Date of Award

Fall 1-1-2025

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Astronomy

First Advisor

Geha, Marla

Abstract

The ΛCDM model for cosmology accurately predicts the large scale structure of our universe. However, at smaller scales (≤ 100 Mpc) observations are at tension with what ΛCDM predicts. Dwarf galaxies which exhibit scale radii on the order of pc to kpc and stellar streams which span sizes of up to several 100 kpc have the potential to shed light into these discrepancies between observations and theory. One particular misconception comes from the distribution of dark matter in the inner regions of dwarf galaxies. ΛCDM dark matter only simulations predict the steeply rising density profiles, while observations of several systems are more consistent with density profiles that are flat near the center. On the other hand, as dwarf galaxies are accreted by larger ones, in accordance with ΛCDM, tidal forces cause stars to be stripped from these systems forming long tidal tails. Although this process takes them out of equilibrium, and therefore rules out equilibrium modeling, other properties such as the timescale in which the formation of tidal tails, as well as the timescale at which these streams become phase-mixed can also place constraints on dark matter models. Fortunately, the proximity of these systems to us makes detailed observations at the individual star scale possible. There is a long history of tools developed in order to study these systems, but in recent years there has been both a significant leap in the computational tools and observations that will allow us to place much stronger constraints on the structural properties of these systems and hence the dark matter models that are consistent with these observations. This dissertation explores the observations and inference tools necessary to investigate these systems. First, we consider current and future observations by studying the information content that can be extracted from our observations given the typical dynamical modeling used in the field, Jeans Modeling. For various years, our observations were limited to measuring one component of the velocity vector of stars, the line-of-sight velocities. These observations have known limitations, particularly the known degeneracy that arises between the velocity anisotropy and the inner distribution of dark matter. Observations from the Gaia telescope have recently made proper motions, the additional two components of the velocity vector. From an information point of view, we show the limitation of only having line-of-sight velocities. We show that even increasing the number of stars and the precision at which we observe these velocities, additional information is needed in order to break degeneracies and better constrain the properties of these systems that we are interested in. Additionally, we show that beyond a certain precision, when the uncertainty on individual velocities is roughly half the velocity dispersion of the system, increasing the precision yields diminishing returns. We then move on to the tools that will be necessary in order to handle these observations. Since the relation between kinematics and underlying structural properties of the dark matter halo we are interested are not straight forward, analysis requires processing (in the form of various integrals) in order to infer the properties we’re interested in. The increased number of stars we have for each system as well as the addition of proper motions to our observational data sets will require us to develop more efficient and flexible methods in order to mitigate the computational cost that comes with our inference methods. Fortunately, access to computational clusters and graphical processing units as well as software packages that allow us to take advantage of this hardware without having to learn specialized languages as in the past has become more common. We develop tools and a framework in order to compute the various dynamical quantities that are required to model these systems. We present a package written completely in python that can readily be used on any computer, as well as in clusters and with GPUs. Finally, we tackle stellar streams using high resolution cosmological simulations. Using the ‘Mint’ DC Justice League simulations, a suite of simulations of Milky Way like galaxies we develop tools to track star particles, and identify and categorize systems as they merge and become accreted by larger systems becoming stellar streams and then continue to become gravitationally disturbed until they are ultimately phase-mixed and become kinematically undistinguishable from other stars in the galaxy. We compare results to other simulations at comparable resolution and find similar results when comparing the velocity dispersions of systems which we classify as intact, stream or phase-mixed.

Download

Share

COinS