Hybrid Reduced Order Models: From Exploiting Physical Principles to Novel Machine Learning Approaches

This webinar presents Dr. Soledad Le Clainche's research on "Hybrid reduced order models: from exploiting physical principles to novel machine learning approaches" as part of the DDPS Talk series from January 31st, 2024. Learn how reduced-order models (ROMs) based on physical principles are being developed using modal decomposition techniques like singular value decomposition (SVD) and higher-order dynamic mode decomposition (HODMD), integrated with machine learning methods such as neural networks. Discover how these hybrid approaches address critical challenges in fluid dynamics with applications in combating climate change through improved combustion efficiency, urban air pollution control, and enhanced aircraft design for better fuel efficiency. Dr. Le Clainche, a Professor of Applied Mathematics at Universidad Politécnica de Madrid, shares insights from her work as PI on several national and EU-funded projects focused on reducing environmental impact through advanced computational fluid dynamics and data analysis techniques.