Fabienne JEZEQUEL
Research interests
Fabienne JEZEQUEL
        

Research axes:
     high performance computing, numerical validation of scientific applications

Research activities:

  • Numerical validation of HPC applications
    estimation of rounding errors in parallel codes running on shared or distributed memory architectures, hardware accelerators,...

  • Precision auto-tuning
    automatic modification of numerical types taking into account accuracy requirements on results

  • Performance optimization of numerical algorithms

  • Dynamical control of numerical algorithms
    optimization of the number of iterations of the step size during the execution

  • Development of libraries for scientific applications:
    • CADNA (Control of Accuracy and Debugging for Numerical Applications)
    • PROMISE (PRecision OptiMISE)
    • SAFE (Stochastic Arithmetic with Flexible Exponent)
    • SAM (Stochastic Arithmetic in Multiprecision)
Home page Degrees Publications Talks Teaching