[M-GI30] Computational sciences on the universe, galaxies, stars, planets and their environments

[MGI30-01] Progress and Future Plans of the Program for Promoting Researches on the Supercomputer Fugaku: Structure and Evolution of the Universe Unraveled by Fusion of Simulation and AI

*Ken Ohsuga¹ (1.University of Tsukuba)

Keywords:AI, universe, simulation

The Program for Promoting Researches on the Supercomputer Fugaku, “Structure and Evolution of the Universe Unraveled by Fusion of Simulation and AI,” is a large-scale project involving more than 110 researchers in astronomy and planetary sciences from 32 institutions. This program aims to elucidate the formation and evolution of various astronomical objects, including the large-scale structure of the universe, galaxies, stars, planets, black holes, and so on, through the utilization of AI to achieve unprecedentedly high-precision and efficient simulations. The project began in the 2023 fiscal year and is scheduled to conclude in the 2025 fiscal year, during which significant progress has already been observed across many fields. For example, a six-dimensional Vlasov simulation has successfully tracked the evolution of neutrino distributions, a plasma particle simulation has clarified particle acceleration phenomena within shock waves, and a general relativistic radiation magnetohydrodynamic simulation has succeeded in studying the precession of black hole accretion disks. Progress has also been made in research utilizing AI. For example, a galaxy evolution simulation employing AI to analyze the effects of supernova explosions has succeeded in advancing the understanding of complex processes that govern galaxy evolution. Additionally, a black hole accretion disk simulation incorporating a machine learning model to estimate the distribution of light has been actively developed. Moreover, a machine learning model capable of estimating X-ray intensities and spectra from observational images of hydrogen emission lines is being developed, contributing to the establishment of techniques that facilitate the interpretation of observational results. In this presentation, I will provide a detailed overview of these research achievements and discuss future plans for the program.