Difference between revisions of "January 21, 2014"
From GlueXWiki
m (Text replacement - "http://argus.phys.uregina.ca/cgi-bin/private" to "https://halldweb.jlab.org/doc-private") |
|||
(14 intermediate revisions by 3 users not shown) | |||
Line 12: | Line 12: | ||
== Agenda == | == Agenda == | ||
# Announcements | # Announcements | ||
− | + | #* [[GlueX-Collaboration-Feb-2014|Collaboration Meeting]] | |
− | # | + | # Tagging calibration data "streams" in Level-3 |
+ | #* Recommendations from [https://halldweb.jlab.org/doc-private/DocDB/ShowDocument?docid=2382 software review report]: | ||
+ | #*# We recommend the inclusion of calibration in the data challenges. Demonstration of a functional prompt calibration loop will be an important element of achieving rapid extraction of physics results. | ||
+ | #*# Event tagging in the HLT is recommended as a mechanism for separating calibration data samples into streams for use in a prompt calibration loop. | ||
+ | #* Requesting feedback from detector groups on what calibration data samples might be useful | ||
+ | # [https://halldweb.jlab.org/wiki/index.php/File:Level3_triggerMtg1.21.14.pdf Level-3 Trigger Update] | ||
+ | |||
+ | == Minutes == | ||
+ | # Collaboration Meeting | ||
+ | #* The tentative schedule is to have three talks from Chris (Trigger Electronics), Alex (L1), and Justin (L3) | ||
+ | # Tagging calibration data "streams" in Level-3 | ||
+ | #* As discussed in the [[January 15, 2014 Calibration | calibration meeting]], the detector groups are requested to start compiling a list of the types of data samples that would be useful for calibration | ||
+ | #** Curtis suggested a presentation on what other experiments do (eg. LHCb) would be useful | ||
+ | #* In terms of implementation we discussed a couple of options: | ||
+ | #*# Setting a flag in the data stream for events which would be useful for calibration | ||
+ | #*# Writing events useful for calibration to separate data files (which could be processed quickly by calibrators online, if desired) | ||
+ | #* From previous discussions with the DAQ group it seems that writing events to multiple streams is possible, but it hasn't been done previously. | ||
+ | #* For the spring online data challenge, the plan will be to implement the event flagging in the data stream and postpone the writing of events to separate streams until later | ||
# Level-3 Trigger Update | # Level-3 Trigger Update | ||
+ | #* Justin presented an update on the L3 algorithm by improving the track momentum resolution using wire-based tracking | ||
+ | #** The track momentum resolution is dramatically improved with wire-based tracking for polar angles < 15 degrees | ||
+ | #** The momentum resolution is noticeably worse for tracks in the CDC with DTrackWireBased relative to DTrackCandidates which has been seen before (by Simon) and is unexpected and not understood | ||
+ | #* Using wire-based tracking for this subset of the track candidates will moderately increase in CPU time while improving the performance of the trigger to ~95% efficiency | ||
+ | #* The plan is to test this algorithm on the online nodes in the counting house to get better estimates of CPU performance for different tracking options and BDT staging |
Latest revision as of 17:13, 24 February 2017
Meeting Time and Place
Tuesday January 21, 2014 at 1:30pm
Connections
To connect from the outside, please use ESNET or SeeVogh
- ) ESNET: 8542553
- ) SeeVogh: A conference has been booked under "GlueX Trigger Working Group" from 1:00pm until 3:00pm (EDT).
- ) To phone into an SeeVogh meeting, from the U.S. call (626) 395-2112 and then enter the EVO meeting code, 159726.
- ) To use the SeeVogh Java script use the SeeVogh link.
Agenda
- Announcements
- Tagging calibration data "streams" in Level-3
- Recommendations from software review report:
- We recommend the inclusion of calibration in the data challenges. Demonstration of a functional prompt calibration loop will be an important element of achieving rapid extraction of physics results.
- Event tagging in the HLT is recommended as a mechanism for separating calibration data samples into streams for use in a prompt calibration loop.
- Requesting feedback from detector groups on what calibration data samples might be useful
- Recommendations from software review report:
- Level-3 Trigger Update
Minutes
- Collaboration Meeting
- The tentative schedule is to have three talks from Chris (Trigger Electronics), Alex (L1), and Justin (L3)
- Tagging calibration data "streams" in Level-3
- As discussed in the calibration meeting, the detector groups are requested to start compiling a list of the types of data samples that would be useful for calibration
- Curtis suggested a presentation on what other experiments do (eg. LHCb) would be useful
- In terms of implementation we discussed a couple of options:
- Setting a flag in the data stream for events which would be useful for calibration
- Writing events useful for calibration to separate data files (which could be processed quickly by calibrators online, if desired)
- From previous discussions with the DAQ group it seems that writing events to multiple streams is possible, but it hasn't been done previously.
- For the spring online data challenge, the plan will be to implement the event flagging in the data stream and postpone the writing of events to separate streams until later
- As discussed in the calibration meeting, the detector groups are requested to start compiling a list of the types of data samples that would be useful for calibration
- Level-3 Trigger Update
- Justin presented an update on the L3 algorithm by improving the track momentum resolution using wire-based tracking
- The track momentum resolution is dramatically improved with wire-based tracking for polar angles < 15 degrees
- The momentum resolution is noticeably worse for tracks in the CDC with DTrackWireBased relative to DTrackCandidates which has been seen before (by Simon) and is unexpected and not understood
- Using wire-based tracking for this subset of the track candidates will moderately increase in CPU time while improving the performance of the trigger to ~95% efficiency
- The plan is to test this algorithm on the online nodes in the counting house to get better estimates of CPU performance for different tracking options and BDT staging
- Justin presented an update on the L3 algorithm by improving the track momentum resolution using wire-based tracking