MPI Parallelisation of a Boltzmann Solver for Supernova Simulations

CI: – Dr. B Mueller

Most massive stars end their lives in a violent explosion known as a core-collapse supernova. These explosions are of broad interest as targets of big astronomical surveys, neutrino telescopes, and gravitational wave detectors like Advanced LIGO, Advanced Virgo, and KAGRA. A sound understanding of core-collapse supernovae and the neutron stars or black holes they make is required to underpin the booming field of compact observations in gravitational waves, X-rays and radio waves. As furnaces for the production of many of the chemical elements in the periodic table, core-collapse supernovae are also of great significance for spectroscopic surveys like GALAH that seek to uncover the chemical evolution of the Milky Way.
Last but not least, core-collapse supernovae have been among the
foremost problems in computational astrophysics for decades. The principal challenge consists in the accurate treatment of neutrinos, which are thought to play a crucial role in energising the explosion. To model how neutrinos stream out of the young neutron star and deposit part of their energy in the surrounding stellar shells, one needs to solve the six-dimensional Boltzmann equation for neutrino transport. We have recently developed an elegant, semi-implicit approach to solve the full Boltzmann equation instead of resorting to the common approximations that supernova models have been forced to make so far. We propose to develop an MPI-parallel version of this new scheme using a domain decomposition approach. We request ADACS developer support to implement and test this MPI version, which will pave the way for the first multi-dimensional (2D) supernova simulations with Boltzmann neutrino transport using computer time resources from ASTAC and NCMAS. This will constitute a major breakthrough in supernova modelling and establish our supernova code CoCoNuT as the leading tool for the next generation of stellar explosion models.