Full gradient stabilized cut finite element methods for surface partial differential equationsShow others and affiliations
2016 (English)In: Computer Methods in Applied Mechanics and Engineering, ISSN 0045-7825, E-ISSN 1879-2138, Vol. 310, p. 278-296Article in journal (Refereed) Published
Resource type
Text
Abstract [en]
We propose and analyze a new stabilized cut finite element method for the Laplace–Beltrami operator on a closed surface. The new stabilization term provides control of the full R 3 gradient on the active mesh consisting of the elements that intersect the surface. Compared to face stabilization, based on controlling the jumps in the normal gradient across faces between elements in the active mesh, the full gradient stabilization is easier to implement and does not significantly increase the number of nonzero elements in the mass and stiffness matrices. The full gradient stabilization term may be combined with a variational formulation of the Laplace–Beltrami operator based on tangential or full gradients and we present a simple and unified analysis that covers both cases. The full gradient stabilization term gives rise to a consistency error which, however, is of optimal order for piecewise linear elements, and we obtain optimal order a priori error estimates in the energy and L 2 norms as well as an optimal bound of the condition number. Finally, we present detailed numerical examples where we in particular study the sensitivity of the condition number and error on the stabilization parameter.
Place, publisher, year, edition, pages
Elsevier, 2016. Vol. 310, p. 278-296
Keywords [en]
Surface PDE, Laplace–Beltrami operator, Cut finite element method, Stabilization, Condition number, A priori error estimates
National Category
Computational Mathematics
Identifiers
URN: urn:nbn:se:hj:diva-31178DOI: 10.1016/j.cma.2016.06.033ISI: 000384859400014Scopus ID: 2-s2.0-84982710231Local ID: JTHProduktutvecklingISOAI: oai:DiVA.org:hj-31178DiVA, id: diva2:951242
2016-08-082016-08-082025-10-13Bibliographically approved