Neural network time-series classifiers for gravitational-wave searches in single-detector periods

GW detectors occasionally go offline, often for extended periods of time. While the two LIGO detectors at Hanford and Livingston are typically operated together, there are many periods when only one detector is operational. During these times, it is not possible to use the standard method of identifying gravitational-wave signals, which relies on coincident observations in multiple detectors to distinguish true signals from noise transients. As a result, many potential gravitational-wave signals that occur during single-detector periods may be missed.

We (A. Trovato, É. Chassande-Mottin, R. Flamary and N.Courty) have looked at the Livingston detector data from the first observation run (O1) of Advanced LIGO, which took place from September 2015 to January 2016. During this period, three GW events were detected: GW150914, GW151012, and GW151226. With the use of neural networks classifiers, we have indentified a fourth candidate event, which occured in the Livingston detector on 2016-01-04 12:24:17 UTC. From the parameter estimation performed with traditional methods, it is consisent with a binary black hole merger with component masses of 50 and 24 solar masses.

See the full paper on arXiv:2307.09268 for more details.