Numerical simulations are now the dominant tool in theoretical cosmology. As advances in computer power and software algorithms continue into the petascale regime, it will become possible to simulate the entire visible universe at high mass and spatial resolutions. This project will develop a new, petascale-optimized version of the gravity/hydrodynamics code GADGET, the most widely used cosmological simulation code, using an efficient and low-risk incremental strategy. Hybrid shared memory, load balancing and other improvements will allow the new p-GADGET to run on (of order) a million cores. p-GADGET will be applied to two fundamental problems for which current numerical methods fall short by a factor of a thousand. This includes the first hydrodynamical simulation capable of studying the formation of the vast range of quasars and galaxies, from the smallest to the rarest and most luminous, using well-tested algorithms for star formation, black hole formation, accretion, and feedback. p-GADGET will also create the ultimate dark matter simulation, one that fully resolves the mass function of dark matter halos by modeling every single clump that hosts a galaxy. Matching these results to our observed Universe will show whether or not we fully understand the astrophysical impact of dark matter on cosmology.
The public version of GADGET has been wildly successful. p-GADGET will similarly be publicly available, opening up the domain of efficient petascale science to a wide audience, and adding strongly to the user base of the first petascale machines. The simulation outputs will also be generally available for use by anyone: an extremely broad base of researchers will be necessary to exploit the detail and enormous amount of information in simulations produced with petascale resources. This new understanding of how structure forms in the Universe on a wide range of scales, with accompanying visualizations, will inform an expansive outreach program, including high school students, middle school students from minorities under-represented in science, undergraduate coursework projects, and public presentations.
