Active learning

Dynamics and control

Fluid–structure interaction

Acoustics and vibrations

In many physical systems, transient instabilities are difficult to control because of the curse of dimensionality. Order-reduction techniques have become customary because design and implementation of controllers in low-dimensional subspaces is computationally more tractable. We explore how reduced-order models can be combined with high-fidelity computational models to shed light on the physical mechanisms responsible for transient instabilities. A key question is the incorporation of compressed sensing, state reconstruction, and machine learning capabilities in the control algorithms for them to be deployable in the wild.

Read more

Learning modes of instability from data: Chaos 29, 113120 (2019)

Reduced-order control of fluid flows: Physical Review Fluids 4, 053902 (2019)

Reduced-order modeling of instabilities: SIAM Journal on Applied Dynamical Systems 18, 1143–1162 (2019)