[Mesa-users] Strange temperature instability near max age

[Bill Paxton]

> On Apr 5, 2019, at 3:33 PM, TUCHOW, NOAH WOLFE <nxt5109 at psu.edu> wrote:
> 
> That does appear to resolve the variability problem (see attached plot). The main reason why I wasn't using it was because I couldn't understand exactly what it was doing based on the description in the documentation. I plotted the time step versus model number, and I think that I have a better understanding of what it is doing now. It appears to be gradually decaying the time step size to the dt_years_for_steps_before_max_age and then running for a specified number of steps.

right.  that turns out to be necessary to avoid the problems you found previously.

> However, I noticed that setting num_adjusted_dt_steps_before_max_age = 200 effectively doubles the computation time required, compared to my previous model. Is this just a trade off for the precision I need, or would there be ways for me to potentially speed up the model without sacrificing much precision?

do the experiment yourself to answer your question.  how small can you make num_adjusted_dt_steps_before_max_age before the results get too ugly for you?  ;)

b


> 
> Thanks,
> Noah Tuchow
> 
> 
> Noah Tuchow
> Penn State University
> Department of Astronomy and Astrophysics
> 532C Davey Lab
> University Park, PA 16802
> 
> 
> On Fri, Apr 5, 2019 at 5:52 PM Bill Paxton <paxton at kitp.ucsb.edu <mailto:paxton at kitp.ucsb.edu>> wrote:
> 
> 
>> On Apr 5, 2019, at 2:39 PM, TUCHOW, NOAH WOLFE <nxt5109 at psu.edu <mailto:nxt5109 at psu.edu>> wrote:
>> 
>> Hello everybody,
>> 
>> I've been running a solar model in MESA r10398 on linux, based off the test suite simplex solar calibration example. Recently I noticed that, near the maximum age of the model, several of the output parameters, such as Teff and log(L), display a strange behavior where they appear to vary substantially from time step to time step (see attached plots). They appear to maintain a general trend, but oscillate very rapidly. As far as I can tell nothing particularly different is happening within the Sun when this behavior starts after around ~4.1 Gyr, and I'm lead to believe that this behavior may be due to some problem in my model rather than an actual feature in the Sun's evolution. 
>> 
>> I though that this may be due to the max_years_for_timestep being too high with a value of 1.0d7 so I halved it, but I still see the wiggles in Teff. I am trying to calibrate a solar model, so I'd like it if the effective temperature of the Sun didn't vary by 10K which is several times the uncertainty in the literature values. Have other MESA users noticed this behavior in the star history outputs prior to the max age? Are there any potential solutions to this instability?
> 
> 
>> Would setting num_adjusted_dt_steps_before_max_age to something other than zero help to alleviate this? 
> 
> excellent idea!
> 
> As you may have noticed, the inlist_simplex_search_controls in the simplex solar calibration test case includes these lines:
> 
>       num_smaller_steps_before_age_target = 200 ! only used if > 0
>       dt_for_smaller_steps_before_age_target = 1d6 ! 1/10 age_sigma
> 
> those are addressing exactly the issue you have noticed.
> 
> since you are not running your model using that inlist, try setting these
> 
> dt_years_for_steps_before_max_age
> num_adjusted_dt_steps_before_max_age
> 
> give it a try with num_adjusted_dt_steps_before_max_age = 200 to match the simplex calibration inlist.
> 
> let us know if that helps.
> 
> b
> 
> 
> 
> 
>> 
>> I'm attaching my inlists, the model file I'm loading from, and two plots of this behavior if anyone would like to replicate it. My model uses diffusion so it might take a few minutes to run. I know that in the Astero and Solar calibration modules, MESA can acheive the precision I'm looking for, but I'm intentionally doing this outside of those modules.
>> 
>> Does anyone have any suggestions as to how I can fix this?
>> 
>> Thanks,
>> Noah Tuchow
>> 
>> Noah Tuchow
>> Penn State University
>> Department of Astronomy and Astrophysics
>> 532C Davey Lab
>> University Park, PA 16802
>> 
>> <init_1Msun.mod><inlist><inlist_1msun><Teff_variability_1e7dt.png><Teff_variability_5e6dt.png>_______________________________________________
>> mesa-users at lists.mesastar.org <mailto:mesa-users at lists.mesastar.org>
>> https://lists.mesastar.org/mailman/listinfo/mesa-users <https://lists.mesastar.org/mailman/listinfo/mesa-users>
>> 
> 
> <Teff_variability_resolved .png><time_step_plot.png>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20190405/675bd094/attachment.html>