Error estimates for a vectorial second-order elliptic eigenproblem by the local \(L^2\) projected \(C^0\) finite element method (Q2845608)
From MaRDI portal
| This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: Error estimates for a vectorial second-order elliptic eigenproblem by the local \(L^2\) projected \(C^0\) finite element method |
scientific article; zbMATH DE number 6203693
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Error estimates for a vectorial second-order elliptic eigenproblem by the local \(L^2\) projected \(C^0\) finite element method |
scientific article; zbMATH DE number 6203693 |
Statements
2 September 2013
0 references
vectorial second-order elliptic eigenproblem
0 references
singular solution
0 references
finite-element method
0 references
Fortin-type interpolation
0 references
spectral correct approximation
0 references
error estimates
0 references
eigenfunctions
0 references
convergence
0 references
Error estimates for a vectorial second-order elliptic eigenproblem by the local \(L^2\) projected \(C^0\) finite element method (English)
0 references
The paper deals with the \(C^0\) finite element method (FEM) for solving the following vectorial second-order elliptic eigenproblem: Find \((\omega^2, u)\) such that NEWLINE\[NEWLINE\alpha \nabla \times \nabla \times u-\beta \nabla \nabla \cdot u=\omega^2 u \quad \text{in} \quad \Omega \in \mathbb{R}^d,~ d=2,3,NEWLINE\]NEWLINE NEWLINE\[NEWLINEu \times {\mathbf n}=0, \quad \nabla \cdot u=0 \quad \text{on} \quad \Gamma,NEWLINE\]NEWLINE where \(\omega^2 \in \mathbb{R}\), \(u\neq0\), is the unknown eigenpair, \(\alpha, \beta\) are given positive constants and \(\nabla \) is the gradient operator. It is known that e.g. for nonsmooth domains \(\Omega\) with reentrant corners and edges some eigenfunctions can be singular and belong to \((H^r(\Omega))^d\) for some real \(r<1\) only which causes the wrong convergence of the FEM to an element outside \((H^r(\Omega))^d\). It is shown that the method proposed guarantees the convergence order \({\mathcal O}(h^{2r})\) for the eigenvalues and \({\mathcal O}(h^r)\) for the singular eigenfunctions. The crutial tools for the theoretical framework including the optimal error bounds for singular solutions in \(H^r\), \(r>1/2\), are Fortin-type interpolations and an inf-sup inequality.
0 references