Cosmological Simulations

Wide-area surveys from LSST will critically test cosmological simulations which underpin our understanding of the physics of galaxy evolution. Over the last two decades, semi-analytical models have been successful in reproducing many of the bulk properties of galaxies over a significant fraction of cosmic time. By employing approximations derived from more detailed numerical simulations, and empirical calibrations from data, the semi-analytical approach has offered a computationally inexpensive route to probing the phenomenology of galaxy formation and, in particular, the theoretical analysis of today’s large survey data sets (e.g. Somerville et al. 2012).

A significant recent advance has been the advent of hydro-dynamical simulations in cosmological volumes (e.g. Horizon-AGN, EAGLE and Illustris), in which baryons and dark matter are evolved self-consistently. These models have opened up a powerful new set of tools with which to understand the physics of galaxy evolution and have found broad success in reproducing the properties of (mostly massive) galaxies that are detectable in current and past survey data (e.g. Kaviraj et al. 2017).

Nevertheless, both the mass and spatial resolutions of such simulations are not ideally suited to the LSB/dwarf regime, which hosts much of the vast untapped discovery space that will be unlocked by LSST. For example, the stellar mass and spatial resolutions of Horizon-AGN are 4 x 10⁶ MSun and 1 kpc respectively. Since 100 star particles are required for detecting structures, the effective galaxy mass resolution is ~10^8.5 MSun, making the dwarf regime mostly inaccessible. Furthermore, a 1 kpc resolution does not fully resolve the scale heights of disks even in massive galaxies (the Milky Way scale height, for comparison, is ~300 pc). Finally, while merger-induced LSB tidal features do appear around massive galaxies in these simulations, the mass resolution does not lend itself well to the detection of such features around lower mass systems.

A new generation of higher-resolution cosmological hydro-dynamical simulations offers an unprecedented opportunity, both to make statistical predictions in the LSB/dwarf regime and to better resolve the internal properties of massive galaxies. An example is the New Horizon simulation (Dubois et al. in preparation), which offers mass and a maximum spatial resolutions of 10⁴ MSun and 40 pc respectively, in a volume that has a radius of 10 Mpc (Fig 2). A significant achievement of these new models is the appearance of realistic thin disks, which have been difficult to produce in previous cosmological simulations due to their lack of resolution (e.g. Fig 3; Park et al 2019).

New Horizon, together with other high-resolution cosmological simulations like Illustris-TNG50 (Pillepich et al. 2019) and cosmological 'zoom-ins' (e.g. Hopkins et al. 2014; Wang et al. 2015), will provide valuable theoretical counterparts to the objects that will be routinely detected in surveys using LSST.

It should be noted, however, that the increase in resolution comes at the expense of the simulation box size (Fig 4). This makes it difficult to probe a large range in environments and means that these simulations do not contain rare objects, like the most massive galaxies. As a result, such high-resolution simulations will likely be used in conjunction with previous generations of hydro-dynamical simulations and also semi-analytical models, which lend themselves particularly well to the efficient, phenomenological exploration of the theoretical parameter space.