ePIC generator data survey

A survey of the ePIC simulation data the collaboration holds — event-generator (EVGEN) samples, FULL and RECO campaign output, and the machine-background inputs needed to mix frames — organised as selection tables: which physics process exists at which beam energy, whether the beam crossing angle has been applied, and where background can be added. Built from a read-only sweep of /volatile, /cache, /mss and Rucio on the JLab farm. See also Available Datasets for the Rucio DID catalogue and Backgrounds for the mixing pipeline.

Date: 2026-10-15 · Scope: 20 beam configurations, 190 physics branches. Marco's two tables are transcribed verbatim (sections 1–3); the availability survey (sections 5 and 8) is built from a read-only sweep of /volatile, /cache, /mss and Rucio on ifarm, with the afterburner state of every branch measured from the beam momenta rather than inferred from path names (section 4).

Summary of findings

Four energies are mixable today — 10x100, 10x130, 10x275 (vacuum and gold-coated 10 µm) and 18x275 (vacuum). The campaign's per-source mixing configs reference 120 background files and 117 are on disk. 5x41 and 5x100 are blocked by exactly three missing files. The 9x points have no rate table — but since the campaign already reuses the 10x275 electron-ring files for 10x100 and 10x130 with only freq changed, 9x is a rate-table problem, not a production problem. Section 5.1.
190 physics branches measured across 20 energies, each with its afterburner state read from the beam momenta rather than inferred from the path. Section 5.2 is the catalogue, keyed on process rather than on Marco's class slugs; section 5.3 lists the processes his table has no row for.
RECO is a first-class source, not a fallback. MCParticles keeps the crossing angle and the full Pythia8 history, so edm2hepmc recovers usable generator events — including at 9x, where FULL has nothing at all, and from tape, where much of EVGEN is /volatile-only. Section 5.6.
Three of Marco's rows that look missing are not (rho, spectroscopy, omega — the energy is in the filename, not the path), and tcs is ambiguous: pure TCS exists only head-on. Section 5.7.

0. Sources

What	Where
Primary slides	Event Builder for EICRecon, O. Partridge & M. Meyer-Conde, SRO Workshop, 2026-08-19 — indico 33464 / contrib 128271 (32 slides)
Machine-background rates	https://github.com/eic/simulation_campaign_datasets/tree/main/config_data
Physics signal dataset listing	https://github.com/eic/simulation_campaign_datasets
generatorStatus encoding issue	https://github.com/AIDASoft/DD4hep/issues/1657#issuecomment-5280917973

Tables below are transcribed from the PDF. Text extraction lost the ff/fi ligatures (Di ractive, Classi cation, o eAu); those are restored here. Nothing else is reworded — Marco's own shorthand and notes are kept verbatim.

1. Marco's ML class table (slide 12, Machine Learning Classes)

Goal of the GNN, as stated on the slide: classification (Bkg vs Phy.) and segmentation (attributed hits). Class definitions grouped as NC/CC DIS, exclusive hadron, exclusive meson, photo-production.

#	Class	dir	Physics Process	Generator / source
0	Background	`bkg`	SynRad (2000), e-beam-gas/brems (3000), Coulomb (4000), Touschek (5000), p-beam-gas (6000)	dataprod / GETaLM / Xsuite / Pythia8
1	NeutralCurrentDISQ1	`ncdisq1`	NC DIS, Q² ≥ 1	Pythia8
2	NeutralCurrentDISQ10	`ncdisq10`	NC DIS, Q² ≥ 10	Pythia8
3	NeutralCurrentDISQ100	`ncdisq100`	NC DIS, Q² ≥ 100	Pythia8
4	NeutralCurrentDISQ1000	`ncdisq1000`	NC DIS, Q² ≥ 1000	Pythia8
5	ChargedCurrentDISQ100	`ccdisq100`	CC DIS, Q² ≥ 100	Pythia8
6	ChargedCurrentDISQ1000	`ccdisq1000`	CC DIS, Q² ≥ 1000	Pythia8
7	DiffractiveDIS	`ddis`	Diffractive DIS, intact proton	RAPGAP
8	DVCS	`dvcs`	Deeply Virtual Compton Scattering	DVCS generator
9	DDVCS	`ddvcs`	Double DVCS (ep→eℓ⁺ℓ⁻p) — gated to BEAM=18x275	EpIC (18x275/10x130 only)
10	DVMP	`dvmp`	Deeply Virtual Meson Production	EpIC
11	DEMPQ3to10	`dempq3to10`	Deep Exclusive Meson Production, Q² ∈ [3,10]	DEMPgen
12	DEMPQ10to20	`dempq10to20`	DEMP, Q² ∈ [10,20]	DEMPgen
13	DEMPQ20to35	`dempq20to35`	DEMP, Q² ∈ [20,35]	DEMPgen
14	TCS	`tcs`	Timelike Compton Scattering	TCS generator
15	DiffractiveJPsi	`jpsi`	Diffractive J/ψ electroproduction, ep	lAger
16	JPsiPhotoproduction	`jpsiphoto`	Coherent J/ψ photoproduction — gated to BEAM=18x275	eSTARlight (18x275 only)
17	UChannelRho	`urho`	u-channel ρ exchange	SLIGHT/GENEX
18	CoherentRho	`cohrho`	Coherent diffractive ρ off eAu (nuclear rapidity gap)	Sartre
19	UChannelPi0	`upi0`	u-channel π⁰ exchange — gated to BEAM=18x275	(18x275 only)
20	Upsilon	`upsilon`	Υ threshold production	eSTARlight-family
21	MesonStructureFn	`mesonsf`	K⁺Λ exclusive meson structure function	eicMesonSFGen
22	CoherentPhi	`cohphi`	Coherent diffractive ϕ→K⁺K⁻ — off eAu (nuclear rapidity gap)	Sartre
23	Rho	`rho`	Exclusive ρ→π⁺π⁻ electroproduction, ep, Q² ∈ [0,5] — npsim SIGSEGVs, unresolved	SLIGHT
24	Spectroscopy	`spectroscopy`	Exotic states X, Y, ψ(2S)	dedicated spectroscopy generators
25	Omega	`omega`	ω neutral-decay photoproduction — gated to BEAM=18x275	SLIGHT (18x275 only)
26	Photoproduction	`photoprod`	Min-bias photoproduction, Q² ∈ [0,1]	Pythia6-eic

Classes 9, 16, 19, 25 are marked by Marco as gated to 18x275 only. Checked: correct for jpsiphoto (16), upi0 (19) and omega (25); ddvcs (9) also has a 10x130 production. Section 5.2 carries the measured picture.

Class 23 (rho) is annotated "npsim SIGSEGVs, unresolved". Confirmed that this is a simulation failure, not missing input: EXCLUSIVE/RHO_ABCONV holds 52 afterburned 10x100 files (719 MiB, slight_10x100_Q2of0to1_rho2Pions_ab_hiAcc_*.hepmc.gz, measured 25.010 ± 0.156 mrad). Generator level is fine; the ddsim step is what broke.

A caution about this table as a target list. Marco's dir slugs are ML class names, not paths, and his row set reflects what he chose to mix in summer 2026 — not the limit of what the collaboration has. The survey below is therefore organised by what exists and is usable, with Marco's slug as a cross-reference column rather than the row key. Three of his rows that look absent from a path-name search (rho, spectroscopy, omega) in fact have data; several usable processes have no row at all.

2. Machine background sources (slide 10, Signal and Background Mixing)

Rates in kHz/GHz are from simulation_campaign_datasets/config_data.

Source	Status	Vacuum	Gold Coating	File size	Effect
Synchrotron Radiation	2000	36.608 GHz	13.277 GHz	~78 MB	1–2 photons per event; unique templates reused 10 000×
Electron bremsstrahlung radiation	3000	3.177 MHz	3.177 MHz	~94 MB	Few-particle QED elastic scatter
Electron Touschek scattering	4000	29.56 kHz	29.56 kHz	~525 MB	Intra-bunch scatter; slightly more complex final state
Electron Coulomb scattering processes	5000	233.5 kHz	233.5 kHz	~6.2 GB	Full DIS final state with hadronisation
Proton beam gas interactions	6000	31.9 kHz	31.9 kHz	~16.7 GB	Soft/hard hadronic production; many secondary tracks

Discrepancy in the slides — resolved from the production configs. Slide 10 labels 4000 = Touschek (29.56 kHz) and 5000 = Coulomb (233.5 kHz), while slide 12 row 0 lists "Coulomb (4000), Touschek (5000)". The campaign's own mixing configs (simulation_campaign_datasets/config_data/synrad_…hgas_…json) settle it: the file named …electron_coulomb_10x275… carries status: 4000 at freq: 29.56, and …electron_touschek_10x275… carries status: 5000 at freq: 233.5. Slide 12 is right; slide 10's two row labels are swapped relative to their own rates. Use 4000 = Coulomb, 5000 = Touschek.
Also note the units: freq in those configs is in kHz, so the SynRad 13277000 is 13.277 GHz and the 3177.25 e-beam-gas entry is 3.177 MHz — consistent with slide 10 once scaled by 10³.
There is no separate bremsstrahlung input file despite the egasbrems in the config filenames: the single GETaLM electron-beam-gas sample at status 3000 is the brems component. Don't go looking for a sixth file.

3. generatorStatus encoding (slide 11)

Generated status range	Meaning
0–99999	bkg / machine background (unchanged) — 10 × 1000-wide sub-bands: SynRad, e-beam-gas/brems, Coulomb, Touschek, p-beam-gas, plus headroom for a future 6th+ source
100000–199999	`ncdisq1`
200000–299999	`ncdisq10`
…	…

Per-class band = 100000 × (class index). Event-level information carried in the physics event weight; frame-level in EventBuilderFrameInfo; AI ground truth encoded in generatedStatus of the *HitLinks for segmentation.

Slide 31 flags an unresolved-class problem: hits with no truth link at all, and an undocumented generatorStatus gap [7000, 10000], suspected integer overflow / signed-unsigned conversion in generation software.

4. Afterburner state: crossing angle vs head-on

Head-on generator output cannot be fed to the DD4hep/ddsim simulation. The ePIC beams cross at 25 mrad; the afterburner (abconv) applies the crossing angle, beam divergence and vertex smearing to a head-on HepMC file. Simulating a head-on file against the ePIC geometry puts the beams down the wrong axis and the result is unusable.

So every entry in the availability survey carries an explicit afterburner state, and a dataset that is head-on only is not usable generator data for our purpose — it is at best a candidate that first needs abconv run on it.

4.1 Naming conventions tested against measurement

The EVGEN tree marks some afterburned branches with an _ABCONV suffix on the process directory, and some files carry ab / _ab / .ab in the filename. Neither is reliable, measured today:

Checked	Beam angle (e⁻ / p), entry 0	Verdict
`EXCLUSIVE/TCS/10x100/hel_plus/tcs_10x100_10.hepmc`	`px=py=0`, `pz=-10` / `px=py=0`, `pz=+100`	head-on
`EXCLUSIVE/TCS_ABCONV/10x100/hel_plus/tcs_bh_int_ab_hiAcc_10x100_hel_plus_1.hepmc`	p `px=-2.525` → 25.2 mrad, event vertex offset present	afterburned
`EXCLUSIVE/DVCS/10x100/DVCS.1.10x100.hepmc`	`px=py=0` both beams	head-on
`EXCLUSIVE/DVCS_ABCONV/10x100/DVCS.1.ab.hiAcc.10x100.hepmc`	p `px=-2.525` → 25.2 mrad	afterburned
`EXCLUSIVE/DVCSD_ABCONV/5x41/DVCS.1-eD-ab-hiAcc-5x41.hepmc`	n `px=-1.038` → 25.3 mrad	afterburned
`DIS/pythia8.316-1.0/NC/noRad/ep/9x275/…_run000.hepmc3.tree.root` — no `ab` token in path or filename	0.08 / 25.245 mrad	afterburned
`DIS/NC/5x41/minQ2=1/pythia8NCDIS_5x41_minQ2=1_beamEffects_xAngle=-0.025_hiDiv_1.hepmc3.tree.root`	0.135 ± 0.070 / 25.021 ± 0.227 mrad	afterburned, hiDiv
`DDIS/rapgap3.310-2.0/noRad/ep/9x275/…_ab.hepmc3.tree.root`	0.200 / 25.087 mrad	afterburned
`EXCLUSIVE/DEMP/DEMPgen-1.2.5/9x275/q2_3_10/pi+/…_ab.hepmc3.tree.root`	0.200 / 25.087 mrad	afterburned
`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep/9x275/…_ab_run0001`	0.362 / 25.161 mrad	afterburned
`EXCLUSIVE/DVMP/EpIC1.1.8-1.0/unpolarised/9x275/hiDiv/…_ab_hiDiv_run001`	0.247 ± 0.128 / 24.995 ± 0.124 mrad	afterburned, hiDiv
`SIDIS/pythia6-eic/1.3.0/ep_noradcor/9x275/q2_10to100/….ab.…`	0.233 ± 0.129 / 24.987 ± 0.127 mrad	afterburned

The takeaways:

_ABCONV directory with a bare sibling of the same process (TCS vs TCS_ABCONV, DVCS vs DVCS_ABCONV) → the bare one is head-on. Here the convention holds.
The DIS, DDIS, SIDIS and EXCLUSIVE/DVMP, EXCLUSIVE/DEMP trees carry no ABCONV segment at all and are afterburned anyway. Judging them by the directory name would have wrongly excluded the main DIS samples — including every 9 GeV-electron DIS dataset.
So the survey classifies by measurement, with the path convention only as a cross-check.

4.2 Measurement method

hepmc3.tree.root files: open with ROOT 6.40 inside eicweb/eic_xl:nightly (gSystem.Load("libHepMC3rootIO")), read the hepmc3_tree / hepmc3_event branch, take the two status == 4 particles and compute atan2(hypot(px,py), |pz|). apptainer needs -B /volatile:/volatile:ro and a writable TMPDIR, or TFile::Open fails with "does not exist". Plain .hepmc (HepMC3 ASCII) files are cheaper: head the file and read the P lines with trailing status 4.

Beam divergence and vertex smearing separate hiAcc (nominal angle, no divergence) from hiDiv (realistic divergence); DIS/NC and DIS/pythia8.316-1.0 samples show vertex-z spread of 34–45 mm, i.e. realistic beam effects. For streaming-readout / timing work hiDiv is the physical choice where both exist.

The smeared interaction point lives in two different places depending on the production chain, and reading only one of them gives a false "no beam effects":

event_pos on the HepMC3 event (the @ x y z t field on the ASCII E line) — this is where the abconv-processed EpIC / DEMPgen / lAger / eSTARlight / rapgap / pythia6-eic files put it;
vertices[0].position — this is where the pythia8.316-1.0 and DIS/NC pythia8 NCDIS files put it, with event_pos left at zero.

The survey reads both and takes whichever is non-zero. Three ASCII branches (DIS/CC/{5x41,10x100,10x275}/minQ2=*) record neither despite being beamEffects_xAngle=-0.025_hiDiv productions; those are reported as "vertex spread not stored in this file" rather than as having none.

4.3 Directory census by afterburner naming

First count of energy-named directories under /volatile/eic/EPIC/EVGEN (depth ≤ 7), split by whether the path contains ABCONV — remembering from 4.1 that a missing ABCONV does not mean head-on:

Energy	energy dirs total	of which under an `ABCONV` branch
5x41	25	9
9x100	8	7
9x130	18	10
9x275	21	12

(Whole-tree token census, any depth: 10x100 85, 5x41 81, 18x275 79, 9x275 43, 9x130 42, 10x130 41, 5x100 24, 10x110 22, 10x275 18, 10x115 16, 10x250 15, 10x166 12, 9x100 10, 9x166 8, 18x110 4, 9x115 3, 18x100 3. The 9 GeV-electron configurations are real and well populated, except 9x100 which is thin.)

5. Survey of available data

Target energies (7), agreed 2026-10-15:

Priority	Energies
Asked for	5x41, 9x100, 9x130, 9x275
Added as the 10 GeV-electron counterparts, for comparison	10x100, 10x130, 10x275

Wanted: event data without background, in order of preference:

Event generator files (HepMC / hepmc3.tree.root) — preferred, and only directly usable if afterburned (section 4).
FULL (ddsim/DD4hep) campaign output without background mixing — less preferred; already simulated, so geometry/campaign version is fixed. Every campaign version that has the class at that energy is reported, with the version named.

