The direct detection of gravitational waves with upcoming second-generation gravitational wave detectors such as Advanced LIGO and Virgo will allow us to probe the genuinely strong-field dynamics of general relativity (GR) for the first time. We present a data analysis pipeline called TIGER (Test Infrastructure for GEneral Relativity), which is designed to utilize detections of compact binary coalescences to test GR in this regime. TIGER is a model-independent test of GR itself, in that it is not necessary to compare with any specific alternative theory. It performs Bayesian inference on two hypotheses: the GR hypothesis $mathcal{H}_{rm GR}$, and $mathcal{H}_{rm modGR}$, which states that one or more of the post-Newtonian coefficients in the waveform are not as predicted by GR. By the use of multiple sub-hypotheses of $mathcal{H}_{rm modGR}$, in each of which a different number of parameterized deformations of the GR phase are allowed, an arbitrarily large number of testing parameters can be used without having to worry about a model being insufficiently parsimonious if the true number of extra parameters is in fact small. TIGER is well-suited to the regime where most sources have low signal-to-noise ratios, again through the use of these sub-hypotheses. Information from multiple sources can trivially be combined, leading to a stronger test. We focus on binary neutron star coalescences, for which sufficiently accurate waveform models are available that can be generated fast enough on a computer to be fit for use in Bayesian inference. We show that the pipeline is robust against a number of fundamental, astrophysical, and instrumental effects, such as differences between waveform approximants, a limited number of post-Newtonian phase contributions being known, the effects of neutron star spins and tidal deformability on the orbital motion, and instrumental calibration errors.
Download