The abstract states that the proposed methods achieve relative solution errors on the order of O(10^{-2}) and parameter estimation errors on the order of O(10^{-3}) across four benchmarks, but supplies no derivation details, baseline comparisons, error-bar analysis, or ablation studies. This absence makes it impossible to verify whether the reported errors support the central claims or result from post-hoc tuning.

Authors: The abstract is intended as a concise high-level summary and therefore omits methodological details that appear in the main text. Section 3.2 derives the DHPO loss and the physics-informed inverse objective; Section 4 reports all numerical experiments, including direct comparisons against PINN and sparse-regression baselines, standard-error bars computed over five independent runs with different random seeds, and ablation tables varying sensor density and noise amplitude. We acknowledge that the abstract could better signpost these validations. We will revise the abstract to add one sentence noting that results are obtained from systematic comparisons and ablations detailed in Sections 3–4. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper introduces two DeepONet-based frameworks (DHPO for hidden physics and a pretrained-operator inverse model for parameters) and validates them empirically on four standard PDE benchmarks, reporting concrete relative errors of O(10^-2) for solutions and O(10^-3) for parameters under sparse/noisy data. No derivation chain is present that reduces a claimed prediction to a fitted input by construction, nor does any load-bearing step rely on a self-citation whose content is itself unverified or tautological. The work is framed as data-driven operator learning plus physics-informed inversion, with results that are externally falsifiable on the cited benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no explicit free parameters, axioms, or invented entities; all modeling choices, network architectures, and loss terms remain unspecified.

pith-pipeline@v0.9.0 · 5794 in / 1078 out tokens · 23507 ms · 2026-05-23T07:41:57.495143+00:00 · methodology

discussion (0)

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Multimodal Neural Operators for Real-Time Biomechanical Modelling of Traumatic Brain Injury
cs.LG 2025-09 unverdicted novelty 5.0

Multimodal neural operators predict full-field brain displacement from MRE data with high accuracy and fast inference by fusing volumetric imaging, demographics, and acquisition parameters.
Harnessing AI for Inverse Partial Differential Equation Problems: Past, Present, and Prospects
cs.AI 2026-05 unverdicted novelty 4.0

A survey organizing AI methods for inverse PDE problems into inverse problems, inverse design, and control categories, covering applications and future challenges like physics-informed models and uncertainty quantification.

Reference graph

Works this paper leans on

46 extracted references · 46 canonical work pages · cited by 2 Pith papers · 1 internal anchor

[1]

Perdikaris, G

P. Perdikaris, G. E. Karniadakis, Model inversion via multi-fidelity Bayesian optimiza- tion: a new paradigm for parameter estimation in haemodynamics, and beyond, Journal of The Royal Society Interface 13 (118) (2016) 20151107

work page 2016
[2]

B. F. Kennedy, P. Wijesinghe, D. D. Sampson, The emergence of optical elastography in biomedicine, Nature Photonics 11 (4) (2017) 215–221

work page 2017
[3]

Gennisson, T

J.-L. Gennisson, T. Deffieux, M. Fink, M. Tanter, Ultrasound elastography: principles and techniques, Diagnostic and interventional imaging 94 (5) (2013) 487–495

work page 2013
[4]

C.-T. Chen, G. X. Gu, Learning hidden elasticity with deep neural networks, Proceed- ings of the National Academy of Sciences 118 (31) (2021) e2102721118

work page 2021
[5]

I. G. Kevrekidis, C. W. Gear, G. Hummer, Equation-free: The computer-aided analysis of complex multiscale systems, AIChE Journal 50 (7) (2004) 1346–1355

work page 2004
[6]

Raissi, Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations, Journal of Machine Learning Research 19 (25) (2018) 1–24

M. Raissi, Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations, Journal of Machine Learning Research 19 (25) (2018) 1–24

work page 2018
[7]

H. J. Motulsky, L. A. Ransnas, Fitting curves to data using nonlinear regression: a practical and nonmathematical review, The FASEB journal 1 (5) (1987) 365–374

work page 1987
[8]

