betaFeedback

WorksComputer Science

Newton-PINet: A fast physics-informed neural network with Newton linearization for meta-learning nonlinear PDEs

Abstract

Scientific machine learning has opened new avenues for solving parameterized partial differential equations (PDEs), enabling models to learn a family of PDEs and generalize to unseen instances. In this context, data-driven operator learning methods typically require large training datasets, while physics-informed neural networks (PINNs) suffer from difficult optimization and limited generalization, especially for nonlinear PDEs. We propose Newton-PINet, a physics-informed network enhanced by Newton linearization, offering an effective meta-learning framework for nonlinear PDEs. Newton-PINet (i) employs a physics-informed multilayer network with skip connections, where the output-layer weights are solved by least squares; (ii) adopts a two-stage learning strategy that first leverages gradient-based training to learn robust representations from the available training tasks, and then performs gradient-free fine-tuning on the output layer for fast task-specific generalization; and (iii) incorporates a Newton linearization method to speed up the least-squares iteration for nonlinear PDE problems. On a challenging nonlinear reaction-diffusion benchmark, Newton-PINet achieves up to three orders of magnitude lower relative error than recent neural solvers, while using 16x fewer training tasks and over an order of magnitude less training time (about 5 minutes versus several hours). This work advances the meta-learning of PINNs toward data-efficient, fast, and generalizable physics solvers. The code and datasets are publicly available at: https://github.com/fycAIwj/Newton-PINet.

Introduction and related works

Solving partial differential equations (PDEs) is fundamental across diverse fields, including fluid dynamics, climate modeling, and biophysics [24, 19]. Traditional numerical solvers for PDEs often incur prohibitive computational costs, particularly when repeated evaluations are required in applications. Recently, scientific machine learning has emerged as a powerful paradigm, leveraging advances in deep neural networks to deliver fast and accurate approximations of PDE solutions, thereby enabling high-fidelity simulations and real-time predictive capabilities that were previously unattainable [4, 11, 22]. Deep learning for PDEs can be broadly categorized into: data-driven operator learning, physics-informed deep learning, and meta-learning physics-informed neural networks (PINNs).

Data-driven operator learning: Data-driven operator learning focuses on training neural networks to directly map input functions, such as PDE parameters or initial conditions, to their corresponding PDE solutions. Well-known examples of this approach include DeepONet [Lu et al., 2021], Derivative-

Preprint. enhanced-DeepONet [35], Transformer-DeepONet [47], Fourier neural operators (FNO) [29], Factorized-FNO [40], Decomposed-FNO [28], and Tucker-FNO [54]. Once trained, these models can generalize to new parameter settings (e.g., new initial conditions), enabling fast prediction of complex dynamical behaviors. However, they lack explicit physics constraints and interpretability, and their generalization depends on adequacy of labeled training data, which are often expensive to obtain [30]. For example, Transformer-DeepONet typically requires at least 1,000 different PDE solutions for adequate performance [47].

Physics-informed deep learning: An alternative paradigm, physics-informed deep learning, incorporates governing PDEs and boundary conditions into the training loss function [49, 53, 46]. By penalizing violation of physical laws, these models can be trained effectively even in data-sparse regimes, often requiring only initial or boundary conditions. Representative approaches include physics-informed variants of operator learning, such as physics-informed DeepONet [41], physics-informed neural operators (PINO) [30], and HyPINO [1]. Another prominent method is PINN [37], which learns to map the spatiotemporal coordinates to the PDE solution for a given system. While these methods alleviate the need for large labeled datasets, their reliance on physics-informed objectives often results in prohibitively long training time, due to highly nonconvex loss landscapes caused by the stiff PDE constraints [7, 43, 13], especially for nonlinear PDEs. For example, training physics-informed DeepONet on a family of Burgers’ equations can take more than 9 hours [41], while PINNs may require over 24 hours to solve a single Kuramoto-Sivashinsky equation [45].

