Presentation #100.31 in the session AGN.
The AGN database (AGN-DB) is the repository of all discovered AGN. It contains every publicly available catalog of AGN, along with their redshift, classification, and spectro-photometric properties from the radio to the X-ray wavebands. Objects from different catalogs are securely cross-matched with LYRA, a robust algorithm that includes Bayesian statistics, photometric priors, and machine learning techniques (k-d tree for nearest neighbor lookup, mean shift clustering for source density). AGN-DB already contains more than 7 million AGN from 237 catalogs. Information currently available in the database includes multiwavelength photometry (up to 48 bands), redshift, AGN type, X-ray obscuring column density, and host galaxy morphology.
As soon as a new AGN catalog is published, our pipeline ingests it: all the tables are formatted to CSV files, and all the properties are converted to the standard cgs system. The ultimate goal of AGN-DB is to sample AGN across the Universe and make them publicly available to the scientific community in a single, uniform, and complete database. Data are accessible through an SQL query system, jupyter notebooks, an intuitive user interface, and an imaging retrieval system interfaced with hips2fits (CDS Strasbourg). Moreover, CIGALE software (Yang et al. 2022) is implemented in our user interface, allowing the user to perform SED fitting directly on the AGN-DB website.
I will present the current content of AGN-DB and the related projects: X-ray spectral analysis (NH focused) and SED fitting over the entire catalog, multi-band and bolometric luminosity functions, AGN (mis)classification tools through machine learning analysis (random forest and stochastic neighbor embedding), and morphological classification of the AGN host-galaxies with the GaMPEN neural network (Ghosh et al. 2022).