Surrogate modelling the Baryonic Universe II: On forward modelling the colours of individual and populations of galaxies

ABSTRACT Among the properties shaping light of a galaxy, star formation history (SFH) is one most challenging to model due variety correlated physical processes regulating formation. In this work, we leverage stellar population synthesis fsps, together with SFHs predicted by hydrodynamical simulation IllustrisTNG and empirical universemachine, study impact variability on galaxy colours. We start introducing model-independent metric quantify burstiness model, use demonstrate that universemachine predicts more relative IllustrisTNG. Using principal component analysis, construct families SFH models adjustable variability, show precision broad-band optical near-infrared colours degrades as level unresolved short-term increases. same technique in metallicity dust attenuation presents practically negligible variability. additionally provide fitting function capturing how translates into imprecision predictions for colours; our can be used determine minimal reproduces some target precision. Finally, modelling individual galaxies per cent-level demands resorting complex models, while producing precise populations achieved using just few degrees freedom.