5.1 Energies with a complete background recipe

(a) There is no FULL simulation at any 9 GeV-electron energy. Checked in all three stores: /volatile/eic/EPIC/FULL, /mss/eic/EPIC/FULL (tape) and Rucio epic:/FULL/*. Zero directories or DIDs for 9x100, 9x130, 9x275. What FULL does have at our energies:

Energy	`FULL` content	Campaign versions
5x41	`DIS/NC` (`minQ2=1`, `=10`, `=100`)	23.05.2 (arches, brycecanyon), 23.06.0/1, 23.07.1/2, 23.08.0, 23.09.1, 23.11.0, 23.12.0, 25.12.0, 26.02.0, `main`
10x100	`DIS/NC` (`minQ2=1`, `=10`, `=100`); `EXCLUSIVE/UCHANNEL_RHO` (23.08.0 only)	23.03.0 … 23.12.0, 25.12.0, 26.02.0, 26.07.1 (bkg-mixed), `main`
10x275	`DIS/NC` (`minQ2=1`)	25.12.0, 26.02.0, 26.07.1 (bkg-mixed)
9x100, 9x130, 9x275, 10x130	nothing	—

Recent campaigns keep no signal FULL at all — FULL/26.04.1, 26.07.0 and 26.07.1 contain only Bkg_Exact1S_2us, BACKGROUNDS and SINGLE. The signal ddsim output is not retained; only the reconstructed RECO partition is.

So the FULL fallback is only available for ncdisq1/10/100 at 5x41, 10x100 and 10x275, from campaigns 23.x–26.02.0. RECO, however, does cover 9x — see section 5.6, which is the practical route for the 9 GeV points.

(b) Per-source background inputs are complete at 10x100, 10x130, 10x275 and 18x275.

The campaign ships a complete mixing recipe per energy in simulation_campaign_datasets/config_data/synrad_egasbrems_egascoulomb_egastouschek_hgas_<E>_<pipe>_10000Ahr_machineruntime_<T>.json. Each entry is {file, freq, skip, status} — the EVGEN background file, its rate in kHz, and the generatorStatus code to stamp on its particles. The five status codes are exactly Marco's slide-10 set: 2000 SynRad, 3000 electron-beam-gas (incl. brems), 4000 Coulomb, 5000 Touschek, 6000 proton-beam-gas. The hgas_-prefixed configs are the ones to use — the non-hgas_ variants drop the proton-beam-gas term.

Checked every file these configs reference against /volatile. The configs contain 120 file entries, but 99 of those are one per-run series (ProtonBeamGas_275GeV_run001…run100, used only by the 18x275 10 ms config), so there are 21 distinct input files — 18 present, 3 missing.

Energy	pipe variants	all five inputs on disk?
10x100	vacuum, gold 10 µm	yes
10x130	vacuum, gold 10 µm	yes
10x275	vacuum, gold 10 µm	yes
18x275	vacuum (50 s and 10 ms runtimes)	yes
5x100	vacuum	no — missing `BEAMGAS/electron/GETaLM1.0.0-1.1/5GeV/…` and `SYNRAD/dataprod_rel_1.0.0/5x100/…preproc_10000repeats…`
5x41	vacuum	no — the two above plus `BEAMGAS/proton/pythia8.306-1.0/41GeV/combined/…`
9x100, 9x130, 9x275	—	no config exists at all

So four energies are mixable today: 10x100, 10x130, 10x275 (each with a vacuum and a gold-coated 10 µm variant) and 18x275 (vacuum). 5x41 and 5x100 are blocked by three absent files, not by a missing recipe.

The campaign already substitutes electron-ring samples across hadron energies. This is the important precedent for 9 GeV. At 10x100 and 10x130 the SynRad, Coulomb and Touschek inputs are literally the 10x275 files, reused unchanged, with only freq adjusted per energy; only the electron-beam-gas (keyed on electron energy, 10GeV) and proton-beam-gas (keyed on proton energy, 100GeV / 130GeV / 275GeV) terms are native. The same was intended at 5x41, where SynRad is taken from 5x100.

That means a 9 GeV mixing config is a rate-table problem, not a sample-production problem: the 10 GeV ESR samples can be reused for 9x275 / 9x130 / 9x100 exactly as they already are for 10x100 and 10x130, as soon as someone supplies the 9 GeV freq numbers. The proton-side terms already exist natively at 275 GeV, 130 GeV and 100 GeV.

Mixing spec per energy and source

native? says whether the input sample belongs to that energy or is borrowed from another. rate is the sum over entries with that status code; ×N marks a status that appears as N separate entries in the config. The 18x275 10 ms config lists proton-beam-gas as 100 separate ~256 MB run0NN files at 0.225 kHz each — summing to 22.5 kHz, the same total as the 50 s config's single combined entry, so it looks like a way to avoid one 22 GiB input rather than an error. Not verified beyond that arithmetic.

energy	pipe	runtime	status	source	rate [kHz]	input sample	native?	on disk	size
10x100	gold 10 µm	50s	2000	SynRad	13,277,000.00	`10x275`	substituted	yes	78.0 MiB
10x100	gold 10 µm	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x100	gold 10 µm	50s	4000	e Coulomb	29.56	`10x275`	substituted	yes	94.1 MiB
10x100	gold 10 µm	50s	5000	e Touschek	233.50	`10x275`	substituted	yes	525.1 MiB
10x100	gold 10 µm	50s	6000	p-beam-gas	31.90	`100GeV`	native	yes	16.7 GiB
10x100	vacuum	50s	2000	SynRad	36,608,000.00	`10x275`	substituted	yes	287.0 MiB
10x100	vacuum	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x100	vacuum	50s	4000	e Coulomb	29.56	`10x275`	substituted	yes	94.1 MiB
10x100	vacuum	50s	5000	e Touschek	233.50	`10x275`	substituted	yes	525.1 MiB
10x100	vacuum	50s	6000	p-beam-gas	31.90	`100GeV`	native	yes	16.7 GiB
10x130	gold 10 µm	50s	2000	SynRad	13,277,000.00	`10x275`	substituted	yes	78.0 MiB
10x130	gold 10 µm	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x130	gold 10 µm	50s	4000	e Coulomb	29.56	`10x275`	substituted	yes	94.1 MiB
10x130	gold 10 µm	50s	5000	e Touschek	233.50	`10x275`	substituted	yes	525.1 MiB
10x130	gold 10 µm	50s	6000	p-beam-gas	31.97	`130GeV`	native	yes	1.8 GiB
10x130	vacuum	50s	2000	SynRad	36,608,000.00	`10x275`	substituted	yes	287.0 MiB
10x130	vacuum	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x130	vacuum	50s	4000	e Coulomb	29.56	`10x275`	substituted	yes	94.1 MiB
10x130	vacuum	50s	5000	e Touschek	233.50	`10x275`	substituted	yes	525.1 MiB
10x130	vacuum	50s	6000	p-beam-gas	31.97	`130GeV`	native	yes	1.8 GiB
10x275	gold 10 µm	50s	2000	SynRad	13,277,000.00	`10x275`	native	yes	78.0 MiB
10x275	gold 10 µm	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x275	gold 10 µm	50s	4000	e Coulomb	29.56	`10x275`	native	yes	94.1 MiB
10x275	gold 10 µm	50s	5000	e Touschek	233.50	`10x275`	native	yes	525.1 MiB
10x275	gold 10 µm	50s	6000	p-beam-gas	32.60	`275GeV`	native	yes	21.0 GiB
10x275	vacuum	50s	2000	SynRad	36,608,000.00	`10x275`	native	yes	287.0 MiB
10x275	vacuum	50s	3000	e-beam-gas + brems	3,177.25	`10GeV`	native	yes	6.2 GiB
10x275	vacuum	50s	4000	e Coulomb	29.56	`10x275`	native	yes	94.1 MiB
10x275	vacuum	50s	5000	e Touschek	233.50	`10x275`	native	yes	525.1 MiB
10x275	vacuum	50s	6000	p-beam-gas	32.60	`275GeV`	native	yes	21.0 GiB
18x275	vacuum	10ms	2000	SynRad	14,000.00	`18x275`	native	yes	194.9 MiB
18x275	vacuum	10ms	3000	e-beam-gas + brems	316.94	`18GeV`	native	yes	3.8 GiB
18x275	vacuum	10ms	4000	e Coulomb	1.30	`18x275`	native	yes	674.3 KiB
18x275	vacuum	10ms	5000	e Touschek	0.72	`18x275`	native	yes	314.5 KiB
18x275	vacuum	10ms	6000	p-beam-gas ×100	22.50	`275GeV`	native	yes	244.1 MiB
18x275	vacuum	50s	2000	SynRad	3,324,000.00	`18x275`	native	yes	286.8 MiB
18x275	vacuum	50s	3000	e-beam-gas + brems	316.94	`18GeV`	native	yes	3.8 GiB
18x275	vacuum	50s	4000	e Coulomb	0.86	`18x275`	native	yes	3.5 MiB
18x275	vacuum	50s	5000	e Touschek	0.55	`18x275`	native	yes	1.7 MiB
18x275	vacuum	50s	6000	p-beam-gas	22.50	`275GeV`	native	yes	21.0 GiB
5x100	vacuum	50s	2000	SynRad	36,608,000.00	`5x100`	native	MISSING	—
5x100	vacuum	50s	3000	e-beam-gas + brems	2,826.38	`5GeV`	native	MISSING	—
5x100	vacuum	50s	4000	e Coulomb	116.57	`5x100`	native	yes	340.9 MiB
5x100	vacuum	50s	5000	e Touschek	1,112.30	`5x100`	native	yes	2.6 GiB
5x100	vacuum	50s	6000	p-beam-gas	22.00	`100GeV`	native	yes	16.7 GiB
5x41	vacuum	50s	2000	SynRad	36,608,000.00	`5x100`	substituted	MISSING	—
5x41	vacuum	50s	3000	e-beam-gas + brems	2,181.97	`5GeV`	native	MISSING	—
5x41	vacuum	50s	4000	e Coulomb	116.57	`5x100`	substituted	yes	340.9 MiB
5x41	vacuum	50s	5000	e Touschek	1,112.30	`5x100`	substituted	yes	2.6 GiB
5x41	vacuum	50s	6000	p-beam-gas	12.20	`41GeV`	native	MISSING	—

Expected events per 2 µs frame

This is the form that matters for occupancy: freq (kHz) × 10³ × 2×10⁻⁶ s.

config	SynRad	e-gas+brems	Coulomb	Touschek	p-gas	total / frame
10x100 vacuum	73,216	6.35	0.06	0.47	0.06	73,223
10x100 gold 10 µm	26,554	6.35	0.06	0.47	0.06	26,561
10x130 vacuum	73,216	6.35	0.06	0.47	0.06	73,223
10x130 gold 10 µm	26,554	6.35	0.06	0.47	0.06	26,561
10x275 vacuum	73,216	6.35	0.06	0.47	0.07	73,223
10x275 gold 10 µm	26,554	6.35	0.06	0.47	0.07	26,561
5x41 vacuum	73,216	4.36	0.23	2.22	0.02	73,223
5x100 vacuum	73,216	5.65	0.23	2.22	0.04	73,224
18x275 vacuum, 50 s	6,648	0.63	0.00	0.00	0.04	6,649
18x275 vacuum, 10 ms	28	0.63	0.00	0.00	0.05	29

Two things this makes obvious that the kHz table hides.

There is no smooth energy dependence — there are three regimes. SynRad is the same number (36.608 GHz vacuum / 13.277 GHz gold) at 10x100, 10x130, 10x275, 5x41 and 5x100, because the configs reuse one SynRad sample with its rate unchanged. Only 18x275 differs. So the available background regimes are vacuum ≈ 73k events/frame, gold ≈ 26.5k, and 18x275 ≈ 6.6k — plus the 18x275 10 ms config at 29/frame, which is a different beast and is not interchangeable with its 50 s sibling.

Occupancy is SynRad and nothing else. Every other source is under 7 events per frame; Coulomb and Touschek under 2.3. Their rate ratios across energies are dramatic but irrelevant to hit counts. The significance of those four is qualitative: SynRad is 1–2 soft photons per event, so it is hit-level noise with no tracks, whereas Coulomb carries a full DIS final state with hadronisation and proton beam gas produces many secondary tracks (section 2). Those rare sources are the track-like background a tracking-based event builder has to reject. Note 5x41 has ~4× the Coulomb and Touschek of the 10 GeV points at the same total occupancy, making it the track-background-richest configuration available.

Beampipe coating: rates and campaign history

vacuum is the bare beampipe; GoldCoating/10um is a 10 µm gold layer on the inside. Synchrotron photons striking the wall are either absorbed or scattered onward into the detector acceptance; high-Z gold absorbs the soft X-rays that the bare low-Z wall would pass along. That is the standard rationale rather than something verified here, but the numbers agree: 2.76× less SynRad, with the other four sources byte-identical. The input files show it too — the bare …synrad_10x275_run001.preproc_10000repeats file is 287 MiB and the GoldCoating/10um/ file of the same name is 78 MiB. That 78 MB is the figure in Marco's slide-10 "File size" column, so his table quotes the gold-coated file.

The campaign's own choice over time:

campaign	coating used for mixed production
25.06.1	vacuum (`Synrad_18GeV_Vac_10000Ahr_…`), 18x275 only
25.12.0, 26.02.0	gold, both 5 µm and 10 µm
26.03.0, 26.04.1, 26.07.0, 26.07.1	gold, 10 µm only

Vacuum, then gold at two thicknesses, then gold 10 µm for four consecutive campaigns. That reads as the coating being adopted and the thickness settled, so gold 10 µm is the configuration to train on and quote results for, not one of two live options. The engineering decision itself is not something this survey can confirm — but nothing in the data contradicts it.

What that leaves us:

Vacuum becomes a tool, not a scenario. 2.76× more SynRad with everything else identical, at all three mixable 10x energies — a free occupancy stress test and the high end of a rate-randomisation range. It should not be reported as an expected operating point.
5 µm is a superseded intermediate, present in FULL/RECO for 25.12.0 and 26.02.0 but with no shipped rate config, so not mixable without writing our own rate table.
There is no 18x275 gold SynRad sample. So the two realistic points are gold-at-10x (26.5k/frame) and vacuum-at-18x275 (6.6k/frame), and the latter is vacuum only because nobody produced the gold version. If an 18x275 gold configuration matters, that is a specific and small thing to request.

5.2 Process catalogue

This is the table to work from. Rows are physics processes as the data defines them, not Marco's class slugs; his slug is a cross-reference column (— means the process has no row in his table at all). Energies listed are those with an afterburned sample; (h) marks an energy where only head-on data exists, so abconv would have to be run first.

