Adaptive clustering procedure for continuous gravitational wave searches

TL;DR

This paper introduces an adaptive clustering method for continuous gravitational wave searches that improves noise rejection and sensitivity by tailoring cluster volumes to data characteristics, enhancing detection capabilities.

Contribution

The paper presents a novel adaptive clustering procedure that adjusts cluster sizes based on data, improving noise rejection and sensitivity in gravitational wave searches.

Findings

Enhanced noise rejection at fixed detection thresholds.

Increased sensitivity in gravitational wave searches.

Successfully applied in Einstein@Home O1 data analysis.

Abstract

In hierarchical searches for continuous gravitational waves, clustering of candidates is an important postprocessing step because it reduces the number of noise candidates that are followed-up at successive stages [1][7][12]. Previous clustering procedures bundled together nearby candidates ascribing them to the same root cause (be it a signal or a disturbance), based on a predefined cluster volume. In this paper, we present a procedure that adapts the cluster volume to the data itself and checks for consistency of such volume with what is expected from a signal. This significantly improves the noise rejection capabilities at fixed detection threshold, and at fixed computing resources for the follow-up stages, this results in an overall more sensitive search. This new procedure was employed in the first Einstein@Home search on data from the first science run of the advanced LIGO detectors (O1) [11].