AI-Driven High-Resolution Galaxy Simulations: Accelerating Galaxy Evolution Modeling from 8 Months to 2

An international collaborative research group, including Keiya Hirashima (Special Postdoctoral Researcher, iTHEMS), has developed a surrogate model using deep learning, a form of artificial intelligence (AI), to predict the complex physical processes of supernova explosions. This model has been integrated into a galaxy simulation code for the first time.

This achievement marks the first instance of accelerating high-resolution "star-by-star" galaxy simulations by performing deep learning inference in real time during the simulation—something that was previously difficult to realize. The new method is expected to contribute to detailed analysis of supernova feedback in the formation and evolution of our own Milky Way galaxy.

For more information, please refer to the related links below.