Gamma-Ray Bursts (GRBs) are the brightest explosions in the Universe, and Core-Collapse Supernovae (CCSN) are the most energetic events in the Universe. CCSN and the long-soft subset of GRBs both originate from massive stars, but the details of their central engines are essentially unknown. The CCSN and GRB problems and the possible CCSN-GRB relationship are major current questions in astrophysics. Modeling GRBs is a true petascale problem, needing a multi-physics approach over an extreme dynamic range in both physical length scale and dynamic time scale. Of central importance in any comprehensive GRB model is dynamical general relativity combined with magnetohydrodynamics, radiation transport of neutrinos and photons, and neutrino and nuclear microphysics. This award supports a five year research program to develop algorithms not present today, able to address stellar core collapse, supernovae and black hole formation, accretion dynamics, jet formation and sustainment, jet propagation and break out. The team will implement adaptive mesh refinement and parallel input/output to augment the Cactus PDE solving framework so that it runs efficiently on petascale computers. This synergistic interdisciplinary team of both young and experienced researchers spans numerical relativity, astrophysics, and computer science. The team consists of 12 senior investigators including computational and computer scientists at LSU's Center for Computation and Technology led by the PI and co-PI Gabrielle Allen, plus several computational astrophysicists from Princeton and Santa Cruz in the US, Southampton in the UK, and Potsdam in Germany. The award also supports at least two postdoctoral researchers at LSU and Princeton. The team will build on already funded work on other aspects of Cactus infrastructure, and includes members involved in the software development for the Blue Waters project. This project is also closely connected to the NSF supported CyberTools Louisiana statewide cyberinfrastructure project.
On the physics side, this research addresses unsolved problems in modeling some of the most energetic events in the Universe. On the computer science side, the tools and methods to be developed to address science problems on petascale architectures such as Blue Waters will be applicable to a wide range of other fields. The time and effort required to develop and debug scientific software has become a bottleneck in many areas of science and engineering, and the difficulty of developing high-performance computing software is recognized as one of the most significant challenges today in the effective use of large scale computers. The developed pieces for dynamic scheduling, scalability, higher order methods, long-term accuracy, support for multiphysics, checkpointing, and input/output, will all be available for other projects, most of them via the open source Cactus distribution mechanism: the Cactus Framework already provides an enabling environment for applications from different disciplines. The research integrates with outreach programs locally and nationally to help prepare other applications for petascale computing and prepare the next generation of peta-savvy researchers. It also includes outreach to students at the K-12, undergraduate and graduate levels, including summer workshops, helping to introduce future researchers at every level to the concepts and challenges of petascale computing.
