“Traditional supercomputers require one to two years to simulate even relatively small dwarf galaxies at proper resolution.”
– RIKEN Center researcher
Keiya Hirashima and colleagues, identifying the computational bottleneck solved by their new
ASURA-FDPS-ML framework. Their approach uses machine learning to predict how supernovae reshape galaxies—a process crucial for understanding how stars form and how the heavy elements essential for life spread through the cosmos.
