[Paper Review] Core-collapse supernovae in the hall of mirrors. A three-dimensional code-comparison project

Rubén M. Cabezón, K. Pan+6 more|arXiv (Cornell University)|Jun 24, 2018

Astrophysics and Cosmic PhenomenaPhysics and Astronomy119 references16 citations

TL;DR

This study performs the first direct 3D code comparison of core-collapse supernova simulations using four independent hydrodynamics codes (ELEPHANT, FLASH, fGR1, SPHYNX) with varying neutrino treatments (IDSA and M1), gravity models (Newtonian, effective GR, full GR), and physical approximations. It demonstrates good agreement (within ~10%) in post-bounce evolution across codes, confirms the effectiveness of parametrized deleptonization for neutrino-electron scattering, validates the use of effective GR potential, and shows SPH codes can match Eulerian codes in accuracy, while rotation enhances shock expansion by 26–70%.

ABSTRACT

Modeling core-collapse supernovae (CCSNe) with neutrino transport in three dimensions (3D) requires tremendous computing resources and some level of approximation. We present a first comparison study of CCSNe in 3D with different physics approximations and hydrodynamics codes. We aim to assess the impact of the hydrodynamics code, approximations for the neutrino and gravity treatments, and rotation on the simulation of CCSNe in 3D. We use four different hydrodynamics codes in this work (ELEPHANT, FLASH, fGR1, and SPHYNX) in combination with two different neutrino treatments, the isotropic diffusion source approximation (IDSA) and two-moment M1, and three different gravity treatments: Newtonian, 1D General Relativity (GR) correction, and full GR). Additional parameters discussed in this study are the inclusion of neutrino-electron scattering via a parametrized deleptonization (PD) and the influence of rotation. The four codes compared in this work include Eulerian and fully Lagrangian (smoothed particle hydrodynamics) codes for the first time. They show agreement in the overall evolution of the collapse phase and early post-bounce within the range of 10% (20% in some cases). The comparison of the different neutrino treatments highlights the need to further investigate the antineutrino luminosities in IDSA, which tend to be relatively high. We also demonstrate the requirement for a more detailed heavy-lepton neutrino leakage. When comparing with a full GR code, including an M1 transport method, we confirm the influence of neutrino-electron scattering during the collapse phase, which is adequately captured by the PD scheme. Also, the effective GR potential reproduces the overall dynamic evolution correctly in all Newtonian codes. Additionally, we verify that rotation aids the shock expansion and estimate the overall angular momentum losses for each code in rotating scenarios.

Motivation & Objective

• To assess the impact of hydrodynamics codes, neutrino treatments, gravity approximations, and rotation on 3D core-collapse supernova simulations.
• To evaluate the consistency and reliability of different numerical approaches in modeling complex neutrino-driven explosion mechanisms.
• To test the validity of approximate neutrino transport schemes (IDSA and M1) and gravity treatments (effective vs. full GR) against each other and against full GR simulations.
• To investigate the role of rotation and angular momentum conservation in 3D supernova dynamics.
• To establish a benchmark for future multi-code comparisons in 3D supernova simulations.

Proposed method

• Four independent 3D hydrodynamics codes (ELEPHANT, FLASH, fGR1, SPHYNX) were used, including both Eulerian and Lagrangian (SPH) approaches.
• Two neutrino transport schemes were employed: isotropic diffusion source approximation (IDSA) and two-moment M1 with variable Eddington factor closure.
• Three gravity treatments were compared: Newtonian, effective general relativistic (GR) potential, and full GR.
• A parametrized deleptonization scheme was implemented to model neutrino-electron scattering during collapse.
• Simulations included rotating progenitors with varying initial rotation profiles to assess angular momentum transport and shock expansion.
• Results were compared across codes in terms of PNS structure, shock radius evolution, neutrino luminosities, and angular momentum conservation.

Experimental results

Research questions

• RQ1 How consistent are the results of 3D core-collapse supernova simulations across four independent hydrodynamics codes with different numerical methods and implementations?
• RQ2 To what extent do different neutrino transport approximations (IDSA vs. M1) and gravity treatments (Newtonian, effective GR, full GR) affect the evolution of the proto-neutron star and shock radius?
• RQ3 Can a smoothed particle hydrodynamics (SPH) code produce results comparable to Eulerian codes in 3D core-collapse supernova simulations with neutrino transport?
• RQ4 How does rotation influence shock expansion and angular momentum transport in 3D supernova models?
• RQ5 Is the effective GR potential sufficient to capture key GR effects in 3D supernova simulations, or is full GR required for accurate dynamics?

Key findings

• The four codes (ELEPHANT, FLASH, fGR1, SPHYNX) show good agreement in the collapse and early post-bounce phases, with deviations of ~10% in key quantities like shock radius and PNS structure.
• The IDSA neutrino treatment tends to produce relatively high antineutrino luminosities, indicating a need for further investigation of this approximation.
• The parametrized deleptonization scheme effectively captures the influence of neutrino-electron scattering during the collapse phase, matching results from full GR simulations.
• The effective GR potential accurately reproduces the dynamic evolution seen in full GR simulations, validating its use in Newtonian codes.
• Rotation enhances shock expansion by 26–70% at 50 ms post-bounce, driven by m=1 spiral modes, despite a cooler and less dense PNS.
• Angular momentum conservation is maintained within 4% across all codes at the current resolution, with SPHYNX showing strong agreement with FLASH in key observables, especially with effective GR.