betaFeedback

WorksComputer Science

Function-Space Certification for Neural Operators

Abstract

Certifying the robustness and physical consistency of scientific machine-learning models ensures that their predictions remain reliable beyond the test set. For neural operators, which learn PDE solution operators by mapping infinite-dimensional function spaces, certification means establishing rigorous, domain-wide guarantees for their output functions. However, standard neural network (NN) verification methods are fundamentally designed for vector-to-vector mappings, and they can only certify neural operators at fixed grid resolutions and provide no guarantees over the entire domain or across another resolution. We close this gap by introducing a function-space certification framework that verifies neural operators across the entire continuous domain. We formulate interval bound propagation (IBP) and CROWN in function space, propagating upper and lower bound functions through each layer using rigorous numerical integration. In addition, we extend bound propagation to higher-order derivatives to certify PDE residuals and solution error bounds. Experiments across several neural operator architectures and scientific problems, including the 572.5-million-parameter spherical Fourier neural operator for global weather forecasting, show that our method produces tight, domain-wide guarantees efficiently, providing a foundation for trustworthy operator learning.

Introduction

Machine learning models are often deployed in real-world applications where test accuracy alone is insufficient, and properties such as safety, stability, and robustness are critical factors in determining model reliability. In scientific machine learning, reliability further requires physical consistency, adherence to physical constraints, and control of worst-case errors. In particular, data-driven approaches to solving PDEs [1, 2, 3] generally lack certified error bounds [4] over the continuous domain, compared to traditional numerical methods [5, 6] which offer convergence criteria and a posteriori error estimates [7]. This has driven research in neural network certification [8], which provides provable guarantees that a desired property holds for every input in a prescribed set, rather than only for a finite collection of test examples.

A major class of certification methods is based on bound propagation. Interval bound propagation (IBP) [9] propagates lower and upper intervals through the layers of a neural network, while CROWN [10, 11, 12, 13] constructs tighter affine upper and lower bounds that retain dependencies through the network. These algorithms establish adversarial robustness, guaranteeing that model outputs remain bounded over a set of inputs. Building on CROWN, ∂\partial-CROWN [4] extends bound propagation to higher-order derivatives of physics-informed neural networks (PINNs), allowing certification of (1) initial conditions, (2) boundary conditions, and (3) PDE residual errors.

However, these methods are primarily developed for finite-dimensional neural networks. Neural operators instead learn mappings between function spaces, for example, mapping a forcing field to the solution field of a partial differential equation (PDE). The Fourier neural operator (FNO) [2] is a widely used operator learning architecture, while physics-informed neural operators (PINOs) [14] can additionally be trained with PDE residual losses to encourage physical consistency. For such models, certification should provide guarantees on the entire predicted function, rather than only on a finite-dimensional vector.

Overview of function-space certification for neural operators

Figure 1. Overview of function-space certification for neural operators. Function-space IBP and CROWN propagate the input through the neural operator. Each spectral convolution is an integral, which is enclosed with a rigorous error bound. The resulting upper and lower bound functions, and their derivatives, hold at every point of the domain and certify robustness, PDE residuals, boundary/initial conditions, and the solution error through a stability estimate.

In practice, neural operators are implemented and evaluated on finite grids. Directly applying standard NN verification methods therefore certifies the model only at the grid points of the chosen resolution. Their certificate does not translate to arbitrary resolutions or to the continuum operator. When perturbations occur between grid points, grid-based verifiers may fail to capture their effects on the output function. This is distinct from discretization convergence: even when a discretized neural operator converges to a continuum operator as the resolution increases, certification at one fixed resolution does not by itself certify all admissible functions consistent with those samples.

We illustrate this with a simple counterexample: consider an input function a(x)a(x) and perturbation r(x)r(x) defined as

a(x)=sin⁡(2πx),r(x)=ε2(1−cos⁡(2πNx)),a(x) = \sin(2\pi x), \qquad r(x) = \frac{\varepsilon}{2}\left(1-\cos(2\pi Nx)\right),

over [0,1)[0,1) with N=8N = 8 and ε=0.2\varepsilon= 0.2, and a neural operator GθG_\theta with Fourier multipliers R0=1R_0 = 1, R−1=R1=1/2R_{-1} = R_1 = 1/2, and all others zero.

Gθ(a)(x)=F−1(R⋅F(a))(x).G_\theta(a)(x) = \mathcal{F}^{-1}\left(R\cdot\mathcal{F}(a)\right)(x).

where F\mathcal{F} denotes the Fourier transform. Then

Gθ(a)(x)=12sin⁡(2πx),Gθ(a+)(x)=12sin⁡(2πx)+ε2.G_\theta(a)(x) = \frac{1}{2}\sin(2\pi x), \qquad G_\theta(a_+)(x) = \frac{1}{2}\sin(2\pi x) + \frac{\varepsilon}{2}.

