Gravitational-wave astronomy has thus far been carried out using ground-based gravitational-wave detectors. These are essentially recycled Michelson interferometers with state of the art technology that allows scientists to detect displacements smaller than one thousandth the diameter of a proton. Currently, there are four such detectors: the twin LIGO interferometers, the European Virgo observatory and the Japanese KAGRA, with LIGO India scheduled to join the terrestrial network by the end of the 2020s.
LIGO has been operating since the mid 2000s. After a major upgrade in the early 2010s, the twin interferometers detected, for the first time, gravitational waves (GWs) from the inspiral and merger of a binary black hole system in September 2015. Since then, there have been roughly 100 confirmed detections and 200 more candidates waiting to be vetted.
Data analysis is an essential part of gravitational astronomy as most of the GW signals are buried under the detector noise. In general, the detector noise can be treated as colored Gaussian noise, though over short timescales there are many non-Gaussian noise artefacts referred to as “glitches”. Be that as it may, the noise is generally well understood. As such, we can extract signals buried in it. This is done using matched filtering, which requires the generation of billions of candidate GW signals (templates) which need to be fast and faithful. The faithfulness is gauged with respect to a few thousand existing numerical relativity simulations of binary black hole mergers. The speediness is necessary as estimating the parameters of the black holes from the detected GWs requires sampling the 15-dimensional parameter space which can take anywhere from days to weeks.
There exist now several families of GW templates referred to as waveform models or approximants. Most of these combine some kind of post-Newtonian evolution for the inspiral with plunge-merger-ringdown regimes calibrated against numerical-relativity simulations. The models have matured to the level of achieving 0.1% accuracy with respect to the simulations and can take as little as 1 millisecond to generate. Despite these advances, neither the current models nor the simulations are at the accuracy deemed necessary to meet the technological advances that will be brought in the 2030s-2040s by the third generation detectors such as the Einstein Telescope and the Cosmic Explorer.