Bayesian pulsar timing and noise analysis with Vela.jl : an overview

Abhimanyu Susobhanan · 2024

Read Bayesian pulsar timing and noise analysis with Vela.jl : an overview by Abhimanyu Susobhanan (2024) 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-2412-15858.txt.

Introduction

Abhimanyu Susobhanan presents, in a 2024 paper, a software package for timing pulsars and sorting signal from noise. Pulsars are spinning neutron stars whose pulses arrive with clock-like regularity. Tiny delays in those arrivals can betray passing gravitational waves and other physics.

His package, Vela, is written in Julia for speed and built in modules for clarity. A Python companion handles file reading, clock corrections, and solar system tables using an established timing tool. A strict test suite guards the package's reliability. The paper explains the design and walks through its use. Fast, trustworthy timing tools are the workhorses behind pulsar astronomy's biggest claims.

Catalog record

CREATORAbhimanyu Susobhanan
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-15858

Browse the full library · About the Tower of Babel

Full text

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

Bayesian pulsar timing and noise analysis with Vela.jl : an overview Abhimanyu Susobhanan Abstract We present Vela.jl , an efficient, modular, easy-to-use Bayesian pulsar timing and noise analysis package written in Julia. Vela.jl provides an independent, efficient, and parallelized implementation of the full non-linear pulsar timing and noise model along with a Python binding named pyvela . One-time operations such as data file input, clock corrections, and solar system ephemeris computations are performed by pyvela with the help of the PINT pulsar timing package. Its reliability is ensured via careful design utilizing Julia’s type system, strict version control, and an exhaustive test suite. This paper describes the design and usage of Vela.jl focusing on the narrowband paradigm. PINT [ 56 , 73 ] , emcee [ 25 ] , nestle [ 11 ] , numpy [ 34 ] , astropy [ 66 ] , matplotlib [ 40 ] , corner [ 24 ] , DoubleFloats.jl [ 68 ] , Distributions.jl [ 12 ] , JuliaCall/PythonCall.jl [ 67 ] , git https://git-scm.com/ I Introduction Pulsars are rotating neutron stars whose electromagnetic radiation is received as periodic pulses. Their high rotational stability makes them excellent celestial clocks, and pulsar timing, the technique of tracking a pulsar’s rotation by measuring the times of arrival (TOAs) of its pulses, is one of the most precise techniques in astronomy [ 55 ] . Over the years, pulsar timing has been applied to study a wide array of astrophysical phenomena, such as neutron star equation of state and internal dynamics [ 15 , 6 , e.g.,] , tests of theories of gravity [ 84 , e.g.,] , solar wind [ 77 , e.g.,] , galactic dynamics [ 62 , e.g.,] , etc. Recently, pulsar timing array [ 26 , PTA:] experiments found evidence [ 3 , e.g.,] for a stochastic gravitational wave background [ 36 , GWB:] in the nanohertz frequency range with the help of pulsar timing. In addition to the pulsar rotation, the measured TOAs are also influenced by several deterministic astrophysical processes such as the orbital motion of the Earth, proper motion of the pulsar, solar wind, interstellar dispersion, the orbital motion of the pulsar, etc [ 20 ] , as well as stochastic processes such as radiometer noise, pulse jitter, rotational irregularities, interstellar medium variability, etc [ 1 ] . High-precision pulsar timing requires accurate modeling of all of these effects. A pulsar timing model or pulsar ephemeris provides a mathematical description of the deterministic processes that affect the measured TOAs and is often accompanied by a noise model that describes the stochastic processes therein. In practice, pulsar timing involves estimating the parameters of a pulsar timing model given a set of TOA measurements, usually in a frequentist setting using a software package such as tempo2 [ 39 , 20 ] or PINT [ 56 , 73 ] . Noise characterization can be done in a few different ways. In PINT , noise parameters can be estimated together with the timing parameters in a frequentist way [ 73 ] . tempo2 provides plugins like fixData and spectralModel that can estimate some