Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

PDE-DKL: PDE-constrained deep kernel learning in high dimensionality

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read PDE-DKL claims that a neural network can compress high-dimensional PDE coordinates into a low-dimensional latent space where a PDE-constrained Gaussian process delivers accurate solutions and uncertainty estimates with sparse data.

desk verdict The math is sound, but the high-dimensional benchmarks are effectively one-dimensional ridge functions, so the curse-of-dimensionality claim is not yet supported. read the letter →

arxiv 2501.18258 v1 pith:7BQHUTBA submitted 2025-01-30 cs.LG cs.AIstat.ML

Weihao Yan , Christoph Brune , Mengwu Guo This is my paper · ORCID
classification cs.LGcs.AIstat.ML
keywords deepkernellearningGaussianprocessregressionphysics-informedmachinehigh-dimensionalPDEsuncertaintyquantificationlatentrepresentationPDE-constrainedprocesseslimiteddata
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

The paper proposes PDE-DKL, a solver that pairs a neural network with Gaussian process regression for high-dimensional linear PDEs under limited data. The network maps the PDE coordinates, including space, time, and parameters, into a low-dimensional latent space, and a squared-exponential GP kernel operates there while the governing PDE enters through the covariance via the linear operator acting on the GP prior. The central claim is that this hybrid avoids the curse of dimensionality: reported relative $L^2$ errors stay below about five percent for Poisson, heat, and advection-diffusion-reaction equations in 50 dimensions, where a plain PDE-constrained GP cannot run. A sympathetic reader would care because the method offers physically consistent predictions together with principled posterior uncertainties without requiring large training sets.

What carries the argument

The central object is the deep kernel $k_{\mathrm{DKL}}(q,q') = \sigma^2 \exp\!\left(-\tfrac12 (h_L(q;\omega)-h_L(q';\omega))^\top \mathrm{diag}(\ell^2)^{-1}(h_L(q;\omega)-h_L(q';\omega))\right)$, a squared-exponential ARD kernel defined not on the raw coordinates $q$ but on the output $h_L(q;\omega)$ of an $L$-layer neural network. The network performs the dimensionality reduction that GP regression alone cannot, and the ARD lengthscales act as a learned sensitivity analysis over the latent variables. Applying the linear operator $\mathcal{A}$ to the GP prior produces cross-covariance terms such as $\mathcal{A} k(q,q')\, \mathcal{A}^\top$, which is how the PDE constraint enters the covariance, and training then minimizes the negative log marginal likelihood over network weights, kernel hyperparameters, and noise variances, with Bayesian optimisation selecting the latent dimension and a PDE-informed projection of forcing data onto solution data providing the initialization.

What would settle it

Solve a 50-dimensional linear PDE whose exact solution genuinely involves all coordinates, for example $u(x) = \prod_{i=1}^{50} \sin(\pi x_i)$, with the same sparse data budget; if the relative $L^2$ error climbs well above the sub-five-percent range reported here, the curse of dimensionality has not actually been avoided.

Watch

Extended reading notes

Core claim

PDE-DKL's central discovery is that composing a PDE-constrained Gaussian process with a neural-network feature map shifts the burden of high dimensions onto a learnable low-dimensional representation. Because the differential operator $\mathcal{A}$ is linear, a GP prior on the solution $u$ induces a GP prior on the forcing term $f = \mathcal{A}[u]$, and the joint covariance can be assembled from the deep kernel and its derivatives; conditioning on both solution data and forcing data yields a posterior GP that satisfies the PDE by construction. In the numerical benchmarks this delivers relative $L^2$ errors between $0.29\%$ and $3.38\%$ for $u$ and between $0.82\%$ and $4.27\%$ for $f$ across ten- and fifty-dimensional problems, while the corresponding PDE-GP baseline becomes computationally infeasible in fifty dimensions. The paper presents this as evidence that neural-network latent representations and physics-constrained GPs are complementary tools for sparse high-dimensional data.

Load-bearing premise

The method assumes the PDE solution can be compressed into a low-dimensional latent representation learned by the neural network; every benchmark solution actually depends on the coordinates only through their sum, $1^\top x$ or $1^\top x/d$, so the true effective dimension of all test problems is one.

