Implicit Active Flux methods for linear advection

Wasilij Barsukow,Raul Borsche

2023-12-12

Abstract:In this work we develop implicit Active Flux schemes for the scalar advection equation. At every cell interface we approximate the solution by a polynomial in time. This allows to evolve the point values using characteristics and to update the cell averages using fluxes obtained by integrating this polynomial. The resulting schemes have order of convergence up to five, but show only moderate oscillations with high frequencies for discontinuous solutions. In numerical experiments we compare the different methods and show an application to network flows.

Numerical Analysis

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to develop an implicit Active Flux method applicable to the linear advection equation. Specifically, the author aims to improve the existing numerical methods in the following ways:

1. **Enhance the flexibility of the time step**: Traditional explicit methods need to follow the CFL condition, which restricts the choice of the time step and is particularly inefficient when dealing with network problems with different scales. Implicit methods can avoid this time - step limitation, thereby improving computational efficiency and accuracy.

2. **Reduce the oscillation of high - frequency discontinuous solutions**: Some of the existing high - order methods will produce large oscillations when dealing with discontinuous solutions. The author hopes that the new implicit Active Flux method can reduce these oscillations and thus obtain more stable numerical solutions.

3. **Apply to network flow problems**: The paper also explores the application of these new methods in network flow problems, such as practical industrial problems in supply chain models or district heating systems. Such problems usually involve complex network structures and require efficient numerical methods to handle grid cells of different scales.

### Overview of the main content

To achieve the above goals, the author proposes two strategies to construct the implicit Active Flux method:

- **Semi - discrete method**: Based on the existing finite - difference or finite - volume methods, update the point values and cell averages by introducing implicit integration in time.

- **Single - stage method**: Update the point values by performing time interpolation at the cell interface, combined with the characteristics method, and use polynomial interpolation to mix the values of the current and the next time layer.

The main difference between these two methods lies in how to handle the discretization of time and space and how to ensure the stability and accuracy of the numerical method. Through detailed numerical experiments, the author compares the performance of different methods and analyzes their diffusion and dispersion characteristics to evaluate their performance in practical applications.

### Key formulas

For the linear advection equation:
\[
\frac{\partial q}{\partial t}+u\frac{\partial q}{\partial x} = 0
\]
where \(u\) is a fixed positive velocity. The update formula for the cell average is:
\[
\frac{d}{dt}\bar{q}_i(t)+\frac{f(q_{i + 1/2}(t))-f(q_{i-1/2}(t))}{\Delta x}=0
\]
The update formula for the point value is:
\[
\frac{d}{dt}q_{i + 1/2}(t)=-u\frac{2q_{i-1/2}(t)-6\bar{q}_i(t)+4q_{i + 1/2}(t)}{\Delta x}
\]

Through these formulas, the author successfully constructs multiple implicit Active Flux methods and verifies their stability and accuracy under different conditions.