170 likes | 320 Views
Frédérique MARION. Status of inspiral search in C6 and C7 Virgo data. for the Collaboration GWDAW 2005. C6 & C7 runs. Optimally oriented [1.4,1.4] M (SNR=8). C6 14 days 86% duty cycle (in science mode)
E N D
Frédérique MARION Status of inspiral search in C6 and C7 Virgo data for the Collaboration GWDAW 2005
C6 & C7 runs Optimally oriented [1.4,1.4] M (SNR=8) • C6 • 14 days • 86% duty cycle (in science mode) • Factor 2 variation in the horizon over the course of the run • C7 • 5 days • 65% duty cycle • Detector sensitivity also not stationary
Inspiral online analysis • Both pipelines, Merlino and MBTA, ran online during C6 and C7, getting data from online h-reconstruction • Monitoring information provided • Lists of loudest events • Plots for monitoring web pages MBTA Hardwareinjections
C6 hardware injections • 56 inspiral hardware injections • 4 different periods, [1.4,1.4] M, SNR ranging from ~15 to 25 • All detected • Check timing and mass estimation accuracy • Check SNR recovery ~ 10% loss might be due to sensitivity non-stationarity MBTA
C7 hardware injections • 2 periods of inspiral injections • [1.32 , 1.36] M, SNR ~ 20 • Detected with good accuracy • Issue with some injections having a bad 2 • under investigation Merlino
± 10% variations on short time scales(~15 min) Smaller variations still present Non-stationarities: C6 • Evolution in time of the frequency at which half of the SNR is integrated Change in the shape of the PSD
Same kind of variations observed in C7 Non-stationarities: C7 • Variations of the PSD on short time scales not obvious to follow with adaptive mechanisms
Dark fringe spectrum C6 analysis: quality cuts (I) • C6 first six days • Horizon very unstable • “Bump” going up and down in 100-300 Hz region of sensitivity curve (stray beam) • Leave this first part of the run aside • Focus on remaining 8 days
C6 analysis: quality cuts (II) • Skip first five minutes after relock • Use auxiliary channels to identify obvious sources of high SNR triggers, e.g. • Saturations in frequency stabilization loop • Glitches on actuator coil drivers
C6 analysis: SNR distribution MBTA Inspiral injections • [0.9 - 3] M analysis • ~ 170 hours of data • Very loud events eliminated by basic quality cuts • Tail of distribution still extends to high values • Even strong, the hardware injections (in red) do not dominate the noise Burst injections
MBTA Merlino A relatively quiet 10 hours long period C7 analysis • [0.9 - 3] M analysis • ~ 80 hours of data • C7 SNR distribution slightly better than C6
C7 software injections • Systematic studies (efficiency, 2 test) with Merlino on software injections performed during C7 quiet period see poster by L.Bosi
noise event hardware injection noise event C6: a posteriori vetoes • SNR distribution between low and high frequency bands • 22 bands • Shawhan-Ochsner veto • Look at output of matched filter in time domain
22 bands < 100 Samples above threshold < 30 22 bands< 100 Samples above threshold < 30 C6: SNR with vetoes • Bulk of distribution unchanged • High SNR tail definitely improved • Hardware injections almost stand out
C7: vetoes • Behavior similar to C6
C7: SNR with vetoes 22 bands < 40 • Loudest surviving event has SNR < 20 Samples above threshold < 30 22 bands< 40 Samples above threshold < 30
Conclusion • C6 and C7 data have extra-galactic sensitivity to neutron star inspirals • Not science quality data • Detector behavior glitchy and not stationary • Preliminary analysis • Two pipelines operational for online and offline analysis • Hardware injections, quality cuts, a posteriori vetoes • Pursue investigations • Learn as much as we can from those data in order to get prepared for the science data to come next year