Modelling galaxies in the high redshift universe

Abstract

This thesis describes the combination of a phenomenological model of galaxy formation with the spectrophotometric dust extinction model of Silva et al.(1998). I use TV-body simulations to predict the underlying dark matter den­sity field in a hierarchal Lambda Cold Dark Matter universe. Analytic pre­scriptions of gas cooling, star formation and stellar feedback describes the formation and evolution of galaxies embedded in the hierarchy of dark matter haloes. Starlight is attenuated by a two-phase medium of graphite and sili­cate dust grains, yielding the observed spectral energy distribution of a galaxy from the UV to the submillimetre. With this in place, the role of dust in galaxy formation can be probed and assessed in a self-consistent manner, improving on the common plane-parallel slab and tunable dust temperature approach employed in many semi-analytic models.

I first present model predictions of the local universe, demanding that an acceptable model must first perform well at z —0 before being tested at high redshift. I can adequately reproduce many key local observational con­straints, although a few stimulating discrepancies are present. Not only will this ground the models in reality, it will then help assess whether a ‘stan­dard* semi-analytic model can reproduce facets of the high redshift universe without recourse to new physics.

I then use the above framework to present predictions at early times, where dust is known to play an important role in influencing the observed magni­tudes, colours and properties of galaxies. Two further models are introduced to assess how different parameter choices that give similar local results di­verge at early times. I first concentrate on the more global, well understood facets of the high redshift universe, namely the galaxy and stellar luminosity functions. I then construct mock surveys in an attempt to account for Lyman Break Galaxies (LBGs) at 2 = 3, of which there is a comprehensive array of high quality data available in the literature. Properties of LBGs can be ade­quately reproduced by the models, although there is a marked overabundance of such objects present, likely due to an underestimate of the dust extinction inherent in such objects.

I finally use the mock survey approach to study Extremely Red Objects (EROs), galaxies with (R — K ) > 5 uncovered in deep K-band surveys, and Submillimetre Galaxies (SMGs) detected in blank field SCUBA surveys. These extreme objects tend to the bane of semi-analytic models, and I find that my model fares no different in this respect: a truncation method used to prevent the formation of overly massive galaxies at low redshift also inhibits their formation at early times. I severely underpredict the number counts of EROs and SMGs, although a few of their observational properties can be matched reasonably well. I then investigate the overlaps between distinct populations, and the fate of high redshift objects is followed through time, shedding light on relationships between evolutionary phases of galaxies.