[Mesa-users] Strange behavior in 15140 with initial mass decimal places

[Roman Gerasimov]

Dear MESA community,

I would like to report a strange issue I am having with 15140 that I was
unsuccessful in investigating myself, likely due to my extremely poor
understanding of the codebase.

The issue occurs with the attached inlist_project. When running this inlist
on my setup (Ubuntu 18.04, SDK x86_64-linux-21.4.1), after step 9 I get
"retry: max residual jumped" multiple times until my dt falls
below min_timestep_limit and the run terminates with "stopping because of
problems dt < min_timestep_limit".

What's weird about this is that this issue only occurs for this particular
initial mass (0.03130689655172414). If I change the last decimal place to
"5" or reduce the number of decimal places, the issue goes away completely.
I find this extremely puzzling, as I would not have expected MESA to be any
sensitive to the 17th decimal place in any of the input values, let alone
sensitive enough to determine whether a run is going to succeed or crash.

For the time being, I circumnavigate this by setting "max_resid_jump_limit
= 1d7" (an order of magnitude increase over the default).

I am not quite sure if this is a numerical issue in MESA or me being stupid
(the latter is far more likely).

Roman
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20210524/ea9a9681/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: inlist_project
Type: application/octet-stream
Size: 2184 bytes
Desc: not available
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20210524/ea9a9681/attachment.obj>