[Mesa-users] pre-main sequence boundary conditions

[Thomas Steindl]

There is some explanation in num/public/num_def.f:

! if max_correction > scale_max_correction, then
! rescale the correction vector by scale_max_correction/max_correction
! the default value (for work(r_scale_max_correction)=0) is 1d99

so it basically limits the size of the maximal correction for an  
iteration in the newton solver. We can run your initial inlist with
report_hydro_solver_progress = .true.
in this way we see a lot of information about the newton iteration.
The output shows that the newton iteration fails because it cannot  
adjust the corrections:

adjust_correction failed in eval_equations -- give up

This results in retries until the timestep is below the min_timestep_limit.

So, my guess is that the corrections are to big, therefore  
overshooting the 'correct solution'. Using scale_max_correction, the  
corrections are rescaled. With the scaled correction adjust_correction  
does not fail and the iterations converge.

I hope this helps.

Thomas

Zitat von "TUCHOW, NOAH WOLFE" <nxt5109 at psu.edu>:

> Thanks, changing the scale_max_correction did seem to fix the  
> problem. Do you know what exactly scale_max_correction does? It  
> wasn't very clear in the documentation or in  
> 'star/private/star_newton'.
>
> Thanks again,
> Noah Tuchow
>
> Hello Noah,
> You can add
> scale_max_correction = 0.1
> to your controls inlist. This helps the Eddington_Grey atmosphere to  
> converge.
> Thomas
> University Innsbruck
>
> Noah Tuchow
> Penn State University
> Department of Astronomy and Astrophysics
> 532C Davey Lab
> University Park, PA 16802
>
>
> On Mon, Oct 7, 2019 at 5:02 PM Noah Tuchow  
> <nxt5109 at psu.edu<mailto:nxt5109 at psu.edu>> wrote:
> Hello everyone,
>
> I have a question about atmospheric boundary conditions for pre-main  
> sequence models. I want to initialize a pre-main sequence model  
> (setting  create_pre_main_sequence_model = .true.), save a model  
> file after Deuterium depletion, and then load this model file for  
> the rest of the star's evolution. Previously, I had been using the  
> 'simple_photosphere'  boundary condition until the deuterium  
> birthline and then changing the atmosphere to Eddington grey after  
> loading the model file. However, looking at the evolutionary tracks  
> on the HR diagram, I can see that changing the boundary condition  
> upon loading a model causes a pretty large discontinuity in T and L  
> (see attached image where I load a model file made with the  
> simple_photosphere boundary condition and change to Eddington_grey).  
> I've noticed that this discontinuity disappears if I keep the same  
> boundary condition for both the initialization and the evolution  
> inlists.
>
> What I would like to do is have the same boundary condition to  
> initialize the model and evolve it through the pre-main sequence and  
> main sequence, so that I can get rid of the discontinuity upon  
> loading. However, I'm having a problem in that I can only get the  
> 'simple_photosphere' and 'photosphere_tables' boundary conditions  
> working for the initialization prior to the deuterium birthline,  
> whereas I would like to get a Grey atmosphere working, such as  
> 'Eddington_grey'. When I try to use the Eddington grey boundary  
> condition in my birthline inlist, MESA exits during the initial  
> relaxation, saying "stopping because of problems dt <  
> min_timestep_limit". I was wondering if there was anything that I  
> could do to get a grey atmosphere working for my birthline model.  
> Are there any tweaks to my inlists which I could do to prevent MESA  
> from exiting during the initial relaxation? Alternatively, are there  
> any ways to reduce the discontinuity upon loading a model with a  
> different boundary condition?
>
> I've attached my inlist_birthline which I've been using with MESA r11701.
>
> Thanks for your assistance,
> Noah Tuchow
>
>
> Noah Tuchow
> Penn State University
> Department of Astronomy and Astrophysics
> 532C Davey Lab
> University Park, PA 16802
>