Numerical modelling of black-hole-binary mergers

The beginning of gravitational wave astronomy started on September 14th 2015 [13]. \nThe event, GW150914, was so loud that the distinct morphological features indicative \nof the merger of two inspiraling black holes was difficult to deny. \nThe estimation of source parameters and parameterised tests of general relativity \nin the strong field regime require the use of gravitational waveform models that \npredict the inspiral, merger and ringdown of binary black holes according to general \nrelativity. This thesis is focused on providing the gravitational wave community \nwith an accurate model for the gravitational wave signal emitted by coalescing, \nnon-precessing binary black holes covering the inspiral, merger and ringdown. \nThe solutions to the Einstein equations for the late inspiral, merger and ringdown \nof binary black holes can only be obtained by using numerical relativity. However, \nthe computational cost of a single simulation is on the order of weeks to months and \nprohibits a dense sampling of the parameter space. Our method is founded on the \nphenomenological modelling program, which was specifically designed to directly incorporate \nresults from numerical relativity and analytic approximations to construct \nglobal models across the parameter space for gravitational wave searches. \nWe have refined the phenomenological method and developed a new waveform \nmodel, IMRPhenomD, which is suitable not only for gravitational wave searches but \nis also sufficiently accurate to be in used to estimate the parameters of gravitational \nwave candidate events without incurring large systematic uncertainties due to waveform \nmodelling errors. Subsequent to the work presented here our waveform model \nwas also extended to include the effects of precession, which was used in the analysis \nof advanced LIGO data during its first observing run (2015-2016), including the \nanalysis of GW150914. \nWe evaluate the current state of the field of waveform modelling by performing \nnumerous comparisons between leading inspiral, merger and ringdown waveform \nmodels and find that independently developed models are largely in agreement. \nThis builds confidence in our models when we use them outside of their respective \ncalibration regions. However, there are still large regions of parameter space where \nthe models are in disagreement and we highlight these regions as urgent targets for \nnew numerical relativity simulations.