Data

Available background-mixed datasets in Rucio. Source of truth:

rucio did list --filter 'is_background_mixed=true' 'epic:*'

The campaign strips below are authored by hand with the <DidStrips> component — see .vitepress/theme/components/DidStrips.vue. To refresh them after a new Rucio campaign, run the update-data-md skill: it queries Rucio, digests the DIDs with scripts/summarize_rucio_dids.py, and guides authoring the strips (labels, colours, and tooltips are set by hand).

Last updated: 2026-09-29

Conventions

• FULL — DD4hep simulation output (sim-level, full Geant4 hits).
• RECO — EICrecon reconstruction output (reco-level).
• Each strip covers one (campaign, data_name) pair. Click a strip to reveal the exact Rucio DIDs with copy buttons; hover any tag for a description.
• main/ branch / CI / Test entries are filtered out — only versioned campaigns are shown.
• epic_craterlake is the only detector configuration.
• Use the per-row 📋 to copy a single DID, the Copy all button to copy every DID in a strip (newline-separated), or click the prefix line to copy the shared DID prefix.
• A strip's CSV path points to our converted CSVs. Files written by eic2ai use the column names on Data Format; files written earlier by the csv_convert/ ROOT macros use the older names (prt_status, event, …), which rename_dataframe(df, role) from eic2ai/column_renames.py maps to the current ones.

Campaign 26.07.1

▸

26.07.1Bkg_Exact1S_2usFULL ·10RECO ·10

DISNCSIDIS10x10010x275minQ2=1minQ2=10minQ2=100minQ2=1000q2_0to1

📋epic:/{FULL|RECO}/26.07.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/Copy prefix

Adds a SIDIS branch (pythia6-eic 1.0.0, q2_0to1) alongside the DIS/NC grid. Every dataset has both a FULL (sim) and a RECO partition.

Campaign 26.07.0

▸

26.07.0Bkg_Exact1S_2usFULL ·8RECO ·8

DISNC10x10010x275minQ2=1minQ2=10minQ2=100minQ2=1000

📋epic:/{FULL|RECO}/26.07.0/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/Copy prefix

Converted for the event-type classifier (simulation-pipeline configs/config-phase12-disnc.yaml): DIS/NC 10x100, minQ2 = 1, 10, 100, FULL + RECO file pairs, 99 frames per file, eic2ai roles trk_hits, cal_hits, reco_dis, event_index; output in /work/eic3/users/romanov/disnc-2026-07-background/csv-2026-09-phase12/10x100_minQ2_<N>/. The files were staged from tape into /cache. They carry edm4eic schema 81000 and need the eic_xl:26.07.0-stable container — a 26.06 container reads them with the collections silently missing.

Campaign 2026-06

Meson-structure k-Λ background

▸

2026-06reco-background

k-Λ10x10010x13018x275background + beam-effects

📋/work/eic3/users/romanov/meson-structure-2026-06/reco-background/Copy prefix

k-Λ signal reconstructed with background + beam effects — not on Rucio, on the JLab work disk (XRootD via root://sci-xrootd.jlab.org, the csvpath host). One leaf per beam energy of k_lambda_<beam>_5000evt_*.edm4eic.root; file counts are in the beam-energy tooltips above. The matching simulation files are in dd4hep-background/<beam> next to reco-background/.

CSVs (simulation-pipeline configs, all under /work/eic3/users/romanov/meson-structure-2026-06/):

Directory	Config	Converter, roles
csv-2026-09-background/<beam>	config-msf-26-06-background.yaml	macros: mc_dis, mcpart_lambda, reco_particles, calo_clusters, trk_hits, cal_hits
csv-2026-09-bkg-phase10/<beam>	config-phase10-cal.yaml, config-phase10-cal-18x275.yaml	eic2ai: trk_hits, cal_hits, reco_dis, event_index; 10x130 (8 files), 10x100, 18x275 (all 107 files)

Campaign 26.05.0

▸

26.05.0Bkg_Exact1S_2us_e_onlyRECO ·16

EW_BSMALPaem-ax-emmupmumaem-axem10x110ma_0.1ma_0.2ma_0.5ma_1.0ma_2.0ma_5.0ma_10.0ma_20.0

📋epic:/RECO/26.05.0/epic_craterlake/Bkg_Exact1S_2us_e_only/GoldCt/10um/EW_BSM/ALP/madgraph5-3.7.0-1.0/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2026-05/csv_eicrecon/Copy CSV path

Campaign 26.04.1

▸

26.04.1Bkg_Exact1S_2usFULL ·8RECO ·8

DISNC10x10010x275minQ2=1minQ2=10minQ2=100minQ2=1000

📋epic:/{FULL|RECO}/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/Copy prefix

Campaign 26.03.0

▸

26.03.0Bkg_Exact1S_2usRECO ·3

DISNCSIDISD0_ABCONV10x10010x275minQ2=1q2_100

📋epic:/RECO/26.03.0/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/Copy prefix

Campaign 25.10

Not available on Rucio, but can be downloaded from JLab xrootd (root://dtn2304.jlab.org:8443//jlab-osdf-ro/eic/EPIC/volatile/...; a strip path /volatile/eic/EPIC/<path> is root://dtn2304.jlab.org:8443//jlab-osdf-ro/eic/EPIC/volatile/<path>). These strips list the directory paths under each background configuration rather than leaf DIDs.

