Difference between revisions of "HDGeant4 Meeting, March 12, 2019"

From GlueXWiki
Jump to: navigation, search
(→‎Agenda)
(Added the minutes.)
 
(One intermediate revision by the same user not shown)
Line 19: Line 19:
 
#** [https://halldweb.jlab.org/wiki/images/3/35/G3_G4_combocontruction.png Combo construction]: timing
 
#** [https://halldweb.jlab.org/wiki/images/3/35/G3_G4_combocontruction.png Combo construction]: timing
 
# Written Report ([https://docs.google.com/document/d/1v7gy5qhHrcdMTeGiWu6aqPxoYKZI7K41o-chk-kwcZo/edit?usp=sharing outline])
 
# Written Report ([https://docs.google.com/document/d/1v7gy5qhHrcdMTeGiWu6aqPxoYKZI7K41o-chk-kwcZo/edit?usp=sharing outline])
 +
# [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=3852 FCAL: comparing G3 and G4] (Richard)
 
# [https://github.com/JeffersonLab/HDGeant4/issues Issues on GitHub] (all)
 
# [https://github.com/JeffersonLab/HDGeant4/issues Issues on GitHub] (all)
 
# [https://github.com/JeffersonLab/HDGeant4/pulls Pull Requests on GitHub] (all)
 
# [https://github.com/JeffersonLab/HDGeant4/pulls Pull Requests on GitHub] (all)
 +
 +
== Minutes ==
 +
 +
Present:
 +
* ''' Glasgow: ''' Peter Pauli
 +
* ''' JLab: ''' Shankar Adhikari, Alex Austregesilo, Thomas Britton, Stuart Fegan, Mark Ito (chair), Simon Taylor, Beni Zihlmann
 +
* ''' UConn: ''' Richard Jones
 +
 +
There is a [https://bluejeans.com/s/1aas1/ recording of the meeting on the BlueJeans site]. Use your JLab credentials to access it.
 +
 +
=== Review of minutes from the last HDGeant4 ===
 +
 +
We reviewed [[HDGeant4_Meeting,_February_26,_2019#Minutes|the minutes from February 26]]. Simon continues to work on single-particle gun comparisons of HDG3 vs. HDG4.
 +
 +
=== Discussion/proposal_for_making_transition_to_G4 ===
 +
 +
Mark led us through the [[GlueX_Software_Meeting,_March_5,_2019#Discussion.2Fproposal_for_making_transition_to_G4|discussion from the last Software Meeting]] on a policy for making the transition to HDG4. He concludes that there was no consensus on what the policy should be. Since studies and comparisons are going forward in any case, he proposed, and the group agreed, that we drop the idea of a formal policy. As a corollary, we are dropping the idea of a document to support collaboration-wide adoption of HDG4; the recommendation that we do so has essentially been rejected, at least for now.
 +
 +
=== G3/G4 comparison for ρ(770) cross section: after ST fix, without trigger emulation ===
 +
 +
Alex repeated his study from two weeks ago with a fix to the start counter geometry and with no trigger emulation as suggested last time. His presentation starts at 12:50 in the recording.
 +
 +
He first showed the [https://halldweb.jlab.org/wiki/images/9/9d/Accpentance_g3_g4_notrigger.pdf overall acceptance as a function of energy]. Removing the trigger emulation had almost no effect. The disagreement at low energy remains.
 +
 +
Next he showed a selection from the [https://halldweb.jlab.org/talks/2019/Independent_rho_G3_G4_noTrigger.pdf large number of comparison plots he produced]. He highlighted the following areas (with page numbers):
 +
 +
# FDC Cathode (32.54)
 +
# B/FCal Shower Energy (76/77, 308/311)
 +
# Timing of charged tracks in FCal/BCal (866/873)
 +
# Tracking below 12 degrees (1041)
 +
 +
In general things look more similar than they did in the previous version of the the study.
 +
 +
* The number of FDC hits on tracks shows a significant difference, 10-20% more hits in the HDG4 simulation. Richard thought this might be due to difference in energy deposition physics in G3 vs. G4.
 +
* The time of charged particles measured in the BCAL vs the real time has a high-side tail much larger for HDG3 than for HDG4. Since these may be hadronic showers, Richard stated that neither is to be trusted.
 +
* Alex showed a [https://halldweb.jlab.org/wiki/images/3/35/G3_G4_combocontruction.png plot showing were events are cut in the analysis chain]. In particular, there is cut that requires all charged particles to agree on the RF bunch that caused the event. The plot shows a much larger number of events eliminated by the timing agreement cut for HDG3 than for HDG4 about a factor of two. This may be related to the BCAL timing effecting the proton time.
 +
** Richard will look at the these BCAL timing distributions in real data. Perhaps neither simulation does a good job and we should smear this quantity with distributions derived from real data.
 +
 +
=== FCAL: comparing G3 and G4 ===
 +
 +
Richard gave [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=3852 an update on the energy response of FCAL], comparing HDG3 and HDG4.
 +
* The work had benefited from interactions with the Calorimeter Group.
 +
* He compares shower energy, after full simulation, smearing, reconstruction, and non-linearity correction.
 +
* The comparison is done in slices of polar angle.
 +
* Agreement between HDG3 and HDG4 is very good.
 +
* He sees a significant remaining non-linearity, both at low energy (below 1 GeV) and at high energy (above 4 GeV). The character of the non-linearity for HDG3 is the same as that for HDG4.
 +
He did not show the corresponding BCAL study but told us that the agreement is goo there as well. He now plans to start looking at timing in the calorimeters.
 +
 +
=== Review of Issues and Pull Requests ===
 +
 +
Jon Zarling reported an issue, [https://github.com/JeffersonLab/HDGeant4/issues/94 Photon Gun Sample Changes Behavior after ~500,000 events? #94] where photons from the gun do not escape the target deep into the HDGeant4 run. Naomi Jarvis was able to reproduce it. The problem was real, Richard got to the bottom of it, it has been reported to the Geant4 Collaboration, and a fix implemented in HDGeant4. For details of the bug, see Issue #94 on GitHub.

