Deniz Bezgin
TU Munich, Germany

Toward Machine-learned Discretizations and Differentiable CFD for Compressible Two-phase Flows

• What is the topic of your PhD thesis?

The topic of my PhD thesis was at the intersection of computational fluid dynamics (CFD) and machine learning. I focused on combining classical numerical methods and data-driven techniques and on developing a diCerentiable CFD solver that can be used both to train machine learning models and to solve inverse problems.

From a fluid mechanics perspective, my work concentrated mainly on compressible singleand two-phase flows. Such flows are central to many engineering applications, but they are also among the most challenging to simulate accurately and eCiciently. My research explored how machine learning and diCerentiable simulations can help address some of these challenges.

• What motivates you in your work?

F​ull interview is available here