Difference between revisions of "HDGeant4 Meeting, December 4, 2018"

From GlueXWiki
Jump to: navigation, search
m (Marki moved page HDGeant4 Meeting, November 20, 2018 to HDGeant4 Meeting, December 4, 2018 without leaving a redirect: meeting delayed)
m (Issues on GitHub =)
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
HDGeant4 Meeting<br/>
 
HDGeant4 Meeting<br/>
Tuesday, November 20, 2018<br/>
+
Tuesday, December 4, 2018<br/>
 
3:00 pm<br/>
 
3:00 pm<br/>
 
JLab: CEBAF Center, A110<br/>
 
JLab: CEBAF Center, A110<br/>
Line 9: Line 9:
 
# [[HDGeant4_Meeting,_October_23,_2018#Minutes|Review of minutes from last time]]
 
# [[HDGeant4_Meeting,_October_23,_2018#Minutes|Review of minutes from last time]]
 
# [https://github.com/JeffersonLab/HDGeant4/issues Issues on GitHub] (all)
 
# [https://github.com/JeffersonLab/HDGeant4/issues Issues on GitHub] (all)
 +
 +
== Minutes ==
 +
 +
Present:
 +
* ''' FSU: ''' Sean Dobbs
 +
* ''' JLab: ''' Alex Austregesilo, Thomas Britton, Nacer Hamdi, Mark Ito (chair), Simon Taylor, Beni Zihlmann
 +
* ''' UConn: ''' Richard Jones
 +
* ''' W&M: ''' Justin Stevens
 +
 +
There is a [https://bluejeans.com/s/cheXd/ recording of this meeting] on the BlueJeans site. Use your JLab credentials.
 +
 +
=== Review of minutes from last time ===
 +
 +
We looked over [[HDGeant4_Meeting,_October_23,_2018#Minutes|the minutes from October 23]].
 +
* Beni's issue with hdgeant3 crashes is still out there.
 +
* Thomas re-iterated his offer to produce truth time distribution relevant to the [https://github.com/JeffersonLab/HDGeant4/issues/66 FDC issue].
 +
* Mark has not made progress on augmented Oasis distributions suitable for development.
 +
* Thomas reported that the problem he was seeing with blocked database server access from the OSG submit host (scosg16.jlab.org) was due to an overloaded network interface card (NIC) on the submit host. The load comes when shipping random trigger files to multiple sites when jobs are freshly submitted. The NIC was only rated at 1 Gb/s. It has since been replaced with two "bridged" 10 Gb/s cards; problem fixed. There was a residual problem with Condor giving non-unique job ID numbers, but that was resolved with a reboot of the host.
 +
 +
=== Issues on GitHub ===
 +
 +
We went over the [https://github.com/JeffersonLab/HDGeant4/issues?q=is%3Aissue list of issues on GitHub]. We skipped over those for which significant progress has not been made.
 +
 +
==== DIRC photon propagation time across boundaries ====
 +
 +
Justin originally submitted [https://github.com/JeffersonLab/HDGeant4/issues/65 this issue]. Cerenkov photon propagation times from points very close to each other, but on opposite sides of a volume boundary, were separated by way more than they should be. Richard traced the problem to a bug in Geant4 having to do with how parallel worlds are treated. See further discussion in [https://github.com/JeffersonLab/HDGeant4/pull/72 his pull request #72]. The bug fix has been submitted to the Geant4 collaboration and should appear in future releases.
 +
 +
==== New Issue: Merging in the DIRC, Software-Wise ====
 +
 +
Currently, to run the DIRC simulation and reconstruction, one must check out special versions of the code that resides on branches of four repositories: hdds, halld_recon, halld_sim, and hdgeant4. This was unavoidable given the nature of the changes. However, it presents a potential challenge when all of these changes are ready to merge onto their respective master branches. We need to look at the issue, see how much of a problem it really is, and make a plan. Mark agreed to open a new issue on this.

Latest revision as of 16:53, 14 December 2018

HDGeant4 Meeting
Tuesday, December 4, 2018
3:00 pm
JLab: CEBAF Center, A110
BlueJeans: 968592007

Agenda

  1. Review of minutes from last time
  2. Issues on GitHub (all)

Minutes

Present:

  • FSU: Sean Dobbs
  • JLab: Alex Austregesilo, Thomas Britton, Nacer Hamdi, Mark Ito (chair), Simon Taylor, Beni Zihlmann
  • UConn: Richard Jones
  • W&M: Justin Stevens

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

Review of minutes from last time

We looked over the minutes from October 23.

  • Beni's issue with hdgeant3 crashes is still out there.
  • Thomas re-iterated his offer to produce truth time distribution relevant to the FDC issue.
  • Mark has not made progress on augmented Oasis distributions suitable for development.
  • Thomas reported that the problem he was seeing with blocked database server access from the OSG submit host (scosg16.jlab.org) was due to an overloaded network interface card (NIC) on the submit host. The load comes when shipping random trigger files to multiple sites when jobs are freshly submitted. The NIC was only rated at 1 Gb/s. It has since been replaced with two "bridged" 10 Gb/s cards; problem fixed. There was a residual problem with Condor giving non-unique job ID numbers, but that was resolved with a reboot of the host.

Issues on GitHub

We went over the list of issues on GitHub. We skipped over those for which significant progress has not been made.

DIRC photon propagation time across boundaries

Justin originally submitted this issue. Cerenkov photon propagation times from points very close to each other, but on opposite sides of a volume boundary, were separated by way more than they should be. Richard traced the problem to a bug in Geant4 having to do with how parallel worlds are treated. See further discussion in his pull request #72. The bug fix has been submitted to the Geant4 collaboration and should appear in future releases.

New Issue: Merging in the DIRC, Software-Wise

Currently, to run the DIRC simulation and reconstruction, one must check out special versions of the code that resides on branches of four repositories: hdds, halld_recon, halld_sim, and hdgeant4. This was unavoidable given the nature of the changes. However, it presents a potential challenge when all of these changes are ready to merge onto their respective master branches. We need to look at the issue, see how much of a problem it really is, and make a plan. Mark agreed to open a new issue on this.