|Department:||The Faculty of Physics and Astronomy|
|Full text PDF:||http://archiv.ub.uni-heidelberg.de/volltextserver/15897/|
In this thesis, I present my PhD work: a multi-phase chemodynamic galaxy formation and evolution model. The model is aimed at treating the dynamics of stars, molecular (cold) clouds, and hot/warm diffuse gas individually and allowing for mass, momentum, and energy exchange between them in a self-consistent way, so as to overcome the difficulties of a single-phase description. I introduce the detailed implementation of physical processes in the model including gravity, gas dynamics, heat conduction, cooling, star formation and stellar feedback. A dwarf galaxy model is evolved for 1 Gyr. The corresponding star formation rate decreases from 1 Msol/Year to 0.1 Msol/Year. The cloud mass distribution follows a power law with a slope of -2.3. The discrepancies of chemical abundance between hot/warm and cold phase are reproduced. As an extension to the classical multi-phase model, I introduce a transition process such that hot/warm gas can collapse to cold clouds, which solves the problem of cold clouds' initial mass fraction and distribution in the multi-phase simulation. This process is proven to be more suitable for low mass systems. Also I implement an individual star formation model, in which individual stars are created analytically inside a molecular cloud with a stellar mass distribution given by a specific initial mass function (IMF). This model reproduces the life cycle of interstellar medium in a galactic scale simulation and realizes the process of star cluster formation inside one molecular cloud. The multi-phase code is parallelized with MPI and shows good scaling relations. GPUs are used to accelerate the most time consuming parts (gravity, SPH and neighbour search), which results in a speedup of one oder of magnitude for the whole program.