However, at the sampled locations xj=j/Nx_j = j/N, we have r(xj)=0r(x_j) = 0, so a(x)a(x) and a+(x)=a(x)+r(x)a_+(x) = a(x) + r(x) produce identical input vectors, thus grid-based verification observes zero input width at resolution NN even though the continuum outputs differ everywhere. Figure 2 illustrates this gap between grid and function-space certification.

Grid versus function-space certification counterexample

Figure 2. Grid versus function-space certification counterexample. Given two input functions that match at grid points, running CROWN on grid points (grid CROWN) can overlook perturbations between points, while function-space CROWN soundly bounds the outputs across the entire domain.

Our approach. This motivates the study of neural operator certification in function space, where we formulate IBP and CROWN to efficiently propagate upper and lower bound functions for the continuum FNO. Following the layer-by-layer propagation principle of standard IBP/CROWN, we derive propagation rules for each component of the continuum FNO, including pointwise activations, lifting/projection layers, spectral convolutions, and normalization. Since our derived bounds are functions, they certify every point in the domain for every function in the input set. Because spectral convolutions are integrals, we introduce a certified integration rule that computes integrals with rigorous error bounds using the Gaussian-Legendre quadrature [15, 16].

We show that residual, boundary, and initial condition errors, which constitute the three PINN correctness conditions in ∂-CROWN [4], can be certified for the neural operator by extending it to higher-order derivatives. For PDEs with suitable stability estimates, we further convert these errors into bounds on the neural operator’s solution error [17] without knowledge of the true PDE solution. This provides a post-training verification mechanism for learned surrogates over the continuous domain, analogous to the tolerance used to assess the accuracy of classical numerical solvers.

We evaluate our framework on neural operators of various sizes and architectures. Across the Poisson, Helmholtz, and Burgers’ equations, we certify adversarial robustness, PDE residuals, and solution errors. To evaluate certification over long-horizon rollouts, we test our framework on the Kuramoto–Sivashinsky equation [18] for nonlinear chaotic dynamics. Finally, we scale our method to real-world global weather forecasting [19], certifying maximum regional temperature predictions for a 572.5-million-parameter pretrained spherical Fourier neural operator (SFNO) [20]. The results show that our framework produces tight, domain-wide bounds across a diverse set of tasks while requiring no more than a few hours on a single GPU.

Methods

Problem formulation

We consider PDEs with boundary and initial conditions:

La[u](x)=0,x∈Ω,Ba[u](x)=0,x∈Γb,Ia[u](x)=0,x∈Γ0.(1)\begin{aligned} \mathcal{L}_a[u](x) &= 0, \quad x \in\Omega, \\ \mathcal{B}_a[u](x) &= 0, \quad x \in\Gamma_b, \\ \mathcal{I}_a[u](x) &= 0, \quad x \in\Gamma_0. \tag*{(1)} \end{aligned}

Here Ω\Omega is the spatial or space–time domain, and La\mathcal{L}_a, Ba\mathcal{B}_a, and Ia\mathcal{I}_a encode the governing equation, boundary conditions, and initial conditions, respectively. Let G:a↦u\mathcal{G}: a \mapsto u denote the solution or forecasting operator. The neural operator Gθ\mathcal{G}_\theta approximates G\mathcal{G}, with uθ=Gθ(a)u_\theta= \mathcal{G}_\theta(a).

We define the continuum FNO as a function-to-function mapping through

z0(x)=P(a(x)),zℓ+1(x)=σℓ(Wℓzℓ(x)+bℓ+∫Ωκℓ(x−y)zℓ(y) dy),z_0(x) = \mathcal{P}(a(x)), \qquad z_{\ell+1}(x) = \sigma_\ell\left(W_\ell z_\ell(x) + b_\ell+ \int_\Omega\kappa_\ell(x-y)z_\ell(y)\,\mathrm{d}y\right),

for ℓ=0,…,n−1\ell= 0,\ldots,n-1, where nn is the number of FNO layers, followed by

u(x)=Q(zn(x)).u(x) = \mathcal{Q}(z_n(x)).