Meta-learning PINNs: Adapting PINN to new PDEs or parameter settings typically requires costly retraining. Transfer learning strategies accelerate convergence by reusing knowledge from related tasks [48, 42]. Parameterized PINNs (P2P^{2}INNs) further incorporate problem parameters as network inputs, enabling a single model to represent families of PDEs without labeled data while preserving physical consistency [9]. However, P2P^{2}INNs still suffer from the optimization challenges of physics-based loss functions, and their adaptation to new tasks remains heavily reliant on gradient-based fine-tuning which is often slow to converge, especially for nonlinear PDEs. Meta-learning methods aim to enable rapid adaptation to unseen PDEs with limited data by learning parameter initializations or update rules [33, 50]. MAML-style approaches learn transferable parameter priors [31, 51], while other methods leverage latent codes or meta-optimizers for task-specific adaptation [23, 8]. To avoid gradient-based adaptation, Baldwinian-PINN introduces a gradient-free meta-learning strategy based on neuroevolution, evolving shared representations and fine-tuning the output layer via least squares [52]. While highly efficient for linear PDEs, its single-layer architecture limits nonlinear expressiveness. Moreover, its Picard (lagging-of-coefficients) iteration exhibits only linear convergence, leading to slow solution refinement for nonlinear problems [6, 34].

See Appendix A for an extended literature review. In summary, existing deep learning approaches for solving nonlinear PDEs remain constrained by substantial data requirements and/or slow, optimization-heavy adaptation. For this reason, we propose a novel meta-learning PINN framework (Figure 1), termed Newton-PINet. The main contributions are as follows:

Newton-PINet model framework

Figure 1. Newton-PINet model framework. Stage 1 (pre-training) meta-learns the task-agnostic parameters Θ={w~,λpde,λreg}\Theta= \{\tilde{w}, \lambda_{\mathrm{pde}}, \lambda_{\mathrm{reg}}\} by gradient descent (outer loop), while the output-layer weights ww of each training task are updated by gradient-free Tikhonov-regularized least squares (inner loop); Stage 2 (inference) freezes Θ\Theta and adapts only ww to new tasks. The bottom panel compares Picard and Newton linearization; the latter accelerates the convergence of nonlinear iterations for solving nonlinear PDEs.

  • We introduce a physics-informed multilayer network with skip connections from early hidden layers to the output, where the final-layer weights are computed using least-squares approach (Tikhonov regularization). This skip-connected Tikhonov regularization PINN architecture improves the model’s nonlinear representation capabilities.

  • We employ a two-stage learning strategy, where the first stage uses a gradient-based method to meta-learn the nonlinear hidden layers’ network weights and essential hyperparameters for task-specific adaptation in an unsupervised or few-shot manner. At inference (test time), task-specific adaptation is confined to the output layer, which can be rapidly updated in a gradient-free manner using Tikhonov regularization.

  • We integrate Newton linearization into the Tikhonov-regularized PINN and establish its equivalence to the classical regularized Gauss-Newton method through mathematical derivation, with experiments validating its near-quadratic convergence. The Newton linearization accelerates nonlinear least-squares convergence, providing a more efficient alternative to the

traditional Picard approach during both meta-learning and inference. We therefore refer to the proposed physics-informed network with Newton linearization as Newton-PINet.

Results on diverse nonlinear PDE benchmarks show that, compared with recent neural solvers, Newton-PINet significantly reduces the reliance on training data while achieving fast and accurate generalization to new PDE tasks through physics-consistent fine-tuning.

Preliminaries

For simplicity, consider a nonlinear PDE with spatial variable xx, time tt, and solution uu defined on Ω×[0,T]\Omega\times[0,T]:

PDE:Nθ[u(x,t)]=q(x,t)x∈Ω, t∈[0,T],IC:u(x,t=0)=u0(x)x∈Ω,BC:B[u(x,t)]=g(x,t)x∈∂Ω, t∈[0,T].(1)\begin{aligned} \mathrm{PDE}: \quad\mathcal{N}_{\theta}[u(x,t)] &= q(x,t) \quad x \in\Omega,\ t \in[0,T],\\ \mathrm{IC}: \quad u(x,t=0) &= u_0(x) \quad x \in\Omega,\\ \mathrm{BC}: \quad\mathcal{B}[u(x,t)] &= g(x,t) \quad x \in\partial\Omega,\ t \in[0,T]. \tag*{(1)} \end{aligned}

where the general differential operator Nθ\mathcal{N}_{\theta} can include both linear and nonlinear combinations of the temporal and spatial derivatives and PDE parameters θ\theta, q(x,t)q(x,t) is the source term, u0(x)u_0(x) is the initial condition (IC), and g(x,t)g(x,t) is the boundary condition (BC).

Single PDE problem: Standard PINNs aim to solve a single PDE instance (a single task) defined by specific PDE parameters, IC, and BC. PINNs are trained by minimizing the discrepancy between Eq. (1) and the model’s prediction.

