Simulated datasets for detector and particle flow reconstruction: CLIC detector, machine learning format

Synopsis

Machine-learning friendly format of tracks, clusters and target particles in electron-positron events, simulated with the CLIC detector. Ready to be used with jpata/particleflow:v2.3.0. Derived from the EDM4HEP ROOT files in https://zenodo.org/record/8260741.

clic_edm_ttbar_pf.zip: e+e- -> ttbar, center of mass energy at 380 GeV

clic_edm_qq_pf.zip: e+e- -> Z* -> qqbar, center of mass energy at 380 GeV

clic_edm_ww_fullhad_pf.zip: e+e- -> WW -> W decaying hadronically, center of mass energy at 380 GeV

clic-tfds.ipynb: an example notebook on how to load the files

Contents

Each .zip file contains the dataset in the tensorflow-datasets, array_record format. We have split the full datasets into 10 subsets, due to space considerations on zenodo, two subsets from each dataset are uploaded. Each dataset contains a train and test split of events.

Dataset semantics (to be updated)

Each dataset consists of events that can be iterated over using the tensorflow-datasets library and used in either tensorflow or pytorch. Each event has the following information available:

X: the reconstruction input features, i.e. tracks and clusters

ytarget: the ground truth particles with the features ["PDG", "charge", "pt", "eta", "sin_phi", "cos_phi", "energy", "jet_idx"], with "jet_idx" corresponding to the gen-jet assignment of this particle

ycand: the baseline Pandora PF particles with the features ["PDG", "charge", "pt", "eta", "sin_phi", "cos_phi", "energy", "jet_idx"], with "jet_idx" corresponding to the gen-jet assignment of this particle

The full semantics, including the list of features for X, are available at https://github.com/jpata/particleflow/blob/v2.3.0/mlpf/heptfds/clic_pf_edm4hep/utils_edm.py and https://github.com/jpata/particleflow/blob/v2.3.0/mlpf/data/key4hep/postprocessing.py.