Stability and convergence of N-body simulations for galaxy formation

Galaxy formation is still a current topic in astronomy. An important tool to understanding it is through simulation, which allows galaxies to be studied from all angles and across time. It allows us to explore the gap between observation and theory, but only if the results are sufficiently accurate. In this thesis I look at the majority of the simulation pipeline from running through the various stages of analysis, and some of the limits of their accuracy, and the fidelity of the subsequent analysis tools. It starts by looking at running simulations from initial conditions, and what influence changing parameters and simulation engines has on the outcome. Then I look in detail at how successful subhalo detection is by comparing a number of substructure finders, and examining their strengths and weaknesses. Following this I focus on a single parameter recovered for such haloes, the spin, and how well it was recovered, and what it tells us about the spin of substructures. Following this I investigated the building of merger trees, by writing my own merger tree program, and comparing it with some of the established ones. Then I look at using these processes as input to semi-analytic models, and how mass changes could affect the outcome. Finally I used a number of these tools to investigate the fate of some of the larger haloes formed at early times in an attempt to show where ultra-compact dwarf galaxies are formed and their fate