HDGeant4 Meeting, November 19, 2019

From GlueXWiki
Jump to: navigation, search

HDGeant4 Meeting
Tuesday, November 19, 2019
2:00 pm EST
JLab: CEBAF Center, A110
BlueJeans: 968 592 007


  1. Announcement: New version set: version_4.11.0.xml
  2. Review of minutes from the last Meeting (all)
  3. Comparison of geant3/geant4 for γ p → K+Σ0 (Nilanga)
  4. Issues on GitHub
  5. Pull Requests on GitHub
  6. Action Item Review



  • CMU: Naomi Jarvis
  • FSU: Sean Dobbs
  • JLab: Alex Austregesilo, Mark Dalton, Mark Ito (chair), Igal Jaegle, Keigo Mizutani, Simon Taylor, Beni Zihlmann
  • ODU: Nilanga Wickramaarachchi
  • UConn: Richard Jones

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


Mark announced a New version set: version_4.11.0.xml. This one incorporates the nuclear ions recently added by Richard.

Alex asked if the new ions definitions render old reconstruction launch halld_recon version incompatible with the latest halld_sim versions. Seems like they are. We will have to patch those halld_recon versions before they can be used with modern simulation.

Review of Minutes from the November 5 Meeting

We went over the minutes.

Visualization with OpenGL

Igal's problem was resolved by sitting down with Richard and adding a missing command option in his vis.mac.

Python Interface Problem

Igal, again working with Richard, got single-threaded builds of both geant4 and hdgeant4, as suggested at the last meeting, and had success using the Python interface.

Neutron Simulation

We reviewed Issue #125. Richard looked into Igal's suggestion from last time and found the "high precision" option in the documentation. Enabling it would like solve the problem for low energy neutrons (less than 20 MeV) but at an unknown cost in execution time. We will give it a try. There may be other less compute intensive solutions as well.

This problem does not occur in hadronic showers because of the "neutron killer" in Geant4. Initial state neutrons are not subject to getting killed.

Difference in Acceptance between G3 and G4

We reviewed Issue #111. Colin Gleason responded to Mark's email. He says that this is still a problem and he will try to repeat the study with a modern version of the software. He did report running into problem when trying with halld_sim 4.9.1.

Terminal cleanup?

We reviewed Issue #102. Mark pointed out that Richard has already completed this request. Mark will mark it as closed.

On a related note, Mark remarked that hdgeant4 is printing out the random number seeds for each event. He will look into removing -DVERBOSE_RANDOMS[?] from the compiler switches for hdg4, as per Naomi and Richard.

Calorimeter timing mismatch between g3 and g4

We reviewed Issue #93, having to do with charged particle timing measurements in our two calorimeters. Since the last meeting Alex did a three-way comparison of timing among G3, G4, and data. See the plots below.


We noted that the FCAL sample is unusual in that if the FCAL is being used for timing, that means that there was no TOF hit to use, which should not be the case most of the time.

There are significant tails being cut by the analysis library, as can be seen on the plots. This might explain differences in high-level efficiency for events with a particular topography, especially if these timing cuts are applied for multiple particles. The high-level efficiency studies should be repeated with wider time windows to see if the G3 vs. G4 comparison is affected.

Comparison of geant3/geant4 for γ p → K+Σ0

Nilanga showed comprehensive set of slides on acceptance for this channel comparing G3 with G4 and with data.

For the t-channel he showed

  • the t-distribution for
    • thrown events
    • events accepted in the G3 simulation
    • events accepted in the G4 simulation
    • acceptance for G3 and G4
    • accepted real data events
  • For single particles, he showed
    • 2-D plots of momentum vs. polar angle, for both thrown and reconstructed, and for both G3 and G4.
    • 1-D plots of momentum and polar angle, G3 and G4
    • 2-D plots of momentum vs. polar angle for data
    • all of these were shown for K+, π, p, and γ

Also all of these plots were shown for the u-channel as well. So that is a lot of plots.

Naomi (and others) noted that the following slide was especially interesting:

G3 vs. G4 features in Nilanga's analysis

She noted

The other pages showed generally fewer particles (pi-, p, gamma) reconstructed through G3 than G4 for the u-channel events - these have the backward low-momentum K+ but the general lack of other particles in a smoother way would be explained by this sharp absence of K+. K+ was the only particle with a dramatic dip. The data also show a dip in u-channel yield vs -u which matches the G3 plot, but not G4, so it looks like something is going awry with G4 somehow.

Richard will have a look at trying to reproduce these odd features with a particle gun.