Searching for Gravitational Waves from
Binary Inspirals with LIGO
Duncan Brown
University of Wisconsin-Milwaukee
for the
LIGO Scientific Collaboration
Inspiral Working Group
LIGO-G030671-00-Z
S1 Binary Neutron Star Search Results
•
B. Abbott et. al. “Analysis of LIGO Data for Gravitational Waves from Binary Neutron Stars”
gr-qc/0308069, submitted to PRD
•
Searched total of 236 hours of LIGO data
»
•
Used LHO 4k and LLO 4k single IFO and double coincident data
Average sensitivity to an optimal 2 x 1.4 Msun binary at r = 8
»
LLO 4k: 176 kpc, LHO 4k: 46 kpc
•
No double coincident inspiral signals were found
•
Loudest event found at r = 15.9 in LLO 4k
»
Not an inspiral signal: due to a photodiode saturation in the interferometer
Upper Limit:
R90% < 1.7 x 102
per year per MWEG
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
2
S2 Analysis Goals
• Binary Neutron Star Search
» Well modeled waveforms and known population
» Detection search for non-spinning binaries
» In absence of detection place upper limit on rate
• Binary Black Hole Search
» Waveforms not well known, no population model
» Search using waveforms from non-spinning BCV detection family
» See talks by Cokelaer and Messaritaki
• Binary Black Hole MACHO search
» Proposed population of ~ 0.5 Msun binaries in galactic halo
» Well modeled waveforms and theoretical population
» Detection search and upper limit on rate
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
3
S2 Binary Search Pipeline
•
LIGO interferometers recorded data from Feb 14 – Apr 14 2003 (S2)
•
Apply Data Quality cuts to raw data to obtain science segments
•
Only use coincident data (either double or triple)
•
Perform a triggered search of the coincident data
•
Apply instrumental vetoes to inspiral triggers
•
Demand coincidence between triggers from different interferometers
•
Use time slides to determine threshold for desired background rate (Brady)
•
Measure pipeline efficiency and set upper limit if no detection is made
•
Tune the pipeline on playground data and then run the search on full data set
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
4
S2 Binary Inspiral Search Sensitivity
• Average sensitivity to
optimally oriented 2 x 1.4 Msun
neutron star binary at r = 8
in playground data:
» LLO 4k: 1.81 Mpc
» LHO 4k: 0.90 Mpc
» LHO 2k: 0.60 Mpc
• Sensitive to inspirals in
»
»
»
»
Milky Way
Magellanic Clouds
Andromeda
M33, M32, M110
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
5
Data Quality Cuts
•
Good IFO data flagged as science mode
by operators during run
•
Extra information used in deciding what
data to analyze:
»
»
»
»
LIGO-G030671-00-Z
Exclude photodiode saturations
Exclude data without calibration lines
Exclude data with invalid timing
Exclude times with excess noise in LHO 4k
LIGO Scientific Collaboration - Inspiral Working Group
6
Binary Neutron Star Templates
• Search for inspiral signals with matched filtering
»
»
»
»
»
Templates: 2 pN stationary phase waveforms 1.0 < (m1,m2) < 3.0 Msun
Generate bank for each chunk with maximum 3% loss in signal-to-noise
Apply a low frequency cutoff of 100 Hz to data
15 x 256 sec data segments overlapped by 128 sec
Median power spectral estimate using 15 segments
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
7
Generation of Inspiral Triggers
• Resample data to 4096 Hz and high pass above 90 Hz
• Compute median PSD for 15 segments of length 256 sec
• Matched filter templates to obtain signal-to-noise ρ
• If SNR ρ > ρ* compute template based veto, c2
» Small values of c2 indicate that ρ was accumulated in a manner consistent
with an inspiral signal: If c2 < c2 * then record trigger at maximum ρ
• Triggers are clustered within duration of each template
• Multiple templates can trigger at same time
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
8
Instrumental Vetoes
•
Construct vetoes to remove spurious inspiral triggers
»
»
Some inspiral triggers are due to “obvious” instrumental glitches
Look for explanation of spurious inspiral triggers in other channels
– Glitch monitors on auxiliary interferometer channels
– Physical environment monitoring channels
Gravitational
Wave Channel
Beam Splitter
Pick off
•
Tune vetoes on playground then apply to inspiral triggers
»
See talk by Christensen and Shawhan
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
9
Trigger Coincidence Test
• Look for coincident triggers
» Present in all interferometers
» Coincident to within 11 ms between sites, 1 ms at the same site
» Coincident in both mass parameters
•
L and H detectors are not
co-aligned, so ratio of effective
distance varies
•
Cannot use an amplitude cut on
coincident signals
•
Could use arrival time of signals
»
»
LIGO-G030671-00-Z
Obtain information about sky position
from time delay between sites
Will not be used in S2
LIGO Scientific Collaboration - Inspiral Working Group
10
Threshold Selection
•
Time slide the data from one interferometer and look for coincidences
between shifted and un-shifted interferometer
»
»
Slide by longer than a chirp length
Any coincidences must be due to background, not events
•
Pick signal-to-noise threshold, ρ* , so only get a coincident trigger
1 in 100 times
•
Look for coincidence in un-shifted data
»
Any event above threshold has 99% chance of being an event
•
Best result of search would be a detection!
•
Follow up candidate events in other interferometer channels
(auxiliary interferometer, PEM, etc.)
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
11
Upper Limit on the Rate
•
In absence of detection, construct an
upper limit on event rate
•
Simulate a population of binary
neutron stars
•
Inject signals from population into data
from all three LIGO interferometers
»
»
•
Determine efficiency, e, for detection of
simulated signals at threshold ρ*
»
•
Inject in software
Validated by hardware injections (Fairhurst)
Efficiency e = Ndet / Ninj
Rate ] ( e(ρ*) T )-1
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
12
Playground Results: Pipeline Efficiency
Detected
Andromeda
Injections
Average L1 Sensitivity
Average H1 Sensitivity
Missed
Andromeda
Injections
(Too far away)
Detected
Milky Way
Injections
LIGO-G030671-00-Z
e = 0.3044
LIGO Scientific Collaboration - Inspiral Working Group
13
Conclusions
• S1 result: R90% < 1.7 x 102 per year per MWEG
• S2 binary neutron start search is almost complete
» Final results will be under internal LSC review in January
• Two other searches underway that are not yet
mature:
» Non-spinning binary black hole search
» Search for binary black holes MACHOs in the galactic halo
LIGO-G030671-00-Z
LIGO Scientific Collaboration - Inspiral Working Group
14