Core Insights - An international team from Japan's RIKEN, the University of Tokyo, and the University of Barcelona has developed the first galaxy model capable of tracking the evolution of over 100 billion stars, achieving a scale 100 times greater than previous models [1][2] - Traditional models were limited to simulating systems of up to 1 billion solar masses, while the Milky Way contains over 100 billion stars, leading to challenges in accurately representing astronomical events [1] - The new model combines deep learning with traditional physical simulations, allowing for significant reductions in computation time for simulating the evolution of the galaxy [2] Summary by Sections - **Model Development** - The new galaxy model can track the evolution of over 100 billion stars, surpassing previous models by a factor of 100 [1] - Previous models could only simulate systems up to 1 billion solar masses, which limited their accuracy in representing the Milky Way [1] - **Computational Efficiency** - The new approach reduces the time required for simulating a million years of evolution to just 2.78 hours, and a billion years to 115 days [2] - Traditional simulations required 315 hours for a million years and 36 years for a billion years, highlighting the efficiency of the new model [1][2] - **Technological Innovation** - The integration of deep learning with traditional simulations marks a significant advancement, moving beyond mere pattern recognition to becoming a genuine scientific discovery tool [2] - The AI system was trained on high-resolution supernova simulation data to predict the movement of interstellar gas following supernova explosions [1]