Editorial extensions

If this is right

  • For linear PDEs whose solutions have low-dimensional structure, PDE-DKL can hold relative $L^2$ error below roughly five percent in 50 input dimensions using limited training data, whereas PDE-GP cannot even be applied at that scale.
  • The fitted posterior mean and variance are PDE-consistent, so the uncertainty estimates inherit the physical constraints rather than being added post hoc.
  • Because the GP operates on the low-dimensional latent space rather than raw high-dimensional inputs, the method reduces the memory and runtime cost of constructing and inverting covariance matrices.
  • The reported comparisons indicate that the hybrid NN-GP architecture is a practical candidate when both scarce data and uncertainty quantification matter for solving forward PDE problems.

Reading between the lines

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

  • An untested extension would feed the posterior variance back into experimental design, using the GP uncertainty to choose where to acquire solution or forcing data next.
  • A natural stress test is a solution that genuinely depends on all coordinates, such as a product of sinusoids; the paper's benchmarks all depend on the coordinates through a single sum $1^\top x$, so the advertised 50-dimensional accuracy has not been shown to survive when the latent structure is nonlinear or higher-dimensional.
  • Extending the same PDE-constraint guarantee to nonlinear PDEs would require linearization or another approximation step, since the argument relies on Gaussianity being preserved under the linear operator $\mathcal{A}$.
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 / 5 minor

Summary. The paper proposes PDE-DKL, a hybrid Gaussian-process/deep-learning method for linear PDEs. A neural network maps the high-dimensional coordinate q to a low-dimensional latent representation h_L(q); an ARD squared-exponential kernel on h_L is used as a GP prior for the solution u, and the linear PDE operator A is applied to obtain a GP prior for the forcing term f. Conditioning on observations of both u and f yields a posterior GP whose mean and covariance satisfy A m* = m*_A and A k* A^T = k*_A, so the PDE constraint is preserved. Training minimizes the negative log marginal likelihood over kernel hyperparameters, network weights, and noise variances, with Bayesian optimization for the latent dimension and a PDE-aware initialization. The method is tested on a parametric heat equation and on Poisson, heat, and advection-diffusion-reaction equations in nominal dimensions 10 and 50, and compared with a PDE-constrained GP baseline.

Significance. If the reported behavior is representative, the framework is a useful combination of existing ingredients: NN-based dimensionality reduction, deep kernel learning, and linearly constrained GPs. The derivation in Section 2 is standard and correct, and Eq (12) is a nice explicit check that the posterior pair (u*, f*) is consistent with A[u]=f. The paper also makes a falsifiable empirical claim and provides a code repository link. However, the high-dimensional benchmarks use target functions that are ridge functions of a single linear functional of x (plus time), so the experiments do not yet establish the paper's central claim of overcoming the curse of dimensionality. The single-run results in Table 1, the uncalibrated UQ statements, and the inconsistency between Eq (20) and Algorithm 2 further weaken the empirical and reproducibility support. With additional benchmarks and a clarified algorithm, the work could be a solid contribution.

