Difference between revisions of "GlueX Offline Meeting, January 12, 2011"

From GlueXWiki
Jump to: navigation, search
(Telephone: new calling instructions)
m (Text replacement - "www/halldweb1/" to "www/halldweb/")
 
(15 intermediate revisions by 2 users not shown)
Line 1: Line 1:
 
GlueX Offline Software Meeting<br>
 
GlueX Offline Software Meeting<br>
Wednesday, January 11, 2011<br>
+
Wednesday, January 12, 2011<br>
 
1:30 pm EST<br>
 
1:30 pm EST<br>
 
JLab: CEBAF Center F326/7
 
JLab: CEBAF Center F326/7
Line 7: Line 7:
  
 
# Announcements
 
# Announcements
#* Collaboration Meeting Offline Agenda
+
#* [[GlueX-Collaboration-Jan-2011|Collaboration Meeting]] Offline Agenda
 
# Review of [[GlueX Offline Meeting, December 15, 2010#Minutes|minutes from the last meeting]]: all
 
# Review of [[GlueX Offline Meeting, December 15, 2010#Minutes|minutes from the last meeting]]: all
 
# Impressions from the Parallelism Workshop: David
 
# Impressions from the Parallelism Workshop: David
 
# IT Readiness for 12 GeV: Graham Heyes
 
# IT Readiness for 12 GeV: Graham Heyes
 
# Changes to TOF code: Beni
 
# Changes to TOF code: Beni
# Tracking Issues: Kei
+
# Tracking Issues: [https://halldweb.jlab.org/talks/2011-1Q/offline-software.Kei-Moriya.2011-01-12.pdf Kei]
 
# Action Item Review: all
 
# Action Item Review: all
 
## [http://tinyurl.com/2beqwlk Resolved issues]
 
## [http://tinyurl.com/2beqwlk Resolved issues]
Line 38: Line 38:
 
==Slides==
 
==Slides==
  
Talks can be deposited in the directory <code>/group/halld/www/halldweb1/html/talks/2010-12</code> on the JLab CUE <ref>You have to be a member of the "halld" Unix group to do this.</ref>. This directory is accessible from the web at https://halldweb1.jlab.org/talks/2010-12/ .
+
Talks can be deposited in the directory <code>/group/halld/www/halldweb/html/talks/2011-1Q</code> on the JLab CUE <ref>You have to be a member of the "halld" Unix group to do this.</ref>. This directory is accessible from the web at https://halldweb.jlab.org/talks/2011-1Q/ .
  
 
=Minutes=
 
=Minutes=
  
 
Present:
 
Present:
* '''IU''': Kei Moriya, Matt Shepherd
+
* '''FSU''': Nathan Sparks
* '''JLab''': Craig Bookwalter, Mark Ito (chair), David Lawrence, Lubomir Pentchev, Dmitry Romanov, Sascha Somov, Simon Taylor, Beni Zihlmann
+
* '''IU''': Jake Bennett, Ryan Mitchell, Kei Moriya, Matt Shepherd
 +
* '''JLab''': Hovanes Egiyan, Graham Heyes, Mark Ito (chair), David Lawrence, Yi Qiang, Dmitry Romanov, Simon Taylor, Beni Zihlmann
  
 
==Announcements==
 
==Announcements==
  
* Simon announced that he has added time and energy smearing to the start counter in mcsmear.
+
Mark encouraged people to start thinking about what talks they want to give at the [[GlueX-Collaboration-Jan-2011|upcoming Collaboration Meeting]] in the Offline session.
  
==Review of minutes from previous meeting==
+
==Parallelism Workshop==
  
We went over the [[GlueX Offline Meeting, December 15, 2010#Minutes|minutes from the December 15th meeting]].
+
David gave some impressions from last week's [http://conferences.jlab.org/Parallelism2011/ Workshop on Parallelism in Experimental Nuclear Physics]. Fifty five people registered and a like number attended. There were lots of interesting and informative talks, featured among them talks from folks outside the JLab community. See [https://halldweb.jlab.org/talks/2011-1Q/Parallelism%202011%20Workshop%20Talks%20available.txt David's email] for his summary for the participants and a link to the virtual machine that has the hands-on examples loaded on it.
  
* David mentioned yet another issue with using GEANT3 for our simulation. There are some CERNLIB routines that we have imported to the GlueX repository because of bug fixes. These can sometimes cause problems with include files.
+
In discussion Matt asked if JLab has licenses for [http://en.wikipedia.org/wiki/IBM_Rational_Purify Purify]. Graham said that it did; he would look into who owns it now.
  
