Machine-learning nonstationary noise out of gravitational-wave detectors

Abstract

Signal extraction out of background noise is a common challenge in high-precision physics experiments, where the measurement output is often a continuous data stream. To improve the signalto-noise ratio of the detection, witness sensors are often used to independently measure background noises and subtract them from the main signal. If the noise coupling is linear and stationary, optimal techniques already exist and are routinely implemented in many experiments. However, when the noise coupling is non-stationary, linear techniques often fail or are sub-optimal. Inspired by the properties of the background noise in gravitational wave detectors, this work develops a novel algorithm to efficiently characterize and remove non-stationary noise couplings, provided there exist witnesses of the noise source and of the modulation. In this work, the algorithm is described in its most general formulation, and its efficiency is demonstrated with examples from the data of the Advanced LIGO gravitational-wave observatory, where we could obtain an improvement of the detector gravitational-wave reach without introducing any bias on the source parameter estimation.

GW190412: Observation of a binary-black-hole coalescence with asymmetric masses

“…In O3, this procedure was completed as a part of the calibration pipeline [38], both in low latency and offline. Additional noise contributions due to nonlinear coupling of the 60 Hz power mains are subtracted for offline analyses using coupling functions that rely on machine learning techniques [42].…”

GW190521: A Binary Black Hole Merger with a Total Mass of 150 M⊙

“…Subsequent to the subtraction conducted within the online calibration pipeline, we perform a secondary offline subtraction [55] on the LIGO data with the goal of ; the point estimate waveform from the CWB search [43] (black lines); the 90% credible intervals from the posterior probability density functions of the waveform time series, obtained via Bayesian inference (LALINFERENCE [49]) with the NRSur7dq4 binary BH waveform model [50] (orange bands), and with a generic wavelet model (BayesWave [51], purple bands). The ordinate axes are in units of noise standard deviations.…”

Sensitivity and performance of the Advanced LIGO detectors in the third observing run

“…As seen in Figure 7, the contribution of these auxiliary channels to the DARM noise is larger than expected based on coherence alone, suggesting nonlinear, bilinear, and/or nonstationary coupling to DARM. Nonstationary coupling has already been observed due to modulation from motion of the angular degrees of freedom, and can be partially removed offline [79,80]. Additional work is required to understand this type of contribution to the interferometer noise floor.…”

“…IV G 1), is expected to improve the nonlinear noise coupling during times of large ground motion and also improve interferometer duty cycle by limiting saturations. Machine learning techniques are being developed that allow offline removal of nonlinear noise contributions [79].…”