Relatori
Paola Cinnella
Paola Cinnella
Sorbonne University, CNRS, Paris
Francia
Sorbonne University, CNRS, Paris
Francia
7 lectures on:
RANS/LES/DNS trade-offs; ML-augmented turbulence closures (field inversion, symbolic/sparse regression); generalization and robustness; intro to Bayesian UQ for modelform and data uncertainty.
Antonio Colanera
Antonio Colanera
SISSA
Italia
SISSA
Italia
with Niccolò Tonicello
12 lectures on:
12 lectures on:
ML for turbulence/LES: surrogate vs embedded SGS models, intrusive/non-intrusive paradigms, PDE-constrained optimization, adjoints, RL/differentiable solvers.
Data-driven modal analysis (SPOD, multiscale POD, DMD/Koopman) and cluster-based ROMs (CNM, ql-ROM) for multi-regime, nonlinear turbulent dynamics.
Hands-on tutorials with EZyRB/PINA: POD/SPOD/DMD workflows, ROM construction, autoencoder-based ROMs, latent-space interpolation and extrapolation assessment.
Luca Magri
Luca Magri
Imperial College London & Politecnico di Torino
UK & Italy
Imperial College London & Politecnico di Torino
UK & Italy
4 lectures on:
Chaos-aware modelling (Lyapunov time, attractors); nonlinear ROMs via autoencoders; latent dynamics learning (ESN/RNN); sequential Bayesian data assimilation; biasaware strategies.
Gianluigi Rozza
Gianluigi Rozza
SISSA, Trieste
Italia
SISSA, Trieste
Italia
4 lectures on:
ROM fundamentals, POD-Galerkin, stability/wellposedness to Navier-Stokes, pressure stabilization, closures, geometry parametrization, non-intrusive/autoencoder ROMs.
Niccolò Tonicello
Niccolò Tonicello
SISSA, Trieste
Italia
SISSA, Trieste
Italia
with Antonio Colanera
12 lectures on:
12 lectures on:
ML for turbulence/LES: surrogate vs embedded SGS models, intrusive/non-intrusive paradigms, PDE-constrained optimization, adjoints, RL/differentiable solvers.
Data-driven modal analysis (SPOD, multiscale POD, DMD/Koopman) and cluster-based ROMs (CNM, ql-ROM) for multi-regime, nonlinear turbulent dynamics.
Hands-on tutorials with EZyRB/PINA: POD/SPOD/DMD workflows, ROM construction, autoencoder-based ROMs, latent-space interpolation and extrapolation assessment.
Masayuki Yano
Masayuki Yano
University of Toronto
University of Toronto
4 lectures on:
Certified and adaptive nonlinear ROMs: a-posteriori error estimation, hyperreduction, space-time/component-based methods, goal-oriented reduction, optimization, real-time CFD.