DIFFER Seminar: From Complex Systems to Simple Models: Preserving Structure for Analysis and Control

Abstract: Large-scale dynamical systems arise in many engineering applications. Although high-fidelity models can accurately describe their dynamics, their computational complexity often limits their use in simulation, monitoring, and control. Model reduction offers a way to construct simpler representations, but reducing model complexity while retaining essential physical and dynamical properties remains a fundamental challenge. In this talk, I will discuss a new perspective in structure-preserving model reduction, with a particular focus on preserving system-theoretic properties and network interconnections. Through examples, I will illustrate possible methods for structure preservation and how they improve the interpretability and suitability of reduced models for subsequent analysis and control.