ColliderML Drifting Beamspot: Single-Muon Tracking Samples with Smeared Digitisation

Provider
CERN, ColliderML team (Open Data Detector)
License
CC-BY-4.0 (source)
Size
2,032 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/CERN/ColliderML-drifting-beamspot)
Access
Free and public. No account with the data provider is needed.
Category
Physics
Source
https://arxiv.org/abs/2609.32479

About this dataset

Single-muon events simulated in the Open Data Detector (ODD), with the primary vertex spread uniformly over a region much wider than the nominal beamspot (plus or minus 5 mm in x and y, 200 mm in z, 0.32 ns in time). It is the training and evaluation set of Renusch et al., Fast and Precise Learned Charged-Particle Trajectory Regression at the Large Hadron Collider (arXiv:2609.32479), and part of the ColliderML series. Detector response uses Gaussian smearing digitisation with fixed per-layer resolution and no cluster formation, unlike the geometric digitisation of ColliderML Release 1. Muons of both charges are generated over |eta| < 3 in six samples: uniform pT 1 to 110 GeV (201.6 million events), log-uniform pT 0.9 to 110 GeV (200 million), and fixed 2, 10, 50 and 100 GeV (100,000 each). Listing data/ (measured 2026-11-13) shows 30 folders, one per sample and table: particles, tracker_hits, tracker_simhits, tracks and truth_tracks, all without pileup (pu0). The repository uses about 538 GB. Rows are events with Parquet list columns; join tables on event_id, lengths in mm, momenta in GeV. Open with the colliderml library, Hugging Face datasets with streaming, or polars. Released under CC-BY-4.0; cite the paper above.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("CERN/ColliderML-drifting-beamspot")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: data/single_muon_100GeV_pu0_particles/train-00000-of-00001.parquet

Analyze this dataset without downloading it

On Scigantic, ColliderML Drifting Beamspot: Single-Muon Tracking Samples with Smeared Digitisation opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Physics datasets