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 (for example, Gaussian Processes with
linear scaling), 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.
To begin viewing slides, click on the first slide below. (Or, view as pdf.)