[Mesa-users] Question about controls in intermediate-mass stars evolution

[Francisco De Geronimo]

Hi Josiah,
thanks for your email.
Now I understand what they mean.  I think i will be not considering them,
its not useful for
what im computing.

I was doing some convergence test, varying these parameters:

      mesh_delta_coeff
      timestep_dt_factor
between 0-1.

If  one of the runs gives different results, for example when smaller steps
on timestep and grid are adopted,
what are the steps to follow?
I read
https://docs.mesastar.org/en/latest/using_mesa/best_practices.html?highlight=convergence#explore-numerical-convergence
but says nothing about when runs gives different results.

For the next simulations, It is safer start with a work/ directory than a
test_suite?

Best wishes,

Francisco


El mar, 17 ago 2021 a las 17:48, Josiah Schwab (<jwschwab at ucsc.edu>)
escribió:

> Hi Francisco,
>
> In this test suite there are two commands included in the test suite for
>> which i didnt find information in the web page:
>>
>> min_cell_energy_fraction_for_dedt_form
>> use_correction_for_KE_plus_PE_with_eps_grav_in_energy_eqn
>>
>> The chemical structure at the end of the C-burning seems to be affected
>> whether these commands are considered or not, and Im not sure what they do.
>> Are these just simple numerical improvements in the equations ?
>> Are they worth considering in v12778?
>>
>
> I disliked both of these options once I learned about
> their existence/implementation and so have removed them in
> subsequent versions. If I were forced to set them I would probably choose
> min_cell_energy_fraction_for_dedt_form = 0 and
> use_correction_for_KE_plus_PE_with_eps_grav_in_energy_eqn = .false. .
>
> The option min_cell_energy_fraction_for_dedt_form controlled whether there
> was per-cell switching between the different forms of energy equation
> ("dedt" or "eps_grav") .  I never really understood the motivation for this
> option and because it was framed in terms of a per-cell fraction of the
> total internal energy, it introduces a dependence of the equations on cell
> size.
>
> The option use_correction_for_KE_plus_PE_with_eps_grav_in_energy_eqn added
> in a correction term when using the eps_grav form that effectively
> amounts to accounting for heating from numerical viscosity.  In a converged
> model, this should approach zero, so I believe that it should not have any
> significant effect.
>
> More specifically, looking at what's in the 7M_prems_to_AGB test case in
> r12778
>
>       use_dedt_form_of_energy_eqn = .true.
>       min_cell_energy_fraction_for_dedt_form = 0
>       use_correction_for_KE_plus_PE_with_eps_grav_in_energy_eqn = .true.
>
> and since 'min_cell...' is 0, MESA should use the 'dedt' form in all
> cells, unless some other switch condition is active, meaning the setting of
> the 'use_correction_...' option shouldn't matter because no cell is using
> the 'eps_grav' form.
>
> If you're seeing significant differences in your models, I would suggest
> confirming that your results for a given run are insensitive to variations
> of the space and time resolution.  Note that the test suite cases are often
> not converged, either due to neglect or in an attempt to reduce their
> runtime.
>
> Hope that helps,
> Josiah
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20210819/b3414208/attachment.htm>