206 branches were measured; 190 are physics (the other 16 are the background sources of section 2). Mixable energies are in bold. † marks an energy taken from the filename because the beams could not be parsed (erhic-cc-yrad-floff_eD_18_137_q2_1_evt.hepmc3 → 18x137, eD per-nucleon).

process	system	generator(s)	energies with afterburned data	Marco slug	branches	files	size
CC DIS, Q² threshold	ep	Pythia8 (campaign)	5x41, 5x100, 10x100, 10x275, 18x275	`ccdisq100/1000`	5	90	639.6 GiB
CC DIS, radiative (eRHIC fmt)	eD	DJANGOH/eRHIC	18x137(h)†	`ccdisq*`	1	42	92.2 GiB
CC DIS, radiative (eRHIC fmt)	ep	DJANGOH/eRHIC	18x275(h)	`ccdisq*`	1	35	97.8 GiB
CC DIS, radiative, polarised	ep	DJANGOH 4.6.10	9x275	`ccdisq* (diff. generator)`	2	6	3.0 GiB
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	ep	DEMPgen	5x41, 5x100, 9x130, 9x275, 10x100, 10x130, 18x275	`dempq3to10/10to20/20to35`	13	840	375.1 GiB
DVCS	ep	DVCS gen (legacy)	5x41, 10x100, 18x275	`dvcs`	6	65	32.9 GiB
DVCS + BH + INT	ep	EpIC 1.1.6	5x41, 10x100, 10x130, 18x275	`dvcs`	6	10	3.5 GiB
DVCS / BH / interference	ep	EpIC 1.1.8	9x130, 9x275	`dvcs`	6	24	11.3 GiB
DVCS on deuteron	eD	EpIC	5x41, 10x100	`— (eD)`	2	26	23.9 GiB
DVMP / Υ (lAger legacy)	ep	lAger	10x100	`dvmp/upsilon`	1	8	1.7 GiB
DVMP J/ψ→μ⁺μ⁻	ep	lAger	9x130, 9x275, 10x130, 18x275	`dvmp`	4	4	3.2 GiB
DVMP π⁰	ep	EpIC	5x41, 9x130, 9x275, 10x100, 10x130, 18x275	`dvmp`	7	8	2.2 GiB
NC DIS, Q² threshold	ep	Pythia8 (campaign)	5x41, 5x100, 10x100, 10x275, 18x275	`ncdisq1/10/100/1000`	8	328	2.3 TiB
NC DIS, inclusive	ep	Pythia6.428, Pythia8.316	5x41, 5x100, 9x130, 9x275, 10x100, 10x130, 10x275, 18x275	`ncdisq1/10/100/1000`	8	175	222.6 GiB
NC DIS, inclusive (EEEMCal cfg)	ep	Pythia8.315	9x275	`ncdisq*`	1	100	1.8 GiB
NC/CC DIS, radiative	ep	DJANGOH 4.6.21	18x275	`ncdisq/ccdisq`	2	8	6.5 GiB
SIDIS / min-bias photoprod.	en	Pythia6-eic	5x41, 10x100, 18x275	`photoprod`	3	715	576.5 GiB
SIDIS / min-bias photoprod.	ep	Pythia6-eic	5x41, 5x100, 9x130, 9x275, 10x100, 10x275, 18x275	`photoprod`	10	1141	1.1 TiB
SIDIS D_s	ep	Pythia8 HFsim	9x275	`—`	1	1200	3.1 GiB
SIDIS D_s (eA)	eAu	BeAGLE HFsim	9x100	`—`	1	500	5.8 GiB
SIDIS dijet (PGF)	ep	Pythia6.428 dijet	18x275	`—`	1	100	972.1 MiB
SIDIS open charm D⁰	ep	Pythia8, Pythia8 HFsim	5x41, 9x130, 9x275, 10x100, 10x130, 18x275	`—`	6	3728	22.7 GiB
SIDIS open charm D⁰ (eA)	eAu	BeAGLE HFsim	5x41, 9x100, 10x100	`—`	5	4900	15.1 GiB
SIDIS Λ	ep	Pythia8.306	18x275	`—`	1	8	10.5 GiB
SIDIS Λ_c	ep	Pythia8 HFsim	9x275, 10x100	`—`	2	3500	3.7 GiB
SIDIS Λ_c (eA)	eAu	BeAGLE HFsim	9x100, 10x100	`—`	3	1498	6.6 GiB
SIDIS, Q²-binned	ep	Pythia6 (IP6 era)	5x41, 10x100, 18x275	`photoprod (older prod.)`	5	5900	3.9 TiB
TCS + BH interference	ep	TCS gen	5x41, 10x100, 18x275	`tcs`	6	860	121.9 GiB
TCS, pure	ep	TCS gen — no afterburned sample anywhere	5x41(h), 5x100(h), 10x100(h), 18x275(h)	`tcs`	4	80	83.1 GiB
coherent J/ψ photoproduction	ep	eSTARlight	18x275	`jpsiphoto`	1	2	679.2 MiB
coherent ρ, eA	eAu	Sartre	18x110(h)	`cohrho`	1	2	23.7 GiB
coherent φ→K⁺K⁻, eA	eAu	Sartre	9x100, 10x100, 18x110	`cohphi`	4	14	61.1 GiB
coherent/incoherent J/ψ, eA	eAu	Sartre	18x110	`— (eA)`	2	8	32.4 GiB
coherent/incoherent ρ, eA	eAu	Sartre	9x100, 10x100, 18x110	`cohrho`	4	2012	48.1 GiB
diffractive DIS	ep	RAPGAP 3.310	9x130, 9x275, 10x100	`ddis`	5	5	14.2 GiB
diffractive J/ψ, eA	eAu	Sartre	18x110(h)	`— (eA)`	1	2	23.6 GiB
diffractive J/ψ, ePb	ePb	BeAGLE	18x108.41	`— (ePb)`	1	54	112.5 GiB
diffractive J/ψ, ep	ep	lAger	5x41, 5x100, 9x130, 9x275, 10x100, 10x130, 18x275	`jpsi`	7	7	437.7 MiB
diffractive φ, eA	eAu	Sartre/BeAGLE, eSTARlight	18x110	`cohphi`	2	3	23.7 GiB
diffractive φ, eA	ep	Sartre/BeAGLE	18x110(h)	`cohphi`	1	200	72.0 GiB
diffractive φ, ePb/eAu	eAu	BeAGLE	9x100, 10x100	`cohphi`	2	200	35.8 GiB
diffractive φ, ePb/eAu	ePb	BeAGLE	18x108.41	`cohphi`	1	54	608.6 GiB
double DVCS	ep	EpIC	10x130, 18x275	`ddvcs`	2	6	2.3 GiB
exclusive ρ→π⁺π⁻	ep	SLIGHT	10x100, unknown(h)	`rho`	3	56	1.5 GiB
inclusive eA DIS	eAu	BeAGLE	5x41, 9x100, 10x100	`—`	5	815	94.3 GiB
inclusive eA DIS	eH2	BeAGLE	9x130, 10x130	`—`	5	221	73.3 GiB
inclusive eA DIS	eHe3	BeAGLE	5x41	`—`	1	3	3.2 GiB
meson structure function K⁺Λ	ep	eicMesonSFGen	5x41, 9x100, 9x130, 9x275, 10x100, 10x130, 18x275	`mesonsf`	7	3800	2.9 GiB
spectroscopy ψ(2S)/X/Y	ep	dedicated	10x100	`spectroscopy`	2	30	2.6 GiB
u-channel π⁰	ep	SLIGHT	18x275	`upi0`	1	1	36.3 MiB
u-channel ρ	ep	SLIGHT/GENEX	10x100	`urho`	1	1	35.6 MiB
u-channel ω (neutral decay)	ep	SLIGHT	18x275(h)	`— (not upi0/omega)`	1	10	250.6 MiB
Υ photoproduction (legacy)	ep	eSTARlight	unknown(h)	`upsilon`	1	3	9.5 MiB
Υ(1S/2S/3S) threshold	ep	eSTARlight 1.3.0	9x130, 9x275	`upsilon`	6	450	14.1 GiB
ω photoproduction	ep	SLIGHT	18x275	`omega`	2	98	1.4 GiB

5.3 Processes with no row in Marco's table

Afterburned, ready to use, and absent from his 27 rows. Ordered by how much there is.

process	system	generator	energies	files	size
diffractive J/ψ, ePb	ePb	BeAGLE	18x108.41	54	112.5 GiB
inclusive eA DIS	eAu	BeAGLE	5x41, 9x100, 10x100	815	94.3 GiB
inclusive eA DIS	eH2	BeAGLE	9x130, 10x130	221	73.3 GiB
coherent/incoherent J/ψ, eA	eAu	Sartre	18x110	8	32.4 GiB
DVCS on deuteron	eD	EpIC	5x41, 10x100	26	23.9 GiB
SIDIS open charm D⁰	ep	Pythia8, Pythia8 HFsim	5x41, 9x130, 9x275, 10x100, 10x130, 18x275	3728	22.7 GiB
SIDIS open charm D⁰ (eA)	eAu	BeAGLE HFsim	5x41, 9x100, 10x100	4900	15.1 GiB
SIDIS Λ	ep	Pythia8.306	18x275	8	10.5 GiB
SIDIS Λ_c (eA)	eAu	BeAGLE HFsim	9x100, 10x100	1498	6.6 GiB
SIDIS D_s (eA)	eAu	BeAGLE HFsim	9x100	500	5.8 GiB
SIDIS Λ_c	ep	Pythia8 HFsim	9x275, 10x100	3500	3.7 GiB
inclusive eA DIS	eHe3	BeAGLE	5x41	3	3.2 GiB
SIDIS D_s	ep	Pythia8 HFsim	9x275	1200	3.1 GiB
SIDIS dijet (PGF)	ep	Pythia6.428 dijet	18x275	100	972.1 MiB

5.4 Per-energy detail

events/file is the entry count of the largest file in the branch. Beam angle is mean ± rms over 20 events, with the beam energy the file itself reports.

10x100 — mixable (background recipe complete)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, Q² threshold	`DIS/CC`	Pythia8 (campaign)	ep	20	133.5 GiB	—	0.162+-0.078@E10.0 / 25.030+-0.226@E100.0	not recorded	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.1.0`	DEMPgen	ep	390	3.2 GiB	29,342	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.0`	DEMPgen	ep	60	93.1 GiB	5,451,937	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.2`	DEMPgen	ep	3	287.5 MiB	457,311	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
DVCS	`EXCLUSIVE/DVCS`	DVCS gen (legacy)	ep	1	823.1 MiB	—	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
DVCS	`EXCLUSIVE/DVCS_ABCONV`	DVCS gen (legacy)	ep	12	5.8 GiB	—	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	EpIC 1.1.6	ep	2	843.6 MiB	1,300,000	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
DVCS on deuteron	`EXCLUSIVE/DVCSD_ABCONV`	EpIC	eD	14	11.8 GiB	—	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
DVMP / Υ (lAger legacy)	`EXCLUSIVE/DVMP/lAger`	lAger	ep	8	1.7 GiB	—	0.000+-0.000@E10.0 / 25.000+-0.000@E100.0	not recorded	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.0.0-1.0`	EpIC	ep	1	24.8 MiB	100,000	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.0.0-1.1`	EpIC	ep	1	48.0 MiB	100,000	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	5	139.7 GiB	1,000,000	0.145+-0.089@E10.0 / 24.995+-0.243@E100.0	z 32 mm, t 34.02 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	62	378.7 GiB	—	0.145+-0.089@E10.0 / 24.995+-0.243@E100.0	not recorded	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	27.6 GiB	1,000,000	0.142+-0.064@E10.0 / 25.004+-0.247@E100.0	z 39 mm, t 39.99 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.0.0`	Pythia6-eic	ep	50	60.2 GiB	1,000,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.1.0/en_noradcor`	Pythia6-eic	en	278	216.6 GiB	2,000,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.2.0/ep_noradcor`	Pythia6-eic	ep	80	85.6 GiB	2,000,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/pythia8.306-1.1`	Pythia8	ep	20	5.5 GiB	99,031	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰ (eA)	`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-1.0/eAu`	BeAGLE HFsim	eAu	998	1.9 GiB	723	0.143+-0.066@E10.0 / 24.992+-0.183@E100.8	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰ (eA)	`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-2.0/eAu`	BeAGLE HFsim	eAu	998	1.9 GiB	723	0.143+-0.066@E10.0 / 24.992+-0.183@E100.8	z 31 mm, t 33.41 ns	—
SIDIS Λ_c	`SIDIS/Lc_ABCONV/HFsim-PYTHIA/pythia8.306-1.2/ep`	Pythia8 HFsim	ep	2500	1.7 GiB	303	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
SIDIS Λ_c (eA)	`SIDIS/Lc_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-1.0/eAu`	BeAGLE HFsim	eAu	499	1.3 GiB	1,321	0.143+-0.066@E10.0 / 24.964+-0.106@E100.8	z 31 mm, t 33.41 ns	—
SIDIS Λ_c (eA)	`SIDIS/Lc_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-2.0/eAu`	BeAGLE HFsim	eAu	499	1.3 GiB	1,321	0.143+-0.066@E10.0 / 24.964+-0.106@E100.8	z 31 mm, t 33.41 ns	—
SIDIS, Q²-binned	`SIDIS/pythia6/ep_10x100/hepmc_ip6/noradcor`	Pythia6 (IP6 era)	ep	240	404.2 GiB	—	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
TCS + BH interference	`EXCLUSIVE/TCS_ABCONV`	TCS gen	ep	320	23.9 GiB	—	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
TCS, pure	`EXCLUSIVE/TCS`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
coherent φ→K⁺K⁻, eA	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/sartre1.39-1.0/eAu/coherent/bsat`	Sartre	eAu	1	2.4 GiB	6,894,738	0.143+-0.066@E10.0 / 25.013+-0.185@E100.0	z 31 mm, t 33.41 ns	—
coherent/incoherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/sartre1.39-1.1/eAu/coherent/bsat`	Sartre	eAu	1	3.5 GiB	10,000,000	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
diffractive DIS	`DDIS/rapgap3.310-1.0/noRad/ep`	RAPGAP 3.310	ep	1	5.4 GiB	5,000,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	74.6 MiB	174,056	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
diffractive φ, ePb/eAu	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/BeAGLE1.03.02-1.1/eAu`	BeAGLE	eAu	100	1.2 GiB	20,000	0.151+-0.075@E10.0 / 25.010+-0.156@E100.9	z 31 mm, t 33.41 ns	—
exclusive ρ→π⁺π⁻	`EXCLUSIVE/RHO`	SLIGHT	ep	2	177.6 MiB	—	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
exclusive ρ→π⁺π⁻	`EXCLUSIVE/RHO_ABCONV`	SLIGHT	ep	52	719.0 MiB	—	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.0/eAu`	BeAGLE	eAu	200	12.6 GiB	50,000	0.143+-0.066@E10.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-2.0/eAu`	BeAGLE	eAu	200	12.6 GiB	50,000	0.143+-0.066@E10.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	200	425.8 MiB	5,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—
spectroscopy ψ(2S)/X/Y	`EXCLUSIVE/SPECTROSCOPY`	dedicated	ep	6	339.9 MiB	—	0.000+-0.000@E10.0 / 0.000+-0.000@E100.0	head-on	—
spectroscopy ψ(2S)/X/Y	`EXCLUSIVE/SPECTROSCOPY_ABCONV`	dedicated	ep	24	2.2 GiB	—	0.151+-0.075@E10.0 / 25.010+-0.156@E100.0	z 31 mm, t 33.41 ns	—
u-channel ρ	`EXCLUSIVE/UCHANNEL_RHO`	SLIGHT/GENEX	ep	1	35.6 MiB	100,000	0.185+-0.092@E10.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—

10x130 — mixable (background recipe complete)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.3`	DEMPgen	ep	9	264.4 MiB	101,871	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.4`	DEMPgen	ep	3	331.3 MiB	400,309	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.0`	EpIC 1.1.6	ep	1	324.9 MiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.1`	EpIC 1.1.6	ep	1	324.9 MiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	EpIC 1.1.6	ep	2	843.7 MiB	1,300,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVMP J/ψ→μ⁺μ⁻	`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	1.1 GiB	2,732,981	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.1.6-1.0/unpolarised`	EpIC	ep	1	374.3 MiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
NC DIS, inclusive	`DIS/pythia6.428-1.0/NC/noRad/ep`	Pythia6.428	ep	40	25.0 GiB	500,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.312-2.0/ep`	Pythia8 HFsim	ep	700	2.7 GiB	1,531	0.185+-0.092@E10.0 / 25.012+-0.190@E130.0	z 31 mm, t 33.41 ns	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	85.3 MiB	199,430	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
double DVCS	`EXCLUSIVE/DDVCS_ABCONV/EpIC1.1.6-1.1`	EpIC	ep	2	774.1 MiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.0/eH2`	BeAGLE	eH2	1	1.3 GiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.3	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.3/eH2/en`	BeAGLE	eH2	10	12.8 GiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.3	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.3/eH2/ep`	BeAGLE	eH2	10	12.8 GiB	1,000,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.1	z 31 mm, t 33.41 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	200	425.5 MiB	5,000	0.151+-0.075@E10.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—

