|Periodicity in Gravitational Waves
|371, Statistical Challenges in Modern Astronomy IV
|The direct detection of gravitational waves from space will herald the start of an entirely new form of astronomy, and this exciting prospect has resulted in a big effort to find the first signal in recent data from gravitational wave telescopes. Although gravitational waves can be generated in many ways, one of the more promising classes of targets for the current generation of detectors is rapidly spinning asymmetric neutron stars. The rotation of these objects will produce well-defined quasi-sinusoidal signals that one might imagine would be straightforward to pull out of the data. However there are several complications: the detector noise is highly coloured and non-stationary and the signals are so weak that many months of data need to be processed. In addition the sources move each day through the antenna pattern of the detector which, combined with Doppler shifts due to the Earth’s motion, give the sinusoids a complicated amplitude and frequency modulation.
The analysis of these data has presented some interesting challenges, and several strategies have been developed to probe different regions of parameter space using both coherent and incoherent steps in Bayesian and frequentist frameworks. Here we discuss these ideas and highlight the Bayesian methods used to search for gravitational signals from radio-loud neutron stars (pulsars). We also discuss how their development is influenced by the practical need to make sensible astrophysical statements based (currently) on the absence of any detectable signal.