AURIGA: the data analysis

 

The data collected with a gravitational wave on the detector are dominated by noise. This means that due to the very small interaction between a gravitational wave and the detector, the signal to noise ratio (SNR) threshold has to be kept as low as possible. However there is a limit in relaxing the threshold for event detection, because of the increasing number of false alarms due to purely statistical fluctuations of the background Gaussian noise (e.g. the thermal motion of the bar). With a hundred false alarms per day against possibly a single g.w. event per century, the only hope for a robust claim comes from coincidences between many detectors operating independently.

The request to the single detector data analysis is to produce a list of candidate events. The way commonly adopted to perform the search is to assume a model for the signal, a model for the noise, and then apply Wiener-Kolmogorov matched filter theory. A threshold passing algorithm is needed to locate the event arrival time in the filtered data stream, and get the corresponding amplitude of the candidate gravitational wave events. The choice of this threshold is somewhat arbitrary, and depend on a balance between false dismissal (efficiency) and false alarms.

The filter setup depends on few parameters based on an analytical model. Once the filtering has start up, an on-line procedure begin to track the difference between the estimated values and the real values of the parameters, and they are consequently updated on a hourly basis. This adaptiveness has fast convergence times (~1 hour), but is sensitive to deviations from the model not just due to parameter mismatch. The most frequent case is when a spurious unmodeled transient excitation enters the system. The algorithm has been amended by adding a validating gate to block the buffers that do not follow a Gaussian statistic. The same criterion is used to clean the data set by removing the most noisy time span (i.e. those in which non-Gaussian buffers are 50% or more).

If you desire more information, please choose your area of interest:

 

 


Previous Next

HOME

Glossary