Review Committee Meeting Monday 27 March 2006 09:00 PST / 12:00 EST
Minutes: Monday 27 March 2006 09:00 PST / 12:00 EST
Agenda and Contact Info
- Take stock of review requirements prior to APS presentations.
- Brief orientation of Q-Pipeline documentation [ HTML ].
- Continue to go through SGR 1806-20 technical document [ PDF ].
AccuConference teleconferencing service: Phone: 1-800-704-9896, participant code: 038621# International callers ++1-404-920-6472 with same code
- Remaining reviewing issues for APS presentations:
- S5 Untriggered Analysis
- Need to review MDC frames. Specifically need to understand bias in time estimation. Hypothesis that it had something to do with calibration was going to be checked by running on h(t). At present, there is no code review for MDC frame production yet.
- CorrPower: Keith Thorne is working on code review. But CorrPower review is not essential for the S5 high-threshold results to be shown at the APS meeting.
- Sanity check on sensitivities is still needed.
- S4 Untriggered Analysis
- Only GWDAW results + remarks on how final results will be different from GWDAW results will be presented. No need to do further review for APS meeting presentation (beyond going over slides as usual).
- Laura has revised fits to ignore points with efficiency less than 10% (?) which provides better fits for the more relevant data (50% and 90% efficiencies). Peter has also done linear interpolation fits which can account for uncertainties in points.
- GRB Analysis
- Need technical report for S5 results if these are to be presented.
- Need to see plots comparing the LPF data conditioning with the "standard" data conditioning for the cc values (not just for sensitivities).
- Sensitivities need to be measured with a full 180s segment so that they are analyzed in the same way as the search itself.
- Signs of calibration: Pretty sure that they are fully understood between H1 and H2 for all final calibrations. Technical report claims that the calibration versions are S2V3 S3V2 and S4V3. Isabel has already re-run the S4 with V4 calibration. Need to rerun S3 with V3 calibration (the final calibration for S3).
- S5 calibration: Good to about 5%. Non-unity factors up to end of Jan now available. On-line GRB search uses unity factors. Not clear if DARM_ERR or AS_Q is used. If DARM_ERR, only gamma factor enters. If AS_Q, only the gamma factor is important for phase calibration unless imaginary part of alpha is important. Gamma is pretty stable for S5 so that the unity factors are probably good enough. From Gaby's talk at the LSC meeting: H1 has gamma=0.97+-2% with errors of 0.6% systematic, 0.2% random; H2 has gamma=1.02+-3% with errors of 1% systematic, 0.4% random; L1 has gamma=1.01+-3% with errors of 0.2% systematic, 0.2% random. Errors estimated from imaginary part.
- Need to understand various statistical tests, e.g., binomial test, ranksum test, loudest event test, etc. Which of these will be more powerful? It may depend on typical strength of signal (i.e., how much GRBs bias the value of cc) but this should be straightforward to model. Are certain tests more powerful for particular classes of signals?
- Population study: what "results" will be shown? Possibly only ranksum statistic. Need to look at Soumya's document in more detail.
- SGR 1806-20 Search
- Really only need to understand sensitivities. This search is essentially an excess power search with a given time-frequency window. Details of search (e.g., data selection) probably only have a limited effect on sensitivity.
- Webpage has new sample upper limits based on median value of background distribution. Technical document is being updated.
- Will need to review conversion of upper limits into GW energy. Note describing this is being prepared.
- To begin review, need to know where to get overview of search. Are the chapters on the Q-pipeline the best starting point? Let's read these first.
- Required: technical documentation describing this S5 search.
- Q: When the Q-pipeline search is substantially reviewed, will this simplify review of analyses that use triggers produced by the Q-ETG? A: Yes ... but ... that is if Q-pipeline is maintained (and review is continuous). Support of ETGs is burst group's responsibility: in the past, ETGs have been dropped when primary maintainers leave collaboration. It looks as if the Q-ETG is high priority. Guestimate is that a detailed review of Q-pipeline will take place this summer.
- S5 Untriggered Analysis