Author(s): , ,
Institution(s): 1. New York University, 2. University of Oxford
Stellar rotation has many applications in the field of exoplanets. High-precision photometry from space-based missions like Kepler and K2 allows us to measure stellar rotation periods directly from light curves. Stellar variability produced by rotation is usually not sinusoidal or perfectly periodic, therefore sine-fitting periodograms are not well suited to rotation period measurement. Autocorrelation functions are often used to extract periodic information from light curves, however uncertainties on rotation periods measured by autocorrelation are difficult to define. A ‘by eye’ check, or a set of heuristic criteria are used to validate measurements and rotation periods are only reported for stars that pass this vetting process. A probabilistic rotation period measurement method, with a suitable generative model bypasses the need for a validation stage and can produce realistic uncertainties. The physics driving the production of variability in stellar light curves is still poorly understood and difficult to model. We therefore use an effective model for stellar variability: a Gaussian process with a quasi-periodic covariance function. By injecting fake signals into Kepler light curves we show that the GP model is well suited to quasi-periodic, non-sinusoidal signals, is capable of modelling noise and physical signals simultaneously and provides probabilistic rotation period measurements with realistic uncertainties.