Long-duration transient, gravitational-wave search pipeline

TL;DR

This paper introduces a new gravitational-wave data analysis pipeline optimized for detecting long-duration, unmodeled transient signals, significantly improving sensitivity and reducing computational costs compared to previous methods.

Contribution

The paper presents a novel pipeline for long-duration gravitational-wave transients that enhances detection efficiency and computational speed, adaptable to multiple detector networks.

Findings

Detection efficiency increased depending on signal morphology

Computing time reduced by at least a factor of 10

Generalized to arbitrary detector networks

Abstract

As the sensitivity and observing time of gravitational-wave detectors increase, a more diverse range of signals is expected to be observed from a variety of sources. Especially, long-lived gravitational-wave transients have received interest in the last decade. Because most of long-duration signals are poorly modeled, detection must rely on generic search algorithms, which make few or no assumption on the nature of the signal. However, the computational cost of those searches remains a limiting factor, which leads to sub-optimal sensitivity. Several detection algorithms have been developed to cope with this issue. In this paper, we present a new data analysis pipeline to search for un-modeled long-lived transient gravitational-wave signals with duration between 10 and 1000 s, based on an excess cross-power statistic in a network of detectors. The pipeline implements several new features that are intended to reduce computational cost and increase detection sensitivity for a wide range of signal morphologies. The method is generalized to a network of an arbitrary number of detectors and aims to provide a stable interface for further improvements. Comparisons with a previous implementation of a similar method on simulated and real gravitational-wave data show an overall increase in detection efficiency depending on the signal morphology, and a computing time reduced by at least a factor 10.