This is a git-lfs repository containing a range of jet-related data sets.
There are 3 types of processes, simulated with Pythia 8.223:
- Top jets (using ttbar events with tops decaying to hadrons)
- W jets (using WW events with W decaying to hadrons)
- QCD jets (using dijet events)
For each of these processes we cluster anti-kt jets, taking the two hardest jets with rapidity |y|<2.5 and transverse momentum above 200 GeV, 500 GeV or 2 TeV (depending on the data set).
A data set is given at parton level (turning off MPI and hadronisation), at hadron level (turning off MI), and at truth level.
The different folders contain:
- train: training data sets of 500k jets each.
- valid: validation data sets of 50k jets each.
- test: test data sets of 50k jets each.
- declusterings: files containing the lists of declusterings of the jets.
- histograms: text file output from the runs, with some associated histograms.
Each file is a zipped json file where each line corresponds to a separate jet, given as a list of four-momenta of its constituent particles.