Contributed Talk - Splinter Computation

Friday, 25 September 2020, 10:20   (virtual room B)

On the importance of the hydrodynamic method in galaxy simulations with resolved feedback

Ulrich P. Steinwandel, Thorsten Naab, Klaus Dolag and Benjamin Moster

Recent simulations of galaxy formation and evolution have reached the resolution where it becomes possible to resolve stellar feedback on the sub-grid scale, removing the constraint enforced by sub-grid models for Supernova-feedback (e.g. terminal momentum injection) by directly resolving the Sedov-Taylor phase of individual SN-remnants (Hu et al. 2016, Steinwandel et al. 2020b). In combination with models for non-equilibrium cooling and chemistry, photo-electric heating, and photo-ionizing radiation this enables to study stellar-feedback driven outflows in a resolved fashion. In this study we will investigate the dependence of the physical result by changing the underlying method for solving the equations of hydrodynamics. Therefore, we will compare the results obtained with modern pressure-energy SPH, that is numerically stabilized via time-dependent artificial diffusion terms and the Riemann based second order method meshless finite mass (MFM, see Hopkins 2015) that are both implemented in our code Gadget-3. We will present the implementation strategies for both, our hydrodynamic solvers and the advanced stellar feedback routines that include the sampling of single stars, photo-electric heating, and photo ionization. We test the impact of the hydrodynamic method in resolved simulations of isolated dwarf-galaxies and galaxy merger simulations. In both cases we find variations not only in the structure of the hot and cold phases of the ISM but also in the star formation histories and the chemical abundances in the ISM. Our results indicate that the new MFM-method shows significant improvement in capturing the multi-phase structure of the ISM compared to SPH in the current implementation in our code Gadget-3. Furthermore, the MFM-method increases the overall runtime of a resolved high resolution simulation by roughly a factor of 2 compared to SPH.