major comments (4)
  1. [§4.2–§4.4 and Table 1] All high-dimensional benchmark solutions depend on at most one spatial linear functional: the Poisson solution is sin(1^T x)+cos(1^T x), the heat solution is e^{-t} cos(1^T x/d), and the advection-diffusion-reaction solution is e^{-t} sin(1^T x/d). Thus the nominal d=10 and d=50 tests have effective input dimension at most 2 (time plus one spatial direction), and the Poisson case has effective dimension 1. These experiments show that the neural network can fit a ridge function and that GP regression on the learned latent coordinate works, but they do not demonstrate that PDE-DKL overcomes the curse of dimensionality for solutions that genuinely depend on many coordinates. The statement in Section 3.1 that 'when n_h is sufficiently small, the GP regression will have guaranteed performance' presumes the NN has learned the true low-dimensional coordinate; that presumption is not tested by ridge benchmarks. At least one d=10 or d=50 problem with genuinely multi-coordinate dependence (e.g., a product or sum of several active directions) is needed to support the abstract's scalability claim.
  2. [§3.3, Eq (20) and Algorithm 2 line 3] The initialization formula in the text and the algorithm do not agree. Equation (20) defines tilde y_{f->u} = (A_{q'} k(Q_f,Q_f))(A_q A_{q'} k(Q_f,Q_f))^{-1} y_f, which with the notation of the paper is (k(Q_f,Q_f)A^T)(A k(Q_f,Q_f)A^T)^{-1} y_f. Algorithm 2 line 3 uses (k(Q_f,Q_f) A^T)^{-1}(A k(Q_f,Q_f) A^T) y_f. These products are not equal in general, and the algorithm's version is not well-defined when k(Q_f,Q_f)A^T is not square or not invertible. Since the proposed initialization is claimed to be an important ingredient of the method, the intended formula must be specified unambiguously and implemented consistently.
  3. [Table 1 and Section 4] Table 1 reports a single relative L2 error for each method and problem, with no repetitions, error bars, or training-data sizes. The experiments involve random sampling of collocation/data points, Bayesian optimization over the latent dimension, and stochastic gradient training, so single runs do not establish that PDE-DKL is consistently more accurate than PDE-GP. At a minimum, a few independent restarts with mean and standard deviation should be reported for the headline numbers, along with the number of training points used in each problem, since the abstract claims 'reduced data requirements'.
  4. [§4.1–§4.5] The paper repeatedly claims 'reliable uncertainty quantification' (e.g., §4.2, §4.3, §4.4 and Figs. 3–8), but no quantitative validation of the posterior uncertainty is given. Posterior standard deviations are displayed as error bars, yet there is no empirical coverage check, calibration curve, or interval score against the known ground-truth solutions. Because uncertainty quantification is one of the two central advertised benefits, the authors should report, for example, the empirical coverage of the 95% predictive intervals on held-out test points.
minor comments (5)
  1. [§3.3 / Algorithm 2] The symbol A^T is used both for the operator acting on the second kernel argument and as a matrix transpose; these two uses should be distinguished, especially because the inconsistency between Eq (20) and Algorithm 2 partly arises from this ambiguity.
  2. [Eq (18)] There is an extra comma inside the determinant term log |K_DKL(theta, omega, sigma_u^2, sigma_f^2, )|; the trailing comma should be removed.
  3. [§3.3 and §4.4] There are dangling cross-references: 'Algorithm ??' appears in the text after Eq (18), and Section 4.4 refers to '10-dimensional (7) and 50-dimensional (8)' without matching numbered equations or figures.
  4. [Algorithm 1] Algorithm 1 computes the marginal likelihood objective using k(Z_u,Z_f) and k(Z_f,Z_f) without applying the PDE operator A, whereas the actual training loss (18) uses the A-transformed kernels; the text should clarify whether the Bayesian optimization step is intentionally PDE-free, since a PDE-constrained objective would be the consistent choice.
  5. [§3.2] The paragraph on 'specialised algorithms' that exploit sparsity and parallelisation promises implementation details, but no such details are given in the paper; either specify them or soften the claim.

Circularity Check

1 steps flagged · score 1.0 of 10

Only by-construction element is f* = A u*; central high-dimensional accuracy claim rests on reported u-errors and is not circular.

  1. self definitional [Section 2.3, Eq. (12); numerical e_f metrics in Table 1]
    "It is evident that m∗A(q) = Am∗(q) , and k∗A(q, q′) = Ak∗(q, q′)A⊤ , implying that the pair of solved u and reconstructed f, both being GPs, satisfies the constraint of linear transformation (5) through A. Such a constraint is preserved throughout the GP surrogate modelling from the prior setting (7) to the posterior prediction (12), confirming the PDE-constrained nature of this approximation method."

    The reconstructed f posterior is not an independent prediction: it is obtained by applying the linear PDE operator A to the posterior of u (Eqs. (10)-(11)). Eq. (12) therefore holds as an algebraic identity by construction, so the sentence claiming that the pair 'confirms' the PDE-constrained nature is a restatement of the linear-GP construction rather than an empirical finding. The e_f entries in Table 1 are likewise deterministic images of the u-posterior errors under A, so they add no independent evidence. This step is minor because the main u-accuracy comparisons are reported against analytical test values and do not reduce to this identity.

full rationale

No load-bearing self-citation, imported uniqueness, or fitted-input-as-prediction pattern was found: the deep kernel is assembled from cited manifold-GP/DKL constructions, the PDE operator is encoded through the exact linear-GP conditional (Eqs. (9)-(15)), and all trainable parameters are fit by NLML before evaluating reported u-errors. The only by-construction relation is that the f-posterior is A applied to the u-posterior, which is a tautological consistency check rather than an independent validation. A legitimate non-circularity concern is that all d=10/50 test solutions are ridge functions (sin(1^T x), e^{-t} cos(1^T x/d), e^{-t} sin(1^T x/d)), so the effective input dimension of the benchmarks is at most 2; that is a benchmark-hardness and correctness issue, not a circularity issue. The score therefore reflects only the minor self-definitional f-reconstruction metric.

