No abstract

The fast evolution of techniques and performance in scientific computing modifies how one approaches high-technology problems. The models are more complex and can take into account the complete phenomenon in its environment (nonlinear equations, unsteady behaviour, coupling between several physical and chemical aspects, for example). Often, research and development give rise to documented software that is part of the deliverables. Engineering practice also evolves. Parametric studies allow the integration of a large number of parameters, and this reduces significantly the number of prototypes.Since scientific computing is at the heart of the development of high technologies in industry, the problem of assessing the robustness of numerical methods becomes increasingly important today.