Here P\mathcal{P} and Q\mathcal{Q} are pointwise lifting and projection maps, σℓ\sigma_{\ell} is the pointwise activation, WℓW_{\ell} is a local linear transformation, and the kernel is κℓ(t)=1L∑k=−(m−1)m−1Rℓ,ke2πikt/L\kappa_{\ell}(t)=\frac{1}{L}\sum_{k=-(m-1)}^{m-1}R_{\ell,k}e^{2\pi ikt/L}. At resolution NN, the discrete FNO Gθ,N\mathcal{G}_{\theta,N} replaces each integral by the sum

LN∑j=0N−1κℓ(xi−xj)zℓ(xj),xj=jL/N,\frac{L}{N}\sum_{j=0}^{N-1}\kappa_{\ell}(x_i-x_j)z_{\ell}(x_j),\qquad x_j=jL/N,

evaluated using the fast Fourier transform (FFT).

Following prior NN verification work [8, 9, 10] and ∂\partial-CROWN [4], our framework verifies the following conditions for a neural operator Gθ\mathcal{G}_{\theta}:

  1. Regional maxima. For a predicted field uθ=Gθ(a)u_{\theta}=\mathcal{G}_{\theta}(a) over a region RR, we certify

max⁡x∈Ruθ(x)≤umax⁡.\max_{x\in R}u_{\theta}(x)\leq u_{\max}.
  1. Adversarial robustness. We consider a smooth input set A⊂X\mathcal{A}\subset\mathcal{X},

A={a∣a‾(x)≤a(x)≤a‾(x) ∀x∈Ω, ∣a′(x)∣≤Ma}.\mathcal{A}=\{a\mid\underline{a}(x)\leq a(x)\leq\overline{a}(x)\ \forall x\in\Omega,\ |a'(x)|\leq M_a\}.

We construct upper/lower bound functions satisfying

u‾(x)≤uθ(x)≤u‾(x)∀a∈A, ∀x∈Ω.\underline{u}(x)\leq u_{\theta}(x)\leq\overline{u}(x)\qquad\forall a\in\mathcal{A},\ \forall x\in\Omega.

The formal theorem statement and proof are in Appendix § 15.

  1. PDE error. For a fixed input aa, we bound the PDE residual over the entire domain:

∥La[Gθ(a)]∥≤εr,(2)\lVert\mathcal{L}_a[\mathcal{G}_{\theta}(a)]\rVert\leq\varepsilon_r, \tag*{(2)}

In addition, we bound the boundary and initial condition errors as

∥Ba[Gθ(a)]∥≤δb,∥Ia[Gθ(a)]∥≤δ0.(3)\lVert\mathcal{B}_a[\mathcal{G}_{\theta}(a)]\rVert\leq\delta_b,\qquad\lVert\mathcal{I}_a[\mathcal{G}_{\theta}(a)]\rVert\leq\delta_0. \tag*{(3)}
  1. Solution error. Combining residual and boundary/initial conditions, we bound the error relative to the exact solution u=G(a)u=\mathcal{G}(a):

∥Gθ(a)−u∥≤δu.(4)\lVert\mathcal{G}_{\theta}(a)-u\rVert\leq\delta_u. \tag*{(4)}

Certified integration

Each spectral convolution layer integrates a hidden function zℓ(y)z_{\ell}(y) against a smooth kernel κℓ(x−y)\kappa_{\ell}(x-y). Standard numerical evaluation simply approximates this by a weighted sum. For rigorous certification, we instead bound the approximation error explicitly so that the true integral is guaranteed to lie within our lower and upper bounds. Inspired by Arb [16], we use Gaussian quadrature, and choose its error bound depending on whether the integrand is smooth.

  • Smooth case. When the integrand is analytic, as for tanh networks, we partition the domain into cells Cj=[xj,xj+h)C_j=[x_j,x_j+h) of width h=L/Qh=L/Q and use the Gauss–Legendre quadrature [15, 21]. It chooses nn points xix_i and weights wiw_i,

∫−11f(x) dx≈∑i=1nwif(xi),\int_{-1}^{1}f(x)\,\mathrm{d}x\approx\sum_{i=1}^{n}w_i f(x_i),

such that the formula integrates all polynomials of degree up to 2n−12n-1 exactly. In this case, xix_i is chosen to be the roots of the Legendre polynomial Pn(x)P_n(x). The error is bounded by a Cauchy remainder that decays as 4−2n4^{-2n}. We define QQ as the quadrature resolution.