10x275 — mixable (background recipe complete)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, Q² threshold	`DIS/CC`	Pythia8 (campaign)	ep	20	165.7 GiB	—	0.232+-0.140@E10.0 / 25.003+-0.124@E275.0	not recorded	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	5	171.3 GiB	1,000,000	0.228+-0.100@E10.0 / 25.023+-0.092@E275.0	z 25 mm, t 25.31 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	60	441.3 GiB	—	0.228+-0.100@E10.0 / 25.023+-0.092@E275.0	not recorded	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	31.5 GiB	1,000,000	0.213+-0.156@E10.0 / 25.026+-0.133@E275.0	z 25 mm, t 28.22 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.0.0`	Pythia6-eic	ep	50	75.5 GiB	1,000,000	0.268+-0.134@E10.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—

18x275 — mixable (background recipe complete)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, Q² threshold	`DIS/CC`	Pythia8 (campaign)	ep	20	184.7 GiB	—	0.284+-0.129@E18.0 / 24.957+-0.133@E274.9	not recorded	—
CC DIS, radiative (eRHIC fmt)	`DIS/Djangoh/ep`	DJANGOH/eRHIC	ep	35	97.8 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.0	head-on	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.0`	DEMPgen	ep	60	37.6 GiB	2,271,515	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.2`	DEMPgen	ep	3	260.8 MiB	452,229	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
DVCS	`EXCLUSIVE/DVCS`	DVCS gen (legacy)	ep	1	823.3 MiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.0	head-on	—
DVCS	`EXCLUSIVE/DVCS_ABCONV`	DVCS gen (legacy)	ep	12	5.8 GiB	—	0.126+-0.058@E18.0 / 25.002+-0.055@E275.1	z 26 mm, t 29.28 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	EpIC 1.1.6	ep	2	648.0 MiB	1,000,000	0.126+-0.058@E18.0 / 25.002+-0.055@E275.1	z 26 mm, t 29.28 ns	—
DVMP J/ψ→μ⁺μ⁻	`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	2.1 GiB	5,329,042	0.126+-0.058@E18.0 / 25.002+-0.055@E275.1	z 26 mm, t 29.28 ns	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.1.6-1.1/unpolarised`	EpIC	ep	1	375.1 MiB	1,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	70	515.9 GiB	—	0.274+-0.144@E18.0 / 24.980+-0.126@E275.0	not recorded	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	33.9 GiB	1,000,000	0.247+-0.132@E18.0 / 24.980+-0.102@E275.0	z 40 mm, t 37.22 ns	—
NC/CC DIS, radiative	`DIS/DJANGOH4.6.21-1.0/NC/Rad/ep`	DJANGOH 4.6.21	ep	4	3.2 GiB	493,566	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
NC/CC DIS, radiative	`DIS/DJANGOH4.6.21-1.0/NC/noRad/ep`	DJANGOH 4.6.21	ep	4	3.3 GiB	499,611	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.0.0`	Pythia6-eic	ep	50	69.9 GiB	1,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.1.0/en_noradcor`	Pythia6-eic	en	258	183.9 GiB	1,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.2.0/ep_noradcor`	Pythia6-eic	ep	320	311.3 GiB	2,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS dijet (PGF)	`SIDIS/DIJET_ABCONV/pythia6.428-dijet-v1.0/PGF/noRC/ep`	Pythia6.428 dijet	ep	100	972.1 MiB	4,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/pythia8.306-1.0`	Pythia8	ep	8	6.8 GiB	180,161	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS Λ	`SIDIS/Lambda_ABCONV/pythia8.306-1.0`	Pythia8.306	ep	8	10.5 GiB	398,185	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS, Q²-binned	`SIDIS/pythia6/ep_18x275/hepmc_ip6/noradcor`	Pythia6 (IP6 era)	ep	960	1.4 TiB	—	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS, Q²-binned	`SIDIS/pythia6/ep_18x275/hepmc_ip6/radcor`	Pythia6 (IP6 era)	ep	300	506.2 GiB	—	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
SIDIS, Q²-binned	`SIDIS/pythia6/ep_5x41/hepmc_ip6/noradcor` ⚠ dir label says ?	Pythia6 (IP6 era)	ep	740	1.4 TiB	2,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.0	head-on	—
TCS + BH interference	`EXCLUSIVE/TCS_ABCONV`	TCS gen	ep	240	17.9 GiB	—	0.126+-0.058@E18.0 / 25.002+-0.055@E275.1	z 26 mm, t 29.28 ns	—
TCS, pure	`EXCLUSIVE/TCS`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.0	head-on	—
coherent J/ψ photoproduction	`EXCLUSIVE/PHOTOPRODUCTION_JPSI_ABCONV/coherent_ep/eSTARlight1.3.0-1.0`	eSTARlight	ep	2	679.2 MiB	1,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.2	z 26 mm, t 29.28 ns	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	161.6 MiB	379,402	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
double DVCS	`EXCLUSIVE/DDVCS_ABCONV/EpIC1.1.6-1.0`	EpIC	ep	4	1.5 GiB	1,000,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	200	430.2 MiB	5,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.1	z 26 mm, t 29.28 ns	—
u-channel π⁰	`EXCLUSIVE/UCHANNEL_PI0`	SLIGHT	ep	1	36.3 MiB	100,000	0.286+-0.131@E18.0 / 25.008+-0.127@E275.2	z 26 mm, t 29.28 ns	—
u-channel ω (neutral decay)	`EXCLUSIVE/omega`	SLIGHT	ep	10	250.6 MiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.1	head-on	—
ω photoproduction	`EXCLUSIVE/OMEGA`	SLIGHT	ep	2	201.8 MiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E275.1	head-on	—
ω photoproduction	`EXCLUSIVE/OMEGA_ABCONV`	SLIGHT	ep	96	1.2 GiB	—	0.126+-0.058@E18.0 / 25.002+-0.055@E275.2	z 26 mm, t 29.28 ns	—

9x100 — no background recipe yet

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
SIDIS D_s (eA)	`SIDIS/Ds_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	BeAGLE HFsim	eAu	500	5.8 GiB	1,358	0.143+-0.066@E9.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰ (eA)	`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	BeAGLE HFsim	eAu	1000	8.5 GiB	1,026	0.143+-0.066@E9.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
SIDIS Λ_c (eA)	`SIDIS/Lc_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	BeAGLE HFsim	eAu	500	4.1 GiB	937	0.143+-0.066@E9.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
coherent φ→K⁺K⁻, eA	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/sartre1.39-1.3/eAu/coherent/bsat`	Sartre	eAu	1	3.5 GiB	10,000,000	0.143+-0.066@E9.0 / 25.013+-0.185@E100.0	z 31 mm, t 33.41 ns	—
coherent/incoherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/sartre1.39-1.4/eAu/coherent/bsat`	Sartre	eAu	1	3.5 GiB	10,000,000	0.143+-0.066@E9.0 / 25.013+-0.185@E100.0	z 31 mm, t 33.41 ns	—
diffractive φ, ePb/eAu	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/BeAGLE1.03.02-3.1/eAu`	BeAGLE	eAu	100	34.6 GiB	50,000	0.143+-0.066@E9.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-3.1/eAu`	BeAGLE	eAu	15	49.2 GiB	397,054	0.143+-0.066@E9.0 / 25.013+-0.185@E100.9	z 31 mm, t 33.41 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	1000	408.9 MiB	1,000	0.185+-0.092@E9.0 / 25.012+-0.190@E100.0	z 31 mm, t 33.41 ns	—

9x130 — no background recipe yet

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.5`	DEMPgen	ep	3	296.2 MiB	400,078	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/BH_ONLY`	EpIC 1.1.8	ep	2	551.4 MiB	1,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_BH`	EpIC 1.1.8	ep	8	2.2 GiB	1,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_ONLY`	EpIC 1.1.8	ep	2	554.3 MiB	1,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVMP J/ψ→μ⁺μ⁻	`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.1`	lAger	ep	1	8.8 MiB	22,458	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.1.8-1.0/unpolarised`	EpIC	ep	1	339.9 MiB	1,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	28.2 GiB	1,000,000	0.160+-0.088@E9.0 / 25.010+-0.223@E130.0	z 32 mm, t 33.89 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.3.0/ep_noradcor`	Pythia6-eic	ep	80	61.3 GiB	2,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.316-1.0/ep`	Pythia8 HFsim	ep	1000	3.0 GiB	1,186	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
diffractive DIS	`DDIS/rapgap3.310-2.0/ewRad/ep`	RAPGAP 3.310	ep	1	2.1 GiB	2,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
diffractive DIS	`DDIS/rapgap3.310-2.0/noRad/ep`	RAPGAP 3.310	ep	1	2.1 GiB	2,000,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.7.0-1.0`	lAger	ep	1	7.7 MiB	19,034	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-3.1/eH2/en`	BeAGLE	eH2	100	23.3 GiB	100,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.3	z 31 mm, t 33.41 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-3.1/eH2/ep`	BeAGLE	eH2	100	23.1 GiB	100,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.1	z 31 mm, t 33.41 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	1000	408.6 MiB	1,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.0	z 31 mm, t 33.41 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon1S`	eSTARlight 1.3.0	ep	50	1.6 GiB	100,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.1	z 31 mm, t 33.41 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon2S`	eSTARlight 1.3.0	ep	50	1.6 GiB	100,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.1	z 31 mm, t 33.41 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon3S`	eSTARlight 1.3.0	ep	50	1.6 GiB	100,000	0.151+-0.075@E9.0 / 25.009+-0.156@E130.1	z 31 mm, t 33.41 ns	—

9x275 — no background recipe yet

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, radiative, polarised	`DIS/DJANGOH4.6.10-2.0/CC/Rad/eMinus-pMinus`	DJANGOH 4.6.10	ep	3	1.5 GiB	999,955	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
CC DIS, radiative, polarised	`DIS/DJANGOH4.6.10-2.0/CC/Rad/eMinus-pPlus`	DJANGOH 4.6.10	ep	3	1.5 GiB	999,949	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.5`	DEMPgen	ep	3	295.7 MiB	400,770	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/BH_ONLY`	EpIC 1.1.8	ep	2	1.3 GiB	2,500,000	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_BH`	EpIC 1.1.8	ep	8	5.4 GiB	2,500,000	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
DVCS / BH / interference	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_ONLY`	EpIC 1.1.8	ep	2	1.3 GiB	2,500,000	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
DVMP J/ψ→μ⁺μ⁻	`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.1`	lAger	ep	1	31.2 MiB	80,058	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.1.8-1.0/unpolarised`	EpIC	ep	2	679.7 MiB	1,000,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	31.1 GiB	999,999	0.234+-0.134@E9.0 / 24.995+-0.104@E275.0	z 28 mm, t 25.74 ns	—
NC DIS, inclusive (EEEMCal cfg)	`DIS/pythia8.315-1.2/EEEMCal/NC/noRad/ep`	Pythia8.315	ep	100	1.8 GiB	10,000	0.228+-0.117@E9.0 / 24.988+-0.120@E275.0	z 24 mm, t 25.11 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.3.0/ep_noradcor`	Pythia6-eic	ep	81	68.9 GiB	2,000,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
SIDIS D_s	`SIDIS/Ds_ABCONV/HFsim-PYTHIA/pythia8.316-1.1/ep`	Pythia8 HFsim	ep	1200	3.1 GiB	897	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.316-1.0/ep`	Pythia8 HFsim	ep	1000	2.8 GiB	1,026	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
SIDIS Λ_c	`SIDIS/Lc_ABCONV/HFsim-PYTHIA/pythia8.316-1.0/ep`	Pythia8 HFsim	ep	1000	2.0 GiB	740	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
diffractive DIS	`DDIS/rapgap3.310-2.0/ewRad/ep`	RAPGAP 3.310	ep	1	2.3 GiB	2,000,000	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
diffractive DIS	`DDIS/rapgap3.310-2.0/noRad/ep`	RAPGAP 3.310	ep	1	2.2 GiB	2,000,000	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.7.0-1.0`	lAger	ep	1	28.0 MiB	68,977	0.148+-0.074@E9.0 / 25.002+-0.056@E275.1	z 26 mm, t 28.84 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	1000	410.7 MiB	1,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon1S`	eSTARlight 1.3.0	ep	100	3.1 GiB	100,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon2S`	eSTARlight 1.3.0	ep	100	3.1 GiB	100,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—
Υ(1S/2S/3S) threshold	`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon3S`	eSTARlight 1.3.0	ep	100	3.1 GiB	100,000	0.268+-0.134@E9.0 / 25.006+-0.102@E275.1	z 26 mm, t 28.84 ns	—

5x41 — background recipe blocked (3 files missing)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, Q² threshold	`DIS/CC`	Pythia8 (campaign)	ep	10	45.6 GiB	—	0.156+-0.081@E5.0 / 24.939+-0.257@E41.0	not recorded	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.1.0`	DEMPgen	ep	180	4.2 GiB	78,936	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.0`	DEMPgen	ep	60	171.9 GiB	9,868,034	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.2`	DEMPgen	ep	6	613.6 MiB	553,488	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DVCS	`EXCLUSIVE/DVCS`	DVCS gen (legacy)	ep	3	2.4 GiB	—	0.000+-0.000@E5.0 / 0.000+-0.000@E41.0	head-on	—
DVCS	`EXCLUSIVE/DVCS_ABCONV`	DVCS gen (legacy)	ep	36	17.3 GiB	—	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DVCS + BH + INT	`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	EpIC 1.1.6	ep	2	649.2 MiB	1,000,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DVCS on deuteron	`EXCLUSIVE/DVCSD_ABCONV`	EpIC	eD	12	12.1 GiB	—	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.1.6-1.1/unpolarised`	EpIC	ep	1	372.6 MiB	1,000,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	5	90.6 GiB	999,990	0.149+-0.082@E5.0 / 25.016+-0.186@E41.0	z 32 mm, t 33.32 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	61	240.6 GiB	—	0.149+-0.082@E5.0 / 25.016+-0.186@E41.0	not recorded	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	15	20.2 GiB	999,994	0.136+-0.055@E5.0 / 25.015+-0.217@E41.0	z 45 mm, t 44.07 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.0.0`	Pythia6-eic	ep	50	59.7 GiB	1,000,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.1.0/en_noradcor`	Pythia6-eic	en	179	175.9 GiB	2,000,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.2.0/ep_noradcor`	Pythia6-eic	ep	330	266.2 GiB	2,000,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
SIDIS open charm D⁰	`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.306-1.2/ep`	Pythia8 HFsim	ep	1000	1.9 GiB	887	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
SIDIS open charm D⁰ (eA)	`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-1.0/eAu`	BeAGLE HFsim	eAu	952	1.4 GiB	612	0.163+-0.071@E5.0 / 25.008+-0.248@E41.3	z 52 mm, t 54.43 ns	—
SIDIS open charm D⁰ (eA)	`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-2.0/eAu`	BeAGLE HFsim	eAu	952	1.4 GiB	612	0.163+-0.071@E5.0 / 25.008+-0.248@E41.3	z 52 mm, t 54.43 ns	—
SIDIS, Q²-binned	`SIDIS/pythia6/ep_5x41/hepmc_ip6/radcor`	Pythia6 (IP6 era)	ep	3660	187.4 GiB	—	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E5.0 / 0.000+-0.000@E41.0	head-on	—
TCS + BH interference	`EXCLUSIVE/TCS_ABCONV`	TCS gen	ep	240	17.9 GiB	—	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
TCS, pure	`EXCLUSIVE/TCS`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E5.0 / 0.000+-0.000@E41.0	head-on	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	29.3 MiB	66,327	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.0/eAu`	BeAGLE	eAu	200	9.9 GiB	50,000	0.163+-0.071@E5.0 / 25.018+-0.253@E41.4	z 52 mm, t 54.43 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-1.2/eHe3`	BeAGLE	eHe3	3	3.2 GiB	1,000,000	0.163+-0.071@E5.0 / 25.018+-0.253@E41.2	z 52 mm, t 54.43 ns	—
inclusive eA DIS	`DIS/BeAGLE1.03.02-2.0/eAu`	BeAGLE	eAu	200	10.0 GiB	50,000	0.163+-0.071@E5.0 / 25.018+-0.253@E41.4	z 52 mm, t 54.43 ns	—
meson structure function K⁺Λ	`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	eicMesonSFGen	ep	200	423.7 MiB	5,000	0.170+-0.073@E5.0 / 25.016+-0.197@E41.0	z 33 mm, t 35.69 ns	—

