Japanese astronomers innovated an artificial intelligence (AI) technique to remove noise in astronomical data that is based on random variations in galaxy shapes. They then applied the new tool to actual data from Japan’s Subaru Telescope, discovering that the mass distribution derived from using this method is consistent with the current models of the Universe. This innovation represents a powerful tool for analyzing big data from current and planned astronomy surveys.
Wide area survey data can be used to study the large-scale structure of the Universe through measurements of gravitational lensing patterns. However, the gravity of a foreground object, can distort the image of a background object, such as a more distant galaxy. The large-scale structure, consisting mostly of mysterious “dark” matter, can distort the shapes of distant galaxies too, but subtly. Averaging many galaxies in an area is necessary to create a map of foreground dark matter distributions.
There are galaxies that are just innately a little funny looking, making it difficult to distinguish between a galaxy image distorted by gravitational lensing and a galaxy that is actually distorted. This is referred to as shape noise.
To compensate for shape noise, a team of Japanese astronomers first used ATERUI II, the world’s most powerful supercomputer to generate 25,000 mock galaxy catalogs based on real data from the Subaru Telescope. They then added realist noise to these perfectly known artificial data sets and trained an AI to statistically recover the lensing dark matter from the mock data. Using AI, the team found a distribution of foreground mass consistent with the standard cosmological model.
The results appeared as Shirasaki et al. “Noise reduction for weak lensing mass mapping: an application of generative adversarial networks to Subaru Hyper Suprime-Cam first-year data” in the June 2021 issue of Monthly Notices of the Royal Astronomical Society.
Original Release: Eureka Alert