Latest revision as of 13:08, 17 March 2019

HDGeant4 Meeting
Tuesday, March 12, 2019
3:00 pm
JLab: CEBAF Center, A110
BlueJeans: 968592007

Agenda

  1. Review of minutes from the last HDGeant4
  2. Discussion/proposal_for_making_transition_to_G4 from last Software Meeting
  3. Do-over on comparison studies
    • G3/G4 comparison for ρ(770) cross section (Alex): after ST fix, without trigger emulation
      • Acceptance
      • Full Comparison with biggest offenders:
        • FDC Cathode (32.54)
        • B/FCal Shower Energy (76/77, 308/311)
        • Timing of charged tracks in FCal/BCal (866/873)
        • Tracking below 12 degrees (1041)
      • Combo construction: timing
  4. Written Report (outline)
  5. FCAL: comparing G3 and G4 (Richard)
  6. Issues on GitHub (all)
  7. Pull Requests on GitHub (all)

Minutes

Present:

  • Glasgow: Peter Pauli
  • JLab: Shankar Adhikari, Alex Austregesilo, Thomas Britton, Stuart Fegan, Mark Ito (chair), Simon Taylor, Beni Zihlmann
  • UConn: Richard Jones

There is a recording of the meeting on the BlueJeans site. Use your JLab credentials to access it.

Review of minutes from the last HDGeant4

We reviewed the minutes from February 26. Simon continues to work on single-particle gun comparisons of HDG3 vs. HDG4.

Discussion/proposal_for_making_transition_to_G4

Mark led us through the discussion from the last Software Meeting on a policy for making the transition to HDG4. He concludes that there was no consensus on what the policy should be. Since studies and comparisons are going forward in any case, he proposed, and the group agreed, that we drop the idea of a formal policy. As a corollary, we are dropping the idea of a document to support collaboration-wide adoption of HDG4; the recommendation that we do so has essentially been rejected, at least for now.

G3/G4 comparison for ρ(770) cross section: after ST fix, without trigger emulation

Alex repeated his study from two weeks ago with a fix to the start counter geometry and with no trigger emulation as suggested last time. His presentation starts at 12:50 in the recording.

He first showed the overall acceptance as a function of energy. Removing the trigger emulation had almost no effect. The disagreement at low energy remains.

Next he showed a selection from the large number of comparison plots he produced. He highlighted the following areas (with page numbers):

  1. FDC Cathode (32.54)
  2. B/FCal Shower Energy (76/77, 308/311)
  3. Timing of charged tracks in FCal/BCal (866/873)
  4. Tracking below 12 degrees (1041)

In general things look more similar than they did in the previous version of the the study.

  • The number of FDC hits on tracks shows a significant difference, 10-20% more hits in the HDG4 simulation. Richard thought this might be due to difference in energy deposition physics in G3 vs. G4.
  • The time of charged particles measured in the BCAL vs the real time has a high-side tail much larger for HDG3 than for HDG4. Since these may be hadronic showers, Richard stated that neither is to be trusted.
  • Alex showed a plot showing were events are cut in the analysis chain. In particular, there is cut that requires all charged particles to agree on the RF bunch that caused the event. The plot shows a much larger number of events eliminated by the timing agreement cut for HDG3 than for HDG4 about a factor of two. This may be related to the BCAL timing effecting the proton time.
    • Richard will look at the these BCAL timing distributions in real data. Perhaps neither simulation does a good job and we should smear this quantity with distributions derived from real data.

FCAL: comparing G3 and G4

Richard gave an update on the energy response of FCAL, comparing HDG3 and HDG4.

  • The work had benefited from interactions with the Calorimeter Group.
  • He compares shower energy, after full simulation, smearing, reconstruction, and non-linearity correction.
  • The comparison is done in slices of polar angle.
  • Agreement between HDG3 and HDG4 is very good.
  • He sees a significant remaining non-linearity, both at low energy (below 1 GeV) and at high energy (above 4 GeV). The character of the non-linearity for HDG3 is the same as that for HDG4.

He did not show the corresponding BCAL study but told us that the agreement is goo there as well. He now plans to start looking at timing in the calorimeters.

Review of Issues and Pull Requests

Jon Zarling reported an issue, Photon Gun Sample Changes Behavior after ~500,000 events? #94 where photons from the gun do not escape the target deep into the HDGeant4 run. Naomi Jarvis was able to reproduce it. The problem was real, Richard got to the bottom of it, it has been reported to the Geant4 Collaboration, and a fix implemented in HDGeant4. For details of the bug, see Issue #94 on GitHub.

Retrieved from "https://halldweb1.jlab.org/wiki/index.php?title=HDGeant4_Meeting,_March_12,_2019&oldid=91808"