Investigating model dependencies for obscured Active Galactic Nuclei: a case study of NGC 3982

TL;DR

This study compares different obscuration models for heavily obscured AGN NGC 3982 using X-ray spectra, revealing how model choice affects inferred properties like column density and luminosity, and demonstrating the potential of HEX-P for improved parameter estimation.

Contribution

The paper systematically evaluates the impact of model dependencies on parameter inference in obscured AGN using global exploration of X-ray spectral data.

Findings

Consistent estimates of Compton-thick column density across models.

Significant variation in intrinsic X-ray luminosity inferred by different models.

Distinct parameter space regions occupied by each model's posterior distributions.

Abstract

X-ray spectroscopy of heavily obscured Active Galactic Nuclei (AGN) offers a unique opportunity to study the circum-nuclear environment of accreting supermassive black holes (SMBHs). However, individual models describing the obscurer have unique parameter spaces that give distinct parameter posterior distributions when fit to the same data. To assess the impact of model-specific parameter dependencies, we present a case study of the nearby heavily obscured low-luminosity AGN NGC 3982, which has a variety of column density estimations reported in the literature. We fit the same broadband XMM-Newton + NuSTAR spectra of the source with five unique obscuration models and generate posterior parameter distributions for each. By using global parameter exploration, we traverse the full prior-defined parameter space to accurately reproduce complex posterior shapes and inter-parameter degeneracies. The unique model posteriors for the line-of-sight column density are broadly consistent, predicting Compton-thick $N_{\rm H}$ $>1.5\times10^{24}\rm cm^{-2}$ at the 3$\sigma$ confidence level. The posterior median intrinsic X-ray luminosity in the 2-10 keV band however was found to differ substantially, with values in the range log $L_{ 2-10\,{\rm keV}}$ergs$^{-1}$ = 40.9-42.1 for the individual models. We additionally show that the posterior distributions for each model occupy unique regions of their respective multi-dimensional parameters spaces, and how such differences can propagate into the inferred properties of the central engine. We conclude by showcasing the improvement in parameter inference attainable with the High Energy X-ray Probe (HEX-P) with a uniquely broad simultaneous and high-sensitivity bandpass of 0.2-80 keV.