Press "Enter" to skip to content

Prof. Giordano Scarciotti visited CDSL

Prof. Giordano Scarciotti from Imperial College London visited CDSL on November 10, 2026, and gave a seminar entitled “One Equation to Rule Them All: Direct Data-Driven Cascade Stabilisation, Regulation, and Reduction.” Prof. Scarciotti is an Associate Professor in Control Theory at Imperial, where he is a member of the Control and Power Group and leads the MAC-X research group (Modelling, Analysis and Control of compleX systems). His research concerns how complex dynamical systems can be analyzed, controlled, and reduced, and spans nonlinear systems, model order reduction, optimal control, and data-driven control. He is known for his contributions to model order reduction by moment matching, including its data-driven formulation.

In his seminar, Prof. Scarciotti presented a framework for direct data-driven control of problems that involve interconnections of dynamical systems. The “one equation” in the title is the Sylvester equation, whose solution characterizes an invariant subspace of a cascade interconnection and thus underlies many analysis and design tools. Instead of first identifying a model, the framework computes this solution directly from data and comes with an analysis of the error caused by noise. Since a design task is decomposed into interconnections of subsystems, the effect of the noise can be followed from one stage to the next, which yields bounds on the resulting designs. Prof. Scarciotti discussed three representative problems that the framework covers, namely cascade stabilization, output regulation, and model order reduction. These results are reported in a two-part paper (Part I, Part II) written with Junyu Mao, Emyr Williams, and Thulasi Mylvaganam.

The seminar was closely related to CDSL’s research interests in output regulation and moment-based analysis and design. In particular, CDSL has recently studied how the steady-state response of an interconnected system can be assigned through a Sylvester equation rather than the regulator equations. Prof. Scarciotti’s seminar showed how the Sylvester equation can be solved directly from data and how noise in the data affects the designs built on it, which provided a valuable opportunity to explore how such designs can be carried out without a model. It was a pleasure to have Prof. Scarciotti visit CDSL, and we hope to have more chances to exchange ideas in the future.