NERSCPowering Scientific Discovery for 50 Years

NERSC Initiative for Scientific Exploration (NISE) 2011 Awards

Large High-Resolution Galaxy Simulation Program

Joel Primack, University of California - Santa Cruz

Associated NERSC Project: Galaxy Formation Simulations (mp363)

NISE Award: 1,500,000 Hours
Award Date: June 2011

A combination of observations with Hubble Space Telescope and simulations on NERSC's largest system Hopper-II will allow us to resolve many mysteries of galaxy formation. One of the key issues is the formation of bulges of spiral galaxies and the formation of elliptical galaxies. Current theory (mostly hydro simulations) has two competing scenarios. Bulges can be formed either in large mergers of galaxies or through violent instabilities developing in disk galaxies in the early Universe. With few high-quality simulations available so far (about a dozen per year by different groups) it is difficult to interpret observations and resolve this critial issue. The situation will be different once we simulate hundreds of galaxies with different masses, merging histories, and environments.

The Cosmic Assembly Near-IR Deep Extragalactic Legacy Survey (CANDELS) is a very large (over 900 orbits) new Hubble Space Telescope (HST) observational program to study the the first third of galactic evolution from redshift $z = 8$ to 1.5 (0.6 to 4.5 billion years after the Big Bang) via deep imaging of more than 250,000 galaxies exploiting the new WFC3 camera on HST. Five premier multi-wavelength sky regions are targeted; each has multi-wavelength data and also extensive spectroscopy of the brighter galaxies. We are participants in CANDELS, and this NISE proposal intends to complement these new observations with a large suite of hydro plus N-body cosmological simulations of galaxies that include the numerous physical processes involved: hydrodynamics, radiative atomic and molecular cooling, stellar feedback, metal enrichment, radiation pressure, and the fueling of and feedback from supermassive black holes. Development of this simulation program was started on the NERSC Bassi system, as described and illustrated in the article ``Baby Brutes'' in the NERSC Annual Report 2008-2009, pp. 64-68. We have continued developing our codes on the NASA Ames Schirra and Pleiades systems. Several improvements have now been added, and the simulations are now even more realistic.

The many 24-core nodes on Hopper-II will allow us to do mass production of extremely high-resolution galaxy simulations, which was unthinkable even a few years ago. We propose to simulate a statistical sample of 300 galaxies, a factor of 20 more than we have at present. This will be by far the best set of cosmological simulations of galaxy formation. With hundreds of galaxies simulated we can also explore numerous issues that were difficult to address before, such as the effects of environment, the frequency and duration of major mergers, and the formation of spiral galaxy bulges and of elliptical galaxies. We now have tools -- including our premier radiative transfer code SUNRISE -- to produce the same quantities that we get from observations: realistic images in many wavebands including the effects of dust, spectral energy distributions, stellar masses and ages, and kinematics. This allows a direct comparison of the numerical simulations with astronomical observations.