A Gaussian Mixture Model for Nulling Pulsars

TL;DR

This paper introduces a Gaussian mixture model-based algorithm that improves the measurement of pulsar nulling behavior, providing higher precision and unbiased results across various pulsars, including those without obvious nulls.

Contribution

The paper presents a novel, statistically robust algorithm for measuring pulsar nulling, outperforming existing methods in accuracy and applicability.

Findings

Better nulling measurement precision

No bias in nulling estimates

Applicable to pulsars without obvious nulls

Abstract

The phenomenon of pulsar nulling -- where pulsars occasionally turn off for one or more pulses -- provides insight into pulsar-emission mechanisms and the processes by which pulsars turn off when they cross the "death line." However, while ever more pulsars are found that exhibit nulling behavior, the statistical techniques used to measure nulling are biased, with limited utility and precision. In this paper we introduce an improved algorithm, based on Gaussian mixture models, for measuring pulsar nulling behavior. We demonstrate this algorithm on a number of pulsars observed as part of a larger sample of nulling pulsars, and show that it performs considerably better than existing techniques, yielding better precision and no bias. We further validate our algorithm on simulated data. Our algorithm is widely applicable to a large number of pulsars even if they do not show obvious nulls. Moreover, it can be used to derive nulling probabilities of nulling for individual pulses, which can be used for in-depth studies.