[Mesa-users] Using make_o_ne_wd to prepare some white dwarf models

Mon Jul 23 09:35:48 EDT 2018

Greetings Yi,

For novae, I don’t think C12+C12 will play much of a role. It’s typically important for carbon burning in massive stares, type Ia supernovae, and superbursts in the oceans of neutron stars.  As such, I would be surprised if your results depend on which 12C+12C rate you choose. Of course you could always run test calculations to be sure.

In any case, the Gasques et al. PRC 2005 paper aimed to provide an improved 12C+12C reaction rate for the pycnonuclear fusion regime (density-driven fusion which happens at relatively high densities and low temperatures). I don’t think this regime is relevant for what you’re simulating.  In the thermonuclear regime, which I believe is what’s relevant for you, the Gasques rate and the CF88 rate aren’t very different (see e.g. Figure 11 of http://adsabs.harvard.edu/abs/2007PhRvC..75a5803J).

For any future simulations where you find the 12C+12C rate is important, it might be interesting to employ the most recent determination, which comes from Tumino et al. (https://www.nature.com/articles/s41586-018-0149-4). Their Figure 3 shows more than two orders of magnitude enhancement relative to the CF88 rate at some places in the thermonuclear regime. They provide their rate in tabular form in their Table 2.  One thing to note if using this rate is that it does not come from a direct measurement and so there is some theory which is involved in going from the experimental data to the final reaction rate. That said, their technique has been validated for other cases (where the indirect and direct measurement data are available), so there is no reason that I’m aware to not trust the results.

Regarding the other reaction rates you mention with regards to “set_rate_”,  there is no need to specify these in the inlist if you plan on using jina_reaclib for these reaction rates and you have opted to use the default reaction rate library (which is jina reaclib). I believe the reason there is an option to set these rates specifically to the one in the JINA reaclib library is that the alternative exists to instead use the NACRE reaction rate library for all of your reaction rates. Presumably one of the MESA developers wanted the easy option for these particular cases to override the NACRE rate and instead use another rate, e.g. from JINA reaclib.

-Zach

----
Zach Meisel
Asst. Prof. of Physics & Astronomy
Ohio University
Email: meisel at ohio.edu
Web: http://inpp.ohiou.edu/~meisel

________________________________
From: Mesa-users <mesa-users-bounces at lists.mesastar.org> on behalf of LAM Yi Hua via Mesa-users <mesa-users at lists.mesastar.org>
Sent: Saturday, July 21, 2018 7:44:34 AM
To: mesa-users at lists.mesastar.org
Subject: [Mesa-users] Using make_o_ne_wd to prepare some white dwarf models

Dear MESA community,

Recently, I'd like to test some reactions associated with nova outbursts using MESA-r10398 and MESA-sdk (revision 245). [Hopefully, older MESA-sdk version won't affect the simulation.] So, most probably, the suitable white dwarf (WD) model for testing the reactions should be ONe WD, but not CO WD.

I have some questions to ask the community.

FIRST step:
I used the "make_o_ne_wd" test_suite to prepare some simple white dwarf models.
For the first test, the simulation of this default model stopped at

model 592
termination code: xa_central_upper_limit

Then, I commented these two lines in the &star_job of inlist_make_o_ne_wd file:

!      xa_central_upper_limit_species(1) = 'ne20'
!      xa_central_upper_limit(1) = 0.35d0

I ran the simulation again, but it stopped at

model 840
stopping because of problems -- too many retries
terminated evolution: nonzero_ierr
termination code: max_number_retries

Perhaps, someone can identify the problem.

Anyway, it is not a big deal. Although the HR diagram shown in Figure 1 of make_o_ne_wd/docs/make_o_ne_wd.pdf could not be reproduced, at least I can confirm that my installation is successful, and a partial HR diagram was produced from log Teff=3.9 up to 5.5 and log L/Ldot = 3.7 up to 4.5. Also, I do not think the problem is due to older MESA-SDK version. After that, I tested the make_o_ne_wd test_suite of MESA-r8118 with MESA-sdk (revision 245). This simulation ran successfully until it terminated with log_L_lower_limit. The produced HR diagram make_o_ne_wd (MESA-r8118) is not exactly the same as the partial HR diagram of MESA-r10398, in particular there is a kink from Teff=3.9 to 4.2 in the HR diagram of make_o_ne_wd of MESA-r8118 during the post-AGB stage. In fact, the HR diagram shown in Figure 1 of make_o_ne_wd.pdf also has that kink.

As long as we can still produce a WD model eventually, shall we change some settings to eliminate/reduce such kink? [Question 1]

Both HR paths produced from MESA-r10398 and MESA-r8118 show some oscillation at the end of post-AGB stage, around Teff=5.1 to 5.4 and log L/Ldot = 4 to 4.4. If we can still produce a WD model eventually, shall we change some settings to eliminate/reduce such oscillation? [Question 2]

SECOND step:
I extend the list of isotopes and reactions stated in co_burn_plus.net<http://co_burn_plus.net> which is used by make_o_ne_wd. The newly extended list includes 78 isotopes (from 1H up to 40Ca) and associated 480 reactions. These 78 isotopes were also included by Politano et al. (1995) and Weiss et al. (1990). Also, I made a few changes to inlist_make_o_ne_wd:

**************************************************************
&star_job
eosDT_use_linear_interp_for_X = .true. < This flag does not exist in MESA-r8118. I compare results produced by MESA-r10398 with and without this flag.
                                       < I found that implementing this linear interpretation increases mass loss.
                                       < However, if I switch it to be .true. for initiall mass 9.5 Mdot, it causes the problem of dt < min_timestep_limit=1d-20 at model 1070.
                                       < Perhaps, this is not the reason causing the problem of dt < min_timestep_limit .
                                       < There is no problem for initial_mass=9.8 or 9.7. Maybe, there are some settings associated with the change of initial mass.

