Heinlein, Alexander, Hochmuth, Christian and Klawonn, Axel ORCID: 0000-0003-4765-7387 (2018). Monolithic Overlapping Schwarz Domain Decomposition Methods with GDSW Coarse Spaces for Saddle Point Problems. Technical Report.

This is the latest version of this item.


Download (2MB) | Preview


Monolithic overlapping Schwarz preconditioners for saddle point problems of Stokes, Navier-Stokes, and mixed linear elasticity ty e are presented. For the first time, coarse spaces obtained from the GDSW (Generalized Dryja-Smith-Widlund) approach are used in such a setting. Numerical results of our parallel implementation are presented for several model problems. In particular, cases are considered where the problem cannot or should not b e reduced using local static condensation, e.g., Stokes, Navier-Stokes or mixed elasticity problems with continuous pressure spaces. In the new monolithic preconditioners, the local overlapping problems and the coarse problem are saddle point problems with the same structure as the original problem. Our parallel implementation of these preconditioners is based on the FROSch (Fast and Robust Overlapping Schwarz) library, which is part of the Trilinos package ShyLU. The implementation is algebraic in the sense that the preconditioners can be constructed from the fully assembled stiffness matrix and information about the block structure of the problem. Parallel scalability results for several thousand cores for Stokes, Navier-Stokes, and mixed linear elasticity model problems are reported. Each of the local problems is solved using a direct solver in serial mo de, whereas the coarse problem is solved using a direct solver in serial or MPI-parallel mode or using an MPI-parallel iterative Krylov solver

Item Type: Preprints, Working Papers or Reports (Technical Report)
Heinlein, Alexanderalexander.heinlein@uni-koeln.deUNSPECIFIED
Hochmuth, Christianchristian.hochmuth@uni-koeln.deUNSPECIFIED
Klawonn, Axelaxel.klawonn@uni-koeln.deorcid.org/0000-0003-4765-7387
URN: urn:nbn:de:hbz:38-83431
Series Name at the University of Cologne: Technical report series. Center for Data and Simulation Science
Volume: 2018,1
Subjects: Data processing Computer science
Technology (Applied sciences)
Uncontrolled Keywords:
monolithic overlapping SchwarzEnglish
domain decompositionEnglish
saddle point problemEnglish
parallel computingEnglish
algebraic preconditionerEnglish
Faculty: Central Institutions / Interdisciplinary Research Centers
Divisions: Central Institutions / Interdisciplinary Research Centers > Center for Data and Simulation Science
Language: English
Date: 2 July 2018
Funders: Deutsche Forschungsgemeinschaft
Projects: Domain-Decomposition-Based Fluid Structure Interaction Algorithms for Highly Nonlinear and Anisotropic Elastic Arterial Wall Models in 3D
Number of Pages: 27
URI: http://kups.ub.uni-koeln.de/id/eprint/8355


Downloads per month over past year


Available Versions of this Item

Actions (login required)

View Item View Item