[mesa-users] Variations with spatial resolution of massive stars core structures

[Mathieu R. <m....@uv...>]

Dear all,

I would like to share the outcome of one of the many resolution tests I 
have been performing lately, to show what can go wrong and encourage 
everybody to spend some time doing these (often painful) exercises.

The attached plot shows the post-oxygen depletion time evolution of the 
compactness parameter,

$$ \xi_{M=2.5} \propto \frac{M}{R(M)}$$ ,

for an initially 25$M_\odot$ model, at solar metallicity, with mass loss 
but no rotation. All the curves start from the same exact model and 
differ only in the adopted mesh_delta_coeff and 
mesh_delta_coeff_for_highT ($\Delta$ and $\Delta_\mathrm{high \ T}$ in 
the legend). The end point is either the onset of core collapse, or in 
some cases (e.g. thick gray curve), failure to find an acceptable 
solution for the stellar structure equations.

I ran these models as a self-consistency test on the spatial meshing, 
using an approx21 nuclear reaction network. This network pre-determines 
the value of $Y_e$, therefore the effective Chandrasekhar mass of the 
stellar core and ultimately the core structure and compactness, so do 
not take the values on the y axis as realistic.

*The key point is that there is a bifurcation point in the evolution, 
and two different classes of solutions appear. Which class of solutions 
MESA finds depends on the spatial resolution, and increasing it, MESA 
can jump back and forth between the 2 classes.* I did not investigate 
further what physically causes this bifurcation, as it was not relevant 
for my scientific question.

This kind of behavior might be an issue also in your research, so beware 
of it!

Best regards,

Mathieu


-- 
Mathieu Renzo <https://staff.fnwi.uva.nl/m.renzo/>
PhD student/KITP grad fellow
Anton Pannekoek Institute,
University of Amsterdam

Contacts:

Address: PO Box 94249, 1090 GE Amsterdam
Office: C4.128a
Telephone: (+31) 020-5258352
Skype name: m4thr3n
Mail: m....@uv... <mailto:m....@uv...>