Simulations of the galaxy population constrained by observations from z = 3 to the present day: implications for galactic winds and the fate of their ejecta

Abstract

We apply Monte Carlo Markov Chain (MCMC) methods to large-scale simulations\nof galaxy formation in a LambdaCDM cosmology in order to explore how star\nformation and feedback are constrained by the observed luminosity and stellar\nmass functions of galaxies. We build models jointly on the Millennium and\nMillennium-II simulations, applying fast sampling techniques which allow\nobserved galaxy abundances over the ranges 7<log(M*/Msun)<12 and z=0 to z=3 to\nbe used simultaneously as constraints in the MCMC analysis. When z=0\nconstraints alone are imposed, we reproduce the results of previous modelling\nby Guo et al. (2012), but no single set of parameters can reproduce observed\ngalaxy abundances at all redshifts simultaneously, reflecting the fact that\nlow-mass galaxies form too early and thus are overabundant at high redshift in\nthis model. The data require the efficiency with which galactic wind ejecta are\nreaccreted to vary with redshift and halo mass quite differently than\npreviously assumed, but in a similar way as in some recent hydrodynamic\nsimulations of galaxy formation. We propose a specific model in which\nreincorporation timescales vary inversely with halo mass and are independent of\nredshift. This produces an evolving galaxy population which fits observed\nabundances as a function of stellar mass, B- and K-band luminosity at all\nredshifts simultaneously. It also produces a significant improvement in two\nother areas where previous models were deficient. It leads to present day dwarf\ngalaxy populations which are younger, bluer, more strongly star-forming and\nmore weakly clustered on small scales than before, although the passive\nfraction of faint dwarfs remains too high.\n