Factorization-Based Graph Repartitionings

0348801 - ÚI 2011 RIV DE eng C - Conference Paper (international conference)
Jurková, K. - Tůma, Miroslav
Factorization-Based Graph Repartitionings.
Large Scale Scientific Computing. Berlin: Springer, 2010 - (Lirkov, I.; Margenov, S.; Wasniewski, J.), s. 771-778. Lecture Notes in Computer Science, 5910. ISBN 978-3-642-12534-8. ISSN 0302-9743.
[LSSC 2009. International Conference on Large-Scale Scientific Computations /7./. Sozopol (BG), 04.06.2009-08.06.2009]
R&D Projects: GA AV ČR IAA100300802
Institutional research plan: CEZ:AV0Z10300504
Keywords : graph partitioning * load balancing * matrix factorization * parallel computation
Subject RIV: BA - General Mathematics

The paper deals with the parallel computation of matrix factorization using graph partitioning-based domain decomposition. It is well-known that the partitioned graph may have both a small separator and well-balanced domains but sparse matrix decompositions on domains can be completely unbalanced. In this paper we propose to enhance the iterative strategy for balancing the decompositions by graph-theoretical tools. We propose the whole framework for the graph repartitioning. In particular, new global and local reordering strategies for domains are discussed in more detail. We present both theoretical results for structured grids and experimental results for unstructured large-scale problems.
Permanent Link: http://hdl.handle.net/11104/0189228