SearcharxivSearch

arXiv subjects

FJ Summers

Publications and source records attributed to FJ Summers.

2 recordsLinked to original sources

GALAXY DYNAMICS IN CLUSTERS

We use high resolution simulations to study the formation and distribution of galaxies within a cluster which forms hierarchically. We follow both dark matter and baryonic gas which is subject to thermal pressure, shocks and radiative cooling. Galaxy formation is identified with the dissipative collapse of the gas into cold, compact knots. We examine two extreme representations of galaxies during subsequent cluster evolution --- one purely gaseous and the other purely stellar. The results are quite sensitive to this choice. Gas-galaxies merge efficiently with a dominant central object while star-galaxies merge less frequently. Thus, simulations in which galaxies remain gaseous appear to suffer an ``overmerging'' problem, but this problem is much less severe if the gas is allowed to turn into stars. We compare the kinematics of the galaxy population in these two representations to that of dark halos and of the underlying dark matter distribution. Galaxies in the stellar representation are positively biased (\ie over-represented in the cluster) both by number and by mass fraction. Both representations predict the galaxies to be more centrally concentrated than the dark matter, whereas the dark halo population is more extended. A modest velocity bias also exists in both representations, with the largest effect, $σ_{gal}/σ_{DM} \simeq 0.7$, found for the more massive star-galaxies. Phase diagrams show that the galaxy population has a substantial net inflow in the gas representation, while in the stellar case it is roughly in hydrostatic equilibrium. Virial mass estimators can underestimate the true cluster mass by up to a factor of 5. The discrepancy is largest if only the most massive galaxies are used, reflecting significant mass segregation.

astro-ph

Cosmological Simulations Using Special Purpose Computers: Implementing P3M on Grape

An adaptation of the Particle-Particle/Particle-Mesh (P3M) code to the special purpose hardware GRAPE is presented. The short range force is calculated by a four chip GRAPE-3A board, while the rest of the calculation is performed on a Sun Sparc 10/51 workstation. The limited precision of the GRAPE hardware and algorithm constraints introduce stochastic errors of the order of a few percent in the gravitational forces. Tests of this new P3MG3A code show that it is a robust tool for cosmological simulations. The code currently achieves a peak efficiency of one third the speed of the vectorized P3M code on a Cray C-90 and significant improvements are planned in the near future. Special purpose computers like GRAPE are therefore an attractive alternative to supercomputers for numerical cosmology.

astro-ph