Automated detection of filaments in the large scale structure of the universe
We present a new method to identify large scale filaments and apply it to a cosmological simulation. Using positions of haloes above a given mass as node tracers, we look for filaments between them using the positions and masses of all the remaining dark-matter haloes. In order to detect a filament, the first step consists in the construction of a backbone linking two nodes, which is given by a skeleton-like path connecting the highest local dark matter (DM) density traced by non-node haloes. The filament quality is defined by a density and gap parameters characterising its skeleton, and filament members are selected by their binding energy in the plane perpendicular to the filament. This membership condition is associated to characteristic orbital times; however if one assumes a fixed orbital timescale for all the filaments, the resulting filament properties show only marginal changes, indicating that the use of dynamical information is not critical for the method. We test the method in the simulation using massive haloes($M>10^{14}$h$^{-1}M_{\odot}$) as filament nodes. The main properties of the resulting high-quality filaments (which corresponds to $\simeq33%$ of the detected filaments) are, i) their lengths cover a wide range of values of up to $150 $h$^{-1}$Mpc, but are mostly concentrated below 50h$^{-1}$Mpc; ii) their distribution of thickness peaks at $d=3.0$h$^{-1}$Mpc and increases slightly with the filament length; iii) their nodes are connected on average to $1.87\pm0.18$ filaments for $\simeq 10^{14.1}M_{\odot}$ nodes; this number increases with the node mass to $\simeq 2.49\pm0.28$ filaments for $\simeq 10^{14.9}M_{\odot}$ nodes.