Robust methods and algorithms for solving large algebraic systems on modern high performance computing systems

Presentation of the school

Objectives

Benefiting from the ever increasing performances of computing facilities, either from the point of view of the characteristics of microprocessors and interconnection networks, or for what concern storage capacities, high performance scientific computing finds nowadays numerous industrial applications and societal applications as well. In particular, parallel computing has become a mandatory path for tackling the complexity of the problems relevant to the scientific research and engineering design communities. A lot of scientific computing applications call for the numerical solution of systems of partial differential equations (PDE) modelling complex steady or unsteady physical phenomena. The discretization of these PDE systems most often leads to the resolution of large algebraic systems of equations or to the extraction of the spectral properties of large matricial operators. In this context, this school aims at presenting recent developments concerning methods and algorithms for solving large algebraic systems on modern high performance computing systems.

Audience

This school will be addressed to students, engineers and researchers interested in high performance scientific computing, in particular for large-sacle numerical simulations in computational physics.

Keywords

Scientific computing, high performance computing, parallel computing, numerical linear algebra, numerical simulation.

Speakers