[Mesa-users] AGB convergence issue

[Francis Timmes]

> it never made it to the MS

then there are other larger issues.

we'll take this off mesa-users until its resolved.

fxt




> On Mar 6, 2018, at 5:20 PM, Champagne, Arthur E <artc at physics.unc.edu> wrote:
> 
> Hi Frank,
>      I wanted to compare the details of your model, which runs successfully, against my original model so I did what you did - change the initial mass to 5 M_sun and Z to 1e-4. The only other changes were that I set the calculation to stop at  logL=4.3 and turned off rotation. It ran until an age of 7.6e+5 y, after which "hydro_newton_step failed to accept” and it stalled. So it never made it to the MS (and that was the reason for turning off rotation - an earlier try didn’t even make it that far). Now I’m wondering if it’s an installation issue, despite the fact that there were no obvious problems with installation and nothing to suggest that something didn’t install properly. It’s version r10108, for now on a Macbook Pro running Yosemite, i7 processor, 4 cores, set for 8 threads to take advantage of hyper-threading. Xcode and developer tools are installed. Have you run into any Mac-specific (or Mac-user-specific) problems in the past?
>      Thanks,
>      Art
> _________________________________________________________________
>  
> Yeah, he filled his head with notions, seemingly (George Harrison)
> _________________________________________________________________
> A.E. Champagne
> Class of 1989 William C. Friday Professor of Physics
> Dept. of Physics and Astronomy, Univ. of North Carolina
> Office: Phillips 230, (919) 962-7205
> and Director, Triangle Universities Nuclear Laboratory, Duke Univ.
> Office: TUNL 414, (919) 660-2607
> aec at tunl.duke.edu or artc at physics.unc.edu
> _________________________________________________________________
> 
> 
> 
> 
> 
>> On Mar 5, 2018, at 9:14 AM, Francis Timmes <fxt44 at mac.com> wrote:
>> 
>> hi art,
>> 
>> i changed the initial mass to 5 msun and the metallicity of z=1e-4 
>> in the inlists i sent. it ran without issue to the specified stopping condition 
>> of envelope_mass_limit < 1e-2 msun. screen shot attached of the final model.
>> 
>> these inlists are modified versions of the ones used in fields et al 2015
>> which you can grab from http://mesastar.org in the inlist section.
>> 
>> fxt
>> 
>> 
>> 
>> 
>> 
>>> On Mar 2, 2018, at 9:34 AM, Francis Timmes <fxt44 at mac.com> wrote:
>>> 
>>> hi art,
>>> 
>>> for a research project i happen to be looking at ~3 msun zams models
>>> evolved from the pre-ms to a hot co wd with r10108.
>>> 
>>> attached are a snapshot of my current exploratory research-grade inlists 
>>> (i.e, lots of things activated, commented out, … etc as i explore sensitivities). 
>>> see if they work for you when the initial mass is changed to 5 msun.
>>> 
>>> fxt
>>> 
>>> 
>>> 
>>> 
>>> 
>>>> On Mar 2, 2018, at 8:46 AM, Champagne, Arthur E <artc at physics.unc.edu> wrote:
>>>> 
>>>> Hi all,
>>>>     I’m fairly new to MESA - I’m running r10108 and various test models run fine. I’ve constructed a 5 M_sun, Z=1e-4 model, which evolves in a reasonable way to the start of the AGB. The first 2 pulses look good, but it bogs down on the 3rd pulse, which occurs after the He shall narrows. The problem is “hydro_newton_step failed to accept” and debugging indicates that the problem is occurring in the zones that include the edges of the He shell. No surprise. I’ve tried various approaches to improving resolution and convergence, without success. These have come from the inlists on the web page as well as the agb and 13C pocket models in the test suite. The most recent attempt included (from the agb model):
>>>> 
>>>> mesh_dlog_pp_dlogP_extra = 0.25
>>>> mesh_dlog_cno_dlogP_extra = 0.25
>>>> mesh_dlog_3alf_dlogP_extra = 0.225
>>>> mesh_dlog_burn_c_dlogP_extra = 0.225
>>>> mesh_dlog_burn_n_dlogP_extra = 0.225
>>>> mesh_dlog_burn_o_dlogP_extra = 0.225
>>>> mesh_logX_species(1) = 'h1'
>>>> mesh_logX_min_for_extra(1) = -6
>>>> mesh_dlogX_dlogP_extra(1) = 0.25
>>>> mesh_logX_species(1) = 'he4'
>>>> mesh_logX_min_for_extra(1) = -6
>>>> mesh_dlogX_dlogP_extra(1) = 0.25
>>>> xtra_coef_os_above_nonburn = 0.5
>>>> xtra_coef_os_below_nonburn = 0.25
>>>> xtra_coef_os_above_burn_h = 0.25
>>>> xtra_coef_os_below_burn_h = 0.15
>>>> xtra_coef_os_above_burn_he = 0.25
>>>> xtra_coef_os_below_burn_he = 0.25
>>>> xtra_coef_os_above_burn_z = 0.25
>>>> xtra_coef_os_below_burn_z = 0.25
>>>> 
>>>> xtra_dist_os_above_nonburn = 0.1
>>>> xtra_dist_os_below_nonburn = 0.2
>>>> xtra_dist_os_above_burn_h = 0.2
>>>> xtra_dist_os_below_burn_h = 0.2
>>>> xtra_dist_os_above_burn_he = 0.2
>>>> xtra_dist_os_below_burn_he = 0.2
>>>> xtra_dist_os_above_burn_z = 0.2
>>>> xtra_dist_os_below_burn_z = 0.2
>>>> 
>>>>    I have tried playing with the parameters some, but I don’t have a good intuition yet for what they’re doing. So I was wondering if someone who has encountered this problem might have some advice on how to proceed?
>>>>    Many thanks,
>>>>    Art Champagne
>>>> 
>>>> 
>>>> _________________________________________________________________
>>>> 
>>>> Yeah, he filled his head with notions, seemingly (George Harrison)
>>>> _________________________________________________________________
>>>> A.E. Champagne
>>>> Class of 1989 William C. Friday Professor of Physics
>>>> Dept. of Physics and Astronomy, Univ. of North Carolina
>>>> Office: Phillips 230, (919) 962-7205
>>>> and Director, Triangle Universities Nuclear Laboratory, Duke Univ.
>>>> Office: TUNL 414, (919) 660-2607
>>>> aec at tunl.duke.edu or artc at physics.unc.edu
>>>> _________________________________________________________________
>>> 
>>> 
>>> <inlist><inlist_project><inlist_pgstar>
>> 
>> 
>> <pms_to_wd_009127.png>
>> 
>> 
>