  • Non-smooth case. In the case where the integrand is not analytic, for example when the model uses ReLU activations or when the input is perturbed, the Cauchy remainder no longer decays exponentially. We instead bound the error using only the slope of the integrand. We apply the two-point Gauss–Legendre rule to each cell, with nodes xj,kx_{j,k} and weights h/2h/2. Suppose we are given f‾j,k\underline{f}_{j,k}, f‾j,k\overline{f}_{j,k}, Mf,jM_{f,j} such that

f‾j,k≤f(xj,k)≤f‾j,k,sup⁡s∈Cj∣f′(s)∣≤Mf,j.\underline{f}_{j,k} \le f(x_{j,k}) \le\overline{f}_{j,k}, \qquad\sup_{s \in C_j} |f'(s)| \le M_{f,j}.

Then the continuum integral satisfies

h2∑j,kf‾j,k−ηf≤∫Ωf(s) ds≤h2∑j,kf‾j,k+ηf,ηf=dh2∑j=0Q−1Mf,j,(5)\frac{h}{2}\sum_{j,k}\underline{f}_{j,k}-\eta_f \le\int_{\Omega} f(s)\,\mathrm{d}s \le\frac{h}{2}\sum_{j,k}\overline{f}_{j,k}+\eta_f,\qquad\eta_f=d h^2\sum_{j=0}^{Q-1}M_{f,j}, \tag*{(5)}

