Abstract:The definition of partial differential equation (PDE) models usually involves a set of parameters whose values may vary over a wide range. The solution of even a single set of parameter values may be quite expensive. In many cases, e.g., optimization, control, uncertainty quantification, and other settings, solutions are needed for many sets of parameter values. We consider the case of the time-dependent Navier-Stokes equations for which a recently developed ensemble-based method allows for the efficient determination of the multiple solutions corresponding to many parameter sets. The method uses the average of the multiple solutions at any time step to define a linear set of equations that determines the solutions at the next time step. To significantly further reduce the costs of determining multiple solutions of the Navier-Stokes equations, we incorporate a proper orthogonal decomposition (POD) reduced-order model into the ensemble-based method. The stability and convergence results for the ensemble-based method are extended to the ensemble-POD approach. Numerical experiments are provided that illustrate the accuracy and efficiency of computations determined using the new approach.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: how to efficiently find multiple solutions to the unsteady Navier - Stokes equations (NSE), especially when a large number of parameter sets need to be processed. Specifically, the paper proposes a new method that combines the ensemble method and the proper orthogonal decomposition (POD) reduced - order model to significantly reduce the computational cost.
### Problem Background
In many engineering and geophysical applications, it is often necessary to calculate multiple solutions to the fluid flow equations, such as quantifying uncertainty, performing sensitivity analysis or prediction, etc. However, these calculations are often very expensive, especially in terms of storage and computation time. Although parallel computing and ensemble data assimilation techniques have developed rapidly in recent years, enabling reliable ensemble predictions to be obtained with a small number of realizations in some cases, the existing computational power is still insufficient for high - precision ensemble calculations in large - scale spatial applications (such as numerical weather prediction). In order to reduce the total computation time, the spatial resolution is usually sacrificed.
### Limitations of Existing Methods
Currently, there are some methods that can be used to generate perturbations of the initial conditions, such as the Bred - vector method, the singular vector method and the ensemble transform Kalman filter, etc. However, these methods still face the problem of high computational cost when dealing with large - scale problems.
### Solution Proposed in the Paper
To solve the above problems, the paper proposes a new algorithm based on the ensemble method and the POD reduced - order model. The main innovations of this method include:
1. **Ensemble Method**: By using the average value of all ensemble members at each time step to define the linear equations, the solutions corresponding to multiple parameter sets can be effectively solved.
2. **POD Reduced - Order Model**: The POD method is introduced to further compress the computational scale. By extracting the optimal basis to construct the reduced - order model, the computation time and storage requirements can be significantly reduced.
### Advantages of the Method
- **Computational Efficiency**: The new method can greatly reduce the computational cost while ensuring the accuracy.
- **Stability and Convergence**: The paper proves the stability and convergence of the new method and verifies its effectiveness and efficiency through numerical experiments.
### Mathematical Formula Representation
The mathematical formulas involved in the paper are as follows:
- Navier - Stokes Equations:
\[
\begin{cases}
u_t+u\cdot\nabla u-\nu\Delta u+\nabla p = f(x,t)&\forall x\in\Omega\times(0,T]\\
\nabla\cdot u = 0&\forall x\in\Omega\times(0,T]\\
u = 0&\forall x\in\partial\Omega\times(0,T]\\
u(x,0)=u_0(x)&\forall x\in\Omega
\end{cases}
\]
- POD Basis Construction Problem:
\[
\min_{\{\phi_i\}_{i = 1}^R}\sum_{k = 1}^{JS}\sum_{l = 0}^{NS}\left\|u_{k,l}^{h,s}-\sum_{j = 1}^R(u_{k,l}^{h,s},\phi_j)\phi_j\right\|^2
\]
where \(\phi_i\) is an orthogonal basis, satisfying \((\phi_i,\phi_j)=\delta_{ij}\).
Through these improvements, the paper provides an efficient and accurate method for solving multiple solutions to the unsteady Navier - Stokes equations, which is suitable for application scenarios where a large number of parameter sets need to be processed.