Gaussian process representation of dispersion measure noise in pulsar wideband datasets 2026 Gaussian process representation of dispersion measure noise in pulsar wideband datasets – C

Abhimanyu Susobhanan, Rutger van Haasteren · 2025

Read Gaussian process representation of dispersion measure noise in pulsar wideband datasets 2026 Gaussian process representation of dispersion measure noise in pulsar wideband datasets – C by Abhimanyu Susobhanan, Rutger van Haasteren (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-2505-05274.txt.

Introduction

Abhimanyu Susobhanan and Rutger van Haasteren clean up pulsar timing noise in a 2025 paper. Pulsars are precise cosmic clocks, but the ionized gas between us and them delays their radio pulses by a jittering amount. That dispersion noise must be modeled well, or it contaminates the search for gravitational waves in pulsar timing arrays.

Modern wideband receivers measure the delay and the arrival time together from each observation, handling several distortions at once and shrinking data volumes. The authors derive a way to fold any model of the delay variation, including flexible Gaussian process models, into wideband timing and noise analysis. Their method generalizes the existing treatment of dispersion noise for these data sets. The audience is the pulsar timing community, where small modeling gains translate into sensitivity to the faintest signals in astrophysics: the background hum of merging giant black holes.

Catalog record

CREATORAbhimanyu Susobhanan, Rutger van Haasteren
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE38 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-05274

