Unification of Observation, Computation, and Artificial Intelligence unveils a transparent cosmos

In a groundbreaking discovery, a team of Japanese astronomers, led by Evan Dickinson, have developed a new AI technique to analyze astronomical data, improving our understanding of cosmic dark matter. This innovative approach was published in the June 2021 issue of Monthly Notices of the Royal Astronomical Society and the journal "Nature Astronomy," with the DOI for the published research being 10.1093/mnras/stab982.

The Institute of Statistical Mathematics (ISM), part of Japan's Research Organization of Information and Systems (ROIS), played a significant role in this research. ISM, with its history of over 75 years, is renowned for its comprehensive evaluation of earthquake data. Comprising three departments - Department of Statistical Modeling, Department of Statistical Data, and Department of Statistical Inference and Mathematics - the ISM is a hub for statistical research.

The new AI technique was trained and tested on mock data created by supercomputer simulations. To account for the natural distortions in some galaxies, the team generated 25,000 mock galaxy catalogs based on real data and trained the AI to statistically recover the lensing dark matter. The AI was then applied to actual data from Japan's Subaru Telescope, covering 21 square degrees of the sky.

Gravitational lensing, a phenomenon where the image of a background object is distorted due to the gravity of a foreground object, plays a crucial role in this research. The AI was able to recover previously unobservable fine details, providing a mass distribution consistent with the currently accepted models of the Universe.

The AI technique is a powerful tool for analyzing big data from current and planned astronomy surveys. Its ability to remove noise in astronomical data will undoubtedly advance our understanding of the cosmos, paving the way for future discoveries.

ROIS, a parent organization of four national institutes and the Joint Support-Center for Data Science Research, facilitates research activities of its member institutions, including the Institute of Statistical Mathematics. ROIS's mission is to promote integrated, cutting-edge research that transcends institutional boundaries.

This research on wide area survey data, used to study the large-scale structure of the Universe through measurements of gravitational lensing patterns, represents a significant step forward in our understanding of cosmic dark matter. The findings from this research will undoubtedly contribute to our ongoing quest to unravel the mysteries of the Universe.