Some observational tests of a minimal galaxy formation model

TL;DR

This paper tests a minimal galaxy formation model based on dark matter simulations, comparing its predictions with observations across redshifts 0-2, and discusses its limitations and potential improvements.

Contribution

It implements and evaluates a simple galaxy formation model using dark matter simulations, highlighting its discrepancies with observations and suggesting avenues for refinement.

Findings

Model shows reasonable galaxy distribution predictions but varies with simulation parameters.

Including redshift dependence may improve agreement with observations.

The model exhibits more scatter in galaxy histories than more complex semi-analytic models.

Abstract

Dark matter simulations can serve as a basis for creating galaxy histories via the galaxy-dark matter connection. Here, one such model by Becker (2015) is implemented with several variations on three different dark matter simulations. Stellar mass and star formation rates are assigned to all simulation subhalos at all times, using subhalo mass gain to determine stellar mass gain. The observational properties of the resulting galaxy distributions are compared to each other and observations for a range of redshifts from 0-2. Although many of the galaxy distributions seem reasonable, there are noticeable differences as simulations, subhalo mass gain definitions, or subhalo mass definitions are altered, suggesting that the model should change as these properties are varied. Agreement with observations may improve by including redshift dependence in the added-by-hand random contribution to star formation rate. There appears to be an excess of faint quiescent galaxies as well (perhaps due in part to differing definitions of quiescence). The ensemble of galaxy formation histories for these models tend to have more scatter around their average histories (for a fixed final stellar mass) than the two more predictive and elaborate semi-analytic models of Guo et al (2013) and Henriques et al (2015), and require more basis fluctuations (using PCA) to capture 90 percent of the scatter around their average histories. The codes to plot model predictions (in some cases alongside observational data) are publicly available to test other mock catalogues at https://github.com/jdcphysics/validation . Information on how to use these codes is in the appendix.