Flatiron Institute’s Center for Computational Astrophysics, New York, USA
Rigorous statistical methodology has been widely adopted across astrophysics, driven in large part by the availability of well-documented and user-friendly open source software. But, as datasets grow and research questions continue to get more ambitious, we need to continuously re-evaluate our tooling choices and learn from methodological developments across astrophysics and other disciplines.
In this talk, I will give some examples of how and why open source tools developed for other purposes (like machine learning) can be used to accelerate and improve our data analysis workflows. I will, in particular, highlight some of my interdisciplinary work to develop computationally efficient and physically motivated methods for time domain astronomy, with specific applications to exoplanets and stellar variability, within modern high-performance model building frameworks. There are, however, some limitations to the broad application of these tools and ideas with astrophysics, so I will discuss some of the challenges and propose some possible approaches for tackling these issues.
2021 May 05, 13:30
Online broadcast (Zoom)