Nowadays, a growing number of industrial applications need more and more intensive computations, with increasingly sophisticated numerical methods and algorithms. During their career, the engineers have to use techniques for which they may not have been trained. Similarly, the freshly graduated engineers must master the latest improvements in order to fulfill the needs. Last, the academic researchers in the University or the specialized laboratories have to be aware of the latest needs in order to develop new solutions and/or participate to their diffusion. Since 1996, one of the ways that the scientific community can use to reach this goal is the Centre d' été Mathématique de Recherche avancées en Calcul Scientifique (CEMRACS). Each year, the topic of the school changes, but the method remains the same. It is a summer school of scientific computing. The training is both theoretical and applied. This action is supported by the French Department of Education, CNRS, European funding, and is hosted by the CIRM (Centre International de Recherches Mathématiques), a privileged resort in the south of France. Last, the CEMRACS is also sponsored by the SMAI (Société de Mathématiques Appliquées et Industrielles), as one of its recurrent activities. This year, we wish a more important participation of European fellows. This summer school is different from more classical ones organized in the academic world. First, it is splitted into two periods of time. First, the participants attend a series of lectures at ASCI (Laboratoire de calcul scientifique intensif du CNRS) in Orsay near Paris, for about two weeks. The goal of these lectures, delivered by well known scientists, is to present a state of the art on the topic of interest. Then the students move to CIRM in Marseille, and start a research period on problems defined in cooperation with industry. For the students, it is an opportunity to work on really applied problems, and to develop useful contacts. What are the objectives? In some situations, the equations of a physical problem are well known, but impossible to solve analytically. What is possible is to approximate the solution via a numerical scheme, but this may not be easy. Assuming that this is possible, it is frequently difficult to analyze the mathematical properties of the scheme, and the implementation of the method on a computer is mandatory in order to guess its properties. Most of the problems arising in scientific computing are at the interface between several fields : applied mathematics, mechanics, chemistry, etc... Before the numerical solution, it is necessary to understand well the physics of the problem of interest. In many cases, this understanding will be a guideline to numerics. Most of the time, modeling is carried out by people with a good background in physics, but not necessarily in numerics. But the numerics are complex by themselves. The ambition of CEMRACS is to provide a training both theoretical and practical, and to give scientists the opportunity to meet in order to study recent problems with modern computer facilities. Seminars are regularly organized, and visitors are invited for shorter periods.

This year, the topics are oriented toward multi-scale problems but other subjects will be considered as well.   Multi-scale problems in time or space are an important challenge in modeling and scientific computing.  Some examples of applications are :  porous media (geophysics, petroleum exploration, under-surface pollution, nuclear and chemical waste disposal, civil engineering)  modeling of materials (metallurgy, concrete, glass, damage modeling) composites :  non linear media  random micro-structures  numerical methods for homogenization propagation in random media fluids with micro-structures (polymers, physiological fluids) meteorology finance Numerical simulations may be either direct, or based on effective behaviors obtained by modeling. For direct simulation of multi-scale problems, the design of efficient and robust algorithms (with a linear complexity) is an important challenge.  Also, intermediate numerical methods which take advantage of the multi-scale structure of the problem are attractive.

Sponsors   - AMIF(ESF)   - ANDRA   - ASCI   - CEA   - CNRS   - EDF   - IFP   - INRIA   - Michelin   - Ministère     de la Recherche   - Péchiney   - TMR Wavelet

News Deadline for registration for the courses : 11th of July 2001 Deadline for registration for the research projects : 1st of June 2001

French Version