Story perspectives

AI-Powered Simulation Revolutionizes Milky Way Evolution Studies

11/21/2025

37 5

Summary

Story summary

Keiya Hirashima's team at the RIKEN Center in Japan built a machine-learning–augmented Milky Way simulation that runs 100 times faster and uses 100 billion particles.
The model maps galactic evolution over billions of years and cuts a billion-year run from 36 years to 115 days by integrating short-term supernova events into processes.
Hirashima says the AI-accelerated approach could revolutionize simulations across fields and deepen understanding of galactic and elemental formation.

1 of 1

Story summary

Keiya Hirashima's team at the RIKEN Center in Japan built a machine-learning–augmented Milky Way simulation that runs 100 times faster and uses 100 billion particles.
The model maps galactic evolution over billions of years and cuts a billion-year run from 36 years to 115 days by integrating short-term supernova events into processes.
Hirashima says the AI-accelerated approach could revolutionize simulations across fields and deepen understanding of galactic and elemental formation.