Towards generalizable PINNs: There is growing interest in models capable of generalizing across a set of tasks belonging to some underlying task-distribution p(T)p(\mathcal{T}), e.g., a family of PDEs spanning different PDE parameters, ICs, and BCs. For meta-learning PINNs, the goal is to learn network initializations using training tasks sampled from p(T)p(\mathcal{T}) that enable fast, accurate, and physics-aware predictions on unseen scenarios, i.e., any new task from the distribution Ti∼p(T)\mathcal{T}_i \sim p(\mathcal{T}) through fine-tuning [50].

Methodology

Skip-connected Tikhonov regularization PINN: We introduce skip connections from all hidden layers to the output layer, with the output-layer weights computed using Tikhonov regularization [17]. The skip-connected neural architecture improves expressivity, ensuring stable and accurate least-squares computation by increasing the output-layer width through stacking additional hidden layers, while still maintaining a moderate number of nodes per layer. As shown in Figure 1, we employ the neural networks with L+1L+1 layers, where the input is x=(x,t)\mathbf{x}=(x,t) (layer 0) and the output is uu (layer LL). Each hidden layer l=1,…,L−1l=1,\ldots,L-1 contains the same number of neurons (NnN_n). Sinusoidal feature embeddings are applied at the first layer [49], and all hidden layers use sin⁡(⋅)\sin(\cdot) activation functions. The pre-final output f\mathbf{f} is constructed as the concatenation of all hidden activations (skip connections). All trainable network parameters up to the pre-final layer are denoted w~\widetilde{\mathbf{w}}. The output uu can be computed by u(x)=f(x;w~)Twu(\mathbf{x})=\mathbf{f}(\mathbf{x};\widetilde{\mathbf{w}})^T\mathbf{w}, where w\mathbf{w} is the output-layer weights. See Appendix § 10.§ 1 for a detailed description of the skip-connected model architecture.

The objective of task-specific learning is to determine the best set of w\mathbf{w} such that u(x)=f(x;w~)Twu(\mathbf{x})=\mathbf{f}(\mathbf{x};\widetilde{\mathbf{w}})^T\mathbf{w} satisfies the PDE, IC, and BC defined in Eq. (1) for a task Ti∼p(T)T_i\sim p(T). This can be formulated as a physics-based least-squares problem. Given collocation points for PDE residual, points at the initial time, and points on the boundary, together with the loss importance hyperparameter λpde\lambda_{\mathrm{pde}}, the following system is obtained:

\begin{bmatrix} \cdots& \lambda_{\mathrm{pde}}\cdot\mathcal{N}_{\theta}[f_j(\mathbf{x}^{\mathrm{pde}}_1,t^{\mathrm{pde}}_1;\widetilde{\mathbf{w}})] & \cdots\\ \cdots& f_j^i(\mathbf{x}^{\mathrm{ic}}_1,0;\widetilde{\mathbf{w}}) & \cdots\\ & \vdots& \\ \cdots& \mathcal{B}[f_j(\mathbf{x}^{\mathrm{bc}}_1,t^{\mathrm{bc}}_1;\widetilde{\mathbf{w}})] & \cdots\\ & \vdots& \end{bmatrix} \begin{bmatrix} \vdots\\ w_j\\ \vdots \end{bmatrix} = \begin{bmatrix} \lambda_{\mathrm{pde}}\cdot q(\mathbf{x}^{\mathrm{pde}}_1,t^{\mathrm{pde}}_1)\\ \vdots\\ u_0(\mathbf{x}^{\mathrm{ic}}_1)\\ g(\mathbf{x}^{\mathrm{bc}}_1,t^{\mathrm{bc}}_1)\\ \vdots \end{bmatrix}, \tag*{(2)}\end{bmatrix}

which can be compactly written as

Aw=b.A\mathbf{w}=\mathbf{b}.

Here AwA\mathbf{w} represents the left-hand side of Eq. (1) at a given set of collocation points, and b\mathbf{b} denotes the corresponding right-hand side of Eq. (1). The number of rows in AA and b\mathbf{b} equals the total number of collocation points. The number of columns in AA corresponds to the product of the network’s pre-final feature dimension and the output dimension. The final output weights w\mathbf{w} are obtained via Tikhonov regularization: arg min⁡w∥Aw−b∥22+λregwTw\mathop{\operatorname{arg\,min}}_{\mathbf{w}}\lVert A\mathbf{w}-\mathbf{b}\rVert_2^2+\lambda_{\mathrm{reg}}\mathbf{w}^T\mathbf{w}. λreg\lambda_{\mathrm{reg}} is the Tikhonov regularization parameter whose proper setting improves the conditioning of the problem. The system has a closed-form solution, either w=(λregI+ATA)−1ATb\mathbf{w}=(\lambda_{\mathrm{reg}}I+A^TA)^{-1}A^T\mathbf{b} for over-determined system or AT(λregI+AAT)−1bA^T(\lambda_{\mathrm{reg}}I+AA^T)^{-1}\mathbf{b} for under-determined one, enabling fast, gradient-free updates. See Appendix § 10.§ 2 for the details.

