S4 LIGO-GEO Burst Review Committee Special Meeting 02 May 2008 10:00 Eastern
Agenda and minutes of the special LIGO-GEO S4 Bursts Review telecon, held Friday 02 May 2008.
Agenda and Contact Info
- Discussion of segment lists. [ web ]
InterCall telecon service: USA (toll-free): 1-866-616-1738 UK (toll-free): 0800 073 8914 Italy (toll-free): 800-906-494 Germany (toll-free): 0800-1014-907 participant code: 251 288 9495#
AttendanceMichele Zanolin, Peter Shawhan, Patrick Sutton (minutes), Keith Riles, Sergei Klimenko, Igor Yakushin.
- Re-check Siong's figure 4 a,b.
- Make new plots with updated iWB trigger sets.
- Check background observation time used for the rate vs. strength plot.
Extracts from Sutton's email before meeting
Patrick: New web page on segment lists set up by Igor, Sergei. Very useful. Did some double-checking and updates to page.
The cat 1 and cat 2 lists you put on the web page are exact copies of the files in the cWB CVS (this is good). To avoid confusion from having multiple copies of the same file, I deleted the files you posted and set the links on the web page to point directly to CVS.
I have also confirmed that the cat 2 segment list is identical to the one posted on the elog entry you reference ( log entry ).
So, the cWB segments look fine so far.
I have found 3 different segment lists posted iWB-CP. Two of them are the lists you and Igor posted to the web page; they are different, and there is no explanation of which one is the right one. The third segment list is the one used to generate the iWB triggers; I got it from Igor's page of S4 trigger sets. That's different from the other two. It's also the one I am inclined to trust, since it came from Igor's trigger repository.
So, we still need to clear up which iWB segment list is the right one, and why the other two aren't the right ones.
Another question: Is it true that for BOTH iWB and cWB in S4, segments as short as 1 sec were analysed? These segment lists all contain segments that short.
- Igor: For iWB, first list on web page is cat 1 segment list with "minimal cuts" gave job segment lists. Chosen to minimize fragmentation of segments. List 2 is cat 2 segs, used in post production. Very fragmented. List 3 is after post-production: intersection of first two lists.
- Igor: Had noted earlier than ~2% of cWB times are not analysed by iWB-CP. Discrepancy due to fact Siong mistakenly required cat 2 segments have minimum duration of 300 sec, which is not necessary for post-processing. Cat 1 segments need min duration 300sec, but no min duration for cat 2 segments.
- Injection table bug: Igor found bug in script that shifted strain factors for iWB in Igor's found vs. missed tables by one step. Means tables showed iWB efficiencies to be slightly better than they actually are.
- Tables now regenerated with bug fixed: see 050108a. Use common segment lists for iWB, cWB, after MDC resampling, after cuts. Use 257 hours intersection segment list, Laura's resmapled injection logs for both pipelines.
- Igor: Livetimes for time lags in iWB pipeline obtained from script which is currently running.
- Action items:
- Get new iWB cat 1 file from LLO (ldas-pcdev1.ligo-la.caltech.edu).
- Verify combined segment list for iWB-CP.
- Igor: One iWB job may have failed, so there may be a small discrepancy in the segment lists for iWB and cWB.
- The "chi2cut" file is the cat 2 list. The iWB-CP segment list should be the intersection (cat 1 AND cat 2), up to the possibly missing job.
- Edge effects: WB does not analyse the first 8 or last 8 sec of data in a segment. Seg lists should overlap by 16 sec. Not clear if this is accounted for already inposted semgent lists.
- Igor: Algorithm for iWB segment lists should be to remove 8 sec off of each end of cat 1 list, then AND with cat 2 list to get combined segment list.
- Sergei: For cWB, drop 8 sec off start and end of cat 1 segments. Since cWB runs on h(t), may be missing ~10 sec at start or end of some segments. Not clear if this time is already accounted for posted segment lists, or if there are gaps in data in posted segment lists.
- Action item: Need to get cWB segment lists from the jobs that actually ran, covering times actually analysed.
- Sergei: The cWB S4 LIGO-GEO page, under "Search Sample" section, has lists of livetime for zero lag and time lags. Already includes cat 1 and 2 flags, and accounts for failed jobs, 8-sec cuts at edges of segments, etc. THese should be the final numbers for the cWB livetimes.
Recap of Action items
- Laura will run her scripts for efficiencies as cross-check of Siong's results.
- Igor will generate livetimes for each lag for iWB (should agree to within few percent to estimate from zero-lag).
- Figure 4: Don't yet know exact iWB background livetime. Will need to regenerate figure with corrected background livetime.
- BUG in iWB-CP post-processing: Siong used incorrect hrss value. Used hrss reconstructed by CorrPower instead of reconstructed amplitude from iWB.
- Cross-check new iWB-CP efficiencies from Laura with Igor's found-missed tables.
- Follow-up Zg = 8-10 outliers (have Gamma ~2-3.)
- Check what happened to Gamma~25 outlier (how was it cut in corrected post-processing).
- Update paper text on outliers.
- Figure 3.