The Milky Way and Local Group are an unparalleled laboratory in which to study a wide range of phenomena, from stellar populations to dark matter subhalos. Yet our view of the Galaxy is obscured by a thick blanket of interstellar dust. In order to gain a full picture of our Galaxy, it is necessary not only to determine the three-dimensional distribution of stars, but to simultaneously map interstellar dust. In this talk, I will discuss a new three-dimensional dust map, based on photometry of nearly a billion stars, as well as Gaia parallaxes for hundreds of millions of stars. This new map is a major step forward, combining accurate distances from Gaia with more advanced priors on the correlated spatial distribution of dust, in order to ensure a smoother, higher signal-to-noise map.

In order to extend this project to the Southern Hemisphere, I have helped lead the Dark Energy Camera Plane Survey (DECaPS), which is imaging the Southern Galactic Plane in the optical and near-infrared. Already, this survey has generated one of the largest photometric catalogs in existence, with two billion unique point sources. I will discuss a particular challenge that arose in the analysis of this data, which we solved with the aid of convolutional neural networks.

The Local Group also gives us a unique environment in which to test the predictions of the standard Lambda CDM cosmology at small scales. I will discuss a framework for approaching the "Missing Satellites Problem" - the possible overabundance of Milky Way satellites predicted by Lambda CDM. This problem is particularly important, because if the perceived tension with Lambda CDM is real, it would point to possible modifications to the properties of dark matter in the standard cosmological model. Finally, I will discuss new possibilities that the rich Gaia dataset has opened up for testing Lambda CDM at small scales.

BIO
Gregory M. Green
Porat Fellow
Kavli Institute for Particle Astrophysics and Cosmology
Stanford University

Gregory is passionate about applying modern statistical techniques - from Bayesian inference to convolutional neural networks - to Galactic structure and near-field cosmology. He is currently a Porat Postdoctoral Fellow at Stanford. When he was a graduate student at Harvard working with Prof. Douglas Finkbeiner, Gregory modeled the distance, reddening and type of a billion stars observed by the Pan-STARRS 1 telescope to create a three-dimensional map of Milky Way dust, covering three quarters of the sky. As a postdoctoral fellow at Stanford, he has worked to extend this project to the Southern Hemisphere with the DECaPS survey, incorporated stellar parallaxes from Gaia into the three-dimensional dust map, and worked on the missing satellites problem, to help understand if Lambda CDM cosmology is consistent with the observed satellite population of the Milky Way. In the coming years, Gregory aims to harness the flood of data that Gaia, LSST and other surveys will provide to uncover Galactic structure and test our cosmological models. When not researching astrophysics or writing code, Gregory can usually be found studying Mandarin, German or French, or reading about Classical history.

Host: Prof. Shude Mao