noise parameters by fixing the timing model parameters [ 38 ] . Bayesian noise characterization can be performed using the ENTERPRISE [ 42 ] and TEMPONEST [ 54 ] packages starting from a post-fit timing model, and is considered standard practice for high-precision timing experiments like PTAs. ENTERPRISE and TEMPONEST differ in how they treat the timing model fit; the former analytically marginalizes a linearized approximation of the timing model [ 81 ] , whereas the latter enables inference over the full non-linear timing and noise model. The PINT package also supports Bayesian parameter estimation through pint.fitter.MCMCFitter and pint.bayesian interfaces although their use on large datasets is hampered by slow performance [ 56 , 73 ] . Other packages that have been used for Bayesian pulsar timing and/or noise analysis include libstempo [ 83 ] , piccard [ 80 ] , PAL2 [ 21 ] , and FortyTwo [ 14 , e.g.] . In this paper, we present Vela.jl 1 1 1 Named after the Vela pulsar [ 52 , J0835–4510:] , the brightest radio pulsar. Also, Vēḻa is a word meaning occasion, time, etc in Malayalam with cognates in several other Indian languages. 2 2 2 The source code is available at https://github.com/abhisrkckl/Vela.jl . The documentation is available at https://abhisrkckl.github.io/Vela.jl/ . This paper corresponds to version 0.0.7 [ 71 ] . , a new framework written in Julia [ 13 ] for performing Bayesian inference over the full non-linear timing and noise model. Vela.jl provides an efficient and modular implementation of the pulsar timing and noise model independent of other pulsar timing packages 3 3 3 Note that Vela.jl relies on PINT to read input files, perform clock corrections, and compute solar system ephemerides. Aside from this, the timing & noise model is implemented independently. See subsection III.5 for more details. , and supports parallelized evaluation of the pulsar timing likelihood function using multi-threading. We also provide a Python interface named pyvela since Python is more popular among astronomers than Julia. This package is developed with a focus on reliability, performance, and modularity (in that order), employing strict version control (using git ) and rigorous testing. The main objective of Vela.jl is to provide a flexible, robust, modular, well-documented tool for performing Bayesian inference over the full non-linear pulsar timing and noise model. While this functionality is already available in the TEMPONEST package, Vela.jl differs from TEMPONEST in the following ways. (1) TEMPONEST can only be used with the nested sampling packages MULTINEST [ 23 ] and PolyChord [ 32 ] , whereas Vela.jl can be used with any Markov Chain Monte Carlo [ 19 , MCMC:] or nested sampler [ 9 ] available in Julia or Python. (2) TEMPONEST uses tempo2 internally to evaluate the timing model, whereas Vela.jl contains an independent implementation of the timing model. (3) TEMPONEST only supports the narrowband timing paradigm, whereas Vela.jl supports both the narrowband and wideband timing paradigms (the application of Vela.jl on wideband data will be discussed in a separate paper). (4) TEMPONEST is a stand-alone application whereas Vela.jl is a library that can be used programmatically alongside various Julia and Python packages. This paper is arranged as follows. In Section II , we provide a brief overview of the pulsar timing and noise analysis. In Section III , we describe the design and implementation of Vela.jl and pyvela . In Section IV , we provide three examples that showcase the functionalities of Vela.jl . Finally, in V , we summarize our work and discuss the future directions. II A brief overview of Bayesian pulsar timing and noise analysis The fundamental measurable quantity in conventional pulsar timing is the TOA. TOAs are measured by folding the time series pulsar data over a known pulsar period and cross-correlating the resulting integrated pulse profile against a noise-free template [ 75 ] , usually after splitting the observation into multiple frequency subbands (this is known as narrowband timing). The more recent wideband timing technique involves measuring a single TOA and a dispersion measure (DM) from an observation by matching the frequency-resolved integrated pulse profile

Read the complete text