Gravitational wave detector characterisation and transient searches

Pearlstone, Brynley (2019) Gravitational wave detector characterisation and transient searches. PhD thesis, University of Glasgow.

Full text available as:
Download (49MB) | Preview
Printed Thesis Information:


Recent gravitational wave observations have allowed for a new avenue through which to observe our universe. The detection of gravitational wave events has been possible by the advent of the advanced detector era and highly sensitive gravitational wave detectors, such as advanced LIGO and advanced Virgo. Spinning non-axisymmetric neutron stars present a promising source of gravitational waves, known as continuous waves. Such gravitational waves have yet to be detected, and are predicted to be extremely weak.
In order to increase the detectors’ sensitivity to gravitational waves, they must be finely tuned to produce data without loud artefacts in either the time or frequency domains. The pursuit of identifying, characterising and removing these artefacts is knows as detector characterisation. In the first portion of this thesis, we present a range of definitions, tools, and techniques that are used in detector characterisation.
I detail three case studies of work undertaken during the course of this research. These case studies draw on a broad knowledge base from staff scientists at both LIGO observatories, the detector characterisation community within the LIGO collaboration, and continuous wave search groups within the LIGO-Virgo collaborations.
First, I investigate a broadening of the 60Hz noise line at the LIGO Livingston observatory during the final two observations of the initial LIGO detectors. Previous investigations by site staff linked short transients in magnetometer channels around the site with the this line broadening. I examine this suspected coupling with more scrutiny, and fail to find an instantaneous coupling between broad magnetometer glitches and the broadening of the 60Hz line. Secondly, we detail the creation of a list of known instrumental and environmental noise lines and combs in advanced LIGO’s second observation. Thirdly, we present an investigation of a comb of noise lines prevalent in the low frequencies (≤ 100Hz) of advanced LIGO’s first observing run. In collaboration with site staff, I identified the source of the comb, and the site staff took measures to reduce the contribution of the com to the strain channel.
In the second portion of the thesis, we focus on data analysis methods FOR “transient” continuous waves. We examine several phenomena that can result in transient-continuous waves, and review a number of existing searches for transient-continuous waves. Then, we present a new method, reduced Bayesian blocks, built around the Bayesian blocks formalism, in order to determine whether a candidate continuous wave signal exhibits transient behaviour, and loosely constrain the beginning and ending times of any transient-continuous emission.
The methods involves coarsely chunking up gravitational wave observations into several chunks of equal time. These chunks are then recompiled into all allowable blocks, where a block is a series of contiguous chunks. In each block, the evidences that the block contains signal from a given source, or only Gaussian noise, are estimated. These evidences are then used to recompile a mixed-model description of the observation. We term each arrangement of signal- and noise-blocks as “intermittencies”. By comparing the posterior probabilities of all intermittencies in a given observation, we can determine whether an observation contains a transient-continuous signal from a given source.
We examine the performance of this new method, first on simple, short, simulated dataset then on a larger set of longer simulated signals. We see that the method produces results in line with our expectations. Then, we attempt to recover the injected intermittencies in the hardware injections of O2. Finally, we follow-up four outlying candidates from the first low-frequency all-sky search for continuous waves in advanced LIGO. In all four candidates, the reduced Bayesian blocks method supported the Gaussian noise model throughout the observation.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: Gravitational wave detector, neutron star, pulsar glitch.
Subjects: Q Science > QC Physics
Colleges/Schools: College of Science and Engineering > School of Physics and Astronomy
Supervisor's Name: Woan, Professor Graham
Date of Award: 2019
Depositing User: Dr Brynley Pearlstone
Unique ID: glathesis:2019-41205
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 30 Apr 2019 09:05
Last Modified: 05 Mar 2020 21:33
Thesis DOI: 10.5525/gla.thesis.41205

Actions (login required)

View Item View Item


Downloads per month over past year