Accurate Modeling of Grazing Transits Using Umbrella Sampling

Grazing transits present a special problem for statistical studies of exoplanets. Even though grazing planetary orbits are rare (due to geometric selection effects), many low moderate signal-to-noise cases, significant fraction the posterior distribution is nonetheless consistent with geometry. A failure accurately model can therefore lead biased inferences even cases where planet not actually on trajectory. With recent advances in stellar characterization, limiting factor scientific applications now quality available transit fits themselves, and so time ripe revisit fitting problem. In this paper, we exoplanet using novel application umbrella sampling geometry-dependent parameter basis that minimizes covariances between parameters. Our technique splits into independent Monte Carlo runs grazing, non-grazing, transition regions space, which then recombine single joint probability robust weighting scheme. method be trivially parallelized requires no increase wall clock needed computations. Most importantly, our produces accurate estimates properties both non-grazing orbits, yielding more results than standard methods common star-planet configurations.