Dissertation Defense
Student: Pierre Augusto Ré Martho
Program: Astronomy
Title: "Search for compact dwarf galaxies in the Fornax cluster"
Orientador: Prof. Dr. Laerte Sodré Junior - IAG/USP
Judging Committee:
- Prof. Dr. Laerte Sodré Junior - Orientador e Presidente - IAG/USP
- Profa. Dra. Claudia Lucia Mendes de Oliveira - IAG/USP
- Profa. Dra. Analía Viviana Smith Castelli - FCAG-UNLP, Argentina
- Profa. Dra. Arianna Cortesi - OV-UFRJ (por videoconferência)
Abstract: This work presents a systematic search for compact dwarf galaxies in the Fornax cluster, combining twelve-band photometry from the S-PLUS survey with machine learning-based classification. Distinguishing UCDs from other compact objects, such as stars or globular clusters, is particularly challenging. To address this difficulty, we employ a supervised method that classifies objects into two categories based solely on photometry: compact and extended. This approach allows for the identification of compact objects with photometric properties similar to galaxies, facilitating the selection of UCD candidates. The analysis was carried out over a region covering five times the virial radius of the cluster. Several classifiers were tested, with a Random Forest algorithm optimized via cross-validation achieving an area under the ROC curve (AUC-ROC) of 0.97 and a Matthews correlation coefficient of 0.84. Based on the probabilities returned by the classifier, we applied morphological cuts and photometric redshift constraints, resulting in the identification of 14 new compact dwarf galaxy candidates in regions extending beyond the estimated virial radius of ∼700 kpc for Fornax. If confirmed spectroscopically, this would suggest the presence of such systems in the cluster outskirts, in environments different from those where UCDs are typically found. Additionally, we explored the use of the narrow-band J0660 filter to detect emission features that could indicate star-forming compact galaxies, such as blue compact dwarfs (BCDs). Although initial observations with GMOS-S on the Gemini telescope revealed redshifts of z∼0.3, consistent with background sources, this experiment validates the inclusion of narrow-band color cuts in future surveys targeting compact emitters. Our approach demonstrates that automated methodologies can significantly enhance the census of compact stellar systems in dense environments.
Keywords: Ultra-compact dwarf galaxies, blue compact galaxies, Fornax cluster, machine learning, S-PLUS