==Tagger hall geometry/simulation==
+
==Review of IT Readiness for the 12 GeV Era==
  
Sascha led us through some recent work he has done, documented in a [http://argus.phys.uregina.ca/cgi-bin/private/DocDB/ShowDocument?docid=1646 recent GlueX Note]. The main conclusion is that we expect that dark rates in SiPM's used in the microscope could increase by a factor of five in about 2.5 months. This estimate includes shielding with polyethylene which reduces the exposure by about a factor of 7. See his note for details.
+
Graham described the upcoming review of all computing plans for 12 GeV era. The review will be comprehensive in that it will include accelerator controls, data acquisition support software, the Computer and Network Infrastructure Department (CNI) and the Management Information Systems Department (MIS). Some of his points:
 +
* want to determine if we can handle data coming from experiments
 +
* need an outline of remarks by all speakers by end of month
 +
* the review will be around the 2nd week of March
 +
* Graham will send around a spreadsheet, asking for updates to computing requirements from various groups
 +
* scheduled speakers: Bob Michaels from Hall A, Dennis Weygand from Hall B, Mark Ito from Hall D, a representative from Hall C--TBA, and Graham with an overview
 +
* current computing requirements:
 +
** Hall D: compute intensive
 +
** Hall B: storage intensive
 +
* storage more of a worry
 +
* this will be an internal (to JLab) review
 +
* a pre-cursor to external review in a year or so
  
The geometry configurations that were used have not been released for public consumption yet. See Sascha if you have a need for them.
+
==Changes to TOF Code==
  
==Calibration Database Update==
+
Beni described recent changes he has made to the HDDM data model to accommodate information from the Monte Carlo about "truth" information related to time-of-flight hits. In some cases, new elements were introduced to be able to keep both generated and detected information, and some elements were expanded to store information about the particle producing the hits.
  
Dmitry gave us a review of the scheme for the calibration database and reported on the current status. See [https://halldweb1.jlab.org/talks/2010-12/db_design_12_15_2010.pptx his slides] for details. Status is as follows:
+
A lot of discussion followed on the general idea of keeping generated quantities ("truth") and detected quantities ("hits") separate in our analysis. Some of Beni's changes seemed to violate this separation. Beni, Mark, and David agreed to meet after the meeting to discuss alternate approaches that preserve the information desired, but make the above mentioned separation more explicit.
  
* File specification is done
+
==Tracking Issues==
* File upload to DB in progress
+
* “Field” C++ API in progress
+
* Core C++ API done
+
* User C++ API mostly done
+
* Python wrapping--mostly done
+
* Console utils--first working prototype is done
+
* Web interface--prototype is done
+
  
There was some discussion during and after the talk:
+
Kei presented a list of bugs/anomalies that he has come across in running our reconstruction code. See [https://halldweb.jlab.org/talks/2011-1Q/offline-software.Kei-Moriya.2011-01-12.pdf his talk] for details. He mentioned four items:
  
* '''Security''': We had a lot of discussion about what level of security was necessary or appropriate. Too little and unauthorized changes can be made to sensitive calibration sets. Too much and making modifications is made difficult and the central collaboration copy might not get all important changes; those changes may languish in private version.
+
'''0.''' event processing hangs every ~1000 events for his event topology<br>
* '''Access outside JLab firewall''': Richard asked about how the constants would be accessed outside JLab.
+
'''1.''' pi+ events reconstructed as proton show “spike” in momentum at ~0.5 GeV/c<br>
*# David has implemented a web service to deliver constants. This was some time ago and needs to be revisited.
+
'''2.''' some events have direction of momentum “reversed”<br>
*# The database can be replicated at other sites and access granted to the replicas as needed. This is useful for the console tools, for example.
+
'''3.''' values of variables are set to non-physical values when they cannot be defined for a given track (leads to side-effects if not checked for explicity)<br>
*# The system will export files that can be used to supply constants from a local filesystem. For example, this could be used on a laptop without a network connection.
+
*# [added in press] There will be a web application that will allow any user with a web browser to access the contents of the database.
+
  
==Random number seed policy==
+
Work continues to try to resolve these mysteries.
  
Richard reviewed the options he outlined in his [https://mailman.jlab.org/pipermail/halld-offline/2010-December/000446.html email to the offline list]. He recalled for us that the issue was raised by [https://mailman.jlab.org/pipermail/halld-offline/2010-December/000443.html Kei getting indentical results] from separate runs of genr8 when they were started within a second of each other.
+
'''Side note''': Mark encouraged folks to write to halld-offline@jlab.org when they encounter problems like this so that (a) we all are aware of the problem (b) others that encounter similar problems can benefit from the conversation.
  
We formed a consensus around using /dev/urandom, at least initially. We commissioned Richard to write an appropriate python script that would produce a seed on standard output as an initial stab at a more robust random seed strategy.
+
==Action Item Review==
  
David noted that we should add better random seed management to mcsmear.
+
We looked over recently resolved items and those outstanding. David has added some items based on feedback from the workshop. Mark got David to volunteer to update the Build Make System (BMS) document to bring it up-to-date.
  
