ALICE-EATS-AN-APPLE

ALICE Experiment and Testbed Simulation: A Novel Algorithmic Particle Physics Learning Experiment

GNNs for HEP Particle Tracking

TODO:

1. GNN in this folder DAVIDo~~!
2. XGBoost to do radial stuff
3. XGBoost + KMeansClustering to identify the 18-sided structures present in data
4. Generalize treeToCSV to any ROOT structure, and use arguments

Data processing

1. ROOT files, stored in roots/*.root and scraped from CERN's open data EOS, are converted into CSVs using root treeToCSV.C after downloading root from https://root.cern/install/. This stores each ROOT tree as a separate csv in csvs/Clusters/ and csvs/RecTracks.
2. Convert CSVs to NPZs using processCSVtoNPZ.py, which stores ../clusters.npz and ../tracks.npz.

Data visualization

Assumes that you have your data of the form ../clusters.npz to store particle clusters. run visualizeData.py --events 0 1 2 3 4 to print the ``fSubdetId'' field of the clusters. This uses plotly in order to plot the clusters and colors them using this field. Specifications to visualizeData are as follows:

• events are indexed 0 through 317, and visualizeData will automatically generate only 300,000 points at a time. Over 500,000 points using plotly can crash your computer.

Learning Data Parameters with XGBoost Framework

In boostData.py, tweaking around with the optimal parameters used for mapping various parameters of the dataset loaded in ../clusters.npz, namely the fields of fDetId, fSubdetId, and fLabel[3]. fSubdetId is the most interesting and shows preliminary track processing.

• an ensemble tree method with XGBoost is found that has 99.6% accuracy to within a single integer classification for fSubdetId, which ranges from 0 to up to 7000 and isn't trivially spatially correlated.