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

S5 Q-Pipeline Review Summary Table

Q-Pipeline S5 Y1 Review Summary Table

** Completed actions

Actions that need completed

Item Description Links Status
Search Basics We read Chapters 3 and 5 of Shourov's thesis, and discussed the following.
  • (t,f,Q) parametization of bursts
  • Basis rationalization
  • Optimal t-f-Q tiling, fractional energy lost
  • SNR conventions, coversions
  • Q transform (cont. and discrete)
  • normalized energy; significance and probability
  • H1H2 Coherent transform and Null stream
**Discussed, questioned, and understood many aspects of the search



Data Conditioning
Data conditioning happens before the Q-transform. The Q-transform basis functions are SG's in the whitened data space. Before Q-transform is applied, data is band-pass filtered and whitened.

**Sample whitened data stream in technical documentation
**Conditioning code covered in code review
**Reviewers read chapter 4 of Shourov's thesis to understand whitening process

Note for HFS: Issues for high freq?



Q-Plane Tiling/Q-Transform The signal space is "tiled" in Q, f, and t for a maxium of 20% mismatch in energy. We have discussed this process in some detail, as well as seen verification that the tiling meets the stated goal. The Q-transform itself has been discussed in detail, and we have seen verification of its ability to represent signals both with hardware injections and software injections. Beyond our checks, Q-scan and Q-event have been routinely used in control room and DetChar studies - in this sense the basic Q-transform has been "reviewed" more extensively than we could ever hope to accomplish.

**Seen verification of space tiling
**Discussed tiling energy mismatch metric
**Seen verification of Q-transform's ability to recover injections (h.w. & s.w.)



H1H2 Coherent Transform The colocated, coaligned nature of the H1 & H2 IFOs makes the coherent transform independant of sky position, and so computationally cheap. We have read about and discussed the coherent transform, and understand its application. We have seen verification of the coherent transform in action in the tech document. Additionally, we have seen the results of the coherent transform on injections plotted in Coherent-Correlated space, and verified the relationship between slope of the trendline in this space and the H1 to H2 sensitivity ratio. This verification has been carried out for the playground day, S4, and for the S5 1st year.

Certain details involved in the actual implementation of the coherent transform have been brought to our attention, but we have not completely understood them.

**Verification that coherent transform is succesful in recovering injections in the technical documentation.
**Verification in 2-D hard-corr space that PG/S4/S5 injections are recovered as expected
**Isabel checked code for colocated coherent transform

S4 Results
Isabel's Notes


Tile Downselection & Thresholding Tiles are projected onto 2-D t-f space, and overlapping tiles are systematically downselected. After downselection, remaining tiles above threshold are written to trigger lists. We have read about this process in Shourov's thesis, and discussed it.

**Read about and discussed.
**Examples can be seen by clicking images under "null stream tuning" in tech doc



Null Stream Veto The Null Stream veto rejects tiles that overlap in the t-f plane with H1H2 inconsistencies. It is well documented in the tech document, including tests of robustness to limited calibration uncertainty (this is by design). There are also demonstrations using S4 data showing that application of the Null Stream Veto does not significantly reduce the sensitivity.

** Discussed veto
** Documented in Tech Doc
** Seen effects on playground analysis

Parameter Documentation


Coincidence Find coincidence in T-F plane between H1H2 and L1

** Discussed
** Details in tech doc
** Careful tests of coincidence window by Shourov
** Seen segments verified by Laura's script
** mask, coincide, and select have been checked out by QHFS review

Jonah: Check Post-processing scipts
Reviewers: Post-processing scripts live on UWM here: /home/shourov/s5_qpipeline_rerun_postprocess

Seg-Tests 1, 2, 3
Jonah's Notes

In Progress

Thresholds Background slides allow the setting of thresholds. The target is a few events per hundred time slides.

** Reviewers have seen background slides for s5 1st year, H1H2L1 & H1H2
** Seen set thresholds and discussed
** Seen efficiency curves
** Made simple estimate of sensitivty - agrees to ~20%



Examine Loud Events The claim in the technical document is to "follow-up 100 loudest events." We should actually be looking at 10-20 loud events.

** Have seen table of loud events in coherent energy
** Loudest H1H2 event has been discussed some (playground day)
** Have table of loud events sorted by coherent AND corr energy

All: Review table of loud events - look for event candidates.


In Progress

Oustanding Action Items A review of the minutes revealed a list of action items and reviewer concerns from May 8 A link to these minutes is at right. I have made an attempt here to list the items, from that list, that I believed are not addressed elsewhere (e.g. in the above table). This is intended as a working list - corrections, additions, or subtractions are welcome.

***Exclusion curves by Laura
** Q Pipeline Only upper limit by Dave

Shourov: Make plots suggested by Isabel's first comment
Shourov: MDC study of calibration uncertainty
Shourov: Plot statistics in terms of event number in addition to event rate

May 8

Not Complete

Write Review Document


Draft (tex)
cWB Document
Writing Assignments July 2008

Not Complete

Back to the top level Review Page

$Id: summaryTable.html,v 1.16 2008/11/24 16:09:26 jkanner Exp $