PhD defence: Precision science with Einstein Telescope - Cleaning glitches and acclerating parameter inference for the third generation of gravitational wave detectors

The Einstein Telescope is a third-generation gravitational wave detector which is expected to be an order of magnitude more sensitive than current detectors. While the current detectors enabled various measurements which were first of their kind (such as the strong field test of general relativity, an independent measurement of the Hubble constant, probing the dense nuclear matter inside the neutron stars, and many more), the third-generation detectors may not only bring about unprecedented accuracy and precision to these measurements but also enable new first-of-their-kind measurements (the detection of primordial black holes or phase transitions in the neutron star matter, for example).

While the improved sensitivity is desirable, it makes it harder to extract physics-related insights from data. Of particular concern are transient noise sources, also known as glitches, overlapping with the gravitational wave signal(s). For the third-generation detectors, we expect many more glitches overlapping with the signals compared to the current scenario, making it crucial to develop a methodology to “de-glitch” the data without corrupting the underlying signal.

To address this challenge, we develop a new framework to accurately remove glitches overlapping with gravitational wave signals. The framework is unique to the triangular geometry of the Einstein Telescope, which means it cannot be realised for other proposed geometries such as the two distant, mis-aligned L-shaped detectors. Exploiting the triangular geometry, our framework outperforms the traditional methodology of de-glitching used for L-shaped detectors, showcasing a clear edge of the triangle geometry over the rest. In this context, our work stands to play an important role in the design studies related to the Einstein Telescope, underway at the time of writing of this thesis.

Bayesian parameter inference plays a key role in the measurements of the source parameters (such as masses, spins, sky-location, and distance) of a gravitational wave signal. Increasing costs of Bayesian parameter inference is another challenge posed by the improved sensitivity of the detectors. To address this challenge, we develop a relative-binning based framework to reduce the cost of Bayesian inferences. We benchmark our methodology on a large number of real as well as simulated signals for the current and next generation detectors. We also test the validity of our rapid inference framework on overlapping signals and lensed gravitational wave signals. In addition, we perform simulation to demonstrate how gravitational lensing can help constrain gravitational wave propagation beyond general relativity.