References

  1. [1]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: 10.1038/s42254-021-00314-5.DOI
  2. [2]Zongyi Li, Nikola Kovachki, Kamyar Azizadehnesheli, Burigede Liu, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Fourier neural operator for parametric partial differential equations. In International Conference on Learning Representations, 2021. URL https://arxiv.org/abs/2010.08895.
  3. [3]Lu Lu, Pengzhan Jin, Guofei Pang, Zhongqiang Zhang, and George Em Karniadakis. Learning nonlinear operators via deeponet based on the universal approximation theorem of operators. Nature Machine Intelligence, 3(3):218–229, March 2021. ISSN 2522-5839. doi: 10.1038/s42256-021-00302-5. URL http://dx.doi.org/10.1038/s42256-021-00302-5.
  4. [4]Francisco Eiras, Adel Bibi, Rudy R. Bunel, Krishnamurthy Dvijotham, Philip Torr, and M. Pawan Kumar. Efficient error certification for physics-informed neural networks. In Proceedings of the 41st International Conference on Machine Learning, volume 235 of Proceedings of Machine Learning Research, pages 12318–12347. PMLR, 2024. URL https://proceedings.mlr.press/v235/eiras24a.html.arxiv.org/abs/2305.10157
  5. [5]Susanne C. Brenner and L. Ridgway Scott. The Mathematical Theory of Finite Element Methods. Springer, 3 edition, 2008. doi: 10.1007/978-0-387-75934-0.DOI
  6. [6]Lloyd N. Trefethen. Spectral Methods in MATLAB. Society for Industrial and Applied Mathematics, 2000. doi: 10.1137/1.9780898719598.DOI
  7. [7]Rüdiger Verfürth. A Posteriori Error Estimation Techniques for Finite Element Methods. Oxford University Press, 04 2013. ISBN 9780199679423. doi: 10.1093/acprof:oso/9780199679423.001.0001. URL https://doi.org/10.1093/acprof:oso/9780199679423.001.0001.
  8. [8]Guy Katz, Clark Barrett, David Dill, Kyle Julian, and Mykel Kochenderfer. Ruleplex: An efficient smt solver for verifying deep neural networks. In Computer Aided Verification, pages 97–117, 2017. doi: 10.1007/978-3-319-63387-9_5. URL https://arxiv.org/abs/1702.01135.
  9. [9]Sven Gowal, Krishnamurthy Dvijotham, Robert Stanforth, Rudy Bunel, Chongli Qin, Jonathan Uesato, Relja Arandjelovic, Timothy Mann, and Pushmeet Kohli. Scalable verified training for provably robust image classification. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019. URL https://arxiv.org/abs/1810.12715.
  10. [10]Huan Zhang, Tsui-Wei Weng, Pin-Yu Chen, Cho-Jui Hsieh, and Luca Daniel. Efficient neural network robustness certification with general activation functions. Advances in Neural Information Processing Systems, 31:4939–4948, 2018. URL https://arxiv.org/pdf/1811.00866.pdf.
  11. [11]Kaidi Xu, Zhouxing Shi, Huan Zhang, Yihan Wang, Kai-Wei Chang, Minlie Huang, Bhavya Kailkhura, Xue Lin, and Cho-Jui Hsieh. Automatic perturbation analysis for scalable certified robustness and beyond. In Advances in Neural Information Processing Systems, volume 33, 2020. URL https://arxiv.org/abs/2002.12920.
  12. [12]Kaidi Xu, Huan Zhang, Shiqi Wang, Yihan Wang, Suman Jana, Xue Lin, and Cho-Jui Hsieh. Fast and Complete: Enabling complete neural network verification with rapid and massively parallel incomplete verifiers. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=nVZtXb6lN.arxiv.org/abs/2011.13824
  13. [13]Shiqi Wang, Huan Zhang, Kaidi Xu, Xue Lin, Suman Jana, Cho-Jui Hsieh, and J. Zico Kolter. Beta-CROWN: Efficient bound propagation with per-neuron split constraints for neural network robustness verification. In Advances in Neural Information Processing Systems, volume 34, 2021. URL https://arxiv.org/abs/2103.06624.
  14. [14]Zongyi Li, Hongkai Zheng, Nikola Kovachki, David Jin, Haoxuan Chen, Burigede Liu, Kamyar Azizadehnesheli, and Anima Anandkumar. Physics-informed neural operator for learning partial differential equations. ACM/IMS Journal of Data Science, 1(3), 2024. doi: 10.1145/3648506. URL https://arxiv.org/abs/2111.03794.
  15. [15]Lloyd N. Trefethen. Is gauss quadrature better than clenshaw–curtis? SIAM Review, 50(1):67–87, 2008. doi: 10.1137/060659831. URL https://doi.org/10.1137/060659831.
  16. [16]Fredrik Johansson. Numerical integration in arbitrary-precision ball arithmetic. In Mathematical Software – ICMS 2018, pages 255–263, 2018. doi: 10.1007/978-3-319-96418-8_30. URL https://arxiv.org/abs/1802.07942.
  17. [17]Birgit Hillebrecht and Benjamin Unger. Rigorous a posteriori error bounds for pde-defined pinns. IEEE Transactions on Neural Networks and Learning Systems, 36(1):1583–1593, January 2025. ISSN 2162-2388. doi: 10.1109/tnnls.2023.3335837. URL http://dx.doi.org/10.1109/TNNLS.2023.3335837.
