Re: [mesa-users] Derivatives of eps_nuc in massive evolved models

[Alfred G. <al...@ga...>]

Ehsan

as far as I can see the partials eps_T and eps_Rho look fine to me.
Notice that eps_T and eps_Rho are not zero as long as the coder decided
nuclear reacations are to be computed. Even though the magnitude of
epsilon_nuc itself will be essentially zero, the partials are not. 

Since in all applications I came across, always the combination of
eps_nuc * eps_T or eps_nuc * eps_Rho occurred, these composed functions
are always very smooth and they are restricted to the nuclear burning
regions of a star. 

It seems that the Mesa coders decided to compute nuclear energy
generation to quite low temperatures (Bill: do you have any numbers
ready?)

Along the pragmatic and heuristic avenue: As star models react very
sensitively to eps_T and eps_nuc during the Henyey iterations, and since
the tests of Mesa apparently run very reliable and compare well with the
literature, I guess that the partials are computed correctly; at least
consistent with eps itself. 

I doubt that there is anything helpful in the literature regarding eps_T
and eps_Rho; it's all auxiliary machinery no referee (and probably
reader) wants to see in the end.   

Regards, Alfred