The accuracy of subhalo detection

With the ever increasing resolution of N-body simulations, accurate subhalo detection is becoming essential in the study of the formation of structure, the production of merger trees and the seeding of semi-analytic models. To investigate the state of halo finders, we compare two different approaches to detecting subhaloes; the first based on overdensities in a halo and the second being adaptive mesh refinement. A set of stable mock Navarro-Frenk-White (NFW) dark matter haloes was produced and a subhalo was placed at different radii within a larger halo. subfind (a friends-of-friends based finder) and ahf (an adaptive mesh based finder) were employed to recover the subhalo. As expected, we found that the mass of the subhalo recovered by subfind has a strong dependence on the radial position and that neither halo finder can accurately recover the subhalo when it is very near the centre of the halo. This radial dependence is shown to be related to the subhalo being truncated by the background density of the halo and originates due to the subhalo being defined as an overdensity. If the subhalo size is instead determined using the peak of the circular velocity profile, a much more stable value is recovered. The downside to this is that the maximum circular velocity is a poor measure of stripping and is affected by resolution. For future halo finders to recover all the particles in a subhalo, a search of phase space will need to be introduced. © 2010 The Authors. Journal compilation © 2010 RAS.