Spinning Black Hole Binaries1 Search for Spinning Black Hole Binaries in Advanced LIGO: Parameter tuning of HACR Speaker: Gareth Jones Cardiff University LSC Meeting March 2005 ASIS session DCC#: G Z

Spinning Black Hole Binaries2 Overview Goal: Detection of SBBH’s in Time Frequency representation of Advanced LIGO data What is HACR? Spinning Black Hole Binaries Tuning HACR in 2 steps Results Conclusions and Future Work

Spinning Black Hole Binaries3 What is HACR? Hierarchical Algorithm for Clusters and Ridges written by R. Balasubramanian Time-frequency search Short bursts of excess power More robust than matched-filtering Implemented in GEO++ as HACRMon

Spinning Black Hole Binaries4 How HACR works 1. Identify pixels with P > η upper Time Frequency

Spinning Black Hole Binaries5 How HACR works 1. Identify pixels with P > η upper Time Frequency “Black pixels”

Spinning Black Hole Binaries6 How HACR works 2. Identify neighbouring pixels with P > η lower Time Frequency “Black pixels”

Spinning Black Hole Binaries7 How HACR works 2. Identify neighbouring pixels with P > η lower Time Frequency “Cluster”

Spinning Black Hole Binaries8 How HACR works 3. Identify clusters with N pixels > N threshold Time Frequency “Cluster”

Spinning Black Hole Binaries9 How HACR works 3. Identify clusters with N pixels > N threshold Time Frequency “Event”

Spinning Black Hole Binaries10 Spinning Black Hole Binaries Require 17 physical parameters to accurately describe waveform Spin-induced precession of orbital plane causes modulation of amplitude and phase Amplitude modulation causes of GW “archipelago” of clusters in TF map

Spinning Black Hole Binaries11 Spinning Black Hole Binaries Time Amplitude Frequency

Spinning Black Hole Binaries12 Tuning HACR For a given false alarm rate maximise probability of detection Identify combinations of threshold parameters (η upper, N threshold ) that produce a constant false alarm rate… … then for each combination of parameters measure probability of detection

Spinning Black Hole Binaries13 False Alarm Analysis 100,000s of simulated Adv. LIGO data Analyse with low values of η upper and N threshold Store maximum-power and number of pixels for each found “event” For range of η upper choose values of N threshold that give specific False Alarm rates

Spinning Black Hole Binaries14 False Alarm Analysis η upper N threshold Contours of constant false alarm rate 1 False alarm per… …hour (27.8 events) …1000s (100 events) …100s (1000 events) …10 mins (166.7 events ) …10s (10,000 events) …min ( events)

Spinning Black Hole Binaries15 Injection of SBBH waveforms Physical waveforms (1-49), (10-40) and (25-25)M solar 1000 injections each with SNR = 8,12,16 Measure number of injections recovered for each combination of (η upper,N threshold )

Spinning Black Hole Binaries16 Results η upper # of recovered injections SNR = 8 SNR = 16 SNR = 12 Expected # of FA = ~54

Spinning Black Hole Binaries17 Results Small spin modulationLarge spin modulation (1-49)M (10-40)M (25-25)M

Spinning Black Hole Binaries18 Results Small spin modulationLarge spin modulation (1-49)M (10-40)M (25-25)M η upper = 23, N threshold = 50

Spinning Black Hole Binaries19 Conclusion and Future Work More injections and at various distances Power law pattern matching for potential event clusters Tune HACR for EMRI signals in LISA (with Jonathan Gair)