Pith. sign in

REVIEW 20 cited by

An Expert's Guide to Training Physics-informed Neural Networks

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 2308.08468 v1 pith:GCHB6OGV submitted 2023-08-16 cs.LG cs.NAmath.NAphysics.comp-ph

classification cs.LGcs.NAmath.NAphysics.comp-ph
keywords trainingpinnsstudiesaccuracychoicesdeepforthfuture
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Physics-informed neural networks (PINNs) have been popularized as a deep learning framework that can seamlessly synthesize observational data and partial differential equation (PDE) constraints. Their practical effectiveness however can be hampered by training pathologies, but also oftentimes by poor choices made by users who lack deep learning expertise. In this paper we present a series of best practices that can significantly improve the training efficiency and overall accuracy of PINNs. We also put forth a series of challenging benchmark problems that highlight some of the most prominent difficulties in training PINNs, and present comprehensive and fully reproducible ablation studies that demonstrate how different architecture choices and training strategies affect the test accuracy of the resulting models. We show that the methods and guiding principles put forth in this study lead to state-of-the-art results and provide strong baselines that future studies should use for comparison purposes. To this end, we also release a highly optimized library in JAX that can be used to reproduce all results reported in this paper, enable future research studies, as well as facilitate easy adaptation to new use-case scenarios.

Discussion (0). Sign in to comment.

Forward citations

Cited by 20 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 66 citations worldwide. Full citation record

  1. Optimal Control with Expectation Constraint in a Smooth Boundary Case

    math.OC 2026-07 conditional novelty 6.0 of 10

    A Dirichlet boundary PDE is derived for expectation-constrained stochastic control, with convergent bounded-control and noise approximations and a neural-network numerical demonstration.

  2. Material-agnostic temperature field prediction for metal additive manufacturing via a parametric PINN framework

    cs.LG 2026-04 unverdicted novelty 6.0 of 10

    One physics-informed network trained on the heat equation, not labeled data, predicts laser-scan temperature fields for unseen alloys including copper with ~1% relative error.

  3. PIDT: Physics-Informed Digital Twin for Optical Fiber Parameter Estimation

    eess.SP 2026-01 conditional novelty 6.0 of 10

    PIDT replaces the neural network in physics-informed neural operators with a parameterized split-step Fourier method, improving fiber parameter estimation accuracy and speed.

  4. LieSolver: PDE-Constrained Learning for IBVPs via Lie Symmetries

    cs.LG 2025-10 conditional novelty 6.0 of 10

    LieSolver fits initial/boundary data with linear combinations of Lie-symmetry-generated base solutions, enforcing linear homogeneous PDEs exactly by construction.

  5. Estimation of Hemodynamic Parameters via Physics Informed Neural Networks including Hematocrit Dependent Rheology

    math.NA 2025-08 conditional novelty 6.0 of 10

    PINNs reconstruct smooth velocity and pressure fields from synthetic 4D-flow MRI data for anemic to polycythemic blood, and combining the PINN velocity field with the vWERP estimator gives the most accurate pressure drops.

  6. BWLer: Barycentric Weight Layer Elucidates a Precision-Conditioning Tradeoff for PINNs

    cs.LG 2025-06 conditional novelty 6.0 of 10

    Adding a barycentric interpolation layer to PINNs lifts their precision ceiling, achieving up to 1e-13 relative error on smooth PDEs, while exposing a tradeoff between accuracy and loss conditioning.

  7. Physics-Regularized Machine Learning for Proprioceptive Vehicle Localization Using Onboard Sensors

    cs.RO 2026-07 conditional novelty 5.5 of 10

    End-to-end training of a transformer measurement model through a differentiable EKF yields lower 60 s onboard-sensor dead-reckoning error and better low-µ generalization than pure ML or classical baselines.

  8. Latent PDE mapping for efficient physics-informed learning across geometries with limited data

    cs.LG 2026-07 conditional novelty 5.0 of 10

    By mapping PDE residuals and boundary conditions to a fixed latent geometry through deformation gradients, latent PDE mapping improves out-of-distribution geometric generalization in PINNs and PI-DONs for the Aliev-Pa...

