New framework to follow up candidates from continuous gravitational-wave searches

authored by: P. B. Covas, R. Prix, J. Martins
Abstract: Searches for continuous gravitational waves from unknown neutron stars are limited in sensitivity due to their high computational cost. For this reason, developing new methods or improving existing ones can increase the probability of making a detection. In this paper we present a new framework that uses Markov chain Monte Carlo (MCMC) or nested sampling methods to follow up candidates of continuous gravitational-wave searches. This framework aims to go beyond the capabilities of pyfstat (which is limited to the ptemcee sampler), by allowing a flexible choice of sampling algorithm (using bilby as a wrapper) and multidimensional correlated prior distributions. We show that MCMC and nested sampling methods can recover the maximum posterior point for much bigger parameter-space regions than previously thought (including for sources in binary systems), and we present tests that examine the capabilities of the new framework: a comparison between the dynesty, nessai, and ptemcee samplers, the usage of correlated priors, and its improved computational cost.
Organisation(s): Institute of Gravitation Physics
External Organisation(s): Max Planck Institute for Gravitational Physics (Albert Einstein Institute)
Type: Article
Journal: Physical Review D
Volume: 110
No. of pages: 16
ISSN: 2470-0010
Publication date: 22.07.2024
Publication status: Published
Peer reviewed: Yes
ASJC Scopus subject areas: Nuclear and High Energy Physics
Electronic version(s): https://doi.org/10.48550/arXiv.2404.18608 (Access: Open)
https://doi.org/10.1103/PhysRevD.110.024053 (Access: Open)

BibTeX

@article{c55a971f25214f09b2e6ad7c3c666b8c,
title = "New framework to follow up candidates from continuous gravitational-wave searches",
abstract = "Searches for continuous gravitational waves from unknown neutron stars are limited in sensitivity due to their high computational cost. For this reason, developing new methods or improving existing ones can increase the probability of making a detection. In this paper we present a new framework that uses Markov chain Monte Carlo (MCMC) or nested sampling methods to follow up candidates of continuous gravitational-wave searches. This framework aims to go beyond the capabilities of pyfstat (which is limited to the ptemcee sampler), by allowing a flexible choice of sampling algorithm (using bilby as a wrapper) and multidimensional correlated prior distributions. We show that MCMC and nested sampling methods can recover the maximum posterior point for much bigger parameter-space regions than previously thought (including for sources in binary systems), and we present tests that examine the capabilities of the new framework: a comparison between the dynesty, nessai, and ptemcee samplers, the usage of correlated priors, and its improved computational cost.",
author = "Covas, {P. B.} and R. Prix and J. Martins",
note = "Publisher Copyright: {\textcopyright} 2024 authors.",
year = "2024",
month = jul,
day = "22",
doi = "10.48550/arXiv.2404.18608",
language = "English",
volume = "110",
journal = "Physical Review D",
issn = "2470-0010",
publisher = "American Institute of Physics",
number = "2",
}