High–order nodal discontinuous Galerkin methods for the Maxwell eigenvalue problem (Q5313488)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2201053
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | High–order nodal discontinuous Galerkin methods for the Maxwell eigenvalue problem |
scientific article; zbMATH DE number 2201053 |
Statements
High–order nodal discontinuous Galerkin methods for the Maxwell eigenvalue problem (English)
0 references
1 September 2005
0 references
spurious solutions
0 references
curl-conforming elements
0 references
0.8459730744361877
0 references
0.8177120089530945
0 references
0.8170033693313599
0 references
0.8106087446212769
0 references
0.8081861138343811
0 references
The authors discuss the benefits of applying discontinuous Galerkin (DG) method to discretizations of Maxwell's equations in curl-curl form. One of the main difficulties with discrete forms of Maxwell's equations is the presence of a large nullspace of the discrete operator. This nullspace is associated with the divergence condition and can be largely eliminated by use of curl-conforming elements such as Nédélec elements or by use of so-called cross-hatch or criss-cross grids. A third possibility is the focus of this paper: DG methods. This paper presents the results of several computations and some analysis that supports the choice of DG methods as efficient and accurate solution methods for Maxwell's equations in curl-curl form. In the case of the eigenvalue problem in two space dimensions, the authors show that a DG method applied to the curl-curl form, a first-order form, and the Helmholz form of the equations all yield the same eigensolution. In particular, spurious eigensolutions do not arise. Numerical examples are presented that demonstrate that the DG method does not give rise to non-physical eigensolutions. These numerical examples include an L-shaped domain, a square with a cut in it, and a square with variable permittivity. The equivalence among three forms of the equations in two dimensions does not extend to include the Helmholz form in three dimensions. The DG method is consequently less robust in the face of spurious eigensolutions. A three-dimensional numerical example is a cubic cavity. The DG method provides the possibility of an upwind flux that can be formulated to reduce to a symmetric flux as a parameter \(\alpha\) approaches zero. This upwind formulation adds dissipation that eliminates many troublesome eigensolutions. Interestingly, choosing \(\alpha\) larger than one can further improve the solution. The authors also consider a penalized formulation of the equations. For time-dependent problems in two space dimensions, the DG method can be applied directly with good results, as has been shown in earlier work of the authors. In the case of a cubic cavity with an initial pulse, the power spectrum of the solution should reveal the eigenvalues of the cavity. The authors employ both penalization and upwinding to accomplish this task.
0 references