Review Committee Meeting Monday 12 June 2006 09:00 PST / 12:00 EST
Minutes: Monday 12 June 2006 09:00 PST / 12:00 EST
Agenda and Contact Info
- First comments on draft paper for the S2-S3-S4 GRB-GWB search [ PDF ]
- Discussion of vetoes used in the SGR1806-20 search [ HTML ]
AccuConference teleconferencing service: Phone: 1-800-704-9896, participant code: 038621# International callers ++1-404-920-6472 with same code
- GRB Draft Paper
[ PDF ]
- Several comments on paper will be circulated on an electronically marked up copy. Some of the comments were discussed in the meeting.
- Biggest thing missing in current draft: analysis of systematic and
- Fig 2 on page 4: error bars are present but not clear. Errors must be propogated to get an error on the "local probability" which will be shown on Figs 3 and 4 on page 6. Then need to figure out how to propogate these through the binomial test.
- Fig 8 on page 8: vertical error bars on plot (from uncertainty in calibration) and perhaps horizontal error bars due to random error in MC. Also, error bars in fit parameters will lead to error bars in UL.
- Other things we need to think about:
- Use circularly polarized waveforms. This would be using a minimal astrophysical input.
- Fig 10 on page 9. Why is it so jagged? Will be corrected.
- Beta distribution: model redshift distribution then have snr ~ 1/z. E.g., choose a so that the distribution of source distances gose as r^3 for nearby sources (homogeneous) then choose b to put a peak in the redshift distribution in roughly the right place.
- For a relatively model-independent (and highly optimistic) scenario, have snr^2 ~ (c^5/G) tau / 4 pi r^2 f0 S(f0) where tau is integration time (25 ms), f0 is some canonical frequency (say 200 Hz). Could multiply this by an ignorance parameter epsilon^2. (For a mass m falling into a black hole M, epsilon ~ m / M.)
- Eq. 15 on page 10. Why not choose a canonical spectrum, e.g., the SRD spectrum, rather than the median spectrum? This might help in extrapolations to the future. Or perhaps just plot median spectrum on the figure.
- Luca described the vetoing studies [ HTML ]
- Veto on fast signatures
- Look for ~1s transients in power (8-sigma). Would these veto target signals? They might veto some power for signals that are short, but these are not the target signals. Nevertheless, need to quantify safety.
- Goal is to look into some optimization of the veto.
- Veto will be applied in on-source data too. Do not anticipate much of the on-source data to be lost.
- Band correlation veto
- Look for excess noise in nearby band. Need to worry about possible leakage of signals from one band into the next (either by relatively broad, i.e., short duration signals or by drifting frequencies). To some extent these would not be the target signals, but still need to quantify.
- Will discuss further next time.