Speaker
Description
Gravitational-wave (GW) data-analysis methods for compact binary
coalescences (CBCs) are generally divided into two categories. On the
one hand, there are modelled techniques, in which a detected signal
is compared to a theoretical model of the CBC source. On the other
hand, unmodelled approaches attempt to reconstruct the signal
without relying on a specific model by using a superposition of
wavelets.
In this work, both techniques are combined, allowing unmodelled
features to be incorporated into a modelled CBC signal. This approach
opens up the possibility to recover new physics from CBC signals that
lack precise models. A possible target is the resonant excitation of
neutron stars during inspiral, which could be represented by wavelets
superimposed on a neutron-star CBC model. The joint inference method
can also improve the estimation of astrophysical source parameters
when detector glitches - transient noise spikes - are present, by
separating the glitch from the GW signal during parameter estimation
(PE).
As a proof of concept, this study focuses on adding wavelets to
binary black hole (BBH) merger signals, as these are relatively
inexpensive to analyze. It is possible to jointly estimate both
wavelet and BBH parameters, particularly for high signal-to-noise
ratio (SNR) signals expected in next-generation detectors such as the
Einstein Telescope. The systematics of the joint inference have been
tested on a small set of injections, which indicate a detectability
threshold for wavelets in next-generation detectors. The results also
highlight a systematic issue related to the sampling priors
implemented for this inference method.
The technique also shows promise for mitigating detector glitches,
even in current detectors, without significantly degrading the quality
of the PE.
