PIRSA Logo


PERIMETER INSTITUTE RECORDED SEMINAR ARCHIVE

Bayesian Inference on Numerical Injections
Speaker(s): Ilya Mandel
Abstract: We describe a Markov-Chain Monte-Carlo technique to study the source parameters of gravitational-wave signals from the inspirals of stellar-mass compact binaries detected with ground-based detectors such as LIGO and Virgo. We can apply this technique to both spinning and non-spinning waveforms and we use a variety of tools like parallel tempering to improve the sampling efficiency of the algorithm in a multi-dimensional parameter space. We describe new developments in model-selection techniques for distinguishing between alternative signal models. We present preliminary results from the application of these techniques to data sets containing injections of numerical-relativity waveforms into simulated Gaussian detector noise. We study the source parameters of signals from the inspirals of stellar-mass compact binaries detected with ground-based gravitational-wave detectors such as LIGO and Virgo. We use automatic adaptation of the step size and take into account the correlations between parameters to efficiently probe the parameter space while keeping the algorithm suitable for a wide range of signals. We shall discuss the performance of the MCMC algorithm and the typical measurement accuracy of the source parameters as a function of the binary parameters and the number of detectors in the network. We will show that despite the lower positional accuracy compared to other astronomical observations an association of a gravitational-wave event with e.g. an electromagnetic detection is possible with three or even two 4-km-size interferometers.
Date: 26/06/2010 - 10:30 am
Valid XHTML 1.0!