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

The framework introduces no new physical entities or forces. It relies on fitted kernel, network, noise, and latent-dimension parameters, and on the assumption that the solution has learnable low-dimensional structure. The mathematical derivation of the PDE-constrained GP is standard and sound; the empirical claims depend on these free parameters and assumptions.

free parameters (4)
  • Neural network weights and biases (omega) = not reported
    Trained by minimizing the negative log marginal likelihood (NLML) to learn the latent representation; values are not listed in the paper.
  • Kernel hyperparameters (sigma^2, lengthscales ell) = not reported
    ARD kernel variance and lengthscales chosen by empirical Bayes through NLML optimization; exact fitted values are not given.
  • Observation noise variances (sigma^2_u, sigma^2_f) = not reported
    Optimized together with the kernel parameters to model noise and improve conditioning of the covariance matrix.
  • Latent dimension n_h^L = varies per problem, not reported
    Selected by Bayesian optimization using the same NLML objective, so it is tuned on the training data rather than chosen a priori.
assumptions (4)
  • standard math Gaussianity is preserved under linear operators applied to GPs.
    Used in Section 2.3 to derive the GP prior for f = A[u] from the GP prior for u.
  • domain assumption The differential operator A in Eq (4) is linear and the kernel k_DKL is sufficiently differentiable to compute A k A^T.
    Needed to define the joint covariance K_DKL in Eq (15); only linear PDEs are considered.
  • ad hoc to paper The PDE solution is well approximated by a GP with a deep kernel whose latent representation is low-dimensional.
    Central modeling assumption in Section 3.1; if the solution lacks such low-dimensional structure, the method's accuracy claim is unsupported.
  • domain assumption Optimizing the negative log marginal likelihood yields a well-generalizing model.
    Used for all parameter training in Section 3.3; the paper itself cites [53] noting NLML can overfit and be prior-sensitive.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PDE-DKL: PDE-constrained deep kernel learning in high dimensionality." pith.science (2026). https://pith.science/paper/7BQHUTBA

@misc{pith2026250118258,
  author       = {Pith},
  title        = {Pith review of: PDE-DKL: PDE-constrained deep kernel learning in high dimensionality},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7BQHUTBA}},
  note         = {Machine review of arXiv:2501.18258}
}
read the original abstract

Many physics-informed machine learning methods for PDE-based problems rely on Gaussian processes (GPs) or neural networks (NNs). However, both face limitations when data are scarce and the dimensionality is high. Although GPs are known for their robust uncertainty quantification in low-dimensional settings, their computational complexity becomes prohibitive as the dimensionality increases. In contrast, while conventional NNs can accommodate high-dimensional input, they often require extensive training data and do not offer uncertainty quantification. To address these challenges, we propose a PDE-constrained Deep Kernel Learning (PDE-DKL) framework that combines DL and GPs under explicit PDE constraints. Specifically, NNs learn a low-dimensional latent representation of the high-dimensional PDE problem, reducing the complexity of the problem. GPs then perform kernel regression subject to the governing PDEs, ensuring accurate solutions and principled uncertainty quantification, even when available data are limited. This synergy unifies the strengths of both NNs and GPs, yielding high accuracy, robust uncertainty estimates, and computational efficiency for high-dimensional PDEs. Numerical experiments demonstrate that PDE-DKL achieves high accuracy with reduced data requirements. They highlight its potential as a practical, reliable, and scalable solver for complex PDE-based applications in science and engineering.

Figures

Figures reproduced from arXiv: 2501.18258 by the authors.

