Active galactic nuclei (AGNs) are powered by accretion onto supermassive black holes (SMBHs), lying at the centres of galaxies, and are one of the most efficient mechanisms of energy release in the Universe. Through their feedback processes, AGNs play an important role in the co-evolution between SMBHs and their host galaxies.
A significant fraction of the radiative output of AGNs is emitted in the X-rays by a hot corona close to the SMBH. This radiation interacts with the surrounding gas and dust, producing distinct absorption and reflection features in observed X-ray spectra that carry information about the structure of the circumnuclear material. According to the unified model of AGNs, many of the observational differences between AGN classes arise from the orientation relative to a parsec-scale dusty torus surrounding the central engine. A substantial fraction of AGNs are heavily obscured by circumnuclear material with hydrogen column densities along the line of sight of NH≳ 1024 cm−2, the so-called Compton-thick (CTK) population. Constraining the demographics of obscured AGNs is therefore essential for tracing the growth of SMBHs across cosmic time.
CTK sources are notoriously challenging to detect, yet they contribute to the cosmic X-ray background (CXB), which is dominated by the integrated X-ray emission of all AGNs. Population synthesis of AGNs has long been used to uncover AGN demographics, with the CXB spectrum serving as the primary observational constraint. Previous studies largely rely on phenomenological prescriptions, typically treating absorption and reflection as independent parameters and the intrinsic fraction of CTK AGNs as a free parameter rather than emerging from the underlying physical conditions. As a result, these models are presented with a high degeneracy between the assumed reflection strength and CTK fraction, while providing limited insight into the structure and physics of the circumnuclear environment. Furthermore, population synthesis studies are constructed from the X-ray luminosity function (XLF), which although useful, does not directly connect AGN demographics to the underlying properties, SMBH mass and Eddington ratio (mass-normalised accretion rate).
The aim of this thesis is to overcome these limitations by developing a self-consistent framework for AGN population synthesis, in which AGN demographics arise from the structure and physics of the circum- nuclear material. In this work, I combine state-of-the-art numerical simulations from the X-ray ray-tracing code RefleX, forward modelling of the population, and simulation-based inference (SBI) to constrain ge- ometrical and physical properties of the dusty torus using X-ray observations. With this approach, the emission, absorption, and reflection of AGN X-ray spectra emerge self-consistently from the simulated dis- tribution of material. SBI is employed for advanced parameter inference, as the forward modelling involves complex stochastic simulations. By reproducing observational constraints covering different regimes of flux, obscuration, and redshift, such as the CXB spectrum and absorption properties of detected AGNs, this framework can test various AGN unification scenarios.