  9. Trainable Spline Representations for Physics-Informed Learning

    cs.LG 2026-07 conditional novelty 5.0 of 10

    A tensor-product B-spline whose coefficients are learned by minimizing PDE residuals achieves lower error than standard PINNs on four benchmark problems with far fewer parameters.

  10. Continuously Tempered Diffusion Samplers

    cs.LG 2025-08 conditional novelty 5.0 of 10

    CTDS trains neural samplers with a controlled Langevin dynamics over both position and a continuous temperature coordinate, and reports improved sampling on a 40-mode Gaussian mixture.

  11. Solved in Unit Domain: JacobiNet for Differentiable Coordinate-Transformed PINNs

    cs.LG 2025-08 conditional novelty 5.0 of 10

    A learned coordinate-mapping neural network is inserted in front of a PINN so that PDEs on irregular domains can be solved in a unit reference domain with autograd Jacobians and hard boundary constraints, improving ac...

  12. Physics-Informed PointNets for Modeling Electromagnetic Scattering from All-Dielectric Metasurfaces with Inclined Nanopillars

    physics.optics 2025-07 conditional novelty 5.0 of 10

    A physics-informed PointNet that encodes spatially varying permittivity predicts near-field and far-field scattering from metasurfaces with inclined nanopillars, reaching 1.7% MAPE for low-contrast 2D cases.

  13. Quantifying data needs in surrogate modeling for flow fields in two-dimensional stirred tanks with physics-informed neural networks

    cs.CE 2025-07 conditional novelty 5.0 of 10

    A vanilla PINN trained with six strategically placed velocity labels, or with approximate velocity labels, can predict 2D stirred-tank flow with roughly 2.5-3% error across most tested Reynolds numbers, far fewer labe...

  14. Physics-informed machine learning surrogate for scalable simulation of thermal histories during wire-arc directed energy deposition

    cs.CE 2025-07 conditional novelty 5.0 of 10

    A data-free physics-informed neural network matches finite-element thermal histories of wire-arc directed energy deposition to about 7% relative L2 error, reporting up to 98.6% compute-time reduction versus a fine-mes...

  15. Geometric flow regularization in latent spaces for smooth dynamics with the efficient variations of curvature

    math.NA 2025-06 conditional novelty 5.0 of 10

    Curvature-flow-regularized latent spaces improve mean out-of-distribution errors for Burger's equation relative to a plain autoencoder, but the flows are heuristic and the evidence is limited.

  16. Leveraging KANs for Expedient Training of Multichannel MLPs via Preconditioning and Geometric Refinement

    cs.LG 2025-05 conditional novelty 5.0 of 10

    Training in a B-spline KAN basis is equivalent to preconditioned gradient descent on a multichannel ReLU MLP, and geometric refinement plus trainable knots accelerate and improve training.

  17. Principled Approaches for Extending Neural Architectures to Function Spaces for Operator Learning

    cs.LG 2025-06 conditional novelty 4.0 of 10

    A practical recipe to convert common neural architectures into discretization-agnostic neural operators, validated by Navier-Stokes experiments showing cross-resolution generalization of FNO-style models.

  18. Physics-Informed Deep Learning for Nonlinear Friction Model of Bow-string Interaction

    eess.AS 2025-05 conditional novelty 4.0 of 10

    PINNs solve the nonlinear bowed mass-spring system for all tested bow forces, PI-DeepONets fail at high bow force, and adding supervised data rescues them.

  19. Uncertainty quantification in mechanics: A unified Bayesian perspective

    physics.comp-ph 2026-07 conditional novelty 2.0 of 10

    Bayesian probability theory is presented as the single framework that unifies forward propagation, inverse calibration, surrogate modeling, model selection, experimental design, and sensitivity analysis in mechanics.

  20. Cardiovascular Digital Twins from Physics Based to Data Driven Approaches

    physics.med-ph 2026-08 unverdicted novelty 1.0 of 10

    A review of cardiovascular digital twin modeling paradigms that concludes hybrid physics-informed and graph-based methods are the most promising direction for clinical deployment.

Pith tools