3D Milky Way Dust Map using a Scalable Gaussian Process

Interstellar dust in the Milky Way corrupts nearly every stellar observation, and accounting for this corruption is crucial to many investigations from within the Milky Way to cosmology. In this short talk I will present Ziggy, our work on modeling the dust distribution as a Gaussian process, a spatially varying latent field, which we then estimate from stellar observations. We develop a new model and inference method that scales to millions of astronomical observations and kpc scales. I will show some results from Ziggy on the Ananke dataset, a mock Gaia catalogue of a Milky Way-like simulation.