Building merger trees from cosmological N-body simulations - Towards improving galaxy formation models using subhaloes

摘要

Context. In the past decade or so, using numerical N-bodysimulations to describe the gravitational clustering of dark matter(DM) in an expanding universe has become the tool of choice fortackling the issue of hierarchical galaxy formation. As mass resolutionincreases with the power of supercomputers, one is able to grasp finerand finer details of this process, resolving more and more of the innerstructure of collapsed objects. This begs one to revisit time and againthe post-processing tools with which one transforms particles into``invisible'' dark matter haloes and from thereon into luminousgalaxies. Aims. Although a fair amount of work has beendevoted to growing Monte-Carlo merger trees that resemble those builtfrom an N-body simulation, comparatively littleeffort has been invested in quantifying the caveats one necessarilyencounters when one extracts trees directly from such a simulation. Tosomewhat revert the tide, this paper seeks to provide its reader with acomprehensive study of the problems one faces when following thisroute. Methods. The first step in building merger historiesof dark matter haloes and their subhaloes is to identify thesestructures in each of the time outputs (snapshots) produced by thesimulation. Even though we discuss a particular implementation of suchan algorithm (called AdaptaHOP) in this paper, we believe that ourresults do not depend on the exact details of the implementation butinstead extend to most if not all (sub)structure finders. To illustratethis point in the appendix we compare AdaptaHOP's results to thestandard friend-of-friend (FOF) algorithm, widely utilised in theastrophysical community. We then highlight different ways of buildingmerger histories from AdaptaHOP haloes and subhaloes, contrasting theirvarious advantages and drawbacks. Results. We find that the best approach to (sub)halomerging histories is through an analysis that goes back and forthbetween identification and tree building rather than one that conductsa straightforward sequential treatment of these two steps. This isrooted in the complexity of the merging trees that have to depict aninherently dynamical process from the partial temporal informationcontained in the collection of instantaneous snapshots available fromthe N-body simulation. However, we also propose asimpler sequential ``Most massive Substructure Method'' (MSM) whosetrees approximate those obtained via the more complicated nonsequential method. Key words: methods: numerical - methods: N-bodysimulations - cosmology: large-scale structure of Universe