Results of the search for burst gravitational waves
Results of the search for burst gravitational waves with the TAMA 300 detector Masaki Ando (Department of Physics, University of Tokyo) and The TAMA Collaboration The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France)
Introduction Target of this work … Burst gravitational waves Predicted waveforms of stellar-core collapse TAMA 300 data Data Taking 9 Data analysis Excess power filter Fake reduction Upper limit for Galactic events Galactic event rate GW energy rate Galactic simulation Signal injection simulation The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 2
Outline Target waveform TAMA 300 data Overview, Observation runs, Noise level Analysis scheme Burst filter, Fake reduction Analysis results Event-trigger rate Galactic simulation Summary and Conclusion The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 3
Analysis target - Target waveforms - Targets : Burst waves from stellar-core collapse Numerical simulations … ~100 waveforms are obtained Not cover all initial conditions Not suitable for templates (matched filtering: not available) Reference waveforms Common characteristics Short burst waves Spike wave ~1 msec Duration time <30 msec H. Dimmelmeier et al, Astron. Astrophys. 393 (2002) 523. Relativistic, axisymmetric simulation by Dimmelmeieret al. 26 waveforms Amplitude : h rss : 4 x 10 -22 /Hz 1/2 (at Galactic center: 8. 5 kpc) Energy : E tot : 9 x 10 -8 Moc 2 The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 4
TAMA 300 (1) - Overview - Data : Observation data by TAMA 300, an interferometric GW detector in Japan Baseline length 300 m : Fabry-Perot-Michelson interferometer with power recycling Placed at. National Astronomical Observatory in Japan First obs. run in 1999 (DT 1) Upgrade and observation runs Baseline 300 m Sufficient sensitivity for Galactic binary inspirals Automatedcrewless operation 9 observation runs (Data : ~3000 hours) National Astronomical Observatory at Mitaka, Japan The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 5
TAMA 300 (2) - Observation runs - TAMA observation runs Data Taking Objective Observation time Typical strain noise level Total data (Longest lock) DT 1 August, 1999 Calibration test 1 night 3 x 10 -19 /Hz 1/2 10 hours (7. 7 hours) DT 2 September, 1999 First Observation run 3 nights 3 x 10 -20 /Hz 1/2 31 hours DT 3 April, 2000 Observation with improved sensitivity 3 nights 1 x 10 -20 /Hz 1/2 13 hours DT 4 Aug. -Sept. , 2000 100 hours' observation data 2 weeks (night-time operation) 1 x 10 -20 /Hz 1/2 (typical) 167 hours (12. 8 hours) DT 5 March, 2001 100 hours' observation with high duty cycle 1 week (whole-day operation) 1. 7 x 10 -20 /Hz 1/2 (LF improvement) DT 6 Aug. -Sept. , 2001 1000 hours' observation data 50 days DT 7 Aug. -Sept. , 2002 Full operation with Power recycling 2 days DT 8 Feb. -April. , 2003 1000 hours Coincidence 2 months DT 9 Nov. 2003 Jan. , 2004 Automatic operation 6 weeks 5 x 10 -21 /Hz 1/2 111 hours 1038 hours (22. 0 hours) 25 hours 3 x 10 -21 /Hz 1/2 1. 5 x 10 -21 /Hz 1/2 1157 hours (20. 5 hours) 558 hours (27 hours) The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 6
TAMA 300 (3) - Data Taking 9 - TAMA DT 9 Data Taking 9 Nov. 28, 2003 – Jan. 10, 2004 558 hours of data Noise level : 2 x 10 -21 /Hz 1/2 This analysis … 2 nd half : 200 hours Noise floor level drift DT 6 DT 8 (Christmas, new-year Holiday terms) Better noise level Stable environment DT 9 The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 7
TAMA 300 (4) - Noise spectrum TAMA noise spectrum with. Dimmelmeier waveforms Detectable range : ~ 300 pc (optimal direction, polarization) The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 8
Burst-wave analysis (1) - Excess power filter - Burst filter : Excess power filter Evaluatesignal power in giventime-freqency regions Raw Data (time series) Spectrogram Total power in given T-F region Signal !! Time- Frequency plane (spectrogram) Freq. sum Few assumptions for signal … time-frequency bands Robust for waveform uncertainties The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 9
