[Mesa-users] timestep_hold > model_number leads to infinite timestep retries

[Warrick Ball]

Hi Erica,

I haven't tried to reproduce this but one thing that stands out to me in the terminal output is that the pre-main-sequence model might not be well-enough relaxed.  The default for `pre_ms_relax_num_steps` in `&star_job` has been 300 since at least r15140, so try increasing it to that as a start.

I don't think the `astero` module overwrites `pre_ms_relax_num_steps` (`grep` didn't turn up anything obvious) but if you find it's stuck at 100 even after changing it in `&star_job`, I'll check more closely.

Physically, it looks (I admit from terminal output, not plots) like the issue is appearing as the star hooks towards higher temperature and the bottom of the Hayashi track, where a radiative core is probably trying to form.  You may want to have a look at where these parameters sit relative to the 15 samples that didn't have any trouble, and to look at the output for those samples to see if they also had a problem but managed to push past it.

Cheers,
Warrick

___________

Warrick Ball
Senior Research Software Engineer
University of Birmingham
W.H.Ball at bham.ac.uk

On Fri, 3 Feb 2023, Erica A Sawczynec via Mesa-users wrote:

> Hi All, I have a model that I’ve been struggling to get to converge for a while. The model runs until it reaches 16 samples and then it gets stuck retrying the same time step until eventually the job times out. I turned on the report_why_dt_limits
> hoping that it would help me figure out why the model is getting stuck, but the output is "timestep_hold > model_number, so no timestep increase” which I don’t fully understand. I’ve attached my two inlists and the terminal output where the model
> begins to get stuck. I use MESA version r12115 with the accompanying SDK.
> Thanks!
> Erica
> 
> 
>