Description du sujet de thèse

Stochastic machine learning model for turbulent flow

Contexte de travail

The proposed thesis aims at constructing a stochastic subgrid scale model with the use of machine learning techniques .The objective of the proposed project is to use direct numerical simulations (DNS) of homogeneous and isotropic flows to examine the joint pdf of the energy flux and the invariants and to use this data set and machine learning techniques to improve deterministic subgrid scale models by optimizing the dependence of the model parameters to the gradients of the resolved fields.