5x100 — background recipe blocked (3 files missing)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, Q² threshold	`DIS/CC`	Pythia8 (campaign)	ep	20	110.1 GiB	—	0.234+-0.092@E5.0 / 25.046+-0.190@E100.1	not recorded	—
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	`EXCLUSIVE/DEMP/DEMPgen-1.2.0`	DEMPgen	ep	60	62.8 GiB	3,663,830	0.236+-0.106@E5.0 / 25.011+-0.179@E100.0	z 31 mm, t 33.41 ns	—
NC DIS, Q² threshold	`DIS/NC`	Pythia8 (campaign)	ep	60	330.5 GiB	—	0.213+-0.090@E5.0 / 24.957+-0.257@E100.0	not recorded	—
NC DIS, inclusive	`DIS/pythia8.316-1.0/NC/noRad/ep`	Pythia8.316	ep	20	25.0 GiB	1,000,000	0.197+-0.109@E5.0 / 25.029+-0.177@E100.0	z 40 mm, t 39.45 ns	—
SIDIS / min-bias photoprod.	`SIDIS/pythia6-eic/1.0.0`	Pythia6-eic	ep	50	64.7 GiB	1,000,000	0.236+-0.106@E5.0 / 25.011+-0.179@E100.0	z 31 mm, t 33.41 ns	—
TCS, pure	`EXCLUSIVE/TCS`	TCS gen	ep	20	20.8 GiB	—	0.000+-0.000@E5.0 / 0.000+-0.000@E100.0	head-on	—
diffractive J/ψ, ep	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	lAger	ep	1	51.0 MiB	117,730	0.236+-0.106@E5.0 / 25.011+-0.179@E100.0	z 31 mm, t 33.41 ns	—

18x108.41 (ePb)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
diffractive J/ψ, ePb	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/BeAGLE/ePb_18x108.41_tau10_B1.1_Jpsi_highstats`	BeAGLE	ePb	54	112.5 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E109.4	z 31 mm, t 33.83 ns	—
diffractive φ, ePb/eAu	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/BeAGLE/ePb_18x108.41_tau10_B1.1_Phi_highstats`	BeAGLE	ePb	54	608.6 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E109.3	z 31 mm, t 33.83 ns	—

18x110 (eAu)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
coherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO/Sartre`	Sartre	eAu	2	23.7 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E110.0	head-on	—
coherent φ→K⁺K⁻, eA	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/Sartre/Coherent`	Sartre	eAu	8	46.1 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
coherent φ→K⁺K⁻, eA	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/Sartre/Incoherent`	Sartre	eAu	4	9.1 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
coherent/incoherent J/ψ, eA	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/Sartre/Coherent`	Sartre	eAu	4	24.8 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
coherent/incoherent J/ψ, eA	`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/Sartre/Incoherent`	Sartre	eAu	4	7.5 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
coherent/incoherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/Sartre/Coherent`	Sartre	eAu	1504	30.7 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
coherent/incoherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/Sartre/Incoherent`	Sartre	eAu	506	10.3 GiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E110.0	z 31 mm, t 33.83 ns	—
diffractive J/ψ, eA	`EXCLUSIVE/DIFFRACTIVE_JPSI/Sartre`	Sartre	eAu	2	23.6 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E110.0	head-on	—
diffractive φ, eA	`EXCLUSIVE/DIFFRACTIVE_PHI/BeAGLE`	Sartre/BeAGLE	ep	200	72.0 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E111.0	head-on	—
diffractive φ, eA	`EXCLUSIVE/DIFFRACTIVE_PHI/Sartre`	Sartre/BeAGLE	eAu	2	23.7 GiB	—	0.000+-0.000@E18.0 / 0.000+-0.000@E110.0	head-on	—
diffractive φ, eA	`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/eSTARLight`	eSTARlight	eAu	1	54.0 MiB	—	0.106+-0.068@E18.0 / 25.014+-0.186@E109.8	z 31 mm, t 33.83 ns	—

Energy not determined (eD, beams did not parse)

process	branch	generator	system	files	size	events/file	e⁻ / hadron beam [mrad @ GeV]	beam effects	tape
CC DIS, radiative (eRHIC fmt)	`DIS/Djangoh/eD`	DJANGOH/eRHIC	eD	42	92.2 GiB	—	? / ?	head-on	—
exclusive ρ→π⁺π⁻	`EXCLUSIVE/rho`	SLIGHT	ep	2	666.2 MiB	—	? / ?	head-on	—
Υ photoproduction (legacy)	`EXCLUSIVE/UPSILON/v1`	eSTARlight	ep	3	9.5 MiB	—	? / 0.536+-0.727@E0.4	head-on	—

5.5 Usability caveats per branch

Head-on branches requiring `abconv`

process	branch	energy	files	size	afterburned alternative at the same energy
CC DIS, radiative (eRHIC fmt)	`DIS/Djangoh/ep`	18x275	35	97.8 GiB	no — only head-on
CC DIS, radiative (eRHIC fmt)	`DIS/Djangoh/eD`	unknown	42	92.2 GiB	no — only head-on
DVCS	`EXCLUSIVE/DVCS`	10x100	1	823.1 MiB	yes
DVCS	`EXCLUSIVE/DVCS`	18x275	1	823.3 MiB	yes
DVCS	`EXCLUSIVE/DVCS`	5x41	3	2.4 GiB	yes
DVMP π⁰	`EXCLUSIVE/DVMP/EpIC1.0.0-1.0`	10x100	1	24.8 MiB	yes
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	10x100	20	20.8 GiB	yes
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	18x275	20	20.8 GiB	yes
TCS + BH interference	`EXCLUSIVE/TCS/TCS_BH_TCS+BH`	5x41	20	20.8 GiB	yes
TCS, pure	`EXCLUSIVE/TCS`	10x100	20	20.8 GiB	no — only head-on
TCS, pure	`EXCLUSIVE/TCS`	18x275	20	20.8 GiB	no — only head-on
TCS, pure	`EXCLUSIVE/TCS`	5x100	20	20.8 GiB	no — only head-on
TCS, pure	`EXCLUSIVE/TCS`	5x41	20	20.8 GiB	no — only head-on
coherent ρ, eA	`EXCLUSIVE/DIFFRACTIVE_RHO/Sartre`	18x110	2	23.7 GiB	no — only head-on
diffractive J/ψ, eA	`EXCLUSIVE/DIFFRACTIVE_JPSI/Sartre`	18x110	2	23.6 GiB	no — only head-on
diffractive φ, eA	`EXCLUSIVE/DIFFRACTIVE_PHI/BeAGLE`	18x110	200	72.0 GiB	no — only head-on
diffractive φ, eA	`EXCLUSIVE/DIFFRACTIVE_PHI/Sartre`	18x110	2	23.7 GiB	yes
exclusive ρ→π⁺π⁻	`EXCLUSIVE/RHO`	10x100	2	177.6 MiB	yes
exclusive ρ→π⁺π⁻	`EXCLUSIVE/rho`	unknown	2	666.2 MiB	no — only head-on
spectroscopy ψ(2S)/X/Y	`EXCLUSIVE/SPECTROSCOPY`	10x100	6	339.9 MiB	yes
u-channel ω (neutral decay)	`EXCLUSIVE/omega`	18x275	10	250.6 MiB	no — only head-on
Υ photoproduction (legacy)	`EXCLUSIVE/UPSILON/v1`	unknown	3	9.5 MiB	no — only head-on
ω photoproduction	`EXCLUSIVE/OMEGA`	18x275	2	201.8 MiB	yes

23 head-on branches; 12 of them have no afterburned equivalent at the same energy, so those are the only ones where running abconv ourselves would actually add coverage:

DIS/Djangoh/ep — CC DIS, radiative (eRHIC fmt), ep, 18x275, 35 files, 97.8 GiB
DIS/Djangoh/eD — CC DIS, radiative (eRHIC fmt), eD, unknown, 42 files, 92.2 GiB
EXCLUSIVE/DIFFRACTIVE_PHI/BeAGLE — diffractive φ, eA, ep, 18x110, 200 files, 72.0 GiB
EXCLUSIVE/DIFFRACTIVE_RHO/Sartre — coherent ρ, eA, eAu, 18x110, 2 files, 23.7 GiB
EXCLUSIVE/DIFFRACTIVE_JPSI/Sartre — diffractive J/ψ, eA, eAu, 18x110, 2 files, 23.6 GiB
EXCLUSIVE/TCS — TCS, pure, ep, 18x275, 20 files, 20.8 GiB
EXCLUSIVE/TCS — TCS, pure, ep, 10x100, 20 files, 20.8 GiB
EXCLUSIVE/TCS — TCS, pure, ep, 5x100, 20 files, 20.8 GiB
EXCLUSIVE/TCS — TCS, pure, ep, 5x41, 20 files, 20.8 GiB
EXCLUSIVE/rho — exclusive ρ→π⁺π⁻, ep, unknown, 2 files, 666.2 MiB
EXCLUSIVE/omega — u-channel ω (neutral decay), ep, 18x275, 10 files, 250.6 MiB
EXCLUSIVE/UPSILON/v1 — Υ photoproduction (legacy), ep, unknown, 3 files, 9.5 MiB

Non-HepMC formats

These hold physics we want but not in a form ddsim reads. Found while sweeping; none were measurable with the HepMC reader.

branch	format	note
`EXCLUSIVE/KLambda/Q2_{3,5}`, `EXCLUSIVE/KSigma/Q2_{3,5}`	LUND `.lund`	DEMPgen legacy, `5on41`; 40 files, 8.6 GiB
`EXCLUSIVE/meson_MC`	LUND `.dat`	K⁺Λ `10on100`; 6 files, 2.5 GiB
`SIDIS/pythia6/ep_5x41/{noradcor,radcor}`	`.hist.root`	histograms only, not events
`SIDIS/pythia6/ep_18x100/ep-18x100`	eic-smear `.root`	not HepMC; 60 files, 112 GiB
`Djangoh/ep`, `DIS/Djangoh/eD`	eic-smear `.root` / `.hepmc3`	eRHIC-era CC DIS; `.hepmc3` beams did not parse — inspect before use
`EXCLUSIVE/UPSILON/v1`	`.hepmc`	records a ~0.4 GeV virtual photon as the beam, with no e/p beam pair — a photoproduction generator convention. Needs the beam pair reconstructed before `abconv`; 3 files, 9.5 MiB
`EXCLUSIVE/DVCS+BH+INT.coherent`	single file literally named `18x275`, no extension	823.3 MiB, excluded from the measurement sweep by the extension filter — not measured, afterburner state unknown

One directory is mislabelled

SIDIS/pythia6/ep_5x41/hepmc_ip6/noradcor is named 5x41 but its files are ep_noradcor.18x275_* and the beams measure 18x275 (0.286 / 25.008 mrad at 18.0 / 275.1 GeV). 740 files, 1.4 TiB. Trust the beams, not the path — this is the one case in 206 branches where the two disagree.

5.6 Generator events recovered from `FULL` / `RECO`

The simulated campaign files keep the generator record, so a background-free FULL or RECO file can be turned back into HepMC3 by reading MCParticles and dropping the Geant4 secondaries. Checked today on two files:

RECO/26.07.1/epic_craterlake/EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_BH/9x275/q2_1_100/emhLm (…emhLm.0000.eicrecon.edm4eic.root, 585 MiB, 6849 events, 1304 branches):

MCParticles is present.
generatorStatus == 4 gives both beams with the crossing angle intact — 0.200 mrad (e⁻) and 25.087 mrad (p), identical to the parent EVGEN file.
The beams' vertex and time carry the smeared interaction point (−0.0620, −0.0047, +7.7080) mm, t = −0.0356 ns — matching the EVGEN event_pos.
generatorStatus histogram of event 0: {0: 99, 1: 3, 4: 2, 13: 1} — 5 of 105 particles are generator-level, the other 99 are Geant4 secondaries.

RECO/main/epic_craterlake/Test/…/DIS/NC/18x275/minQ2=1 (pythia8 NCDIS, beamEffects_xAngle=-0.025_hiDiv), 20 events:

generatorStatus totals {0: 21, 1: 175, 2: 94, 4: 20, 21: 20, 23: 13, 41: 4, 42: 4, 43: 4, 44: 6, 61: 10, 62: 14, 63: 13, 71: 13} — the full Pythia8 record survives: hard process (21, 23), ISR/FSR (41–44), hadronisation (61–63, 71), final state (1). Not just final-state particles, so the HepMC3 event can be rebuilt faithfully, not merely approximated.
Beam angles preserved (0.142 / 24.732 mrad), but here the beam particles' vertex and time are exactly zero even though the production is hiDiv. Checked where it went instead: the generatorStatus == 1 particles carry it — over 10 events their production vertex has z mean +19.6 mm, rms 39.5 mm, and time rms 0.127 ns, matching the 35–45 mm vertex-z spread of the parent EVGEN files. In event 0 every primary shares one vertex (+0.1387, +0.0073, −49.6093) mm at t = +0.116 ns while both beams sit at exactly (0, 0, 0, 0). So in this chain the interaction point is on the daughters, not on the beams.

So, for the recipe:

Select on generatorStatus: keep 1–4 and the Pythia8 process codes, drop 0 (Geant4 secondaries). 0 is the only status that is unambiguously not from the generator.
The crossing angle comes for free — it is already in the momenta, so no abconv step and no risk of double-applying it.
Read the interaction point per production, don't assume. Beam-particle vertex/time for the EpIC-style chains, daughter vertices for pythia8 NCDIS. For timing work this has to be checked for each branch used.
Only background-free branches. The background-mixed campaign output lives under a Bkg_* or BACKGROUNDS path segment (Bkg_Exact1S_2us, Bkg_Exactly1SignalPer2usFrame, Bkg_RealisticSignalPer2usFrame, Bkg_OnlyPer2usFrame, Bkg_1SignalPer2usFrame). Everything in the table below has no such segment. 35 background-mixed RECO directories at these energies were excluded from it.
Marco's generatorStatus banding (section 3: 0–99999 background, 100000 × class for signal) is his own merged production's convention, not what these campaign files carry — here generatorStatus is the plain HepMC/Pythia8 status. Don't decode one with the other's scheme.

Background-free `RECO` coverage

Campaign versions listed per cell; main/try*/Test paths excluded, and the 35 background-mixed (Bkg_* / BACKGROUNDS) directories at these energies left out — we mix ourselves, so only the background-free branches are listed.

Scanned for 5x41, 9x100, 9x130, 9x275, 10x100, 10x130, 10x275 across campaigns 25.10.4 → 26.07.3 and main.

18x275, 5x100 and 18x110 were scanned separately over 25.12.0 → 26.07.1. 18x275 turns out to have the broadest background-free RECO coverage of any energy — 15 distinct branches, more than the 9x set:

RECO branch (background-free)	18x275	5x100	18x110
`DIS/BeAGLE1.03.02-1.2/eHe3`	—	—	26.02.0, 26.03.1, 26.04.1, 26.07.1
`DIS/CC`	26.02.0, 26.04.1, 26.07.0, 26.07.1	26.02.0, 26.04.1, 26.07.0, 26.07.1	—
`DIS/NC`	26.02.0, 26.04.1, 26.07.0, 26.07.1	26.02.0, 26.04.1, 26.07.0, 26.07.1	—
`DIS/pythia8.316-1.0/NC/noRad/ep`	26.05.0, 26.06.0, 26.07.1	26.05.0, 26.07.1	—
`EXCLUSIVE/DDVCS_ABCONV/EpIC1.1.6-1.0`	26.02.0, 26.04.1	—	—
`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	26.02.0, 26.04.1, 26.07.1	26.02.0, 26.04.1, 26.07.1	—
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	26.02.0, 26.04.1	—	—
`EXCLUSIVE/DVMP/EpIC1.1.6-1.1/unpolarised`	26.02.0, 26.07.1	—	—
`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.0`	26.02.0, 26.04.1, 26.07.1	—	—
`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	26.02.0, 26.03.1, 26.04.1, 26.07.1	—	—
`EXCLUSIVE/PHOTOPRODUCTION_JPSI_ABCONV/coherent_ep/eSTARlight1.3.0-1.0`	26.02.0, 26.04.1	—	—
`SIDIS/D0_ABCONV/pythia8.306-1.0`	26.02.0, 26.04.1	—	—
`SIDIS/DIJET_ABCONV/pythia6.428-dijet-v1.0/PGF/noRC/ep`	26.02.0, 26.03.1, 26.04.1	—	—
`SIDIS/pythia6-eic/1.2.0/ep_noradcor`	26.04.1	—	—

DIS/NC and DIS/CC at 18x275 go back to 22.11.x and EXCLUSIVE/{DVCS,TCS}_ABCONV/18x275 to 22.11.3, so older campaigns add more still. Note that 26.02.0 and 26.04.1 carry processes that 26.07.1 dropped (DDVCS_ABCONV, DVCS_ABCONV, PHOTOPRODUCTION_JPSI_ABCONV, SIDIS/D0_ABCONV, SIDIS/DIJET_ABCONV, SIDIS/pythia6-eic/1.2.0) — for 18x275 the newest campaign is not the most complete one.

