[Mesa-users] Changing reaction rate issu in PPISN
farag.11 at buckeyemail.osu.edu
Fri Jan 21 00:44:34 UTC 2022
Hello Vaishnav and Harshda,
Perhaps one of the following approaches can help in your debugging!
Most c12ag occurs during and after Helium burning, before Carbon Ignition. This means one could test the effect of changing the c12ag rate by looking at the mass fraction of Carbon in the core before the Ignition of central Carbon. The mass fraction of Carbon in the center cell is reported as "center_c12" in the pgstar and "C_center" in the terminal output.
I have found in my tests that most C12 conversion to O16 occurs in the core between central temperatures of 2d8 - 4d8 Kelvin, and central carbon ignition occurs closer to 8d8 Kelvin. See Figure 13 of https://doi.org/10.3847/1538-4357/ac3130 .
All else equal, two models with different adopted c12ag rates should produce different values of "center_c12" at a central temperature of say 4e8.
One can also check if the rate is being read in and interpolated correctly in MESA by checking MESA's cache version of the rate.
>run a model for 1 timestep with the chosen c12ag rate then
>./show_rates ../../data/rates_data/cache/r_c12_ag_o16_0.bin > rate.txt
>to get a text file of what mesa thinks the rate is
Best of Luck,
From: Mesa-users <mesa-users-bounces at lists.mesastar.org> on behalf of Mathieu Renzo via Mesa-users <mesa-users at lists.mesastar.org>
Sent: Thursday, January 20, 2022 2:52 PM
To: mesa-users at lists.mesastar.org <mesa-users at lists.mesastar.org>
Subject: Re: [Mesa-users] Changing reaction rate issu in PPISN
Hi Vaishnav and Harshda,
I believe you want to set num_special_rate_factors=1 in your inlist.
The exact setup to reproduce Farmer et al. results is available on zenodo: https://zenodo.org/record/3346593<https://urldefense.com/v3/__https://zenodo.org/record/3346593__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_TOekJYmE$>
and https://zenodo.org/record/3559859<https://urldefense.com/v3/__https://zenodo.org/record/3559859__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_Tq34d2cM$> (and I believe https://ui.adsabs.harvard.edu/abs/2022ApJ...924...39M/abstract<https://urldefense.com/v3/__https://ui.adsabs.harvard.edu/abs/2022ApJ...924...39M/abstract__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_Tg7JmYtg$> successfully used these).
Regarding running PPISN at low-Z: I would look at how sharp the core edges are in these models. Once wind-mass loss becomes too small, the cores become sharper and it is hard for pulses to get through them (as I found in models with no-winds at finite Z). If you really want low-Z model an idea could be add some small extra mixing (D_mix = 1d-2 for example) to smooth the edges a bit without messing up the stars too much. N.B: I have not tested this myself.
Hope this helps!
On 1/20/22 2:17 PM, Vaishnav Rao via Mesa-users wrote:
Dear MESA users,
We have been playing around with the PPISN test suite for the past year in an attempt to understand and reproduce the black hole mass gap. We were facing difficulties in incorporating custom reaction rates for the C12-O16 reaction in our test suite.
We have been attempting to reproduce the plots of the effect of CO reaction rates on the final BH mass as in Farmer et al., 2019<https://urldefense.com/v3/__https://arxiv.org/abs/1910.12874__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_TG348ESU$> and Farmer et al., 2020<https://urldefense.com/v3/__https://arxiv.org/abs/2006.06678__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_TQtCZ-pA$> . Using the reaction rates from the files of the latter, we tried to run a modified test suite with recommended settings (which had already run for the default rates for us before). However, upon comparing the final BH masses for 0 sigma and -1 sigma rate tables, there was no major difference, even though there is an expected difference of 8M in the final BH masses. Clearly the lines
special_rate_factor(1)=1.0d0 ! Used for linear scaling of rates
which specify the path to the modified rate tables do not seem to be having any major effect on the simulations.
Our simulations are for metallicities of 1d-3 (in accordance with Farmer et al., 2019<https://urldefense.com/v3/__https://arxiv.org/abs/1910.12874__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_TG348ESU$>). We chose this metallicity because our simulations for 1d-5 ( as in Farmer et al., 2020<https://urldefense.com/v3/__https://arxiv.org/abs/2006.06678__;!!KGKeukY!i0LO4Z6a7wTRPLszCUDHL2R_4MoPOjZ9q6fNlkDxGOAm_0hPMoXyqErxtA4yamXs02_TQtCZ-pA$>) faced multiple convergence issues (mostly ‘hydro_failed’ or too many retries). According to the other posts we saw on the MESA mailing list archives, many others too faced convergence issues of this sort for such low metallicities. We are not sure whether using the same rate tables for different metallicities is correct.
Our inlist, run_star_extras, and rate tables (zip file) have been attached for reference. Any advice on solving this issue would be appreciated.
MESA version: r12778
SDK version: March 25, 2020
OS: Cluster, 64-bit Linux
Thanking you in advance,
Vaishnav V. Rao & Harshda Saxena
Department of Physics
Indian Institute of Technology Bombay
mesa-users at lists.mesastar.org<mailto:mesa-users at lists.mesastar.org>
Flatiron Research Fellow
Center for Computational Astrophysics
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Mesa-users