The context of this work falls within the scope of ASTER project that intends to develop and implement methods to improve the simulation of MHD instabilities that are needed to evaluate mechanisms to control the energy losses observed in the standard tokamak operating scenario (ITER).

The applications we study are complex in that they require simulations at a very large scale; several tens or hundreds teraflops of computation using several terabyte of data are often needed. To carry out these more and more accurate full-scale simulations without increasing the number of unknowns in a uniform way, a technique consists in handling a finer grid where the solution varies abruptly and a coarser grid at other places accordingly to quite specific criteria.
This approach is broadly used when explicit numerical schemes are considered. Here, the characteristic time scales may be very different so that an implicit scheme is needed.

The goal of this work is first to make a state of the art of the mesh refinement methods based on numerical implicit schemes in particular for irregular grids, then to study within a parallel framework the principal difficulties encountered in these methods such as: management of the hierarchy of the grids which themselves are distributed, criteria of “refinement/unrefinement”, management of the versatility of the parallel solvers. The effectiveness of such a method will therefore require the use of dynamic load balancing methods over the set of processors to
compensate the variability of the grid during the simulation.

Please send your CV, a cover letter describing your research interests, and the names and email addresses of at least 2 references to:
O. Coulaud and/or P. Ramet, INRIA, Univ. Bordeaux 1, 33405 Talence, FRANCE
e-mail: Olivier.Coulaud@labri.fr, Pierre.Ramet@labri.fr