Reconstructing AGN X-ray spectral parameter distributions with Bayesian methods

We present a new Bayesian method for reconstructing the parent distributions of X-ray spectral parameters active galactic nuclei (AGN) in large surveys. The uses probability distribution function (PDF) posteriors obtained by fitting consistent physical model to each object with method. PDFs are often broadly distributed and may systematic biases, such that naive point estimators or even some standard parametric modeling not sufficient reconstruct population without obvious bias. Our transfer computed from realistic simulation same selection as actual sample redistribute stacked PDF then forward-fit nonparametric it way, so biases properly taken into account. In this we able accurately distributions. apply our inference methods XMM-COSMOS survey pilot study. For 819 AGN detected COSMOS field, 663 (8%) which have spectroscopic redshifts (spec-z) others high-quality photometric (photo-z), find prominent bi-modality widely separated peaks absorbing hydrogen column density (N_H) an indication absorbed harder photon indices. A clear decreasing trend fraction versus intrinsic 2-10keV luminosity is observed, but there no evolution redshift. designed be readily applicable samples XXL survey, eventually eROSITA.