Optimal strategies for continuous wave detection in pulsar timing arrays: Realistic pulsar noise and a gravitational wave background

Shashwat C. Sardesai, Gabriel E. Freedman, Sarah J. Vigeland, Caitlin A. Witt · 2025

Read Optimal strategies for continuous wave detection in pulsar timing arrays: Realistic pulsar noise and a gravitational wave background by Shashwat C. Sardesai, Gabriel E. Freedman, Sarah J. Vigeland, Caitlin A. Witt (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-01919.txt.

Introduction

Shashwat Sardesai, Gabriel Freedman, Sarah Vigeland, and Caitlin Witt tune the hunt for individual gravitational wave sources inside pulsar timing array data. Arrays of pulsars can sense the background hum of countless supermassive black hole binaries, and possibly single loud ones, continuous waves from individual pairs. The Fp-statistic is the standard frequentist detector for those continuous waves.

The authors study how two complications, each pulsar's own red noise and the background itself, affect the statistic's behavior. Handling the red noise by marginalizing over it versus plugging in its maximum-likelihood values made little difference, a useful simplification. They then ran the statistic on the NANOGrav 12.5-year dataset and found no evidence for continuous waves, agreeing with the previously published Bayesian analysis of the same data.

Catalog record

CREATORShashwat C. Sardesai, Gabriel E. Freedman, Sarah J. Vigeland, Caitlin A. Witt
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-01919

