ABSTRACT We investigate the conditional colour–magnitude distribution (CCMD), namely the colour–magnitude distribution at fixed halo mass, of the central galaxies in semi-analytical galaxy formation model (SAM) and hydrodynamic simulations. We analyse the CCMD of central galaxies in each halo mass bin with the Gaussian mixture model and find that it can be decomposed into red and blue components nearly orthogonal to each other, a red component narrow in colour and extended in magnitude and a blue component narrow in magnitude and extended in colour. We focus on the SAM galaxies to explore the origin of the CCMD components by studying the relation between central galaxy colour and halo or galaxy properties. Central galaxy colour is correlated with halo assembly properties for low-mass haloes and independent of them for high-mass haloes. Galaxy properties such as central supermassive black hole mass, cold gas mass, and gas specific angular momentum can all impact central galaxy colour. These results are corroborated by an alternative machine learning analysis in which we attempt to predict central galaxy colour with halo and galaxy properties. We find that the prediction for colours of central galaxies can be significantly improved using both halo and galaxy properties as input compared to using halo properties alone. With the halo and galaxy properties considered here, we find that subtle discrepancies remain between predicted and original colour distribution for low-mass haloes and that no significant determining properties are identified in massive haloes, suggesting modulations by additional stochastic processes in galaxy formation.
Read full abstract