Researchers Unveil Groundbreaking Milky Way Simulation Model

Researchers at the RIKEN Center for Interdisciplinary Theoretical and Mathematical Sciences (iTHEMS) in Japan have achieved a significant milestone in astrophysics. Collaborating with colleagues from the University of Tokyo and the Universitat de Barcelona, the team has created the first Milky Way simulations that accurately depict over 100 billion stars across a span of 10,000 years. This groundbreaking simulation is notable for incorporating 100 times more individual stars than previous models while being produced 100 times faster.

The innovative approach utilized by the research team combined the power of 7 million CPU cores, advanced machine learning algorithms, and sophisticated numerical simulations. This model not only represents a leap forward for astrophysics but also enhances the fields of supercomputing and artificial intelligence. 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,” which will be presented at the *Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis* (SC ’25).

Advancing Galactic Studies

The enhanced simulations allow astronomers to test theories of galactic formation and evolution through detailed comparisons with astronomical observations. Historically, creating simulations of such complexity has faced challenges due to the need to accurately model various forces, including gravity, fluid dynamics, supernovae, element synthesis, and the influence of supermassive black holes (SMBHs). Until now, the available computing power limited scientists to modeling galaxies up to one billion solar masses, representing less than 1% of the Milky Way’s total stellar population.

Current state-of-the-art supercomputing systems require approximately 315 hours (over 13 days) to simulate just 1 million years of galactic evolution, which is a mere 0.00007% of the Milky Way’s estimated age of 13.61 billion years. Moreover, simulating the desired 1 billion years would take more than 36 years. Consequently, only large-scale events could be accurately represented, making it challenging to gain insights into finer details of galactic dynamics.

Introducing AI Solutions

To overcome these limitations, lead researcher Hirashima and his team implemented an AI-driven surrogate model, which significantly optimized resource usage. This model was trained on high-resolution simulations of supernovae, allowing it to predict the effects of these explosions on surrounding gas and dust over a period of 100,000 years post-explosion. By integrating this AI component with traditional physical simulations, the team successfully modeled both the large-scale dynamics of a Milky Way-sized galaxy and small-scale stellar phenomena concurrently.

“Our new method can simulate the evolution of galaxies with over 100 billion stars, achieving a remarkable 1 million years of evolution in just 2.78 hours,”

said Hirashima in a press release from RIKEN.

This efficiency means that a full 1 billion years of galactic history could potentially be simulated in just 115 days. The implications of this research extend beyond astrophysics; the AI approach may also enhance simulations in other complex fields, including meteorology, ocean dynamics, and climate science.

The successful validation of this approach through extensive testing on the Fugaku and Miyabi Supercomputer Systems demonstrates its potential to revolutionize the way scientists study our galaxy and the universe at large. As computational demands increase, such innovative methods will be crucial for advancing our understanding of cosmic phenomena.