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Optimal strategies for continuous wave detection in pulsar timing arrays: Realistic pulsar noise and a gravitational wave background Shashwat C. Sardesai, Gabriel E. Freedman, Sarah J. Vigeland, Caitlin A. Witt Abstract Pulsar timing arrays are sensitive to low-frequency gravitational waves (GWs), such as those produced by supermassive binary black holes at subparsec separations. The incoherent superposition of GWs emitted by a cosmological population of these sources produces a gravitational wave background (GWB), while some individual sources may be resolvable as deterministic signals with slowly varying GW frequencies, which are often referred to as “continuous waves” (CWs). The ℱ p \mathcal{F}_{p} -statistic is a frequentist method of detecting these CWs. In this paper, we study how the presence of pulsar red noise and a GWB affect the ℱ p \mathcal{F}_{p} -statistic. We compare results when marginalizing over the red noise and using the maximum-likelihood values of the red noise, and find little difference between the two. We also present results of using the ℱ p \mathcal{F}_{p} -statistic to analyze the NANOGrav 12.5-year data set, where we find no evidence for CWs in agreement with the previously published Bayesian results. I Introduction Pulsar timing arrays (PTAs) use millisecond pulsars to detect the presence of low-frequency gravitational waves (GWs) Sazhin 1978 ; Detweiler 1979 ; Foster and Backer 1990 . When a passing GW passes between the line-of-sight between the Earth and the pulsar, it induces a redshift which results in a delay in the times of arrival of the pulsar signals. By measuring the time-delay (or residuals) \delta t , we can detect the presence of GWs. Millisecond pulsars are ideal for this purpose owing to their exceptional rotational stability – they can be timed to sub-microsecond precision over decades Matsakis et al. 1997 ; Verbiest et al. 2009 Many astrophysical and cosmological sources are predicted to generate low-frequency GWs that can be observed by PTAs, including supermassive black hole binaries (SMBHBs) at sub-parsec separations ( f\sim 10^{-9}-10^{-7} Hz) Rajagopal and Romani 1995 ; Jaffe and Backer 2003 ; Wyithe and Loeb 2003 ; Rosado et al. 2015 . Recently, PTA collaborations around the world have published the first evidence of nanohertz GWs, which is consistent with predictions for a gravitational wave background (GWB) produced by a population of SMBHBs Agazie et al. 2023a ; EPTA Collaboration et al. 2023 ; Reardon et al. 2023 ; Xu et al. 2023 ; Miles et al. 2025 . If the GWB is coming from a population of SMBHBs, we expect that some binaries can be resolved as individual sources Bécsy et al. 2022 . These sources would appear as continuous waves (CWs) because the GW frequency evolves very slowly compared to the timespan of the data set. So far searches of NANOGrav data sets have found no evidence of CWs ( Aggarwal et al. 2019 ; Arzoumanian et al. 2023 ; Agazie et al. 2023b ) , but we expect that future PTA data sets will detect an individual binary within the next 5-10 years Bécsy et al. 2022 ; Gardiner et al. 2025 . PTAs have primarily performed Bayesian searches for CWs, in which the pulsars’ intrinsic noise is simultaneously fit along with GW signals. These searches are computationally expensive owing to the large number of parameters in the model and can take weeks to run. In contrast, frequentist searches are significant faster to perform, although they use simplified signal models. Therefore, frequentist methods complement Bayesian methods and can be used to identify potential signals which can then be followed up with a full Bayesian search. The \mathcal{F} -statistic is a frequentist method to search for CWs that was initially developed for the Laser Interferometer Gravitational Wave Observatory (LIGO) ( Jaranowski et al. 1998 ) . The same approach was then applied to the Laser Interferometer Space Antenna (LISA) ( Cornish and Porter 2007 ) and PTAs Sesana and Vecchio 2010 ; Babak and Sesana 2012 . The \mathcal{F}_{p} - and \mathcal{F}_{e} -statistics, introduced in Ellis et al. 2012 , are frequentist statistics for detecting continuous waves in PTA data. The \mathcal{F}_{p} -statistic was used to analyze the NANOGrav 5-year and 11-year data sets ( Arzoumanian et al. 2014 ; Aggarwal et al. 2019 ) , but has not been used to analyze subsequent data sets owing to the emergence of a common red noise signal. In the derivation of the \mathcal{F}_{p} - and \mathcal{F}_{e} -statistics, the authors assumed that only white noise and a CW signal were present, and did not consider the presence of either intrinsic red noise or a GWB. Recently, work has been done to use the \mathcal{F}_{e} -statistic to analyze PTA data containing common red noise. The European Pulsar Timing Array (EPTA) used the \mathcal{F}_{e} -statistic to search for CWs in their second data release in addition to a common process EPTA Collaboration et al. 2024 , and Furusawa et al. 2025 looked at how the presence of a GWB effects the \mathcal{F}_{e} -statistic. In this paper, we look at the performance of the \mathcal{F}_{p} -statistic in the presence of pulsar red noise and common red noise. In previous analyses Arzoumanian et al. 2023 ; Aggarwal et al. 2019 ; Arzoumanian et al. 2014 , we used the maximum likelihood posterior parameters to calculate the \mathcal{F}_{p} -statistic (MLFP). However, in the case of the optimal statistic, a frequentist detection statistic for the GWB Chamberlin et al. 2015 , it was found that noise marginalization leads to a more accurate distribution of recovered values Vigeland et al. 2018 . In this paper, we investigate whether marginalizing over the red noise posteriors affects the \mathcal{F}_{p} -statistic. We find that there is not a significant difference between the mean noise-marginalized \mathcal{F}_{p} -statistic (NMFP) values and the maximum-likelihood \mathcal{F}_{p} -statistic (MLFP) values. We use simulations to show that, in the absence of a CW signal, the NMFP values follow the expected chi-squared distribution for frequencies up to where the red noise is modeled, and in the presence of a CW the \mathcal{F}_{p} -statistic can recover the signal at the correct frequency. This paper is organized as follows. In Section II , we discuss our methods and software, including the implementation of the \mathcal{F}_{p} -statistic and its differences from the \mathcal{F}_{e} -statistic as described in Ellis et al. 2012 . We tested the \mathcal{F}_{p} -statistic on simulated PTA data sets to study its performance in the presence of intrinsic pulsar red noise, common red noise, and a continuous wave; we present the results of our analyses on simulated data sets in Section III . In Section IV , we present the noise-marginalized results for the NANOGrav 12.5-year data set ( Alam et al. 2021 ) , and compare these results to the results of Bayesian searches for individual binaries published in Arzoumanian et al. 2023 . We give a brief summary of our work and our conclusions in Section V . II Methods and software In this section, we review the signal model for a CW in PTA data. We assume the source is a circular binary evolving only due to GW emission. We also review the \mathcal{F}_{p} -statistic, a frequentist statistic for detecting CWs. More details can be found in Ellis et

Read the complete text