Note that the Tikhonov-regularized formulation provides a closed-form result in a single computation step for any linear PDE. For nonlinear PDEs, the nonlinear terms must be linearized so that the system can be cast into a linear form. An initial guess is then required, followed by a few nonlinear iterations (a process that can also be interpreted as gradient-free fine-tuning) to update w\mathbf{w} toward the optimal solution. Hereafter, we use NN to denote the number of nonlinear iterations.

The proposed Newton-PINet model framework consists of two stages (Figure 1): (i) meta-learning on training tasks and (ii) gradient-free fine-tuning on new tasks.

Meta-learning on training tasks: The meta-learning objective is to optimize the network weights w~\widetilde{\mathbf{w}}, importance hyperparameter λpde\lambda_{\mathrm{pde}} of PDE loss relative to IC/BC, and the regularization parameter λreg\lambda_{\mathrm{reg}}, collectively denoted Θ\Theta, to learn task-agnostic representations spanning a family of PDEs and enable fast generalization to unseen tasks via fine-tuning only the output-layer weights using the Tikhonov regularization update. In our framework, the outer loop of meta-learning updates the learnable parameters via gradient descent with learning rate η\eta, while task-specific adaptation in the inner loop, or generalization, is performed through gradient-free Tikhonov regularization applied to the output-layer weights. The meta-learning objective (outer-loop loss) can be either the physics-based least-squares error lLSE(w∗)=∥Aw∗−b∥22l_{\mathrm{LSE}}(\mathbf{w}^*)=\lVert A\mathbf{w}^*-\mathbf{b}\rVert_2^2, or the data-driven mean squared error lMSE(w∗)=1n∑s(uslabel−f(x;w~)Tw∗)2l_{\mathrm{MSE}}(w^*) = \frac{1}{n}\sum_s (u_s^{\mathrm{label}} - f(x;\widetilde{w})^T w^*)^2 given labeled data {uslabel}s=1n\{u_s^{\mathrm{label}}\}_{s=1}^n where nn is the total number of collocation points (PDE residual, IC, and BC). Here, Tikhonov regularization is used to compute the optimal task-specific output-layer weights w∗w^*, enabling the model to specialize to any realization of the task. lLSEl_{\mathrm{LSE}}-based learning is termed unsupervised learning. lMSEl_{\mathrm{MSE}}-based learning is termed hybrid learning, since it couples physics-based Tikhonov updates in the inner loop with data-driven minimization in the outer loop. See Algorithm 1 for the pseudocode and Appendix B.3 for a detailed mathematical description.

Gradient-free fine-tuning to new tasks: After meta-learning, Θ\Theta is fixed. For a new task Ti∼p(T)\mathcal{T}_i \sim p(\mathcal{T}) with different PDE parameters or IC/BC, Tikhonov regularization is used to compute the task-specific output-layer weights, enabling fast gradient-free adaptation to the new PDE instance. Since this update is independent of gradient-based backpropagation optimizers such as stochastic gradient descent (SGD), the resulting adaptation is extremely fast while remaining physics-compliant. See Algorithm 2 for the pseudocode.

Algorithm 1 : Meta-learning on training tasks
Require: Initialize Θ; p(T ); η
Require: C, constructor of least-squares system (A, b); see Eq. (2)
while not done do
    Sample training tasks {Ti} ∼ p(T )
    for each task Ti do
        Get initial guess u(1)
        for k = 1 to N do
            (A, b) = C(Ti, Θ, uk)
            w = { (λregI + AᵀA)−1Aᵀb, or
                  Aᵀ(λregI + AAᵀ)−1b
            u(k+1) = fᵀw
        end for
        Compute loss lTi
    end for
    Θ ← Θ − η∇Θ ∑i lTi
end while
Algorithm 2 : Gradient-free fine-tuning to new tasks
Require: Freeze Θ
for each test task Ti do
    Get initial guess u(1)
    for k = 1 to N do
        (A, b) = C(Ti, Θ, uk)
        w = { (λregI + AᵀA)−1Aᵀb, or
              Aᵀ(λregI + AAᵀ)−1b
        u(k+1) = fᵀw
    end for
    Store solution uTi = u(N+1)
end for
return all task solution {uTi}

