Efficient Shallow Ritz Method: 1D Semi-Linear Diffusion-Reaction and Allen-Cahn Equations

DOI10.48550/arXiv.2610.17553MaRDI QIDQ7395234zbMATH OpenarXivFDO

Authors Zhiqiang Cai, César Herrera

Publication date 13 October 2026

Published in arXiv

Tldr We extend the shallow Ritz method to a nonlinear diffusion-reaction equation with cubic reaction term, and develop a reduced block Newton (rBN) solver for the resulting non-convex optimization problem with $\mathcal{O}(n)$ computational cost. We also propose a time-stepping neural network approach with unconditional energy stability for Allen-Cahn equations.

Full work available at URL https://doi.org/10.48550/arXiv.2610.17553

arXiv classification

math.NA

Cites work

• A Discussion on Solving Partial Differential Equations using Neural Networks
• A neuron-wise subspace correction method for the finite neuron method
• A New Conservative Allen-Cahn Type Ohta-Kawaski Phase-Field Model for Diblock Copolymers and Its Numerical Approximations
• A second-order unconditionally stable method for the anisotropic dendritic crystal growth model with an orientation-field
• Active neuron least squares: a training method for multivariate rectified neural networks
• Adaptive Two-Layer ReLU Neural Network: II. Ritz Approximation to Elliptic PDEs
• Algorithms for Separable Nonlinear Least Squares Problems
• Convergence Analysis of Block Newton Methods for 1D Shallow Neural Network Approximation
• Deep least-squares methods: an unsupervised learning-based numerical method for solving elliptic PDEs
• DGM: a deep learning algorithm for solving partial differential equations
• Fast Iterative Solver For Neural Network Method: I. 1D Diffusion Problems
• Long time numerical simulations for phase-field problems using \(p\)-adaptive spectral deferred correction methods
• Neural‐network‐based approximations for solving partial differential equations
• Numerical approximations of Allen-Cahn and Cahn-Hilliard equations
• On the separable nonlinear least squares problems
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Problems
• Robust Training and Initialization of Deep Neural Networks: An Adaptive Basis Viewpoint
• Second order convex splitting schemes for periodic nonlocal Cahn-Hilliard and Allen-Cahn equations
• Splines (with optimal knots) are better
• The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
• The Differentiation of Pseudo-Inverses and Nonlinear Least Squares Problems Whose Variables Separate
• The Global Dynamics of Discrete Semilinear Parabolic Equations

This page was built for publication: Efficient Shallow Ritz Method: 1D Semi-Linear Diffusion-Reaction and Allen-Cahn Equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7395234)