Abstract

We investigate whether the galaxy and star formation model used for the Auriga simulations can produce a realistic globular cluster (GC) population. We compare statistics of GC candidate star particles in the Auriga haloes with catalogues of the Milky Way (MW) and Andromeda (M31) GC populations. We find that the Auriga simulations do produce sufficient stellar mass for GC candidates at radii and metallicities that are typical for the MW GC system (GCS). We also find varying mass ratios of the simulated GC candidates relative to the observed mass in the MW and M31 GCSs for different bins of galactocentric radius metallicity (r(gal)-[Fe/H]). Overall, the Auriga simulations produce GC candidates with higher metallicities than the MW and M31 GCS and they are found at larger radii than observed. The Auriga simulations would require bound cluster formation efficiencies higher than 10 per cent for the metal-poor GC candidates, and those within the Solar radius should experience negligible destruction rates to be consistent with observations. GC candidates in the outer halo, on the other hand, should either have low formation efficiencies, or experience high mass-loss for the Auriga simulations to produce a GCS that is consistent with that of the MW or M31. Finally, the scatter in the metallicity as well as in the radial distribution between different Auriga runs is considerably smaller than the differences between that of the MW and M31 GCSs. The Auriga model is unlikely to give rise to a GCS that can be consistent with both galaxies.