Software developed in the Scientific Computing Group

Neural networks for ODEs (including closure modeling for turbulent flows)

• NeuralClosureTutorials: Neural closure modeling tutorials from the 2023 SciML semester programme.

• IncompressibleNavierStokes.jl: Incompressible Navier-Stokes solver in Julia with neural networks to model turbulence.

• DEEPDIP project: Project to integrate neural networks in physics applications through SciML libraries. In particular, the CNODE is a coupling library in which multiple coupled ODEs with neural networks in the right-hand side can be learned from data.

• Energy-conserving neural networks

• Neural closure models

General 2D/3D Navier-Stokes solvers

• IncompressibleNavierStokes.jl: Incompressible Navier-Stokes solver in Julia.

• INS2D: Incompressible Navier-Stokes solver in MATLAB.

• Spectral solver: spectral solver.

1D compressible flow solvers

• HELMEOS: An entropy stable solver for the compressible Euler equations using a real gas equations of state based on the Helmholtz free energy.

• SPADES: An entropy stable solver for the shallow water equations as part of our collaboration with Inria MEMPHIS team.

Reduced-order models

• INS2D: Incompressible Navier-Stokes solver with energy-conserving POD-Galerkin.

Uncertainty quantification

• EasySurrogate: Toolkit for creating surrogate models.

• EasyVVUQ: Toolkit for Verification, Validation and Uncertainty Quantification.

• UQ4WIND: Uncertainty quantification for aeroelastic wind turbine calculations, based on the UQLab tools.