[mesa-users] overshooting problem

[Jakub Ostrowski]

Hi!

I'm struggling with a problem related to overshooting and wonder if someone of you has experienced something similar.

I noticed unexpected behavior of Brunt-Vaisala frequency in my 15 M_S post-MS models (shell hydrogen burning phase). If overshooting is enabled, there is a noise in the region of \mu-gradient and a very high peak in the vicinity of intermediate convective zone which definitely shouldn't be there . The larger the value of overshooting the more visible noise becomes. Everything is alright if overshooting is disabled. The attached propagation diagram shows this behavior (only region of \mu-gradient is shown). There are two models: the first one without overshooting (logTeff = 4.1860, logL = 4.6080) and the second one with \alpha_ov = 0.1 (overshoot_step_fraction = 1.0, logTeff  = 4.1836, logL = 4.6738).
-------------- next part --------------
A non-text attachment was scrubbed...
Name: propag.pdf
Type: application/pdf
Size: 36119 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20130628/a846da7a/attachment.pdf>
-------------- next part --------------


I'm also attaching runs of \mu and hydrogen and helium profiles in the same region of logT. This large peak is clearly related to this "break" in \mu and abundance profiles. This is no discontinuity, there are points within this "break". Without overshooting profiles are nice and smooth. To be honest I would expect the opposite - models with overshooting enabled should be smoother, but this is not the case here.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: mu.pdf
Type: application/pdf
Size: 41302 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20130628/a846da7a/attachment-0001.pdf>
-------------- next part --------------

-------------- next part --------------
A non-text attachment was scrubbed...
Name: abundance.pdf
Type: application/pdf
Size: 57633 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20130628/a846da7a/attachment-0002.pdf>
-------------- next part --------------


I have no idea what's going on here. The presented models are calculated with Ledoux criterion, \alpha_mlt = 1.8, \alpha_sc = 0.0. There are more details in attached inlist. As expected, the problem starts during main sequence evolution. I attached propagation diagram for two MS models (\alpha_ov = 0.1, X = 0.357 and X = 0.136). The noise is clearly visible as well as the peak in the more evolved model. There is also very thin intermediate convective zone in the second model, which is rather unreal in my opinion.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: propag-ms.pdf
Type: application/pdf
Size: 42480 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20130628/a846da7a/attachment-0003.pdf>
-------------- next part --------------


I tried probably everything from mixing settings in order to fix it but I didn't succeed. I also reset the inlist - I mean I left at the default value almost every control not directly related to this particular problem and still I couldn't fix it. The presented results are obtained with the steep overshooting. If exponential overshooting is used the noise is a little bit smaller but it's still there as well as all of these sharp features of abundance profiles. I use MESA 5118 with the latest SDK on OS X 10.8.4.

Have someone of you experienced similar problem or do you have some clues how to explain or fix such behavior? 

Cheers,
Jakub


-------------- next part --------------
A non-text attachment was scrubbed...
Name: inlist_basic_15m_ov01
Type: application/octet-stream
Size: 19345 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20130628/a846da7a/attachment.obj>