Goal-oriented a posteriori error control for time discretizations involving incompressible flows

Lade...
Vorschaubild

Datum

Zeitschriftentitel

ISSN der Zeitschrift

Bandtitel

Verlag

Sonstige Titel

Zusammenfassung

We use the dual weighted residual (DWR) method to apply goal-oriented error control for time discretizations. We consider the heat equation and the Navier-Stokes equations and use finite difference and Galerkin methods to discretize in time. We propose a uniform approach to take quadrature errors into account. For finite difference methods, we build upon the works of Meidner and Richter [48, 49] and give new proofs for their DWR error representations. Here, θ time-stepping schemes are considered. In contrast to classical DWR error estimates, an additional term is included here, which the authors refer to as Galerkin defect or quadrature error. For the Navier-Stokes equations, we modify their discretization so that the entire test space is θ dependent, which improves the error representation. For continuous and discontinuous Galerkin methods, we modify the error representations known from literature by taking quadrature errors into account. This leads to an additional term in the error representations, which makes the Galerkin method error representations more similar to those derived with the θ method. We demonstrate the importance of the additional terms in the modified error representations by numerical results. We focus on discontinuous Galerkin methods and the usage of the implicit Euler method. The error estimators are implemented in the software FEAT3.

Beschreibung

Inhaltsverzeichnis

Schlagwörter

Schlagwörter nach RSWK

Zitierform

Befürwortung

Review

Ergänzt durch

Referenziert von