Visit the repository's website for more information
These tutorials were first used for the 2022 MT ARD ST3 pre-meeting Machine Learning Workshop , part of the 10th MT ARD ST3 Meeting 2022 in Berlin.
You can run the notebooks in this repository either locally or on the cloud.
- To run in the cloud, click on the Binder badge at the top left of this page. This may take a minute to spawn.
- To run the notebooks for this workshop locally, you will need a Python 🐍 installation. You can then clone the workshop repository and access it:
git clone https://github.com/ansantam/2022-MT-ARD-ST3-ML-workshop.git
cd 2022-MT-ARD-ST3-ML-workshopWith Python installed, run the following command to install the needed packages:
pip install -r requirements.txtThen start the Jupyter Notebook by running
jupyter notebookYou are now ready to execute the workshop notebooks 🎉
Note: The reinforcement learning notebook has a number of additional requirements and cannot be run on Binder. Please refer to the instructions in the notebook itself for the installation.
- Introduction to machine learning in accelerator physics
- Introduction to artificial neural networks
- Introduction to Bayesian optimization
- Application of Bayesian optimization to improve the injection efficiency at KARA
- Reinforcement learning for particle accelerators
- Basic reinforcement learning introduction without ML libraries (dynamic programming)
- Quadrupole surrogate model (tensorflow & pytorch)
It may be that you cannot clone the repository to your local machine because git is not installed. In that case, instead of using git, simply visit the repository home page, click the green Code button and select Download ZIP.
You might encounter this error on a Linux distribution that does not come with gcc installed by default (such as Ubuntu). Run the following commands to fix, then re-run the command to create the environment.
sudo apt install build-essential
conda env remove --name mt-ard-st3-ml-workshop