A Massively Parallel Multigrid Method for Finite Elements

Hülsemann F, Bergen B, Gradl T, Rüde U (2006)


Publication Type: Journal article

Publication year: 2006

Journal

Publisher: Institute of Electrical and Electronics Engineers (IEEE)

Book Volume: 8

Pages Range: 56-62

Journal Issue: 6

URI: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1717316

DOI: 10.1109/MCSE.2006.102

Abstract

The hierarchical hybrid grid (HHG) framework supports the parallel implementation of multigrid solvers for finite element problems. The HHG framework attempts to combine some of the flexibility of unstructured grid representations with the performance and efficiency of structured grid data structures. HHG uses a combination of regular refinement and grid decomposition that lets the solver treat structured regions of the refined grid hierarchy with stencil-based data structures. These data structures allow for the efficient implementation of standard multigrid component algorithms, in terms of both performance and memory usage. The grid is distributed among the available message-passing interface (MPI) processes to parallelize the computation. Regular refinement, applied in HHG framework, results in a nested grid hierarchy, which makes geometric multigrid methods easier to adapt to this approach.

Authors with CRIS profile

Related research project(s)

How to cite

APA:

Hülsemann, F., Bergen, B., Gradl, T., & Rüde, U. (2006). A Massively Parallel Multigrid Method for Finite Elements. Computing in Science & Engineering, 8(6), 56-62. https://dx.doi.org/10.1109/MCSE.2006.102

MLA:

Hülsemann, Frank, et al. "A Massively Parallel Multigrid Method for Finite Elements." Computing in Science & Engineering 8.6 (2006): 56-62.

BibTeX: Download