Progenitor stars calculated with small reaction networks should not be used as initial conditions for core collapse

Main take home point

If you need progenitors to study multi-dimensional neutrino magneto-hydrodynamics of stellar explosions, whether including rotation or not, do not waste your precious CPUh/GPUh on progenitors computed with approx nuclear reaction networks.

These are great for many things, but not to calculate the weak reactions that determine the fraction of free electrons in the core. They introduce a known systematic uncertainty comparable to unknown in many other physical ingredients, and thus muddy the waters and prevent learning robust conclusion from the explosion of progenitors computed with these.

How to reproduce

Chemically homogeneous stellar models with varying nuclear reaction networks

We computed the stellar models with MESA (version r24.03.1). The folder MESA_templates/ contains template work directories, you just need to edit new_net_name to pick the nuclear reaction network of choice and run the models. The exact template used for the submitted manuscript is also available on zenodo, the folder here may evolve later on.

Figure

From zenodo download the MESA results, stored in 40_rot0.6_large_net.xz for the 128-isotope model run with mesa_128.net and 40_rot0.6_small_net.tar.xz for the 22-isotope (approx21_plus_cr56.net) model.

Download also scripts.tar.xz from zenodo and unpack all in the same folder (or use the scripts/ folder here).

Run compare_two_models.py to produce the figure in the research note. It depends on functions defined in the other *.py files in scripts/. The figure in the manuscript was made with python 3.11.9 and the environment provided in scripts/.