A physical model for cosmological simulations of galaxy formation

Mark Vogelsberger, Shy Genel, Debora Sijacki, Paul Torrey, Volker Springel, and Lars Hernquist


We present a new comprehensive model of the physics of galaxy formation designed for large-scale hydrodynamical simulations of structure formation using the moving mesh code AREPO. We describe first results from cosmological simulations with our new implementation, highlighting the impact of different feedback mechanisms on the simulated galaxy population. Our model includes primordial and metal line cooling with self-shielding corrections, stellar evolution and feedback processes, gas recycling, chemical enrichment, a novel subgrid model for the metal loading of outflows, black hole (BH) seeding, BH growth and merging procedures, quasar- and radio-mode feedback, and a prescription for radiative electro-magnetic (EM) feedback from active galactic nuclei (AGN). Our stellar evolution and chemical enrichment scheme follows nine elements (H, He, C, N, O, Ne, Mg, Si, Fe) independently, and tracks the overall metallicity and the total mass return from stars to gas. Stellar feedback is realised through kinetic outflows, driven either directly from interstellar medium gas and tied to the star formation rate or generated locally around evolving stellar populations. The scaling of the mass loading of galactic winds can be phenomenologically set to be either energy or momentum driven, or a mixture of both. The metal mass loading of outflows can be adjusted independently of the wind mass loading. This is required to simultaneously reproduce the stellar mass content of low mass haloes and their gas oxygen abundances. Radiative EM AGN feedback is implemented assuming an average spectral energy distribution and a luminosity-dependent scaling of obscuration effects. This form of feedback suppresses star formation more efficiently than continuous thermal quasar-mode feedback alone, but is less efficient than mechanical radio-mode feedback in regulating star formation in massive haloes. We have also implemented inlined analysis techniques for gas tracking based on velocity field or Monte Carlo tracer particles and on-the-fly visualisation routines for direct volume rendering with a ray casting technique. We contrast simulation predictions for different variants of our galaxy formation model with key observations, allowing us to constrain the importance of different modes of feedback and their uncertain efficiency parameters. We identify a fiducial best match model and show that it reproduces, among other things, the cosmic star formation history, the stellar mass function, the stellar mass - halo mass relation, g-, r-, i-, z-band SDSS galaxy luminosity functions, and the Tully-Fisher relation. We can achieve this success only if we invoke very strong forms of stellar and AGN feedback such that star formation is adequately reduced in both low and high mass systems. In particular, the strength of radio-mode feedback needs to be increased significantly compared to previous studies to suppress efficient cooling in massive, metal-enriched haloes.

Contact: Mark Vogelsberger


Rotation around a massive halo at z=0
(movie format: 1024x768 (MPEG-4))


Figure 2: redshift evolution Figure 3: feedback effect


Figure 4: rendered gas temperature field Figure 5: rendered gas metallicity field