Figure 1
Figure 1. Conceptual diagram of PDE-constrained deep kernel learning (PDE-DKL): The initialisation training involves [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Comparison of PDE-GP (black dots) and PDE-DKL (blue dots) in predicting the solution term [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗
Figure 3
Figure 3. Comparison of PDE-GP and PDE-DKL methods in solving the ten-dimensional Poisson equation. Subplots [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Accuracy of PDE-DKL and PDE-GP in solving high-dimensional Poisson equations. Subfigure ( [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Subfigures ( [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Predictive accuracy of PDE-DKL versus PDE-GP in solving the high-dimensional heat equation. Subfigure [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]
Figure 7
Figure 7. Figure 7: Comparison of PDE-GP and PDE-DKL in solving the ten-dimensional advection-diffusion-reaction equation. [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Solution fidelity of PDE-DKL versus PDE-GP for the high-dimensional advection-diffusion-reaction equation. [PITH_FULL_IMAGE:figures/full_fig_p017_8.png]
Figure 9
Figure 9. Figure 9: The figure presents contour maps of predicted values for high-dimensional, 50D advection-diffusion-reaction [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. A unified perspective of Gaussian process approximation for differential equations

    math.NA 2026-07 accept novelty 6.0 of 10

    A unified Bayesian framework based on derivative matching shows that diverse Gaussian process methods for differential equations are instances of a common probabilistic structure.

Reference graph

Works this paper leans on

54 extracted references · 39 canonical work pages · cited by 1 Pith paper

  1. [1]

    Finite element method—the natural approach

    JH Argyris, H Balmer, J St Doltsinis, PC Dunne, M Haase, M Kleiber, GA Malejannakis, H-P Mlejnek, M Müller, and DW Scharpf. Finite element method—the natural approach. Computer Methods in Applied Mechanics and Engineering, 17:1–106, 1979

    work page 1979

  2. [2]

    Finite difference method for numerical computation of discontinuous solutions of the equations of fluid dynamics

    Sergei K Godunov and I Bohachevsky. Finite difference method for numerical computation of discontinuous solutions of the equations of fluid dynamics. Matematiˇ ceskij Sbornik, 47(3):271–306, 1959

    work page 1959

  3. [3]

    Finite volume methods

    Robert Eymard, Thierry Gallouët, and Raphaèle Herbin. Finite volume methods. Handbook of Numerical Analysis, 7:713–1018, 2000

    work page 2000

  4. [4]

    Numerical Analysis of Spectral Methods: Theory and Applications

    David Gottlieb and Steven A Orszag. Numerical Analysis of Spectral Methods: Theory and Applications. SIAM, 1977

    work page 1977

  5. [5]

    Approximation capabilities of multilayer feedforward networks

    Kurt Hornik. Approximation capabilities of multilayer feedforward networks. Neural Networks, 4(2):251–257, 1991

    work page 1991

  6. [6]

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations

    Maziar Raissi, Paris Perdikaris, and George 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, 378:686–707, 2019. 19 PDE-DKL: PDE-constrained deep kernel learning in high dimensionality

    work page 2019

  7. [7]

    Physics- informed machine learning

    George Em Karniadakis, Ioannis G Kevrekidis, Lu Lu, Paris Perdikaris, Sifan Wang, and Liu Yang. Physics- informed machine learning. Nature Reviews Physics, 3(6):422–440, 2021

    2021

  8. [8]

    Physics-informed deep learning for computational elastodynamics without labeled data

    Chengping Rao, Hao Sun, and Yang Liu. Physics-informed deep learning for computational elastodynamics without labeled data. Journal of Engineering Mechanics, 147(8):04021043, 2021

    work page 2021

Show all 54 references
  1. [9]

    fPINNs: Fractional physics-informed neural networks

    Guofei Pang, Lu Lu, and George Em Karniadakis. fPINNs: Fractional physics-informed neural networks. SIAM Journal on Scientific Computing, 41(4):A2603–A2626, 2019

    2019

  2. [10]

    Physics-guided neural networks (pgnn): An application in lake temperature modeling

    Anuj Karpatne, William Watkins, Jordan Read, and Vipin Kumar. Physics-guided neural networks (pgnn): An application in lake temperature modeling. arXiv preprint arXiv:1710.11431, 2, 2017

    2017 arXiv

  3. [11]

