Abstract:Machine learning has been successfully applied to various fields of scientific computing in recent years. In this work, we propose a sparse radial basis function neural network method to solve elliptic partial differential equations (PDEs) with multiscale coefficients. Inspired by the deep mixed residual method, we rewrite the second-order problem into a first-order system and employ multiple radial basis function neural networks (RBFNNs) to approximate unknown functions in the system. To aviod the overfitting due to the simplicity of RBFNN, an additional regularization is introduced in the loss function. Thus the loss function contains two parts: the $L_2$ loss for the residual of the first-order system and boundary conditions, and the $\ell_1$ regularization term for the weights of radial basis functions (RBFs). An algorithm for optimizing the specific loss function is introduced to accelerate the training process. The accuracy and effectiveness of the proposed method are demonstrated through a collection of multiscale problems with scale separation, discontinuity and multiple scales from one to three dimensions. Notably, the $\ell_1$ regularization can achieve the goal of representing the solution by fewer RBFs. As a consequence, the total number of RBFs scales like $\mathcal{O}(\varepsilon^{-n\tau})$, where $\varepsilon$ is the smallest scale, $n$ is the dimensionality, and $\tau$ is typically smaller than $1$. It is worth mentioning that the proposed method not only has the numerical convergence and thus provides a reliable numerical solution in three dimensions when a classical method is typically not affordable, but also outperforms most other available machine learning methods in terms of accuracy and robustness.

Application of multivariate bilinear neural network method to fractional partial differential equations

Interaction phenomenon for variable coefficient Kadomtsev-Petviashvili equation by utilizing variable coefficient bilinear neural network method

Bilinear neural network method to obtain the exact analytical solutions of nonlinear partial differential equations and its application to p-gBKP equation

Lump solution, interaction solution, and interference wave for the (3+1)-dimensional BKP-Boussinesq equation as well as analysis of BNNM model degradation

Fractal Solitons, Arbitrary Function Solutions, Exact Periodic Wave and Breathers for a Nonlinear Partial Differential Equation by Using Bilinear Neural Network Method

Legendre Neural Network Method for Several Classes of Singularly Perturbed Differential Equations Based on Mapping and Piecewise Optimization Technology

Multiple exact solutions for the dimensionally reduced p-gBKP equation via bilinear neural network method

Bilinear residual network method for solving the exactly explicit solutions of nonlinear evolution equations

M-lump solutions, lump-breather solutions, and N-soliton wave solutions for the KP-BBM equation via the improved bilinear neural network method using innovative composite functions

Physics-Informed Kernel Function Neural Networks for Solving Partial Differential Equations

Neural networks for bifurcation and linear stability analysis of steady states in partial differential equations

A neural networks-based numerical method for the generalized Caputo-type fractional differential equations

The exact analytical solutions of the [math]-dimensional extended Korteweg–de Vries equation using bilinear neural network method and bilinear residual network method

Application of Intelligent Paradigm through Neural Networks for Numerical Solution of Multiorder Fractional Differential Equations

Artificial neural network for solving the nonlinear singular fractional differential equations

An application of artificial neural networks for solving fractional higher-order linear integro-differential equations

A reduced order Schwarz method for nonlinear multiscale elliptic equations based on two-layer neural networks

Neural Operator: Graph Kernel Network for Partial Differential Equations

Subspace method based on neural networks for solving the partial differential equation

Deep neural network for solving differential equations motivated by Legendre-Galerkin approximation

Solving multiscale elliptic problems by sparse radial basis function neural networks