Section: New Software and Platforms

MUMPS

A MUltifrontal Massively Parallel Solver

Keywords: High-Performance Computing - Direct solvers - Finite element modelling

Functional Description

MUMPS is a software library to solve large sparse linear systems (AX=B) on sequential and parallel distributed memory computers. It implements a sparse direct method called the multifrontal method. It is used worldwide in academic and industrial codes, in the context numerical modeling of physical phenomena with finite elements. Its main characteristics are its numerical stability, its large number of features, its high performance and its constant evolution through research and feedback from its community of users. Examples of application fields include structural mechanics, electromagnetism, geophysics, acoustics, computational fluid dynamics. MUMPS is developed by INPT(ENSEEIHT)-IRIT, Inria, CERFACS, University of Bordeaux, CNRS and ENS Lyon. In 2014, a consortium of industrial users has been created (http://mumps-consortium.org).

• Participants: Patrick Amestoy, Alfredo Buttari, Jean-Yves L'Excellent, Chiara Puglisi, Mohamed Sid-Lakhdar, Bora Uçar, Marie Durand, Abdou Guermouche, Maurice Bremond, Guillaume Joslin, Stéphane Pralet, Aurélia Fevre, Clément Weisbecker, Theo Mary, Emmanuel Agullo, Jacko Koster, Tzvetomila Slavova, François-Henry Rouet, Philippe Combes and Gilles Moreau

• Partners: CERFACS - CNRS - ENS Lyon - INPT - IRIT - LIP - Université de Bordeaux - Université de Lyon - Université de Toulouse

• Latest public release: MUMPS 5.0.2 (July 2016)

• Contact: Jean-Yves L'Excellent

• URL: http://mumps-solver.org/

• Next MUMPS User Days: we have started organizing the next MUMPS User days, which will be hosted by Inria on June 1 and 2, 2017 near Grenoble, France (see http://mumps.enseeiht.fr/ud_2017.php)

In the context of the MUMPS consortium (see Section 8.1 and http://mumps-consortium.org), we had scientific exchanges and collaborations with industrial members and released two versions in advance for the consortium (in July 2016 and November 2016), containing major improvements for large-scale problems and many other improvements. Much effort was also put on developing features and algorithms to improve the quality and performance of MUMPS, especially in the context of problems offering potential for low-rank compression. This work is done in close collaboration with the partners who co-develop MUMPS, in particular in Toulouse.