    Hard encoding of physics for learning spatiotemporal dynamics

    Chengping Rao, Hao Sun, and Yang Liu. Hard encoding of physics for learning spatiotemporal dynamics. arXiv preprint arXiv:2105.00557, 2021

    2021 arXiv

  4. [12]

    Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data

    Yinhao Zhu, Nicholas Zabaras, Phaedon-Stelios Koutsourelakis, and Paris Perdikaris. Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data. Journal of Computational Physics, 394:56–81, 2019

    2019

  5. [13]

    Three ways to solve partial differential equations with neural networks—A review

    Jan Blechschmidt and Oliver G Ernst. Three ways to solve partial differential equations with neural networks—A review. GAMM-Mitteilungen, 44(2):e202100006, 2021

    2021

  6. [14]

    Gaussian processes for machine learning

    Christopher KI Williams and Carl Edward Rasmussen. Gaussian processes for machine learning. MIT press Cambridge, MA, 2006

    2006

  7. [15]

    Bayesian interpolation

    David JC MacKay. Bayesian interpolation. Neural Computation, 4(3):415–447, 1992

    1992

  8. [16]

    A survey of constrained Gaussian process regression: Approaches and implementation challenges

    Laura P Swiler, Mamikon Gulian, Ari L Frankel, Cosmin Safta, and John D Jakeman. A survey of constrained Gaussian process regression: Approaches and implementation challenges. Journal of Machine Learning for Modeling and Computing, 1(2), 2020

    2020

  9. [17]

    Linear operators and stochastic partial differential equations in Gaussian process regression

    Simo Särkkä. Linear operators and stochastic partial differential equations in Gaussian process regression. In Artificial Neural Networks and Machine Learning–ICANN 2011: 21st International Conference on Artificial Neural Networks, Espoo, Finland, June 14-17, 2011, Proceedings...

    2011

  10. [18]

    Machine learning of linear differential equations using Gaussian processes

    Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Machine learning of linear differential equations using Gaussian processes. Journal of Computational Physics, 348:683–693, 2017

    2017

  11. [19]

    Physics-informed Gaussian process regression generalizes linear PDE solvers

    Marvin Pförtner, Ingo Steinwart, Philipp Hennig, and Jonathan Wenger. Physics-informed Gaussian process regression generalizes linear PDE solvers. arXiv preprint arXiv:2212.12474, 2022

    2022 arXiv

  12. [20]

    Stationary and related stochastic processes: Sample function properties and their applications

    Harald Cramér and M Ross Leadbetter. Stationary and related stochastic processes: Sample function properties and their applications. Courier Corporation, 2013

    2013

  13. [21]

    Deep kernel learning

    Andrew Gordon Wilson, Zhiting Hu, Ruslan Salakhutdinov, and Eric P Xing. Deep kernel learning. In Artificial Intelligence and Statistics, pages 370–378. PMLR, 2016

    2016

  14. [22]

    Manifold Gaussian processes for regression

    Roberto Calandra, Jan Peters, Carl Edward Rasmussen, and Marc Peter Deisenroth. Manifold Gaussian processes for regression. In 2016 International Joint Conference on Neural Networks (IJCNN), pages 3338–3345. IEEE, 2016

    2016

  15. [23]

    Guided deep kernel learning

    Idan Achituve, Gal Chechik, and Ethan Fetaya. Guided deep kernel learning. In Uncertainty in Artificial Intelligence, pages 11–21. PMLR, 2023

    2023

  16. [24]

    Artificial neural networks for solving ordinary and partial differential equations

    Isaac E Lagaris, Aristidis Likas, and Dimitrios I Fotiadis. Artificial neural networks for solving ordinary and partial differential equations. IEEE Transactions on Neural Networks, 9(5):987–1000, 1998

    1998

  17. [25]

    Numerical solution for high order differential equations using a hybrid neural network—optimization method

    Alaeddin Malek and R Shekari Beidokhti. Numerical solution for high order differential equations using a hybrid neural network—optimization method. Applied Mathematics and Computation, 183(1):260–271, 2006. 20 PDE-DKL: PDE-constrained deep kernel learning in high dimensionality

    2006

  18. [26]

    A unified deep artificial neural network approach to partial differential equations in complex geometries

