Robust Algebraic Multigrid Methods and their Parallelization

  • Pechstein, Clemens (Researcher)
  • Pohoata, Alin (Researcher)
  • Reitzinger, Stefan (Researcher)
  • Wabro, Markus (Researcher)
  • Yang, Huidong (Researcher)
  • Langer, Ulrich (PI)

Project: Funded researchFWF - Austrian Science Fund

Project Details

Description

The aim of this project is the development and construction of Algebraic MultiGrid (AMG) methods. AMG methods try to mimic geometric multigrid methods just by knowing single grid information but retaining the optimality of the method. Since the construction of an universal AMG method for all arising situations in practice is an unrealistic dream. We will assemble efficient and robust solver for special classes of matrices such as large scale systems arising from the (FEM, FDM, FIT, BEM) discretization of: * Anisotropic Potential Equation * Maxwell's Equations * Stokes / Navier-Stokes Equations Moreover, parallelization of AMG method is an important task, because it is the main source for a further enhancement of efficiency in real life applications.
StatusFinished
Effective start/end date01.10.200130.09.2004

Funding

  • FWF

Fields of science

  • 101020 Technical mathematics
  • 102009 Computer simulation
  • 101 Mathematics
  • 102022 Software development
  • 102023 Supercomputing
  • 101016 Optimisation
  • 101014 Numerical mathematics

JKU Focus areas

  • Digital Transformation