Loose Ends for the Exomoon Candidate Host Kepler-1625b

Abstract

The claim of an exomoon candidate in the Kepler-1625b system has generated substantial discussion regarding possible alternative explanations for the purported signal. In this work, we examine these possibilities in detail. First, the effect of more flexible trend models is explored, and we show that sufficiently flexible models are capable of attenuating the signal—although this is an expected byproduct of invoking such models. We also explore trend models using x- and y-centroid positions, and show that there is no data-driven impetus to adopt such models over temporal ones. We quantify the probability that the 500 ppm moon-like dip could be caused by a Neptune-sized transiting planet to be <0.75%. We show that neither autocorrelation, Gaussian processes, nor a Lomb–Scargle periodogram are able to recover a stellar rotation period, demonstrating that K1625 is a quiet star with periodic behavior <200 ppm. Through injection and recovery tests, we find that the star does not exhibit a tendency to introduce false-positive dip-like features above that of pure Gaussian noise. Finally, we address a recent reanalysis by Kreidberg et al. and show that the difference in conclusions is not from differing systematics models but rather the reduction itself. We show that their reduction exhibits, in comparison to the original analysis: (i) slightly higher intraorbit and post-fit residual scatter, (ii) ≃900 ppm larger flux offset at the visit change, (iii) ≃2 times larger y-centroid variations, and (iv) ≃3.5 times stronger flux-centroid correlation coefficient. These points could be explained by larger systematics in their reduction, potentially impacting their conclusions.