RECO branch (background-free)	5x41	9x100	9x130	9x275	10x100	10x130	10x275	detector
`DDIS/rapgap3.310-1.0/noRad/ep`	—	—	—	—	25.12.0, 26.02.0, 26.03.1, 26.04.1, 26.07.1	—	—	craterlake
`DDIS/rapgap3.310-2.0/ewRad/ep`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`DDIS/rapgap3.310-2.0/noRad/ep`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`DIS/BeAGLE1.03.02-1.0/eAu`	25.12.0	—	—	—	25.12.0	—	—	craterlake_without_zdc
`DIS/BeAGLE1.03.02-1.0/eH2`	—	—	—	—	—	25.12.0, 26.03.1	—	craterlake
`DIS/BeAGLE1.03.02-1.2/eHe3`	25.12.0, 26.02.0, 26.03.1, 26.04.1, 26.07.1	—	—	—	—	—	—	craterlake
`DIS/BeAGLE1.03.02-1.3/eH2/en`	—	—	—	—	—	26.03.1, 26.04.1	—	craterlake
`DIS/BeAGLE1.03.02-1.3/eH2/ep`	—	—	—	—	—	26.03.1, 26.04.1	—	craterlake
`DIS/BeAGLE1.03.02-2.0/eAu`	26.02.0, 26.03.1, 26.04.1	—	—	—	26.02.0, 26.03.1, 26.04.1	—	—	craterlake_without_zdc
`DIS/BeAGLE1.03.02-3.1/eAu`	—	26.07.1	—	—	—	—	—	craterlake
`DIS/BeAGLE1.03.02-3.1/eH2/en`	—	—	26.07.1	—	—	—	—	craterlake
`DIS/BeAGLE1.03.02-3.1/eH2/ep`	—	—	26.07.1	—	—	—	—	craterlake
`DIS/CC`	25.12.0, 26.02.0, 26.04.1, 26.07.0, 26.07.1	—	—	—	25.12.0, 26.02.0, 26.04.1, 26.07.0, 26.07.1	—	25.12.0, 26.02.0, 26.04.1, 26.07.0, 26.07.1	craterlake
`DIS/DJANGOH4.6.10-2.0/CC/Rad/eMinus-pMinus`	—	—	—	26.07.1	—	—	—	craterlake
`DIS/DJANGOH4.6.10-2.0/CC/Rad/eMinus-pPlus`	—	—	—	26.07.1	—	—	—	craterlake
`DIS/NC`	25.12.0, 26.02.0, 26.03.0, 26.04.1, 26.07.1	—	—	—	25.12.0, 26.02.0, 26.03.0, 26.04.1, 26.07.0, 26.07.1	—	25.12.0, 26.02.0, 26.03.0, 26.04.1, 26.07.0, 26.07.1	craterlake
`DIS/pythia6.428-1.0/NC/noRad/ep`	—	—	—	—	—	25.12.0, 26.03.1, 26.04.1	—	craterlake
`DIS/pythia8.315-1.2/EEEMCal/NC/noRad/ep`	—	—	—	26.07.1	—	—	—	craterlake
`DIS/pythia8.316-1.0/NC/noRad/ep`	26.05.0	—	26.05.0, 26.06.0, 26.07.1	26.05.0, 26.06.0, 26.07.1	26.05.0, 26.06.0, 26.07.1	—	26.05.0, 26.06.0, 26.07.1	craterlake
`EXCLUSIVE/DDVCS_ABCONV/EpIC1.1.6-1.1`	—	—	—	—	—	26.03.1, 26.04.1	—	craterlake
`EXCLUSIVE/DEMP/DEMPgen-1.2.2`	25.12.0	—	—	—	25.12.0	—	—	craterlake
`EXCLUSIVE/DEMP/DEMPgen-1.2.3`	—	—	—	—	—	25.12.0	—	craterlake
`EXCLUSIVE/DEMP/DEMPgen-1.2.4`	—	—	—	—	—	25.12.0, 26.02.0, 26.04.1	—	craterlake
`EXCLUSIVE/DEMP/DEMPgen-1.2.5`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.6.1-1.0`	25.12.0, 26.02.0, 26.04.1, 26.07.1	—	—	—	25.12.0, 26.02.0, 26.04.1, 26.07.1	25.12.0, 26.02.0, 26.04.1	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_JPSI_ABCONV/lAger3.7.0-1.0`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/BeAGLE1.03.02-1.1/eAu`	—	—	—	—	25.12.0, 26.02.0, 26.04.1	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/BeAGLE1.03.02-3.1/eAu`	—	26.07.1	—	—	—	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/sartre1.39-1.0/eAu/coherent/bsat`	—	—	—	—	25.12.0, 26.02.0, 26.04.1	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_PHI_ABCONV/sartre1.39-1.3/eAu/coherent/bsat`	—	26.07.1	—	—	—	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/sartre1.39-1.1/eAu/coherent/bsat`	—	—	—	—	25.12.0, 26.02.0, 26.04.1	—	—	craterlake
`EXCLUSIVE/DIFFRACTIVE_RHO_ABCONV/sartre1.39-1.4/eAu/coherent/bsat`	—	26.07.1	—	—	—	—	—	craterlake
`EXCLUSIVE/DVCS_ABCONV`	25.12.0	—	—	—	25.12.0	—	—	craterlake
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.1`	—	—	—	—	—	25.12.0	—	craterlake
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.6-1.2`	26.02.0, 26.04.1	—	—	—	26.02.0, 26.04.1	26.02.0, 26.04.1	—	craterlake
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/BH_ONLY`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_BH`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/DVCS_ONLY`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DVMP/EpIC1.1.6-1.0/unpolarised`	—	—	—	—	—	25.12.0, 26.02.0	—	craterlake
`EXCLUSIVE/DVMP/EpIC1.1.6-1.1/unpolarised`	25.12.0, 26.02.0, 26.07.1	—	—	—	—	—	—	craterlake
`EXCLUSIVE/DVMP/EpIC1.1.8-1.0/unpolarised`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.0`	—	—	—	—	—	25.12.0, 26.02.0, 26.04.1	—	craterlake
`EXCLUSIVE/DVMP/JPsi_MuMu_ABCONV/lAger3.6.1-1.1`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/MESON_SF_ABCONV/eicMesonSFGen-1.0.0/ep`	26.02.0, 26.03.1, 26.04.1, 26.07.1	26.07.1	26.07.1	26.07.1	26.02.0, 26.03.1, 26.04.1, 26.07.1	26.02.0, 26.03.1, 26.04.1	—	craterlake
`EXCLUSIVE/UCHANNEL_RHO`	—	—	—	—	25.12.0	—	—	craterlake
`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon1S`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon2S`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`EXCLUSIVE/UPSILON_ABCONV/eSTARlight1.3.0-1.0/Upsilon3S`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-2.0/eAu`	25.12.0, 26.03.1, 26.04.1	—	—	—	25.12.0, 26.03.1, 26.04.1	—	—	craterlake_without_zdc
`SIDIS/D0_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	—	26.07.1	—	—	—	—	—	craterlake
`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.306-1.2/ep`	25.12.0, 26.02.0, 26.03.1, 26.04.1	—	—	—	—	—	—	craterlake
`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.312-2.0/ep`	—	—	—	—	—	26.03.1, 26.04.1	—	craterlake
`SIDIS/D0_ABCONV/HFsim-PYTHIA/pythia8.316-1.0/ep`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`SIDIS/D0_ABCONV/pythia8.306-1.1`	—	—	—	—	25.12.0, 26.02.0, 26.03.0, 26.04.1	—	—	craterlake
`SIDIS/Ds_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	—	26.07.1	—	—	—	—	—	craterlake
`SIDIS/Ds_ABCONV/HFsim-PYTHIA/pythia8.316-1.1/ep`	—	—	—	26.07.1	—	—	—	craterlake
`SIDIS/Lc_ABCONV/HFsim-BeAGLE/BeAGLE1.03.01-2.0/eAu`	—	—	—	—	25.12.0, 26.03.1, 26.04.1	—	—	craterlake_without_zdc
`SIDIS/Lc_ABCONV/HFsim-BeAGLE/BeAGLE1.03.02-1.0/eAu`	—	26.07.1	—	—	—	—	—	craterlake
`SIDIS/Lc_ABCONV/HFsim-PYTHIA/pythia8.306-1.2/ep`	—	—	—	—	25.12.0, 26.02.0, 26.03.1, 26.04.1	—	—	craterlake
`SIDIS/Lc_ABCONV/HFsim-PYTHIA/pythia8.316-1.0/ep`	—	—	—	26.07.1	—	—	—	craterlake
`SIDIS/pythia6-eic/1.0.0`	—	—	—	—	26.03.0	—	26.03.0	craterlake
`SIDIS/pythia6-eic/1.2.0/ep_noradcor`	—	—	—	—	26.04.1	—	—	craterlake
`SIDIS/pythia6-eic/1.3.0/ep_noradcor`	—	—	26.07.1	26.07.1	—	—	—	craterlake
`Test/2025-10-17_14-30-00/EXCLUSIVE/DVCS_ABCONV`	main	—	—	—	—	—	—	craterlake
`Test/2025-11-24_07-22-00/1Bkg_1SignalPer2usFrame/Synrad_10GeV_Vac_10000Ahr_Runtime_50s_Egas_10GeV_Hgas_275GeV/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-24_15-22-00/1Bkg_1SignalPer2usFrame/Synrad_10GeV_Vac_10000Ahr_Runtime_50s_Egas_10GeV_Hgas_275GeV/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-25_15-22-00/1Bkg_1SignalPer2usFrame/Synrad_10GeV_Vac_10000Ahr_Runtime_50s_Egas_10GeV_Hgas_275GeV/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-25_22-22-00/1Bkg_1SignalPer2usFrame/Synrad_10GeV_Vac_10000Ahr_Runtime_50s_Egas_10GeV_Hgas_275GeV/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-25_22-43-00/1Bkg_1SignalPer2usFrame/Synrad_10GeV_Vac_10000Ahr_Runtime_50s_Egas_10GeV_Hgas_275GeV/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-27_23-43-00/Bkg_1SignalPer2usFrame/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-28_23-50-00/Bkg_1SignalPer2usFrame/DIS/NC`	—	—	—	—	—	—	main	craterlake
`Test/2025-11-28_23-55-00/Bkg_1SignalPer2usFrame/DIS/NC`	—	—	—	—	—	—	main	craterlake

This is what makes 9 GeV workable: FULL has nothing at 9x (section 5.1a), but RECO 26.07.1 covers essentially the whole 9x130 / 9x275 generator set, and RECO 26.05.0 / 26.06.0 / 26.07.1 cover DIS/pythia8.316-1.0/NC at both.

Staging caveat. These files are largely tape-resident. The 9x RECO directories on /volatile and /cache exist but are mostly empty — EXCLUSIVE/DVCS_ABCONV/EpIC_v1.1.8-1.4/{DVCS_BH,DVCS_ONLY,BH_ONLY}/{9x130,9x275} were the only 9x RECO branches with files on disk today. The XRootD door root://dtn2304.jlab.org:8443//jlab-osdf-ro/eic/EPIC/{volatile,cache,work}/… serves only the disk stores, so it cannot reach the purged files either — anything else has to be staged from /mss with jcache first.

5.7 Recommendations and open requests

Framed as decided: we add background ourselves, so what we need from the collaboration is background-free physics events plus the per-source background inputs — not anybody's pre-mixed production.

1. Work at 10x100, 10x130, 10x275 and 18x275 now. These are the four energies where all five background sources are on disk (section 5.1b), and between them they carry the widest signal coverage in the catalogue. 18x275 has the largest statistics by far (the SIDIS/pythia6 18x275 branches alone are 1.4 TiB + 506 GiB) and the lowest SynRad rate; 10x275 vacuum is the busiest occupancy. Having both lets us train and test across an order of magnitude in background rate rather than at a single point.

2. 9 GeV is signal-ready and background-blocked. 9x275 and 9x130 have afterburned generator data for a broad set of processes; 9x100 is an eA point with almost nothing in ep. What is missing is only the rate table: the campaign already reuses the 10x275 SynRad / Coulomb / Touschek files for 10x100 and 10x130 with nothing changed but freq, so the same reuse at 9x needs numbers, not new simulation. Worth asking for explicitly, since the samples are already there.

3. 5x41 and 5x100 are blocked by three files, not by a missing recipe: BEAMGAS/electron/GETaLM1.0.0-1.1/5GeV/…, SYNRAD/dataprod_rel_1.0.0/5x100/…preproc_10000repeats…, and (5x41 only) BEAMGAS/proton/pythia8.306-1.0/41GeV/combined/…. If a low-energy point matters, asking for those three is a much smaller request than a new production.

4. edm2hepmc is worth building — RECO keeps the generator record.RECO keeps MCParticles with the crossing angle already applied and the full Pythia8 history intact (section 5.6, measured on two files). That makes RECO a first-class source, not a fallback: it covers 9x where FULL has nothing, and it is tape-backed where much of EVGEN is not. Selection rule is simply generatorStatus != 0; the interaction-point offset has to be read per production (beam particles for the EpIC-style chains, daughter vertices for pythia8 NCDIS).

5. Nothing is being staged. Per instruction, no jcache. Everything above was established from directory metadata plus files already resident on /volatile / /cache.

6. Durability — most of EVGEN is not archived. Branches whose directory also exists under /mss. Counted over the seven energies of the first pass only (18x275, 5x100 and the eA points are not in this tally):

Energy	tape-backed branches	`/volatile`-only
5x41	5 / 25	20
9x100	0 / 8	8
9x130	5 / 18	13
9x275	8 / 21	13
10x100	5 / 31	26
10x130	4 / 15	11
10x275	2 / 4	2

/volatile is purgeable scratch. This is another argument for the edm2hepmc route: RECO/26.07.1 is on tape and covers nearly the whole 9x set, so the generator record survives a purge even where the EVGEN copy does not.

7. Corrections to carry forward. Two things in Marco's table are wrong as a guide to availability, and one of my own earlier readings was wrong:

rho (23), spectroscopy (24) and omega (25) all have afterburned generator data — EXCLUSIVE/RHO_ABCONV (10x100), SPECTROSCOPY_ABCONV (10x100), OMEGA_ABCONV (18x275). They are invisible to a path-name search because the energy is in the filename, not a directory segment; I reported them as empty earlier in this survey and that was wrong.
tcs is ambiguous in his table. TCS_ABCONV is the TCS+BH interference sample (files tcs_bh_int_ab_hiAcc_*); pure TCS exists only head-on, at 5x41, 5x100, 10x100 and 18x275, 20 files / 20.8 GiB each. If the class means pure TCS, abconv has to be run; if it means the interference sample, it is ready at 5x41, 10x100 and 18x275.
cohrho / cohphi are eAu-only by definition, and the bulk of them sit at 18x110, an energy not in the original request.

