Abstract:Geophysical flow simulations using hyperbolic shallow water moment equations require an efficient discretization of a potentially large system of PDEs, the so-called moment system. This calls for tailored model order reduction techniques that allow for efficient and accurate simulations while guaranteeing physical properties like mass conservation. In this paper, we develop the first model reduction for the hyperbolic shallow water moment equations and achieve mass conservation. This is accomplished using a macro-micro decomposition of the model into a macroscopic (conservative) part and a microscopic (non-conservative) part with subsequent model reduction using either POD-Galerkin or dynamical low-rank approximation only on the microscopic (non-conservative) part. Numerical experiments showcase the performance of the new model reduction methods including high accuracy and fast computation times together with guaranteed conservation and consistency properties.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to improve computational efficiency while ensuring that physical properties (such as mass conservation) are maintained when solving the hyperbolic shallow - water moment equations (HSWME). Specifically:
1. **Computational efficiency problem**: Due to the introduction of polynomial velocity profiles in HSWME, the number of nonlinear equations increases, resulting in increased computational cost and memory usage. Therefore, an effective method is required to reduce the complexity of these equations in order to improve computational efficiency.
2. **Maintenance of physical properties**: When performing model order reduction, it is necessary to ensure that important physical properties (such as mass conservation) are not violated. Traditional methods may lead to the loss of these properties, thus affecting the accuracy of simulation results.
To solve these problems, the author proposes the following methods:
- **Macro - micro decomposition**: Decompose the model into a macroscopic (conservative) part and a microscopic (non - conservative) part. The macroscopic part includes water height and momentum, and the microscopic part includes higher - order moments of the velocity profile. In this way, the microscopic part can be reduced independently without affecting the physical characteristics of the macroscopic part.
- **POD - Galerkin and dynamic low - rank approximation (DLRA)**: These two methods are used to reduce the order of the microscopic part. The POD - Galerkin method reduces the dimension by projecting the basis functions generated in the offline stage; the DLRA method achieves order reduction by adaptively updating the basis functions in the online stage.
Through the above methods, the paper achieves efficient and accurate simulations and ensures important physical properties such as mass conservation. Numerical experiments show that the new method not only has high accuracy but also significantly reduces the computation time.
### Formula presentation
To better understand the methods in the paper, the following are some key formulas involved:
- **Hyperbolic shallow - water moment equations (HSWME)**:
\[
\partial_t q+A(q)\partial_x q = g_{\nu}(q)
\]
where \(q = [h, hu_m, h\alpha_1,\ldots, h\alpha_N]^T\in\mathbb{R}^{N + 2}\), \(A(q)\in\mathbb{R}^{(N + 2)\times(N + 2)}\) is the transport matrix, and \(g_{\nu}(q)\in\mathbb{R}^{N+2}\) is the friction term.
- **Friction term \(g_{\nu}(q)\)**:
\[
g_{\nu}^i(q)=-\frac{\nu}{\lambda(2i + 1)}\left(u_m+\sum_{j = 1}^N\alpha_j\right)-\frac{\nu}{h}\sum_{j = 1}^N a_{i,j}\alpha_j,\quad i=0,\ldots,N
\]
where \(a_{i,j}\) are constants, defined as:
\[
a_{i,j}=\begin{cases}
0&\text{if }i + j\text{ is even},\\
\frac{\min(i - 1,j)(\min(i - 1,j)+ 1)}{2}&\text{if }i + j\text{ is odd}.
\end{cases}
\]
- **Basis functions in the POD - Galerkin method**:
\[
v(x,t)\approx\hat{v}(x,t)=\sum_{k = 1}^r\hat{\alpha}_k(x,t)w_k = W\hat{v}(x,t)
\]
where \(W = [w_1,\ldots,w_r]\in\mathbb{R}^{N\times r}\) is the basis vector and \(\hat{v}=[\hat{\alpha}_1,\ldots,\hat{\alpha}_r]^T\in\mathbb{R}^r\) is the reduced.