The First Star-by-star $N$-body/Hydrodynamics Simulation of Our Galaxy Coupling with a Surrogate Model

2025-10-27

AI Summary

Traditional N-body/hydrodynamics simulations of galaxies face severe scalability bottlenecks due to prohibitively small timesteps imposed by subgrid-scale, short-timescale processes—such as supernova feedback—limiting both computational efficiency and achievable particle resolution. Method: This work introduces, for the first time, a deep learning–based surrogate model for subgrid physics, replacing computationally expensive explicit subgrid treatments while preserving essential astrophysical feedback mechanisms. The framework integrates N-body dynamics, self-gravitating hydrodynamics, and machine-learned surrogates, and is designed for massively parallel execution across heterogeneous architectures (A64FX, x86-64, CUDA). Contribution/Results: We achieve the highest-resolution galaxy-scale simulation to date: a full-Milky-Way simulation with 300 billion particles and resolved stellar populations numbering in the tens of millions. This constitutes the first numerically resolved, star-by-star “virtual laboratory” for galaxy evolution studies.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Search and Optimization: Sampling/Simulation-based Search
• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

A major goal of computational astrophysics is to simulate the Milky Way Galaxy with sufficient resolution down to individual stars. However, the scaling fails due to some small-scale, short-timescale phenomena, such as supernova explosions. We have developed a novel integration scheme of $N$-body/hydrodynamics simulations working with machine learning. This approach bypasses the short timesteps caused by supernova explosions using a surrogate model, thereby improving scalability. With this method, we reached 300 billion particles using 148,900 nodes, equivalent to 7,147,200 CPU cores, breaking through the billion-particle barrier currently faced by state-of-the-art simulations. This resolution allows us to perform the first star-by-star galaxy simulation, which resolves individual stars in the Milky Way Galaxy. The performance scales over $10^4$ CPU cores, an upper limit in the current state-of-the-art simulations using both A64FX and X86-64 processors and NVIDIA CUDA GPUs.

Problem

• Simulating Milky Way with individual star resolution
• Overcoming supernova-induced timestep limitations in simulations
• Breaking billion-particle barrier in galactic scale simulations

Innovation

• Coupling N-body hydrodynamics with machine learning
• Using surrogate model to bypass supernova timesteps
• Achieving star-by-star galaxy simulation with 300 billion particles