High performance domain decomposition methods on massively parallel architectures with freefem++

P. Jolivet 1 , 2 , V. Dolean 3 , F. Hecht 1 , F. Nataf 1 , C. Prud’Homme 2  and N. Spillane 1
  • 1 Laboratoire Jacques-Louis Lions, CNRS UMR 7598, Universit´e Pierre et Marie Curie, 75005 Paris, France
  • 2 Laboratoire Jean Kuntzmann, CNRS UMR 5224, Universit´e Joseph Fourier, 51 rue des Mat´ematiques, BP53, 38041 Grenoble Cedex 9, France
  • 3 Laboratoire Jean-Alexandre Dieudonn´e, CNRS UMR 6621, Universit´e Nice Sophia Antipolis, Parc Valrose, 06108 Nice Cedex 2, France

Abstract

- In this document, we present a parallel implementation in freefem++ of scalable two-level domain decomposition methods. Numerical studies with highly heterogeneous problems are then performed on large clusters in order to assert the performance of our code.

Purchase article
Get instant unlimited access to the article.
$42.00
Log in
Already have access? Please log in.


Journal + Issues

The Journal of Numerical Mathematics contains high-quality papers featuring contemporary research in all areas of Numerical Mathematics. This includes the development, analysis, and implementation of new and innovative methods in Numerical Linear Algebra, Numerical Analysis, Optimal Control/Optimization, and Scientific Computing.

Search