  18. [18]Chuwei Wang, Boris Bonev, Julius Berner, Zongyi Li, Di Zhou, Jiayun Wang, Thorsten Kurth, Jane Bae, and Anima Anandkumar. Coarse graining with neural operators for simulating chaotic systems, 2026. URL https://arxiv.org/abs/2408.05177.
  19. [19]Earth2Studio Contributors. NVIDIA Earth2Studio, April 2024. URL https://github.com/NVIDIA/earth2studio.
  20. [20]Boris Bonev, Thorsten Kurth, Christian Hundt, Jaideep Pathak, Maximilian Baust, Karthik Kashinath, and Anima Anandkumar. Spherical fourier neural operators: Learning stable dynamics on the sphere. In Proceedings of the 40th International Conference on Machine Learning, volume 202 of Proceedings of Machine Learning Research, pages 2806–2823, 2023. URL https://arxiv.org/abs/2306.03838.
  21. [21]Lloyd N. Trefethen. Six myths of polynomial interpolation and quadrature. Math. Today (Southend-on-Sea), 47(4):184–188, 2011. URL https://people.maths.ox.ac.uk/trefethen/mythspaper.pdf.
  22. [22]Yousef Saad. Iterative Methods for Sparse Linear Systems. Society for Industrial and Applied Mathematics, 2 edition, 2003. doi: 10.1137/1.9780898718003.DOI
  23. [23]Ernst Hairer, Syvert P. Nørsett, and Gerhard Wanner. Solving Ordinary Differential Equations I: Nonstiff Problems, volume 8 of Springer Series in Computational Mathematics. Springer, 2 edition, 1993. doi: 10.1007/978-3-540-78862-1.
  24. [24]Tamara G. Grossmann, Urszula Julia Komorowska, Jonas Latz, and Carola-Bibiane Schönlieb. Can physics-informed neural networks beat the finite element method? IMA Journal of Applied Mathematics, 89(1):143–174, 2024. doi: 10.1093/imamat/hxae011.DOI
  25. [25]Nick McGreivy and Ammar Hakim. Weak baselines and reporting biases lead to overoptimism in machine learning for fluid-related partial differential equations. Nature Machine Intelligence, 6(10):1256–1269, 2024. doi: 10.1038/s42256-024-00897-5.DOI
  26. [26]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. doi: 10.1093/imanum/drab093.
  27. [27]Ivo Babuška and Werner C. Rhe inboldt. Error estimates for adaptive finite element computations. SIAM Journal on Numerical Analysis, 15(4):736–754, 1978. doi: 10.1137/0715049.
  28. [28]Mark Ainsworth and J. Tinsley Oden. A Posteriori Error Estimation in Finite Element Analysis. John Wiley & Sons, 2000. doi: 10.1002/9781118032824.DOI
  29. [29]Mitsuhiro T. Nakao, Michael Plum, and Yoshitaka Watanabe. Numerical Verification Methods and Computer-Assisted Proofs for Partial Differential Equations, volume 53 of Springer Series in Computational Mathematics. Springer Singapore, 2019. doi: 10.1007/978-981-13-7669-6.DOI
  30. [30]Tim De Ryck and Siddhartha Mishra. Generic bounds on the approximation error for physics-informed (and) operator learning. In Advances in Neural Information Processing Systems, volume 35, pages 10945–10958, 2022. URL https://papers.neurips.cc/paper_files/paper/2022/hash/46f0114c06524debc60ef2a72769f7a9-Abstract-Conference.html.arxiv.org/abs/2205.11393
  31. [31]Lewin Ernst and Karsten Urban. A certified wavelet-based physics-informed neural network for the solution of parameterized partial differential equations. IMA Journal of Numerical Analysis, 45(1):494–515, 2025. doi: 10.1093/imanum/drae011. URL https://doi.org/10.1093/imanum/drae011.
  32. [32]Lawrence C. Evans. Partial Differential Equations, volume 19 of Graduate Studies in Mathematics. American Mathematical Society, 2 edition, 2010. doi: 10.1090/gsm/019.DOI
  33. [33]David Gilbarg and Neil S. Trudinger. Elliptic Partial Differential Equations of Second Order. Classics in Mathematics. Springer, 2001. doi: 10.1007/978-3-642-61798-0.DOI
  34. [34]Nicholas Metropolis and Stanislaw Ulam. The Monte Carlo method. Journal of the American Statistical Association, 44(247):335–341, 1949. doi: 10.1080/01621459.1949.10483310.DOI
  35. [35]Zongyi Li, Miguel Liu-Schiaffini, Nikola Kovachki, Burigede Liu, Kamyar Azizzadenesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Learning dissipative dynamics in chaotic systems. In Advances in Neural Information Processing Systems, volume 35, 2022. URL https://arxiv.org/abs/2106.06898.
  36. [36]Thorsten Kurth, Shashank Subramanian, Peter Harrington, Jaideep Pathak, Morteza Mardani, David Hall, Andrea Miele, Karthik Kashinath, and Anima Anandkumar. Fourcastnet: Accelerating global high-resolution weather forecasting using adaptive fourier neural operators. In Proceedings of the Platform for Advanced Scientific Computing Conference, 2023. doi: 10.1145/3592979.3593412. URL https://arxiv.org/abs/2208.05419.
