Inpainting over the cracks: challenges of applying pre-merger searches for massive black hole binaries to realistic LISA datasets Preprint: APS/123-QED
Gareth Cabourn Davies, Ian Harry · 2026
Read Inpainting over the cracks: challenges of applying pre-merger searches for massive black hole binaries to realistic LISA datasets Preprint: APS/123-QED by Gareth Cabourn Davies, Ian Harry (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-13738.txt.
Introduction
Gareth Cabourn Davies and Ian Harry demonstrate two techniques for premerger identification of massive black hole binaries in realistic LISA data. They apply both to the LISA Data Challenge dataset Sangria-HM, recovering all 14 signals expected visible at least half a day before merger. The inpainting method identifies premerger signals even with data gaps, recovering a signal despite 3 day-long gaps added to the 14 days before merger.
Catalog record
| CREATOR | Gareth Cabourn Davies, Ian Harry |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44797 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.13738
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Inpainting over the cracks: challenges of applying pre-merger searches for massive black hole binaries to realistic LISA datasets Preprint: APS/123-QED Gareth Cabourn Davies, Ian Harry Abstract A key science target of the Large Interferometer Space Antenna (LISA) is to carry out multi-messenger observations of massive black hole binaries, observing the merger simultaneously in gravitational waves and with electromagnetic observatories. Identifying that a merger is happening and providing an updating estimate of the sky location in the hours, days and weeks before the merger is critical to enable electromagnetic observations of the merger event. In this work we demonstrate and compare two methods for premerger identification of massive black hole binaries; a zero-latency filter approach and, for the first time, an approach using an “inpainting” technique. We apply these methods to the LISA Data Challenge dataset 2a–Sangria-HM–and demonstrate the successful recovery of the 14 signals in the dataset that we expected to be identifiable at least half a day before merger. We demonstrate that the inpainting method can identify premerger signals even when gaps are present in the data, demonstrating the recovery of a signal even when three day-long data gaps are added to the 14 days preceding merger. Finally, we explore the challenge of overlapping signals, using a region of overlapping signals in the Sangria-HM dataset, all of which merge within a 10-day window, and show how removing signals that have been confidently identified from the data allows us to identify quieter signals in the same period. † † footnotetext: Both authors contributed equally to this work. I Introduction The LISA ( LISA ) gravitational-wave observatory, planned for launch in the mid-to-late 2030s 3 ; 17 , will unlock our ability to observe the universe in the low-frequency gravitational-wave band 33 . A key science target for LISA is the observation of MBHB . These signals are expected to have a very large signal-to-noise ratio, making the problem of identifying them trivial 33 . A significant challenge lies in reliably inferring the parameters of these sources, especially given a large population of other sources present in the data, and this problem has been extensively explored over the last two decades, primarily through the LISA Data Challenges 7 ; 6 ; 5 ; 8 ; 9 ; 4 ; 10 ; 12 ; 36 . The large signal-to-noise ratio that MBHB are expected to have unlocks a particularly interesting possibility; that of observing these systems in the weeks to months before they merge 13 ; 23 ; 31 . Such pre-merger observations provide an early warning that a loud merger event is about to happen and can provide sky-localization estimates 13 ; 26 ; 28 ; 32 for these incredibly science-rich events 19 . This advanced notice is critical to enable electromagnetic facilities to be directed toward the source region rapidly to maximize the possibility of multi-messenger observation and to potentially reschedule any planned repositioning operations of LISA itself to reduce the chance that it is not taking data when the merger arrives 17 ; 16 . The technique that we introduced in 13 offers a method for immediate analysis of LISA data. It allows for the accurate computation of a matched filter using an acausal, zero-latency whitening filter that alleviates the need to wait (up to a day) for additional data to arrive before being able to filter. If one has to wait an additional day, the observation window may have passed, or LISA itself may not be taking data while repositioning 17 ; 16 . Figure 1 : Comparison of power spectral densities used in this work. The power spectral density is estimated from the Sangria-HM dataset using the Welch method with a segment duration of 18.25 days. The model shown is the noise model used to produce the Sangria-HM dataset, and is shown with and without unresolvable Galactic binary sources. We show the power spectral density model and estimate smoothed using the method described in Section II.2 . We also show the characteristic strain for a representative MBHB signal from Sangria-HM (Signal 2), which shows that at 14, 7 and 4 days-before-merger, the MBHB does not overlap the Galactic binary confusion noise in frequency, but when the signal reaches around one day before merger (blue line), the Galactic binary confusion noise starts to overlap with the signal. However, our method is not robust to gaps in the LISA data and only searches for signals at discrete values of the gap to merger. We therefore introduce a second possibility for premerger observation, applying the inpainting technique of 40 as an alternative to the zero-latency whitening filter. In this paper, we present the application of both the zero-latency and inpainting techniques to the “Sangria-HM dataset” produced as part of the LISA Data Challenges. We demonstrate that, where there are no gaps in the data, both methods compare very well, and identify 14 of the 15 MBHB signals in the dataset at least 0.5 days before merger–the remaining signal being too quiet for detection 0.5 days before merger. We demonstrate how the inpainting technique works well when gaps are added to the dataset, allowing premerger identification even in an example where the data contains three separate day-long gaps. The paper is organized as follows: In section II we describe the Sangria-HM dataset and evaluate the potential for premerger detection of the 15 MBHB signals added to it. In section III we present the results of our zero-latency search on this dataset. In section IV we describe how we adapt the inpainting technique of 40 to the problem of premerger observation of MBHB before presenting results of that search in section V . In section VI we discuss the analysis of Signals 2 - 5, whose merger times are all contained within a single ten-day window, causing some overlap in their identification. Finally we conclude in section VII . All of our results and figures are fully reproducible, with the code used to make them publicly available. Please see the data release at https://icg-gravwaves.github.io/lisa_premerger_sangria/ for more detail. II The Sangria-HM Dataset The LISA Data Challenges were designed to simulate comprehensively modeled data as an introduction to LISA data analysis 12 ; 36 . Here we use the dataset known as Challenge 2a, also known as Sangria-HM 37 . Sangria-HM consists of both a “training” and “blind” dataset, we analyze here the training dataset to more easily compare the values of the signals to our results. The Sangria-HM training dataset is a year long and contains Gaussian instrumental noise, a foreground of Galactic binaries and MBHB systems, both with parameters derived from astrophysical models. The MBHB are simulated using the PhenomHM waveform model 20 . In our previous work 13 , we demonstrated the zero-latency technique on a dataset that did not include a foreground of Galactic binaries. This foreground, made up of millions of white dwarf binaries in our galaxy which are nearly monochromatic in frequency, occupy the same frequency range as mergers of MBHB . Many of the Galactic binary systems are not expected to be resolvable, and so must be considered an extra source of noise for MBHB searches. The contribution of this ‘noise’ will vary throughout the year, as