Physical modelling, structure-aware learning, and consulting.

ROSLAB is a research and consulting group, working on dynamical systems, fluid mechanics, data-driven physical modelling, and general machine learning.

Research focus

From first-principles physical modelling to learning representations from experimental and simulated data, with emphasis on spectral properties, symmetries, thermodynamic consistency, and the detection of latent structure in high-dimensional systems.

Consulting

Applied work with industry partners on model development, uncertainty quantification, and deployment of learning systems grounded in physical constraints. See the Consulting page for selected engagements.