George Haller

Manifold-based reduced models of nonlinear systems can be significantly extended using Padé approximation. This yields more accuracy with fewer unknowns on larger domains, even from experimental data: nature.com/articles/s4146…

We now have another open Ph.D. position at ETH Zurich in data-driven nonlinear reduced-order modelling, with applications in system ID and control. Interested candidates may apply here: jobs.ethz.ch/job/view/JOPG_…

A smooth, data-driven reduced model using Spectral Submanifolds (SSMs) outperform all other competing approaches on a non-smooth tribomechadynamics benchmark challenge problem: link.springer.com/article/10.100…

Just published by @TheSIAMNews: An introduction to the theory and applications of spectral submanifolds, with a large collection of equation- and data-driven examples in model reduction, system identification, and control. epubs.siam.org/doi/10.1137/1.…

Data-driven nonlinear modelling on spectral submanifolds (SSMs) can uncover problems in the experimental data collection, rather than just fit to the available data. Here, a mismatch between the SSM-predicted forced response and results from the actual experiment revealed a…

Data-driven, nonlinear, SSM-based model-predictive control of soft robots outperforms other model reduction approaches by a large margin, as shown in rdcu.be/ecgH4. Collaborators at @stanford and @ETH_en: @heateralora, @mattiacenedese, and @drmapavone

I'm humbled and honored. Thank you!

🙂🙂

A 2D SSM-reduced model accurately predicts the motion of the full, nonlinear, finite-element model of a plate in a random pressure field. General theory of random SSMs (random spectral submanifolds) with examples here: doi.org/10.1016/j.jsv.…

Data-Driven Model Reduction for Dynamical Systems Montestigliano Spring School for Graduate Students 13th - 19th April 2025, Montestigliano, Italy Register till February 3, 2025 ercoftac.org/events/ercofta…

We can now extract simple and predictive reduced-order models for nonlinear dynamical systems directly from videos! doi.org/10.1007/s11071…

Puzzling period-3 response in forced fluid sloshing turns out to arise from a nonlinear resonance. Predicted by a data-driven, 4D SSM-reduced model, extracted directly from a video of the experiment. nature.com/articles/s4159…

A Ph.D. position is available at ETH Zurich in Data-Driven Nonlinear Model Reduction (a.k.a. Dynamics-Based Machine Learning) and its applications to large-scale physical problems. Interested candidates may apply here: jobs.ethz.ch/job/view/JOPG_…

Data-driven linearization (DDL) explains dynamic mode decomposition (DMD) as a first-order, approximate linearization of the dynamics on a dominant spectral submanifold. DDL then refines this linearization to higher orders with increased accuracy. link.springer.com/article/10.100…

Model Reduction and Machine Learning for Solids, Fluids and Controls (Udine, Italy, Sept. 9-13). Summer school by Bala Balachandran, Charbel Farhat, Michael Graham, George Haller, Shobhit Jain and Gianluigi Rozza. Register here: lnkd.in/eTDmhZc6

Bacterial Barriers (Part 1): New manuscript with @rrjs29 in @JFluidMech shows how interactions between swimming bacteria and flow elliptic Lagrangian coherent structures @GeorgeHallerETH form Lagrangian vortex boundaries (i.e., mixing barriers). See it at bit.ly/3RaXOWI

Non-smooth dynamical systems, such as an oscillating beam rubbing against a wall, also have low-dimensional attracting spectral submanifolds (SSMs). But those SSMs, and the reduced models they carry, are only piecewise smooth. Movie: @leonardo_bttn Paper: doi.org/10.1016/j.ijno…