Faster search for long gravitational-wave transients: GPU implementation of the transient F-statistic

TL;DR

This paper introduces a GPU-based implementation of the transient F-statistic, significantly accelerating searches for long gravitational-wave transients and enabling broader parameter space exploration.

Contribution

A novel GPU implementation of the transient F-statistic that enhances search efficiency for gravitational-wave transients with complex models.

Findings

Achieved substantial speedup using GPU computing.

Enabled wider parameter space coverage for transient signals.

Facilitated more comprehensive searches for neutron star signals.

Abstract

The F-statistic is an established method to search for continuous gravitational waves from spinning neutron stars. Prix et al. (2011) introduced a variant for transient quasi-monochromatic signals. Possible astrophysical scenarios for such transients include glitching pulsars, newborn neutron stars and accreting systems. Here we present a new implementation of the transient F-statistic, using pyCUDA to leverage the power of modern graphics processing units (GPUs). The obtained speedup allows efficient searches over much wider parameter spaces, especially when using more realistic transient signal models including time-varying (e.g. exponentially decaying) amplitudes. Hence, it can enable comprehensive coverage of glitches in known nearby pulsars, improve the follow-up of outliers from continuous-wave searches, and might be an important ingredient for future blind all-sky searches for unknown neutron stars.