Required Education : Master2
Start date : 1 February 2020
Mission duration : 6 months
Deadline for applications : 20 December 2019
Salary : 650 Euros/month
The European Center for Advanced Research and Training in Scientific Computing (CERFACS) is working on the resolution, through modeling and numerical simulation, of scientific problems requiring the use of the most powerful means of calculation. It combines interdisciplinary research, as well as advanced training, with physicists, applied mathematicians and engineers. This job will take place in the CFD team which is working on the development and improvement of numerical methods for numerical simulations (Aerodynamics and Combustion essentially) on parallel computers.
Since the beginning of the 90s, several probabilistic approaches have been possible to treat the uncertainties in the stochastic angle. Thus, methods based on response surfaces, such as mathematical approximation representations, are developed at CERFACS, in particular within the framework of the BATMAN project, to answer the problem of UQ (uncertainties quantification). These techniques are also known under the term Meta-Models (Surrogate Model in English) when it comes to studying a global response surface. They rely on constructions of functional representations of random variables and seem promising in the multidisciplinary framework. Indeed, the global response surface is a much more flexible “black-box” evaluator than numerical simulation codes.
The subject of the internship will be the analysis of the sensitivity of different numerical simulations of flows to the characteristics of their meshes. First of all, it will be necessary to determine the parameters of the meshes that are decisive for the prediction of the variables of interest, then to carry out a study of the propagation of the uncertainties using the BATMAN Open Source software. Mesh construction will take place with CENTAUR-Soft, while the flow simulation will be performed by the AVBP combustion code.
– Ability to work in a team.
– Good knowledge of the UNIX system.
– Essential knowledge of Python programming
– Knowledge of applied mathematics
– Thematic knowledge: uncertainties, approximations, model reduction, sampling.
Knowledge of CFD would be a plus.
Name: Bénédicte CUENOT
Name: Jean-Christophe JOUHAUD