[Mesa-users] Model terminating after main sequence?

[Bill Paxton]

> On May 28, 2020, at 5:08 PM, Francis Timmes via Mesa-users <mesa-users at lists.mesastar.org> wrote:
> 
> i invite bill to pile on about misuse of varcontrol_target.

just a quick comment:  varcontrol_target is useful as long as the value is well above the floor of changes that is set by the fact that the newton solver is only providing approximate solutions.  the approximations will give a small level of step-to-step variation that cannot be reduced by smaller timesteps.  if you have a small value of varcontrol_target (< 1d-4 for example) and you are getting tiny timesteps all reporting varcontrol as the reason in the terminal output, then it is time to increase varcontrol_target and find more specific timestep controls that apply to things that are derived from the basic variables such as nuclear burning rates.  those items will be much more sensitive than restricting yourself to the basic variables (logRho, logT, logR …) that are considered in varcontrol.

the next public release will have a general time step control analogous to the general mesh control mesh_delta_coeff.  until then, you’ll need to selectively set the controls that matter for your problem.  the terminal output includes a phrase in the far right of the last line such as “lgT_cntr”.  If you search in controls.defaults for that, you’ll find that the corresponding control is delta_lgT_cntr_limit and that sets the limit for magnitude of change in log10 temperature at center.  So you will get smaller timesteps if you reduce that control from its default setting of 0.01.

well, i’ve gone beyond a “quick comment” so i’ll stop here.

cheers,
bill


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20200528/fda2e707/attachment.htm>