Speaker
Description
Galaxy groups and clusters represent systems of galaxies residing within the same dark matter halos. This inherent connection to dark matter halos positions galaxy groups as a direct avenue for addressing pivotal questions in astrophysics, such as the role of the environment in galaxy formation and evolution, as well as the cosmic density field. Various group-finding methodologies have been proposed and implemented across different redshift surveys, including the Friends-of-Friends (FoF) algorithm, halo-based group finders, and the C4 algorithm.
In this study, we introduce a novel group-finding approach that leverages machine learning techniques. Our group finder comprises two distinct machine learning models: a classification model designed to identify central galaxies from their neighboring counterparts and a regression model for estimating the mass of groups formed by galaxies sharing the same central galaxy. These two models are integrated into an iterative process, ultimately producing the final galaxy catalog.
To validate our method, we conducted extensive training and testing using the Millennium Simulation. Our results demonstrate remarkable accuracy, even under extensibility testing scenarios such as raising the apparent magnitude limitation of mock catalogs, utilizing high-redshift galaxy samples, and applying our method to a different simulation (TNG300). Furthermore, we applied our approach to real observational data from the NASA-Sloan Atlas Catalog, based on the Sloan Digital Sky Survey (SDSS). The group catalog will be made available in the near future.
