Difference between revisions of "GlueX Offline Meeting, September 3, 2014"

From GlueXWiki
Jump to: navigation, search
(Agenda)
m (Text replacement - "www/halldweb1/" to "www/halldweb/")
 
(12 intermediate revisions by 2 users not shown)
Line 9: Line 9:
 
# [[GlueX-Collaboration-Oct-2014|Collaboration Meeting]]
 
# [[GlueX-Collaboration-Oct-2014|Collaboration Meeting]]
 
# Review of [[GlueX Offline Meeting, August 20, 2014#Minutes|minutes from August 20]] (all)
 
# Review of [[GlueX Offline Meeting, August 20, 2014#Minutes|minutes from August 20]] (all)
 +
# [https://mailman.jlab.org/pipermail/halld-offline/2014-August/001771.html Named Pipes]
 
# Tagger Simulation Update (Richard)
 
# Tagger Simulation Update (Richard)
# [https://mailman.jlab.org/pipermail/halld-offline/2014-August/001759.html Re-Creating Tracks] (all)
+
#* [https://mailman.jlab.org/pipermail/halld-offline/2014-September/001789.html email message]
# Commissioning Geometry Simulation (Sean, Simon)
+
#* [http://clasweb.jlab.org/websvn/prod/listing.php?repname=GlueX&path=%2F&rev=14984&sc=1 changes to trunk]
# [https://halldweb1.jlab.org/talks/2014-3Q/SixFinalStates.pdf Analysis of six final states using data challenge 2 MC] (Ryan)
+
# Simulating the Commissioning Geometry (Simon)
 +
# Comparing Simulated HDDM and EVIO files (Sean) [https://halldweb.jlab.org/wiki/images/7/7e/Hddm_evio_comp.pdf slides]
 
# BCAL Time Reconstruction Update
 
# BCAL Time Reconstruction Update
# Data Challenge 3 (Mark)
+
# BCAL Energy Reconstruction (Mark D.)
# Action Items
+
# Data Challenge 3 (Mark I.)
 +
# Software Package Versioning Control (Mark I.)
 +
# Meta-data catalog
 +
# [https://mailman.jlab.org/pipermail/halld-offline/2014-August/001785.html Requirements Scrubbing]
 +
# Action Item Review
  
 
=Communication Information=
 
=Communication Information=
Line 26: Line 32:
 
==Slides==
 
==Slides==
  
Talks can be deposited in the directory <code>/group/halld/www/halldweb1/html/talks/2014-3Q</code> on the JLab CUE. This directory is accessible from the web at https://halldweb1.jlab.org/talks/2014-3Q/ .
+
Talks can be deposited in the directory <code>/group/halld/www/halldweb/html/talks/2014-3Q</code> on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2014-3Q/ .
  
