The search for gravitational wave bursts in data from the second LIGO science run

Abstract

The network of detectors comprising the Laser Interferometer Gravitational-wave Observatory (LIGO) are among a new generation of detectors that seek to make the first direct observation of gravitational waves. While providing strong support for the General Theory of Relativity, such observations will also permit new tests of physical theory in regions of strong space-time curvature and high matter-energy density. However, the observed signals are expected to occur near the limit of detector sensitivity. The problem of identifying such small signals is the primary focus of this work. This work presents a novel method for the identification of astrophysically unmodeled bursts of gravitational radiation in data from networks of interferometric detectors. The method is based on the Q transform, a multiresolution time-frequency transform that efficiently targets waveforms within a finite region of time, frequency, and Q space. The method is also based on a modification of linear prediction that greatly simplifies the resulting statistical analysis by whitening interferometric detector data prior to Q transform analysis. Together, these techniques form the basis of a complete analysis pipeline that is equivalent to a template-based matched filter search for minimum uncertainty waveforms in the whitened data stream. This method is then applied to search for gravitational-wave bursts with duration less than 1 second and frequency content between 64 and 1024 Hz in coincident data from two detectors during second LIGO science run. Although no gravitational-wave bursts are identified,

(cont.) upper bounds are reported for the rate of gravitational-wave bursts as a function of signal strength for isotropic and galactic populations of sources with both abstract and astrophysically motivated waveform. The results indicate a maximum of 0.09 events per day at the 90% confidence level for bursts with characteristic strain amplitude in excess of 10-20 to 10-19 strain Hz-l/2 depending on waveform. A comparison with previous searches demonstrates that this search is one of the most sensitive to date for gravitational-wave bursts of unknown waveform, and is inconsistent with recent indications for an statistical excess of events by the ROG collaboration at above the 99% confidence level.