Core-collapse supernova simulations in one and two dimensions: comparison of codes and approximations

We present spherically symmetric (1D) and axisymmetric (2D) supernova simulations for a convection-dominated 9 M⊙ and a 20 M⊙ progenitor that develops violent activity by the standing-accretion-shock instability (SASI). We compare in detail the Aenus–Alcar code, which uses fully multidimensional two-moment neutrino transport with an M1 closure, with a ray-by-ray-plus (RbR +) version of this code and with the Prometheus–Vertex code that employs RbR+ two-moment transport with a Boltzmann closure. Besides testing consequences of ignored non-radial neutrino-flux components in the RbR + approximation, we also discuss the influence of various transport ingredients applied or not applied in recent literature, namely simplified neutrino-pair processes, neutrino-electron scattering, velocity-dependent and gravitational-redshift terms, and strangeness and many-body corrections for neutrino-nucleon scattering. Alcar and Vertex show excellent agreement in 1D and 2D despite a slightly but systematically smaller radius (∼1 km) and stronger convection of the proto-neutron star with Alcar. As found previously, the RbR + approximation is conducive to explosions, but much less severely in the convection-dominated 9 M⊙ case than in the marginally exploding 20 M⊙ model, where the onset time of explosion also exhibits big stochastic variations, and the RbR + approximation has no distinctly stronger supportive effect than simplified pair processes or strangeness and many-body corrections. Neglecting neutrino-electron scattering has clearly unfavourable effects for explosions, while ignoring velocity and gravitational-redshift effects can both promote and delay the explosion. The ratio of advection time-scale to neutrino-heating time-scale in 1D simulations is a sensitive indicator of the influence of physics ingredients on explosions also in multidimensional simulations.