Research interests

Since 2004, my research activities within the Dolphin project-team (Inria Lille - CRIStAL) were first mainly focused on cooperative parallel optimization for large scale platforms (computational grids). Several issues have been addressed including checkpointing, optimized coding of work units to deal with wide-area communications, and dynamic adaptive load balancing. My research activities have recently evolved towards parallel optimization on heterogeneous environments including multi-core processors and accelerators (GPU).

The major addressed problems are thread divergence, memory optimization, CPU-GPU data transfer optimization, and the definition of new data structures to store and manage efficiently large amounts of generated subproblems. Within the context of the BONUS team, I am interested in big optimization (high-dimensional expensive optimization) using decomposition, surrogate-based Machine Learning and extreme-scale computing. Big clusters including various multi-core processors and GPU accelerators are mainly considered. From the application point of view, I am interested in aerospace design engineering and Tuberculosis and Covid-19 transmission control in epidemiology.