Sparsely connected neural network representation of Lagrange finite element function

DOI10.48550/arXiv.2609.23299MaRDI QIDQ7353451arXivFDO

Authors Yunqing Huang, Nianyu Yi, Jiaxiong Hao

Publication date 20 September 2026

Published in arXiv

Tldr We construct a sparsely connected neural network that exactly reproduces arbitrary-order Lagrange finite elements on simplicial meshes by incorporating local basis functions via barycentric coordinate transformations and shared degrees of freedom, enabling mesh-free interpolation and adaptivity in finite element methods for PDEs.

Full work available at URL https://doi.org/10.48550/arXiv.2609.23299

arXiv classification

math.NA

Cites work

• Approximation by superpositions of a sigmoidal function
• De Rham compatible deep neural network FEM
• Deep Neural Networks and Finite Elements of Any Order on Arbitrary Dimensions
• Deep ReLU networks and high-order finite element methods
• Deep ReLU networks and high-order finite element methods. II: Chebyšev emulation
• Finite Element Neural Network Interpolation. Part I: Interpretable and Adaptive Discretization for Solving PDEs
• HiDeNN-FEM: a seamless machine learning approach to nonlinear finite element analysis
• Hierarchical deep learning neural network (HiDeNN): an artificial intelligence (AI) framework for computational science and engineering
• Hierarchical deep-learning neural networks: finite elements and beyond
• Multilayer feedforward networks are universal approximators
• Neural Network Element Method for Partial Differential Equations
• Neural network-enhanced $hr$-adaptive finite element algorithm for parabolic equations
• On a Class of Finite Elements Generated by Lagrange Interpolation
• ReLU deep neural networks and linear finite elements
• Shallow ReLU neural networks and finite elements
• The Mathematical Theory of Finite Element Methods
• Unfitted finite element interpolated neural networks

This page was built for publication: Sparsely connected neural network representation of Lagrange finite element function

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7353451)