[mesa-users] creating preSN model

Bill Paxton paxton at kitp.ucsb.edu
Fri Aug 26 12:11:41 EDT 2011


On Aug 25, 2011, at 4:58 PM, Tuguldur Sukhbold wrote:

> - why this slight difference in numerical values of Y and Z making this difference in the runs?
> - what should I do to prevent these extremely small timesteps during the Si burning phase?
> Please enlighten me! I am a new graduate student and am very confused with this situation.

in my experience, stellar evolution codes are confusing, disappointing, and depressing.
so my 1st suggestion is that you change research areas.

if in spite of that, you decide to continue, you should at least change your expectations:
expect the code to fail and be amazed and delighted if it somehow manages to work.

at this moment I am trying to fix a typical problem with mesa -- I have all of the test cases running
fine on my mac with ifortran and gfortran, but when I run them on a linux machine, there is one
subcase of one of the tests that fails; all the rest are fine.  same code, same data, same inlists,
but one out of a huge set of tests fails for no apparent reason.   I'll make some random changes
until something makes the problem go away, but i'm not likely to ever understand what is going on.

for your particular question about massive stars, my only suggestion to is start from something
that works and make small changes to see if you can find other things that work.  but expect
that the small changes will trigger some nasty non-linearity that causes things to break.
then redefine your research topic to focus on the things that work and skip those that don't.
if you fix your topic 1st and expect to force the code to cooperate, you're doomed.

you might start from the 15M_si_burn test case in star/test_suite.  It should run to onset of core collapse.
Once you get that to work, change it to start from ZAMS instead of from a saved model.
If that works, try changing the mass -- I've been able to make 15, 20, and 25 work with Z=0.02;
outside of that range, I've run into problems and haven't put in the time to search for a solution.
then try changing Z -- and try changing mixing (overshoot/semiconvection/rotational) --
and try changing mass loss (superEddington/supersonic/rotationally enhanced/...).

if you manage to make progress on this, please let me know how you've done it!

good luck,

More information about the Mesa-users mailing list