On The Detection And Characterization Of Exomoons Through Survey And Targeted Observations

The following chapter further investigates a number of alternative hypotheses relating to the candidate moon, though the conclusion that an exomoon best explains the data in hand remains unchanged. Finally, I present the results of an effort to identify candidate exomoon signals in the Kepler data by developing a convolutional neural network trained on O(105) Kepler light curves injected with simulated planet and moon transit signals. The most promising exomoon candidates identified by the neural network are examined in detail, undergoing a full photodynamical model fit and Bayesian model selection. I conclude by discussing the outlook for the moon search, highlighting strategies for future work and myriad unanswered questions that should be pursued in the coming years.