A Bayesian method for pulsar template generation

TL;DR

This paper introduces a Bayesian approach to generate pulsar templates, improving the accuracy of pulsar timing by quantifying uncertainties and fluctuations in the templates derived from noisy data.

Contribution

The paper presents a novel Bayesian inference method for pulsar template generation, offering a statistically significant alternative to classical averaging techniques.

Findings

Bayesian templates quantify fluctuations and uncertainties.

The method reconstructs statistically significant templates from 10-50 pulses.

It accurately reconstructs phase information with acceptable uncertainty estimates.

Abstract

Extracting Times of Arrival from pulsar radio signals depends on the knowledge of the pulsars pulse profile and how this template is generated. We examine pulsar template generation with Bayesian methods. We will contrast the classical generation mechanism of averaging intensity profiles with a new approach based on Bayesian inference. We introduce the Bayesian measurement model imposed and derive the algorithm to reconstruct a "statistical template" out of noisy data. The properties of these "statistical templates" are analysed with simulated and real measurement data from PSR B1133+16. We explain how to put this new form of template to use in analysing secondary parameters of interest and give various examples: We implement a nonlinear filter for determining ToAs of pulsars. Applying this method to data from PSR J1713+0747 we derive ToAs self consistently, meaning all epochs were timed and we used the same epochs for template generation. While the average template contains fluctuations and noise as unavoidable artifacts, we find that the "statistical template" derived by Bayesian inference quantifies fluctuations and remaining uncertainty. This is why the algorithm suggested turns out to reconstruct templates of statistical significance from ten to fifty single pulses. A moving data window of fifty pulses, taking out one single pulse at the beginning and adding one at the end of the window unravels the characteristics of the methods to be compared. It shows that the change induced in the classical reconstruction is dominated by random fluctuations for the average template, while statistically significant changes drive the dynamics of the proposed method's reconstruction. The analysis of phase shifts with simulated data reveals that the proposed nonlinear algorithm is able to reconstruct correct phase information along with an acceptable estimation of the remaining uncertainty.