A HYDRODYNAMIC APPROACH TO COSMOLOGY - METHODOLOGY

An accurate and efficient hydrodynamic code for evolving self-gravitating cosmological systems is described. The code may be coupled with a standard particle code, e.g., a particle-mesh code, if point-mass collisionless components are needed in the model. The hydrodynamic code is a flux-based mesh code originally designed for engineering hydrodynamical applications. A variety of checks have been made which indicate that the resolution of the code is a few cells, giving accuracy for integral energy quantities in our simulations of 1%-3% over the whole runs. The advantages of using a true hydrodynamic code over an N-body code are manifold, since the observed universe is baryonic, and its early evolution should be treated via the classic Euler equations of hydrodynamics. In addition, the distribution of baryonic matter in temperature and density and the background radiation fields can be directly obtained, enabling comparison between these distributions and astronomical observations. A standard particle-mesh code is also described, for the reason that we will use it when pointlike components are introduced. We track separately six species (e-, H I, H II, He I, He II, He III), computing in detail (non-LTE) relevant ionization and recombination processes, as well as line and continuum heating and cooling. The background radiation field is simultaneously determined in the range 1 eV to 100 keV allowing, in a spatially averaged fashion, for absorption, emission, and cosmological effects. We show how the inevitable numerical inaccuracies can be estimated and to some extent overcome. For example, the mean Zel'dovich-Sunyaev y-parameter is calculated to be 6.0 x 10(-6) in a simulation with 128(3) dark matter particles and 128(3) cells, a maximum scale size of L = 64 h-1 Mpc, and a minimum wavelength of lambda(min) = 1 h-1 Mpc, with h = 0.5 and b = 1.0. By comparing the results of runs with different values of (N, L(max), L(min)), we extrapolate to y = (1.25 +/- 0.65) x 10(-5) as (N-1, L(max)-1, L(min)) --> 0.