Semileptonic ttbar Neutrino Regression Dataset

A set of MC simulated training/testing events for the training/evaluation of neutrino regression models.

• Training set contains 585600 events split across 5 files
  • We leave it up to the individual how to split this dataset for cross validation
• Testing set contains 94293 events and should only be used for final evaluation

Simulation Details

• Events are generated from simulated proton-proton collisions at a center-of-mass energy of 13 TeV.
• Mass of the top quark is set to 173 GeV.
• Hard interactions are simulated using MadGraph5 (v3.1.0), with decays of top quarks and W bosons modeled with MadSpin.
• Event generation is interfaced to Pythia (v8.243) to model parton shower and hadronization.
• Detector response is simulated using Delphes (v3.4.2) with a parametrization that mimics the response of the ATLAS detector.

Event Reconstruction and Selection

• Jets are reconstructed using energy-flow objects and the anti-kt algorithm with R=0.4.
• Jet b-tagging corresponding to an inclusive signal efficiency of 70%.
• At least four jets with pT > 25 GeV in the range |eta|<2.5 are required.
• Exactly one reconstructed electron or muon with pT > 15 GeV in the range |eta|<2.5 is required.

File Contents

Each HDF file contains the “delphes” table which holds multiple arrays and structured arrays.

The reconstructed information includes:

• MET: The missing transverse momentum of the event (stored using polar cords)
  • Keys: MET, phi
• leptons: The single reconstructed lepton in the event:
  • Keys: pt, eta, phi, energy, charge, type
• jets: A zero padded table of the leading 10 jets in the event
  • Keys: pt, eta, phi, energy, is_tagged
• njets: Numpy array holding the number of reconstructed jets in the event
• nbjets: Numpy array holding the number of b-tagged jets

The truth/generator level information includes:

• truth_neutrinos: Truth level information of the single neutrino in the event
  • Keys: PDGID, pt, eta, phi, mass
• truth_quarks: Table of information for the four signal quarks in the semileptonic ttbar decay (b_lep, b_had, W_q1, W_q2)
  • Keys: PDGID, pt, eta, phi, mass
• jets_indices: A numpy array containing the identity of each of the jets in the reconstructed jet table
  • 0 = b_lep, 1 = b_had, 2 = W_q1, 3 = W_q2