Numer. Math. Theor. Meth. Appl., 7 (2014), pp. 288-316.
Published online: 2014-07
Cited by
- BibTex
- RIS
- TXT
We present quantitative studies of transfer operators between finite element spaces associated with unrelated meshes. Several local approximations of the global $L^2$-orthogonal projection are reviewed and evaluated computationally. The numerical studies in 3D provide the first estimates of the quantitative differences between a range of transfer operators between non-nested finite element spaces. We consider the standard finite element interpolation, Clément's quasi-interpolation with different local polynomial degrees the global $L^2$-orthogonal projection, a local $L^2$-quasi-projection via a discrete inner product, and a pseudo-$L^2$-projection defined by a Petrov-Galerkin variational equation with a discontinuous test space. Understanding their qualitative and quantitative behaviors in this computational way is interesting per se; it could also be relevant in the context of discretization and solution techniques which make use of different non-nested meshes. It turns out that the pseudo-$L^2$-projection approximates the actual $L^2$-orthogonal projection best. The obtained results seem to be largely independent of the underlying computational domain; this is demonstrated by four examples (ball, cylinder, half torus and Stanford Bunny).
}, issn = {2079-7338}, doi = {https://doi.org/10.4208/nmtma.2014.1218nm}, url = {http://global-sci.org/intro/article_detail/nmtma/5876.html} }We present quantitative studies of transfer operators between finite element spaces associated with unrelated meshes. Several local approximations of the global $L^2$-orthogonal projection are reviewed and evaluated computationally. The numerical studies in 3D provide the first estimates of the quantitative differences between a range of transfer operators between non-nested finite element spaces. We consider the standard finite element interpolation, Clément's quasi-interpolation with different local polynomial degrees the global $L^2$-orthogonal projection, a local $L^2$-quasi-projection via a discrete inner product, and a pseudo-$L^2$-projection defined by a Petrov-Galerkin variational equation with a discontinuous test space. Understanding their qualitative and quantitative behaviors in this computational way is interesting per se; it could also be relevant in the context of discretization and solution techniques which make use of different non-nested meshes. It turns out that the pseudo-$L^2$-projection approximates the actual $L^2$-orthogonal projection best. The obtained results seem to be largely independent of the underlying computational domain; this is demonstrated by four examples (ball, cylinder, half torus and Stanford Bunny).