SEAGLE III: Towards resolving the mismatch in the dark-matter fraction in early-type galaxies between simulations and observations

The central dark-matter fraction of galaxies is sensitive to feedback processes during galaxy formation. Strong gravitational lensing has been effective in the precise measurement of the dark-matter fraction inside massive early-type galaxies. Here, we compare the projected dark-matter fraction of early-type galaxies inferred from the SLACS (Sloan Lens ACS Survey) strong-lens survey with those obtained from the Evolution and Assembly of GaLaxies and their Environment (EAGLE), Illustris, and IllustrisTNG hydrodynamical simulations. Previous comparisons with some simulations revealed a large discrepancy, with considerably higher inferred dark-matter fractions -by factors of approximate to 2-3- inside half of the effective radius in observed strong-lens galaxies as compared to simulated galaxies. Here, we report good agreement between EAGLE and SLACS for the dark-matter fractions inside both half of the effective radius and the effective radius as a function of the galaxy's stellar mass, effective radius, and total mass-density slope. However, for IllustrisTNG and Illustris, the dark-matter fractions are lower than observed. This work consistently assumes a Chabrier initial mass function (IMF), which suggests that a different IMF (although not excluded) is not necessary to resolve this mismatch. The differences in the stellar feedback model between EAGLE and Illustris and IllustrisTNG are likely the dominant cause of the difference in their dark-matter fraction and density slope.

Automated summary

The central dark-matter fraction of galaxies, influenced by feedback processes during formation, has been accurately measured using strong gravitational lensing techniques. Comparisons of observational data with simulations show good agreement between the EAGLE and SLACS results, while results from IllustrisTNG and Illustris simulations are lower than observed values. Differences in the stellar feedback model are likely the main cause of this discrepancy.