[Mesa-users] entropy relaxation skipped

[Bradley Munson]

Akaash,

I have not played with those options much. In my inlists, they are:


    set_initial_dt = .true.

    years_for_initial_dt = 0.01


If you are having trouble with composition now, you may try changing the following option:


    timescale_for_relax_composition = 1d10


Changing the relaxation timescale tends to be a more robust fix in my experience. This is roughly the order of magnitude I have had to use when the composition does not converge during relaxation. I hope it helps!


Brad



________________________________
From: Akaash Conhye <ac01283 at surrey.ac.uk>
Sent: Thursday, December 17, 2020 9:37 AM
To: Bradley Munson <bmunso2 at lsu.edu>
Cc: mesa-users <mesa-users at lists.mesastar.org>
Subject: Re: [Mesa-users] entropy relaxation skipped

Hi Brad,

thanks for your email. I've tried your suggestion and the result is the same: entropy relaxation is skipped. There is something ive noticed, setting the line
set_initial_dt = .true.
years_for_initial_dt = 1d-20

or something else very small allows the entropy to relax without issue. However, it messes up the composition relaxation i.e. MESA runs into convergence problems where the min_timestep is reached, even when this value is set to 0 (instead, the error message is dt_is_zero). Letting MESA choose the initial time step allows for the composition to relax without issue, but the entropy relaxation is skipped. May I ask what initial dt you usually use?

Thanks,

Akaash
________________________________
From: Bradley Munson <bmunso2 at lsu.edu>
Sent: 16 December 2020 20:34
To: Conhye, Akaash (UG - Physics) <ac01283 at surrey.ac.uk>
Cc: mesa-users <mesa-users at lists.mesastar.org>
Subject: Re: [Mesa-users] entropy relaxation skipped

Hi Akaash,

It's hard to diagnose the problem when we do not have the necessary files to reproduce it. I think you are interpreting the output correctly (the entropy relaxation struggles for a bit and then ultimately gets "skipped" without taking a successful step).

Have you tried playing around with the following options? The defaults for both are 1d-9, but the following values typically work for me with similar looking entropy profiles:

timescale_for_relax_entropy = 1d-12
max_dt_for_relax_entropy = 5d-13

​Try adjusting those options and see if your model at least takes some steps in the right direction.

Also, what is the resolution of your input entropy file like? Could it be too sparse?

Best,

Brad
________________________________
From: Mesa-users <mesa-users-bounces at lists.mesastar.org> on behalf of Akaash Conhye via Mesa-users <mesa-users at lists.mesastar.org>
Sent: Wednesday, December 16, 2020 12:13 PM
To: Francis Timmes <fxt44 at mac.com>
Cc: mesa-users <mesa-users at lists.mesastar.org>
Subject: Re: [Mesa-users] entropy relaxation skipped

Hi Francis,

Removing this max_model_number line allows for proper time evolution to occur i.e. the white dwarf goes about its cooling. This line is something I found in the relax_composition_j_entropy template on the test suite. It is a means to stop the simulation right when relaxation finishes, at least as far as I know. The entropy relaxation process is just skipped over after a few retries and backups when the line is removed. This is precisely what I find quite strange.

Thanks,

Akaash
________________________________
From: Francis Timmes <fxt44 at mac.com>
Sent: 16 December 2020 14:01
To: Conhye, Akaash (UG - Physics) <ac01283 at surrey.ac.uk>
Cc: Frank Timmes <fxt44 at mac.com>; mesa-users <mesa-users at lists.mesastar.org>
Subject: Re: [Mesa-users] entropy relaxation skipped

the run is terminating as instructed:

> stop because model_number >= max_model_number


try setting max_model_number to a larger value in your inlist(s).

fxt




> On Dec 16, 2020, at 6:13 AM, Akaash Conhye via Mesa-users <mesa-users at lists.mesastar.org> wrote:
>
> Hi all,
>
> I'm trying to relax composition and entropy for a white dwarf model. The profiles come from hydrodynamic simulations with a Helmholtz EOS without coulomb corrections.
>
> composition relaxation occurs just fine. However, my problem is entropy relaxation. In the terminal output file I've attached, MESA tries a few times with entropy relaxation and gives up instead of the usual timescale reduction to the minimum timestep. Does anyone know why this might be occurring?
>
> Also attached is a comparison of the entropy profiles pre-relaxation (default) post relaxation (relax363) and the target profile (from hydro simulation)
>
> Thanks,
>
> Akaash
>
> <relax363_s_and_mu_no_initial_dt_max_dt_s_1d0_mlt_cox_mesh_coeff_0.5_timescale_default_eddington_t_tau_iterated.out><sprofile_default_relax363_hydro_loglog.png>_______________________________________________
> mesa-users at lists.mesastar.org
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.mesastar.org%2Fmailman%2Flistinfo%2Fmesa-users&data=04%7C01%7Cac01283%40surrey.ac.uk%7C726c5d348f2b4d3d31c208d8a1cb26c2%7C6b902693107440aa9e21d89446a2ebb5%7C0%7C0%7C637437241188199657%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=NfuXykXuLCuGFCq%2BEbFJT%2FCthWojEgoE3s8L12ThyJ0%3D&reserved=0<https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.mesastar.org%2Fmailman%2Flistinfo%2Fmesa-users&data=04%7C01%7Cbmunso2%40lsu.edu%7Ceb7f8d697dad4a23f09008d8a2a1b7f4%7C2d4dad3f50ae47d983a09ae2b1f466f8%7C0%7C0%7C637438162758469386%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=2D4Ld4DjzxLKHCpdvgvlTHBjIMCvchC8aS3YNCI7X8M%3D&reserved=0>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20201217/a79234b6/attachment.htm>