The nature of assembly bias - I. Clues from a LCDM cosmology

Abstract

We present a new proxy for the overdensity peak height for which the large-scale clustering of haloes of a given mass does not vary significantly with the assembly history. The peak height, usually taken to be well represented by the virial mass, can instead be approximated by the mass inside spheres of different radii, which in some cases can be larger than the virial radius and therefore include mass outside the individual host halo. The sphere radii are defined as r = $a$ delta_t + $b$ log_10(M_vir/M_nl), where delta_t is the age relative to the typical age of galaxies hosted by haloes with virial mass M_vir, M_nl is the non-linear mass, and $a$=0.2 and $b$=-0.02 are the free parameters adjusted to trace the assembly bias effect. Note that $r$ depends on both halo mass and age. In this new approach, some of the objects which were initially considered low-mass peaks belong to regions with higher overdensities. At large scales, i.e. in the two-halo regime, this model properly recovers the simple prescription where the bias responds to the height of the mass peak alone, in contrast to the usual definition (virial mass) that shows a strong dependence on additional halo properties such as formation time. The dependence on the age in the one-halo term is also remarkably reduced. The population of galaxies whose "peak height" changes with this new definition consists mainly of old stellar populations and are preferentially hosted by low-mass haloes located near more massive objects. The latter is in agreement with recent results which indicate that old, low-mass haloes would suffer truncation of mass accretion by nearby larger haloes or simply due to the high density of their surroundings, thus showing an assembly bias effect. The change in mass is small enough that the Sheth et al. (2001) mass function is still a good fit to the resulting distribution of new masses.