I come from a machine learning background but working in applied research has made me aware of the limitations of data-based modeling. In real-world applications we must be able to trust our models, and that is not possible if the data does not fully capture the dynamics we want to model or is too noisy. However, integrating physics knowledge can fix it, while still preserving the efficiency of machine learning models.

The CRUNCH group from Brown University, led by prof. Karniadakis, has pioneered some of the most known methods in the field of physics-informed machine learning, such as PINNs (physics-informed neural networks) or DeepONets (deep operator neural networks). I have been working as a part of the group for half a year now and I can say that it has been the most valuable time in my PhD. 

Physics-informed machine learning is a rapidly evolving field which requires collaborative and interdisciplinary approach. Undertaking my research stay here has allowed me to fully participate in the supportive and collaborative environment of the group and discuss new ideas and approaches with researchers who are involved in the field, though they come from a variety of backgrounds. The people whose papers I used to study now guide me in my own research on a daily basis. The CRUNCH group also fosters collaboration within a whole wider research community through weekly seminars, which allows me to connect with, discuss, and learn about the work of researchers in the field from around the world. 

CRUNCH seminar: https://www.youtube.com/@CrunchGroup

Brown University provides ample opportunities to engage with other students, both academically and personally. Pursuing a PhD is characterized by long hours of studying and experimentation, but these often turn into social study sessions with fellow students. The Brown community also offers plenty of activities, such as social nights, cultural events and participating in sports, and some of which have been a brand new experience for me, such as attending football games or celebrating Halloween.

My research stay so far has been an incredibly enriching experience and has given me practical knowledge and understanding of the latest methodologies in the field of physics-informed machine learning. With half a year of my stay remaining, I am eager to apply the methods that I have been researching and developing to more complex real-life applications from NorwAI partners, such as wind turbine operation modeling.