Improving Statistical Sensitivity of X-ray Searches for Axion-Like Particles

TL;DR

This paper introduces a new analysis method for X-ray observations of AGNs that enhances the detection sensitivity for axion-like particles by isolating their characteristic modulations using residual analysis and machine learning, doubling previous sensitivity.

Contribution

The paper develops a novel analysis technique combining residual analysis in wavelength space and machine learning to improve ALP detection sensitivity in X-ray data.

Findings

Simulations show a twofold increase in sensitivity for microcalorimeter-resolution telescopes.

The method effectively isolates ALP-induced modulations from other spectral features.

Enhanced sensitivity could lead to tighter constraints on ALP properties.

Abstract

X-ray observations of bright AGNs in or behind galaxy clusters offer unique capabilities to constrain axion-like particles (ALPs). Existing analysis technique rely on measurements of the global goodness-of-fit. We develop a new analysis methodology that improves the statistical sensitivity to ALP-photon oscillations by isolating the characteristic quasi-sinusoidal modulations induced by ALPs. This involves analysing residuals in wavelength space allowing the Fourier structure to be made manifest as well as a machine learning approach. For telescopes with microcalorimeter resolution, simulations suggest these methods give an additional factor of two in sensitivity to ALPs compared to previous approaches.