Physics-Based Model Discovery with Symbolic Regression
Speaker: Dr. Alessandro Lucantonio
Associate Professor at Aarhus University (DK)
When: May, 08, 2025 | 9:00 AM – 10:00 AM
Location: CEME Building, room 1203 (6250 Applied Science Lane, Vancouver)
Zoom: https://ubc.zoom.us/j/97206574075?pwd=b0tpc3czNmV0YmJ5dUhmcGwya0RjZz09
This talk is part of the Structural Matter Seminar Series.
Abstract:
I will present our research on advancing symbolic regression for model discovery, with a focus on developing new algorithms to uncover interpretable and accurate models for complex physical systems. Key areas of exploration include reduced-order modeling for fluid mechanics, robotic control, and fracture mechanics. Our work also introduces a pioneering framework that integrates evolutionary algorithms with discrete calculus primitives, enabling the discovery of spatio-temporal models for diverse applications. In particular, we are applying this framework to traffic flow modeling, demonstrating its versatility in addressing real-world challenges.
Biography:
Alessandro Lucantonio is Associate Professor at Aarhus University (DK), where he leads the Computational Physics and Machine Learning Lab. He was previously Assistant Professor at the BioRobotics Institute of Scuola Superiore Sant’Anna (Italy) and postdoc at SISSA (Italy). In 2022, he received an ERC Starting Grant focused on using machine learning and symbolic methods to discover interpretable models for complex physical systems.