Test Sets for Jet Anomaly Detection at the LHC

A few datasets are updated in Version 2.1. These datasets are tagged with 'new' in the file names.

- pT~1.2TeV top jets, the mother Z' mass is slightly adjusted to better match pT peaks around 1.2TeV

- Especially there was a bug in the previous version for top jets with mass 80 GeV `*top_m80_100k*`  (the top jet mass was not correctly set). So please be careful when you use those datasets. 

Data Description

These datasets are generated as a series of test sets for anomalous jet tagging at the LHC. They include boosted W jets, Top jets, and Higgs jets. Jet transverse momentum is focused around 600 GeV and 1200 GeV (with prefix "pt1200_" in file names).  Each file includes 100k original events from MadGraph, but might have slightly less events in the final h5 files due to fatjet pre-selection. Production processes include:

• pp -> W' -> W (jj) Z($\nu \nu$); $m_W=59,80,120,174\text{GeV}$
• pp -> Z' -> t t~; $m_t=80,174\text{GeV}$. For m_t=80 GeV, the decay product W mass is set to 20 GeV.
• pp -> HH -> (hh) (hh), (h -> jj); $m_H=174\text{GeV}$, $m_h=20,80\text{GeV}$

Data Generation

Jet samples in this dataset are generated with MadGraph, Pythia8, and Delphes (no pile-up effects simulated). Particle flow objects are used to cluster jets. FastJet was used for jet clustering. Jets are clustered using the anti-kt algorithm with the cone size R=1.0.

•  Leading jet: $p_T>450\text{GeV}$; sub-leading jet: $p_T>200\text{GeV}$

Data Structure

• To get jets: f['objects/jets']
• For jets, there are two datasets: ['constituents', 'obs']. (jets information is stored with the higher-pt jet first)
  • `obs[:, n_j - 1]`: jet four vectors and n-subjettiness for the $n_j$-th jet (pt, eta, phi, m, tau1, tau2, tau3, tau4, tau5) 
  • pt-sorted (highest first) jet constituents information are stored in variable length arrays for the $n_j$-th jet `constituents[:, n_j - 1]`: $\{E_i,P_{xi},P_{yi},P_{zi},{\text{PID}}_i\}$ (PID: PDG for tracks; [22] for photons; [0] for neutral hadrons) 

Extra Notes

• Since the dataset is structured as events, for W jet samples, only leading jet is available; while for Top and Higgs jets, leading and sub-leading jets are both valid. One might need to restrict the jet $p_T$ range at use.
• e.g. to get leading jet constituents: `f["objects/jets/constituents"][:, 0]`
• The file names are self-explanatory on the corresponding generation process. Each file was generated in 100K original Madgraph events. After the preselection, a small fraction of events is discarded.

Contact

• we are welcoming any feedback, suggestions, or requests on new test samples. Please contact chengtaoli.1990@gmail.com for more information.