Resources for Getting Started in Research





Some videos students in our group have found useful for an introduction to particle physics.


In each of the 4 sections below, the tutorials are listed roughly in the chronological order you should go through them. To get started, we suggest you understand all the topics in (2) well, go through the tutorials relevant to your project in (3), and then jump directly to (4). Tutorials in (1) can be used if/when needed for your project.

We suggest you go through this README before you embark on your journey.

Also, feel free to look at this for a complete list of ML activities in particle physics.

(1) Particle Physics (HEP)

  • CMS related terminology: this twiki probably contains everything there is to know about CMS. We suggest you skip it for now but always know that it's there for you :)
  • uproot tutorial: a python package that can read and write files in the .root format without actually requiring or running the ROOT software at all.
  • ROOT tutorial: an open-source data analysis framework used in HEP, which lets you save and access your experiment's data, allows you to process the data in a computationally efficient and statistically sound way and gives you access to all tools to produce publication-quality results. Basically it is C++ with many predefined classes and function.
  • Fermilab-LPC-HATS: a set of tutorials covering different aspects of HEP. Topics include: MET, Trigger, jet-algorithms, etc.

(2) Coding/Computing

(3) Machine Learning (ML)

(4) ML in HEP

  • Top-tagging part-I: introduces you to "top-tagging", a physics task, that can be tackled by ML techniques.
  • Top-tagging part-II: explores a more complicated (but hence rewarding) ML approach to tackle the same task, top-tagging.
  • CMSDAS-2020-ML: a similar run through top-tagging.
  • UCSD particle physics and ML data science capstone project: a set of 9-week material (8-notebooks) that guides you to the task of jet-classification. It takes you from data exploration to building classifiers to making the classifiers robust to improving those classifiers by using a GNN architecture called "interaction network."
  • Fermilab-LPC-HATS-2020-ML: a similar set of 10-notebooks that provides you with a comprehensive guide to the task of jet-classification.


If you need computing resources, sign up for an account on the PRP Kubernetes portal and tell me once you have.