  37. [37]Boris Bonev, Thorsten Kurth, Ankur Mahesh, Mauro Bisson, Jean Kossaifi, Karthik Kashinath, Anima Anandkumar, William D. Collins, Michael S. Pritchard, and Alexander Keller. Fourcastnet 3: A geometric approach to probabilistic machine-learning weather forecasting at scale, 2025. URL https://arxiv.org/abs/2507.12144.
  38. [38]Hans Hersbach, Bill Bell, Paul Berrisford, Shoji Hirahara, András Horányi, Joaquín Muñoz-Sabater, Julien Nicolas, Carole Peubey, Raluca Radu, Dinand Schepers, Adrian Simmons, Cornel S00ci, Saleh Abdalla, Xavier Abellan, Gianpaolo Balsamo, Peter Bechtold, Gionata Biavati, Jean-Raymond Bidlot, Massimo Bonavita, Giovanna De Chiara, Per Dahlgren, Dick Dee, Michail Diamantakis, Rossana Dragani, Johannes Flemming, Richard Forbes, Manuel Fuentes, Alan Geer, Leopold Haimberger, Sean Healy, Robin J. Hogan, Elías Holm, Marta Janisková, Sarah Keeley, Patrick Lalaou, Philippe Lopez, Cristina Lupu, Gabor Radnoti, Patricia de Rosnay, Iryna Rozum, Freja Vamborg, Sébastien Villaume, and Jean-Noël Thé́paut. The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730):1999–2049, 2020. ISSN 1477-870X. doi: 10.1002/qj.3803.DOI
  39. [39]Ehsan Kharazmi, Zhongqiang Zhang, and George Em Karniadakis. hp-vpinns: Variational physics-informed neural networks with domain decomposition. Computer Methods in Applied Mechanics and Engineering, 374:113547, 2021. doi: 10.1016/j.cma.2020.113547.DOI
  40. [40]Yaohua Zang, Gang Bao, Xiaojing Ye, and Haomin Zhou. Weak adversarial networks for high-dimensional partial differential equations. Journal of Computational Physics, 411:109409, 2020. doi: 10.1016/j.jcp.2020.109409.DOI
  41. [41]Richard Courant, Kurt Friedrichs, and Hans Lewy. Über die partiellen differenzengleichungen der mathematischen physik. Mathematische Annalen, 100(1):32–74, 1928. doi: 10.1007/BF01448839.DOI
  42. [42]Peter D. Lax and Robert D. Richtmyer. Survey of the stability of linear finite difference equations. Communications on Pure and Applied Mathematics, 9(2):267–293, 1956. doi: 10.1002/cpa.3160090206.DOI
  43. [43]Zhouxing Shi, Qirui Jin, Zico Kolter, Suman Jana, Cho-Jui Hsieh, and Huan Zhang. Neural network verification with branch-and-bound for general nonlinearities. In International Conference on Tools and Algorithms for the Construction and Analysis of Systems, pages 315–335, 2025. doi: 10.1007/978-3-031-90643-5_17. URL https://arxiv.org/abs/2405.21063.
  44. [44]Gagandeep Singh, Timon Gehr, Matthew Mirman, Markus Püschel, and Martin Vechev. Fast and effective robustness certification. In Advances in Neural Information Processing Systems (NeurIPS), volume 31, 2018. URL https://proceedings.neurips.cc/paper_files/paper/2018/hash/f2f446980d8e971ef3da97af089481c3-Abstract.html.
  45. [45]Hadi Salman, Greg Yang, Huan Zhang, Cho-Jui Hsieh, and Pengchuan Zhang. A convex relaxation barrier to tight robustness verification of neural networks. Advances in Neural Information Processing Systems, 32:9835–9846, 2019. URL https://proceedings.neurips.cc/paper_files/paper/2019/hash/246a3c55441eb054f3ea718f61adfa16-Abstract.html.arxiv.org/abs/1902.08722
  46. [46]Huan Zhang, Shiqi Wang, Kaidi Xu, Linyi Li, Bo Li, Suman Jana, Cho-Jui Hsieh, and J Zico Kolter. General cutting planes for bound-propagation-based neural network verification. Advances in Neural Information Processing Systems, 2022. URL https://proceedings.neurips.cc/paper_files/paper/2022/hash/0b06c8673ebb453e5e468ff7743d8f54e-Abstract-Conference.html.arxiv.org/abs/2208.05740
  47. [47]Duo Zhou, Christopher Brix, Grani A Hanasusanto, and Huan Zhang. Scalable neural network verification with branch-and-bound inferred cutting planes. In The Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024. URL https://proceedings.neurips.cc/paper_files/paper/2024/hash/33d93e4dc57453e7667b20f62e7c0681-Abstract-Conference.html.arxiv.org/abs/2501.00200
  48. [48]Duo Zhou, Jorge Chavez, Hesun Chen, Grani A Hanasusanto, and Huan Zhang. Clip-and-verify: Linear constraint-driven domain clipping for accelerating neural network verification. In The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025. URL https://proceedings.neurips.cc/paper_files/paper/2025/hash/ffa977364ab7046c803da0e04dbb2832-Abstract-Conference.html.arxiv.org/abs/2512.11087
  49. [49]Liam Davis, Duo Zhou, Huan Zhang, Guy Katz, Clark Barrett, and Haoze Wu. Lookahead branching for neural network verification. In Proceedings of the International Joint Conference on Artificial Intelligence, 2026. URL https://arxiv.org/abs/2607.17290.
  50. [50]Suhas Kotha, Christopher Brix, J. Zico Kolter, Krishnamurthy Dvijotham, and Huan Zhang. Probably bounding neural network preimages. In A. Oh, T. Neumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems, volume 36, pages 80270–80290. Curran Associates, Inc., 2023. URL https://proceedings.neurips.cc/paper_files/paper/2023/file/fe061ec0ae03c5cf5b5323a2b912bfd-Paper-Conference.pdf.arxiv.org/abs/2302.01404
