Minutes of 2007-Nov-15 S5 QPipeline Review Teleconference
Shourov Chatterji, Jonah Kanner, Isabel Leonor, Dave Reitze
Minutes by Dave Reitze.
DiscussionStatus of the code review from Isabel: Have been a read through of the main parts of the code at telecons, and individuals have looked at it; Shourov has made changes to the code and those have to be reviewed. No major issues (bugs), looking at implementation issues that will come out during this review
Shourov: ˝ of code is documentation. Lots of code for interacting with Condor
Code is in CVS; not using tags.
Shourov informs code reviewers of changes during weekly telecons.
Document for code review – html document of comment reviewers posted at http://www.lsc-group.phys.uwm.edu/bursts/projects/q/codereview/comments.html
Jonah: what exactly are we supposed to do in this review? Shourov and Dave: look at output of the search; does it make sense, logical? If algorithmic problems found, that impacts code review. Basically, complete validation of the search results, look at everything to look for problems.
Shourov overview – goes through Project summary page; http://www.lsc-group.phys.uwm.edu/bursts/projects/q/s5/
All sky burst search, focus on 1st year S5 data, will be extended to 2nd year (how to fold in Virgo?)
Hope for detection, but will do upper limits
Look at H1H2 and H1H2L1 only for reasons of DQ and live time.
Discussion of analysis approach on the summary page.
Jonah: how to follow up events? Time consuming for even one event. Shourov: can bore in on top level events with some threshold. (Can also look for events in the background – complicated but possible.)
Question from Shourov: does it make sense to fold in high frequency search into this review? Really a question for Jolien
Presentation of Analysis methods – discussion of Q-transform. Dave: why should bursts look like SGs? Answer – lots of reasons, but basically SGs are a minimum time-freq basis. Look at Ch 3 of Shourov’s thesis.
--> Presentation of background in Shourov’s thesis, Ch 3.
At the end of the hour, the review team felt that they needed to go and read ch 3 and 5 to better understand the burst detection methods and the Q-transform.
Jonah: Question about timeline? How long should it take? Answer: analysis not completely done yet. Reviews can be long (from experience).
We agreed to meet in two weeks, November 29, at noon, EST time.