6. Access verified 2026-10-15

Item	Result
ifarm tunnel	`ssh -p 4333 romanov@localhost` → `ifarm2401.jlab.org`, user `romanov`. Works.
Rucio client	`apptainer exec /cvmfs/singularity.opensciencegrid.org/eicweb/eic_xl:nightly rucio …` → rucio 40.2.0
Rucio identity	`rucio whoami` → account `eicread` (GROUP, ACTIVE) — read-only, which is what we want
Rucio scopes	`ying_test`, `test`, `eic`, `panta`, `rahmans`, `epic`
Rucio content	`epic:` indexes campaign output (`epic:/RECO/<ver>/…`, `epic:/FULL/…`) and generator datasets (`epic:/EVGEN/…`) — but only partially: 121 `EVGEN` DIDs and 197 `FULL` DIDs in total. Whole process trees that exist on disk are absent from Rucio (`EXCLUSIVE/TCS`, `OMEGA_ABCONV`, `SPECTROSCOPY_ABCONV`, `RHO_ABCONV`, `DEMPgen-1.2.4`, `DEMPgen-1.2.5`, …). Rucio is not the inventory.
Store hierarchy, measured	`/volatile` ⊇ tape ⊇ Rucio. 414 target-energy EVGEN directories on `/volatile` vs 105 on `/mss`, and every `/mss` target-energy EVGEN directory also exists on `/volatile` (`comm -23` empty). So: `/volatile` = the inventory, `/mss` = the durability check (a `/volatile`-only dataset is purgeable), Rucio = a partial index that adds replica/RSE information for the DIDs it does carry.
EVGEN on disk	`/volatile/eic/EPIC/EVGEN/{BACKGROUNDS,CI,DDIS,DIS,Djangoh,EW_BSM,EXCLUSIVE,SIDIS,SINGLE,Test}`; partial mirror in `/cache/eic/EPIC/EVGEN/{BACKGROUNDS,DIS,EXCLUSIVE,SIDIS,Test}`
Campaign config repo	`github.com/eic/simulation_campaign_datasets` cloned locally; has `config_data/.json` (background rates per energy incl. 10x130, 10x100, 10x275, 5x41, 5x100, 18x275) and per-process `config.yml` + `.csv` dataset listings
Local space	`/data` 951 GB free (bulk); scratchpad is tmpfs, 7.8 GB free (text only)
Farm write space	`/w/eic-scshelf2104/users/romanov/optim-out` exists and is writable — not needed for listing work

7. Appendix: reproducing the sweep

All of it is read-only on the data. The only writes anywhere on the farm were a handful of ~kB helper scripts in ifarm's /tmp, removed afterwards; nothing was written under /work, /volatile, /cache, /mss or /w/…. Commands go over the tunnel as ssh -p 4333 -o BatchMode=yes romanov@localhost bash -s <<'EOF' … EOF (ifarm's login shell is tcsh, so the script must arrive on stdin), and Rucio / ROOT run inside /cvmfs/singularity.opensciencegrid.org/eicweb/eic_xl:nightly.

7.1 Inventory dumps

bash

# Rucio (partial index): 121 EVGEN DIDs, 197 FULL DIDs, 105 background-mixed
apptainer exec /cvmfs/singularity.opensciencegrid.org/eicweb/eic_xl:nightly \
  rucio did list 'epic:/EVGEN/*'

# the real inventory
find /volatile/eic/EPIC/EVGEN -maxdepth 10 -type d
find /volatile/eic/EPIC/FULL  -maxdepth 9  -type d
find /mss/eic/EPIC/RECO/26.07.1 -maxdepth 8 -type d -name '9x*'

# per-leaf file count and size
find "$d" -maxdepth 1 -type f | wc -l; du -sb "$d"

7.2 Afterburner measurement

Needs -B /volatile:/volatile:ro (otherwise TFile::Open reports does not exist) and a writable TMPDIR. Do not add python3 -I — isolated mode hides ROOT from the interpreter.

bash

apptainer exec -B /tmp:/tmp -B /volatile:/volatile:ro \
  /cvmfs/singularity.opensciencegrid.org/eicweb/eic_xl:nightly \
  env TMPDIR=/tmp HOME=/tmp python3 /tmp/measure_remote.py --root < list.tsv

measure_remote.py (reads .hepmc3.tree.root with --root, plain .hepmc ASCII without it; input is TSV file<TAB>dir<TAB>energy<TAB>nfiles<TAB>bytes):

python

import sys, math, statistics as st, os, gzip
def ascii_scan(path, nev=20):
    op = gzip.open if path.endswith(".gz") else open
    beams={}; pos=[]; nE=0
    with op(path, "rt", errors="replace") as fh:
        for line in fh:
            if line.startswith("E "):
                nE += 1
                if nE > nev: break
                f=line.split()
                if "@" in f:
                    i=f.index("@")
                    try: pos.append(tuple(float(x) for x in f[i+1:i+5]))
                    except Exception: pass
            elif line.startswith("P ") and nE>0:
                f=line.split()
                # P <id> <parent> <pid> px py pz e m status
                if len(f)>=10 and f[9]=="4":
                    pid=int(f[3]); px,py,pz=float(f[4]),float(f[5]),float(f[6])
                    beams.setdefault(pid,[]).append(math.atan2(math.hypot(px,py),abs(pz))*1000.0)
    return beams, pos, None
def root_scan(path, nev=20):
    import ROOT
    f=ROOT.TFile.Open(path)
    tname=next(k.GetName() for k in f.GetListOfKeys() if k.GetClassName()=="TTree")
    t=f.Get(tname); bn=t.GetListOfBranches()[0].GetName(); n=t.GetEntries()
    beams={}; pos=[]
    for i in range(min(nev,n)):
        t.GetEntry(i); ev=getattr(t,bn)
        for p in ev.particles:
            if p.status==4:
                m=p.momentum
                beams.setdefault(p.pid,[]).append(math.atan2(math.hypot(m.px(),m.py()),abs(m.pz()))*1000.0)
        got=False
        try:
            ep=ev.event_pos
            if any(abs(v)>1e-12 for v in (ep.x(),ep.y(),ep.z(),ep.t())):
                pos.append((ep.x(),ep.y(),ep.z(),ep.t())); got=True
        except Exception: pass
        if not got:
            try:
                v=ev.vertices[0].position
                pos.append((v.x(),v.y(),v.z(),v.t()))
            except Exception: pass
    f.Close()
    return beams, pos, n
def fmt(beams,pos,n):
    out=[]
    for pid in sorted(beams, key=lambda k: -abs(k)):
        a=beams[pid]
        out.append(f"{pid}:{st.mean(a):.3f}+-{(st.pstdev(a) if len(a)>1 else 0):.3f}")
    ang="|".join(out)
    if pos:
        zs=[p[2] for p in pos]; ts=[p[3] for p in pos]; xs=[p[0] for p in pos]
        vtx=f"x_rms={st.pstdev(xs) if len(xs)>1 else 0:.4f};z_rms={st.pstdev(zs) if len(zs)>1 else 0:.3f};t_rms={st.pstdev(ts) if len(ts)>1 else 0:.3f}"
    else:
        vtx="no_event_pos"
    return ang, vtx, (n if n is not None else "")
if __name__=="__main__":
    if len(sys.argv)>1 and sys.argv[1]=="--root":
        import ROOT; ROOT.gSystem.Load("libHepMC3rootIO"); ROOT.gErrorIgnoreLevel=ROOT.kFatal
    for line in sys.stdin:
        line=line.rstrip("\n")
        if not line: continue
        fp, d, e, nf, b = line.split("\t")
        isroot = fp.endswith(".root")
        want_root = (len(sys.argv)>1 and sys.argv[1]=="--root")
        if isroot != want_root: continue
        try:
            beams,pos,n = (root_scan if isroot else ascii_scan)(fp)
            ang,vtx,nn = fmt(beams,pos,n)
            print(f"{d}\t{e}\t{nf}\t{b}\t{os.path.basename(fp)}\t{nn}\t{ang}\t{vtx}", flush=True)
        except Exception as ex:
            print(f"{d}\t{e}\t{nf}\t{b}\t{os.path.basename(fp)}\tERR\t{type(ex).__name__}: {ex}\t", flush=True)

7.3 Reading `MCParticles` out of a RECO file

python

import math, collections, ROOT
f = ROOT.TFile.Open(path)                  # …eicrecon.edm4eic.root
t = f.Get("events")
t.GetEntry(0)
mc = t.MCParticles
for p in mc:
    if p.generatorStatus == 4:             # the two beams
        m, v = p.momentum, p.vertex
        ang = math.atan2(math.hypot(m.x, m.y), abs(m.z)) * 1000   # mrad
        print(p.PDG, ang, (v.x, v.y, v.z), p.time)
print(collections.Counter(p.generatorStatus for p in mc))
# keep generatorStatus != 0 to rebuild the generator record; 0 = Geant4 secondary

No podio or edm4hep dictionary load is needed — ROOT 6.40 in eic_xl picks them up from the rootmap.

8. Processes with crossing-angle HepMC3 data

The single table to pick datasets from. Every row has HepMC3 files with the 25 mrad crossing angle measured present on the beams, so each is ready for ddsim with no abconv step. Head-on branches, non-HepMC formats and the one unmeasured branch are deliberately absent — see sections 5.4 and 5.5 for those.

Energies in bold are the four where a complete background mixing recipe also exists (section 5.1b), i.e. where a full signal+background dataset can be built today. format: root = .hepmc3.tree.root, ascii = .hepmc / .hepmc3, ascii.gz = gzipped. 167 branches, grouped into 47 rows.

8.1 Processes per beam energy

The same data rolled up by energy, ordered by how many processes each one carries. This is the trade-off in one view: breadth of physics against whether a background recipe exists.

mixable means all five machine-background sources for that energy are present on disk, so we can add background ourselves there today — SynRad (status 2000), electron beam gas + brems (3000), Coulomb (4000), Touschek (5000) and proton beam gas (6000), as listed in the campaign's config_data/…hgas_<E>_<pipe>….json. It says nothing about the signal data; it is purely whether background can be mixed in. Full detail in section 5.1b.

energy	processes with crossing-angle HepMC3	mixable
10x100	27	yes
18x275	22	yes
5x41	18	—
9x275	15	—
9x130	12	—
10x130	10	yes
9x100	8	—
5x100	6	—
18x110	4	—
10x275	4	yes
18x108.41	2	—

10x100 and 18x275 are the only energies that are both broad and mixable. 10x275 is mixable but carries just 4 processes (the DIS and SIDIS lines), so its value is as a second occupancy point — SynRad is 36.6 GHz there in vacuum against 3.3 GHz at 18x275 — rather than as a second process set.

5x41 is third by breadth (18 processes) but its background recipe is blocked by three missing files, and 9x275 / 9x130 are fourth and fifth with no rate table at all. Both are worth asking for on the strength of the signal coverage already sitting on disk.

8.2 Process catalogue, crossing-angle data only

process	system	energies with crossing-angle HepMC3	generator(s)	format	branches	files	size	Marco slug
CC DIS, Q² threshold	ep	5x41 5x100 10x100 10x275 18x275	Pythia8 (campaign)	ascii	5	90	639.6 GiB	`ccdisq100/1000`
CC DIS, radiative, polarised	ep	9x275	DJANGOH 4.6.10	root	2	6	3.0 GiB	`ccdisq* (diff. generator)`
coherent J/ψ photoproduction	ep	18x275	eSTARlight	root	1	2	679.2 MiB	`jpsiphoto`
coherent φ→K⁺K⁻, eA	eAu	9x100 10x100 18x110	Sartre	ascii, root	4	14	61.1 GiB	`cohphi`
coherent/incoherent J/ψ, eA	eAu	18x110	Sartre	ascii	2	8	32.4 GiB	`— (eA)`
coherent/incoherent ρ, eA	eAu	9x100 10x100 18x110	Sartre	ascii.gz, root	4	2012	48.1 GiB	`cohrho`
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	ep	5x41 5x100 9x130 9x275 10x100 10x130 18x275	DEMPgen	root	13	840	375.1 GiB	`dempq3to10/10to20/20to35`
diffractive DIS	ep	9x130 9x275 10x100	RAPGAP 3.310	root	5	5	14.2 GiB	`ddis`
diffractive J/ψ, ep	ep	5x41 5x100 9x130 9x275 10x100 10x130 18x275	lAger	root	7	7	437.7 MiB	`jpsi`
diffractive J/ψ, ePb	ePb	18x108.41	BeAGLE	ascii	1	54	112.5 GiB	`— (ePb)`
diffractive φ, eA	eAu	18x110	eSTARlight	ascii	1	1	54.0 MiB	`cohphi`
diffractive φ, ePb/eAu	eAu	9x100 10x100	BeAGLE	root	2	200	35.8 GiB	`cohphi`
diffractive φ, ePb/eAu	ePb	18x108.41	BeAGLE	ascii	1	54	608.6 GiB	`cohphi`
double DVCS	ep	10x130 18x275	EpIC	root	2	6	2.3 GiB	`ddvcs`
DVCS	ep	5x41 10x100 18x275	DVCS gen (legacy)	ascii	3	60	28.9 GiB	`dvcs`
DVCS + BH + INT	ep	5x41 10x100 10x130 18x275	EpIC 1.1.6	root	6	10	3.5 GiB	`dvcs`
DVCS / BH / interference	ep	9x130 9x275	EpIC 1.1.8	root	6	24	11.3 GiB	`dvcs`
DVCS on deuteron	eD	5x41 10x100	EpIC	ascii	2	26	23.9 GiB	`— (eD)`
DVMP / Υ (lAger legacy)	ep	10x100	lAger	ascii	1	8	1.7 GiB	`dvmp/upsilon`
DVMP J/ψ→μ⁺μ⁻	ep	9x130 9x275 10x130 18x275	lAger	root	4	4	3.2 GiB	`dvmp`
DVMP π⁰	ep	5x41 9x130 9x275 10x100 10x130 18x275	EpIC	root	6	7	2.1 GiB	`dvmp`
exclusive ρ→π⁺π⁻	ep	10x100	SLIGHT	ascii.gz	1	52	719.0 MiB	`rho`
inclusive eA DIS	eAu	5x41 9x100 10x100	BeAGLE	root	5	815	94.3 GiB	`—`
inclusive eA DIS	eH2	9x130 10x130	BeAGLE	root	5	221	73.3 GiB	`—`
inclusive eA DIS	eHe3	5x41	BeAGLE	root	1	3	3.2 GiB	`—`
meson structure function K⁺Λ	ep	5x41 9x100 9x130 9x275 10x100 10x130 18x275	eicMesonSFGen	root	7	3800	2.9 GiB	`mesonsf`
NC DIS, inclusive	ep	5x41 5x100 9x130 9x275 10x100 10x130 10x275 18x275	Pythia6.428, Pythia8.316	root	8	175	222.6 GiB	`ncdisq1/10/100/1000`
NC DIS, inclusive (EEEMCal cfg)	ep	9x275	Pythia8.315	root	1	100	1.8 GiB	`ncdisq*`
NC DIS, Q² threshold	ep	5x41 5x100 10x100 10x275 18x275	Pythia8 (campaign)	ascii, root	8	328	2.3 TiB	`ncdisq1/10/100/1000`
NC/CC DIS, radiative	ep	18x275	DJANGOH 4.6.21	root	2	8	6.5 GiB	`ncdisq/ccdisq`
SIDIS / min-bias photoprod.	en	5x41 10x100 18x275	Pythia6-eic	root	3	715	576.5 GiB	`photoprod`
SIDIS / min-bias photoprod.	ep	5x41 5x100 9x130 9x275 10x100 10x275 18x275	Pythia6-eic	root	10	1141	1.1 TiB	`photoprod`
SIDIS D_s	ep	9x275	Pythia8 HFsim	root	1	1200	3.1 GiB	`—`
SIDIS D_s (eA)	eAu	9x100	BeAGLE HFsim	root	1	500	5.8 GiB	`—`
SIDIS dijet (PGF)	ep	18x275	Pythia6.428 dijet	root	1	100	972.1 MiB	`—`
SIDIS open charm D⁰	ep	5x41 9x130 9x275 10x100 10x130 18x275	Pythia8, Pythia8 HFsim	root	6	3728	22.7 GiB	`—`
SIDIS open charm D⁰ (eA)	eAu	5x41 9x100 10x100	BeAGLE HFsim	root	5	4900	15.1 GiB	`—`
SIDIS Λ	ep	18x275	Pythia8.306	root	1	8	10.5 GiB	`—`
SIDIS Λ_c	ep	9x275 10x100	Pythia8 HFsim	root	2	3500	3.7 GiB	`—`
SIDIS Λ_c (eA)	eAu	9x100 10x100	BeAGLE HFsim	root	3	1498	6.6 GiB	`—`
SIDIS, Q²-binned	ep	5x41 10x100 18x275	Pythia6 (IP6 era)	ascii, root	5	5900	3.9 TiB	`photoprod (older prod.)`
spectroscopy ψ(2S)/X/Y	ep	10x100	dedicated	ascii	1	24	2.2 GiB	`spectroscopy`
TCS + BH interference	ep	5x41 10x100 18x275	TCS gen	ascii	3	800	59.6 GiB	`tcs`
u-channel π⁰	ep	18x275	SLIGHT	root	1	1	36.3 MiB	`upi0`
u-channel ρ	ep	10x100	SLIGHT/GENEX	root	1	1	35.6 MiB	`urho`
Υ(1S/2S/3S) threshold	ep	9x130 9x275	eSTARlight 1.3.0	root	6	450	14.1 GiB	`upsilon`
ω photoproduction	ep	18x275	SLIGHT	ascii	1	96	1.2 GiB	`omega`

Rows are keyed on (process, collision system) — ep, en, eD, eAu, eH2, eHe3, ePb are not interchangeable, and several processes exist in more than one. Energies are the measured ones, so SIDIS, Q²-binned is listed at 18x275 for the branch whose directory is misnamed ep_5x41 (section 5.5).

