Detailseite
Projekt Druckansicht

Optimal preconditioners of spectral Discontinuous Galerkin methods for elliptic boundary value problems

Fachliche Zuordnung Mathematik
Förderung Förderung von 2012 bis 2015
Projektkennung Deutsche Forschungsgemeinschaft (DFG) - Projektnummer 218348188
 
Erstellungsjahr 2016

Zusammenfassung der Projektergebnisse

The central objective of this project has been to develop and analyze “optimal” preconditioners for “fully flexible” DG discretizations of second order elliptic boundary value problems. By “fully flexible” we mean that arbitrary local mes refinements and variable degrees are permitted under mild grading conditions. “Optimal” then means that the condition numbers remain uniformly bounded. In particular, spectral methods are covered as a special case. Employing a cascaded Auxiliary Space Method as a central strategy, this goal has been achieved for geometrically conforming meshes and arbitrary varying polynomial degrees. The core obstructions that had so far prevented the availability of optimal preconditioners ultimately stem from the non-nestedness of LGL-grids and the anisotropy of corresponding low order finite element auxiliary spaces. Corresponding conceptional remedies have been developed in terms of associated dyadic grid hierarchies and certain anisotropic multi-waveletpreconditoners. The theoretical investigations have been accompanied and complemented by a systematic software development which, on the one hand, provided numerical realizations of the methods and, on the other hand, aimed to gaining deeper insight by monitoring and quantifying the performance of the various algorithmic constituents. An extension to covering as well local refinements with hanging nodes is currently in progress and the main theoretical ingredients are already available.

Projektbezogene Publikationen (Auswahl)

 
 

Zusatzinformationen

Textvergrößerung und Kontrastanpassung