LHC Olympics 2020 (LHC Olympics 2020 Anomaly Detection Challenge)

这些是 2020 年 LHC 奥运会异常检测挑战赛的官方数据集。每个“黑匣子”包含 100 万个事件，旨在代表实际 LHC 数据。这些事件可能包括信号，挑战包括使用您选择的方法找到这些信号。我们总共上传了三个用于挑战的黑匣子。_x000D_
_x000D_
此外，我们还包含一个 1M 事件的背景样本，旨在帮助应对挑战。背景样本由使用 Pythia8 和 Delphes 3.4.1 模拟的 QCD dijet 事件组成。请注意，此模拟的物理和探测器建模可能无法准确反映黑匣子中的“数据”。对于背景和黑盒数据，使用单个 fat-jet (R=1) 触发器选择事件，pT 阈值为 1.2 TeV._x000D_
_x000D_
这些事件存储为保存为压缩 h5 格式的 pandas 数据帧。对于每个事件，假设所有重建的粒子都是无质量的，并记录在探测器坐标 (pT, eta, phi) 中。不包括更详细的信息，例如粒子电荷。事件被零填充到 700 个粒子的恒定大小数组。因此数组格式为 (Nevents=1M, 2100)。

These are the official datasets for the 2020 LHC Olympics Anomaly Detection Challenge. Each "black box" contains 1 million events intended to represent real LHC data. These events may include signal, and the challenge requires participants to identify these signals using methods of their choice. We have uploaded a total of three "black boxes" for the challenge.

Additionally, we include a background sample of 1M events to assist with the challenge. The background sample consists of QCD dijet events simulated using Pythia8 and Delphes 3.4.1. Please note that the physics and detector modeling of this simulation may not accurately reflect the "data" within the "black boxes". For both the background and black box datasets, events are selected using a single fat-jet (R=1) trigger with a pT threshold of 1.2 TeV.

These events are stored as pandas DataFrames saved in compressed HDF5 format. For each event, all reconstructed particles are assumed to be massless, and their properties are recorded in detector coordinates (pT, eta, phi). More detailed information such as particle charge is not included. Events are zero-padded to a constant-size array of 700 particles, resulting in an array format of (Nevents=1M, 2100).

搜集汇总

背景与挑战

背景概述

该数据集是LHC Olympics 2020异常检测挑战赛的官方数据，包含三个黑匣子，每个有100万个模拟LHC事件，可能包含信号用于检测挑战，并提供一个背景样本辅助分析。数据以压缩h5格式存储，记录粒子坐标，事件数组格式固定为1M x 2100，由多个研究机构于2020年发布。