DaMI metadata fields

Discussion

Pantelis Kontaxakis (University of Geneva), Juraj Smieško (CERN)

Training of the MC Production contacts

CERN, 13/3-005

11 November 2026

Current LFN schema

  • Directory
    
    /fcc/<machine>/<campaign>/<energy>/<process>/[<detector>/]<datatype>/<prodID>/
                        
    • Datatype
      • Generator level: stdhep, hepmc, lhef
      • Fast sim: delphes
      • Full sim: sim, rec
    • <detector> only for fast and full sim
      • KKMC has the generator there instead
  • File
    
    <prodID>/000/<process>_<datatype>_<prodID>_<task>.<ext>
                        
    • <datatype> matches the directory, except two hepmc productions with gen in the file name
  • EOS path: /eos/experiment/fcc/prod + LFN
    • For productions, the CERN EOS path will change to /eos/experiment/fcc/rucio

LFN examples

  • Generator level: file format as datatype, generator only for KKMC (without campaign)
    
    /fcc/ee/test_spring2024/240gev/ZH/stdhep/00016879/000/ZH_stdhep_16879_1.stdhep
    /fcc/ee/spring2024/50gev/nu/hepmc/00800000/000/nu_hepmc_00800000_0.hepmc
    /fcc/ee/winter2023/91.19gev/tautau/lhef/00016140/000/tautau_lhef_16140_1.lhe
    /fcc/ee/91.2gev/tautau/kkmcee/lhef/00015639/15639_weights_events_info.tar.gz
                        
  • Fast sim (Delphes): detector label and delphes
    
    /fcc/ee/summer2026/240gev/ZH/IDEA/delphes/00017123/000/ZH_delphes_17123_1.root
                        
  • Full sim (CLD): detector version and sim/rec, one production ID per step
    
    /fcc/ee/test_spring2024/240gev/ZH/CLD_o2_v05/sim/00016880/000/ZH_sim_16880_1.root
    /fcc/ee/test_spring2024/240gev/ZH/CLD_o2_v05/rec/00016881/000/ZH_rec_16881_1.root
                        
  • Energy written as 240gev, 91gev, 91.188gev, 91.2gev; some productions without campaign

Candidate metadata fields

Identification dataset name, campaign, accelerator, stage, detector, file type, energy group (Z, WW, ZH, Top)
Physics process description, centre-of-mass energy, generator and version, generator-level cuts
Normalisation cross section (unit, source), k-factor, matching efficiency, sum of weights
Statistics number of events, number of files, size
Provenance Key4hep release, generator card, detector configuration (repository + version), DRS request, transformation IDs, FCCDIRAC attempt directory, computing element(s)
Location LFN, storage elements (SEs) holding the dataset
People physics group, requester, MC contact, submitted by
Status validation, known issues, comment

To settle

  • Which fields are required, which are optional?
  • Names and units of the keys, e.g. cross section in pb
  • Where does the cross section come from, and who fills it in?
  • One record per process, or per process and step?
  • Validation: a simple status, or a reference to what was checked?
  • Regular checks of the records: what is checked, and how often?
  • Common process definitions, e.g. what “qq” means (uds, ud or udscb)

The MC contacts should keep the agreed fields up to date and compatible

Backup

LFN schema proposals


# A: generator in place of the detector at generator level
/fcc/<machine>/<campaign>/<energy>/<process>/<detector|generator>/<datatype>/<prodID>/

# B: generator and detector always present
/fcc/<machine>/<campaign>/<energy>/<process>/<generator>/<detector|none>/<datatype>/<prodID>/

# C: datatype first
/fcc/<machine>/<campaign>/<datatype>/<energy>/<process>/<detector|generator>/<prodID>/

# D: DaMI dataset name in place of the energy and process
/fcc/<machine>/<campaign>/<dataset>/<datatype>/<prodID>/
              
  • A: <datatype> = file format at generator level, stage (delphes/sim/rec) after it, EDM4hep implied
  • B: generator recorded for every datatype, fixed depth
  • C: one subtree per datatype, easier cleanup and quotas
  • D: path follows the DaMI record, e.g. p8_ee_ZH_ecm240
  • In any case: one energy format, one capitalisation of the detector labels