Phase-space structures – II. Hierarchical structure finder

A new multidimensional Hierarchical Structure Finder (hsf) to study the phase-space structure of dark matter in N-body cosmological simulations is presented. The algorithm depends mainly on two parameters, which control the level of connectivity of the detected structures and their significance compared to Poisson noise. By working in six-dimensional phase space, where contrasts are much more pronounced than in three-dimensional (3D) position space, our hsf algorithm is capable of detecting subhaloes including their tidal tails, and can recognize other phase-space structures such as pure streams and candidate caustics. If an additional unbinding criterion is added, the algorithm can be used as a self-consistent halo and subhalo finder. As a test, we apply it to a large halo of the Millennium Simulation, where 19 per cent of the halo mass is found to belong to bound substructures, which is more than what is detected with conventional 3D substructure finders, and an additional 23–36 per cent of the total mass belongs to unbound hsf structures. The distribution of identified phase-space density peaks is clearly bimodal: high peaks are dominated by the bound structures and show a small spread in their height distribution; low peaks belong mostly to tidal streams, as expected. However, the projected (3D) density distribution of the structures shows that some of the streams can have comparable density to the bound structures in position space. In order to better understand what hsf provides, we examine the time evolution of structures, based on the merger tree history. Given the resolution limit of the Millennium Simulation, bound structures typically make only up to six orbits inside the main halo. The number of orbits scales approximately linearly with the redshift corresponding to the moment of merging of the structures with the halo. At fixed redshift, the larger the initial mass of the structure which enters the main halo, the faster it loses mass. The difference in the mass loss rate between the largest structures and the smallest ones can reach up to 20 per cent. Still, hsf can identify at the present time at least 80 per cent of the original content of structures with a redshift of infall as high as z≤ 0.3, which illustrates the significant power of this tool to perform dynamical analyses in phase space.