  51. [51]Zhouxing Shi, Yihan Wang, Huan Zhang, J Zico Kolter, and Cho-Jui Hsieh. Efficiently computing local lipschitz constants of neural networks via bound propagation. Advances in Neural Information Processing Systems, 35:2350–2364, 2022. URL https://proceedings.neurips.cc/paper_files/paper/2022/hash/0ff54b4ec4f70b3ae12c8621ca8a49f4-Abstract-Conference.html.arxiv.org/abs/2210.07394
  52. [52]Robert Joseph George, Jennifer Cruden, Will Adkisson, Xiangru Zhong, Huan Zhang, and Anima Anandkumar. Torchlearn: Formalizing neural networks in lean, 2026. URL https://arxiv.org/abs/2602.22631.
  53. [53]Abolaji D. Adesoji and Pin-Yu Chen. Evaluating the adversarial robustness for fourier neural operators, 2022. URL https://arxiv.org/abs/2204.04259.
  54. [54]Nisha Chandramoorthy, Daniel Sanz-Alonso, and Nathan Waniorek. From spectral methods to sample complexity bounds for fourier neural operators, 2026. URL https://arxiv.org/abs/2607.00320.
  55. [55]Ziqi Ma, David Pitt, Kamyar Aziz zad e hnesheli, and Anima Anandkumar. Calibrated uncertainty quantification for operator learning via conformal prediction. Transactions on Machine Learning Research, 2024. URL https://openreview.net/forum?id=cGpegxy12T.arxiv.org/abs/2402.01960
  56. [56]Johannes Gründler, Moritz Maibaum, and Philipp Petersen. Certified and accurate computation of function space norms of deep neural networks, 2026. URL https://arxiv.org/abs/2603.06431.
  57. [57]Ghifari Adam Faza, Jolan Wauters, Fabio Cuzzolin, Hans Hallez, and David Moens. Direct interval propagation methods using neural-network surrogates for uncertainty quantification in physical systems surrogate model. Knowledge-Based Systems, 341:115824, 2026. doi: 10.1016/j.knosys.2026.115824. URL https://arxiv.org/abs/2603.21308.
  58. [58]Ali Baheri and Ignacio Laguna Peralta. Can we formally verify neural pde surrogates? smt compilation of small fourier neural operators, 2026. URL https://arxiv.org/abs/2605.08938.
  59. [59]Gianluca Fabiani, Hannes Vandecasteele, Somdatta Goswami, Constantinos Siettos, and Ioannis G. Kevrekidis. Enabling local neural operators to perform equation-free system-level analysis. Nature Machine Intelligence, 8:1127–1141, 2026. doi: 10.1038/s42256-026-01265-1. URL https://doi.org/10.1038/s42256-026-01265-1.
  60. [60]Diab W. Abueida, Bilal Ahmed, Panos Pantidis, and Mostafa E. Mobasher. Physics-audited agentic discovery in scientific machine learning, 2026. URL https://arxiv.org/abs/2607.07379.
  61. [61]Randall J. LeVeque. Finite Difference Methods for Ordinary and Partial Differential Equations: Steady-State and Time-Dependent Problems. Society for Industrial and Applied Mathematics, 2007. doi: 10.1137/1.9780898717839.DOI
  62. [62]Alfio Quarteroni and Alberto Valli. Numerical Approximation of Partial Differential Equations, volume 23 of Springer Series in Computational Mathematics. Springer, 1994. doi: 10.1007/978-3-540-85268-1.DOI
  63. [63]Claudio Canuto, M. Yousuff Hussaini, Alfio Quarteroni, and Thomas A. Zang. Spectral Methods: Fundamentals in Single Domains. Scientific Computation. Springer, 2006. doi: 10.1007/978-3-540-30726-6.DOI
  64. [64]Nikola Kovachki, Zongyi Li, Burigede Liu, Kamyar Aziz zad e hnesheli, Kaushik Bhattacharya, Andrew Stuart, and Anima Anandkumar. Neural operator: Learning maps between function spaces with applications to pdes. Journal of Machine Learning Research, 24:1–97, 2023. URL https://jmlr.org/papers/v24/21-1524.html.
  65. [65]Samuel Lanthaler, Andrew M. Stuart, and Margaret Trautner. Discretization error of fourier neural operators. In International Conference on Learning Representations, 2025. URL https://arxiv.org/abs/2405.02221.
  66. [66]DLMF. NIST Digital Library of Mathematical Functions. https://dlmf.nist.gov/, Release 1.2.0 of 2024-03-15. URL https://dlmf.nist.gov/. F. W. J. Olver, A. B. Olde Daalhuis, D. W. Lozier, B. I. Schneider, R. F. Boisvert, C. W. Clark, B. R. Miller, B. V. Saunders, H. S. Cohl, and M. A. McClain, eds.
  67. [67]C. W. Clenshaw and A. R. Curtis. A method for numerical integration on an automatic computer. Numer. Math., 2(4):197–205, 1960. ISSN 0029-599X. doi: 10.1007/BF01386223.DOI

Paper details

vlnc:W_yakwurw4
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 11

downloads · 0

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

  • Computer Science

Large language models

View statement
Looking for related papers…
Contents