Probing the Parameter Space of Axion-Like Particles Using Simulation-Based Inference

TL;DR

This work tackles constraining axion-like particle (ALP) parameters, specifically the mass $m_a$ and photon coupling $g_{a\gamma}$, from gamma-ray spectra shaped by photon--ALP mixing in astrophysical magnetic fields. It adopts a simulation-based inference framework using Truncated Marginal Neural Ratio Estimation (TMNRE) implemented in Swyft to infer posteriors from simulated CTAO observations of NGC 1275, incorporating extragalactic background light attenuation and the full instrument response. The study demonstrates that TMNRE can recover posteriors that peak near injected values, though the contours remain broad due to limited training density and parameter degeneracies, with calibration tests showing good $g_{a\gamma}$ coverage but mild undercoverage for small $m_a$. This establishes a viable SBI pipeline for ALP searches with CTAO and outlines a roadmap for incorporating more realistic systematics and multi-source analyses ahead of applying to real data.

Abstract

Axion-like particles (ALPs), hypothetical pseudoscalar particles that couple to photons, are among the most actively investigated candidates for new physics beyond the Standard Model. Their interaction with gamma rays in the presence of astrophysical magnetic fields can leave characteristic, energy-dependent modulations in observed spectra. Capturing such subtle features requires precise statistical inference, but standard likelihood-based methods often fall short when faced with complex models, large number of nuisance parameters and limited analytical tractability. In this work, we investigate the application of simulation-based inference (SBI), specifically Truncated Marginal Neural Ratio Estimation (TMNRE), to constrain ALP parameters using simulated observations from the upcoming Cherenkov Telescope Array Observatory (CTAO). We model the gamma-ray emission from the active galactic nucleus NGC 1275, accounting for photon-ALP mixing, extragalactic background light (EBL) absorption, and the full CTAO instrument response. Leveraging the Swyft framework, we infer posteriors for the ALP mass and coupling strength and demonstrate its potential to extract meaningful constraints on ALPs from future real gamma-ray data with CTAO.

Figures (3)

• Figure 1: Simulated counts for NGC 1275, showing spectra with ALP-induced oscillations (red) and without (black), together with error bars from simulated counts.
• Figure 2: Posterior distribution for simulated true values, $m_a = 10$ neV and $g_{a\gamma} = 3 \times 10^{-11}$ GeV$^{-1}$. The contours peak near the true value but exhibit broad uncertainty due to limited simulation density and parameter degeneracies.
• Figure 3: Expected Coverage Probability (ECP) for $m_a$ and $g_{a\gamma}$. While $g_{a\gamma}$ shows good calibration, $m_a$ is slightly underconfident at low credibility.