[Mesa-users] Comparison of models between MESA v10398 and v11554

Tue Apr 23 18:24:13 EDT 2019

Yes, this looks more reasonable. Ignoring the difference from shifting
resolution you have a sub 1% spread in lifetimes. I would also think that
your 11554 results would line up much better between each other if you're
careful about how the initial rotation is set (see my previous email).

Cheers

On Tue, Apr 23, 2019 at 4:31 PM Ying Qin <Ying.Qin at unige.ch> wrote:

> Hi Pablo,
>
> Thank you very much for your time.
> In order to 100% make sure what I was doing correctly, I reinstalled
> v10398. The bigger discrepancy in my previous test was due to a slight
> modification I did a test myself before.
>
> Here I tried 6 models, and it shows how much both resolution and new
> rotation routine have an impact on the convective core and age. I can see
> for v11554 the resolution has a slightly larger
> impact on the results when compared to new rotation routine.  Also, it’s
> good to know the discrepancy between the two versions comes from the
> ongoing work of EOS.
>
> Thanks again,
> Ying
>
>
>
> On 23 Apr 2019, at 06:27, Pablo Marchant <pamarca at gmail.com> wrote:
>
> Hi Ying, Manos,
>
> there is currently ongoing work on the EOS, and my first thought was that
> this could be the source of your discrepancy.
>
> However, I was also worried that you might not be comparing quite the same
> stars. The reason for this is that the relaxation routine that sets the
> rotation does not specify unambiguously the starting point of the
> evolution. The star you're modeling undergoes a phase of thermal adjustment
> as it approaches CN equilibrium, during which the radius rapidly changes.
> Because of this, the total angular momentum of your star might differ
> slightly depending on things such as the size of the timesteps used during
> the relaxation routine. So doing the sort of comparisons you're trying to
> do here can be tricky if you don't ensure the rotation has a well-defined
> value at a clear point in the evolution. In particular, the
> high_rot_darkening test has a run_star_extras that ensures the star remains
> at a fixed omega/omega_crit until a certain amount of hydrogen has been
> burnt.
>
> In addition, the default resolution used by the test case is not that good
> in either space or time. So the differences you're checking in terms of
> core masses might correspond to not that many cells, and be essentially
> unresolved.
>
> So I went and adjusted the inlist to increase the resolution, and when
> using the new rotation routines I adjusted by hand the initial
> omega/omega_crit in the inlist such that omega at the surface matches after
> a bit of hydrogen is burnt in the core. I ran this model with three
> different MESA versions, 10398 using the old rotation routines, 11554 using
> the updated rotation routines and 11677 using both the updated and the old
> rotation routines. Purpose of testing 11677 is that it includes changes to
> the EOS that in principle should result in a better match with the results
> of 10398. Attached are the inlists for these tests, the ones marked with
> "newversion" where used for 11554 and 11677.
>
> Attached is a plot showing the evolution of the convective core wrt to
> time for these three cases. Did not include 11677 with the old rotation
> routines because it pretty much overlaps with 11677 with the new rotation
> routines and 10398 with the old rotation routines. What I see is that 11554
> has a small discrepancy from all the rest which I attribute to the EOS
> differences. However the discrepancy I find is much smaller than what you
> showed, and this could be a consequence of poor resolution in your runs.
>
> Can you please verify this is the case?
>
> Cheers
>
> On Mon, Apr 22, 2019 at 2:25 PM Ying Qin via Mesa-users <
> mesa-users at lists.mesastar.org> wrote:
>
> Dear MESA users,
>
> We were trying to compare non-rotating and rotating models between MESA
> v10398 and v11554. I describe below how we did this comparison and the
> result we got.
>
> We started with test_suite/high_rot_darkening and the inlist is attached
> for reproducing our test.
> For each MESA version, we tried two models, i.e., a non-rotating one and
> one with initial rotation rate of 0.5 w_crit. All 4 models were terminated
> at TAMS.
>
> In the attached plot (he_core_mass vs. star_age), in the MESA v10398 the
> star has a longer lifetime for the rotating model and its final helium core
> mass at TAMS is higher (Turning off fitted_fp_ft_i_rot does not change this
> trend). This is what I expected before this test due to rotational mixing.
> However, in the MESA v11554, we got the opposite results, with the helium
> core mass being even slightly lower for the rotating model. Also, the plot
> of mass_conv_core vs. star_age is attached. It shows for v11554 the
> convective core mass of rotating models was almost same with the one of
> non-rotating models. But this is also different from the results in v10398.
>
> I do not know if any users have already found similar results, or am I
> missing something ?
> I’d be very grateful if any user could give me some help.
>
> For v11554, the latest mesasdk was used, while I use MESA v10398 with
> mesasdk-x86_64-linux-20171120. The inlist I used is attached below.
>
> Best regards,
> Ying and Manos
>
>
> _______________________________________________
> mesa-users at lists.mesastar.org
> https://lists.mesastar.org/mailman/listinfo/mesa-users
>
>
>
> --
> Pablo Marchant Campos
> M.Sc on Astrophysics, Universidad Católica de Chile
> PhD on Astrophysics, Argelander-Institut für Astronomie, Universität Bonn
> <inlist_project_10398><inlist_project_newversion_newrot>
> <inlist_project_newversion_oldrot><mass_conv_core.png>
>
>
>

-- 
Pablo Marchant Campos
M.Sc on Astrophysics, Universidad Católica de Chile
PhD on Astrophysics, Argelander-Institut für Astronomie, Universität Bonn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20190423/e7432d5f/attachment.html>