Minutes of 2008-May-08 S5 QPipeline Review Teleconference
Shourov Chatterji, Jonah Kanner, Isabel Leonor, Dave Reitze
Minutes by Isabel Leonor.
* Dave: Jolien sent out email with request for review closure. Plan for Monday's burst review telecon: Dave will write a few paragraphs about status of QPipeline review, and include list of action items. Dave will not be able to attend Monday telecon, but Jonah and Isabel will attend. * Went on to discuss Shourov's posted agenda for today. Shourov: This will be a short call. Dave: Paper is taking shape. Can you walk us through Patrick's method for combining upper limits? Shourov: Patrick is writing CGQ paper about this. For the Orsay meeting, goal is to have a picture of the paper, and present the results. Shourov then shows us where to find link to Patrick's presentation at a burst telecon. It can be found on the agenda page for the February 13 burst telecon: http://www.lsc-group.phys.uwm.edu/bursts/investigations/general/misc/incoming/1202925.637-MultiPipe.pdf.pdf Shourov briefly goes through some of the slides, i.e. slide #3, 4, 5, 8, 9. A key point seems to be the plot on slide #5 which shows single-pipeline efficiencies and combined efficiencies. Dave: On slide five, the purple curve shows that the end result is worse than that from either pipeline. Jonah: There is potentially a lot of information here. It gives ability to say which injections are seen by one pipeline but not another. This will be put to the test. Using the same MDCs, the same amplitude, this is good. Shourov: That will feature in the paper. Isabel: In this procedure of combining upper limits, do the different pipelines use the same false alarm threshold? Shourov: The burst group decided that the threshold be the same for the different pipelines, but this is not necessary. Dave (to Shourov): Is there anything else to point out in paper? It is in pretty rudimentary state. Shourov: It needs cleaning up. * Shourov then mentioned cross-validation of QPipeline Matlab segment tools with segwizard tools. The results from the two are the same. Shourov: This is just a piece of the check of post-processing scripts. There are still the C++ scripts to be checked. * Shourov then mentioned the availability of final segment lists. This includes most recent version of DQ flags and also veto flags. The QPipeline results need to be updated using these final segments. * Shourov: Have not had chance to address list of past action items. Maybe we should not have meeting next week so these things can be worked on. Dave: Will be on a plane next week. Jonah: Will be in a car. All: Agree to not have a telecon next week. Isabel (to Shourov): For the update using the final segments, will you have to re-run QPipeline on the data? Shourov: No, even though there may be slight changes in Category 1 segments. Category 2 and Category 3 flags are applied post analysis, so this only requires masking triggers. Small changes in Category 1 segments probably do not matter much, especially since QPipeline uses 64-second analysis blocks, unlike for example the 1024-second blocks used by inspiral group. Category 1 flag means data is so bad. Jonah: What are the differences in livetimes for the different pipelines? Shourov: Venn diagrams in paper shows the different livetimes. Jonah: It is awkward to try and explain why different pipelines ran on different Category 1 data sets. If the difference is small, then it doesn't matter. It would be nice to know this. The different pipelines will have slightly different livetimes anyway, because of different ways of processing data, i.e. different ways of handling edges, etc. The effect is the same in that the livetime is slightly different for different pipelines. So if effect of different Category 1 segments is marginal, then it's okay.