Newton linearization: Note that the Tikhonov regularization applies to linear PDEs, where a linear matrix system can be constructed and solved in a single step. For nonlinear PDEs, the nonlinear terms must first be linearized so that the system can be cast into a linear form. The previous approaches typically relied on Picard method (lagging-of-coefficients), which linearizes the nonlinear terms using the solution from the previous iteration and then performs least-squares solves for multiple nonlinear iterations. Although simple, it is only linearly convergent [14]. To address this limitation, we use the Newton linearization approach to approximate the nonlinear term by Taylor expansion around the current iterate. For a function FF of one state variable uu, this gives F(uk+1)=F(uk)+dFdu∣k(uk+1−uk)+H.O.T.F(u^{k+1}) = F(u^k) + \left.\frac{dF}{du}\right|_k (u^{k+1} - u^k) + \text{H.O.T.}, where kk is the nonlinear iteration step, and H.O.T denotes the higher-order terms of Taylor expansion. For a function of two variables uu and vv, the expansion becomes F(uk+1,vk+1)=F(uk,vk)+∂F∂u∣k(uk+1−uk)+∂F∂v∣k(vk+1−vk)+H.O.T.F(u^{k+1},v^{k+1}) = F(u^k,v^k) + \left.\frac{\partial F}{\partial u}\right|_k (u^{k+1} - u^k) + \left.\frac{\partial F}{\partial v}\right|_k (v^{k+1} - v^k) + \text{H.O.T.} Based on this principle, the nonlinear terms commonly arising in PDEs can be expressed in a Newton-linearized form amenable to Tikhonov regularization. Table 1 summarizes several representative nonlinear terms derived by our work.

Table 1: Newton-linearized expressions of several nonlinear terms derived by our work. mm denotes the exponent. kk and k+1k+1 refer to the current and next nonlinear iterations, respectively.

Nonlinear termNewton-linearized expression
(um)k+1(u^m)^{k+1}m(uk)m−1uk+1+(1−m)(uk)mm(u^k)^{m-1}u^{k+1}+(1-m)(u^k)^m
(umux)k+1(u^m u_x)^{k+1}m(uk)m−1uxkuk+1+(uk)muxk+1−m(uk)muxkm(u^k)^{m-1}u_x^k u^{k+1}+(u^k)^m u_x^{k+1}-m(u^k)^m u_x^k
[sinh⁡(u)]k+1[\sinh(u)]^{k+1}cosh⁡(uk)uk+1+sinh⁡(uk)−cosh⁡(uk)uk\cosh(u^k)u^{k+1}+\sinh(u^k)-\cosh(u^k)u^k
[exp⁡(u)]k+1[\exp(u)]^{k+1}exp⁡(uk)uk+1+exp⁡(uk)−exp⁡(uk)uk\exp(u^k)u^{k+1}+\exp(u^k)-\exp(u^k)u^k
[uln⁡(u)]k+1[u\ln(u)]^{k+1}(ln⁡(uk)+1)uk+1+ukln⁡(uk)−(ln⁡(uk)+1)uk(\ln(u^k)+1)u^{k+1}+u^k\ln(u^k)-(\ln(u^k)+1)u^k

Table 1.

A detailed mathematical derivation is provided in Appendix B.5. In Appendix C, we further establish that the Newton linearization used in the Tikhonov-regularized PINN is essentially equivalent to the classical regularized Gauss-Newton method, with experiments validating its near-quadratic convergence.

Experiment results

We compare the performance of the proposed Newton-PINet (Newton linearization), PINet (Picard linearization) and purely data-driven deep neural network (DNN) on several representative classes of nonlinear PDEs:

