Dwarf galaxies with AGN and their environments in observations and simulations

This thesis is a study of dwarf galaxies with active galactic nucleis (AGN) characteristics, their environments, and identification of them in both observations and simulations. More specifically, it attempts to answer the questions of what environmental conditions are favourable for AGN activity, if environmental has any influence at all, and to what degree current AGN identification tools are suitable for dwarf galaxies. Using the observational catalogue NASASloan Atlas and the Baldwin-Philips-Terlevich (BPT) and WHAN diagrams as diagnostics, no connection between AGN activity and environment is found based on 62 258 dwarf galaxies, although a weak connection cannot be refuted in a redshift-limited sample of BPT galaxies, while the IllustrisTNG simulation shows an increase in AGN occupation fraction of its 6 771 dwarf galaxies if they have recent mergers. Additionally, dense environments are found to be detrimental for AGN activity, but this finding may be due to numerical reasons. Machinelearning does not rank environmental features highly for identifying AGN, but predicted AGN galaxies reside closer to a massive galaxy and denser neighbourhoods. Preliminary results indicate that the best model relies internal features. Other studies find multi-wavelength data provide the best venue to obtain a complete set of AGN in dwarf galaxies, and simulations are now utilising higher resolution and improved black hole (BH) modelling, enabling accurate evolutionary paths of dwarf galaxies. The seemingly contradictory results between different approaches can in part be explained selection bias (e.g BPT favours unobscured AGN), numerical effects (e.g overmassive BH seeding), and statistical framework used toquantify differences. Future work involves constructing a more complete and accurate sample of dwarf AGN, achieved through using multi-wavelength data, higher sensitivity observations like integrated field unit spectroscopy, and simulations with improved dwarf galaxy and BHmodelling, tying together the many strings by a fine tuned machine learning approach