1. A High Frequency Burst Search with the LIGO Interferometers
Brennan Hughey
MIT-LIGO and the
LIGO Scientific Collaboration
February 15th 2008
MIT Postdoc Symposium

2. Gravitational waves: In GR, propagate the gravitational force
Result from changing quadrupole moment of mass
(thus requiring a system that isn't spherically symmetric)
Propagate at the speed of light.
Distort space itself, stretching in one direction and squeezing in the perpendicular direction, then vice versa.
Are extremely weak: only hope of detecting them is via huge gravitational wave generators (astrophysical objects)

3. The Evidence for Gravitational Waves
Radio pulsar B , discovered in 1974 by Hulse and Taylor as part of a binary system
Long-term radio observations have yielded neutron star masses and orbital parameters
System shows very gradual orbital decay just as general relativity predicts!
Very strong indirect evidence
for gravitational radiation

4. Interferometer Concept
Orthogonal arm lengths change in different ways as they interact with a gravitational wave
Use laser to measure relative lengths L/L by observing the changes in interference pattern at the anti-symmetric port, for example, for L ~ 4 km and for a hypothetical wave of h ~ 10–21 L ~ m !
Power-recycled Michelson interferometer with Fabry-Perot arm cavities

5. Ground interferometers’ noise budget
Best strain sensitivity ~3x /Hz1/2 at 200 Hz
Displacement Noise
Seismic motion
Thermal Noise
Radiation Pressure
Sensing Noise
Photon Shot Noise
Residual Gas
Facilities limits much lower
Several ground interferometers are currently operating at or near design sensitivity

6. Interferometric Detectors
TAMA 300m
Japan
VIRGO 3km
Italy
LIGO Louisiana 4km
USA
CLIO 100m
Japan
GEO 600m
Germany
LIGO Washington 2km& 4km
USA

7. LIGO Laser Interferometer Gravitational-wave Observatory
Hanford, Washington: 2 km and 4 km detectors
Livingston, Louisiana: 4 km detector
10 ms light travel time
Managed and operated by Caltech and MIT with NSF funding
LIGO Scientific Collaboration – 500+ researchers from 45 institutions
worldwide run and analyze data from the LIGO and GEO instruments

8. Sources And Methods Template-less methods Matched filter Short
Bursts
Stochastic Background
Template-less methods
Compact Binary Inspirals
Pulsars
Matched filter
Short
duration
Long

9. LIGO Science Runs and Sensitivities
S1: 23 Aug – 9 Sep ‘02
S2: 14 Feb – 14 Apr ‘03
S3: 31 Oct ‘03 – 9 Jan ‘04
S4: 22 Feb – 23 Mar ‘05
S5: 4 Nov ‘05 – Oct '07
S5 represents a full year
livetime of triple-coincident
science quality data

10. High Frequency Burst Search
Analyzes data in kHz range
whereas previous burst searches
have been limited to region < 2 kHz
Currently being performed over
1st calendar year of LIGO's 5th
science run and will be extended to
2nd calendar year.
A “Burst” search – looking for short
transient signal without relying on
a specific model of emission
Mirrors lower frequency analysis
but with needed adjustments for
higher frequency
Work conducted with Erik Katsavounidis
Michele Zanolin and others

11. Sources of High Frequency Waves
Models which have resulted in predictions
of high frequency burst signals include:
Stellar collapse as predicted
by Baiotti, Rezzola et al.
Phys.Rev.Lett. 97 (2006)
Class. Quant. Grav. 24 (2006) S187-S206
Burrows-Ott supernovae
Ap J 600 (2004)
Low mass black hole mergers
Phys.Rev.Lett. 91 (2003)
Nonaxisymmetric hypermassive neutron stars
as predicted by Oeschlin and Janka
astro-ph/ v

12. Analysis Pipeline Qpipeline Hanford Postprocessing CorrPower
Environmental
Vetoes
Event
Candidates
Qpipeline
Livingston
Next few slides will address these steps individually

13. QPipeline
The QPipeline is a multi-resolution time-frequency search for statistically significant excess signal energy
Targets gravitational wave bursts of unknown waveform
Projects whitened data onto an overlapping bank of complex valued sinusoidal Gaussians characterized by central time , central frequency , and Q (ratio of central frequency to bandwidth):
Equivalent to a templated matched filter search for waveforms that are sinusoidal Gaussians after whitening
Measures the normalized tile energy Z, matched filter SNR , and white noise significance P, where
Reports the minimal set of non-overlapping templates that best describes the signal.

14. Postprocessing Qpipeline generates thousands of triggers,
Livingston Observatory
4 km interferometer L1
Hanford Observatory
4 km and 2 km interferometers
H1 and H2
Qpipeline generates thousands of triggers,
which must then be subjected to postprocessing
A coincidence window is applied between sites
GW's propagate as plane wave traveling at c
We require each Hanford trigger have a matching trigger
at Livingston within 20 ms (longer than light travel time)
We also require consistent central frequencies
The surviving triggers are clustered
Triggers are combined into clusters of 1 second with the characteristics of the
most dominant trigger, otherwise the same gravitational wave or other
disturbance would generally produce multiple triggers
Data quality flags are applied to remove time segments wherein we don't
trust the data sufficiently
Flags include periods with known seismic or wind activity or problems
originating within the detectors themselves
There is a fundamental set of data quality flags applied to both high and low
frequency analyses, but we studied the effects of less obvious flags
specifically at high frequencies to decide which to use

15. CorrPower
CorrPower finds correlated power between waveforms in the various interferometers.
It produces an output variable (Gamma) based on the maximum observed correlation using Pearson's linear correlation statistic.
CorrPower output is normalized so that it doesn't consider overall energy, just degree
of correlation, and thus provides a method of background rejection which is largely
independent of Qpipeline
Unphysical time lags
(on order 5 to a few
hundred seconds) between
the two sites were used to
tune the cuts

16. Environmental Vetoes
Elimination of triggers with known external causes
Trigger-by-trigger basis rather than specific time periods as in data quality flags
Seismic/wind: seismometers, accelerometers, wind monitors
Sonic/acoustic: microphones
Magnetic fields:
magnetometers
Line voltage fluctuations:
volt meters

17. Status and Future Plans
Finalizing 1st year analysis, in particular environmental vetoes
Will soon move on to second year
Will incorporate Virgo, which has sensitivities equivalent to that of LIGO at high frequencies and began its 1st science run near the end of LIGO's S5

18. Advanced LIGO Factor 10 better amplitude sensitivity (Reach)3 = rate
Factor 4 lower frequency bound
Infrastructure of initial LIGO but replace many detector components with new designs
Advanced LIGO
Increase laser power in arms.
Better seismic isolation.
Quadruple pendula for each mass
Larger mirrors to suppress thermal noise.
Silica wires to suppress suspension thermal noise.
“New” noise source due to increased laser power: radiation pressure noise.
Signal recycling mirror: Allows tuning sensitivity for a particular frequency range. 18

19. Enhanced LIGO LIGO today ~2009 Advanced LIGO ~2014 100 million
light years
Advanced LIGO
~2014 19