Comparison of neural and spline representations for physics-informed learning

DOI10.48550/arXiv.2610.06294MaRDI QIDQ7367843arXivFDO

Authors R{é}gis Duvigneau

Publication date 5 October 2026

Published in arXiv

Tldr This paper conducts a comparison between neural networks and B-Splines within physics-informed learning framework for solving PDEs, finding that both approaches have their own merits and drawbacks.

Full work available at URL https://doi.org/10.48550/arXiv.2610.06294

arXiv classification

math.APmath.NAmath.OC

Cites work

• A comprehensive study of non-adaptive and residual-based adaptive sampling for physics-informed neural networks
• A practical guide to splines
• An algorithm for low-rank approximation of bivariate functions using splines
• An Expert's Guide to Training Physics-informed Neural Networks
• Characterizing possible failure modes in physics-informed neural networks
• Data-driven prediction of unsteady flow over a circular cylinder using deep learning
• Deep NURBS -- Admissible Physics-informed Neural Networks
• DGM: a deep learning algorithm for solving partial differential equations
• Exact imposition of boundary conditions with distance functions in physics-informed deep neural networks
• Fourier Neural Operator for Parametric Partial Differential Equations
• Higher order B-spline collocation at the Greville abscissae
• Isogeometric analysis for compressible flows using a discontinuous Galerkin method
• Isogeometric analysis. Toward integration of CAD and FEA
• Low rank tensor methods in Galerkin-based isogeometric analysis
• Multilayer feedforward networks are universal approximators
• On the limited memory BFGS method for large scale optimization
• Operator preconditioning
• Physics-informed neural networks with adaptive loss weighting algorithm for solving partial differential equations
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next
• Some properties of LR-splines
• The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
• Trainable Spline Representations for Physics-Informed Learning
• Transformers for modeling physical systems
• When and why PINNs fail to train: A neural tangent kernel perspective

This page was built for publication: Comparison of neural and spline representations for physics-informed learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7367843)