Blind source separation for decomposing X-ray pulsar profiles

Abstract

Accretion-powered X-ray pulsars offer a unique opportunity to study physics under extreme conditions. To fully exploit this potential, the interrelated problems of modelling radiative transport and the dynamical structure of the accretion flow must, however, be solved. This task is challenging both from a theoretical and observational point of view and is further complicated by a lack of direct correspondence between the properties of emission emerging from the neutron star and observed far away from it. In general, a mixture of emission from both poles of the neutron star viewed from different angles is indeed observed at some or even all phases of the pulse cycle. It is essential, therefore, to reconstruct the contributions of each pole to the observed flux in order to test and refine models describing the formation of the spectra and pulse profiles of X-ray pulsars. In this paper we propose a novel data-driven approach to address this problem using the pulse-to-pulse variability in the observed flux, and demonstrate its application to RXTE observations of the bright persistent X-ray pulsar Cen X-3. We then discuss the comparison of our results with previous work attempting to solve the same problem and how they can be qualitatively interpreted in the framework of a toy model describing emission from the poles of a neutron star.