Short gravitational wave signal searches in TAMA300 data : stellar - PowerPoint PPT Presentation

Short gravitational wave signal searches in TAMA300 data : stellar core collapse and black hole Nobuyuki Kanda TAMA collaboration @ TAUP2007, 11th Sep. 2007, Sendai Special Thanks to M.Ando, T.Akutsu, R.Honda and Y.Tsunesada

1-1 Excess Power Filter 1-2 ALF 1-3 TF-Cluster Matched filter TAMA's searches for Short GW 1. Stelar-core collapse (SN) : Burst GW 2. Black-hole quasi-normal mode : Ringdown GW 3. Keyword for short signal searches

Observational runs and data Data Taking period actual data amount remarks DT1 8/6 - 7/1999 ~3 + ~7 hours continuous lock first whole system test DT2 9/17 - 20/1999 31 hours first Physics run DT3 4/20 - 23/2000 13 hours h ~ 5x10-21 [1/ √ Hz] -- 8/14/2000 World best sensitivity DT4 8/21 - 9/3/2000 167 hours stable long run DT5 3/1 - 3/8/2001 111 hours Longest stretch of continuous lock is Test Run 1 6/4 - 6/6/2001 keep running all day 24:50 1038 hours DT6 8/1 - 9/20/2001 full-dressed run duty cycle 86% Recycling, h ~ 3x10-21 [1/ √ Hz], DT7 8/31 - 9/2/2002 24 hours with duty cycle 76.7% Simultaneous obs with LIGO & GEO 1168 hours, coincidence obs with DT8 2/14 – 4/14/2003 duty cycle 81.1% LIGO S2 10/31(Actually 558 hours, partial coincidence run with LIGO S3 DT9 11/28)/2003 (weekday: night time, ‘ crewless ’ operation – 1/5/2003 weekend: full time)

1. Burst Gravitational Waves from Stellar-core collapse Numerical Simulation Predicts GW Waveform. Komatsu et al. (1989) Zwerger & Müler (1997) Dimmelmeier et.al., (2001,2002) 1.5 A1B1G1 A3B3G1 1 A4B1G2 ] –20 Amplitude [x 10 0.5 0 –0.5 Gravitational waveforms –1 from stellar–core collapse (10kpc from the earth) –1.5 0 10 20 30 40 50 Time [msec]

TAMA300 Sensitivity : Range of Detection for Burst GW from Stellar-Core Collapse 1/2 10 –18 Detector Noise Level [1/Hz TAMA noise level (DT9) GW RSS Amplitude and 10 –20 100pc events 10 –22 10kpc events LCGT design sensitivity 10 –24 10 2 10 3 Frequency [Hz] Figure by M.Ando, GW signals by Dimmelmeier, et al. (2002)

1-1. Excess Power Filter by M.Ando (Tokyo Univ.)

by M.Ando Phy. Rev. D71, 082002 (2005)

1-2. ALF (Alternative Linear Filter) by Tomomi Akutsu (ICRR, Tokyo Univ. / Osaka City Univ.),

U.L. for h rss >10 -17 0.55 [events/day] , C.L.90% by Tomomi Akutsu , et al. Class. Quantum Grav. 23 (2006) S715

1-3 TF (Time-Frequency) - cluster 4 strain amplitude [x 10 -20 ] 2 0 -2 power [x 10 -42 / Hz] power [x 10 -42 / Hz] -4 10 -6 A1B1G1 @ 1kpc 8 -8 6 -10 -10 -5 0 5 10 15 20 25 30 4 time [msec] 2 3000 x 10 -42 3000 2000 0 -10 -5 1000 10 0 2500 5 0 frequency [Hz] 8 10 frequency [Hz] time [msec] 6 2000 4 1500 2 0 1000 power [/Hz] 500 0 -10 -5 0 5 10 15 20 time [msec] by R.Honda (Osaka City Univ.)

Example A4B2G2 type I/II @100pc 100pc Type I 5000 5000 300 300 4000 frequency [Hz] 4000 frequency [Hz] 200 200 3000 100 3000 100 2000 0 1000 2000 0 0 1000 -10 -5 0 5 10 time [msec] @500pc 0 500pc 5000 -15 -10 -5 0 5 10 15 12 4000 frequency [Hz] time [msec] 9 6 3000 3 Spike noise 2000 0 TAMA noise 1000 5000 power 0 4500 12000 -10 -5 0 5 10 10000 4000 time [msec] 8000 @1kpc 3500 frequency [Hz] 1000pc 6000 3000 5000 4000 6 2000 2500 4000 frequency [Hz] 0 4 2000 3000 2 1500 2000 0 1000 1000 500 -8 -6 -4 -2 0 2 4 6 8 0 -10 -5 0 5 10 time [msec] time [msec] by R.Honda (Osaka City Univ.)

