UCLA

In this thesis, I expand on existing methodologies for identifying distant planetary and stellar companions to known planet hosts. First, I outline my discovery of a distant planetary companion to Kepler-1656b, a highly eccentric sub-Saturn. To make this discovery, I used a combination of traditional radial velocity detection methods and dynamical modeling of the system to resolve the properties of an outer planetary companion. In doing so, I characterized the dynamical environment of the Kepler-1656 system and found that, surprisingly, the outer planet is exciting the inner planet eccentricity in situ, i.e. without inducing inward migration of the inner planet. I also identified signatures of similar companions, which I call “gentle giants”, in the broader sub-Saturn and giant planet populations.

The second portion of my thesis explores more experimental methods of detecting stellar companions to known planet hosts using data-driven spectroscopy. Recent advances in data-driven spectroscopy have enabled more modeling of stellar spectra, which may in turn allow us to detect signatures of stellar companions to planet hosts that are missed by more traditional binary detection methods. I trained data-driven models on spectra from the Gaia mission and Keck-I telescope, and in doing so, developed a novel wavelet-based method for removing non-astrophysical contamination in stellar spectra. I found that while our models are excellent at label transfer– a standard application of data-driven spectroscopy– they are limited in their ability to accurately model stellar spectra and identify hidden stellar companions. I also comment on the state of the field of data-driven spectroscopy and outline possible paths towards more accurate spectroscopic models and surveys for planet-hosting binaries.