Abstract:

Star clusters play a key role in structure formation because they mark a fundamental transition from structures governed by gravity to those driven by the complex physics of baryons. They emerge from giant molecular clouds assembled through the fragmentation and collapse of gas within galaxies. Massive clusters formed in the early Universe may survive to the present day as globular clusters, which are fossils of early galaxy assembly. Less massive clusters, on the other hand, gradually lose stars via tidal disruption and ultimately dissolve into stellar streams that populate galactic stellar halos. Cluster formation, migration, and dynamical evolution are governed by different physical processes operating on distinct spatial and time scales. These enormous dynamical ranges prevent theorists from building a purely analytical framework describing the complete lifecycle of star clusters.

Numerical modeling provides a powerful tool for disentangling this complexity. However, no single model can resolve all relevant scales, and constructing a unified framework remains an ongoing challenge. This dissertation presents my contributions toward such a framework. I first developed a semi-analytical model applied to cosmological simulations, where clusters are tagged to simulation particles to follow their migration through galaxies. The model reproduces the chemical, spatial, and kinematic properties of cluster populations across a broad range of galaxy masses. To improve its prescription for dynamical disruption, I developed a particle-spray algorithm calibrated against direct N-body simulations to trace the distribution of escaping stars. Based on this algorithm, I developed an automated stellar stream-detection method designed to achieve both high purity and high completeness. Because streams are stellar debris of dynamical mass loss, balancing both quantities is important for obtaining unbiased measurements of cluster mass loss rates.

Looking forward, parsec-resolution hydrodynamic simulations, combined with deep space-based observations, will constrain cluster formation in early galaxies, while larger stream samples from next-generation wide-field surveys will improve models of dynamical evolution. Together, these advances will ultimately connect the birth of clusters in the earliest galaxies to the surviving clusters and stellar streams observed today.