LSC-Virgo Burst Analysis Working Group


Burst Group Home
ExtTrig Home
CBC Group Home
LSC, LIGO, Virgo


'How-to' docs
[2008, 2007, earlier]
Talks/Posters [pre-wiki]
Paper plans
White papers


Analysis projects
Old notebook [General, S2, S3, S4, S5]
Virgo workarea
External collabs


Main review page

X-Pipeline Review Meeting Monday 08 Dec 2008 09:00 Pacific / 12:00 Eastern

Agenda: Monday 08 Dec 2008 09:00 Pacific / 12:00 Eastern

  1. Getting ready to open some boxes: checklist on review summary page.
  2. Code review: Reviewing likelihood types in lines 706--1192 of xtimefrequencymap.


Attendance: Patrick Sutton, Jolien Creighton, Gareth, Xavi, Peter Raffai, Antony
  1. Going through items in Review Status review summary page.
    1. Data conditioning -- update: next time.
    2. No changes required.
    3. Changes required?
    4. H1H2 Likelihood choice: Fixed.
    5. Calibration study: Fixed set of injection amplitudes. Too coarse, to large a range. Want to account for coarsness by putting it into the uncertainties. This can be done after the fact. Being done.
    6. Tuning:
      • Followup with significance of 95% (around 5ish).
      • Note: I_null does have contributions from GW (not really null). E_null is null combination, I_null is the same combination but with off-diagonal terms thrown out. Looked at Fig. 4 p. 16 of technical report. Why are significant boxes below diagonal. Could be a H1 glitch?
      • Tuning: median loudest off-source event as dummy loudest event. Gives best upper limit on average. Like to try pulling out the 95% loudest surviving background event -- try to set the tuning to suppress the tails of the background distribution. Trying this out. Tuning does an UL 90% analysis. Suggest doing a 50% UL tuning for the 95% loudest event for a detection search. Still under investigation.
    7. UL statement. Propose E_GW rather than h_rss. Simple formula. Seems funny since circular polarization and isotropic emission is not consistent. WNB instead?
    8. Population statement: Same as S2-S3-S4 paper. Binomial test and Wilcoxian rank-sum test. Propose loudest 10 events. Is one gold-plated event more likely than 10 marginal events? In some sense, don't need to do a binomial test for 1 event since follow-up of all loudest events.
$Id: minutes-2008-12-08.html,v 1.3 2008/12/08 19:57:53 jolien Exp $