Clustering (recognization of connected region) peak hight : P(t 0 ,f 0 ) cluster threshold : P (t 0 ,f 0 ) 1/2 f 0 t 0 cluster characteristics parameters : � t � tP ( t, f ) t1s = t1v = S V � ( t − t1s) 2 � ( t − t1v) 2 P ( t, f ) t2s = t2v = S V � = 1 S � ( t − t1s) 3 � ( t − t1v) 3 P ( t, f ) t3s = t3v = t,f V (t2v) 3 / 2 S (t2s) 3 / 2 � = P ( t, f ) V � ( t − t1v) 4 P ( t, f ) � ( t − t1s) 4 t4v = t4s = t,f V (t2v) 4 / 2 S (t2s) 4 / 2 by R.Honda (Osaka City Univ.)

TF-cluster : Event Selection exclude Gauss noise exclude TAMA noises f1s vs f2s t1s/S vs f1s/S 25 0.4 TAMA SNR>100 gauss noise 0.3 DFM 1pc DFM 10pc 20 DFM 50pc 0.2 DFM 100pc 0.1 15 f1s/S t2s 0.0 10 -0.1 TAMA SNR>100 -0.2 gauss noise 5 DFM 1pc DFM 10pc -0.3 DFM 50pc DFM 100pc 0 -0.4 -10.0 -8.0 -6.0 -4.0 -2.0 0.0 2.0 4.0 6.0 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 t1s t1s/S -2.0 ≤ f1s ≤ 2.0 S ≥ 4 -1.5 ≤ t1v ≤ 1.5 f2s ≤ 5.0 F ≤ 4 t2v ≤ 3.0 f4v ≤ 6.0 (1250Hz) (t1s 2 +f1s 2 ) 1/2 /S ≤ 0.15 t2v 1/2 /S ≥ 0.04 by R.Honda (Osaka City Univ.)

Efficiency and Selected Events efficiency = 86 % within 10pc (SNR > 70) N = 152 event for 1.26 x 10 5 sec data Rate = N / (T x efficiency) = 1.4 x 10 -3 events/sec = 4.9 events/hour by R.Honda (Osaka City Univ.), Master Thesis, Feb. 2007

Echeverria (1989) central frequency Quality factor 2. Ringdown GW from black-hole quasi-normal mode inspiral-merger Ringdown Binary, BH formation Kerr BH SN expl. QNMs core collapse perturbed BH Waveform: Damped sinusoid (Quasi-normal modes) h ( t ) = exp( − πf c t/Q ) sin(2 πf c t ) f c = 3 . 2 × 10 4 [Hz] 1 − (1 − a ) 0 . 3 � � M/M � M: Mass Q = 2 . 0(1 − a ) − 0 . 45 a: Spin * Probe for BH direct observation * BH physics in inspiral-merger, core collapses, ... * Good SNR expected, ~ 100@10kpc (TAMA sensitivity) by Y.Tsunesada (NAO, Tokyo Institute of Technology)

Matched Filter Design for BH Ringdown � s ( f ) h ∗ ( f ; f c , Q ) s(f): signal + noise ρ = d f h(f): template S n ( f ) Sn(f): Weight (noise power) Template construction in (fc, Q) plane (Nakano, Takahashi, Tagoshi, Sasaki, PRD 2003) 35 fc = 100 ~ 2500 [Hz] 30 Q (a = 0 ~ 0.998) Q = 2 ~ 33.3 25 20 682 templates (SNR loss < 2%) Q 15 10 CPU Time � N tmplt � T 1 50s = 130 [sec] 5 682 Intel PenIV 2.5GHz 0 � N tmplt � � � 16 400 450 500 550 600 T 1000h = 6 . 5 [days] f c 682 f c [Hz] N CPU by Y.Tsunesada

Trigger Rate of the Ringdown Search R ( f c ) = N trg ( f c ) 1 1 � ( f c ) 1 − (false dismissal) T obs 100 Preliminary DT6 diff f c > 1500Hz: DT6 cum (M < 20M solar ) DT8 diff 10 DT8 cum Rate [H -1 ] (SNR > 20) DT9 diff DT6: DT9 cum R < 4 . 6 [H − 1 ] 1 DT8: R < 1 . 8 × 10 − 1 [H − 1 ] 0.1 DT9: R < 3 . 4 × 10 − 2 [H − 1 ] 0.01 (SNR > 20) 500 1000 1500 2000 2500 Trigger Rate (DT9) < 1ev/day Ringdown Frequency [Hz] by Y.Tsunesada

Q = Kerr parameter fc = Mass of BH BH Mass Spectroscopy ... Ringdown GW detection can measure; % % % % Tsunesada, Kanda et al. Phys.Rev.D 71, 103005 (2005)

Burst : only numerical predicted, power filter Ringdown : anlytical prediction, matched filter Spike, Glitch, Steps... Non-Gaussianity Instabilit Short GW search requires ‘silent detector’. 3. Keyword for short signal searches Different types of ‘waveforms’ and search methods Even so, there are same noise sources !

Burst GW Excess power filter, ALF, TF cluster BH ringdown GW Matched Filter Summary TAMA searched for short GW signals, and derive upper limits: The data analysis evaluated a kind of TAMA detector noise characteristics.