Supermassive black holes require a reservoir of cold gas at the centre of their host galaxy in order to accrete and shine as active galactic nuclei (AGN). Major mergers have the ability to drive gas inwards through the strong tidal torques involved, but observations trying to link mergers with AGN have found mixed results due to the difficulty of consistently identifying galaxy mergers in large surveys, with much past work in this area relying on visual classification of small samples. Deep learning has made many advances in recent years with identifying galaxy morphology and merger state from imaging, but this approach has yet to reach its full potential in the study of AGN merger triggering.

We present here our work analysing morphology and merger status of AGN host galaxies. First, we use a supervised learning approach to identify likely post-mergers Type 2 Seyferts, training a convolutional neural network on survey-realistic imaging of IllustrisTNG galaxies whose merger status is known. We find no significant difference in merger fraction between the Seyferts and mass- and redshift-matched inactive control galaxies. However, when separated by mass and star formation rate, we find that the AGN hosts in the star-forming blue cloud exhibit significant merger enhancements over controls, while those in the quiescent red sequence have marginally lower merger fractions relative to controls. Hence we show that the impact of galaxy mergers on AGN activity depends on star formation rate and implicitly on cold gas fraction within the galaxy.

Second, we discuss a more generalised approach to identifying AGN host galaxy morphology, using a contrastive learning technique to study galaxies hosting either Type 1 or Type 2 AGN. This tool gives accurate morphological classifications of galaxies even with a central point source outshining other features. Using this method, we are able to compare morphological properties of galaxies hosting obscured and unobscured AGN to each other and to controls. We will show the first results of our morphological analysis of AGN hosts regardless of obscuration.