mutual transits

I have developed a new photodynamical transit model for overlapping bodies such as an exoplanet/exomoon pair. This work was a response to a lack of open-source transit models for multiple, potentially overlapping bodies. This model is exact and includes analytic derivatives with respect to all input parameters which makes it suitable for gradient-based inference methods such as Hamiltonian MCMC. It is currently the only exact and differentiable transit code available for modeling exomoon-planet or planet-planet mutual transit events. Applications of this code include searching for exomoon transits and modeling simultaneous planetary transits in coplanar multiplanetary systems. These planet-planet mutual events are important because they allow us to constrain the mutual inclination -- an important parameter for understanding the dynamical history and present-day architecture of these systems.

code // paper

gaussian processes for transiting exoplanets

I developed an extension to the celerite algorithm ( Foreman-Mackey et al., 2017) algorithm which allows for the computation of two-dimensional GPs for certain covariance structures. This method can be used to model stellar variability across multiple wavelength bands, making it a useful noise model for multiband transit lightcurves. This includes spectral time-series from JWST, and multiband lightcurves from future transit missions such as PLATO and Ariel. I demonstrated that the precision of measured transit parameters can be significantly improved by modeling correlations in stellar variability across wavelengths.

code // paper

stellar rotation

Knowing the age of a star is important for understanding stellar evolution, investigating star formation histories, and studying their planets. Because stars lose angular momentum over time, we can estimate their ages by measuring their rotational periods. I've used a Gaussian processes to model light curves and measure stellar rotation periods for thousands of K2 targets. Our sample shows strong evidence of a bimodal period distribution which we interpret as indicative of a broken spindown law.