coherent WaveBurst Review Meeting 02 July 2009 09:00 Pacific / 12:00 Eastern
Agenda: Thursday 02 July 12:00 Eastern
- Discussion of how to deal with calibration uncertainties
Dial-in number: 1 800 704 9896 International dial-in: +1 404 920 6472 Conference code: 5374 2349 #
AttendancePatrick (minutes), Marco, Romain, Virginia, Francesco, Sergey, Igor, Stas.
Calibration Uncertainty Studies
- FS: Ran 9 new miscalibrated MDCs. 4X and LIGO 3X sigmoid curve
plots are updated, also hrss10/50/90% vs. frequency plots.
- The 2756 Hz curve has one of the 18 miscalibrated runs removed; only removed for this one waveform.
- FS: Other sections (V1H1L1, etc.) have not had extra MDC runs.
- FS: Used different draws for each frequency of the SGs. This is why things went wrong for only one SG.
- Not including antenna pattern long wavelength approximation correction or error in converting h(f) to h(t); expect both of these to be negligible (were in analytic treatment of uncertainties in LFS). Also not including phase jitters in these injections, as smaller effect than amplitude shifts.
- FS: Also did single Gaussian simulation with independent draws from Gaussian distribution for each injection (PJS: the correct way). Mean from 18 miscalibrated sets very similar to mean from this run: "fully consistent".
- PJS: At LVS F2F, think we decided to just show mean curve for ULs, with no extra 1-sigma allowance. May report error bars for sample sigmoid curves.
- FS: Important to note that "typical" or "mean" curve shows a little bit better performance than the 11% analytically (quadrature-sum) adjusted curves.
- FS: Propose not doing extra sims for the other network combinations; small livetime (about 20%) so the effect of not including them will be small.
SK: ULs to come from H1H2L1V1, H1H2L1, H1H2, H1L1V1.
PJS: Not H1H2V1?
PJS: H1H2 plot goes wonky above 4 kHz.
FS: Don't know what's happening.
FS: This plot ("Gauss Dist") is one with independent draws for each injection. Must be that some injections are incompatible in H1H2 due to miscalibration.
SK: Looks like lose maybe 5% of loud injections due to these calibration errors. Rejected as glitches due to inconsistency.
FS: Monte Carlo estimates conservative in that they include implausibly large relative miscalibrations of H1-H2.
FS: 2 main problems with thsi study:
- RMS calibration uncertainty in HF regime very large.
- Using coherent technique, which can be very sensitive to miscalibrations in some situations (e.g., relative amplitude miscalibration in H1H2). Can only be properly handled using miscalibrated injections.
- RG: LMI follow-up for HFS not done yet, Maybe for next week.
- FS to update wiki for next week; PJS to study.