Gravitational wave searches with
Advanced LIGO and Advanced Virgo
Chris Van Den Broeck
on behalf of the LIGO Scientific Collaboration and the Virgo Collaboration
Rencontres de Moriond, La Thuile, March 2015
LISA
LISA
From initial to advanced detectors
§ Initial LIGO, initial Virgo
●
Active 2002-2011
●
No detections
●
Design sensitivity reached
–
Proof of the technology!
§ Advanced detectors
●
●
●
●
LISA
arXiv:1003.2480, 1304.0670
O1: September-December 2015
Advanced LIGO, limited sensitivity
O2: 2016-2017
Advanced LIGO + Advanced Virgo
O3: 2017-2018
Advanced LIGO + Advanced Virgo
(+ KAGRA?)
Design sensitivity: 2019+
Sources of gravitational waves
LISA
Coalescing binary neutron stars and black holes
Fast-spinning neutron stars
Bursts (e.g. supernovae)
“Stochastic” gravitational waves
Coalescence of compact binaries
§ NS-NS, NS-BH, BH-BH
●
Possibly intermediate mass black
holes
§ Signal is determined by 15
parameters
●
Masses, spins
●
Distance, sky position, orientation
●
Time and phase at arrival
§ Search by “matched filtering”
LISA
Coalescence of compact binaries
§ Search by “matched filtering”
●
For given choice of masses, integrate
waveform against the data
–
“Extrinsic” parameters (sky position,
orientation, distance) absorbed in
overall amplitude
–
Fold in the known properties of the
detector noise (“spectral density”)
→ “Signal-to-noise ratio”
§ Do this for a large number of mass
choices
●
●
LISA
“Template bank”
Density of templates determined by
how different waveforms are when
parameters are slightly varied
Coalescence of compact binaries
After confident detection:
Reconstruction of the source
§ Measure masses, spins, distance, sky
location, orientation, ...
§ Techniques to explore 15-dimensional
parameter space and find most likely values
§ Example: nested sampling
arXiv:0911.3820
●
●
Likelihood function
= prob(data | parameter values)
Identify nested hypervolumes in
parameter space bounded by
increasingly larger values of likelihood
§ Arrive at probability densities for parameters
●
LISA
Various (well-understood)
degeneracies
“Burst” sources
Transient sources
(other than coalescing binaries):
§ Supernovae in or near the galaxy
§ Cosmic strings
§ Long gamma-ray bursts
§ Soft gamma-ray repeater giant flares
§ Neutron star instabilities
§ Accretion disk instabilities
...
§ ... the unknown
Many of these are poorly modeled
§ Can't necessarily use matched filtering
LISA
C. Ott, LIGO-G1000171
“Burst” sources
Detection without a (good) signal model
●
Look for “coherent” signals:
●
●
Courtesy P. Sutton
●
Present in multiple detectors
Look for signal that is consistent given
different detector responses
Look for “excess power”
●
Decompose data
–
●
Make time-frequency map
●
Identify and cluster loudest pixels
arXiv:0210043
LISA
Fourier, wavelets, ...
“Burst” sources
Parameter estimation?
●
General case:
Try to characterize signal rather than source
●
●
For e.g. supernovae, some amount of
theoretical modeling has been done
●
●
arXiv:1202.3256
●
●
Time-frequency content, polarization
Large-scale numerical simulations with different
assumptions give “catalog” of waveforms
Extract main features through e.g. principal
component analysis
Bayesian model selection to infer which
physical mechanism dominates
Cosmic strings: matched filtering is possible!
●
Cosmic string cusps h ∝ f-4/3, kinks h ∝ f-5/3
–
LISA
Amplitude depends on
string tension Gμ, loop size ε, distance
arXiv:1408.5299
Continuous gravitational waves
§ Emission by fast-spinning neutron stars
●
Isolated neutron stars: asymmetry due to “starquakes”
●
Neutron stars in binaries: accretion
§ Signals are weak:
§ But, very long-lived, can integrate for long time
§ Searches:
●
●
All-sky
–
Includes stars that have not (yet) been seen
electromagnetically
–
Computationally challenging
“Directed”
–
●
“Targeted”
–
LISA
Sky position know but not frequency
(e.g. Cas A supernova remnant)
Sky position and frequency known
(e.g Crab pulsar, Vela pulsar)
Continuous gravitational waves
All-sky searches for isolated neutron stars
§ Account for Doppler modulation due to motion of
the Earth:
●
Binning in sky position:
Up to (few)x105 points
§ Given a sky position, account for change in
frequency due to spin-down:
●
Binning in spin-down coefficients:
Typically additional O(106) points
§ Hierarchical approach
arXiv:1407.8333
LISA
●
Refined grids once candidate events
●
Signal reconstruction
Continuous gravitational waves
“Targeted” searches: Known position, frequency
arXiv:1309.4027
* Upper limits on strain sensitivity for 195 pulsars from initial LIGO/Virgo runs
–
▾
LISA
Crab pulsar: GW emission no more than 1% of spin-down
Upper limits on GW emission from other known pulsars
Stochastic gravitational waves
§ Gravitational wave backgrounds of a
fundamental nature
●
Inflation: period of exponential
growth of the Universe
–
●
E.g. axion inflation
Phase transitions: fundamental
forces splitting off
●
Cosmic strings
●
...
§ Searched for by cross-correlating
between detectors:
Courtesy T. Regimbau
●
LISA
Optimal filter:
Stochastic gravitational waves
§ Stochastic background of weak
signals from far-away sources
(e.g. coalescing binaries) that are
not individually resolvable
§ Might be detectable after ~1 year of
operations at design sensitivity!
“High” event rate
“Realistic” event rate
“Low” event rate
arXiv:1112.1898
LISA
Summary
§ Advanced gravitational wave observatories are getting ready to make
first direct detections in the next few years
§ Wealth of possible sources:
●
Coalescing binaries composed of neutron stars or black holes
●
“Burst” events: supernovae, cosmic strings, ..., the unknown!
●
Continuous waves from fast-spinning neutron stars
●
Stochastic backgrounds
–
Fundamental, cosmological
–
Confusion backgrounds
(e.g. weak signals from coalescing binaries out to large distances)
§ Robust data analysis techniques are in place
More information in the LIGO-Virgo Data Analysis Whitepaper
https://dcc.ligo.org/LIGO-T1400054
LISA
Backup slides
LISA
Coalescence of compact binaries
§ Triggered searches
●
●
Short, hard gamma ray bursts
believed to be caused by
NS-NS or NS-BH mergers
Deeper searches around the time of
GRBs discovered by dedicated EM
telescopes
§ All-sky, all-time searches
●
●
arXiv:1404.5623
LISA
“Low latency”
–
Triggers to EM partners in minutes
–
Sky maps from detector network
“High latency”
–
Better estimate of significance of
candidate events
–
Better characterization of
instrument
Coalescence of compact binaries
§ Determine consistency of data with real
signal
●
●
●
Pick a number of frequency intervals
Check whether content of each interval
consistent with there being a signal
(χ2 test)
Check for coincidence between
detectors
–
in time, and in parameter space
§ Time-slide data streams with respect to
each other and analyze
●
●
●
LISA
Arrive at “background distribution” of
spurious events (because not coincident
in time)
Genuine events (or simulated ones)
will be separated from background in
terms of signal-to-noise ratio ρ, and χ2
Assign significance to candidate events
–
False alarm probability