A robust automated machine-learning method for the identification of star clusters in the central region of the Small Magellanic Cloud

¹ Department of Physics, National and Kapodistrian University of Athens, Panepistimiopolis, Zografos 15784, Greece
e-mail: a_strantzalis@yahoo.gr
² University of Crete, Physics Department & Institute of Theoretical & Computational Physics, 71003 Heraklion, Crete, Greece
³ Foundation for Research and Technology-Hellas, 71110 Heraklion, Crete, Greece
⁴ Department of Physics, Box 41051, Science Building, Texas Tech University, Lubbock, TX 79409-1051, USA
⁵ Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA

Received: 18 January 2023
Accepted: 30 August 2023

Abstract

Aims. We developed a cluster-detection method based on the code DBSCAN to identify star clusters in the central region of the Small Magellanic Cloud (SMC).

Methods. Two approaches were used to determine the values of the free parameters of DBSCAN. They agree well with each other and can be used in the fields that are studied without any a priori knowledge of clustering, characteristic scales, or background density. We validated the success of the DBSCAN cluster-detection method on recent cluster catalogues after introducing a cluster-classification scheme based on three diagnostics that relie on colour-magnitude diagrams and growth curves. We used data from the Magellan Telescope at the Las Campanas Observatory in Chile and from Gaia Data Release 3.

Results. As a byproduct of the validation process, we revisited objects that were classified as clusters in recent compilations. We found that 40% fail all diagnostics and most probably are not clusters. DBSCAN was very successful in recovering actual clusters with high precision and recall.