USI - Email

Archive / INF Seminars / INF_2024_03_07_Chiara_Segala

USI - Email

	Moment-driven predictive control and kernel methods for mean-field collective dynamics

Host: Prof. Michael Multerer

Thursday

07.03

USI Campus Est, room D1.14, sector D // Online on Microsoft Teams

11:00 - 12:00


	Chiara Segala

Abstract:
In the first part of the talk the synthesis of control laws is presented for interacting agent-based dynamics and their mean-field limit. A linearization-based approach is used for the computation of suboptimal feedback laws obtained from the solution of differential matrix Riccati equations. The feedback laws are embedded into a nonlinear model predictive control framework where the control is updated adaptively in time according to dynamic information on moments of the dynamics. The performance and robustness of the proposed methodology is assessed through different numerical experiments.
In the second part of the presentation, evidence of the kernel methods efficacy is provided for a variety of multi-agent systems. Motivated by recent developments of learning approaches in the context of interacting particles, an investigation is conducted on kernel methods acting on data with many measurement variables. The mean field limit of kernels and numerical results are presented in order to reduce the computational effort for simulation and control of interacting particle systems.

Biography:
Chiara Segala started her academic journey by earning a Master's degree in Mathematics from the University of Verona in 2017.
She then received a Predoc Research Grant from the University of Verona, focusing her work on geometric control of mechanical systems.
Supervised by Prof. Giacomo Albi, she earned a Ph.D. in Mathematics at the University of Trento and successfully defended her thesis "Robust control strategies for mean-field collective dynamics" in 2022.
Currently, she is a Postdoctoral Researcher at RWTH Aachen University. Her research in numerical analysis and optimization is conducted within the research group led by Prof. Michael Herty.
Her research interests include multiagent systems, crowd dynamics, uncertainty quantification, optimal control, model predictive control, sparse control, neural networks, kernel methods, multiscale models, and meanfield theory.

Click here to join online