Bridging science and machine learning at the Physics-Informed Neural Networks Summer School
The Summer School on Physics Informed Neural Networks (PINNs) and Applications from 18 to 30 June 2023 was organised under the direction of Kateryna Morozovska of the Division of Decision and Control Systems.
"Thanks to the support of KTH Digitalisation Platform , we have been able to scale the event size and accept 100 participants to the summer school," says Kateryna Morozovska .
The summer school was led by renowned experts Professor George Karniadakis and Dr. Khemraj Shukla from Brown University. Both of them are internationally recognised experts in Scientific Machine Learning, and, in particular, Professor Karniadakis is considered the inventor of Physics Informed Neural Networks (PINNs). Such an excellent networking and learning opportunity has attracted a large number of applications, resulting in 96 total participants; 47 from KTH, the rest from Europe, the USA, and Asia.
The course was split into two parts. In the first, lectures on scientific machine learning techniques with an emphasis on PINNs were given. In addition to covering theoretical and mathematical topics, these lectures included practical Python instruction using Jupiter notebooks. The curriculum was further enhanced by guest lectures given by Dr. Tor Laneryd and Associate Professor Ricardo Vinuesa.
Beyond academic pursuits, the Summer School provided ample networking opportunities through social engagements, fostering interaction and discussions among participants. Notable events included a dinner at Skansen, offering a relaxed atmosphere for networking and socialising.
"The organisation of the Summer School has been a challenging but yet rewarding experience,” says co-organiser Marco Laudato , a postdoctoral researcher at the Department of Engineering Mechanics at KTH. “The great positive feedback received by both the teachers and the participants is a measure of the success of the event”.
The conducted summer school program
Text: Sofia Tatsis