Fitting Multi-Planet Transit Models to Photometric Time-Data Series by Evolution Strategies

摘要

In this paper we present the application of an evolution strategy to the problem of detecting multi-planet transit events in photometric time-data series. Planetary transits occur when a planet regularly eclipses its host star, reducing stellar luminosity. The transit method is amongst the most successful detection methods for exo-planets and is presently performed by space telescope missions.The goal of the presented algorithm is to find high quality fits of multi-planet transit models to observational data, which is a challenging computational task. In particular we present a method for an effective objective function evaluation and show how the algorithm can be implemented on graphics processing units. Results on artificial test data with three artificial planets are reported.