Review Committee Meeting Monday 8 October 2007 08:00 Pacific / 11:00 Eastern
Minutes: Monday 8 October 2007 08:00 Pacific / 11:00 Eastern
Agenda and Contact Info
- GRB070201 Search
- Review of BurstMDC Simulation
InterCall telecon service: Phone: 1-866-380-5536, participant code: 435 672 9587 # International callers ++1-816-249-4731 with same code International access numbers (~10 cents/min surcharge for LIGO Lab): Paris 017 080 7156 ; Lyon 042 603 0036 ; Germany toll-free 0800 182 1591 ; Berlin 030 726 167 371 ; Rome 00645 217 080 ; UK toll-free 0808 234 7914 ; London 0203 107 0293 Handy participant command: Press *6 ["*M"] to mute, #6 to unmute
- GRB070201 Search
- More discussion about systematics. Vicky argues that the 10% distance uncertainty is not unreasonable.
- Discussion about the 1.6- vs. 1.2-sigma choice. Depends
on wording. Somewhat confused by overloaded use of the
word "confidence". For explicitness, Patrick Sutton
described the example of the 1.4-10 inspiral results from
Isabel: 99.8% of the injections had ccmax larger than
ccmaxobserved. When distance to Andromeda was adjusted
(according to 1.2-simga) then this dropped to 99.5%.
Interpretation of this is that the confidence in the
hypothesis that there was an inspiral of a 1.4-10 system
in Andromeda is less than 0.5%, with a 90% confidence in
this number. That is, the fraction has a distribution,
and we are looking at the conservative tail of this
distribution. In addition, a correction for MC error would
However, if we said that the +-1-sigma error bar on distance is 10%, then we would say that the +-90% error bar on distance is 16% -- that is, use the number 1.6 rather than 1.2. This error could then be adopted with a statement that all values are conservatively shifted to the less favorable value of the 90% uncertainty range.
Concrete proposal: Since it looks like 10% is an OK value for M31 distance uncertainty, and there is also something like a 10% uncertainty due to calibration, then just choose the 1-sigma error -- 14% -- which is what Isabel has already done. That way there is no rerunning needed.
- Review of BurstMDC Simulation: Discussion of networkgraven.
- Synopsis: graven runs on one IFO and produces a logfile. Graven can also read in a logfile to run on a second IFO. This routine does a loop, calling graven and producing a logfile for the first IFO and then calling graven using this logfile for remaining IFOs.
- We did not look at the code to add to backgrounds.
- getifolist appears similar to other functions we've looked at.
- getdetrate uses "incorrect" V2 for Virgo rather than V1. This has been changed in newer version of this code. However V2 is then used for Virgo with LIGO sample rate. This sounds like it could cause confusion. Suggest something else. In any case, word "VIRGO" is used and this should perform correctly.
- General philosophy choice: either have reasonable default values or have code crash and burn if anything unexpected happens. Would prefer latter. But Matlab did not handle errors gracefully in earlier versions. Now it does. Gradually changing things to do better error checking. We support this.
- Issue with logfiles: name is constructed from gpsstart and duration. But what about GRBs when you want to have different injection files for the same time? Issue will need to be addressed. Often this is accommodated by running in different directories. Perhaps could be done by adding a hash of the random params to logfilename. Not currently an issue.
- Not sure what is happening on lines 192 and 193. Keith will do some research and we'll begin here next time.
- S5 SGR analysis. Discussed possibility of reviewing Flare portion now, cWB portion later.