Geom workshop 151014

Room: Kelvin Building 506

Physics Block challenge 11.00 am

Issue #1552

  1. Simulation of Step IV on GRID - Bayes, Martyniak
  2. Analysis of GRID simulation data - Hunt
  3. Work required to converge - All

Geometry 14.00 pm

  1. CAD import geometry (Step IV), summary, status - Bayes
  2. Mice Modules geometry (Step IV), summary, status - Hunt
  3. Validation - Rogers
  4. Work required for conversion - All
    • Beamline, incl proton absorbers and beamline magnets
    • Cooling channel magnets
    • Diffuser
    • Detectors
    • Survey integration
    • Validation

Phone Line

  • UK: 0844 4 73 73 73
  • Switzerland: 0848 560 347
  • US: 1 415 363 0833 (or 1 646 652 0772)
  • Italy: 0821 230 749
  • Others listed below
  • Skype +49 1803 001 178

PIN: 746868 (PIMUNU)


Simulation of Step IV on GRID

  • Simulation of full beam propagated through MAUS 6, 200 MeV/c - DONE, see slides
    • Memory usage, seems okay
    • Disk usage, 3 GB for 100,000 muons at TOF1
    • Quadrupole currents massaged by hand
    • Cooling currents were maybe massaged by hand
      • CDB has CoolingChannel table for superconducting magnets
        SSD M1, M2, E1, E2, C
        SSU M1, M2, E1, E2, C
    • CDB has Beamline table for normal conducting magnets
      • This can be used in the geometry but was not
    • Geometry does come from CDB
      • Some corrections uploaded yesterday except asymmetric vacuum windows
    • Diffuser does come from CDB
    • Proton absorber hard coded into the G4BL deck
      • Proton absorber in/out comes from CDB
      • Proton absorber thickness comes from CDB
    • All fields come from CDB
    • script (or whatever) to return 0 if successful, take a
      few command line arguments; software must never write into MAUS/tmp or /tmp/;
      must only write into cwd
      • Note suds writes to /tmp

DECISION (REMINDER): Propose use negative run numbers and 1974 in the not-tagged beamline
DECISION (REMINDER): Propose use decimal integer representation for binary number between 0 and 15 inclusive for diffuser

ACTION: (Janusz/Hanlet) to modify CoolingChannel to use FC for Step IV magnets NOT CC - pass it by Pierrick and then hand to Janusz
ACTION: (Ryan) Ryan to modify code to use CoolingChannel table for Step IV magnets
ACTION: (Ryan) Ryan to modify diffuser code to use binary
ACTION: (MAUS Coordinator) investigate suds writing to /tmp/ - this may be illegal on GRID
ACTION: (Ryan) Ryan to check he is happy with Monte Carlo Serial Number then Janusz push to production
ACTION: (Rogers) Speak to Jason/Roy re: mysterious windows around AFC (in LiH mode)
ACTION: (Janusz) fix FTS transfer credentials issue with write-to-CASTOR
ACTION: (Ryan) script to drive monte carlo
ACTION: (Durga) new MAUS version including
ACTION: (Janusz) install new MAUS version to web
ACTION: (Janusz) trivial update to existing reco script to use


CAD geometry status

  • Model from Jason as of last July, with addendum in May
  • Some modifications
    • changing default material to vacuum in cooling channel
    • adding vacuum window
  • Geometry communicates with beamline and cooling channel components
  • Few modifications required
  • The CAD describes the material components of quads, diffuser, solenoids in cooling channel
  • The CAD does not describe detector elements e.g. TOF, KL, etc
    • Use GDML and convert to MICE Modules for geometry
  • The CAD does not include diffuser irises
  • CAD is described using tesselated solids, issues with respect to loading time
  • The fields are converted from CDB information, cooling channel positions, to MICE Modules
  • Cryostat volume is a cylinder with end caps, cryostat jacket, each copper coil, bobbin, flanges, lots of detail
  • Tracker coil misalignment in Hunt geometry arises due to
  • Note tracker position is probably nominal, not surveyed in

Hunt geometry status

  • Cryostat bodies are hollow shields may from thin steel
  • Windows are possibly correct
  • AFC windows are from legacy and wrong
  • AFC from legacy
  • No beamline
  • Missing everything outside of the tracker volume e.g. windows, TOFs, etc
  • Magnet volumes are approximate
  • "Cylinders with coils in"
  • Need "apertures and coils" level of fit
  • Note global reconstruction needs similar level of fit, but going out to TOF0 -> EMR
    Ryan Bayes is proposing either:
  • Maintain full CAD geometry and use GDML parser
  • From the CAD geometry, create fake object to mimic tesselated solid
GDML parser is almost ready for prime time
  • Overlapping geometry objects shaken out
  • Need to check
Issue: Systematics studies
  • Helium windows
  • Absorber thicknesses
  • Detector
    Are all handled manually

Issue: Step 3pi/2 geometry should use the same versioning system

Windows are done by tessellated solid
Apertures are done by tessellated solid

DECISION: We will use gdml going forward for both by-hand and automated geometry; work towards improved usability and greater validation of automated geometry, seeking feedback from users - an "orderly retreat" until we find a happy place where the users can use the geometry and the geometry folks can generate the geometry in a robust way

ACTION: (Rogers) Push validation further - field maps and beam optics
ACTION: (Rogers) Compare this geometry with that geometry #1567 DONE
  • geometry_validation 1D plot will show both geometries overlaid
  • geometry_validation 2D plot will show "Diffs"

ACTION: (Rogers) Calculate volumes and bounding boxes for each volume #1567 DONE

ACTION: (Bayes) Read gdml files from datacards - 40 hours
ACTION: (Bayes) Finalise a geometry which reads gdml files from datacards - 8 hours
ACTION: (Bayes) interface with download script - location dependence - 8 hours
ACTION: (Bayes) migrate gdml geometry objects into modules e.g. "beamline", "Q4", "Q5", etc, "SSU", "FCM", "SSD" - 80 hours
ACTION: (Bayes) identify gdml documentation for reference from MAUS should be linked into main documentation, document how auxiliaries works - 8 hours
ACTION: (Bayes) identify different physical objects by gdml file - 40 hours

ACTION: (Hunt) Migrate his geometry to gdml

ACTION: organise follow-up workshop in the new year, including consultation with analysis users

Detector construction:
  • GDML files need to be self contained
  • G4Detector volumes need to be built in the GDML itself, not using the existing C++
    • All SDs have been rewritten
  • ISSUE: Reopening geometry
  • User limits etc are handled by Auxiliary field

Updated by Rogers, Chris over 6 years ago ยท 22 revisions