Search for black hole hyperbolic encounters with gravitational wave detectors

In recent years, the proposal that there is a large population of primordial black holes living in dense clusters has been gaining popularity. One natural consequence of these dense clusters will be that the black holes inside will gravitationally scatter off each other in hyperbolic encounters, emitting gravitational waves that can be observed by current detectors. In this paper we will derive how to compute the gravitational waves emitted by black holes in hyperbolic orbits, taking into account up to leading order spin effects. We will then study the signal these waves leave in the network of gravitational wave detectors currently on Earth. Using the properties of the signal, we will detail the data processing techniques that can be used to make it stand above the detector noise. Finally, we will look for these signals from hyperbolic encounters in the publicly available LIGO-Virgo data. For this purpose we will develop a two step trigger. The first step of the trigger will be based on looking for correlations between detectors in the time-frequency domain. The second step of the trigger will make use of a residual convolutional neural network, trained with the theoretical predictions for the signal, to look for hyperbolic encounters. With this trigger we find 8 hyperbolic encounter candidates in the 15.3 days of public data analyzed. Some of these candidates are promising, but the total number of candidates found is consistent with the number of false alarms expected from our trigger. (c) 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Automated summary

This paper discusses the emission of gravitational waves by primordial black holes in dense clusters through hyperbolic encounters, and how to analyze these signals using gravitational wave detectors. By using specific data processing techniques and triggers, 8 hyperbolic encounter candidates were found in the analyzed public data.