PELLEGRINI, François

doi:10.1007/978-3-540-74466-5_22

Métadonnées

Licence d’utilisation du document

PELLEGRINI, François
Laboratoire Bordelais de Recherche en Informatique [LaBRI]
Algorithms and high performance computing for grand challenge applications [SCALAPPLIX]

Langue

Communication dans un congrès

Ce document a été publié dans

Euro-Par 2007 Parallel Processing, Euro-Par 2007 Parallel Processing, EuroPar, 2007-08-28, Rennes. 2007-08, vol. 4641, p. 195-204

Springer

Résumé en anglais

Graph partitioning algorithms have yet to be improved, because graph-based local optimization algorithms do not compute smooth and globally-optimal frontiers, while global optimization algorithms are too expensive to be of practical use on large graphs. This paper presents a way to integrate a global optimization, diffusion algorithm in a banded multi-level framework, which dramatically reduces problem size while yielding balanced partitions with smooth boundaries. Since all of these algorithms do parallelize well, high-quality parallel graph partitioners built using these algorithms will have the same quality as state-of-the-art sequential partitioners< Réduire

Mots clés en anglais

graph partitioning

band

diffusion

aspect ratio

jug

danaides

URI

https://oskar-bordeaux.fr/handle/20.500.12278/198756

DOI

http://dx.doi.org/10.1007/978-3-540-74466-5_22

Project ANR

SOLveurs et SimulaTIons en Calculs Extrême - ANR-06-CIS6-0010

Origine

Importé de hal

Unités de recherche

Laboratoire Bordelais de Recherche en Informatique (LaBRI) - UMR 5800