Abstract:In this article, we present a new learning method called sl-PINN to tackle the one-dimensional viscous Burgers problem at a small viscosity, which results in a singular interior layer. To address this issue, we first determine the corrector that characterizes the unique behavior of the viscous flow within the interior layers by means of asymptotic analysis. We then use these correctors to construct sl-PINN predictions for both stationary and moving interior layer cases of the viscous Burgers problem. Our numerical experiments demonstrate that sl-PINNs accurately predict the solution for low viscosity, notably reducing errors near the interior layer compared to traditional PINN methods. Our proposed method offers a comprehensive understanding of the behavior of the solution near the interior layer, aiding in capturing the robust part of the training solution.

What problem does this paper attempt to address?

This paper attempts to address the problem of solving the one-dimensional viscous Burgers equation under low viscosity conditions when singularities appear in the interior layer. Specifically, when the viscosity parameter ε is very small, the solution to the Burgers equation changes rapidly within a certain region, forming an interior layer. Traditional Physics-Informed Neural Network (PINN) methods have significant errors when dealing with such interior layers. Therefore, the authors propose a new method—Singular Layer Physics-Informed Neural Network (sl-PINN) to more accurately predict the solution of the Burgers equation in this scenario. ### Main Contributions of the Paper: 1. **Introduction of Corrective Functions**: Through asymptotic analysis, corrective functions describing the flow behavior within the interior layer were determined. 2. **Construction of the sl-PINN Structure**: Using these corrective functions, the sl-PINN model was constructed to improve the prediction accuracy of the solution near the interior layer. 3. **Numerical Experiment Validation**: Numerical experiments demonstrated that sl-PINN has higher accuracy in handling the Burgers equation under low viscosity conditions compared to traditional PINN methods, especially with significantly reduced errors near the interior layer. ### Main Methods and Techniques: - **Asymptotic Analysis**: Used to determine the corrective functions within the interior layer. - **sl-PINN Structure**: Integrates the corrective functions into the PINN to improve the prediction of the solution near the interior layer. - **Numerical Experiments**: Validated the effectiveness of sl-PINN through numerical experiments under different initial and boundary conditions. ### Experimental Results: - **Smooth Initial Data**: Under viscosity parameters of 10^(-1/π), 10^(-2/π), and 10^(-3/π), sl-PINN showed higher accuracy in both L2 norm and L∞ norm. - **Non-Smooth Initial Data**: In the Riemann problem, sl-PINN also demonstrated better capability in capturing the solution near the interior layer. ### Conclusion: The sl-PINN method proposed in the paper can more accurately predict the solution near the interior layer when dealing with the Burgers equation under low viscosity conditions, significantly reducing the errors of traditional PINN methods in this region. This method provides new insights for understanding and solving the interior layer problem in the Burgers equation.

Singular layer PINN methods for Burgers' equation

Singular layer physics informed neural network method for plane parallel flows

Solving Viscous Burgers' Equation: Hybrid Approach Combining Boundary Layer Theory and Physics-Informed Neural Networks

Semi-analytic PINN methods for boundary layer problems in a rectangular domain

Singular Layer Physics-Informed Neural Network Method for Convection-Dominated Boundary Layer Problems in Two Dimensions

Semi-analytic PINN methods for singularly perturbed boundary value problems

Burgers' pinns with implicit euler transfer learning

Physics-informed neural networks with parameter asymptotic strategy for learning singularly perturbed convection-dominated problem

Theory-guided physics-informed neural networks for boundary layer problems with singular perturbation

Multistep Asymptotic Pre-Training Strategy Based on Pinns for Solving Steep Boundary Singular Perturbation Problems

Investigating the Ability of PINNs To Solve Burgers' PDE Near Finite-Time BlowUp

A Combination of Deep Neural Networks and Physics to Solve the Inverse Problem of Burger's Equation

General-Kindred Physics-Informed Neural Network to the Solutions of Singularly Perturbed Differential Equations

Physics-Informed Neural Networks with Adaptive Localized Artificial Viscosity

Binary structured physics-informed neural networks for solving equations with rapidly changing solutions

Finite Volume Physical Informed Neural Network (FV-PINN) with Reduced Derivative Order for Incompressible Flows

Solving free-surface problems for non-shallow water using boundary and initial conditions-free physics-informed neural network (bif-PINN)

Estimation of the Characteristic Wavelength Parameter in 1D Leray-Burgers Equation with PINN

Multi-Grade Deep Learning for Partial Differential Equations with Applications to the Burgers Equation

A novel discretized physics-informed neural network model applied to the Navier-Stokes equations

Approximate solution of KdV-Burgers equation using improved PINNs algorithm