Chang, M

C.-W. Chang, M. Ushio, C.-h. Hsieh, Empirical dynamic modeling for beginners, Eco- logical research 32 (2017) 785–796

work page 2017
[9]

B. C. Daniels, I. Nemenman, Automated adaptive inference of phenomenological dy- namical models, Nature communications 6 (1) (2015) 8133

work page 2015
[10]

Koza, On the programming of computers by means of natural selection, Genetic programming (1992)

J. Koza, On the programming of computers by means of natural selection, Genetic programming (1992)

work page 1992
[11]

S. L. Brunton, M. Budiˇ si´ c, E. Kaiser, J. N. Kutz, Modern koopman theory for dynamical systems, arXiv preprint arXiv:2102.12086 (2021)

work page arXiv 2021
[12]

S. L. Brunton, J. L. Proctor, J. N. Kutz, Discovering governing equations from data by sparse identification of nonlinear dynamical systems, Proceedings of the national academy of sciences 113 (15) (2016) 3932–3937

work page 2016
[13]

Raissi, P

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

work page 2019
[14]

S. Cai, Z. Mao, Z. Wang, M. Yin, G. E. Karniadakis, Physics-informed neural networks (PINNs) for fluid mechanics: a review, Acta Mechanica Sinica 37 (12) (2021)

work page 2021
[15]

W. Wang, T. P. Wong, H. Ruan, S. Goswami, Causality-Respecting Adaptive Refine- ment for PINNs: Enabling Precise Interface Evolution in Phase Field Modeling, arXiv preprint arXiv:2410.20212 (2024). 22

work page arXiv 2024
[16]

Kissas, Y

G. Kissas, Y. Yang, E. Hwuang, W. R. Witschey, J. A. Detre, P. Perdikaris, Machine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non- invasive 4D flow MRI data using physics-informed neural networks, Computer Methods in Applied Mechanics and Engineering 358 (2020) 112623

work page 2020
[17]

Raissi, A

M. Raissi, A. Yazdani, G. E. Karniadakis, Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations, Science 367 (6481) (2020) 1026–1030

work page 2020
[18]

Zhang, H

H. Zhang, H. Wang, Z. Xu, Z. Liu, B. C. Khoo, A physics-informed neural network- based approach to reconstruct the tornado vortices from limited observed data, Journal of Wind Engineering and Industrial Aerodynamics 241 (2023) 105534

work page 2023
[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

work page 2021
[20]

Goswami, A

S. Goswami, A. Bora, Y. Yu, G. E. Karniadakis, Physics-Informed Neural Operators, arXiv preprint arXiv:2207.05748 (2022)

work page arXiv 2022
[21]

Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, A. Anand- kumar, Fourier Neural Operator for Parametric Partial Differential Equations (2021). arXiv:2010.08895

work page internal anchor Pith review Pith/arXiv arXiv 2021
[22]

Tripura, S

T. Tripura, S. Chakraborty, Wavelet Neural Operator for solving parametric partial differential equations in computational mechanics problems, Computer Methods in Ap- plied Mechanics and Engineering 404 (2023) 115783

work page 2023
[23]

Anandkumar, K

A. Anandkumar, K. Azizzadenesheli, K. Bhattacharya, N. Kovachki, Z. Li, B. Liu, A. Stuart, Neural operator: Graph kernel network for partial differential equations, in: ICLR 2020 Workshop on Integration of Deep Neural Models and Differential Equations, 2020

work page 2020
[24]

Raonic, R

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

work page 2023
[25]

Q. Cao, S. Goswami, G. E. Karniadakis, Laplace neural operator for solving differential equations, Nature Machine Intelligence 6 (6) (2024) 631–640

work page 2024
[26]

Ingebrand, A

T. Ingebrand, A. J. Thorpe, S. Goswami, K. Kumar, U. Topcu, Basis-to-basis operator learning using function encoders, arXiv preprint arXiv:2410.00171 (2024)

work page arXiv 2024
[27]

Bahmani, S

B. Bahmani, S. Goswami, I. G. Kevrekidis, M. D. Shields, A resolution independent neural operator, arXiv preprint arXiv:2407.13010 (2024)

work page arXiv 2024
[28]