Totals: 33,502 files, 10.3 TiB.

Several processes exist in more than one collision system with different energy coverage — inclusive eA DIS splits across eAu / eH2 / eHe3, diffractive φ across eAu / ePb, and SIDIS min-bias photoproduction has an en (electron–neutron) variant alongside ep. Those are not interchangeable, so they stay as separate rows rather than being merged into one process line.

8.3 Coverage including degraded sources

Same rows as 8.2, but no longer restricted to ready-to-use data. Each energy carries a marker for the best source available at that energy:

marker	meaning	what it costs to use
(none)	crossing-angle HepMC3 on `/volatile`	nothing — feed to `ddsim`
`(R)`	only in `RECO`, as `MCParticles`	needs `edm2hepmc`; crossing angle already applied
`(h)`	only head-on HepMC3	needs `abconv` run first
`(L)`	only in LUND / eic-smear / other non-HepMC form	needs a format conversion
`(?)`	present but beams not determined	needs inspection

A marker describes the best option for that branch family, so a plain energy may also have head-on or RECO copies. Note that a few (h) rows do have an afterburned counterpart, but under a differently-named row because the productions are distinct — coherent ρ, eA (head-on, DIFFRACTIVE_RHO/Sartre) versus coherent/incoherent ρ, eA (afterburned, DIFFRACTIVE_RHO_ABCONV/Sartre/{Coherent,Incoherent}). Section 5.5 pairs them explicitly. Energies in bold have a complete background recipe.

This table has 57 rows against 8.2's 47: the ten extra are processes that exist only head-on, only in a non-HepMC format, or only in RECO.

process	system	energies, by best available source	generator(s)	branches	files	size	Marco slug
CC DIS, Q² threshold	ep	5x41 5x100 10x100 10x275 18x275	Pythia8 (campaign)	5	90	639.6 GiB	`ccdisq100/1000`
CC DIS, radiative (eRHIC fmt)	eD	18x137 (h)	DJANGOH/eRHIC	1	42	92.2 GiB	`ccdisq*`
CC DIS, radiative (eRHIC fmt)	ep	18x275 (h)	DJANGOH/eRHIC	2	36	102.4 GiB	`ccdisq*`
CC DIS, radiative, polarised	ep	9x275	DJANGOH 4.6.10	2	6	3.0 GiB	`ccdisq* (diff. generator)`
coherent J/ψ photoproduction	ep	18x275	eSTARlight	1	2	679.2 MiB	`jpsiphoto`
coherent ρ, eA	eAu	18x110 (h)	Sartre	1	2	23.7 GiB	`cohrho`
coherent φ→K⁺K⁻, eA	eAu	9x100 10x100 18x110	Sartre	4	14	61.1 GiB	`cohphi`
coherent/incoherent J/ψ, eA	eAu	18x110	Sartre	2	8	32.4 GiB	`— (eA)`
coherent/incoherent ρ, eA	eAu	9x100 10x100 18x110	Sartre	4	2012	48.1 GiB	`cohrho`
DEMP K⁺Λ / K⁺Σ (LUND fmt)	ep	5x41 (L) 10x100 (L)	DEMPgen (legacy)	5	46	11.1 GiB	`dempq*`
DEMP π⁺ / K⁺Λ / K⁺Σ⁰	ep	5x41 5x100 9x130 9x275 10x100 10x130 18x275	DEMPgen	13	840	375.1 GiB	`dempq3to10/10to20/20to35`
diffractive DIS	ep	9x130 9x275 10x100	RAPGAP 3.310	5	5	14.2 GiB	`ddis`
diffractive J/ψ, eA	eAu	18x110 (h)	Sartre	1	2	23.6 GiB	`— (eA)`
diffractive J/ψ, ep	ep	5x41 5x100 9x130 9x275 10x100 10x130 18x275	lAger	7	7	437.7 MiB	`jpsi`
diffractive J/ψ, ePb	ePb	18x108.41	BeAGLE	1	54	112.5 GiB	`— (ePb)`
diffractive φ, eA	eA	18x110 (h)	Sartre/BeAGLE	1	200	72.0 GiB	`cohphi`
diffractive φ, eA	eAu	18x110	Sartre/BeAGLE, eSTARlight	2	3	23.7 GiB	`cohphi`
diffractive φ, ePb/eAu	eAu	9x100 10x100	BeAGLE	2	200	35.8 GiB	`cohphi`
diffractive φ, ePb/eAu	ePb	18x108.41	BeAGLE	1	54	608.6 GiB	`cohphi`
double DVCS	ep	10x130 18x275	EpIC	2	6	2.3 GiB	`ddvcs`
DVCS	ep	5x41 10x100 18x275	DVCS gen (legacy)	6	65	32.9 GiB	`dvcs`
DVCS + BH + INT	ep	5x41 10x100 10x130 18x275	EpIC 1.1.6	6	10	3.5 GiB	`dvcs`
DVCS + BH + INT, coherent	ep	18x275 (L)	EpIC	1	1	823.3 MiB	`dvcs`
DVCS / BH / interference	ep	9x130 9x275	EpIC 1.1.8	6	24	11.3 GiB	`dvcs`
DVCS on deuteron	eD	5x41 10x100	EpIC	2	26	23.9 GiB	`— (eD)`
DVMP / Υ (lAger legacy)	ep	10x100	lAger	1	8	1.7 GiB	`dvmp/upsilon`
DVMP J/ψ→μ⁺μ⁻	ep	9x130 9x275 10x130 18x275	lAger	4	4	3.2 GiB	`dvmp`
DVMP π⁰	ep	5x41 9x130 9x275 10x100 10x130 18x275	EpIC	7	8	2.2 GiB	`dvmp`
exclusive ρ→π⁺π⁻	ep	10x100 18x275 (h)	SLIGHT	3	56	1.5 GiB	`rho`
inclusive eA DIS	eAu	5x41 9x100 10x100	BeAGLE	5	815	94.3 GiB	`—`
inclusive eA DIS	eH2	9x130 10x130	BeAGLE	5	221	73.3 GiB	`—`
inclusive eA DIS	eHe3	5x41 18x110 (R)	BeAGLE	1	3	3.2 GiB	`—`
meson structure function K⁺Λ	ep	5x41 9x100 9x130 9x275 10x100 10x130 18x275	eicMesonSFGen	7	3800	2.9 GiB	`mesonsf`
NC DIS, inclusive	ep	5x41 5x100 9x130 9x275 10x100 10x130 10x275 18x275	Pythia6.428, Pythia8.316	8	175	222.6 GiB	`ncdisq1/10/100/1000`
NC DIS, inclusive (EEEMCal cfg)	ep	9x275	Pythia8.315	1	100	1.8 GiB	`ncdisq*`
NC DIS, Q² threshold	ep	5x41 5x100 10x100 10x275 18x275	Pythia8 (campaign)	8	328	2.3 TiB	`ncdisq1/10/100/1000`
NC/CC DIS, radiative	ep	18x275	DJANGOH 4.6.21	2	8	6.5 GiB	`ncdisq/ccdisq`
SIDIS / min-bias photoprod.	en	5x41 10x100 18x275	Pythia6-eic	3	715	576.5 GiB	`photoprod`
SIDIS / min-bias photoprod.	ep	5x41 5x100 9x130 9x275 10x100 10x275 18x275	Pythia6-eic	10	1141	1.1 TiB	`photoprod`
SIDIS D_s	ep	9x275	Pythia8 HFsim	1	1200	3.1 GiB	`—`
SIDIS D_s (eA)	eAu	9x100	BeAGLE HFsim	1	500	5.8 GiB	`—`
SIDIS dijet (PGF)	ep	18x275	Pythia6.428 dijet	1	100	972.1 MiB	`—`
SIDIS open charm D⁰	ep	5x41 9x130 9x275 10x100 10x130 18x275	Pythia8, Pythia8 HFsim	6	3728	22.7 GiB	`—`
SIDIS open charm D⁰ (eA)	eAu	5x41 9x100 10x100	BeAGLE HFsim	5	4900	15.1 GiB	`—`
SIDIS Λ	ep	18x275	Pythia8.306	1	8	10.5 GiB	`—`
SIDIS Λ_c	ep	9x275 10x100	Pythia8 HFsim	2	3500	3.7 GiB	`—`
SIDIS Λ_c (eA)	eAu	9x100 10x100	BeAGLE HFsim	3	1498	6.6 GiB	`—`
SIDIS, Q²-binned	ep	5x41 10x100 18x100 (L) 18x275	Pythia6 (IP6 era)	8	7320	4.0 TiB	`photoprod (older prod.)`
spectroscopy ψ(2S)/X/Y	ep	10x100	dedicated	2	30	2.6 GiB	`spectroscopy`
TCS + BH interference	ep	5x41 10x100 18x275	TCS gen	6	860	121.9 GiB	`tcs`
TCS, pure	ep	5x41 (h) 5x100 (h) 10x100 (h) 18x275 (h)	TCS gen	4	80	83.1 GiB	`tcs`
u-channel π⁰	ep	18x275	SLIGHT	1	1	36.3 MiB	`upi0`
u-channel ρ	ep	10x100	SLIGHT/GENEX	1	1	35.6 MiB	`urho`
u-channel ω (neutral decay)	ep	18x275 (h)	SLIGHT	1	10	250.6 MiB	`— (not upi0/omega)`
Υ photoproduction (legacy)	ep	? (h)	eSTARlight	1	3	9.5 MiB	`upsilon`
Υ(1S/2S/3S) threshold	ep	9x130 9x275	eSTARlight 1.3.0	6	450	14.1 GiB	`upsilon`
ω photoproduction	ep	18x275	SLIGHT	2	98	1.4 GiB	`omega`

Cells by source: 128 ready, 1 RECO-only, 12 head-on-only, 4 needing format conversion, 0 undetermined.

What the RECO route adds. Exactly one (process, system, energy) combination exists in a background-free RECO campaign with no crossing-angle HepMC3 on /volatile — so RECO's value is durability and the 9x points, not extra process coverage:

process	system	energy	RECO campaign(s)
inclusive eA DIS	eHe3	18x110	26.02.0, 26.03.1, 26.04.1, 26.07.1

8.4 Processes per beam energy, all sources

Section 8.1 counted only ready-to-use data. This counts a process at an energy if it is reachable at all — ready, via RECO, via abconv, or via a format conversion. Ordered by total, as in 8.1.

Legend:
mixable — all five machine-background sources for that energy are on disk, so we can add background ourselves there today: SynRad (status 2000), electron beam gas + brems (3000), Coulomb (4000), Touschek (5000), proton beam gas (6000). Only 10x100, 10x130, 10x275 and 18x275 qualify; 5x41 and 5x100 are short three files and the 9x points have no rate config at all (section 5.1b). This is about background only — it says nothing about the signal data.
The other columns say what form the signal data takes, i.e. what it costs to use:
column the data is work needed
ready crossing-angle HepMC3 on /volatile none — feed straight to ddsim
(R) only in RECO, as MCParticles needs the edm2hepmc tool; the crossing angle is already applied (section 5.6)
(h) only head-on HepMC3 needs abconv first — head-on beams cannot go into ddsim (section 4)
(L) only in LUND or eic-smear format needs a format conversion (section 5.5)
A process is counted once per energy, under its best available source.

column	the data is	work needed
`ready`	crossing-angle HepMC3 on `/volatile`	none — feed straight to `ddsim`
`(R)`	only in `RECO`, as `MCParticles`	needs the `edm2hepmc` tool; the crossing angle is already applied (section 5.6)
`(h)`	only head-on HepMC3	needs `abconv` first — head-on beams cannot go into `ddsim` (section 4)
`(L)`	only in LUND or eic-smear format	needs a format conversion (section 5.5)

energy	processes, any source	ready	`(R)` RECO only	`(h)` head-on only	`(L)` convert	mixable
10x100	29	27	—	1	1	yes
18x275	27	22	—	4	1	yes
5x41	20	18	—	1	1	—
9x275	15	15	—	—	—	—
9x130	12	12	—	—	—	—
10x130	10	10	—	—	—	yes
9x100	8	8	—	—	—	—
18x110	8	4	1	3	—	—
5x100	7	6	—	1	—	—
10x275	4	4	—	—	—	yes
18x108.41	2	2	—	—	—	—
18x100	1	—	—	—	1	—
18x137	1	—	—	1	—	—
?	1	—	—	1	—	—
total cells	145	128	1	12	4

? is EXCLUSIVE/UPSILON/v1 — a legacy Υ photoproduction sample whose only recorded beam is a ~0.4 GeV virtual photon, so no configuration can be read off it. 18x137 is eD (erhic-cc-yrad-floff_eD_18_137_*), 18x100 is the eic-smear-format SIDIS/pythia6/ep_18x100 set, and 18x108.41 is the BeAGLE ePb production.

Against 8.1 the ordering barely moves: 10x100 and 18x275 stay first and second and stay the only broad-and-mixable pair. What the extra sources buy is small and concentrated — 18x110 nearly doubles (4 → 8) because the Sartre eA head-on samples and the eHe3 RECO branch live there, and 18x275 gains 5 (pure TCS, u-channel ω, eRHIC-format CC DIS, the coherent DVCS+BH+INT file). Neither changes which energy to start at.

ePIC generator data survey ​

0. Sources ​

1. Marco's ML class table (slide 12, Machine Learning Classes) ​

2. Machine background sources (slide 10, Signal and Background Mixing) ​

3. generatorStatus encoding (slide 11) ​

4. Afterburner state: crossing angle vs head-on ​

4.1 Naming conventions tested against measurement ​

4.2 Measurement method ​

4.3 Directory census by afterburner naming ​

5. Survey of available data ​

5.1 Energies with a complete background recipe ​

Mixing spec per energy and source ​

Expected events per 2 µs frame ​

Beampipe coating: rates and campaign history ​

5.2 Process catalogue ​

5.3 Processes with no row in Marco's table ​

5.4 Per-energy detail ​

10x100 — mixable (background recipe complete) ​

10x130 — mixable (background recipe complete) ​

10x275 — mixable (background recipe complete) ​

18x275 — mixable (background recipe complete) ​

9x100 — no background recipe yet ​

9x130 — no background recipe yet ​

9x275 — no background recipe yet ​

5x41 — background recipe blocked (3 files missing) ​

5x100 — background recipe blocked (3 files missing) ​

18x108.41 (ePb) ​

18x110 (eAu) ​

Energy not determined (eD, beams did not parse) ​

5.5 Usability caveats per branch ​

Head-on branches requiring abconv ​

Non-HepMC formats ​

One directory is mislabelled ​

5.6 Generator events recovered from FULL / RECO ​

Background-free RECO coverage ​

5.7 Recommendations and open requests ​

6. Access verified 2026-10-15 ​

7. Appendix: reproducing the sweep ​

7.1 Inventory dumps ​

7.2 Afterburner measurement ​

7.3 Reading MCParticles out of a RECO file ​

8. Processes with crossing-angle HepMC3 data ​

8.1 Processes per beam energy ​

8.2 Process catalogue, crossing-angle data only ​

8.3 Coverage including degraded sources ​

8.4 Processes per beam energy, all sources ​