Campaign 26.04.1-stable
Converted (CSV) datasets produced from the 26.04.1-stable ePIC reconstruction campaign.
Naming note
The campaign is labelled 26.04 (software release 26.04.1-stable), but the DIDs were actually registered in May 2026 (created_at: 2026-05-05). The 26.04 in the name refers to the software/release line, not the production date.
What this campaign contains
Reconstructed background-mixed simulation, mixed at one signal event per 2 µs frame (Bkg_Exact1S_2us), with the gold-coated 10 µm beampipe configuration. Source files are EICrecon outputs (*.eicrecon.edm4eic.root) held in Rucio under DIDs like:
epic:/RECO/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/10x100/minQ2=1Key parameters (from each dataset's rucio_metadata):
| Field | Value |
|---|---|
software_release | 26.04.1-stable |
data_level | reconstruction |
geometry_config | craterlake_<energy> |
generator | pythia8 |
is_background_mixed | true |
ion_species | p |
| beams | electron × ion GeV (e.g. 10×100) |
How the CSVs are produced
The conversion streams each ROOT file directly over XRootD (no local copy) and runs the edm4eic_* converter macros inside the campaign container. See CSV Convert for the converters themselves and Available Datasets for the full Rucio catalogue.
Two pipeline steps drive it:
generate_datasets csv_eicrecon --rucio— queries Rucio (is_background_mixed=true, patternepic:*RECO/26.04.1*) and writes one dataset card YAML per DID intodatasets/.40_csv_convert.py csv_eicrecon— reads those cards and generates the conversion jobs, one result directory per dataset, running the config'scsv_eicrecon.macros.
Output layout
The Rucio path tree is not mirrored. Each dataset collapses to one flat directory (slug), and each ROOT file becomes one CSV named by its file index:
csv_eicrecon/
Bkg_Exact1S_2us_GoldCt_10um_DIS_NC_10x100_q2-gt-1/
0000.trk_hits.csv
0000.trk_hits.csv.zip
0001.trk_hits.csv
...
Bkg_Exact1S_2us_GoldCt_10um_DIS_NC_10x275_q2-gt-1/
...- Directory slug mirrors the DID path below
epic_craterlake/, withminQ2=1rewritten toq2-gt-1(no=in names). - File id is the per-file index (
0000,0001, …) taken from the ROOT filename's trailing number.
Dataset cards
Each dataset has a YAML card written by generate_datasets --rucio carrying its DID, slug, parsed metadata, authoritative rucio_metadata, and the full file list:
did: epic:/RECO/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/10x100/minQ2=1
slug: Bkg_Exact1S_2us_GoldCt_10um_DIS_NC_10x100_q2-gt-1
metadata:
beam_energy: 10x100
beam_crossing_angle: -0.025
beam_divergence: hidiv_1
beam_effects: true
physics: pythia8-nc-dis
rucio_metadata:
software_release: 26.04.1-stable
electron_beam_energy_gev: 10
ion_beam_energy_gev: 100
q2_min_gev2: 1
is_background_mixed: true
n_files: 123
files:
- root://.../0000.eicrecon.edm4eic.root