Peter S. Behroozi, Risa H. Wechsler, Hao-Yi Wu, Michael T. Busha, Anatoly A. Klypin, Joel R. Primack
We present a new algorithm for generating merger trees and halo catalogs
which explicitly ensures consistency of halo properties (mass, position,
velocity, radius) across timesteps. Our algorithm has demonstrated the ability
to improve both the completeness (through detecting and inserting otherwise
missing halos) and purity (through detecting and removing spurious objects) of
both merger trees and halo catalogs. In addition, our method is able to
robustly measure the self-consistency of halo finders; it is the first to
directly measure the uncertainties in halo positions, halo velocities, and the
halo mass function for a given halo finder based on consistency between
snapshots in cosmological simulations. We use this algorithm to generate merger
trees for two large simulations (Bolshoi and Consuelo) and evaluate two halo
finders (ROCKSTAR and BDM). We find that both the ROCKSTAR and BDM halo finders
track halos extremely well; in both, the number of halos which do not have
physically consistent progenitors is at the 1-2% level across all halo masses.
Our code is publicly available at http://code.google.com/p/consistent-trees.
Our trees and catalogs will made publicly available once the referee process is
complete.
View original:
http://arxiv.org/abs/1110.4370
No comments:
Post a Comment