 
=Minutes=
 
=Minutes=
 +
 +
Present:
 +
* '''CMU''': Paul Mattione, Curtis Meyer
 +
* '''IU''': Kei Moriya, Matt Shepherd
 +
* '''JLab''': Alex Barnes, Mark Dalton, Mark Ito (chair), David Lawrence, Will McGinley, Elton Smith, Mike Staib, Simon Taylor
 +
* '''MIT''': Justin Stevens
 +
* '''NU''': Sean Dobbs
 +
 +
==Collaboration Meeting Agenda==
 +
 +
We took a look at the [[GlueX-Collaboration-Oct-2014|Collaboration Meeting page]]. There are nominally 4 30-min slots for the offline starting at 4:00 pm Friday. Mark will make a proposal.
 +
 +
David suggested that part of someone's presentation should include a list of software projects so that collaborators can volunteer themselves or others for needed tasks.
 +
 +
==Review of Minutes from the Last Meeting==
 +
 +
We went over the [[GlueX Offline Meeting, August 20, 2014#Minutes|minutes from the August 20 meeting]].
 +
 +
In the context of Ryan's presentation on multi-pion final states, Matt commented that he, Ryan, and Paul did a detailed comparison of Ryan's software and Paul's analysis library, down to comparing single events. With a lot of back and forth to clarify/resolve differences the agreement between the two is "perfect in some very difficult to reproduce, slightly obscure way".
 +
 +
==Tagger Simulation Update==
 +
 +
Richard Jones was in transit to JLab with the tagger microscope and could not participate in the meeting, but we noted [https://mailman.jlab.org/pipermail/halld-offline/2014-September/001789.html his email] announcing the merge from his development branch onto the trunk and looked at the [http://clasweb.jlab.org/websvn/prod/listing.php?repname=GlueX&path=%2F&rev=14984&sc=1 long list of files] changed on the trunk.
 +
 +
Sean did some checking with the new code; see his report below.
 +
 +
==Simulating the Commissioning Geometry==
 +
 +
Simon is done with putting in the geometry changes associated with the commissioning configuration. He as not done a check for overlaps yet.
 +
 +
Sean is putting together a set of configuration files do drive large-scale simulation using this geometry. Simon has done some tests as well.
 +
 +
==Comparing Simulated HDDM and EVIO Files==
 +
 +
Sean has done comparisons between hit information between simulation native output (HDDM format) and EVIO data derived from same using the new trunk from Richard (see above). In most cases the agreement is very good. An exception is the number of tracks at the wire-based stage which shows about twice as many tracks for EVIO. This needs to be tracked down.
 +
 +
For the complete set of comparisons see [https://halldweb.jlab.org/wiki/images/7/7e/Hddm_evio_comp.pdf Sean's slides].
 +
 +
We discussed a feature of the HDDM to EVIO conversion where a 100 ns offset is added to all times so that they come out positive (and thus TDC-like). Richard added an offset to each of the appropriate "hit" factories in sim-recon to take out this offset to restore times to their original values. We discussed alternate solutions to this problem, including adding a global offset or putting values in the CCDB for the Monte Carlo variation (which Sean has already implemented), but decided not to decide until Richard can be included in the discussion.
 +
 +
==Calibration Constant Style==
 +
 +
Mark D. raised the issue of a policy on how we implement constants for multi-channel systems. There are two styles being used:
 +
 +
# One global factor to get all channels "into the ballpark" and channel-by-channel factors for fine adjustment.
 +
# Channel-by-channel factors only.
 +
 +
(Not all constants are factors, but this exemplifies the issue.)
 +
 +
Elton argued that style (1) leads to confusion in the long term since the "true meaning" of the global factor can change from generation to generation of calibrators, making historical comparisons difficult. Mark D. pointed out that for the BCAL, since different layers have different "ballpark" factors, style (1) is unwieldy.
 +
 +
Discussed whether or not we should have a uniform style across all detector systems. In the end, we decided that each detector system should be able to choose its own style as appropriate. The main argument was that either style is workable and we already have a mix of both as one goes from system to system.
 +
 +
==Refrain from Excessive Use of Cosmetics==
 +
 +
David strongly encouraged us to try and preserve the original style of code as much as possible. In particular, indentation, spacing conventions, line-break choices should be left in their as-found state. Wholesale changes that have no functional effect defeat the "blame" feature of Subversion, where one can obtain a line-by-line report of the last person to change the code. Purely cosmetic changes cause gratuitous blame shifting.
 +
 +
==Action Items==
 +
 +
# Propose an agenda for the Collaboration Meeting. (Mark)
 +
# Track down factor of two in wire-based track multiplicity between "HDDM" and "EVIO".
 +
# Decide on how to handle the 100 ns EVIO conversion shift.

Latest revision as of 10:35, 31 March 2015

GlueX Offline Software Meeting
Wednesday, September 3, 2014
1:30 pm EDT
JLab: CEBAF Center F326/327

Agenda

  1. Announcements
  2. Collaboration Meeting
  3. Review of minutes from August 20 (all)
  4. Named Pipes
  5. Tagger Simulation Update (Richard)
  6. Simulating the Commissioning Geometry (Simon)
  7. Comparing Simulated HDDM and EVIO files (Sean) slides
  8. BCAL Time Reconstruction Update
  9. BCAL Energy Reconstruction (Mark D.)
  10. Data Challenge 3 (Mark I.)
  11. Software Package Versioning Control (Mark I.)
  12. Meta-data catalog
  13. Requirements Scrubbing
  14. Action Item Review

Communication Information

Remote Connection

Slides

Talks can be deposited in the directory /group/halld/www/halldweb/html/talks/2014-3Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2014-3Q/ .

