Presentation #205.01 in the session Distinguished Career Talk: Meg Urry.
My early training in X-ray astronomy, being inherently Bayesian, shaped the way I interpret observations of the Universe. For example, my PhD thesis explained that blazars were dominated by relativistic jets pointing at us and described how beaming strongly affects the observed numbers of blazars and parent radio galaxies with luminosity. More recently, I have used X-rays from Active Galactic Nuclei to study the growth of supermassive black holes over the past 12 billion years. X-ray samples are far more complete than UV-optical surveys and constrain population synthesis models far better than the integral X-ray background. The Accretion History of AGN (AHA) survey is a multiwavelength survey that combines IR, optical, and X-ray samples over a wide range of survey volumes, allowing independent constraints on luminosity and redshift. We have generated the only black hole growth model that explains all published X-ray luminosity functions, number counts, redshift distributions, and Compton-thick fractions. The AHA clustering analyses show that AGN reside in all environments and that selection effects dominate previous clustering measurements. Using convolutional neural networks, we analyze host galaxy morphologies quantitatively so they can be compared to large-scale environment and AGN activity. We do the same for AGN host galaxies, first using a generative adversarial network to remove the AGN point source. Finally, spectral energy distributions for X-ray selected samples show a much broader distribution of shapes than optical-UV-selected samples, confirming the latter are a small subset of the full population.