▸

25.10.0Bkg_1SignalPer2usFrameRECO ·1

Synrad

📋/volatile/eic/EPIC/RECO/25.10.0/epic_craterlake/Bkg_1SignalPer2usFrame/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2025-10/csv_reco/Copy CSV path

▸

25.10.4Bkg_1SignalPer2usFrameRECO ·8

DIS/NCDDVCS_ABCONVDVMPSIDISDVCS_ABCONV10x10010x27518x275cocktailsynrad-onlygt-10to101to10000to1

📋/volatile/eic/EPIC/RECO/25.10.4/epic_craterlake/Bkg_1SignalPer2usFrame/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2025-10/csv_reco/Copy CSV path

The six signal channels above resolve to eight leaf datasets (the directories that actually hold the RECO .edm4eic.root files). The DDVCS_ABCONV channel splits into hminus/hplus, and the Synrad_18GeV_… folder is a self-contained subtree (DIS/NC + DVCS_ABCONV) mixed with synchrotron radiation only rather than the full cocktail:

Leaf dataset	Beam	Background	Files
DIS/NC/10x100/minQ2=1	10x100	cocktail	4 978
DIS/NC/10x275/minQ2=1	10x275	cocktail	5 922
EXCLUSIVE/DDVCS_ABCONV/…/q2_0_10/edecay/hminus	18x275	cocktail	7 517
EXCLUSIVE/DDVCS_ABCONV/…/q2_0_10/edecay/hplus	18x275	cocktail	7 517
EXCLUSIVE/DVMP/…/unpolarised/18x275/q2_1_1000	18x275	cocktail	6 803
SIDIS/pythia6-eic/1.0.0/18x275/q2_0to1	18x275	cocktail	1 ⚠️
Synrad_18GeV_…/DIS/NC/18x275/minQ2=1	18x275	synrad-only	46 696
Synrad_18GeV_…/EXCLUSIVE/DVCS_ABCONV/18x275	18x275	synrad-only	20 103

⚠️ SIDIS currently has a single reconstructed file (…run10.ab.0173…) — the rest of that dataset is not (yet) reconstructed on XRootD. Treat it as a placeholder until the campaign completes.

▸

25.10.4Bkg_Exactly1SignalPer2usFrameRECO ·2

EXCLUSIVESIDIS

📋/volatile/eic/EPIC/RECO/25.10.4/epic_craterlake/Bkg_Exactly1SignalPer2usFrame/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2025-10/csv_reco/Copy CSV path

▸

25.10.4Bkg_OnlyPer2usFrameRECO ·1

proton-beamgas

📋/volatile/eic/EPIC/RECO/25.10.4/epic_craterlake/Bkg_OnlyPer2usFrame/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2025-10/csv_reco/Copy CSV path

▸

25.10.4Bkg_RealisticSignalPer2usFrameRECO ·1

SIDIS

📋/volatile/eic/EPIC/RECO/25.10.4/epic_craterlake/Bkg_RealisticSignalPer2usFrame/Copy prefix

📋CSVroot://sci-xrootd.jlab.org//work/eic/users/romanov/epic-background-2025-10/csv_reco/Copy CSV path

Using these DIDs

A DID resolves to a set of files (PFNs) on Rucio Storage Elements. The EIC file-access tutorial covers this in full; this section is the minimum.

• 📚 Rucio usage tutorial — setup, metadata filters, command reference.
• 📚 Use-cases tutorial — download vs. stream, batch workflows, common patterns.

1. Check the client (inside eic-shell)

rucio whoami
xrdcp --version     # must be v5.9 or newer

XRootD needs a 5.9+ client with TLS

JLab XRootD (root://dtn2304.jlab.org:8443) accepts only XRootD clients 5.9 and newer, and the login runs over TLS, so the client needs a CA directory. Check with xrdcp --version. eic_xl 26.06-stable and later work as shipped (5.9.1+, X509_CERT_DIR=/opt/local/certificates; 26.10.0-stable: 5.9.2). Both problems show as Login failed: Got invalid login response: an older client, or a 5.9 client without CA certificates (XRD_LOGLEVEL=Debug then prints Failed to initialize TLS context: CA directory does not exist). For the second, set export X509_CERT_DIR=/etc/ssl/certs. The old host dtn-eic.jlab.org is retired.

The Rucio client in eic-shell comes configured; rucio whoami prints the identity it uses. If it fails, follow the setup on the Rucio usage tutorial.

2. See what's inside a dataset

rucio did content list 'epic:/RECO/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/10x100/minQ2=1'

List the XRootD URLs of its files:

rucio replica list file --protocols root --pfns --rses isopenaccess 'epic:/RECO/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/10x100/minQ2=1'

The simulation pipeline uses the same command to build dataset cards for official campaigns (generate_datasets <stage> --rucio).

3. Download files locally

rucio download 'epic:/RECO/26.04.1/epic_craterlake/Bkg_Exact1S_2us/GoldCt/10um/DIS/NC/10x100/minQ2=1'

(!) A full dataset is large — download a sample first.

Downloads land in a directory tree mirroring the DID. Add --nrandom N to pull only a sample, or --rse <RSE> to pin a storage element.

4. Stream over XRootD (no local copy)

Most analyses can read directly from XRootD — faster on the JLab farm and saves disk:

# List replica URLs for a DID
rucio replica list file --protocols root --pfns --rses isopenaccess 'epic:/RECO/26.04.1/.../minQ2=1' | head

# Then open them directly, e.g. in ROOT / uproot / EICrecon:
root -l 'root://dtn2304.jlab.org:8443//jlab-osdf-ro/eic/EPIC/volatile/RECO/.../file.edm4eic.root'

This needs an XRootD 5.9+ client (see step 1). The same files are readable over HTTPS at https://dtn2304.jlab.org:8443/jlab-osdf-ro/eic/EPIC/volatile/....

See use-cases for the recommended patterns when running many jobs over a dataset.