[France] Postdoc position Scientific Grid Computing at INRIA Sophia Antipoli
Postdoctoral project at INRIA Sophia Antipolis Hierarchical mesh partitioning for parallel scientific computing on a Grid
Grid computing is currently the subject of a lot of research activities world-wide. Most of these activities aim at providing the necessary tools for dealing with aspects such as administration, security, performance simulation, discovery, scheduling and volatility of resources, etc. Few studies are centred on applications. Computational grids are particularly well suited to compute intensive, embarrassingly parallel applications.
The situation is less clear for non- embarrassingly parallel scientific computing applications such as those involving the numerical resolution of system of Partial Differential Equations using mesh based (finite difference, finite volume or finite element) methods. The standard parallelization strategy for mesh based numerical solvers for the solution of systems of PDEs combines a partitioning of the computational mesh with a message passing programming model. In most cases, grid-enabled numerical simulation tools are essentially a direct porting of parallel software previously developed for clusters of PCs or parallel supercomputers, thanks to the availability of appropriate MPI implementations such as MPICH-G2. Moreover, these grid-enabled simulation software rarely take into account important architectural issues characterizing computational grids, such as heterogeneity both of processing nodes and interconnection networks, that greatly impact the parallel performances.
The DiscoGrid project aims at studying and promoting a hierarchical paradigm for programming non-embarrassingly parallel scientific computing applications on distributed, heterogeneous, computing platforms. The target applications require the numerical resolution of systems of partial differential PDEs modeling electromagnetic wave propagation (CEM) and fluid flow (CFD) problems. More importantly, the underlying numerical methods share the use of unstructured meshes and are based on well known finite element and finite volume formulations.
The general objective of this postdoctoral project will be to design and experiment mesh partitioning strategies that take into account: (a) the heterogeneity of processing nodes and interconnection networks of a grid computing platform and, (b) the hierarchical parallel programming paradigm considered in the DiscoGrid project. The proposed work will also include the adaptation of existing parallel simulation software in order to deal with the hierarchical programming model.
For more details including informations on how to apply to this post-doctoral project please refer to the announcement.
Filed under: Postdoctoral