Browse the full library · About the Tower of Babel

Full text

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

Gaussian process representation of dispersion measure noise in pulsar wideband datasets 2026 Gaussian process representation of dispersion measure noise in pulsar wideband datasets – C Abhimanyu Susobhanan, Rutger van Haasteren Abstract The ionized interstellar medium disperses pulsar radio signals, resulting in a stochastic time-variable delay known as the dispersion measure (DM) noise. In the wideband paradigm of pulsar timing, we measure a DM together with a time of arrival from a pulsar observation to handle frequency-dependent profile evolution, interstellar scintillation, and radio frequency interference more robustly, and to reduce data volumes. In this paper, we derive a method to incorporate arbitrary models of DM variation, including Gaussian process models, in pulsar timing and noise analysis and pulsar timing array analysis. This generalizes the existing method for handling DM noise in wideband datasets. 1 Introduction Pulsars, rotating neutron stars emitting electromagnetic radiation that appears as periodic pulses to terrestrial observers, are some of the most rotationally stable objects in the universe ( Lorimer & Kramer 2012 ) . Radio waves emitted by pulsars are dispersed as they travel through the ionized interstellar medium (ISM), inducing delays that are proportional to the electron column densities along their lines of sight (also known as the dispersion measure, DM) and inversely proportional to the square of the observing frequency ( Backer & Hellings 1986 ) . The DM varies stochastically as a function of time due to the dynamic nature of the ISM and the fact that the pulsar and the Earth are moving relative to each other ( Donner et al. 2020 , e.g.) . Other astrophysical effects that influence the times of arrival (TOAs) of pulsar pulses include the motion of the Earth through the solar system ( Edwards et al. 2006 ) , the motion of the pulsar in a binary system ( Damour & Deruelle 1986 ) , the proper motion of the pulsar, and gravitational waves passing across the line of sight ( Estabrook & Wahlquist 1975 ) . Pulsar timing is the technique of tracking a pulsar’s rotation by measuring the TOAs of its pulses, allowing the pulsar to be used as a celestial clock ( Hobbs et al. 2006 ) . Pulsar timing has been used to probe a wide range of phenomena, from neutron star equation of state ( Cromartie et al. 2020 , e.g.) to nanohertz gravitational waves ( Agazie et al. 2024 , e.g.) . Radio pulsar timing is conventionally performed by folding the frequency-resolved pulsar light curve using the known rotational period to obtain a frequency-resolved integrated pulse profile. The integrated pulse profile is then split into multiple frequency sub-bands and a TOA is measured independently from each sub-band by cross-correlating the profile against a noise-free template ( Taylor 1992 ) . While this paradigm, known as narrowband timing, is widely used, it has two major drawbacks: it produces a large number of TOAs leading to large data volumes, and it does not adequately account for frequency-dependent profile evolution ( Hankins & Rickett 1986 ) in many cases. The more recently developed wideband timing technique handles these issues by treating the frequency-resolved integrated pulse profile as a single two-dimensional entity and simultaneously measures a single TOA and a single dispersion measure (DM) by cross-correlating it against a two-dimensional template ( Liu et al. 2014 ; Pennucci et al. 2014 ) . The two-dimensional template, known as a portrait, is usually derived from data using a principal component analysis-based algorithm ( Pennucci 2019 ) . An extension of the wideband method for combining simultaneous multi-frequency observations during TOA and DM estimation was presented in Paladi et al. 2023 . Remarkable applications of this method include Alam et al. 2021 , Tarafdar et al. 2022 , Curyło et al. 2023 , Agazie et al. 2023 , and Tan et al. 2024 . Since wideband timing is a relatively new method, data analysis techniques and software tools for handling wideband datasets are still being developed. Wideband TOA and DM measurements are performed using PulsePortraiture ( Pennucci et al. 2014 ; Pennucci 2019 ) . Interactive timing of wideband datasets can be performed using tempo ( Nice et al. 2015 ) and PINT ( Luo et al. 2021 ; Susobhanan et al. 2024 ) . Bayesian noise analysis of wideband datasets assuming a linearized timing model (commonly known as single-pulsar noise analysis, SPNA) is available in a limited capacity via the ENTERPRISE package ( Ellis et al. 2020 ; Johnson et al. 2024 ) , and Bayesian timing & noise analysis using the full non-linear timing model (commonly known as single-pulsar noise & timing analysis, SPNTA) can be performed using Vela.jl ( Susobhanan 2025b ; Susobhanan 2025a ) . ENTERPRISE also enables pulsar timing array ( Foster & Backer 1990 , PTA:) data analysis using wideband datasets, which combines multiple pulsar datasets to search for signals correlated across pulsars. The wideband analyses available in the literature have generally employed a piecewise-constant model for DM variations, known as the DMX model ( Arzoumanian et al. 2015 ; Alam et al. 2021 ) , and this is the only DM variability model currently available in ENTERPRISE for wideband datasets. However, recent studies have shown that Gaussian process (GP) models ( van Haasteren & Vallisneri 2014a ) for interstellar DM (commonly known as DMGP) and solar wind variability may be more suitable in many cases ( Larsen et al. 2024 ; Iraci et al. 2024 ; Susarla et al. 2024 ) . While Vela.jl provides the DMGP model in the context of SPNTA, it is also advantageous to incorporate it in SPNA since the latter is usually computationally less expensive. SPNA can also be generalized to include correlations between different pulsars, such as those produced by the stochastic gravitational wave background ( Hellings & Downs 1983 , GWB:) , in a computationally tractable way ( van Haasteren et al. 2009 ) . In this paper, we develop a formalism to incorporate the DMGP model in SPNA and PTA analysis from first principles. This paper is arranged as follows. We provide a brief overview of the wideband timing & noise model, used for SPNTA, in Section 2 . We describe the linearization of the wideband timing model and the analytic marginalization of its posterior distribution, used for SPNA, in Section 2 . In Section 4 , we demonstrate our results using a simulated dataset. Finally, we summarize our results in Section 5 . 2 The wideband timing & noise model In the wideband timing paradigm, each observation yields a TOA-DM pair \left(t_{i},d_{i}\right) with uncertainties \left(\sigma_{i},\epsilon_{i}\right) measured simultaneously at a fiducial observing frequency \nu_{i} within the observing band ( Pennucci et al. 2014 ; Liu et al. 2014 ) . The measurement covariance \left\langle t_{i}d_{i}\right\rangle is usually made to vanish by adjusting the fiducial observing frequency \nu_{i} . The time of emission \tau_{i} can be estimated from the TOA up to an additive constant using the expression ( Hobbs et al. 2006 ; Edwards et al. 2006 ) \displaystyle\tau_{i} \displaystyle=t_{i}-\Delta_{\text{clock}}(t_{i})-\Delta_{\text{jump};i}-\Delta_{\odot}(t_{i})-\Delta_{\text{DM}}(t_{i},\nu_{i})-\Delta_{\text{GW}}(t_{i}) \displaystyle\quad-\Delta_{\text{B}}(t_{i})-...-\mathcal{N}_{\text{R};i}\,, (1) where \Delta_{\text{clock}} is a clock correction that converts t_{i} from the observatory timescale to a timescale defined at the solar system barycenter, \Delta_{\text{jump}} contains instrumental delays introduced by the observing

Read the complete text