The effects of LIGO detector noise on a 15-dimensional Markov-chain Monte-Carlo analysis of gravitational-wave signals

TL;DR

This paper evaluates how LIGO detector noise affects the accuracy of parameter estimation for gravitational-wave signals from binary inspirals using a 15-dimensional Markov-chain Monte-Carlo method, highlighting challenges in distinguishing signals from noise artifacts.

Contribution

It demonstrates the robustness of parameter estimation in detector noise and explores the difficulty of differentiating real signals from glitches using MCMC analysis.

Findings

Parameter estimation accuracy is similar in Gaussian noise and real detector noise.

Markov chains do not converge in the presence of glitches, but PDFs may still resemble true signals.

Further work is needed to reliably distinguish signals from noise artifacts.

Abstract

Gravitational-wave signals from inspirals of binary compact objects (black holes and neutron stars) are primary targets of the ongoing searches by ground-based gravitational-wave (GW) interferometers (LIGO, Virgo, and GEO-600). We present parameter-estimation results from our Markov-chain Monte-Carlo code SPINspiral on signals from binaries with precessing spins. Two data sets are created by injecting simulated GW signals into either synthetic Gaussian noise or into LIGO detector data. We compute the 15-dimensional probability-density functions (PDFs) for both data sets, as well as for a data set containing LIGO data with a known, loud artefact ("glitch"). We show that the analysis of the signal in detector noise yields accuracies similar to those obtained using simulated Gaussian noise. We also find that while the Markov chains from the glitch do not converge, the PDFs would look consistent with a GW signal present in the data. While our parameter-estimation results are encouraging, further investigations into how to differentiate an actual GW signal from noise are necessary.