Find research datasets worth reusing

HLS4ML LHC Jet dataset (30 particles)

Dataset of high-pT jets from simulations of LHC proton-proton collisions Prepared for FastML/HLS4ML studies: https://fastmachinelearning.org Includes: High level features (see https://arxiv.org/abs/1804.06913) Images: jet images with up to 30 particles/jet (see https://arxiv.org/abs/1908.05318) List: list of jet features with up to 30 particles/jet (see https://arxiv.org/abs/1908.05318)

HLS4ML LHC Jet dataset (50 particles)

Dataset of high-pT jets from simulations of LHC proton-proton collisions Prepared for FastML/HLS4ML studies: https://fastmachinelearning.org Includes: High level features (see https://arxiv.org/abs/1804.06913) Images: jet images with up to 50 particles/jet (see https://arxiv.org/abs/1908.05318) List: list of jet features with up to 50 particles/jet (see https://arxiv.org/abs/1908.05318)

HLS4ML LHC Jet dataset (150 particles)

Dataset of high-pT jets from simulations of LHC proton-proton collisions Prepared for FastML/HLS4ML studies: https://fastmachinelearning.org Includes: High level features (see https://arxiv.org/abs/1804.06913) Images: jet images with up to 150 particles/jet (see https://arxiv.org/abs/1908.05318) List: list of jet features with up to 150 particles/jet (see https://arxiv.org/abs/1908.05318)

HLS4ML LHC Jet dataset (100 particles)

Dataset of high-pT jets from simulations of LHC proton-proton collisions Prepared for FastML/HLS4ML studies: https://fastmachinelearning.org Includes: High level features (see https://arxiv.org/abs/1804.06913) Images: jet images with up to 100 particles/jet (see https://arxiv.org/abs/1908.05318) List: list of jet features with up to 100 particles/jet (see https://arxiv.org/abs/1908.05318)

opencc-by-4.0Feb 2022View details →

Particle-based Fast Jet Simulation at the LHC with Variational Autoencoders: generator-level and reconstruction-level jets dataset

Jets at generator and reconstruction level saved in .npy format. Each jet is represented as an array of jet constituents characterized by their particle momentum in Cartesian coordinates, i.e., (px, py, pz). For both generator-level and reconstruction-level jets, jet constituents are ordered by decreasing pT. The shape of the datasets is [N, 50, 3], where N is the total number of jets, 50 is the number of particles per jet, and 3 is the number of particle features (in order): [px, py, pz]. About 1.7M jets split into training, validation and testing sets at 60%, 20% and 20% respectively.

opencc-by-4.0Apr 2019View details →

R&D Dataset for LHC Olympics 2020 Anomaly Detection Challenge

This is the first R&D dataset for the LHC Olympics 2020 Anomaly Detection Challenge. It consists of 1M QCD dijet events and 100k W'->XY events, with X->qq and Y->qq. The W', X, and Y masses are 3.5 TeV, 500 GeV and 100 GeV respectively. The events are produced using Pythia8 and Delphes 3.4.1, with no pileup or MPI included. They are selected using a single fat-jet (R=1) trigger with pT threshold of 1.2 TeV.  The events are randomly shuffled together, but for the purposes of testing and development, we provide the user with a signal/background truth bit for each event. Obviously, the truth bit will not be included in the actual challenge. These events are stored as pandas dataframes saved to compressed h5 format. For each event, all Delphes reconstructed particles in the event are assumed to be massless and are recorded in detector coordinates (pT, eta, phi). More detailed information such as particle charge is not included. Events are zero padded to constant size arrays of 700 particles, with the truth bit appended at the end. The array format is therefore (Nevents=1.1M, 2101). For more information, including an example Jupyter notebook illustrating how to read and process the events, see the official LHC Olympics 2020 webpage. <a href="https://lhco2020.github.io/homepage/">https://lhco2020.github.io/homepage/</a> UPDATE May 18 2020 We have uploaded a second signal dataset for R&D, consisting of 100k W'->XY with X,Y->qqq (i.e. 3-prong substructure). Everything else about this signal dataset (particle masses, trigger, Pythia configuration, detector simulation) is the same as the previous one described above.  UPDATE November 23 2020 We now include high-level feature files for the background and 2-prong signal (events_anomalydetection_v2.features.h5) and for the 3-prong signal (events_anomalydetection_Z_XY_qqq.features.h5). To produce the features, we have clustered every event into R=1 jets using the anti-kT algorithm. The features (calculated using fastjet plugins) are the 3-momenta, invariant masses, and n-jettiness variables tau1, tau2 and tau3 for the highest pT jet (j1) and the second highest pT jet (j2): 'pxj1', 'pyj1', 'pzj1', 'mj1', 'tau1j1', 'tau2j1', 'tau3j1', 'pxj2', 'pyj2', 'pzj2', 'mj2', 'tau1j2', 'tau2j2', 'tau3j2' The rows (events) in each feature file should be ordered exactly the same as in their corresponding raw event file. For convenience, we have also included the label (1 for signal and 0 for background) as an additional column in the first feature file (events_anomalydetection_v2.features.h5). UPDATE February 11 2021 We have included the Delphes detector card and the Pythia8 command files used to produce the R&D datasets. UPDATE April 17 2022 It was brought to our attention that somehow the raw events file events_anomalydetection.h5 was never updated to v2, which had a lower generator-level pT threshold (PhaseSpace:pTHatMin = 500) for QCD events to minimize artificial trigger sculpting. This v2 is the version that the features file (events_anomalydetection_v2.features.h5) corresponds to, as well as the Pythia cmnd file (pythia_RnD_qcd.cmnd). Now the raw events file has been brought up to date as well.