References

  1. [1]Rafael Bischof, Michal Piovarči, Michael A Kraus, Siddhartha Mishra, and Bernd Bickel. Hypino: Multi-physics neural operators via hyperpinns and the method of manufactured solutions. arXiv preprint arXiv:2509.05117, 2025.arxiv.org/abs/2509.05117
  2. [2]Lise Le Boudec, Emmanuel De Bézenac, Louis Serrano, Ramon Daniel Regueiro-Espino, Yuan Yin, and Patrick Gallinari. Learning a neural solver for parametric pde to enhance physics-informed methods. 2025. URL https://openreview.net/forum?id=jqVj8vCQsT.
  3. [3]James Bradbury, Roy Frostig, Peter Hawkins, Matthew James Johnson, Chris Leary, Dougal Maclaurin, George Necula, Adam Paszke, Jake VanderPlas, Skye Wanderman-Milne, et al. Jax: composable transformations of python+ numpy programs. 2018.
  4. [4]Shengze Cai, Zhiping Mao, Zhicheng Wang, Minglang Yin, and George Em Karniadakis. Physics-informed neural networks (pinns) for fluid mechanics: A review. Acta Mechanica Sinica, 37(12):1727–1738, 2021.arxiv.org/abs/2105.09506
  5. [5]Pao-Hsiung Chiu and Hee Joo Poh. Development of an improved divergence-free-condition compensated coupled framework to solve flow problems with time-varying geometries. International Journal for Numerical Methods in Fluids, 93(1):44–70, 2021.DOI
  6. [6]Pao-Hsiung Chiu, Tony WH Sheu, and Rui-Kuo Lin. An effective explicit pressure gradient scheme implemented in the two-level non-staggered grids for incompressible navier–stokes equations. Journal of Computational Physics, 227(8):4018–4037, 2008.DOI
  7. [7]Pao-Hsiung Chiu, Jian Cheng Wong, Chinchun Ooi, My Ha Dao, and Yew-Soon Ong. Can-pinn: A fast physics-informed neural network based on coupled-automatic–numerical differentiation method. Computer Methods in Applied Mechanics and Engineering, 395:114909, 2022.DOI
  8. [8]Woojin Cho, Kookjin Lee, Donsub Rim, and Noseong Park. Hypernetwork-based meta-learning for low-rank physics-informed neural networks. Advances in Neural Information Processing Systems, 36:11219–11231, 2023.arxiv.org/abs/2310.09528
  9. [9]Woojin Cho, Minju Jo, Haksoo Lim, Kookjin Lee, Dongeun Lee, Sanghyun Hong, and Noseong Park. Parameterized physics-informed neural networks for parameterized pdes. arXiv preprint arXiv:2408.09446, 2024.arxiv.org/abs/2408.09446
  10. [10]Steven M Cox and Paul C Matthews. Exponential time differencing for stiff systems. Journal of Computational Physics, 176(2):430–455, 2002.DOI
  11. [11]Salvatore Cuomo, Vincenzo Schiano Di Cola, Fabio Giampaolo, Gianluigi Rozza, Maziar Raissi, and Francesco Piccialli. Scientific machine learning through physics-informed neural networks: Where we are and what’s next. Journal of Scientific Computing, 92(3):88, 2022.arxiv.org/abs/2201.05624
  12. [12]Eric C Cyr, Mamikon A Goulian, Ravi G Patel, Mauro Perego, and Nathaniel A Trask. Robust training and initialization of deep neural networks: An adaptive basis viewpoint. In Mathematical and Scientific Machine Learning, pages 512–536. PMLR, 2020.arxiv.org/abs/1912.04862
  13. [13]Chinmay Datar, Taniya Kapoor, Abhishek Chandra, Qing Sun, Erik Lien Bolager, Iryna Burak, Anna Veselovska, Massimo Fornasier, and Felix Dietrich. Fast training of accurate physics-informed neural networks without gradient descent. arXiv preprint arXiv:2405.20836, 2024.arxiv.org/abs/2405.20836
  14. [14]John E Dennis Jr and Robert B Schnabel. Numerical methods for unconstrained optimization and nonlinear equations. SIAM, 1996.
  15. [15]Suchuan Dong and Zongwei Li. Local extreme learning machines and domain decomposition for solving linear and nonlinear partial differential equations. Computer Methods in Applied Mechanics and Engineering, 387:114129, 2021.DOI
  16. [16]Tobin A Driscoll, Nicholas Hale, and Lloyd N Trefethen. Chefbun guide, 2014.
  17. [17]Gene H Golub, Per Christian Hansen, and Dianne P O’Leary. Tikhonov regularization and total least squares. SIAM journal on matrix analysis and applications, 21(1):185–194, 1999.DOI
