MINDS

Machine learning for computational fluid dynamics

About MINDS

MINDS is a Carnot-Mines project aiming at bringing machine learning techniques to the service of computational fluid dynamics. The project is hosted in the CFL team at the Mines ParisTech CEMEF research center.

Development Team

Jonathan Viquerat

Research engineer

Aurélien Larcher

Researcher

Elie Hachem

Professor

PhD students

Junfeng Chen

PhD student

Hassan Ghraieb

PhD Student

Gallery

DRL for conjugate heat transfer Optimization of heat transfer systems using deep reinforcement learning

Drag prediction for arbitrary 2D shapes Training a CNN on a dataset of arbitrary shapes to learn to predict drag coefficients at low Reynolds number

DRL for passive flow control Exploiting single-step DRL for open-loop flow control

DRL with multiple agents for active control of unstable flows Exploiting translational invariance to control unstable 1D flow with deep reinforcement learning

Shape optimization using DRL Generate aerodynamics shapes from scratch using deep reinforcement learning

Field prediction using U-nets Training a U-net on a dataset of arbitrary shapes to learn to predict velocity and pressure fields at low Reynolds number

Open positions

There are no open positions currently. You can still submit an unsolicited application to jonathan.viquerat@mines-paristech.fr