Generator
- physics:
- define samples:
- DBD
+ WIMP analysis samples (signal & background)?
+ Higgsinos? Jackies or Hales? - how about 6f? Not generated for DBD at 250 GeV, but processes like WWZ and vvWW etc should in principle be open?
- general purpose Bhabhas?
- question from Daniel: status of tau polarisation in Whizard2 => can be checked on existing small test samples
transverse tau polarisation: Akiya did some work, unclear status
- DBD
- Number of events for each processes. Especially large Xsect. processes, such as 2f, 1f, 3f.
- Radiative return to Z events. Generate separately or generate all process together.
- Naming convention of whizard2 samples and directory structure → ok up to 4f, iterate 6f
- define samples:
- general:
- verify batch mode
- tag Whizard version?
- ee:
- generate events
- 2f
- 4f
- 6f
- generate events
- ea / ae / aa:
- create lumi spectra and z-position of vertex distributions => done
- generate events!
- aa overlay : waiting for Tim... => remind....
- seeable pairs: fine, on Dirac
Simulation
- selection / adjustment of DD4HEP detector model
- use the existing ILD_l5_vo2 => no validation needed
- create ddsim steering / config files for 250 GeV
- z vertex distributions for ee / ea / ae / aa => Remi has put them in ILDConfig:
- verify that photon cluster position effect is not due to cell geometry problem => expected? Then calibration issue?
- simulate full pairs => done => tune BeamCal reco!
- need to create BeamCal bg-map for large (and small) 250 GeV models
- w/ and w/o anti-DID
- for standard reco use same as for 500 GeV (w/ anti-DID ?)
- simulate seeable pairs
- simulate aa overlay
- simulate single particles for calibration => verify that LumiCal / LHCal are covered!
Reconstruction
- solution unknown:
- muon reconstruction failure at costheta = 0.6 => needs 4 FTE weeks
- photon cluster position / angle bias => theta "just" a calibration/correction issue, phi unclear?
- photon energy calibration consistent with Pandora calibration
=> needs deep thinking on Pandora calibration strategy, Daniel will take a look once tautau benchmark is done - z0/d0 errors in fwd region => on a good way, fix requires time?
- solution / procedure known:
- tune BeamCal reconstruction → ongoing (Remi/Moritz)
- calibrate LumiCal & LHCal
- update parameters for beam spot constraint (LCFI)
- validation!
Production
- DISK SPACE ! (Cannot use tape back-end anymore) => estimate need: ~200 TB!
=> mc-opt3 currently is 600-700 TB!
But: reduce to one detector model => 350 TB
keep only 10% of REC files (but all SIM files) => 25% less ?
simulate only SiW Ecal + AHCal? - Save sim files of all events, but save REC files only for 10% of events
→ How to operate/select/control 10% files? # of split generator file base? Set additional meta key to save REC file.
→ Need to investigate DIRAC codes how to save only DST files w/o producing REC files in production jobs. - Required statistics : 2ab-1 for each channel?
→ Estimate required disk space and CPU times for required statistics.
→ How about the large cross-section channel? (2f, 4f)
→ How about the 6f samples? - update production scripts
- make sure whizard2 lcio file splitting works
- Update for new directory structure and file name conventions.
→ Revisit file name convention : "wizard2" as "w2"? - simulate aa_lowpt and seeable pair background files. Update scripts to 250 GeV background files.
- Need small whizard2 samples of various process type for development of production scripts.
- Save tar-gzipped log files on tape directory.
- File save location : DESY-SRM as primary and KEK-SRM as secondary, need to purchase TAPEs at KEK.
BG samples situation
Gen TDR
Gen Set A
SIM TDR
Sim Set A
aa_lowpt
Produced
Not yet (Tim)
Produced
Not yet
seeable pair
No
Produced (Mikael)
No
Not yet v02
IP smear
No
Apply
Solved issues (for new productions)