  18. [18]Somdatta Goswami, Aniruddha Bora, Yue Yu, and George Em Karniadakis. Physics-informed deep neural operator networks. In Machine learning in modeling and simulation: methods and applications, pages 219–254. Springer, 2023.arxiv.org/abs/2207.05748
  19. [19]Jan Hagnberger, Daniel Musekamp, and Mathias Niepert. CALM-PDE: Continuous and Adaptive Convolutions for Latent Space Modeling of Time-dependent PDEs. In Advances in Neural Information Processing Systems, volume 38, 2025.arxiv.org/abs/2505.12944
  20. [20]Xu Han, Han Gao, Tobias Pfaff, Jian-Xun Wang, and Li-Ping Liu. Predicting physics in mesh-reduced space with temporal attention. arXiv preprint arXiv:2201.09113, 2022.arxiv.org/abs/2201.09113
  21. [21]Jingren Hou, Hong Wang, Pengyu Xu, Chang Gao, Huafeng Liu, and Liping Jing. Learning neural operators from partial observations via latent autoregressive modeling. arXiv preprint arXiv:2601.15547, 2026.arxiv.org/abs/2601.15547
  22. [22]Peiyan Hu, Haodong Feng, Hongyuan Liu, Tongtong Yan, Wenhao Deng, Tianrun Gao, Rong Zheng, Haoren Zheng, Chenglei Yu, Chuanrui Wang, Kaiwen Li, Zhi-Ming Ma, Dezhi Zhou, Xingcai Lu, Dixia Fan, and Tailin Wu. Realpdebench: A benchmark for complex physical systems with real-world data. In The Fourteenth International Conference on Learning Representations, 2026. URL https://openreview.net/forum?id=y3oHMcoItR. Oral Presentation.arxiv.org/abs/2601.01829
  23. [23]Xiang Huang, Zhanhong Ye, Hongsheng Liu, Shi Ji, Zidong Wang, Kang Yang, Yang Li, Min Wang, Haotian Chu, Fan Yu, et al. Meta-auto-decoder for solving parametric partial differential equations. Advances in Neural Information Processing Systems, 35:23426–23438, 2022.arxiv.org/abs/2111.08823
  24. [24]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.DOI
  25. [25]Jungeun Kim, Kookjin Lee, Dongeun Lee, Sheo Yon Jhin, and Noseong Park. Dpm: A novel training method for physics-informed neural networks in extrapolation. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 8146–8154, 2021.arxiv.org/abs/2012.02681
  26. [26]Aditi Krishnapriyan, Amir Gholami, Shandian Zhe, Robert Kirby, and Michael W Mahoney. Characterizing possible failure modes in physics-informed neural networks. Advances in neural information processing systems, 34:26548–26560, 2021.arxiv.org/abs/2109.01050
  27. [27]Varun Kumar, Somdatta Goswami, Katiana Kontolati, Michael D Shields, and George Em Karniadakis. Synergistic learning with multi-task deeponet for efficient pde problem solving. Neural Networks, 184:107113, 2025.DOI
  28. [28]Kangjie Li and Wenjing Ye. D-fno: A decomposed fourier neural operator for large-scale parametric partial differential equations. Computer Methods in Applied Mechanics and Engineering, 436:117732, 2025.
  29. [29]Zongyi Li, Nikola Kovachki, Kamyar Azizadehnesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier neural operator for parametric partial differential equations. arXiv preprint arXiv:2010.08895, 2020.arxiv.org/abs/2010.08895
  30. [30]Zongyi Li, Hongkai Zheng, Nikola Kovachki, David Jin, Haoxuan Chen, Burigede Liu, Kamyar Azizade-nesheli, and Anima Anandkumar. Physics-informed neural operator for learning partial differential equations. ACM/IMS Journal of Data Science, 1(3):1–27, 2024.arxiv.org/abs/2111.03794
  31. [31]Xu Liu, Xiaoya Zhang, Wei Peng, Weien Zhou, and Wen Yao. A novel meta-learning initialization method for physics-informed neural networks. Neural Computing and Applications, 34(17):14511–14534, 2022.
  32. [33]Michael Penwarden, Shandian Zhe, Akil Narayan, and Robert M Kirby. A metalearning approach for physics-informed neural networks (pinns): Application to parameterized pdes. Journal of Computational Physics, 477:111912, 2023.
  33. [34]Richard H Pletcher, John C Tannehill, and Dale Anderson. Computational fluid mechanics and heat transfer. CRC press, 2012.
  34. [35]Yuan Qiu, Nolan Bridges, and Peng Chen. Derivative-enhanced deep operator network. Advances in Neural Information Processing Systems, 37:20945–20981, 2024.arxiv.org/abs/2402.19242
