The following archive corresponds to the data products of the paper "Accretion properties of X-ray AGN: Evidence for radiation-regulated obscuration with redshift-dependent host galaxy contribution" by Laloux et al. Inside the "public_data" folder, there are: the Jupyter notebook "plot_SARD.ipynb" which uses the data of the folder to reproduce all the figures of the papers. python script class_cube.py, containing the class CUBE used to load the data hypercubes the fits table best_fit_all_surveys.fits containing the best-fit values and uncertainties from X-ray spectroscopy and SED fitting (see below). a folder extra_data containing different files corresponding to previous studies results (used for plots) a folder data_cubes containing two folders: "blowout" and "trioNH_monoM95". The former is the STAN output of the SARD with NH=[20-22-23-24], used for the forbidden region study. The latter is the STAN output of the SARD with NH=[20-22-24-26], used for the rest of the analyses. The table best_fit_all_surveys.fits contains the measurements and uncertainties for the 3882 sources of our sample from COSMOS-legacy, AEGIS and CDFS Chandra deep fields. The table have 16 columns: "ID": ID of the source "z_lo", "z", "z_hi": 16th, 50th and 84th percentile of the redshift posterior distribution "Lx_lo", "Lx", "Lx_hi": 16th, 50th and 84th percentile of the logarithmic intrinsic 2-10keV X-ray luminosity posterior distribution "nH_lo", "nH", "nH_hi": 16th, 50th and 84th percentile of the logarithmic column density NH posterior distribution "Mstar_lo", "Mstar", "Mstar_hi": 16th, 50th and 84th percentile of the logarithmic stellar mass posterior distribution "SAR_lo", "SAR", "SAR_hi": 16th, 50th and 84th percentile of the logarithmic specific accretion rate posterior distribution (= (logLx - logMstar) + log10(25 / (1.3 1e38 0.002))

Abstract of Laloux et al. (2024): We adopt a Bayesian X-ray spectral approach to investigate the accretion properties of unobscured ($20<\log(N_{\rm H}/{\rm cm}^{-2}<22$) and obscured ($22< \log(N_{\rm H}/{\rm cm}^{-2}<24$) active galactic nuclei (AGN) to shed light on the orientation vs evolution scenarios for the origin of the obscuring material. For a sample of 3882 X-ray-selected AGN from the {\it Chandra} COSMOS Legacy, AEGIS and CDFS extragalactic surveys, we constrain their stellar masses, $M_\star$, intrinsic X-ray luminosities, $L_{\rm X}$, obscuring column densities, $N_{\rm H}$, and specific accretion rates $\lambda\propto L_{\rm X}/M_\star$.By combining these observables within a Bayesian non-parametric approach, we infer, for the first time, the specific accretion rate distribution (SARD) of obscured and unobscured AGN to $z\approx3$, i.e. the probability of a galaxy with mass $M_\star$ at redshift $z$ hosting an AGN with column density $N_{\rm H}$ and specific accretion rate $\lambda$. Our findings indicate that (1) both obscured and unobscured SARDs share similar shapes, shifting towards higher accretion rates with redshift, (2) unobscured SARDs exhibit a systematic offset towards higher $\lambda$ compared to obscured SARD for all redshift intervals, (3) the obscured AGN fraction declines sharply at $\log\lambda_{\rm break} \sim-2$ for $z <0.5$, but shifts to higher $\lambda$ values with increasing redshift, (4) the incidence of AGN within the theoretically unstable blow-out region of the $\lambda-N_{\rm H}$ plane increases with redshift. These observations provide compelling evidence for AGN "downsizing" and radiation-regulated nuclear-scale obscuration with an increasing host galaxy contribution towards higher redshifts.