Abstract:This paper aims to analyze errors in the implementation of the Physics-Informed Neural Network (PINN) for solving the Allen--Cahn (AC) and Cahn--Hilliard (CH) partial differential equations (PDEs). The accuracy of PINN is still challenged when dealing with strongly non-linear and higher-order time-varying PDEs. To address this issue, we introduce a stable and bounded self-adaptive weighting scheme known as Residuals-RAE, which ensures fair training and effectively captures the solution. By incorporating this new training loss function, we conduct numerical experiments on 1D and 2D AC and CH systems to validate our theoretical findings. Our theoretical analysis demonstrates that feedforward neural networks with two hidden layers and tanh activation function effectively bound the PINN approximation errors for the solution field, temporal derivative, and nonlinear term of the AC and CH equations by the training loss and number of collocation points.

What problem does this paper attempt to address?

This paper aims to analyze the errors when using the physics - informed neural network (PINN) to solve the Allen - Cahn (AC) and Cahn - Hilliard (CH) partial differential equations (PDEs). Specifically, the paper focuses on: 1. **Error Estimation**: The paper explores the possible errors that may occur when using PINN to solve AC and CH equations and proposes a new adaptive weight scheme - Residuals - Region - Activation - Evaluation (Residuals - RAE) to improve the stability of training and the accuracy of prediction. 2. **Challenges of Non - linear and High - order Time - varying PDEs**: Traditional PINN faces challenges when dealing with PDEs with strong non - linearity and high - order time - variation. By introducing the Residuals - RAE scheme, the paper ensures a fair training process and effectively captures the solutions of these complex equations. 3. **Theoretical Analysis and Numerical Experiments**: Through theoretical analysis, the paper proves that the feed - forward neural network (with two hidden layers and the tanh activation function) can effectively limit the approximation errors of PINN for the solution fields, time derivatives and non - linear terms of AC and CH equations. In addition, the paper also conducts numerical experiments on 1D and 2D AC and CH systems to verify the theoretical results. 4. **Error Decomposition**: The paper decomposes the error in PINN into three parts: approximation error, generalization error and optimization error. The approximation error reflects the ability of the network to approximate the exact solution; the generalization error measures the difference between the prediction of the model on sample data and the true solution; the optimization error is caused by the optimization algorithm used to update the network weights. 5. **Adaptive Weight Scheme**: The paper proposes a residual - based point - weight scheme (Residuals - RAE). This scheme first uses residuals to obtain the weights of each residual point, then averages the weights of the nearest neighbor points, and finally uses this average as the weight of this point. This method can keep the weight distribution as close as possible to the residual distribution during the training process while avoiding frequent changes in point weights. In summary, the main objective of this paper is to improve the accuracy and stability of PINN in solving strongly non - linear and high - order time - varying PDEs (such as AC and CH equations) by introducing the Residuals - RAE scheme, and the effectiveness of this scheme is verified through theoretical analysis and numerical experiments.

A Priori Error Estimation of Physics-Informed Neural Networks Solving Allen--Cahn and Cahn--Hilliard Equations

Error Analysis of Physics-Informed Neural Networks for Approximating Dynamic PDEs of Second Order in Time

A Unified Framework for the Error Analysis of Physics-Informed Neural Networks

Higher-order error estimates for physics-informed neural networks approximating the primitive equations

Physics Informed Neural Networks (PINNs) as intelligent computing technique for solving partial differential equations: Limitation and Future prospects

Residual-based error bound for physics-informed neural networks

Self-adaptive loss balanced Physics-informed neural networks

Physics-informed neural networks based on adaptive weighted loss functions for Hamilton-Jacobi equations

Efficient Error Certification for Physics-Informed Neural Networks

Estimates on the generalization error of Physics Informed Neural Networks (PINNs) for approximating PDEs

Understanding the Difficulty of Solving Cauchy Problems with PINNs

Adaptive Learning Rate Residual Network Based on Physics-Informed for Solving Partial Differential Equations

Estimates on the generalization error of Physics Informed Neural Networks (PINNs) for approximating a class of inverse problems for PDEs

Recipes for when physics fails: recovering robust learning of physics informed neural networks

Enhancing PINNs for solving PDEs via adaptive collocation point movement and adaptive loss weighting

A Physics Informed Neural Network for Time-Dependent Nonlinear and Higher Order Partial Differential Equations

A simple remedy for failure modes in physics informed neural networks

Loss-attentional physics-informed neural networks

Is Physics Informed Loss Always Suitable for Training Physics Informed Neural Network?

Error estimates and physics informed augmentation of neural networks for thermally coupled incompressible Navier Stokes equations