DS0708 - Données massives et calcul intensif : enjeux et synergies pour la simulation numérique

Parallelisation methods for complex cinetics – CINE-PARA

Submission summary

The need for faster numerical simulations of complex phenomena, and the definition in this context of what a complex phenomenon is, is evolving in line with the improvement of the platforms that are available for High Performance Computing. Indeed, what used to require hours or days of numerical simulations on large computers can now be run in fractions of seconds on laptops. Nevertheless the understanding of real phenomena, the control and optimization of processes and the monitoring of industrial problems propose new challenges where i) better accuracy, ii) use of more involved mathematical models, iii) simulations on bigger object or iv) on longer period of time for unsteady phenomena are required. The evolution of the computing platforms helps in addressing bigger problems but is not sufficient. The forthcoming large scale parallel computing architectures, indeed, have a number of processors (up to million of computer cores) for which the standard divide and conquer approaches for parallelism, i.e. the domain decomposition methods or task partitioning approaches reach their limits in their ability to use the entire computational resource with the same efficiency as currently achieved on existing smaller systems. It is thus crucial to develop new, complementary and robust algorithms that allow for an improved parallel efficiency and scalability. For time dependent problems, either pure differential systems or coupled with partial differential equations, the time direction leads to new families of algorithms that might allow to provide full efficiencies and speed ups. The parareal (parallel in time) algorithm (introduced in 2000 by Lions, Maday and Turinici with further advances and modifications proposed by Bal and Maday and Baffico, Bernard, Maday, Turinici and Zerah both published in 2002) and the waveform relaxation methods have been introduced to fill this gap and the partners of this proposal count among the leading experts (or even the initiators) in this novel direction of research. Many applications have already been handled proving all the potential of these approaches, both for simple evolution problems including long time simulations for Hamiltonian systems and for evolution PDE's with combination with other approaches (domain decomposition, control iteration..).

Based on this leadership at the root of this consortium together with the adjunction of pertinent partners (scientific computing, large scale problem with target societal interest), this proposed project aims to expand research in this area and more precisely bring these methods for parallelization in time to larger classes of parallel architectures, increase the theoretical understanding, expand the domains of application including industrial size problem, and increase their robustness with respect to memory access requirements , latency and failure.

Project coordination

Yvon MADAY (Laboratoire Jacques Louis Lions)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

ENPC/NAVIER Laboratoire Navier
Inria - Centre de Paris Institut National de Recherche en Informatique et Automatique
UMR 7534 Centre de Recherche en Mathématiques de la Décision Université Paris-Dauphine
DM2S CEA-Saclay Département de Modélisation des Systèmes et Structures, CEA Saclay
LJLL Laboratoire Jacques Louis Lions
LAGA-Université Paris 13 Laboratoire Analyse, Géométrie et Applications

Help of the ANR 572,208 euros
Beginning and duration of the scientific project: September 2015 - 48 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter