Research Fellow, SISSA - International School for Advanced Studies
Nicola Demo is a researcher and co-founder of Fast Computing Srl. With a focus on reduced order modeling, data-driven modeling, and machine learning, he has made significant contributions to these fields during his tenure as a Researcher Fellow at SISSA mathLab. His research extends to data-driven modeling, where he leverages machine learning algorithms and statistical methods to extract valuable insights from extensive datasets. Moreover, he actively collaborates on industrial projects, employing his computational expertise to tackle real-world challenges. Through his publications, industrial partnerships, and role as CTO in the Fast Computing startup, he continues to advance computational methodologies, fostering the integration of academia and industry for practical applications.
Real-time simulations for enabling digital twin
Numerical simulations are nowadays a consolidated approach in many industrial and academic fields. Its accuracy however is inversely linked to the required computational cost, resulting typically in a high demand for resources, both in terms of time and energy. Reduced order modeling has emerged in the last decades as a possible solution to accelerate simulations by reducing the high dimensionality of the most used numerical solver. In this lecture, we present a historical overview of these techniques, presenting in detail the current state-of-the-art of proper orthogonal decomposition frameworks, which are integrated also with data science and machine learning approaches. Thanks to their capability, these methods play a fundamental role in digital twin construction, enabling the possibility to quickly compute the physical model and providing so an (almost) real-time digital replica.
The final part of the lecture will concentrate on a more practical session, where we will focus on some of the open-source libraries that implement reduced order model techniques, showing how to tackle some example problems also from the software perspective.