T. Chen, H. Chen, Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems, IEEE transactions on neural networks 6 (4) (1995) 911–917. 23

work page 1995
[29]

P. C. Di Leoni, L. Lu, C. Meneveau, G. Karniadakis, T. A. Zaki, DeepONet prediction of linear instability waves in high-speed boundary layers, arXiv preprint arXiv:2105.08697 (2021)

work page arXiv 2021
[30]

Borrel-Jensen, S

N. Borrel-Jensen, S. Goswami, A. P. Engsig-Karup, G. E. Karniadakis, C.-H. Jeong, Sound propagation in realistic interactive 3D scenes with parameterized sources using deep neural operators, Proceedings of the National Academy of Sciences 121 (2) (2024) e2312159120

work page 2024
[31]

Kontolati, S

K. Kontolati, S. Goswami, M. D. Shields, G. E. Karniadakis, On the influence of over- parameterization in manifold based surrogates and deep neural operators, Journal of Computational Physics (2023) 112008

work page 2023
[32]

Q. Cao, S. Goswami, G. E. Karniadakis, S. Chakraborty, Deep neural operators can predict the real-time response of floating offshore structures under irregular waves, arXiv preprint arXiv:2302.06667 (2023)

work page arXiv 2023
[33]

P. Jin, S. Meng, L. Lu, MIONet: Learning multiple-input operators via tensor product, arXiv preprint arXiv:2202.06137 (2022)

work page arXiv 2022
[34]

Goswami, D

S. Goswami, D. S. Li, B. V. Rego, M. Latorre, J. D. Humphrey, G. E. Karniadakis, Neural operator learning of heterogeneous mechanobiological insults contributing to aortic aneurysms, Journal of the Royal Society Interface 19 (193) (2022) 20220410

work page 2022
[35]

Zhang, A

E. Zhang, A. Kahana, E. Turkel, R. Ranade, J. Pathak, G. E. Karniadakis, A Hybrid Iterative Numerical Transferable Solver (HINTS) for PDEs Based on Deep Operator Network and Relaxation Methods, arXiv preprint arXiv:2208.13273 (2022)

work page arXiv 2022
[36]

Kahana, E

A. Kahana, E. Zhang, S. Goswami, G. Karniadakis, R. Ranade, J. Pathak, On the ge- ometry transferability of the hybrid iterative numerical solver for differential equations, Computational Mechanics 72 (3) (2023) 471–484

work page 2023
[37]

Goswami, K

S. Goswami, K. Kontolati, M. D. Shields, G. E. Karniadakis, Deep transfer operator learning for partial differential equations under conditional shift, Nature Machine Intel- ligence (2022) 1–10

work page 2022
[38]

S. Wang, H. Wang, P. Perdikaris, Learning the solution operator of parametric partial differential equations with physics-informed DeepONets, Science advances 7 (40) (2021) eabi8605

work page 2021
[39]

Mandl, S

L. Mandl, S. Goswami, L. Lambers, T. Ricken, Separable deeponet: Breaking the curse of dimensionality in physics-informed machine learning, arXiv preprint arXiv:2407.15887 (2024)

work page arXiv 2024
[40]

Kumar, S

V. Kumar, S. Goswami, K. Kontolati, M. D. Shields, G. E. Karniadakis, Synergis- tic learning with multi-task deeponet for efficient pde problem solving, arXiv preprint arXiv:2408.02198 (2024). 24

work page arXiv 2024
[41]

Karumuri, L

S. Karumuri, L. Graham-Brady, S. Goswami, Efficient training of deep neural operator networks via randomized sampling, arXiv preprint arXiv:2409.13280 (2024)

work page arXiv 2024
[42]

J. He, S. Koric, D. Abueidda, A. Najafi, I. Jasiuk, Geom-deeponet: A point-cloud-based deep operator network for field predictions on 3d parameterized geometries, Computer Methods in Applied Mechanics and Engineering 429 (2024) 117130

work page 2024
[43]

Kontolati, S

K. Kontolati, S. Goswami, G. Em Karniadakis, M. D. Shields, Learning nonlinear oper- ators in latent spaces for real-time predictions of complex dynamics in physical systems, Nature Communications 15 (1) (2024) 5101

work page 2024
[44]

