Applications of Markov Chain Monte Carlo methods to continuous gravitational wave data analysis

QB Astronomy;QC Physics

Veitch, John D. ; Woan, Graham

University:University of Glasgow

Department:School of Physics and Astronomy

关键词: MCMC, Bayesian Inference, Bayesian, gravitational waves, data analysis;   

Others  :  http://theses.gla.ac.uk/35/1/2007VeitchPhD.pdf

来源: University of Glasgow

【 摘 要 】

A new algorithm for the analysis of gravitational wave data from rapidly rotating neutron stars has been developed. The work is based on the Markov Chain Monte Carlo algorithm and features enhancements specifically targeted to this problem. The algorithm is tested on both synthetic data and hardware injections in the LIGO Hanford interferometer during its third science run ("S3''). By utilising the features of this probabilistic algorithm a search is performed for a rotating neutron star in the remnant of SN1987A within in frequency window of 4 Hz and a spindown window of 2E-10 Hz/s. A method for setting upper limits is described and used on this data in the absence of a detection setting an upper limit on strain of 7.3E-23.A further application of MCMC methods is made in the area of data analysis for the proposed LISA mission. An algorithm is developed to simultaneously estimate the number of sources and their parameters in a noisy data stream using reversible jump MCMC. An extension is made to estimate the position in the sky of a source and this is further improved by the implementation of a fast approximate calculation of the covariance matrix to enhance acceptance rates. This new algorithm is also tested upon synthetic data and the results are presented here.Conclusions are drawn from the results of this work, and comments are made on the development of MCMC algorithms within the field of gravitational wave data analysis, with a view to their increasing usage.