Use LePHARE library-fit quality to discriminate galaxies from AGN

Determine whether the quality of photometric-redshift fitting with different LePHARE template libraries can be used to discriminate between galaxies and active galactic nuclei in LSST source catalogs.

Background

The paper evaluates three independent LePHARE template libraries—GAL, AGN, and QSO—for photometric-redshift estimation of six AGN samples selected through X-ray, radio, infrared, variability, color, and spectroscopic criteria. Because the optimal library varies with AGN selection and morphology, the authors investigate whether differences in fit quality could themselves identify whether an LSST source is a galaxy or an AGN.

The authors explicitly state that they tried this discrimination in multiple ways without success. They note that future machine-learning methods, trained on larger and more representative samples, might instead assign each LSST source a probability of being a galaxy or an AGN, but that probabilistic classification remains unresolved in the discussed approach.