    Jens Berg and Kaj Nyström. A unified deep artificial neural network approach to partial differential equations in complex geometries. Neurocomputing, 317:28–41, 2018

    2018

  19. [27]

    Finite-element neural networks for solving differential equations

    Pradeep Ramuhalli, Lalita Udpa, and Satish S Udpa. Finite-element neural networks for solving differential equations. IEEE Transactions on Neural Networks, 16(6):1381–1392, 2005

    2005

  20. [28]

    Deep learned finite elements

    Jaeho Jung, Kyungho Yoon, and Phill-Seung Lee. Deep learned finite elements. Computer Methods in Applied Mechanics and Engineering, 372:113401, 2020

    2020

  21. [29]

    DGM: A deep learning algorithm for solving partial differential equations

    Justin Sirignano and Konstantinos Spiliopoulos. DGM: A deep learning algorithm for solving partial differential equations. Journal of Computational Physics, 375:1339–1364, 2018

    2018

  22. [30]

    The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems

    Bing Yu et al. The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems. Communications in Mathematics and Statistics, 6(1):1–12, 2018

    2018

  23. [31]

    Deep Nitsche method: Deep Ritz method with essential boundary conditions

    Yulei Liao and Pingbing Ming. Deep Nitsche method: Deep Ritz method with essential boundary conditions. arXiv preprint arXiv:1912.01309, 2019

    1912 arXiv

  24. [32]

    Learning the solution operator of parametric partial differential equations with physics-informed DeepONets

    Sifan Wang, Hanwen Wang, and Paris Perdikaris. Learning the solution operator of parametric partial differential equations with physics-informed DeepONets. Science Advances, 7(40):eabi8605, 2021

    2021

  25. [33]

    Neural operator: Learning maps between function spaces with applications to pdes

    Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Learning maps between function spaces with applications to pdes. Journal of Machine Learning Research, 24(89):1–97, 2023

    2023

  26. [34]

    Fourier neural operator for parametric partial differential equations

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

    2010 arXiv

  27. [35]

    Error bounds for approximations with deep relu networks

    Dmitry Yarotsky. Error bounds for approximations with deep relu networks. Neural networks, 94:103–114, 2017

    2017

  28. [36]

    Understanding and mitigating gradient flow pathologies in physics-informed neural networks

    Sifan Wang, Yujun Teng, and Paris Perdikaris. Understanding and mitigating gradient flow pathologies in physics-informed neural networks. SIAM Journal on Scientific Computing, 43(5):A3055–A3081, 2021

    2021

  29. [37]

    Estimates on the generalization error of physics-informed neural networks for approximating pdes

    Siddhartha Mishra and Roberto Molinaro. Estimates on the generalization error of physics-informed neural networks for approximating pdes. IMA Journal of Numerical Analysis, 43(1):1–43, 2023

    2023

  30. [38]

    Pinnacle: A comprehensive benchmark of physics-informed neural networks for solving PDEs

    Zhongkai Hao, Jiachen Yao, Chang Su, Hang Su, Ziao Wang, Fanzhi Lu, Zeyu Xia, Yichi Zhang, Songming Liu, Lu Lu, et al. Pinnacle: A comprehensive benchmark of physics-informed neural networks for solving PDEs. arXiv preprint arXiv:2306.08827, 2023

    2023 arXiv

  31. [39]

    Deep splitting method for parabolic PDEs

    Christian Beck, Sebastian Becker, Patrick Cheridito, Arnulf Jentzen, and Ariel Neufeld. Deep splitting method for parabolic PDEs. SIAM Journal on Scientific Computing, 43(5):A3135–A3154, 2021

    2021

  32. [40]

    Solving stochastic differential equations and Kolmogorov equations by means of deep learning

    Christian Beck, Sebastian Becker, Philipp Grohs, Nor Jaafari, and Arnulf Jentzen. Solving stochastic differential equations and Kolmogorov equations by means of deep learning. arXiv preprint arXiv:1806.00421, 1(1), 2018

    2018 arXiv

  33. [41]

    Numerically solving parametric families of high-dimensional Kolmogorov partial differential equations via deep learning

    Julius Berner, Markus Dablander, and Philipp Grohs. Numerically solving parametric families of high-dimensional Kolmogorov partial differential equations via deep learning. Advances in Neural Information Processing Systems, 33:16615–16627, 2020