Smoller, Shock waves and reaction diffusion equations, 2nd Edition, Springer Science, Bussiness media LLC, 2012

J. Smoller, Shock waves and reaction diffusion equations, 2nd Edition, Springer Science, Bussiness media LLC, 2012

work page 2012
[45]

M. D. McKay, R. J. Beckman, W. J. Conover, A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code, Technometrics 21 (2) (1979) 239–245

work page 1979
[46]

D. R. Sarkar, C. Annavarapu, P. Roy, Adaptive Interface-Pinns (Adai-Pinns) for Inverse Problems: Determining Material Properties for Heterogeneous Systems, Available at SSRN 4993297. Appendix A. Hyperparameter tuning In the deep learning community, Weights & Biases (WandB) is a popular online platform that provides tools for monitoring and controlling hyp...

work page

[1] [1]

Perdikaris, G

P. Perdikaris, G. E. Karniadakis, Model inversion via multi-fidelity Bayesian optimiza- tion: a new paradigm for parameter estimation in haemodynamics, and beyond, Journal of The Royal Society Interface 13 (118) (2016) 20151107

work page 2016

[2] [2]

B. F. Kennedy, P. Wijesinghe, D. D. Sampson, The emergence of optical elastography in biomedicine, Nature Photonics 11 (4) (2017) 215–221

work page 2017

[3] [3]

Gennisson, T

J.-L. Gennisson, T. Deffieux, M. Fink, M. Tanter, Ultrasound elastography: principles and techniques, Diagnostic and interventional imaging 94 (5) (2013) 487–495

work page 2013

[4] [4]

C.-T. Chen, G. X. Gu, Learning hidden elasticity with deep neural networks, Proceed- ings of the National Academy of Sciences 118 (31) (2021) e2102721118

work page 2021

[5] [5]

I. G. Kevrekidis, C. W. Gear, G. Hummer, Equation-free: The computer-aided analysis of complex multiscale systems, AIChE Journal 50 (7) (2004) 1346–1355

work page 2004

[6] [6]

Raissi, Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations, Journal of Machine Learning Research 19 (25) (2018) 1–24

M. Raissi, Deep Hidden Physics Models: Deep Learning of Nonlinear Partial Differential Equations, Journal of Machine Learning Research 19 (25) (2018) 1–24

work page 2018

[7] [7]

H. J. Motulsky, L. A. Ransnas, Fitting curves to data using nonlinear regression: a practical and nonmathematical review, The FASEB journal 1 (5) (1987) 365–374

work page 1987

[8] [8]

Chang, M

C.-W. Chang, M. Ushio, C.-h. Hsieh, Empirical dynamic modeling for beginners, Eco- logical research 32 (2017) 785–796

work page 2017

[9] [9]

B. C. Daniels, I. Nemenman, Automated adaptive inference of phenomenological dy- namical models, Nature communications 6 (1) (2015) 8133

work page 2015

[10] [10]

Koza, On the programming of computers by means of natural selection, Genetic programming (1992)

J. Koza, On the programming of computers by means of natural selection, Genetic programming (1992)

work page 1992

[11] [11]

S. L. Brunton, M. Budiˇ si´ c, E. Kaiser, J. N. Kutz, Modern koopman theory for dynamical systems, arXiv preprint arXiv:2102.12086 (2021)

work page arXiv 2021

[12] [12]

S. L. Brunton, J. L. Proctor, J. N. Kutz, Discovering governing equations from data by sparse identification of nonlinear dynamical systems, Proceedings of the national academy of sciences 113 (15) (2016) 3932–3937

work page 2016

[13] [13]

Raissi, P

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

work page 2019

[14] [14]

S. Cai, Z. Mao, Z. Wang, M. Yin, G. E. Karniadakis, Physics-informed neural networks (PINNs) for fluid mechanics: a review, Acta Mechanica Sinica 37 (12) (2021)

work page 2021

[15] [15]

W. Wang, T. P. Wong, H. Ruan, S. Goswami, Causality-Respecting Adaptive Refine- ment for PINNs: Enabling Precise Interface Evolution in Phase Field Modeling, arXiv preprint arXiv:2410.20212 (2024). 22

work page arXiv 2024

[16] [16]

