Top quark pair production at the LHC, with all hadronic resolved decays for solving event combinatorics

R&D Datasets for solving event combinatorics in all hadronic top quark pair events at the LHC.

Used in the development of Topographs: Topological Reconstruction of Particle Physics Processes using Graph Neural Networks

The datasets contain 5.8M ttbar events in the all hadronic decay channel, with jets matched to the truth partons in the top quark decays.

Event generation

Centre of Mass energy: 13 TeV

MC Generator: MadGraph5_aMC@NLO v3.1.0, with MadSpin modelling the decays of the top quarks and W bosons.

Parton Shower: Pythia v .243

Detector response: Delphes v3.4.2 using ATLAS-like geometry

Jets reconstructed with anti-kt algorithm, R=0.4, using FastJet

b-Tagging corresponds to inclusive 70% b-jet efficiency

Event selection and truth matching

All events are required to have at least six reconstructed jets and exactly zero leptons (electrons or muons)

Partons are matched to jets using \(\Delta R\) matching, with \(\Delta R < 0.4\)

Events with partons matched to multiple jets or jets to multiple partons are discarded

Up to 16 jets are stored per event

In the training dataset 1,340,000 events have all partons from the ttbar decays matched to jets.

In the validation dataset, 71,000 events have all partons from the ttbar decays matched to jets.

In the testing dataset 76,000 events have all partons from the ttbar decays matched to jets.

Dataset format

The dataset is in h5 format and the key 'delphes' has the following numpy arrays:

jets (16), jets_indices (16), matchability, nbjets, njets, partons (10)

Jets structured numpy array per event:

(pt, eta, phi, energy, is_tagged)

Jets_indices:

Integer corresponding to the parton the jet is matched to

From 0 to 5: b1 W1j1 W1j2 b2 W2j1 W2j2 (1= from top, 2=from antitop)

-1 indicates not matched to a parton

Properties of matched partons can be obtained from the partons array

matchability:

Which partons are matched to jets in event

Binary representation with bits corresponding to each parton (length 6) 0b111111

From left to right: b1 W1j1 W1j2 b2 W2j1 W2j2

0b111000 (56) is one top fully matched, 0b000111 (7) is the other top fully matched, 0b111111 (63) is both tops fully matched

njets, nbjets:

How many jets/bjets in event

partons:

List of truth particles from ttbar decay: tops, Ws, quarks, ordered by top quark and its decays followed by anti-top and its decays

PDGID, pt, eta, phi, mass