Burst-wave analysis (2) - Filter parameters (time-frequency window) Burst GW → Short duration, wide frequency band (c. f. Cont. waves: long duration, narrow freq. band) For higher SNR… Short time window ~Signal duration Requires line removal Wide frequency window Limited by worse noise level at low and high freq. Dt = 12. 8 [msec] Df = 2300 [Hz] The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 10
Burst-wave analysis (3) - Data conditioning Line removal Without Line Removal Filter freq. resolution : ~80 Hz line removal is required (AC line, Violin mode peak, Calibration peak) Method FFT 72 sec data Reject line freq. components Inverse FFT With Line Removal Normalization Track the drift of noise level Each spectrum is normalized by averaged noise spectrum 30 min (5. 6 x 10 -4 Hz) Use 30 min-averaged spectrum The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 11
Burst-wave analysis (4) - Fake reduction, Injection test Two veto methods Time-scale selection Burst signal < 100 msec Most detector noises > a few seconds Effective to long-duration noises Hardware and software injections Safety check : not to reject real signals Confirm that monitor channel bursts were not caused by real GW signal Veto with monitor channels Correlated bursts in intensity monitor channel Effective to short spikes Event threshold DT 8 analysis results (before veto) veto threshold False-dismissal rate estimation Less than 2% Hardware injection results Calibration : SNR(filter output) hrss The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 12
TAMA burst analysis - Analysis results Trigger rate with vetoes Improvement in rates with veto analyses Better in DT 9 than DT 8 Fake rate : 30 – 100 times Sensitivity : 3 -6 times Much larger than results with Gaussian noise DT 8 DT 9 (before veto ) DT 9 DT 6 Still many fake events The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 13
Galactic simulation (1) - Simulation method - Galactic simulation Monte-Carlo simulation Random events Inject signal to real data Analyze data with same codes Investigate ‘what happens with real signals’ compare with obs. results Position : Somewhere in Galaxy Exponential Disk model Galactic-event distribution R = (x 2+y 2) 1/2 Rd : 3. 5 kpc, hd : 325 pc Time : Somewhere in DT 9 (200 hr) Detector angular depend. Source : 26 Dimmelmeier waveforms Random angle Source angular depend. Effective distance The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 14
Galactic simulation (2) - Results of Galactic injection test Event-selection threshold : SNR>2. 9 -5 Detection efficiency : 1 x 10 -2 events/sec Observation result : 7 x 10 Upper limit Assume a Poisson distribution for the observed event number N obs N ul Galactic event rate 6 x 103 events/sec GW energy rate 6 x 10– 4 Moc 2/sec (90%C. L. ) The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 15
Summary Burst-wave analysis with TAMA 300 data TAMA 300 DT 9, 200 hours of data Excess Power filter, Fake reduction Galactic event simulation (Simulated-signal injection test) Galactic event rate 6 x 103 events/sec Galactic GW energy rate 6 x 10– 4 Moc 2/sec (90%C. L. ) Too large for real events Originate in residual fake triggers The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 16
Conclusion This work … Scheme to set upper limits for Galactic events from observation data Current works TAMA detector improvement (Noise hunting, Better isolation) Better filter(Filter tuning, better burst filter) Coincidence analysis (LIGO, ROG) We need … More realistic waveform catalog More realistic Galactic model Next-generation detector to cover our Galaxy (Ad-LIGO, LCGT) The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 17
End The 9 th Gravitational Wave Data Analysis Workshop (December 15 -18, 2004, Annecy, France) 18
- Slides: 18