opencc-by-4.0Oct 2020View details →

Data Augmentation at the LHC through Analysis-specific Fast Simulation with Deep Learning: W+jet large test dataset

W+jet events at generator and reconstruction level, used to train analysis-specific generative models. Events are represented as an array of relevant high-level features. Reco objects are matched to Gen objects and a minimal selection is applied to define the generator support in the N-dim space identified by the input features. About 2M events, used for large-scale testing Details in https://arxiv.org/abs/2010.01835

opencc-by-4.0Aug 2020View details →

Probing Dark Matter with Disappearing Tracks at the LHC

This data set includes: 1. Tables with efficiencies and cross section limits from disappearing charged tracks  in ( life time - DM mass) plane for  i2HDM (Inert Doublet Model), MFDM (Minimal Fermion DM model) and VTDM (Vector Triplet DM model). These are *_efficiency_and_xs_limit.dat files 2. Tree-level cross sections VS DM mass for Pair DM production: D+D- and D+/- D These are * _prod_xs.dat files.   See details  in https://arxiv.org/pdf/2008.08581.pdf

opencc-by-4.0Dec 2023View details →

Top quark pair events for heavy flavour tagging and vertexing at the LHC

This data contains jets from top anti-top decays. The event and parton shower are simulated in Pythia 8 with a centre of mass energy of 13 TeV, with detector response modelled in the Delphes framework. The detector response is modelled on the ATLAS detector, and a mean pileup of 50 was used. The dataset consists of jets, jet constituents, and truth heavy-flavour hadrons. For each jet, up to 50 charged constituents and 5 truth hadrons are included. Each constituent includes a link to the truth hadron associated, if such a link exists.  Provided are 5 files, which are detailed below <ul> <li>class_dict.yaml - Details the relative weights for classification labels, based on the frequency of occurrence for a given entry. Labels are detailed below.</li> <li>norm_dict.yaml - Contains the means and standard deviations of variables that can be used for training, allowing for scaling.</li> <li>pp_output_train.h5 - 13.5 million training jets, consisting of 4.5 million b-jets, c-jets, and light-flavoured jets. Resampling is applied over the jet pT and eta, to ensure equivalent kinematic distributions</li> <li>pp_output_val.h5 - 1.35 million jets for validation, consisting of 450,000 of each jet flavour. Kinematics are resampled in the same way as the training file.</li> <li>pp_output_test_ttbar.h5 - 1.35 million jets for evaluation, consisting of 450,000 of each jet flavour, with no kinematic resampling applied.</li> </ul> Each h5 file contains the following groups: <ul> <li>Jets - (N,) - N jets, including variables such as jet kinematics, flavours, and summary statistics on the number of hadrons and constituents in the jet.</li> <li>Consts (N,50) - Up to 50 charged constituents per jet. Includes details on constituent kinematics and identification. A variable 'valid' is True for tracks in the jet, and False for all other tracks. The additional variable 'truth_hadron_idx' details which hadron (if any) a constituent is associated to.</li> <li>Hadrons (N, 5) - Up to 5 truth heavy-flavour hadrons per jet. Each hadron includes details on kinematics. The variable 'hadron_idx' represents an ID for the hadron, and matches to the constituent variable 'truth_hadron_idx'.</li> </ul> Each group contains both variables that can be used in training, and truth labels, which are as follows: <ul> <li>Jets <ul> <li>flavour - Flavour ID of the jet, 5 for b-jets (containing at least 1 b-hadron within a 0.4 dR(jet, hadron) match), 4 for c-jets (no b-hadrons, and contains at least 1 c-hadron), 0 for light-flavoured jets (contains no b- or c-hadrons)</li> </ul> </li> <li>Consts <ul> <li>truth_hadron_idx - integer that refers to hadron that produced the constituent. '-1' for padded tracks, or tracks with no truth heavy flavour hadron (e.g, pileup, hadronisation).</li> <li>truth_vertex_idx - integer that refers to the vertex a constituent came from - if two tracks have an equivalent truth_vertex_idx, they originate from the same vertex.</li> </ul> </li> <li>Hadrons <ul> <li> <div>hadron_idx - The idx of the hadron, value is '-1' for padded hadrons, 0-4 for remaining hadrons. If a constituent 'truth_hadron_idx' matches with this value, then the constituent came from this heavy flavour hadron decay.</div> </li> <li>flavour - The flavour of the hadron. '5' if the hadron contains at least 1 b-quark, '4' if there is no b-quark but a c-quark is present, '-1' for padded hadrons</li> <li>pt, lxy, dr, mass - The transverse momentum (pt) [GeV], transverse distance between hadronic decay vertex and the primary vertex (lxy) [mm], dR(Jet, Hadron), and the hadron truth mass [GeV]</li> </ul> </li> </ul> This dataset allows for studies into algorithms that aim to perform vertex reconstruction.

