Pith. sign in

REVIEW 17 cited by

Physics-Informed Neural Operator for Learning Partial Differential Equations

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2111.03794 v4 pith:CGQWYTNP submitted 2021-11-06 cs.LG cs.NAmath.NA

classification cs.LGcs.NAmath.NA
keywords operatorpinoconstraintsdataneuraltrainingphysics-informedresolution
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

In this paper, we propose physics-informed neural operators (PINO) that combine training data and physics constraints to learn the solution operator of a given family of parametric Partial Differential Equations (PDE). PINO is the first hybrid approach incorporating data and PDE constraints at different resolutions to learn the operator. Specifically, in PINO, we combine coarse-resolution training data with PDE constraints imposed at a higher resolution. The resulting PINO model can accurately approximate the ground-truth solution operator for many popular PDE families and shows no degradation in accuracy even under zero-shot super-resolution, i.e., being able to predict beyond the resolution of training data. PINO uses the Fourier neural operator (FNO) framework that is guaranteed to be a universal approximator for any continuous operator and discretization-convergent in the limit of mesh refinement. By adding PDE constraints to FNO at a higher resolution, we obtain a high-fidelity reconstruction of the ground-truth operator. Moreover, PINO succeeds in settings where no training data is available and only PDE constraints are imposed, while previous approaches, such as the Physics-Informed Neural Network (PINN), fail due to optimization challenges, e.g., in multi-scale dynamic systems such as Kolmogorov flows.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 17 Pith papers

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

  1. CaLiSym: Learning Symplectic Dynamics of Real-World Systems through Structured Canonical Lifts

    cs.RO 2026-07 conditional novelty 7.0 of 10

    Lifting non-conservative, actuated, and contact-constrained robot dynamics into an exactly symplectic phase-space map yields state-of-the-art out-of-distribution autoregressive rollout error at low parameter and FLOP cost.

  2. Solver Exactness, Learned Flexibility: Equivariant Boundary-Correction Operators for Stokes Flow

    physics.flu-dyn 2026-06 unverdicted novelty 7.0 of 10

    Learning only the Stokes boundary correction on an exact free-space core yields a 5–16× more data-efficient operator than black-box DeepONet, with geometric generalization controlled by descriptor invariance and train...

  3. Hybrid Lagrangian-Eulerian Model for Lagrangian Fluid Simulation

    cs.CE 2026-08 conditional novelty 6.0 of 10

    A hybrid Lagrangian-Eulerian graph neural simulator with adaptive downsampling and cross-attention achieves state-of-the-art accuracy and rollout stability on particle-based fluid benchmarks.

  4. A Physics-Informed Neural Operator for Thermal Ranking of Low-Cost Wall Materials in Hot-Dry Climates

    cs.LG 2026-07 accept novelty 6.0 of 10

    A physics-informed neural operator, trained on 1,500 wall-simulation runs, reproduces the exact material ranking of a finite-difference solver and points to clay-straw adobe as the best low-cost choice for hot-dry rur...

  5. Physics-Informed Neural Operator for Warm-Starting Background-Decomposed and Preconditioned PSFD: Enabling Scalable 3-D EUV Mask Simulation

    physics.optics 2026-07 conditional novelty 6.0 of 10

    A physics-informed neural operator trained on the PSFD residual, not precomputed data, matches mask scattered-field predictions to MAE ~7e-3 and warms up the iterative solver to reach practical accuracy in ~2 minutes.

  6. Neptuna: A Comprehensive Machine Learning Framework for Benchmarking Complex Multiphase Flows

    physics.flu-dyn 2026-07 conditional novelty 6.0 of 10

    A new 2.4 TB benchmark shows no single ML surrogate dominates on shock-driven multiphase flows, and composite losses with SoftAdapt weighting improve interface and spectral fidelity.

  7. Quadrature-Aware Complex-Linear Neural Operator for Boundary-to-Field Prediction in Resonant Acoustics

    physics.flu-dyn 2026-07 conditional novelty 6.0 of 10

    A quadrature-aware complex-linear neural operator halves field error versus DeepONet and enforces exact source superposition for resonant cavity acoustics.

  8. Forward and Inverse Mantle Convection with Neural Operators

    physics.geo-ph 2026-01 conditional novelty 6.0 of 10

    Neural operators trained on 2D Rayleigh-Benard convection serve as fast surrogates for forward and reverse mantle convection, and a joint inversion with terminal temperature plus surface velocity history gives the mos...

  9. Missing Physics Discovery through Fully Differentiable Finite Element-Based Machine Learning

    cs.CE 2025-07 conditional novelty 6.0 of 10

    FEML couples a differentiable finite element solver with neural networks to learn missing constitutive and thermal laws from indirect observations, with demonstrations on synthetic problems.

  10. Neural Operators for Forward and Inverse Potential-Density Mappings in Classical Density Functional Theory

    physics.chem-ph 2025-06 conditional novelty 6.0 of 10

    In 1D hard-rod cDFT, Fourier neural operators learn the density-to-direct-correlation-function map more accurately than DeepONet variants and dense networks, with squared ReLU giving the best extrapolation.

  11. PDE-Transformer: Efficient and Versatile Transformers for Physics Simulations

    cs.LG 2025-05 conditional novelty 6.0 of 10

    PDE-Transformer, a diffusion-transformer variant with shifted-window attention, multi-scale token processing, and per-channel tokens, outperforms leading transformer and operator baselines for PDE surrogate modeling a...

  12. Feature Interaction Modeling for Physics-Informed Neural Networks and Neural Operators

    cs.LG 2026-07 conditional novelty 5.0 of 10

    Adding NFM-style bi-interaction layers to PINNs and DeepONets improves accuracy on several high-dimensional smooth PDEs and shock-dominated conservation laws, but not on low-dimensional smooth problems.

  13. DRIFT: Direct Reduced Fourier Transforms for Distributed Spectral Neural Operators

    cs.DC 2026-07 conditional novelty 5.0 of 10

    Distributed Fourier Neural Operators can compute their truncated spectra with local partial DFTs and two collectives on the kept modes, giving exact results with communication independent of grid resolution.

  14. Operator learning for models of tear film breakup

    math.NA 2026-01 conditional novelty 5.0 of 10

    Operator learning can approximate the inverse mapping from fluorescence intensity to tear film thickness and osmolarity on synthetic data, but predictions diverge from ODE-based reference fits on experimental data.

  15. Operator-based machine learning framework for generalizable prediction of unsteady treatment dynamics in stormwater infrastructure

    cs.CE 2025-07 conditional novelty 5.0 of 10

    A composite operator neural network predicts 3D unsteady flow and particle concentrations in a stormwater separator, matching CFD closely on most held-out storm events.

  16. Graph-Based Operator Learning from Limited Data on Irregular Domains

    cs.LG 2025-05 reject novelty 5.0 of 10

    GOLA combines attention-based graph message passing with a learnable Fourier encoder and reports lower relative L2 error than GKN on four 2D PDE benchmarks, especially with few training samples.

  17. The Fourier Spectral Transformer Networks For Efficient and Generalizable Nonlinear PDEs Prediction

    cs.LG 2025-07 reject novelty 3.0 of 10

    The paper trains a Transformer on Fourier spectral coefficients to surrogate 1D Burgers and 2D Navier-Stokes dynamics, but its claimed superiority over numerical and ML baselines is not demonstrated.

Pith tools