coherent WaveBurst Review Meeting 05 Feb 2009 09:00 Pacific / 12:00 Eastern
Agenda: Thursday 05 Feb 2009 09:00 Pacific / 12:00 Eastern
- Update on action items from previous week:
- (reviewers): Reviewers to look more closely at formula for damped sinusoids.
- Code review: high-frequency search
- Finish Code Review page.
- Code review: low-frequency search
- LFS driver scripts [web].
Dial-in number: 1 800 704 9896 International dial-in: +1 404 920 6472 Conference code: 5374 2349 #
AttendancePatrick (minutes), Stas, Giovanni, Marco, Gabriele, Romain, Francesco, Igor, Sergey.
- Romain will begin looking at loud missed injection and loud background triggers for the various network combinations.
- Action item: Igor will produce tables of found/missed injections for each network combination.
- Action item: Igor will make sure all results pages are updated to include CED follow-ups for loud background events (already done for H1H2L1, perhaps some others).
- Action item: Igor will circulate links to pages with more info on simulations performed.
Code review: high-frequency search
- Looking at "Pre-Processing" section of HFS review wiki page: [wiki].
- Gabriele: 4-step procedure. Steps 1-3 use same scripts as in LFS for segment generation. Difference is that the raw input lists are the single-IFO lists, so segments are "inclusive". For example, H1H2L1 segments include H1H2L1V1 times. Therefore need 4th step of excising higher-order (e.g., 4X) times from lower-order (e.g., 3X) lists.
- Exclusion procedure : drop segments entirely within higher-order list. For those that overlap higher-order list, apply overlap time as an effective cat 2 veto.
- Igor: Most LFS trigger production done before Virgo segments were available, therefore processed all LIGO data. After-the-fact excluded triggers that fell in higher-order times.
- Action item: We'll need to check exclusion procedure for LFS (when we get that far). Really only important for zero-lag triggers.
- HFS used later DQ version for LIGO, 2.5, while LFS used 2.3. Both used VH2 Virgo flags. Gabriele produced table comparing livetimes using the two sets of flags.
- Action item: Patrick to check segment list production. Did similar job for LFS in first year already.
- Action item: Gabriele/other to make sure all input segment lists are archived somewhere and linked from the review page.
- Action item: Igor will do the same for the LFS.
- Category 2 lists are good times.
- Looking at "Post-Processing" section of HFS review wiki page: [wiki].
- Francesco: All plots on wiki pages produced by netplot.C and simplot.C scripts (linked). Procedure: In ROOT, pass in netInit file for appropriate network, along with netplot.C or simplot.C.
- Looked at netInit file for H1H2L1V1 (background):
- USEVETO: flag to apply cat 2 DQ vetoes. Defined for low frequency, but were asked to apply anyway. Results of vetoes consistent with random coincidence.
- H1H2: flag indicating that H1 and H2 are both in the network. Causes H1H2 specific tests (e.g., energy disbalance) to be applied.
- Tuned thresholds are T_cor, T_cut. Other thresholds are not set by the tuning study, but are hard-wired.
- T_out only affects how plots are made.
- line 33: lags are defined to always be >= 0.
- line 35: Manually exclude triggers in 5 Hz band centered in 5482.5 Hz when H1H2 are present. In tuning saw excess correlated background due to H1H2 at this frequency. Will see examples later.
- line 234: nil, nill variables. Sergei: Expectation value of L is nonzero in Gaussian noise, in earlier versions of cWB. To make an unbiased measure of the SNR, the expectation value for Gaussian noise was subtracted from the measured likelihood and added to the null energy. These variables held that bias value. No longer used, since S5 version of code uses unbiased statistic.
- Will continue with review of netplot.C and simplot.C at next call.
- Action item: Patrick will circulate proposal for extra review call for Europeans to push through review of post-processing scripts (~2000 lines total). (Sergey recommends focusing on parts where thresholds applied to triggers.)