    2020

  34. [42]

    A deep learning solution approach for high-dimensional random differential equations

    Mohammad Amin Nabian and Hadi Meidani. A deep learning solution approach for high-dimensional random differential equations. Probabilistic Engineering Mechanics, 57:14–25, 2019

    2019

  35. [43]

    An extreme learning machine-based method for computational PDEs in higher dimensions

    Yiran Wang and Suchuan Dong. An extreme learning machine-based method for computational PDEs in higher dimensions. Computer Methods in Applied Mechanics and Engineering, 418:116578, 2024

    2024

  36. [44]

    Linearly constrained Gaussian processes

    Carl Jidling, Niklas Wahlström, Adrian Wills, and Thomas B Schön. Linearly constrained Gaussian processes. Advances in Neural Information Processing Systems, 30, 2017. 21 PDE-DKL: PDE-constrained deep kernel learning in high dimensionality

    2017

  37. [45]

    Inferring solutions of differential equations using noisy multi-fidelity data

    Maziar Raissi, Paris Perdikaris, and George Em Karniadakis. Inferring solutions of differential equations using noisy multi-fidelity data. Journal of Computational Physics, 335:736–746, 2017

    2017

  38. [46]

    Gaussian process port-Hamiltonian systems: Bayesian learning with physics prior

    Thomas Beckers, Jacob Seidman, Paris Perdikaris, and George J Pappas. Gaussian process port-Hamiltonian systems: Bayesian learning with physics prior. In 2022 IEEE 61st Conference on Decision and Control (CDC), pages 1447–1453. IEEE, 2022

    2022

  39. [47]

    Learning physical models that can respect conservation laws

    Derek Hansen, Danielle C Maddix, Shima Alizadeh, Gaurav Gupta, and Michael W Mahoney. Learning physical models that can respect conservation laws. arXiv preprint arXiv:2302.11002, 2023

    2023 arXiv

  40. [48]

    The statistical finite element method (statFEM) for coherent synthesis of observation data and model predictions

    Mark Girolami, Eky Febrianto, Ge Yin, and Fehmi Cirak. The statistical finite element method (statFEM) for coherent synthesis of observation data and model predictions. Computer Methods in Applied Mechanics and Engineering, 375:113533, 2021

    2021

  41. [49]

    Gaussian process latent force models for learning and stochastic control of physical systems

    Simo Särkkä, Mauricio A Alvarez, and Neil D Lawrence. Gaussian process latent force models for learning and stochastic control of physical systems. IEEE Transactions on Automatic Control, 64(7):2953–2960, 2018

    2018

  42. [50]

    Neural-net-induced Gaussian process regression for function approximation and PDE solution

    Guofei Pang, Liu Yang, and George Em Karniadakis. Neural-net-induced Gaussian process regression for function approximation and PDE solution. Journal of Computational Physics, 384:270–288, 2019

    2019

  43. [51]

    On the generalized distance in statistics

    Prasanta Chandra Mahalanobis. On the generalized distance in statistics. Sankhy¯a: The Indian Journal of Statistics, Series A (2008-), 80:S1–S7, 2018

    2008

  44. [52]

    Physics-guided machine learning for scientific discovery: An application in simulating lake temperature profiles

    Xiaowei Jia, Jared Willard, Anuj Karpatne, Jordan S Read, Jacob A Zwart, Michael Steinbach, and Vipin Kumar. Physics-guided machine learning for scientific discovery: An application in simulating lake temperature profiles. ACM/IMS Transactions on Data Science, 2(3):1–26, 2021

    2021

  45. [53]

    Bayesian model selection, the marginal likelihood, and generalization

    Sanae Lotfi, Pavel Izmailov, Gregory Benton, Micah Goldblum, and Andrew Gordon Wilson. Bayesian model selection, the marginal likelihood, and generalization. In International Conference on Machine Learning, pages 14223–14247. PMLR, 2022

    2022

  46. [54]

    Self-normalizing neural networks

    Günter Klambauer, Thomas Unterthiner, Andreas Mayr, and Sepp Hochreiter. Self-normalizing neural networks. Advances in neural information processing systems, 30, 2017. 22

    2017

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.