Skip to content

khwajawisal/Particle-Track-Reconstruction

 
 

Repository files navigation

Code for project on Particle Track Reconstruction - trackml dataset

The repository has code for project done under Dr. Kinjal Banerjee

This is all sharad's work, I have just used his repo. and parts of it to improve my own implemenations. Current Progress:

  • Initial data exploration
  • Clustering
  • Neural Network - FC: 86%
  • Random Forest: 93%
  • Gradient Boosted Classifiers: 96%
  • XGBoost Classifier, 500 trees and (max_depth = 25), Trained on 1 event: 98.1%
  • Exploration of different Neural Network architectures

Particle Physics and Quantum Mechanics:

  • Chapter 1 Griffiths
  • Chapter 2 Griffiths
  • Introductory Quantum Mechanics

Current Approach:

  1. Classification of 2 hits as promising or not
  2. Classification of a third promising hit
  3. Reconstruction of the trajectory based on the three hits classified as promising
  • The current models are trained 1st step(i.e., classification of 2 hits as promising or not), since the same model can be extended in the second step
  • In the final step, the hits that are closest to the reconstructed trajectory will be selected

About

Using Machine Learning to reconstruct trajectories of particles detected as hits by ATLAS experiment at LHC, Cern.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 98.1%
  • C++ 1.4%
  • Other 0.5%