Department of Scientific Computing : projects, actions and teams
CADNA means ''Control of Accuracy and Debugging for Numerical Applications''. This project focuses on the development of the CADNA library and the SOFA toolbox (Stochastic Optimization for Fixed-point Arithmetic) to estimate and study the accuracy of numerical results obtained with a finite floating-point or fixed-point arithmetic. This project includes the study of the numerical quality of scientific codes like, for instance, the Computer Physics Communications Program Library or chaotic systems.
Remodelage means ''REconstruction 3D+t par fusion de données hétérogènes et multiMODales pour L'Assimilation dans des modèles bioloGiques''. The goal is to extract from sequences as much information as we can to incorporate them into physical models. Giving heterogeneous and multimodal data, not available (or missing) at the same time, we want to reconstruct the whole complete 3D sequence, using, in particular, non linear filtering techniques.
The ''SYmbolic and NUmerical Solving'' action involves the two teams (PEQUAN and SPIRAL) of the Scientific Computing department at LIP6. They conduct a joint research project involving both symbolic and numerical algorithms. This project aims to combine and develop the two types of algorithms for scientific computing to improve the efficiency of symbolic algorithms by performing numerical computatoins when finite precision is appropriate.
Heterogeneous and distributed grids provide a large computational power but bring new algorithmic problems: communication times are significant and the performance of the processors differs. Consequently, the performances of numerical algorithms designed to be run on parallel homogeneous computers are not satisfactory on such grids. The aim of the ''Numerical Applications on GRID'' project is to develop new parallel numerical algorithms having to be coarse grained and asynchronous.
SALSA means ''Software for ALgebraic Systems and Applications''. This research project is a joint project-team with INRIA (Paris, Rocquencourt unit). It aims to designing and implementing certified algorithms for solving polynomial systems. Among the fields of applications in which some results have already been obtained, we can point out simulation control and diagnostic of parallel manipulatorsn celestial mechanics, cryptography.
Random Generation Action
In the field of random generation of structures, Boltzmann model, which comes from analytic combinatorics, shows to be particularly efficient for generating objects obeying complex constraints. The goal of the action ''Random Generation'' is to develop Boltzmann methods for the automatic generation of tests for intensive and large scale applications, particularly in the context of testing and checking software.