The SciML GPU Bootcamp (, co-organised by ENCCS,, and NVIDIA and held over two half-days on 21-22 February 2023, was aimed to teach researchers and engineers how to use scientific machine learning (SciML) to address domain-specific data challenges and extract insights from scientific datasets. The bootcamp focused on physics-informed neural networks (PINNs), a class of deep learning (DL) networks that can solve linear and non-linear equations with demanding accuracy and computational performance requirements. 

You can watch the recording here

The bootcamp offered hands-on experience with NVIDIA Modulus, a neural network framework that integrates scientific equations into the DL network training. Participants were guided through step-by-step instructions to study the core concepts of neural operators, PINNs, and how to apply them to scientific problems. Pedagogical lectures and guided sessions were given by Dr. Niki Loppi, AI & HPC Solution Architect at NVIDIA. 

A unique feature of the bootcamp was that teams were given access to the Meluxina EuroHPC cluster for the duration of the hackathon, enabling them to work with cutting-edge compute resources and experience the power of A100 GPUs. The ENCCS team would once again like to thank our friends at Meluxina in Luxembourg for their excellent support and generous allocation of GPU compute hours which provided an additional learning experience to the participants. 

Jupyter notebooks adapted for the Meluxina system were provided as training material (see, and a recording of all presentations during the bootcamp will be published on the ENCCS lessons page (, providing a valuable resource also for those who could not attend. 

As in all previous events we have co-organised with NVIDIA the workshop was well received and participants found it to be highly interesting and useful. We thank NVIDIA for the excellent work and look forward to many more fruitful collaborations in the future!