Fully Reliable Localized Error Control in the FEM (Q4509783)

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 1519777
Language Label Description Also known as
default for all languages
No label defined
    English
    Fully Reliable Localized Error Control in the FEM
    scientific article; zbMATH DE number 1519777

      Statements

      Fully Reliable Localized Error Control in the FEM (English)
      0 references
      0 references
      0 references
      19 October 2000
      0 references
      a posteriori error estimates
      0 references
      error control
      0 references
      computable error bounds
      0 references
      adaptive algorithms
      0 references
      reliability
      0 references
      numerical examples
      0 references
      If the first task in numerical analysis is the calculation of an approximate solution, the second is to provide a guaranteed error bound and is often of equal importance. The standard approaches in the a posteriori error analysis of finite element methods suppose that the exact solution has a certain regularity or the numerical scheme enjoys some saturation property. For coarse meshes those asymptotic arguments are difficult to recast into rigorous error bounds. The aim of this paper is to provide reliable computable error bounds which are efficient and complete in the sense that constants are estimated as well. The main argument is a localization via a partition of unity which leads to problems on small domains. Two fully reliable estimates are established: The sharper one solves an analytical interface problem with residuals following \textit{I. Babuška} and \textit{W. C. Rheinboldt} [SIAM J. Numer. Anal. 15, 736-754 (1978; Zbl 0398.65069)]. The second estimate is a modification of the standard residual-based a posteriori estimate with explicit constants from local analytical eigenvalue problems. For some class of triangulations the authors show that the efficiency constant is smaller than 2.5. According to the authors' numerical experience, the overestimation of their computable estimates proved to be reasonably small, with an overestimation by a factor between 2.5 and 4 only.
      0 references

      Identifiers