a small part of the via lactea halo

high resolution Milky Way dark matter halos simulated on NASA's Columbia and ORNL's Jaguar supercomputers









The via lactea project uses cosmological N-body simulations to study the dark matter halo of the Milky Way. This site provides links to press articles and scientific publications, and additional material, like simulation data, movies and images.

The first run (VL-1) was completed in April 2006 on NASA's Columbia supercomputer. It contains over 200 million resolution elements with a mass of 21 thousand solar masses and took about 350 thousand CPU hours to run. The second run (VL-2) has over one billion particles with a mass of only 4.1 thousand solar masses and uses an impropved, physical time-stepping method (Zemp et al. 2007). VL-2 took about one million CPU hours to finish and was run in November 2007 at the Oakridge National Lab on the Jaguar supercomputer. VL-2 is an INCITE project awarded by the U.S. Department of Energy.

Both VL-1 and VL-2 focus on Milky Way size halos (about 2e12 solar masses) which had no recent major mergers, making them suitable hosts for a Milky Way-like disk galaxy. The simulations were performed with the MPI-parallel tree-code PKDGRAV2 (principal author Joachim Stadel).

in the news

VL-2:   VL-2 made the top ten breakthroughs of 2008 selected by the DOE's office of advanced scientific computer research
UC Santa Cruz - press release, August 2008
NCCS - press release, August 2008

VL-1:   UC Santa Cruz - press release, November 2006
NewScientist.com online article, December 2006
NASA Advanced Supercomputing (NAS) Division
Mid-County Post (local newspaper)
NewScientist article about gamma rays, 29 April 2007
and others ...

cosmological parameters and inital conditions

We use multiple mass particle grid initial conditions generated with the GRAFIC-2 package (Berschinger 2001). The high resolution region is embedded within a periodic box of comoving size 90 Mpc (VL-1, 40 Mpc for VL-2) which is sampled at lower resolution to account for the large scale tidal forces. VL-1 used the original version of GRAFIC-2, which incorrectly uses the baryonic instead of the CDM power spectrum for the refinement levels. This slighly reduces the small scale power (like having a somewhat smaller spectral index of about n_s=0.90), but it does not affect our conclusions from VL-1. To generate the VL-2 initial conditions we used a corrected, parallel version of GRAFIC-2 developed by Mike Kuhlen and Doug Potter.

For both VL-1 and VL-2 we use the best-fit cosmological parameters from version 1 of the WMAP three-year data release (Spergel et al. 2006): Omega_M = 0.238, Omega_Lambda = 0.762, H_0= 73 km/s/Mpc, n_s=0.951, and sigma_8=0.74. These n_s and sigma_8 values are lower than the latest 5-year WMAP only parameters (Komatsu et al. 2008), but they lie within 1.55 and 0.85 sigma. Note that many cosmological simulations adopt sigma_8=0.9 and n_s=1.0 (both about 3 sigma above the 5 year WMAP only values), which results in significantly more small scale power than used for the via lactea runs and leads to about 30 percent higher subhalo velocity functions (see e.g. Zentner & Bullock 2003, Diemand et al. 2008).

last updated: November 24, 2008, by J. Diemand

Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 2.5 License,
please acknowledge the via lactea project and/or its authors wherever material from this web site is used.