DD4hep Developers Meeting

Thursday 22 Oct 2020, 15:30 → 17:00 Europe/Zurich

2/1-034 (CERN)

Marko Petric (CERN)

Please Join DD4hep developer's meeting via Vidyo

DD4hep meeting 22/10/2020
-------------------------

A.Sailer, C.Vuosalo, I.Osborne, M.Frank, S.Ko, V.Voelkl

1) New guest Sang Hyun Ko from Seoul University. 
   He is working for FCC on simulation of the dual readout calorimeter.
   In the presentation he introduced us into his working environment
   using dd4hep and Geant4.
   - He is very happy with the optical surfaces and it's properties.
   - He spotted a setup problem for setting Birk's constant to the material properties.
   - He also noted that there is no analogy to G4PVParametrized. It required extensive use
     of segmentations to overcome this deficiency.
   - Using colors in DDG4 can end up using significant CPU overhead.
     Need to turn off vis attributes during simulation.
   - Adding many daughters to a volume is very time consuming up to non-usability.
     Actually strcmp is getting significant.

2) Closed issues:
  -- Help for beginners (Andre/Marko)
     PR   https://github.com/AIDASoft/DD4hep/issues/703
     Python on MAC was a major issue for this problem. 
     Ianna will make a recipe public to sort out this issue. 
     Closed

3) Open issues:
  -- Unit system: Do we want to use Geant4 units?

4) Round table
  Ianna:  - Cannot process XML files in parallel, because they are interlinked.
            Hence, the TBB dependency is removed from DD4hep.
          - They see some problems with reflections.
          - There are some hotspots in the xml handling at the level of string handling.
  Carl:     --
  Andre:  - Started with the python mismatch will update ci.
  Markus: - Looking into new browser based EVE.
            Started to look into reflections issue.
  Sang Hyun: See comments to the presentation
  Valentin:  --

5) AOB
  None


Next meeting in 2 weeks: 05/11/2020

 

15:30 → 16:00 DD4hep migration of IDEA dual-readout calorimeter

Experience of porting the IDEA dual-readout calorimeter to DD4hep. Repeat of a presentation at the FCC SW meeting
https://indico.cern.ch/event/960882/

Speaker: Sang Hyun Ko (Seoul National University (KR))

201009_shKo_dd4hep.pdf

• Issue 1 (Birk's constant): Setting Birk's constant is a bit different from other material properties in GEANT4 (G4MaterialPropertiesTable) - Birk's constant is set for another material property of G4IonisParamMat.
• Issue 2 (Parameterised volume): unfortunately, since TGeo (which DD4hep depends on) does not support it, it would not be possible to implement unless having dedicated method in DD4hep for placing volumes with parameterisation matrix.
• Issue 3 (setVisAttributes): potential bug of ROOT? (https://github.com/root-project/root/blame/331efa4c00fefc38980eaaf7b41b8e95fcd1a23b/core/base/src/TColor.cxx#L1966). It should be not necessary to create a new instance when the same colour already described in a compact file.
• Issue 4 (std::strcmp): simply wrapping _addNode would not be possible since TGeoVolume->_addNode returns void, still may be resolved by having an index to the nodes (using the size of the container).