Reconstructing the Big Bang with Physics-Guided Machine Learning
Cosmological observations and theories answer some of the most fundamental questions of humanity such as how the universe began and what it is made of. In the coming decade, cosmology will be blessed with a vast influx of new experimental data. Galaxy surveys will probe the universe in incredible detail, but connecting their measurements to fundamental physics is extremely difficult due to the complex non-linear physics of cosmic structure formation. This interdisciplinary project will develop and apply new methods from machine learning to solve this problem. Physicists and computer scientists at UW–Madison will develop a new technique to look back in time and reconstruct what happened at the big bang from the data we have today. Several recent developments in machine learning will significantly enhance our ability to perform this reconstruction and perhaps lead to new discoveries about the origin of the universe.
Moritz Muenchmeyer, assistant professor of physics
Kangwook Lee, assistant professor of electrical and computer engineering
Gary Shiu, professor of physics