Data Guide | Dark Matter Simulation Basics
An efficient way to simulate the universe inside a supercomputer is to focus on the dominant mass distribution and its evolution. This usually involves running a collisionless N-body simulation in a volume that is large enough to be representative of the Universe as a whole and provides a significant reduction in computational effort at fixed resolution compared with hydrodynamic galaxy formation simulations. Hydrodynamic effects are complicated and slow to compute numerically relative to the rather simple calculations required in a gravity-only simulation. Hence, gas and galaxies are often added later in post-processing using semi-analytic or other statistical techniques.
As the universe evolves, gravity pulls small structures together to assemble larger structures (i.e. hierarchical growth). Within the numerical simulation, such “halos” are typically identified using a Friends-of-Friends (FoF) algorithm which detects gravitationally bound systems of particles and determines their properties (e.g. Davis et al. 1985; More et al. 2011). Structures within structures (i.e. sub-structures) can be found using a variety of methods (e.g. Springel et al. 2001; Behroozi, Wechsler & Wu 2013). Such sub-halos are typically expected to host smaller satellite galaxies; the evolution of these satellite galaxies is greatly influenced by the central galaxy and its host dark matter halos. More massive halos accrete less massive halos, along with their galaxies. These galaxies will eventually merge with the more massive central galaxy.
This information, calculated across all time-steps in a simulation for a particular object, defines its merger tree. The collection of such trees is then used as input to construct a galaxy formation model. An example halo merger tree from the Millennium Simulation (Springel et al. 2005) is shown here. Here, the top panel shows the tree itself for a 1.9 × 10^13 Msun halo at z = 0 (assuming h = 0.73), while the corresponding mass growth history with time is shown in the lower panel.