| The goal of this work is to show how large cosmological N-body simulations are computed, how structures within such simulations can be identified with observed galaxies, and how the weak gravitational lensing effect of large scale structure introduces changes in the observed brightness of high redshift sources. We describe PKDGRAV, a fully parallel N-body code that can both spatially adapt to large ranges in particle densities, and temporally adapt to large ranges in dynamical timescales. Care has been taken to insure that the code runs efficiently on a variety of parallel architectures. This has led us to using a non-standard data structure for efficiently calculating the gravitational forces, a variant on the k-D tree, and a new method for treating periodic boundary conditions. A major problem with the interpretation of cosmological N-body simulations has always been identifying the positions and velocities of the "galaxies" in simulations. Observational constraints almost always concern the position and velocities of galaxies, not of the dark matter background. We present our grouping algorithm, SKID, which can identify galaxy halos independent of the environment in which the halo is found, a property that is very important given the large dynamic range in background densities present in cosmological simulations. PKDGRAV has been used to perform many very high resolution cosmological N-body simulations. We give an overview of some of the scientific program that has been enabled by this code. We present the magnification distributions due to weak lensing through all the intervening large scale structure from an observer to a source at z = 1. We show that the distribution is a highly skewed one with a shifted mode implying a slight demagnification bias and an extended tail toward high magnification. We investigate the implications of these results on the observations of high redshift type-Ia supernovae and show how these can be used to provide a new cosmological test. |