Kissas, Y

G. Kissas, Y. Yang, E. Hwuang, W. R. Witschey, J. A. Detre, P. Perdikaris, Machine learning in cardiovascular flows modeling: Predicting arterial blood pressure from non- invasive 4D flow MRI data using physics-informed neural networks, Computer Methods in Applied Mechanics and Engineering 358 (2020) 112623

work page 2020

[17] [17]

Raissi, A

M. Raissi, A. Yazdani, G. E. Karniadakis, Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations, Science 367 (6481) (2020) 1026–1030

work page 2020

[18] [18]

Zhang, H

H. Zhang, H. Wang, Z. Xu, Z. Liu, B. C. Khoo, A physics-informed neural network- based approach to reconstruct the tornado vortices from limited observed data, Journal of Wind Engineering and Industrial Aerodynamics 241 (2023) 105534

work page 2023

[19] [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

work page 2021

[20] [20]

Goswami, A

S. Goswami, A. Bora, Y. Yu, G. E. Karniadakis, Physics-Informed Neural Operators, arXiv preprint arXiv:2207.05748 (2022)

work page arXiv 2022

[21] [21]

Z. Li, N. Kovachki, K. Azizzadenesheli, B. Liu, K. Bhattacharya, A. Stuart, A. Anand- kumar, Fourier Neural Operator for Parametric Partial Differential Equations (2021). arXiv:2010.08895

work page internal anchor Pith review Pith/arXiv arXiv 2021

[22] [22]

Tripura, S

T. Tripura, S. Chakraborty, Wavelet Neural Operator for solving parametric partial differential equations in computational mechanics problems, Computer Methods in Ap- plied Mechanics and Engineering 404 (2023) 115783

work page 2023

[23] [23]

Anandkumar, K

A. Anandkumar, K. Azizzadenesheli, K. Bhattacharya, N. Kovachki, Z. Li, B. Liu, A. Stuart, Neural operator: Graph kernel network for partial differential equations, in: ICLR 2020 Workshop on Integration of Deep Neural Models and Differential Equations, 2020

work page 2020

[24] [24]

Raonic, R

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

work page 2023

[25] [25]

Q. Cao, S. Goswami, G. E. Karniadakis, Laplace neural operator for solving differential equations, Nature Machine Intelligence 6 (6) (2024) 631–640

work page 2024

[26] [26]

Ingebrand, A

T. Ingebrand, A. J. Thorpe, S. Goswami, K. Kumar, U. Topcu, Basis-to-basis operator learning using function encoders, arXiv preprint arXiv:2410.00171 (2024)

work page arXiv 2024

[27] [27]

Bahmani, S

B. Bahmani, S. Goswami, I. G. Kevrekidis, M. D. Shields, A resolution independent neural operator, arXiv preprint arXiv:2407.13010 (2024)

work page arXiv 2024

[28] [28]

T. Chen, H. Chen, Universal approximation to nonlinear operators by neural networks with arbitrary activation functions and its application to dynamical systems, IEEE transactions on neural networks 6 (4) (1995) 911–917. 23

work page 1995

[29] [29]

P. C. Di Leoni, L. Lu, C. Meneveau, G. Karniadakis, T. A. Zaki, DeepONet prediction of linear instability waves in high-speed boundary layers, arXiv preprint arXiv:2105.08697 (2021)

work page arXiv 2021

[30] [30]

Borrel-Jensen, S

N. Borrel-Jensen, S. Goswami, A. P. Engsig-Karup, G. E. Karniadakis, C.-H. Jeong, Sound propagation in realistic interactive 3D scenes with parameterized sources using deep neural operators, Proceedings of the National Academy of Sciences 121 (2) (2024) e2312159120

work page 2024

[31] [31]

Kontolati, S

K. Kontolati, S. Goswami, M. D. Shields, G. E. Karniadakis, On the influence of over- parameterization in manifold based surrogates and deep neural operators, Journal of Computational Physics (2023) 112008

work page 2023

[32] [32]