opencc-by-4.0Apr 2022View details →

Datasets for Simulation-based Anomaly Detection for Multileptons at the LHC

The simulated background and signal data used for a signal model agnostic machine learning search. This search examined the decay of the Higgs boson to leptons working off of LHC data from the Atlas experiment. Details are provided in the paper entitled "Simulation-based Anomaly Detection for Multileptons at the LHC".

opencc-by-4.0Jan 2024View details →

2401.10320 LHC Dataset

Raw energy-momentum tensor data for <a href="https://arxiv.org/abs/2401.10320">https://arxiv.org/abs/2401.10320</a> The accompanying RHIC dataset is provided in: DOI: <a href="https://doi.org/10.5281/zenodo.10820531">https://doi.org/10.5281/zenodo.10820531</a> Plots based on these data are provided in: DOI: <a href="https://doi.org/10.5281/zenodo.10809853">https://doi.org/10.5281/zenodo.10809853</a> Each data file is accompanied by a file containing the physical coordinates for each of the coordinate array dimensions. The arrays have dimension: (taus, etas, x, y, mu, nu). All data files are npz (compressed numpy arrays) that can be read via <code>numpy.load()</code>. The tar archive extracts to the following folder structure with results for 10 events per parameter set: <pre><code>LHC_data_for_2401.10320 ├── ir_cutoff=0.2000 │ ├── xi_ratio=0.1000 │ │ ├── b_ratio=0.0000 │ │ │ ├── tau=0.2000 │ │ │ ├── tau=0.4000 │ │ │ ├── tau=0.6000 │ │ │ ├── tau=0.8000 │ │ │ └── tau=1.0000 │ │ └── b_ratio=1.0000 │ │ └── tau=0.4000 │ └── xi_ratio=2.0000 │ ├── b_ratio=0.0000 │ │ └── tau=0.4000 │ └── b_ratio=1.0000 │ └── tau=0.4000 └── ir_cutoff=2.0000 ├── xi_ratio=0.1000 │ ├── b_ratio=0.0000 │ │ └── tau=0.4000 │ └── b_ratio=1.0000 │ └── tau=0.4000 └── xi_ratio=2.0000 ├── b_ratio=0.0000 │ └── tau=0.4000 └── b_ratio=1.0000 └── tau=0.4000</code></pre>

opencc-by-4.0Jun 2024View details →

Probing conversion-driven freeze-out at the LHC - Code and Data

The file LLP-CDFO-main.zip contains all the code and processed data for reproducing the results in the  <a href="https://arxiv.org/abs/2404.16086">Probing conversion-driven freeze-out at the LHC</a> paper. The raw data is provided as a separate tarball. Additional instructions can be found in the README file contained in LLP-CDFO.zip or in the <a href="https://github.com/andlessa/LLP-CDFO">GitHub repository</a>.

opencc-by-4.0Nov 2022View details →

Data-driven modeling of beam loss in the LHC

LHC diagnostics datasets from runs 2017 & 2018 used for training of beam loss machine-learning models for research paper "Data-driven modeling of beam loss in the LHC".

opencc-by-4.0Feb 2023View details →

Dataset for Quantum anomaly detection in the latent space of proton collision events at the LHC

Dataset used for https://arxiv.org/abs/2301.10780. The initial dataset is compressed to a low-dimensional latent space using a deep autoencoder. Files with compressed data are provided here in HDF5 format. Different sets of files are given, for different choices of dimensionality for the latent space. A description of the dataset is provided in https://arxiv.org/abs/2301.10780