Minutes

Present:

  • CMU: Paul Mattione, Curtis Meyer
  • IU: Kei Moriya, Matt Shepherd
  • JLab: Alex Barnes, Mark Dalton, Mark Ito (chair), David Lawrence, Will McGinley, Elton Smith, Mike Staib, Simon Taylor
  • MIT: Justin Stevens
  • NU: Sean Dobbs

Collaboration Meeting Agenda

We took a look at the Collaboration Meeting page. There are nominally 4 30-min slots for the offline starting at 4:00 pm Friday. Mark will make a proposal.

David suggested that part of someone's presentation should include a list of software projects so that collaborators can volunteer themselves or others for needed tasks.

Review of Minutes from the Last Meeting

We went over the minutes from the August 20 meeting.

In the context of Ryan's presentation on multi-pion final states, Matt commented that he, Ryan, and Paul did a detailed comparison of Ryan's software and Paul's analysis library, down to comparing single events. With a lot of back and forth to clarify/resolve differences the agreement between the two is "perfect in some very difficult to reproduce, slightly obscure way".

Tagger Simulation Update

Richard Jones was in transit to JLab with the tagger microscope and could not participate in the meeting, but we noted his email announcing the merge from his development branch onto the trunk and looked at the long list of files changed on the trunk.

Sean did some checking with the new code; see his report below.

Simulating the Commissioning Geometry

Simon is done with putting in the geometry changes associated with the commissioning configuration. He as not done a check for overlaps yet.

Sean is putting together a set of configuration files do drive large-scale simulation using this geometry. Simon has done some tests as well.

Comparing Simulated HDDM and EVIO Files

Sean has done comparisons between hit information between simulation native output (HDDM format) and EVIO data derived from same using the new trunk from Richard (see above). In most cases the agreement is very good. An exception is the number of tracks at the wire-based stage which shows about twice as many tracks for EVIO. This needs to be tracked down.

For the complete set of comparisons see Sean's slides.

We discussed a feature of the HDDM to EVIO conversion where a 100 ns offset is added to all times so that they come out positive (and thus TDC-like). Richard added an offset to each of the appropriate "hit" factories in sim-recon to take out this offset to restore times to their original values. We discussed alternate solutions to this problem, including adding a global offset or putting values in the CCDB for the Monte Carlo variation (which Sean has already implemented), but decided not to decide until Richard can be included in the discussion.

Calibration Constant Style

Mark D. raised the issue of a policy on how we implement constants for multi-channel systems. There are two styles being used:

  1. One global factor to get all channels "into the ballpark" and channel-by-channel factors for fine adjustment.
  2. Channel-by-channel factors only.

(Not all constants are factors, but this exemplifies the issue.)

Elton argued that style (1) leads to confusion in the long term since the "true meaning" of the global factor can change from generation to generation of calibrators, making historical comparisons difficult. Mark D. pointed out that for the BCAL, since different layers have different "ballpark" factors, style (1) is unwieldy.

Discussed whether or not we should have a uniform style across all detector systems. In the end, we decided that each detector system should be able to choose its own style as appropriate. The main argument was that either style is workable and we already have a mix of both as one goes from system to system.

Refrain from Excessive Use of Cosmetics

David strongly encouraged us to try and preserve the original style of code as much as possible. In particular, indentation, spacing conventions, line-break choices should be left in their as-found state. Wholesale changes that have no functional effect defeat the "blame" feature of Subversion, where one can obtain a line-by-line report of the last person to change the code. Purely cosmetic changes cause gratuitous blame shifting.

Action Items

  1. Propose an agenda for the Collaboration Meeting. (Mark)
  2. Track down factor of two in wire-based track multiplicity between "HDDM" and "EVIO".
  3. Decide on how to handle the 100 ns EVIO conversion shift.