Story perspectives

AI-Powered Milky Way Simulation Cuts Time from 36 Years to 115 Days

11/24/2025

23 5

Summary

Story summary

Researchers at RIKEN's Center for Interdisciplinary Theoretical and Mathematical Sciences developed a Milky Way simulation modeling over 100 billion stars across 10,000 years.
Keiya Hirashima led team, combining artificial intelligence with traditional physics to overcome computational limits.
The AI predicts gas expansion after supernovae, enabling tracking of galaxy-wide dynamics and stellar events.
The method reduces run time from 36 years to 115 days and was verified on the Fugaku supercomputer.

1 of 1

Story summary

Researchers at RIKEN's Center for Interdisciplinary Theoretical and Mathematical Sciences developed a Milky Way simulation modeling over 100 billion stars across 10,000 years.
Keiya Hirashima led team, combining artificial intelligence with traditional physics to overcome computational limits.
The AI predicts gas expansion after supernovae, enabling tracking of galaxy-wide dynamics and stellar events.
The method reduces run time from 36 years to 115 days and was verified on the Fugaku supercomputer.