Numerical approximation of the spectrum of the curl operator (Q2871176)
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: Numerical approximation of the spectrum of the curl operator |
scientific article; zbMATH DE number 6248919
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | Numerical approximation of the spectrum of the curl operator |
scientific article; zbMATH DE number 6248919 |
Statements
Numerical approximation of the spectrum of the curl operator (English)
0 references
22 January 2014
0 references
eigenvalue problems
0 references
curl operator
0 references
solenoidal eigensolutions
0 references
simply-connected domains
0 references
finite elements
0 references
spectral convergence
0 references
optimal-order error estimates
0 references
numerical experiments
0 references
0 references
For bounded, simply-connected domains, an eigenvalue problem for the curl operator with divergence-free eigensolutions is investigated. Two different weak formulations are presented and discretized by finite elements.NEWLINENEWLINEThe first one directly delivers the eigenvalues and eigensolutions of the original problem, but its discretization leads to a degenerate generalized eigenvalue problem with both matrices being non-definite, so that an application of standard solvers is not possible.NEWLINENEWLINEThe second formulation leads to an eigenvalue problem, whose eigenvalues are the square of the eigenvalues of the curl operator and the corresponding eigenspaces are invariant subspaces of the original problem. The finite element discretization results in a generalized eigenvalue problem with Hermitian matrices, the one on the right hand side being positive definite. For this approach, spectral convergence, optimal-order error estimates and absence of spurious modes are established.NEWLINENEWLINEIn the case of eigenspaces of dimension larger than 1, the eigensolutions of the curl operator cannot be reconstructed from the eigensolutions of the second formulation. In this case, one can resort to the first variational method.NEWLINENEWLINEFinally, some numerical experiments confirm the theoretical results.
0 references