Adaptive and frugal BDDC coarse spaces for virtual element discretizations of a Stokes problem with heterogeneous viscosity

Tommaso Bevilacqua,Axel Klawonn,Martin Lanser

2024-09-05

Abstract:The virtual element method (VEM) is a family of numerical methods to discretize partial differential equations on general polygonal or polyhedral computational grids. However, the resulting linear systems are often ill-conditioned and robust preconditioning techniques are necessary for an iterative solution. Here, a balancing domain decomposition by constraints (BDDC) preconditioner is considered. Techniques to enrich the coarse space of BDDC applied to a Stokes problem with heterogeneous viscosity are proposed. In this framework a comparison between two adaptive techniques and a computationally cheaper heuristic approach is carried out. Numerical results computed on a physically realistic model show that the latter approach in combination with the deluxe scaling is a promising alternative.

Numerical Analysis

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: when using the Virtual Element Method (VEM) to discretize the Stokes problem with heterogeneous viscosities, how to effectively construct robust and efficient pre - conditioners. Specifically, the authors focus on how to improve the efficiency and stability of solving linear systems by enhancing the coarse space of the BDDC (Balancing Domain Decomposition by Constraints) method.

### Problem Background

1. **Virtual Element Method (VEM)** : This is a numerical method for discretizing partial differential equations (PDEs) and can be carried out on very general polygonal or polyhedral computational meshes. However, compared with the standard Finite Element Method (FEM), the linear systems generated by VEM are usually in worse condition, so robust pre - conditioning techniques are required for iterative solution.

2. **Stokes Problem** : This is a mathematical model for describing fluid motion, which involves the solution of the velocity field \( \mathbf{u} \) and the pressure field \( p \). When the viscosity function \( \nu(x) \) has significant differences in different regions, the problem becomes more complicated.

### Main Contributions of the Paper

1. **Proposing New Coarse - Space Enhancement Techniques** : The authors propose two adaptive techniques and a computationally less - costly heuristic method (frugal approach) for enhancing the coarse space of the BDDC method. These techniques aim to capture the main features on the boundaries of each sub - domain, thereby improving the performance of the solver.

2. **Comparing the Effects of Different Methods** : Through numerical experiments with physically reasonable models, the authors compare the performance of the above three methods. The results show that the heuristic method combined with deluxe scaling is a promising alternative, which can provide good convergence while maintaining computational efficiency.

### Mathematical Model

The continuous form of the Stokes problem can be expressed as:

\[
\begin{cases}
- \nabla \cdot (\nu(x) \nabla \mathbf{u}) + \nabla p = \mathbf{f} & \text{in } \Omega, \\
\nabla \cdot \mathbf{u} = 0 & \text{in } \Omega, \\
\mathbf{u} = 0 & \text{on } \Gamma.
\end{cases}
\]

where:
- \( \mathbf{u} \) is the velocity field,
- \( p \) is the pressure field,
- \( \mathbf{f} \) is the external force term,
- \( \nu(x) \) is the heterogeneous viscosity function.

### Discretization and Solution

The Virtual Element Method is used to discretize the above problem, and the BDDC pre - conditioner is used to accelerate the solution of the discretized linear system. The paper pays special attention to how to enhance the coarse space through adaptive and heuristic methods to improve the solution efficiency and stability.

### Conclusion

By introducing new coarse - space enhancement techniques, especially the heuristic method combined with deluxe scaling, the authors show that when dealing with the Stokes problem with heterogeneous viscosities, the performance and robustness of the solver can be significantly improved.