!      change_D_omega_flag = .true.  < These flags do not exist in MESA-r8118. I switch off these flags because they do not smoothen the post-AGB HR path.
!      new_D_omega_flag = .true.     < Bill Paxton said he is "experimenting a new scheme to reduce the noise in the rotational mixing diffusion coefficients
                                     < by doing spatial and temporal smoothing."
                                     < The default flags are .false.

&controls
       max_number_retries = 500                     < I increased it for more attempts solving the hydro-Newton equation before termination.
       max_model_number = 6000                      <

!      xa_central_upper_limit_species(1) = 'ne20'   < I switched off these two flags due to reason mentioned above.
!      xa_central_upper_limit(1) = 0.35d0           <

      min_timestep_limit=1d-20                      <
      log_L_lower_limit = 0                         <

**************************************************************

I did not apply any "stellar engineering" (trick) during mass loss at end-AGB stage when star mass drops from 10 Mdot to 5.5 Mdot.

With defining initial_mass= 10, the simulation stops at log_L_lower_limit = 0 and eventually, the star becomes a white dwarf with 16O and 20Ne rich core, and log_cntr_T=8.08 (center temperature). The mass fraction of 16O is 0.545 from 0 to 1.195 Mdot, whereas the mass fraction of 12C is about 0.427 at the mass coordinate of 1.195 Mdot and drops to around 0.010 at the mass coordinate of 1.175 Mdot. I suppose the small mass region with high fraction of 12C is the so-called C-rich buffer zone separating the bottom of the convective envelope from the WD surface according to some literatures.

If I am not mistaken, to further cool the WD and to make a naked WD, we need to remove this buffer zone first, otherwise the code cannot proceed due to unresolved hydro-Newton equation causing termination of dt < min_timestep_limit=1d-20. I did try to cool the WD (restart the simulation) after the stage of log_L_lower_limit = 0 and log_cntr_T=8.08 (center temperature). But it does not work, it keeps stopping with limit of dt < min_timestep_limit=1d-20, even I tried to change some settings.

I think, to proceed the cooling of a WD to the desired temperature, artificially removing this buffer zone will be more effective than changing various settings of simulation.
To do so, do we just simply change the mass fractions of 12C and other isotopes in the freshly saved *.mod and *.data files to 0 after the simulation stop?  [Question 3]
Do we have to rescale the dq (fraction of total mstar in cell) column?  [Question 4]
And, how about other physical properties recorded in columns of lnd, lnT, lnR, L, omega, etc. do we need to follow some procedure/method to change them?  [Question 5]

Besides, I did try with initial_mass=9.5d0, but the simulation stops at model 1120 due to unresolved hydro-Newton equation and finally it reached dt < min_timestep_limit=1d-20. There is no problem to set initial_mass=9.8d0. Perhaps, there are some settings directly related to the change of initial star mass. May I know the best way to overcome this problem?  [Question 6]

If we would like to use the default EOS and opacity defined in Paxton et al MESA-steering paper [Astrophys. J. Suppl. 192 (2011) 3], do we set the EOS and opacity as what are defined in the default inlist of make_o_ne_wd as below? [Question 7]

      eos_file_prefix = 'mesa'
      kappa_file_prefix = 'gs98'

If this is not the correct way, may I know how to define the default EOS and opacity?  [Question 8]

If I am not mistaken, the default reaction library of MESA-r10398 is JINA REACLIB, do we still need to include these few lines below in inlist file?  [Question 9]
      set_rates_preference = .true.
      new_rates_preference = 2
      set_rate_c12ag = 'jina reaclib'
      set_rate_n14pg = 'jina reaclib'
      set_rate_3a = 'jina reaclib'

May I know the best option for the below reactions associated with 12C? [Question 10]
Is there anyone would like to share her/his experience?

         ! + `'CF88_basic_1212'` : the single rate approximation from CF88.
         ! + `'CF88_multi_1212'` : combines the rates for the n, p, and a channels.
         ! c12(c12,n)mg23, c12(c12,p)na23, and c12(c12,a)ne20 and
         ! uses neutron branching from dayras, switkowski, and woosley, 1976.
         ! + `'G05'` : based on Gasques, et al. Phys Review C, 72, 025806 (2005)

         set_rate_1212 = ''

Thanks for your precious time...
Your responses are very appreciated...

best regards,

Yi Hua
LAM

-- CaTCh EVEry SiNGle OppORtuNItY WhEN U r StILl alIVe --
---- Success Belonged To Those Who's Willing To Strive For It ----
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20180723/7f0cc39a/attachment.html>