Most of my works on the sparse linear algebra field concerns algorithms and methods that are implemented in these two softwares :          

HIPSHIPS (Hierarchical Iterative Parallel Solver) is a C-library for solving large sparse linear systems on parallel platforms. The code provides a hybrid method which blends direct and iterative solvers. HIPS exploits the partitioning and multistage ILU techniques developed in the PHIDAL algorithm [see,  Tech. reports] to enable a highly parallel scheme whereby several subdomains can be assigned a process.

Joint work with  :  Jérémie Gaidamour,  Yousef Saad  (University of Minneapolis)

PaStiXPaStiX is  a Parallel Direct Solver for Sparse linear system. PaStiX can also be used as an iterative solver with a preconditioner based on block ILU(k) factorizations. PaStiX use a MPI/thread implementation that is usefull on multicore and SMP architectures (reduce the great overhead of communication in direct solver).

Joint work with : Mathieu Faverge, Xavier Lacoste,  Pierre RametJean Roman

Other people have contributed to PaStiX : David Goudin, Dimitri Lecas,  François Pellegrini , and many internship students that we warmly thank for their great work.