Difference between revisions of "GlueX PID Meeting, November 8, 2018"

From GlueXWiki
Jump to: navigation, search
(Agenda)
(Agenda)
Line 15: Line 15:
 
#Standards
 
#Standards
 
#Work Packages
 
#Work Packages
 +
 +
==Minutes==
 +
# Announcements
 +
##Really well attended (~20 people), so naturally our initial room B101 would not let the chair log in.  Mark I. found a new room and after five minutes we began.
 +
##Attendees were encouraged to watch a short video, linked above, for inspiration
 +
#PID FOM
 +
##Simon briefly mentioned the existence of the variable PIDFOM which sought to combine chisq info from various detectors. Uncertainties in question...
 +
#General discussion
 +
##Thomas mentioned a TOF discussion when both he and Beni touched on the solutions used by LHCB and HERMES.  Mark I. suggests a reprisal of the discussion as this is ultimately a "solved problem"
 +
##Mark I suggests starting with some brainstorming.  At this point discussions were had:
 +
##Sean suggests an incremental approach with links to the workpackages.  Thomas agrees and tends to follow a "first make it work, then make it good" approach. Sean suggests we try to come up with some simple variables we can use now to get us going.
 +
##Thomas mentioned he had talked to Daniel L. over shift and had mentioned working on some BDT/MVA/NN solutions to PID. 
 +
## Daniel described briefly what he has been playing around with and agreed to give a presentation next time
 +
## Daniel also suggested having two methods early on would be beneficial and that different variables may in-fact need to utilize different methods for the best results.  There was general agreement surrounding this.  Also speaking to Sean's earlier comment, focusing per hypothesis and then collating those results later may be better than a single global classifier.
 +
## Sean suggests groups working both top down and bottom up.
 +
## Daniel asks about the interface with the sub detector groups.  Thomas points out a majority of both the TOF and SC groups were present.

Revision as of 10:28, 9 November 2018

Meeting Connections

  1. To join via a Web Browser, go to the page [1] https://bluejeans.com/198066682.
  2. To join via Polycom room system go to the IP Address: 199.48.152.152 (bjn.vc) and enter the meeting ID: 198066682.
  3. To join via phone, use one of the following numbers and the Conference ID: 198066682.
    • US or Canada: +1 408 740 7256 or
    • US or Canada: +1 888 240 2560


Agenda

  1. Announcements
    1. identification
  2. Resources
    1. PID FOM
  3. Standards
  4. Work Packages

Minutes

  1. Announcements
    1. Really well attended (~20 people), so naturally our initial room B101 would not let the chair log in. Mark I. found a new room and after five minutes we began.
    2. Attendees were encouraged to watch a short video, linked above, for inspiration
  2. PID FOM
    1. Simon briefly mentioned the existence of the variable PIDFOM which sought to combine chisq info from various detectors. Uncertainties in question...
  3. General discussion
    1. Thomas mentioned a TOF discussion when both he and Beni touched on the solutions used by LHCB and HERMES. Mark I. suggests a reprisal of the discussion as this is ultimately a "solved problem"
    2. Mark I suggests starting with some brainstorming. At this point discussions were had:
    3. Sean suggests an incremental approach with links to the workpackages. Thomas agrees and tends to follow a "first make it work, then make it good" approach. Sean suggests we try to come up with some simple variables we can use now to get us going.
    4. Thomas mentioned he had talked to Daniel L. over shift and had mentioned working on some BDT/MVA/NN solutions to PID.
    5. Daniel described briefly what he has been playing around with and agreed to give a presentation next time
    6. Daniel also suggested having two methods early on would be beneficial and that different variables may in-fact need to utilize different methods for the best results. There was general agreement surrounding this. Also speaking to Sean's earlier comment, focusing per hypothesis and then collating those results later may be better than a single global classifier.
    7. Sean suggests groups working both top down and bottom up.
    8. Daniel asks about the interface with the sub detector groups. Thomas points out a majority of both the TOF and SC groups were present.