Activities of the LIGO Scientific Collaborations Continuous Wave Group

1
Activities of the LIGO Scientific Collaborations
Continuous Wave Group
Xavier Siemens for the LSCUniversity of
Wisconsin -- Milwaukee
2
The CW Group

• http//www.lsc-group.phys
  .uwm.edu/pulgroup/
• For the first science run (S1) set upper limit
  on a single known pulsar (J19392134) using two
  independent methods (Frequency domain, optimal
  for large parameter space searches, and Time
  domain, optimal for targeted searches).
• For the second and third science runs (S2, S3)
  the group has been involved in a variety of
  activities, which include the coherent methods
  used for S1 as well as incoherent methods

(NASA/CXC/SAO)
3
The Signal

• We search for gravitational waves generated by
  neutron star with deviations from axisymmetry.
• For example, isolated neutron star parameters

• Frequency f of source in Solar System Barycenter
  (SSB)
• Rate of change of frequency df/dt in SSB
• Sky coordinates (?, ?) of source
• Strain amplitude h0
• Spin-axis inclination ?
• Phase, Polarization ?, ?

Phase evolution Amplitude modulation
4
The Searches
Coherent searches

• Time-domain - Targeted - Markov Chain
  Monte Carlo - Frequency-domain -
  Isolated - Binary, Sco X-1

Searches over narrow parameter space
Searches over wide parameter space
Ultimately, would like to combine these two in a
hierarchical scheme
Incoherent searches

• Hough transform - Stack-Slide - Powerflux

Excess power searches
5
Time domain searches
Targeted search

• Targeting radio pulsars with rotational
  frequencies gt 25 Hz at known locations with phase
  inferred from radio data
• Upper limits defined in terms of Bayesian
  posterior probability distributions for the
  pulsar parameters
• For S2 looked at 28 known isolated pulsars
  (gr-qc/0410007) best strain upper limits were
  at the level of a few x 10-24
• For S3 will soon extend search to known neutron
  stars in binaries and will benefit from much
  improved sensitivity.
• TALK Matt Pitkin

Markov Chain Monte Carlo

• Computational Bayesian technique
• MCMC can both estimate parameters and generate
  summary statistics (pdfs, cross-correlations,
  etc)
• Initial Applications searches in restricted
  parameter space and SN1987a (location known but
  other parameters not known)
• POSTER John Veitch

6
Isolated Search

• The detection statistic is the so-called
  F-Statistic, which is the log of the likelihood
  maximized over the unknown parameters
  Jaranowski, Krolak and Schutz, PRD58 063001
  Jaranowski and Krolak, PRD59 063003
• For S2 we are doing a wide band (160 Hz-460 Hz)
  all-sky search for the 10 most sensitive hours of
  data for both the Livingston and the Hanford
  (4km) interferometers.
• We are combining the results of the two searches
  incoherently using a coincidence scheme and
  setting upper limits on 1Hz bands using a
  frequentist scheme based on the loudest
  coincident event.
• For S2 we expect upper limits in the strain to be
  in the range of several x 10-23 for quiet bands
  and around 10-22 for noisy bands
• TALK Yousuke Itoh

7
Binary Search

• S2 analysis upper-limit on Sco X-1 using a
• one-stage coherent search over short
  integration
• time (Tobs 6 hr).
• Also uses the F-statistic and a coincidence
  scheme,
• and sets a frequentist upper limit based on the
• loudest coincident event
• TALK Chris Messenger
• In the future search Sco X-1 and other known
  LMXBs (20 targets) with a hierarchical scheme
• Coherent analysis over short data chunks
• Add incoherently (stack-slide) chunks
• POSTER Virginia Re

(Artists impression NASA)
8
Stack-Slide

• An incoherent search method that stacks and
  slides normalised power to search for periodic
  sources.
• Starting point is a set of Short Fourier
  Transforms (SFTs). Then we
• Stack the (normalised) power
• Slide to correct for Doppler shifts and spin-down
• Sum and search for significant peaks
• Can be used as part of a hierarchical search with
  coherent incoherent stages

9
Hough transform

• An incoherent search method conceptually similar
  to stack-slide
• Select a threshold for the normalised power and
  label frequency bins with a 1 if the threshold is
  exceeded or 0 otherwise
• Slide to correct for Doppler shifts and spin-down
• Compute the number count for a given frequency
  bin by summing the 0s and 1s
• For S2 All-sky search using entire data set with
  one spin-down parameter for the Livingston and
  Hanford 4km interferometers. Set an frequentist
  upper limit using software injections
• Can be used as part of a hierarchical search with
  coherent incoherent stages
• TALK
  Badri Krishnan

10
Powerflux

• An incoherent search method, similar to
  stack-slide and Hough
• The main differences are that for this search the
  sum of the power for each frequency bin is
• weighted with the antenna pattern functions of
  the interferometer, and,
• weighted by the noise, which is estimated from
  neighboring frequency bins. The weighting scheme
  is chosen to minimize the variance for the
  estimator of the gravitational wave amplitude
• Can be used as part of a hierarchical search with
  coherent incoherent stages

11
Einstein_at_home

• Like SETI_at_home, but for LIGO/GEO data. When
  computer idle, screensaver does real work.
• American Physical Society (APS) will publicize as
  part of World Year of Physics (WYP) 2005
  activities, and will help with web site
  graphics.
• Use infrastructure/help from SETI_at_home developers
  for the distributed computing parts.
• Goal pulsar searches using 1 million clients.
  Support for Windows, Mac OSX, Linux clients.
• Why?
• From our own clusters we can get thousands of
  CPUs. From Einstein_at_home hope to get order(s)
• of magnitude more at low cost.
• Great outreach and science education tool.

It runs the coherent isolated F-statistic code.
Still testing but Read all about it and sign up
at http//einstein.phys.uwm.edu
12
Upcoming talks/posters

• Talks
• 1018 Badri Krishnan The Hough transform search
  for gravitational waves from pulsars
• 1055 Chris Messenger Bounding the strength of
  gravitational radiation from SCO-X1
• 1113 Mathew Pitkin Searching for gravitational
  waves from known milisecond pulsars
• 1131 Yousuke Itoh All-sky broad band search
  for continuous waves using LIGO
• Posters
• Reinhard Prix Matched filtering properties of
  all-sky searches for continuous
  gravitational waves
• Virginia Re Stack-slide hierarchical searches
  for gravitational waves from accreting
  neutron stars
• John Veitch A time-domain MCMC search technique
  for gravitational radiation of
  uncertain frequency from a targeted neutron star.