Case studies with GPBilby of glitch-contaminated transient gravitational waves

Abstract

In their fourth observing run, the LIGO--Virgo--KAGRA gravitational-wave observatories have found hundreds of new signals, but many are contaminated by non-Gaussian transient noise artefacts known as glitches. Left unaddressed, glitches can bias parameter inference and lead to misleading astrophysical conclusions. We present a series of case studies using GPBilby, a parameter estimation tool that employs a time-domain likelihood jointly modelling the astrophysical signal with a physical waveform and non-Gaussian noise with a Gaussian process. We first show that when the detector noise is Gaussian, GPBilby produces results consistent with those obtained with the standard Gaussian-noise likelihood, and then consider events affected by non-Gaussian features. For GW231123, the highest-mass binary black hole candidate observed to date, analyses using IMRPhenomXPHM reveal coherent residual structure that leads to measurable shifts in inferred source parameters. In contrast, analyses employing NRSur7dq4 show no significant excess residual power and remain consistent across likelihood choices. This demonstrates that waveform systematics and flexible noise modelling are intrinsically coupled, as the Gaussian process terms can partially absorb coherent waveform mismatches. For GW191109, we find that evidence for spin misalignment remains robust despite glitches in both LIGO detectors. For GW230630_070659, excluded from GWTC-4.0 owing to poor data quality, we find the data to be consistent with a BBH waveform model, with no additional residual power identified by the Gaussian process component. Overall, these results highlight how GPBilby can be used to perform glitch-robust inference and as a tool to understand waveform modelling systematics.

Figures (20)

• Figure 1: The whitened time-domain strain from the LHO data surrounding https://gwosc.org/eventapi/html/GWTC-2.1-confident/GW150914/v4/. In the upper panel, we show the raw whitened strain as a black curve and the measurement errors (estimated from the calibration envelope) as a one-standard-deviation blue band around the whitened strain. In the lower panel, we show the measurement error as a blue band, and we show the 68% non-symmetric interval estimated from the residual directly to validate that the errors are approximately symmetric. Differences in the whitened strain relative to Fig. 6 of LIGOScientific:2016vlm arise from the use of a different PSD and data-conditioning procedures.
• Figure 2: Violin plots of selected https://gwosc.org/eventapi/html/GWTC-2.1-confident/GW150914/v4/ source parameter posteriors for the Whittle likelihood (WL), GPBilby with a single jitter term (GP-J), and GPBilby with a jitter and SHO term (GP-JS). The left side of each violin shows the posterior obtained using the GWTC-2.1 PSD, while the right side shows the posterior obtained using the newly computed PSD and the GWTC-4.0 settings.
• Figure 3: Plots of selected https://gwosc.org/eventapi/html/GWTC-2.1-confident/GW150914/v4/ source parameter posteriors, with respect to the fraction of data notched. The blue filled regions show how the 90% interval changes as the proportion of notched data increases. The orange dots represent the median values and the blue shaded scatter show the full distribution of posterior samples.
• Figure 4: Posterior distributions for the SHO term parameters of the GPBilby GP-JS analyses of https://gwosc.org/eventapi/html/GWTC-2.1-confident/GW150914/v4/ (left panel: LHO, right panel: LLO). The analyses using the GWTC-2.1 PSD are shown in green, while those using the newly computed PSD are shown in purple. For the LHO analysis with the newly computed PSD, the frequency posterior is tightly peaked at $f^{0}_{\rm H1} = 60.53^{+0.98}_{-1.17}$ Hz.
• Figure 5: Violin plots of selected https://gwosc.org/eventapi/html/GWTC-2.1-confident/GW170814/v4/ source parameter posteriors for the Whittle likelihood (WL), GPBilby with a single jitter term (GP-J), and GPBilby with a jitter and SHO term (GP-JS).