==New Action Items==
+
===New Action Items===
  
# Write a python script that uses /dev/urandom to produce pretty darn random numbers.
+
# Look into the Purify license. -> Graham
# Random seed management for mcsmear.
+
# Send email to the list on recent problems in the offline code. -> Kei
 +
# Prepare revised computing requirements for IT review. -> Mark
 +
# Revise BMS document. -> David
  
 
----
 
----

Latest revision as of 06:35, 1 April 2015

GlueX Offline Software Meeting
Wednesday, January 12, 2011
1:30 pm EST
JLab: CEBAF Center F326/7

Agenda

  1. Announcements
  2. Review of minutes from the last meeting: all
  3. Impressions from the Parallelism Workshop: David
  4. IT Readiness for 12 GeV: Graham Heyes
  5. Changes to TOF code: Beni
  6. Tracking Issues: Kei
  7. Action Item Review: all
    1. Resolved issues
    2. Outstanding issues
    • To view these links, login first. Anonymous login is available.
  8. Review of recent repository activity: all

Communication Information

Telephone

To connect by telephone:

  1. dial:
  2. enter access code followed by the # sign: 3421244#

Video Conferencing

Slides

Talks can be deposited in the directory /group/halld/www/halldweb/html/talks/2011-1Q on the JLab CUE [1]. This directory is accessible from the web at https://halldweb.jlab.org/talks/2011-1Q/ .

Minutes

Present:

  • FSU: Nathan Sparks
  • IU: Jake Bennett, Ryan Mitchell, Kei Moriya, Matt Shepherd
  • JLab: Hovanes Egiyan, Graham Heyes, Mark Ito (chair), David Lawrence, Yi Qiang, Dmitry Romanov, Simon Taylor, Beni Zihlmann

Announcements

Mark encouraged people to start thinking about what talks they want to give at the upcoming Collaboration Meeting in the Offline session.

Parallelism Workshop

David gave some impressions from last week's Workshop on Parallelism in Experimental Nuclear Physics. Fifty five people registered and a like number attended. There were lots of interesting and informative talks, featured among them talks from folks outside the JLab community. See David's email for his summary for the participants and a link to the virtual machine that has the hands-on examples loaded on it.

In discussion Matt asked if JLab has licenses for Purify. Graham said that it did; he would look into who owns it now.

Review of IT Readiness for the 12 GeV Era

Graham described the upcoming review of all computing plans for 12 GeV era. The review will be comprehensive in that it will include accelerator controls, data acquisition support software, the Computer and Network Infrastructure Department (CNI) and the Management Information Systems Department (MIS). Some of his points:

  • want to determine if we can handle data coming from experiments
  • need an outline of remarks by all speakers by end of month
  • the review will be around the 2nd week of March
  • Graham will send around a spreadsheet, asking for updates to computing requirements from various groups
  • scheduled speakers: Bob Michaels from Hall A, Dennis Weygand from Hall B, Mark Ito from Hall D, a representative from Hall C--TBA, and Graham with an overview
  • current computing requirements:
    • Hall D: compute intensive
    • Hall B: storage intensive
  • storage more of a worry
  • this will be an internal (to JLab) review
  • a pre-cursor to external review in a year or so

Changes to TOF Code

Beni described recent changes he has made to the HDDM data model to accommodate information from the Monte Carlo about "truth" information related to time-of-flight hits. In some cases, new elements were introduced to be able to keep both generated and detected information, and some elements were expanded to store information about the particle producing the hits.

A lot of discussion followed on the general idea of keeping generated quantities ("truth") and detected quantities ("hits") separate in our analysis. Some of Beni's changes seemed to violate this separation. Beni, Mark, and David agreed to meet after the meeting to discuss alternate approaches that preserve the information desired, but make the above mentioned separation more explicit.

Tracking Issues

Kei presented a list of bugs/anomalies that he has come across in running our reconstruction code. See his talk for details. He mentioned four items:

0. event processing hangs every ~1000 events for his event topology
1. pi+ events reconstructed as proton show “spike” in momentum at ~0.5 GeV/c
2. some events have direction of momentum “reversed”
3. values of variables are set to non-physical values when they cannot be defined for a given track (leads to side-effects if not checked for explicity)

Work continues to try to resolve these mysteries.

Side note: Mark encouraged folks to write to halld-offline@jlab.org when they encounter problems like this so that (a) we all are aware of the problem (b) others that encounter similar problems can benefit from the conversation.

Action Item Review

We looked over recently resolved items and those outstanding. David has added some items based on feedback from the workshop. Mark got David to volunteer to update the Build Make System (BMS) document to bring it up-to-date.

New Action Items

  1. Look into the Purify license. -> Graham
  2. Send email to the list on recent problems in the offline code. -> Kei
  3. Prepare revised computing requirements for IT review. -> Mark
  4. Revise BMS document. -> David

Footnotes

  1. You have to be a member of the "halld" Unix group to do this.