ND Software Integration

Friday 6 Sep 2019, 14:00 → 15:45 US/Central

Connect via: https://fnal.zoom.us/j/100733281

14:00 → 14:20 Geometry Update

Speaker: Perri Zilberman

Slides 9-6-19_ND_Hall_Geometry_Update_Presentation_Perri_Zilberman.pdf

14:20 → 14:40 Software Discussion Points

Speaker: Dr Thomas Junk (Fermilab)

LArSoft Offline Leads meetings are held approximately once/month.  DUNE ND (Mat) should be part of the ongoing LArSoft disucssion.  At yesterday's meeting we discussed

• Software heirarchy:  An umbrella Redmine project should be created for integrated DUNE ND work, as a place to put the combined geometry files for example.  These can simply include component GDML files of the pieces that need also to be able to run by themselves.
• B-field maps:  Currently LArSoft has just a single constant B-field parameter, and it's set to zero.  GArSoft uses a copy of this feature and sets it to 0.5 T.  The actual field will be much more complex.  See Vladimir Kashikhin's talk at the DUNE ND magnet workshop which just wrapped up today:
  https://indico.fnal.gov/event/21272/session/0/contribution/1/material/slides/0.pdf
  and the workshop link at:
  https://indico.fnal.gov/event/21272/other-view?view=standard
  It's not too urgent for the physics of the CDR (or even the TDR), but we should convince ourselves that the fringe fields won't bend muons coming out of the liquid too far away from the MPD.  Mat asked if the field maps would be separate for the three detectors.  They haven't been sent to us yet, and they'd be preliminary maps anyway, just for practice, but they'd have to be separate in order for us to accommodate off-axis detector positions.

The GENIE version is of some concern.  The FD TDR was built using samples of GENIE v2_12_10 (or possibly earlier), and LArSoft has migrated to GENIE v3.x in the last few weeks on MicroBooNE's request.  DUNE discussed this with FD and long-baseline folks and physics coordinators, and we proposed that anyone doing future LBL work would use frozen LArSoft releases with GENIE 2.x as needed.  This is fine for the FD as the sim and reco algorithms are adequate for TDR-level studies.  But we need to develop new algorithms for ArgonCube and run them along with an older GENIE.  We get our GENIE set up through nutools.  In the future, a wrapper will decouple the versions so that we can flexibly mix and match LArSoft versions and GENIE versions, but this will start with GENIE v3.x.  To discuss.

LArSoft will migrate to GitHub away from Redmine, and move to a pull-request model.  Reviewers will have to inspect and approve or disapprove commits.  dunetpc will have to follow suit, and likely the ND code efforts too.

There is talk within SCD to replace the art framework with one that is a stitched version of CMSSW with art features that have been added over the years.  The idea is for SCD not to have to support both art and CMSSW.