Abstract
The laser interferometer space antenna (LISA) is due for launch in the mid 2030’s and the gravitational wave (GW) community is contributing significant efforts to refine, explore and develop the tools needed for accurate identification and parameter estimation of signals which LISA will be sensitive to. LISA data analysis is still in its infancy and the focus of this thesis is to contribute to this effort by developing open-source software tools to enable the search and inference of massive black hole binary systems.We begin with an overview of GW theory which describes the mathematical background of how we model a GW signal, specifically emitted from compact binary systems. We then overview the current data analysis applications used in laser interferometer detectors to search for and infer properties of compact binary mergers. We also outline the waveform models used in this work and their description to higher order multipoles. We then move on to apply GW theory to the LISA constellation and demonstrate the conventions/approximations being made when modelling the arm response from LISA as a GW signal passing through.
In Chapter 3, we introduce a full detection and inference pipeline for massive black hole binary signals. This section concentrates on (2,2) multipole only signals in simulated LISA noise with galactic binary confusion noise. We demonstrate that the application of a template bank search is a viable approach to detecting these signals at a match with the injected signal of ≈ 0.92 and above. We then demonstrate that this match is sufficient to inform our application of the heterodyne likelihood in the parameter estimation stage of the pipeline. We recover all intrinsic parameters within a 90% confidence interval for all signals injected into a blind data challenge, and demonstrate a well known degeneracy in sky position.
In Chapter 4, we demonstrate the recovery of higher order multipole massive black hole binary signals in zero noise. In particular we outline modifications made to an application of a heterodyne likelihood model, which considers each multipole separately whilst also appropriately dealing with multipole cross terms ensuring the correct computation of the likelihood. We show the importance of the inclusion of higher order multipole content and the consequences of not accounting for multi- poles which contain significant signal to noise (SNR) information, leading to biased parameter estimation. We recover the parameters of 6 injections within a 90% confi- dence interval and verify that our inference application works on signals with higher order multipoles.
In Chapter 5, we develop a full pipeline to search and infer higher order multi- pole massive black hole binary signals building on the works found in the previous sections. We incorporate the Simple PE algorithm to act as a refinement stage from the search to the inference, to ensure the heterodyne likelihood is well informed as to not bias parameter estimation. By implementing Simple PE as a refinement stage to inform the heterodyne likelihood, we recover injections made in simulated LISA noise with galactic binary confusion noise within 90% confidence intervals and note that further development is needed to deal with the degeneracy found within the extrinsic parameter space.
Finally, we summarise the full work in the conclusion and explore the potential further work that could follow up what this thesis has begun to explore. In par- ticular, exploration in the significance of higher order multipoles has on parameter estimation and the accuracy of fiducial waveforms in the heterodyne likelihood.
| Date of Award | 25 Feb 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Laura Nuttall (Supervisor), Ian Harry (Supervisor) & David Wands (Supervisor) |