Training materials
Presentations:
Machine Learning-based methodologies for combustion – 1
Interactive HPC with Jupyter
—————————————–
Venue: Online via Zoom
Date: 22-23 March 2022
Description: South-East Europe Combustion Spring School 2022 intends to give a high view of fundamental aspects and current challenges in combustion, as well as to introduce the use of Machine Learning (ML) and High Performance Computing (HPC) to approach Exa-scale simulations of turbulent reacting flows. In this course, different methodologies to enhance the computational performance of high-fidelity combustion simulations will be introduced. The methods cover from node to system-level performance optimisations and algorithms for combustion simulations. This is the first training course in a series of two CoEC seasonal schools that will take place in 2022. This introductory school to combustion will allow participants to acquire new knowledge and strengthen their understanding in a range of topics including laminar flames, turbulent combustion, ML and HPC methodologies in combustion and interactive supercomputing for in-situ analysis. In the following up CoEC School for 2022, we will build on the ground of the first school by extending the scope and topics covered and by including a number of hands-on exercises.
The CoEC Spring School 2022 is a joint effort by NCSA Bulgaria, CERFACS (France), BSC (Spain), Institute of Combustion Technology (ITV) – RWTH Aachen University (Germany) and Juelich Supercomputing Centre (Germany).
What you will learn
- Basics of laminar combustion including equations, combustion chemistry, flame regimes and structures;
- Understanding of phenomenology of turbulence and turbulent combustion;
- Presentation of turbulent combustion models for the simulation of combustion chambers;
- Introduction to Center of Excellence in Combustion (CoEC): Applying exascale computing technologies to promote and develop advanced simulation software;
- Introduction to ML algorithms and to combustion models that will benefit from ML;
- Presentation of modeling of chemical mechanisms through artificial neural network;
- Presentation of modeling turbulence and scalar fluxes in numerical simulations through a convolutional neural network and super-resolution generative adversarial network.
Target audience: Under-graduate and graduate students, Ph.D. students, junior researchers (academy and industry) with some knowledge in Computational Fluid Dynamics (CFD).
Selection process: There is a threshold number of participants (no more than 90 registrations), a timely registration is essential. We will stick to the following criteria:
- 80% for attendees from institutions (universities, research centers, and companies) based in the EU or Associated Countries to the Horizon 2020 Programme and 20% for others.
- If the threshold is exceeded, we will prioritise attendees from South-East Europe, Israel, and Ukraine as well as participants that come from EU13 countries.
Programme: Please click here
Cost: There is no registration fee. The CoEC School is free of charge.
Registration: Please click here
Registration deadline: March 16, 2022.
For any further information, please feel free to contact us at g.prangov@ncsa.bg
Find more information on the NCSA website.