Instability, Perturbations, and Substructure in Protoplanetary Disks

This work focuses on the analysis of protoplanetary disks, the sites of planet formation. Numerical, analytic, and machine learning techniques are applied to the topics of gravitational instability, planet-induced perturbations, and non-Keplerian motion in general. This is largely done through kinematic analysis: the quantification of the motion within the disk using line emission observations. Gravitational instability is a process that is thought to be of potential importance early in a disk's lifetime and may facilitate the formation of specific types of planets. I demonstrate that the signatures of gravitational instability contain a rich amount of information that can be difficult to disentangle. Planets in the disk may induce perturbations that are visible in observations. The identification of these perturbations is an important avenue for testing models of planet formation. I use machine learning techniques to find these perturbations and report a previously unidentified planet in the disk HD 142666. Finally, I demonstrate that unsupervised machine learning can be used to identify general non-Keplerian motion or anomalous regions in line emission observations. These results pave the way for a more complete understanding of the influences and effects of gravitational instability, demonstrates the ability of machine learning to identify important features, and introduces a novel method to apply machine learning to astronomical observations across subfields.