Abstract:We propose and analyze a finite element method for the Oseen eigenvalue problem. This problem is an extension of the Stokes eigenvalue problem, where the presence of the convective term leads to a non-symmetric problem and hence, to complex eigenvalues and eigenfunctions. With the aid of the compact operators theory, we prove that for inf-sup stable finite elements the convergence holds and hence, error estimates for the eigenvalues and eigenfunctions are derived. We also propose an a posteriori error estimator which results to be reliable and efficient. We report a series of numerical tests in two and three dimension in order to assess the performance of the method and the proposed estimator.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to propose and analyze a finite - element method for solving the Oseen eigenvalue problem. Specifically:
1. **Asymmetry problem**: The Oseen eigenvalue problem is different from the Stokes eigenvalue problem. Due to the existence of the convection term, the problem is asymmetric. Therefore, eigenvalues and eigenfunctions naturally exist in the complex space, which is significantly different from the Stokes eigenvalue problem.
2. **Numerical approximation**: The author proposes a conforming finite - element method to approximate the eigenvalues and eigenfunctions of the Oseen eigenvalue problem. In particular, they focus on whether the spectrum of the Oseen eigenvalue problem converges to the spectrum of the Stokes eigenvalue problem when the convection vector tends to zero in norm.
3. **Error estimation**: By introducing the compact operator theory, the author proves that for finite elements satisfying the inf - sup condition, the convergence holds and derives the error estimates of eigenvalues and eigenfunctions.
4. **A posteriori error estimator**: The author also proposes a reliable and efficient a posteriori error estimator to evaluate the performance of the numerical method.
5. **Numerical experiments**: To verify the effectiveness of the method, the author conducts a series of 2 - D and 3 - D numerical tests and analyzes the behavior of the a posteriori error estimator.
### Mathematical formula representation
- The strong form of the Oseen equation is:
\[
\begin{cases}
-\nu\Delta u+(\beta\cdot\nabla)u + \nabla p=\lambda u, & \text{in }\Omega,\\
\operatorname{div}u = 0, & \text{in }\Omega,\\
\int_{\Omega}p = 0, & \text{in }\Omega,\\
u = 0, & \text{on }\partial\Omega,
\end{cases}
\]
where \(u\) is the displacement, \(p\) is the pressure, \(\beta\) is a given vector field representing the steady - state flow velocity, and \(\beta\in W^{1,\infty}(\Omega,\mathbb{C})^d\) is divergence - free, and \(\nu> 0\) is the kinematic viscosity.
- The weak form is: Look for \(\lambda\in\mathbb{C}\) and \((u, p)\in X\) such that
\[
\begin{cases}
a(u, v)+b(v, p)=\lambda(u, v), & \forall v\in H_0^1(\Omega,\mathbb{C})^d,\\
b(u, q) = 0, & \forall q\in L_0^2(\Omega,\mathbb{C}),
\end{cases}
\]
where \(a(u, v)=\int_{\Omega}\nu\nabla u:\nabla v+\int_{\Omega}(\beta\cdot\nabla)u\cdot v\), \(b(v, q)=-\int_{\Omega}q\operatorname{div}v\).
Through these works, the author aims to provide an effective numerical solution method for the Oseen eigenvalue problem and ensure its convergence and accuracy.