"Data-driven discovery in the astronomical time domain"

Abstract: Thousands of exoplanets have been discovered over the past few years using observations from large and homogeneous time domain surveys of nearby stars. To push the exoplanet detection threshold to the smallest planets or the longest orbital periods using these data, we combine physical models of exoplanets with data-driven models of the stars and the spacecraft. As part of this process, we also constrain the physical properties of the target stars using the same time series. Scaling these models to hundreds of thousands of stars with tens of thousands of measurements each poses an interesting technical challenge that we have solved through interdisciplinary collaboration, the development of new scalable time series algorithms, and the release of well-tested and easy-to-use open source implementations of these methods. In this talk, I will describe some current and future astronomical time domain surveys and the new methods that we have developed for analyzing these data. I will discuss the implications of this work and present some of the open source software that we have released for analyzing time domain datasets.