[Mesa-users] Problem with 'prepare_for_new_try: bad num omega error

[Ian Foley]

Thanks Francis,

I have tried a number of mesh_coeff values that give anywhere from 2000 to 6600 cells. The error still occurs.

I will try plots with rotational quantities.

There are 48 isotopes which is a minimal net that will run to core collapse in 14M and 15M simulations. The reason is because I am running in a docker container with windows pro and 8 GB memory. Net’s above 80 isotopes cause significant performance issues.

 It runs fine until I hit these complex situations near core collapse when the abundance of si28 is low and fe26 high. It occurs with both my 13M and 12M runs where the off-centre burning occurs and only after many models of it. Central degeneracy is about 30 and the inner 0.6M seems to balance the pressure from above, but the temp and mass is sufficient for the off-centre burning to proceed. In the png files attached, the off-centre temp is about logT = 9.55 at the off-centre boundary.

Another factor which may be important is that the energy at the off-centre cell can vary hugely. Thus, quite often , when logdt -10 and less, log energy can vary from 15 to 18 in 1 timestep - that’s more than 1000 times in less than 1/1000 of a second. So I suspect the computational processes are being tested to their limit.

Turning rotation off seems to prevent the error, but I don’t really want to do that as a general work around. Rotation can be important in massive stars. I only did it for a few hundred models.

Kind regards
Ian

> On Fri, 31 May 2019 at 10:29 am, Francis Timmes <fxt44 at mac.com> wrote:
> since the plots do not appear to show rotational quantities, 
> perhaps plot the profiles of some at several timepoints to help 
> convince oneself there are, in fact, no large values.
> 
> i see there are ~6600 cells. how many isotopes?
> 
> the evolution of the core will be much faster than the angular momentum transport timescale.
> as an experiment, one could turn off rotation or the transport.
> 
> fxt
> 
> 
> 
> 
> 
> > On May 30, 2019, at 12:29 AM, Ian Foley via Mesa-users <mesa-users at lists.mesastar.org> wrote:
> > 
> > Hi,
> > 
> > I'm having problems with the following error killing my run
> > 
> > stop 'prepare_for_new_try: bad num omega'. 
> > 
> > Checking the source code in evolve.f90 tells me that one or more cells has an angular momentum (omega(k)) > 1d50. Examining the profile output near and just before the error shows that omega, i_rot and j_rot has no values that are near or could cause a value of 1d50.
> > 
> > It has occurred about 10 times while evolving a 12M star in the si28 burning
> > stage and beyond. inlist has z=0.02, no overshooting, blocker=0.7 for mass loss and low rotation. Timesteps are varying between logdt -6 to -13. Most of si28 has burned off-center and fe56 abundance is up around 0.6.
> > 
> > Without supplying any inlists etc but can if needed, my question is what to do about this error to prevent it occurring. I started the run with only low rotation. I really don't understand what is causing it. mess_coeff=1.5 and there are added zones around the si28 and fe56 burning regions. At this stage the error is blockiing further progress.
> > 
> > At this stage I am asking if there is some general advice that can be given. If there are specifics needed with my actual run, I can supply the necessary files.
> > 
> > I have attached 4 png files which will help to capture the context.
> > 
> > I am fascinated by the possibility that core collapse may occur off-center. Can't find anything about that in the literature. It's assumed that the off-center burning will migrate to the centre, but there's no evidence of it in this run and similar behavior at the 13M level - also not completed.
> > 
> > Thanks in advance.
> > ian 
> > 
> > 
> > <abund_012940.png><conv_012940.png><profile_Panels3_012940.png><grid6_012940.png>_______________________________________________
> > mesa-users at lists.mesastar.org
> > https://lists.mesastar.org/mailman/listinfo/mesa-users
> >
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20190531/9565506b/attachment.html>