Key Challenges: Couple hydrodynamics, nuclear reactions, and — eventually — general relativity and magnetism, to the non-thermal transport of six kinds of neutrinos (electron, muon and tauon and their antiparticles) and do it all with high spatial resolution in 3-D.
Why it Matters: Few events in the cosmos are as energetic as the explosion of a massive star in a supernova. For several seconds, the neutrino luminosity of a single event exceeds that of the rest of the visible universe. The output in light can rival that of an entire galaxy. An event on the other side of our galaxy would be as bright as the sun. Corecollapse supernovae are also responsible for making most of the elements heavier than helium in nature. In their interiors, conditions exist that test our understanding of novel particle physics and the behavior of matter at high density. They are also poorly understood. No supernova has even been completely modeled from first principles.
Accomplishments: Used the CASTRO radiation-hydrodynamics code; explored the spatial dimension dependence of neutrino heating. Found that likelihood of an explosion is a direct function of dimensionality in the simulations: 3D>2D>1D. The effect is larger than many science effects such as relativity & scattering. This is believed to represent a quantum leap in supernova understanding and it suggests that available computer power is a key limit in further understanding. The runs for this used Franklin with 256-16,384 cores, typically 8,192; all visualization was done using ViSIT on Franklin.
Investigators: Stan E Woosley and Candace Church (UC Santa Cruz); John Bell and Ann Almgren (LBNL); Adam Burrows and Jason Nordhaus (Princeton University); A. Heger (University of Minnesota
NERSC Contribution: NERSC chose this project as a NISE recipient in 2009.