Section: Partnerships and Cooperations

International Initiatives

Inria Associate Teams not involved in an Inria International Labs

R-LAS

• Title: Randomized Linear Algebra Software

• International Partner (Institution - Laboratory - Researcher):

  • University of Tennessee, Knoxville (United States) - Innovative Computing Laboratory (ICL) - Jack Dongarra

• Start year: 2014

• See also: https://www.lri.fr/~baboulin/r-las.html

• The objective of the associate team between Inria and University of Tennessee is to develop a class of fast algorithms and software based on randomization to enhance linear algebra calculations in high-performance computing (HPC) applications. The first application will focus on FFT-like randomization techniques to avoid pivoting in dense and sparse matrix factorizations and thus removing the communication cost due to pivoting. The second application is related to the computation of statistical condition estimates for linear algebra problems in order to assess the numerical quality of solutions computed by HPC applications. The targeted architectures are large scale multicore systems with accelerators. The ultimate goal of the project is to make the randomized solvers designed by the associate team accessible to end-users thanks to a public domain software library.