Researchers from the RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences (iTHEMS) in Japan have achieved a groundbreaking feat by creating the world’s first comprehensive simulation of the Milky Way galaxy, accurately representing over 100 billion stars over a span of 10,000 years. This simulation, a collaboration with experts from the University of Tokyo and the Universitat de Barcelona, significantly surpasses previous models by representing 100 times more stars and achieving this feat 100 times faster.
The innovative simulation was made possible through the integration of 7 million CPU cores, advanced machine learning algorithms, and numerical simulations. This combination allows researchers to explore stellar and galactic evolution on an unprecedented scale, marking a significant advancement in the fields of astrophysics, supercomputing, and artificial intelligence.
Breakthrough in Simulations of Galactic Dynamics
The findings were detailed in a paper titled “The First Star-by-star N-body/Hydrodynamics Simulation of Our Galaxy Coupling with a Surrogate Model,” published in the *Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis* (SC ’25). The ability to simulate the Milky Way at this level of detail enables astronomers to test theories regarding galactic formation, structure, and evolution, allowing for direct comparisons with astronomical observations.
Historically, scientists faced significant challenges in simulating galactic dynamics due to the complexity of the forces at play, including gravity, fluid dynamics, supernovae, and the influence of supermassive black holes. Current capabilities have limited the mass of galaxies that can be realistically modeled to approximately one billion solar masses, which is less than 1% of the stars in the Milky Way. Moreover, existing supercomputers would require about 315 hours (over 13 days) to simulate just one million years of galactic evolution, a mere fraction of the Milky Way’s estimated age of 13.61 billion years.
Innovative Solutions Using AI Technology
To overcome these limitations, the team, led by researcher Hirashima, implemented an AI-driven approach utilizing a machine learning surrogate model. This model, trained on high-resolution supernova simulations, predicts the impact of these cosmic explosions on surrounding gas and dust over a span of 100,000 years after the event. By integrating this AI model with physical simulations, the team successfully captured both the large-scale dynamics of a Milky Way-sized galaxy and the intricate details of smaller stellar phenomena.
Testing was conducted on the Fugaku and Miyabi Supercomputer systems at RIKEN and the University of Tokyo, respectively. The results demonstrated that the new method could accurately simulate star resolution in galaxies with more than 100 billion stars, achieving the simulation of one million years of evolution in just 2.78 hours. This breakthrough suggests that an entire 1 billion years of galactic history could be simulated in as little as 115 days, providing astronomers with a powerful tool for investigating galactic evolution.
This research not only opens new avenues for understanding the Milky Way but also illustrates the potential of AI technologies in enhancing complex simulations across various scientific fields, including meteorology, ocean dynamics, and climate science. The incorporation of surrogate AI models could significantly reduce the time and energy demands associated with advanced simulations, paving the way for more efficient and comprehensive studies in the future.
