Evolving Neural Network (ENN) Method for One-Dimensional Scalar Hyperbolic Conservation Laws: I Linear and Quadratic Fluxes

Zhiqiang Cai,Brooke Hejnal

2023-12-12

Abstract:We propose and study the evolving neural network (ENN) method for solving one-dimensional scalar hyperbolic conservation laws with linear and quadratic spatial fluxes. The ENN method first represents the initial data and the inflow boundary data by neural networks. Then, it evolves the neural network representation of the initial data along the temporal direction. The evolution is computed using a combination of characteristic and finite volume methods. For the linear spatial flux, the method is not subject to any time step size, and it is shown theoretically that the error at any time step is bounded by the representation errors of the initial and boundary condition. For the quadratic flux, an error estimate is studied in a companion paper. Finally, numerical results for the linear advection equation and the inviscid Burgers equation are presented to show that the ENN method is more accurate and cost efficient than traditional mesh-based methods.

Numerical Analysis

What problem does this paper attempt to address?

### Problems the paper attempts to solve

This paper aims to propose and study an evolving neural network (ENN) - based evolutionary method for solving one - dimensional scalar hyperbolic conservation law problems, especially those with linear and quadratic spatial fluxes. Specifically, the paper attempts to solve the following key problems:

1. **Treatment of discontinuous solutions**:
- Solutions to hyperbolic conservation laws usually contain discontinuities, which may be caused by initial conditions, inflow boundary conditions, or shock formation. Traditional numerical methods have difficulties in dealing with these discontinuities because they require a sufficiently fine grid to capture the positions of discontinuous points.
- The ENN method can handle discontinuous solutions more effectively by using neural networks to represent initial data and inflow boundary data and evolving these representations in the time direction.

2. **Control of time step**:
- For linear spatial fluxes, the ENN method is not limited by the time step. It can be theoretically proven that the error at any time step is determined by the representation errors of the initial and boundary conditions.
- For quadratic spatial fluxes, an accompanying paper is proposed in the paper to study error estimation, and a characteristic finite - volume scheme and a time - step control procedure are introduced to ensure the stability and accuracy of the algorithm.

3. **Improvement of computational efficiency**:
- Traditional grid - based methods require a very fine grid when dealing with discontinuous solutions, which will lead to high computational costs. The ENN method can significantly reduce the computational cost while maintaining high accuracy by avoiding the use of a fine uniform grid.
- The paper shows through numerical experiments that the ENN method is more accurate and computationally more efficient than traditional grid methods when solving the linear convection equation and the inviscid Burgers equation.

### Main contributions

- **Innovative numerical method**: The ENN method is proposed, which combines the advantages of the characteristic method and the finite - volume method and can effectively handle discontinuous solutions.
- **Theoretical analysis**: For linear spatial fluxes, a theoretical analysis of error estimation is given; for quadratic spatial fluxes, a research framework for error estimation is proposed.
- **Numerical verification**: The effectiveness and superiority of the ENN method are verified through numerical experiments, especially its performance in handling discontinuous solutions.

### Summary

By introducing the ENN method, this paper overcomes the shortcomings of traditional numerical methods in handling discontinuous solutions of hyperbolic conservation laws and provides a new and efficient numerical solution method.