Investigating Planet Formation Through Simulation, Observation, and Machine Learning
Restricted (Penn State Only)
- Author:
- Zawadzki, Brianna
- Graduate Program:
- Astronomy and Astrophysics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 09, 2023
- Committee Members:
- Bradford Foley, Outside Unit & Field Member
Rebekah Dawson, Major Field Member
Kevin Luhman, Major Field Member
Robin Ciardullo, Program Head/Chair
Ian Czekala, Chair & Dissertation Advisor
Eric Ford, Major Field Member - Keywords:
- astronomy
protoplanetary disks
exoplanets
planet formation
machine learning
interferometry
imaging - Abstract:
- A robust understanding of the planet formation process is needed to contextualize the diverse population of known exoplanets. In this dissertation I used recent observations of protoplanetary disks and exoplanetary systems in combination with planet formation simulations to study multiple stages of planet formation. To understand how various planetary architectures arise, I ran N-body simulations with the inclusion of disk torques that drive planet migration. The first simulations were focused around low-mass stars; I discovered that M dwarf systems tend to evolve rapidly, with planets forming resonant chains and migrating into the disk cavity before the gas disk dissipates. This process completely rearranged the disk solids, making it impossible to reliably reconstruct the initial distribution of disk solids from observations of fully-formed planetary systems. The other simulations were centered around Sun-like stars, and the resulting systems resembled those detected by the Kepler space telescope; I examined potential causes of the apparent excess of Kepler systems with only a single transiting planet, and present evidence that this phenomenon is caused by migration traps (disk locations where planets of certain masses cannot migrate). In this scenario, trapped planets do not migrate or become massive enough to be detectable by Kepler, resulting in an overabundance of single-transiting systems. Planets that avoid migration traps tend to form a compact system of inner planets similar to close-in multi-planet Kepler systems. I also determined best practices for high-resolution imaging of sub-mm interferometric protoplanetary disk observations, particularly from the Atacama Large Millimeter-submillimeter Array (ALMA). I used regularized maximum likelihood (RML) imaging techniques on both simulated and real ALMA protoplanetary disk observations, characterizing the behavior and outcomes of various types of regularization. In order to synthesize RML images at the highest angular resolution and image fidelity possible, I also studied and applied cross-validation methods to these datasets. I found that RML imaging methods performed exceptionally well on both real and simulated ALMA protoplanetary disk observations, resulting in up to a threefold improvement in angular resolution compared to traditional image synthesis methods like CLEAN without sacrificing sensitivity. I applied these RML imaging techniques to new observations of two protoplanetary disks from exoALMA, an ALMA large program designed for detecting deviations from Keplerian rotation ("velocity kinks") in disks that may be indicative of active planet formation. I found that RML imaging methods reproduced the tentative velocity kinks seen in the CLEAN images of the 12CO emission of the protoplanetary disks around LkCa 15 and SY Cha, suggesting that these kinks are real features rather than mere artifacts of a specific imaging procedure.