A Multi-Method Approach to membership selection in Ultra-Faint Dwarf Galaxies: The role of RR Lyrae Stars

The formation of the Milky Way (MW) and its stellar halo remains a fundamental open problem in astrophysics. In recent decades, deep wide-field surveys have revealed numerous ultra-faint dwarf galaxies (UFDs) around the MW - stellar systems much smaller and fainter than the classical dwarfs such as Draco, Fornax, and Leo I. UFDs typically host old, metal-poor stellar populations, similar to those in the Galactic halo, suggesting a potential evolutionary link. RR Lyrae stars, as pulsating variables tracing ancient, metal-poor populations, are powerful tools for investigating the structure and content of UFDs. The ESA’s Gaia mission has revolutionised this field by delivering all-sky, high-precision astrometric and photometric data, enabling the detection and characterisation of RR Lyrae stars even in the sparse environments of UFDs. This thesis analyses six UFD galaxies using Gaia Data Release 3 (DR3), with the aim of identifying likely RR Lyrae members. Two complementary approaches are employed: a classical method based on proper motion selection, colour-magnitude diagrams, and the Period-Wesenheit-Metallicity (PWZ) relation; and a statistical method using machine learning clustering algorithms to associate stars with their host systems based on astrometric and photometric properties. This dual strategy enhances membership reliability and demonstrates the value of machine learning in astronomical classification tasks. The thesis is organised as follows: Chapter 1 introduces the scientific context; Chapter 2 details the methodology; Chapter 3 presents results for Boötes I, Boötes III, Carina II, Coma Berenices, Sagittarius II, and Ursa Major I; Chapter 4 concludes with future prospects.