  35. [36]Aniruddh Raghu, Maithra Raghu, Samy Bengio, and Oriol Vinyals. Rapid learning of feature reuse? towards understanding the effectiveness of maml. arXiv preprint arXiv:1909.09157, 2019.arxiv.org/abs/1909.09157
  36. [37]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.DOI
  37. [38]Tony WH Sheu and RK Lin. Newton linearization of the incompressible navier–stokes equations. International Journal for Numerical Methods in Fluids, 44(3):297–312, 2004.DOI
  38. [39]Tony WH Sheu and RK Lin. On a high-order newton linearization method for solving the incompressible navier–stokes equations. International journal for numerical methods in engineering, 62(11):1559–1578, 2005.
  39. [40]Alasdair Tran, Alexander Mathews, Lexing Xie, and Cheng Soon Ong. Factorized fourier neural operators. In The Eleventh International Conference on Learning Representations, 2023. URL https://openreview.net/forum?id=tmiMP14IPa.
  40. [41]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.
  41. [42]Sifan Wang, Shyam Sankaran, and Paris Perdikaris. Respecting causality is all you need for training physics-informed neural networks. arXiv preprint arXiv:2203.07404, 2022.
  42. [43]Sifan Wang, Shyam Sankaran, Hanwen Wang, and Paris Perdikaris. An expert’s guide to training physics-informed neural networks. arXiv preprint arXiv:2308.08468, 2023.
  43. [44]Sifan Wang, Bowen Li, Yuhan Chen, and Paris Perdikaris. Piratenets: Physics-informed deep learning with residual adaptive networks. Journal of Machine Learning Research, 25(402):1–51, 2024.
  44. [45]Sifan Wang, Ananyae Kumar Bhartari, Bowen Li, and Paris Perdikaris. Gradient alignment in physics-informed neural networks: A second-order optimization perspective. arXiv preprint arXiv:2502.00604, 2025.
  45. [46]Chang Wei, Yuchen Fan, Jian Cheng Wong, Chin Chun Ooi, Heyang Wang, and Pao-Hsiung Chiu. Ffv-pinn: A fast physics-informed neural network with simplified finite volume discretization and residual correction. Computer Methods in Applied Mechanics and Engineering, 444:118139, 2025a.
  46. [47]Zhi-Feng Wei, Wengian Chen, and Panos Stinis. Efficient transformer-inspired variants of physics-informed deep operator networks. arXiv preprint arXiv:2509.01679, 2025b.
  47. [48]Jian Cheng Wong, Abhishek Gupta, and Yew-Soon Ong. Can transfer neuroevolution tractably solve your differential equations? IEEE Computational Intelligence Magazine, 16(2):14–30, 2021.
  48. [49]Jian Cheng Wong, Chin Chun Ooi, Abhishek Gupta, and Yew-Soon Ong. Learning in sinusoidal spaces with physics-informed neural networks. IEEE Transactions on Artificial Intelligence, 5(3):985–1000, 2022.
  49. [50]Jian Cheng Wong, Abhishek Gupta, Chin Chun Ooi, Pao-Hsiung Chiu, Jiao Liu, and Yew-Soon Ong. Evolutionary optimization of physics-informed neural networks: Evo-pinn frontiers and opportunities. IEEE Computational Intelligence Magazine, 21(1):16–36, 2026a.
  50. [51]Jian Cheng Wong, Isaac Yin Chung Lai, Pao-Hsiung Chiu, Chin Chun Ooi, Abhishek Gupta, and Yew-Soon Ong. Transferable physics-informed representations via closed-form head adaptation. In 2026 International Joint Conference on Neural Networks (IJCNN), pages 1–8, 2026b. doi: 10.1109/IJCNN69854.2026.11729888.
  51. [52]Jian Cheng Wong, Chin Chun Ooi, Abhishek Gupta, Pao-Hsiung Chiu, Joshua Shao Zheng Low, My Ha Dao, and Yew-Soon Ong. Evolutionary optimization of physics-informed neural networks: Advancing generalizability by the baldwin effect. IEEE Transactions on Evolutionary Computation, 2026c.
  52. [53]Chenhui Xu, Dancheng Liu, Amir Nassereldine, and Jinjun Xiong. Fp64 is all you need: Rethinking failure modes in physics-informed neural networks. Advances in Neural Information Processing Systems, 2025.arxiv.org/abs/2505.10949
  53. [54]Guancheng Zhou, Zelin Zeng, Yisi Luo, Qi Xie, and Deyu Meng. Tucker-fno: Tensor tucker-fourier neural operator and its universal approximation theory. In The Fourteenth International Conference on Learning Representations, 2026.

Paper details

vlnc:W_sywbg687
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents