Bayesian Inference on Milky Way Datasets

A week-long course at the Heidelberg Physics Graduate Days.

Instructor: Gregory Green, Max Planck Institute for Astronomy

Course Abstract:

The Milky Way offers an unparalleled laboratory for studying the processes that shape galaxies, at a star-by-star level of detail. Over the last two decades, large ground- and space-based surveys have revolutionized our understanding of the Milky Way. The Sloan Digital Sky Survey, Pan-STARRS 1, LAMOST, Gaia and other surveys have provided photometry and astrometry of billions of stars and spectra of millions of stars.

However, extracting information from these large datasets about the Milky Way requires a rigorous approach. In this course, I will first describe several modern astronomical surveys, and the type of data they provide about the Milky Way. Next, I will give an introduction to Bayesian inference, a framework for thinking about and quantifying uncertainty. Finally, we will work through a few examples of how these methods apply to Milky Way datasets, including:

  1. Inferring distances to stars using parallaxes.
  2. Inferring the properties of stars and the three-dimensional distribution of interstellar dust using stellar photometry and parallaxes.
  3. “Weighing” the Milky Way using the orbits of its satellite galaxies.

Through this course, my hope is that students will gain familiarity not only with Milky Way datasets, but that they will also gain hands-on experience with statistical methods that are applicable throughout many areas of science.

Lecture Notes / Presentations:


There are various Jupyter notebooks that students can use to explore some of the topics covered in this course. The notebooks can be found here.

If you don’t have Jupyter installed on your computer, these notebooks can be run in your browser directly, on Google Colab (requires a Google account) or JupyterLab (click “Try JupyterLab” and then upload the notebooks).