Projects
Externally funded projects currently running in the group.
SYMBIOSIS — entropy-driven model reduction for fluid flows
ERC Consolidator Grant, €2 million, five years.
Turbulent flows cannot be simulated in full, so simulations depend on closure models for what the grid does not resolve. Physics-based closures are stable but often inaccurate; data-driven ones can be accurate but unstable and poor at generalizing. SYMBIOSIS aims to combine the two through the concept of entropy, giving closure models that are at once mathematically stable and physically faithful. The methods target wind energy optimization, CO2 transport and climate modelling.
Discovering the equations of turbulence
NWO Open Competition Domain Science — XS, €50,000.
An attempt to uncover the hidden mathematical equations behind turbulence using neural stochastic differential equations. The XS scheme funds speculative, high-risk work in which negative results count as much as positive ones.
Generative AI for physics simulations
AiNed XS Europa, €80,000.
Development of a physics-aware generative AI model able to generate physics simulations, in collaboration with the I-X Centre for AI in Science at Imperial College London and the School of Computation, Information and Technology at the Technical University of Munich.
DEEPDIP
Integrating neural networks into physics applications through scientific machine learning libraries, including CoupledNODE.jl, a library for learning several coupled ODEs with neural networks in the right-hand side.