Q. Cao, S. Goswami, G. E. Karniadakis, S. Chakraborty, Deep neural operators can predict the real-time response of floating offshore structures under irregular waves, arXiv preprint arXiv:2302.06667 (2023)

work page arXiv 2023

[33] [33]

P. Jin, S. Meng, L. Lu, MIONet: Learning multiple-input operators via tensor product, arXiv preprint arXiv:2202.06137 (2022)

work page arXiv 2022

[34] [34]

Goswami, D

S. Goswami, D. S. Li, B. V. Rego, M. Latorre, J. D. Humphrey, G. E. Karniadakis, Neural operator learning of heterogeneous mechanobiological insults contributing to aortic aneurysms, Journal of the Royal Society Interface 19 (193) (2022) 20220410

work page 2022

[35] [35]

Zhang, A

E. Zhang, A. Kahana, E. Turkel, R. Ranade, J. Pathak, G. E. Karniadakis, A Hybrid Iterative Numerical Transferable Solver (HINTS) for PDEs Based on Deep Operator Network and Relaxation Methods, arXiv preprint arXiv:2208.13273 (2022)

work page arXiv 2022

[36] [36]

Kahana, E

A. Kahana, E. Zhang, S. Goswami, G. Karniadakis, R. Ranade, J. Pathak, On the ge- ometry transferability of the hybrid iterative numerical solver for differential equations, Computational Mechanics 72 (3) (2023) 471–484

work page 2023

[37] [37]

Goswami, K

S. Goswami, K. Kontolati, M. D. Shields, G. E. Karniadakis, Deep transfer operator learning for partial differential equations under conditional shift, Nature Machine Intel- ligence (2022) 1–10

work page 2022

[38] [38]

S. Wang, H. Wang, P. Perdikaris, Learning the solution operator of parametric partial differential equations with physics-informed DeepONets, Science advances 7 (40) (2021) eabi8605

work page 2021

[39] [39]

Mandl, S

L. Mandl, S. Goswami, L. Lambers, T. Ricken, Separable deeponet: Breaking the curse of dimensionality in physics-informed machine learning, arXiv preprint arXiv:2407.15887 (2024)

work page arXiv 2024

[40] [40]

Kumar, S

V. Kumar, S. Goswami, K. Kontolati, M. D. Shields, G. E. Karniadakis, Synergis- tic learning with multi-task deeponet for efficient pde problem solving, arXiv preprint arXiv:2408.02198 (2024). 24

work page arXiv 2024

[41] [41]

Karumuri, L

S. Karumuri, L. Graham-Brady, S. Goswami, Efficient training of deep neural operator networks via randomized sampling, arXiv preprint arXiv:2409.13280 (2024)

work page arXiv 2024

[42] [42]

J. He, S. Koric, D. Abueidda, A. Najafi, I. Jasiuk, Geom-deeponet: A point-cloud-based deep operator network for field predictions on 3d parameterized geometries, Computer Methods in Applied Mechanics and Engineering 429 (2024) 117130

work page 2024

[43] [43]

Kontolati, S

K. Kontolati, S. Goswami, G. Em Karniadakis, M. D. Shields, Learning nonlinear oper- ators in latent spaces for real-time predictions of complex dynamics in physical systems, Nature Communications 15 (1) (2024) 5101

work page 2024

[44] [44]

Smoller, Shock waves and reaction diffusion equations, 2nd Edition, Springer Science, Bussiness media LLC, 2012

J. Smoller, Shock waves and reaction diffusion equations, 2nd Edition, Springer Science, Bussiness media LLC, 2012

work page 2012

[45] [45]

M. D. McKay, R. J. Beckman, W. J. Conover, A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code, Technometrics 21 (2) (1979) 239–245

work page 1979

[46] [46]

D. R. Sarkar, C. Annavarapu, P. Roy, Adaptive Interface-Pinns (Adai-Pinns) for Inverse Problems: Determining Material Properties for Heterogeneous Systems, Available at SSRN 4993297. Appendix A. Hyperparameter tuning In the deep learning community, Weights & Biases (WandB) is a popular online platform that provides tools for monitoring and controlling hyp...

work page