GlueX Offline Meeting, June 15, 2018

GlueX Offline Software Meeting
Friday, June 15, 2018
10:00 am EDT
JLab: CEBAF Center A110
BlueJeans: 968 592 007

Agenda

1. Announcements
   1. Upgrade of OSG Software on scosg16
   2. Software Review will be November 27-28
   3. This meeting ends at 11 am.
2. Review of minutes from the June 1 meeting (all)
3. Splitting up sim-recon into sim and recon (Sean, Mark)
4. Computing Model (David)
5. Running on the OSG (all)
6. Review of recent pull requests (all)
7. Review of recent discussion on the GlueX Software Help List (all)
8. Action Item Review (all)

Communication Information

Remote Connection

• The BlueJeans meeting number is 968 592 007 .
• Join the Meeting via BlueJeans

Slides

Talks can be deposited in the directory /group/halld/www/halldweb/html/talks/2018 on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2018/ .

Minutes

Present:

• FSU: Sean Dobbs
• JLab: Alex Austregesilo, Thomas Britton, Mark Ito (chair), David Lawrence, Simon Taylor, Beni Zihlmann
• Regina: Zisis Papandreou
• Yerevan: Hrach Marukyan

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

Announcements

1. Upgrade of OSG Software on scosg16. The upgrade went without incident on June 6. Thomas noticed that condor_q is different now.
2. Software Review will be November 27-28. Rolf Ent announced the date at the last Physics Senior Staff Meeting at JLab. No other details are known.

Review of Minutes from the June 1 Meeting

We went over the minutes.

Comparisons of HDGeant4 and HDGeant

Sean reported on studies he is doing with a student at FSU comparing various quantities between the HDGeant4 (Geant4) and HDGeant (GEANT3) versions of our simulation. They are using the Python HDDM interface. They have looked at:

1. Difference in CDC hits. Overall efficiency is the same. There may be some spurious hits in G3
2. Big differences are seen in TOF hit times and energies.

Launch Parameter Page

Alex reported that most of the recent launch links are OK now.

Splitting up sim-recon into sim and recon

Mark went over further thinking and development on the split of sim-recon. The wiki page is reproduced here:

• Process has been automated and thus documented.
• Testing: identical hdgeant_smeared.hddm generated from the standard b1pi test, 1000 events, from two systems: traditional and split
• Needed: build_scripts 2.0 to handle the two new repositories.
• Proposal, with details:
  1. Name new recon repository "gluex-recon"
     • Created by a fork of sim-recon
       • gluex-recon will have full history of sim-recon
       • sim-recon will live on as read-only repository to support past builds
  2. Name new sim repository "gluex-sim"
     • Created by git clone and git filter-branch
     • "recon" directories removed along with their history
  3. Deadline for changes to sim directories of sim-recon: June 29
  4. No deadline needed for recon directories of sim-recon
     • Redirecting existing sim-recon repositories to push to the new gluex-recon repository needs testing.

Computing Model

David presented slides using a new "Computing Model" software approach (a script, not a spreadsheet), previewing estimates of computing needs going forward. He also showed post-dictions of previous years and compared them with statistics we have collected. This work is in preparation for the readiness review for high-intensity GlueX running, coming up a week from Monday. See David's slides for the many details.

There is a large fraction of our past farm used due to individuals users that is not captured in the model. Some of this is calibration, and therefore of benefit for the collaboration. Although this activity has been accounted for in the past, We might have to think more about how to capture this load.

Reproducibility of Reconstruction Results

Beni reported that there are active efforts to run these anomalies to ground. He will include Mark in future emails to the group.