About
Abstract:
The CERN accelerators generally use a modular control system to deal with the resulting complexity of hundreds or thousands of tuneable parameters. Low level hardware parameters are combined into higher level accelerator physics parameters defined by simulation results. For correction and tuning, low-level feedback systems are available, together with high-level physics algorithms to correct beam parameters based on observables from instrumentation. Still not all effects and processes are covered by this controls approach - only partly because of limited instrumentation or modelling. In recent years numerical optimisers, sometimes combined with machine learning techniques, have led to many improvements and successful implementations in some of these areas. For a certain class of optimisation problems, the methods of Reinforcement Learning (RL) can bring further advantages. In this talk I will show some of the numerical optimisation techniques used at the CERN accelerator complex with successful implementations and finally introduce the key concepts of reinforcement learning with applications and results from the CERN injectors.
Short Bio Verena Kain
Verena Kain received her PhD in accelerator physics with the design of the Injection Protection System for the LHC. She worked on the commissioning of the SPS and CNGS fast extraction systems as CERN fellow and was then Engineer in Charge on the LHC during LHC Run 1.
In 2014 she moved back to the SPS as machine physicist where she contributed with novel techniques on controlling slow extraction and improving slow extracted spill quality.
She has been in charge of coordinating and preparing the commissioning and start-up of the 6 accelerators in the LHC injector chain after the LHC Injector Upgrade (LIU project). Verena is now section leader and responsible for the SPS and LEIR performance and operation.
In 2019 Verena founded the ML coffee as informal meeting ground for topics on ML and advanced…
The talk
Credits and licence
- Institution
- CERN
- Department
- Academic Training Lecture Regular Programme,2021-2022
- Speaker
- Verena Kain
- Year
- 2022
- Level
- Graduate
- Language
- English
- Subjects
- Computer science
- Source
- CERN
- Licence
- CERN, as CERN states it. CERN's terms
- Attribution
- © CERN. CERN provides its audiovisual media free of charge for educational and informational use, credited to CERN; this lecture is shown in CERN's own player.
- Who can watch
- CERN gives no rating. Invisible Network shows it to Tweens profiles and older, judged from what the source says about it. How courses are placed