Empirical Study of Simulated Two-planet Microlensing Event

Abstract

We undertake the first study of two-planet microlensing models recovered from simulations of microlensing events generated by realistic multi-planet systems in which 292 planetary events including 16 two-planet events were detected from 6690 simulated light curves. We find that when two planets are recovered, their parameters are usually close to those of the two planets in the system most responsible for the perturbations. However, in one of the 16 examples, the apparent mass of both detected planets was more than doubled by the unmodeled influence of a third, massive planet. This fraction is larger than, but statistically consistent with, the roughly 1.5% rate of serious mass errors due to unmodeled planetary companions for the 274 cases from the same simulation in which a single planet is recovered. We conjecture that an analogous effect due to unmodeled stellar companions may occur more frequently. For seven out of 23 cases in which two planets in the system would have been detected separately, only one planet was recovered because the perturbations due to the two planets had similar forms. This is a small fraction (7/274) of all recovered single-planet models, but almost a third of all events that might plausibly have led to two-planet models. Still, in these cases, the recovered planet tends to have parameters similar to one of the two real planets most responsible for the anomaly.