Encoding the nearby Universe: a unified observational picture of nearby galaxies across wavelength and scale
Presenter: Dayla Baron, Research Scientist, Stanford University
Abstract: Modern astronomy is entering an era in which the same astrophysical systems are mapped across wavelengths and scales by numerous surveys. This panchromatic view expands the discovery space, and enables joint constraints on the physics of stars, gas, dust, and black holes. However, these datasets are highly heterogeneous, with differing spatial resolutions, point-spread functions, noise properties, and survey strategies, leaving much of the available information fragmented. In this talk, I will describe our ongoing effort to address this challenge through FM-ISM, a probabilistic framework designed to unify multi-wavelength observations of nearby galaxies into a single, coherent model of emission in space and wavelength. By jointly modeling images and spectral cubes across the electromagnetic spectrum, the framework aims to capture the rich correlation structure linking stars, dust, and multiphase gas on scales of ~10–500 pc. This approach seeks to unlock the full information content of existing observations, enabling data-driven discovery and inference using all available data without degrading their native resolution.
Bio: Dalya Baron is an AI+Astro research scientist at Stanford University and a member of the Center for Decoding the Universe, which uses data science and AI to address fundamental questions about the Universe. Previously, she was a Carnegie–Princeton Fellow at Carnegie Observatories. She is an observational astronomer studying galaxy evolution, focusing on how galaxies transition from starburst to quiescence and the role of accreting supermassive black holes. Her work also explores the multiphase interstellar medium on ~100 pc scales in nearby galaxies. She develops and applies machine-learning methods to discover new phenomena and trends in large astronomical datasets.